Biosimilars for the Treatment of Cancer: A Systematic Review of Published Evidence

Transcription

1 BioDrugs (2017) 31:1 36 DOI /s SYSTEMATIC REVIEW Biosimilars for the Treatment of Cancer: A Systematic Review of Published Evidence Ira Jacobs 1 Reginald Ewesuedo 2 Sadiq Lula 3 Charles Zacharchuk 2 Published online: 11 January 2017 The Author(s) This article is published with open access at Springerlink.com Abstract Background Biologic treatments for cancer continue to place a significant economic burden on healthcare stakeholders. Biosimilar therapies may help reduce this burden through cost savings, thereby increasing patient access. Objectives The purpose of this study was to collate all published data to assess the weight of available evidence (quantity and quality) for proposed monoclonal antibody biosimilars and intended copies, for the treatment of cancer. Methods MEDLINE, Embase, and ISI Web of Science databases were searched to September Conference proceedings (17) were searched (2012 to July 2015). Searches of the United States National Library of Medicine ClinicalTrials.gov registry were also conducted. Risk of bias assessments were undertaken to assess data strength and validity. Results Proposed biosimilars were identified in 23 studies (36 publications) in oncology and ten studies in 14 publications in oncology and chronic inflammatory diseases for bevacizumab, rituximab, and trastuzumab originators. Based on our review of the included published studies, and as inferred from the conclusions of study authors, the identified proposed biosimilars exhibit close similarity to their originators. Published data were also retrieved on intended copies of rituximab. It remains unclear what role these agents may have, as publications on rigorous clinical studies are lacking for these molecules. Conclusion While biosimilar products have the potential to improve patient access to important biologic therapies, robust evidence of outcomes for monoclonal antibody biosimilars in treating cancer patients, including data from comparative efficacy and safety trials, is not yet available in the published literature. Significant data gaps exist, particularly for intended copies, which reinforces the need to maintain a clear differentiation between these molecules and true biosimilars. As more biosimilars become available for use, it will be important for stakeholders to understand fully the robustness of overall evidence used to demonstrate biosimilarity and gain regulatory approval. Electronic supplementary material The online version of this article (doi: /s ) contains supplementary material, which is available to authorized users. & Ira Jacobs ira.jacobs@pfizer.com Pfizer Inc, Pfizer Essential Health, 235 East 42nd Street, New York, NY , USA Pfizer Inc, Biosimilars Development, Cambridge, MA, USA Envision Pharma Group, London, UK

2 2 I. Jacobs et al. Key Points Monoclonal antibody drugs account for a significant proportion of oncology spending in the USA and are associated with high out-of-pocket costs for patients. Biosimilar therapies have the potential to improve access to these specialist oncology drugs, but knowledge gaps may slow their adoption. The degree of biosimilarity is ultimately determined by regulatory authorities and is based on the totality of evidence, which includes data on molecular and functional characterization, other nonclinical data, and the safety, pharmacokinetic, immunogenicity, and efficacy clinical trial data. Based on this review of published nonclinical and clinical oncology studies, and as inferred from the conclusions of study authors, proposed biosimilars of bevacizumab, rituximab, and trastuzumab exhibit close similarity to their originators. However, at present, robust evidence of outcomes for monoclonal antibody biosimilars in cancer, including data from comparative efficacy and safety trials, is not yet widely available in the published literature. 1 Introduction The treatment of cancer continues to place a significant burden on healthcare systems, with the number of cancer cases continuing to rise due to an aging population. Improvements in cancer diagnosis and disease management are now extending survival, and consequently increasing the length of time patients remain on treatment. As a result, there is a need to control current levels of expenditure, which are unsustainable. IMS Health recently reported a snapshot of USA expenditure on cancer medicines: Global spending on oncology and supportive care drugs reached $100 billion in 2014, with targeted therapy expenditures accounting for almost 50% of total spending [1]. Spending on oncology medicines in the USA increased 18.0% to $39.1 billion in 2015 [2]. The fastest-growing classes of oncology therapy are monoclonal antibodies (mabs) and protein kinase inhibitors; mabs account for 35% of oncology spending due to the introduction of new treatments [2]. USA sales figures in 2015 for two of the top 20 global products were $6.2 billion for bevacizumab and $5.6 billion for trastuzumab [3]. Given the economic burden of cancer treatments, healthcare systems around the world have devised a range of methods to try to contain these costs, often resulting in seemingly arbitrary access restrictions for patients. Patient access to oncology medicines has been shown to vary significantly even at the regional level [4]. A lack of consensus among healthcare professionals on the most reliable economic drug evaluation methods to employ has led to inconsistency in treatment guidelines. This was demonstrated in a 2015 systematic review by Park et al. that examined the cost-effectiveness of mabbased orphan drugs [5]. Patient access and reimbursement decisions can vary greatly between regions as a consequence of the different evaluation methods employed by each agency [5]. In the USA, patient out-of-pocket costs for intravenous cancer drugs have increased substantially in recent years, in part due to the integration of small community-based practices into larger hospital systems [1]. Across publicly funded healthcare systems in Europe and other parts of the world, a lack of reimbursement also may limit access to effective oncology medicines, with reimbursement often contingent upon evidence of cost effectiveness. Biosimilars are biologics that are highly similar to biologics already approved for the treatment of disease. The first biosimilar was authorized for use in the EU in A greater adoption of biosimilars may help to alleviate the substantial burden on healthcare systems by stimulating price competition and improving patient access to important treatments [6]. Regulatory frameworks for the development of biosimilars were first established by the European Medicines Agency (EMA), followed by the World Health Organization (WHO) and the USA Food and Drug Administration (FDA) [7]. These regulatory frameworks specify the requirements for approval of biosimilars, including important foundational analytical studies to compare the biosimilar with the approved biologic originator product. In addition, comparative nonclinical and clinical studies are required to assess toxicity and pharmacokinetics (PK)/ pharmacodynamics (PD), and clinical studies are required to demonstrate an efficacy profile that is comparable to the originator, as well as comparable safety and immunogenicity profiles. Robust evidence of similarity provided from analytical, PK, and nonclinical studies form the foundation to demonstrate the comparability of a biosimilar to the originator and are required to meet regulatory standards and requirements for regulatory approval. It is also important to ensure that these data are made available in

3 Biosimilars for Cancer: A Systematic Review of Published Evidence 3 the public domain, to facilitate awareness and understanding of biosimilar treatments among physicians and other healthcare professionals. Intended copies are copies of originator biologics that have not undergone rigorous comparative evaluations as stipulated by major regulatory agencies, but are nevertheless being commercialized by manufacturers in some countries. There is a lack of published information about the efficacy and safety of intended copies compared with the originator. Furthermore, these products may have clinically significant differences in formulation, dosages, efficacy, and safety [8]. A comprehensive systematic literature review (SLR) was undertaken in 2015 to identify, collate, and synthesize all published evidence on named biosimilars and intended copies of originator mabs and fusion proteins [9]. The aim of that analysis was to summarize the quantity and quality of data available and the number and diversity of publications describing biosimilars for mabs or fusion proteins across all indications (including chronic inflammatory diseases, oncology, cardiovascular and ophthalmology). Here, we explore the findings for biosimilars indicated for oncology disease in more detail. 2 Methods 2.1 Systematic Literature Review (SLR) A detailed description of the methods used in this SLR can be found in the manuscript by Jacobs et al. [9]. MEDLINE /Medline in process and Embase (searched using the OVIDSP interface), and ISI Web of Science were searched from database inception to September 3, The search was executed on April 27, 2015 and repeated on September 3, 2015 to capture more recent full-text publications. The search strategy consisted of the following: (1) terms that captured mab and fusion protein terms; and (2) terms that included the different terminologies for biosimilar products, such as biosimilars, subsequent entry biologics, follow-on biologics, follow-on proteins, biocomparables, biogenerics, similar biotherapeutic products, and intended copies, or biobetters (which were analyzed separately). Included publications were required to contain both a mab and/or fusion protein term and a biosimilars term. Proposed biosimilars were differentiated from intended copies, based on them meeting the established rigorous regulatory requirements for biosimilarity, as outlined by major regulatory health authorities such as the EMA, FDA, WHO, Pharmaceuticals Medical Devices Agency/Japan Ministry for Health Labor and Welfare, Health Canada or Korean Ministry of Food and Drug Safety (MFDS). Other markets have issued guidance on biosimilars, although the evaluation of the biosimilar approval pathways by regulatory authorities outside of the major markets is considered beyond the scope of this review. Controlled vocabulary and free-text terms were used, and the search results were filtered using the study designs of interest. Search results from each database were limited to references published in the English language. In addition, a hand-search of relevant conference proceedings (17 conferences) was conducted for the period January 1, 2012 to July 31, 2015 in order to capture the latest studies not yet published as full-text articles and/or supplement results of previously published studies (Jacobs et al. [9]). Oncology-focused conference proceedings (n = 10) are shown in Supplementary Table S1 (see the electronic supplementary material, online resource 1). For the SLR analysis, conference proceedings of interest included disease-specific (i.e., for oncology), health economics and outcomes research, regulatory-/payer-focused, and manufacturing-/development-themed meetings. In order to identify biosimilars in development that did not appear in the published literature or in the identified congresses, searches were also conducted (on September 21, 2015) using the USA National Library of Medicine ClinicalTrials.gov registry. Hand-screening was used to identify relevant records because of the limited extent of the searches available for ClinicalTrials.gov. 2.2 Components of the SLR This SLR had two components: the empirical analysis, which focused on peer-reviewed publications of analytical, nonclinical, clinical, pharmacovigilance, and observational empirical data; and the non-empirical analysis, which included opinion pieces or commentaries, publications describing product-related patient support programs, and those on manufacturing and supply issues, which were further classified into general thematic categories to summarize key topics being published on biosimilars. These two components were included to assess the diversity and extent of these types of publications, as well as to identify emergent themes and knowledge gaps in the published literature. Empirical studies were categorized by type into one of the following areas: human studies subdivided as randomized controlled trials (RCTs), observational/postmarketing studies, and health economic studies; nonclinical (in vitro/in vivo) studies; and analytical studies. Non-empirical publications were classified into one of the following categories: manufacturing or supply topics and themes, review articles, opinion pieces or commentaries, regulatory-/policy-related content, published descriptions of product-related patient support programs, and any other

4 4 I. Jacobs et al. non-empirical publication type relevant to biosimilars meeting the inclusion criteria. 2.3 Risk of Bias (Quality) Assessment A risk of bias assessment was undertaken for each classified study (i.e., RCTs, observational studies, studies published in conference proceedings, and animal studies) using a validated tool matched to study type to assess the strength/validity of the empirical data and in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [10, 11]. In cases where multiple publications were retrieved for the same study, quality assessments were only conducted on the first original publication or first full-text publication. The quality of RCTs was assessed using the National Institute for Health and Care Excellence (NICE) single technology appraisal (STA) manufacturer s template [12] and the Jadad scoring system [13] (Supplementary Table S2; Supplementary Fig. S1). Non-randomized studies were assessed using the Downs and Black instrument [14] (Supplementary Table S3). The Downs and Black instrument was modified to include only the most critical qualifying parameters (12 of 26) for quality assessment of conference proceeding abstracts (Supplementary Table S4). Detailed parameters related to process were excluded as these data were not available in abstract formats, e.g., suitability of statistical method employed. Animal studies were assessed using SYRCLE s risk of bias tool [15]. Conference abstracts of analytical/ nonclinical studies were not evaluated, as suitable tools were not available at the time of analysis. Analytical and cell-based studies published full-text were also not evaluated for the same reason. 3 Results 3.1 Literature and Conference Search A total of 768 publications relevant to the topic of biosimilars were retained from a total of 1991 publications identified through a title and abstract screen. Of the 768 references, 147 (19%) reported mab biosimilars for use in oncology. The number of publications included in the analysis is shown in Supplementary Fig. S2. Where encore (or duplicate) publications were retrieved for studies, the information was compared with the original (first published article) and excluded if no additional data were provided. If new data were identified, encore publications were included along with the original publication. However, this did not affect the overall study count. In some instances, a biosimilar was indicated for multiple disorders or diseases. For example, rituximab biosimilar publications were reported for both the oncology and inflammatory disorders categories, where relevant, as rituximab is indicated for both therapeutic areas. Proposed biosimilars or intended copies (i.e., where a unique identifier was provided) were identified in 23 studies (36 publications) in oncology and ten studies in 14 publications in oncology and chronic inflammatory diseases (Fig. 1). In this analysis, biosimilar trials were identified for the following indications: follicular lymphoma, diffuse large B-cell lymphoma (DLBCL), breast cancer (BC), and non-squamous non-small cell lung cancer (NSCLC). 3.2 Preclinical Data Proposed Bevacizumab Biosimilars Published preclinical studies of proposed bevacizumab biosimilars are presented in Table 1. ABP 215 (Amgen) The functional similarity of ABP 215 compared with bevacizumab was assessed in vitro using human umbilical vein endothelial cells (HUVEC) and characterized using analytical methods [16 18]. Binding to vascular endothelial growth factor (VEGF) and VEGF isoforms, neonatal Fc receptor and FcgRIIa, inhibition of proliferation, and composition between the biosimilar and originator were similar. Hutterer et al. investigated the structural similarity of ABP 215 compared with bevacizumab [19]. Based on results of structural assessment (including higher order structure, impurities, and stability), the authors concluded that ABP 215 was analytically highly similar to bevacizumab. PF (Pfizer) The functional similarity (biological activity or mode of action) of PF and bevacizumab were assessed in vitro using cell-based assays [20, 21]. The results of these studies indicated that PF was functionally similar to bevacizumab (in HUVEC and another unspecified cell line). Peraza et al. evaluated the functional similarity of PF and bevacizumab in cynomolgus monkeys [21]. The authors reported that PF was well tolerated and displayed similar PK properties compared with bevacizumab. Anti-drug antibodies (ADAbs) were not detected. The charge heterogeneity, post-translational modifications, and hydrodynamic size heterogeneity of PF compared with bevacizumab were investigated using various biochemical analytical techniques [20]. The biochemical properties were also confirmed using complementary analyses. The authors reported that PF displayed similar structural properties compared with bevacizumab.

5 Biosimilars for Cancer: A Systematic Review of Published Evidence 5 Bevacizumab 0 ABP BCD PF RPH Rituximab 1B8 1 1 BCD GP PF RTXM SAIT Kikuzubam (IC) Reditux (IC) Trastuzumab BCD CT-P6 2 2 FTMB 1 2 PF Nonclinical studies Nonclinical publications RCT studies RCT publications Observational/post-marketing studies Observational/post-marketing publications Analytical studies Analytical publications Fig. 1 Frequency of publications of reported named proposed biosimilars and ICs in oncology. IC intended copy, RCT randomized controlled trial Peraza et al. evaluated the analytical (structural) similarity of PF compared with the originator [21]. The primary sequence of PF , bevacizumab-eu, and bevacizumab-usa was reported to be identical, as delineated by liquid chromatography (LC)/mass spectrometry (MS)/MS peptide mapping. RPH-001 (Alphamab/R-Pharm) Archuadze et al. evaluated the PK profiles of RPH-001 and bevacizumab following a single intravenous administration at three different doses in cynomolgus monkeys [22]. The tested PK parameters were reported to be comparable between RPH-001 and bevacizumab, and neither drug was associated with any toxicity Proposed Rituximab Biosimilars Published preclinical studies of proposed rituximab biosimilars are presented in Table 2. 1B8 (Center of Molecular Immunology) Dorvignit et al. evaluated the functional comparability between 1B8 and rituximab in human Burkitt s lymphoma, Ramos, Daudi, and Raji cell lines [23]. Similar biological potency was reported [as measured by complement-dependent cytotoxicity (CDC), antibody-dependent cellular cytotoxicity (ADCC), and apoptosis assays]. GP2013 (Sandoz) The pharmacological comparability of GP2013 and the originator rituximab was assessed in a preclinical study of moderate quality (SYRCLE s risk of bias tool) [24]. The authors reported similar in vitro ADCC potency when compared in a dose-response manner against two lymphoma cell lines using human natural killer (NK) cells [24]. In a separate analysis, in vivo efficacy was demonstrated in mouse xenograft models [25]. PK/PD (CD20 cell depletion) profiles were reportedly comparable based on analysis in cynomolgus monkeys [26]. Visser et al. reported in vitro functional bioequivalence between GP2013 and the originator rituximab in Raji B cells [27]. The Visser et al. study did not undergo quality assessment, because of the unavailability of suitable risk of bias assessment tools for this type of study. Visser et al. [27] (not assessed) and da Silva et al. [24] (moderate quality using SYRCLE s risk of bias tool) compared GP2013 with rituximab using combined analytical and characterization methods. Intact mass analysis of GP2013 revealed the same molecular mass to that of rituximab [24 26]. The primary sequence and higher order structure of GP2013 was reported by authors as indistinguishable from the originator. GP2013 and originator rituximab were comparable with regard to post-

6 6 I. Jacobs et al. Table 1 Outcomes for proposed bevacizumab biosimilars Study/patients (n) References Outcome/time point Biosimilar a Bevacizumab a Statistical comparison Quality assessment rating ABP 215 RCT PK/safety study/nhv (NR) Nonclinical CA [16]; CA [17]; CA [18] Safety, single dose Modified D&B: All AEs, % 22.1 USA: 16.4; EU: 22.4 excellent ADAb, % 0 0 Score: 10/12 PK/PD, single dose AUC, lg h/ml 29,400 USA: 29,600; EU: 30,600 Ratio: NR ( ) b C max, single dose, lg/ ml Functional assessment (in vitro) 87.2 USA: 89.1; EU: 84.7 Ratio: NR ( ) b Binding to VEGF, pm 117 USA: 126; EU: 112 Binding to FcRn, ratio USA: ; EU: % Binding to FcgRIIIa, USA: 85 93; EU: ratio % Inhibition of USA: 91 96; EU: proliferation, ratio % Autophosphorylation, lg/ml Anti-tumor activity Similar Similar Nonclinical (cell based)/analytical CA [19] Functional assessment Highly similar Highly similar Not evaluated c (in vitro) Target binding NR NR Effector function NR NR Binding to FcRn NR NR Composition: Highly similar Highly similar Primary structure NR NR Post-translational mod NR NR Carb structure NR NR Higher order structure NR NR Aggregates NR NR Impurities NR NR Stability NR NR

7 Biosimilars for Cancer: A Systematic Review of Published Evidence 7 Table 1 continued Study/patients (n) References Outcome/time point Biosimilar a Bevacizumab a Statistical comparison Quality assessment rating BCD-021 RCT PK/safety study/nsclc (28) RCT Comparative safety/efficacy study/ NSCLC (138) PF Nonclinical (cell based)/analytical CA [20] Functional assessment (in vitro) CA [45] Safety, 3 wk Modified D&B: All AEs, % NR NR p = NS excellent PK/PD, 3 wk Score: 10/12 Single dose, NR NR Ratio: AUC 0 504h AUC, NR NR NR p = NS Half-life, NR NR NR p = NS T max,nr NR NR p = NS CA [46] Efficacy, 18 wk Modified D&B: ORR, % (95% CI) ( ) ( ) p = NS excellent CR, % p = NS Score: 11/12 PR, % p = NS Stable disease, % p = NS Progressed, % p = NS Safety, 18 wk All AEs, % NR NR p = NS Serious AEs, % NR NR p = NS Transient ADAbs, 1 (NR) 1 (NR) n (%) Similar Similar Not evaluated c Biological activity NR NR Mode of action NR NR Composition Similar Similar Peptide mapping NR NR Post-translational NR NR modifications Biochemical NR NR Separation Similar Similar Size heterogeneity NR NR Charge heterogeneity NR NR

8 8 I. Jacobs et al. Table 1 continued Study/patients (n) References Outcome/time point Biosimilar a Bevacizumab a Statistical comparison Quality assessment rating Nonclinical (cynomolgus monkeys and cell based)/analytical RPH-001 Nonclinical (cynomolgus monkeys and cell based) CA [21] Functional assessment (in vitro) Inhibition of VEGF Similar Similar binding PK/PD (monkey), 1 mo AUC, ratio % C max, ratio % Mortality Well tolerated Well tolerated Clinical parameters Incidence and severity of side Incidence and severity of side Clinical pathology effects were similar effects were similar Toxicology Histopathology ADAb Not detected Not detected Composition Amino acid sequence Identical Identical CA [22] PK/PD (monkey), single dose AUC last, NR Similar Similar C max, NR Similar Similar Clearance, NR Similar Similar Volume of distribution, Similar Similar NR T max elimination rate, Similar Similar NR Toxicology Not associated Not associated Not evaluated c Not evaluated c ADAb antidrug antibody, AE adverse event, AUC area under the curve, AUC 0 504h area under the curve from 0 to 504 h, AUC area under the curve, AUC last area under the curve to the last measurable concentration, CA conference abstract, CI confidence interval, Cmax maximum serum concentration, CR clinical remission, D&B Downs and Black (tool), FcRn neonatal Fc receptor, mo month(s), NHV normal healthy volunteers, NR not reported, NS not significant, NSCLC non-small cell lung cancer, ORR overall response rate, PD pharmacodynamics, PK pharmacokinetics, PR partial remission, RCT randomized controlled trial, T max time that the drug is present at its maximum concentration in the serum, VEGF vascular endothelial growth factor, wk week(s) Qualitative data for biosimilarity as stated by the corresponding study authors a b 90% CIs shown in parentheses Quality assessment not conducted, because of the absence of validated tools specific for the study type, at the time of analysis c

9 Biosimilars for Cancer: A Systematic Review of Published Evidence 9 Table 2 Outcomes for proposed rituximab biosimilars Study type/patients (n) References Outcome/time point Biosimilar a Rituximab a Statistical comparison Quality assessment rating 1B8 Analytical/nonclinical (cell based) BCD-020 RCT Comparative PK/safety/efficacy study/nhl (92) [23] Functional assessment (in vitro) Not evaluated d CA [47]; CA [48]; CA [49] ADCC, binding Same Same CDC, potency Similar Similar Efficacy, day 50 ± 5: Modified D&B: excellent ORR, % p = Score: 9/12 Biologic experienced, ORR, % p = 1.00 Biologic naïve, ORR, % p = 1.00 CR, % p = CR, unconfirmed, % p = 1.00 PR, % p = Stable disease, % p = 1.0 Progressed, % p = CD20 CD19 count Undetectable Undetectable p [ 0.05 CD19 count, time Depleted \1 wk Safety, day 50 ± 5 Depleted \1 wk All AEs, n Treatment AEs p = Serious AEs ADAb, % 0 2 PK/PD AUC h, 4 doses, ratio % NR NR Ratio: NR ( ) b NS AUC 0 168h, 1 dose, ratio % NR NR Ratio: NR ( ) b

10 10 I. Jacobs et al. Table 2 continued Study type/patients (n) References Outcome/time point Biosimilar a Rituximab a Statistical comparison Quality assessment rating GP2013 Analytical/nonclinical (cynomolgus monkeys, mouse xenograft models, and cell based) [24]; CA [25]; CA [26] Functional assessment (in vitro) SYRCLE s risk of bias: ADCC Overlapping Overlapping NS moderate Efficacy (mouse model) Similar Similar 12 unclear, 10 low risk Tumor growth inhibition, 3 mg, SU- DHL Tumor growth inhibition, 30 mg, SU- DHL Ratio: 1.07 ( ) c Ratio: 1.08 ( ) c Tumor growth inhibition, 0.1 mg, Jeko-1 Ratio: 1.06 ( ) c Tumor growth inhibition, 0.3 mg, Jeko-1 Ratio: 0.95 ( ) c Safety ADAb, d [9 [9 PK/PD (monkey), single/multiple doses AUC (%) NR NR Ratio: NR ( ) b C max, lg/ml 13% lower 13% higher CD20 levels (%) NR NR Ratio: NR ( ) c Nonclinical (cell based)/analytical Composition Similar Similar Glycan quantification NR NR Charge variation NR NR Specific amino acid modifications NR NR Size heterogeneity NR NR [27] Functional assessment (in vitro) Bioequivalent Not evaluated d Cell-based competitive binding, ratio % p \ ADCC, ratio % p \ CDC, ratio % p \ Apoptosis, ratio % p \

11 Biosimilars for Cancer: A Systematic Review of Published Evidence 11 Table 2 continued Study type/patients (n) References Outcome/time point Biosimilar a Rituximab a Statistical comparison Quality assessment rating PF Nonclinical (cynomolgus monkeys and cell based)/analytical Rituximab-EU [29]; CA [28]; CA [30] Functional assessment (in vitro) SYRCLE s risk of bias: CDC Similar Similar moderate Safety (monkey) Similar Similar 9 unclear, 13 low risk ADAb, single dose ADAb, repeated dose, day 22, % ADAb, repeated dose, day 30, % 7 14 ADAb, repeated dose, day 114, % PK/PD (monkey) AUC, 2 mg/kg, lg h/ml, mean (SD) 4720 (966) 4940 (890) AUC, 10 mg/kg, lg h/ml, mean (SD) 34,700 (8650) 37,100 (6010) AUC, 20 mg/kg, lg h/ml, mean (SD) 64,000 (14,600) AUC, repeated dose, day 1, lg h/ml, mean (SD) AUC, repeated dose, day 22, lg h/ml, mean (SD) 56,800 (15,400) 53,200 (18,700) 51,700 (11,900) 54,600 (8800) 79,500 (39,900) C max, 2 mg/kg, lg/ml, mean (SD) 74.0 (15.5) 80.3 (8.0) C max, 10 mg/kg, lg/ml, mean (SD) (70.4) (62.2) C max, 20 mg/kg, lg/ml, mean (SD) (198.0) (138.0) C max, repeated dose, day 1, lg/ml, mean (SD) (241.0) (292.0) C max, repeated dose, day 22, lg/ml, mean (SD) (118.0) (313.0) T max, 2 mg/kg, h, mean (SD) 1.22 (2.37) 0.08 (0.0) T max, 10 mg/kg, h, mean (SD) 0.24 (0.37) 0.24 (0.37) T max, 20 mg/kg, h, mean (SD) 0.24 (0.37) 0.39 (0.47) Half-life, repeated dose, h, mean (SD) 0.70 (1.57) 0.35 (0.43) Half-life, repeated dose, h, mean (SD) 0.48 (0.47) 0.67 (0.46) Tmax Similar Similar Reduction in splenic weight, % Single dose CD3 - CD20?, 2, 10, 20 mg, day 4, % 99.3, 99.7, 99.2 CD3 - CD20?, 2, 10, 20 mg, day 15, % 92.1, 97.4, 98.7 CD3 - CD20?, 2, 10, 20 mg, day 92, % 27.7, 28.7, , 99.5, , 99.7, , 18.3, 24.1

12 12 I. Jacobs et al. Table 2 continued Study type/patients (n) References Outcome/time point Biosimilar a Rituximab a Statistical comparison Quality assessment rating Nonclinical (cell based)/analytical RTXM83 RCT PK/safety study/dlbcl (24) CD3 - CD20? CD40?, 2, 10, 20 mg, day 4, % 100, 100, , 100, 100 CD3 - CD20? CD40?, 2, 10, 20 mg, day 15, % 91.1, 97.0, , 99.2, 95.0 CD3 - CD20? CD40?, 2, 10, 20 mg, day 92, % 29.1, 8.3, , -15.5, 18.8 CD3 - CD19?, 2, 10, 20 mg, day 4, % 86.3, 90.2, , 85.2, 87.1 CD3 - CD19?, 2, 10, 20 mg, day 15, % 81.0, 85.6, , 88.7, 83.1 CD3 - CD19?, 2, 10, 20 mg, day 92, % 9.8, 2.1, , -5.2, -8.4 CD3 - CD40?, 2, 10, 20 mg, day 4, % 85.9, 91.4, , 87.3, 88.3 CD3 - CD40?, 2, 10, 20 mg, day 15, % 81.0, 84.8, , 88.2, 88.0 CD3 - CD40?, 2, 10, 20 mg, day 92, % 25.3, 1.5, , -30.0, 5.3 Repeated dose CD3 - CD20?, day 4, 15, 121, % 100, 99.9, , 99.7, 80.9 CD3 - CD20? CD40?, day 4, 15, 121, % 100, 99.9, , 99.4, 80.5 CD3 - CD19?, day 4, 15, 121, % 96.1, 90.3, , 89.6, 84.0 CD3 - CD40?, day 4, 15, 121, % 97.3, 97.3, , 96.8, 77.1 Composition Tryptic peptide mapping Superimposable Superimposable CA [31] Functional assessment (in vitro) Not evaluated d Biologic activity, ratio % USA/EU: Composition Similar Similar Proteolytic peptide mapping Similar Similar Glycan quantification Similar Similar Purity NR NR Charge heterogeneity NR NR Major post-translational modifications NR NR Hydrodynamic size heterogeneity NR NR High molecular mass species, % USA/EU: CA [50] Safety, 18 wk Modified D&B: good ADAb, % 0 0 Score: 7/12 PK/PD, 18 wk AUC, lg h/ml Similar Similar C max, lg/ml Similar Similar Half-life, h Similar Similar

13 Biosimilars for Cancer: A Systematic Review of Published Evidence 13 Table 2 Outcomes for proposed rituximab biosimilars Study type/patients (n) References Outcome/time point Biosimilar a Rituximab a Statistical comparison Quality assessment rating Nonclinical (cynomolgus monkeys and cell based)/analytical Rituximab-EU SAIT101 RCT Comparative PK/safety/efficacy study/dlbcl (24) CA [32] Functional assessment (in vitro) Not evaluated d ADCC, binding Similar Similar CDC, potency Similar Similar PK/PD (monkey), NR AUC, ratio % C max, ratio % Half-life, repeated dose, ratio % CD20 and CD40 depletion Similar Similar Composition Peptide mapping Similar Similar Glycan quantification Similar Similar Charge variant Similar Similar CA [51] Efficacy, up to 8 cycles Modified D&B: excellent Cycle 2, CR, %, n = Score: 9/12 Cycle 2, PR, %, n = Cycle 6, CR, %, n = Cycle 6, PR, %, n = Cycle 8, CR, %, n = Cycle 8, PR, %, n = Safety All AEs No difference No difference Serious AEs No difference No difference PK/PD AUC, lg h/ml, n = 11 36,377 28,657 Ratio: 0.92 ( ) b C max, lg/ml Ratio: 0.93 ( ) b Change from baseline in CD19? B-cell count Difference: 0.3 (-0.9 to 1.4) b? Positive, - negative, ADAb antidrug antibody, ADCC antibody-dependent cellular cytotoxicity, AE adverse event, AUC area under the curve, AUC 0 168h area under the curve from 0 to 168 h, AUC h area under the curve from 0 to 1176 h, CA conference abstract, CDC complement-dependent cytotoxicity, CI confidence interval, C max maximum serum concentration, CR clinical remission, d day(s), D&B Downs and Black (tool), DLBCL diffuse large B-cell lymphoma, NHL non-hodgkin lymphoma, NR not reported, NS not significant, ORR overall response rate, PD pharmacodynamics, PK pharmacokinetics, PR partial remission, RCT randomized controlled trials, SD, standard deviation, SU-DHLsouthwestern university diffuse histiocytic lymphomat max time that the drug is present at its maximum concentration in the serum, wk week(s) Qualitative data for biosimilarity as stated by the corresponding study authors a b 90% CIs shown in parentheses 95% CIs shown in parentheses c d Quality assessment not conducted, because of the absence of validated tools specific for the study type, at the time of analysis

14 14 I. Jacobs et al. translational modifications, glycan pattern, purity, and aggregate levels. PF (Pfizer) A comparative nonclinical study of moderate quality (SYRCLE s risk of bias tool) was conducted with PF and rituximab-eu [28 30]. CDC assay results were reported to be similar. As part of the same study, the safety of PF and rituximab was also evaluated in cynomolgus monkeys [28 30]. Both molecules were well tolerated at doses up to 20 mg/kg. There were no reported effects on food consumption, body weight, body weight gain, coagulation, urinalysis, or clinical chemistry parameters. No cardiac or ophthalmologic effects were observed in the repeat-dose study (not evaluated in the single-dose study). Emesis was noted in animals administered with either PF or rituximab (PF vs. control, n = 6 vs. 1; rituximab-eu vs. control, n = 7 vs. 2). The PK and PD profiles for both molecules were purportedly comparable. Karnik et al. reported biological activity data (from in vitro cell-based assays, cell line unspecified) for PF versus rituximab [31]. The reported biological activity was % for PF versus % for rituximab (combined range for rituximab-eu and rituximab-usa). The Karnik et al. study did not undergo quality assessment, because of the unavailability of suitable risk of bias assessment tools for this type of study at the time of analysis. As part of a comparative nonclinical assessment study evaluating PF versus rituximab, the authors investigated the physicochemical similarity of both products using peptide mapping [reverse-phase (RP) high-performance LC (HPLC)] [28 30]. The identity of each peptide was confirmed by LC/MS using an ultra highresolution quadrupole time-of-flight MS. The chromatographic profiles of both products were reported as identical. The totality of the data (including a separate LC/MS analysis) confirmed identical primary sequences. In a separate study, the primary sequence of PF and rituximab were confirmed using proteolytic digests and peptide mapping via LC/MS/MS [31]. Structure, with respect to glycosylation profile and presence of high molecular mass species ( vs for PF and rituximab-eu/-usa, respectively), was reported as similar. Product purity, charge heterogeneity, post-translational modifications, and hydrodynamic size heterogeneity were reportedly undertaken (individual analyses not shown; Table 2). RTXM83 (mabxience) Seigelchifer et al. evaluated the functional comparability of RTXM83 and rituximab-eu in an unspecified cell line and in cynomolgus monkeys [32]. Similar biological potency was reported (as measured by CDC, ADCC, and apoptosis assays). The in vitro binding affinities to neonatal receptors were also described as similar. Data from the in vivo PK/PD study indicated that the area under the curve (AUC), half-life, and maximum serum concentration (C max ) of RTXM83 were all within the prespecified % range of rituximab. The structural similarity between RTXM83 and rituximab was evaluated using mass fingerprinting and MS analysis, ultraviolet circular dichroism spectroscopy, glycan and charge variant analysis, sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis, and size-exclusion HPLC analysis [32]. The results reportedly indicated similar primary structures, comparable secondary and tertiary structures, and similarity in the extent of post-translational modifications. The level of purity was observed to be comparable between the proposed biosimilar and the originator Proposed Trastuzumab Biosimilars Published preclinical studies of proposed trastuzumab biosimilars are presented in Table 3. PF (Pfizer) Boyle et al. evaluated the biological activity of PF as part of a wider analytical characterization study [33]. The mode of action of PF was compared with trastuzumab using in vitro cell-binding assays (methods and cell line not described or specified). The biological activity data indicated functional similarity to trastuzumab-usa or trastuzumab-eu. The statistical significance of the results was not reported. The binding kinetics to the ligand and Fc-gamma receptor III, involved in immune-effector functions, were also assessed (individual data not reported). This study did not undergo quality assessment, because no suitable risk of bias assessment tool was available at the time of analysis. A nonclinical comparability study between PF and trastuzumab was conducted using in vitro structural and functional analyses [tumor cell growth inhibition assay in the human epidermal growth factor receptor 2 (HER2) overexpression metastatic breast carcinoma cell line, SKBR- 3, and CDC assay], and in vivo PK and immunogenicity assessments in male CD-1 mice [34 37]. Results of the in vitro assessments indicated functional similarity between PF and trastuzumab. PK parameters [C max,auc from time zero to infinity (AUC 0? ), clearance, and steadystate volume of distribution] in mice were reported to be similar and dose dependent. The incidence of ADAbs was reported to be low (*10%) and similar between PF and trastuzumab across all dose levels tested. All animals survived to their terminal blood collection (up to 2880 h) with no mab-related differences in clinical signs, mean body weight, or body weight change. The study by Hurst et al. was considered to be of moderate quality using SYRCLE s risk of bias tool [34].

15 Biosimilars for Cancer: A Systematic Review of Published Evidence 15 Table 3 Outcomes for proposed trastuzumab biosimilars Study type/patients (n) References Outcome/time point Biosimilar a Trastuzumab a Statistical comparison Quality assessment rating BCD-022 RCT PK/safety study/mbc (46) CA [52] Safety, 3 wk Modified D&B: All AEs NR NR p = NS excellent PK/PD, 3 wk Score: 9/12 AUC 0 504h NR NR Ratio: NR ( ) b C max NR NR Ratio: NR ( ) b CT-P6 RCT Comparative PK/safety/efficacy study/her2? mbc (174) Half-life Similar Similar Clearance Similar Similar Volume of distribution Similar Similar T max Similar Similar CA [53] Safety, 8 cycles Modified D&B: All AEs, % excellent Serious AEs, % Score: 10/12 Infusion reaction, % Hypersensitivity, % Cardiotoxicity, % Infection (any grade), % PK/PD, 8 cycles AUC, lg h/ml, mean 32,000 30,600 Ratio: 1.05 ( ) b C trough, lg/ml, mean Ratio: 1.01 ( ) b C max Similar Similar Half-life Similar Similar Clearance Similar Similar Volume of distribution Similar Similar Mean residence time Similar Similar Peak trough fluctuation Similar Similar

16 16 I. Jacobs et al. Table 3 continued Study type/patients (n) References Outcome/time point Biosimilar a Trastuzumab a Statistical comparison Quality assessment rating RCT Comparative safety/efficacy study/her2? mbc (475) FTMB RCT PK/safety study/nhv (118) CA [54] Efficacy, 8 cycles Modified D&B: ORR, % excellent Time to progression, median, p = 0.1 Score: 10/12 mo Time to response, median, mo p = 0.37 Safety, 8 cycles All AEs, % Serious AEs, % Infusion reaction/ hypersensitivity, % Cardiotoxicity, % Infection, % [40]; CA [39] Safety, 9 wk NICE STA: All AEs, 0.5 mg/kg, % 100 NA excellent All AEs, 1.5 mg/kg, % 66.7 NA 7 low risk All AEs, 3.0 mg/kg, % 100 NA Jadad: excellent All AEs, 6.0 mg/kg, % (PBO: 50) Score: 5/5 PK/PD, 9 wk AUC, 1 mg/kg, lg d/ml, mean AUC, 6 mg/kg, lg d/ml, mean (SD) C max, 6 mg/kg, lg/ml, mean (SD) Ratio: 0.89 ( ) b (228.7) (208.3) (35.78) (31.52) Half-life, d, mean (SD) (2.514) (2.362) Ratio: 0.89 ( ) b Clearance, mean (SD) (0.722) (0.5422) Volume of distribution Similar Similar T max, d, mean (SD) (0.0557) (0.0476)

17 Biosimilars for Cancer: A Systematic Review of Published Evidence 17 Table 3 continued Study type/patients (n) References Outcome/time point Biosimilar a Trastuzumab a Statistical comparison Quality assessment rating PF Clinical (safety/pk)/nonclinical (CD-1 mice and cell based)/analytical/nhv (105) [43]; CA [42]; CA [44]; CA [41]; CA [36] Safety, 10 wk NICE STA: All AEs, % 80 USA: 82.9; EU: 82.9 excellent Serious AEs, % unclear, 6 low Infusion reaction, % 37.1 USA: 20; EU: 28.6 risk ADAb, % 0 USA: 0; EU, 2.9 Jadad: good PK/PD, 10 wk Score: 3/5 AUC, lg h/ml, mean (SD) 37,130 (6305) USA: 37,310 (6728); EU: 40,330 (6994) Ratio: 0.99 ( ) b Ratio: 0.92 ( ) b C max, lg/ml, mean (SD) 159 (26) USA: 164 (31); EU: 174 (31) Ratio: 0.97 ( ) b Ratio: 0.91 ( ) b Half-life, h, mean (SD) 213 (42) USA: 212 (47); EU: 220 (42) Clearance, ml/h/kg, mean (SD) Volume of distribution, ml/kg, mean (SD) (0.026) USA: (0.032); EU: (0.025) 56.1 (8.2) USA: 55.7 (8.8); EU: 51.7 (6.9) Comparative PK/safety/efficacy protocol/ HER2? CA [55] Protocol: no outcomes reported Not applicable mbc (690) Clinical study protocol/her2? invasive BC (220) CA [36] Protocol: no outcomes reported Not applicable Nonclinical (cynomolgus monkeys, mouse xenograft models, and cell based)/analytical CA [33] Functional assessment (in vitro) Biological activity, ratio % USA/EU: Composition Peptide mapping Identical Identical Charge isoform Similar Similar Glycan quantification Similar Similar Purity Similar Similar Major post-translational Similar Similar modifications High molecular mass, % USA/EU: Not evaluated c

18 18 I. Jacobs et al. Table 3 continued Study type/patients (n) References Outcome/time point Biosimilar a Trastuzumab a Statistical comparison Nonclinical (CD-1 mice and cell based)/analytical [34]; CA [35]; CA [37]; CA [36] Functional assessment (in vitro) Tumor growth inhibition Superimposable Superimposable CDC Similar Similar Safety (mice), single dose ADAb, 1 mg/kg, % 20.0 USA: 20.0; EU: 20.0 ADAb, 10 mg/kg, % 8.0 USA: 0; EU: 4.0 ADAb, 100 mg/kg, % 4.2 USA: 4.0; EU: 8.0 Tolerability Similar Similar PK/PD (mice), single dose AUC, 1 mg/kg, lg h/ml, mean AUC, 10 mg/kg, lg h/ml, mean AUC, 100 mg/kg, lg h/ml, mean C max, 1 mg/kg, lg/ml, mean (SD) C max, 10 mg/kg, lg/ml, mean (SD) C max, 100 mg/kg, lg/ml, mean (SD) 4220 USA: 4080; EU: 4650 Ratio: 1.03; Ratio: ,800 USA: 50,000; EU: 51, ,000 USA: 289,000; EU: 298, (1.90) USA: 26.3 (5.83); EU: 18.6 (8.55) 318 (49.4) USA: 269 (52.7); EU: 281 (49.9) 2520 (219) USA: 2620 (322); EU: 2700 (450) Half-life, 1 mg/kg, h, mean 380 USA: 416; EU: 536 Half-life, 10 mg/kg, h, mean 440 USA: 352; EU: 392 Half-life, 100 mg/kg, h, mean 309 USA: 320; EU: 280 Clearance, 1 mg/kg, ml/h/kg, USA: 0.245; EU: mean Clearance, 10 mg/kg, ml/h/kg, USA: 0.200; EU: mean Clearance, 100 mg/kg, ml/h/ USA: 0.346; EU: kg, mean Volume of distribution, 1 mg/ 104 USA: 129; EU: 113 kg, lg h/ml, mean Volume of distribution, 10 mg/ 84.9 USA: 86.7; EU: 85.4 kg, lg h/ml, mean Volume of distribution, 120 USA: 130; EU: mg/kg, lg h/ml, mean Ratio: 1.04; Ratio: 1.0 Ratio: 0.99; Ratio: 0.96 Ratio: 0.87; Ratio: 1.2 Ratio: 1.18; Ratio: 1.13 Ratio: 0.96; Ratio: 0.93 Quality assessment rating SYRCLE s risk of bias: moderate 12 unclear, 10 low risk

19 Biosimilars for Cancer: A Systematic Review of Published Evidence 19 Table 3 continued Study type/patients (n) References Outcome/time point Biosimilar a Trastuzumab a Statistical comparison Quality assessment rating Composition Tryptic peptide mapping Similar Similar? Positive, ADAb antidrug antibody, AE adverse event, AUC area under the curve, AUC 0 504h area under the curve from 0 to 504 h, BC breast cancer, CA conference abstract, CDC complement-dependent cytotoxicity, CI confidence interval, Cmax maximum serum concentration, Ctrough plasma trough concentration, d day(s), D&B Downs and Black (tool), HER2 human epidermal growth factor receptor 2, mbc metastatic breast cancer, mo month(s), NA not applicable, NHV normal healthy volunteers, NICE STA National Institute for Health and Care Excellence Single Technology Appraisal (manufacturer s template), NR not reported, NS not significant, ORR overall response rate, PBO placebo, PD pharmacodynamics, PK pharmacokinetics, RCT randomized controlled trial, SD standard deviation, T max time that the drug is present at maximum concentration in the serum, wk week(s) Qualitative data for biosimilarity as stated by the corresponding study authors a 90% CIs shown in parentheses b Quality assessment not conducted, because of the absence of validated tools specific for the study type, at the time of analysis c An analytical study published by Boyle et al. compared the structural and functional attributes of PF with those of trastuzumab [33]. Peptide mapping via LC/MS/MS confirmed identical primary structures for both PF versus the licensed products. Charge heterogeneity, major post-translational modifications, hydrodynamic size heterogeneity, and glycan quantification of PF versus trastuzumab were evaluated using a combination of biochemical techniques, and all analyses were reported by authors as similar (individual data not reported). High molecular mass species ranged from in the PF sample versus in the trastuzumab sample. This study did not undergo quality assessment, because no suitable risk of bias assessment tool was available at the time of analysis. Hurst and colleagues evaluated the structural attributes of PF as part of a wider nonclinical comparability PK/immunogenicity assessment in mice [34 37]. Using peptide mapping, the authors reported structural similarity, with detected peptides constituting C98% of the amino acid sequence of trastuzumab. The complete amino acid sequence was obtained from a separate LC/MS analysis at the subunit level, which confirmed structural similarity between PF and trastuzumab at the primary sequence level. The study by Hurst et al. was considered to be of moderate quality using SYRCLE s risk of bias tool [34]. 3.3 Pharmacokinetic/Safety Data in Healthy Subjects Proposed Bevacizumab Biosimilars Published PK/safety studies of proposed bevacizumab biosimilars in healthy subjects are presented in Table 1. ABP 215 (Amgen) The PK equivalence of ABP 215 (vs. bevacizumab-eu and bevacizumab-usa) was evaluated in healthy adult male subjects [16, 17]. The authors reported PK equivalence of ABP 215 compared with bevacizumab (within the prespecified equivalence criteria of %). Similar treatment-related adverse events (AEs) were reported for the ABP 215, bevacizumab-usa, and bevacizumab-eu groups; no subjects developed ADAbs. The study was considered to be of excellent quality Proposed Rituximab Biosimilars No published PK/safety studies in healthy subjects have been reported for proposed biosimilars of rituximab. However, planned and completed PK/safety studies (with

20 20 I. Jacobs et al. published data) have been identified for rituximab biosimilars in rheumatoid arthritis patients [38] Proposed Trastuzumab Biosimilars Published PK/safety studies of proposed trastuzumab biosimilars in healthy subjects are presented in Table 3. FTMB (Allergan/Amgen) FTMB was evaluated and compared with trastuzumab or placebo in a dose-escalation/ bioequivalence study in healthy male subjects (n = 118) [39, 40]. The authors reported bioequivalence of FTMB compared with trastuzumab [primary endpoint, AUC 0? ; ratio 0.89; 90% confidence interval (CI) ]. Nonlinear, target-mediated PK were also observed. FTMB was reported to be well tolerated up to doses of 6.0 mg/kg, with no additional AEs occurring beyond those already documented for trastuzumab. The study was considered to be of excellent quality, as measured by the NICE STA and Jadad scoring tools. PF (Pfizer) The safety and PK properties of PF were evaluated and compared with trastuzumab in a study involving 105 healthy male subjects [36, 41 44]. Reported AEs were comparable between treatments. However, the incidence of pyrexia was higher in the PF arm [ten participants (28.6%); mild in all but one participant] versus the trastuzumab arms [two to three participants ( %)]. In the context of biosimilar development, a PK similarity study is not necessarily designed for a comparative assessment of safety. This observation appears to be due to chance and not to any inherent difference between PF and trastuzumab [43]. Only one subject was ADAb positive (ADAb? ) in the trastuzumab-eu group. The authors reported PK similarity between PF and trastuzumab-eu and trastuzumab-usa, with a 90% CI of C max, time from 0 to the last time point with quantifiable concentration (AUC 0 T ), and AUC 0? within the prespecified range (80 125%) of PK equivalence. The study was of excellent quality as measured by the NICE STA; however, using the Jadad scoring tool, the study was considered to be only of good quality because the method of randomization and blinding was not described. 3.4 Clinical Evidence in Cancer Patients Proposed Bevacizumab Biosimilars Published clinical studies of proposed bevacizumab biosimilars in cancer patients are presented in Table 1. BCD-021 (Biocad) The PK and safety of BCD-021 were assessed in a comparative trial involving 28 patients with non-squamous NSCLC (stage IIIb/IV) [45]. PK analysis revealed that 90% CIs for the ratio of geometric means of AUC from 0 to 504 h (AUC 0 504h ) of bevacizumab after single BCD-021 and bevacizumab administrations were %. The safety characteristics of BCD-021 and bevacizumab were considered equivalent by authors. BCD-021 or bevacizumab in combination with paclitaxel and carboplatin was evaluated in a comparative safety/efficacy study in 138 patients with advanced nonsquamous NSCLC (stage IIIb/IV) [46]. No statistically significant difference was observed between groups for overall response rate (ORR) as the primary endpoint and other efficacy endpoints [clinical remission (CR), partial remission (PR), stable disease, and progression rate]. The AE profiles of BCD-021 and bevacizumab were declared equivalent, with the majority of incident AEs reportedly associated with chemotherapy. Binding and neutralizing ADAbs were transient and detected only in one patient in each group. Both studies were of excellent quality Proposed Rituximab Biosimilars Published clinical studies of proposed rituximab biosimilars in cancer patients are presented in Table 2. BCD-020 (Biocad) BCD-020 was evaluated in a comparative safety/efficacy study in 92 patients with non-hodgkin lymphoma (NHL) and compared with the parameters of rituximab. The study was considered to be of excellent quality [47 49]. ORR on day 50 ± 5 (primary endpoint) in both arms was reported as equivalent (p = ). The lower limit of the 95% CI for difference in proportion of ORR between arms was 17.81%, exceeding the predefined noninferiority margin ( 20%). BCD-020 was considered noninferior to rituximab in all other measured efficacy outcomes. The proportion of AEs (including grades 3 4) was equal across both arms. The PK/PD and immunogenicity profiles of BCD-021 were comparable with that of the originator. RTXM83 (mabxience) In a PK and safety study, RTXM83 was evaluated and compared with rituximab in combination with cyclophosphamide, doxorubicin, vincristine, prednisolone (CHOP) chemotherapy in 24 patients with DLBCL; the study was considered to be of good quality [50]. The authors reported comparable PK and (null) immunogenicity profiles across both treatment arms. SAIT101 (Samsung) The efficacy, safety, and PK/PD parameters of SAIT101 (plus CHOP chemotherapy) were compared with rituximab in a PK/safety study in 24 patients with DLBCL [51]. The study was considered to be of excellent quality. SAIT101 displayed efficacy (tumor response as measured by CR and PR parameters) and

21 Biosimilars for Cancer: A Systematic Review of Published Evidence 21 safety comparable with rituximab, although one case of mortality (drug-induced pneumonitis) in the SAIT101 group was recorded. Results of the PK/PD assessment indicated a high probability of similarity between SAIT101 and the originator Proposed Trastuzumab Biosimilars Published clinical studies of proposed trastuzumab biosimilars in cancer patients are presented in Table 3. BCD-022 (Biocad) BCD-022 was evaluated and compared with trastuzumab in a comparative PK substudy in 46 patients with HER2? metastatic BC [52]. The study was considered to be of excellent quality using the modified Downs and Black instrument. The primary endpoint of the study was AUC 0 504h. The authors reported PK equivalence of BCD-022 compared with trastuzumab in accordance with the prespecified equivalence criteria of %. BCD-022 and trastuzumab were found to be well tolerated, without any significant differences in AE frequency between the groups. CT-P6 (Celltrion) CT-P6 has been investigated in a PK/ safety study in patients with HER2? metastatic BC (n = 174) [53]. The limits of the 90% CIs for the ratio of AUC at steady-state geometric means were reportedly contained within the established margin (80 125%) required for bioequivalence. There was no reported difference between other studied PK parameters. Serious AEs were reported in 15.8% of CT-P6 and 20.9% of trastuzumab patients; 2.6 and 3.0% were treatment related, respectively. Hypersensitivity, cardiotoxicity, and infection were noted in 1.3, 2.6, and 1.3% of CT-P6 patients and in 1.5, 7.5, and 0% of trastuzumab patients, respectively. Results of a comparative safety/efficacy study comparing the efficacy and safety of CT-P6 versus trastuzumab in patients (n = 475) with HER2? metastatic BC were also published [54]. The primary outcome measure was to demonstrate equivalence of CT-P6 and trastuzumab, both given in combination with paclitaxel, in terms of efficacy determined by ORR. A numerically greater number of patients achieved ORR in the trastuzumab group, compared with the CT-P6 group (62 vs. 57%, respectively). The limits of the 95% CIs for the difference in the proportion of responders (from the pooled intention-to-treat population) were reported to be within the range required for bioequivalence ( 0.15, 0.15). No significant differences were observed between the two groups in the other measured efficacy parameters. The tolerability and safety of CT-P6 were reported as comparable with those of trastuzumab. Both studies were considered to be of excellent quality. PF (Pfizer) Two clinical studies comparing PF with trastuzumab were ongoing at the time of review. One was for first-line treatment of patients (estimated n = 690) with HER2? metastatic BC and the other for neoadjuvant treatment of operable HER2? BC (estimated n = 220) [55, 56]. These studies did not undergo quality assessment, as both were found to be trial protocols. 3.5 Preclinical and Clinical Data for Intended Copies of Rituximab: Kikuzubam and Reditux TM Published preclinical and clinical studies of rituximab intended copies are presented in Table 4. Kikuzubam (Probiomed) Flores-Ortiz et al. validated a series of analytical methods for biosimilar characterization purposes [57]. In vitro comparability assessments (CDC and ADCC) were conducted using cell-based assays (CD20-expressing cells and primary NK cells). Kikuzubam displayed comparable potency to rituximab (and intended copy Reditux TM ), as indicated by the results of the CDC and ADCC assays. The physicochemical properties of Kikuzubam were evaluated and compared with both Reditux TM and rituximab [57]. The sequence comparability of Kikuzubam to both comparators was confirmed using peptide mapping (RP-HPLC). Kikuzubam and rituximab were reported to have the identical glycoform profiles (Reditux TM displayed heterogeneity in theoretical mass due to the C-terminal lysine content). Significant differences were observed among all three products with respect to charge variants, as determined by cation exchange (CEX) and hydrophobic interaction chromatography (HIC). Because the CEX- HPLC technique reportedly revealed nearly two times more acidic variants than standard HPLC methods, the authors urged careful consideration in the chosen analytical methods for comparability studies. The structure [determined by differential scanning calorimetry (DSC) and matrix-assisted lasers (MALS)] and purity [determined by capillary gel electrophoresis (CGE) and size-exclusion chromatography (SEC)] were reportedly similar across all three products evaluated. Reditux TM (Dr Reddy s Laboratories) Thakral et al. investigated the biodistribution profile of a radiolabeled immunoconjugate of Reditux TM [177Lu-DOTA-antiCD20 antibody-rituximab (BioSim)] in three patients with B-cell NHL [58]. The study was considered to be of good quality (Downs and Black score). Maximum uptake of the radioimmunoconjugate was reportedly observed in the liver, indicating a favorable biodistribution profile.

22 22 I. Jacobs et al. Table 4 Outcomes for intended copies of rituximab Study type/patients (n) References Outcome/time point Biosimilar a Rituximab a Statistical comparison Quality assessment rating IC Kikuzubam Nonclinical (cell based)/analytical [57] Functional assessment (in vitro) (RTX/ Reditux TM ) ADCC, ratio % IC Reditux TM Clinical (PK/PD)/nonclinical (cell based)/nhl (3) CDC (3 batches), ratio % Composition 98, 102, 112 NR/81, 111, 108 Peptide mapping Same Same Glycan Same Same quantification Mass spectrometry, intact mass Heterogeneous Heterogeneous DSC analysis Similar Similar Cation exchange, /7.0 acid, % Cation exchange, /20.6 main, % Cation exchange, /72.4 basic, % Hydrophobic interaction (main isoform), % \3.0 \2.0/\24.1 Multiangle laser light scattering [58] PK/PD (biodistribution) Liver, %, mean (SD) Kidney, %, mean (SD) Spleen, %, mean (SD) Heart, %, mean (SD) Functional assessment (in vitro) Immunoreactivity Similar Similar 22.0 (8.0) at 96 h NR 3.8 (0.8) at 48 h NR 2.5 (1.3) at 48 h NR 3.5 (1.5) at 24 h NR Highly specific \2% Nonspecific binding Immunoreactive fraction, NR Apyrogenicity Pyrogen free NR Not evaluated b D&B: good Score: 13/26

23 Biosimilars for Cancer: A Systematic Review of Published Evidence 23 Table 4 continued Study type/patients (n) References Outcome/time point Biosimilar a Rituximab a Statistical comparison Quality assessment rating Post-marketing/observational (retrospective)/nr (223) Post-marketing/observational (prospective)/dlbcl (133) (and rheumatoid arthritis, scleroderma, and dermatomyositis) [60]; CA [59] CA [61] Efficacy CR, % p = PR, % p = CR/PR? p = progressed Overall survival, % p = Progression-free p = survival, % Safety; grades 3 and 4 (n = 30): Febrile p = neutropenia, % Mucositis, % 10 5 p = Diarrhea, % Peripheral 7 3 neuropathy, % Infusion reactions, 7 5 p = % Dilated cardiomyopathy, % 3 3 CMV viremia, % 3 0 Herpes zoster 3 0 reactivation, % Intestinal 0 3 perforation, % Urinary tract 0 3 infection, % Pneumonia, % 0 8 D&B: good Score: 16/26 CA [62] Safety Modified All AEs, % 14.3 NA D&B: fair Chills, % 20 NA Score: 4/12 Headache, % 16.7 NA Fever, % 13.0 NA Urticaria, % 10.0 NA Possibly treatmentrelated 66.7 NA AEs, % Probable 13.3 NA treatment-related AEs, % Proven treatmentrelated 20.0 NA AEs, % All treatmentrelated 73.0 NA AEs, % Mild AEs, % 90.0 NA Moderate AEs, % 6.7 NA Severe AEs, % 3.3 NA Mortality, % 0 NA

24 24 I. Jacobs et al. Table 4 continued Study type/patients (n) References Outcome/time point Biosimilar a Rituximab a Statistical comparison Quality assessment rating Post-marketing/observational (prospective)/dlbcl (21) CA [63] Efficacy Modified B-cell count, day 1.75 (0.27) cells/ NA D&B: 3, median (SD) ll good B-cell count, day 3, median (SD) Progression-free survival, % Residence time, d, mean (SD) Nonclinical (cell based)/analytical [57] Functional assessment (in vitro): 5.56 (1.24) cells/ ll NA 70.3 NA Safety No toxicity NA PK/PD: AUC, lg h/ml, 54,236 (47,555) NA mean (SD) C max, lg/ml, (141.46) 408 (literature) mean (SD) Half-life, d, mean 10.9 (8.6) 22 (literature) (SD) Clearance, ml/h/ 0.15 (0.16) 0.14 (literature) kg, mean (SD) Volume distribution, L/kg, mean (SD) 1.3 (0.64) 2.7 (literature) 2.78 (3.08) NA (RTX/ Kikuzubam ) ADCC, ratio % CDC (3 batches), ratio % Composition Multiangle laser light scattering Nonclinical (cell based) CA [64] Safety (rat and rabbit cell lines) ADAb 81, 111, 108 NR/98, 102, 112 Peptide mapping Same Same Glycan Same Same quantification Mass spectrometry, intact mass Heterogeneous Heterogeneous DSC analysis Similar Similar Cation exchange, /37.8 acid, % Cation exchange, /56.6 main, % Cation exchange, /5.6 basic, % Hydrophobic interaction (main isoform), % \24.1 \2.0/\3.0 Similar Similar Comparable Comparable (USA/EU) Score: 5/12 Not evaluated b Not evaluated b

25 Biosimilars for Cancer: A Systematic Review of Published Evidence 25 Table 4 continued Study type/patients (n) References Outcome/time point Biosimilar a Rituximab a Statistical comparison Quality assessment rating Analytical CA [65] Composition Not IdeS digestion Similar Similar evaluated b Peptide mapping (trypsin and pepsin) Similar Similar Isotope Similar Similar Analytical CA [66] Composition Not SDS-PAGE Similar Similar evaluated b ice NR NR Sig. diff. CE NR NR Sig. diff. CEX-HPLC NR NR Sig. diff. ADAb antidrug antibody, ADCC antibody-dependent cellular cytotoxicity, AE adverse event, AUC area under the curve, CA conference abstract, CDC complement-dependent cytotoxicity, CE capillary electrophoresis, CEX-HPLC cation exchange high-performance liquid chromatography, C max maximum concentration in serum, CMV cytomegalovirus, CR clinical remission, d day(s), D&B Downs and Black (tool), DLBCL diffuse large B-cell lymphoma, IC intended copy, ice imaged capillary electrophoresis, DSC differential scanning calorimetry, NA not applicable, NHL non-hodgkin lymphoma, NR not reported, PD pharmacodynamics, PK pharmacokinetics, PR partial remission, RTX rituximab, SD standard deviation, SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis, sig. diff. significantly different a Qualitative data for biosimilarity as stated by the corresponding study authors b Quality assessment not conducted, because of the absence of validated tools specific for the study type, at the time of analysis A retrospective observational study comparing the efficacy and safety of Reditux TM versus rituximab in patients with DLBCL (n = 223) has been reported, and was considered to be of good quality (Downs and Black score) [59 61]. CR rates were similar between Reditux TM and rituximab originator, and overall survival at 5 years was also reportedly comparable. There were no reported differences in infusion-related reactions or grades 3 and 4 neutropenia and oral mucositis between the two interventions. A prospective observational study was conducted to determine the causality and factors associated with onset drug reactions during the administration of Reditux TM in patients with cancer (excluding Hodgkin lymphoma and chronic lymphocytic leukemia indications) [62]. The study also assessed patients with chronic inflammatory disease, and the analyses were pooled across indications. Of the total number of patients (n = 133 for combined oncology and chronic inflammatory conditions), only 19 patients experienced AEs, of which the majority were perceived as mild and related to the speed of infusion. The study was considered to be of fair quality (modified Downs and Black score). Menon et al. evaluated the PK/PD properties of Reditux TM versus rituximab in a prospective observational study in patients (n = 21) with DLBCL [63]. Analyses were undertaken and compared with published data on rituximab. The authors concluded that the PK profile and B-cell kinetics of Reditux TM were comparable with rituximab from the published literature. The study was considered to be of good quality. Flores-Ortiz et al. compared the physicochemical properties and biological activity of both Reditux TM and Kikuzubam with rituximab [57]. Biological comparability assessments were conducted in vitro using cellbased assays (CD20-expressing cells and primary NK cells). Reditux TM displayed similar biological effects to rituximab (and Kikuzubam ), as indicated by the results of the CDC and ADCC assays (Table 4). The Florez- Ortiz et al. study did not undergo quality assessment, because of the unavailability of suitable risk of bias assessment tools for this type of study at the time of analysis. Kumar et al. investigated the suitability (sensitivity and specificity) of complementary assays for immunogenicity assessment in comparative clinical trials for biosimilars, using Reditux TM and rituximab as prototypic agents [64]. The study objectives were not aimed at delineating the comparability of Reditux TM versus rituximab. However, the authors commented that the analytical similarity (as implied by this study) was sufficient to justify the use of biosimilar drug-based assays for immunogenicity assessment. The Kumar et al. study did not undergo quality assessment, because of the

26 26 I. Jacobs et al. unavailability of suitable risk of bias assessment tools for this type of study at the time of analysis. The immunoreactivity and apyrogenicity of an immunoconjugate of Reditux TM were investigated in vitro using RAMOS cells [58]. The immunoreactivity of the mab conjugate, as reported by the authors, showed high and specific binding ability to target cells (nonspecific binding showed \2% of the total radioactivity). The immunoreactive fraction was reported as r = R = , where R = The radiolabeled doses of the mab conjugate were reported to be pyrogen free. The bacterial endotoxin level was \0.1 EU/mL (the authors reported a permissible limit of \0.5 EU/mL; Table 4). The Thakral et al. study [58] assessed human subjects as well, and thus underwent quality assessment using the Downs and Black instrument. The study was considered to be of good quality. As described previously, the physicochemical properties of Reditux TM were evaluated and compared with Kikuzubam and rituximab [57]. The sequence comparability of Reditux TM to both comparators was confirmed using peptide mapping (RP-HPLC). Results of the chemical composition analysis showed heterogeneity of Reditux TM (compared with rituximab and Kikuzubam ) in theoretical mass due to the C-terminal lysine content. Significant differences were observed for both Reditux TM and Kikuzubam versus rituximab with respect to charge variants, as determined by CEX and HIC. The structure of Reditux TM (determined by DSC and MALS) and purity (determined by CGE and SEC) were said to be similar compared with both the originator and Kikuzubam. Mekhssian et al. developed a workflow to aid in a more comprehensive biosimilar comparability assessment using high-resolution MS methodologies [65]. Although this study was not aimed at the characterization of Reditux TM, it was utilized as a prototypic agent along with the originator, and some analyses of the data are relevant to this review. Three main glycoforms, G0F, G1F, and G2F, were observed in the mass spectra corresponding to the Fc/2 fragment of both Reditux TM and rituximab, with variations in their corresponding ratios. The authors observed that the C-terminal lysine (heavy chain) was completely clipped for rituximab and only partially clipped for Reditux TM. Moreover, the presence of a light chain without N-terminal pyroglutamate formation was identified only in Reditux TM. The authors reported on the suitability of IdeS digestion as a method to generate mab fragments that are more amenable for LC separation. The authors postulated that the addition of guanidine hydrochloride (GdHCl) in turn permits the complete reduction of disulfide bridges. Results of a comparability study by Lin et al. indicated similar structural profiles for rituximab-us and rituximab- EU [66]. However, significant differences were reported between rituximab-usa/-eu and Reditux TM as delineated by imaged capillary electrophoresis, capillary electrophoresis, and CEX-HPLC methods [66]. 3.6 Quality Assessment of the Studies The quality of analytical and cell-based studies, including abstract publications of analytical and animal studies, could not be evaluated as validated tools to assess the risk of bias for these types of studies or publications were unavailable at the time of the analysis. Thirteen studies were not evaluated [19 23, 27, 31 33, 57, 64 66]. An additional two published abstracts were found to be trial protocols and thus were not eligible for quality assessment [36, 55]. Two RCTs were assessed using the NICE STA manufacturer s template and Jadad scoring tool (Supplementary Fig. S3) [40, 43]. Both studies were considered to be of excellent quality. Using the NICE STA assessment criteria, although both studies were randomized, only one study provided information to assess the randomization process (Yin et al. [43] did not). Despite the lack of details for randomization in the Yin et al. study, both RCTs were reported as being concealed and blinded. However, details of the blinding method were not provided in Yin et al. [43]. Patient withdrawals, outcome selection, reporting bias, and statistical analyses were appropriately reported in both studies according to the NICE STA criteria. The two RCTs were also evaluated using the Jadad scoring tool, with Wisman et al. [40] (5 points) reporting on the randomization and method, blinding and methodology, and withdrawals and dropouts (Supplementary Fig. S3), whereas Yin et al. [43] (3 points), as mentioned above, failed to report on the method of randomization and blinding. Two observational studies were assessed using the Downs and Black scoring tool (Supplementary Fig. S4) [58, 60]. Of these, both were considered to be of good quality. Roy et al. scored 16 points from a maximum of 27 points [60], while Thakral et al. scored 13 points [58]. Word count in abstracts from conference proceedings and meetings is restricted, and thus abstracts generally provide limited information on study methodologies and outcomes. For this reason, the Downs and Black instrument was adapted to assess the quality of the 11 identified abstracts for original studies (Supplementary Fig. S5) [16, 45 47, 50 54, 62, 63]. The average score for abstract publications was 8.5 out of a possible 12 points. The total score was fair quality (3 4) for one study [62], good quality (5 8) for two studies [50, 63], and excellent quality (9 12) for eight studies [16, 45 47, 51 54]. The majority of studies published as conference abstracts were of good or excellent quality (90.9%). Three animal studies were assessed using SYRCLE s risk of bias tool (Supplementary Fig. S6) [24, 29, 34]. The three studies were of moderate quality.

27 Biosimilars for Cancer: A Systematic Review of Published Evidence Weight and Breadth of Evidence for Biosimilarity The available comparative data from full-text publications of RCTs are currently limited. In addition, the only postmarketing/observational studies that were identified were for intended copy Reditux TM. Biosimilarity (based on combined evidence from all related published studies) was graded as identical, highly similar, similar, and dissimilar, which was directly inferred from the conclusions of the study authors (Table 5). For biosimilars, based on study author conclusions from the literature at the time the search was undertaken, the current data from clinical and nonclinical studies suggest a moderate to high degree of similarity between the described biosimilars and their originators (Table 5). The degree of biosimilarity is ultimately determined by regulatory authorities and is based on the totality of evidence, which includes data on molecular and functional characterization, other nonclinical data, and the safety, PK, immunogenicity, and efficacy clinical trial data. However, given that regulatory submissions are not publicly available and the full data set therein cannot be systematically reviewed, the authors have instead based all analysis on information from the peer-reviewed literature. The total number of studied variables and total reported patient numbers were identified from analytical/nonclinical and clinical studies, respectively. These parameters were then mapped (Fig. 2a, b) against the degree of similarity (as observed by the study author) in order to depict the depth of the research programs and the overall quantity of available evidence for each agent. Table 5 Summary of published evidence for the degree of similarity between proposed biosimilars or intended copies and originators by study type Biologic originator Biosimilar (name[s]; company) Comparator studies Analytical studies Nonclinical studies PK/ safety studies Comparative safety/efficacy studies Post-marketing/ observational studies Bevacizumab ABP 215 (Amgen, USA) 44*** NA NA BCD-021 (Biocad, Russia) NA NA NA PF (Pfizer, USA) 4 or 44 4 NA NA NA RPH-001 (Alphamab, China/R-Pharm, NA 4 NA NA NA Russia) Rituximab BCD-020 (AcellBia TM ; Biocad, Russia) NA NA NA 44 NA GP2013 (Sandoz, Switzerland) NA NA NA PF (Pfizer, USA) 44 or 44 NA NA NA 444 RTXM83 (mabxience, Switzerland) NA NA SAIT101 (Samsung, South Korea) NA NA 44 NA NA 1B8 (Center of Molecular Immunology, Cuba) NA 4 NA NA NA Intended Copies of Rituximab Kikuzubam (Probiomed, Mexico) 3 or NA NA NA Reditux TM (Dr. Reddy s Laboratories, 3 4 NA NA 4 India) Trastuzumab BCD-022 (Biocad, Russia) NA NA 44 NA NA CT-P6 (Celltrion, South Korea) NA NA NA FTMB (ABP 980; Allergan, USA/ Amgen, USA/Synthon, the Netherlands) NA NA 4 NA NA PF (Pfizer, USA) 44 or 444 NA not applicable, evidence from published studies not available, PK pharmacokinetics 4 Similar (based on combined evidence from all related published studies) 44 Highly similar (based on combined evidence from all related published studies) 444 Identical (based on combined evidence from all related published studies) 3 Dissimilar (based on combined evidence from all related published studies) NA NA

28 28 I. Jacobs et al. a Highly dissimilar Dissimilar Similar Highly 10,000 similar Identical Healthy subjects/patients (N, logarithmic scale) 1, Highly dissimilar 1 Dissimilar Similar Highly Degree of similarity similar Identical BCD-021 (BEV) Reditux (IC; RTX) BCD-020 (RTX) RTXM83 (RTX) SAIT101 (RTX) BCD-022 (TRA) CT-P6 (TRA) FTMB (TRA) PF (TRA) b Highly dissimilar Dissimilar Similar Highly similar 100 Identical Number of reported variables (N, logarithmic scale) 10 Highly dissimilar 1 Dissimilar Similar Highly Degree of similarity similar Identical ABP 215 (BEV) RPH-001 (BEV) PF (BEV) Kikuzubam (IC; RTX)* Reditux (IC; RTX) 1B8 (RTX) GP2013 (RTX) PF (RTX) RTXM83 (RTX) PF (TRA)

29 Biosimilars for Cancer: A Systematic Review of Published Evidence 29 b Fig. 2 Biosimilarity and a total number of healthy subjects or patients for proposed biosimilars and ICs in clinical trials; and b breadth of data for proposed biosimilars and ICs in analytical and nonclinical studies. Totality of evidence from all available published studies (up to September 3, 2015) was used to assess degree of similarity for proposed biosimilars and ICs, and is based on the original conclusions made by the study authors. The scale of reference used by each author was not accounted for, as this was not uniformly reported in the literature. * based on author interpretation of study data, Kikuzubam exhibits, in some cases, dissimilar and, in other cases, identical physicochemical characteristics compared with the originator. BEV bevacizumab, IC intended copy, RTX rituximab, TRA trastuzumab Studies (n = 16) of the proposed biosimilars and intended copies that were identified included 2514 patients or healthy subjects (Fig. 2a). Biosimilars CT-P6 and PF reported the largest combined study populations from PK/safety and comparative efficacy/safety studies [649 and 1015 (two planned studies) patients, respectively]. Based on clinical reports (Fig. 2a), development candidates FTMB and RTXM83 were reported to be similar to their originators. All other development candidates were reported by study authors to be highly similar, based on clinical studies. The extent of reported information varied greatly across studies when considering the breadth of data available for preclinical studies (based on number of variables reported from structural, functional, and nonclinical studies) for named biosimilars (Fig. 2b). PF (rituximab), PF (trastuzumab), and GP2013 (rituximab) all had published data on a larger number of investigated variables for analytical and nonclinical biosimilarity (ranging from 29 to 54), compared with the other development candidates, which published on average only five variables across their preclinical programs, as reported in the literature. Findings from Flores-Ortiz et al. [57] revealed heterogeneity with respect to MS and CEX data for intended copy Kikuzubam. All other variables (DSC analysis, peptide mapping, glycan quantification, etc.) were, however, reported to be the same. Thus the positioning on the x-axis (Fig. 2b) was determined to be both dissimilar and identical across selected variables. Although the conclusions put forward by study authors were that the majority of molecules exhibited biosimilarity to their originator, comparative data were not provided for all attributes studied. 3.8 Non-empirical Data A limited number of non-empirical publications that referenced a named biosimilar were retrieved overall. These results are shown in Supplementary Table S5. A large percentage of these studies described approaches or methodologies for the characterization and development of biosimilars, as well as how to demonstrate biosimilarity with the originator molecule [67 71]. Additional studies provided examples of biosimilars and opinions on regulatory aspects of biosimilar development [72 80]. 3.9 Planned and Ongoing Trials of Bevacizumab, Rituximab, and Trastuzumab Biosimilars The results of the ClinicalTrials.gov search (conducted September 21, 2015) are presented in Supplementary Table S6. A total of six oncology clinical trials were identified for proposed biosimilars of bevacizumab, one of which was complete at the time of analysis. PF represented the most commonly studied bevacizumab biosimilar (two trials identified, one PK/safety study complete). Comparative efficacy and safety studies were reportedly active for ABP 215 and for BCD-021 at the time of analysis. The search also identified additional oncology biosimilars not yet appearing in the published literature. These were bevacizumab biosimilars SB8 (Samsung Bioepis) and BEVZ92 (mabxience/laboratorio Elea/ LIBBS). For proposed biosimilars of rituximab, six trials, all in lymphoma, were reported. CT-P10 (Celltrion) had the highest number of studies (two comparative efficacy and safety trials identified, one actively recruiting). Single comparative efficacy and safety studies were also listed on the ClinicalTrials.gov registry for BCD-020, GP2013, PF and RTXM83. Within oncology and compared across all originator classes, trastuzumab biosimilars appeared most frequently in nine studies (two PK/safety in healthy subjects, one PK/ safety study in patients with BC, and six comparative safety/efficacy trials). Of the trastuzumab biosimilars, Pfizer s PF had the most studies listed (one PK/ safety study in healthy subjects, and two further studies in BC patients [55, 56]), followed by Celltrion s CT-P6 (two comparative efficacy/safety studies). The search also identified a novel trastuzumab biosimilar, SB3 (Samsung Bioepis), not yet appearing in the published literature at the time of review Publishing Trends on Biosimilars Journals A total of 58 unique journals publishing material on biosimilars in oncology that met our inclusion criteria were identified during this review (23 of which published empirical studies; Supplementary Table S7). The top publishing journal was mabs, which has published seven articles since 2010, five of which were published between 2013 and Before 2005, no journal articles relevant to

30 30 I. Jacobs et al. the topic of mabs or fusion protein biosimilars were published relevant to the field of oncology. The greatest number of articles was published in 2013 (n = 8), of which mabs, Seminars in Oncology, and Annals of Oncology published two each. Since 2011 (up to and including 2014), the number of publications per year has reached similar totals (n = 5, 5, 8, 4, and 6 for 2011, 2012, 2013, 2014, and 2015, respectively). The mabs journal published the greatest number of empirical studies in oncology (n = 7), all of which were published in the years (three were published between January and September in 2015) Congresses A total of 27 unique congresses were identified as publishing abstracts of mab biosimilars relevant to oncology between 2007 and August Between January and August 2015, 14 abstracts were identified, highlighting the increasing interest in biosimilars for oncology among academic and industry professionals. The top three congresses identified with the highest number of published abstracts were the American Association of Pharmaceutical Scientists National Biotechnology Conference (AAPS-NBC) (n = 13), the American Society of Clinical Oncology (ASCO) Annual Meeting (n = 11), and the American Society of Hematology (ASH) Annual Meeting (n = 4). The diverse range of congresses identified [i.e., scientific (analytical), oncology specific, regulatory/payer focused] provides evidence of the importance researchers and manufacturers put on disseminating their data to a broad audience. Before 2007, no congress abstracts relevant to the topic of mab biosimilars in oncology were identified. Among the top identified congresses (and up to the last complete year, 2014), the highest number of abstracts was published in 2013 (n = 18), a threefold rise from the previous year (n = 6). In 2013, the congress with the greatest number of relevant published abstracts was the AAPS-NBC (n = 6), followed by ASCO (n = 2), the International Society of Pharmacovigilance (n = 2), the European Cancer Congress (n = 2), the American Association for Cancer Research (n = 2), and the European Society for Animal Cell Technology (n = 2) (Supplementary Table S8). 4 Discussion 4.1 Challenges Relevant to Biosimilar Development Identified in the Literature While biosimilars will eventually play a greater role in the care of patients with cancer, there remain challenges that will temper their immediate impact. We identified several surveys in our review of the literature that highlight some of these challenges. For example, a survey published in 2013 of 400 healthcare professionals (including oncologists, nurses, and pharmacists) demonstrated that most participants had a lack of understanding of the regulatory pathways for biosimilars, as well as a lack of familiarity with biosimilars in the current drug pipeline [81]. A survey of 277 healthcare providers (including physicians, nurses, and pharmacists) conducted by the National Comprehensive Cancer Network (NCCN) showed similar results [6]. When asked about the level of interest in prescribing, dispensing, or administering biosimilars in their practices, 21% of physicians, 31% of nurses, and 26% of pharmacists said they needed more information on biosimilars to make a decision. Among the NCCN recommendations made by the multidisciplinary Biosimilars Working Group was a recommendation for healthcare professionals and policy makers to be educated on biosimilars [6]. The areas that respondents most wanted information on were related to comparative clinical endpoints (safety and efficacy), PK similarity (absorption, distribution, metabolism, and elimination), chemical/physical similarity, and guideline and compendia inclusion. In addition to the need for robust educational programs on biosimilar therapies, several additional factors may influence integration of biosimilars into oncology practices [82]. These include the choice of patient population for efficacy evaluation, the choice of efficacy endpoints, extrapolation of indication, post-approval safety monitoring for immunogenicity, and interchangeability and automatic substitution. Physicians, payers, and other healthcare stakeholders will all influence the uptake and widespread adoption of biosimilars in clinical practice. As indicated in the NCCN survey, oncologists want access to the data used to support FDA approval of biosimilars, and furthermore, they need to be confident in the evidence supporting biosimilarity [6]. It is therefore incumbent on manufacturers to provide as much transparency in their development programs as possible. 4.2 Biosimilars The majority of data comparing an investigational biosimilar with the originator in oncology imply a moderate to high degree of similarity compared with the originator products bevacizumab and trastuzumab, as inferred from study author conclusions (Table 5; Fig. 2a, b). This trend appears to be reflected throughout the stages of biosimilar development, showing that the biosimilar of interest showed high similarity at the structural level (as characterized by biochemical methods), displayed acceptable levels of biological activity (i.e., functional similarity)

31 Biosimilars for Cancer: A Systematic Review of Published Evidence 31 in vitro, and met the required PK thresholds in vivo (usually in mice or monkeys; Table 5; Fig. 2a, b). The evidence generated in preclinical stages supports progression to clinical stages of development. Of the proposed biosimilars identified in this review, a number of molecules have followed the same logical path, including Amgen s ABP 215 (bevacizumab), which, at the time of review, was at more advanced stages of clinical development in advanced NSCLC, having shown a high degree of structural similarity to bevacizumab along with achieving the desired outcomes (performing at the level of the originator or better) in animal models and in early safety and PK trials in humans. Pfizer s PF (bevacizumab) was also pursuing the same course, with comparative safety/efficacy trials underway in NSCLC, having satisfied all data requirements during its preclinical phase. For proposed rituximab biosimilars in development for oncology, GP2013 (Sandoz), PF (Pfizer), and RTXM83 (mabxience) were reportedly in active comparative safety/efficacy studies in lymphoma. Both molecules have shown adequate degrees of similarity to rituximab during preclinical stages of development. Various press releases (in 2012) alluded to the halt in development of SAIT101 (Samsung), owing to concerns over biosimilar regulations in several markets [83]. Results in a PK/safety study of SAIT101 in patients with lymphoma were reported in a single publication in 2014; however, no publications were identified before or since then. For all proposed biosimilars, it should be noted that the number and quality of studies providing evidence of structural and functional similarity at the preclinical phase may not necessarily be sufficient to predict behaviors or performance in humans. Manufacturers are required to provide high-quality evidence of comparable PK/PD and comparative efficacy and safety outcomes (including a comprehensive immunogenicity assessment) for these biosimilars in well-designed trials if they are to satisfy the requirements of the EMA and FDA for approval. According to this review, three of Biocad s molecules [BCD-020 (rituximab), BCD-021 (bevacizumab), and BCD- 022 (trastuzumab)] have entered the clinical development phase without published evidence of underlying structural and functional similarity to their originators, for which a preliminary nonclinical assessment is required. While the authors concede that unpublished data may exist for these molecules, the intent of this research was to uncover the gaps in the literature that may contribute to stakeholder uncertainty about the depth and breadth of evidence underpinning these molecules. In the case of the trastuzumab biosimilars, at the time of this review, PF (Pfizer) was the only molecule that had published data from all nonclinical and clinical stages. CT-P6 (Celltrion) was being investigated in PK/safety and comparative safety/efficacy trials in metastatic BC; FTMB (Amgen/Allergan/Synthon) had data from PK/safety trials (in healthy subjects), but plans for further trials were not reported; and the aforementioned BCD-022, at the time of the analysis, was being investigated in a comparative safety/efficacy study (in metastatic BC). 4.3 Intended Copies In addition to the biosimilars identified in the literature, the authors also reviewed available data for intended copies. The extent of available data between proposed biosimilars and intended copies differs between products [7]. Nonclinical studies on intended copy Kikuzubam (Probiomed) have, thus far, provided some evidence of its structural and functional similarity to rituximab. However, Kikuzubam was marketed as a biosimilar or biocomparable in Mexico, gaining approval without published comparative clinical studies to provide evidence of its safety or efficacy. In 2012, the Mexican Pharmacovigilance Program issued a warning of anaphylactic reactions that occurred in several patients. Subsequently, Kikuzubam was withdrawn from the market by the regulatory authority, owing to the documented anaphylactic reactions and the lack of clinical data [84]. Some countries are now only just beginning to establish formal regulatory guidelines for the approval of biosimilars to cover requirements for preclinical, clinical, or other analyses that should be used to demonstrate the safety, quality, and effectiveness of a biosimilar (including studies required for immunogenicity and AEs), as well as demonstration of similar modes of action or PD properties to that of the originator. Other countries are yet to provide any guidelines. Reditux TM (Dr. Reddy s Laboratories) represents a further example of an intended copy that has been marketed in Latin American markets, India, and Iran, without rigorous data to indicate structural or functional similarity or clinical effectiveness compared with rituximab. Based on the totality of evidence from studies identified during this review, Reditux TM shows heterogeneity with respect to structure, and RCTs are lacking [84, 85]. Although several observational studies have been reported for Reditux TM, only one study (retrospective in design) included rituximab as a comparator. A clear distinction must therefore be made with regard to appropriate terminology for these products, subject to the level and quality of evidence on each, such that they may be easily classified into one of the following categories: approved biosimilars (from a regulatory standpoint), proposed biosimilars (i.e., they are currently in development, but not yet approved by regulatory authorities), and intended copies (i.e., have not undergone rigorous comparative evaluations according to the WHO

32 32 I. Jacobs et al. recommendations, but are being commercialized in some countries). As is evident from this review, manufacturers and researchers do not always clearly state whether they are developing a true biosimilar, or whether they are planning to introduce an intended copy (and appropriately name it as such). 4.4 Knowledge Gaps Biosimilar products require comprehensive testing to ensure biosimilarity in terms of PK/PD, safety, and efficacy. This is to ensure patients are receiving efficacious treatment and are not being put at unnecessary risk of treatment complications, as has occurred in the case of some intended copies being commercialized in emerging markets without having undergone rigorous comparative evaluations according to the WHO recommendations [84]. A perceived advantage of biosimilars is their potential to reduce the acquisition cost of cancer drugs, thereby increasing affordability and patient access. Thus, economic factors will influence the use of these drugs in clinical practice. It has been estimated that biosimilars will cost 20 40% less in the USA, while in the EU, savings of 10 35% have been reported [7]. This is in contrast to the typical 70 80% discount associated with small-molecule generic drugs [7]. Evaluation of cost and potential cost savings will need to be performed by oncologists, institutions, and payers. From a broader societal impact, costeffectiveness analyses will need to be performed in order to manage the differences in perceived value of these drugs when comparing the payer s versus the patient s and caregiver s perceptions. In addition, pharmacovigilance and risk management, as well as safety and tolerability considerations, will need to be taken into account when assessing the value of biosimilars. 4.5 Limitations of this SLR Several limitations of the study should be noted. These include the potential of database searches not capturing all terms related to oncology or mab biosimilars. Also, the determination of proposed biosimilar versus intended copy is limited by the inability to discern the intentions of manufacturers with development candidates. The term proposed biosimilar is employed in this analysis as a blanket term for all development candidates pending final determination of their status as biosimilars. In addition, a relatively small number of selected conference proceedings were searched. Thus, it is possible that data presented in other conference proceedings may not have been included in this analysis. Furthermore, at the time of analysis, limited outcomes data were available from published conference abstracts only, with no full-text publications available at the time of the review. The search for clinical trials was done using the ClinicalTrials.gov results database; no other clinical trial registries were used in this analysis. Therefore, it is possible that trials not registered on ClinicalTrials.gov but registered on other clinical trial databases may have been missed. The degree of biosimilarity, as determined by regulatory authorities, is based on the totality of evidence approach, which includes data on molecular and functional characterization, other nonclinical data, and the safety, PK, immunogenicity, and efficacy clinical trial data. We did not attempt to assess the agents against the criteria used by regulatory authorities, but instead based our analysis on the totality of evidence in the public domain, with biosimilarity determined on the basis of study authors conclusions. Consequently, our analysis is based on the scale of reference (i.e., identical, highly similar, similar, or dissimilar ) used by the study authors. Conclusions on biosimilarity were also drawn collectively from a variety of clinical study types [e.g., RCTs and observational (prospective or retrospective) studies], without accounting for any variation in the overall quality of evidence provided by each study type. One of the original aims of our research was to identify gaps in the published literature. The results presented therefore contain only the outcomes data and statistical comparisons available from the published abstracts or fulltext articles retrieved from the search. Thus, the information collated may not be representative of the full extent of data available for each study, only that which has been published. It should additionally be noted that the search strategy was designed to capture only articles published by MEDLINE or Embase indexed journals. Therefore, studies published by non-indexed journals were not included. A full Internet search of all online content was not included in the methodology. As this review represents a cross-sectional analysis of available published evidence over a defined period, the molecules reviewed were at various stages of development and therefore cannot be compared like for like. Since completion of this review, the authors have identified new biosimilar studies in oncology. These include clinical trials for ABP798 (rituximab; Amgen; CD20? NHL) and for HLX01 (rituximab; Shanghai Henlius Biotech; CD20? DLBCL). The JASMINE study is a randomized, doubleblind study assessing the safety and efficacy of ABP798 compared with rituximab in patients with CD20? NHL. The safety and efficacy of HLX01 is being compared with rituximab in combination with CHOP in previously untreated subjects with CD20? DLBCL.

33 Biosimilars for Cancer: A Systematic Review of Published Evidence 33 5 Conclusions This SLR collates and summarizes all published evidence up to September 2015 on named biosimilars and intended copies of originator mabs that have been developed in the field of oncology. The review identified biosimilars for the marketed biologics bevacizumab, trastuzumab, and rituximab, and many of these biosimilar molecules are in latestage clinical development. Intended copies in development for rituximab were also identified. At present, robust evidence of outcomes for mab biosimilars in cancer, including data from comparative efficacy and safety trials, is not yet substantially available in the published literature. The authors acknowledge that there are a number of planned and ongoing clinical trials that will help inform the current knowledge gaps. While biosimilar treatments are expected to reduce drug costs and increase patient access to much-needed therapies, the potential impact of these drugs in oncology is currently unclear. Only when oncology biosimilars are approved and adopted in routine clinical practice will their impact be fully recognized. Acknowledgements The SLR to support this report was sponsored by Pfizer Inc. Medical writing support was provided by Robyn Fowler, PhD, of Engage Scientific Solutions, and was funded by Pfizer Inc. Carole Jones of Envision Pharma Group was involved with the development of the SLR, which was funded by Pfizer Inc. Author contributions All authors were involved in drafting the article and revising it critically for important intellectual content. All authors read and approved the final manuscript submitted for publication. Compliance with Ethical Standards Conflict of interest IJ, RE, and CZ are full-time employees and shareholders of Pfizer. SL is a full-time employee of Envision Pharma Group, who were paid consultants to Pfizer in connection with the development of the SLR report that forms the basis of this manuscript. He was not compensated for his role in the development of this manuscript. Funding The SLR to support this manuscript was sponsored by Pfizer Inc. Open Access This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License ( which permits any noncommercial use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. References 1. IMS Health. Developments in cancer treatments, market dynamics, patient access and value. Global oncology trend report May 2015 [cited June]; Available from: org/sites/default/files/iihi_oncology_trend_report_2015.pdf. 2. IMS Health. Medicines use and spending in the U.S. A review of 2015 and outlook to April 2016 [cited June]; Available from: /04/IMS-Institute-US-Drug-Spending-2015.pdf. 3. IMS Health. Top 20 global products December 2015 [cited June]; Available from: files/web/corporate/news/top-line%20market%20data/top_20_ Global_Products_2015.pdf. 4. Gori S, Di Maio M, Pinto C, Alabiso O, Baldini E, Beretta GD, et al. Differences in the availability of new anti-cancer drugs for Italian patients treated in different regions. Results of analysis conducted by the Italian Society Of Medical Oncology (AIOM). Tumori. 2010;96(6): Park T, Griggs SK, Suh DC. Cost effectiveness of monoclonal antibody therapy for rare diseases: a systematic review. Bio- Drugs. 2015;29(4): Zelenetz AD, Ahmed I, Braud EL, Cross JD, Davenport-Ennis N, Dickinson BD, et al. NCCN biosimilars white paper: regulatory, scientific, and patient safety perspectives. J Natl Compr Cancer Netw. 2011;9(suppl 4):S Camacho LH, Frost CP, Abella E, Morrow PK, Whittaker S. Biosimilars 101: considerations for U.S. oncologists in clinical practice. Cancer Med. 2014;3(4): Mysler E, Pineda C, Horiuchi T, Singh E, Mahgoub E, Coindreau J, et al. Clinical and regulatory perspectives on biosimilar therapies and intended copies of biologics in rheumatology. Rheumatol Int. 2016;36(5): Jacobs I, Petersel D, Shane LG, Ng CK, Kirchhoff C, Finch G, et al. Monoclonal antibody and fusion protein biosimilars across therapeutic areas: a systematic review of published evidence. BioDrugs. 2016;30(6): Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. J Clin Epidemiol. 2009;62(10): Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. J Clin Epidemiol. 2009;62(10):e National Institute for Health Care Excellence. Single technology appraisal: user guide for company evidence submission template Jan 2015 [cited June]; Available from: nice.org.uk/article/pmg Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan DJ, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials. 1996;17(1): Downs SH, Black N. The feasibility of creating a checklist for the assessment of the methodological quality both of randomised and non-randomised studies of health care interventions. J Epidemiol Community Health. 1998;52(6): Hooijmans CR, Rovers MM, de Vries RB, Leenaars M, Ritskes- Hoitinga M, Langendam MW. SYRCLE s risk of bias tool for animal studies. BMC Med Res Methodol. 2014;14: Markus R, Kaur PP, Chow V, Zhang N, Born TL, Huynh Q, et al. Results of functional testing and pharmacokinetics comparing ABP 215 to bevacizumab. J Clin Oncol. 2015;33(15 suppl): Markus RA, Born T, Chow V, Zhang N, Huynh Q, Maher G, et al. Functional similarity and human pharmacokinetic (PK) equivalence of ABP 215 and bevacizumab. J Clin Oncol. 2015;33(15 suppl):e Born TL, Huynh Q, Mathur A, Velayudhan J, Canon J, Reynhardt K, et al. Functional similarity assessment results comparing

34 34 I. Jacobs et al. bevacizumab to biosimilar candidate ABP 215. Ann Oncol. 2014;25(suppl 4):iv Hutterer K, Liu J, Martinez T, Pace D, Yates Z, Velayudhan J, et al. Analytical similarity assessment of a biosimilar to bevacizumab. In: 19th Symposium on the interface of regulatory and analytical sciences for biotechnology health products (WCBP); January 2015; Washington, DC; p Grunder B. Characterization and similarity assessment of bevacizumab and a proposed biosimilar. In: 2014 American Association of Pharmaceutical Scientists (AAPS) annual meeting and exposition; 2 6 November 2014; San Diego; p. W Peraza M, Shiue M, Phenix S, Rule K, Finch GL, Thibault S, et al. Comparative nonclinical assessment of the potential biosimilar PF and bevacizumab. In: 54th annual meeting of the Society of Toxicology (SOT); March 2015; San Diego; p. Abstr: Archuadze S, Shipaeva P, Dwivedi Y, Lavrovsky M, Samsonov M, Oshodi T. Pharmacokinetic profile of RPH-001, a recombinant humanized monoclonal antibody to human VEGF following administration by intravenous infusion in the cynomolgus monkey. Toxicologist. 2015;144(1): Dorvignit D, Palacios JL, Merino M, Hernández T, Sosa K, Casaco A, et al. Expression and biological characterization of an anti-cd20 biosimilar candidate antibody: a case study. mabs. 2012;4(4): da Silva A, Kronthaler U, Koppenburg V, Fink M, Meyer I, Papandrikopoulou A, et al. Target-directed development and preclinical characterization of the proposed biosimilar rituximab GP2013. Leuk Lymphoma. 2014;55(7): da Silva A, Kronthaler U, Meyer I, Papandrikopoulou A, Stangler T, Visser JM. Target-directed development of a proposed biosimilar rituximab (GP2013): comparability of antibody-dependent cellular cytotoxicity activity and pre-clinical pharmacokinetics and pharmacodynamics with originator rituximab. In: American College of Rheumatology/Association of Rheumatology Health Professionals (ACR/ARHP) Annual Meeting; 9 14 November 2012; Washington, DC; p. S da Silva A, Kronthaler U, Meyer I, Fritsch C, Schneiderer T, Stangler T, et al. Physicochemical, functional, and pharmacologic comparability between the proposed biosimilar rituximab GP2013 and originator rituximab as the foundation for biosimilarity. J Clin Oncol. 2013;31(15 suppl): Visser J, Feuerstein I, Stangler T, Schmiederer T, Fritsch C, Schiestl M. Physicochemical and functional comparability between the proposed biosimilar rituximab GP2013 and originator rituximab. BioDrugs. 2013;27(5): Hurst SI, Ryan AM, Ng C-K, Ploch SA, Finch GL, Leach MW. Nonclinical assessments demonstrating the similarity of the proposed biosimilar PF and rituximab. Ann Rheum Dis. 2013;72(suppl 3):A Ryan AM, Sokolowski SA, Ng C-K, Shirai N, Collinge M, Shen AC, et al. Comparative nonclinical assessments of the proposed biosimilar PF and rituximab (MabThera ). Toxicol Pathol. 2014;42(7): Hurst S, Ryan AM, Ng CK, Thompson M, Plock S, Finch GL, et al. Comparative nonclinical assessments of rituximab-eu (MabThera ) and the potential biosimilar PF In: 2013 American Association of Pharmaceutical Scientists (AAPS) national biotechnology conference; May 2013; San Diego; p. M Karnik S, Thompson M, De Gruttola H, Ferrari S, Gu L, Aggarwal P, et al. Characterization and comparison of PF a proposed rituximab biosimilar to the licensed product. In: 2013 American Association of Pharmaceutical Scientists (AAPS) national biotechnology conference; May 2013; San Diego; p. M Seigelchifer M, Corley E, Fresnillo G, Pesce A, Bes C, Elise M, et al. Development of RTXM83 (a potential rituximab biosimilar): in vitro and in vivo comparability with MabThera. J Clin Oncol. 2014;32(15 suppl):e Boyle PJ, DeGruttola H, Ferrari S, Gu L, Aggarwal P, Thompson M, et al. Characterization and comparison of PF a proposed trastuzumab biosimilar to the licensed product. In: 2013 American Association of Pharmaceutical Scientists (AAPS) National Biotechnology Conference; May 2013; San Diego; p. M Hurst S, Ryan AM, Ng C-K, McNally JM, Lorello LG, Finch GL, et al. Comparative nonclinical assessments of the proposed biosimilar PF and trastuzumab (Herceptin ). Bio- Drugs. 2014;28(5): Hurst S, Ryan A, Ng C, Thompson M, McNally J, Ploch S, et al. Comparative nonclinical assessments of trastuzumab-us and trastuzumab-eu (Herceptin ) and the potential biosimilar PF In: 2013 American Association of Pharmaceutical Scientists (AAPS) national biotechnology conference; May 2013; San Diego; p. M Jacobs IA, Ewesuedo R, Yin D, Li R, Hurst S, Ryan AM, et al. Development of PF , a potential biosimilar to trastuzumab. In: 2015 American Society of Clinical Oncology (ASCO) annual meeting; 29 May 2 June 2015; Chicago; p Ryan AM, Hurst S, McNally JM, Lorello LG, Finch GL, Leach MW, et al. Comparative pharmacokinetics of trastuzumab-us and trastuzumab-eu and the potential biosimilar trastuzumab- Pfizer in male CD-1 mice. Eur J Cancer. 2012;48(suppl 6): Jacobs I, Petersel D, Isakov L, Lula S, Lea Sewell K. Biosimilars for the treatment of chronic inflammatory diseases: a systematic review of published evidence. BioDrugs. 2016;30(6): Wisman L, De Cock E, Reijers J, De Visser I, De Kam M, van Os S, et al. A phase I dose escalation and bioequivalence study of a trastuzumab biosimilar (FTMB) in healthy male volunteers. In: 2012 European Society for Medical Oncology (ESMO) congress; 28 September 2 October 2012; Vienna; p. ix103 ix Wisman LAB, De Cock EPM, Reijers JAA, Kamerling IMC, Van Os SHG, de Kam ML, et al. A phase I dose-escalation and bioequivalence study of a trastuzumab biosimilar in healthy male volunteers. Clin Drug Investig. 2014;34(12): Ricart AD, Zacharchuk C, Reich SD, Meng X, Barker KB, Taylor CT, et al. A phase I pharmacokinetics trial comparing PF and trastuzumab in healthy volunteers (REFLEC- TIONS B327-01). In: 35th Annual Cancer Therapy & Research Center and American Association for Cancer Research (CTRC- AACR) San Antonio Breast Cancer Symposium (SABCS); 4 8 December 2012; San Antonio: Cancer Res; p. OT Yin D, Barker KB, Li R, Meng X, Reich SD, Ricart AD, et al. A phase I pharmacokinetics trial comparing PF (a potential biosimilar) and trastuzumab in healthy volunteers (REFLECTIONS B327-01). In: 2013 American Society of Clinical Oncology (ASCO) annual meeting; 31 May 4 June 2013; Chicago; p Yin D, Barker KB, Li R, Meng X, Reich SD, Ricart AD, et al. A randomized phase 1 pharmacokinetic trial comparing the potential biosimilar PF with trastuzumab in healthy volunteers (REFLECTIONS B327-01). Br J Clin Pharmacol. 2014;78(6): Yin D, Cai CH, Taylor CT, Zacharchuk CM, Rudin D, Reich SD, et al. Immunogenicity assessment of PF , a potential biosimilar to trastuzumab, in healthy subjects (REFLECTIONS B327-01). Eur J Cancer. 2013;49(suppl 2):S Orlov SV, Burdaeva ON, Nachaeva MP, Kopp MV, Kotiv BN, Sheveleva LP, et al. Pharmacokinetics and safety of BCD-021, bevacizumab biosimilar candidate, compared to Avastin in patients. J Clin Oncol. 2014;32(15 suppl):e13500.

35 Biosimilars for Cancer: A Systematic Review of Published Evidence Filon O, Orlov SV, Burdaeva ON, Kopp MlV, Kotiv BN, Alekseev S, et al. Efficacy and safety of BCD-021, bevacizumab biosimilar candidate, compared to Avastin: results of international multicenter randomized double blind phase III study in patients with advanced non-squamous NSCLC. In: 2015 American Society of Clinical Oncology (ASCO) annual meeting; 29 May 2 June 2015; Chicago: J Clin Oncol; p Alexeev S, Zaritskey A, Volodicheva E, Loginov A, Orlova R, Dvornichenko V, et al. Clinical comparability of BCD-020 to innovator rituximab in patients with indolent non-hodgkin s lymphoma. Haematologica. 2014;99(suppl 1): Kaplanov K, Zaritskiy A, Alexeev S, Volodicheva E, Loginov A, Orlova R, et al. Key results of international randomized openlabel clinical study of BCD-020 (rituximab biosimilar candidate) in patients with B-cell non-hodgkin s lymphoma. Blood. 2014;124(21): Poddubnaya I, Babicheva L, Kaplanov K, Zaritskey A, Volodicheva E, Alexeev S, et al. Comparison of pharmacokinetics and pharmacodynamics of BCD-020 with innovator rituximab to patients with indolent non-hodgkin lymphoma. J Clin Oncol. 2014;32(15 suppl):e Florez A, Di Matteo T, Fresnillo G, Tudela C, Seigelchifer M, Corley E, et al. Clinical pharmacokinetic (PK) and safety (immunogenicity) of rituximab biosimilar RTXM83 in combination with chemotherapy CHOP in patients with diffuse large B-cell lymphoma (DLBCL). Blood. 2014;124(21): Kim SJ, Kim WS, Kang HJ, Kim JS, Choi CW, Lee SI, et al. Safety, pharmacokinetic/pharmacodynamic profiles and efficacy of SAIT101, a biosimilar of rituximab in patients with diffuse large B-cell lymphoma. Haematologica. 2012;97(suppl 1):S Stenina MB, Ignatova E, Frolova MA, Burdaeva ON, Nechaeva MP, Kopp MV, et al. Pharmacokinetics and safety of BCD-022, trastuzumab biosimilar candidate, compared to Herceptin in patients. J Clin Oncol. 2014;32(15 suppl):e Im Y, Krasnozhon D, Bondarenko I, Zvirbule Z, Jung K, Oliynychenko P, et al. Phase I/IIb clinical trial comparing PK and safety of trastuzumab and its biosimilar candidate CT-P6. Breast. 2013;22(suppl 1):S Im Y-H, Odarchenko P, Grecea D, Komov D, Anatoliy CV, Gupta S, et al. Double-blind, randomized, parallel group, phase III study to demonstrate equivalent efficacy and comparable safety of CT-P6 and trastuzumab, both in combination with paclitaxel, in patients with metastatic breast cancer (MBC) as first-line treatment. J Clin Oncol. 2013;31(15 suppl): Ewesuedo R, Barker KB, Taylor CT, Jacobs I. A phase 3 randomized, double-blind trial comparing PF ? paclitaxel vs. trastuzumab? paclitaxel for treatment of HER2? metastatic breast cancer. Cancer Res. 2013;73(24 suppl). 56. Jacobs I, Coiro J, Hilton F, Orazem J, Abbas R, Zacharchuk C. A phase 3 randomized, double-blind trial comparing PF ? docetaxel and carboplatin vs. trastuzumab? docetaxel and carboplatin for neoadjuvant treatment of operable HER2? breast cancer. Cancer Res. 2015;75(9 suppl):abstract OT Flores-Ortiz LF, Campos-García VR, Perdomo-Abúndez FC, Pérez NO, Medina-Rivero E. Physicochemical properties of rituximab. J Liq Chromatogr Relat Technol. 2014;37(10): Thakral P, Singla S, Yadav MP, Vasisht A, Sharma A, Gupta SK, et al. An approach for conjugation of 177 Lu-DOTA-SCN- rituximab (BioSim) & its evaluation for radioimmunotherapy of relapsed & refractory B-cell non Hodgkins lymphoma patients. Indian J Med Res. 2014;139(4): Roy PS, John S, Kannan S, Gujral S, Gawande J, B. B, et al. Comparison of efficacy and safety of rituximab (Mabthera) and its biosimilar (Reditux) in diffuse large B cell lymphoma (DLBCL) patients treated with chemo-immunotherapy: a retrospective analysis. In: 53rd National Conference of Indian Society of Hematology & Blood Transfusion (ISHBT); 9 11 November 2012; Puri: Springer; p. Abstr: Roy PS, John S, Karankal S, Kannan S, Pawaskar P, Gawande J, et al. Comparison of the efficacy and safety of rituximab (Mabthera TM ) and its biosimilar (Reditux TM ) in diffuse large B-cell lymphoma patients treated with chemo-immunotherapy: a retrospective analysis. Indian J Med Paediatr Oncol. 2013;34(4): Roy PS, Sengar M, Menon H, et al. A retrospective single centre analysis of safety, toxicity and efficacy of rituximab (original) and its biosimilar in diffuse large B-cell lymphoma patients treated with chemo-immunotherapy. In: 2012 European Society for Medical Oncology (ESMO) Congress; 28 September 2 October 2012; Vienna; p. Abstr: ix Aliaga L, Fernandez I, Sanchez M, Saavedra H, Espinoza C. Pharmacovigilance of anti CD 20 monoclonal antibody biosimilar at the Edgardo Rebagliati Martins Hospital-Peru. Drug Saf. 2013;36(9): Menon H, Yadav A, Subramanian PG, Sengar M, Rath S, Kavathiya K, et al. Pharmacokinetic and pharmacodynamic properties of a biosimilar rituximab (Reditux ) are identical to the innovator brand MabThera experience from a tertiary cancer centre in Western India. Blood. 2014;124(21): Kumar V, Vemuri S, Lella RK, Babu MN, Atluri A, Gowravaram M, et al. Suitability of one assay for immunogenicity assessment of both proposed biosimilar and innovator rituximab. In: 2014 American Association of Pharmaceutical Scientists (AAPS) national biotechnology conference; 2 6 November 2014; San Diego; p. M Mekhssian K, Mess JN, Montalibet J, Garofolo F. Application of complementary HRMS methodologies for a thorough biosimilar comparability assessment. In: 2015 American Association of Pharmaceutical Scientists (AAPS) national biotechnology conference; 8 10 June 2015; San Francisco; p. T Lin S, Ramatlapeng D, Hanning J, Bakewell W. Comparability study of monoclonal biosimilar drug products using various analytical methods. In: 18th symposium on the interface of regulatory and analytical sciences for biotechnology health products (WCBP); January 2014; Washington, DC; p Beck A, Sanglier-Cianférani S, Van Dorsselaer A. Biosimilar, biobetter, and next generation antibody characterization by mass spectrometry. Anal Chem. 2012;84(11): da Silva A, Grau R, Stangler T, Summer H, Meyer I, Rohde A, et al. Comprehensive target-directed approach for the development of a highly-comparable rituximab biosimilar. Ann Rheum Dis. 2012;71(suppl 3): Ebbers HC, van Meer PJ, Moors EH, Mantel-Teeuwisse AK, Leufkens HG, Schellekens H. Measures of biosimilarity in monoclonal antibodies in oncology: the case of bevacizumab. Drug Discov Today. 2013;18(17 18): Kurbanova EK, Chulkova TY, Byrikhina DV, Dano AO, Nikolaeva VA, Gusarov DA. PHARMA-2020: technological aspects of production of some strategical biopharmaceuticals. Russ J Biopharm. 2012;4(3): Li E, Subramanian J, Anderson S, Thomas D, McKinley J, Jacobs IA. Development of biosimilars in an era of oncologic drug shortages. Drug Des Dev Ther. 2015;9: What are biosimilars and are they important? Drug Ther Bull. 2013;51(5): Cragg MS. CD20 antibodies: doing the time warp. Blood. 2011;118(2):

36 36 I. Jacobs et al. 74. Dhingra K. Biosimilars: the industry perspective. J Crohns Colitis. 2009;3(suppl 1): Kay J. Biosimilars: a regulatory perspective from America. Arthritis Res Ther. 2011;13(3): Kudrin A, Knezevic I, Joung J, Kang HN. Case studies on clinical evaluation of biosimilar monoclonal antibody: scientific considerations for regulatory approval. Biologicals. 2015;43(1): Nelson KM, Gallagher PC. Biosimilars lining up to compete with Herceptin opportunity knocks. Expert Opin Ther Pat. 2014;24(11): Rioufol C, Salles G. Biosimilar monoclonal antibodies in lymphoma: a critical appraisal. Expert Rev Anticancer Ther. 2015;15(5): Thill M. New frontiers in oncology: biosimilar monoclonal antibodies for the treatment of breast cancer. Expert Rev Anticancer Ther. 2015;15(3): Vital EM, Kay J, Emery P. Rituximab biosimilars. Expert Opin Biol Ther. 2013;13(7): Krejcik C. NACCME Releases Results of survey assessing medical education needs in biosimilars [news release] May [cited 2016 October 20]; Available from: prweb.com/releases/2013/5/prweb htm. 82. Rugo HS, Linton KM, Cervi P, Rosenberg JA, Jacobs I. A clinician s guide to biosimilars in oncology. Cancer Treat Rev. 2016;46: Macdonald G. Samsung halts biosimilar rituximab trial on regulatory concerns [news release] Oct 2012 [cited June]; Available from: Developments/Samsung-halts-biosimilar-rituximab-trial-onregulatory-concerns. 84. Castañeda-Hernández G, González-Ramírez R, Kay J, Scheinberg MA. Biosimilars in rheumatology: what the clinician should know. RMD Open. 2015;1(1):e Espinosa-de la Garza CE, Perdomo-Abúndez FC, Padilla-Calderón J, Uribe-Wiechers JM, Pérez NO, Flores-Ortiz LF, et al. Analysis of recombinant monoclonal antibodies by capillary zone electrophoresis. Electrophoresis. 2013;34(8):

37 Current Medical Research and Opinion ISSN: (Print) (Online) Journal homepage: Comparability and biosimilarity: considerations for the healthcare provider Jaymi F. Lee, Jason B. Litten & Gustavo Grampp To cite this article: Jaymi F. Lee, Jason B. Litten & Gustavo Grampp (2012) Comparability and biosimilarity: considerations for the healthcare provider, Current Medical Research and Opinion, 28:6, , DOI: / To link to this article: Accepted author version posted online: 20 Apr Published online: 06 Jun Submit your article to this journal Article views: 1486 View related articles Citing articles: 12 View citing articles Full Terms & Conditions of access and use can be found at Download by: [F. Hoffmann - La Roche Inc] Date: 22 December 2016, At: 13:34

38 Current Medical Research & Opinion Vol. 28, No. 6, 2012, Article ST-0010.R1/ doi: / All rights reserved: reproduction in whole or part not permitted Brief review Comparability and biosimilarity: considerations for the healthcare provider Jaymi F. Lee Drug Product Development, Amgen Inc., Thousand Oaks, CA, USA Jason B. Litten Hematology/Oncology, Amgen Inc., Thousand Oaks, CA, USA Gustavo Grampp Regulatory Affairs, Amgen Inc., Longmont, CO, USA Address for correspondence: Jaymi F. Lee, MPH, Scientist, Drug Product Development, Amgen Inc., One Amgen Center Drive, MS 8-1-C, Thousand Oaks, CA 91320, USA. Tel.: þ ; Fax: ; Keywords: Biologic Biosimilar Biosimilarity Comparability Innovator Manufacturing Accepted: 17 April 2012; published online: 31 May 2012 Citation: Curr Med Res Opin 2012; 28: Abstract Background: Healthcare providers use recombinant biologics such as monoclonal antibodies to treat a variety of serious illnesses. Manufacturing of approved biotechnology products is complex, and the quality of the resulting biologic is dependent on careful control of process inputs and operating conditions. Biosimilars, which are similar but not identical to innovator biologics, are entering regulatory evaluation, approval, and marketing in regions with biosimilar approval pathways. Scope and findings: This article describes the evaluation and potential impact of manufacturing process changes and biosimilar product development, and explores the similarities and distinctions between the two. Regulatory agencies generally require a comparability exercise following a manufacturing process change. This comparability is focused primarily on analytical characterization of the approved product before and after the manufacturing process change, with non-clinical and clinical confirmation required when determined necessary. When developing a biosimilar, the manufacturer does not have access to key information including the innovator manufacturer s cell line, cell culture conditions, purification procedures, and fill and finish processes. Further, the biosimilar manufacturer does not have access to information about the innovator manufacturer s product development history, including knowledge about the quality attributes of lots used in non-clinical and clinical development. We define the biosimilar manufacturer s lack of information as the knowledge gap. As a result, a biosimilarity exercise to compare a biosimilar to an approved innovator biologic requires a rigorous evaluation to ensure the safety and efficacy of the biosimilar. Copyright 2012 Informa UK Limited Not for Sale or Commercial Distribution Conclusion: Given the knowledge gap under which biosimilars are developed, data to establish biosimilarity should go beyond a simple comparability exercise. Unauthorized use prohibited. Authorised users can download, display, view and print a single copy for personal use Background Currently, more than 250 biologic molecules derived from recombinant DNA technology have been approved for use in patients to treat many serious illnesses 1. While use of these products has resulted in improved patient care and outcomes, healthcare providers (HCPs) may not recognize the complexity of the design, manufacturing and delivery of these medicines. Biologics are a diverse group of medicines that are produced and isolated from living systems, including bacteria, yeast and mammalian cells. Modifications of the primary, secondary, tertiary, or quaternary structure of an approved biologic may impact its potency, purity and safety that underlie its approval. As a result, meticulous attention is! 2012 Informa UK Ltd Comparability and biosimilarity Lee et al. 1053

39 Current Medical Research & Opinion Volume 28, Number 6 June 2012 Biologic Manufacturing Cell Culture/Fermentation & Purification Cell Expansion Product Recovery Purification Bulk Drug Substance Fill & Finish Bulk Freeze/ Thaw Formulation Filtration Fill Inspection Labeling and Packaging Figure 1. The complexity of commercial biologic manufacturing. required to assure the proper function and integrity of the manufacturing environment for producing approved biologics. Factors that impact the quality of the manufactured commercial product include but are not limited to: source and type of raw materials used; chemical substances that can leach from containers and equipment; and conditions such as temperature, ph and agitation 2. These factors become more critical in the context of glycosylated proteins and monoclonal antibodies that often consist of various structural isoforms. Identification and characterization of isoforms are particularly important because some variants are considered impurities that may alter biologic activity, safety and/or immunogenicity 3. Thus, the presence of these isoforms and structural variants results in an end product that is a complex mixture. This complexity stands in contrast to the well-defined homogeneous (498%) molecular structures that characterize traditional small-molecule drugs, e.g. statins. Biologics are up to 1000 times larger in molecular weight and more structurally complex than traditional small-molecule drugs. Due to this heterogeneity, there is no single test of identity that can verify the distribution and optimal proportion of a biologic s isoforms 4. In the context of prescribing any approved biologic, it is important for HCPs to understand the challenges faced in biologic manufacturing and the potential clinical implications. Information that can assist prescribing decisions concerning the utilization of similar biologics includes clinical trial results, regulatory approvals and product labels. In regions with a biosimilar approval pathway, biosimilars, which are similar but not identical to innovator biologics, may be entering regulatory evaluation, approval and marketing. This article explores the evaluation and potential impact of commercial manufacturing process changes and biosimilar development, and the similarities and differences between the two exercises. The complexity of commercial biologic manufacturing The typical steps in the manufacture of an approved recombinant biologic are illustrated in Figure 1. The cell culture and fermentation processes are particularly critical and sensitive in terms of defining the identity, purity and potency of the approved biologic. Modifications of parameters in any of these steps may impact cell culture performance, leading to variability in the quality of the commercial recombinant protein 5 7. Modern analytical science and process design and controls provide manufacturers and regulators with a high degree of assurance that a process, once validated, will perform consistently over time, and that unexpected excursions of product characteristics will be detected and investigated appropriately. These considerations are important for ongoing manufacturing; however, the structural complexity and process sensitivity of biologics are particularly relevant in the context of a manufacturing process change. A manufacturer may introduce a change to the manufacturing process of an approved biologic for a variety of reasons, including the need to: comply with regulatory commitments, improve product quality and yield, and improve manufacturing efficiency and reliability Comparability and biosimilarity Lee et al Informa UK Ltd

40 Current Medical Research & Opinion Volume 28, Number 6 June 2012 Nature of Process Change Change filter supplier Move equipment within same facility Move to new production facility (same manufacturer) Change cell culture media New cell line or major formulation change Risk Level / Data Requirements Low Risk Moderate Risk High Risk Analytical date Analytical data Analytical data Process data Process data Process data Stability data Stability data Non-clinical data Clinical data Figure 2. The nature of a manufacturing change determines the amount and type of supporting data required to evaluate comparability. Manufacturing process changes are governed by the ICH Q5E comparability guidance 11,12. Depending on the nature and extent of the manufacturing change, routine control measures and analytical tests may not be sufficient to assess the impact of the change on a product s quality, safety and efficacy; this may necessitate non-clinical and clinical evaluations 8,9. Comparability exercises following a manufacturing process change Historically, recombinant biologics were considered so complex that changes in the design of any critical steps in a manufacturing process were rarely, if ever, considered. However, in the time following approval of the first recombinant biologics, regulatory agencies accumulated sufficient experience with analytical characterization and, by proxy, the structure function relationship of these approved products to propose what is known as a comparability exercise following a manufacturing change 8,9. In a 1996 guidance on comparability, the Food and Drug Administration (FDA) provided the biotechnology industry with the opportunity to modernize and improve manufacturing processes 10. The European Medicines Agency (EMA) and other regulatory bodies provided analogous guidelines, culminating in the International Conference on Harmonization comparability guidance ICH Q5E 11,12, now recognized by regulatory authorities throughout the world. In certain situations, a comparability evaluation permits a manufacturer to provide a robust package of analytical data comparing the properties of the products before and after the manufacturing process change to sufficiently conclude that there has been no adverse impact to the product s quality (Figure 2). For biologic manufacturers, the result in many cases has been the opportunity to improve efficiency and reliability of manufacturing processes without repeating clinical trials for minor changes. In all cases, manufacturers make these decisions in direct consultation with regional regulatory agencies, focusing on patient safety as the primary consideration. The comparability evaluation is informed by a comprehensive risk assessment that considers knowledge of the existing product, scope of the process change, potential impact of the process change, and limitations of analytical methods to detect potentially relevant changes in the product. For example, if a change to a given processing step could potentially result in formation of partially oxidized product impurities, one may ask whether there has been prior experience with such impurities during the product s development history, whether analytical methods exist to detect subtle changes in the product s oxidation state, and whether the impact of oxidation can be adequately evaluated with structural and biological characterization studies. These manufacturing changes often involve one or a few discrete steps, e.g. raw material change or process modification. In most cases, the risks of such a process change can either be completely mitigated or otherwise justified. However, if these considerations are not fully addressed with analytical studies, it may be necessary to perform nonclinical and clinical studies to confirm the product s continued safety and efficacy. While such clinical bridging studies are often successful, there are published examples where unexpected clinical findings have been observed following a major process change. Non-equivalent pharmacokinetics of products before and after process changes were observed following development of a new cell bank 13, transfer of manufacturing from one organization to another 14, and scale-up of a manufacturing process within the same organization 15. Due to the unexpected clinical outcomes, the process changes were not pursued in one case 13, while in the other two cases the manufacturers performed additional clinical studies to justify the process changes 14,15. These examples demonstrate the value of clinical studies in assessing the potential impact of manufacturing changes when the risks cannot be completely assessed based on analytical studies.! 2012 Informa UK Ltd Comparability and biosimilarity Lee et al. 1055

41 Current Medical Research & Opinion Volume 28, Number 6 June 2012 Abbreviated Comparability (same manufacturer) Modified Process Prior Findings of Safety and Efficacy Licensed Innovator Product and Process Moderate risk manufacturing change (ICH Q5E)* Knowledge Gap Comprehensive Comparability (same manufacturer) High risk manufacturing change (ICH Q5E)* Biosimilarity New Process Biosimilar Process Figure 3. Data requirements for comparability exercises for a moderate risk and a high risk manufacturing change by the same manufacturer compared with a biosimilar manufacturing process. *Manufacturing process changes are governed by the ICH Q5E comparability guidance 11,12. Comparability and biosimilarity Innovator manufacturers begin with a deep institutional knowledge of the development of their product, including cell line, culture conditions, purification, formulation, and delivery of the biologic. This knowledge guides the design of bridging studies aimed at demonstrating comparability of the product before and after a manufacturing change (Figure 3). Depending on the extent of the manufacturing change, analytical, non-clinical and/or clinical studies may be required as informed by ICH Q5E guidance 11,12 and in discussions with regulatory agencies. Most of the innovator manufacturer s product information is proprietary and not available in the public domain for use by a biosimilar manufacturer; thus, the biosimilar manufacturer experiences what we refer to as a knowledge gap (Figure 3). Due to this knowledge gap, the biosimilar manufacturer must start with development of a new cell line, and then design and develop the manufacturing process in its entirety. Further, the biosimilar manufacturer must demonstrate analytical and preclinical biosimilarity of its product to the innovator biologic. Finally, to meet biosimilarity requirements, clinical studies are required for approval in the EU 16,17. In February of 2012, FDA issued draft guidance for industry titled Scientific Considerations in Demonstrating Biosimilarity to a Reference Product 18. In this document, FDA acknowledges the difficulty of demonstrating biosimilarity and clarifies the limitations of comparing this exercise with a comparability exercise after a manufacturing process change; stating: Demonstrating that a proposed product is biosimilar to a reference product typically will be more complex than assessing the comparability of a product before and after manufacturing changes made by the same manufacturer. This is because a manufacturer who modifies its own manufacturing process has extensive knowledge and information about the product and the existing process, including established controls and acceptance parameters. In contrast, the manufacturer of a proposed product will likely have a different manufacturing process (e.g., different cell line, raw materials, equipment, processes, process controls, and acceptance criteria) from that of the reference product and no direct knowledge of the manufacturing process for the reference product. Therefore, even though some of the scientific principles described in ICH Q5E may also apply in the demonstration of biosimilarity, in general, more data and information will be needed to establish biosimilarity than would be needed to establish that a manufacturer s post-manufacturing change product is comparable to the 1056 Comparability and biosimilarity Lee et al Informa UK Ltd

42 Current Medical Research & Opinion Volume 28, Number 6 June 2012 pre-manufacturing change product 18. From these statements, it is clear that both FDA and EMA approach biosimilarity determinations with scientific rigor focused on protecting patient safety. European regulatory experience with biosimilars European biosimilar manufacturers have been working with EMA to establish standards for biosimilarity for over a decade, and EMA has provided an abbreviated biosimilar approval pathway. This early experience defining biosimilarity is likely to inform biosimilar regulations in other regions. It is important to note that the first approved EU biosimilars were small proteins related to naturally occurring proteins 19,20, e.g. human growth hormone, insulin, erythropoietin, and filgrastim. In all instances to date, clinical studies have been required for biosimilar approvals 19, likely due to the previously defined knowledge gap. In these cases, biosimilarity was assessed using robust and sensitive efficacy endpoints. Some biosimilar products approved in the EU have shown different physical, chemical or immunogenicity profiles compared with the innovator biologics 19, Of the biosimilar products that have completed EMA review (from 2006 to 2011) 19, six produced outcomes that were not expected based on non-clinical testing. Of these six, four failed to demonstrate similar safety and/or efficacy to the innovator products and were withdrawn or rejected For the remaining two to achieve marketing authorizations, one manufacturer modified its manufacturing process and repeated a clinical study 29 and the other was denied expansion of the label for subcutaneous use 21. Building on this early experience, the next generation of biosimilar marketing applications will be for large, structurally complex, and heterogeneous protein drugs, including monoclonal antibodies and fusion proteins 19. Therefore, analytical, non-clinical and/or clinical studies will likely be required to show biosimilarity to the innovator products. These studies will be critical to minimize risk of adverse outcomes to patients and to confirm comparable efficacy. Summary and conclusion: considerations for the healthcare provider Throughout the product life cycle of an approved biologic molecule, a manufacturer may implement process changes to incorporate technological advances or efficiencies. Regulators evaluate these changes carefully and use scientific comparability criteria to determine whether there is a potential impact on the safety or efficacy that underlies its approval. Most of these manufacturing changes are evaluated with analytic studies designed to assess the potential for an unexpected change in the quality of the commercial product. Information on the changes introduced and the data used to support these changes are generally not available in the public domain. Occasionally, however, a major manufacturing change will require that the manufacturer completes a clinical trial to demonstrate that there is no impact on the product s safety or efficacy. While the evaluation of a new manufacturing process for a biosimilar may parallel a manufacturing process change for a biologic, there are important distinctions to consider. The biosimilar manufacturer must develop the manufacturing process in its entirety, from cell line selection to fill and finish. These activities are performed without full access to the innovator manufacturer s product development history. These important variables constitute what we define as the knowledge gap. Therefore, the potential for differences between an innovator biologic and a biosimilar is greater than that between a biologic before and after a manufacturing change. Due to the potential for differences, a biosimilar manufacturer should provide data to ensure that differences between the biosimilar and the innovator biologic will not impact the efficacy or safety of its product. Considering the current limitations of analytics, these data should include comparative clinical testing to confirm biosimilarity. Transparency Declaration of funding This work was funded by Amgen Inc. Declaration of financial/other relationships J.F.L., J.B.L., and G.G. are Amgen employees and own Amgen stock. Authorship All authors were involved in the development of this manuscript and have fulfilled the ICMJE authorship criteria. Additionally, all authors have approved the final version of the manuscript for submission to CMRO. CMRO peer reviewers may have received honoraria for their review work. The peer reviewers on this manuscript have disclosed that they have no relevant financial relationships. Acknowledgments The authors thank Martha Mutomba from Amgen Inc. for editing assistance. References 1. Biotechnology Industry Organization website. Available at: node/517 [Last accessed 19 December 2011]! 2012 Informa UK Ltd Comparability and biosimilarity Lee et al. 1057

43 Current Medical Research & Opinion Volume 28, Number 6 June Rathore N, Rajan RS. Current perspectives on stability of protein drug products during formulation, fill and finish operations. Biotechnol Prog 2008;24: Sharma B. Immunogenicity of therapeutic proteins. Part 3: Impact of manufacturing changes. Biotechnol Adv 2007;25: Kleinberg M, Mosdell KW. Current and future considerations for the new classes of biologicals. Am J Health Syst Pharm 2004;61: Kirdar AO, Chen G, Weidner J, et al. Application of near-infrared (NIR) spectroscopy for screening of raw materials used in the cell culture medium for the production of a recombinant therapeutic protein. Biotechnol Prog 2010;26: Lubiniecki A, Volkin DB, Federici M, et al. Comparability assessments of process and product changes made during development of two different monoclonal antibodies. Biologicals 2011;39: Luo Y, Chen G. Combined approach of NMR and chemometrics for screening peptones used in the cell culture medium for the production of a recombinant therapeutic protein. Biotechnol Bioeng 2007;97: Lewis RM, Cosenza ME. Summary of DIA Workshop: Comparability challenges: Regulatory and scientific issues in the assessment of biopharmaceuticals. Drug Inf J 2010;44: Putnam WS, Prabhu S, Zheng Y, et al. Pharmacokinetic, pharmacodynamic and immunogenicity comparability assessment strategies for monoclonal antibodies. Trends Biotechnol 2010;28: Demonstration of comparability of human biological products, including therapeutic biotechnology-derived products. Rockville, MD: FDA/Center for Biologics Evaluation and Research (CBER)/Center for Drug Evaluation and Research (CDER), Available at: Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm htm [Last accessed 19 December 2011] 11. Q5E Comparability of biotechnological/biological products subject to changes in their manufacturing process. Rockville, MD: FDA/Center for Drug Evaluation and Research (CDER)/Center for Biologics Evaluation and Research (CBER), Available at: RegulatoryInformation/Guidances/ucm pdf [Last accessed 19 December 2011] 12. Comparability of biotechnological/biological products subject to changes in their manufacturing process, Q5E. International Conference on Harmonization/ICH Expert Working Group, Available at: Step4/Q5E_Guideline.pdf [Last accessed 9 March 2012] 13. Avonex Õ Summary Basis for Approval. Rockville, MD: FDA/Center for Biologics Evaluation and Research (CBER)/Center for Drug Evaluation and Research (CDER), Available at: DevelopmentApprovalProcess/HowDrugsareDevelopedandApproved/ ApprovalApplications/TherapeuticBiologicApplications/ucm pdf [Last accessed 19 December 2011] 14. Raptiva Õ European Public Assessment Report Scientific Discussion. London: European Medicines Agency, Available at: docs/en_gb/document_library/epar_-_scientific_discussion/human/000542/ WC pdf [Last accessed 16 March 2012] 15. Meeting of the Endocrinologic and Metabolic Drugs Advisory Commitee (EMDAC) to Evaluate Myozyme (alglucosidase alfa) for the treatment of late onset Pompe Disease. Rockville, MD: FDA/Center for Biologics Evaluation and Research (CBER)/Center for Drug Evaluation and Research (CDER), Available at: [Last accessed 19 December 2011] 16. Guideline on similar biological medicinal products. London: European Medicines Agency, Available at: human/biosimilar/043704en.pdf [Last accessed 19 December 2011) 17. Rossert J. EMEA guidelines on biosimilars and their clinical implications. Kidney Blood Press Res 2007;30(Suppl 1): Scientific Considerations in Demonstrating Biosimilarity to a Reference Protein Product; Draft Guidance; 9 February Rockville, MD: FDA/Center for Biologics Evaluation and Research (CBER)/Center for Drug Evaluation and Research (CDER), Available at: Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM pdf [Last accessed 6 March 2012] 19. Miletich J, Eich G, Grampp G, et al. Biosimilars 2.0: Guiding principles for a global patients first standard. MAbs 2011;3: Schellekens H. Biosimilar therapeutics-what do we need to consider? NDT Plus 2009;2:i Haag-Weber M, Eckardt KU, Hörl WH, et al. Safety, immunogenicity and efficacy of subcutaneous biosimilar epoetin-a (HX575) in non-dialysis patients with renal anemia: a multi-center, randomized, double-blind study. Clin Nephrol 2012;77: Niederwieser D, Schmitz S. Biosimilar agents in oncology/haematology: From approval to practice. Eur J Haematol 2010;86: Schellekens H. Recombinant human erythropoietins, biosimilars and immunogenicity. J Nephrol 2008;21: Schellekens H. Assessing the bioequivalence of biosimilars: The Retacrit case. Drug Discov Today 2009;14: Refusal assessment report for Alpheon. London: European Medicines Agency, Available at: WC pdf [Last accessed 16 March 2012] 26. Withdrawal assessment for Insulin Human Rapid Marvel. London: European Medicines Agency, Available at: en_gb/document_library/application_withdrawal_assessment_report/ 2010/01/WC pdf [Last accessed 16 March 2012] 27. Withdrawal assessment for Insulin Human Long Marvel. London: European Medicines Agency, Available at: en_gb/document_library/application_withdrawal_assessment_report/ 2010/01/WC pdf [Last accessed 16 March 2012] 28. Withdrawal assessment for Insulin Human 30/70 Marvel. London: European Medicines Agency Available at: en_gb/document_library/application_withdrawal_assessment_report/ 2010/01/WC pdf [Last accessed 16 March 2012] 29. Omnitrope Õ European Public Assessment Report Scientific Discussion. London: European Medicines Agency, Available at: europa.eu/docs/en_gb/document_library/epar_-_scientific_discussion/ human/000607/wc pdf [Last accessed 16 March 2012] 1058 Comparability and biosimilarity Lee et al Informa UK Ltd

44 4 August 2016 EMA/168402/2014 Guideline on good pharmacovigilance practices (GVP) Product- or Population-Specific Considerations II: Biological medicinal products Draft finalised by the Agency in collaboration with Member States 17 November 2015 Draft agreed by the European Risk Management Strategy Facilitation Group (ERMS FG) 24 November 2015 Draft adopted by Executive Director 8 December 2015 Start of public consultation 15 December 2015 End of consultation (deadline for comments) 29 February 2016 Revised draft finalised by the Agency in collaboration with Member States 9 June 2016 Revised draft agreed by ERMS FG 26 July 2016 Revised draft adopted by Executive Director as final 4 August 2016 Date for coming into effect 16 August 2016 See websites for contact details European Medicines Agency Heads of Medicines Agencies The European Medicines Agency is an agency of the European Union European Medicines Agency and Heads of Medicines Agencies, Reproduction is authorised provided the source is acknowledged.

45 Table of contents P.II.A. Introduction... 4 P.II.A.1. Pharmacovigilance aspects specific to biologicals... 5 P.II.A.1.1. Immunogenicity... 5 P.II.A.1.2. Manufacturing variability... 6 P.II.A.1.3. Stability and cold chain... 7 P.II.A.1.4. Product traceability... 8 P.II.B. Structures and processes... 8 P.II.B.1. Risk management system... 8 P.II.B.1.1. Content of the risk management plan (RMP)... 9 P.II.B RMP part I Product overview... 9 P.II.B RMP part II Safety specification... 9 P.II.B RMP module SVII Identified and potential risks and RMP module SVIII Summary of the safety concerns... 9 P.II.B RMP module SVI Additional EU requirements for the safety specification.. 10 P.II.B RMP part III Pharmacovigilance plan P.II.B RMP part III section Routine pharmacovigilance activities P.II.B RMP part III section Additional pharmacovigilance activities P.II.B RMP part V Risk minimisation measures P.II.B.1.2. Updates to the risk management plan due to manufacturing changes P.II.B Potential impact of a manufacturing change P.II.B Risk analysis P.II.B Update of the risk management plan P.II.B.2. Management and reporting of adverse reactions P.II.B.3. Periodic safety update report (PSUR) P.II.B.3.1. PSUR section Estimated exposure and use patterns P.II.B.3.2. PSUR section Overview of signals: new, ongoing, or closed and Signal and risk evaluation P.II.B.4. Signal management P.II.B.5. Additional monitoring P.II.B.6. Safety communication P.II.C. Operation of the EU network P.II.C.1. Roles and responsibilities P.II.C.1.1. Marketing authorisation holder and applicant in the EU P.II.C Risk management plan (RMP) P.II.C Reporting of adverse reactions and signal management P.II.C Periodic safety update report (PSUR) P.II.C Additional monitoring P.II.C Safety communication P.II.C.1.2. Competent authorities in Member States P.II.C Risk management plan (RMP) P.II.C Reporting of adverse reactions P.II.C Periodic safety update report (PSUR) P.II.C.1.3. European Medicines Agency Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 2/19

46 P.II.C Pharmacovigilance Risk Assessment Committee P.II.C.2. Safety communication about biologicals in the EU Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 3/19

47 P.II.A. Introduction A biological medicinal product (hereon referred to as biological ) is a medicinal product that contains an active substance that is produced by or extracted from a biological source and that needs for its characterisation and the determination of its quality a combination of physio-chemical-biological testing, together with the production process and its control [Directive 2001/83/EC, Annex 1, Part I, Section (b)]. Biologicals encompass a very wide and diverse array of medicines. These include medicinal substances derived from blood and plasma, biotechnology-derived medicines (e.g. using recombinant DNA technology), all types of prophylactic vaccines and advanced therapy medicinal products (ATMPs). This GVP Module does not apply to vaccines and ATMPs as separate specific guidance already exists for these products (see GVP Module P.I. and the Guideline on Safety and Efficacy Follow-up and Risk Management of Advanced Therapy Medicinal Products 1 ). Unless specified otherwise in particular Sections, this Module applies to all biological medicinal products regardless of the regulatory pathway of approval or market exclusivity status, i.e. it applies to reference biological medicinal products (hereafter referred to reference products ), to similar biological medicinal products (hereafter referred to as biosimilars ) and to products which contain the same or closely related active substance but not authorised as biosimilars (e.g. different versions of interferon beta-1a, factor VIII or normal human immunoglobulin) (hereafter referred to as related products ). A biosimilar is a biological medicinal product that contains a version of the active substance of an already authorised reference product in the EEA, and which has shown similarity to the reference product in terms of quality characteristics, biological activity, safety and efficacy based on a comprehensive comparability exercise (see Guideline on Similar Biological Medicinal Products 2 ). The legal requirements for pharmacovigilance and the good pharmacovigilance practices (GVP) apply to biologicals just as they do for other medicines. The guidance of this Module does not replace any of these. However, as outlined below, biologicals are associated with several specific challenges in pharmacovigilance. This Module P.II. is therefore intended to be read and followed alongside the process-related GVP Modules when developing and implementing pharmacovigilance for biologicals to ensure that these challenges are addressed. P.II.A. describes some of the specific issues and challenges, P.II.B. provides guidance on addressing these in the context of the main pharmacovigilance processes described in the GVP and P.II.C. provides guidance related to operation of the EU network. Although separate guidance exists on donor traceability of medicinal substances derived from blood and plasma (see Guideline on Plasma-derived Medicinal Products 3 ), the general principles of pharmacovigilance and patient traceability in this Module also apply to such products. Relevant guidelines to be considered include the Guideline on Immunogenicity Assessment of Biotechnology-derived Therapeutic Proteins, the Guideline on Comparability of Biotechnology-derived Medicinal Products after a Change in the Manufacturing Process, the Guideline on Similar Biological Medicinal Products Containing Biotechnology-derived Proteins as Active Substance: Non-clinical and Clinical Issues, the Guideline on Similar Biological Medicinal Products Containing Biotechnology-derived Proteins as Active Substance: Quality Issues and the Guideline on Process Validation for the Manufacturer of Biotechnology-derived Active Substances and Data to Be Provided in the Regulatory 1 See 2 See 3 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 4/19

48 Submission 4. Other guidelines with pharmacovigilance requirements for specific biosimilars should also be considered. In this Module, all applicable legal requirements are referenced in the way explained in the Good Pharmacovigilance Practices (GVP) Introductory Cover Note 5 and are usually identifiable by the modal verb shall. Guidance for the implementation of legal requirements is provided using the modal verb should. References to the legislation are provided as follows: Directive 2001/83/EC as amended is referenced as DIR, Regulation (EC) No 726/2004 as amended as REG and the Commission Implementing Regulation (EU) No 520/2012 on the Performance of Pharmacovigilance Activities provided for in Regulation (EC) No 726/2004 and Directive 2001/83/EC as IR. As regards the use of the term competent authority in GVP, in particular in Section B, the term is to be understood in its generic meaning of an authority regulating medicinal products and/or a national authority appointed for being in charge of all or individual pharmacovigilance processes. For the purpose of applying GVP in the EU, the term competent authority, used anywhere in GVP, covers the competent authorities in Member States and the European Medicines Agency (hereafter the Agency). The term organisation in GVP covers marketing authorisation holders, competent authorities of Member States and the Agency. P.II.A.1. Pharmacovigilance aspects specific to biologicals Unlike chemically synthesised medicines which can usually be easily characterised and reproduced across different manufacturers, biological active substances are complex molecules produced usually using complex manufacturing processes with many upstream or downstream steps that shape the overall safety, quality and efficacy profile. The manufacturing process (including choice of cell line, raw or starting materials, fermentation and purification process, final formulation) is as much a determinant of the product s quality as the active substance, and minor changes in any manufacturing step can affect the product quality, and subsequently its safety and efficacy. Advances in biotechnology and analytical sciences will continue to allow greater characterisation and control of biologicals, but it is this fundamental complexity that creates the specific challenges for biologicals in pharmacovigilance. P.II.A.1.1. Immunogenicity As with any medicinal product, the safety profile of a biological is determined partly by the direct or indirect pharmacological, including immunogenic, properties of the active substance (e.g. exaggerated immunomodulation or immunosuppression), of the excipients and of process-related impurities (e.g. host cell proteins), or by host or disease-related susceptibility (e.g. medicine-induced allergic reactions, auto-immunity, inflammatory events). For biologicals and non-biologicals, the basic principles of benefit-risk assessment of other GVP Modules apply to potential or identified risks. However, due to their much more complex nature, biologicals pose a greater potential risk of immunogenicity compared to non-biologicals and require specific consideration. This is discussed in detail in the Guideline on Immunogenicity Assessment of Biotechnology-derived Therapeutic Proteins 6. In most cases, immunogenicity to a biological will be without clinical significance, such as a transient appearance of antibodies, and will not impact on the risk-benefit balance of the product. However, on some occasions, immunogenicity could result in serious and life-threatening reactions. 4 See 5 See GVP webpage of the EMA website: 0b01ac058058f32c 6 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 5/19

49 For the purpose of this Chapter, immunogenicity refers to an unwanted immune response that is considered potentially clinically relevant and may require product-specific pharmacovigilance and risk management activities. Sources of immunogenicity for biologicals are multi-factorial and involve one or more product-related factors (e.g. choice of cell line, post-translational changes and alterations to the 3D structure during downstream processing, impurities, choice of product containers), treatment-related factors (e.g. route of administration, dosing frequency) and patient or disease-related factors (e.g. genetic background, concomitant medications, nature of the underlying disease and immune status). The clinical consequences of immunogenicity may include partial or complete loss of efficacy of the product due to induction of neutralising antibodies, altered pharmacokinetics due to antibody binding, general immune effects such as anaphylaxis, formation of immune complexes and potential induction of cross-reactivity with endogenous proteins or other auto-antibodies. Specific evaluation of immunogenicity is required during product development and prior to authorisation of biotechnological medicines (see Guideline on Immunogenicity Assessment of Biotechnology-derived Therapeutic Proteins 7 ). However, non-clinical models and analytical methods/bioassays can usually not predict immunogenicity in humans. Furthermore, the limited sample size of pre-authorisation studies or the rarity of the disease to be treated may not allow rare consequences of immunogenicity to be evaluated prior to authorisation. Uncertainty in relation to immunogenicity should be reflected in the risk management plan (RMP) (see P.II.B.1.) and requires specific activities or surveillance in the post-authorisation phase as appropriate. For biosimilars in particular, initial marketing authorisation is based on demonstrated and accepted biosimilarity of quality, safety and efficacy in accordance with the comprehensive comparability exercise. This exercise is designed to exclude any relevant differences between the biosimilar and the reference product. However, the Guideline on similar biological medicinal products containing biotechnology-derived proteins as active substance: non-clinical and clinical issues 8 notes that Data from pre-authorisation clinical studies are usually insufficient to identify rare adverse effects. Therefore, clinical safety of biosimilars must be monitored closely on an ongoing basis during the postapproval phase including continued benefit-risk assessment. Following on from characterisation of immunogenicity at the time of initial marketing authorisation, the next challenge relevant to any biological relates to changes to manufacturing or quality, and the fact that immunogenicity can potentially be introduced or altered at any time post-authorisation potentially resulting in an altered safety or efficacy profile of a product. P.II.A.1.2. Manufacturing variability Marketing authorisation holders of medicinal products make frequent changes to the manufacturing process of their products post-authorisation. This happens for many reasons including for example changes in source materials, facilities or regulatory requirements. Manufacturing changes may be more complex for biologicals. They need to be supported by a comparability exercise and submitted by the marketing authorisation holder as a variation or as an extension to the marketing authorisation to determine that the pre-and post-change products are comparable, to the extent that quality, safety and efficacy are not adversely affected. In accordance with the Guideline on Comparability of Biotechnology-derived Medicinal Products after a Change in the Manufacturing Process 9, demonstration of comparability is a sequential process, beginning with quality 7 See 8 See 9 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 6/19

50 studies. If a marketing authorisation holder can provide evidence of comparability through physicochemical/analytical and biological assays, then non-clinical or clinical studies with the post-change product are not warranted. In other cases, the process change may require supportive non-clinical and/or clinical data and specific pharmacovigilance requirements. Recital (17) of Regulation (EU) No 1235/2010 states that Risk management plans are normally required for new active substances, biosimilars, medicinal products for paediatric use and for medicinal products for human use involving a significant change in the marketing authorisation, including a new manufacturing process of a biotechnologically-derived medicinal product. The Guideline on Immunogenicity Assessment of Biotechnology-derived Therapeutic Proteins 10 also refers to the need to consider risk management planning if changes in immunogenicity (see P.II.A.1.1.) are possible. Judgements on what constitutes a significant change in the manufacturing process can only be made on a case-by-case basis, based on the comparability exercise. Most manufacturing changes should result in a comparable product, and the need, extent and nature of non-clinical and clinical comparability studies will be determined on a case-by-case basis. However, it will not be possible to predict immunogenicity based on physico-chemical/analytical and biological assays alone, and supportive clinical studies (if requested) will not always be able to detect rare consequences of any altered immunogenicity before approval of a manufacturing change. Biologicals are therefore potentially subject to this dynamic quality profile, with the potential for serious new risks (safety or efficacy) to emerge at any time point in the product life-cycle due to changes in product quality or characteristics (which may also be related to product handling and patient characteristics). These potential changes are relevant not only within a product (e.g. change in quality specifications over time), but also across products with the same international non-proprietary name (INN). In the long-term post-authorisation period, the reference product, biosimilars and related products may potentially exhibit different safety profiles as these products evolve through their life-cycle. Whether or not an updated risk management plan (RMP) (see P.II.B.1.2.) was implemented to support approval of a given manufacturing change, it underlines the importance for biologicals of continuous, life-cycle pharmacovigilance and risk management to rapidly detect any important changes in product safety and efficacy over time. P.II.A.1.3. Stability and cold chain Strict process controls are in place for biologicals to ensure that manufacturing processes and standards remain within the authorised specification. Beyond the point of manufacture and release, overall product stability is maintained by adherence to appropriate storage and handling conditions, cold chain and good distribution practices (see the Guidelines on Good Distribution Practice of Medicinal Products for Human Use 11 ). More so than for non-biologicals, non-adherence to these processes and standards may affect the stability and quality of biologicals, which in turn may introduce or alter immunogenicity (see P.II.A.1.1.) or contamination. Though very rare, particularly for a product that has already been released, such defects and deviations would usually affect specific batches. Life-cycle pharmacovigilance at the levels of products and batches is therefore an important issue for biologicals (see P.II.A.1.4.). 10 See 11 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 7/19

51 P.II.A.1.4. Product traceability As a consequence of manufacturing variability over time in the post-authorisation phase within and across products with similar active substances, a key requirement for pharmacovigilance of biologicals is the need to ensure continuous product and batch traceability in clinical use. This is especially important for biologicals compared to chemically-synthesised medicines due to a greater inherent variability in product characteristics. Whether reference product, biosimilar or related product, it is essential that different products with the same INN can be readily distinguishable in order that newly emerging and product-specific safety concerns and immunogenicity (see P.II.A.1.1.) are rapidly detected and evaluated throughout a product life-cycle, and that supply can be traced to locations and patients if necessary. As any given product usually retains the same product name following a significant change to manufacturing process, batch traceability is an important aspect to be considered in any associated updates to risk management plans (see P.II.B.1.). As product name and batch information is included in the product packaging, this information is available to be recorded and reported at all levels in the supply chain from manufacturer release to prescription, dispensing and patient administration. Biologicals constitute a very diverse array of products for a wide range of therapeutic areas and the clinical settings for prescription, dispensing, supply and administration are equally diverse. Traceability needs therefore to be fully integrated in different healthcare settings and infrastructure that may vary across products and between countries, such as the infrastructure for electronic data recording and record linkage. Most products will be supplied in a hospital setting and, if record linkage does not exist, other methods need to be used to collect exposure information, such as routine bar code scanning at all points in the supply chain. National health authorities should also work towards better integration and automation of prescription information. It should be noted that prescribing practice and product interchangeability, and particularly switching and substitution between biologicals, are beyond the scope of this Module as they fall under the scope of the individual Member States. The product name and batch number of an administered biological should be recorded by the healthcare professional and be provided to the patient. This is particularly important in cases when different versions of the same active substance are available concomitantly on the market and interchangeably used by the same patient. P.II.B. Structures and processes P.II.B.1. Risk management system All marketing authorisation applications submitted in the EU after 2 July 2012 (through the centralised marketing authorisation procedure) or 21 July 2012 (through the mutual recognition marketing authorisation procedure or the decentralised marketing authorisation procedure) should contain a risk management plan (RMP) that must be approved by the competent authorities prior to the granting of the marketing authorisation. The submission of a RMP, or an update thereof, is also normally required for medicinal products for which the initial application was submitted before the above dates if a significant change in the marketing authorisation, including a new manufacturing process of a biotechnology-derived medicinal product [Recital (17) of Regulation (EU) No 1235/2010] (see GVP Module V). As a general principle, any post-authorisation update to the RMP for a reference product should be similarly applied to the relevant biosimilars and related products, and vice-versa, unless justified, e.g. where available information suggests that the clinical concern prompting the update was product- Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 8/19

52 specific (i.e. not related to the active substance or other common excipients). All parts of a RMP are required for a biosimilar, with the exception of RMP part II, module SI Epidemiology of the target population. P.II.B.1.1. Content of the risk management plan (RMP) P.II.B RMP part I Product overview The origin of an active substance of a biological should be included as important information about its composition (see GVP Module V, with biological as a stated example). P.II.B RMP part II Safety specification P.II.B RMP module SVII Identified and potential risks and RMP module SVIII Summary of the safety concerns In accordance with the requirements of GVP Module V, the safety specification should include important identified risks, important potential risks and missing information. The potential for immunogenicity and associated clinical consequences (see P.II.A.1.1.) should be fully evaluated and discussed as part of the initial marketing authorisation application (or variation) in the relevant sections of the Summary of clinical safety of the application for marketing authorisation. Immunogenicity may occur during the life-cycle of a biological, but is not in itself a specific safety concern. It should be included in the safety specification of the RMP only if the conclusion of the discussion warrants its classification as an important risk (identified or potential) or as an area of missing information. In such instances, this concern should be defined as precisely as possible (including any specific potential clinical risks with case definitions), so that specific pharmacovigilance measures to address the uncertainty can be developed (see P.II.B ). The Guideline on Immunogenicity Assessment of Biotechnology-derived Therapeutic Proteins 12 as well as any relevant available product/class-specific guidance on immunogenicity evaluation (e.g. the Guideline on immunogenicity assessment of monoclonal antibodies intended for in vivo clinical use 13 ) should be used in order to determine the most appropriate strategy to further evaluate the potential risk. In case of a significant change to the manufacturing process requiring an amendment of the RMP (see P.II.B.1.2.), potential immunogenicity and clinical consequences should be included in the safety specification. If no specific potential clinical concern has been identified (other than the significant manufacturing change with uncertain clinical consequence), the missing information listed in the updated safety specification should make reference to immunogenicity following a significant change to the manufacturing process. For biosimilars and related products, the summary of safety concerns should, as a minimum, be the same as the reference product unless otherwise justified. Such justification may include the situations where a particular risk associated with the reference product was known to be associated with a component, manufacturing process (other than the active substance) or other factor that is not associated with the biosimilar or related product, or where elements of the safety specification are specific to a particular use (e.g. indication or route of administration) that is absent in some products (but potential for off-label use would need to be considered). Important risks or missing information relating to uncertainties identified from the comparability exercise with regard to seriousness and frequency of adverse reactions for the biosimilar as compared 12 See 13 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 9/19

53 to the reference product should be included in the RMP and the need for additional pharmacovigilance or risk minimisation measures should be assessed. Any other proposed differences in the safety specification of a biosimilar compared to the reference product should be duly justified based on the outcome of the comprehensive comparability exercise. P.II.B RMP module SVI Additional EU requirements for the safety specification For all biologicals, the potential for infections caused by residuals of biological material used in the manufacturing process as well as contaminations introduced by the manufacturing process should be presented in relation to the potential for transmission of infectious agents. P.II.B RMP part III Pharmacovigilance plan The need and plans for continuous life-cycle signal detection and pharmacovigilance specific to the product including batch-specific issues, particularly following a significant change to the manufacturing process, should be discussed. In this context, the pharmacovigilance plan should include a discussion around clinical settings of product use and how this may impact on routine product name and batch recording and reporting (e.g. whether used in primary or tertiary care) and what additional activities or risk minimisation measures may be required to support product traceability (e.g. provision of sticky labels, bar coding). P.II.B RMP part III section Routine pharmacovigilance activities In this section, the marketing authorisation applicant or holder should discuss: the clinical settings of product use and how this may impact on product name and batch recording and reporting; measures that will be introduced to routinely follow-up on case reports to obtain information on product name and batch number(s) (see also GVP Module VI Appendix 1); signal detection activities performed to identify batch-specific safety issues; any adverse events of special interests (AESIs), with definitions, identified as important potential risks for which specific safety surveillance will be put in place (see also GVP Considerations P.I.). P.II.B RMP part III section Additional pharmacovigilance activities In this section, the marketing authorisation applicant or holder should discuss: any additional measures introduced in collaboration with the national competent authorities to support traceability of the product (e.g. provision of sticky labels, bar coding); activities performed to measure background rates for AESIs, preferably by indication, in the age group targeted by the product; activities performed to continuously monitor suspected adverse reaction reporting frequencies or rates for AESIs based on available data on exposure and comparing such rates to relevant defined background rates (using methods such as observed vs expected analyses) (see also GVP Module P.I.); use of existing patient registries or other data sources (or establishment of a new registry if existing data sources are inadequate) (see GVP Module VIII Appendix 1); Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 10/19

54 for a biosimilar, any specific safety monitoring imposed to the reference product or product class and its relevance for the concerned product. For significant changes to the manufacturing process that require an RMP update (see P.II.B.1.2.), given that the product name usually does not change, there should be a particular emphasis on batchspecific pharmacovigilance for an agreed time period at the time of submission of manufacturing change variation. This period of surveillance should start after approval of the variation once new batches are on the market. Immunogenicity If immunogenicity is included in the safety specification (see P.II.B ), relevant strategies for the evaluation of immunogenicity and associated clinical consequences in the post-authorisation setting should be proposed as an additional pharmacovigilance activity. Where applicable, the principles for immunogenicity evaluation should follow the Guideline on Immunogenicity Assessment of Biotechnology-derived Therapeutic Proteins 14 as well as any relevant available product or class-specific guidance on immunogenicity evaluation (e.g. the Guideline on immunogenicity assessment of monoclonal antibodies intended for in vivo clinical use 15 ). Depending on the nature of any potential immunogenicity and the data that generated the concern, or the nature of the missing information, the additional pharmacovigilance activities should have clearlydefined objectives. The plan may include bio-analytical methods (e.g. in vitro assays, serology studies), non-clinical studies, interventional clinical studies or observational epidemiological approaches. Any analytical and clinical endpoints relevant to the potential risk, including those related to safety and efficacy (e.g. in order to evaluate potential effects of neutralising antibodies), should be clearly defined to support their characterisation in passive surveillance (e.g. via targeted follow up), additional pharmacovigilance activities or epidemiological studies. For these reasons, determination of the optimal strategy for evaluation of immunogenicity in the RMP should be a multidisciplinary approach, with input from experts in the quality, non-clinical, clinical, pharmacovigilance and epidemiological fields. If a new clinical risk is identified that may have an immunogenic aetiology, it should be fully explored in any subsequent risk evaluation. Whether the risk is specific to a specific product or batch, the potential root cause should be assessed in order to evaluate the ability for risk minimisation or elimination (e.g. improved assays, manufacturing steps). Post-authorisation safety studies The most optimal study design should be used considering the objective of the post-authorisation safety study (PASS) (see GVP Module VIII Appendix 1). If an existing registry is to be used or a new registry is to be established, a comparator or non-exposed group should preferably be included. Joint disease registries should be encouraged. Biosimilars and related products Any specific safety monitoring imposed on the reference product or product class should be adequately addressed in the pharmacovigilance plan, unless otherwise justified (e.g. if the safety concern was specific to the reference product and not included in the safety specification of the biosimilar or related product). Where applicable and feasible, competent authorities should encourage marketing authorisation holders of biosimilars and related products to participate in any pharmacoepidemiological studies already in place for the reference product, unless otherwise justified (see P.II.B ). 14 See 15 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 11/19

55 P.II.B RMP part V Risk minimisation measures Evaluation of any new clinical risk associated with a biological product should include a root cause analysis in order to evaluate the ability for risk minimisation or elimination via analytical studies or bioassays (e.g. improved assays, manufacturing steps). As a general principle in order to improve traceability of biological medicines, all summaries of product characteristics (SmPCs) for biologicals (also with relevant appropriate wording in the package leaflets (PLs)) should include a prominent statement that the name and batch number of the administered product should be clearly recorded in the patient file. Related wording should also be included in relevant educational material, direct healthcare professional communication (see P.II.B.6.) and product promotional material as applicable. Use of other tools such as sticky/tear-off labels in the product packaging should also be considered to facilitate accurate recording in patient files and provision of information to patients. Use of available bar code-scanning technology and infrastructure should also be encouraged where appropriate. Risk minimisation activities in place for the reference product should, in principle, be included in the RMP of the biosimilars and related products, and vice-versa. Any deviation from this (e.g. when the risk minimisation is linked specifically to the reference product) should be justified. P.II.B.1.2. Updates to the risk management plan due to manufacturing changes P.II.B Potential impact of a manufacturing change If the comparability evaluation identifies a potential impact of the manufacturing change in terms of clinical relevance, the change requires submission of an updated RMP, unless otherwise justified. This justification would need to be made on a case-by-case basis. It is not possible to give specific guidance on what may constitute a clinically relevant impact of a manufacturing change in every situation, and judgements have to be made based on the findings of the comparability exercise or other quality or clinical evaluation that supports the variation to the process, as well as any other relevant precedents or experience. Even minor changes to a manufacturing process can potentially have unpredicted significant clinical effects. In cases when the comparability exercise or evaluation has not necessarily identified a potential impact of clinical relevance, submission of an updated RMP with the variation to the manufacturing process may still be appropriate based on a risk analysis or previous experience. P.II.B Risk analysis To support this process and ensure that Recital (17) of Regulation (EU) No 1235/2010 is adhered to, all applications for a variation to the manufacturing process of a biological should routinely include a RMP update if the marketing authorisation holder has already decided that it is required, or a risk analysis on the potential significance and the need, or not, for an update to the RMP. This process is in line with the concepts envisaged in ICH-Q5E (Comparability of Biotechnological/Biological Products Subject to Changes in their Manufacturing Process) and ICH-Q10 (Pharmaceutical quality system) 16. The risk analysis from the marketing authorisation holder may be a short statement with appropriate justifications or a more complex evidence-based analysis if required by the nature of the change (particularly if there is precedent for the type of change resulting in a clinically significant impact). 16 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 12/19

56 P.II.B Update of the risk management plan If the marketing authorisation holder considers that an update of the RMP is required, it should be provided with the application warranting such update. Otherwise, if the competent authority concludes on the need for an RMP update, it should provide the marketing authorisation holder with recommendations on the nature of the changes expected in the RMP. A RMP update should be submitted as soon as possible to allow for its approval in the context of the variation to the manufacturing change. Updates to the RMP should address the safety specification, pharmacovigilance plan and risk minimisation measures. For cases when the comparability evaluation identifies a potential impact of the manufacturing change in terms of clinical relevance, particular attention should be paid to describe as a routine pharmacovigilance activity how batch-specific evaluation can be done in order that the pre- and post-change products can be easily distinguished during a relevant time period after the manufacturing change. Following an update to the RMP, subsequent periodic safety update reports (PSURs) (see P.II.B.3.) should specifically evaluate reports and any other information that might indicate a new clinical risk related to a process change. This evaluation should relate to the specific concern included in any updated safety specification of the RMP based on the manufacturing change. The cycle of submission of the PSURs may also be amended (and re-instated) accordingly in line with the updated RMP. P.II.B.2. Management and reporting of adverse reactions The requirements for the management and reporting of suspected adverse reactions outlined in GVP Module VI apply equally to biologicals and non-biologicals. In addition, through the methods for collecting information and where necessary through the follow-up of suspected adverse reaction reports, competent authorities shall ensure that all appropriate measures are taken to identify clearly any biological prescribed, dispensed or sold in their territory which is the subject of a suspected adverse reaction report, with due regard to the name of the medicinal product (see GVP Annex I) and the batch number [DIR Art 102(e)]. When reporting suspected adverse reactions, competent authorities and marketing authorisation holders shall provide all available information on each individual case (see GVP Module VI), including the product name and batch number(s) [IR Art 28(3)h)]. For this purpose, Member States and marketing authorisation holders should encourage health care professionals to provide patients and carers with information on the product name and batch number(s) of any biological administered, regardless of the point of prescription, supply or administration and technical infrastructure that may exist. Competent authorities and marketing authorisation holders should also encourage reporters to record information on product names and batch numbers. A follow-up procedure should be put in place to obtain the batch number where it is not indicated in the initial report. The business process map included in GVP Module VI Appendix 1 should be followed. If the RMP of a biological specifies certain activities to be performed to collect information on defined clinical endpoints (e.g. immunogenicity endpoints), specific laboratory/assay data, case definitions and questionnaires may be developed and referred to in the RMP for the follow-up of targeted adverse reactions, in addition to the capture of product name and batch information. Where marketing authorisation holders and competent authorities consider utilising their website to facilitate the collection of reports of suspected adverse reactions by providing reporting forms or appropriate contact details for direct communication (see GVP Module VI), any such activities should be used to communicate, promote and facilitate the capture of product names and batch information in reports of adverse reactions. Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 13/19

57 P.II.B.3. Periodic safety update report (PSUR) The requirements for signal evaluation as part of the PSUR in GVP Module VII apply equally to biologicals and non-biologicals (see P.II.C.1.2. for the assessment of PSURs for biosimilars). P.II.B.3.1. PSUR section Estimated exposure and use patterns To support the processes for signal management (see P.II.B.4.), marketing authorisation holders should make every effort to obtain data on actual usage of the product (i.e. rather than relying exclusively on aggregated sales data). Real-world data sources are important to estimate overall exposure and patterns of use. P.II.B.3.2. PSUR section Overview of signals: new, ongoing, or closed and Signal and risk evaluation The guidance in P.II.B.4. should be applied to the signal evaluation process within PSURs, i.e. case-bycase judgements are required on whether or not the signal applies to a single product or to all products with the same active substance. In reference to P.II.B.1.2., and in accordance with the Guideline on Comparability of Biotechnology-derived Medicinal Products after a Change in the Manufacturing Process 17, following a significant change to the manufacturing process (which will normally require submission of an updated RMP), PSURs should specifically evaluate reports and any other information that might indicate a new clinical risk related to a process change. The required data on batch-specific exposure patterns will support such evaluation. This should be presented in the context of the specific concern that is included in any updated safety specification of the RMP on account of the manufacturing change. Following a significant change to the manufacturing process, the cycle of submission of the PSURs may also be amended (and re-instated) accordingly in line with the updated RMP (providing that the merits of this outweigh the requirement for a harmonised cycle across biosimilars and related products). P.II.B.4. Signal management The requirements for signal management in GVP Module IX apply equally to biologicals and nonbiologicals. As with all medicinal products, biologicals require continuous pharmacovigilance in order to detect and evaluate potential new clinical risks (safety or efficacy) that may emerge during a product life-cycle. However, this is especially important for biologicals for the reasons described in P.II.A.1., particularly due to the inherent variability in manufacturing process that may potentially alter the immunogenicity of a product and induce clinical consequences. Signal detection for biologicals should therefore be specific to the product, as well as the active substance. All steps of signal management should be performed at the level of the product name, as well as the active substance. In case of a signal any effort should be made to identify any common root cause such as batch. Processes should be particularly sensitive to detect any acute and serious new risks that may emerge following a change in the manufacturing process or quality of a biological and important differences between batches of the same product (this is particularly important following a significant change to the manufacturing process given that the product name usually does not change). Important differences between reference products and biosimilars or related products should be identified during the product(s) life-cycle based on the available information. Any clinical consequences of potential emerging immunogenicity (as a theoretical risk) should be monitored throughout the product life-cycle. 17 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 14/19

58 Post-authorisation exposure information is needed for signal management for biologicals, but biologicals are often prescribed or dispensed in the hospital setting and the required exposure information may not be available in population-based databases. Marketing authorisation holders should make every effort to obtain data on actual usage specific to a product (see P.II.B.3.) and explore all methods and data sources to obtain reliable and updated information. Denominator data and data of suspected adverse reactions (see GVP Module IX) should be analysed to support continuous signal detection and particularly detection of any apparent changes in suspected adverse reaction reporting rates or trends that could indicate new signals (particularly following manufacturing changes). Some active substances or medicinal products may also be subject to an increased frequency of data monitoring and a significant change in the manufacturing process of a biological may, on a case-by-case basis, justify specific signal detection activities (see GVP Module IX). Any such requirements should be specified in the risk management plan (see P.II.B and P.II.B.1.2.). Continuous disproportionality analysis and observed vs expected methods (see GVP Module P.I., the GVP Module IX Addendum I and the ENCePP Guide on Methodological Standards in Pharmacoepidemiology 18 ) should also be consulted as needed. Any signal should be evaluated in the context of batch-specific exposure data, including numbers/codes of delivered or sold batches, their size and the regions or countries where the respective batches have been delivered. Implementation of strengthened processes for routine pharmacovigilance will facilitate earlier detection of new risks and changes in product safety or quality over time. For new signals, case-by-case judgements are required on whether or not the signal may apply to the concerned product or to all products with the same active substance. However, on a precautionary basis, inadequate evidence on the specificity of a signal detected for a biosimilar or related product may justify application of a regulatory action to the reference product, and vice versa. Any new clinical risk suspected to have an immunogenic aetiology should be fully investigated to determine whether the risk is specific to a product name or batch and evaluate its potential root cause in order to determine the potential for risk minimisation or elimination (e.g. improved assays, manufacturing steps). P.II.B.5. Additional monitoring Biologicals authorised after 1 January 2011 shall be included in the list of medicinal products that are subject to additional monitoring [REG Art 23(1)(b)]. They shall be removed from the list under the mandatory scope five years after the Union reference date unless the period of additional monitoring is extended [REG Art 23(3)]. P.II.B.6. Safety communication This guidance addresses specific aspects of communications for biologicals due to their complex manufacturing processes and compositions as well as to the complex effects they have on the human body including possible adverse reactions caused by immunogenicity (see P.II.A.1.). It does not address general principles and methods for safety communication. They are described in GVP Module XV and also apply to biologicals. There should be awareness of specific concerns that patients and healthcare professionals frequently have in relation to biologicals, so that they can be addressed. Communicating about risks of biologicals poses challenges for presenting scientifically, technically and medically complex issues in a language understandable to patients and the general public, and also to healthcare professionals of various 18 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 15/19

59 specialities. Some technical terms and concepts require careful explanation in order to ensure their proper understanding and avoid social risk amplification 19 due to e.g. biotechnological methods, mainly recombinant DNA technology, which are not commonly known by non-specialists and which may be perceived by some individuals or populations as not natural and negatively interfering with nature, the human body or genes. Hence, information on the manufacturing process and its variability, the active substance and its mode of action as well as the excipients and possible residues should be communicated to patients and health care professionals for their good understanding. Immunogenicity is a specific source of concerns for biologicals, resulting in information needs to be fulfilled consistently. Issues around previous exposure to the same or cross-immunogenic products may also have to be addressed in communication documents. For biosimilars, consultations with patients and healthcare professionals have shown information needs relating to quality, safety, efficacy, extrapolation, comparability and interchangeability. The EMA Questions and Answers on Biosimilar Medicines (similar biological medicinal products) 20, drawn up in consultation with patient and healthcare professional representatives, and the European Commission s Consensus Information Document What you need to know about biosimilar medicinal products 21 may be used as a source for explanations when drafting product-specific communication documents. Building confidence of users in biologicals requires not only communication on product-specific aspects, but also on the mechanisms in place for safety surveillance. The relevant risk management plan summary (see GVP Module V) may be referred to in communications. If applicable, comparability data may be provided. Encouraging reporting of suspected adverse reactions requires some specific information for biologicals. It should be communicated to patients and healthcare professionals that adverse reactions may arise even if a medicinal product has previously been well tolerated, e.g. due to a manufacturing variability or changes or long-term or delayed onset effects, and that reporting of suspected adverse reactions occurring even after long-term use or with unknown features is important. With a view to adverse reaction reporting and effective risk management, traceability is a major objective in managing the appropriate use and pharmacovigilance of biologicals (see P.II.A.1.4.) and hence constitutes a specific communication objective for biologicals vis-à-vis patients and healthcare professionals. Communication should therefore emphasize the importance of providing the product name (or INN and name of the marketing authorisation holder) and batch number(s) when reporting suspected adverse reactions. Other specific safety communication objectives in relation to biologicals may aim at avoiding errors in storage and handling, in particular as regards cold chain requirements (see P.II.A.1.3.) and administration which frequently requires specific medical devices. In order to ensure proper understanding, consultation of draft communication documents with patients and healthcare professionals should be undertaken (see GVP Modules XI and XV). 19 The concept of social risk amplification describes changes in risk perceptions at various stages of dissemination of information, e.g. through scientific debates or discussion in the general media. 20 See 21 See Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 16/19

60 P.II.C. Operation of the EU network P.II.C.1. Roles and responsibilities P.II.C.1.1. Marketing authorisation holder and applicant in the EU Medicinal products developed by means of one of the biotechnology processes listed in the Annex of Regulation (EC) No 726/2004, or fulfilling any other criteria of the Annex, shall be authorised in the EU through the centralised authorisation procedure. P.II.C Risk management plan (RMP) The marketing authorisation applicant is responsible for the submission of the RMP in line with the format and content presented in GVP Module V and P.II.B.1.1. In case of significant changes to the manufacturing process, a risk analysis and updated RMP should be submitted (see P.II.B.1.2.). P.II.C Reporting of adverse reactions and signal management When reporting suspected adverse reactions, marketing authorisation holders shall provide all available information on each individual case, including, for biologicals, the name and batch number(s) of the administered product [IR Art 28(3)(h)]. Signal management should be performed as described in GVP Module IX. Signal detection processes for biologicals should be particularly sensitive to detect any acute and serious new risks that may emerge following a change in the manufacturing process or quality and important differences between batches of the same product. Any important differences between reference products and biosimilars or related products should be identified during the product(s) life-cycle based on the available information. Any clinical consequences of potential emerging immunogenicity (as a theoretical risk) should be monitored throughout the product life-cycle. P.II.C Periodic safety update report (PSUR) Where relevant to the interpretation of safety data, including a new safety signal that has been detected in the interval covered by the PSUR, marketing authorisation holders should include in the PSUR a summary of relevant information on the batches delivered during the PSUR reporting period, including batch numbers, countries (EU Member States) and regions where such batches have been delivered, size of the batches and any available information on the number of batches that were delivered per country. All assumptions used for calculations should be provided. P.II.C Additional monitoring For biologicals included in the list of medicinal products subject to additional monitoring according to the mandatory or optional scope [REG Art 23 (1) and (1a)], it is the responsibility of the marketing authorisation holder to perform the activities described in GVP Module X. P.II.C Safety communication Further to the guidance in P.II.B.6., safety communication is an important activity to be considered by the marketing authorisation holder throughout the life-cycle of biologicals, and P.II.C.2. should be followed for additional EU-specific guidance. Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 17/19

61 P.II.C.1.2. Competent authorities in Member States P.II.C Risk management plan (RMP) When assessing RMPs and their updates for biosimilars, the safety specification, pharmacovigilance plan and risk minimisation plan introduced in the RMP for the reference product should be taken into consideration (see P.II.B.1.1.). The risk analysis submitted by the marketing authorisation holder of a biological medicinal product in the case of a change in the manufacturing process should be assessed and a conclusion should address the need to update the RMP based on this assessment (see P.II.B.1.2.). P.II.C Reporting of adverse reactions Member States shall ensure, through the methods for collecting information and where necessary through the follow-up of suspected adverse reaction reports, that all appropriate measures are taken to identify clearly any biological prescribed, dispensed or sold in their territory which is the subject of a suspected adverse reaction report, with due regard to the name of the medicinal product (as defined in DIR Art 1(20)], and the batch number [DIR Art 102(e)]. To fulfil this obligation, national competent authorities should agree with marketing authorisation holders, where applicable, a system to ensure the traceability of the biologicals that are prescribed, dispensed or sold, inform healthcare professionals and patients of the need to provide the product name (i.e. brand/invented name or, as appropriate, INN accompanied by the name of the marketing authorisation holder) and batch number/code when reporting a suspected adverse reaction and make this information available to assessors for signal detection and evaluation of individual case reports. Member States shall facilitate in their territory the reporting of suspected adverse reactions by means of alternative reporting systems, accessible to healthcare professionals and consumers, in addition to web-based formats [DIR Art 102]. If electronic and web-based reporting forms and data capture tools are developed, consideration should be given to optimise the ability of these to encourage provision of product and batch information. This may include automatic prompts if the product name or batch is not provided or drop-down list of available products when a particular active substance is selected. P.II.C Periodic safety update report (PSUR) For the assessment of PSURs for biosimilars, it is critical that the data can be assessed in parallel to the safety data collected for the reference product. For the assessment of PSURs for biologicals subject to different marketing authorisations, authorised in more than one Member State, containing the same active substance or the same combination of active substances whether or not held by the same marketing authorisation holder, the PSUR EU single assessment procedure should be followed further to harmonisation of the frequency and dates of submission of PSURs in the list of EU reference dates (EURD list) [DIR Art 107e-g]. This assessment should be performed by a Member State appointed by the CMDh when none of the marketing authorisations concerned has been granted in accordance with the centralised procedure (see GVP Module VII). P.II.C.1.3. European Medicines Agency As for all medicinal products, the European Medicines Agency has the responsibility for coordinating the existing scientific resources for the pharmacovigilance of biologicals such as the coordination of: the assessment of the risk analysis submitted by the marketing authorisation holder of a biological in the case of a change in the manufacturing process and, based on this assessment, provision on a recommendation on the need to update the RMP (see P.II.B.1.5.); Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 18/19

62 the PSUR EU single assessment procedure for biologicals containing the same active substance or the same combination of active substances where at least one of the marketing authorisations concerned has been granted in accordance with the centralised procedure (see GVP Module VII). For signal detection of biologicals, the Agency should provide rapporteurs, lead Member States and national competent authorities with electronic reaction monitoring reports and other data outputs and statistical reports at the product level rather than at the substance level and provide marketing authorisation holders with appropriate support for the monitoring of the EudraVigilance database at the product level. The Agency shall maintain and publish the list of biologicals subject to additional monitoring under the mandatory or optional scope [REG Art 23]. P.II.C Pharmacovigilance Risk Assessment Committee The Pharmacovigilance Risk Assessment Committee (PRAC) shall: recommend, upon a request from the European Commission or a competent authority of a Member State, as appropriate, whether a biological medicinal product which is subject to the conditions set out in Article 23(1a) of Regulation (EC) No 726/2004 should be included in the additional monitoring list; appoint a rapporteur for the PSUR EU single assessment procedure for biological medicinal products containing the same active substance where at least one of the marketing authorisations concerned has been granted in accordance with the centralised procedure [DIR Art 107e to 107g] (see GVP Module VII); adopt a recommendation on the PSUR EU single assessment procedure for biological medicinal products as identified in the EURD list [DIR Art 107e]; provide advice on the RMP [REG Art 61a(6)]; for RMPs for biosimilars, the PRAC should ensure as appropriate that the pharmacovigilance plan and risk minimisation plan include similar activities as for the reference product. P.II.C.2. Safety communication about biologicals in the EU Further to the guidance in P.II.B.6., the following should be considered for safety communications about biologicals in the EU. Operational details of communication processes may differ according to different scenarios among Member States regarding the use of biologicals, in particular regarding interchangeability and interchange practices of biosimilars. These differences should be accounted for during the EU-wide coordination of safety communication, while maintaining overall consistency of scientific benefit-risk messages across the EU Member States. Competent authorities in Member States should publish in the local language explanations of biological-related terms and concepts and other information for patients, in particular comparability assessments, and should support healthcare professionals with communication material. This should facilitate timely communication with patients with a view to ensuring informed therapeutic choice (including possible change of treatment), adequate risk minimisation and reporting of suspected adverse reactions. Communication in the EU should be underpinned by transparency on how regulatory decisions were reached and on the roles and responsibilities of each stakeholder in the EU (see P.II.C.1.). Guideline on good pharmacovigilance practices (GVP) P. II EMA/168402/2014 Page 19/19

63 Considerations in Demonstrating Interchangeability With a Reference Product Guidance for Industry DRAFT GUIDANCE This guidance document is being distributed for comment purposes only. Comments and suggestions regarding this draft document should be submitted within 60 days of publication in the Federal Register of the notice announcing the availability of the draft guidance. Submit electronic comments to Submit written comments to the Division of Dockets Management (HFA-305), Food and Drug Administration, 5630 Fishers Lane, rm. 1061, Rockville, MD All comments should be identified with the docket number listed in the notice of availability that publishes in the Federal Register. For questions regarding this draft document, contact (CDER) Ebla Ali-Ibrahim, , or (CBER) Office of Communication, Outreach and Development, or U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER) Center for Biologics Evaluation and Research (CBER) January 2017 Biosimilars 11192dft 01/12/17

64 Considerations in Demonstrating Interchangeability With a Reference Product Guidance for Industry Additional copies are available from: Office of Communications, Division of Drug Information Center for Drug Evaluation and Research Food and Drug Administration New Hampshire Ave., Hillandale Bldg., 4 th Floor Silver Spring, MD Phone: or ; Fax: druginfo@fda.hhs.gov and/or Office of Communication, Outreach and Development Center for Biologics Evaluation and Research Food and Drug Administration New Hampshire Ave., Bldg. 71, Room 3128 Silver Spring, MD Phone: or ocod@fda.hhs.gov U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER) Center for Biologics Evaluation and Research (CBER) January 2017 Biosimilars

65 Contains Nonbinding Recommendations Draft Not for Implementation TABLE OF CONTENTS I. INTRODUCTION... 1 II. BACKGROUND... 2 III. GENERAL PRINCIPLES... 2 IV. SCOPE... 4 V. FACTORS IMPACTING THE TYPE AND AMOUNT OF DATA AND INFORMATION NEEDED TO SUPPORT A DEMONSTRATION OF INTERCHANGEABILITY... 5 A. Product-Dependent Factors That May Impact the Data Needed to Support a Demonstration of Interchangeability Product Complexity and the Extent of Comparative and Functional Characterization Product-Specific Immunogenicity Risk Totality of Factors to Consider in Assessing the Data and Information Needed to Support a Demonstration of Interchangeability... 7 B. Biosimilar Product Postmarketing Data That May Impact the Data Needed to Support a Demonstration of Interchangeability... 8 VI. DATA AND INFORMATION NEEDED TO SUPPORT A DEMONSTRATION OF INTERCHANGEABILITY... 9 A. Considerations for the Design and Analysis of a Switching Study or Studies Needed to Support a Demonstration of Interchangeability Study Endpoints Study Design and Analysis Study Population Condition of Use To Be Studied Route of Administration B. Extrapolation of Data VII. USE OF A U.S.-LICENSED REFERENCE PRODUCT IN A SWITCHING STUDY OR STUDIES VIII. CONSIDERATIONS FOR DEVELOPING PRESENTATIONS FOR PROPOSED INTERCHANGEABLE PRODUCTS A. General Considerations B. Analysis of Proposed Presentations of Proposed Interchangeable Products Threshold Analyses Studies to Evaluate Differences That May Not Be Minor as Observed in Threshold Analyses IX. POSTMARKETING SAFETY MONITORING CONSIDERATIONS APPENDIX A: COMPARATIVE USE HUMAN FACTORS STUDIES Study Design Considerations Sample Size Considerations... 26

66 Contains Nonbinding Recommendations Draft Not for Implementation Considerations in Demonstrating Interchangeability With a Reference Product Guidance for Industry 1 This draft guidance, when finalized, will represent the current thinking of the Food and Drug Administration (FDA or Agency) on this topic. It does not establish any rights for any person and is not binding on FDA or the public. You can use an alternative approach if it satisfies the requirements of the applicable statutes and regulations. To discuss an alternative approach, contact the FDA staff responsible for this guidance as listed on the title page. I. INTRODUCTION This guidance is intended to assist sponsors in demonstrating that a proposed therapeutic protein product is interchangeable with a reference product for the purposes of submitting a marketing application or supplement under section 351(k) of the Public Health Service Act (PHS Act) (42 U.S.C. 262(k)). The Biologics Price Competition and Innovation Act of 2009 (BPCI Act) amends the PHS Act and other statutes to create an abbreviated licensure pathway in section 351(k) of the PHS Act for biological products shown to be biosimilar 2 to or interchangeable with an FDA-licensed biological reference product 3 (see sections 7001 through 7003 of the Patient Protection and Affordable Care Act (Affordable Care Act) (Public Law )). Although the 351(k) pathway applies generally to biological products, this guidance focuses on therapeutic protein products and gives an overview of important scientific considerations in demonstrating interchangeability of a proposed therapeutic protein product (proposed interchangeable product or proposed product) with a reference product. This guidance is one in a series of guidances that FDA is developing to implement the BPCI Act and includes references to information from other FDA guidances, where appropriate. In general, FDA s guidance documents do not establish legally enforceable responsibilities. Instead, guidances describe the Agency s current thinking on a topic and should be viewed only as recommendations, unless specific regulatory or statutory requirements are cited. The use of 1 This guidance has been prepared by the Office of Medical Policy in the Center for Drug Evaluation and Research (CDER) in cooperation with the Center for Biologics Evaluation and Research (CBER) at the Food and Drug Administration. 2 Section 351(i)(2) of the PHS Act defines biosimilar or biosimilarity to mean that the biological product is highly similar to the reference product notwithstanding minor differences in clinically inactive components (highly similar provision) and that there are no clinically meaningful differences between the biological product and the reference product in terms of the safety, purity, and potency of the product (no clinically meaningful differences provision). 3 Section 351(i)(4) defines reference product to mean the single biological product licensed under subsection (a) against which a biological product is evaluated in an application submitted under subsection (k). 1

67 Contains Nonbinding Recommendations Draft Not for Implementation the word should in Agency guidances means that something is suggested or recommended, but not required. II. BACKGROUND Section 351(k) of the PHS Act, as amended by the BPCI Act, sets forth the requirements for an application for a proposed biosimilar product and an application or a supplement for a proposed interchangeable product. Section 351(k)(4) of the PHS Act further provides that upon review of an application submitted under section 351(k) or any supplement to such an application, FDA will determine the biological product to be interchangeable with the reference product if FDA determines that the information submitted in the application or the supplement is sufficient to show that the biological product is biosimilar to the reference product and can be expected to produce the same clinical result as the reference product in any given patient 4 and that for a biological product that is administered more than once to an individual, the risk in terms of safety or diminished efficacy of alternating or switching between use of the biological product and the reference product is not greater than the risk of using the reference product without such alternation or switch. 5 Section 351(i) of the PHS Act states that the term interchangeable or interchangeability, in reference to a biological product that is shown to meet the standards described in section 351(k)(4) of the PHS Act, means that the biological product may be substituted for the reference product without the intervention of the health care provider who prescribed the reference product. 6 III. GENERAL PRINCIPLES FDA intends to consider the totality of the evidence provided by a sponsor when the Agency evaluates the sponsor s demonstration of interchangeability according to the criteria set forth in section 351(k). To support a demonstration of interchangeability, section 351(k)(4)(A) of the PHS Act provides, among other things, that a sponsor must show that the proposed product is biosimilar to the reference product. Where a product is first licensed as a biosimilar, that licensure may be referenced to support a showing for this statutory criterion for demonstrating interchangeability. In addition, section 351(k)(4)(A) of the PHS Act provides that an application for an interchangeable product must include information sufficient to show that the proposed interchangeable product can be expected to produce the same clinical result as the reference 4 Section 351(k)(4)(A) of the PHS Act. 5 Section 351(k)(4)(B) of the PHS Act. 6 The terms interchangeable or interchangeability in this guidance have the same meaning as defined in section 351(i)(3) of the PHS Act. 2

68 Contains Nonbinding Recommendations Draft Not for Implementation product in any given patient. FDA expects that sponsors will submit data and information to support a showing that the proposed interchangeable product can be expected to produce the same clinical result as the reference product in all of the reference product s licensed conditions of use. The data and information to support a showing that the proposed interchangeable product can be expected to produce the same clinical result as the reference product in all of the reference product s licensed conditions of use may vary depending on the nature of the proposed interchangeable product and may include, but need not be limited to, an evaluation of data and information generated to support a demonstration of a biological product s biosimilarity, such as: The identification and analysis of the critical quality attributes The identification of analytical differences between the reference product and the proposed interchangeable product, and, in addition, an analysis of the potential clinical impact of the differences An analysis of mechanism(s) of action in each condition of use for which the reference product is licensed, which may include the following: - The target receptor(s) for each relevant activity/function of the product - The binding, dose/concentration response, and pattern of molecular signaling upon engagement of target receptor(s) - The relationship between product structure and target/receptor interactions - The location and expression of target receptor(s) The pharmacokinetics and biodistribution of the product in different patient populations The immunogenicity risk of the product in different patient populations Differences in expected toxicities in each condition of use and patient population (including whether the expected toxicities are related to the pharmacological activity of the product or to off-target activities) Any other factor that may affect the safety or efficacy of the product in each condition of use and patient population for which the reference product is licensed Where applicable, the data and information should include a scientific justification as to why any differences that exist between the reference product and the proposed interchangeable product, with respect to the factors described, do not preclude a showing that the proposed interchangeable product can be expected to produce the same clinical result as the reference product in any given patient. As previously noted, the data and information may vary depending on the nature of the proposed interchangeable product, and not all factors will necessarily be relevant to a given scientific justification. The data and information may also include a scientific rationale to extrapolate data and information supporting a demonstration of interchangeability in an appropriate condition of use to the remaining conditions of use for which the reference product is licensed. Extrapolation of data is further described in section VI.B of this guidance. Generally, the data and information to support a showing under the can be expected to produce 3

69 Contains Nonbinding Recommendations Draft Not for Implementation the same clinical result as the reference product in any given patient standard will likely not involve additional clinical studies other than those necessary to support other elements of demonstrating interchangeability. We note that although a sponsor may seek licensure for a proposed interchangeable product for fewer than all conditions of use for which the reference product is licensed, we recommend that a sponsor seek licensure for all of the reference product s licensed conditions of use when possible. In addition, section 351(k)(4)(B) of the PHS Act provides that another of the criteria for FDA to make a determination of interchangeability is a finding that information in the application is sufficient to show that for a biological product that is administered more than once to an individual, the risk in terms of safety or diminished efficacy of alternating or switching between use of the biological product and the reference product is not greater than the risk of using the reference product without such alternation or switch. FDA expects that applications generally will include data from a switching study or studies 7 in one or more appropriate conditions of use. FDA anticipates that data and information acquired from a switching study or studies will be useful in assessing the risk, in terms of safety and diminished efficacy, of alternating or switching between the products. Considerations for the design of a switching study, including study endpoints, study design and analysis, study population, condition(s) of use, and routes of administration to be studied, are discussed in detail in section VI.A of this guidance. An interchangeable product may be substituted for the reference product without the intervention of the health care provider who prescribed the reference product. 8 Sponsors of proposed interchangeable products should evaluate the proposed product s presentation, including product design and user interface, relative to the reference product. Considerations for developing presentations, container closure systems, and delivery device constituent parts for proposed interchangeable products are discussed in detail in section VIII of this guidance. IV. SCOPE This guidance provides an overview of important scientific considerations in demonstrating interchangeability with a reference product, including the following: Data and information needed to support a demonstration of interchangeability Considerations for the design and analysis of a switching study or studies to support a demonstration of interchangeability Recommendations regarding the use of a U.S.-licensed reference product in a switching study or studies 7 The term switching study or studies as used throughout this guidance refers to a clinical study or studies used to determine the impact of alternating or switching between the proposed interchangeable product and the reference product. 8 Section 351(i) of the PHS Act. 4

70 Contains Nonbinding Recommendations Draft Not for Implementation Considerations for developing presentations, container closure systems, and delivery device constituent parts for proposed interchangeable products 9,10 V. FACTORS IMPACTING THE TYPE AND AMOUNT OF DATA AND INFORMATION NEEDED TO SUPPORT A DEMONSTRATION OF INTERCHANGEABILITY The data and information needed to support a demonstration of interchangeability, beyond that needed to demonstrate biosimilarity, 11 may be dependent on and influenced by multiple factors, which are discussed in this section. A. Product-Dependent Factors That May Impact the Data Needed to Support a Demonstration of Interchangeability 1. Product Complexity and the Extent of Comparative and Functional Characterization Consistent with the guidance for industry Scientific Considerations in Demonstrating Biosimilarity to a Reference Product, the Agency recommends that sponsors use a stepwise approach generating data and information to address residual uncertainty about demonstrating interchangeability during product development. At each step, the sponsor should evaluate the extent to which there is residual uncertainty about the interchangeability of the proposed product with the reference product, and identify next steps to try to address that uncertainty. Section 351(k)(4)(A)(i) of the PHS Act provides that one of the criteria for FDA to make a determination of interchangeability is a finding that information in the application is sufficient to 9 Products that include both a biological product and a device constituent part to deliver the biological product are combination products (see 21 CFR parts 3 and 4). The delivery device constituent part and the biological product constituent part may be a single entity (e.g., a prefilled syringe) or the two constituent parts may be co-packaged (e.g., a biologic in a vial packaged in the same box with a syringe). The primary mode of action of these combination products is provided by the biological product constituent part, which is regulated by CDER or CBER. CDER or CBER, therefore, will have primary jurisdiction for these combination products; and these Centers and the Center for Devices and Radiological Health (CDRH) will coordinate as appropriate. 10 Considerations specific to demonstrating interchangeability under section 351(k)(4) of the PHS Act with respect to container closure systems and delivery device constituent parts are addressed in section VIII of this guidance. This guidance does not address other information generally necessary to support the proposed container closure system and/or the delivery device constituent part of a proposed product. Sponsors should also refer to relevant FDA guidance documents and resources from CBER, CDRH, CDER, and the Office of Combination Products (OCP) to assess what other data and information should be included to support the proposed container closure system(s) and/or delivery device constituent part(s). (Some of the FDA guidances and other resources that address these topics are referenced at appropriate places in section VIII of this guidance.) 11 Data and information needed to demonstrate biosimilarity are discussed in section VII of the guidance for industry Scientific Considerations in Demonstrating Biosimilarity to a Reference Product. We update guidances periodically. To make sure you have the most recent version of a guidance, check the FDA Drugs guidance Web page at 5

71 Contains Nonbinding Recommendations Draft Not for Implementation show that the proposed interchangeable product is biosimilar to the reference product. Such information would include, in part, a showing that the proposed interchangeable product meets the highly similar standard for demonstrating biosimilarity. 12 FDA acknowledges that there is a continuum of comparative analytical data that could support a demonstration that the highly similar standard is satisfied. 13 For example, a fingerprint-like characterization 14 may reduce residual uncertainty regarding interchangeability and inform the data and information needed to support a demonstration of interchangeability, which may lead to a more selective and targeted approach to clinical studies necessary to demonstrate interchangeability. Despite significant improvements in analytical techniques, current analytical methodologies may not detect or characterize all relevant structural and functional differences between the reference product and the proposed interchangeable product. 15 There may also be some structural features that specifically impact interchangeability (e.g., features that influence patient response to one product after exposure to another product). Data sets that include highly sensitive analytics and/or sequential analytical methods that can identify molecules with different combinations of attributes (e.g., charge variants and glycoforms), as well as a comprehensive assessment of the relationships between attributes, may provide information that reduces the residual uncertainty about interchangeability and thus inform the data and information needed to support a demonstration of interchangeability between the two products. The extent to which data and information provided by these advanced analytical approaches helps to reduce residual uncertainty about interchangeability depends on the degree of analytical similarity between the products and the strength of the evidence for the clinical relevance of the analytical data. Evidence of clinical relevance may range from a risk assessment describing the potential importance of the additional attributes evaluated to advanced modeling supported by functional and/or in vivo data. A clinically relevant and thus meaningful fingerprint-like characterization may reduce residual uncertainty regarding interchangeability and may lead to a more selective and targeted approach to the clinical studies necessary to demonstrate interchangeability. 12 Section 351(i)(2) of the PHS Act defines biosimilarity, in part, to mean that the biological product is highly similar to the reference product notwithstanding minor differences in clinically inactive components. 13 See the guidance for industry Quality Considerations in Demonstrating Biosimilarity of a Therapeutic Protein Product to a Reference Product for the Agency s current thinking on factors to consider when demonstrating that a proposed therapeutic protein product is highly similar to a reference product. 14 For information regarding fingerprint-like characterization, see the guidance for industry Quality Considerations in Demonstrating Biosimilarity of a Therapeutic Protein Product to a Reference Product. Also see the draft guidance for industry Clinical Pharmacology Data to Support a Demonstration of Biosimilarity to a Reference Product. When final, this guidance will represent the Agency s current thinking on this topic. Also see Kozlowski S, Woodcock J, Midthun K, Sherman RB, 2011, Developing the Nation s Biosimilars Program, N Engl J Med, 365: See Section IV.A. Nature of Protein Products and Related Scientific Considerations in the guidance for industry Scientific Considerations in Demonstrating Biosimilarity to a Reference Product. 6

72 Contains Nonbinding Recommendations Draft Not for Implementation The product s degree of structural and functional complexity may also influence the data and information needed to support a demonstration of interchangeability, because the product s structural complexity can affect the residual uncertainty regarding interchangeability. For example, products expected to have a single target (e.g., a receptor) may have less residual uncertainty regarding interchangeability than those acting on multiple or less-defined biological pathways. Along the continuum of possible data sets that could support a demonstration of the highly similar standard, there may be extensive characterization approaches that have some, but not all, of the features of a meaningful fingerprint-like characterization. These approaches could be of greater importance for more-complex products because these products would have a larger number of attributes and thus a potential for greater residual uncertainty regarding interchangeability. Such extensive characterization approaches may reduce the residual uncertainty regarding interchangeability for complex products. Reducing residual uncertainty can impact what additional data and information would be needed to support a demonstration of interchangeability. 2. Product-Specific Immunogenicity Risk Clinical experience with the reference product and comprehensive product risk assessments (e.g., regarding immunogenicity) 16 may also affect the data and information needed to support a demonstration of interchangeability. For example, products with a documented history of inducing detrimental immune responses may require more data to support a demonstration of interchangeability than products with an extensive documented history that immunogenicity does not impact clinical outcomes. 3. Totality of Factors to Consider in Assessing the Data and Information Needed to Support a Demonstration of Interchangeability The factors discussed in sections V.A.1 and V.A.2 of this guidance need to be considered together to inform a consideration regarding residual uncertainty about the data and information needed to support a demonstration of interchangeability. Consider the following illustrative examples: Product A has relatively low structural complexity, has been demonstrated to have meaningful fingerprint-like analytical similarity to the reference product as a part of demonstrating biosimilarity, and has a low incidence of serious adverse events related to immunogenicity. Here, data derived from an appropriately designed switching study (see section VI.A) may be sufficient to support a demonstration of interchangeability. Product B has high structural complexity, has been demonstrated to be highly similar to the reference product as a part of demonstrating biosimilarity but has no demonstration of meaningful fingerprint-like analytical similarity, and has known serious adverse events related to immunogenicity. Here, postmarketing data for the product as a licensed 16 Section VII.D.2 in the guidance for industry Scientific Considerations in Demonstrating Biosimilarity to a Reference Product provides a discussion on clinical immunogenicity assessment. 7

73 Contains Nonbinding Recommendations Draft Not for Implementation biosimilar, in addition to an appropriately designed switching study (see section VI.A), may provide additional data and information necessary to support a demonstration of interchangeability. The collection of biosimilar postmarketing data is described further in section V.B of this guidance. Based on the factors discussed in sections V.A.1 and V.A.2, the residual uncertainty regarding the interchangeability of the respective proposed products (described in the preceding examples) would be different. Therefore, the data and information necessary to support a demonstration of interchangeability needs to be considered on a case-by-case basis. B. Biosimilar Product Postmarketing Data That May Impact the Data Needed to Support a Demonstration of Interchangeability New tools and improved epidemiological approaches to evaluating postmarketing exposures and outcomes lend promise to the continued improvement of the capabilities of postmarketing surveillance and the collection of data related to the actual use of drug products in general. However, our current thinking is that postmarketing data collected from products first licensed and marketed as a biosimilar, without corresponding data derived from an appropriately designed, prospective, controlled switching study or studies, generally would not be sufficient to support a demonstration of interchangeability. For example, we generally would not expect postmarketing data to provide sufficient information related to the impact on clinical pharmacokinetics (PK) and pharmacodynamics (PD) of switching or alternating between the use of the proposed interchangeable product and the reference product, which we think are important study endpoint considerations in the switching studies for the reasons described in section VI.A.1 of this guidance. Notwithstanding these limitations, however, we recognize that in certain circumstances, postmarketing data from a licensed biosimilar product may be helpful as a factor when considering what data is necessary to support a demonstration of interchangeability. For example, some postmarketing data may describe the real-world use of the biosimilar product, including certain safety data related to patient experience with some switching scenarios. Such data may impact residual uncertainty about interchangeability and thus the data needed to support a demonstration of interchangeability. In certain situations, postmarketing surveillance data from the licensed biosimilar product in addition to data from an appropriately designed switching study may be needed to address residual uncertainty regarding a demonstration of interchangeability and add to the totality of the evidence to support a demonstration of interchangeability. Further, there may be situations where a postmarketing study, in addition to postmarketing surveillance data, from the licensed biosimilar product may be needed to address residual uncertainty regarding a demonstration of interchangeability. For example, as a scientific matter, where there is residual uncertainty regarding interchangeability based on immunogenicity-related adverse events that could affect use of the product as an interchangeable, a sponsor may need to first obtain licensure as a biosimilar product and collect postmarketing data before interchangeability can be demonstrated. In such cases, the type and amount of biosimilar product postmarketing data needed would depend on the residual uncertainty regarding the demonstration of interchangeability. Sponsors are encouraged to discuss with FDA their plans for the use of postmarketing data to address 8

74 Contains Nonbinding Recommendations Draft Not for Implementation residual uncertainty about interchangeability and add to the totality of the evidence to support a demonstration of interchangeability. VI. DATA AND INFORMATION NEEDED TO SUPPORT A DEMONSTRATION OF INTERCHANGEABILITY FDA advises sponsors intending to develop a proposed interchangeable product to meet with FDA to discuss their proposed product development plan. Early discussions with FDA about product development plans, including adequate scientific justification for the proposed development program, will facilitate development of interchangeable products. 17 A. Considerations for the Design and Analysis of a Switching Study or Studies Needed to Support a Demonstration of Interchangeability For biological products that are intended to be administered to an individual more than once, sponsors generally will be expected to conduct a switching study or studies to address the statutory provision for a biological product that is administered more than once to an individual, the risk in terms of safety or diminished efficacy of alternating or switching between use of the biological product and the reference product is not greater than the risk of using the reference product without such alternation or switch set forth in section 351(k)(4)(B) of the PHS Act. The main purpose of a switching study or studies is to demonstrate that the risk in terms of safety or diminished efficacy of alternating or switching between use of the proposed interchangeable product and the reference product is not greater than the risk of using the reference product without such alternation or switch. A switching study or studies should evaluate changes in treatment that result in two or more alternating exposures (switch intervals) to the proposed interchangeable product and to the reference product. For biological products that are not intended to be administered to an individual more than once, FDA expects that switching studies would generally not be needed. However, FDA expects that a sponsor will provide a justification for not needing data from a switching study as a part of the demonstration of interchangeability, and sponsors are encouraged to meet with FDA to discuss their planned development approach. Design of switching studies may be informed by how the proposed interchangeable product will be used in clinical practice, taking into consideration scenarios where alternating or switching products might cause the most clinical concern. For treatments that have a long course of therapy, sponsors should anticipate dropouts in the study and should use a scientifically justifiable method to address the increased possibility of missing data. It is important to note that if patients experience an immune response or adverse event during the course of a switching study, a carryover effect may make it difficult to determine whether the proposed interchangeable product or the reference product caused the event. If an apparent 17 See the guidance for industry Formal Meetings Between the FDA and Biosimilar Biological Product Sponsors or Applicants, which provides recommendations to industry on all formal meetings between the FDA and sponsors or applicants for biosimilar biological products intended to be submitted under 351(k) of the PHS Act. 9

75 Contains Nonbinding Recommendations Draft Not for Implementation difference in immune response or adverse events is noticed between the switching and nonswitching arms of the study (see section VI.A.2.a of this guidance), it would raise concerns as to whether the proposed interchangeable product is interchangeable, regardless of whether the proposed interchangeable product or the reference product or the switching of the two products actually caused the event. FDA has outlined a flexible approach regarding the design of any necessary switching study. FDA will address program-specific scientific matters (e.g., the impact of small patient populations) on a case-by-case basis in interactions with sponsors. To facilitate development of interchangeable products, FDA encourages sponsors to have early discussions with FDA about their product development plans. 1. Study Endpoints The primary endpoint in a switching study or studies should assess the impact of switching or alternating between use of the proposed interchangeable product and the reference product on clinical pharmacokinetics and pharmacodynamics (if available), because these assessments are generally most likely to be sensitive to changes in immunogenicity and/or exposure that may arise as a result of alternating or switching. FDA recommends that clinical PK and PD test methods and assays be developed and validated early in product development. The validation should consider both the proposed interchangeable product and the reference product. Although assessments of efficacy endpoints can be supportive, at therapeutic doses many clinical efficacy outcomes would only be sensitive to large changes in exposure or immunogenicity, which may not be observed in a study of limited duration and with a limited number of switches. Biologically relevant PD measures, if available, may be useful as shorter-term, more-sensitive indicators of the potential impact of alternating or switching on the risk of diminished efficacy as compared to efficacy endpoints. Relevant PD measures may also be useful to reflect multiple domains of activity, which could reduce residual uncertainty about interchangeability. Selection of PD endpoints should be scientifically justified for the intended purpose. When PD endpoints that are sensitive to changes in drug concentration can be identified, PD analysis, in addition to PK analysis, may be useful to address residual uncertainty with respect to interchangeability. In addition to PK and PD parameters, a switching study or studies would also be expected to assess immunogenicity and safety. 2. Study Design and Analysis a. Dedicated Switching Study Design A study with a lead-in period of treatment with the reference product, followed by a randomized two-arm period with one arm incorporating switching between the proposed interchangeable product and the reference product (switching arm) and the other remaining as a non-switching arm receiving only the reference product (non-switching arm) may be appropriate when designing a switching study. Considerations for the design and analysis of such a study are discussed as follows: 10

76 Contains Nonbinding Recommendations Draft Not for Implementation Sample size: The sample size of the switching study should generally be based on PK considerations (inter-subject variability in AUC tau or C max should be primary considerations) and should be appropriately justified. As the switching study will likely require repeated patient monitoring, the study designers should anticipate the possibility of a considerable dropout rate for reasons unrelated to the study treatment arms. An anticipated high dropout rate due solely to an influence affecting all treatment arms could be assumed to be random. The negative impact on the statistical power of such a random influence could be precluded by factoring such influences into the sample size calculation. It should be noted that dropout rates or missing data rates that differentially affect the study treatment arms could represent treatment arm differences, and sponsors should provide adequate justification to FDA about any such differences and their possible causes. In addition, FDA will investigate possible causes of the noted differences in treatment arms. Number and duration of switches: The number and duration of switches between the reference product and the proposed interchangeable product should take into consideration the clinical condition to be treated, the therapeutic dosing of the product, and the duration of the exposure interval to each product that would be expected to cause the greatest concern in terms of immune response and resulting impact on safety and efficacy, if any. - The lead-in period should be of sufficient duration to ensure an adequate baseline with respect to the study (e.g., steady state of pharmacokinetics) before randomization to the switching period of the study. - The switching arm is expected to incorporate at least two separate exposure periods to each of the two products (i.e., at least three switches with each switch crossing over to the alternate product). - The last switching interval should be from the reference product to the proposed interchangeable product, where the duration of exposure to the proposed interchangeable product after the last switch is sufficiently long to allow for washout of the reference product (i.e., at least three or more half-lives) to assess the pharmacokinetics of the proposed interchangeable product in the switching arm and compare it to the pharmacokinetics of the reference product in the non-switching arm. PK, PD, and immunogenicity sampling: To capture the full PK profile, intensive PK sampling should be performed during the last switch interval following the dose after which at least three half-lives of the reference product have elapsed. Trough PK sampling should be conducted after each switch to ensure that steady state is attained. The timing of PD 18 and immunogenicity 19 sampling should be appropriately justified. 18 See Section IV.H. Defining the Appropriate Pharmacodynamic Time Profile in the draft guidance for industry Clinical Pharmacology Data to Support a Demonstration of Biosimilarity to a Reference Product. When final, this guidance will reflect FDA s current thinking on this topic. 11

77 Contains Nonbinding Recommendations Draft Not for Implementation Study Analysis: - Primary analysis: The following PK data obtained during the intensive sampling interval should be reported: C max, T max, C trough, and AUC tau. The log-transformed AUC tau and C max data should be statistically analyzed using an analysis of the variance. The 90% confidence interval for the geometric mean ratio of AUC tau and C max between the proposed interchangeable product and the reference product should be within %. C trough and T max should also be analyzed as secondary endpoints. PD endpoints, when evaluated, should be measured at appropriate times during the PK sampling interval. - Safety, immunogenicity, and efficacy should be descriptively analyzed as secondary endpoints. b. Integrated Study Design If a sponsor is considering a study design using a single study intended to (1) support a demonstration of no clinically meaningful differences between the reference product and the proposed product for biosimilarity 20 and (2) evaluate the impact of switching or alternating between the reference product and the proposed product for interchangeability, an integrated, two-part study design may be appropriate. Following the time point(s) for evaluation of the appropriate endpoint(s) to support the demonstration of no clinically meaningful differences for biosimilarity between the proposed product and the reference product in the first part of the study, the subjects in the reference product arm should be re-randomized in the second part of the study to continue to receive the reference product (non-switching reference product arm) or to switch to the proposed product (switching arm) as described in section VI.A.2.a of this guidance. FDA recommends continuing the proposed product arm (non-switching proposed product arm) from the inception of the study, through the duration of the switching portion of the integrated study, to the completion of the study. An integrated study needs to be adequately powered to evaluate the appropriate endpoint(s) to support the demonstration of no clinically meaningful differences for biosimilarity, where the primary comparison is between the proposed product arm and the reference product arm. In addition, the study needs to be adequately powered to evaluate pharmacokinetics and pharmacodynamics (if available), following the last switch to support a demonstration of interchangeability, where the primary comparison is between the switching arm and the nonswitching reference product arm. 19 See Section VII.A. Obtaining Patient Samples in the draft guidance for industry Assay Development for Immunogenicity Testing of Therapeutic Proteins. Also see Section IV. Recommendations for Mitigating Immunogenicity Risk in the Clinical Phase of Development of Therapeutic Protein Products in the guidance for industry Immunogenicity Assessment of Therapeutic Protein Product. 20 Data and information needed to demonstrate biosimilarity are discussed in section VII of the guidance for industry Scientific Considerations in Demonstrating Biosimilarity to a Reference Product. 12

78 Contains Nonbinding Recommendations Draft Not for Implementation 3. Study Population The study population for switching studies should be adequately sensitive to allow for detection of differences in pharmacokinetics and/or pharmacodynamics, common adverse events, and immunogenicity between the switching and non-switching arms. Even though it is likely that the study population will generally have characteristics that are consistent with those of the population studied for licensure of the reference product for the same indication, sponsors may conduct switching studies in a patient population that is different from that used to support licensure of the reference product. Sponsors should provide adequate scientific justification to support that such a population is adequately sensitive to detect the impact of switching (e.g., differences in clinical pharmacokinetics and/or pharmacodynamics, common adverse events, and immunogenicity). FDA strongly recommends that sponsors use patients in switching studies because these studies are designed to mimic how the proposed interchangeable product will be used in clinical practice. In a circumstance where a sponsor considers using healthy subjects, the sponsor should weigh the benefit of exposing healthy subjects to a proposed interchangeable product during the course of a clinical study against the risk of having them develop antibodies to the product, which in turn may preclude them from being able to receive the treatment in the future, if needed. However, there may be some limited situations where it is clinically and ethically appropriate to use healthy subjects in switching studies. Sponsors are strongly encouraged to discuss with FDA their rationale for conducting switching studies in healthy subjects before initiating studies, preferably before submitting a proposed protocol or protocol amendment. 4. Condition of Use To Be Studied As described in section VI.B of this guidance, sponsors should consider choosing a condition of use that would support subsequent extrapolation of data to other conditions of use. In addition, it is important to note that a sponsor may obtain licensure only for a condition of use (or uses) for which the reference product is licensed. If a reference product has multiple conditions of use and one of those conditions of use was licensed under section 506(c) of the Federal Food, Drug, and Cosmetic Act and 21 CFR part 601, subpart E (accelerated approval), and the reference product s clinical benefit in this condition of use has not yet been verified in postmarketing studies, then sponsors should consider studying another condition of use for which the reference product is licensed, to avoid complications in the event that postmarketing studies fail to verify the reference product s clinical benefit for the condition of use being considered under the accelerated approval provisions. 5. Route of Administration If a product is approved for more than one route of administration, sponsors should study the route of administration that will best assess how a patient s immune response will impact the clinical performance of the proposed interchangeable product, including changes in safety risk and efficacy. Choosing a more immunogenic route of administration (e.g., subcutaneous rather 13

79 Contains Nonbinding Recommendations Draft Not for Implementation than intravenous) for use in switching studies may help sponsors anticipate the clinical implications of real-world use in clinical practice. B. Extrapolation of Data If the proposed product meets the statutory requirements for licensure as an interchangeable product under section 351(k) of the PHS Act based on, among other things, data and information sufficient to demonstrate interchangeability in an appropriate condition of use, the sponsor may seek licensure of the proposed product as an interchangeable product for one or more additional conditions of use for which the reference product is licensed. The sponsor would need to provide sufficient scientific justification for extrapolating data to support a determination of interchangeability for each condition of use for which the reference product is licensed and for which licensure as an interchangeable product is sought. The scientific justification for extrapolation should address, for example, the following issues for the tested and extrapolated conditions of use: The mechanism(s) of action in each condition of use for which the reference product is licensed, which may include the following: - The target receptor(s) for each relevant activity/function of the product - The binding, dose/concentration response, and pattern of molecular signaling upon engagement of target receptor(s) - The relationship between product structure and target/receptor interactions - The location and expression of target receptor(s) The pharmacokinetics and biodistribution of the product in different patient populations (Relevant PD measures may also provide important information on the mechanism(s) of action.) The immunogenicity risk of the product in different patient populations Differences in expected toxicities in each condition of use and patient population (including whether the expected toxicities are related to the pharmacological activity of the product or to off-target activities) Any other factor that may affect the safety or efficacy of the product in each condition of use and patient population for which the reference product is licensed 21 Differences between conditions of use with respect to the factors described do not necessarily preclude extrapolation. A scientific justification should address these differences in the context of the totality of the evidence supporting a demonstration of interchangeability. Advanced 21 These factors are also discussed in Section VII.D.4. Extrapolation of Clinical Data Across Indications in the guidance for industry Scientific Considerations in Demonstrating Biosimilarity to a Reference Product. 14

80 Contains Nonbinding Recommendations Draft Not for Implementation structural and functional characterization may also provide additional support for the justification for extrapolation. In choosing a condition of use to study that would permit subsequent extrapolation of data to other conditions of use, FDA recommends that a sponsor consider choosing a condition of use that would be adequately sensitive to assess the risk of alternating or switching between the products, in terms of safety or diminished efficacy, in a switching study and subsequently support extrapolation based on the factors described in this section. VII. USE OF A U.S.-LICENSED REFERENCE PRODUCT IN A SWITCHING STUDY OR STUDIES In the context of demonstrating biosimilarity to a reference product, FDA has advised that sponsors may seek to use data derived from animal or clinical studies comparing a proposed product with a non-u.s.-licensed comparator product to address, in part, the requirements under section 351(k)(2)(A) of the PHS Act. 22,23 In clinical studies used to support a demonstration of no clinically meaningful differences as a part of demonstrating biosimilarity, the comparator product (whether it is a non-u.s.-licensed product or a U.S.-licensed reference product) serves as a control against which the proposed product is evaluated. However, in a switching study that is designed to evaluate the impact of switching or alternating to support a determination of interchangeability, the comparator product plays a different role. Rather than being used only as a control, the comparator product is used in a switching study in both the active switching arm and the control non-switching arm. Switching studies are designed to assess whether one product will affect the immune system s response to the other product, once the switch occurs, and whether this will result in differences in immunogenicity or PK profiles. Thus, using a non-u.s.-licensed comparator product generally would not be appropriate in a switching study for the following reasons: 24 It is possible that the proposed interchangeable product and the non-u.s.-licensed comparator product have, for example, subtle differences in levels of specific structural features (e.g., acidic variants, deamidations). The immune system reaction in terms of the overall level of antibody produced to each product could be similar, thereby supporting a demonstration of no clinically meaningful differences. Thus, these subtle differences would not preclude a demonstration of 22 See section V on U.S.-licensed reference product and other comparators in the guidance for industry Scientific Considerations in Demonstrating Biosimilarity to a Reference Product. 23 See Q.I.8 in the guidance for industry Biosimilars: Questions and Answers Regarding Implementation of the Biologics Price Competition and Innovation Act of 2009, which discusses use of a non-u.s.-licensed product to support a demonstration that the proposed product is biosimilar to the reference product. 24 See Q.I.8 in the guidance for industry Biosimilars: Questions and Answers Regarding Implementation of the Biologics Price Competition and Innovation Act of 2009, which explains that [a]t this time, as a scientific matter, it is unlikely that clinical comparisons with a non-u.s.-licensed product would be an adequate basis to support the additional criteria required for a determination of interchangeability with the U.S.-licensed reference product. 15

81 Contains Nonbinding Recommendations Draft Not for Implementation biosimilarity. However, with switching, multiple exposures to each product can prime the immune system to recognize subtle differences in structural features between products, and the overall immune response could be increased under these conditions. This immunologic response is highly dependent on the structural differences between the proposed interchangeable product and the comparator product used in the switching study, in addition to other potential differences between the products (e.g., impurities). Because there may be subtle differences between the U.S.-licensed reference product and the non-u.s.-licensed comparator product, there is uncertainty as to whether the results observed in a switching study using a non-u.s.-licensed comparator product would also be observed if the U.S.-licensed reference product had been used instead. Under the BPCI Act, an interchangeable product may be substituted for the reference product without the prescribing health care provider s intervention. There may be multiple versions of a non-u.s.-licensed comparator product on the international market, each approved for use by the relevant national regulatory authority and each with possible subtle differences in levels of structural features from the U.S.-licensed reference product and between each other. The goal of a switching study or studies is to determine a biosimilar product s interchangeability with a reference product that is licensed for use in U.S. clinical settings, thus establishing interchangeability with a product that patients will not receive in the United States would generally not be appropriate. For these reasons, FDA strongly recommends that sponsors use a U.S.-licensed reference product in a switching study or studies. Sponsors are encouraged to contact FDA early in the product development process to discuss the design of a switching study, including any proposal to provide adequate scientific justification to support the use of data generated in a switching study using a non-u.s.-licensed comparator product to support a demonstration of interchangeability. VIII. CONSIDERATIONS FOR DEVELOPING PRESENTATIONS FOR PROPOSED INTERCHANGEABLE PRODUCTS The data and information needed to support a demonstration of interchangeability, beyond that needed to demonstrate biosimilarity, 25 may also be influenced by the proposed product s presentation. 26 This section provides a framework for sponsors to determine the types of data and information related to a proposed presentation that might be necessary to support a demonstration of interchangeability. The considerations described in this section are intended to provide clarity and support flexibility, where appropriate. 25 Data and information needed to demonstrate biosimilarity are discussed in section VII of the guidance for industry Scientific Considerations in Demonstrating Biosimilarity to a Reference Product. 26 For the purposes of this guidance, the term presentation means the container closure system and/or delivery device constituent part of the product. 16

82 Contains Nonbinding Recommendations Draft Not for Implementation Clarity: The framework outlined in this section is designed to provide clear recommendations to guide the development of interchangeable product presentations. Often decisions regarding a presentation are made early in product development. The approach described is intended to reduce potential uncertainty during product development with respect to a proposed presentation and enable sponsors to conduct a product-specific evaluation of their proposed presentation. Flexibility: FDA anticipates that sponsors of proposed interchangeable products may develop presentations that have some differences in design from the presentations licensed for the reference product. FDA does not expect that all differences in the design of the presentation of a proposed interchangeable product, when compared to the presentation of a reference product, would negatively impact the appropriate use 27 of the product when substituted for the reference product. We intend this section to assist sponsors in tailoring the data and information needed to support a demonstration of interchangeability of their proposed product. The threshold analyses described in section VIII.B.1.a of this guidance are recommended for all proposed interchangeable products to identify any differences in design between the proposed interchangeable product and the reference product. If there are differences other than minor as observed in the threshold analyses described in section VIII.B, sponsors can use the results from the threshold analyses to determine the need, if any, for additional data or information, such as data and information from a comparative human factors study. FDA expects that such additional studies will likely not be needed for many interchangeable products. A. General Considerations When developing a product for licensure as interchangeable under section 351(k) of the PHS Act, it is important that sponsors carefully consider the presentation of the proposed interchangeable product relative to the reference product. 28 A sponsor developing an interchangeable product generally should not seek licensure for a presentation for which the reference product is not licensed. For example, if the reference product is only marketed in a vial and a prefilled syringe, a sponsor should not seek licensure for the proposed interchangeable product for a different presentation, such as an auto-injector. A sponsor planning to develop a presentation for which the reference product is not licensed should discuss its proposed presentation with FDA. In such cases, FDA will evaluate whether the proposed presentation could support a demonstration of interchangeability. As applicable, a general description of the entire container closure system should be provided in the chemistry, manufacturing, and controls (CMC) section of the application. There should be 27 The terms appropriate use or appropriately use are sometimes used in this section for brevity to refer to use of the proposed interchangeable product in a manner that supports a demonstration of interchangeability under section 351(k) of the PHS Act. 28 See Q.I.4 and Q.I.6 in the guidance for industry Biosimilars: Questions and Answers Regarding Implementation of the Biologics Price Competition and Innovation Act of

83 Contains Nonbinding Recommendations Draft Not for Implementation complete CMC information for the proposed product, including delivery device constituent part design and development information. The presentation should be shown to be compatible for use with the final formulation of the proposed product through appropriate studies, including, for example, extractable/leachable studies, performance testing, and stability studies. Data and information supporting the appropriate use and performance testing of the delivery device constituent part of the proposed product should be submitted. B. Analysis of Proposed Presentations of Proposed Interchangeable Products The use of a biological product generally involves a sequence of administration steps because biological products are generally injected or infused into the body. In addition, these products are administered by a variety of end users, including health care providers, patients, caregivers, or a combination of these end users. 29,30 The design of the presentation determines the specific tasks necessary to administer the product. These tasks can vary considerably depending on the type of presentation and its design characteristics. Differences in the design of the container closure system or delivery device constituent part between the proposed interchangeable product and the reference product may be acceptable provided that the design differences are analyzed appropriately and data are provided to demonstrate that the changes do not negatively impact the ability of end users, including patient and caregiver end-user groups, to appropriately use these products when the interchangeable product is substituted for the reference product without the intervention of the prescribing health care provider or additional training before use. Because a proposed interchangeable product may be substituted for the reference product without the intervention of the health care provider who prescribed the reference product, a proposed interchangeable product with a differently designed presentation than the reference product may raise uncertainty about whether the difference in presentations would impact the ability of end users, including patients or caregivers, to appropriately use the proposed product. Therefore, FDA recommends that sponsors analyze the presentations of a proposed interchangeable product to identify differences in design compared to the presentations licensed for the reference product using the threshold analysis outlined in this section. These threshold analyses may be used in the development of the proposed presentation to minimize differences between the proposed interchangeable product and the reference product as well as to identify whether additional data, including data from comparative use human factors studies (as described further in this section), may be needed in certain circumstances. To conduct the analysis of the presentations of the proposed interchangeable product and reference product for the purposes of identifying differences between the presentations, sponsors should examine the external critical design attributes of the proposed interchangeable product in 29 Administration steps at a high level include the aseptic technique to manipulate the product to prepare it for injection and to ensure that the right dose is administered, followed by a physical manipulation to inject the biologic in the correct site and by the correct route. 30 As a scientific matter, FDA recognizes that the end users of biological products may have different training and expertise, and we provide some technical considerations in this section and in Appendix A for sponsors to consider, as appropriate. 18

84 Contains Nonbinding Recommendations Draft Not for Implementation comparison to those of the reference product. External critical design attributes are those features that directly affect the performance of critical tasks 31 that end users perform to appropriately use or administer the product. To identify these attributes, a sponsor should examine the overall external operating principles of the container closure system or delivery device constituent part by evaluating all the tasks that an end user needs to perform to prepare and administer the product. The external critical design attributes of the product would be those features that end users rely on to perform the tasks identified as critical to the appropriate use of the product. Because these attributes may impact appropriate use of the product, FDA recommends that sponsors consider the external critical design attributes of the reference product as part of their development program for a proposed interchangeable product. The technical description of the threshold analysis appears in the next section, which may be of general use for sponsors in the development of proposed interchangeable products. In those circumstances where a threshold analysis indicates that further data may need to be gathered from comparative use human factors studies, Appendix A provides a technical description of comparative use human factors studies intended to support a demonstration of interchangeability. 1. Threshold Analyses Three types of threshold analyses can be used in the development program for the purposes of identifying and evaluating differences in design and should be conducted after the presentation of the proposed interchangeable product, including product design and user interface, 32 have been finalized by the sponsor and are believed to be representative of the commercial product. FDA recommends that sponsors carefully evaluate the risks associated with differences in the container closure system(s) and/or the delivery device constituent part(s) for proposed interchangeable products that may affect the patient or caregiver as the end user, 33 especially because interchangeable products may be substituted for the reference product without the intervention of the prescribing health care provider or additional training before use. The patient or caregiver end-user groups may not receive additional training in such circumstances and may lack the expertise that a health care provider user group is expected to possess. Patient and caregiver end-user groups may be less accustomed to navigating differences in container closure 31 For additional information on critical tasks, see Section III.B.1. Critical Tasks in the draft guidance for industry Human Factors Studies and Related Clinical Study Considerations in Combination Product Design and Development. When final, this guidance will reflect FDA s current thinking on this topic. 32 The user interface includes all components of the delivery device constituent part with which the user interacts, such as controls and displays (i.e., those parts of the delivery device constituent part that users see, touch, and hear). The user interface also includes the delivery device constituent part labeling, which includes package labels, any instructions for use in user manuals, package inserts, instructions on the delivery device constituent part itself, and any accompanying informational materials. For additional insight, see the guidance for industry and FDA staff Applying Human Factors and Usability Engineering to Medical Devices. 33 For additional information about end-user group considerations, see Section III.B.2. Intended Users and Use Environment in the draft guidance for industry Human Factors Studies and Related Clinical Study Considerations in Combination Product Design and Development. When final, this guidance will reflect FDA s current thinking on this topic. 19

85 Contains Nonbinding Recommendations Draft Not for Implementation systems and/or delivery device constituent parts for biological products than health care providers. As a result, there is concern that patients or caregivers who encounter different external critical design attributes between the container closure system and/or delivery device constituent part of a reference product and a proposed interchangeable product may be at increased risk for a use-related error that may impact their ability to appropriately use these products. a. Types of threshold analyses The following three types of analyses are recommended as part of the threshold analyses of proposed product presentation for all proposed interchangeable products: i. Labeling comparison FDA recommends a side-by-side, line-by-line comparison (between the reference product and the proposed interchangeable product) of the full prescribing information, instructions for use, and descriptions of the container closure systems and/or delivery device constituent parts. ii. Comparative task analysis FDA recommends that sponsors conduct a comparative task analysis between the reference product and the proposed interchangeable product. 34 iii. Physical comparison of the interchangeable product and the reference product, along with their respective container closure system and/or delivery device constituent part FDA recommends that sponsors of proposed interchangeable products acquire the reference product to examine (e.g., visual and tactile examination) the physical features of the reference product and compare them to those of the proposed interchangeable product. b. Outcomes of threshold analyses After completing the threshold analyses, the following outcomes are possible: i. No design differences When no differences are identified in the design of the presentation of the proposed interchangeable product and the reference product after the threshold analyses, it is likely that additional data to support the appropriate use of the proposed interchangeable product by the end 34 To conduct a comparative task analysis, sponsors should systematically dissect the use process for each product (i.e., both the proposed interchangeable product and the reference product) and analyze and compare the sequential and simultaneous manual and intellectual activities for end users interacting with both products. FDA recommends that sponsors analyze the differences, with the goal of characterizing the potential for use error. Also see the American National Standards Institute/Association for the Advancement of Medical Instrumentation HE75, 2009(R)2013 Human Factor Engineering Design of Medical Devices. The standard can be accessed at 20

86 Contains Nonbinding Recommendations Draft Not for Implementation users, including data from comparative use human factors studies, will not be necessary to support licensure as an interchangeable product. The sponsor should provide any analyses comparing the presentations for FDA s review and concurrence. ii. Differences in design If differences are identified between the design of the presentations of the proposed interchangeable product and the reference product, the sponsor should focus on whether the difference(s) involves an external critical design attribute that can negatively impact appropriate use by the patient and caregiver end-user groups and should seek to establish and categorize the differences as follows: Minor design differences: FDA views a design difference in product presentation as minor if the differences in the user interface of the proposed interchangeable product, in comparison to the user interface of the reference product, do not affect an external critical design attribute. Minor differences in design are likely to be viewed by FDA as acceptable provided that the data and information submitted by the sponsor demonstrate that the differences are in fact minor. For example, such data and information may be collected from thorough threshold analyses (described in sectionviii.b.1.a of this guidance) that demonstrate that the differences in design do not involve an external critical design attribute that could negatively impact appropriate use. Similarly, for those products that would be expected to be administered only by a health care provider, the risks associated with substitution may be adequately addressed through threshold analyses rather than a comparative use human factors study. As mentioned previously, patient and caregiver end-user groups may be less accustomed to navigating differences in container closure systems and/or delivery device constituent parts for biological products than health care providers. The sponsor should provide this data and information for FDA s review and concurrence. Other design differences: FDA may not view a design difference as minor if any aspect of the threshold analyses suggests that differences in the design of the presentation of the proposed interchangeable product as compared to the reference product may impact an external critical design attribute that involves patient use or caregiver administration of the product. In such cases, the sponsor may consider modifying the design of the proposed presentation to minimize differences from the reference product, which could reduce the data that might be needed to support a demonstration of interchangeability. 35 Alternatively, if such differences are present 35 FDA recognizes that, in certain circumstances, a sponsor may elect to retain a difference in design of an interchangeable product compared to the reference product to reduce difficulty with use or to minimize risk associated with the design of the reference product s presentation. FDA generally encourages the optimization of the design of the delivery device constituent part to enhance the safety of the product. However, there may be circumstances where an interchangeable product may be substituted for the reference product without additional end-user training. Thus, it is important that sponsors identify differences between their proposed presentation and the reference product s presentation, as described in this guidance, including identifying any differences intended to optimize the design of the delivery device constituent part to enhance the safety of the product. The sponsor should also characterize any identified differences in design that may impact the end user s ability (particularly patients and 21

87 Contains Nonbinding Recommendations Draft Not for Implementation in the final design of the presentation of the proposed interchangeable product, FDA recommends that sponsors provide appropriate data from additional studies, such as from a comparative use human factors study, to address whether such differences might negatively affect the appropriate use of the biological product in circumstances where the interchangeable product is substituted for the reference product. The data from additional studies should seek to characterize whether the difference(s) could negatively affect the appropriate use of the products by patients and caregivers (see section VIII.B.2 of this guidance for types of studies). Based on the results of additional studies, FDA may or may not determine that the design difference between the presentation of the proposed interchangeable product and the reference product is acceptable for a proposed interchangeable product. 2. Studies to Evaluate Differences That May Not Be Minor as Observed in Threshold Analyses If the threshold analyses determine that a design difference may not be minor, as described in section VIII.B.1.b.ii of this guidance, sufficient evidence should be provided to permit FDA to evaluate the design difference for purposes of interchangeability. Alternatively, as mentioned previously, the sponsor may consider modifying the design of the proposed presentation to minimize differences from the reference product, which could reduce the data that might be needed to support a demonstration of interchangeability. However, if differences that may not be minor are present in the final design of the presentation of the proposed interchangeable product, FDA recommends that sponsors provide appropriate data from additional studies to support these differences. Such data may be gathered in a focused comparative use human factors study that evaluates the critical tasks related to the external critical design attributes that are found to be different or to focus on the patient and caregiver end-user group(s) that are most likely to be negatively impacted by the differences in the design of the presentation of the proposed interchangeable product and the reference product. a. Comparative use human factors studies Comparative use human factors studies may be needed to provide the evidence necessary to assess whether differences that may not be minor in the design of the presentation of the proposed interchangeable product prevent licensure of the proposed product as interchangeable with the reference product. The objective of the comparative use human factors studies described in this guidance is to assess any differences in the use error rate between the reference product and the proposed interchangeable product. This objective differs from the objective of human factors validation studies, which are conducted to evaluate how a product s user interface supports safe and effective use; such studies are not designed to assess differences in use error rates between two products. Therefore, the human factors validation studies described in the guidance for industry and FDA staff Applying Human Factors and Usability Engineering to Medical Devices generally do not apply when evaluating interchangeability. caregivers) to appropriately use an interchangeable product when substituted for the reference product. (See sectionviii.b.2 of this guidance.) 22

88 Contains Nonbinding Recommendations Draft Not for Implementation See Appendix A of this guidance for considerations for comparative use human factors studies (if needed) to evaluate differences that may not be minor, as observed in threshold analyses. b. Additional studies The need for additional data or information to support a presentation beyond what is described in this guidance may depend on a risk-based analysis and will be determined on a case-by-case basis. Additional studies such as comparative in vivo or in vitro performance testing may be warranted to support a demonstration of interchangeability under section 351(k) of the PHS Act. 36 FDA recommends that sponsors define specifications for each testing parameter before studies are initiated. Sponsors are encouraged to discuss appropriate testing with FDA as early during product development as feasible. IX. POSTMARKETING SAFETY MONITORING CONSIDERATIONS Robust postmarketing safety monitoring is an important component in ensuring the safety and effectiveness of biological products, including biosimilar and interchangeable products. Postmarketing safety monitoring for interchangeable products should first take into consideration any particular safety or effectiveness concerns associated with the use of the reference product and its class, the proposed interchangeable product in its development and clinical use (if marketed outside the United States), the specific condition of use and patient population, and patient exposure in the interchangeability development program. Postmarketing safety monitoring for an interchangeable product should also have adequate pharmacovigilance mechanisms in place. 37 Rare but potentially serious safety risks may not be detected during preapproval clinical testing because the size of the population exposed likely will not be large enough to assess rare events. In particular cases, such risks may need to be evaluated through postmarketing surveillance or studies. In addition, as with any other biological product, FDA may require a postmarketing study or a clinical trial to evaluate certain safety risks. 38 Because some aspects of postmarketing safety monitoring are product-specific and dependent upon the risk that is the focus of monitoring, FDA encourages sponsors to consult with appropriate FDA divisions to discuss the sponsor s proposed approach to postmarketing safety monitoring. 36 Comparative in vivo or in vitro performance testing may include the critical elements in establishing dose accuracy; for example, extended needle length, needle integrity, activation force, dispensing time, and dispensing volume. Additional types of performance specification testing, such as activation force, breakloose force, extrusion force, needle gauge, and needle protrusion, may need to be considered as well. 37 For general pharmacovigilance considerations, see the guidance for industry Good Pharmacovigilance Practices and Pharmacoepidemiologic Assessment and the guidance for industry Postmarketing Adverse Experience Reporting for Human Drug and Licensed Biological Products: Clarification of What to Report. 38 See section 505(o)(3) and 505(p)(1)(A)(ii) of the Federal Food, Drug, and Cosmetic Act. 23

89 Contains Nonbinding Recommendations Draft Not for Implementation APPENDIX A: COMPARATIVE USE HUMAN FACTORS STUDIES Considerations for Comparative Use Human Factors Studies, if needed, to evaluate differences that may not be minor as observed in threshold analyses: 1. Study Design Considerations To support a demonstration of interchangeability under section 351(k) of the PHS Act, a comparative use human factors study should be designed to provide data supporting that the use error rate for the proposed interchangeable product is not worse than the use error rate for the reference product when used by patients and caregivers (as applicable) in representative use scenarios and use environments. The comparative use human factors studies described in this guidance would generally be simulated-use studies 39 where the participants, who are representative of the patients and caregivers, are asked to simulate the use of the product presentations (container closure systems and/or delivery device constituent parts) without actually administering the product. For many aspects of demonstrating interchangeability under section 351(k) of the PHS Act, data is best collected using an equivalence study design. For example, it would be unlikely for FDA to determine that a proposed product is interchangeable with a reference product if data showed it to have lower or higher exposure than the reference product. In such cases, there could be a negative impact to the patient associated with a substantial deviation from equivalence. However, for the purpose of the comparative use human factors studies described in this appendix, the risks associated with container closure systems and delivery device constituent parts are derived from errors that occur in using the container closure system and/or delivery device constituent part. FDA would generally accept a proposed interchangeable product that had the same rates of error as the reference product, as demonstrated by an adequately designed comparative use human factors study or studies. However, we also recognize that lower error rates for a proposed interchangeable product compared to error rates for the reference product would likely not be considered to negatively impact the interchangeability assessment. Therefore, lower bounds on error rates are generally not necessary in comparative use human factors studies described in this appendix. For this reason, instead of using equivalence designs, noninferiority (NI) study designs are generally appropriate in such situations. NI tests comparing use of the presentation of a proposed interchangeable product to that of the reference product are similar to usual statistical tests for a difference, but translated to account for allowable differences in design between the presentation of the proposed interchangeable product and the reference product. In comparing pharmaceutical products, NI tests are often conducted to indirectly demonstrate that a proposed product is more efficacious than a placebo. A standard way of approaching this goal is to fix an NI margin (referred to as d in this appendix) at some fraction of the difference 39 For more information on simulation techniques, see Section D.1. Human Factors Simulated Use Validation Studies in the draft guidance for industry Human Factors Studies and Related Clinical Study Considerations in Combination Product Design and Development. When final, this guidance will reflect FDA s current thinking on this topic. 24

90 Contains Nonbinding Recommendations Draft Not for Implementation between the placebo effect and the effect of the proposed product. Showing the effect of the proposed product to be significantly better than the margin demonstrates NI. The draft guidance for industry Non-Inferiority Clinical Trials discusses meta-analyses and margin selection in detail. 40 A comparative human factors study with an NI design for the purpose of demonstrating interchangeability under section 351(k) of the PHS Act will typically be less complicated than those described in the guidance on NI clinical trials because the endpoints of these NI studies will not be dependent on therapy and the placebo effect will not be a confounding factor. The choice of comparative use human factors study endpoint(s) will depend on the nature of the presentation being evaluated and the associated differences between presentation of the proposed interchangeable product and the reference product identified in the threshold analyses. Although there may be many possible endpoints, a study evaluating performance of a critical task can often be reduced to a binary endpoint that considers whether or not the end user makes an error in performing the task. One possible endpoint for such a study may be the rates of errors observed when using the presentations of the proposed interchangeable product and the reference product. In this guidance, we show ER IP and ER RP as error rates observed when using the presentation associated with the proposed interchangeable product and that of the reference product, respectively. Using the result of the threshold analyses described earlier as a guide, a risk assessment should be done to identify the external critical design attributes, end-user group(s), use scenarios, and use environments on which to focus the comparative use human factors study intended to support a demonstration of interchangeability. FDA recommends that patient and caregiver end users (as applicable) of the reference product be considered for inclusion in the comparative use human factors study. The risk assessment should explore risks for the various subgroups of the current patient and caregiver end-user groups and may identify an appropriate subpopulation on which to focus the comparative use human factors study. For example, in some cases, the risk assessment may determine that only a certain patient subpopulation (or subpopulations) is likely to experience difficulty administering the product, and thus the comparative use human factors study may be most appropriately focused on the identified patient subpopulation(s). The goal of a comparative use human factors study intended to support a demonstration of interchangeability with an NI design is to demonstrate that ER IP is no greater than ER RP +d, where d is some acceptable deviance above ER RP. In determining the margin d, the variability in ER RP should be considered as well as the risk any difference in outcomes will pose to patients. The results of the risk assessment should be considered when determining the NI margin (d) between ER RP and ER IP. An example of a simple and direct approach to an NI test comparing ER IP and ER RP can be summarized as follows: Determine the allowable margin (d) by which ER IP could exceed ER RP. 40 For additional insight, see the draft guidance for industry Non-Inferiority Clinical Trials. When final, this guidance will reflect FDA s current thinking on this topic. 25

91 Contains Nonbinding Recommendations Draft Not for Implementation Calculate the study sample size, considering assumed error rates and d. Observe error rates for the critical task(s) during the experiment. Perform the statistical hypothesis test: o H 0 : ER IP - ER RP >d 978 o H A : ER IP - ER RP d Rejecting the null hypothesis (H 0 ) in favor of the alternative hypothesis (H A ) supports the claim of NI as defined by d. Typically, the acceptable Type I error probability (α) will be set at 5%. The NI test may be performed by comparing the upper bound of the appropriate confidence interval level for the difference in event rates to d. If the upper bound is less than d, NI is demonstrated. Paired designs and parallel designs are appropriate approaches to the NI studies discussed in this appendix. A paired design in which each end user uses both presentations and acts as his or her own control will generally be applicable and more efficient with respect to resources than a parallel design. Parallel group designs in which end users are randomized to groups using one or the other presentation are also viable in situations where paired designs are not possible. Sponsors are advised to propose and discuss study designs with FDA before initiating studies. 2. Sample Size Considerations Sample sizes for a comparative use human factors study should be adequate to support a demonstration of interchangeability. Sample sizes needed to adequately compare the presentations of the reference product and the proposed interchangeable product may be larger than those described generally in the human factors study literature. In general, small sample sizes are likely to be inadequate in this context because the goals of the comparative use human factors studies to support a demonstration of interchangeability may be different than the goals of typical human factors/usability studies discussed in the literature and certain FDA guidances. The literature on human factors studies holds a variety of opinions with respect to sample sizes. Some references dedicated to qualitative, non-comparative human factors studies suggest small samples sizes (5 to 25 participants), while other studies suggest that greater numbers of test subjects should be included. 41,42,43 41 Faulkner, L, 2003, Beyond the five-user assumption: Benefits of increased sample sizes in usability testing, Behavior Research Methods, Instruments, and Computers, 35(3), Nielsen, J, 2000, Why You Only Need to Test With 5 Users, Jakob Nielsen s Alertbox, Retrieved October 13, 2015, from 43 Spool, J, Schroeder, W, 2001, Testing web sites: Five users is nowhere near enough, in CHI 2001 Extended Abstracts (p ), New York: ACM Press. 26

92 Contains Nonbinding Recommendations Draft Not for Implementation The comparative use human factors studies described in this appendix are intended to ensure that design differences other than minor design differences found in the threshold analyses (described in section VIII.B.1 of the guidance) do not preclude a demonstration of interchangeability under section 351(k) of the PHS Act. Thus, as a scientific matter, a larger sample size may be necessary. Consider, for example, a failure in a critical task that can result in under-dosing that has a negative impact on a patient. If a study of 50 users showed no such failures, the upper bound of the 90% confidence interval for the error rate would be This means that an error rate of approximately 6% at a reasonable confidence level could not be ruled out. If 50 subjects were asked to operate two delivery device constituent parts, one being the reference product and another being the proposed interchangeable product, and no subject failed using either delivery device constituent part, the 90% confidence interval for the difference in error rates would be (-0.051, 0.051). Although the observed difference in error rates is zero, such a demonstration does not rule out a 5% difference in error rates at the 90% confidence level. Putting these numbers into the context of the under-dosing example, the question is whether it is acceptable to expect up to 58 out of 1,000 users to be under-dosed. The risk associated with the specific error in question will determine the acceptability of error rates. In some cases, a 6% error rate or a 5% difference in error rates would be untenable, even though there could be other contexts in which such rates or differences would be acceptable. This example illustrates the importance of the risk analysis portion of the comparative use human factors study design, as well as the importance of properly sizing the study. If extremely small error rates are expected, an adaptive design may be used to minimize sample size, while allowing for an adequate sample size if error rates are higher than initially assumed. Group sequential designs that are designed to stop early or a study design that adapts sample size may be considered. Consult the appropriate FDA guidance documents for detailed advice on designing studies using adaptive designs For detailed advice on designing studies using adaptive designs, see the draft guidances for industry Adaptive Designs for Medical Device Clinical Studies and Adaptive Design Clinical Trials for Drugs and Biologics. When final, these guidances will reflect FDA s current thinking on this topic. 27

93 Non-Inferiority Clinical Trials to Establish Effectiveness Guidance for Industry U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER) Center for Biologics Evaluation and Research (CBER) November 2016 Clinical/Medical

94 Non-Inferiority Clinical Trials to Establish Effectiveness Guidance for Industry Additional copies are available from: Office of Communications, Division of Drug Information Center for Drug Evaluation and Research Food and Drug Administration New Hampshire Ave., Hillandale Bldg., 4 th Floor Silver Spring, MD Phone: or ; Fax: druginfo@fda.hhs.gov and/or Office of Communication, Outreach and Development Center for Biologics Evaluation and Research Food and Drug Administration New Hampshire Ave., Bldg. 71, Room 3128 Silver Spring, MD Phone: or ocod@fda.hhs.gov U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER) Center for Biologics Evaluation and Research (CBER) November 2016 Clinical/Medical

95 Contains Nonbinding Recommendations TABLE OF CONTENTS I. INTRODUCTION... 1 II. BACKGROUND... 2 III. GENERAL CONSIDERATIONS FOR NON-INFERIORITY STUDIES... 3 IV. A. The Non-Inferiority Hypothesis... 3 B. Reasons for Using a Non-Inferiority Design... 7 C. The Non-Inferiority Margin... 8 D. Assay Sensitivity E. Statistical Inference F. Regulatory Conclusions G. Alternative Designs H. Number of Studies Needed I. Choice of Active Control CHOOSING THE NON-INFERIORITY MARGIN AND TESTING THE NON- INFERIORITY HYPOTHESIS A. Introduction B. Statistical Uncertainties and Quantification of the Active Control Effect C. Fixed Margin and Synthesis Methods D. Considerations for Selecting the Clinical Margin (M 2 ) E. Estimating the Sample Size F. Study Quality and Choice of Analysis Population G. Testing Non-Inferiority and Superiority in a Single Trial V. FREQUENTLY ASKED QUESTIONS AND GENERAL GUIDANCE APPENDIX EXAMPLES REFERENCES FOR EXAMPLES REFERENCES... 51

96 Contains Nonbinding Recommendations Non-Inferiority Clinical Trials to Establish Effectiveness Guidance for Industry 1 This guidance represents the current thinking of the Food and Drug Administration (FDA or Agency) on this topic. It does not establish any rights for any person and is not binding on FDA or the public. You can use an alternative approach if it satisfies the requirements of the applicable statutes and regulations. To discuss an alternative approach, contact the FDA office responsible for this guidance as listed on the title page. I. INTRODUCTION This document provides guidance to sponsors and applicants submitting investigational drug applications (INDs), new drug applications (NDAs), biologics licensing applications (BLAs), or supplemental applications on the appropriate use of non-inferiority (NI) study designs to provide evidence of the effectiveness of a drug or biologic, usually because a superiority study design (drug versus placebo, dose response, or superiority to an active drug) cannot be used. 2 The guidance gives advice on when NI studies intended to demonstrate effectiveness of an investigational drug can provide interpretable results, how to choose the NI margin, and how to test the NI hypothesis. This guidance does not provide recommendations for the use of NI study designs to evaluate the safety of a drug. In general, FDA s guidance documents do not establish legally enforceable responsibilities. Instead, guidance documents describe the Agency s current thinking on a topic and should be viewed only as recommendations, unless specific regulatory or statutory requirements are cited. The use of the word should in FDA guidance documents means suggested or recommended, but not required. This guidance finalizes the draft guidance for industry, Non-Inferiority Clinical Trials, published in In addition, it supersedes the guidance for industry, Antibacterial Drug Products: Use of Noninferiority Trials to Support Approval, also published in 2010, which will be withdrawn. 1 This guidance has been prepared by the Office of Biostatistics and the Office of New Drugs in the Center for Drug Evaluation and Research (CDER) and the Center for Biologics Evaluation and Research (CBER) at the Food and Drug Administration. 2 For the purposes of this guidance, all references to drugs include both human drugs and therapeutic biologic products unless otherwise specified. While most concepts discussed will be broadly applicable, certain issues related to vaccines, such as the choice of the NI margin when the study endpoint is the level of antibodies, would call for consultation from CBER. 1

97 II. BACKGROUND FDA s regulations on adequate and well-controlled studies (21 CFR ) describe four kinds of concurrently controlled trials that provide evidence of effectiveness. Three of them placebo, no treatment, and dose-response controlled trials are superiority trials that seek to show that a test drug is superior to the control (placebo, no treatment, or a lower dose of the test drug). The fourth kind, comparison with an active treatment (active control), can also be a superiority trial, if the intent is to show that the new drug is more effective than the control. More commonly, however, the goal of such studies is to show that the difference between the new and active control treatment is small small enough to allow the known effectiveness of the active control, based on its performance in past studies and the assumed effectiveness of the active control in the current study, to support the conclusion that the new test drug is also effective. How to design and interpret the results of such studies so that they can support a conclusion about effectiveness of the new drug is challenging. Active controlled trials that are not intended to show superiority of the test drug but rather to show that the new treatment is not inferior to an unacceptable extent were once called clinical equivalence trials. The intent of an NI trial, however, is not to show that the new drug is equivalent, but rather that it is not materially worse than the control. Therefore, the interest is one-sided. The new drug could be better than the control, and therefore at a minimum noninferior, but it would not be equivalent. The critical difference between superiority and NI trials is that a properly designed and conducted superiority trial, if successful in showing a difference, is entirely interpretable without further assumptions (other than lack of bias) that is, the result speaks for itself and requires no extra-study information. In contrast, the NI study is dependent on knowing something that is not measured in the study, namely, that the active control had its expected effect in the NI study. When this occurs, the trial is said to have assay sensitivity, defined as the ability to have shown a difference from placebo of a specified size. A successful NI trial, one that shows what appears to be an acceptably small difference between treatments, may or may not have had assay sensitivity and therefore may or may not support a conclusion that the test drug was effective. Thus, if the active control had no effect at all in the NI trial (i.e., did not have any of its expected effect), then even ruling out a very small difference between control and test drug is meaningless and provides no evidence that the test drug is effective. (See Section III.D. for further discussion on assay sensitivity.) In the absence of a placebo arm, knowing whether the trial had assay sensitivity relies heavily on external (not within-study) information, giving NI studies some of the characteristics of a historically controlled trial. FDA regulations have recognized since 1985 the critical need to know, for an NI trial to be interpretable, that the active control had its expected effect in the trial. Thus, 21 CFR (a)(2)(iv), unchanged since 1985, says: If the intent of the trial is to show similarity of the test and control drugs, the report of the study should assess the ability of the study to have detected a difference between treatments. Similarity of test drug and active control can mean either that both drugs were effective or that neither was effective. The analysis of the study should explain why the 2

98 drugs should be considered effective in the study, for example, by reference to results in previous placebo-controlled studies of the active control drug. This guidance consists of four parts: A general discussion of regulatory, study design, scientific, and statistical issues associated with the use of NI studies to establish the effectiveness of a new drug A focus on some of these issues in more detail, notably the statistical approaches used to determine the NI margin and to test for non-inferiority Commonly asked questions about NI studies Four examples of successful and unsuccessful efforts to define NI margins and test for non-inferiority 3 III. GENERAL CONSIDERATIONS FOR NON-INFERIORITY STUDIES A. The Non-Inferiority Hypothesis In a placebo-controlled trial, the null hypothesis (H o ) is that the beneficial response to the test drug (T) is less than or equal to the response to the placebo (P); the alternative hypothesis (H a ) is that the response to the test drug is greater than P. Thus: H o : T P; T P 0 H a : T > P; T P > 0 In most cases, the test for a treatment effect corresponds to showing that the lower bound of the two-sided 95% confidence interval (equivalent to the lower bound of a one-sided 97.5% confidence interval) for T-P is > 0. This result shows that the effect of the test drug is greater than 0 (see Figure 1). 3 References: In this guidance, references to methods or studies are not included in the text; rather they are included in the general references section in the Appendix and are grouped by topic. A separate references section is also provided for the examples in the Appendix. 3

99 Figure 1. Possible Results of a Placebo-Controlled Superiority Study (Point Estimate and 95% Confidence Interval (CI)) Test Placebo (T-P) 1. Point estimate of effect is 2; 95% CI lower bound is 1. Conclusion: Drug is effective and has an effect of at least Point estimate of effect is 2; 95% CI lower bound is <0. Conclusion: Drug is not shown to be effective. 3. Point estimate of effect is 0; 95% CI lower bound is well below 0. Conclusion: Drug is not shown to be effective. Although there is no difference in the conclusions of scenarios 2 and 3, the magnitude of the treatment difference and width of the confidence interval in scenario 2 may encourage the conduct of another study (possibly larger) before deciding that the test drug has no effect. In an NI study, the goal is to demonstrate that the test drug has an effect by showing that its effect is sufficiently close to the effect of an active control. There is no placebo arm in the study; therefore, the effect of the active control is not measured in the study but must be assumed. The goal of the study is to show that the effect of the test drug (T) is not inferior to the effect of the active control (C) by a specified amount, called the NI margin, or M. The null and alternative hypotheses correspond to a null hypothesis of inferiority and an alternative hypothesis of non-inferiority, as follows: H o : C T M (T is inferior to the control (C) by M or more) H a : C T < M (T is inferior to the control (C) by less than M) 4

100 A statistical test of the above hypothesis is provided by comparing the upper bound of the twosided confidence interval for C-T with the NI margin, M, which is specified in advance. If the upper bound is less than M, non-inferiority of T relative to C is established. One choice for M (the largest possible value) is to set it equal to the entire known effect of the active control relative to placebo, based on past randomized controlled trials. With this choice for M, called M 1, and assuming the control drug attains this level of efficacy in the NI study, a finding of non-inferiority means that the test drug has an effect greater than 0 (see Figure 2). A more usual choice is to set M equal to some clinically relevant portion of M 1, namely, the portion of the control drug effect it is important to preserve with the test drug, based on clinical judgment. 5

101 Figure 2. Possible Results of an NI Study Showing Control Drug Test Drug Differences (Point Estimate and 95% CI) Control Test (C-T) 1. Point estimate of C-T is 0, suggesting equal effect of C and T; upper bound of the 95% CI for C-T is 1, well below M 1 ; NI is demonstrated. 2. Point estimate of C-T favors C; the upper bound of the 95% CI for C-T is >2, above M 1 ; NI is not demonstrated. 3. Point estimate of C-T is zero, which suggests an equal effect; but the upper bound of the 95% CI for C-T is >2 (i.e., above M 1 ), so that NI is not demonstrated. 4. Point estimate favors T; NI is demonstrated, but superiority is not demonstrated. 5. Point estimate favors T; superiority and NI are demonstrated. 6. Point estimate of C-T is 1, favoring the control. The upper bound of the 95% CI for C-T is <M 1, demonstrating NI (the entire effect of C has not been lost) but at the same time the 95% CI for C-T is above zero, indicating that T is actually inferior to C, even while meeting the NI standard. The determination of M 1 is a critical step in designing an NI trial and is often difficult; this determination is therefore a major focus of this guidance. M 1 cannot be directly measured in the NI study, because there is no concurrent placebo group. It must be estimated based on the past performance of the active control, preferably in placebo-controlled trials, and then assumed to hold for the NI study based on a comparison of prior test conditions to the current test environment (see section III.D). 6

102 The choice of the margin M 1 has important practical consequences. The smaller the margin, the lower the upper bound of the 95% two-sided confidence interval for C-T must be, and the larger the sample size needed to establish non-inferiority. Showing that the upper bound of the 95% CI of C-T is less than M 1 demonstrates that the test drug has some effect (i.e., an effect > 0). The margin of interest, however, as noted above, is usually smaller than M 1 (to show that an adequate portion of the clinical benefit of the control is preserved), in which case it is called M 2. The basis for this expectation is described below in section III.C.4. B. Reasons for Using a Non-Inferiority Design The usual reason for using an NI active control study design instead of a superiority design is an ethical one. Specifically, this design is chosen when it would not be ethical to use a placebo, or a no-treatment control, or a very low dose of an active drug, because there is an effective treatment that provides an important benefit (e.g., life-saving or preventing irreversible injury) available to patients for the condition to be studied in the trial. Whether a placebo control can be used depends on the nature of the benefits provided by available therapy. The International Conference on Harmonisation guidance E10: Choice of Control Group and Related Issues in Clinical Trials (ICH E10) states: In cases where an available treatment is known to prevent serious harm, such as death or irreversible morbidity in the study population, it is generally inappropriate to use a placebo control. There are occasional exemptions, however, such as cases in which standard therapy has toxicity so severe that many patients have refused to receive it. In other situations, where there is no serious harm, it is generally considered ethical to ask patients to participate in a placebo-controlled trial, even if they may experience discomfort as a result, provided the setting is non-coercive and patients are fully informed about available therapies and the consequences of delaying treatment [ICH E10; pps.13-14]. Aside from this ethical reason, there may be other reasons to include an active control, possibly in conjunction with a placebo control, either to compare treatments or to assess assay sensitivity (see section III.D). Caregivers, third party payers, and some regulatory authorities have increasingly placed an emphasis on the comparative effectiveness of treatments, leading to more studies that compare two treatments. Such studies can provide information about the clinical basis for comparative effectiveness claims, which may be helpful in assessing cost effectiveness of treatments. If a placebo group is included in addition to the active comparator, it becomes possible to judge whether the study could have distinguished treatments that differed substantially, e.g., active drug versus placebo. Such comparative effectiveness studies must be distinguished from NI studies, which are the main focus of this document. The word noninferior is used here in a special sense. The methods described in this document are intended to show that a new treatment that demonstrates non-inferiority is effective, not that it is as effective as the active comparator. A new treatment may meet the standard of effectiveness (that it is superior to placebo) without justifying a conclusion that it is as effective or even nearly as effective as the active comparator. 7

103 C. The Non-Inferiority Margin As described above, the NI study seeks to show that the amount by which the test drug (T) is inferior to the active control (C), C-T, is less than some prespecified NI margin (M). M can be no larger than the presumed entire effect of the active control in the NI study, and the margin based on the entire active control effect is generally referred to as M 1. It is critical to recognize that M 1 is not measured in the NI trial (in the absence of a placebo arm), but rather is estimated based on past performance of the active control. The effect is assumed to be present in the current study based on a thorough comparison of the characteristics of the current NI study with those of prior studies and assessment of the quality of the NI study. The validity of any conclusion from the NI study depends on the choice of M 1 and its relevance to the current NI study. If, for example, the NI margin is chosen as 10, and the study does indeed rule out a difference of 10 (but not a smaller difference), seeming to demonstrate effectiveness of T, but the true effect of C in this study was actually less than 10, then a conclusion that the study demonstrated non-inferiority would have been incorrect. The choice of M 1, together with reasonable assurance that this effect occurred in the trial (i.e., the presence of assay sensitivity), is thus critical to obtaining a meaningful, correct answer in an NI study. Because assay sensitivity can never be proven in the absence of a placebo group, historical evidence of assay sensitivity is essential, but measures should be taken to make it likely that the active control will have the presumed effect in the NI study (see section III.A.5). This, together with careful selection of the NI margin, will increase the likelihood of valid and interpretable results. Because the consequence of choosing a margin greater than the actual treatment effect of the active control in the study may be a false conclusion that a new drug is effective, an undesirable public health outcome, there is a powerful tendency to be conservative both in the choice of margin and in the statistical analysis used to rule out a degree of inferiority of the test drug to the active control of more than that margin. This is generally done by ensuring that the upper bound of the 95% two-sided confidence interval for C-T in the NI trial is smaller than M 1. Use of this interval corresponds to a one-sided test size (alpha level) of in testing the NI hypothesis stated above. The upper bound of the confidence interval for C-T is not, however, the only measurement of interest, just as the lower bound of a 95% confidence interval for the effect of drug versus placebo in a superiority trial is not the only value of relevance. The point estimate of the treatment effect and the width of its confidence interval are also relevant. Nonetheless, the upper bound of the 95% confidence interval is typically used to judge the effectiveness of the test drug in the NI study, just as a two-sided test size (alpha level) of 0.05 is traditionally the standard used for defining success in a superiority trial. The 95% confidence interval upper bound for C- T is used, in conjunction with M 1, to provide a reasonably high level of assurance that the test drug does, in fact, have an effect greater than zero (i.e., ruling out loss of all the effect of the active control). Although the NI margin used in a trial can be no larger than the entire assumed effect of the active control in the NI study (M 1 ), it is usual and generally desirable to choose a smaller value, called M 2, for the NI margin. Showing non-inferiority to M 1 would provide assurance that the test drug had an effect greater than zero, but in many cases that would not be sufficient to conclude that the test drug had a clinically acceptable effect. Recall that the main reason an NI study is conducted is that it is not ethical to include a placebo arm. The active control has a 8

104 beneficial effect, and denying that benefit to subjects with a serious illness is not ethical. For the same reason, it would not usually be acceptable for a test drug to lose most of that active control s effect. It is therefore usual in NI studies to choose a smaller margin (M 2 ) that reflects the largest loss of effect that would be clinically acceptable. This can be described as an absolute difference in effect (typical of antibiotic trials) or as a fraction of the risk reduction provided by the control (typical in cardiovascular outcome trials). Note that a larger value for M 2 may be justified clinically, if the test drug were shown to have some important advantage (e.g., on safety or on a secondary endpoint). Concern has been expressed that use of M 2 represents an FDA comparative effectiveness standard that is not included in the Federal Food, Drug, and Cosmetic Act. In explaining the role of relative effectiveness under law in April 1995, President Clinton and Vice President Gore ( Reinventing Regulation of Drugs and Medical Devices, part of the National Performance Review) stated the following: In certain circumstances, however, it may be important to consider whether a new product is less effective than available alternative therapies, when less effectiveness could present a danger to the patient or to the public. For example, it is essential for public health protection that a new therapy be as effective as alternatives that are already approved for marketing when: 1. the disease to be treated is life-threatening or capable of causing irreversible morbidity (e.g., stroke or heart attack); or 2. the disease to be treated is a contagious illness that poses serious consequences to the health of others (e.g., sexually transmitted disease). The reinvention statement was placed in the Federal Register of August 1, 1995 (60 FR at 39181), as an FDA position by FDA s Deputy Commissioner for Policy, William Schultz. The definitions used to describe these two versions of M are: M 1 = the entire effect of the active control assumed to be present in the NI study M 2 = the largest clinically acceptable difference (degree of inferiority) of the test drug compared to the active control M 1 is estimated based on the historical experience with the active control drug. Its relevance to the current NI trial is based on: 1. Assessment of the likelihood that the current effect of the active control is similar to that observed in the past studies used to estimate the active control effect (the constancy assumption) 2. Assessment of the quality of the NI trial, particularly looking for deficiencies in study design and/or conduct that could reduce a difference between the active control and the new drug Note that in a trial seeking to show a difference (i.e., superiority), this diminution of the betweentreatment difference in the second element is a bias toward the null, but in a non-inferiority 9

105 trial, it is a bias toward the alternative (i.e., non-inferiority). Because of this second element, in some situations M 1, although prespecified, has to be discounted (i.e., a smaller value would be used), but the amount of discounting needed may not be known until the NI study has been completed. The choice of M 2 is a matter of clinical judgment, but M 2 can never be greater than M 1, even for a situation in which the active control drug has a small effect, and clinical judgment might argue that a larger difference is not clinically important. Even if that clinical judgment were reasonable, choosing an M 2 greater than M 1 as the NI margin would not allow a conclusion that the test drug has any effect. As explained above, ruling out a difference between the active control and test drug larger than M 1 is the critical finding that supports a conclusion of effectiveness. This analysis is approached with great rigor; that is, a difference (C-T) larger than M 1 needs to be ruled out with a high degree of statistical assurance. As M 2 represents a clinical judgment, there may be a greater flexibility in interpreting a 95% upper bound for C-T that is slightly greater than M 2, as long as the upper bound is still well less than M 1 (see Figure 3). 10

106 Figure 3. Possible Results of an NI Study Showing Active Control Test Drug Differences (Point Estimate and 95% CI) Control Test (C-T); (degree of inferiority of test drug) 1. C-T point estimate = 0 and upper bound of 95% CI < M 2, indicating test drug is effective and adequately rules out an unacceptable loss of the control effect (NI demonstrated). 2. Point estimate of C-T favors C; upper bound of 95% CI < M 1 but > M 2, indicating test drug effect > 0 but an unacceptable loss of the control effect has not been ruled out. 3. Point estimate of C-T is zero and upper bound of 95% CI < M 1 but it is slightly greater than M 2. Loss of the pre-specified M 2 has thus not been ruled out, but whether the study has shown adequate preservation of the control effect would be a matter of clinical judgment. 4. C-T point estimate favors C and upper bound of 95% CI > M 1, indicating the study does not provide evidence of effectiveness for test drug. D. Assay Sensitivity Assay sensitivity is an essential property of an NI clinical trial. Assay sensitivity is the ability of the trial to have detected a difference between treatments of a specified size. Stated in another way, assay sensitivity means that had the study included a placebo, a control drug-placebo difference of at least M 1 would have been present. 11

107 As explained above, the choice of M 1, and the conclusion that a trial has assay sensitivity (i.e., the active control would have had an effect of at least M 1 ), is based on three considerations: 1. Historical evidence of sensitivity to drug effects 2. The similarity of the new NI trial to the historical trials (the constancy assumption) 3. The quality of the new trial (ruling out defects that would tend to minimize differences between treatments) 1. Historical Evidence of Sensitivity to Drug Effects (HESDE) HESDE means that prior studies, which were appropriately designed and conducted trials in the past and that used a specific active treatment (generally the one that is to be used in the new NI study or, in some cases, one or more pharmacologically closely related drugs), regularly showed this treatment to be superior to placebo (or some other treatment). Consistent findings in past studies allow for a reliable estimate of the drug s effect compared to placebo. The estimate of the size of the effect must take the variability of past results into account; one should not presume that the largest effect seen in any trial, or even the point estimate of a meta-analysis of all relevant trials, is certain to be repeated. Analysis of historical data will be discussed further in section IV. HESDE cannot be determined for many symptomatic treatments (e.g., treatments for depression, anxiety, insomnia, angina, symptomatic heart failure, symptoms of irritable bowel disease, and pain), as even well-designed and conducted studies for such indications often fail to distinguish an effective drug from placebo. In those cases, it cannot be assumed that an active control would have shown superiority to a placebo (had there been one) in any given NI study, and NI studies of drugs for these indications may therefore be non-informative. These issues can also negatively affect the assessment of effectiveness in outcome studies. For example, in the case of aspirin, the largest placebo-controlled trial (AMIS, the Aspirin Myocardial Infarction Study; see Example 2) did not show an effect of aspirin even though other trials all favored aspirin. Similarly, of more than 30 post-infarction beta-blocker trials, only a small number showed significantly improved survival or other cardiovascular benefit. 2. Similarity of the Current NI Trial to the Historical Studies and Its Relationship to the Constancy Assumption The conclusion that HESDE can be used to estimate the effect of the active control, which will then serve as the basis for choosing M 1 for the new NI study, can be reached only when it is appropriate to conclude that the NI study is sufficiently similar to the past studies with respect to all important study design and conduct features that might influence the active control effect. This conclusion is referred to as the constancy assumption. The design features of interest include: The characteristics of the patient population Important concomitant treatments Definitions and ascertainment of study endpoints Dose of active control 12

108 Entry criteria Analytic approaches For example, the effect of an angiotensin-converting enzyme (ACE) inhibitor on heart failure mortality has repeatedly been shown in studies where the drugs were added to diuretics and (frequently) digoxin, establishing HESDE, but evolution in treatment since those studies were conducted raises questions about our understanding of the present-day effect of ACE inhibitors. Since the time of those studies, other medications (beta blockers, spironolactone) have come into standard use. We do not know whether the past effect would still be present when ACE inhibitors are added to a regimen including drugs from these two classes. Similarly, the effect of a thrombolytic on cardiovascular mortality could depend on how soon after symptoms the drug was given, concomitant use of anticoagulants and platelet inhibitors, and use of lipid-lowering drugs. To provide a sound basis for choosing M 1, the historical studies and the new NI study should be as nearly identical as possible in all important respects. Providing reasonable assurance that endpoints in the historical trial will be similar to, and will have been evaluated similarly to, endpoints in the new trial is easier when the endpoints are standardized and objective. The effect of the active control could be on a single endpoint (e.g., mortality) or on a composite (e.g., death, heart attack, and stroke), but, again, it is critical that measurement and assessment of these be reasonably consistent over time. The endpoint used in the NI study need not necessarily be the one used in the original trials of the active control if data from the historical studies are available to estimate the size of the effect of the active control on the new endpoint used in the NI study. For example, even if the historical studies used a mortality endpoint, the studies could be used, if high quality data could be obtained, to calculate the magnitude of an effect for an endpoint of death plus hospitalization, as long as it was possible to be confident that the circumstances leading to the hospitalization were similar in the historical studies and the NI study. Note, however, that it would not be acceptable to search through a range of endpoints to find the largest historical effect, as this could represent an overestimate of the effect to be expected in the NI study. In general, where there has been substantial evolution over time in disease definition and treatment, or where the methodology used in the historical trials has become outdated, the assumption of constancy may not be supported, and, therefore, the use of an NI design may not be justified. Although an NI study can be designed to be similar in most aspects to the historical studies, it may not be possible to assess that similarity fully until the NI study is completed and various characteristics of the study population and response are evaluated. When there is known heterogeneity of the active control treatment effect related to patient characteristics (e.g., age, gender, severity) and that heterogeneity can be quantified, it may be necessary to adjust the estimate of the size of the active control effect in the NI study if the mix of patient characteristics in the historical studies and the NI study differ substantially. The property of constancy of the treatment effect may depend on which metric is chosen to represent the treatment effect. This issue is discussed in more depth in section IV.B.2.c. Experience suggests that when background rates of outcome events differ among study 13

109 populations, metrics like hazard ratio or relative risk may be more stable than metrics like absolute risk difference, which are more sensitive to changes in event rates in the population. 3. Good Study Quality Conducting any poor quality studies should always be avoided, but with NI studies, sloppiness in study design/conduct is particularly problematic, because it introduces bias towards the alternative hypothesis of non-inferiority (see ICH E10; pp and section IV.F for further discussion). Deficiencies such as imprecise or poorly implemented entry criteria, poor compliance, use of concomitant treatments whose effects may overlap with the drugs under study, inadequate measurement techniques, errors in delivering assigned treatments, high attrition, or poor follow-up may reduce the difference C-T observed in the study, potentially leading to a false conclusion of non-inferiority. It should also be appreciated that intent-to-treat approaches, which preserve the principle that all patients are analyzed according to the treatment to which they have been randomized even if they do not receive it, although conservative in superiority trials, are not necessarily conservative in an NI study and can lead to an incorrect finding of non-inferiority. In a superiority trial, sloppiness can lead to study failure. In contrast, poor quality in an NI trial can sometimes lead to an apparent finding of non-inferiority that is incorrect. Therefore, particular attention to quality is critical when planning and conducting an NI study. Adjustment for poor quality after the fact is usually not possible. E. Statistical Inference The various approaches to calculating the NI margin and analyzing an NI study will be discussed in detail in section IV. A commonly used fixed margin method is generally referred to as the 95%-95% method. The first 95% refers to the confidence interval of the estimated effect of the control based on the historical studies demonstrating the effect, and the second 95% refers to the confidence interval used to test the null hypothesis in the NI study. Note that the first 95% lower confidence bound discussed here is a bound for the average effect of the active comparator in historical studies, or the true effect in the NI study if the constancy assumption holds. It is sometimes supposed that this is a lower bound for the actual effect of the comparator in the NI study, and the draft guidance (issued in 2010) suggests this, but this is not correct. The 95% lower bound does indeed define a lower bound for the true control effect, but the actual effect in the NI study would still be liable to sampling variation. To bound the actual effect, a different type of interval would be required; these intervals are known as prediction intervals and are much wider than the corresponding confidence intervals. Confidence intervals nevertheless suffice for the present purpose, because the confidence interval from the historical studies bounds the true effect of the comparator relative to placebo, and the confidence interval from the NI study bounds the true effect of the test drug relative to the comparator. Relying always on the constancy assumption, these can be combined to infer a positive effect of the test drug relative to placebo. The 95%-95% fixed margin approach (using in this case a risk difference rather than a risk ratio for the margin) can be illustrated by FDA s evaluation of a new thrombolytic product, reteplase, 14

110 for treatment of acute myocardial infarction (AMI). To calculate the NI margin, results of all available placebo-controlled trials of streptokinase, the active comparator (control) for the NI study, were pooled by means of an appropriate meta-analytic method. The pooled results provided a point estimate for the effect on survival of an absolute 2.6% difference in mortality rates, with a 95% lower bound of 2.1% (i.e., M 1 ). A clinical decision was made that any new thrombolytic should rule out a loss of more than half of the benefit of streptokinase to be regarded as an acceptable alternative. The NI study would therefore have had to rule out an absolute 1.05% increase in mortality rate (i.e., M 2 ) in the reteplase-treated patients compared to those treated with streptokinase. The NI analysis for this study was to show that the 95% confidence interval (one-sided for this particular case) of the difference in mortality rates excluded an increase of 1.05%. The INJECT study accomplished this, and the product was approved for marketing. An alternative to the fixed margin approach is known as the synthesis approach because it combines or synthesizes the data from the historical trials and the current NI trial, reflecting the variability in both data sources. Although the 95%-95% method was developed in a different way, it has been noted that it can be viewed as mathematically equivalent to the synthesis method but with the standard error of the effect of the test drug relative to placebo, estimated by SS H + SS N, where SE H and SE N are the standard errors from the historical studies and the NI study, respectively, rather than by SS H 2 + SS N 2, which is the standard error for the synthesis method. The first formula always gives a larger standard error than the second. Thus, compared to the synthesis method, the 95%-95% method is conservative. Both methods rely crucially on the assumption of constant effect of the control (on average) between the historical studies and the NI study. The use of the 95%-95% method might be seen as an allowance for possible deviation from this assumption, a desirable feature for evaluating whether loss of M 1 (the whole effect of the control) has been ruled out. Use of the synthesis method, along with a careful justification of the constancy assumption and, when appropriate, an explicit allowance for deviation from it, would also be acceptable and may be recommended for determining whether a loss of effect greater than M 2 has been ruled out (see Section IV.C and Example 1(B) for more detail). F. Regulatory Conclusions A successful NI study shows rigorously that the test drug has an effect greater than zero if it excludes an NI margin of M 1, as long as M 1 is well chosen and represents an effect that the control drug actually would have had (versus a placebo, had there been a placebo group). The NI study can also be used to show that the test drug had an effect greater than some fraction of the control drug effect, depending on the M 2 that is used. Note, however, that although a successful NI study supports effectiveness of the test drug, it will only rarely support a conclusion that the drug is equivalent or similar to the active control, a concept that has not 15

111 been well defined for these situations. It should be appreciated that in addition to the rigorous demonstration of effectiveness based on significance testing, the trial provides additional information, just as a placebo-controlled trial supporting the effectiveness of a drug does. The point estimate of the drug effect and its confidence interval provide information about how large the difference between the test and control drug is likely to be. As noted before, a successful NI study will usually not be sufficient to support a conclusion that the test drug is equivalent or similar to the active control. The methods discussed in this document, especially with regard to choosing the margin, are not intended to demonstrate equivalence or similarity but only to show that the test drug is effective. However, if the lower bound of the confidence interval for the effect of the test drug relative to the active control were only slightly negative, a judgment that similarity had been shown would be possible. If studies are intended to support equivalence, the margin against which the confidence interval is to be judged should be justified in advance, and the margin will usually be smaller than M 2. G. Alternative Designs ICH E10 identifies a wide variety of study designs that may be better than an NI design in situations where there is difficulty or uncertainty in setting the NI margin or where the NI margin needs to be so small that the NI study sample size becomes impossibly large. 1. Add-on Study In many cases, for a pharmacologically novel treatment, the most interesting question is not whether it is effective alone but whether the new drug can add to the effectiveness of treatments that are already available. The most pertinent study would therefore be a comparison of the new agent and placebo, each added to established therapy. Thus, new treatments for heart failure have added new agents (e.g., ACE inhibitors, beta blockers, and spironolactone) to diuretics and digoxin. As each new agent became established, it became part of the background therapy to which any new agent and placebo would be added. This approach is also typical in oncology, in the treatment of seizure disorders, and, in many cases, in the treatment of AIDS. 2. Identifying a Population Not Known to Benefit From Available Therapy in Which a Placebo-Controlled Trial Is Acceptable In many outcome study settings, effectiveness is established for some clinical settings (e.g., severe disease) but not others. Therefore, it may be possible to study less severely ill patients in placebo-controlled trials. The demonstration that simvastatin was effective in hypercholesterolemic post-infarction patients (4S), for example, did not preclude studies of statins in hypercholesterolemic non-infarction patients (WOSCOPS) or in patients with lesser degrees of hypercholesterolemia (TEXCAPS). This approach is appropriate as long as there is uncertainty as to whether the treatment is of value in the new study population. It may also be possible to study patients intolerant to the known effective therapy. For example, it was possible to study angiotensin receptor blockers in a placebo-controlled trial in heart failure patients intolerant of ACE inhibitors, but it would not have been possible to deny a more general 16

112 population of heart failure patients an ACE inhibitor, as it had already been established that ACE inhibitors improved survival in a general heart failure population. 3. Early Escape, Rescue Treatment, Randomized Withdrawal In symptomatic conditions, there may be reluctance to leave people on placebo for prolonged periods when effective therapy exists. It is possible to incorporate early escape/rescue provisions for patients who do not respond by a particular time, or to use a design that terminates patients on first recurrence of a symptom such as unstable angina, grand mal seizure, or paroxysmal supra-ventricular tachycardia. To evaluate the persistence of effects over time, where conducting a long-term placebo-controlled trial would be difficult, a randomized withdrawal study can be used. In these studies, patients successfully treated with a drug are randomly assigned to placebo or continued drug treatment. As soon as symptoms return, the patient is considered to have had an endpoint. H. Number of Studies Needed Ordinarily, with exceptions allowed by the FDA Modernization Act of 1997 (the Modernization Act), FDA expects that there will be more than one adequate and well-controlled study supporting effectiveness. The Modernization Act allows one study plus confirmatory evidence to serve as substantial evidence in some cases, and FDA has issued a guidance that discusses when a single study might be sufficient. 2 Where there is uncertainty about the magnitude of the historical treatment effect (and thus M 1 ) because of variability or reliance on a single historical study, more than one NI study is usually needed to support effectiveness. Where the studies are of relatively modest size, there is usually no impediment to conducting more than one NI trial if that appears necessary. Conducting two trials that are very large (to have adequate statistical power), however, may be infeasible, and it is worth considering what might make a single trial persuasive. Generally, two considerations might do so: (1) availability of other relevant information and (2) a statistically persuasive result. 1. Other Relevant Information In NI trials, the test drug is generally pharmacologically similar to the active control (i.e., if they were not pharmacologically similar, an add-on study would usually have been more persuasive and more practical). In these cases, the expectation of similar performance (but still requiring confirmation in a trial) might make it possible to accept a single trial and perhaps could also 2 See the guidance for industry Providing Clinical Evidence of Effectiveness for Human Drug and Biological Products (Providing Clinical Evidence of Effectiveness guidance), available on the FDA Drug Web page at We update guidances periodically. To make sure you have the most recent version of a guidance, check the FDA Drug Web page. 17

113 allow less conservative choices in choosing the NI margin. A similar conclusion might be reached when other types of data are available, for example: If there were a very persuasive biomarker confirming similar activity of the test drug and active control (e.g., tumor response, ACE inhibition, or extent of beta blockade) If the drug has been shown to be effective in closely related clinical settings (e.g., effective as adjunctive therapy with an NI study of monotherapy) If the drug has been shown to be effective in distinct but related populations (e.g., adult versus pediatric) 2. Statistically Persuasive Result A conclusion that an NI trial can be considered statistically persuasive may be based on the internal consistency of the NI finding or on the margin that is ruled out with a two-sided 95% confidence interval. It is important to recognize that there are two margins of interest, M 1 and M 2. In an NI study, the clinically determined margin M 2 is smaller, often considerably smaller, than M 1, which is used to determine whether the test drug has any effect. For example, M 2 might be chosen to be 40% of M 1. By meeting this M 2 criterion, ruling out a loss of 40% of the effect of the control, a single NI study provides reasonable assurance that the test drug preserves a clinically sufficient fraction (at least 60%) of the effect of the control treatment. At the same time, it provides strong assurance that the test drug has an effect greater than zero. In such cases, a single trial would usually be a sufficient basis for approval. Where the preservation of effect is particularly critical, of course, demonstrating that loss of M 2 has been ruled out in more than one study might be considered necessary. In some cases, a study planned as an NI study may show superiority to the active control. Recommendations in International Conference on Harmonisation guidance E9: Statistical Principles for Clinical Trials (ICH E9) and FDA policy have been that this superiority finding arising in an NI study can be interpreted without adjustment for multiplicity. Showing superiority to an active control is a very persuasive demonstration of the effectiveness of the test drug, because demonstrating superiority to an active drug is much more difficult than demonstrating superiority to placebo. Similarly, a finding of less than superiority, but with a 95% confidence interval upper bound for C-T considerably smaller than M 2, is also statistically persuasive for demonstrating effectiveness. I. Choice of Active Control The active control must be a drug whose effect is well-defined. The most obvious choice is a single drug for which historical placebo-controlled trials are available to define the active control effect assumed for the NI trial. When there are several pharmacologically similar drugs, it may be possible to use a combined analysis to choose an NI margin, and in that case, any one of the related drugs may be considered as the active control in the NI trial (see section IV.B.2.b). 18

114 IV. CHOOSING THE NON-INFERIORITY MARGIN AND TESTING THE NON- INFERIORITY HYPOTHESIS A. Introduction In this section, various sources of uncertainty that come into play for NI trials are discussed, with particular emphasis on those affecting the choice of margin. We also discuss in more detail the choice of statistical test for the NI hypothesis. As described briefly in section III, there are two different approaches to analysis of the NI study, one called the fixed margin method and the other called the synthesis method. Both use the same data from the historical studies and NI study, but in different ways. With the fixed margin method, the margin M 1 is based upon estimates of the effect of the active comparator in previously conducted studies. The NI margin to be ruled out in the NI study is then prespecified, and it is usually chosen as a quantity smaller than M 1 (i.e., M 2 ) to ensure that a reasonable fraction of the effect of the control is preserved. The NI study is successful if the results rule out with a sufficient level of confidence (e.g., 97.5%) inferiority of the test drug to the control by an amount equal to the NI margin or more. It is referred to as a fixed margin method because the past studies comparing the control drug with placebo are used to derive a single fixed value for M 1. The value typically chosen is the lower bound of the 95% confidence interval about the treatment effect of a single placebo-controlled trial or metaanalysis of such trials and represents a conservative estimate of the effect the active control drug is expected to have in the NI study. If M 1 is ruled out by the 95% confidence interval upper bound for C-T, then the conclusion is made that the test drug has an effect. If M 2 is ruled out, then the effect of the test drug has been shown to preserve a clinically important fraction of the effect of the control drug. The synthesis method, derived from the same data, combines (or synthesizes) the estimate of treatment effect relative to the control from the NI trial with the estimate of the control effect from a meta-analysis of historical trials. This method treats both sources of data as if they came from the same randomized trial, to project what the placebo effect would have been had the placebo been present in the NI trial. The process makes use of the variability from both the NI trial and the historical trials and yields one confidence interval for testing the NI hypothesis that the treatment rules out loss of a prespecified fixed fraction of the control effect without actually specifying that control effect or a specific fixed NI margin based on the control effect. In other words, the fraction M 2 /M 1 of the control effect that is to be preserved in evaluating NI is specified in advance, but neither quantity (M 1 or M 2 ) is specified. The multiple steps involved and assumptions made in evaluating the NI hypothesis using either the fixed-margin or synthesis approach are all potential sources of uncertainty that may be introduced into the results and conclusions of an NI study. In section IV.B, we attempt to identify these sources and suggest approaches to accounting for uncertainty to reduce the possibility of drawing false conclusions from an NI study. Section IV.C provides further discussion of the statistical analysis methods for evaluating non-inferiority. 19

115 B. Statistical Uncertainties and Quantification of the Active Control Effect (M 1 ) 1. What Are the Sources of Uncertainty in an NI Study? The interpretation of an NI study depends on three linked critical conclusions: 1. That there is reliable information about the effect the active control drug had in past studies 2. That there is reason to believe the effect the active control drug has in the current NI study is similar to the effect observed in past studies 3. That the NI study provides reliable information about the effect of the test drug relative to the comparator All three sources of information are subject to uncertainty. For the first and third, the uncertainty is largely of a statistical nature, measured by standard errors (for the synthesis method) and confidence intervals (for the fixed margin method). The second conclusion is subject to scientific uncertainty that is largely unquantifiable. The first source of uncertainty is the estimation of the effect of the active control in past studies. Particular problems arise when there is only a single historical study of the active control versus placebo because there is then no information about study-to-study variability; when there are multiple studies but substantial inconsistency in the estimates of the size of the effect among them; and when data from several pharmacologically related drugs are used to develop the estimate for the effect of the active control. All three of these potential problems need to be considered when choosing M 1. The second source of uncertainty is not statistically based but rather arises from the concern that the magnitude of effect estimated from past studies will be larger than the effect of the active control in the current NI study. Assuming that the effect will be unchanged is often referred to as the constancy assumption. If the assumption is incorrect and the magnitude of the effect in the current NI study is smaller than the estimated effect from historical studies, M 1 will have been incorrectly chosen (too large) and an apparently successful study showing NI could have given an erroneous result. Lack of constancy can occur for many reasons, including advances in adjunctive medical care, differences in the patient populations, or changes in the assessment of the endpoints under study. As discussed in section IV.B.2.c, there is some experience to support the view that in cardiovascular outcome studies, the absolute size of the treatment effect is more likely to be variable and sensitive to the background rates in the control group than is the risk reduction. The risk reduction may thus be a more constant measure of control drug effect than the absolute effect. How to adjust the NI margin for concerns about constancy is inevitably a matter of judgment. The third source of uncertainty involves the risk of making a false conclusion based on the results of the test of the NI hypothesis in the NI study (i.e., concluding that C-T < M 1 when it is not). This uncertainty is referred to as the Type I error, or the false positive conclusion risk, and is similar to the concern in a placebo-controlled superiority trial that one might mistakenly conclude that a drug is more effective than placebo. It is, in other words, present in any 20

116 hypothesis-testing situation. In the NI case, the statistical test is intended to ensure that the difference between control and test drug (C-T, the degree of superiority of the control over the test drug) is smaller than the NI margin, meaning that some of the effect of the control is preserved (if C-T < M 1 ) or that a sufficient amount is preserved (if C-T < M 2 ). Typically, the one-sided Type I error is set at by asking that the upper bound of the 95% confidence interval for C-T be less than the NI margin; this is roughly similar to the usual statistical test of superiority in a placebo-controlled trial. If only one NI study is going to be conducted, the probability of a Type I error can be made smaller by requiring that the upper bound of a confidence interval greater than 95% (e.g., 99%) be calculated and be less than the margin. This approach is similar to what is a commonly done for a single placebo-controlled trial (e.g., testing at an alpha of instead of 0.05). As noted earlier, however, there may be prior information that eases this concern, and a single study at the usual Type I error boundary (0.025) may be considered sufficient if, for example, the drug and active control are pharmacologically similar. In the following subsections, we elaborate on the impact of the first two sources of uncertainty as well as the choice of margin to use in hypothesis testing. 2. Quantification of the Active Control Effect Past controlled studies provide the empirical data for estimating the treatment effect of the active control. The magnitude of that treatment effect, which will be the basis for determining the control drug effect that can be assumed to be present in the NI study, is critical to determining whether conducting an NI study is feasible. If the active control has a small treatment effect, an effect only marginally distinguishable from placebo, or an inconsistent effect, an active controlled study designed to show non-inferiority is likely to require a very large sample size or may not be practical at all. The magnitude of the treatment effect of the active control may be determined in several ways, depending upon the amount of data and the number of separate studies of similar design available to support this determination. Having many independent historical studies is generally more informative than having only one or two, because the estimate of the active control effect can be more precise and less subject to uncertainty, and because it becomes possible to judge the constancy of the effect for at least the time period spanned by the studies. a. Determining HESDE from a single study The most common situation in which an NI design is used to demonstrate effectiveness involves outcome studies where the active control drug has been approved for use to reduce the risk of major events (e.g., death, stroke, heart attack, continued infection, or tumor progression). It is not unusual for such approval to have been based on a single study in a specific setting, although there may be other pertinent data in related conditions or in different populations, or with pharmacologically similar drugs. Generally, basing an NI margin on a single randomized placebo-controlled superiority study would need to take into account the variability of the data in that study. The estimate of the treatment effect is usually represented by some metric such as the difference between the event rate in the active treatment group and the placebo control group, 21

117 which can be a difference in event rates or a risk ratio. The treatment effect has an uncertainty that is usually represented by the confidence interval. The lower bound of the 95% confidence interval provides a conservative estimate of the effect the active control had in the historical study and can be assumed to have in a future NI study, recognizing the possibility that changes in practice could reduce the historical effect somewhat. It is critical that a conservative estimate of the effect of the active control be used as the basis of the NI margin, because the logic of the NI study depends on the assumption that the active control has an effect at least equal to M 1 in the NI study. If the p-value of the estimated treatment effect is much smaller than 0.05, for instance in the range of 0.01 or or even smaller, the lower bound of the 95% confidence interval will generally be well above zero (in absolute value) or well below 1.0 (for hazard ratio and other risk estimates). In this case, we are more certain that the treatment effect is real and that the effect of the control in the NI study will be of reasonable size. When there is only a single placebo-controlled trial of the active control, there is no objective assessment of study-to-study variability of the treatment effect and, inevitably, there is concern about the level of assurance we can have that the control will have an effect of a particular size in the NI study. A potential cautious approach to account for this situation is to use the lower bound of a wider confidence interval, such as the 99% confidence interval. This approach is possible where the effect is very large, but will often yield an M 1 that necessitates a very large NI trial. It may be reassuring in these cases if closely related drugs, or the control drug in closely related diseases, have similar effects. A high level of internal consistency in subpopulations (e.g., if the effect of the control drug is similar in subgroups based on gender or age) could also provide some reassurance about the reproducibility of the result. These findings might support use of the 95% confidence interval lower bound even if only a single study of the active control drug in the population is available for the NI trial. b. Determining HESDE from multiple trials Identical clinical trials in identical populations can produce different estimates of treatment effect by chance alone. The extent to which two or more studies produce estimates of treatment effect that are close is a function of the sample size of each study, the similarity of the study populations, the conduct of the studies (e.g., dropout rates), and other factors that are probably not measurable. Therefore, another source of uncertainty to be considered when choosing a margin for the current NI study is the study-to-study variability in the estimate of the active control treatment effect. Multiple studies of the active control relative to a placebo, ineffective treatment, or no treatment provide an opportunity to obtain not only an estimate of the average effect of the active control and its uncertainty, but also a measure of the study-to-study variability of the effect. Study-tostudy variability in the active control effect is a critical consideration because one of the basic assumptions in NI studies is the consistency of the effect between the historical studies and the current NI study. 22

118 Several cases illustrate the use of multiple studies to estimate the active control effect and problems that can arise. The ideal case is one where (1) there are a number of studies, each of sufficient size to demonstrate the effect of the active control and (2) the effects derived from these studies are reasonably consistent, so that an average effect derived from a meta-analysis provides a reasonable estimate of the control effect and supports a credible choice of M 1. If there are many small studies, some of which have not demonstrated an effect of the active control, it may still be possible to estimate the active control effect combining the studies using appropriate meta-analysis methods, but care should be taken if there is evidence of a large amount of study-to-study heterogeneity in the treatment effects. If there are multiple large studies, it would be troublesome if one or more of the studies showed little or no effect. Unless there is a convincing explanation as to why one of the studies did not demonstrate an effect, a failed study may argue against use of an NI design. There are sometimes several trials of different drugs in the same pharmacologic class. Pooling them with appropriate meta-analytic methods may allow calculation of a 95% confidence interval that is narrower than that from any single study. The presumption that the pharmacologically similar drugs would have similar effects may be reasonable, but care should be exercised in extending this assumption too far. If the effects of these different drugs vary considerably in the trials, it may be reasonable to use the combined data to estimate the average effect with a meta-analysis and then select the drug with the largest effect (point estimate) as the active control in the NI study. When multiple studies of the active control are available, it is important to consider all studies and all patients when estimating the average effect. Dropping a study that does not show an effect, unless there is a very good reason, can overestimate the control drug effect and give a falsely high M 1. As noted above, the existence of properly designed and sized studies that show no treatment effect of the active control may preclude conducting NI studies with that control unless there are valid reasons to explain the disparate results. When combining data from multiple studies to estimate the active control effect, some variation of effect is expected. Random-effects meta-analysis can be used that allow for both betweenstudy and within-study heterogeneity. Both frequentist and Bayesian methods are available that differ in the assumptions made about the distribution of the random effects. These methods should be used to explore whether the variation between studies is more than would be expected by chance. If so, the between-study variation should be taken into account and will result in a more conservative margin for the NI study. Examples 1 and 2 in the Appendix illustrate in more detail how multiple historical placebo-controlled trials of the active control are evaluated in designing an NI trial. c. Metrics 23

119 As previously discussed, the assumption of constancy of the effect of the active comparator between the historical studies and the NI study is crucial to a conclusion of efficacy based on non-inferiority. There are, however, technical difficulties in defining what is meant by a constant effect across studies in different populations and at different times. The mathematical way in which the treatment effect is formulated will affect the meaning and plausibility of the constancy assumption. Consider, for example, studies of fairly infrequent, binary outcomes, typical of most cardiovascular studies, a situation in which NI trials have been successfully employed. Suppose the chance of an event in the placebo group of a historical study was p 1, and in the active treatment group p 2. Then the treatment effect in the historical study can be expressed in several ways: 1. The ratio p 2 /p 1, called the relative risk or risk ratio 2. The relative risk reduction 1 p 2 /p 1 3. The odds ratio [p 2 /(1 p 2 )]/[p 1 /(1 p 1 )] 4. The risk difference p 1 p 2 5. The number needed to treat (NNT), 1/(p 1 p 2 ) If p 1 and p 2 vary across studies, they might nevertheless vary in such a way that p 2 /p 1 is constant. For example, if p 1 and p 2 are 0.10 and 0.05 in one study, and 0.06 and 0.03 in another study, the ratio p 2 /p 1 is 0.5 in both studies. The relative risk reduction will then also be the same in both studies, and the odds ratio will be approximately the same. The risk difference, however, is 0.05 in one study and 0.03 in the other, and the numbers needed to treat are 20 and 33. In contrast, if two studies have different p 1 and p 2 but the same risk difference, they will have the same NNT but cannot have the same relative risk, relative risk reduction, or odds ratio. Thus, the very definition of constancy depends on how the effect is formulated mathematically. Because the interpretation of a NI study relies crucially on the assumption of constancy, careful consideration must be given to the question of what, if anything, is likely to be constant across historical studies and between the historical studies and the NI study. As noted, based on both experience and expectation, the constancy assumption for outcome studies has generally been based on expected constancy of relative and not absolute effects. The difference between expressing the treatment effect as the absolute difference between success rates in treatment groups and as the relative risk or risk ratio for success on the test treatment relative to the active comparator is illustrated in the following two examples. For the first example, consider a disease where the cure rate is at least 40% in patients receiving the selected active control and 30% for those on placebo, a 10% difference in cure rates. If the purpose of an NI study is to demonstrate that the test product is effective (i.e., superior to a placebo), then the difference between the test product and active control in the NI study must be less than 10%. The margin M 1 would then be 10%. If the additional clinical objective is to establish that the test product preserves at least half of the active control s effect, then the cure rate of the test product must be shown to be less than 5% lower than the control, the M 2 margin. 24

120 This approach requires that the control drug have an effect of at least 10% relative to placebo (had there been one) in the NI study. If the population in the NI study did not have such a benefit (e.g., if the patients all had illnesses not susceptible to the test drug such that the benefit was less than 10%), then even if the 5% difference were ruled out, that would not demonstrate the desired effectiveness (although it would seem to). Note that in this case, if the true effect of the control in the study were 8%, then ruling out a 5% difference would in fact show some effect of the test drug, just not the desired 50% of control effect. The second example illustrates a NI margin selected for the risk ratio (test/control) metric. Let C and P represent the true rates of an undesirable outcome for the control and a placebo, respectively. The control s effect compared to placebo is expressed by the risk ratio, C/P. A risk ratio of 1 represents no effect; a ratio of less than 1 shows an effect, a reduction in rate of undesirable outcomes. Metrics like the risk ratio may be less affected by variability in the event rates in a placebo group that would occur in a future study. For example, a risk ratio for the event of interest of 3/4 =0.75 can be derived from very different absolute failure rates from different studies, as shown in the table below. While the risk ratio is similar in all four hypothetical studies, the absolute difference in failure rates ranges from 5% to 20%. Suppose that the NI margin were based on historical studies showing control drug effects like those in the fourth study. The NI margin (M 1 ) would then be chosen as 20%. Now suppose that under more modern circumstances the NI study had a control rate more like Study 1 and the size of the treatment effect compared to placebo is far less than 20%. An NI margin (M 1 ) of 20% would then be far greater than the drug effect in the NI study, and ruling out a difference of 20% would not demonstrate effectiveness at all. Thus, if the NI margin were chosen as ruling out an inferiority of 33% (or a relative risk of 1.33, i.e., ), and the control rate were 15%, the difference (M 1 ) between test and control would need to be less than 5% (15% x 1.33 = 20%, or 5% > the 15% rate in the active control group). Study Number Risk Ratio (C/P) Control rate Placebo rate Study 1 ¾ 15% 20% Study 2 ¾ 30% 40% Study 3 ¾ 45% 60% Study 4 ¾ 60% 80% In this case, where absolute effect sizes vary but risk reductions are reasonably constant, the risk ratio metric provides a better adjustment to the lower event rate in the NI study. Revisiting the example from section III.E, it was clear in the placebo-controlled trials of streptokinase that the mortality rate in the placebo group was decreasing, so that even if the mortality reduction were constant as a relative reduction, the absolute reduction would decrease and the expected size of the treatment effect (M 1 ) would no longer be an absolute 2.1% reduction. The relative risk of mortality (ratio of mortality rates) was therefore used for subsequent thrombolytic product development programs because thrombolytic efficacy expressed as the relative mortality rates appeared to be more stable over time as improved standards of care were provided to all patients in a study. 25

121 For the development of tenecteplase in the treatment of AMI, evaluation of the relevant historical studies led to the conclusion that the comparator (90 minute infusion of alteplase) would reduce the relative risk of mortality compared to placebo by at least 0.22 (to about 0.78, upper limit of the relative risk confidence interval). The NI margin was chosen, again based upon clinical judgment, so the new drug would retain at least half of the efficacy of the comparator (M 1 = 0.22), i.e., at least 0.11 (M 2 ). The relative mortality risk of tenecteplase compared to placebo should be not greater than This provided a value for the limit of the relative risk comparison of tenecteplase to alteplase of 1.14 ([tenecteplase/placebo]/[alteplase/placebo] = 0.89/0.78), which had to be excluded by the 95% confidence interval (one-sided in this case) of the ratio of mortality rates in the NI study. The ASSENT II study results provided a 95% onesided upper limit of relative risk of 1.104, which led to the marketing approval of tenecteplase. These examples illustrate the importance of understanding how a particular metric will perform. The choice between a relative metric (e.g., risk ratio) and an absolute metric (e.g., a difference in rates) in characterizing the effects of treatments may also be based upon clinical interpretation, medical context, and previous experience with the behavior of the rates of the outcome. d. Discounting One strategy employed in choosing the margin M 1 for the NI study design is discounting or reducing the magnitude of the active control s effect that is estimated from the HESDE analysis. Such discounting is done to account for the uncertainties in the assumptions that need to be made in estimating, based on past performance, the effect of the active control in the NI study. This concept of discounting focuses on M 1 determination and is distinct from a clinical judgment that the effect that can be lost on clinical grounds should be some fraction of M 1 (i.e., M 2 ). As discussed above, there are uncertainties associated with translating the historical effect of the active control (HESDE) to the new situation of the active control NI trial, and it is tempting to deal with that uncertainty in the constancy assumption by discounting the effect (e.g., take half ). Rather than applying automatic discounting in choosing M 1, concerns about the active control effect should, to the extent possible: Be specific Make use of available data (e.g., magnitude of possible differences in effect in different patient populations, consistency of past studies, and consistency within studies across population subsets) Take into account factors that reduce the need for a conservative estimate, such as the pharmacologic similarity of the test and control drugs and pharmacodynamic effects of the new drug A closely related issue is the possible need for adjustment of M 1 to reflect any differences that were observed between the population in the NI study and the historical study. For example, a finding of a smaller effect in women than in men in the historical studies would need to be considered in assessing the validity of M 1 if the NI study had a substantially greater proportion of women than the historical studies. In general, the assessment of the historical data should identify differences in the outcomes for important subgroups so that the statistical analysis plan 26

122 for the NI study can be designed prospectively to take these factors into account (e.g., through covariate adjustments) or so that the value of M 1 can be revisited in light of the baseline characteristics of the study population that was enrolled in the NI study. C. Fixed Margin and Synthesis Methods Conceptually, the conclusion of non-inferiority is a synthesis of information from the NI trial itself and the historical evidence that the comparator in that trial was effective. The historical trials assessed C h P, the effect of the active control compared to placebo. The NI trial assesses T C n, the effect of the test drug compared to active control. If the outcome for the active control is constant across studies, then C n = C h, and the sum T C n + C h P = T P represents the effect of the test drug compared to placebo. The indispensable purpose of a NI trial is to show with high confidence that this effect is positive. Additionally, NI trials are usually intended to estimate this effect, at least roughly, in comparison to the historical effect of the active control. The historical studies will furnish an estimate of C h P with an associated standard error SE H. Similarly, the NI study will estimate T C n with a standard error SE N. The two estimates are independent, so that the standard error of the sum is given by SS H 2 + SS N 2. Notwithstanding that the interpretation of an NI study is fundamentally a synthesis, we recommend a statistical method, the fixed-margin method, that treats the problem in two separate steps. This approach offers two practical advantages. First, it allows separation of the problem of calculating, justifying, and possibly adjusting the NI margins M 1 and M 2 from the problem of analyzing the NI study. Second, it is equivalent to applying a somewhat larger standard error to the synthesized estimate, which can be seen as an allowance for the possibility of minor deviations from constancy. 1. The Fixed Margin Approach The fixed margin approach to analysis in an NI trial is well known (see, e.g., ICH E9, section 3.3.2). This approach relies upon the choice of a fixed NI margin that is prespecified at the design stage of the NI trial and used to determine the sample size needed to provide sufficient power for a test of the hypothesis of non-inferiority. Operationally, the fixed margin approach usually proceeds in the following manner. First, the active control effect is estimated from an analysis of past placebo-controlled studies of the active control (see section IV.B.2.a-b). The goal of this analysis is to define the margin M 1, a fixed value based on the past effect of the active control that is intended to be no larger than the effect the active control is expected to have in the NI study. The variability of the treatment effects observed across past studies as well as the constancy assumption should be taken into account in choosing M 1. 27

123 The selection of M 2 is then based on clinical judgment regarding how much of the active comparator treatment effect needs to be retained to demonstrate sufficient benefit for drug approval. The exercise of clinical judgment for the determination of M 2 should be applied after the determination of M 1 has been made. All relevant studies of the active comparator and all randomized patients within these studies should generally be used in determining the margin M 1 because that provides a more reliable estimate of the active control effect and avoids any potential for prejudice in selecting which historical studies to include. Determination of the relevance of past studies, however, may not be straightforward, and Examples 1(A) and 2 in the Appendix illustrate this point. The lower bound of the confidence interval of the estimated active control effect based on past studies is typically selected as M 1. Concerns about past variability and constancy may lead to a determination to discount this lower bound in choosing M 1 to account for any sources of uncertainty and dissimilarities between the historical data and the NI study to be conducted, as discussed in earlier sections. Following this, a clinical judgment is made as to how much of this effect should be preserved. Choosing M 2 as 50% of M 1 has become usual practice for cardiovascular (CV) outcome studies, where effects may be small, and it is important to retain at least half of the effect of the original treatment. In antibiotic trials, effects tend to be quite large relative to placebo or ineffective treatments, and as a result, a 10-15% NI margin for the treatment difference (M 2 ) is commonly chosen. Note that the M 2 of 50% of M 1 is on a relative scale, whereas the 10-15% is on the risk difference scale for antibiotic drugs. The formal test of the NI hypothesis also involves the use of a 95% confidence interval (i.e., to rule out C-T > M 2 ). Thus, there are two confidence intervals involved in the fixed margin approach, one for the historical data, where the lower bound is selected as M 1, and one for the NI study (to rule out C-T > M 2 ); both intervals are typically 95% confidence intervals. That is why this fixed margin approach is sometimes called the 95%-95% method. Separating the process of estimating the treatment effect of the active comparator based upon the historical data (i.e., choice of M 1 ) from the analysis of the NI study has two advantages. It gives a single number that is clinically understandable for M 1 (and derived M 2 ) and provides a basis for planning the sample size of the NI study to achieve control of Type I error and the power needed for the NI study to meet its objective for the prespecified NI margin. Decisions to discount M 1 further or, where appropriate, to use a narrower confidence interval, are easily explained, and will make the fixed margin approach either more or less conservative. Deciding on the NI clinical margin M 2 is a matter of judgment about how large a loss of the control treatment effect would be acceptable, a consideration that may reflect the seriousness of the outcome, the benefit of the active comparator, and the relative safety profiles of the test and comparator. Choice of M 2 also has major practical implications. For example, in large cardiovascular studies, it is unusual to have an M 2 that reflects a loss of less than 50% of the control drug effect, even if this might be clinically reasonable, because doing so will usually make the study size infeasible. Of course, allowing too much inferiority of the test drug to the standard, especially for endpoints of mortality and serious morbidity, would clearly not be acceptable. 28

124 The fixed margin approach considers the NI margin as a single number, fixed in advance of the NI study. The hypothesis tested in the NI study determines whether the comparison of the test drug to the active control meets the specified NI criterion, assuming, of course, that the active control had at least its expected effect (equal to M 1 ) (that is, the study had assay sensitivity). A successful NI conclusion, ruling out a difference > M 1, shows that the test drug is effective (just as a superiority study showing a significant effect at p 0.05 does) and, if a difference > M 2 is also ruled out, shows that the new drug preserves the desired fraction of the control drug s effect. 2. The Synthesis Approach An alternative statistical approach is known as the synthesis approach because it combines or synthesizes the data from the historical trials and the current NI trial, reflecting the variability in the two sources of data (the current NI study and the past studies used to determine HESDE). The synthesis method is designed to directly address the question of whether the test product would have been superior to a placebo had a placebo been in the NI study and also to address the related question of what fraction of the active control s effect is maintained by the test product. In the synthesis approach, M 1 is not prespecified, but the percent of active control effect to be preserved is prespecified. Based on the observed outcome of the test drug vs. active control comparison in the NI study, an evaluation is made as to whether the test agent has demonstrated preservation of the clinically relevant effect of the active control, without determining M 1. Although the synthesis approach combines the data from the historical trials into the comparison of the concurrent active control and the test drug in the NI study, a direct randomized concurrent comparison with a placebo is of course not possible, because the placebo group is not a concurrent control and there is no randomization to such a group within the NI study. The imputed comparison with a placebo group that is not part of the NI study thus rests on the validity of several assumptions, just as the fixed margin approach does. The critical assumption of the constancy of the active control effect derived from the historical placebo-controlled trials is just as important when the synthesis method is used as when the fixed margin method is used. The use of the synthesis approach can lead to a more efficiently designed study (e.g., by allowing for a reduction in sample size or achieving greater power for a given sample size) than the fixed margin approach, provided the constancy assumption holds. The greater statistical efficiency of the synthesis approach derives from how this method deals with the standard error of the comparison of test product to active control. See Appendix, Example 1(B), for a comparison of the two methods and the variance calculations. The synthesis approach does not specify a fixed NI margin. Rather, the method combines (or synthesizes) the estimate of the treatment effect relative to the control from the NI trial with the estimate of the control effect from a meta-analysis of historical trials. The synthesis process makes use of the variability from the NI trial and the historical trials and yields one confidence interval for testing the NI hypothesis that the treatment preserves a fixed fraction of the control effect, without actually specifying that control effect or a specific fixed NI margin based on the control effect. Clinical judgment is used to prespecify an acceptable fraction of the control therapy s effect that should be retained by the test drug, regardless of the magnitude of the 29

125 control effect. The disadvantage of the synthesis approach is that it is not possible to use clinical judgment to choose M 2, based on the magnitude of M 1, in advance of the NI trial. D. Considerations for Selecting the Clinical Margin (M 2 ) M 2 is the prespecified NI margin that is to be ruled out in an NI study. The determination of M 2 is based on clinical judgment and is usually calculated by taking a percentage or fraction of M 1. The clinical judgment in determining M 2 may take into account the actual disease incidence or prevalence and its impact on the practicality of sample sizes that would have to be accrued for a study, as noted above with respect to cardiovascular outcome studies. There can be flexibility in selecting the M 2 margin, choosing a wider margin, for example, when: 1. The primary endpoint does not involve an irreversible outcome such as death (in general, the M 2 margin will be more stringent when treatment failure results in an irreversible outcome) 2. The test product is associated with fewer serious adverse effects or better tolerability than other therapies already available 3. The test product has another advantage over available therapies that warrants use of a less stringent margin (M 2 ) A more stringent choice for M 2 may be required, however, when the difference between the active comparator response rate and the untreated response rate is large, making it feasible to demonstrate retention of a larger fraction of M 1 without requiring an impractical sample size for the NI study. The implication of failing to rule out inferiority relative to M 1 and M 2 differs. Failure to exclude inferiority relative to M 1 means there is no assurance of any effect. Just as it would be unusual to accept a placebo-controlled study as positive (i.e., a finding of superiority) with p > 0.05, it would be unusual to accept an NI study as positive (i.e., a finding of non-inferiority) where the upper bound of the 95% confidence interval was > M 1. On the other hand, failing to exclude M 2 by a small amount (e.g., ruling out a loss of 52% of M 1 rather than a prespecified loss of 50% of M 1 ) may be acceptable, as the small amount would not suggest the absence of an effect of the drug. E. Estimating the Sample Size It is important to plan the sample size for an NI clinical trial so that the trial will have adequate statistical power to conclude that the NI margin is ruled out if the test drug is truly non-inferior. At the protocol planning stage, using the fixed margin approach, the NI margin (M 2 ) will be specified, and the sample size will be based on the need to rule out that margin. Both the size of the margin and the estimated variance of the treatment effect (T vs C) will affect the sample size determination. For event-driven trials, if the event rate in the NI study is lower than anticipated, power to demonstrate non-inferiority will be reduced. It may be important, therefore, to monitor overall (blinded) event rates during the study and to adjust the sample size if the interim event rates are unexpectedly low, a practice that is common in superiority trials. There is one further consideration in sample size planning. If, in reality, the test drug is somewhat more effective 30

126 than the control, it will be easier to rule out any given NI margin than if the test drug is equivalent or slightly inferior to the control, and a smaller sample size could be used. A somewhat less effective test drug will, of course, require a larger sample size. It may be difficult to adequately plan the sample size for any study, including an NI study, because of uncertainty about assumptions such as event rates or endpoint variability at the time of study planning. For this reason, adaptive study designs that can allow for the prospective reestimation of a larger sample size at one or more planned interim looks may be considered. F. Study Quality and Choice of Analysis Population Traditionally, the primary analysis of a randomized clinical superiority trial follows the intention-to-treat (ITT) principle, namely, all randomized patients are analyzed according to the treatment to which they were randomized, including patients who leave the study prematurely. This approach is intended to avoid various biases associated with patients switching treatment or patients being excluded from the analysis because of protocol violations or attrition. Adhering to the ITT principle in superiority trials is generally considered conservative, in that poor study quality resulting in a large number of protocol violations will tend to bias the results towards the null hypothesis of no difference between treatments. The opposite is true for NI trials. Quality issues could result in treatment groups appearing similar (i.e., biasing the results towards the alternative hypothesis for NI trials), when, in fact, the test drug may be inferior, as mentioned in section III.D.3. Many problems that may cause a superiority trial to fail, such as non-adherence, misclassification of the primary endpoint, or attrition, can bias the results toward no treatment difference (success) and undermine the validity of the trial, creating apparent non-inferiority when the test drug is in fact inferior. Imputation of missing data under the inferiority null hypothesis is one possible approach to countering the bias due to attrition. The best advice for conducting an NI study is to emphasize study quality at the planning stage and continuously monitor the trial during the conduct and analysis stages to minimize the problems described above. If the NI trial is open label, attention to quality is all the more important because it may be very difficult to prove, after the fact, that enrollment, assessment of endpoints, and other study procedures were conducted in an unbiased way. G. Testing Non-Inferiority and Superiority in a Single Trial In general, when there is only one primary endpoint and one dose of the test treatment, a trial that is planned to demonstrate non-inferiority may also be used to test for superiority without concern about inflating the Type I error rate. This sequential testing procedure has the Type I error rates for tests of both non-inferiority and superiority controlled at the 2.5% level. A study designed primarily to show superiority, however, would yield credible evidence of non-inferiority only if the study had the key features of a NI study (e.g., the NI margin must be prespecified, and assay sensitivity and HESDE must be established). An unplanned determination of non-inferiority following failure to show superiority, when the margin was not determined until results of the trial were known, would not be sufficient for demonstrating non-inferiority of the test drug. 31

127 When multiple endpoints or multiple doses of the test treatment are to be evaluated in an NI study that also includes a test for superiority, careful planning of the order in which hypothesis tests are to be carried out is needed. For example, will superiority of the primary endpoint be tested before or after non-inferiority of a key secondary endpoint? A decision tree can be helpful in determining whether adjustments for multiplicity are needed and to what tests they should be applied to ensure that adequate control of the Type I error rate is achieved for the trial. In general, it would not be appropriate to apply 95% confidence intervals for multiple tests of noninferiority and superiority across multiple endpoints or multiple doses because of the potential to inflate the Type I error rate for the trial; use of larger intervals that reflect adjustment for multiple tests (e.g., 97.5% confidence intervals or higher) may be needed. V. FREQUENTLY ASKED QUESTIONS AND GENERAL GUIDANCE 1. Can a margin be defined when there are no historical placebo-controlled trials of the active control for the disease being assessed in the NI study? If the active control has shown superiority to other active treatments in the past, the difference demonstrated represents a conservative estimate of HESDE, one that can certainly serve as a basis for choosing M 1. It may also be possible that trials of the active control in related diseases are relevant. The more difficult question is whether historical experience from nonconcurrently controlled trials can be used to define the NI margin. The answer is that it can, but only in situations that meet the following three general criteria for assessing the persuasiveness of a historically controlled trial (see ICH E10). First, there should be a good estimate of the historical untreated response rate or outcome without treatment. Examination of medical literature and other sources of information may provide data upon which to base these estimates (e.g., historical information on natural history or the results of ineffective therapy). Second, the cure rate of the active control should be estimated from historical experience, preferably from multiple experiences in various settings and possibly including observational studies. Third, the untreated and treated patients should be comparable. Then, if the treated and untreated response rates are substantially different, it may be possible to determine a NI margin. For example, if the spontaneous cure rate of a disease is 10-20% and the cure rate with an active control is 70-80%, these rates are substantially different and can be used to determine M 1. The clinically acceptable loss of this effect can then be determined for M 2. Some examples of margins determined in this way have been presented in guidance on antibiotic trials, e.g., trials in community-acquired bacterial pneumonia (see Example 4 in the Appendix). Identifying a margin is more difficult when the difference between the spontaneous cure rate and active drug cure rate is smaller. For example, if the historical spontaneous cure rate is 40% and the active control rate is 55%, identifying the NI margin in this case as 15% would not be reasonable because such a small difference could easily be the result of a different disease definition or ancillary therapy. When the historical cure rates for the active control and the cure rate in patients who receive no treatment are not known at all from actual studies (i.e., are just based on clinical impressions), it will be difficult or impossible to define an NI margin. 32

128 2. Can the margin M 2 be flexible? As discussed in detail in sections III and IV, there is a critical difference between demonstrating that the margins M 1 and M 2 have been met. M 1 is used to determine whether the NI study shows that the test drug has any effect at all. Accepting a result in which the 95% confidence interval did not rule out a loss of M 1 or greater would be similar to accepting as evidence of effectiveness a superiority study whose estimated treatment effect was not significant at p M 2, in contrast, represents clinical judgment about the amount of the active control effect that must be retained. A typical value for M 2 is often 50% of M 1, at least partly because the sample sizes needed to retain a larger amount, e.g., 60% or more, of the active control effect become impractically large. In this case, there is a better argument for some degree of flexibility if the study did not quite rule out the M 2 margin; there might be reason to consider, for example, assurance of 48% retention as acceptable. We have also concluded that the more conservative fixed margin method should generally be used in ensuring that M 1 is ruled out, but that the synthesis method can be used to assess noninferiority with respect to M 2. Of course, allowing the test drug to be inferior to the standard by too large an amount, especially for endpoints of mortality and serious morbidity, would not be acceptable. 3. Can prior information or other data (e.g., studies of related drugs, pharmacologic effects) be considered statistically in choosing the NI margins or in deciding whether the NI study has demonstrated its objective? Prior information can be incorporated into a statistical model or within a Bayesian framework to take into account such factors as evidence of effects in other related indications or on other endpoints. As discussed in section IV.B.2.b, a meta-analysis is often used to estimate the average effect of the active control for purposes of setting the NI margin, and in certain circumstances, trials from related indications or for other drugs in the same class may be included in the meta-analysis conducted for this purpose. Some methods of meta-analysis allow the down-weighting of less relevant studies or studies that are not randomized or controlled (e.g., observational studies), which can be particularly important if few placebocontrolled trials are available. Bayesian methods that incorporate historical information from past active control studies through the use of prior distributions of model parameters provide an alternative approach to evaluating non-inferiority in the NI trial itself. Although discussed in the literature and used in other research settings, CDER and CBER have not had much experience to date in evaluating NI trials of new drugs or therapeutic biologics that make use of a Bayesian approach for design and analysis. If a sponsor is planning to conduct a Bayesian NI trial, early discussions with the Agency are advised. If important covariates are distributed differently in the historical studies than in the current NI study, model-based approaches may be used to adjust for these covariates in the NI analysis. Such covariates should be identified prior to the NI trial, and the methods for covariate adjustment should be specified prospectively in the NI trial protocol. Applying 33

129 post-hoc adjustments developed at the time of analyzing the NI trial would not be appropriate. 4. Can a drug product be used as the active comparator in a study designed to show noninferiority if the product s labeling does not have the indication for the disease being studied, and could published reports in the literature be used to support a treatment effect of the active control? The active control does not have to be labeled or approved in the United States for the indication being studied in the NI study as long as there are adequate data that are reliable and reproducible to support the chosen NI margin. FDA does, in some cases, rely on published literature and has done so in carrying out the meta-analyses of the effect of the active control that are used to define NI margins. The FDA guidance, Providing Clinical Evidence of Effectiveness, describes the approach to considering the use of literature in providing evidence of effectiveness, and similar considerations would apply here. Among these considerations are: The quality of the publications (the level of detail provided) The difficulty of assessing the endpoints used Changes in practice between the present and the time of the studies Whether FDA has reviewed some or all of the studies Whether FDA and the sponsor have access to the original data As noted above, the endpoint for the NI study could be different (e.g., death, heart attack, and stroke) from the primary endpoint (cardiovascular death) in the studies if the alternative endpoint is well assessed and data on that endpoint are available for determining M 1 (see also question 6). 5. If the active control drug is approved for the indication that is being studied, does the margin need to be justified, or if the active control drug has been used as an active comparator in the past in another study of design similar to the current study and a margin has been justified previously, can one simply refer to the previous margin used? When an active control drug is approved for the same indication as that of the NI trial, the size of the effect appearing in the label of the active control is usually based on the pivotal trials rather than a meta-analysis of all past studies. Further, the variability of the effect across past studies may not be available. In general, approval of a drug is based on showing superiority to placebo, usually in at least two studies, but FDA may not have critically assessed whether the effect is consistent across past studies and may not have analyzed any failed studies. It is therefore essential to use the data from all available controlled trials (unless a trial has a significant defect), including trials conducted after marketing, to calculate a reasonable estimate of the actual control effect. If the active control data have been used to define the NI margin for another study of the same indication, and if a determination is made that the trials involved are relevant to the new trial, then use of the same margin in the new trial should be acceptable. 34

130 6. What factors should be considered when selecting an endpoint for a NI trial? The endpoints chosen for clinical trials (superiority or non-inferiority) should be clinically meaningful measures of the way patients feel, function, or survive that can be reliably assessed in the target population. The endpoints generally reflect the event rate or other measure of disease in the population but also must take into account practical considerations, such as the size of the study required to demonstrate superiority or non-inferiority with respect to the endpoint. In NI studies, the endpoint must be one for which there is a good basis for knowing the effect of the active control. The endpoint need not necessarily be the endpoint used in the historical trials or the effectiveness endpoint appearing in labeling of the active control. Past trials that were successful in showing an effect on a mortality endpoint could, for example, serve as the basis for estimating an effect on a composite endpoint (cardiovascular mortality, myocardial infarction, and stroke), if that were the desired endpoint for the NI study. The use of a different endpoint in the NI study might be desirable because it would permit a smaller study, but it would be important not to include components that were not affected by the active control or did not represent an important clinical benefit. 7. Are there circumstances where conducting an NI study may not be feasible? Unfortunately, these are several, including some where a placebo-controlled study would not be considered ethical. Some examples include the following: The treatment effect of the active comparator may be so small that the sample size required to conduct a NI study may not be feasible (unless the test drug is assumed to be superior). There is large study-to-study variability in the treatment effect. In this case, the treatment effect may not be sufficiently reproducible, calling into question the assumption of assay sensitivity. This is often the case for symptomatic treatments (e.g., treatments for depression, anxiety, insomnia, angina, symptomatic heart failure, symptoms of irritable bowel disease, and pain). There is no historical evidence available to determine a NI margin, and assuming a zero response rate for untreated patients is not reasonable. Medical practice has changed so much (e.g., the active control is always used with additional drugs) that the effect of the active control in the historical studies is not relevant to the current study. 8. In a situation where a placebo-controlled trial would be considered unethical, but a NI study cannot be performed, what are the options? In that case it may be possible to design a superiority study in an appropriate population that would be considered ethical. Several possibilities are discussed in ICH E10 and include the following: 35

131 When the new drug and an established treatment are pharmacologically distinct, an addon study where the test drug and placebo are each added to the established treatment. A study in patients who do not respond to the established therapy. It may be possible to conduct a placebo-controlled trial in these patients. Alternatively, nonresponders could be randomly assigned to the test drug or the failed therapy, and superiority assessed to demonstrate effectiveness in the nonresponder population. A study in patients who cannot tolerate the established effective therapy. A study of a population for which there is no available effective therapy. A dose-response study without placebo, if the new drug is known to have dose-related side effects, and a dose lower than the usual dose would be considered ethical. 9. When can a single NI study be sufficient to support effectiveness? Several sections above touch on this question, notably section III.H, which discusses this issue in detail. Briefly, reliance on a single study in the NI setting is based on considerations similar to reliance on a single study in the superiority setting, with the additional consideration of the stringency of showing non-inferiority using the M 2 margin. Many of these considerations are described in the FDA guidance, Providing Clinical Evidence of Effectiveness, and include supportive information, such as results with pharmacologically similar agents (a very common consideration, because the NI study will often compare drugs of the same pharmacologic class), support from credible biomarker information (tumor responses, ACE inhibition, beta blockade), and a statistically persuasive result from the single NI study. With respect to the latter, it is noted above that a finding of NI based on excluding a treatment difference > M 2 provides very strong evidence that the test treatment has an effect > 0 when M 2 is substantially smaller than M 1. For all these reasons, most NI studies with mortality or serious morbidity as endpoints, if clearly successful, will be sufficient to demonstrate NI as single studies. 36

132 APPENDIX EXAMPLES The following four examples derived from publicly available information (see references following examples) illustrate determination of the NI margin, application of methods of NI analysis, and other considerations relevant to determining whether it is possible to conduct and interpret the results of an NI study. Example 1(A): Determination of an NI Margin for a New Anticoagulant Fixed Margin Approach This example will demonstrate the following: Determination of the NI margin (M 1 ) using the fixed margin approach How to select and assess the randomized trials of the active control on which to base the estimate of active control effect How to assess whether the assumption of assay sensitivity is appropriate and whether the constancy assumption is reasonable for this drug class The use of 95% confidence intervals for both margin determination and in the NI study for C-T to demonstrate non-inferiority, i.e., the 95% - 95% method SPORTIF V is an NI study that tested the novel anticoagulant ximelagatran against the active control warfarin. Warfarin is a highly effective, orally active anticoagulant that is approved in the United States for the treatment of patients with nonvalvular atrial fibrillation at risk of thromboembolic complications (e.g., stroke, TIA). There are six placebo-controlled studies of warfarin involving the treatment of patients with nonvalvular atrial fibrillation, all published between the years 1989 and The primary results of these studies are summarized in Table 1 and provide the basis for choosing the NI margin for SPORTIF V. The point estimate of the event rate (ischemic and hemorrhagic strokes plus systemic embolic events) on warfarin compared to placebo is favorable to warfarin in each of the six studies. The upper bound of the 95% confidence interval of the risk ratio calculated in each study is less than one in five of the six studies, indicating a significant treatment effect in favor of warfarin. The one exception is the CAFA study. However, this study was reportedly stopped early because of favorable results published from the AFASAK and SPAF I studies (Connolly et al. 1991). Although the CAFA study was stopped early, a step that can sometimes lead to an overestimate of effect, the data from this study appear relevant in characterizing the overall evidence of effectiveness of warfarin because there is no reason to think it was stopped for early success, which might have introduced a favorable bias. These placebo-controlled studies of warfarin in patients with nonvalvular atrial fibrillation show a fairly consistent effect. Based on the positive results from five of the six studies, it can reasonably be assumed that were placebo to be included in a warfarin-controlled NI study involving a novel anticoagulant, warfarin would have been superior to placebo. 37

133 Contains Nonbinding Recommendations Table 1. Placebo-Controlled Trials of Warfarin in Nonvalvular Atrial Fibrillation Study Summary Events/Patient Years Risk Ratio (95% CI) Warfarin Placebo AFASAK open label. 1.2 yr follow-up 9/413 = 2.18% 21/398 = 5.28% 0.41 (0.19, 0.89) BAATAF open label. 2.2 yr follow-up 3/487 = 0.62% 13/435 = 2.99% 0.21 (0.06, 0.72) EAFT open label. 2.3 yr follow-up patients with recent TIA 21/507 = 4.14% 54/405 = 13.3% 0.31 (0.19, 0.51) CAFA* double blind. 1.3 yr follow-up 7/237 = 2.95% 11/241 = 4.56% 0.65 (0.26, 1.64) SPAF I open label. 1.3 yr follow-up 8/260 = 3.08% 20/244 = 8.20% 0.38 (0.17, 0.84) SPINAF double blind. 1.7 yr follow-up 9/489 = 1.84% 24/483 = 4.97% 0.37 (0.17, 0.79) * CAFA was stopped early because of favorable results observed in other studies. As can be seen from the summary table, most of these studies were open label. It is not clear how great a concern this should be given the reasonably objective endpoints in the study (see Table 2), but to the extent there was judgment involved in assessing the endpoints, there may have been the potential for bias to be introduced. The event rate for placebo in the EAFT study was strikingly high, perhaps because the patient population in that study was different from the patient population in the remaining five studies, in that only patients with a recent TIA or stroke were enrolled in EAFT. That factor would clearly increase the event rate, but in fact the risk reduction in EAFT was very similar to the four trials other than CAFA, which is relatively reassuring with respect to constancy of risk reduction in various AF populations. Even if the historical studies are consistent, a critical consideration in deciding upon the NI margin is whether the constancy assumption is reasonable. That is, we must consider whether the magnitude of the effect of warfarin relative to placebo in the previous studies would be present in the new NI study, or whether changes in medical practice (e.g., concomitant medications, skill at reaching the desired International Normalised Ratio (INR)), or changes in the population being studied may make the effect of warfarin estimated from the previous studies not relevant to the current NI study. To evaluate the plausibility of this constancy assumption, one might compare some features of the six placebo-controlled warfarin studies with the NI study, SPORTIF V. There is considerable heterogeneity in the demographic characteristics of these studies. While some characteristics can be compared across the studies (e.g., age, race, and target INR), some cannot (e.g., concomitant medication use, race, mean blood pressure at baseline) if they are not consistently reported in the study publications. Whether these are critical to outcomes is the question. Table 2 shows that for some characteristics, such as a history of stroke or TIA, there are some differences across the studies. An important inclusion criterion in the EAFT study was that subjects have a prior history of stroke or TIA. None of the other studies had such a requirement. Subjects enrolled into the EAFT study were thus at higher risk than subjects in the other studies, presumably leading to the higher event rates in both the warfarin and placebo arms, shown in Table 1. The higher event rates in the EAFT study may also have been influenced by the relatively long duration of follow-up or the fact that the primary endpoint definition was broader (including vascular deaths and nonfatal myocardial infarctions), or both. Even with the 38

134 higher event rates observed in this one study, however, the risk ratios are quite consistent (with the exception of CAFA), a relatively reassuring outcome. Table 2. Demographic Variables, Clinical Characteristics, and Endpoints of Warfarin Atrial Fibrillation Studies AFASAK BAATAF CAFA SPAF VA EAFT SPORTIF V Age years (mean) Sex (%) Male 53% 75% 76% 74% 100% 59% 70% h/o stroke or 6% 3% 3% 8% 0% 100% 18.3% TIA (%) h/o HTN (%) 32% 51% 43% 49% 55% 43% 81% >65 years old 8% 10-16% 12-15% 7% 17% 7% 41% & CAD (%)* >65 years old 7-10% 14 16% 10-14% 13% 17% 12% 19% & DM (%)* h/o LV dysfunction (%)* 50% 24-28% 20-23% 9% 31% 8% 39% Mean BP at NA NA NA 130/78 NA 145/84 133/77 BL (mm Hg) Target INR Primary endpoint Ischemic stroke Ischemic stroke Stroke, TIA, systemic embolism Ischemic stroke and systemic embolism Ischemic stroke and systemic embolism Vascular death, NF MI, stroke, systemic embolism Stroke (ischemic + hemorrhagic) and systemic embolism * = Not possible to verify whether definitions of CAD, DM, and LV dysfunction were the same in comparing the historic studies and SPORTIF V. NA = Not available To calculate M 1, the relative risks in each of the six studies were combined using a random effects meta-analysis to give a point estimate of for the relative risk with a confidence interval of (0.248, 0.527). The 95% confidence interval upper bound of represents a 47% risk reduction, which translates into a risk increase of about 90% from not being on warfarin (1/0.527 = 1.898) (i.e., what would be seen if the test drug had no effect). Thus, M 1 (in terms of the hazard ratio favoring the control to be ruled out) is It was considered clinically important to show that the test drug preserved a substantial fraction of the warfarin effect. The clinical margin M 2 representing the largest acceptable level of inferiority was therefore set at 50% of M 1. M 2 was calculated on the log hazard scale as 1.378, so that NI would be demonstrated provided the upper bound for the 95% confidence interval for C vs. T < In the SPORTIF V study, the point estimate of the relative risk was 1.39, and the two-sided 95% confidence interval for the relative risk was (0.91, 2.12). Thus, in this example, the noninferiority of ximelegatran to warfarin is not demonstrated because the upper limit (2.12) is greater than M 2 (=1.378). Indeed, it does not even demonstrate that M 1 (=1.898) has been excluded. 39

135 Example 1(B): Application of the Synthesis Method to Example 1(A) This example demonstrates the following: The critical features of the synthesis method for demonstrating non-inferiority of a new anticoagulant The calculations and sources of statistical variability that are incorporated in the synthesis approach The main differences in interpretation of results from the fixed margin and the synthesis approaches when applied to the same set of studies and data In this example, we illustrate the synthesis method using the same data as Example 1(A), which consist of six studies comparing warfarin to placebo and one NI study comparing ximelegatran to warfarin. In contrast to the fixed margin method in Example 1(A), the synthesis method does not use a separate 95% confidence interval for this historical estimate of the effect of warfarin versus placebo and for the comparison in the NI study. Rather, the synthesis method is constructed to address the questions of whether ximelegatran preserves a specified percent, in this case 50% of the effect of warfarin, and whether ximelegatran would be superior to a placebo, if placebo had been included as a randomized treatment group in the NI study. To accomplish this goal, the synthesis method makes a comparison of the effect of ximelegatran in the NI study to historical placebo data, an indirect comparison that is not based upon a randomized concurrent placebo group. The synthesis method combines the data from the placebo-controlled studies of warfarin with the data from the NI study in such a way that a test of hypothesis is made to demonstrate that a certain percent of the effect of warfarin is retained in the NI study. A critical point distinguishing the synthesis method from the fixed margin method is that the effect (M 1 ) of warfarin is not specified in advance. But to carry out the analysis, an assumption needs to be made regarding the placebo comparison, namely, that the difference between control drug and placebo (had there been one) in the NI trial is the same as that seen in the historical placebo-controlled trials of warfarin. The assumption is needed because there is no randomized comparison of warfarin and placebo in the NI trial. As a point of reference, we know from example 1(A) that the warfarin effect M 1 was estimated from the historical placebo-controlled studies to be a 47% risk reduction. In this case, the synthesis method statistically tests the null hypothesis that the inferiority of ximelegatran compared to warfarin is less than 50% or one half of the risk reduction of warfarin compared to placebo, a question that the fixed margin method does not directly address because in the fixed margin method, the placebo is only present in the historical studies and not in the NI study. We carry out this test on the log relative risk scale, so that the null hypothesis can be written as: 40

136 H 0 : {log-relative risk of ximelegatran versus warfarin} ½ {log-relative risk of warfarin versus placebo} A test of this hypothesis is performed by the expression below (the statistical test) that has the form of a quotient where the numerator is an estimate of the parameter defined in the null hypothesis by {log-relative risk of ximelegatran versus warfarin} + ½ {log-relative risk of warfarin versus placebo} and the denominator is an estimate of the standard error of the numerator. In this case, the estimated log-relative risk of ximelegatran versus warfarin is (log of 1.39) with a standard error of while the estimated log-relative risk of warfarin versus placebo is 1.02 (log of.361) with a standard error of These estimates are combined, and the synthesis test statistic is calculated as: ( 1.02) [ 1 ( 0.154) ] = Assuming the statistic is normally distributed, it is then compared to 1.96 (for a one-sided Type I error rate of 0.025). For this case, the value, 0.789, is not less (more negative) than 1.96, so we cannot reject the null hypothesis. Therefore, it cannot be concluded that an NI margin of 50% retention is satisfied using the synthesis method. The fixed margin test statistic for this example is: ( 1.02) ( 0.154) = This value, , is also greater than Neither method leads to a conclusion of noninferiority for this example. 41

137 Example 2: Aspirin to Prevent Death or Death/MI After Myocardial Infarction This example demonstrates the following: When it may not be possible to determine the NI margin because of the limitations of the data available By 1993, the effect of aspirin in preventing death after myocardial infarction had been studied in six large randomized placebo-controlled clinical trials. A seventh trial, ISIS-2, gave the drug during the first day after the AMI and is not included because it addressed a different question. The results are summarized and presented in Table 3. Table 3. Results of Six Placebo-Controlled Randomized Studies of the Effect of Aspirin in Preventing Death After Myocardial Infarction Year Study published Aspirin Placebo Relative Risk (95% CI) N Death rate N Death rate MRC % % 0.74 (0.52, 1.05) CDP % % 0.70 (0.48, 1.01) MRC % % 0.83 (0.65, 1.05) GASP % % 0.82 (0.53, 1.28) PARIS % % 0.82 (0.59, 1.13) AMIS % % 1.12 (0.94, 1.33) The results suggest: 1. The effect of aspirin on mortality as measured by the relative risk appears to diminish over the years in which the studies were conducted. 2. The largest trial, AMIS, showed a numerically higher mortality rate in the aspirintreated group. The relative risk in the AMIS study is significantly different from the mean relative risk in the remaining studies (p 0.005). The validity of pooling the results of AMIS with those of the remaining studies is therefore a concern. It would be invalid to exclude AMIS from the metaanalyses because the effect in AMIS differed from the effect in the remaining studies unless there were adequate clinical or scientific reasons for such exclusion. At a minimum, any metaanalysis of all studies would need to reflect this heterogeneity by using a random effects metaanalysis. Although a fixed effects meta-analysis of the six studies gives a point estimate of 0.91 (95% confidence interval 0.82 to 1.02), the random-effects analysis gives a point estimate of 0.86 with 95% confidence interval (0.69, 1.08). The effect of aspirin on prevention of death after myocardial infarction in these historical studies is thus inconclusive (i.e., the upper bound of the 95% confidence interval for effect is > 1.0). Therefore, it is not possible to select aspirin as the active control for evaluating the mortality effect of a test drug in an NI trial. Even if the upper 42

138 bound excluded 1.0, it would be difficult to rely on an NI margin that is not supported by AMIS, the largest of the six trials. The same six studies can also be examined for the combined endpoint of death plus AMI in patients with recent AMI. This endpoint reflects the current physician-directed claim for aspirin based on the positive finding in two studies (MRC-2, PARIS). Table 4. Results of Six Placebo-Controlled Randomized Studies of the Effect of Aspirin in Secondary Prevention of Death or MI After Myocardial Infarction Year Aspirin Placebo Relative Risk (95% Study published CI) N Event rate* N Event rate* MRC % % 0.75 (0.55, 1.03) CDP % % 0.76 (0.57, 1.02) MRC % % 0.72 (0.59, 0.88) GASP % % 0.78 (0.54, 1.12) PARIS % % 0.77 (0.61, 0.97) AMIS % % 0.97 (0.86, 1.09) *The event rates of both groups need further verification from each article. The results indicate that the effect of aspirin on death or MI after myocardial infarction is small to absent in the latest trial (AMIS). Random effects meta-analyses give point estimates of the relative risk of approximately 0.8 with 95% confidence interval upper bounds of approximately 1.0. The NI margin based on these six studies is close to zero (without reducing it further to represent M 2 ) and is so small that an NI trial would be unrealistically large. Again, as with the mortality endpoint, it would be troubling even to consider an NI approach when the largest and most recent trial showed no significant effect. 43

139 Example 3: Xeloda to Treat Metastatic Colorectal Cancer the Synthesis Method This example of Xeloda for first-line treatment of metastatic colorectal cancer illustrates: The use of the synthesis method to demonstrate a loss of no more than 50% of the historical control treatment s effect and a relaxation of this criterion when two NI studies are available The use of additional endpoints in the decision-making process The use of a conservative estimate of the active control effect because (1) a subset of the available studies to estimate the margin was selected and (2) the effect was measured relative to a previous standard of care instead of placebo The U.S. regulatory standard for first-line treatment of metastatic colorectal cancer, the use sought for Xeloda, is the demonstration of improvement in overall survival. Two separate clinical trials, each using an NI study design, compared Xeloda to a Mayo Clinic regimen of 5- fluorouracil with leucovorin (5-FU+LV), the standard of care at the time. Xeloda is an oral fluoropyrimidine, while 5-fluorouracil (5-FU) is an infusional fluoropyrimidine. By itself, bolus 5-FU had not demonstrated a survival advantage in first-line treatment of metastatic colorectal cancer. But with the addition of leucovorin to bolus 5-FU, the combination had demonstrated improved survival. A systematic evaluation of approximately 30 studies that investigated the effect of adding leucovorin to a regimen of 5-FU identified 10 clinical trials that compared a regimen of 5-FU+LV similar to the Mayo Clinic regimen to 5-FU alone, thereby providing a measure of the effect of LV added to 5-FU a conservative estimate of the overall effect of 5-FU+LV as it is likely 5-FU has some effect. Table 5 summarizes the overall survival results, using the metric log hazard ratio for the 10 studies identified that addressed the comparison of interest. Table 5. Selected Studies Comparing 5-FU to 5-FU+LV Study Hazard Ratio 1 Log Hazard Ratio 1 Standard Error Historical Study Historical Study Historical Study Historical Study Historical Study Historical Study Historical Study Historical Study Historical Study Historical Study All log hazard ratios are 5-FU/5-FU+LV A random effects model applied to the survival results of these 10 studies yielded the historical estimate of the 5-FU versus 5-FU+LV survival comparison of a hazard ratio of with a 95% 44

140 confidence interval of (1.09, 1.46) and a log hazard ratio of The NI margin is therefore 1.09 for a fixed margin approach ruling out M 1. A summary of the survival results based on the intent-to-treat populations for each of the two Xeloda NI trials is presented in Table 6. Study 2 rules out M 1 using a fixed margin approach, but Study 1 does not. Table 6. Summary of the Survival Results Study Hazard Ratio 1 Log Hazard Ratio 1 Standard Error 95% confidence interval for the Hazard Ratio 1 NI Study (0.84, 1.18) NI Study (0.78, 1.09) 1 Hazard ratios and log hazard ratios are Xeloda/5-FU+LV The clinical choice of how much of the effect on survival of 5-FU+LV should be retained by Xeloda was determined to be 50%. The synthesis approach was used to analyze whether the NI criteria of 50% loss was met. This synthesis approach for each study combines the results of that NI study with the results from the random effects meta-analysis to produce a normalized test statistic. Based on this NI synthesis test procedure, NI Study 1 failed to demonstrate that Xeloda retained at least 50% of the historical effect of 5-FU+LV versus 5-FU on overall survival, but NI Study 2 did demonstrate such an effect. It was then decided to determine what percent retention was demonstrated. By adapting the synthesis test procedure for retention of an arbitrary percent of the 5-FU+LV historical effect, it was determined that NI Study 1 demonstrated that Xeloda lost no more than 90% of the historical effect of 5-FU+LV on overall survival and that NI Study 2 demonstrated no more than a 39% loss of the historical effect. The evidence of effectiveness of Xeloda was supported by the observation that the tumor response rates were statistically significantly greater for the Xeloda arm and the fact that Xeloda and 5-FU were structurally and pharmacologically very similar. 45

141 Example 4: Determination of an NI margin for Community-Acquired Bacterial Pneumonia (CABP) When No Historical Trials Are Available This example illustrates the following points: The use of risk differences as the metric for estimated treatment effects The determination of the NI margin using observational data when no randomized, placebo-controlled trials of active control drugs are available Community-acquired bacterial pneumonia (CABP) is an acute bacterial infection of the pulmonary parenchyma associated with chest pain, cough, sputum production, difficulty breathing, chills, rigors, fever, or hypotension, and is accompanied by the presence of a new lobar or multi-lobar infiltrate on a chest radiograph. FDA issued revised draft guidance for developing treatments of CAPB in 2014, and the appendix in that guidance describes in detail the justification for NI margins with respect to two endpoints: clinical response and mortality. 3 Historical data from nonrandomized studies of bacteremic and nonbacteremic patients with pneumococcal or lobar pneumonia were evaluated to justify NI margins for use in future CABP studies. For the clinical response endpoint, historical data are relied upon not only to determine an acceptable NI margin but also to decide the time point following treatment at which noninferiority should be assessed. For the mortality endpoint, margin determination is more straightforward and provided here for illustration. An area of uncertainty in evaluating historical data is the spectrum of bacterial pathogens that cause CABP today. In most of the historical studies, CABP was considered synonymous with pneumococcal pneumonia because S. pneumoniae was regularly identified. CABP is also caused by other pathogens such as H. influenzae, H. parainfluenzae, S. aureus, and M. catarrhalis, as well as atypical bacteria such as M. pneumoniae, C. pneumoniae, and Legionella species. A fundamental assumption is that historical response rates in infections such as S. pneumoniae CABP are relevant to response rates in modern infections with sensitive organisms. Table 7 provides an overview of the historical studies available for estimation of mortality rates among both treated and untreated patients with pneumococcal pneumonia. In all of the studies, a significant mortality benefit was shown for patients treated with antibacterial drugs (including sulfonamides, penicillin, and tetracyclines) compared to patients who received no specific therapy (untreated). 3 See the revised draft guidance for industry Community-Acquired Bacterial Pneumonia: Developing Drugs for Treatment. When final, this guidance will represent the FDA s current thinking on this topic. 46

142 Table 7. Mortality in Observational Studies of Pneumococcal Pneumonia Publication Population Mortality (%): Untreated Patients (Study Years) Mortality (%): Antibacterial- Treated Patients (Study Years) Treatment Difference Untreated-Treated (95% CI) Finland (1943) Dowling and Lepper (1951) Austrian and Gold (1964) * Historical control patients 12 years old bacteremic and nonbacteremic 10 years old bacteremic and nonbacteremic 12 years old bacteremic N=2,832 ( )* 41% N=1,087 (1939, 1940)* 30.5% N=17 ( ) 82% N=1,220 ( ) 17% (sulfonamides) N=1,274 ( ) 12.3% (sulfonamides) N=920 ( ) 5.1% (penicillins and tetracyclines) N=437 ( ) 17% 24% (21%, 27%) 18.2% (15%, 21%) 25.4% (22%, 28%) 65% (41%, 79%) There are two limitations in the use of these data to determine an NI margin. First, only observational data are presented, and second, no recent studies were available, with the most recent dating to the 1960s. Significant improvements in the standard of care and the availability of better treatment options for patients with CABP compared to the preantibiotic era bring into question the relevance of these historical studies in estimating an active control effect with respect to mortality. In spite of these limitations, the mortality rates among the treated patient cohorts are reasonably consistent, ranging from 5% to 17% across the three decades represented. Mortality rates in the untreated cohorts are more variable. The older references provide estimates of 31% and 41%, while the more recent study gives an estimated mortality rate of 82%, but this estimate is based on a very small sample size (n=17). In the absence of any placebo-controlled trials of active control drugs in this disease area, it is not possible to estimate M 1 using the methods advocated in this guidance document (e.g., the 95%-95% method), but it is clear that the untreated mortality is substantial. Thus, based on the historical evidence described above, with the caveats noted about the nature and age of the studies, it is reasonable to assume that the mortality rate due to CABP, if left untreated, will be substantially higher than the rates observed among the treated cohorts. Use of an NI margin (M 2 ) of approximately 10% is therefore a valid approach for evaluating new treatments of CABP and would clearly represent an effect superior to no treatment as well as, based on clinical judgment, an appropriate clinical margin. 47

143 REFERENCES FOR EXAMPLES Example 1(A and B) The Boston Area Anticoagulation Trial for Atrial Fibrillation Investigators (1990). The Effect of Low-Dose Warfarin on the Risk of Stroke in Patients with Nonrheumatic Atrial Fibrillation. New Engl J Med 323, Connolly, S.J., Laupacis, A., Gent, M., Roberts, R.S., Cairns, J.A., Joyner, C. (1991). Canadian Atrial Fibrillation Anticoagulation (CAFA) Study. J Am Coll Cardiol 18, EAFT (European Atrial Fibrillation Trial) Study Group (1993). Secondary Prevention in Non- Rheumatic Atrial Fibrillation After Transient Ischemic Attack or Minor Stroke. Lancet 342, Ezekowitz, M.D., Bridgers, S.L., James, K.E., Carliner, N.H., et al. (1992). Warfarin in the Prevention of Stroke Associated with Nonrheumatic Atrial Fibrillation. N Engl J Med 327, Food and Drug Administration, Dockets home page. Available at: Halperin, J.L., Executive Steering Committee, SPORTIF III and V Study Investigators (2003). Ximelagatran Compared with Warfarin for Prevention of Thromboembolism in Patients with Nonvalvular Atrial Fibrillation: Rationale, Objectives, and Design of a Pair of Clinical Studies and Baseline Patient Characteristics (SPORTIF III and V). Am Heart J 146, Jackson, K., Gersh, B.J., Stockbridge, N., Fleming, T.R., Temple, R., Califf, R.M., Connolly, S.J., Wallentin, L., Granger, C.B. (2005). Participants in the Duke Clinical Research Institute/American Heart Journal Expert Meeting on Antithrombotic Drug Development for Atrial Fibrillation (2008). Antithrombotic Drug Development for Atrial Fibrillation: Proceedings. Washington, D.C., July 25-27, American Heart Journal 155, Petersen, P., Boysen, G., Godtfredsen, J., Andersen, E.D., Andersen, B. (1989). Placebo-controlled, Randomised Trial of Warfarin and Aspirin for Prevention of Thromboembolic Complications in Chronic Atrial Fibrillation. The Lancet 338, Stroke Prevention in Atrial Fibrillation Investigators (1991). Stroke Prevention in Atrial Fibrillation Study: Final Results. Circulation 84, Example 2 Aspirin Myocardial Infarction Study Research Group (1980). A Randomized Controlled Trial of Aspirin in Persons Recovered from Myocardial Infarction. JAMA 243,

144 Breddin, K., Loew, D., Lechner, K., Uberia, E.W. (1979). Secondary Prevention of Myocardial Infarction. Comparison of Acetylsalicylic Acid, Phenprocoumon and Placebo. A Multicenter Two-Year Prospective Study. Thrombosis and Haemostasis 41, Coronary Drug Project Group (1976). Aspirin in Coronary Heart Disease. Journal of Chronic Disease 29, Elwood, P.C., Cochrane, A.L., Burr, M.L., Sweetnam, P.M., Williams, G., Welsby, E., Hughes, S.J., Renton, R. (1974). A Randomized Controlled Trial of Acetyl Salicylic Acid in the Secondary Prevention of Mortality from Myocardial Infarction. British Medical Journal 1, Elwood, P.C., Sweetnam, P.M. (1979). Aspirin and Secondary Mortality After Myocardial Infarction. Lancet ii, Fleiss, J.L. (1993). The Statistical Basis of Meta-Analysis. Statistical Methods in Medical Research 2, ISIS-2 Collaborative Group (1988). Randomized Trial of Intravenous Streptokinase, Oral Aspirin, Both, or Neither Among Cases of Suspected Acute Myocardial Infarction: ISIS- 2. Lancet 2, Persantine-Aspirin Reinfarction Study Research Group (1980). Persantine and Aspirin in Coronary Heart Disease. Circulation 62, Example 3 FDA Guidance for Industry: Oncologic Drugs Advisory Committee Discussion on FDA Requirements for the Approval of New Drugs for Treatment of Colon and Rectal Cancer. FDA Medical-Statistical Review for Xeloda (NDA ) dated April 23, Example 4 FDA Draft Guidance for Industry (Revision 2): Community-Acquired Bacterial Pneumonia: Developing Antimicrobial Drugs for Treatment, January Singer, M., Nambiar, S., Valappil, T., Higgins, K., and Gitterman, S. (2008). Historical and Regulatory Perspectives on the Treatment Effect of Antibacterial Drugs for Community-Acquired Pneumonia. Clin Infect Dis 47 (Suppl 3), S216-S224. Finland, M. (1943). Chemotherapy in the Bacteremia. Conn State Med J 7, Dowling, H.G., and Lepper, M.H. (1951). The Effect of Antibiotics (Penicillin, Aureomycin and Terramycin) on the Fatality Rate and Incidence of Complications in Pneumococcic Pneumonia: A Comparison With Other Methods of Therapy. AM J Med Sci 222,

145 Austrian, R., and Gold, J. (1964). Pneumococcal Bacteremia With Especial Reference to Bacteremic Pneumococcal Pneumonia. Ann Intern Med 60,

146 REFERENCES General Blackwelder, W.C. (1982). Proving the Null Hypothesis in Clinical Trials. Controlled Clinical Trials 3, Blackwelder, W.C. (2002). Showing a Treatment Is Good Because It Is Not Bad: When Does Noninferiority Imply Effectiveness? Control Clinical Trials 23, Chow, S.C., Shao, J. (2006). On Non-Inferiority Margin and Statistical Tests in Active Control Trial. Statistics in Medicine 25, D Agostino, R.B., Campbell, M., Greenhouse, J. (2005). Non-Inferiority Trials: Continued Advancements in Concepts and Methodology. Statistics in Medicine 25, Fleming, T.R. (2008). Current Issues in Non-inferiority Trials. Statistics in Medicine 27, Fleming, T.R. (2011). Some Essential Considerations in the Design and Conduct of Noninferiority Trials. Clinical Trials 8, Hauschke, D., Pigeot, I. (2005). Establishing Efficacy of a New Experimental Treatment in the Gold Standard Design (with discussions). Biometrical Journal 47, Koch, G.G. (2008). Comments on Current Issues in Non-Inferiority Trials. Statistics in Medicine 27, Peterson, P., Carroll, K., Chuang-Stein, C., Ho, Y-Y., Jiang, Q., Gang, L., Sanchez, M., Sax, R., Wang, Y-C., Snapinn, S. (2010). PISC Expert Team White Paper: Toward a Consistent Standard of Evidence When Evaluating the Efficacy of an Experimental Treatment From a Randomized Active-Controlled Trial. Statistics in Biopharmaceutical Research 2, Rothmann, M., Li, N., Chen, G., Chi, G.Y.H., Temple, R.T., Tsou, H.H. (2003). Noninferiority Methods for Mortality Trials. Statistics in Medicine 22, Snapinn, S.M., Jiang, Q. (2013). Remaining Challenges in Assessing Non-inferiority. Therapeutic Innovation & Regulatory Science 48(1), Regulatory Committee for Proprietary Medicinal Products (CPMP)(2001). Points to Consider on Switching between Superiority and Non-Inferiority. Br J Clin Pharmacol 52, Committee for Medicinal Products for Human Use (CHMP) (2006). Guideline on the Choice of the Non-Inferiority Margin. Statistics in Medicine 25,

147 Ellenberg, S.S., Temple, R. (2000). Placebo-Controlled Trials and Active-Control Trials in the Evaluation of New Treatments - Part 2: Practical Issues and Specific Cases. Annals of Internal Medicine 133, Huitfeldt, B., Hummel, J., on behalf of European Federation of Statisticians in the Pharmaceutical Industry (EFSPI) (2011). The Draft FDA Guideline on Non-Inferiority Clinical Trials: A Critical Review From European Pharmaceutical Industry Statisticians. Pharmaceutical Statistics 10, Hung, H.M.J., Wang, S.J., O Neill, R.T. (2005). A Regulatory Perspective on Choice of Margin and Statistical Inference Issue in Non-Inferiority Trials. Biometrical Journal 47, International Conference on Harmonisation: Statistical Principles for Clinical Trials (ICH E-9), Food and Drug Administration, DHHS, International Conference on Harmonisation: Choice of Control Group and Related Issues in Clinical Trials (ICH E10), Food and Drug Administration, DHHS, July Temple, R., Ellenberg, S.S. (2000). Placebo-Controlled Trials and Active-Control Trials in the Evaluation of New Treatments - Part 1: Ethical and Scientific Issues. Annals of Internal Medicine 133, Synthesis Methods Brittain, E.H., Fay, M.P., Follmann, D.A. (2012). A valid formulation of the analysis of noninferiority trials under random effects meta-analysis. Biostatistics 13, Hasselblad, V., Kong, D.F. (2001). Statistical Methods for Comparison to Placebo in Active- Control Trials. Drug Information Journal 35, Rothmann, M.D., Wiens, B.L., and Chan, S.F. (2012). Design and Analysis of Non-Inferiority Trials (Chapter 5.3). Boca Raton, FL:Taylor and Francis Group. Meta-analysis DerSimonian, R., Laird, N. (1986). Meta-Analysis in Clinical Trials. Controlled Clinical Trials 7, Follmann, D.A., Proschan, M.A. (1999). Validity Inference in Random-Effects Meta- Analysis. Biometrics 55, Muthukumarna, S. and Tiwari, R.C. (2012). Meta-analysis Using Dirichlet Process. Statistical Methods in Medical Research D0I: / Bayesian Approaches Gamalo, M., Tiwari, R.C., LaVange, L.M. (2014). Bayesian Approach to the Design and Analysis of Non-inferiority Trials for Anti-infective Products. Pharmaceutical Statistics 13(1),

148 Simon, R. (1999). Bayesian Design and Analysis of Active Control Clinical Trials. Biometrics 55, Study Quality and Analysis Population Brittain, E., Lin, D. (2005). A Comparison of Intent-to-Treat and Per Protocol Results in Antibiotic Non-Inferiority Trials. Statistics in Medicine 24, Sanchez, M.M., Chen, X. (2006). Choosing the Analysis Population in Non-Inferiority Studies: Per Protocol or Intent-to-Treat. Statistics in Medicine 25, In-Text Examples Retavase (reteplase) label available on the Internet at: Scandinavian Simvastatin Survival Study Group (1994). Randomized Trial of Cholesterol Lowering in 4444 Patients with Coronary Heart Disease: The Scandinavian Simvastatin Survival Study (4S). Lancet 344, Shepherd J., Cobbe S.M., Ford L., Isles C.G., Lorimer A.R., Macfarlane P.W., McKillop J.H., Packard C.J., for the West of Scotland Coronary Prevention Group (1995). Prevention of Coronary Heart Disease with Pravastatin in Men with Hypercholesterolemia. N Engl J Med 333, Downs J.R., Clearfield M., Weis S.,Whitney E., Shapiro D.R., Beere P.A., Langendorfer A., Stein E.A., Kruyer W., Gotto A.M., for the AFCAPS/TexCAPS Research Group (1998). Primary Prevention of Acute Coronary Events with Lovastatin in Men and Women with Average Cholesterol Levels. JAMA 279(20),

149 Nonproprietary Naming of Biological Products Guidance for Industry U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER) Center for Biologics Evaluation and Research (CBER) January 2017 Labeling OMB control number XXXX-XXXX Expiration Date: xx/xx/xxxx The information collection provisions in this guidance regarding submission of proposed suffixes are under OMB review and are not for current implementation. See additional PRA statement in section VII of this guidance.

150 Nonproprietary Naming of Biological Products Guidance for Industry Additional copies are available from: Office of Communications, Division of Drug Information Center for Drug Evaluation and Research Food and Drug Administration New Hampshire Ave., Hillandale Bldg., 4 th Floor Silver Spring, MD Phone: or ; Fax: druginfo@fda.hhs.gov and/or Office of Communication, Outreach and Development Center for Biologics Evaluation and Research Food and Drug Administration New Hampshire Ave., Bldg. 71, Room 3128 Silver Spring, MD Phone: or ocod@fda.hhs.gov U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER) Center for Biologics Evaluation and Research (CBER) January 2017 Labeling

151 Contains Nonbinding Recommendations TABLE OF CONTENTS I. INTRODUCTION... 1 II. SCOPE... 2 III. BACKGROUND... 3 A. The Biologics Price Competition and Innovation Act of B. Evaluation of the Appropriate Naming Convention... 3 IV. CONSIDERATIONS FOR NONPROPRIETARY NAMING OF ORIGINATOR BIOLOGICAL PRODUCTS, RELATED BIOLOGICAL PRODUCTS, AND BIOSIMILAR PRODUCTS... 4 A. Enhancing Biological Product Pharmacovigilance... 4 B. Ensuring Safe Use for Biological Products... 5 C. Advancing Appropriate Practices and Perceptions Regarding Biological Products... 6 D. Prospective and Retrospective Application of Naming Convention... 7 V. FRAMEWORK FOR DESIGNATING THE PROPER NAME OF A BIOLOGICAL PRODUCT... 7 A. Prospective Naming of Biological Products Submitted Under Section 351(a) of the PHS Act... 9 B. Retrospective Naming of Biological Products Licensed Under Section 351(a) of the PHS Act... 9 C. Naming of Biosimilar Products Submitted Under Section 351(k) of the PHS Act... 9 VI. PROPOSING A SUFFIX FOR THE PROPER NAME OF AN ORIGINATOR BIOLOGICAL PRODUCT, A RELATED BIOLOGICAL PRODUCT, OR A BIOSIMILAR PRODUCT VII. PAPERWORK REDUCTION ACT OF GLOSSARY i

152 Contains Nonbinding Recommendations Nonproprietary Naming of Biological Products Guidance for Industry 1 This guidance represents the current thinking of the Food and Drug Administration (FDA or Agency) on this topic. It does not establish any rights for any person and is not binding on FDA or the public. You can use an alternative approach if it satisfies the requirements of the applicable statutes and regulations. To discuss an alternative approach, contact the FDA office responsible for this guidance as listed on the title page. I. INTRODUCTION This guidance describes FDA s current thinking on the need for biological products licensed under the Public Health Service Act (PHS Act) to bear a nonproprietary name 2 that includes an FDA-designated suffix. Under this naming convention, the nonproprietary name designated for each originator biological product, related biological product, and biosimilar product will be a proper name that is a combination of the core name and a distinguishing suffix that is devoid of meaning and composed of four lowercase letters. 3 The suffix format described in this guidance is applicable to originator biological products, related biological products, and biosimilar products previously licensed and newly licensed under section 351(a) or 351(k) of the PHS Act. FDA is continuing to consider the appropriate suffix format for interchangeable products. This naming convention will facilitate pharmacovigilance for originator biological products, related biological products, and biosimilar products containing related drug substances when other means to track a specific dispensed product are not readily accessible or available, as described in section IV.A of this guidance. Distinguishable nonproprietary names will also facilitate accurate identification of these biological products by health care practitioners and patients. Further, distinguishing suffixes should help minimize inadvertent substitution of any such products that have not been determined to be interchangeable. Application of the naming convention to biological products licensed under the PHS Act should (1) encourage routine use of designated suffixes in ordering, prescribing, dispensing, recordkeeping, and pharmacovigilance practices and (2) avoid inaccurate perceptions of the safety and effectiveness of biological products based on their licensure pathway, as described in detail in this guidance. 1 This guidance has been prepared by the Office of Medical Policy in the Center for Drug Evaluation and Research in cooperation with the Center for Biologics Evaluation and Research at the Food and Drug Administration. 2 See the Glossary for definitions and usage of specific terms used throughout this guidance. 3 The nonproprietary name designated by FDA in the license for a biological product licensed under the PHS Act is its proper name (section 351(a)(1)(B)(i) of the PHS Act (42 U.S.C. 262(a)(1)(B)(i)) and 600.3(k) (21 CFR 600.3(k)). 1

153 Contains Nonbinding Recommendations This guidance provides information to industry, the health care community, other regulatory agencies, and the public on FDA s rationale for this naming convention. It is also intended to assist applicants and license holders in proposing the suffix to be incorporated in the nonproprietary name (referred to throughout this guidance as the proper name) for an originator biological product, a related biological product, or a biosimilar product. In general, FDA s guidance documents do not establish legally enforceable responsibilities. Instead, guidances describe the Agency s current thinking on a topic and should be viewed only as recommendations, unless specific regulatory or statutory requirements are cited. The use of the word should in Agency guidances means that something is suggested or recommended, but not required. II. SCOPE This guidance describes FDA s approach to designating the proper name for originator and related biological products licensed under section 351(a) of the PHS Act and for biosimilar products licensed under section 351(k) of the PHS Act. FDA intends to apply a naming convention to interchangeable products that will feature a core name and a suffix included in the proper name; however, FDA is continuing to consider the appropriate format of the suffix for these products. FDA intends to apply the naming convention discussed in this guidance to both newly licensed and previously licensed biological products. As discussed further in section V of this guidance, the revised proper name of biological products previously licensed under the PHS Act generally would include the product s original proper name serving as the core name plus the distinguishing suffix attached with a hyphen. FDA is continuing to consider the process for implementation of this naming convention for previously licensed products but, in the near term, intends to assign distinguishing suffixes to a limited group of these products 4 and also will accept submissions of prior approval labeling supplements that include proposed suffixes. This guidance also will apply to those biological products that are approved under the Federal Food, Drug, and Cosmetic Act (FD&C Act) on or before March 23, 2020, when such products are deemed to be licensed under section 351 of the PHS Act on March 23, 2020 (section 7002(e)(2) through (e)(4) of the Biologics Price Competition and Innovation Act of 2009 (BPCI Act)). 5 FDA intends to provide additional guidance regarding administrative issues associated with the transition (including the process for implementing the naming convention described in this guidance). 4 The Agency published a proposed rule in the Federal Register of August 28, 2015 (80 FR 52224) ( Designation of Official Names and Proper Names for Certain Biological Products ). 5 See the draft guidance for industry Implementation of the Deemed to be a License Provision of the Biologics Price Competition and Innovation Act of When final, this guidance will represent FDA s current thinking on this topic. For the most recent version of a guidance, check the FDA Drugs guidance Web page at 2

154 Contains Nonbinding Recommendations For the purposes of this document, unless otherwise specified, references to biological products include biological products licensed under the PHS Act, such as therapeutic protein products, vaccines, allergenic products, and blood derivatives, and do not include certain biological products that also meet the definition of a device in section 201(h) of the FD&C Act (21 U.S.C. 321(h)), such as in vitro reagents (e.g., antibody to hepatitis B surface antigen, blood grouping reagents, hepatitis C virus encoded antigen) and blood donor screening tests (e.g., HIV and hepatitis C). Also, for the purposes of this document, unless otherwise specified, references to biological products do not include products for which a proper name is provided in the regulations (e.g., 21 CFR part 640) or to certain categories of biological products for which there are well-established, robust identification and tracking systems to ensure safe dispensing practices and optimal pharmacovigilance (e.g., ISBT 128 for cord blood products and blood components). III. BACKGROUND A. The Biologics Price Competition and Innovation Act of 2009 With the passage of the BPCI Act, 6 which established an abbreviated licensure pathway for products demonstrated to be biosimilar to or interchangeable with an FDA-licensed reference product, a growing number of biological products will be entering the marketplace. Section 351(k) of the PHS Act (42 U.S.C. 262(k)), added by the BPCI Act, sets forth the requirements for an application for a proposed biosimilar product and an application or a supplement for a proposed interchangeable product. Section 351(i) defines biosimilarity to mean that the biological product is highly similar to the reference product notwithstanding minor differences in clinically inactive components and that there are no clinically meaningful differences between the biological product and the reference product in terms of the safety, purity, and potency of the product (see section 351(i)(2) of the PHS Act). To meet the additional standard of interchangeability, an applicant must provide sufficient information to demonstrate biosimilarity and also to demonstrate that the biological product can be expected to produce the same clinical result as the reference product in any given patient and, if the biological product is administered more than once to an individual, the risk in terms of safety or diminished efficacy of alternating or switching between the use of the biological product and the reference product is not greater than the risk of using the reference product without such alternation or switch (see section 351(k)(4) of the PHS Act). Interchangeable products may be substituted for the reference product without the intervention of the prescribing health care provider (see section 351(i)(3) of the PHS Act). B. Evaluation of the Appropriate Naming Convention The proper name of a biological product reflects certain scientific characteristics of the product, such as chemical structure and pharmacological properties. This name is different from a 6 Sections 7001 through 7003 of the Patient Protection and Affordable Care Act (Public Law ). 3

155 Contains Nonbinding Recommendations proprietary name, which generally is trademarked and registered for private use. For biological products licensed under the PHS Act, FDA designates the proper name in the license for use upon each package of the biological product (see section 351(a)(1)(B)(i) of the PHS Act and 21 CFR 600.3(k)). Among other things, the proper name of a biological product helps health care providers identify the product s drug substance and distinguish biological products from one another. As part of FDA s implementation of the BPCI Act, the Agency requested public comment on its development of a framework for safe use and optimal pharmacovigilance for biosimilar products and interchangeable products that is informed by current experience and industry best practices, including the role of a product s proper name. FDA has evaluated comments received on the approaches to naming biosimilar products and interchangeable products. 7 In light of the issues considered for biosimilar products and interchangeable products, FDA also evaluated its approach to designating proper names for biological products licensed under section 351(a) of the PHS Act. In implementing the BPCI Act, FDA has carefully considered the appropriate naming convention to maximize the success of biosimilar products and interchangeable products and to help ensure the safety of patients receiving biological products licensed under the PHS Act. IV. CONSIDERATIONS FOR NONPROPRIETARY NAMING OF ORIGINATOR BIOLOGICAL PRODUCTS, RELATED BIOLOGICAL PRODUCTS, AND BIOSIMILAR PRODUCTS This section discusses the main considerations that led FDA to adopt the naming convention described in section V of this guidance. A. Enhancing Biological Product Pharmacovigilance The Agency considers appropriate pharmacovigilance fundamentally important for biological products. Although safety of biological products is rigorously assessed before approval, safety issues that are specific to a manufacturer may arise after approval with any marketed product. To help ensure patient safety and allow the Agency and the manufacturer to swiftly identify and address a problem, FDA aims to track adverse events to a specific manufacturer (and as appropriate, to a lot or manufacturing site for a particular biological product) and allow surveillance systems to detect safety signals throughout the life cycle of a product. Identifying a biological product s manufacturer can help target remedial action (including recall) to avoid implicating a broader set of products for which no such problem exists. 7 See, for example, notices that published in the Federal Register, Approval Pathway for Biosimilar and Interchangeable Biological Products; Public Hearing; Request for Comments (75 FR 61497, October 5, 2010); Draft Guidances Relating to the Development of Biosimilar Products; Public Hearing; Request for Comments (77 FR 12853, March 2, 2012); Nonproprietary Naming of Biological Products; Draft Guidance for Industry; Availability (80 FR 52296, August 28, 2015); and other public dockets established by FDA. 4

156 Contains Nonbinding Recommendations Pharmacovigilance systems, both active and passive, vary in their use of identifiers to differentiate among biological products. These identifiers may include the proprietary name, proper name, manufacturer, national drug code (NDC) number, lot number, and billing codes. However, many active pharmacovigilance systems, which generally identify adverse events by querying privately held electronic health care data such as administrative and billing data, have limited ability to track to its manufacturer a biological product that shares the same proper name with other biological products. Other product identifiers, such as NDC numbers, are not routinely recorded in billing and patient records in many clinical settings in which biological products are dispensed and administered, and therefore the utility of these alternative identifiers in active pharmacovigilance is limited. Similarly, proprietary names and NDC numbers are often not included in adverse event reports. As a result, the use of alternative identifiers, including distinct proprietary names or NDC numbers, is insufficient to address concerns regarding pharmacovigilance. Nonproprietary names that include distinguishing suffixes can serve as a key element to identify specific products in spontaneous adverse event reporting and to reinforce accurate product identification in billing and claims records used for active pharmacovigilance. Other productspecific identifiers, such as proprietary names or NDCs, may not be available or could change over time. A distinguishing suffix will also support the tracking of product-specific events over time, thereby enhancing the accurate attribution of product-specific adverse event reports. 8 The Agency s approach to nonproprietary naming of biological products will provide another critical tool for accurately identifying and facilitating pharmacovigilance for originator biological products, related biological products, and biosimilar products. B. Ensuring Safe Use for Biological Products Biological products generally consist of large, complex molecules and raise unique safety concerns related to immunogenicity. FDA believes the nonproprietary naming convention for originator biological products, related biological products, or biosimilar products should help prevent inadvertent substitution. Inadvertent substitution may lead to unintended alternating or switching of biological products that are not determined by FDA to be interchangeable with each other. This naming convention should facilitate safe use and help to protect the safety of patients. Related biological products may be licensed for different indications. Biosimilar products may be licensed for fewer than all indications for which the reference product is licensed. Likewise, related biological products and biosimilar products may be licensed for fewer than all routes of administration and may be packaged in different delivery systems than those approved for the 8 See the draft guidance for industry Postmarketing Safety Reporting for Human Drugs and Biological Products Including Vaccines. When final, this guidance will represent FDA s current thinking on this topic. 5

157 Contains Nonbinding Recommendations originator biological product. If originator biological products, related biological products, and biosimilar products all share the same proper name, inadvertent substitution may lead to medication errors. For example, a patient could inadvertently receive a product with a different delivery system or route of administration than was prescribed, which may lead to confusion among patients and result in dosing errors. Confusion may also arise among health care providers who, based on their experience with small-molecule drugs and generic versions of those drugs, may incorrectly assume that FDA has determined biological products with the same proper name to be interchangeable. Information on alternating or switching between a proposed product and its reference product is required to support a demonstration of interchangeability, but is not required to support a demonstration of biosimilarity (see section 351(k)(4) of the PHS Act). Applications for related biological products are not required to include any comparative data to any other biological product in support of licensure (see section 351(a) of the PHS Act). Although many biological products may have proprietary names, many health care systems mainly use proper names instead of proprietary names for ordering, prescribing, and dispensing products. The naming convention discussed in this guidance will also facilitate use of the Purple Book 9 for biological products. The Purple Book enables a user to readily see all licensed biological products and identify whether a biological product licensed under section 351(k) of the PHS Act has been determined by FDA to be biosimilar to or interchangeable with a reference product (a previously licensed biological product). Biosimilar products and interchangeable products licensed under section 351(k) of the PHS Act will be listed under the reference product to which biosimilarity or interchangeability was demonstrated. C. Advancing Appropriate Practices and Perceptions Regarding Biological Products With the introduction of more biological products, FDA believes it is important to encourage routine use of designated suffixes in ordering, prescribing, dispensing, recordkeeping, and pharmacovigilance practices for biological products, irrespective of their licensure pathway and date of licensure. The designated suffix will provide a consistent, readily available and recognizable mechanism for patients and health care professionals, including providers and pharmacists, to correctly identify these products. FDA believes it is likely that FDA-designated suffixes will be used routinely when identifying, describing, and recording use of biological products if such suffixes are present in the proper names of all biological products licensed under the PHS Act. 9 FDA published the Purple Book: Lists of Licensed Biological Products With Reference Product Exclusivity and Biosimilarity or Interchangeability Evaluations in September 2014, which is publicly available at erapeuticbiologicapplications/biosimilars/ucm htm. The Purple Book is updated periodically to reflect FDA licensure of a biological product under section 351(a) or section 351(k) of the PHS Act and/or to reflect a determination regarding date of first licensure for a biological product licensed under section 351(a) of the PHS Act. 6

158 Contains Nonbinding Recommendations The inclusion of an FDA-designated suffix in the nonproprietary name of biological products licensed under section 351(a) or 351(k) of the PHS Act should have the added benefit of helping to avoid inaccurate perceptions of the safety and effectiveness of biological products based on their licensure pathway. The safety and effectiveness of biological products is rigorously assessed before approval. Through FDA s implementation of the BPCI Act s standards for biosimilarity and interchangeability, FDA can ensure that the products it determines to be biosimilar to or interchangeable with a reference product can be relied upon by providers and patients to be safe and effective. Applying this naming convention only for products licensed under section 351(k) of the PHS Act but not for the reference product licensed under 351(a) of the PHS Act could adversely affect health care provider and patient perceptions of these new products. Specifically, such an approach could be misinterpreted as indicating that biosimilar products differ from their reference products in a clinically meaningful way or are inferior to their reference products for their approved conditions of use. D. Prospective and Retrospective Application of Naming Convention FDA s current thinking is that a proper name that includes a distinguishing suffix is warranted for both newly licensed and previously licensed originator biological products, related biological products, and biosimilar products. As with prospective application of the naming convention, retrospective application will help (1) prevent a patient from receiving a product different from what was intended to be prescribed; (2) facilitate manufacturer-specific pharmacovigilance by providing a means of determining which biological product is dispensed to patients; (3) encourage routine use of FDA-designated suffixes in ordering, prescribing, dispensing, and recordkeeping practices for these products; and (4) advance accurate perceptions of these biological products. V. FRAMEWORK FOR DESIGNATING THE PROPER NAME OF A BIOLOGICAL PRODUCT FDA s naming convention for biological products licensed under the PHS Act will be a proper name consisting of a core name 10 and an FDA-designated suffix. Proper names designated by FDA for originator biological products, related biological products, and biosimilar products will include a combination of a core name and a distinguishing suffix. For originator biological products, FDA intends to use a core name that is the adopted name designated by the USAN Council 11 for the relevant biological substance when available. If the biological product is a related biological product, a biosimilar product, or an interchangeable 10 Two examples of a core name are filgrastim and epoetin alfa. The proper name for all biological products will include a distinguishing suffix composed of four lowercase letters attached to the core name with a hyphen. 11 The United States Pharmacopeial Convention, 2016, Guiding Principles for Coining United States Adopted Names for Drugs (2016 USP Dictionary of USAN and International Drug Names at 7

159 Contains Nonbinding Recommendations product, the core name will be the same as the core name identified in the proper name of the relevant previously licensed product. 12 A distinguishing suffix that is devoid of meaning and composed of four lowercase letters will be attached with a hyphen to the core name of each originator biological product, related biological product, or biosimilar product. 13 Use of a shared core name will indicate a relationship among products. The placement of the identifier as a suffix, rather than a prefix, should result in biological products with the same core name being grouped together in electronic databases to help health care providers locate and identify these products. 14 To illustrate, the proper names for products sharing the core name replicamab may be displayed as follows: replicamab-cznm replicamab-hjxf To illustrate, the proper names for products sharing the core name putonastim alfa may be displayed as follows: putonastim alfa-jnzt putonastim alfa-kngx In designating proper names for related biological products, the Agency has in some instances designated a proper name that includes an identifier attached as a prefix to distinguish the products from previously licensed biological products; for example, ado-trastuzumab emtansine. In this case, designation of a proper name that includes a unique prefix was necessary to minimize certain medication errors and to facilitate pharmacovigilance. FDA determined that a unique proper name including a prefix was necessary for ado-trastuzumab emtansine to distinguish the product from trastuzumab, a previously licensed biological product submitted in a different BLA. 15 FDA may continue such practices on a limited basis, where appropriate, when the Agency determines that the designation of a prefix, in addition to a suffix as contemplated by this guidance, is necessary to ensure patient safety. 12 FDA will work with stakeholders that play a role in national drug naming and listing to help ensure that the suffixes added to the core name of biological products are recorded appropriately in drug listing systems. 13 FDA determined that a hyphen should separate the shared core name from the suffix. A hyphen is a common punctuation mark used in writing and electronic systems; it is a readily recognized mark. Another punctuation mark, such as an underscore, may not be normally used in handwriting and may not be readily seen in handwriting, electronic systems, or both. 14 The license holder and all distributors of a biological product should use the proper name designated by FDA in the license for that product. 15 As described in the BLA submission for ado-trastuzumab emtansine, medication errors involving administration of the wrong drug (trastuzumab emtansine versus trastuzumab) during clinical trials resulted in serious adverse events. 8

160 Contains Nonbinding Recommendations A. Prospective Naming of Biological Products Submitted Under Section 351(a) of the PHS Act An applicant should propose a suffix composed of four lowercase letters for use as the distinguishing identifier included in the proper name designated by FDA at the time of licensure (see section VI of this guidance). Such submissions can be made during the investigational new drug application (IND) phase 16 or at the time of BLA submission. An applicant should submit up to 10 proposed suffixes, as described in this section, in the order of the applicant s preference. We recommend including any supporting analyses of the proposed suffixes for FDA s consideration based on the factors described in this guidance. B. Retrospective Naming of Biological Products Licensed Under Section 351(a) of the PHS Act A BLA holder may propose a suffix, as described in this guidance, for use in the proper name of currently licensed biological products held by the company by submitting a prior-approval labeling supplement to its BLA (see section VI of this guidance). As part of that labeling supplement, a BLA holder should submit up to 10 proposed suffixes, as described in this section, in the order of the applicant s preference. We recommend including any supporting analyses of the proposed suffixes for FDA s consideration based on the factors described in this guidance. C. Naming of Biosimilar Products Submitted Under Section 351(k) of the PHS Act An applicant for a proposed biosimilar product submitted under section 351(k) of the PHS Act should propose a suffix composed of four lowercase letters for use as the distinguishing identifier included in the proper name designated by FDA at the time of licensure (see section VI of this guidance). Such submissions can be made during the investigational new drug application (IND) phase 17 or at the time of BLA submission. An applicant should submit up to 10 proposed suffixes, as described in this section, in the order of the applicant s preference. We recommend including any supporting analyses of the proposed suffixes for FDA s consideration based on the factors described in this guidance. 16 A request for FDA review of a proposed suffix submitted during the investigational new drug application (IND) phase should be submitted no earlier than at the request for a pre-biologics license application (pre-bla) meeting for biological products to be submitted under section 351(a) of the PHS Act. 17 A request for FDA review of a proposed suffix submitted during the investigational new drug application (IND) phase should be submitted no earlier than at the request for a biosimilar biological product development (BPD) type 4 meeting for biological products to be submitted under section 351(k) of the PHS Act. 9

161 Contains Nonbinding Recommendations VI. PROPOSING A SUFFIX FOR THE PROPER NAME OF AN ORIGINATOR BIOLOGICAL PRODUCT, A RELATED BIOLOGICAL PRODUCT, OR A BIOSIMILAR PRODUCT The proposed suffix should: Be unique Be devoid of meaning Be four lowercase letters of which at least three are distinct Be nonproprietary Be attached to the core name with a hyphen Be free of legal barriers that would restrict its usage The proposed suffix should not: Be false or misleading, such as by making misrepresentations with respect to safety or efficacy Include numerals and other symbols aside from the hyphen attaching the suffix to the core name Include abbreviations commonly used in clinical practice in a manner that may lead the suffix to be misinterpreted as another element on the prescription or order Contain or suggest any drug substance name or core name Look similar to or be capable of being mistaken for the name of a currently marketed product (e.g., should not increase the risk of confusion or medical errors with the product and/or other products in the clinical setting) Look similar to or otherwise connote the name of the license holder Be too similar to any other FDA-designated nonproprietary name suffix FDA encourages applicants to conduct due diligence on their proposed suffixes to ensure that no other restrictions apply to use of the proposed suffix in this context. Any supporting information can be provided to FDA with the submission of the proposed suffix(es). The final determination on the acceptability of a proposed suffix is based on FDA s review of all information and analyses described in this guidance, along with any information submitted by the sponsor. 10

162 Contains Nonbinding Recommendations FDA will evaluate proposed suffixes against the factors described in this section and may consider other factors if they impact the utility of the suffix in meeting the goals of the naming convention articulated in this guidance. Upon completion of the Agency s evaluation, FDA will notify applicants if a proposed suffix is acceptable or if all of the proposed suffixes are determined to be unacceptable. If all of the proposed suffixes are determined to be unacceptable, applicants may submit additional proposed suffixes for FDA s consideration. If an applicant does not submit a suffix that FDA finds acceptable or does not propose suffix candidates within an appropriate time frame to allow sufficient time for FDA review, FDA may elect to assign a four-letter suffix for inclusion in the proper name designated in the license at the time FDA approves the application. VII. PAPERWORK REDUCTION ACT OF 1995 This guidance contains information collection provisions that are subject to review by the Office of Management and Budget (OMB) under the Paperwork Reduction Act of 1995 (44 U.S.C ). Specifically, the guidance recommends that applicants and application holders submit up to 10 proposed suffixes, in the order of the applicant s preference. FDA also recommends including any supporting analyses for FDA s consideration, demonstrating that the proposed suffixes meet the factors described in the final guidance. FDA estimates that the time required to complete this information collection will average 420 hours per response, including the time to review instructions, search existing data sources, gather the data needed, and complete and review the information collection. Send comments regarding this burden estimate or suggestions for reducing this burden to: Office of Medical Policy, Center for Drug Evaluation and Research, Food and Drug Administration, New Hampshire Avenue, Bldg. 51, rm. 6337, Silver Spring, MD This guidance also refers to previously approved collections of information found in FDA regulations. The collections of information related to the submission of a BLA under section 351(k) of the PHS Act have been approved under OMB control number , and the collections of information in 21 CFR part 601 have been approved under OMB control number An Agency may not conduct or sponsor, and a person is not required to respond to, a collection of information unless it displays a currently valid OMB control number. The information collection provisions in this guidance, including resulting proposed modifications to the information collections approved under OMB control number , have been submitted to OMB for review as required by section 3507(d) of the Paperwork Reduction Act of 1995 and are not for current implementation. Before implementing the information collection provisions contained in this guidance, we will publish a notice in the Federal Register announcing OMB's decision to approve, modify, or disapprove those information collection provisions. 11

163 Contains Nonbinding Recommendations GLOSSARY Biosimilar Product means a biological product submitted in a 351(k) application that has been shown to be highly similar to the reference product notwithstanding minor differences in clinically inactive components, and for which there are no clinically meaningful differences between the biological product and the reference product in terms of the safety, purity, and potency of the product (see section 351(i)(2) of the PHS Act). Core Name means the component shared among an originator biological product and any related biological product, biosimilar product, or interchangeable product as part of the proper names of those products. Two examples of a core name are filgrastim and epoetin alfa. Interchangeable Product means a biological product that has been shown to meet the standards described in section 351(k)(4) of the PHS Act and may be substituted for the reference product without the intervention of the health care provider who prescribed the reference product (see section 351(i)(3) of the PHS Act). Nonproprietary Name means a name unprotected by trademark rights that is in the public domain. It may be used by the public at large, both lay and professional. Originator Biological Product means a biological product submitted in a BLA under section 351(a) of the PHS Act (i.e., a stand-alone BLA) that is not a related biological product. Proper Name means the nonproprietary name designated by FDA in the license for a biological product licensed under the PHS Act. 18 Proprietary Name means the trademark or brand name. Reference Product means the single biological product licensed under section 351(a) of the PHS Act against which a biological product is evaluated in a 351(k) application (section 351(i)(4) of the PHS Act). Related Biological Product means a biological product submitted in a BLA under section 351(a) of the PHS Act (i.e., a stand-alone BLA) for which there is a previously licensed biological product submitted in a different section 351(a) BLA that contains a drug substance for which certain nomenclature conventions (e.g., United States Adopted Names (USAN) Guiding Principles 19 ) would be expected to provide for use of the same drug substance name Section 351(a)(1)(B)(i) of the PHS Act (42 U.S.C. 262(a)(1)(B)(i) and 600.3(k)( 21 CFR 600.3(k)). 19 The United States Pharmacopeial Convention, 2016, Guiding Principles for Coining United States Adopted Names for Drugs (2016 USP Dictionary of USAN and International Drug Names at 20 FDA s description of a biological product as a related biological product in this guidance is separate from any determination FDA may make about whether a related biological product is eligible for a period of exclusivity under section 351(k)(7) of the PHS Act. 12

164 Guidance Document Information and Submission Requirements for Biosimilar Biologic Drugs Date Adopted Revised Date Health Products and Food Branch

165 Our mission is to help the people of Canada maintain and improve their health. Health Canada Our mandate is to take an integrated approach to managing the health-related risks and benefits of health products and food by: minimizing health risk factors to Canadians while maximizing the safety provided by the regulatory system for health products and food; and, promoting conditions that enable Canadians to make healthy choices and providing information so that they can make informed decisions about their health. Health Products and Food Branch Her Majesty the Queen in Right of Canada as represented by the Minister of Health, 2016 Available in Canada through Health Canada Publications Address Locator 0900C2 Ottawa, Ontario K1A 0K9 Telephone: , Web address: Également disponible en français sous le titre : Ligne directrice Exigences en matière de renseignements et de présentation relatives aux médicaments biologiques biosimilaires

166 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs Foreword Guidance documents are meant to provide assistance to industry and health care professionals on how to comply with governing statutes and regulations. Guidance documents also provide assistance to staff on how Health Canada mandates and objectives should be implemented in a manner that is fair, consistent, and effective. Guidance documents are administrative instruments not having force of law and, as such, allow for flexibility in approach. Alternate approaches to the principles and practices described in this document may be acceptable provided they are supported by adequate justification. Alternate approaches should be discussed in advance with the relevant programme area to avoid the possible finding that applicable statutory or regulatory requirements have not been met. As a corollary to the above, it is equally important to note that Health Canada reserves the right to request information or material, or define conditions not specifically described in this document, in order to allow the Department to adequately assess the safety, efficacy, or quality of a therapeutic product. Health Canada is committed to ensuring that such requests are justifiable and that decisions are clearly documented. This document should be read in conjunction with the accompanying notice and the relevant sections of other applicable guidance documents. Revised Date: i

167 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs Version Guidance Document: Information and Submission Requirements for Biosimilar Biologic Drugs Document Change Log Replaces Guidance for Sponsors: Information and Submission Requirements for Subsequent Entry Biologics (SEBs) Date November 14, 2016 Date March 5, 2010 Revised Date: ii

168 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs Table of Contents 1. Introduction Objective Scope and application Policy statements Definitions Background Guidance for implementation General Applicable regulations Patents, intellectual property, and data protection Reference biologic drug Review time Information requirements for clinical trial applications (CTA) Information requirements for new drug submissions (NDS) Organization of data Quality information Non-clinical and clinical information Authorization of indications Labelling requirements Product Monograph Risk Management Plan Post-market requirements Adverse drug reaction (ADR) reporting and periodic reports Post-notice of compliance (NOC) changes Consultation with the Biologics and Genetic Therapies Directorate (BGTD) Additional Information Revised Date: iii

169 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs 1. Introduction Health Canada, the federal regulatory authority that evaluates the safety, efficacy, and quality of drugs available in Canada, recognises that with the expiration of patents for biologic drugs, manufacturers may be interested in pursuing subsequent entry versions of these biologic drugs. The term biosimilar biologic drug, hereafter referred to as biosimilar, is used by Health Canada to describe subsequent entry versions of a Canadian approved innovator biologic with demonstrated similarity to a reference biologic drug. Biosimilars were previously referred to as Subsequent Entry Biologics (SEBs) in Canada. 1.1 Objective The objective of this document is to provide guidance to sponsors to enable them to satisfy the information and regulatory requirements under the Food and Drugs Act and Part C of the Food and Drug Regulations for the authorization of biosimilars in Canada. 1.2 Scope and application This guidance document applies to all biologic drug submissions where the sponsor seeks authorization for sale based on demonstrated similarity to a previously approved biologic drug and relies, in part, on prior information regarding that biologic drug in order to present a reduced clinical and non-clinical package as part of the submission. The following criteria determine the scope of products that are eligible to be authorized as biosimilars: a suitable reference biologic drug exists that: a) was originally authorized for sale based on a full quality, non-clinical and clinical data package; and b) has been used in the post market setting such that the demonstration of similarity will bring into relevance a substantial body of reliable data on safety, efficacy and effectiveness the biosimilar and reference biologic drug can be well characterized by a set of modern analytical methods the biosimilar, through extensive characterization and analysis, can be judged similar to the reference biologic drug by meeting an appropriate set of predetermined criteria The demonstration of similarity depends upon detailed and comprehensive product characterization. This guidance applies to biologic drugs that contain, as their active substances, well characterized proteins derived through modern biotechnological methods such as use of recombinant DNA and/or cell culture. Biosimilars employing clearly different approaches to manufacture than the reference biologic drug may be eligible; however, careful consideration should be given to Revised Date:

170 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs expression system differences that may present challenges to demonstrating similarity to the reference biologic drug. In this guidance document, "must" is used to express a requirement that the user is obliged to satisfy in order to comply with the regulatory requirements; "should" is used to express a recommendation or that which is advised but not required; and "may" is used to express an option or that which is permissible within the limits of the guidance document. 1.3 Policy statements The following statements outline the fundamental concepts and principles of the regulatory framework for biosimilars: The sponsor is responsible for providing the necessary evidence to support all aspects of a biosimilar submission A biosimilar sponsor is eligible to apply for the indication(s) and condition(s) of use that are held by the reference biologic drug authorized in Canada Biosimilars are new drugs subject to the Food and Drugs Act and Part C of the Food and Drug Regulations. The concepts and the scientific and regulatory principles within the existing regulatory frameworks for biologic, pharmaceutical, and generic pharmaceutical drugs are used as the basis for the regulatory framework for biosimilars The basis for accepting a reduced non-clinical and clinical data package for a biosimilar hinges on demonstrated similarity between the biosimilar and the suitable reference biologic drug. A final determination of similarity will be based on the entire submission, including data derived from comparative structural, functional, non-clinical and clinical studies Biosimilars are not generic biologics and many characteristics associated with the authorization process and marketed use for generic pharmaceutical drugs do not apply. Authorization of a biosimilar is not a declaration of pharmaceutical equivalence, bioequivalence or clinical equivalence to the reference biologic drug A biosimilar submission involves a comparison to another product. Hence all biosimilars are subject to the laws, and patent and intellectual property principles outlined within the Food and Drug Regulations (Data Protection), Patented Medicines (Notice of Compliance) Regulations, and the Patent Act As a biosimilar is authorized using a reduced non-clincial and clinical package, it should not be used as a reference biologic drug for another biosimilar submission. Revised Date:

171 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs 1.4 Definitions Biologic drug (Médicament biologique) A drug listed in Schedule D to the Food and Drugs Act. Schedule D lists individual products (such as insulin), product classes (such as immunizing agents), references to particular sources (such as drugs, other than antibiotics, prepared from microorganisms ), and methodology (such as drugs obtained by recombinant DNA procedures ). Biologic drugs are derived through the metabolic activity of living organisms and tend to be significantly more variable and structurally complex than chemically synthesized drugs. Biosimilar biologic drug (Médicament biologique biosimilaire) A biologic drug that obtains market authorization subsequent to a version previously authorized in Canada, and with demonstrated similarity to a reference biologic drug. A biosimilar relies in part on prior information regarding safety, efficacy and effectiveness that is deemed relevant due to the demonstration of similarity to the reference biologic drug and which influences the amount and type of original data required. Biosimilar biologic drugs were previously referred to as Subsequent Entry Biologics. Reference biologic drug (Médicament biologique de référence) A biologic drug authorized on the basis of a complete quality, non-clinical, and clinical data package, to which a biosimilar is compared to demonstrate similarity. Specification (Spécification) A specification is defined as a list of tests, references to analytical procedures, and appropriate acceptance criteria which are numerical limits, ranges, or other criteria for the tests described. It establishes the set of criteria to which a drug substance, drug product or materials at other stages of its manufacture should conform to be considered acceptable for its intended use. Conformance to specification means that the drug substance and drug product, when tested according to the listed analytical procedures, will meet the acceptance criteria. Specifications are critical quality standards that are proposed and justified by the manufacturer and approved by regulatory authorities as conditions of approval. Abbreviations and acronyms ADA = Anti-Drug Antibody ADR = Adverse Drug Reaction BGTD = Biologics and Genetic Therapies Directorate CTA = Clinical Trial Application CTD = Common Technical Document ICH = International Council for Harmonisation NDS = New Drug Submission NOC = Notice of Compliance PK/PD = Pharmacokinetic/Pharmacodynamic Revised Date:

172 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs PBRER = Periodic Benefit-Risk Evaluation Reports PSUR = Periodic Safety Update Reports RMP = Risk Management Plan SAM = Scientific Advice Meeting SNDS = Supplemental New Drug Submission 1.5 Background Biologic drugs have contributed to the health of Canadians through their use as treatments in the management of various complex diseases and medical conditions. Unlike pharmaceutical drugs, biologic drugs are derived through the metabolic activity of living organisms and are variable and structurally complex. Biologics tend to be labile and sensitive to changes in manufacturing processes. Biological source materials, production cells, or their fermentation media can present risks, such as the presence of pathogens or the growth of adventitious agents such as viruses. Due to these risks, careful attention is paid to raw material controls, viral/bacterial inactivation or clearance during product purification, and product testing. Changes to source materials, manufacturing processes, equipment, or facilities can result in significant unexpected changes to the intermediate and/or final product. The expiration of patents and/or data protection for some biologic drugs is creating opportunities for subsequent entry versions. The term biosimilar is used by Health Canada to describe a biologic drug that receives market authorization subsequent to a version previously authorized in Canada and with demonstrated similarity to a reference biologic drug. Demonstration of similarity enables the sponsor to rely partially on relevant information about the reference biologic drug and to seek authorization based on a reduced non-clinical and clinical data package tailored to a particular class of products or to a specific case. In order to clearly distinguish between the regulatory process and product characteristics for biosimilars and those for generic pharmaceutical drugs, the terms biogeneric or generic biologic are not used. The Biologics and Genetic Therapies Directorate (BGTD) within the Health Products and Food Branch of Health Canada is the regulator of biologic drugs for human use and provides regulatory oversight for biologics with its comprehensive reviews of biologic drug submissions covering quality, safety and efficacy. 2. Guidance for implementation 2.1 General Applicable regulations Biosimilars, like all new biologic drugs, are subject to Part C of the Food and Drug Regulations for authorization and oversight. Conforming to the guidance provided in this document, along with other guidance for biologics, should enable a sponsor to satisfy the following notable requirements in Part C, Division 8 of the Food and Drug Regulations: Revised Date:

173 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs C (1)(a): No person shall sell or advertise a new drug unless the manufacturer of the new drug has filed with the Minister a new drug submission relating to the new drug that is satisfactory to the Minister. C (2): A new drug submission shall contain sufficient information and material to enable the Minister to assess the safety and effectiveness of the new drug Patents, intellectual property, and data protection All biosimilars enter the market subsequent to a biologic drug product previously authorized in Canada and to which the biosimilar is considered similar. As such, biosimilars are subject to existing laws and regulations outlined in the Patented Medicines (Notice of Compliance) Regulations and C of the Food and Drug Regulations, and related guidance documents entitled, Guidance Document: Data Protection under C of the Food and Drug Regulations and Guidance Document: Patented Medicines (Notice of Compliance) Regulations. In the New Drug Submission (NDS), the biosimilar sponsor must clearly identify the biologic drug authorized in Canada to which it is subsequent and to which it is considered to be making a direct or indirect comparison according to the Patented Medicines (Notice of Compliance) Regulations and C of the Food and Drug Regulations. In addition, where there is a non-canadian reference biologic drug (refer to section 2.1.3), the sponsor must address the relationship between the non-canadian reference biologic drug and the Canadian version Reference biologic drug A biosimilar must be subsequent to a biologic drug that is authorized in Canada and to which a reference is made. Sponsors may use a non-canadian sourced version as a proxy for the Canadian drug in the comparative studies. The onus is on the sponsor to demonstrate that the chosen reference biologic drug is suitable to support the submission. The sponsor should consult with BGTD early in the drug development process to ensure the suitability of the reference biologic drug. The following should be considered when selecting a reference biologic drug: The dosage form(s), strength(s), and route(s) of administration of the biosimilar should be the same as that of the reference biologic drug. The same reference biologic drug should be used throughout the studies supporting the quality, safety and efficacy of the product (i.e. in the developmental programme for the biosimilar). In certain circumstances, it may be possible to use more than one reference biologic drug (e.g. versions of the reference biologic drug sourced from more than one jurisdiction) in clinical studies. As a scientific matter, the type of Revised Date:

174 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs bridging data needed will always include structural and functional data from analytical studies that directly compares all the products (e.g. the proposed biosimilar product, the U.S.-authorized product, and the EU-authorized product) and may also include clinical pharmacokinetic (PK) and, if appropriate, clinical pharmacodynamic (PD) study data for all the products. The active substances (medicinal ingredients) of the biosimilar and the reference biologic drug must be shown to be similar. The reference biologic drug should have accumulated adequate safety, efficacy, and effectiveness data in the post market setting such that the demonstration of similarity will bring into relevance a substantial body of reliable data. A biosimilar should not be used as a reference biologic drug, as it was authorized using a reduced non-clinical and clinical data package Considerations for the use of a non-canadian reference biologic drug In addition to section 2.1.3, the following should be considered when selecting a non-canadian reference biologic drug used for the purposes of demonstrating similarity: The non-canadian reference biologic drug should have the same medicinal ingredient(s), dosage form and route of administration as the version authorized in Canada. Information on the Canadian version can be found in Health Canada s Drug Product Database. The non-canadian reference biologic drug should be marketed in a jurisdiction that formally adopts International Council for Harmonization (ICH) guidelines and that has regulatory standards and principles for evaluation of medicines, post-market surveillance activities, and approaches to comparability that are similar to Canada. If the non-canadian reference biologic drug is used in clinical studies in Canada, data must be provided to satisfy chemistry and manufacturing (quality) information as per C of the Food and Drug Regulations. Refer to section 2.2. for more information Review time The target time for review of a biosimilar is the same as that for an NDS. Refer to the Health Canada Guidance for Industry: Management of Drug Submissions for details on review timelines. Revised Date:

175 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs 2.2 Information requirements for clinical trial applications (CTA) Clinical trials conducted in Canada involving biosimilars are subject to Part C, Division 5 of the Food and Drug Regulations, which outlines the requirements applicable to the sale and importation of drugs for use in human clinical trials in Canada. Clinical Trial Applications (CTAs) should be submitted in accordance with Health Canada s Guidance for Clinical Trial Sponsors: Clinical Trial Applications and the Clinical Trials Manual. If a non-canadian reference biologic drug is used in clinical studies in Canada, data must be provided to support its safety and to satisfy the intent of the regulatory requirements for chemistry and manufacturing (quality) information. At a minimum, this should include confirmation that the non-canadian reference biologic drug is sourced from an ICH country and that there is evidence of a history of safe use in the country of origin. If the comparative structural, functional and non-clinical in vitro studies are considered satisfactory and no issues are identified that would preclude administration into humans, in vivo animal studies may not be necessary. Sponsors are encouraged to request scientific advice meetings and pre-cta consultation meetings for biosimilars. Refer to Section 3 for information on biosimilar scientific advice meetings and Health Canada s Guidance for Clinical Trial Sponsors: Clinical Trial Applications for instructions on how to request a pre-cta meeting. 2.3 Information requirements for new drug submissions (NDS) Part C, Division 8 of the Food and Drug Regulations sets out the requirements for the sale of new drugs in Canada, which include biosimilars, and prohibits the sale of new drugs unless the manufacturer has filed a submission that is satisfactory to the Minister. Section C of the Food and Drug Regulations outlines the requirements for an NDS Organization of data Electronic documents should be provided in electronic common technical document (ectd) format. The regulatory activities provided in ectd format should be prepared using applicable sections of the Guidance Document: Preparation of Drug Regulatory Activities in the Electronic Common Technical Document Format. Alternatively, Health Canada will also accept electronic documents in non-ectd electronic-only format in accordance with the applicable sections of the Guidance Document: Preparation of Drug Regulatory Activities in the Non-eCTD Electronic-Only Format. The assessment of similarity should be organized as a distinct collection of data in module 3 with an associated section in the Quality Overall Summary containing appropriate cross-references. However, the reorganization of modules 2 and 3 of a regulatory activity already prepared for another regulatory jurisdiction should not Revised Date:

176 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs be necessary. Sponsors should provide a note to reviewers indicating the location and organization of the similarity assessment. The biosimilar sponsor is encouraged to consult with BGTD for further guidance Quality information In addition to a typical chemistry and manufacturing data package that is expected for a standard new biologic drug, the biosimilar submission should include extensive data demonstrating similarity with the reference biologic drug. This should include characterization studies conducted in a side-by-side format. For consideration as a biosimilar, similarity should be deduced primarily from comprehensive and well rationalized quality studies. If excipients do not limit the sensitivity of assays used for characterization, it may be feasible to undertake studies to demonstrate similarity using the formulated drug products. Frequently, studies comparing the drug substance will be beneficial or may be the only scientific option. If the reference drug substance used for characterization is isolated from a formulated reference drug product, additional studies should demonstrate that the drug substance is not changed by the isolation process. One approach to potentially qualifying the isolation process is to use the process on the formulated biosimilar drug product and compare the isolated (deformulated) biosimilar drug substance to the biosimilar drug substance obtained prior to formulation. The approach used should be justified Quality considerations for demonstrating similarity Although the comparison of two independent products is outside of the scope of ICH Q5E: Comparability of Biotechnology/Biological Products Subject to Changes in their Manufacturing Process, many of the principles and approaches are applicable. The sponsor should demonstrate whether the biosimilar and the reference biologic drug can be judged highly similar in terms of quality attributes and thus support a possible conclusion of similarity for safety and efficacy. The product should be evaluated at the process steps most appropriate to detect a difference in the quality attributes. However, in most situations this evaluation will be limited to the drug substance, the drug product, or both. The extent of the studies necessary to demonstrate similarity will depend on the following: The nature of the product. The availability of suitable analytical techniques to detect potential product differences. The relationship between quality attributes and safety and efficacy based on overall non-clinical and clinical experience. Differences between the expression systems used to manufacture the biosimilar and the reference biologic drug. Revised Date:

177 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs When considering the similarity of products, the manufacturer should evaluate, for example: Relevant physicochemical and biological characterization data regarding quality attributes. Results from analysis of relevant samples from the appropriate stages of the manufacturing process (i.e. drug substance and drug product). Stability data, including those generated from accelerated or stress conditions, to provide insight into potential product differences in the degradation pathways of the drug product and, hence, potential differences in product-related substances and product-related impurities. Data obtained from multiple batches of the biosimilar and the reference biologic drug to help generate an understanding of ranges in variability. This evaluation may not entail performing all tests on all batches; a matrix approach may be possible but should be rationalized. In addition to evaluating the data, the manufacturer should consider if the results provide insights regarding the following: Critical control points in the manufacturing process that affect product characteristics. Adequacy of the in-process controls including critical control points and in-process testing: in-process controls for the biosimilar should be confirmed, modified, or created, as appropriate, to maintain the quality of the product. The comparative structural and functional studies will determine the type and extent of data to be derived from non-clinical and clinical studies on the drug product Quality considerations Analytical Techniques Analytical tests should be carefully selected and optimised to maximise the potential for detecting relevant differences in the quality attributes of the biosimilar and the reference biologic drug. It may be appropriate to modify existing tests used in biosimilar product development or to add new tests. To address the full range of physicochemical properties or biological activities, it may be appropriate to apply more than one analytical procedure to evaluate the same quality attribute. In such cases, each method should employ different physicochemical or biological principles to collect data for the same parameter to maximise the possibility that differences between the biosimilar and the reference biologic drug may be detected. Revised Date:

178 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs Measurement of quality attributes in characterization studies does not necessarily entail use of validated assays, but assays used should be scientifically sound and provide results that are reliable. Methods used to measure quality attributes for batch release should be validated in accordance with ICH guidelines, ICH Q2(R1): Validation of Analytical Procedures: Text and Methodology, ICH Q5C: Quality of Biotechnological Products: Stability Testing of Biotechnological/Biological Products, and ICH Q6B: Specifications: Test Procedures and Acceptance Criteria for Biotechnological/Biological Products as appropriate. Characterization Characterization of a biotechnological/biological product by appropriate techniques, as described in ICH Q6B, includes the determination of physicochemical properties, biological activity, immunochemical properties (if any), purity, impurities, contaminants and quantity. A complete side-by-side characterization should be performed to directly compare the biosimilar and the reference biologic drug. The significance of any differences should be evaluated. The following criteria should be considered as a key point when demonstrating similarity: Physicochemical Properties The manufacturer should consider the concept of the desired product (and its variants) as defined in ICH Q6B when designing and conducting studies to demonstrate similarity. The complexity of the molecular entity with respect to the degree of molecular heterogeneity should also be considered. The manufacturer should attempt to determine that the higher order structure (secondary, tertiary, and, where applicable, quaternary) is comparable. If appropriate higher order structural information cannot be obtained, a relevant biological activity assay (see biological activity below) could indicate a correct conformational structure. Biological Activity Biological assay results can serve multiple purposes in the confirmation of product quality attributes that are useful for characterization and batch analysis, and in some cases, could serve as a link to clinical activity. The manufacturer should consider the limitations of biological assays, such as high variability, that may prevent detection of differences between two highly similar products. Revised Date:

179 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs In cases where the biological assay also serves as a complement to physicochemical analysis (e.g. as a surrogate assay for higher order structure), the use of a relevant biological assay with appropriate precision and accuracy may provide a suitable approach to confirm that a change in specific higher order structure has not occurred. Where physicochemical or biological assays are not considered adequate to confirm that the higher order structure is maintained, data from non-clinical or clinical studies may be supportive. However, too much reliance on such studies may indicate that consideration as a biosimilar is not appropriate. When the products being compared have multiple biological or functional activities, a set of relevant functional assays designed to evaluate the range of activities should be utilized. Where any of the multiple activities is not sufficiently correlated with clinical safety or efficacy, or if the mechanism of action is not understood, justification should be provided that non-clinical or clinical activity of the biosimilar associated with the clinical indication being sought is not compromised. Immunochemical Properties When immunochemical properties are part of the characterization (e.g. for antibodies or antibody-based products), the manufacturer should confirm that the specific properties of the biosimilar are comparable to those of the reference biologic drug. Purity and Impurity The combination of analytical procedures selected should provide data to allow evaluation of relevant differences in the purity and impurity profiles. Differences observed in the purity and impurity profiles of the biosimilar relative to the reference biologic drug should be evaluated to assess their potential impact on safety and efficacy. Where the biosimilar exhibits different impurities, those impurities should be identified and characterized when possible. Depending on the impurity type and amount, the conduct of non-clinical and clinical studies will help to confirm that there is no adverse impact on safety or efficacy of the biosimilar. The potential impact of differences in the impurity profile upon safety should be addressed and supported by appropriate data. Specifications The tests and analytical procedures chosen to define drug substance or drug product specifications alone are not considered adequate to assess product Revised Date:

180 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs differences since they are chosen to confirm the routine quality of the product rather than to fully characterize it. The manufacturer should confirm that the specifications chosen for the biosimilar are appropriate to demonstrate product quality. Stability For certain manufacturing processes, even slight differences in the production procedures used for the biosimilar and reference biologic drug may cause differences in the stability of the products. Proteins are frequently sensitive to changes, such as those made to buffer composition, processing and holding conditions, and the use of organic solvents. Therefore, real-time/real temperature stability studies should be conducted on the biosimilar and reference biologic drug to compare the stability behaviour of both using the same storage conditions and analytical methods. In some cases, it may be possible and beneficial to conduct side-byside stability studies on samples that have been matched, as far as possible, with respect to date of manufacture. Such stability studies may be able to detect subtle differences between the biosimilar and the reference biologic drug that are not readily detectable by the characterization studies. For example, the presence of trace amounts of a protease may only be detected by product degradation that occurs over an extended time period. Or in some cases, divalent ions leached from the container closure system may change the stability profile because of the activation of trace proteases. Accelerated and stress stability, forced degradation, and photostability studies are often useful tools to establish degradation profiles and can therefore contribute to a direct comparison of a biosimilar and the reference biologic drug. The results may show product differences that warrant additional evaluation. The results may also identify conditions indicating that additional controls should be employed in the manufacturing process and during storage of the biosimilar to eliminate these unexpected differences. Appropriate studies should be considered to confirm that suitable storage conditions and controls are selected. ICH Q5C and ICH Q1A (R2): Stability Testing of New Drug Substances and Products should be consulted to determine the conditions for stability studies that provide relevant data for a product-to-product comparison Manufacturing process considerations A well-defined manufacturing process with its associated process controls assures that an acceptable product is produced on a consistent basis. Revised Date:

181 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs Approaches to determining the impact of any process differences will vary with respect to the specific process, the product, the extent of the manufacturer s knowledge of and experience with the process and the development data generated. Where details of the manufacturing process for the reference biologic drug are available to the biosimilar sponsor, a comparison with those for the biosimilar may help to identify which tests should be performed Determination of similarity The demonstration of similarity does not signify that the quality attributes of the two products being compared are identical, but that they are highly similar with two consequences: 1) that the existing knowledge of both products is sufficient to predict that any differences in quality attributes should have no adverse impact upon safety or efficacy of the biosimilar; and 2) that nonclinical and clinical data previously generated with the reference biologic drug are relevant to the biosimilar. A final determination of similarity will be based on all relevant data from structural, functional, non-clinical and clinical studies. To be considered a biosimilar, the weight of evidence should be provided by the structural and functional studies. The degree of similarity at the quality level will determine the scope and the breadth of the required non-clinical and clinical data. The non-clinical and clinical programs should be designed to complement the structural and functional studies and address potential areas of residual uncertainty. Consideration as a biosimilar may not be appropriate in situations where extensive reliance on the contribution of non-clinical and clinical studies would be expected, such as: i) the analytical procedures used are not sufficient to discern relevant differences that can impact the safety and efficacy of the product; or ii) the relationship between specific quality attributes and safety and efficacy has not been established, and differences between quality attributes of the biosimilar and the reference biologic drug are likely to be observed Manufacturing changes following issuance of market authorization A biosimilar is a new drug with all of the associated regulatory requirements. For any changes to the manufacturing process that warrant a demonstration of comparability, the products to be compared will be the pre-change and postchange versions of the biosimilar. Studies should be conducted in accordance Revised Date:

182 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs with ICH Q5E. Comparisons with the original reference biologic drug are not required Non-clinical and clinical information General Non-clinical and clinical requirements outlined in this guidance document are applicable to biosimilars that have been demonstrated to be similar to the reference biologic drug based on the results of the comparative structural and functional studies included in the chemistry and manufacturing data package. If similarity has not been established, reduced non-clinical and clinical data cannot be justified and the product cannot be considered a biosimilar. This section provides general guidance on non-clinical and clinical information required for biosimilars. Specific requirements for drug classes (e.g. insulin and growth hormone) may differ depending on the class. Requirements may also differ depending on various clinical parameters related to each specific drug product or class, including elements such as therapeutic index and the type and number of indications for which biosimilar sponsors apply. A biosimilar product sponsor is eligible to apply for one or more clinical indications granted to the reference biologic drug in Canada. Any claims made by the biosimilar sponsor must be supported by suitable scientific data. Proposals for indications and uses that are not supported by clinical data specific to the biosimilar can be considered for authorization; refer to Section for additional information. Clinical data should be generated based on the product for which market authorization is sought. Chemistry and manufacturing changes introduced during the clinical development phase or at the end of the clinical development program may result in the need for additional bridging data. Sponsors should refer to ICH Q5E and consult with BGTD for additional guidance Non-clinical studies Where similarity is well established by structural and functional studies, and where extensive in vitro mechanistic studies are indicative of similarity, in vivo non-clinical studies may not be necessary. Refer to the Biological Activity section within for more information on in vitro studies. Sponsors should provide a scientific justification for their approach and should consult with BGTD at the scientific advice and/or pre-submission stage. Revised Date:

183 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs If filed in more than one module, sponsors should provide a note to reviewers that communicates where in the e-ctd non-clinical studies are located. Specialized toxicological studies, including safety pharmacology, reproductive toxicology, mutagenicity and carcinogenicity studies, are not generally required for a biosimilar submission Clinical studies The purpose of the clinical program is to show that there are no clinically meaningful differences between the biosimilar and the reference biologic drug. The clinical program should begin with a PK/PD study(ies) which may be followed by an additional clinical trial(s). Differences observed between the biosimilar and reference biologic drug, such as differences in immunogenicity, should be addressed. If differences cannot be addressed, the sponsor should consider whether the biosimilar submission route is still appropriate or whether the traditional new drug submission route would be more suitable. Pharmacokinetic (PK) studies Comparative PK studies should be conducted to rule out differences in PK characteristics between the biosimilar and the reference biologic drug. PK studies should be carried out in healthy subjects when appropriate as they are usually considered to be a homogeneous and sensitive population. A low or sub-therapeutic dose residing on the linear part of the dose response curve should be considered if studies are performed in healthy subjects. Studies should be conducted in the patient population when the PK or PK/PD in the patient population is known to be substantially altered by the disease states for which authorisation is requested. The dose(s) used in the PK studies in a relevant patient population should be within the therapeutic dosing range specified in the product monograph of the reference biologic drug. The following factors should be taken into consideration during comparative PK study design (e.g. when choosing between cross-over versus parallelgroup study): half-life of the biologics linearity of PK parameters where applicable, the endogenous levels and diurnal variations of the protein under study conditions and diseases to be treated Revised Date:

184 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs route(s) of administration, and indications for which the biosimilar sponsor is applying. General principles of study design and statistical methods for comparative bioavailability studies should be considered when assessing the similarity of the PK parameters between the biosimilar and the reference biologic drug. The PK comparison should not be limited to parameters reflecting absorption only. Parameters representing elimination (e.g. clearance and terminal halflife) should also be compared. Data should not be excluded from the analysis unless the exclusion can be justified and is considered acceptable by BGTD. Acceptable criteria for the determination of similarity in comparative pharmacokinetics should be defined and justified prior to the initiation of PK study(ies). The acceptance criteria should be defined taking into consideration the PK parameters being evaluated and their variations, assay methodologies, and available safety and efficacy information regarding the reference biologic drug and the biosimilar. The criteria for comparative bioavailability studies as outlined in Health Canada s Guidance Document: Conduct and Analysis of Comparative Bioavailability Studies and Comparative Bioavailability Standards: Formulations used for Systemic Effects should be considered. When such criteria are not employed or not met in the comparative PK studies, a discussion should be provided regarding the implication of the findings in conjunction with data obtained from other clinical studies. Pharmacodynamic (PD) studies As for all other studies in the biosimilar developmental program, PD studies should be comparative in nature. Parameters investigated in PD studies should be clinically relevant. Use of a particular PD marker should be scientifically justified. PD markers should be relevant to the mechanism of action of the drug but may not need to be established surrogates for efficacy. In general, the principles regarding study design, conduct, analysis and interpretation that are relevant to equivalence trials with a clinical outcome as the primary endpoint are applicable to equivalence trials with a PD marker as the primary outcome. PD studies should be combined with PK studies, in which case the PK/PD relationship should be characterized. Clinical efficacy trial(s) In most cases, a comparative clinical trial(s) is important to rule out clinically meaningful differences in efficacy and safety between the biosimilar and the Revised Date:

185 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs reference biologic drug. A clinical efficacy trial may not always be necessary, e.g. where there is a clinically relevant PD endpoint. In such cases, a scientific justification is needed and safety as well as comparative immunogenicity data are still required. The comparative clinical trial should be adequately sensitive to rule out clinically meaningful differences within predefined comparability margins. Sponsors should consider the following factors when designing an adequately sensitive clinical trial: The characteristics of the studied population(s) (e.g. underlying disease, immune competence). The characteristics of the clinical trials, such as study duration, route of administration, dosage regimen, clinical endpoint(s) and time of assessment. The risk and impact of immunogenicity. The impact of concomitant therapies (e.g. monotherapy vs. combination therapy). The use of appropriate comparability margins. In some instances, evaluation of more than one sensitive population may be necessary. Careful consideration should be given to the design of the study(ies) including the choice of primary efficacy endpoint(s) and clinical comparability margin. Each of these aspects are important and should be justified on clinical grounds. The study should be conducted using a clinically relevant and sensitive endpoint to show that there are no clinically meaningful differences between the biosimilar and reference biologic drug. The chosen endpoint could be different from the original study endpoint for the reference biologic drug (e.g. a well-established surrogate or a more sensitive endpoint). In all cases, an acceptable comparability margin should be defined taking into account the smallest effect size that the reference biologic drug would reliably be expected to have based on publicly available historical data. If multiple endpoints are used, then the principles described above should apply. In line with the principle of similarity, equivalence trials are generally preferred. If non-inferiority trials are considered, they should be clearly justified and sponsors are advised to consult with BGTD prior to study initiation. Sponsors should be aware that the results from such trials could suggest statistical superiority of the biosimilar relative to the reference biologic drug. In such instances, the superiority observed should be assessed for clinical relevance including its impact on safety. In the event that the superiority observed is considered clinically meaningful and/or is associated with increased adverse drug reactions over those seen with the reference Revised Date:

186 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs biologic drug, the product would no longer be considered as a biosimilar. In addition, demonstration of non-inferiority of the biosimilar to the reference biologic drug might not provide strong support for the authorization of other indications, particularly if the other indications include different dosages than those tested in the clinical trial. Safety The nature, severity and frequency of adverse events should be compared between the biosimilar and the reference biologic drug. Efforts should be made to ensure that comparative clinical studies have a sufficient number of patients treated for an acceptable period of time in order to rule out clinically meaningful differences in safety between the biosimilar and the reference biologic drug. Immunogenicity The purpose of the comparative immunogenicity study(ies) is to rule out clinically meaningful differences in immunogenicity between the biosimilar and the reference biologic drug. Of most concern are those antibodies that have the potential to impact safety and/or efficacy; for example, by altering PK, inducing anaphylaxis, or by neutralising the product and/or its endogenous protein counterpart. For each treatment arm, the comparative study(s) should characterise the incidence and magnitude of the anti-drug antibody (ADA) response, the time-course of ADA development, ADA persistence, and the impact of ADA on safety, efficacy and PK. A suitable population should be selected in which to compare immunogenicity. In selecting an appropriate population, factors such as immunocompetence, prior or concomitant use of immunosuppressant therapies, and historical data with respect to the immunogenicity of the reference biologic drug should be considered. Because the investigation of immunogenicity is usually undertaken as part of the pivotal comparative safety and efficacy study(ies), it is important that the aforementioned factors are considered during the design of the clinical portion of the program to demonstrate biosimilarity. Comparative immunogenicity testing should be conducted using a tiered approach that involves screening assays, confirmatory assays and assays to determine whether binding ADA are neutralising. Independent binding ADA assays incorporating the biosimilar or the reference biologic drug as the capture ligand should be developed in parallel. Each assay should be validated and have demonstrated ability to sensitively detect ADA in the presence of drug. Samples from both treatment arms should be tested for ADA using both assays to demonstrate ADA cross-reactivity against the biosimilar and the Revised Date:

187 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs reference biologic drug. Deviation from this approach should be scientifically justified. Patient samples that test positive for binding ADA in confirmatory binding assays should be tested for their ability to neutralise the drug unless there exists a strong rationale for not doing so. The selection of an appropriate format for neutralising antibody testing is important and should take into account the mechanism of action of the drug. Depending on the mechanism of action, competitive ligand binding assays or cell based assays may be appropriate Authorization of indications A biosimilar sponsor may request authorization for all indications held by the biologic drug authorized in Canada to which a reference is made. The decision to authorize the requested indications is dependent on the demonstration of similarity between the biosimilar and reference biologic drug based on data derived from comparative structural, functional, non-clinical and clinical studies. Where similarity has been established, indications may be granted even if clinical studies are not conducted in each indication. A detailed rationale that scientifically justifies authorization of the biosimilar in each indication should be provided taking into consideration mechanism(s) of action, pathophysiological mechanism(s) of the disease(s) or conditions involved, safety profile, dosage regimen, clinical experience with the reference biologic drug, and any case-bycase considerations. Certain situations may warrant additional clinical data for a particular indication. Situations where a clinical indication being sought is not authorized in Canada for the reference biologic drug fall outside the scope of this guidance document Labelling requirements Product Monograph The product monograph for a biosimilar should be developed in a manner consistent with the principles, practices and processes outlined in the Guidance Document: Product Monograph. The contents of the product monograph for biosimilars should include the following information: A statement indicating that the product is a biosimilar and that similarity between the biosimilar and the reference biologic drug has been established based on comparative structural and functional studies, non-clinical studies, human PK/PD studies and clinical trials, as applicable. A statement that indications have been granted on the basis of similarity between the biosimilar and the reference biologic drug and taking into Revised Date:

188 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs consideration the mechanism(s) of action, disease pathophysiology, safety profile, dosage regimen and clinical experience with the reference biologic drug. Comparative data generated by the biosimilar sponsor on which the decision for market authorization was made summarized in a tabular format. Relevant safety and efficacy information from the product monograph of the biologic drug authorized in Canada to which a reference is made, including warnings and precautions, Adverse Drug Reactions/Adverse Drug Effects and key post-market safety information for all indications that are authorized for the biosimilar. There should be no claims for bioequivalence or clinical equivalence between the biosimilar and the reference biologic drug Risk Management Plan A risk management plan (RMP) should be submitted as part of the drug submission. The RMP should be designed to monitor and detect both known inherent safety concerns and potentially unknown safety signals that may result from the impurity profile and other characteristics of the biosimilar. The biosimilar sponsor should continue the assessment of unwanted immunogenicity and its clinical significance following market authorization. The RMP should consider all identified and potential risks associated with the use of the reference biologic drug and should provide details on how these risks will be addressed in a post-market setting. Health Canada will work with sponsors to ensure a suitable RMP is developed prior to authorization of the product. The minimum surveillance criteria for the biosimilar should be described in accordance with requirements for a new biologic drug. The RMP should include detailed information on systematic evaluation of the immunogenicity potential of the biosimilar. The RMP should include specific (routine or additional) pharmacovigilance and risk minimisation activities similar to those in place for the reference biologic drug or justify that these activities are not relevant for the biosimilar. A discussion about methods to distinguish adverse event reports for the biosimilar from those for other licensed products, including the reference product should be included. The RMP should be maintained and implemented throughout the lifecycle of the product. For more information on RMPs, please refer to the Guidance Document - Submission of Risk Management Plans and Follow-up Commitments. Revised Date:

189 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs 2.4 Post-market requirements Adverse drug reaction (ADR) reporting and periodic reports Sponsors are required to comply with sections C to C of the Food and Drug Regulations, which includes ADR reporting. On an annual basis or as requested by the Director, the manufacturer will conduct a concise, critical analysis of the adverse drug reactions and serious adverse drug reactions after such an occurrence. After an analysis, the Director may request written summary reports where safety is questionable. Periodic safety update reports (PSURs) or periodic benefit-risk evaluation reports (PBRERs) should be prepared and/or submitted as discussed in the risk management plan. The periodicity for the submission of PSURs or PBRERs should be consistent with appropriate ICH guidelines for marketed products, or as determined by the Minister, when the product is authorized for market. For more information on PBRERs, please refer to the Health Canada Guidance Document - Periodic Benefit-Risk Evaluation Report (PBRER) International Conference on Harmonisation (ICH) Topic E2C(R2) Post-notice of compliance (NOC) changes A biosimilar is a new drug with all of the associated regulatory requirements. For guidance on post- NOC changes, refer to the applicable Health Canada guidance documents Post-Notice of Compliance (NOC) Changes: Framework Document, Post-Notice of Compliance (NOC) Changes: Safety and Efficacy Document and Post-Notice of Compliance (NOC) Changes: Quality Document. Biosimilar sponsors should follow labelling requirements set out in the post-noc guidance documents referenced above. This includes monitoring any product class type-specific safety information that may indicate the need for a change in labelling. There may be situations post-noc where biosimilar sponsors seek authorization of indications held by the reference biologic drug authorized in Canada. A Supplemental New Drug Submission (SNDS) for a biosimilar that relies on the previously demonstrated similarity provided in the original biosimilar NDS may be considered by Health Canada on a case-by-case basis. Biosimilar sponsors should consult with BGTD for regulatory guidance for their specific SNDS. 3. Consultation with the Biologics and Genetic Therapies Directorate (BGTD) Biosimilar sponsors are encouraged to consult with BGTD for regulatory guidance as early as possible in the development of their biosimilar submission package. Consultation can occur at any stage of the development of a biosimilar. Revised Date:

190 Health Canada Information and Submission Requirements for Guidance document Biosimilar Biologic Drugs BGTD is undergoing a 3 year pilot to explore a stepwise review approach that would be complementary to the biosimilar development process. During the pilot, a biosimilar sponsor may request a Scientific Advice Meeting (SAM) in order to receive advice from BGTD on their quality package early in the development process. Sponsors should ensure their drug meets the eligibility criteria for a biosimilar as outlined in this guidance. Sponsors wishing to participate in the pilot should contact the Office of Regulatory Affairs. For more information, please refer to the Notice - Subsequent Entry Biologics Scientific Advice Meeting Pilot. 4. Additional Information Health Canada will review this guidance document on an ongoing basis in response to new scientific knowledge, best practices and/or experience gained by the Department. Contact Information: Questions concerning biosimilar submissions should be directed to: Office of Regulatory Affairs Biologics and Genetic Therapies Directorate Health Products and Food Branch Health Canada BGTD.ORA@hc-sc.gc.ca Questions or comments on this guidance document should be directed to: Office of Policy and International Collaboration Biologics and Genetic Therapies Directorate Health Products and Food Branch Health Canada BGTD.OPIC@hc-sc.gc.ca Revised Date:

191 Therapeutic Advances in Drug Safety Review Immunogenicity and other problems associated with the use of biopharmaceuticals Michael G. Tovey and Christophe Lallemand Ther Adv Drug Saf (2011) 2(3) DOI: / ! The Author(s), Reprints and permissions: journalspermissions.nav Abstract: Biopharmaceuticals are used widely for the treatment of cancer, chronic viral hepatitis, inflammatory, and autoimmune diseases. Biopharmaceuticals such as interferons are well tolerated for the most part with the most common adverse events observed being flu-like symptoms that resolve rapidly after initial treatment. Prolonged treatment is associated, however, with more serious adverse events including leucopenia, thrombocytopenia, and neuropsychiatric effects, which may necessitate dose reduction or even cessation of treatment in some patients. Recombinant growth factors, such as erythropoietin (EPO), granulocyte colony-stimulating factor, or granulocyte macrophage colony-stimulating factor, are for the most part well tolerated, although severe complications have been reported in patients with cancer or chronic kidney disease treated with EPO. Similarly, treatment of patients with cancer with high doses of interleukin-2 is associated with significant toxicity. Treatment of chronic inflammatory diseases, such as rheumatoid arthritis, psoriasis, and Crohn s disease, with antitumor necrosis factor-alpha monoclonal antibodies is associated with an increased risk of granulomatous infections and, in particular, tuberculosis. The monoclonal antibody, natalizumab, that targets alpha4 integrins is effective in the treatment of multiple sclerosis but is associated with the activation of JC virus and development of progressive multifocal leukoencephalopathy. Repeated administration of recombinant proteins can cause a break in immune tolerance in some patients resulting in the production of a polyclonal antibody response that can adversely affect pharmacokinetics and clinical response. In addition, neutralizing antibodies that cross react with nonredundant essential proteins such as EPO can cause severe autoimmune reactions. Keywords: adverse events, biopharmaceutical, cytokine, growth factor, immunogenicity, monoclonal antibody Introduction Biopharmaceuticals are an important class of pharmaceuticals that are used widely in clinical medicine. Among the most widely used biopharmaceuticals are monoclonal antibodies used in the therapy of cancer or inflammatory and autoimmune diseases, and recombinant analogues of growth factors such as insulin or erythropoietin (EPO), or cytokines such as type I interferons (IFNs) or interleukin 2 (IL-2). This is reflected by sales of some US$92 billion in 2009 (Table 1) [La Merie Business Intelligence, 2010]. The safety and efficacy of recombinant biopharmaceuticals can be severely impaired, however, by the ability of patients to tolerate the drug. Type I IFNs and IL-2 were among the first biopharmaceuticals to be licensed for use in man and their toxicity profile is well characterized. A consideration of the clinical experience with the use of these molecules as well as with the use of some of the first monoclonal antibodies to be used extensively in the clinic, serves to illustrate the types of adverse events that can be encountered with biopharmaceuticals. Immunogenicity can also limit the use of biopharmaceuticals, particularly for the treatment of chronic inflammatory and autoimmune diseases, such as rheumatoid arthritis (RA) or multiple sclerosis (MS). Although it is widely accepted that injection of foreign proteins can elicit an immune reaction leading to the production of antidrug antibodies (ADAs), it is Correspondence to: Michael G. Tovey, PhD Laboratory of Viral Oncology, FRE3238 CNRS, Institut André Lwoff, 7 rue Guy Moquet, Villejuif, France tovey@vjf.cnrs.fr Christophe Lallemand, PhD Laboratory of Viral Oncology, FRE3238 CNRS, Institut André Lwoff, Villejuif, France 113

192 Therapeutic Advances in Drug Safety 2 (3) Table 1. Biopharmaceutical sales in Relative position Therapeutic class Sales 2009 (US$ billions)* 1 Anticancer monoclonal antibodies Tumor necrosis factor-alpha antagonists Insulin and insulin analogues Erythropoietins Interferon-beta Interferon-alpha 2.61 Total sales of all biopharmaceuticals *Taken from La Merie Business Intelligence, 10 March becoming increasingly apparent that repeated injection of recombinant homologues of authentic human proteins, such as IFNs or EPO, especially when aggregated or partially denatured, can result in a break in immune tolerance to self-antigens leading to the production of an ADA response. ADAs can adversely affect the pharmacokinetics, bioavailability, and efficacy of biopharmaceuticals and in some cases may neutralize the activity of the drug. ADAs can also cause immune complex disease, allergic reactions and, in some cases, severe autoimmune reactions. Assessment of immunogenicity is therefore an important component of drug safety evaluation. Adverse events associated with the administration of biopharmaceuticals In the following sections the type of adverse events that have been observed in the clinical use of bipharmaceuticals will be reviewed by reference to the use of type I IFNs (IFNa2 and IFNb-1), IL-2, EPO, and the monoclonal antibodies, rituximab, natalizumab, daclizumab, and alemtuzumab. Emphasis will be given to an understanding of the emerging pattern of common adverse events observed with different classes of drug products. Type I IFNs Type I IFNs play a determining role in the innate immune response and in the establishment of the subsequent adaptive immune response and resistance to virus infection [O Neill and Bowie, 2010; Content, 2009]. Type I IFNs bind to a common high-affinity cell-surface receptor composed of two transmembrane polypeptide chains, IFNAR1 and IFNAR2. IFN binding results in the phosphorylation and activation of two Janus kinases, Tyk2 and Jak1, and activation of the latent cytoplasmic signal transducers and activators of transcription (STAT) 1 and 2, and the formation of a transcription complex in association with IFN regulatory factor 9 (IRF9). Translocation of this complex to the nucleus results in the transcriptional activation of a specific set of genes that encode the effector proteins responsible for mediating the biological activities of the type I IFNs [O Neill and Bowie, 2010; Content, 2009]. Recombinant IFNa2 and more recently pegylated IFNa2a and pegylated IFNa2b, have found wide application for the treatment of chronic viral hepatitis [McHutchison et al. 2009] and some neoplastic diseases [Kiladjian et al. 2008], while IFNb-1 is used extensively as first-line therapy for the treatment of RRMS [Farrell and Giovannoni, 2010]. Type I IFNs are for the most part well tolerated although mild adverse events, including acute flu-like symptoms, are observed in most patients. More serious chronic side effects, such as myelosuppression, induction or exacerbation of underlying autoimmune disease, and neuropsychiatric effects, are also observed in some patients. Adverse events are usually dose-dependent and a function of the particular IFN used, the disease state, and the genetic makeup of the individual being treated. The majority of patients treated with standard therapeutic doses of native or pegylated IFNa2 or native IFNb-1 exhibit flu-like symptoms, including fever, headache, chills, and myalgia, within 1 3 h of the initiation of IFN treatment. These symptoms decrease in intensity over time, and are believed to be due to both the release of the inflammatory cytokines IL-1, IL-6, and tumor necrosis factor alfa (TNFa), and induction of prostaglandin-e 2 in the hypothalamus [Dinarello et al. 1984]. Neutropenia is one of the most common adverse events observed in patients with chronic hepatitis C treated with native or pegylated IFNa2a or 114

193 MG Tovey and C Lallemand IFNa2b, and can be sufficiently severe to necessitate dose reduction or even discontinuation of treatment [Sleijfer et al. 2005]. Anemia and thrombocytopenia is also observed in approximately 6% of patients with chronic hepatitis C treated with PEG-IFNa2a or PEG-IFNa2b and often leads to dose reduction or discontinuation of treatment [Harris et al. 2001]. Hematological adverse events can be particularly severe in patients co-infected with hepatitis C virus and human immunodeficiency virus and treated with pegylated IFNa2 and ribavirin [Mira et al. 2007; Simin et al. 2007]. An association between polymorphisms in the genes encoding IFNAR1 and STAT2, and IFN-induced neutropenia has recently been reported in certain populations with chronic hepatitis C treated with type I IFN [Wada et al. 2009]. Similarly, IFN-induced thrombocytopenia was found to be associated with a polymorphism in the gene encoding IRF7 [Wada et al. 2009]. In contrast, these polymorphisms were not associated with a sustained virologic response to IFN therapy that was dependent upon independent factors including viral load, viral genotype, and the degree of liver fibrosis on initiation of IFN treatment. Elevated levels of liver transaminases are also a common side effect observed during the first weeks of treatment of patients with IFNa. The effects are dose dependent and resolve rapidly on cessation of IFN therapy [Sleijfer et al. 2005]. Induction or exacerbation of pre-existing autoimmune disease is observed in some patients treated with type I IFNs [Kawaguchi et al. 2009; Nordmark et al. 2009]. Recent genome-wide association studies have identified alleles associated with susceptibility to a number of autoimmune diseases some of which encode proteins involved in IFN induction or IFN signaling [Hellquist et al. 2009; Kyogoku et al. 2009; Miceli-Richard et al. 2009; Nordmark et al. 2009; Suarez-Gestal et al. 2009a, 2009b; Cunninghame Graham et al. 2007; Miceli- Richard et al. 2007; Graham et al. 2006]. Thus, there are a number of reports of the induction or exacerbation of systemic lupus erythematosus (SLE) in patients treated with recombinant IFNa [Kawaguchi et al. 2009; Nordmark et al. 2009; Yilmaz and Cimen, 2009]. Patients with risk factors (such as particular major histocompatibility complex (MHC) haplotypes, female gender, and others) that predispose an individual to the development of autoimmune disease, are particularly susceptible to the development of IFN-induced SLE, the symptoms of which usually resolve after discontinuation of IFN therapy [Kawaguchi et al. 2009; Nordmark et al. 2009; Yilmaz et al. 2009]. Although the mechanism of IFN-induced SLE remains unclear, IFNa can be detected in the peripheral circulation of untreated SLE patients and a number of studies have shown an association between activation of the type I IFN pathway and disease activity [Ronnblom et al. 2006; Baechler et al. 2003]. Thus, the expression of a number of IFNinduced genes, a so-called IFN signature, is up regulated in peripheral blood mononuclear cells from SLE patients with active disease [Ronnblom et al. 2006; Baechler et al. 2003]. Furthermore, polymorphisms in the genes encoding IRF-5, Tyk2, and STAT4 that are involved in the induction and activity of IFNs, have been identified as risk factors for the development of SLE [Suarez- Gestal et al. 2009a; Graham et al. 2006]. Treatment of patients with type I IFNs has also been reported to induce the symptoms of a number of other autoimmune diseases including Sjögren s syndrome [De Santi et al. 2005; Unoki et al. 1996], systemic sclerosis [Powell et al. 2008], dermatomyositis [Somani et al. 2008; Dietrich et al. 2000], polymyositis [Somani et al. 2008; Dietrich et al. 2000], RA [Burdick et al. 2009; Fabris et al. 2003; Levesque et al. 1999; Alsalameh et al. 1998], type I diabetes [Burdick et al. 2009; Fabris et al. 2003], and autoimmune hepatitis [Duchini, 2002; Heathcote, 1995]. Thyroid dysfunction is also a relatively common side effect of IFN treatment, particularly in patients with pre-existing thyroid autoimmunity [Monzani et al. 2004]. As in the case of SLE, polymorphisms in the genes encoding proteins involved in the activation or activity of type I IFNs are also associated with a predisposition to develop certain of these diseases. For example, polymorphisms in the STAT4 and IRF5 genes are associated with an increased risk of developing systemic sclerosis, an autoimmune connective tissue disorder characterized by fibrosis of multiple organs and a type I IFN signature [Dieude et al. 2009; Ito et al. 2009; Rueda et al. 2009]. Treatment of patients with native or pegylated IFNa is associated with a number of neuropsychiatric side effects including fatigue, depression, cognitive disturbances and, in some cases, 115

194 Therapeutic Advances in Drug Safety 2 (3) suicidal tendencies [Dafny and Yang, 2005; Valentine et al. 1998]. A number of mechanisms have been implicated in these effects including effects of IFNs on neurons, serotonergic activity, neuro-endocrine cytokine secretion, tryptophan metabolism, and neurotransmitters [Dafny and Yang, 2005; Valentine et al. 1998]. The results of recent clinical trials in patients with MS suggest that depression is a less common adverse effect of IFNb therapy than was once thought and that flu-like symptoms and injection site reactions, particularly when IFNb is administered by the subcutaneous route, are among the most persistent side effects leading to cessation of treatment in some patients [Farrell and Giovannoni, 2010]. Systemic administration of type I IFN results in the binding of IFN to receptors present on essentially all types of nucleated cells, including neurons and hematopoietic stem cells, in addition to target cells or target organ(s) involved in a particular disease. This may well explain the wide spectrum of adverse events associated with IFN therapy. IL-2 IL-2 was approved in the USA in 1992 for the treatment of metastatic renal cell carcinoma and subsequently for the treatment of metastatic melanoma. Although IL-2 is effective in the treatment of metastatic renal cell carcinoma and metastatic melanoma, and can lead to complete remission in a small subset of patients [Antony and Dudek, 2010], its widespread application has been limited by a spectrum of life-threatening side effects including hypotension, renal dysfunction, peripheral and pulmonary edema, and vascular leakage syndrome [Antony and Dudek, 2010]. IL-2-related toxicity is, however, generally reversible and only rarely causes permanent organ failure [Antony and Dudek, 2010]. EPO After over two decades of use in the treatment of chronic anemia in patients with chronic renal disease and in patients with cancer, recombinant EPO is considered to be safe and well tolerated. In spite of the improved erythropoiesis and reduced need for transfusions in cancer patients receiving recombinant EPO or its analogue, darbepoetin-alpha, meta-analysis has identified an increased risk of thrombo-embolic events during chemotherapy in patients receiving EPO, without increased on-study mortality or reduced overall survival [Dicato and Plawny, 2010]. Furthermore, patients with tumor cells that express EPO or EPO receptors have an unfavorable prognosis, raising concerns about tumor progression and the safety of erythropoiesis-stimulating agents [Dicato and Plawny, 2010]. Rituximab Rituximab (Rituxan Õ ) is a humanized monoclonal antibody that targets the B-cell differentiation antigen CD20. Rituximab is used extensively to treat non-hodgkin s lymphoma and other B-cell lymphomas, as well as inflammatory diseases such as RA [Keating, 2010; Tak et al. 2010]. The action of rituximab is thought to be due at least in part to the depletion of CD20-bearing B cells [Weiner, 2010], through antibody-dependent cellular cytotoxicity and complement-dependent cytotoxicity. Rituximab is in general well tolerated although treatment is associated with infusion reactions, which in some cases may be fatal, tumor lysis syndrome, and some cases of acute renal failure [Kunkel et al. 2000]. Rituximab treatment is also associated with serious infections, including sepsis [Furst, 2010; Kunkel et al. 2000], and some cases of progressive multifocal leukoencephalopathy (PML), a rare demyelinating disease caused by reactivation of the polyoma virus JC virus in immunosuppressed individuals [Berger, 2010]. TNF antagonists Several types of TNFa antagonists, including a chimeric antibody (infliximab), a human antibody (adalimumab), a pegylated F(ab ) fragment (certolizumab), and etanercept, a fusion protein composed of the p75-soluble TNF receptor and the Fc fragment of IgG1, are currently used for the treatment of RA, a debilitating disease characterized by persistent synovitis, systemic inflammation, and the presence of auto-antibodies [Nam et al. 2010]. The anti-tnfa monoclonal antibodies infliximab, adalimumab, and certolizumab are also used to treat inflammatory bowel diseases, including Crohn s disease [Peyrin-Biroulet, 2010]. Although TNFa antagonists are, for the most part, well tolerated, their use is associated with an increased risk of granulomatous infections, particularly the reactivation of tuberculosis [Dixon et al. 2010]. The incidence of infection is higher in patients treated with infliximab and adalimumab than in patients treated with etanercept [Dixon et al. 2010]. There are also a number of reports of reactivation 116

195 MG Tovey and C Lallemand of latent virus infections including hepatitis B virus (HBV) [Carroll and Forgione, 2010]. Treatment of patients with TNFa antagonists is also associated with induction of inflammatory and autoimmune diseases, such as SLE [Costa et al. 2008], and psoriasis [Cohen et al. 2007]. The results of a recent meta-analysis reveal an increased incidence in malignancies in patients treated with TNFa antagonists [Bongartz et al. 2006]. Thus, a statistically significant increase in the incidence of solid tumors was observed in patients with Wegener granulomatosis treated with etanercept [Stone, 2005], and nonmelanoma skin cancers in patients with RA treated with TNFa antagonists [Chakravarty et al. 2005]. Lymphomas and nonhematological malignancies did not appear to occur with an increased frequency, however, when compared with population-based incidence data [Bongartz et al. 2006]. There are reports of the development of demyelinating disease in patients with inflammatory disease treated with TNFa antagonists [Alshekhlee et al. 2010; Bernatsky et al. 2010]. TNFa is associated, however, with both the development of MS in man [Sharief and Hentges, 1991], and experimental autoimmune encephalomyelitis (EAE) in animals. TNFa antagonists have also been shown to decrease the severity of EAE in rodents [Selmaj and Raine, 1995]. A systematic review of the literature and the BIOBADASER pharmacovigilance database has shown that the number of cases of demyelinating disease in patients treated with TNFa antagonists does not differ from the background population suggesting that demyelination may not reflect a true association with the treatment regimen [Fernandez-Espartero et al. 2011]. The authors suggest, however, that the use of TNFa antagonists in patients with both a rheumatic and a demyelinating disease is counter indicated [Fernandez-Espartero et al. 2011]. Natalizumab Natalizumab (Tysabri Õ ) is a humanized monoclonal antibody that targets a4-integrins preventing leukocytes adhering to endothelial vascular cell adhesion molecule-1, and hence reducing trafficking across the blood brain barrier [Coyle, 2010]. Natalizumab has been shown to be effective for the treatment of RRMS [Coyle, 2010]. It is approved for the therapy of RRMS based on its efficacy in the SENTINEL trial in combination with IFNb-1a and its efficacy as monotherapy in the AFFIRM trial [Coyle, 2010; Rudick and Panzara, 2008]. Natalizumab monotherapy has been shown to reduce the risk of disability progression by 42 54%, the annualized relapse rate by 68%, and to reduce significantly central nervous system (CNS) lesions detected by magnetic resonance imaging (MRI) [Coyle, 2010; Rudick et al. 2008]. Although natalizumab is for the most part well tolerated, side effects include anaphylactic reactions (<1.0% of patients) and other hypersensitivity reactions, increased risk of infection, depression, and rash. A number of cases of PML have also been reported in natalizumabtreated patients, particularly those treated for 2 years or more [Clifford et al. 2010], necessitating particular vigilance in regard to its safety profile. Approximately 10% of patients develop neutralizing antibodies to natalizumab, which persist for 12 months or more in 6% of patients [Calabresi et al. 2007]. There are a number of case reports of melanoma in patients treated with natalizumab with no previous history of melanoma, and in patients with pre-existing nevi that developed into melanoma [Laroni et al. 2011; Prinz Vavricka et al. 2010; Bergamaschi and Montomoli, 2009; Ismail et al. 2009]. Although it has been suggested that the target of natalizumab, a(4)b1 integrin, may play a role in the invasiveness of melanoma [Castela et al. 2011], it remains to be shown whether the incidence of melanoma is increased in natalizumabtreated patients relative to the background population. Results from the AFFIRM and SENTINEL trials show that some 9% of patients with MS develop antibodies to natalizumab of which some 6% were persistent and, for the most part, neutralizing [Calabresi et al. 2007]. The presence of persistent antibodies was associated with reduced clinical efficacy and infusionrelated adverse events including hypersensitivity reactions [Calabresi et al. 2007]. Most of those patients who develop an ADA response did so within 12 weeks (88% in the AFFIRM trial). A small group of patients (3%) developed a transient ADA response that lead to a delay in the clinical response to natalizumab therapy, although clinical efficacy was restored once antinatalizumab antibodies were no longer detectable after 6 months of continued treatment [Calabresi et al. 2007]. The authors recommended that patients with disease activity or persistent infusion reaction after 6 months of natalizumab therapy should be tested for ADAs. Obviously 117

196 Therapeutic Advances in Drug Safety 2 (3) therapy should be discontinued in patients who develop hypersensitivity reactions. Daclizumab Daclizumab is a humanized monoclonal antibody that targets the high affinity IL-2 receptor, CD25, which is up regulated upon T-cell activation. Daclizumab has been used extensively for the prevention of allograft rejection [Campara et al. 2010], and more recently for the treatment of RRMS [Wynn et al. 2010]. Daclizumab is thought to act at least in part by expansion of a regulatory subset of CD56 bright NK cells that kill activated T cells present in inflammatory lesions [Bielekova et al. 2006; Campara et al. 2010]. Daclizumab is for the most part well tolerated with a relatively low incidence of infections and other adverse events including mild hypersensitivity reactions observed in renal transplant patients [Campara et al. 2010]. Increased mortality has been observed, however, in cardiac transplant recipients treated with daclizumab together with standard immunosuppressive therapy [Hershberger et al. 2005]. Alemtuzumab Alemtuzumab (Campath-1H Õ and Campath Õ )is a humanized monoclonal antibody that targets CD52, a leukocyte-differentiation antigen, present on most leukocytes including B cells, T cells, NK cells, and monocytes/macrophages [Bates, 2009]. Alemtuzumab is used for the treatment of various neoplastic diseases including chronic lymphocytic leukemia [Schweighofer and Wendtner, 2010], and more recently for the treatment of RRMS [Coles et al. 2008]. Alemtuzumab treatment depletes B cells and T cells for extended periods and its use is associated with significant side effects, including frequent infusion reactions such as pyrexia, malaise, and rash, and an increased rate of infections [Minagar et al. 2010; Coles et al. 2008]. Alemtuzumab treatment is also associated with the development of autoimmunity including thyroid disorders, predominately Grave s disease, in approximately 30% of patients with MS but intriguingly not in patients with cancer [Minagar et al. 2010]. Among the most serious adverse event observed in MS patients treated with alemtuzumab are Goodpasture s syndrome reported in three patients [Farrell and Giovannoni, 2010], and immune thrombocytopenic purpura (ITP). Six cases of ITP, including one fatal case, were reported in one clinical trial [Coles et al. 2008]. Adverse events: underlying mechanisms and clinical implications It is clear from a consideration of different types of biopharmaceuticals ranging from recombinant analogues of authentic human growth factors such as EPO, or cytokines such as type I IFNs or IL-2 to chimeric or human monoclonal antibodies that a pattern of some common side effects does emerge. Thus, administration of a number of these biopharmaceuticals is associated with infusion reactions, increased rates of infection, activation of latent viruses, and induction or reactivation of autoimmune disease. Given the key role played by type I IFNs in the innate immune response it is perhaps not surprising that dysregulation of the type I IFN pathway can under certain circumstances lead to induction or exacerbation of autoimmune disease in predisposed individuals [Burdick et al. 2009]. Thus, polymorphisms in the genes encoding key intermediates in the type I IFN pathway have been identified as risk factors for the development of autoimmune disease [Dieude et al. 2009; Ito et al. 2009; Rueda et al. 2009], and a number of IFN-induced proteins are up regulated in active disease [Ronnblom et al. 2006; Baechler et al. 2003]. Furthermore, a recent report has shown that a loss of function variant in the gene encoding MAVS, a key antiviral molecule downstream of the cytosolic doublestranded RNA sensors RIG-I and MDA-5, is associated with low type I IFN production and the absence of anti-rna-binding protein autoantibodies in a novel subphenotype of SLE [Pothlichet et al. 2011]. It may appear paradoxical that antagonists of the pro-inflammatory cytokine TNFa that are used widely for the treatment of inflammatory and autoimmune diseases can induce SLE [Costa et al. 2008], or psoriasis [Ko et al. 2009; Cohen et al. 2007], in some individuals. Plasmacytoid dendritic cells (pdcs) that infiltrate the target organs of a number of autoimmune diseases, including SLE and psoriasis, secrete large amounts of IFNa, which is thought to play a role in the disease process [Burdick et al. 2009]. TNFa suppresses the generation of pdcs and the production of IFNa [Palucka et al. 2005], which may explain at least in part the observed exacerbation of autoimmune disease in some individuals treated with TNFa antagonists. Reactivation of tuberculosis in patients treated with TNFa antagonists [Dixon et al. 2010], reflects the important role played by TNFa in 118

197 MG Tovey and C Lallemand concert with IFNg in the adaptive immune response and the containment and eradication of Mycobacterium tuberculosis infection. Similarly, reactivation of HBV in patients treated with TNFa antagonists may be related to a reduced virus-specific CD8þ T-cell response in the presence of lower intrahepatic levels of TNFa. Thus, polymorphisms in the TNFa gene that lead to lower levels of TNFa production are related to an increased risk of progression to chronic HBV infection, while reduced levels of intrahepatic levels of TNFa lead to reduced expression of MHC class I antigens and a reduced virus-specific CD8þ T-cell response [Carroll and Forgione, 2010]. It has been suggested, based on cumulative clinical experience that patients chronically infected with HBV who require therapy with a TNFa antagonist should be treated with antiviral therapy 1 2 weeks prior to treatment with a TNFa antagonist. In addition, baseline liver function, based on serum albumin and alanine transaminase levels and HBV DNA viral load, should be determined at the start of antiviral therapy and antitnfa therapy and every 1 2 months thereafter [Carroll and Forgione, 2010]. The detection of some 85 cases of PML in patients with RRMS treated with natalizumab as of January 2011 [Gryta, 2011], equivalent to an incidence of approximately 1 case per 1000, necessitates particular vigilance especially in patients treated for 2 years or more [Ryschkewitsch et al. 2010]. Furthermore, the observation that JC virus DNA can persist in the cerebrospinal fluid (CSF) for years, even in the presence of high levels of anti-jc virus antibody, may warrant periodic testing for the presence of viral DNA in the CSF as well as in serum [Ryschkewitsch et al. 2010]. Three cases of PML were also reported in patients with psoriasis treated with the monoclonal antibody, efalizumab [Lysandropoulos and Du Pasquier, 2010], which targets the integrin, LFA-1 (CD11a). Cases of PML have also been detected in RA patients treated with the anticd20 antibody, rituximab [Berger, 2010; Paues and Vrethem, 2010; Bongartz et al. 2006], suggesting a possible causal association between the increased number of pre-b cells, some of which contain JC virus, detected in the circulation of patients treated with either rituximab or natalizumab [Allison, 2010]. As of March 2009, some 57 cases of PML in rituximab-treated patients have been reported [Carson et al. 2009], with some additional cases reported since that time [Lysandropoulos and Du Pasquier, 2010; Paues and Vrethem, 2010]. Thus, vigilance is again called for when treating patients with RA or non-hodgkin s lymphoma with rituximab. The dose-dependent increase in the risk of malignancies in RA patients treated with high doses of TNFa antagonists suggests that the marginal if any gain in efficacy should be weighed against the increase risk on an individual basis [Bongartz et al. 2006]. The absence of an accumulation of malignancies with longer study duration is suggestive of a potentiation of pre-existing malignancies rather than induction suggesting that systematic screening for pre-existing malignancies and subsequent surveillance may provide a means of improving the safety of antitnfa therapy [Bongartz et al. 2006]. Immunogenicity of biopharmaceuticals It is well established that repeated injection of even native human proteins can result in a break in immune tolerance to self-antigens in some patients leading to a humoral response against the protein that is enhanced when the protein is aggregated or partially denatured [Van Beers et al. 2010]. Although in most cases an immune response to a biopharmaceutical has little or no clinical impact, ADAs do, however, pose a number of potential risks for the patient, particularly in the case of a neutralizing antibody response. Firstly, an ADA response can adversely affect the pharmacokinetics and bioavailability of a drug thereby reducing the efficacy of treatment and necessitating either dose escalation or switching to alternative therapy if such therapy is available. An ADA response can also adversely affect the safety of treatment and cause immune complex disease, allergic reactions and, in some cases, severe autoimmune reactions. Serious and lifethreatening adverse events can occur when ADAs cross react with an essential nonredundant endogenous protein such as EPO or thrombopoietin [Casadevall et al. 2002; Li et al. 2001]. Thus, several cases of pure red cell aplasia were associated with the development of antibodies to recombinant EPO following a change in formulation [Casadevall et al. 2002]. Similarly, the development of antibodies to pegylated megakaryocyte growth and development factor (MGDF) cross reacted with endogenous MGDF resulting in several cases of severe thrombocytopenia [Li et al. 2001]

198 Therapeutic Advances in Drug Safety 2 (3) Figure 1. Factors affecting the immunogenicity of biopharmaceuticals. 1. Hayakawa and Ishii [2011]; 2. Armstrong et al. [2007]; 3. Gribben et al. [1990]; 4. Van Beers et al. [2010]; 5. De Groot and Scott [2007]; 6. Kirshner [2011]; 7. Rosenberg [2003]; 8. Wadhwa et al. [1999]; 9. Barbosa et al. [2006]; 10. Fakharzadeh and Kazazian [2000]; 11. Schernthaner [1993]; 12. Maini et al. [1998]; 13. Li et al. [2001]; 14. Giovannoni [2003]. Factors influencing the immunogenicity of biopharmaceuticals Currently available techniques do not permit one to predict with a sufficient degree of accuracy whether a biopharmaceutical will be immunogenic and if so, to what extent. It is also difficult to predict which patients will develop an immune response to a particular drug, and at what time during treatment an immune response will occur. There are, however, a number of both drugrelated and patient-related factors that are known to influence the immunogenicity of biopharmaceuticals (Figure 1). Drug-related factors include the presence of nonhuman sequences or novel epitopes generated by amino acid substitution designed to enhance stability, or novel epitopes created at the junction of fusion proteins. Molecular structure, and in particular, changes in glycosylation, can also influence the immunogenicity of a biopharmaceutical. Thus, the absence of glycosylation or an altered pattern of glycosylation can expose cryptic B-cell and T-cell epitopes in the protein, or cause the protein to appear foreign to the immune system. Carbohydrate moieties present upon biopharmaceuticals can elicit the production of IgE antibodies that can cause serious adverse reactions including anaphylaxis even upon the first treatment exposure. Pre-existing antibodies against galactose-a-1,3-galactose have been shown to be responsible for IgE-mediated anaphylactic reactions in patients treated with cetuximab [Chung et al. 2008]. Pegylation can reduce the immunogenicity of some proteins although patients produce antibodies to the PEG residue adversely affecting efficacy [Armstrong et al. 2007]. In addition to attributes that can induce a classical immune response, repeated administration of even authentic human proteins such as albumin can under certain circumstances cause a break in 120

199 MG Tovey and C Lallemand immune tolerance leading to the development of an immune response. Thus, the presence of degradation products resulting from oxidation or deamination of the protein, aggregates, or the intrinsic immunomodulatory properties of the molecule can also influence the immunogenicity of a biopharmaceutical. Protein aggregation in particular has long been associated with increased immunogenicity [Moore and Leppert, 1980], although the mechanisms underlying this effect remain poorly understood. It has been suggested that aggregated proteins form repetitive arrays that can lead to efficient cross-linking of B-cell receptors, leading to B-cell activation in the absence of T-cell help, thereby resulting in a break in immune tolerance to self-proteins [Felipe et al. 2010]. The relatively high incidence of ADAs in patients treated with recombinant granulocyte macrophage colony-stimulating factor (GM-CSF) may be related at least in part to the immunostimulatory properties of the molecule itself [Rini et al. 2005]. Thus, GM-CSF can recruit antigen-presenting cells to the site of antigen processing, stimulate the maturation of myeloid DCs, and enhance an antigen-specific CD8þ T-cell response, suggesting that repeated administration of GM-CSF may function as an adjuvant. Indeed, GM-CSF has been used as an immunological adjuvant in a number of vaccination protocols designed to elicit an immune response to self-antigens [Parmiani et al. 2007]. Process-related impurities, including traces of residual DNA or proteins from the expression system, or contaminants that leach from the product container, can also influence the immunogenicity of recombinant biopharmaceuticals [Wadhwa et al. 1999]. Patient-related factors, such as genetic makeup, age, gender, disease status, concomitant medication, and route of administration, can also influence the immune response to a particular biopharmaceutical. For example, a common MHC class II allele, DRB1*0701, is associated with the antibody response to IFNb in MS patients [Barbosa et al. 2006]. Disease state and immune competency also influence the immune response of an individual to a treatment with a biopharmaceutical. Thus, development of antibodies to pegylated MGDF is less frequent in cancer patients who tend to be immunosuppressed than in healthy individuals [Li et al. 2001]. Concomitant therapy with immunosuppressive drugs can also influence a patient s immune response to a biopharmaceutical. Thus, administration of methotrexate together with the chimeric monoclonal antibody infliximab has been shown to reduce the immune response to infliximab and improve the clinical response in patients with RA [Maini et al. 1998]. The duration of treatment and the route of administration also influence the immune response to a biopharmaceutical. Typically, administration of a protein in a single dose results in the production of low-affinity IgM antibodies, while repeated administration results in the production of high-affinity and high-titer IgG antibodies, which may be neutralizing. Thus, in patients with MS treated with IFNb neutralizing antibodies to IFNb often do not appear until after several months of therapy [Giovannoni, 2003]. The intravenous route of administration is considered to be least likely to generate an immune response to a biopharmaceutical compared with intramuscular or subcutaneous administration. The complexity of the humoral response to biopharmaceuticals and the difficulty in establishing the effect on ADAs on drug efficacy is illustrated by the response of patients to treatment with IFNb, first licensed over a decade ago for the treatment of RRMS. Five products are currently available in the USA and Europe as first-line disease-modifying agents for the treatment of RRMS, IFNb-1a (Avonex Õ and Rebif Õ ), IFNb- 1b (Betaseron Õ and Betaferon Õ ), and more recently, the IFNb-1b biosimilar Extavia Õ. Avonex Õ and Rebif Õ are both glycosylated forms of native human IFNb-1a produced in Chinese hamster ovary cells. Betaseron Õ and Extavia Õ are a nonglycosylated form of IFNb-1a produced in Escherichia coli that has a serine substitution for the unpaired cystine at position 17 of the native protein. These drugs are partially effective; they reduce the number of relapses by about one third, reduce the number of CNS lesions detected by MRI by approximately 70%, and may also delay disease progression. Most patients develop an antibody response to IFNb products, and as many as up to 45% of patients develop neutralizing antibodies to IFNb, in some cases as early as 3 months after initiation of therapy. Overall, some 25% of patients develop antiifnb-neutralizing antibodies usually within 6 18 months. ADAs are more frequent in patients treated with IFNb-1b than IFNb-1a, 121

200 Therapeutic Advances in Drug Safety 2 (3) while subcutaneous IFNb-1a (Rebif Õ ) is more immunogenic than intramuscular IFNb-1a (Avonex Õ ) [Malucchi et al. 2004]. The immunogenicity of IFNb varies among individuals, both as a function of the presence of particular MHC class II alleles, and as a function of IFN-receptor expression. Thus, patients who process the DRB1*0701 allele, or who express low levels of IFNAR2, one of the two chains of the type I IFN receptor, upon initiation of treatment, have a significantly higher risk of developing antiifnbneutralizing antibodies [Farrell and Giovannoni, 2010]. Although it has been shown in numerous trials that patients who develop antibodies against IFNb have higher relapse rates, increased number of lesions detected by MRI, and higher rates of disease progression [Francis et al. 2005; Kappos et al. 2005], the significance of antiifnb ADAs remains controversial [Goodin et al. 2007]. This is due to the difficulty in establishing a temporal correlation between the presence of antiifnb ADAs and the loss of drug efficacy due to the variable nature of the disease, the partial effectiveness of the drug, the delay between initiation of treatment and the detection of an effect of the drug on the course of the disease, and the difference in the immunogenicity of different IFNb products. The lack of standardized ADA assays has also rendered direct comparisons of immunogenicity between different products and different studies difficult, which has contributed to the difficulty in establishing a correlation between ADAs and loss of drug efficacy. The development of a standardized cell-based assay for the quantification of antidrug-neutralizing antibodies that allows direct comparisons between different IFNb products [Lallemand et al. 2010, 2008] is discussed in the following. Regulatory guidance Assessment of immunogenicity is an important component of drug safety evaluation in preclinical, clinical, and post-marketing studies. Draft Guidance for Industry Assay Development for Immunogenicity Testing of Therapeutic Proteins has recently been published by the US Food and Drug Administration ( GuidanceComplianceRegulatoryInformation/Guid ances/ucm pdf). Similarly, guidelines on the immunogenicity assessment of biotechnology-derived therapeutic proteins established by the Committee for Medicinal Products for Human Use of the European Medicines Agency came into effect in April 2008 ( emea.europa.ed/pdfs/human/biosimilar/ en.pdf). These guidelines provide a general framework for a systematic and comprehensive evaluation of immunogenicity that should be modified as appropriate, on a case-by-case basis. Although differences in approach and emphasis exist between the USA, European Union, and Japanese regulatory authorities there is, nevertheless, a large degree of consensus on the type of approach that should be adopted; namely, a risk-based approach that is clinically driven and takes into account pharmacokinetic data. Thus, biopharmaceuticals with no endogenous counterpart are considered to be of relative low risk while drugs with a nonredundant endogenous counterpart are considered to present a high risk. A multi-tiered approach to testing samples is also recommended. This consists of an appropriate screening assay capable of detecting both IgM and IgG ADAs, the sensitivity of which is such that a percentage of false-- positive samples would be detected. The specificity of the samples that test positive in the screening assay are then re-assayed in a confirmatory assay usually by competition with an unlabelled drug using the same assay format as that used for the screening assay. Samples that test positive in the screening and confirmatory assays are then tested for the presence of neutralizing ADAs using a cell-based assay whenever possible. Although it may be appropriate to use ligand-binding assays for certain monoclonal antibodies that target soluble antigens, some ADAs may not be detected using a ligand-binding assay. Thus, antibodies have been described that inhibit the antiviral activity of type I IFNs by inducing hyperphosphorylation of a receptor-associated tyrosine kinase without inhibiting binding of IFN to its receptor [Novick et al. 2000]. Cell-based assays for the quantification of neutralizing antidrug antibodies Although regulatory authorities recommend the use of cell-based assays for the detection and quantification of neutralizing antibodies, cellbased assays often give variable results and are notoriously difficult to standardize. Conventional cell-based assays for neutralizing antibodies are based upon the assessment of a drug-induced response in a drug-sensitive cell 122

201 MG Tovey and C Lallemand line, and the ability of an antibody to inhibit that response. The form of the assay can vary considerably, however, reflecting the diversity of the biopharmaceuticals currently employed in the clinic. Drug-induced responses vary from stimulation of cell proliferation in the presence of a growth factor such as EPO or GM-CSF, or induction of apoptosis in the presence of TNFa to inhibition of virus replication in IFNtreated cells [Meager, 2006]. Such drug-induced responses are complex events involving the transcriptional activation or modulation of numerous genes. Drug-induced biological responses also often take several days to develop and are influenced by a number of factors that are difficult to control. A series of engineered reporter cell lines have been developed for the quantification of the activity and neutralizing antibody response to biopharmaceuticals that eliminate many of the limitations of conventional cell-based assays. In order to improve the performance of conventional cell-based assays and to develop assays that allow more direct comparisons of immunogenicity data, reporter cell lines were established based on transfection of cells with the firefly luciferase reporter gene regulated by a drug-responsive chimeric promoter [Lallemand et al. 2008]. These assays allow drug activity, and the neutralization of drug activity, to be determined selectively and with a high degree of precision within a few hours by measuring light emission [Lallemand et al. 2008]. In addition, the use of a single common drug-induced response for a variety of different drugs facilitates comparisons of immunogenicity data. Conventional cell-based assays are difficult to standardize due in part to assay variation resulting from changes in culture conditions and genetic and epigenetic changes that can occur as cells are cultivated continuously in the laboratory. Assay variation can be minimized by the use of cell banks and preparation of each lot of assay cells under standardized conditions from an individual frozen vial [Lallemand et al. 2008]. Each lot of assay cells is thus in an identical physiological condition. Assay-ready cells can be manufactured under conditions of current good manufacturing practices, and stored frozen for several years without loss of drug sensitivity [Lallemand et al. 2008]. In order to render cell-based assays less labor intensive and more amenable to automation, an assay was developed that allows the level of drug activity and antidrug-neutralizing antibodies to be determined in a serum sample simply by the addition of reporter cells without further dilution, or the addition of exogenous drug [Lallemand et al. 2010]. The so called one-step assay is based on a cell line that has been engineered to express and secrete the biopharmaceutical of interest together with a reporter-gene transcribed from the same inducible promoter. Expression of the drug is thus strictly proportional to expression of the reporter gene, thereby allowing drug production to be quantified with precision. The cells also contain a second reporter gene [Grossberg et al. 2009], controlled by a drugresponsive chimeric promoter that allows drug activity to be quantified. The presence of antidrug-neutralizing antibodies in a sample will neutralize a quantity of drug secreted from the reporter cell, proportional to the neutralizing capacity of the antibody. This will prevent the drug from binding to its specific cell-surface receptor, thereby resulting in a corresponding reduction in the expression of the drug-responsive reporter gene. Expression of the drug-responsive reporter gene in the presence or absence of the ADA-containing sample will allow the relative neutralizing titer of the sample to be quantified by the constant antibody method, as a function of the quantity of drug required to neutralize a sample, determined from the level of expression the reporter gene transcribed from the same promoter as the drug [Lallemand et al. 2010]. Results are normalized relative to the expression of an internal standard, and are consequently independent of cell number, thus affording a high degree of assay precision and reducing serum matrix effects to a minimum. The one-step assay is applicable to a wide range of biopharmaceuticals and is ideally suited to high throughput analysis of antidrugneutralizing antibodies during clinical development or routine patient monitoring. Conclusion Biopharmaceuticals are widely used for the treatment of numerous indications including many serious debilitating and life-threatening diseases. Biopharmaceuticals are for the most part well tolerated and represent an important addition to the modern drug armamentarium. The use of some biopharmaceuticals is limited, however, 123

202 Therapeutic Advances in Drug Safety 2 (3) by serious adverse events that occur in some individuals. A number of these adverse events, including infusion reactions, increased rates of infection, activation of latent viruses, and induction or reactivation of autoimmune disease, are observed with different types of biopharmaceuticals. Undesired immunogenicity can also limit the use of biopharmaceuticals, particularly for the treatment of chronic diseases that necessitate repeated treatments over long periods. The development of improved assays, particularly cell-based assays for the detection of neutralizing antibodies, that allow immunogenicity to be determined with precision and the comparison of immunogenicity data between biopharmaceuticals, are critical for the development of less immunogenic and safer biopharmaceuticals. Funding Some of the research cited in this article was funded in part by Biomonitor LTD. Conflict of interest statement The authors declare no conflict of interest in preparing this manuscript. References Allison, M. (2010) PML problems loom for rituxan. Nat Biotechnol 28: Alsalameh, S., Manger, B., Kern, P. and Kalden, J. (1998) New onset of rheumatoid arthritis during interferon beta-1b treatment in a patient with multiple sclerosis: comment on the case report by Jabaily and Thompson. Arthritis Rheum 41: 754. Alshekhlee, A., Basiri, K., Miles, J.D., Ahmad, S.A. and Katirji, B. (2010) Chronic inflammatory demyelinating polyneuropathy associated with tumor necrosis factor-alpha antagonists. Muscle Nerve 41: Antony, G.K. and Dudek, A.Z. (2010) Interleukin 2 in cancer therapy. Curr Med Chem 17: Armstrong, J.K., Hempel, G., Koling, S., Chan, L.S., Fisher, T., Meiselman, H.J. et al. (2007) Antibody against poly(ethylene glycol) adversely affects pegasparaginase therapy in acute lymphoblastic leukemia patients. Cancer 110: Baechler, E.C., Batliwalla, F.M., Karypis, G., Gaffney, P.M., Ortmann, W.A., Espe, K.J. et al. (2003) Interferon-inducible gene expression signature in peripheral blood cells of patients with severe lupus. Proc Natl Acad Sci U S A 100: Barbosa, M.D., Vielmetter, J., Chu, S., Smith, D.D. and Jacinto, J. (2006) Clinical link between MHC class II haplotype and interferon-beta (IFN-beta) immunogenicity. Clin Immunol 118: Bates, D. (2009) Alemtuzumab. Int MS J 16: Bergamaschi, R. and Montomoli, C. (2009) Melanoma in multiple sclerosis treated with natalizumab: Causal association or coincidence? Mult Scler 15: Berger, J.R. (2010) Progressive multifocal leukoencephalopathy and newer biological agents. Drug Saf 33: Bernatsky, S., Renoux, C. and Suissa, S. (2010) Demyelinating events in rheumatoid arthritis after drug exposures. Ann Rheum Dis 69: Bielekova, B., Catalfamo, M., Reichert-Scrivner, S., Packer, A., Cerna, M., Waldmann, T.A. et al. (2006) Regulatory CD56(Bright) natural killer cells mediate immunomodulatory effects of IL-2Ralpha-targeted therapy (daclizumab) in multiple sclerosis. Proc Natl Acad Sci U S A 103: Bongartz, T., Sutton, A.J., Sweeting, M.J., Buchan, I., Matteson, E.L. and Montori, V. (2006) Anti-TNF antibody therapy in rheumatoid arthritis and the risk of serious infections and malignancies: systematic review and meta-analysis of rare harmful effects in randomized controlled trials. JAMA 295: Burdick, L.M., Somani, N. and Somani, A.K. (2009) Type I IFNs and their role in the development of autoimmune diseases. Expert Opin Drug Saf 8: Calabresi, P.A., Giovannoni, G., Confavreux, C., Galetta, S.L., Havrdova, E., Hutchinson, M. et al. (2007) The incidence and significance of antinatalizumab antibodies: Results from Affirm and Sentinel. Neurology 69: Campara, M., Tzvetanov, I.G. and Oberholzer, J. (2010) Interleukin-2 receptor blockade with humanized monoclonal antibody for solid organ transplantation. Expert Opin Biol Ther 10: Carroll, M.B. and Forgione, M.A. (2010) Use of tumor necrosis factor alpha inhibitors in hepatitis B surface antigen-positive patients: A literature review and potential mechanisms of action. Clin Rheumatol 29: Carson, K.R., Evens, A.M., Richey, E.A., Habermann, T.M., Focosi, D., Seymour, J.F. et al. (2009) Progressive multifocal leukoencephalopathy after rituximab therapy in HIV-negative patients: A report of 57 cases from the research on adverse drug events and reports project. Blood 113: Casadevall, N., Nataf, J., Viron, B., Kolta, A., Kiladjian, J., Martin-Dupont, P. et al. (2002) Pure redcell aplasia and antierythropoietin antibodies in patients treated with recombinant erythropoietin. N Engl J Med 346: Castela, E., Lebrun-Frenay, C., Laffon, M., Rocher, F., Cohen, M., Leccia, N.C. et al. (2011) Evolution of nevi during treatment with 124

203 MG Tovey and C Lallemand natalizumab: A prospective follow-up of patients treated with natalizumab for multiple sclerosis. Arch Dermatol 147: Chakravarty, E.F., Michaud, K. and Wolfe, F. (2005) Skin cancer, rheumatoid arthritis, and tumor necrosis factor inhibitors. J Rheumatol 32: Chung, C.H., Mirakhur, B., Chan, E., Le, Q.T., Berlin, J., Morse, M. et al. (2008) Cetuximab-induced anaphylaxis and IgE specific for galactose-alpha-1,3- galactose. N Engl J Med 358: Clifford, D.B., De Luca, A., Simpson, D.M., Arendt, G., Giovannoni, G. and Nath, A. (2010) Natalizumab-associated progressive multifocal leukoencephalopathy in patients with multiple sclerosis: Lessons from 28 cases. Lancet Neurol 9: Cohen, J.D., Bournerias, I., Buffard, V., Paufler, A., Chevalier, X., Bagot, M. et al. (2007) Psoriasis induced by tumor necrosis factor-alpha antagonist therapy: A case series. J Rheumatol 34: Coles, A.J., Compston, D.A., Selmaj, K.W., Lake, S.L., Moran, S., Margolin, D.H. et al. (2008) Alemtuzumab vs. interferon beta-1a in early multiple sclerosis. N Engl J Med 359: Content, J. (2009) Mechanisms of induction and action of interferons. Verh K Acad Geneeskd Belg 71: Costa, M.F., Said, N.R. and Zimmermann, B. (2008) Drug-induced lupus due to anti-tumor necrosis factor alpha agents. Semin Arthritis Rheum 37: Coyle, P.K. (2010) The role of natalizumab in the treatment of multiple sclerosis. Am J Manag Care 16: S164 S170. Cunninghame Graham, D.S., Akil, M. and Vyse, T.J. (2007) Association of polymorphisms across the tyrosine kinase gene, Tyk2 in UK SLE families. Rheumatology (Oxford) 46: Dafny, N. and Yang, P.B. (2005) Interferon and the central nervous system. Eur J Pharmacol 523: De Groot, A.S. and Scott, D.W. (2007) Immunogenicity of protein therapeutics. Trends Immunol 28: De Santi, L., Costantini, M.C. and Annunziata, P. (2005) Long time interval between multiple sclerosis onset and occurrence of primary Sjögren s syndrome in a woman treated with interferon-beta. Acta Neurol Scand 112: Dicato, M. and Plawny, L. (2010) Erythropoietin in cancer patients: Pros and cons. Curr Opin Oncol 22: Dietrich, L.L., Bridges, A.J. and Albertini, M.R. (2000) Dermatomyositis after interferon alpha treatment. Med Oncol 17: Dieude, P., Guedj, M., Wipff, J., Ruiz, B., Hachulla, E., Diot, E. et al. (2009) STAT4 is a genetic risk factor for systemic sclerosis having additive effects with IRF5 on disease susceptibility and related pulmonary fibrosis. Arthritis Rheum 60: Dinarello, C.A., Bernheim, H.A., Duff, G.W., Le, H.V., Nagabhushan, T.L., Hamilton, N.C. et al. (1984) Mechanisms of fever induced by recombinant human interferon. J Clin Invest 74: Dixon, W.G., Hyrich, K.L., Watson, K.D., Lunt, M., Galloway, J., Ustianowski, A. et al. (2010) Drugspecific risk of tuberculosis in patients with rheumatoid arthritis treated with anti-tnf therapy: Results from the British Society for Rheumatology Biologics Register (BSRBR). Ann Rheum Dis 69: Duchini, A. (2002) Autoimmune hepatitis and interferon beta-1a for multiple sclerosis. Am J Gastroenterol 97: Fabris, P., Floreani, A., Tositti, G., Vergani, D., De Lalla, F. and Betterle, C. (2003) Type 1 diabetes mellitus in patients with chronic hepatitis C before and after interferon therapy. Aliment Pharmacol Ther 18: Fakharzadeh, S.S. and Kazazian Jr., H.H. (2000) Correlation between factor VIII genotype and inhibitor development in hemophilia A. Semin Thromb Hemost 26: Farrell, R.A. and Giovannoni, G. (2010) Current and future role of interferon beta in the therapy of multiple sclerosis. J Interferon Cytokine Res 30: Felipe, V., Hawe, A., Schellekens, H. and Jiskoot, W. (2010) Aggregation and immunogenicity of therapeutic proteins, In: Wang, W. and Roberts, C. (eds), Aggregation of Therapeutic Proteins, John Wiley & Sons: Hoboken, NJ. Fernandez-Espartero, M.C., Perez-Zafrilla, B., Naranjo, A., Esteban, C., Ortiz, A.M., Gomez- Reino, J.J. et al. (2011) Demyelinating disease in patients treated with TNF antagonists in rheumatology: data from BIOBADASER, a pharmacovigilance database, and a systematic review. Semin Arthritis Rheum 40: Francis, G.S., Rice, G.P. and Alsop, J.C. (2005) Interferon beta-1a in MS: Results following development of neutralizing antibodies in prisms. Neurology 65: Furst, D.E. (2010) The risk of infections with biologic therapies for rheumatoid arthritis. Semin Arthritis Rheum 39: Giovannoni, G. (2003) Strategies to treat and prevent the development of neutralizing anti-interferon-beta antibodies. Neurology 61: S13 S17. Goodin, D.S., Hurwitz, B. and Noronha, A. (2007) Neutralizing antibodies to interferon beta-1b are not associated with disease worsening in multiple sclerosis. J Int Med Res 35: Graham, R.R., Kozyrev, S.V., Baechler, E.C., Reddy, M.V., Plenge, R.M., Bauer, J.W. et al. (2006) 125

204 Therapeutic Advances in Drug Safety 2 (3) A common haplotype of interferon regulatory factor 5 (IRF5) regulates splicing and expression and is associated with increased risk of systemic lupus erythematosus. Nat Genet 38: Gribben, J.G., Devereux, S., Thomas, N.S., Keim, M., Jones, H.M., Goldstone, A.H. et al. (1990) Development of antibodies to unprotected glycosylation sites on recombinant human GM-CSF. Lancet 335: Grossberg, S.E., Kawade, Y. and Grossberg, L.D. (2009) The neutralization of interferons by antibody III. The constant antibody bioassay, a highly sensitive quantitative detector of low antibody levels. J Interferon Cytokine Res 29: Gryta, T. (2011) Biogen. Six more cases of brain infection in Tysabri patients. Dow Jones Newswires. Harris, J.M., Martin, N.E. and Modi, M. (2001) Pegylation: a novel process for modifying pharmacokinetics. Clin Pharmacokinet 40: Hayakawa, T. and Ishii, A. (2011) Japanese regulatory perspective on immunogenicity, In: Tovey, M.G. (ed.), Detection and Quantification of Antibodies to Biopharmaceuticals: Practical and Applied Considerations, John Wiley & Sons: New York. Heathcote, E.J. (1995) Autoimmune hepatitis and chronic hepatitis C: latent or initiated by interferon therapy? Gastroenterology 108: Hellquist, A., Jarvinen, T.M., Koskenmies, S., Zucchelli, M., Orsmark-Pietras, C., Berglind, L. et al. (2009) Evidence for genetic association and interaction between the Tyk2 and IRF5 genes in systemic lupus erythematosus. J Rheumatol 36: Hershberger, R.E., Starling, R.C., Eisen, H.J., Bergh, C.H., Kormos, R.L., Love, R.B. et al. (2005) Daclizumab to prevent rejection after cardiac transplantation. N Engl J Med 352: Ismail, A., Kemp, J. and Sharrack, B. (2009) Melanoma complicating treatment with natalizumab (Tysabri) for multiple sclerosis. J Neurol 256: Ito, I., Kawaguchi, Y., Kawasaki, A., Hasegawa, M., Ohashi, J., Hikami, K. et al. (2009) Association of a functional polymorphism in the IRF5 region with systemic sclerosis in a Japanese population. Arthritis Rheum 60: Kappos, L., Clanet, M., Sandberg-Wollheim, M., Radue, E.W., Hartung, H.P., Hohlfeld, R. et al. (2005) Neutralizing antibodies and efficacy of interferon beta-1a: A 4-year controlled study. Neurology 65: Kawaguchi, Y., Ota, Y., Kawamoto, M., Ito, I., Tsuchiya, N., Sugiura, T. et al. (2009) Association study of a polymorphism of the CTGF gene and susceptibility to systemic sclerosis in the Japanese population. Ann Rheum Dis 68: Keating, G.M. (2010) Rituximab: a review of its use in chronic lymphocytic leukaemia, low-grade or follicular lymphoma and diffuse large B-cell lymphoma. Drugs 70: Kiladjian, J.J., Chomienne, C. and Fenaux, P. (2008) Interferon-alpha therapy in bcr-abl-negative myeloproliferative neoplasms. Leukemia 22: Kirshner, S. (2011) Immunogenicity of therapeutic proteins: A regulatory perspective, In: Tovey, M.G. (ed.), Detection and Quantification of Antibodies to Biopharmaceuticals: Practical and Applied Considerations, John Wiley & Sons: New York. Ko, J.M., Gottlieb, A.B. and Kerbleski, J.F. (2009) Induction and exacerbation of psoriasis with TNFblockade therapy: a review and analysis of 127 cases. J Dermatolog Treat 20: Kunkel, L., Wong, A., Maneatis, T., Nickas, J., Brown, T., Grillo-Lopez, A. et al. (2000) Optimizing the use of rituximab for treatment of B-cell non- Hodgkin s lymphoma: a benefit-risk update. Semin Oncol 27: Kyogoku, C., Morinobu, A., Nishimura, K., Sugiyama, D., Hashimoto, H., Tokano, Y. et al. (2009) Lack of association between tyrosine kinase 2 (Tyk2) gene polymorphisms and susceptibility to SLE in a Japanese population. Mod Rheumatol 19: Lallemand, C., Meritet, J.F., Blanchard, B., Lebon, P. and Tovey, M.G. (2010) One-step assay for quantification of neutralizing antibodies to biopharmaceuticals. J Immunol Methods 356: Lallemand, C., Meritet, J.F., Erickson, R., Grossberg, S.E., Roullet, E., Lyon-Caen, O. et al. (2008) Quantification of neutralizing antibodies to human type 1 interferons using division-arrested frozen cells carrying an interferon-regulated reportergene. J Interferon Cytokine Res 28: La Merie Business Intelligence. (2010) R & D pipeline. News 10 March 2010, Laroni, A., Bedognetti, M., Uccelli, A., Capello, E. and Mancardi, G.L. (2011) Association of melanoma and natalizumab therapy in the Italian MS population: a second case report. Neurol Sci 32: Levesque, M.C., Ward, F.E., Jeffery, D.R. and Weinberg, J.B. (1999) Interferon-beta1a-induced polyarthritis in a patient with the HLA-DRB1*0404 allele. Arthritis Rheum 42: Li, J., Yang, C., Xia, Y., Bertino, A., Glaspy, J., Roberts, M. et al. (2001) Thrombocytopenia caused by the development of antibodies to thrombopoietin. Blood 98: Lysandropoulos, A.P. and Du Pasquier, R.A. (2010) Demyelination as a complication of new immunomodulatory treatments. Curr Opin Neurol 23: Maini, R.N., Breedveld, F.C., Kalden, J.R., Smolen, J.S., Davis, D., Macfarlane, J.D. et al. (1998) Therapeutic efficacy of multiple intravenous infusions of anti-tumor necrosis factor alpha monoclonal antibody combined with low-dose weekly methotrexate 126

205 MG Tovey and C Lallemand in rheumatoid arthritis. Arthritis Rheum 41: Malucchi, S., Sala, A., Gilli, F., Bottero, R., Di Sapio, A., Capobianco, M. et al. (2004) Neutralizing antibodies reduce the efficacy of betaifn during treatment of multiple sclerosis. Neurology 62: McHutchison, J.G., Lawitz, E.J., Shiffman, M.L., Muir, A.J., Galler, G.W., Mccone, J. et al. (2009) Peginterferon alfa-2b or alfa-2a with ribavirin for treatment of hepatitis C infection. N Engl J Med 361: Meager, A. (2006) Measurement of cytokines by bioassays: Theory and application. Methods 38: Miceli-Richard, C., Comets, E., Loiseau, P., Puechal, X., Hachulla, E. and Mariette, X. (2007) Association of an IRF5 gene functional polymorphism with Sjögren s syndrome. Arthritis Rheum 56: Miceli-Richard, C., Gestermann, N., Ittah, M., Comets, E., Loiseau, P., Puechal, X. et al. (2009) The CGGGG insertion/deletion polymorphism of the IRF5 promoter is a strong risk factor for primary Sjögren s syndrome. Arthritis Rheum 60: Minagar, A., Alexander, J.S., Sahraian, M.A. and Zivadinov, R. (2010) Alemtuzumab and multiple sclerosis: therapeutic application. Expert Opin Biol Ther 10: Mira, J.A., Lopez-Cortes, L.F., Merino, D., Arizcorreta-Yarza, A., Rivero, A., Collado, A. et al. (2007) Predictors of severe haematological toxicity secondary to pegylated interferon plus ribavirin treatment in HIV-HCV-coinfected patients. Antivir Ther 12: Monzani, F., Caraccio, N., Dardano, A. and Ferrannini, E. (2004) Thyroid autoimmunity and dysfunction associated with type I interferon therapy. Clin Exp Med 3: Moore, W.V. and Leppert, P. (1980) Role of aggregated human growth hormone (HGH) in development of antibodies to HGH. J Clin Endocrinol Metab 51: Nam, J.L., Winthrop, K.L., Van Vollenhoven, R.F., Pavelka, K., Valesini, G., Hensor, E.M. et al. (2010) Current evidence for the management of rheumatoid arthritis with biological disease-modifying antirheumatic drugs: a systematic literature review informing the Eular recommendations for the management of RA. Ann Rheum Dis 69: Nordmark, G., Kristjansdottir, G., Theander, E., Eriksson, P., Brun, J.G., Wang, C. et al. (2009) Additive effects of the major risk alleles of IRF5 and STAT4 in primary Sjögren s syndrome. Genes Immun 10: Novick, D., Nabioullin, R.R., Ragsdale, W., McKenna, S., Weiser, W., Garone, L. et al. (2000) The neutralization of type I IFN biologic actions by anti-ifnar-2 monoclonal antibodies is not entirely due to inhibition of Jak-Stat tyrosine phosphorylation. J Interferon Cytokine Res 20: O Neill, L.A. and Bowie, A.G. (2010) Sensing and signaling in antiviral innate immunity. Curr Biol 20: R328 R333. Palucka, A.K., Blanck, J.P., Bennett, L., Pascual, V. and Banchereau, J. (2005) Cross-regulation of TNF and IFN-alpha in autoimmune diseases. Proc Natl Acad Sci U S A 102: Parmiani, G., Castelli, C., Pilla, L., Santinami, M., Colombo, M.P. and Rivoltini, L. (2007) Opposite immune functions of GM-CSF administered as vaccine adjuvant in cancer patients. Ann Oncol 18: Paues, J. and Vrethem, M. (2010) Fatal progressive multifocal leukoencephalopathy in a patient with non- Hodgkin lymphoma treated with rituximab. J Clin Virol 48: Peyrin-Biroulet, L. (2010) Anti-TNF therapy in inflammatory bowel diseases: A huge review. Minerva Gastroenterol Dietol 56: Pothlichet, J., Niewold, T.B., Vitour, D., Solhonne, B., Crow, M.K. and Si-Tahar, M. (2011) A loss-of-function variant of the antiviral molecule MAVS is associated with a subset of systemic lupus patients. EMBO Mol Med 3: Powell, A., Myles, M.L. and Yacyshyn, E. (2008) The development of systemic sclerosis in a female patient with multiple sclerosis following beta interferon treatment. Clin Rheumatol 27: Prinz Vavricka, B.M., Baumbegrer, P., Russmann, S. and Kullak-Ublick, G.A. (2010) Diagnosis of melanoma under concomitant natalizumab therapy. Mult Scler 17: Rini, B., Wadhwa, M., Bird, C., Small, E., Gaines- Das, R. and Thorpe, R. (2005) Kinetics of development and characteristics of antibodies induced in cancer patients against yeast expressed rdna derived granulocyte macrophage colony stimulating factor (GM-CSF). Cytokine 29: Ronnblom, L., Eloranta, M.L. and Alm, G.V. (2006) The type I interferon system in systemic lupus erythematosus. Arthritis Rheum 54: Rosenberg, A.S. (2003) Immunogenicity of biological therapeutics: A hierarchy of concerns. Dev Biol (Basel) 112: Rudick, R.A. and Panzara, M.A. (2008) Natalizumab for the treatment of relapsing multiple sclerosis. Biologics 2: Rueda, B., Broen, J., Simeon, C., Hesselstrand, R., Diaz, B., Suarez, H. et al. (2009) The STAT4 gene influences the genetic predisposition to systemic sclerosis phenotype. Hum Mol Genet 18:

206 Therapeutic Advances in Drug Safety 2 (3) Visit SAGE journals online Ryschkewitsch, C.F., Jensen, P.N., Monaco, M.C. and Major, E.O. (2010) JC virus persistence following progressive multifocal leukoencephalopathy in multiple sclerosis patients treated with natalizumab. Ann Neurol 68: Schernthaner, G. (1993) Immunogenicity and allergenic potential of animal and human insulins. Diabetes Care 16(Suppl 3): Schweighofer, C.D. and Wendtner, C.M. (2010) First-line treatment of chronic lymphocytic leukemia: role of alemtuzumab. Onco Targets Ther 3: Selmaj, K.W. and Raine, C.S. (1995) Experimental autoimmune encephalomyelitis: immunotherapy with anti-tumor necrosis factor antibodies and soluble tumor necrosis factor receptors. Neurology 45: S44 S49. Sharief, M.K. and Hentges, R. (1991) Association between tumor necrosis factor-alpha and disease progression in patients with multiple sclerosis. N Engl J Med 325: Simin, M., Brok, J., Stimac, D., Gluud, C. and Gluud, L.L. (2007) Cochrane Systematic Review: pegylated interferon plus ribavirin vs. interferon plus ribavirin for chronic hepatitis C. Aliment Pharmacol Ther 25: Sleijfer, S., Bannink, M., Van Gool, A.R., Kruit, W.H. and Stoter, G. (2005) Side effects of interferon-alpha therapy. Pharm World Sci 27: Somani, A.K., Swick, A.R., Cooper, K.D. and McCormick, T.S. (2008) Severe dermatomyositis triggered by interferon beta-1a therapy and associated with enhanced type I interferon signaling. Arch Dermatol 144: Stone, J.H. (2005) Etanercept plus standard therapy for Wegener s granulomatosis. N Engl J Med 352: Suarez-Gestal, M., Calaza, M., Dieguez-Gonzalez, R., Perez-Pampin, E., Pablos, J.L., Navarro, F. et al. (2009a) Rheumatoid arthritis does not share most of the newly identified systemic lupus erythematosus genetic factors. Arthritis Rheum 60: Suarez-Gestal, M., Calaza, M., Endreffy, E., Pullmann, R., Ordi-Ros, J., Domenico Sebastiani, G. et al. (2009b) Replication of recently identified systemic lupus erythematosus genetic associations: a case-control study. Arthritis Res Ther 11: R69. Tak, P.P., Rigby, W.F., Rubbert-Roth, A., Peterfy, C.G., Van Vollenhoven, R.F., Stohl, W. et al. (2011) Inhibition of joint damage and improved clinical outcomes with rituximab plus methotrexate in early active rheumatoid arthritis: the Image Trial. Ann Rheum Dis 70: Unoki, H., Moriyama, A., Tabaru, A., Masumoto, A. and Otsuki, M. (1996) Development of Sjögren s syndrome during treatment with recombinant human interferon-alpha-2b for chronic hepatitis C. J Gastroenterol 31: Valentine, A.D., Meyers, C.A., Kling, M.A., Richelson, E. and Hauser, P. (1998) Mood and cognitive side effects of interferon-alpha therapy. Semin Oncol 25: Van Beers, M.M., Sauerborn, M., Gilli, F., Brinks, V., Schellekens, H. and Jiskoot, W. (2010) Aggregated recombinant human interferon beta induces antibodies but no memory in immune-tolerant transgenic mice. Pharm Res 27: Wada, M., Marusawa, H., Yamada, R., Nasu, A., Osaki, Y., Kudo, M. et al. (2009) Association of genetic polymorphisms with interferon-induced haematologic adverse effects in chronic hepatitis C patients. J Viral Hepat 16: Wadhwa, M., Skog, A.L., Bird, C., Ragnhammar, P., Lilljefors, M., Gaines-Das, R. et al. (1999) Immunogenicity of granulocyte-macrophage colonystimulating factor (GM-CSF) products in patients undergoing combination therapy with GM-CSF. Clin Cancer Res 5: Weiner, G.J. (2010) Rituximab: mechanism of action. Semin Hematol 47: Wynn, D., Kaufman, M., Montalban, X., Vollmer, T., Simon, J., Elkins, J. et al. (2010) Daclizumab in active relapsing multiple sclerosis (choice study): a phase 2, randomised, double-blind, placebocontrolled, add-on trial with interferon beta. Lancet Neurol 9: Yilmaz, S. and Cimen, K.A. (2009) Pegylated interferon alpha-2b induced lupus in a patient with chronic hepatitis B virus infection: case report. Clin Rheumatol 28:

207 BioDrugs DOI /s CURRENT OPINION Interchangeability of Biosimilars: A European Perspective Pekka Kurki 1 Leon van Aerts 2 Elena Wolff-Holz 3 Thijs Giezen 4 Venke Skibeli 5 Martina Weise 6 Springer International Publishing Switzerland 2017 Abstract Many of the best-selling blockbuster biological medicinal products are, or will soon be, facing competition from similar biological medicinal products (biosimilars) in the EU. Biosimilarity is based on the comparability concept, which has been used successfully for several decades to ensure close similarity of a biological product before and after a manufacturing change. Over the last 10 years, experience with biosimilars has shown that even complex biotechnology-derived proteins can be copied successfully. Most best-selling biologicals are used for chronic treatment. This has triggered intensive discussion on the interchangeability of a biosimilar with its reference product, with the main concern being immunogenicity. We explore the theoretical basis of the presumed risks of switching between a biosimilar and its reference product and the available data on switches. Our conclusion is that a switch between comparable versions of the same active substance approved in accordance with EU legislation is not expected to trigger or enhance immunogenicity. On the & Pekka Kurki pekka.kurki@fimea.fi Finnish Medicines Agency, FIMEA, Mannerheimintie 103b, P.O. Box 55, Helsinki, Finland Medicines Evaluation Board, P.O. Box 8275, 3503 RG Utrecht, The Netherlands Paul-Ehrlich-Institut (PEI), Paul-Ehrlich-Straße 51-59, Langen, Germany Foundation Pharmacy for Hospitals in Haarlem, Boerhaavelaan 24, 2035 RC Haarlem, The Netherlands Norwegian Medicines Agency, Sven Oftedalsvei 6, 0950 Oslo, Norway Federal Institute for Drugs and Medical Devices, Kurt-Georg- Kiesinger-Allee 3, Bonn, Germany basis of current knowledge, it is unlikely and very difficult to substantiate that two products, comparable on a population level, would have different safety or efficacy in individual patients upon a switch. Our conclusion is that biosimilars licensed in the EU are interchangeable. Key Points Biosimilars are copy versions of an already existing biological medicinal product. They are high-quality products and as efficacious and safe as the original biological medicines. Because of the high similarity, there is no reason to believe that the body s immune system would react differently to the biosimilar compared with the original biological upon a switch. This view is supported by the current experience with biosimilars on the market and by literature data. In our opinion, switching patients from the original to a biosimilar medicine or vice versa can be considered safe. 1 Introduction A biological medicine can be developed to be highly similar to an existing originator biological medicine (the reference product) according to EU legislation and guidelines issued by the European Medicines Agency (EMA). Similar biological medicinal products (biosimilars) can

208 P. Kurki et al. only be marketed following expiry of the data and patent protection of the reference product [1]. The quality, safety, and efficacy profiles of biosimilars are comparable to those of their reference product. Thus, they are therapeutic alternatives that can be used instead of the reference products. Most best-selling biological products are intended for long-term use in chronic diseases. In this case, most patients who would be eligible for biosimilars are receiving long-term treatment with the reference products. Whether patients can be switched from a reference product to the corresponding biosimilar product is of major importance both in containing pharmacotherapy costs and in promoting patients access to current and future biologicals [2 4]. The interchangeability of biosimilars is controversial [5 9], and the pharmaceutical industry is fueling the discussion [10]. Therefore, it is important to critically evaluate the potential risks of switching from a reference biological product to a corresponding biosimilar product. In this article, interchangeability means the medical practice of changing one medicine for another that is expected to achieve the same clinical effect in a given clinical setting and in any patient on the initiative, or with the agreement of, the prescriber. The decision by the treating physician to exchange one medicine with another medicine with the same therapeutic intent in a given patient is referred to as switching [1]. This article does not deal with the automatic substitution of biosimilars, which is a practice of dispensing one medicine instead of another equivalent and interchangeable medicine at the pharmacy level without consulting the prescriber [1]. 2 Comparability in the Context of Manufacturing Changes The manufacturing process of each biotechnological medicinal product undergoes several changes during its life cycle [11]. Changes in the manufacturing process may have a substantial impact on the product because they give rise to a new version of the product. Therefore, the new and previous versions need to be compared by appropriate tests, usually physico-chemical, structural, and in vitro functional tests, before a change in the manufacturing process can be approved [12]. The demonstration of comparability does not mean that the pre-change and post-change product are identical but that they are highly similar and that the existing knowledge is sufficient to conclude that the observed differences have no adverse impact upon safety or efficacy of the medicinal product. When differences in physico-chemical and structural properties between the pre- and post-change products are observed and their clinical impact remains unknown, additional non-clinical and/or clinical studies need to be conducted. The comparability approach has successfully been applied for more than 2 decades in hundreds of manufacturing changes [11]. When comparability has been demonstrated, the new version can be introduced to the market without informing prescribers, pharmacists, or patients. 3 Comparability in the Context of Biosimilarity From the regulatory and scientific viewpoints, a biosimilar and its reference product contain different versions of the same active substance [13]. Just as comparability needs to be demonstrated in the context of manufacturing changes, the development of biosimilars is based on a comparability exercise with the biosimilar and the reference product. This comparability exercise is built on thorough physico-chemical and structural analyses and in vitro functional tests complemented with clinical studies that are specifically designed to address remaining uncertainties after the preceding analyses and tests [13]. The practical difference between the development of a new version of the same product and the development of a biosimilar product is the much larger scale of comparisons, including clinical data, in the biosimilar scenario, because a biosimilar is developed by a different manufacturer using a different manufacturing process. The long experience with manufacturing changes of marketed biological products in general, and of the reference products in particular, are very useful in the assessment of the potential clinical implications of differences and the magnitude of risks associated with transition from one version of the biological to another. 4 Considerations for Interchangeability Experience with manufacturing changes of biological medicines suggests that switching from pre- to post-manufacturing change versions will very rarely trigger adverse reactions [11]. In addition, switches from one biological to another biological product that is structurally clearly different but has the same therapeutic intent, are common in healthcare [14]. Prescribers should not be misled by publications that do not distinguish biosimilars developed according to the strict requirements of the EU from other less defined copies of biological products used elsewhere [15, 16].

209 Interchangeability of Biosimilars 4.1 Switches between Comparable Products: Pharmacokinetics and Pharmacodynamics Biosimilars and their reference products have the same mechanism of action; highly similar physico-chemical, structural, and in vitro functional properties; and comparable pharmacokinetics/pharmacodynamics, safety, and efficacy, as summarized in the respective European Public Assessment Reports (EPARs) [17]. Therefore, it is unlikely that they would behave differently in a single patient. Suggestions that two comparable versions of the same active substance with comparable pharmacokinetic profiles at the population level would have different pharmacokinetics in individual subjects are theoretical. The scientific situation in this respect is similar to that of generics where, based on demonstrated bioequivalence at the population level, individual patients can be switched between the generic and the respective reference product, even without consultation of the prescriber. The goal of pharmacokinetic studies in the context of a comparability exercise is the detection of potential product-related differences (e.g., due to differences in formulation) and should be distinguished from patient-related factors such as day-to-day intra-subject variability. There are no studies on sources of variance in comparative studies investigating biosimilars and their reference products. However, it seems likely that, as for generics, intra-subject variability rather than product-related variation plays a crucial and decisive role in the variation of drug exposure [18]. 4.2 Immunogenicity Manufacturing changes of biotechnology-derived proteins have not, except for very rare cases, triggered significant immune responses (see the following sections). The current EMA regulatory guideline for immunogenicity assessment of a biotechnology-derived protein, including biosimilars, requires that immunogenicity will always be investigated pre-approval using validated state-of-the-art methods to measure the incidence, titer, neutralizing capacity, and persistence of anti-drug antibodies (ADAs) and their correlations with drug exposure, safety, and efficacy outcomes [19]. It is expected that the immune system will recognize most therapeutic proteins [20]. However, the recognition rarely leads to harmful immune responses [21]. Biosimilars are highly similar to their reference products, and the active substance of most currently approved biosimilars mimic closely or at least partly endogenous substances of the body to which there is an immunological tolerance. Therefore, it is not unexpected that the licensed biosimilars were shown to exhibit immunogenicity comparable with their reference products Switching between Immunogenic Products Nevertheless, some reference products are immunogenic, and an immune response might theoretically evolve towards a class switch of ADAs to immunoglobulin E (IgE). IgE-class antibodies to therapeutic proteins may lead to acute hypersensitivity [22]. Another type of evolution is epitope spreading and subsequently enhanced antibody production, leading to cross-reactive neutralizing ADAs that also target the natural counterpart of the therapeutic protein, such as erythropoietin, as has been observed in patients treated with epoetins [23]. In both cases, T-cell help is required for an enhanced immune reaction after the product switch. T-cell activation is not expected to result from a switch between comparable products since the active substances of the reference and biosimilar products have an identical amino acid sequence and because the T-cell epitopes are linear peptides. For example, infliximab antibodies against the reference product are suggested to recognize the same epitopes in the biosimilar infliximab [24, 25]. In the absence of other T-cell stimulation, aggregates and impurities might be able to bypass T cells and generate activation (danger) signals to B cells [26 29]. This possibility is well known, and the current quality standards exclude products with an unfavorable profile of product- and process-derived impurities and aggregates Immunogenicity Cannot be Excluded for Biologicals Harmful immunogenicity is not expected to be triggered by a switch unless the new version of the reference product, after a manufacturing change or creation of a biosimilar, is of inferior quality, i.e., is not truly comparable. This risk is built into all biological medicinal products because they typically undergo several manufacturing changes during their life cycle [11, 30]. The most prominent example of a rare immunological problem is anti-epoetin antibody-induced pure red cell aplasia (PRCA) in more than 200 patients with chronic kidney disease who were switched from a previous to a new version of epoetin alfa (Eprex ) after a simultaneous change in the product formulation and the administration route [23]. Neutralizing cross-reactive anti-epoetin antibodies were also demonstrated during the clinical development of a biosimilar epoetin alfa in two patients with chronic kidney disease [31]. The antibodies evolved into neutralizing antibodies with subsequent development of PRCA-type clinical findings in one patient. It was shown that the patients had received a product containing increased amounts of aggregates due to tungsten that migrated from the syringe into the product and that mediated unfolding and aggregation of epoetin alfa [32].

210 P. Kurki et al. Thus, both developers of biosimilars and regulators can avoid the entry of an inferior product to the market by clinical comparability studies and by learning from past experience gained with the reference product. Interestingly, the current examples of switch-related immunological adverse drug reactions were found to be caused by differences between drug formulations [23, 32] as well as by improper storage and transport conditions [33] but not by differences in active substances as often suggested in the public discussion on interchangeability and risk of immunogenicity. Thus, an enhanced immune reaction after a switch between products containing different versions of the same active substance is unlikely but may occur very rarely towards new versions of biological products, including biosimilars Switch-Related Immunogenicity: Case Studies Induction of an immune reaction after a switch to a biosimilar will require a difference in antigenicity between the products and lack of immunological tolerance in the host. Switches between non-comparable products provide an exaggerated model for switch-related immunogenicity. In this respect, hemophilia A may represent the worst case scenario. The development of neutralizing ADAs (inhibitors) is a serious and relatively common problem in the treatment of hemophilia A with coagulation factor products. In this situation, immunogenicity is not surprising because the patients lack the immunological tolerance to normal coagulation factor VIII (FVIII) and to coagulation factor analogs. The structures of recombinant FVIII products may be very different, but they still contain elements shared by the normal endogenous FVIII. Switching from one product to another is discouraged to avoid inhibitors, i.e., neutralizing ADAs. However, recent clinical studies suggest that the risk of neutralizing ADAs is not significantly increased upon switching between different coagulation factor products [34, 35]. Another example of a potential switch-related risk is that of interferon (IFN)-b-1b and IFN-b-1a in patients with multiple sclerosis [36]. These two proteins have different amino acid sequences, post-translational modification profiles, and administration routes. In a small study, patients with pre-existing neutralizing IFN-b antibodies were randomized to be switched from subcutaneous administration of IFN-b-1b or IFN-b-1a to either receive intramuscular IFN-b-1a or to continue receiving subcutaneous therapy. Neutralizing antibody titers did not change upon the switch compared with in those who did not switch. Similarly, switching treatment from intravenous to subcutaneous formulations may be viewed as a risky scenario because of anticipated increased immunogenicity. A study investigated switching between the intravenous and subcutaneous formulations of trastuzumab (Herceptin ) in patients with breast cancer. The switch was associated with an increased incidence of ADAs but not adverse events [37]. It has been stated that ADA-positive patients should not switch to a biosimilar [24]. However, clinical studies of ADA-positive patients suggest that the switch from the reference product to the biosimilar has no impact on the incidence of ADAs or the clinical outcome [25, 38 40]. Thus, a switch between comparable versions of the same active substance (as with biosimilars) does not appear to be problematic in clinically stable ADA-positive patients. In conclusion, the observations that switching between products containing structurally different active substances, even between high-risk products, or in ADA-positive/susceptible patients did not enhance immune responses suggest that the risk of exaggerated immune reactions as a result of switching between a biosimilar and its reference products is substantially overrated. 5 Experience from Sequential Use of Biologicals Ebbers et al. [14] reviewed the literature for studies in which patients were switched from a biological product either to a new version of the product (e.g., a new formulation) or to a related product by another manufacturer. The studies involved more than 11,000 patients. Most of the studies did not report any switch-related adverse effects, except for increased or decreased injection site reactions in two studies. The sequential use of biological products is not uncommon in clinical practice. Switches have been reported in more than 20% of patients receiving epoetins and about 10% receiving granulocyte colony-stimulating factors (G-CSFs) in less than 2 years. Switches may involve competing products from different manufacturers or the original and modified second-generation products from the same manufacturer [14]. 5.1 Switching From a Reference Product to its Biosimilar Version The development programs of some biosimilars included studies that involved switches from the reference product to the biosimilar and vice versa. The EPARs available on the EMA website describe the development programs of the authorized biosimilars and provide substantial evidence for the safety of a switch [17].

211 Interchangeability of Biosimilars Omnitrope (Somatropin) During development of the first version of the biosimilar, 44 patients treated with the reference product and 45 patients treated with the biosimilar were compared in a clinical trial. The efficacy and safety of the products were comparable, but the biosimilar was more immunogenic because of impurities. In the next part of the study, the same patients were switched to new improved versions of the biosimilar. No changes in efficacy or safety were observed, and ADAs continuously decreased after the switch to the improved biosimilar Epoetin Alfa Hexal, Binocrit, Abseamed (Epoetin Alfa, HX575) In a randomized pivotal efficacy and safety study, 314 patients with renal anemia treated with the reference product intravenously were switched to the HX575 biosimilar and followed for 54 weeks. Additionally 117 patients were later switched from the reference product to the biosimilar and followed for 26 weeks. Overall, no differences in safety, immunogenicity, or efficacy profiles were demonstrated following the switches Silapo, Retacrit (Epoetin Alfa, SB309) A randomized crossover phase III trial in 313 patients with renal anemia found similar safety, immunogenicity, and efficacy profiles between the biosimilar and the reference product. In this study, half of the study population was switched twice: at randomization and again at crossover Zarzio (Filgrastim) Two pharmacokinetic and pharmacodynamic crossover studies of subcutaneous administration involved 96 healthy volunteers; two pharmacokinetic and pharmacodynamic studies with single administration and crossover studies involved 50 patients. The biosimilar and the reference products had similar safety, immunogenicity, and efficacy profiles after switching Nivestim (Filgrastim) A randomized, multiple-dose, active comparator-controlled, two-way crossover study was conducted. Subjects (n = 24) received five doses of subcutaneous filgrastim during each treatment cycle. The pharmacokinetic, pharmacodynamic, and safety profiles were not dependent on the treatment sequence Abasaglar (Insulin Glargine) The phase III study in type I diabetes mellitus compared the biosimilar insulin glargine against the reference product in 536 adult patients in combination with mealtime insulin lispro. In total, 85% of the patients were receiving the reference product before randomization into the biosimilar or the reference product arms. In this subgroup, no relevant differences in efficacy, safety, or immunogenicity were observed between the switch-to-biosimilar group and the reference group. 5.2 Other Data for Switches Involving Biosimilars Omnitrope (Somatropin) Flodmark et al. [41] studied 98 children who switched to biosimilar somatropin (Omnitrope ) from its reference product. The switch was not associated with deviation from the expected height velocity during the monitored period as predicted by modelling. There were no serious or unexpected adverse events following the switch. Similar results were reported by Romer et al. [42] and Rashid et al. [43] Erythropoiesis-Stimulating Agents The efficacy of treatment with erythropoiesis-stimulating agents (ESA) is monitored and the treatment adjusted according to response. On the basis of a population analysis in patients with renal anemia, it was concluded that ESA consumption and persistence with treatment was not affected following a switch from the reference product to a biosimilar, indicating that efficacy and tolerability was similar for biosimilar and reference ESAs [44] Remsima, Inflectra (Infliximab) Two randomized parallel-group studies were conducted, one in patients with ankylosing spondylitis and the other in patients with rheumatoid arthritis. The long-term extension included a switch from the reference product to the biosimilar. Preliminary results at week 52 of the long-term follow-up studies (until week 102) for those who switched from the reference to the biosimilar suggest that safety, efficacy, and immunogenicity were comparable in the biosimilar maintenance and the biosimilar switch (from reference to biosimilar) groups [45, 46]. Small prospective and retrospective clinical studies have explored the safety and efficacy of switching from a reference product to biosimilar infliximab products in a mixed rheumatological cohort [47], in those with ankylosing spondylitis [39] or

212 P. Kurki et al. psoriasis [48] and in pediatric [49] and adult patients [38, 50, 51] with inflammatory bowel disease (IBD). Results of these studies do not raise concerns about the safety or feasibility of switching. A large controlled switch study of biosimilar infliximab (NOR-SWITCH) with a primary endpoint of disease worsening at 52 weeks involved 481 adult patients with rheumatism, psoriasis, or IBD on stable treatment with reference infliximab. Disease worsening occurred in 26.2 and 29.6% in the reference and biosimilar arms, respectively. The adverse event profiles were comparable [52] Flixabi (Infliximab) Phase III results of the second biosimilar infliximab have been reported as a poster [53]. After 54 weeks, 94 patients treated with the reference product (Remicade ) were switched to SB2 (Flixabi ), while 101 patients continued treatment with the reference product and 201 patients continued SB2 treatment until week 78. The efficacy and safety profiles as well as the incidence of ADAs were comparable in the treatment arms during the switch period Emerging Interchangeability Data on Etanercept and Adalimumab Controlled data on switching back and forth between original etanercept (Enbrel ) and biosimilar etanercept (GP2015) as well as data on switching from original adalimumab (Humira ) to a biosimilar adalimumab (ABP 501) were presented at the US FDA arthritis advisory committee meeting in July 2016 [54]. The data did not give cause for concern. These biosimilars have not been licensed in the EU at the time of the writing of this article European Database for Suspected Serious Adverse Reactions Ebbers et al. [14] reviewed the EudraVigilance database, which covers suspected serious adverse drug reactions reported to the regulatory authorities in the EU, Norway, and Iceland. The EU has a very good signal-detection system via the EudraVigilance database. Search for possible switch-related serious adverse reactions produced only three reports of possible adverse effects in which both an originator and a biosimilar product were used in a patient. The lack of safety signals provides further reassurance of the safety of switching between the reference medicinal product and the biosimilar. 6 Discussion 6.1 Interchangeability Studies In the USA, interchangeability is defined in the legislation and corresponds to automatic substitution in the EU terminology, where interchangeability means changing one medicine for another that is expected to achieve the same clinical effect in a given clinical setting and in any patient on the initiative, or with the agreement of, the prescriber [1]. Thus, the European type of interchangeability is not a legal but a scientific and medical term. For the time being, there are no plans at the EU level to introduce new legal or regulatory requirements for interchangeability studies and thus create two classes of biosimilars. Nevertheless, there is ongoing discussion on the need for specific interchangeability studies in the EU. Proponents of interchangeability studies claim that two independently developed biologicals cannot be classified as interchangeable without specific studies evaluating multiple switches between the reference product and the biosimilar in comparison with a group not undergoing switches. In the absence of data on interchangeability between biosimilars of the same reference product, such biosimilars are also deemed not interchangeable [8, 55]. Unfortunately, there are no detailed descriptions of the proposed interchangeability studies of biosimilars since there are major scientific and practical challenges. The considerations and switch data presented in previous sections suggest the potential impact, if any, of a switch is subtle or rare. Therefore, and because of confounding factors, such as the fluctuating course of chronic diseases, varying intra-patient pharmacokinetics, intra- and inter-observer variability in assessing disease activity, concomitant diseases and/or medications, and batch-tobatch variation, specific interchangeability studies would need to be of substantial size [3, 56]. In our opinion, the feasibility and benefit of such studies is questionable but they might discourage biosimilar development. Instead, we believe that interchangeability can be supported adequately by state-of-the-art physico-chemical, structural, and in vitro functional testing complemented by clinical equivalency in a representative therapeutic indication at the population level. In addition, active post-marketing surveillance of switch-related adverse events by registries and by improved adverse event reporting and analysis will provide the necessary safety net. Such an approach will require action not only by developers and regulators but also by prescribers.

213 Interchangeability of Biosimilars 6.2 Practical Aspects of Interchangeability Clinically significant differences in efficacy between the biosimilar and its reference product are ruled out with reasonable certainty by the comparability exercise. Thus, no change in dosage or dosing regimen is warranted when a patient is switched from a reference product to its biosimilar. In chronic diseases in which biosimilar use is anticipated, patients are already monitored regularly, supporting a safe switch between a biosimilar and the reference product. Patients should receive information about the switch in the same way as for any new medication. Patient education would be needed for different administration devices, such as autoinjectors. In the EU, physicians and/or pharmacists should always document the specific biological medicinal products they prescribe for or dispense to their patients, including trade name, international nonproprietary name (INN), and batch number. This also applies to the reporting of adverse events to allow for proper root cause analysis. Traceability of biosimilar products based on trade name has been reported as very good in the EU [57], whereas recording of batch numbers was poor for biologicals in general and should be improved. Switching between biological medicinal products is common in routine healthcare [14]. There are recommendations on switches between non-biosimilar biological products, including those targeting the same antigen/receptor [58 60], whereas several position papers of medical societies do not regard current biosimilars as interchangeable. However, the latter view is changing with improved understanding of the biosimilarity concept among prescribers and with the increasing experience in switching [61]. There is no official position on interchangeability of a biosimilar at the EU level. Instead, several national regulatory authorities, including the Dutch Medicines Evaluation Board (MEB), the Finnish Medicines Agency Fimea, Healthcare Improvement Scotland, the Irish Health Products Regulatory Authority, and Paul Ehrlich Institute in Germany, have already taken national positions to endorse the interchangeability of biosimilars under the supervision of the prescriber [62 66]. 7 Conclusions The analysis of the theoretical grounds of potential switchrelated adverse effects, data on switching between nonbiosimilar biological products, and experience with switches between biosimilars and their reference products suggests that the potential risks have been exaggerated. In the EU, specific switching studies are required neither by legislation nor by regulatory guidelines. Attempts to provide proof of lack of any switch-related changes in efficacy or safety, including immunogenicity, by specific interchangeability studies would be very demanding and likely still unable to provide definite answers. Our conclusion is that a state-of-the-art demonstration of biosimilarity, together with intensified post-marketing surveillance, is a sufficient and realistic way of ensuring interchangeability of EU-approved biosimilars under supervision of the prescriber. In the authors opinion, biosimilars licensed in the EU are interchangeable if the patient is clinically monitored, will receive the necessary information, and, if needed, training on the administration of the new product. Acknowledgements The authors thank Drs. Meenu Wadhwa and Christian Schneider for their valuable expert comments during the preparation of the manuscript. The generous help of Ms. Mira Juppi is also gratefully acknowledged. Author contributions PK conceptualized the idea for the study. The manuscript was written by PK, LvA, EW-H, TG, VS, and MW. Compliance with Ethical Standards Conflict of interest PK, LvA, EW-H, TG, VS, and MW have no conflicts of interest. They are employees of national regulatory agencies. However, the opinions expressed in this article represent the personal opinions of the authors. Funding No funding was received for the preparation of this manuscript. References 1. European Commission, DG Enterprise and industry. What you Need to Know about Biosimilar Medicinal Products. Process on Corporate Responsibility in the Field of Pharmaceuticals Access to Medicines in Europe. A Consensus Information Document, GROWTH&QueryText=biosimilars&op=Search&swlang=en& form_build_id=form-ca4kk1hs9th2qw_ayevywfkauulcnv Y6kFflINvXlD4&form_id=nexteuropa_europa_search_search_ form. Accessed 30 Sept Rak Tkaczuk K, Jacobs I. Biosimilars in oncology: from development to clinical practice. Semin Oncol. 2014;41(Suppl 3):S Ebbers HC, Chamberlain P. Interchangeability. An insurmountable fifth hurdle? GaBI J 2014;3: Ebbers H, Crow S, Vulto A, Schellekens H. Interchangeability, immunogenicity and biosimilars. Nat Biotechnol. 2012;30: Hlavaty T, Letkovsky J. Biosimilars in the therapy of inflammatory bowel diseases. Eur J Gastroenterol Hepatol. 2014;26: The Alliance for Safe Biologic Medicines (ASBM). Interchangeability and Physician Notification. resources/in-the-states/. Accessed 30 Sept Dranitsaris G, Amir E, Dorward K. Biosimilars of biological drug therapies: regulatory, clinical and commercial considerations. Drugs. 2011;71:

214 P. Kurki et al. 8. Declerck P. Biologicals and biosimilars: a review of the science and its implications. GaBI J. 2012;1: Cortes J, Curigliano G, Dieras V. Expert perspectives on biosimilar monoclonal antibodies in breast cancer. Breast Cancer Res Treat. 2014;144: No authors listed. Building a wall against biosimilars. Nat Biotechnol. 2013;31: Vezér B, Buzás Z, Sebeszta M, Zrubka Z. Authorized manufacturing changes for therapeutic monoclonal antibodies (mabs) in European Public Assessment Report (EPAR) documents. Curr Med Res Opin. 2016;32: European Medicines Agency (EMA). ICH Q5E, CPMP/ICH/ 5721/03. Comparability of Biotechnological/Biological Products: Note for guidance on biotechnological/biological products subject to changes in their manufacturing process. europa.eu/ema/index.jsp?curl=pages/regulation/general/general_ content_ jsp&mid=wc0b01ac b. Accessed 30 Sept European Medicines Agency (EMA): Guideline on Similar Biological Medicinal Products. GB/document_library/Scientific_guideline/2014/10/WC pdf. Accessed 30 Sept Ebbers H, Münzenberg M, Schellekens H. The safety of switching between therapeutic proteins. Expert Opin Biol Ther. 2012;12: Jelkmann W. The ESA scenario gets complex: from biosimilar epoetins to activin traps. Nephrol Dial Transplant. 2015;30: Weise M, Bielsky MC, De Smet K, et al. Biosimilars: why terminology matters. Nat Biotechnol. 2011;29(8): European Medicines Agency (EMA). European Public Assessment reports. pages/medicines/landing/epar_search.jsp&mid=wc0b01ac d124. Accessed 30 Sept Yu Y, Teerenstra S, Neef C, Burger D, Maliepaard M. A comparison of the intrasubject variation in drug exposure between generic and brand-name drugs: a retrospective analysis of replicate design trials. Br J Clin Pharmacol. 2016;81(4): European Medicines Agency (EMA). Immunogenicity assessment of biotechnology-derived therapeutic proteins, rev general_content_ jsp&mid=wc0b01ac c. Accessed 1 Sept Stephens S, Emtage S, Vetterlein O, Chaplin L, Bebbington C, Nesbitt A, et al. Comprehensive pharmacokinetics of a humanized antibody and analysis of residual anti-idiotypic responses. Immunology. 1995;85: Chamberlain P. Multidisciplinary approach to evaluating immunogenicity of biosimilars: lessons learned and open questions based on 10 years experience of the European Union regulatory pathway. Biosimilars. 2014;4: Chung CH, Mirakhur B, Chan E, Le QT, Berlin J, Morse M, et al. Cetuximab-induced anaphylaxis and IgE specific for galactosealpha-1,3-galactose. N Engl J Med. 2008;358: Casadevall N, Nataf JM, Viron B, et al. Pure red-cell aplasia and antierythropoietin antibodies in patients treated with recombinant erythropoietin. N Engl J Med. 2002;346: Ruiz-Argüello MB, Maguregui A, Ruiz Del Agua A, Pascual- Salcedo D, Martínez-Feito A, Jurado T, et al. Antibodies to infliximab in Remicade-treated rheumatic patients show identical reactivity towards biosimilars. Ann Rheum Dis. 2016;75(9): Ben-Horin S, Yavzori M, Benhar I, Fudim E, Picard O, Ungar B, et al. Cross-immunogenicity: antibodies to infliximab in Remicade-treated patients with IBD similarly recognise the biosimilar Remsima. Gut. 2016;65(7): Pate M, Kelly Smith J, Chi D, Krishnaswamy G. Regulation and dysregulation of immunoglobulin E: a molecular and clinical perspective. Clin Mol Allergy. 2010;8: Smith-Garvin J, Koretzky G, Jordan M. T Cell Activation. Annu Rev Immunol. 2009;27: Gallucci S, Matzinger P. Danger signals: SOS to the immune system. Curr Opin Immunol. 2001;13: Wolfert MA, Boons G-J. Adaptive immune activation: glycosylation does matter. Nat Chem Biol. 2013;9: Schneider C. Biosimilars in rheumatology: the wind of change. Ann Rheum Dis. 2013;72: Haag-Weber M, Vetter A, Thyroff-Friesinger U, INJ-study group. Therapeutic equivalence, long-term and safety of HX575 in the treatment of anemia in chronic renal failure patients receiving hemodialysis. Clin Nephrol. 2009;72: Seidl A, Hainzl O, Richter M, Fischer R, Böhm S, Deutel B, et al. Tungsten-induced denaturation and aggregation of epoetin-alfa during primary packaging as a cause of immunogenicity. Pharm Res. 2012;29: Fotiou F, Aravind S, Wang PP, Nerapusee O. Impact of illegal trade on the quality of epoetin alfa in Thailand. Clin Ther. 2009;31: Gouw S, van der Born J, Ljung R, Escuriola C, Cid AR, Claeyssens-Donadel S, et al. Factor VIII products and inhibitor development in severe hemophilia A. N Engl J Med. 2013;368: Santagostino E, Auerswald G, Benson G, Dolan G, Jiménez- Yuste V, Lambert T, et al. Switching treatments in haemophilia: is there a risk of inhibitor development? Eur J Haematol. 2015;94: Gneiss C, Koudouovoh-Tripp PM, Ropele S, Gotwald T, Ehling R, Lutterotti A, et al. Influence of interferon-beta therapy switching on neutralizing antibody titres: results from the Austrian Switch Study. Mult Scler. 2009;15: European Medicines Agency (EMA). CHMP assessment report: herceptin subcutaneous formulation. docs/en_gb/document_library/epar_-_assessment_report_-_ Variation/human/000278/WC pdf. Accessed 30 Sept Smits LJ, Derikx LA, de Jong DJ, Boshuizen RS, van Esch AA, Drenth JP, Hoentjen F. Clinical outcomes following a switch from Remicade to the biosimilar CT-P13 in inflammatory bowel disease patients: a prospective observational cohort study. J Crohns Colitis. 2016;10(11): doi: /ecco-jcc/jjw Benucci M, Gobbi FL, Bandinelli F, Damiani A Infantino M, Grossi V, et al. Safety, efficacy and immunogenicity of switching from innovator to biosimilar infliximab in patients with spondyloarthritis: a 6-month real-life observational study. Immunol Res doi: /s (Epub 23 Jul 2016). 40. Buer LC, Moum BA, Cvancarova M, Warren DJ, Medhus AW, Høivik ML. Switching from Remicade to Remsima is safe and feasible: a prospective, open-label study. J Crohns Colitis doi: /ecco-jcc/jjw166 (Epub 22 September 2016). 41. Flodmark C-E, Lilja K, Woehling H, Järvholm K. Switching from originator to biosimilar human growth hormone using dialogue teamwork: single-center experience from Sweden. Biol Ther. 2013;3: Romer T, Zabransky M, Walczak M, Szalecki M, Balzer S. Effect of switching recombinant human growth hormone: comparative analysis of phase 3 clinical data. Biol Ther. 2011;1: Rashid N, Saenger P, Wu YL, et al. Switching to Omnitrope from other recombinant human growth hormone therapies: retrospective study in an integrated healthcare system. Biol Ther. 2014;4: Hörbrand F, Bramlage P, Fischaleck J, Hasford J, Brunkhorst R. A population-based study comparing biosimilar versus originator

215 Interchangeability of Biosimilars erythropoiesis-stimulating agent consumption in 6,117 patients with renal anaemia. Eur J Clin Pharmacol. 2013;69: Yoo DH, Prodanovic N, Jaworski J, Miranda P, Ramiterre E, Lanzon A, et al. Efficacy and safety of CT-P13 (biosimilar infliximab) in patients with rheumatoid arthritis: comparison between switching from reference infliximab to CT-P13 and continuing CT-P13 in the PLANETRA extension study. Ann Rheum Dis. 2017;76(2): doi: /annrheumdis McKeage K. A review of CT-P13: an infliximab biosimilar. BioDrugs. 2014;28: Nikiphorou E, Kautiainen H, Hannonen P, Asikainen J, Kokko A, Rannio T, Sokka T. Clinical effectiveness of CT-P13 (Infliximab biosimilar) used as a switch from Remicade (infliximab) in patients with established rheumatic disease. Report of clinical experience based on prospective observational data. Expert Opin Biol Ther. 2015;15(12): Dapavo P, Vujic I, Fierro MT, Pietro Q, Sanlorenzo M. The infliximab biosimilar in the treatment of moderate to severe plaque psoriasis. J Am Acad Dermatol. 2016;75(4): doi: /j.jaad Sieczkowska J, Jarzebicka D, Banaszkiewicz A, Plocek A, Gawronska A, Toporowska-Kowalska E, et al. Switching between infliximab originator and biosimilar in paediatric patients with inflammatory bowel disease. Preliminary observations. J Crohns Colitis. 2016;10(2): Park SH, Kim YH, Lee JH, Kwon HJ, Lee SH, Park DI, et al. Post-marketing study of biosimilar infliximab (CT-P13) to evaluate its safety and efficacy in Korea. Expert Rev Gastroenterol Hepatol. 2015;9(Suppl 1): Kolar M, et al. P1410 switching of patients with inflammatory bowel disease from original infliximab (Remicade ) to biosimilar infliximab (Remsima ) is effective and safe one-year followup. In: Abstract presented at the United European Gastroenterology (UEG) Week meeting, October 2016, Vienna, Austria. 52. Jørgensen K, et al. LB15 biosimilar infliximab (CT-P13) is not inferior to originator infliximab: results from the 52-week randomized NOR-SWITCH trial. In: Abstract presented at the United European Gastroenterology (UEG) Week meeting, October 2016, Vienna, Austria. ueg2016/confcal/switch. Accessed 27 Oct Smolen JS, Choe J-Y, Prodanovic N, Niebrzydowski J, Staykov I, Dokoupilova E, et al. (FRI0162) comparable safety and immunogenicity and sustained efficacy after transition To SB2 (an infliximab biosimilar) vs ongoing infliximab reference product in patients with rheumatoid arthritis: results of phase III transition study (a poster. Ann Rheum Dis. 2016;75: US Department of Health and Human Services. US Food and Drug Administration Meeting materials, arthritis advisory committee. Briefing Information for the July 12 13, 2016, Meeting of the Arthritis Advisory Committee (AAC). Drugs/ArthritisAdvisoryCommittee/ucm htm. Accessed 30 Dec Declerck P, Mellstedt H, Danese S. Biosimilars: terms of use. Curr Med Res Opin. 2015;31: Minghetti P, Rocco P, Schellekens H. The constrained prescription interchangeability and substitution of biosimilars. Nat Biotechnol. 2015;33: Vermeer N, Straus S, Mantel-Teeuwisse A, Domergue F, Egberts ACG, Leufkens HGM, de Bruin ML. Traceability of biopharmaceuticals in spontaneous reporting systems: a cross-sectional study in the FDA adverse event reporting system (FAERS) and EudraVigilance data bases. Drug Saf. 2013;36: Valensi P, Benroubi M, Borzi V, Gumprecht J, Kawamori R, Shaban J, et al. Initiating insulin therapy with, or switching existing insulin therapy to, insulin aspart 30/70 (Novomix 30) in routine care: safety and effectiveness in patients with type 2 diabetes in the IMPROVE observational study. Int J Clin Pract. 2009;63: Ormerod AD. Switching biologicals for psoriasis. Br J Dermatol. 2010;163: D Haens GR, Panaccione R, Higgins PD, et al. The London position statement of the world congress of gastroenterology on biological therapy for IBD with the European Crohn s and colitis organisation: when to start, when to stop, which drug to choose and how to predict response? Am J Gastroenterol. 2011;106: Annese V, Avendaño-Solá C, Breedveld F, Ekman N, Giezen TJ, Gomollón F, et al. Roundtable on biosimilars with European regulators and medical societies, Brussels, Belgium, 12 January GaBI J. 2016;5(2): Medicines Evaluation Board (MEB). Are biosimilar medicinal products interchangeable? /08/17/clarification-of-stance-on-biological-and-biosimilar medicines. Accessed 19 Jan Finnish Medicines Agency Fimea. Are biosimilars interchangeable? Biosimilaarien_vaihtokelpoisuus_EN.pdf. Accessed 19 Jan Healthcare Improvement Scotland and NHS. Biosimilar medicines: a national prescribing framework. May healthcareimprovementscotland.org/our_work/technologies_and_ medicines/programme_resources/biosimilar_medicines_frame work.aspx. Accessed 2 Oct Health Products Regulatory Agency (Ireland). Guide to biosimilars for healthcare professionals and patients. December ance-documents/guide-to-biosimilars-for-healthcare-profession als-and-patients-v2.pdf?sfvrsn=18. Accessed 2 Oct Position of Paul-Ehrlich-Institute regarding the use of biosimilars. Update December Accessed 13 Oct 2016.

216 J Toxicol Pathol 2015; 28: Concise Review Therapeutic antibodies: their mechanisms of action and the pathological findings they induce in toxicity studies Masami Suzuki 1, Chie Kato 1, and Atsuhiko Kato 1 1 Research Division, Chugai Pharmaceutical Co., Ltd., Komakado, Gotemba, Shizuoka , Japan Abstract: Antibodies can swiftly provide therapeutics to target disease-related molecules discovered in genomic research. Antibody engineering techniques have been actively developed and these technological innovations have intensified the development of therapeutic antibodies. From the mid-1990 s, a series of therapeutic antibodies were launched that are now being used in clinic. The disease areas that therapeutic antibodies can target have subsequently expanded, and antibodies are currently utilized as pharmaceuticals for cancer, inflammatory disease, organ transplantation, cardiovascular disease, infection, respiratory disease, ophthalmologic disease, and so on. This paper briefly describes the modes of action of therapeutic antibodies. Several non-clinical study results of the pathological changes induced by therapeutic antibodies are also presented to aid the future assessment of the toxic potential of an antibody developed as a therapeutic. (DOI: /tox ; J Toxicol Pathol 2015; 28: ) Key words: therapeutic antibody, mode of action, pathological findings, toxicity study Received: 27 May 2015, Accepted: 28 May 2015 Published online in J-STAGE: 15 June 2015 Corresponding author: M Suzuki ( suzukimsm@chugai-pharm.co.jp) 2015 The Japanese Society of Toxicologic Pathology This is an open-access article distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives (by-ncnd) License < Antibodies can swiftly provide therapeutics to target the disease-related molecules that have been discovered in genomic research because 1) the high level of specificity and affinity to the target molecule or antigen achieves a high level of efficacy and fewer adverse events, 2) their ability to target diverse molecules and the modes of action of the antibodies allow them to be applied to a wide range of therapeutic targets, and 3) modification and refinement by genetic engineering technology and the establishment of recombinant manufacturing technology has made industrial manufacturing possible. Development of therapeutic antibodies boomed in the 1980 s, and the first therapeutic antibody, a mouse antibody, was launched in 1986 as an immunosuppressive agent used during organ transplantation 1 3. Although problems, such as mouse antibodies expressing antigenicity in humans, prevented any therapeutic antibodies being launched in the next 10 years, antibody engineering techniques continued to be actively developed and resulted in techniques to produce chimeric antibodies and humanized antibodies from mouse antibodies 4 8. In chimeric antibodies, 33% of the structure originates from mouse, with variable regions from mouse and constant regions from human, and in humanized antibodies, up to 90% of the structure originates from human, with only the antigen binding site in the variable region (complementarity-determining region) originating from mouse. Furthermore, new techniques made it possible to obtain human antibodies from human antibody phage libraries and human antibody producing mice These technological innovations intensified the development of therapeutic antibodies, and from the mid-1990 s, a series of therapeutic antibodies were launched that are now being used in clinic. The disease areas that therapeutic antibodies can target have subsequently expanded, and antibodies are currently utilized as pharmaceuticals for cancer, inflammatory disease, organ transplantation, cardiovascular disease, infection, respiratory disease, ophthalmologic disease, and so on (Table 1). This paper briefly describes the modes of action of therapeutic antibodies. Several non-clinical study results of the pathological changes induced by therapeutic antibodies are also presented to aid the future assessment of the toxic potential of an antibody that is being developed as a therapeutic. Mechanisms of Action of Therapeutic Antibodies The efficacy of therapeutic antibodies stems from various natural functions of antibodies neutralization, antibody-dependent cell-mediated cytotoxic (ADCC) activity, or complement-dependent cytotoxic (CDC) activity, or the antibody can be utilized as a drug delivery carrier (missile therapy) 1 (Fig. 1). Neutralization: Many therapeutic antibodies utilize neutralization to block the pathophysiological function of

217 134 MoA and Pathological Findings of Therapeutic Antibodies Table 1. Antibody Type, Target Molecule, Mechanism of Action, and Major Indication of Antibody Pharmaceuticals Scientific name Trade name Approval Origin and isotype Target MoA * Licensed indication Cancer Rituximab Rituxan, 1997 Chimeric IgG1 CD20 ADCC, B cell non-hodgkin lymphoma MabThera CDC Trastuzumab Herceptin 1998 Humanized IgG1 HER-2 ADCC, HER-2 positive breast cancer CDC, Blocking Gemtuzumab Mylotarg 2000 Humanized IgG4 CD33 ADC Targeting Leukemia ozogamicin Alemtuzumab Campath, 2001 Humanized IgG1 CD52 ADCC, B-CLL MabCampath CDC Ibritumomab Zevalin 2002 Murine IgG1 CD20 RIT Targeting NHL tiuxetan Tositumomab Bexxar 2003 Murine IgG2 CD20 RIT Targeting NHL iodine 131 Cetuximab Erbitux 2004 Chimeric IgG1 EGFR ADCC, Colorectal, head and neck cancer CDC, Blocking Bevacizumab Avastin 2004 Humanized IgG1 VEGF Blocking Colorectal, lung, breast cancer Panitumumab Vectibix 2006 Human IgG2 EGFR ADCC, Colorectal cancer CDC, Blocking Catumaxomab Removab 2009 Chimeric IgG2a/b ** CD3, EpCAM ADCC, Malignant ascites CDC Denosumab Prolia, Xgeva 2010 Human IgG2 RANKL Blocking Osteoporosis, bone metastasis Ofatumumab Arzerra 2009 Human IgG1 CD20 CDC CLL Brentuximab Adcetris 2011 Chimeric IgG1 CD30 ADC Targeting ALCL and Hodgkin lymphoma vedotin Ipilimumab Yervoy 2011 Human IgG1 CTLA4 Blocking Advanced melanoma Pertuzumab Perjeta 2012 Humanized IgG1 HER-2 Blocking HER-2 positive breast cancer Mogamulizumab Poteligeo 2012 Humanized IgG1 CCR4 ADCC T cell leukemia-lymphoma Obinutuzumab Gazyva 2013 Humanized & glycoengineered CD20 ADCC Chronic lymphocytic leukemia IgG1 Trastuzumab emtansine Kadcyla 2013 Humanized IgG1 HER-2 ADC Targeting HER-2 positive, metastatic breast cancer Vedolizumab Entyvio 2014 Humanized integrin α4β7 Blocking Crohn s disease, ulcerative colitis Pembrolizumab Keytruda 2014 Humanized IgG4κ PD-1 Blocking Unresectable or metastatic melanoma Ramucirumab Cyramza 2014 Human IgG1 VEGFR2 Blocking Metastatic gastric or gastroesophageal junction adenocarcinoma, NSCLC Nivolmab Opdivo 2014 Human IgG4 PD-1 Blocking Malignant melanoma Inflammation Infliximab Remicade 1998 Chimeric IgG1 TNF Blocking RA, ankylosing spondylitis, Crohn s disease, ulcerative colitis Adalimumab Humira 2002 Human IgG1 TNF Blocking RA, Crohn s disease, plaque psoriasis Tocilizumab Actemra, 2005 Humanized IgG1 IL-6R Blocking Castleman s syndrome, RA Roactemra Certolizumab Cimzia 2008 Humanaized Fab TNF Blocking Rheumatoid arthritis, Crohn s disease pegol Canakinumab Ilaris 2009 Human IgG1 IL-1β Blocking Muckle-Wells syndrome Golimumab Simponi 2009 Human IgG1 TNF Blocking RA, psoriatic arthritis, ankylosing spondylitis Belimumab Benlysta 2011 Human IgG1 Blys Blocking Systemic lupus erythematosus Raxibacumab Raxibacumab 2012 Human IgG1 Bacillus anthracis protective antigen Blocking Inhalation anthrax from bacillus anthracis Siltuximab Sylvant 2014 Chimeric IgG1κ IL-6 Blocking Castleman s disease Transplant Muromonab- Orthoclone 1986 Murine IgG2a CD3 Blocking Transplant rejection CD3 OKT3 Daclizumab Zenapax 1997 Humanized IgG1 CD25 Blocking Prophylaxis for transplant rejection Basiliximab Simulect 1998 Chimeric IgG1 CD25 Blocking Prophylaxis for transplant rejection

218 Suzuki, Kato, Kato Table 1. Continued Scientific name Trade name Approval Origin and isotype Others Abciximab 135 Target MoA* ReoPro 1994 Chimera Fab GPIIb/IIIa Blocking Palivizumab Synagis 1998 Humanized IgG1 RSV F protein Blocking Omalizumab Efalizumab*** Natalizumab Ranibizumab Eculizumab Xolair Raptiva Tysabri Lucentis Soliris Humanized IgG1 Humanized IgG1 Humanized IgG4 Humanized Fab Humanized IgG2/4 IgE CD11a α4β1 integrin VEGF Complement 5 Blocking Blocking Blocking Blocking Blocking Stelara 2009 Human IgG1 IL12, IL23-p40 Blocking Ustekinumab Licensed indication Prevention of cardiac ischemic complications Prevention of RSV infection in neonates Severe asthma Psoriasis Multiple sclerosis Macular degeneration Paroxysmal nocturnal hemoglobinuria, atypical hemolytic-uremic syndrome Plaque psoriasis *MOA, mode of action; **bi-specific antibody; *** Approved in 2003 and withdrawn from the market in 2009 because of side effect. CD, cluster of differentiation; CDC, complement-dependent cytotoxicity; ADCC, antibody-dependent cell-mediated cytotoxicity; HER-2, human epidermal growth factor receptor 2; ADC, antibody drug conjugate; B-CLL, B-cell chronic lymphocytic leukemia; RIT, radioimmunotherapy; NHL, non-hodgkin lymphoma; EGFR, epidermal growth factor receptor; VEGF, vascular endothelial growth factor; EpCAM, epithelial cell adhesion molecule; RANKL, receptor activator of nuclear factor kappa-b ligand; ALCL, anaplastic large cell lymphoma; CTLA4, cytotoxic T-lymphocyte antigen 4; NSCLC, non-small cell lung cancer; TNF, tumor necrosis factor; RA, rheumatoid arthritis; IL-6R, interleukin 6 receptor; IL-1β, interleukin 1β; BLys, B lymphocyte stimulator; PSA, prostate antigen; RSV, respiratory syncytial virus; IL-12p40, interleukin 12 p40 subunit. Fig. 1. Mechanisms of action of therapeutic antibodies. their target molecules1. In this case, antibodies bind to the ligand or receptor that is expressed on the cell surface and block the target signaling pathway. When the signaling in the tumor through these ligands or receptors is diminished, it can result in cellular activity being lost, proliferation being inhibited, pro-apoptotic programs being activated, or cells being resensitized to cytotoxic agents16. ADCC: To trigger ADCC, the Fv binding domain of an antibody binds to a specific antigen expressed on the surface of a target cell. The antibody is then able to recruit immune-effector cells (such as macrophages and NK cells) that express various receptors able to bind to the Fc and thus