CMC Strategy Forum NA. Production Cell Line Development and Control of Product Consistency During Cell Cultivation Myths, Risks and Best Practices

Transcription

1 CMC Strategy Forum NA Production Cell Line Development and Control of Product Consistency During Cell Cultivation Myths, Risks and Best Practices

2 Welcome and Introductory Remarks Anthony Lubiniecki - CASSS Intro: Mission and Purpose Barry Cherney - Why are we here? Q5D and Q6B have very high level details on requirements for cell lines. Little detail, not sufficient guidance. Objective: How to characterize cell substrate (analytics), how to manage cell line and expression constructs?

3 Industry View on the Relative Importance of Clonality of Biopharmaceutical-producing Cell Lines Anthony Lubiniecki, Janssen R&D, LLC, Malvern, PA USA PRODUCTION CELL LINE DEVELOPMENT WORKSHOP

4 Overview Definition of clonality, purpose of clonality, health authority requirements and role of clonality in evaluation of product quality. (SEE SPEAKER S SLIDES) Clonality Aside from classical definitions, what is truly meant by clonality is the degree of certainty that cell line represents the progeny of the original cells. Production cell lines are intended to be properly identified, make the appropriate product with the right quality attributes and eliminate the preexisting random genetic variation. Likelihood of genetic mutations being introduced in production cell line increases the earlier the mutation appears in population doubling process.

5 Can clonality be achieved in its true sense? The level of proof to ascertain a cell line comes from a single cell clone varies with the methodology used. It is impossible to create a clonal cell line from immortalized cells. By the time enough cells are generated for culture, random mutations are likely to generate a cell population that is no longer identical to the parental cells. Random/spontaneous mutations are a natural event. We can reasonably generate a cell line with a high degree of clonality with a certain degree of certainty, but not exactly a population identical to the parental cell.

6 Role of clonality in evaluation of product quality Developing an integrated control strategy is key. A suitable control strategy involves, for example, understanding of cell line history, control of raw materials, characterization of product. These factors also impact random mutation events. A version of this talk is available in IABS journal Biologicals.

7 Evolving Biological Product Expression Systems with Host Cell Engineering Linachun Fan, Bristol-Myers Squibb Company, Bloomsbury, NJ USA PRODUCTION CELL LINE DEVELOPMENT WORKSHOP

8 Cell Line Development has a short window of 7 mos. To 1 year, but is the cornerstone for the CMC Development Process. Successful CLD requires ability to reduce timeline, increase productivity and generate acceptable product quality attributes. Requirements: suitable expression systems, transgene DNA and FTE/equipment support.

9 Existing expression systems include bacteria, mammalian cells, yeast, insect cells, plant cells, transgenic animals. There are pro s and con s of each expression system (See table on speaker s slides). Mammalian cell line expression systems are the most used in industry The rest of this talk will focus on presenting aspects of mammalian expression systems.

10 Typical Mammalian Host Cell Process Overview Transfection is more of a black box, which in turns requires a bulk selection activity to come up with the cells that express the desired gene. Next steps: sincle cell cloning, HTS, clone selection, banking and released for use for GMP activities (e.g early dev supply, tox material, FIH material) Ideal timeline: weeks

11 Evolution of cell line development in the past 25 years 1990s longer timeline, clonality issues, low productivity. 2000s HTS and FACS, better selection markers, shorter timelines higher productivity 2010s targeted DNA integration, FACS, much shorter timelines, very high and reliable productivity % shortened timeline 8-10 fold increased productivity

12 The ideal HCL Robust behavior 9growth) Highly productive Acceptable PQ Energy efficient Downstream friendly eg. lower HCPs Stable expression How to achieve ideal HCL Improve plasmid engineering Utilize gene editing and targeted integration technologies Utilize cell marker engineering to facilitate screening Utilize metabolic engineering to select high producing cell lines Tools: HTS scale-down models, PTMs analytics Streamlining the HCL technologies facilitates management of diverse portfolio of biologics (eg. mabs, large molecules, combined therapies)

13 Case Studies Case #1: HCE enables afucosylated mab production Eliminate fucose sources Gene knock-out or blockage of GDP-4-keto mannose Does not produce a pure population, but high percentage of cell population that do not express fucos (> 80%) Outcome: ADCC activity is enhanced with afucosylated mab

14 Case #2: Transition to new cell line platform New platform increases titer, robust product quality, enables HTS and reduced timelines for generating a cell line

15 Case #3:Targeted integration in CHO cells Recombinase-based expression vector Current process is random, unpredictable genetic heterogeneity, need aggressive screening process (time and $) Reduction of timeline weeks with titers of g/l New approach allows to screen different host cells. The end result of combining technologies is ideal HCE.

16 Characterization of Production Cell Lines Part One: Industry Luhong He, Eli Lilly and Company, Indianapolis, IN USA PRODUCTION CELL LINE DEVELOPMENT WORKSHOP

17 Introduction concepts of production cell lines They are not clones, but clonally-derived Objective is to have suitable product quality and manufacturing consistency Current platforms cannot achieve genetic homogeneity Transgenes extensively engineered, highly expressed CHO hosts have genetic plasticity

18 Minimization of heterogeneity risks through characterization Screening of cell line candidates Characterize on risk-based approach Prior to commercialization characterization per Q5B Currently there is no regulatory guidance on how to execute characterization

19 Characterization approach for registration batches Demonstrate genetic stability of MCB and WCB at intended EOP. EOP (in vitro cell age) is a critical inprocess control parameter. Not necessary to demonstrate genetic stability earlier than EOP on WCB. Characterization Confirm coding gene sequence Demonstrate structure integrity Transgene copy number analysis

20 Characterization for early stage Primary goal is to eliminate cell lines that have heterogeneous population (population drift) by phenotypic analysis of fed batch from early generation through late generation Assess productivity under different process conditions (eg. Titer) Assess transgene population in cell line using single cell qpcr. Confirm assessment with orthogonal methods (eg. intracellular staining, sc RT-PCR). Differences in Ct values for transgene (higher Ct values) render population unsuitable Using these approaches, homogeneous populations can be identified. About 80% of tested cell lines are now considered suitable.

21 Transgene integrity evaluation during early stages To reduce risk of selecting cells that have additional DNA sequences that did not splice appropriately, or mutations that could produce sequence variants and aberrant protein production Phase I - Cryptic aberrant mrna splicing can result in intron reading in-frame. For mabs it leads to impact to LC/HC, bispecificty or protein productions. Detection is through RT- PCR, Northern blots and LC/MS Cell culture conditions may impact the ratio of aberrant spliced mrna. Presence of aberrant protein sequences impact purification process. Phase II Cryptic splices are repaired and new production cell lines generated with the repaired transgenes.

22 Host-target characterization Host target engineering aims to improve the development process and reduce impurities that may impact immunogenicity/safety (eg. HCPs) and increase product stability Characterization strategy Assessment of genotype and phenotype for production cell line Use of Sanger sequencing to confirm the right sequences are stable Assess target product quality attributes

23 New Developments and challenges Suitable expression systems for new types of molecules such as co-expression of transgene with enzymes or hetero mabs How to approach characterization of these expression systems Host cells for targeted integration Detection of sequence variants at low levels

24 Emerging technologies include nextgeneration sequencing (NGS) and advancement in screening and investigational tools (Example: Illumina sequencing systems) With emerging technologies being highly sensitive, how can we determine if what we are detected is considered a relevant mutation or it is a component of increased sensitivity?

25 Use of NGS to characterize cell lines Sanger sequencing requires high levels of mutation for suitable detection. These are usually detected in early stages of transfection. NGS can be used to detect mutations transfection process later on (bulk).

26 Characterization of Production Cell Lines Part Two: Health Authority Regulatory Perspective on the Evaluation of Clonality of Mammalian Cell Banks Rachel Novak, CDER, FDA, Silver Spring, MD USA PRODUCTION CELL LINE DEVELOPMENT WORKSHOP

27 Focus is to discuss regulatory expectations for clonality, how regulators assess clonality, management of cell lines. Expectations of clonality Scientific considerations Minimization of heterogeneity Non-clonal cell banks present potential to impact manufacturing process when upstream changes occur ICH Q5D

28 Frye s paper on industry position presents a mutual understanding between industry and regulatory agency Generating a safe product Nature of the cells, complexity, potential heterogeneity and control is well understood Assurance of clonality is only a single aspect of the control strategy

29 It s all about minimizing risk for the patient Start with a cell line that has high likelihood of monoclonality Perform small-scale studies to support stability Develop a robust control system for cell culture to minimize changes that could impact clonality in the production cell line Non-clonality can also impact process consistency

30 Expectation for Cell Line Development for INDs Cell bank testing and characterization Parental cell line history Description of cloning process Adaptations to serum-free conditions should be performed PRIOR to the cloning step, otherwise justified. Reviewers will perform initial assessment of information for clonality of MCB. Clonality is NOT a requirements. Lack of clonality does not entail hold or refusal. Probability (numerical calculations) and assurance (tools and data) are evaluated

31 Acceptable PROBABILITY Two rounds of limited dilution or use of supplemental technology such as FACS Acceptable ASSURANCE If acceptable PROBABILITY, ASSURANCE is not required ASSURANCE may be provided with additional supporting data that can be used to justify no additional control strategy modifications Control strategy modifications are needed if low probability and low assurance Reviewers look at total evidence in package to determine if additional information is needed.

32 Expectations for Cell Line Development for BLA If low probability and assurance, not necessarily refusal. Control strategy system changes can be made Example strategies to manage non-clonal cell banks Additional specifications such as SV, glycosylation (even if MOA not expected) Tighter IPCs (eg. in vitro cell age, requalification of WCB) Establishing additional CPPs (eg. cell growth) Expectations for Cell Line Development Post Licensure Report potential non-clonality of MCBs ASAP Identify sub-clones and effect to CQAs and process attributes Establish additional process controls. Statistical evaluation of process drift (prospective). May result in limiting manufacturing until issue is addressed Re-cloning MCB Usually advised in early stages Late stage development approach recommended is usually establishing tighter control strategies

33 Lianchun Fan, Bristol-Myers Squibb Company, USA Steffen Gross, Paul-Ehrlich-Institut, Germany Luhong He, Eli Lilly and Company, USA Michael Laird, Genentech, a Member of Roche Group, USA Anthony Lubiniecki, Janssen R&D, LLC, USA Trent Munro, Amgen Inc., USA Rachel Novak, CDER, FDA, USA PRODUCTION CELL LINE DEVELOPMENT WORKSHOP: PANEL DISCUSSION QUESTIONS AND ANSWERS

34 Q&A Q to R. Novak Concerns and issues from FDA perspective were nicely presented and clear. Can you elaborate more on what constitutes assurance? What kind of experience have you had with properly justified assurance? It always depends on the product and process. Gene sequencing, FISH, sub-clone sequencing are examples of suitable tools. Depending on how the cell line is cloned and the relevance of the data. The recommendation is to provide as much information as possible on process parameters and analytical parameters in addition to simple calculations or numeric output.

35 Q to R. Novak What happens if high assurance of single clonality is not achieved, but the intended product meets quality expectations? Understanding the process and the potential uncertainty that can be introduced is important. It is important to look at future issues (mitigating risk) that could arise during and post-commercialization. Clonality is NOT a requirement and does not entail hold or refusal to file. Clonality is only a single aspect of control strategy. CQAs and processes need to be well understood and controls.

36 Q to L. He Do we need to front load more characterization work in early stages of cell line development? In early phases it is more appropriate to focus on screening. Later on, characterization becomes more comprehensive and focused. Since commercial process is rarely defined at early stages, detailed characterization is appropriate once the process is more or less locked and defined.

37 Q to R. Novak One of the bigger risks identified by FDA is sub-cloning: What is the concern with subcloning, if the target profile and comparability to the original clone are also considered requirement and are demonstrated appropriately by the sponsor? How is sub-cloning considered a more risky change than other changes? The potential risk of getting a different product is the biggest concern. In general, sub-cloning is not recommended, given the time needed to complete is long and the risk of not being able to achieve product quality and target product profile is high.

38 Q to R. Novak and S. Gross How can we harmonize expectations across regulatory agencies? S. Gross: Accepting changes, even if considered subtle, requires extensive characterization. Assessment in changes of copy numbers and using other techniques are examples for suitable demonstration of clonality. Acceptance criteria for comparability, consistency of product quality should be applied. We have accepted submissions for changes, as long as consistency of starting material is fully considered and described. Full characterization of cell line is expected at early stages as well. R. Novak Full genetic characterization is expected at BLA. For IND full characterization to demonstrate safety is required. Normally EOP data is not provided in IND.

39 Q to Panel For older cell lines not cloned under current standards, sub-cloning may be an option: Can anyone speak to experience with submissions and expectations of acceptable product quality from these sub-clones? Is the experience that no gross inconsistencies are detected? M. Laird: Genentech has had some experience. When clonality is assessed, sufficient genetic similarities are seen; productivity usually is low, but clones can be found across the range of production. Product quality may shift slightly. Some sob-clones are non-producers. T. Munro Amgen has had same experience as Genetech with finding some heterogeneity and PQs outside the established range. Clonality usually can be justified, but evaluation of the total package is necessary. L. Fan Experience with sub-cloning legacy cell lines. Characterization of analytical differences is important. S. Gross Asking to characterize fully the sub-cloning product

40 Q to Panel and T. Munro- What is the opinion on using pooled cells for Phase I provided demonstration of a controlled process and developing a suitable comparability protocol for later stages? T. Munro: This is a good way of accelerating development. It is very important to understand the process and product quality assessments are suitably qualified. M. Laird: Providing a full description of the pools in the Phase I submission has been accepted by the regulators. However, using pools has not been considered or submitted to generate commercial material. R. Novak: The agency normally does not recommend use of pools. However, if lot-to-lot consistency and comparability are demonstrated within the submission package, it is acceptable. The agency highlights this approach is very risky for comparability. S. Kennett (CDER, FDA) Even for Phase I clinical studies, pools are not recommended.

41 Q to Panel- Is it realistic to use transient production cells in lieu of pools for clinical studies? T. Munro: Work is underway to disclose the science used to generate such kind of material. Transient material, analytical methods for PQ determination need to be greatly enhanced. There are novel technologies to generate pools: semi-targeted integration, etc. S. Gross: The use of transient transfects is not generally accepted for now. F. Cornel (Health Canada): FC Comment: Actually Health Canada is not currently looking into this because we have not had any sponsor proposing this type of approach. However, what I mentioned during the panel discussion is that Health Canada does allow transient expression in plant cells ( if anyone is interested we have a guidance document on plant-derived biologics: What I said was that in principle, some of the same framework/principles applied to transient expression in plants would likely also apply to mammalian cells. Of course, there would be other issues specific to mammalian cells that would also have to be addressed.

42 Q to R. Novak: If an impurity is found to originate from more than one possible source, how should that be characterized? R. Novak: It is not necessary to determine the origin. It is more relevant to perform risk assessment of whether change is important to PQ.

43 Q based to Regulators: How do we get regulatory consensus on general philosophy for characterization, PTM s and PQ assessments? (There is a recurrent comment from EU regulators comment to refer to the EU monograph for mabs) S. Gross: Safety comes first. The reference helps drive the correct approach for characterization. Always notify agency of changes and any differences in PTMs and PQ as soon as possible. T. Munro: There should be a concerted scientific discussion for regulations on advancements in CMC changes. A dialog on CQA should be established between agency representatives.

44 Q to Panel: ICH 5 guidance was put in place in the 90s, do you consider the documents are ready for revision? A. Lubiniecki: The guidance was written specifically with the objective of leaving out the details on how. The purpose of the guidance was for requirements for filing and not how to generate the data. R. Novak: The regulatory agencies don t want to hinder innovation by adding too much details to ICH5.

45 Q to R. Novak Provide examples of other types of adaptations(other than serum free)and considerations the reviewers take when evaluating. R. Novak: Example: Feed changes. Approach is similar to evaluations for serum-free adaptation. S. Gross Changes in media composition is another example.

46 E. Shacter: Comment to regulators - There is a concern with letting clinical divisions make decisions on CQA and CMC process. The concept of interfacing only with clinical division for decision on clonality is counterproductive, not adequate, misconception. Thoughts?

47 Q to Panel On host cell engineering: Provide your thoughts on approaches presented by L. Fan for characterization? T. Munro: Need to balance the use of novel technologies and their relevance to the end result. Sequence information may or may not be relevant to the gene of interest. R. Novak: Be careful what you are looking for and determine how meaningful the findings are to the product and process.

48 Q to Regulator from L. He For transgenes and host cell targets our general strategy is to consider all aspects of characterization. Setting acceptance criteria up front for characterization helps. Thoughts? S. Gross Will not accept single data set. It has to be submitted as a comprehensive data set. It is important to characterize the transgene early on. L. Fan The technology keeps improving over time and the sensitivity and detection is greater than methodologies applied on legacy processes. As long as we have analytical tools to understand changes, we should keep an open mind of all changes that can be detected.

49 Q from C. Frye: CHO cell lines are prone to generate homologs to hamster antigens with detection levels ppm. Is that a concern? S. Gross: This has been previously reported (co-purification) impacting final product quality. We have requested to restrict shelf life and introduce additional purification steps in the process. R. Novak: No experience with this type of case. M. Laird: Should sponsors knock-out that gene? A. Lubiniecki: Long time ago, an unnamed agency had the same question to a sponsor, before the available technology was sensitive enough to detect these homologs. When technology improved, these homologs were found. The sponsor sought to purify the hamster homolog. The response from agency was unexpected, as their thought was that the homologs may have been the reason why the product was safe and effective in the first place and purifying them out was probably not a good idea. F. Cornel (Health Canada): This depends on the effect of the co-purified protein. If the protein is just another HCP, or if it elicits undesired response. Look for reactive antibodies in patients. FC Comment: Should also carefully look at stability studies to determine if the co-purified protein has any type of activity which could change the stability profile of the product.

50 Q for Panel: If host cell engineering (HCE) is as characterized as transgene, why would there be additional complexities to employ their use? T. Munro: Look at history of cells, like development of DHFR defective cells. Now that we have more precise molecular tools for characterization, we should be fine with using aggressive engineering. R. Novak: The same expectations apply to all cell lines in terms of demonstration of clonality and additional characterization. Include additional details of how the cell line was engineered. S. Gross: Humanized cell lines require full exploration of safety and full characterization.

51 Q to Panel and audience: If an unstable cell line is detected when comparing MCB to EOP, what additional work is needed to justify their suitability for use? Audience member (from Industry): In our case, Southern blot showed differences and there were confirmed variations in product quality. To determine what happened, sub-cloning was used. The sub-cloned material had the same issues. Details were filed in Phase 3 IND with acceptable outcome from agency. R. Novak: Slides presented contain guidance on management of non-clonal cell bank. Characterize impact, detail changes in quality attributes, justify no impact. Understanding the impact is the key message. S. Gross Establishing controls and demonstrating effectiveness of controls is necessary.

52 Q to Panel: On consistency of MCB to EOP - We have seen some changes in genetic RNA expression. However clonality demonstrated and PQ remains the same. When is the DNA integrity an issue vs. RNA/protein expression remaining acceptable/consistent over time? How to pose this in the IND/IMPD? M. Laird: Recommend to build new cell line, if feasible. S. Gross: No clear rules. If the copy number is decreasing over time, it is a risk. R. Novak: Characterize and justify no impact. T. Munro: Depending of the case, lengthy molecular biology investigations and efforts are always outlined in the filing. L. He: Investigate and determine impact. Provide supporting data demonstrating control of the process as part of the package.

53 Q for Panel: Clonality and diagnostics. If a variant has been consistently identified in the population and fully characterized, could the presence of that variant be used as a marker for consistency instead of providing a full comparability study? R. Novak: If found at late stage (close to licensure), control around qualification of WCBs is recommended. Perform a risk assessment. If the variant has always been present and clinical data is available, showing no impact is acceptable. T. Munro: Scientific understanding of the observation is important to justify this approach.

54 Q for Panel: Feedback on guidance to sponsors on expectations of suitable analytical techniques S. Gross/R. Novak: New technologies and their acceptance take time. The more detail from Sponsors in packages submitted, the better for track to acceptance.

55 Screening Approaches for Product Quality to Enable Attribute-driven Cell Line Development with an Eye towards Commercialization Christopher Sellick, MedImmune Limited, Cambridge, United Kingdom CONTROL OF PRODUCT CONSISTENCY DURING CELL CULTURE CULTIVATION

56 Main objective is balancing product titer and product quality. With growth in portfolio diversity (mabs vs nonmabs) it becomes a challenge to engage in characterization. Need to find efficiencies. Need to find a good interaction between developability, target profile and molecular models. Parallel discovery (R&D) with early development. Involve Research and CMC early on.

57 Attribute-focused analytical development R&D and Developability Gain product knowledge Define initial pcqas Define high throughput methods for each pcqa and evaluate methods Implementation of methods for cell line and process development. Typical timeline for cell-line development (CLD) > 20 weeks Demonstrate clonality and suitable product quality outputs. Process for CLD at MedImmune is aligned with processes described in today s morning session. Titer screening is used as the end measurement for clone selection

58 Need high throughput to analyze 100s-1000s of samples: multi-well screening, AMBR, 3-500L bioreactors Analysis of cell culture methods Low sample volumes and protein concentration Rapid turn around time

59 Case Study #1: Changing a production cell line for improved yield Higher titer required for commercialization. ~1 g/l for Phase I-II Analytical tools ProA HPLC for titer and phenotypic stability Charge Heterogeneity (IEC) Monomer content 9SEC) Glycosylation (Gal/a-Gal and Man5 levels) Static vs. shaking culture conditions. AMBR analysis to compare new cell line against original cell line. New cell line was developed with an improved in-house expression system, new platform for feed and media and improved analytics.

60 Case Study #2: Increase half life of molecule Legacy molecule had high fragmentation rate and inconsistent glycosylation profile Approach Updated media and feed processes to reduce fragmentation and sialylation state of the molecule Analytical tools Used for screening and CLD Octet-based lectin assay: Lectins are used to bind captured IGG with a biosensor; correlation between MAL 1 binding and silalylated glycans SDS-CE GXII Touch Aggregation and terminal sialic acid were directly proportional to titer Process development completed 3 months ahead of schedule.

61 Case Study #3: For a multimeric fusion Fc, confirm correct assembly, reduce LMWs and HMWs, consistent glycosylation Analytical tools Octet for Titer UPLC-SEC for aggregation and oligomers Glycan analysis AMBR screening Select cells producing less fragmentation and misassembly and high Man5 levels (High % of hexameric product)

62 Advances in Product Characterization During Cell Cultivation Jason Rouse, Pfizer, Inc., Andover, MA USA CONTROL OF PRODUCT CONSISTENCY DURING CELL CULTURE CULTIVATION

63 Sequence integrity is vital to product quality Assume 100% fidelity, but not achievable Confirm protein expression is the desired outcome Development of a method to detect sequence variants Use of FAbricator enzyme and LC/MS Peptide Map tools to identify peaks by increasing the resolution of the mass analysis. This technology is applied to the entire portfolio using defined workflows for proteomics UHR HCPs -Spike known 48 proteins into IgG1 mab to test MS LOD (~0.1%) SVs detected through genetic sequencing and LC/UV/MS in fed shakes for high levels (1-3%) and in 1L bioreactors for low levels (~0.1%) Subunit analysis lead to detection of misincorporations that cannot be detected through genetic analysis.

64 Mis-incorpration analysis is very sensitive. Therefore, a thorough understanding of the source is important (Incorporated by fed shake or real mis-incorporations?) Tests are applied at early and late stages. PQ assessments for screening allow for selection of suitable clones. Example: Detection of trisulfides (common PTM found in LC/HC in IGg1s) using intact mab SEC/MS Are trisulfides produced by clone, media and/or cell culture? Managing the levels can be achieved by controlling the cell culture process

65 MedImmune has developed a process post-fih and product characterization to Optimize and deliver a commercial-ready mfg process Using platform analytical methods Case Study: Depletion of valine and other aminoacids detected using LC-MS/MS LC MS/MS showed mis-incorporations of aminoacids Mis-incorporations were correlated to depletion of amino acids in the media Solution: Improve media amino acid content by increasing the feed rate

66 Lessons Learned Streamline and evolve using new technology. Reassess work consistently. Understand and control trisulfide levels and misincorporations to ensure product consistency.

67 Product Consistency during Cell Cultivation Regulatory Expectations Steffen Gross, Paul-Ehrlich-Institut, Langen, Germany CONTROL OF PRODUCT CONSISTENCY DURING CELL CULTURE CULTIVATION

68 Similar regulatory points of view between EU agencies and FDA. ICH Q5 D, B and A guidance Confirm aminoacid sequence to 100% coverage Two-tiered cell bank: MCBs and WCBs All WCBs must be well characterized and qualified for Identity Purity

69 Case Study #1: Limit determined for PDLs Gene copy number indicated the more suitable criteria for cell age was 60 PDLs Where and how to set controls? Focus on production bioreactor upstream (cell viability, etc.) and dowsntream purification Case study #2: No data available for small-scale studies for cell seeding density. Recommendation was to reduce limit until smallscale data and development was submitted

70 Case Study #3 Introduction of new CHO cell line with more productivity and modification of USP and DSP. Comparability showed presence of SV between Phase I and Ph III material Sponsor submitted good package including crystal structure, folding patters, potency etc. to justify change Root cause: Increased method sensitivity Solution: Introduce relevant controls in the process

71 Case Study #4: HC Variant identified during process development. Process1 vs Process 2 increased 3X Sponsor provided details on population doubling and confirmed clonality No evidence of potency, effector function or binding effects Toxicity was not impacted Response from Agency: Acceptable process change as long as a control strategy is put in place to control population doubling time and establishment of action limits for the HC variant

72 Other aspects to consider Differences in glycosylation profiles (eg. Mannose) impacting PK, immunogenicity Insertion of amino acid sequences increasing sequence variants PLBL2 co-elution in ProA Chromatography step

73 Recommendations for Process Development: Change in Cell Line Define pcqas early one Compare development and QC data Compare IPCs Confirmation when transitioning to commercial space Perform human PK studies for comparability Provide pivotal efficacy studies It is essential to correlate nucleic acid analysis with evaluation of the final purified protein to ensure quality of a product.

74 Regulatory Expectations and Case Studies for Product Cell Line Development Juhong Liu, CDER, FDA, Silver Spring, MD USA CONTROL OF PRODUCT CONSISTENCY DURING CELL CULTURE CULTIVATION

75 Production Cell Banks: Expectations Consistency during development and lifetime of the product Transition to transgenic animals and mammalian expression systems Transgenic high yield, not suitable for variations in engineering Mammalian (CHO Cells) reliable, well established safety, suitable for variations in engineering, able to control clone selection and process changes

76 Traditional approach to selection of cell banks Generate expression vector Transfection with selectable markers Growth in single cell suspension Clone cell with desired product Screen for suitability and scale-up Pro s Well characterized, stable system Scalability Productivity Defined CQAs Con s Time consuming and labor intensive Platforms may not be suitable for commercial production

77 Is there a Phase-Appropriate approach for cell line development to parallel clinical studies? Analogy: Initial days of cell phones ( non-smart ) required pre-payment for usage. Smart phones require post-payment for data usage. For cell line development, do we have system in place to avoid post payment: Pay now or pay later? Defined analytical target profile? Prior knowledge of product class? Do we have a near locked manufacturing process? DO we have well qualified methods for CQAs? Do we have a defined comparability protocol?

78 Unplanned changes driving cell line changes Monoclonality is lost over time Scale up produces different CQA profiles Insufficient titer to support commercial production Contamination of MCB/WCB Adaptation to serum-free processes Use of a different host cell line Biosimilars are a good example where cell bank change risk is low, as the CQAs and limits are predefined by the CQAs of the priginal product

79 Expectations for changes in cell line Very detailed comparability package mabs Affinity and potency Glycosylation profile and non-impact to ADCC and CDC Same product-related impurity profile Non-Mab Glycosylation profile, half-life Assembly Monomer, cleavage profile Biological activity Non clinical and clinical studies Additional studies depend on the changes detected otherwise with cell line change

80 Case Study #1: mab production cell line change Sponsor proposes to change after Phase II Data to support the change Fucntional analysis and MOA demonstrated Analytical Comparability for glycosylation, cell-based function and other attributes did not reveal significant differenes Nonclinical studies reveal PK not changed Outcome: Pivotal trail proceeded as planned. New clinical material used to validate the change

81 Case Study #2: Post-approval WCB change MCB established with outdated technology that produced slightly diverging WCBs Sponsor provided comprehensive comparability studies package with process improvements. While the root cause of the divergence in WCBs was not found, the sponsor had extensive stability studies and manufacturiong experience. Thus, no additional clinica/non-clinical studies required.

82 Case Study #3- Post-Approval MCB Change Sponsor provided full packages, however bioequivalence failed. Change was not approved Case Study # 4 _ Post Approval MCB change Sponsor presented package with changes in glycoform distribution, but bioequivalence was demonstrated. Change was approved.

83 Concluding remarks ile changes in production cell bank are considered high risk and evaluated conservatively by regulators, the following aspects are considered when Sponsor requests a change Timing Quality of the justification package Available historical and experience with process Incorporation of additional process controls

84 Fiona Cornel, Health Canada, Canada Steffen Gross, Paul-Ehrlich-Institut, Langen, Germany Juhong Liu, CDER, FDA, USA Ilona Reischl, AGES-Austrian Agency for Health and Food Safety, Austria Jason Rouse, Pfizer, Inc., USA Christopher Sellick, MedImmune Limited, United Kingdom PANEL DISCUSSION QUESTIONS AND ANSWERS: CONTROL OF PRODUCT CONSISTENCY DURING CELL CULTURE CULTIVATION

85 Q to Regulators: What about changes in cell bank in the middle of a pivotal trial? Would anyone agree to it? S. Gross: Advised not to change. If absolutely necessary, evaluate differences and provide a package that is sufficient to determine if efficacy and safety are impacted or not. J. Liu: FDA has similar position as EU regulators. If necessary, need clinical data derived from this change and a very comprehensive comparability study.

86 Q to J. Rouse: Did detected analytical changes for SV correlate with changing strategy of the expression system? J. Rouse: Favorite clones are selected first. SV are a fairly new finding and they do not appear consistently and do not grant Q to FDA: What are considerations to change cell banks after licensure? J. Liu: Start with comparability. If we have a good manufacturing history for clinical lots and no significant differences are detected and within tolerability of known attributes, minimal work is required otherwise. S. Gross: Filing to replace would require to file for a variation. No filing required now, if no impact to PQ. If the analytical tests are changes, filing is required.

87 Q to S. Gross: One example where SV was increased with pop doubling? What was the root cause? No clear root cause was determined. Q. to J. Rouse: Can you comment on how was NGS method determined to be robust for? LC MS/MS was analyzed side-by-side with NGS to obtain sufficient evidence of robustness. No comments on precision for NGS. Comment from audience: Precision for NGS is very good. Orthogonal analysis with MS/MS and Sanger sequencing.

88 Q for C. Sellick: When shaking vs static analysis? Were there any findings of correlations between quality attributes from clones generated? Sometimes. Depending of the cause. Sometimes the correlations were more process dependent. If process mas the most likely cause, then focus on screening was deprioritized.

89 How do reviewers view presence of SV vs misincorporation? Are they considered equivalent? What levels Sequence Variants would be considered concerning? J. Liu: Using clinical material available, mapping the levels of sequence variants in the clinic is useful. If levels are consistent and no safety, efficacy issues, no concern. Mis-incoporations can be controlled and easier to justify. SV are genetic in nature and require more comprehensive work. F. Cornel: The more you know about the SV the better.

90 Q for Regulators: Sequence detection by NGS when amino acid is not changing, but not nucleotides, what are the considerations on handling? S. Gross: If nucleic acid mutation or exchange is found in sequence, this is not seen in the dossier. Industry: Clone is not used

91 Q to J. Liu: It is easier to change cell line on biosimilar than on an original? J. Liu: Not necessarily. If during scale-up, attributes are not within range of original product, Sponsor may choose another clone. The product at that point has not gone into patients at all. Comment meant during clone selection, where CQAs are known from original product. If it happens during clinical studies, and PQs are outside range, there is an issue on biosimilarity.

92 Q to Regulators: If new variants found at low levels with no effect on potency, what are requirements to address potential issues with immunogenicity? Audience response: Immunogenicity study is performed. Q to S. Gross: Case study on sig drop in copy number? Was the Sponsor asked to limit pop doubling or to submit characterization package? Sponsor to submit full characterization package. Since the probability and risk is high for patients.

93 Q to S. Gross: Why is a drop in productivity a concern to regulator (> 25%), if PQ is not altered? It is more of a Sponsor risk. S. Gross: It is a matter of process consistency if productivity changes drastically. FDA in Audience: It is usually an indicator of loss of control in manufacturing process and the Sponsor needs to address it. J. Liu: Indicates a loss of process control. A comparability study is recommended.

94 Q to Panel: When talking about types of CQAs for a given molecule, what is the opinion about simpler molecules (a-glycosylated) that have direct mechanisms of action, does that play on simplifying assessments of cell banks and when they need changing S. Gross: These aspects are taken into consideration for requests of complexity for packages. I. Reischl: Consistency must be demonstrated F. Cornel: Comparability package may be reduced, but stull necessary

95 Q to Panel: What is considered as biological noise for sequence variants? There was an example of 3% vs 0.1% in J. Rouse s presentation. J. Rouse: Definition of level is not quite there. Seems there is a break point at 0.1%, mostly related to instrument capability at first, but now confirmed since instruments have become more sensitive. At least for the example of mis-incorporations. Amgen published a paper in 2013 defining biological noise. Part of the rationale of keeping the levels low is to accommodate contributions of scaling-up the processes. J. Liu: If you pick an arbitrary number based on the type of characterization assay, the assay should be developed to suitable to demonstrate precision and robustness to show process control at that level F. Cornel: Also depends on the actual SV and it s effect on safety and efficacy. A strong rationale for selected limits should be submitted by the Sponsor.

96 Q to J. Rouse: Have you found correlations between different methodologies for SV detection (NGS vs. Sanger)? What is the preferred methodology? J. Rouse: Both methodologies have similar responses and outcomes; similar LOD. NGS is currently preferred to extrapolate MS data further in the process. NGS is quick and reliable. At lo levels it is difficult to determine if SV or mis-incorporation D. Gaser: Small changes at genomic level may translate to higher levels in proteomics.

97 Q to Panel: Is yield correlated to finding more problems on finding variants in CHO cell lines? Should we think about using another cell line as a standard with higher yield, less problems? S. Gross: New approaches include changing expression systems, achieving high yields and less toxicity to cells.

98 Q to Regulators: Have you seen real issues with SV reported by Sponsors? F. Cornel: Yes. It has to do with better technologies to detect. Sponsors deal with them in different ways. Early recognition and proposing a strategy to manage, has a higher chance of acceptance. If found at later stages in development, less likely to accept presence of SV. I. Reischl: Usually not visible to regulators, unless seen at inspection. The most sensitive flag comes from analytical outputs. S. Gross: We have not seen it yet. FC Comment: This discussion was difficult because I think each of us that responded to the question had interpreted the question differently. I think eventually the question was re-worded as Have safety signals been observed that have been linked to SV?. I think all the regulators agreed that the answer to that would be not that we know of but pharmacovigilence is not likely sensitive enough to pick up a signal from something like a SV. Definitely, we ve seen some quality issues around SV but so far, safety issues are theoretical. In terms of my comment above, it s not that we would be less likely to accept the presence of an SV later in development, it s more that the sponsor is less likely to have the data characterzing the SV which would make the submission acceptable. So if the SV is discovered late in development, the sponsor is often left scrambling to do studies to fully understand the extent of the SV and its impact on product quality.

99 Q to Panel: Have expectations for characterization of cell lines grown by expansion vs. continuous culture processes different? S. Gross: Continuous process cell lines are not very common. Interesting question J. Liu: No new applications received using continued culture process. CHO cell lines have limited lifetime. Would need validation of genetic stability on continuous cell line. More characterization would likely be requested. R. Deshpande: Amgen is starting to use continued production cells. Information in submission package includes more characterization DURING production. The end point of the continued process cells depends on the stability of product quality attributes. Audience process: The concepts between both type of cell lines are about the same. Variability detected in analytics is normalized by pooling cell media from continued production cell lines.

100 Q to Panel: Regulators were heard mentioning problems when scaling up. Do you have examples of such issues? Are we penalizing the Sponsors for scale-up? J. Liu: Not problems, but risks have been identified by noting differences between scales that could impact product quality attributes. Not a penalty. Simply asking to provide a suitable cell age validation package and product quality. S. Gross: Many studies performed at small-scale used to compare commercial scale. As long as small-scale process is qualified, the risks are perceived as less. FDA in Audience: We have seen changes with cell age, or variations in media metabolites at scale-up. Not clear whether changes are due to process or clonality issues.

101 Q to Panel: Thoughts on points to consider with cell line changes. J. Liu: Most times changes are translated into glycan levels. The comparability package will be essential to justify those changes. S. Gross: A good example are biosimilars. Some are produced in different cell lines than the original. For example, if Fc function in a mab is not relevant, parameters can be changed without much impact to package.

102 Q to S. Gross: Bullet point on showing comparability in-process and at EOP. It becomes problematic when the Sponsors are using in-process controls to establish control without having to demonstrate at EOP S. Gross: I agree in essence form the standpoint of QbD and the move to establish more process controls, such as P.A.T. Agency has accepted reduction of controls.

103 Q to Regulators: What considerations for cell line change to reduce HCPs? S. Gross: Sponsors are encouraged to qualify/validate an HCP assay with sufficient sensitivity and specificity.