Validation of LC-MS/MS Methods for the Determination of Large Molecules: Promises and Challenges

Transcription

1 Validation of LC-MS/MS Methods for the Determination of Large Molecules: Promises and Challenges Mark Rose, Linh Tran, Amir Sharifi, and Hongyan Li Amgen Inc., Thousand Oaks, CA

2 Presentation Outline 1. Why do we need validated LC-MS/MS methods for LM? 2. Key components of a robust validation a. Extraction b. LC-MS/MS 3. Validated methods in the literature 4. Amgen case study 5. Other considerations a. Structure/Modality b. Project stage 6. Bioanalytical strategy - decision tree 7. Regulatory concerns 2

3 Why do we need LC-MS/MS methods for large molecules? 1. When reagents are unavailable 2. When reagents are non-specific 3. Cases where complimentary specificity is required. 4. Non-traditional molecules a. Antibody drug conjugates b. Bi-specific molecules c. Endogenous biomarkers 5. When ADA's interfere with the LBA 3

4 Biotherapeutics are becoming more complicated. Bioactive Peptide 20kDa PEG Peptide Fc-bivalent peptide Fc-monovalent peptide Multiple constructs with multiple sites for catabolism or degradation All exhibit different in vivo PK properties 4 Fc Loop-peptide Fc-bivalent peptide/ Anti KLH Ab

5 Construct Development Requires a Comprehensive Understanding of Catabolism Serum conc (ng/ml) Total antibody concentration Intact conjugate concentration Time (hr) LBA LC-MS/MS Measuring just the antibody with an LBA would give a different conclusion of stability.

6 A Stable Peptide Warhead Increases Circulation Half-life of the Intact Molecule Serum conc (ng/ml) Total antibody concentration Intact conjugate concentration Time (hr) LBA Selective assays are important for optimizing complex conjugates LC-MS/MS

7 Clean-up Strategies and Achieved LC-MS/MS Sensitivity 46 Publications IC: Immunocapture, PP: Protein Precipitation, SPE: Solid Phase Extraction, LC: Liquid Chrom., SEC: Size Exclusion Chrom. Bioanalytical LC MS/MS of protein-based biopharmaceuticals Irene van den Broeka, et.al., J. Chrom. B, (2013) 929,

8 Detecting Hepcidin (DTHFP ICIFC CGCCH RSKCG MCCKT) Using LC-MS/MS (m/e ) b Hepcidn[4+] [4+] b y23[4+] y24[4+] b y21[3+] y22[3+] y23[3+] y19[2+] y21[2+] [3+] b3 b4 b5 b6 b23 b24 Hepcidin DTHFP ICIFC CGCCH RSKCG MCCKT HPLC Electrospray m/e N 2 -CAD Quad 1 Quad 2 Quad 3 D T H F P I C I F C C G C C H R S K C G M C C K T Highly Specific Detection 8 So what s the 8 problem? y23 y22 y21 y20 y19 m/e Channel Electron Multiplier

9 Endogenous Proteins Identified in Plasma 20,433 distinct peptides from1929 identified protein sequences 9 Farrah et. al., A High-Confidence Human Plasma Proteome Reference Set with Estimated Concentrations in PeptideAtlas,, Mol. Cell. Prot. (2011) 10 (9) 1-14.!The same twenty amino acids rearranged > 20,000 ways! Conclusion: Having an extraction method specific to your analyte is important.

10 Examples of Qualified or Validated Methods Published Since 2012 Year Author Analyte Internal Standard 2012 Hongyan Li (Amgen) 2012 Mireia Fernandez Ocana (Pfizer) 2013 Hao Jiang (BMS) 2013 Daniel Wilffert (U. of Groningen) 2013 Guowen Liu (BMS) 2014 Ichio Onami (Chugai Pharm Company) 2014 Hao Jiang (BMS) mab Free and bound mab Coadministered mabs SIL Peptide SIL Flanking Peptide SIL mab SIL Flanking Peptide SIL Flanking Peptide 10 LLOQ Matrix Extraction Technique 100 ng/ml Rat and Monkey Serum Magnetic Beads 7.03 ng/ml Human Serum Magnetic Beads 5 ug/ml 25 ug/ml Monkey Serum rhtrail SIL rhtrail 20 ng/ml Human and Mouse Serum Pellet Digestion (no extraction) IMAC Chromatography mabx SIL Peptide 1 ug/ml Mouse Serum Pellet Digestion (no extraction) RANKL Lysozyme 3.13 ng/ml Mouse Serum Magnetic Beads ADA mab hu IgG SIL Flanking Peptide 1 ug/ml 1 ug/ml 1 ug/ml Human Serum Not many validated LM LC-MS/MS methods published during the last two years. BEAD Extraction and Acid Dissociation

11 General Observations for Recently Validated Methods Precision and accuracy are dependent on choice of internal standard and extraction/enrichment method. Non-specific extraction method = high LLOQ Very specific extraction method = low LLOQ Appropriate internal standard = good precision and accuracy Method formats affect LLOQ Pellet digestion methods can be validated when a high LLOQ is sufficient. Affinity extraction is required for methods with LLOQ < 0.5 ug/ml 11

12 Limited Sample Preparation (Pellet Digestion) with LC-MS/MS Analysis Biological sample Protein Precipitation Add SIL-Peptide internal standard Denature, Cys reduction Tryptic digestion UPLC-MS/MS 12

13 Validated Methods Using Pellet Digestion and LC-MS/MS Pellet digestion: a simple and efficient sample preparation technique for LC MS/MS quantification of large therapeutic proteins in plasma Protein 1 (150 kda) LLOQ : 2.6 µg/ml Intra-assay Precision: <16.2% Accuracy: < 10% Zheng Ouyang, et.al., Bioanalysis (2012) 4(1), Fully Validated LC-MS/MS Assay for the Simultaneous Quantitation of Coadministered Therapeutic Antibodies in Cynomolgus Monkey Serum mab-a LLOQ : 5 µg/ml mab-b LLOQ: 25 µg/ml Intra-assay Precision: <10% Accuracy: < 5.4% Hao Jiang, et.al., Anal. Chem. (2013) 85, Acceptable precision and accuracy possible without immunoaffinity reagents 13 - High LLOQ

14 More Sensitive LC-MS/MS Assays for mabs Incorporate Immunoaffinity Extraction Biological sample Immunoaffinity protein capture Streptavidin Magnetic Beads Streptavidin Tips Add SIL-Peptide internal standard Denature, Cys reduction Tryptic digestion UPLC-MS/MS Format Chosen at Amgen for Validation 14

15 LC-MS/MS Method Validation for the Determination of Hu-IgG2 mab in Preclinical Serum Immunoaffinity Extraction Enzyme digestion LC-MS/MS 60 min 45 min + 10 min 5 min/injection Compound Q1 Q3 Dwell time DP CE EP CXP (ms) (volts) (volts) (volts) (volts) VVSVLTVVHQDWLNGK VVSVL*TVVHQDWL*NGK Standard calibration curve range: 50 10,000 ng/ ml Column: Phenomenex Aeris Peptide XB C18, 3.6 mm, 2.1 x 100 mm Mobile phase: A: 5% ACN/water/0.1% formic acid B: 95% ACN/water/0.1% formic acid Gradient elution: 5 min Injection volume: 10 ul Slide: Hongyan Li 15 I n t e n s i t y, c p I n t e n s i t y, c p e4 2.5e4 2.0e4 1.5e4 1.0e LLOQ: 50 ng/ml Time, min SIL-VVSV IS Time, min

16 LC-MS/MS Method Validation for the Determination of Hu-IgG2 mab in Preclinical Serum Hu-IgG2 Conc. (ng/ml) Calibration Standards QC Samples 50 X X and LBA 16 Assays QC Sample Description LLOQ QC (validation only) 100 X 150 X Low QC 250 X 500 X 750 X Mid QC 1000 X 2500 X 5000 X 7500 X High QC 10,000 X X ULOQ QC 100,000 X Dilution QC Validation Format Fundamentally Similar to LC-MS/MS

17 LC-MS/MS Method Validation for the Determination of Hu-IgG2 mab in Preclinical Serum Hu-IgG2 Method validation parameters using LC-MS/MS assay with immunoaffinity extraction Accuracy Calibration curve and regression Precision Dilution linearity and hook effect Sensitivity Selectivity Stability Matrix effect Reproducibility Parameter Cyno Rat Mouse Validated Range Precision % CV and % Bias (intra-assay) Precision % CV and % Bias (inter-assay) 50-10,000 ng/ml 50-10,000 ng/ml 50-10,000 ng/ml <9.8% and -6.4 to 9.3% <7.0% and -1.1 to 8.5% <7.6% and -5.0 to 4.6% <6.5% and -2.0 to 0.0% <11.3 and to 6.0% <10.5% and -5.9 to 0.2% Validation results were similar to standard small molecule LC-MS/MS assays

18 Considerations in Validating a Large Molecule LC-MS/MS Assay 1. Large and small molecule assay components require a hybridization of regulatory guidances. a. Large molecule, e.g., affinity extraction and large molecule standardization b. Small Molecule, e.g., UPLC, MS/MS, solid phase extraction 2. Little regulatory interactions reported thus far concerning validated large molecule LC-MS/MS data gathering required. 3. Caution recommended due to the multistep complicated nature of the assays 4. Some validation experiments seem scientifically unwarranted, but will still be performed until a knowledge base is obtained. 18

19 Hybrid Parameters for Validating a mabs Method Validation Parameter Accuracy & Precision (A&P) Selectivity Matrix Effect Dilution Accur. Benchtop Stability (BTS) Conditions Specific Details n Acceptance Criteria Calibration Standards (CS)/ QC Preparation Analyze with IS (blank) and without IS (double blank). Spike 6 different sources of biomatrix at the LOQ QC conc. Dilution accuracy for each dilution factor used will be assessed using QC samples. Analyze 6 replicates for the dilution scheme. 2 conc. (low and high QC), n 6 for each conc. Store at least 4 hours at RT. A minimum of 5 of 6 results (>80%) must be available at each level. LBA Based CSs and QCs are prepared separately. Freshly prepared CSs used for each analytical batch. Single use QCs prepared, and stored at -60 to -80 C. 6 different sources/subjects/animals. Evaluate LOQ in 6 different sources/ subjects/animals. Dilution factor of 10, 100 and 1000 for dilution QC at100,000 ng/ml. Low and high QC samples are thawed at room temperature and kept at this temperature for at least 4 hours. The stability samples are processed and quantitated using freshly prepared calibration samples. LC-MS/MS Based 19-3 A&P batches from a minimum of 2 separate days. - 8 freshly prepared nonzero CSs in duplicate (n=2) in each analytical batch. - n=6 at each QC level (in each analytical batch) - n=1 for each lot (B) - n=1 for each lot (DB). analyze in triplicate (n=3) for each source n=6 for each dilution factor n=6 for each level (HQC and LQC) and for each condition Common to Both Minimum 75% of CS ± 20.0%* of target conc. (* ± 25.0% at LLOQ). At least one calibrant at the LLOQ and ULOQ must meet the acceptance criteria. Mean intra- and inter-assay bias ± 20.0%*, CV 20.0%*. (*25.0% at LLOQ). For 5 of 6 sources, no response >20.0% of mean response of LLOQ. For IS, no response >5.00% of mean IS response. For 5 of 6 sources, the measured conc. for 2 of 3 (>66%) samples for each source ±20.0% of target spiked conc.. Mean intra-assay bias ±20.0%, CV 20.0%. The mean measured conc. at each level ±20.0% of target (nominal conc.). CV 20.0%.

20 Hybrid Parameters for Validating a mabs Method Validation Parameter Long Term Stability (LTS) Freeze/Thaw Stability (FTS) Carryover Autosampler Reinjection Reproducibility Conditions Specific Details n Acceptance Criteria - 2 conc. (low and high QC). Analyze n 6 for each conc. at each stability time-point. - Samples must be analyzed using freshly prepared QSs; batches contain QCs to accept/reject. - A min of 5 of 6 results (>80%) available for evaluation at each level Storage at -15 to -25 C targeted this must meet acceptance criteria. for approximately 4 weeks - 2 conc. (LQC and HQC), 3 F/T cycles (min 12 hours frozen for each cycle). - Analyze n 6 for each conc. Samples must be analyzed using freshly prepared CSs. A min of 5 of 6 (>80%) must be available for at each level. Two control blanks, each injected after a ULOQ CS in one batch. - Bulk HQC and LQC levels will be prepared, aliquoted and stored at -60 to - 80 C. - Each condition is subjected to only 1 freeze/thaw cycle. - Storage 3 weeks, 1 month, 3 months, and 6 months (or as needed to support study samples). - HQC and LQC from bulk preparation will be used for this experiment. - HQC and LQC aliquots will be transferred to -15 to -25 C for approximately 4 week storage stability. - Bulk HQCs and LQCs will be stored at -60 to -80 C. - All F/T conditions will be assayed on the same plate. If this spec. cannot be met, alternative spec to control carryover may be employed. Store a batch consisting of a minimum of CSs and a total of 18 Allows for reinjection of samples when the QCs (n=6, low, mid, high conc.). original batch fails for instrumental or Inject after preparation, storage at the reasons unrelated to sample preparation. autosampler temperature for at least 24 hours and re-inject. LBA Based LC-MS/MS Based 20 n=6 for each level (HQC and LQC) and for each condition n=6 for each level (HQC and LQC) and for each condition 1 A&P batch Common to Both The mean measured conc. at each level ±20.0% of target (nominal conc.). CV 20.0%. The mean measured conc. at each level ±20.0% of target (nominal conc.). CV 20.0%. No response >20.0% of the mean response of LLOQ CS. Both original and reinjected batches meet routine acceptance criteria

21 Hybrid Parameters for Validating a mabs Method Validation Parameter Interference Parallelism Hook Effect Robustness Conditions Specific Details n Acceptance Criteria Anti Hu-IgG2 PAb If ADA is observed, incurred samples with ADA will be tested. 3 conc. Use of multiple instruments Different affinity capture incubation times LBA Based - Approximate Molar Ratios: 0.01:1, 0.05:1, 0.1:1, 1:1, 1:5, 10:1, 100:1. - QC level of Hu-IgG2 will be used. - Prepare all conditions in 100% cynomolgus monkey serum - Perform at least 4 dilutions, and each dilution is prepared in 100% cynomolgus monkey serum. - Conc. after dilutions must cover span of standard range. - Used pooled incurred samples - A sample will be prepared at 2 mg/ml. - The sample will be serially diluted to conc. of 100,000, and 10,000 ng/ml in 100% cynomolgus monkey serum. - All diluted and the neat samples will be assayed. Two different LC-MS/MS systems. - 3 incubation times: Long (75 min), middle (60 min), and short (45 min). - Run with Dilution Linearity run. LC-MS/MS Based n= 1 per conc., including blanks (positive and negative controls). n=1 for each dilution n=1 for each dilution 1 A&P batch n=6 of each level n=6 for each level (HQC and LQC) and for each condition. Common to Both %Bias will be calculated from nominal spiked Hu-IgG2 (MQC level). - %Diff at each dilution will be calculated from the undiluted pooled incurred samples. - For the sample to show parallelism, the measured conc. for 2 of 3 (>66%) dilution factors is tested. ±20.0% of target conc. - %Bias for sample at 10,000 ng/ ml must meet acceptance criteria from nominal. - All other samples must be AQL for no Hook effect. - Minimum 75% of CS ± 20.0%* of target conc. (* ± 25.0% at LLOQ). - At least one calibrant at the LLOQ and ULOQ must meet the acceptance criteria. The mean measured conc. at each level ±20.0% of target (nominal conc.). CV 20.0%.

22 Choosing Between Ligand Binding Assays and LM LC-MS/MS in an Overall BA Strategy LBA or LM LC-MS/MS Discovery and Pre-GLP 1. Fast moving 2. Multiple compounds /false starts 3. Inability to generate highly specific reagents 4. No need for parallel processing Time-bound Factors 1. Program stage 2. Reagent availability 3. Sample numbers 4. Outsourcing strategy Early Regulated Studies 1. Limited studies on a single molecule 2. Some tolerance for resource spend 3. Parallel processing desired Late Stage Regulated Studies 1. Large studies on a single molecule 2. Less risk involved for investing in specific and rugged methods 3. Many samples requiring a high throughput method. 4. Parallel processing and analysis preferred. 5. Outsource-able methods Time-independent Factors 1. Type of matrix 2. Clearance mechanism 3. Catabolic sites/degradation 4. Molecule complexity (ADA, conjugates) 5. ADA interference 6. Pegylation 7. Required sensitivity

23 Bioanalytical Platforms Evolve with Project Stage Discovery Screen Hit-to- Lead Lead opt Preclinical Phase 1 Phase 2 Phase 3 Filing Launch Large Molecule LC-MS/MS Ligand Binding Assays Monoclonal Antibodies: Some confirmatory use of LC-MS/MS Antibody Drug Conjugates: No transition due to complexity of the molecules Fusion Proteins: Key Factors to be Considered: 1. Attributes of the modality 2. Cost 3. Speed 4. Risk tolerance 5. Outsourcing strategy Time dependent transition from 100% LC-MS/MS to 100% LBA 23

24 The Argument Against Preclinical Use of LC-MS/MS for mabs Generic Human IgG PK Assays Used for Rat Serum Beth Leary, et.al., J. Immunol. Meth. (2013) 4(1), Is there a justification to use LM LC-MS/MS for mabs development? Perhaps not a compelling one. 24

25 LBA vs. LM LC-MS/MS Decision Tree LM BA Required MW< 10,000 Da MW> 10,000 Da Conventional LC-MS/MS 1. Highly specific 2. Well established 3. Sensitivity may be limiting (LLOQ > 10 ng/ml), so Inappropriate for potent compounds, low expressing biomarkers LBA Reagents Available Yes Reagent Pairs Yes No LBA 1. ELISA and GYROS 2. Sensitive 3. Lack of specificity may be limiting 4. High throughput 5. Cannot do multiplexing 6. Better for free/bound measurements. 25 No 25 Other Affinity Capture Yes No Affinity LM LC-MS/MS 1. Some form of affinity extraction required (mab, IMAC, Prot A,G). 2. Expensive instrumentation 3. Specific 4. Multiplexing possible 5. Resolution may be limiting 6. Limited validation experience Direct LM LC-MS/MS 1. Simple procedure 2. Sensitivity may be limiting 3. Expensive instrumentation 4. Limited validation experience

26 LM LC-MS/MS Features that Would Cause a Regulatory Concern Feature Risk Precedent Standardization Multistep Method Stability Specificity Paradigm Change for Assay Format Low/Medium Medium There does not seem to be anything special about LM LC-MS/MS that would cause a regulatory concern. 26 Low Medium Medium Ligand Binding Assays SM Derivatization Methods Ligand Binding Assays Ligand Binding Assays Ligand Binding Assays

27 Acknowledgements Tim Heath Peng Luan All work was funded by Amgen Inc. 27