Antibody-drug Conjugates: Characterization and Control Strategies of Lysine-linked Products Fred Jacobson Protein Analytical Chemistry Genentech, Inc. CASSS CMC Strategy Forum Japan 2013 December 9-10, 2013 2010 Genentech, Inc.
Structure of T-DM1 Slide 2 Trastuzumab-MCC-DM1 = T-DM1
Mechanism of Action T-DM1 Slide 3
US BLA Approved 22 Feb 2013 Slide 4
Global Approvals/Recommendations Slide 5 Switzerland Kuwait Georgia Australia Ecuador Uruguay Canada (11 Sep) Japan (20 Sep) CHMP (20 Sep) N.Z. (26 Sep) Taiwan (7 Nov)
ADC; Marriage of small and large molecules Cell-culture & recovery process Complex biologic intermediate Simple Post Translational Modification Chemistry Complex, well controlled drug substance Slide 6 Small molecule intermediates, starting materials Synthetic or semi-synthetic process
ADC; High level control strategy Slide 7 KADCYLA not a formal QbD submission; Application of QbD elements (i.e., enhanced understanding) CMA s Complex biologic intermediate CPP s CQA s Complex, well controlled drug substance Small molecule intermediates, starting materials Safety Bioactivity Pharmacokinetics Immunogenicity
Critical Quality Attributes of an ADC Slide 8 Antibody/ADC Specific Statutory Drug to Ab Ratio (DAR) Drug Distribution Size variants (HMWS) Unconjugated drug Charge variants Linked drug variants Cross-linked species Safety Endotoxin Sterility & Bioburden Residual solvents Identity Strength Moisture content (lyo DP)
T-DM1 Conjugation process Input Quality CMAs Critical process steps to control CPPs (Linker) (Drug) Purification CQAs DAR/distribution aggregation and cross-linking
Analytical challenges of lysine ADCs Slide 10 Added complexity due to conjugation heterogeneity Charge, distribution, sites of attachment Superimposed on the Mab heterogeneity Methods developed for MAbs don t always work for conjugates, or they provide different information Charge-based assays for lysine-linked ADCs Demonstrating comparability from process changes during development (and now, post approval) 2010 Genentech, Inc.
2 8 0 n m A b s o r b a n c e 2 8 0 n m A b s o r b a n c e SEC of MAb vs ADC Slide 11 Standard SEC H e r c e p t i n H e r c e p t i n - D M 1 Trastuzumab Solvent modulated SEC T-DM1 6 8 1 0 1 2 1 4 T i m e Trastuzumab T-DM1 H e r c e p t i H e r c e p t i n n - D M Similar behavior for cysteine-linked ADCs; despite lack of intact interchain disulfides fragmentation not seen 6 8 1 0 1 2 1 4 T i m e 2010 Genentech, Inc.
Absorbance UV Spectroscopy for average DAR Slide 12 2 GLP Scale Average DAR Range (N) 1.5 1 30C Small 3.44 3.3 3.6 (7) Intermediate 3.55 3.4 3.8 (20) 0.5 DM1 lmax 252nm Large 3.45 3.4 3.5 (8) 0 240 260 280 300 320 340 Wavelength (nm) A 252 /A 280 is used to calculate the DM1 to antibody ratio (moles/mole.) 2010 Genentech, Inc.
Slide 13 Drug distribution of a lysine linked ADC Small Scale MS of deglycosylated T-DM1 Intermediate Scale Cross-Linked Species Large Scale Comparable distribution across manufacturing scales = macroheterogeneity 1-3% MAb with no bound DM1 Unconjugated Mab is a function of DAR Spec on DAR controls unconjugated Mab ~8% cross-linked in all lots by MS 2010 Genentech, Inc.
Evidence of intra-chain cross-linking Slide 14 Reduced CE-SDS as lot release assay for cross-links Examples of possible cross-linked species 2010 Genentech, Inc.
Non-reduced CE-SDS of T-DM1 Slide 15 - SDS-dissociable fragmentation & aggregation - similar changes to what is seen with un-conjugated MAbs - characterization method Fragments HMW species 97% intact monomer 2010 Genentech, Inc.
Charge-based analysis (imaged cief) of a lysine-linked ADC Slide 16 Shows charge variants associated with the MAb peptide structure Charge separation mostly shows drug distribution; not Mab charge Main Peak Acidic Variants pi Marker 0-Drug MAb peak pi Marker pi Marker pi Marker Method can be used as basis of assay for unconjugated MAb 2010 Genentech, Inc.
Heterogeneity of a Lys-linked ADC Slide 17 Trastuzumab has 88 lysines & 4 N-terminii for possible linkage. 76 sites identified by LC/MS
Heterogeneity Randomness!! Slide 18 Peptide map fingerprint of drug-containing peptides shows consistency between processes run at different scales and sites Well controlled and robust chemical process leads to reproducible product Lot-to-lot the same lysines are modified at the same % occupancy
DAR strongly correlates to distribution Process DOE studies
DAR strongly correlates to distribution
DAR strongly correlates to distribution f(k;l) = lk e -l k! Poisson distribution DAR (UV) Drug Distribution
Reproducibility of DAR/Distribution Slide 22 Based on N = 32 large scale batches (now >100 lots) Average DAR = 3.46 ± 0.06 Analyzed by ESI-MS or by icief - % of each drug loaded species is highly reproducible (even for 0-drug species ~7% RSD) - Orthogonal method (icief) gives comparable results Regulatory expectation (some HAs) for a direct measurement & specification for 0-drug species in the DS and DP (Control of surrogate (DAR) accepted by some HA s)
Free drug analysis by RP-HPLC Slide 23 MeOH Spin HPLC ADC Two diastereomers 2010 Genentech, Inc.
Potency assays for ADCs Slide 24 24 For a therapeutic MAb, a single potency assay, typically cell-based (or other MOA based), is sufficient for lot release & stability
Potency assays for ADCs Slide 25 25 For a therapeutic MAb, a single potency assay, typically cell-based (or other MOA based), is sufficient for lot release & stability For ADCs FDA is requesting two potency assays, at least at early stage A cell potency assay demonstrates the MOA; shows target binding, internalization, drug release, and killing An ELISA shows only target binding Part of product characterization and comparability
ELISA and cell-based assays both require receptor binding Slide 26 26 Cell Based; Represents MOA ELISA Based (Binding only) Enz Color development Enz-HER2-ECD T-DM1 T-T-DM1 internalized and degraded No change in receptor binding is seen between T-DM1 and the starting trastuzumab intermediate (ELISA and Biacore)
Impact of conjugation on HER2 binding Slide 27 No change in EC 50 by ELISA M Tejada & Y Wang
RFU No effect of free MAb on T-DM1 potency Slide 28 F u ll C u rv e s o f T ra s tu zu m a b E m ta n s in e w ith /o u t T ra s tu zu m a b Sp ik e -in 14500 Trastuzumab Emtansine 12500 Trastuzumab Emtansine + 20% Trastuzumab 10500 8500 6500 4500 2500 1 10 100 1000 ng/ml 4-P Fit: y = (A - D)/( 1 + (x/c)^b ) + D: A B C D R^2 Trastuzumab Emtansine Only (Standard: Concentration... 1.33e+04 3.93 36.1 3.38e+03 0.991 Trastuzumab Emtansine + 20% Trastuzumab (Sample: Co... 1.34e+04 4.36 34.1 3.55e+03 0.986 M Tejada & Y Wang Curv e Fit Option - Fixed Weight Value ADC concentration (ng/ml)
Potency Assays; continued Slide 29 Both methods show; -Comparable loss in activity during high temp storage -Equivalent results for samples exposed to other chemical stresses Cell-based assay is sensitive to drug load; ELISA is not; Free antibody does not inhibit ADC cell-killing Some HA s (US FDA) requiring both assays for commercial ADC release and stability
Conclusions Slide 30 Kadcyla is a complex molecule characterized by reproducible heterogeneity of drug-linker content and modification site Despite this complexity, analytical methods are capable of demonstrating lot-to-lot consistency, stability, and process comparability ADC quality is established and maintained through a combination of input, process, and product controls
Acknowledgements Slide 31 Protein Analytical Chemistry Yan Chen Charity Bechtel Chien Lee Laura Zheng Mike Lee Lan Dai Biological Technologies Max Tejada Yaning Wang