Sundar Ramanan, PhD, USA

GaBI Educational Workshops 2 3 March 2016, Hacettepe University, Ankara, Turkey First Turkish Interactive Workshop on Regulation and Approval of SIMILAR BIOTHERAPEUTIC PRODUCTS/BIOSIMILARS Sundar Ramanan, PhD, USA Director, Global R & D and Regulatory Policy, Amgen Inc, USA

GaBI Educational Workshops 2 3 March 2016, Hacettepe University, Ankara, Turkey First Turkish Interactive Workshop on Regulation and Approval of SIMILAR BIOTHERAPEUTIC PRODUCTS/BIOSIMILARS Biologicals and biosimilars the complexity of structure and function Sundar Ramanan, PhD 3 March 2016

Biologicals and biosimilars the complexity of structure and function Sundar Ramanan, PhD Director, Policy R&D and Regulatory Affairs Amgen Inc GaBI Turkey SBP Workshop, March 2 3, 2016, Ankara, Turkey

Discussion Topics Overview of biosimilar development Elements and limitations of analytical studies Role of structure-function studies

Discussion Topics Overview of biosimilar development Elements and limitations of analytical studies Role of structure-function studies

Biosimilar development proceeds through a stepwise similarity exercise Approval is Based on the Totality of Evidence 1 Biosimilar? Originator BLA Biosimilar 3. Clinical 2. Non clinical in vitro and in vivo testing Biosimilar? Biosimilar? 3. Clinical 2. Non- Clinical Cross reference Cross reference 1. Foundation- Structural and functional comparisons 1. Quality Cross reference Prior findings of safety and efficacy Integrated Biosimilarity Exercise Quality, Safety and Efficacy Food and Drug Administration. http://www.fda.gov/downloads/drugs/guidancecomplianceregulatoryinformation/guidances/ucm291128.pdf. Accessed 24 January 2013.

Process design and analytical studies form the foundation of biosimilar development 3. Clinical 2. Non clinical in vitro and in vivo testing Before non-clinical and clinical testing can proceed: Define the target quality profile Design the process Compare structural and functional attributes Biosimilar? 1. Foundation- Structural and functional comparisons Use state-of-the-art analytical characterization and functional assays to assess any structural difference Understand the importance and limitation of functional assays

Biologics may have 4 orders of structure plus modifications that affect in vivo characteristics Sugar side chains (can affect safety and efficacy) β Pleated sheet Carbohydrates & other posttranslational modifications (a) Primary structure α Helix (b)secondary structure (c)tertiary structure Chemical modifications & PEGylation Chemical modifications (can affect purity, safety, or potency) (d) Quaternary structure Synthesis and Folding [Image Source: Tim Osslund; Amgen Usage Rights: Unlimited world-wide usage rights for an unlimited time; http://kvhs.nbed.nb.ca/gallant/biology/protein_structure.html. Data source: USP-NF 1045. Biotechnology-derived articles: 3-20]

Biological products have very complex structures Monoclonal antibody Peptide modifications Deamidation Succinimide Oxidation N & C-terminal variants Amino acid substitution Disulfide isoforms Glycan modifications G0, G1, G2 Core fucosylation High mannose etc Folding/Size Truncation Half molecules Dimer Multimers Aggregates Particles Figure adapted from D. Kelner (Amgen), Comparability and Biosimilarity: Two Sides of the Same (or a Different) Coin? presented at IBC Analytical Technologies,San Diego, CA (March 2012)

Typical analytical similarity assessment evaluates 90 to 100 unique attributes Results from a wide breadth of assay combinations compares the analytical footprint of the biosimilar to the reference product. Is it possible to match all attributes? Figure adapted from J. Liu et al. (Amgen), Analytical Similarity Assessment of Biosimilars presented at the Spring ACS Meeting, Dallas, TX (March 2014)

Biosimilar development can use a Qualityby-Design (QbD) approach QbD for biosimilars Assess criticality based on literature & experience Prior Knowledge Clinical Studies Animal Studies Safety and Efficacy Data In-Vitro Studies High Criticality Attributes Focus on the most important attributes Characterize reference product quality attributes Product Quality Attributes Criticality Assessment Design biosimilar to minimize differences for high criticality attributes Assess potential clinical relevance of remaining differences Low Criticality Attributes Product Understanding Lower criticality attributes are not forgotten Figure adapted from G. Grampp (Amgen), Challenges of Structure-Function Studies for Assessing Similarity presented at the Spring ACS Meeting, New Orleans, LA (April 2013)

Discussion Topics Overview of biosimilar development Elements and limitations of analytical studies Role of structure-function studies

Analytical studies should assess several aspects of structure Primary structure (sequence and linkages) Higher order structures (folding, aggregates) Covalent modifications (glycosylation and chemical modifications) Impurities (product and process) Stability profile

AU Biosimilar product should have identical amino acid sequence to the innovator 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 110.00 120.00 130.00 140.00 150.00 160.00 170.00 180.00 Minutes Peptide mapping 100% sequence confirmation Search for any low level amino acid substitution (sequence variant) due to translational errors, misincorporation, or mutation Post-translational modifications, such as glycosylation, acetylation, sulfation, phosphorylation, glycation, etc Amgen unpublished data Figure adapted from J. Liu (Amgen), Analytical Testing and Characterization of Biosimilars presented at the Health Canada SEB/Biosimilar Scientific Forum, Ottawa, Canada (November 2013)

Primary structure and covalent modifications can be assessed to high fidelity Mass spectroscopy combined with separation based methods can address many uncertainties Amino acid sequences confirmed to ~100% coverage Covalent modifications, sequence variants and glycan structures detected to <1% resolution Example: LC ESI MS/MS Figure adapted from G. Grampp (Amgen), Analytical Similarity Assessments presented at the DIA/FDA Biosimilars Conference, Washington DC (September 2012) Image obtained from University of Kentucky Mass Spectrometry Facility http://www.research.uky.edu/ukmsf/example2a.html

Advanced mass spectroscopy methods still leave some uncertainties Examples of some remaining challenges Accurate quantitation of minor species Sample ( 14 N) Mix Std+Sample Proteolysis m/z Std ( 15 N) Stable Isotope Labeled Internal Standard (SILIS) Figures courtesy of Jiang et al, PEGS 2011 Correctly folded Disulfide misfolds 1 and 2 Identifying and quantifying disulfide bonding patterns Accounting for combinatorial effects Figure adapted from G. Grampp (Amgen), Analytical Similarity Assessments presented at the DIA/FDA Biosimilars Conference, Washington DC (September 2012)

Higher order structure and size variants are characterized by orthogonal methods 4 2 Circular Dichroism 0 CD Ellipticity -2-4 -6-8 -10 US Lot 081492E US Lot 092852E US Lot 092872E 0.00012 ABP Lot 0010095541 ABP Lot 0010112898-12 -14 Cp (cal/ C) 260 280 300 320 Wavelength (nm) 0.00008 0.00004 0.00000 EU Lot 9208 EU Lot 0212 EU Lot 0213 ABP Lot 001 ABP Lot 001 ABP Lot 001 US Lot 0316 US Lot 0422 US Lot 0506 40 60 80 100 Temperature ( C) Calorimetry 10 Absorbance 0.8 0.4 0.0-0.4-0.8-1.2-1.6-2.0 10-4 1700 1680 1660 1640 Wavenumbers (cm-1) FTIR 1620 US Lot 031622E US Lot 042212E US Lot 050662E ABP Lot 0010085288 ABP Lot 0010085295 ABP Lot 0010085297 Figure adapted from J. Liu (Amgen), Analytical Testing and Characterization of Biosimilars presented at the Health Canada SEB/Biosimilar Scientific Forum, Ottawa, Canada (November 2013) 1600 8 Analytical 7 Ultracentrifugation c(s) 9 6 5 4 3 2 1 0 2 4 6 8 10 12 14 16 18 20 Sedimentation coefficient (Svedberg) EU lot H0134B01 EU lot H0021B06 EU lot H0138B01 ABP lot 0010095534 ABP lot 0010112870 ABP lot 0010133673 US lot 970032 US lot 450657 US lot 970031 (e) Monomer, dimer, oligomers Amgen unpublished data

A common limitation of spectroscopic methods is sensitivity to mixtures Eg, unfolded protein spiked into product Limit of detection is 8% by near UV circular dichroism How sensitive to partially unfolded species? Weighted Spectral Difference (WSD) Near UV CD 8.3% unfolded Spectral changes detectable at WSD of 4.5 0 20 40 60 80 100 Percent Unfolded Amgen unpublished data Figure adapted from G. Grampp (Amgen), Analytical Similarity Assessments presented at the DIA/FDA Biosimilars Conference, Washington DC (September 2012)

Particulate characterization technology is improving Focus on characterizing particles (0.1 m to 10 m) Size, composition, quantity, structure Relevance to immunogenicity Aggregation (<0.1 m) c(s) c(s) Monomer Monomer Aggregate Aggregate AUC Improving sensitivity, accuracy, and specificity Protein vs. container Emerging nanotechnology-based approaches for < 1 m particles Quantitative and qualitative comparisons remain difficult Figure adapted from G. Grampp (Amgen), Analytical Similarity Assessments presented at the DIA/FDA Biosimilars Conference, Washington DC (September 2012) MFI Particulation (>1 m) Particle Concentration (#/ml) 5 10 15 5 10 sedimentation coefficient (S) 15 sedimentation coefficient (S) Amgen unpublished data Light obscuration 10000 1000 100 10 LOQ 1 0 5 10 15 20 Diameter (µm) Amgen unpublished data

Glycosylation is a critical quality attribute that can impact biological functions Glycan mapping by HILIC and Mass Spectrometry Over 25 mab glycans identified Correlate glycan attributes with biological function Glycan Type Impact to function No glycan No ADCC Bisecting GN Increase ADCC 40 High mannose Clearance and effector function 30 20 Hydophobic Interaction LC (HILIC) Terminal Gal NANA Increase CDC Anti-inflammatory 10 Afucosylated Increase ADCC 0 5 7.5 10 12.5 15 17.5 20 22.5 25 min Amgen unpublished data Figure adapted from J. Liu (Amgen), Analytical Testing and Characterization of Biosimilars presented at the Health Canada SEB/Biosimilar Scientific Forum, Ottawa, Canada (November 2013)

0.22 0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 UV - 280nm ABP 494 14011712-2 A52Su 1 UV - 280nm Erbitux EU 133839 A52Su UV - 280nm Erbitux US 10C00383 A52Su 12 13 14 15 16 17 18 19 20 21 22 23 Minutes 0.22 0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 Product isoforms need to be fully characterized using separation methods Size variants Truncation Dimer Multimers Charge and hydrophobic variants N-terminal modification C-terminal modification Deamidation Oxidation Size Exclusion HPLC Isoelectric Focusing Ion Exchange HPLC mau K K K AU AU 5 10 15 20 25 30 min Amgen unpublished data Amgen unpublished data Amgen unpublished data Figure adapted from J. Liu (Amgen), Analytical Testing and Characterization of Biosimilars presented at the Health Canada SEB/Biosimilar Scientific Forum, Ottawa, Canada (November 2013)

Separation methods also used to examine the integrity of covalent structure 6 4 2 0-2 mau Internal Std 5 4 3 2 1 Non-reducing SDS Partial molecules Half molecules Fragments Non-disulfide linked aggregates Amgen unpublished data 5 10 15 20 25 min min mau Heavy Chain 40 30 20 Internal Std LC HC Reducing SDS Truncation Clipped species Light Chain 10 0 5 10 15 20 25 min Figure adapted from J. Liu (Amgen), Analytical Testing and Characterization of Biosimilars presented at the Health Canada SEB/Biosimilar Scientific Forum, Ottawa, Canada (November 2013)

Product-related and process-related impurities must be well characterized High resolution and orthogonal methods are required to characterize product-related species. Size variants: Truncation Dimer Multimers Clipped species Non-glycosylated HC 0.11 Partial molecules Half molecules Fragments Non-disulfide linked aggregates Heavy Chain 0.40 Charge and Hydrophobic Variants: N-terminal modification C-terminal modification Deamidation Oxidation Main AU 0.10 0.09 0.08 0.07 0.06 0.05 LMW Light Chain MMW NGHC HMW Amgen unpublished data Signal (280 nm) 0.35 0.30 0.25 0.20 Acidic Basic 0.04 0.15 0.03 0.10 0.02 0.01 0.05 0.00 0.00 12 13 14 15 16 17 18 19 20 21 22 23 24 Minutes 15.0 20.0 25.0 30.0 35.0 40.0 Time (min) Process-related impurities (HCP, DNA, leachables, etc) need to be characterized to ensure product quality. Particles and aggregates of various sizes need to be evaluated and characterized. Figure adapted from J. Liu et al. (Amgen), Analytical Similarity Assessment of Biosimilars presented at the Spring ACS Meeting, Dallas, TX (March 2014)

Proteins undergo complex degradation and are sensitive to storage and handling Biosimilar stability is impacted by its manufacturing process and formulation Chemical stress Physical stress Thermal stress Protein stability Enzymatic stress Degradation contributes to eventual loss of biological activity and/or potential immunogenicity Figure adapted from J. Liu (Amgen), Analytical Testing and Characterization of Biosimilars presented at the Health Canada SEB/Biosimilar Scientific Forum, Ottawa, Canada (November 2013)

Forced degradation studies should demonstrate similar stability profiles Multiple accelerated thermal stress conditions (25, 40, 50 C) provide a quantitative, reproducible, and sensitive comparison of degradation profiles and rates Example: Size Exclusion Chromatography profiles 50 C, T=15 days Rate comparisons: 0-15 days Amgen unpublished data Figure adapted from J. Liu et al. (Amgen), Analytical Similarity Assessment of Biosimilars presented at the Spring ACS Meeting, Dallas, TX (March 2014)

Discussion Topics Overview of biosimilar development Elements and limitations of analytical studiess Role of structure-function studies

Structural comparisons leave residual uncertainties Sources of uncertainty Potential consequences Assay limitations (limit of detection, specificity, etc.) Lot to lot variability and population statistics Observed differences in critical attributes Observed differences in less critical attributes Unobserved differences could potentially impact efficacy or safety Equivalence of means does not prove that individual lots are biologically equivalent Could impact safety or efficacy if differences are large enough Are assumptions about criticality correct? Could combinations of attributes become significant? Functional studies are the first step in addressing these residual uncertainties

Why functional characterization? Part 1: Required by regulators Functional characterization required To confirm quality and potency of the product To address limitations of structural assays To confirm similar mechanism(s) of action presence of expected function, absence of new function specificity of target binding Relevant passage from FDA guidance Depending on the structural complexity of the protein and available analytical technology, the physicochemical analysis may be unable to confirm the integrity of the higher order structures. Instead, the integrity of such structures can be inferred from the product s biological activity. (Emphasis added) FDA Draft Guidance, Quality Considerations in Demonstrating Biosimilarity to a Reference Product, February 2012 Figure adapted from G. Grampp (Amgen), Challenges of Structure-Function Studies for Assessing Similarity presented at the Spring ACS Meeting, New Orleans, LA (April 2013)

Matching all biological and functional properties is essential IgG1 Amgen unpublished data Light chain Heavy chain Target binding to Complementarity- Determining Regions (CDRs) N-glycan Inter-chain disulfide Intra-chain disulfide Binding Region for Fc R (ADCC, CDC) Binding Region for FcRn (PK ½ life) 2e6 1.5e6 1e6 500000 0 0.1 1 10 100 1000 10000 Concentration (nm) 4-P Fit: y = (A - D)/( 1 + (x/c)^b ) + D: A B C D R^2 EU Lot H0134B01 - (H0134B01: Concentr...1.65e+06 1.31 49.4-4.43e+04 0.997 EU Lot H0021B06 - (H0021B06: Concentr...1.64e+06 1.38 73.9-4.51e+04 0.998 EU Lot H0138B01 - (H0138B01: Concentr...1.63e+06 1.38 65-4.21e+04 0.997 ABP Lot 0010095534 - (GMP1 0010095534:... 1.68e+06 1.29 62.9-5.2e+04 0.998 ABP Lot 0010112870 - (GMP2 0010112870:... 1.62e+06 1.43 71-4.68e+04 0.992 ABP Lot 0010133673 - (GMP3 0010133673:... 1.64e+06 1.53 64.9-3.38e+04 0.997 US Lot 970032 - (970032: Concentratio...1.63e+06 1.41 74.4-3.61e+04 0.998 US Lot 450657 - (450657: Concentratio...1.64e+06 1.31 79.2-5.64e+04 0.998 US Lot 970031 - (970031: Concentratio...1.62e+06 1.65 72.5-1.76e+04 0.998 Weighting: Fixed Biological functions are dependent on the target antigen and the class of antibody Figure adapted from J. Liu et al. (Amgen), Analytical Similarity Assessment of Biosimilars presented at the Spring ACS Meeting, Dallas, TX (March 2014)

Why functional characterization? Part 2: May be essential to justify differences QbD for biosimilars Assess criticality based on literature & experience (where available) Minimize differences for high criticality attributes Prior Knowledge Product Quality Attributes Clinical Studies Animal Studies Safety and Efficacy Data Criticality Assessment In-Vitro Studies High Criticality Attributes Focus on the most important attributes Perform structure-function studies to assess remaining differences Lower criticality attributes are not forgotten Relate findings to potential clinical impact Low Criticality Attributes Product Understanding Figure adapted from G. Grampp (Amgen), Challenges of Structure-Function Studies for Assessing Similarity presented at the Spring ACS Meeting, New Orleans, LA (April 2013)

Prepared samples can increase sensitivity of structure-function studies Cation exchange profile of a mab Potency Observed range Slope estimate Enriched sample % variant Charge Variants % Relative Potency Notional data for illustration purposes only Acidic variants 82 Figure adapted from G. Grampp (Amgen), Challenges of Structure-Function Studies for Assessing Similarity presented at the Spring ACS Meeting, New Orleans, LA (April 2013) Main peak 101 Basic variants 84 Amgen unpublished data Improved estimate of slope informs potential criticality and permitted magnitude of differences

Studies must provide relevant conclusions 1) Evaluate in vitro functional data a) Test functional equivalence of actual batches Is a 25% difference really acceptable? b) Relate attribute difference to parameters from structure function studies Measured difference in means Estimated quantitative effect Relate to clinically meaningful differences Figure adapted from G. Grampp (Amgen), Challenges of Structure-Function Studies for Assessing Similarity presented at the Spring ACS Meeting, New Orleans, LA (April 2013) Potency Attribute Level Reference Lots Attribute Level Biosimilar Lots Reference Lots Biosimilar Lots Test for equivalence of means 80% < < 125% Notional data for illustration purposes only x coefficient = estimated effect Notional data for illustration purposes only

Studies must provide relevant conclusions 2) Evaluate PK and drug metabolism where feasible In vivo vs. in vitro conversion of Glu to pyroglu Serum incubation in vitro: is a variant formed under physiologic conditions? pe by UV 14 12 10 8 6 4 2 0 0 10 20 30 Time (day) Adapted from Liu et al. 2011, J.Biol. Chem. 286, 11211-11217 6% G0 M5 increases initially G1 5% M5 Product recovery from PK samples Figure adapted from G. Grampp (Amgen), Challenges of Structure-Function Studies for Assessing Similarity presented at the Spring ACS Meeting, New Orleans, LA (April 2013) Percent of total glycopeptide Percent of total glycopeptide 6% 4% 5% 3% 4% 2% 3% 1% 2% 0% G0 G1 M5 0 10 20 30 followed by decrease 1% Time post injection (Days)

Small effects can combine in unexpected ways Process change case study Change resulted in shifts in 2 attributes (see figure) In vitro potency of prepared fractions Bioassays predicted equivalent potency Equivalent PK shown in human clinical study Potency difference detected in clinical PD study Post hoc studies with prepared fractions identified additive effects on potency LE/N 1.20 1.00 Attribute 2 0.80 0.60 0.40 0.20 0.00 Figure adapted from G. Grampp (Amgen), Challenges of Structure-Function Studies for Assessing Similarity presented at the Spring ACS Meeting, New Orleans, LA (April 2013) 101% 98% 101% Attribute 1 106% 106% 90% 106% 106% 113% 99% 3.20 3.30 3.40 3.50 3.60 3.70 3.80 Amgen unpublished data SA/N Pre-change Post-change Prepared fraction

Additional challenges in structurefunction studies Predicting human PK/PD Animal studies may not account for species specific clearance mechanisms Insufficient power due to small number of animals Predicting human immune response In silico, in vitro, and in vivo methods are insufficient to rule-out clinically relevant differences

Summary and Conclusions Analytical advances permit high resolution similarity assessments for many attributes Higher order structure and particle assessments still subject to uncertainty Orthogonal approaches partially compensate for lower sensitivity Assessing impact of differences remains challenging Not all clinically relevant effects can be evaluated preclinically (e.g., PK and immunogenicity) Small effects and combinations difficult to assess

Thank you! Questions?