Report on a collaborative study for proposed candidate 1st international standard for the biological activities of Rituximab

WHO/BS/2017.2309 ENGLISH ONLY

EXPERT COMMITTEE ON BIOLOGICAL STANDARDIZATION Geneva, 17 to 20 October 2017 Report on a Collaborative Study for Proposed Candidate 1st International Standard for the biological activities of Rituximab Sandra Prior1, Bernard Fox, Thomas Dougall, Peter Rigsby and Simon Hufton National Institute for Biological Standards and Control, Blanche Lane, South Mimms, Potters Bar, HERTS EN6 3QG, UK 1 Email address: sandra.prior@nibsc.org NOTE: This document has been prepared for the purpose of inviting comments and suggestions on the proposals contained therein, which will then be considered by the Expert Committee on Biological Standardization (ECBS). Comments MUST be received by 25 September 2017 and should be addressed to the World Health Organization, 1211 Geneva 27, Switzerland, attention: Technologies, Standards and Norms (TSN). Comments may also be submitted electronically to the Responsible Officer: Dr Hye Na Kang at email: kangh@who.int. © World Health Organization 2017

All rights reserved. Publications of the World Health Organization are available on the WHO web site (www.who.int) or can be purchased from WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland (tel.: +41 22 791 3264; fax: +41 22 791 4857; e-mail: bookorders@who.int). Requests for permission to reproduce or translate WHO publications – whether for sale or for noncommercial distribution – should be addressed to WHO Press through the WHO web site: (http://www.who.int/about/licensing/copyright_form/en/index.html). The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of the World Health Organization concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. Dotted lines on maps represent approximate border lines for which there may not yet be full agreement. The mention of specific companies or of certain manufacturers’ products does not imply that they are end orsed or recommended by the World Health Organization in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of proprietary products are distinguished by initial capital letters. All reasonable precautions have been taken by the World Health Organization to verify the information contained in this publication. However, the published material is being distributed without warranty of any kind, either expressed or implied. The responsibility for the interpretation and use of the material lies with the reader. In no event shall the World Health Organization be liable for damages arising from its use. The named authors alone are responsible for the views expressed in this publication.

WHO/BS/2017.2309 Page 2

Summary A candidate rituximab preparation was formulated and lyophilised at NIBSC prior to evaluation in a collaborative study for its suitability to serve as an international standard for the biological activities of rituximab. The candidate was tested at the collaborating laboratories alongside a coded duplicate, a second rituximab lyophilised preparation and the laboratory’s in house reference standard when available. The preparations were tested for their complement dependent cytotoxic activity (CDC) by sixteen laboratories and their antibody dependent cytotoxic activity (ADCC) were tested by eleven laboratories. Different in vitro cell-based bioassays were used in the study as per collaborator qualified methodologies. A limited number of laboratories also performed cell-based antibody binding and apoptosis assays. The results suggest that the candidate preparation coded as NIBSC 14/210 is suitable to serve as an international potency standard for rituximab. In general, the potency estimates were in poor agreement when expressed relative to the in house reference standards with the greatest variability between laboratories observed for the ADCC assay. However, the data show that the use of the candidate international standard helps with harmonising the reporting of rituximab bioactivities by different laboratories using their in house potency assays for each of the activities studied, namely CDC, ADCC and cell binding. In view of the results, it is proposed that the preparation NIBSC 14/210 is established as the 1 st International Standard for the biological activities of rituximab with the following assigned in vitro bioactivities: 1,000 International Units (IU) of CDC activity per ampoule 1,000 IU of ADCC activity per ampoule 1,000 IU of cell-binding activity per ampoule 1,000 IU of apoptotic activity per ampoule This international standard is intended to support bioassay characterisation, calibration and validation by different stakeholders and also to define international units of bioactivity. However, it is important to note that this international standard does not define specific product activity. Further, the characteristics of the material used to make this reference standard, are not intended to serve any regulatory role in defining biosimilarity, and should not be extrapolated, calculated or inferred as serving this purpose in anyway whatsoever. The proposed unitage is neither intended to form basis of revised labelling or dosing requirements as any decisions regarding the use of these units for specific activity specifications rely solely on the competent authorities. Finally, it is highlighted that the role of the international standard is different to that of the reference product that is used to define biosimilarity given its tracked clinical record of safety and efficacy.

Responses from study participants Responses from twelve of the sixteen participants in the study were obtained. Eleven participants agreed that their data had been reported correctly and agreed with the proposed unitage of bioactivities. Any comments were individually discussed with the relevant laboratories. Minor

WHO/BS/2017.2309 Page 3 editing in the text was made and typographical errors were corrected. Questions related to statistical analysis by three laboratories were addressed. One participant had comments regarding the analysis of assay validity, and the intended use and suitability of the IS. These were discussed in detail but the laboratory did not return any further confirmation on their agreement to the proposal (for these comments see Appendix 3).

Introduction Rituximab is a chimeric mouse-human monoclonal antibody (Mab) used in the treatment of CD20-positive B-cell lymphoproliferative malignancies, transplant rejection and autoimmune disorders. Current therapeutic indications include non-Hodgkin’s lymphoma, chronic lymphocytic leukemia, rheumatoid arthritis, granulomatosis with polyangiitis and microscopic polyangiitis (EMA, WC500025821); (Dotan, Aggarwal et al. 2010) and there is growing interest in its use for the treatment of autoimmune diseases (Sanz 2009). Rituximab is administered both as a monotherapy or in combination with chemotherapy regimens (Hiddemann, Kneba et al. 2005);(Forstpointner, Unterhalt et al. 2006); EMA-WC500025817). Its molecular weight is 144,544 Da and is comprised of 1328 aminoacids (DrugBank DB00073). The molecule consists of a human kappa constant region, a human IgG1 Fc portion and a murine variable region, and contains a conserved N-glycosylation site at Asn297 of both heavy chains occupied by biantennary glycan structures (Jefferis 2009). The exact antitumor mechanism of rituximab remains unclear; however, is assumed that it exerts its effects by various mechanisms comprising binding of its Fab domain to CD20+ Blymphocytes for the induction of apoptosis, either directly or by the immune effector functions of its Fc domain. Thus it mediates B-cell lysis through complement-dependent cytotoxicity (CDC) after binding to C1q, or antibody-dependent cellular cytotoxicity (ADCC) upon engagement to Fcγ receptors (FcγR) on effector cells that include natural killers, granulocyte and macrophages (Glennie, French et al. 2007) (Smith 2003). Rituximab is on the WHO model list of essential medicines for a basic health-care system (19th WHO model List of Essential Medicines, 2015). Currently, there are two approved rituximab products, the so-called originator or innovator, commercialised as Mabthera® in EU (June 1998) and Rituxan® in USA (November 1997) (Biogen/Genentech/Roche) and its first approved biosimilar in EU (February 2017) Truxima® (Celltrion). Large sales of the innovator product reaching 8.7 billion US dollars in 2014 (Udpa and Million 2016) and the prospects of the patent expiration of this “blockbuster” in 2013 for Europe and in 2018 for US (Derbyshire, 2015), have driven a rapid growth in biosimilar drug development which is expected to widen market competition and increase global patient accessibility. To date, at least 44 biosimilars to rituximab are under development (Udpa and Million 2016) and some at late-stages or under regulatory review in Europe. Furthermore, a number of non-originator product versions are also approved in some countries under local regulatory pathways that might not be as strict as those required for the approval of biosimilars monoclonal antibodies in the EU or USA which requires a rigorous comparability exercise (GaBI online, 2016).

WHO/BS/2017.2309 Page 4 Monoclonal antibodies derived by recombinant DNA technology are structurally complex molecules sensitive to small changes in the manufacturing process. The WHO has recognised a global need for standardisation of biotechnology products as a control measure to ensure safety, quality and efficacy (WHO Technical Report Series, 56th Report, 941: 12-13, 2007; WHO Technical Report Series, 66th Report, 999:13-15, 2016) (Wadhwa, Kang et al. 2011). The proposed WHO international standards for biological activity of therapeutic monoclonal antibodies are intended for the evaluation of bioassay performance including calibration and validation of potency assays in alignment with these recognized needs (Thorpe and Wadhwa 2011). Determination of the in vitro bioactivity of rituximab products is routinely performed by the manufacturer for lot release using the licence-holder’s proprietary reference materials. However, rituximab products are dosed and labelled in mass units with no reference to its biological activity; and to date, no higher order reference preparation is available. Bioassays are also an essential part of the comparability exercise against the innovator product, so-called reference product, during biosimilar development and approval process. The reference product defines biosimilarity during the comparative exercise, that is the critical quality attributes that have granted the safe and efficacious clinical history of the product. Conversely, it is recognised that an international standard for rituximab should meet different demands and is needed to control, define and calibrate the performance of potency bioassays but cannot be used to define biosimilarity (EMA, WC500127960). A WHO international standard would facilitate the assessment of the biological activities of rituximab by different stakeholders ensuring test system suitability, assisting harmonisation of bioactivity data and in turn enabling the development of products that are consistent in quality and efficacy pre- and post-marketing across the world. Further, a WHO international potency standard for rituximab will define bioactivity units but will not define specific activity (IU/mg) requirements. In the international collaborative study described in this report we have evaluated the suitability of a candidate rituximab lyophilised preparation to serve as an international standard for the biological activites of rituximab. This project was endorsed by the WHO Expert Committee on Biological Standarisation (ECBS) in October 2014 and seeks consideration by WHO ECBS in October 2017.

Aims of the Study The aim of the study was to assess the suitability of a candidate lyophilised rituximab monoclonal antibody as the 1st WHO international standard for biological activities of rituximab intended for the characterisation, calibration and validation of potency assays. To support this, the bioactivities of three preparations, the candidate, a coded duplicate and a second rituximab preparation were tested and compared with participant laboratories in house standards when available. The preparations were evaluated for their CDC activity (current batch release potency test) and, where possible, for their ADCC activity. Other bioactivities (cell-based binding and apoptosis) were also explored by a limited number of participants and will be presented here to support the suitability of the candidate preparation. The assignment of a value in units of biological activity per ampoule of the candidate reference material is also proposed.

WHO/BS/2017.2309 Page 5

Participants A total of sixteen participants from nine different countries kindly contributed to the bioassay data used in the study (table 1). Amongst the participants 9 were manufacturers, 5 were control laboratories, 1 was a Pharmacopoeia and 1 was a contract research organisation. Table 1. List of participants in the collaborative study Analía Pesce, pharmADN S.A., Carlos Villate 5148, Buenos Aires, B1605AXL, Argentina Chunyu Liu, Division of monoclonal antibody products, National Institute for Food and Drugs Control (NIFDC), No. 2 Tiantan Xili, Beijing 100050, China Disha Dadke, Global Biologics, United States Pharmacopeia – India (P) Ltd, IKP Knowledge Park, Genome Valley, Shameerpet, Hyderabad 500 078, India Jennifer Lawson, Operations, Sartorius Stedim Biooutsource, Reid Building, Block 1, Todd Campus, West of Scotland Science Park, Glasgow G20 0XA, United Kingdom Joanne Sun, Innovent Biologics, INC, 168 Dongping Street, Suzhou Industrial Park, Jiangsu 215123, China Masato Kiyoshi, Minoru Tada, Akiko Ishii-Watabe, Division of Biological Chemistry and Biologicals National Institute of Health Sciences, 1-18-1 Kamiyoga, Setagaya-ku, Tokyo 158-8501, Japan Nancy D. Ramírez-Ibañez, Luis F. Flores-Ortiz, Emilio Medina-Rivero, Analytical Development, R&D Unit, Probiomed S.A. de C.V., Cruce de carreteras Acatzingo-Zumpahuacán s/n, Estado de México, Tenancingo C.P. 52400, Mexico Paulo Roky Bamert, BTDM, Biologics Process R&D, Novartis Pharma AG, WKL-681.3.42, Klybeckstrasse 141, Basel 4057, Switzerland Sandra Prior and Chris Bird, Division of Biotherapeutics, National Institute for Biological Standards and Control, Blanche Lane, South Mimms, Potters Bar, Hertfordshire EN63QG, United Kingdom Scott Kuhns, Amgen Effector Function Characterization, Amgen Corp., One Amgen Center Dr., B30E-1B, Thousand Oaks, CA 91320, USA Sookyung Suh, Division of Advanced Therapy Product Research, National Institute of Food & Drug Safety, 187, Osongsaengmyeong 2-ro, Heungdeok-gu, Cheongju-si, Chungcheongbuk-do, 363-700, South Korea C. Nirmala Raju, Product Development, Biologics, Dr. Reddy’s Laboratories Ltd., Survey No. 47, Bachupally, Qutubullapur, R.R. District, Hyderabad 500090, India Sudha V Gopinath, Subhash Chand, Birender Kumar, PS Chandranand, National Institute of Biologicals, a-32, sector-62, Institutional Area, Noida, Uttar Pradesh 201309, India Tiffany Zhai, Song Zhao, Shanghai CPGuojian Pharmaceutical Co.,Ltd., No.399 Libing Road, Zhangjiang Hi-tech Park, Shanghai 201203, P.R. China Yang Cao, Quality Control, Shanghai Henlius Biotech Co., Ltd., 1289 Yishan Road, Shanghai 200233, P. R. China Yong Suk Yang, QC C&I3 Team, Celltrion, Celltrion Plant 2, 20, Academy-ro 51, Yeonsu-gu, Incheon 406-840, Korea

WHO/BS/2017.2309 Page 6 Note that the participants are identified in the study by a number (from 1 to 16) which is in no way related to the order of the above listing.

Materials and Processing A preparation of recombinant chimeric Rituximab expressed in CHO cells was kindly donated to WHO by Martin Schiestl (Sandoz GmbH, Austria). Suitable Certificate of Analysis, Safety Data sheet and product specification details were also provided (not shown). Trial fills were carried out and the biological activity of different formulated lyophilised preparations were compared with the bulk material using an in house CDC assay to select a suitable formulation (not shown). The rituximab standard candidate production fill was prepared at NIBSC following standardised procedures for the preparation of International Biological Standards (ECBS guidelines –WHO Technical Reports, 55th Report, 932:73-130, 2006). A clinical batch of Mabthera® (Roche) was purchased from ADAllen Pharma Ltd (UK), reformulated as per candidate optimised conditions and a number of ampoules of lyophilised material were prepared to include in the collaborative study. Details of the preparations used in the study are shown in table 2. Buffers and excipients were prepared using water for irrigation (Baxter, Switzerland) and filtered using sterile and non-pyrogenic 0.22 μm filters (Nalgene®, Nalge Nunc International, USA). The preparations were lyophilised from 1 mL of solution containing 100 μg rituximab protein and 1% (v/v) clinical grade human serum albumin (HSA) in formulation buffer. The nominal content of protein in the ampoules is calculated from the dilution of the bulk material and the assumed protein mass content as per manufacturer’s data. Briefly, 1 mL of solution was lyophilised under optimised and controlled conditions; the glass ampoules were sealed under dry nitrogen by heat fusion and stored at -20 °C in the dark until shipment at room temperature. Coded duplicates were prepared from the candidate preparation (Sample A and B) and the lyophilised preparation of the innovator product was used for comparison and is referred here as to Sample C (table 2). Table 2. Preparations used in the collaborative study Study Code NIBSC Ampoule Code 14/210 Fill date No of ampoules in stock ~5,000 Excipients 1% (v/v) Human Serum Albumin 1% (v/v) Human Serum Albumin Rituximab nominal weight per ampoule* 100 μg

Samples A & B

31/10/2014

Sample C

SS-573

15/01/2016

~300

100 μg

*Note that the rituximab ampoule content in μg is nominal and has no status as a declared mass content.

WHO/BS/2017.2309 Page 7

Characterization of the lyophilized preparations The lyophilised preparation characteristics comply to reviewed and approved specifications suitable for WHO reference materials and are described in table 3. No evidence of microbial contamination was found by total viable count method. Table 3. Fill production details of the preparations used in the collaborative study NIBSC Mean Fill Ampoule weight (n)a Code 14/210 SS-573 a b

CV fill weight (%)b 0.24 0.02

1.0079 (187) 1.0108 (3)

CV % Mean residual residual moisture moistureb (n)a (%)b 0.73 (12) 22.86 0.23 (3) 6.68

% Mean headspace oxygenb (n)a 0.18 (12) 0.40 (3)

CV headspace oxygen (%)b 55.6 26.6

Number of fills measured Percentage (%) w/w

Study design and bioassay methods A questionnaire sent to the participants gathered availability and capability of the laboratories to perform cell-based assays for the bioactivities of rituximab for the study. Sixteen laboratories performed complement dependent cytotoxicity assays (CDC), 11 laboratories performed antibody dependent cellular cytotoxicity assays (ADCC), 5 participants performed cell-based antibody binding assays and one laboratory performed an apoptosis assay. The laboratories were encouraged to use their qualified bioassay methods, include routine positive and negative controls and their qualified in house reference standard where possible. Participants were asked to perform three assay runs on three different days using fresh ampoules each day, with each assay run including at least 3 assay plates and at least 2 independent dilution series. Each of the preparations were to be tested concurrently per plate and with sample randomisation as much as reasonably possible within plates and between days. The participants received the collaborative study protocol that included bioassay protocols and layouts as examples only (Appendix 4), instructions for use, excel template sheets to record data and methodological details, and the ampoules of the lyophilised test materials coded as Sample A, Sample B and Sample C. Samples A and B were coded duplicates of the candidate material (NIBSC 14/210) and Sample C was included for comparison (NIBSC SS-573) (table 2). Enough ampoules were provided to perform the three assay runs, conduct preliminary assays to establish suitable working dilution ranges for the test materials and in case of accidental loss. The preparations were shipped at room temperature. It was instructed that ampoules were stored at 20 °C upon arrival and reconstituted with 1 mL of sterile distilled/deionised water on the day of the assay as described per provided instructions. Table 4 shows a summary of the assay methods used by the participants in the study.

WHO/BS/2017.2309 Page 8

Statistical analysis An independent statistical analysis of all bioassay data was performed at NIBSC. Analysis of dose-response curve data was performed using a four-parameter logistic model: y=αδ 1 + 10 β(log10 x- log10 γ)

where y denotes the assay response, x is the concentration, α is the upper asymptote, δ is the difference between upper and lower asymptotes,  is the slope factor and γ is the EC50 (50% effective concentration). Models were fitted using the R package ‘drc’. Parallelism (similarity) for a pair of dose-response curves was concluded by demonstrating equivalence of the parameters α,  and δ independently for each of the 3 bioassays (CDC, ADCC and cell binding). For this approach, ratios of these parameters for the two samples under consideration were calculated and log transformed. Approximate 90% confidence limits for these values were determined using the delta method. Extreme values (defined as the maximum of the upper confidence limit and the reciprocal of the lower confidence limit) were calculated and equivalence concluded in cases where these were below pre-defined upper equivalence bounds. The calculated upper equivalence bound values and the rationale for determining them is shown in the Results section. As apoptosis assays were performed by a single laboratory only, the validity of the assays was concluded when no significant non-parallelism was found by analysis of variance. Relative potency estimates were calculated as the ratio of EC50 estimates in assays where acceptable parallelism was concluded. All relative potency estimates were combined to generate unweighted geometric mean (GM) potencies for each laboratory and these laboratory means were used to calculate overall unweighted geometric mean potencies. Variability between assays and laboratories has been expressed using geometric coefficients of variation (GCV = {10 s-1} × 100% where s is the standard deviation of the log10 transformed potencies).

Stability studies Accelerated degradation studies are performed to predict the long-term stability of the candidate preparation (NIBSC 14/210) and are currently ongoing in NIBSC. Ampoules of the lyophilised preparation were stored at elevated temperatures, namely 56 °C, 45 °C, 37 °C, 20 °C and 4 °C and tested at indicated time points together with ampoules stored at the recommended temperature of -20 °C and -70 °C as baseline reference temperature. To date, tests for CDC and ADCC bioactivities have been performed after 10, 20 and 27 months of storage. The relative bioactivities of the accelerated thermal degradation samples wereill be used to fit an Arrhenius equation relating degradation rate to absolute temperature assuming first-order decay (Kirkwood 1977), and hence predict the degradation rates when stored at -20 °C. Real time stability studies are also conducted for stability monitoring.

WHO/BS/2017.2309 Page 9 Table 4. Summary of the bioassays that contributed to the study. Laboratory code 1 2 3 4 5 6 Bioactivity CDC ADCC CDC CDC CDC CDC CDC ADCC Binding Apoptosis CDC ADCC CDC ADCC Binding CDC CDC ADCC Binding CDC ADCC CDC ADCC Binding CDC ADCC CDC ADCC CDC ADCC CDC ADCC Binding a

7 8

9 10

11 12

13 14 15 16

Target cell line WIL2-S WIL2-S WIL2-S WIL2-S WIL2-S WIL2-S Raji Raji Raji Z-138 WIL2-S WIL2-S Raji WIL2-S Raji Raji WIL2-S Raji WIL2-S WIL2-S WIL2-S Daudi Daudi Daudi WIL2-S WIL2-S Jeko Jeko WIL2-S Raji WIL2-S WIL2-S WIL2-S

Source of Complement Human N/A Human Human Human Rabbit Rabbit N/A N/A N/A Human N/A Human N/A N/A Human Human N/A N/A Rabbit N/A Human N/A N/A Rabbit N/A Rabbit N/A Human N/A Human N/A N/A

Ratio E:Ta N/A 6 N/A N/A N/A N/A N/A 19 N/A N/A N/A 7.6 N/A 5 N/A N/A N/A 1 N/A N/A 5:1 N/A 5 N/A N/A 25 N/A 8 N/A 16 N/A 8 N/A

Assay Type Viability Reporter gene Viability Viability Viability Viability Viability NK cell line Killing Competition binding Apoptosis Viability Reporter gene Viability Reporter gene Binding Toxicity Viability NK cell line Killing Binding Viability PBMC based Killing Toxicity Reporter gene Binding Toxicity NK cell line Killing Viability Primary NK cell Killingb Viability PBMC based Killingb Viability Dual-reporter gene Binding

Assay duration (h)b 18 6 18 20 ± 4 18 18 ± 2 2 1 ~1 24 ± 1 6 19 7 6 0.5 5 4-4.5 2.5 1-1.5 2 4 2 4 0.5 1 1 2 18-22 2 4 18 18 1

In house reference standard Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Assay read out (reagent) Fluorescence (Alamar Blue) Luminescence (Luciferase) Fluorescence (Alamar Blue) Fluorescence (Alamar Blue) Fluorescence (Alamar Blue) Fluorescence (Alamar Blue) Luminescence (Cell-titre Glo) Fluorescence (Calcein AM release) Fluorescence (FACs) Fluorescence (Annexin-V by FACs) Colorimetric (MTS) Luminescence (Luciferase) Colorimetric (CCK-8) Luminescence (Luciferase) Fluorescence (2ry Ab detection by FACs) Colorimetric (LDH) Colorimetric (CCK-8) Colorimetric (LDH) Colorimetric (2ry Ab detection) Fluorescence (Alamar Blue) Fluorescence (Cyto-tox reagent) Luminescence (Cell-titre Glo) Luminescence (Luciferase) Fluorescence (2ry Ab detection by FACs) Fluorescence (Calcein AM release) Fluorescence (Calcein AM release) Luminescence (Cell-titre Glo) Luminescence (Cyto-tox Glo) Colorimetric (CCK-8) Fluorescence (Calcein AM release) Fluorescence (Alamar Blue) Luminescence (Firefly & Renilla Luciferase) Fluorescence (2ry Ab detection by FACs)

E:T- effector: target cell ratio; b (h) hours

WHO/BS/2017.2309 Page 10

Results Data returned for analysis The sixteen different participant laboratories returned data for one, two, three or four bioassays and are represented in this report with a number from 1 to 16 allocated at random that has no relation to the listing order in table 1. The individual laboratory data remain confidential. For the purposes of analysis, assay methods were categorised as “CDC” (a total of 152 plates from 16 laboratories), “ADCC” (110 plates from 11 laboratories), “Binding” (27 plates from 5 laboratories) and “Apoptosis” (9 plates from 1 laboratory). All of the laboratories provided data from three independent runs performed on three different days as requested. For most of the laboratories that was a total of 9 plates, each plate accommodating the three study preparations, the in house reference when available and including two independent dilution series per sample with some randomisation. A few exceptions to the above usually in alignment with laboratory in house procedures were laboratory 6 (CDC: 2 plates per assay run for samples A, B & C, 3 plates per assay run for in house reference, triplicate independent dilutions - data from a total of 6 plates per preparation; ADCC: 3 assay runs with 2, 4 and 3 plates per run respectively - data from a total of 9 plates), laboratory 7 (for both CDC and ADCC: 6 plates for Sample A and in house reference and 3 plates for Samples B & C - data for a total of 18 plates), laboratory 8 (Cell binding assay: 1 plate per assay run - data from a total of 3 plates), laboratory 12 (Cell binding assay: 1 plate per assay run - data from a total of 3 plates), laboratories 10 and 15 (CDC: no sample randomisation), laboratory 16 (ADCC: data from 4 assay runs – data from a total of 12 plates; no sample randomisation but unused plate edges; Cell binding assay: 1 plate per assay run - data from a total of 3 plates) and laboratory 14 (ADCC: data from a total of 8 plates; changed dilution range in assay runs 2 & 3). The data also included in house controls (negative controls and positive controls) used by the laboratories which were reviewed but not used in the data analysis. When the collaborators reported failed in house laboratory system suitability criteria and raw data were provided, the data were included in the independent analysis performed for this report. In some cases, one or two doses were removed at the upper asymptote of the dose-response curve; for example, around half of ADCC assays showed a decrease in activity at the highest antibody concentrations (Laboratories 1, 7, 8, 10 and 12).

Assay validity (parallelism of dose-response curves) The approach taken for the assessment of parallelism in this study requires equivalence bounds to be set, as described above, for each model parameter (α, β and δ) in each assay method. As these parameters are expected to be equivalent when testing the same sample against itself, upper equivalence bounds were set as the 90th percentile of observed extreme values (defined as the maximum of the upper 90% confidence limit and the reciprocal of the lower 90% confidence limit) for the coded duplicate candidate preparation (Samples A & B). The choice of upper equivalence bounds and the impact on validity of assays is illustrated for each assay type independently in the supplementary data (Appendix figures A1, A2 and A3). Upper equivalence bounds and the percentage of valid assays obtained using these values are summarised in the Appendix (Table A1). For an assay to be concluded as valid, equivalence

WHO/BS/2017.2309 Page 11 had to be demonstrated for all three parameters (α, β and δ). It should be noted that the equivalence bounds were intended for use in the analysis of data from this study only and they should not be interpreted as suitable values for routine use in the assessment of assay validity within the collaborating laboratories. Overall, the coded duplicate candidate preparations showed the lowest number of invalid assays. The percentage of invalid assays per lab, bioactivity assay and parameter is shown in the Appendix (tables A2-A5). The relative performance of the laboratories using the defined equivalence criteria ranged from none to 78% invalid assays, when individual laboratory data was assessed and preparations compared to the candidate preparation (Sample A). Relative performance ranged from 0 to 100% invalid assays when any of the preparations were compared to the in house reference standard. Although the data are limited when the ADCC assay is stratified by assay type (5 reporter assays, 3 NK cell line killing assays and 3 primary cell-based killing assays) and depend on the laboratory, in general, the NK cell line killing assays seem to be associated with a lower number of invalid assays and the reporter assays seems to show a trend towards more pronounced differences between the candidate preparation (and coded duplicate) (Samples A & B) and the in house reference samples. No invalid assays (no significant non-parallelism) were found for the apoptosis assay performed by laboratory 6. Further illustration of the number of invalid assays by assay, laboratory, curve parameter and sample preparation comparison can be found in Appendix (figures A4A6 and Table A5). All of the laboratories in the study produced assays acceptable under the set criteria and these assays were used to calculate the sample potency estimates with the exception of laboratory 10 that produced no valid assays for the CDC assay when comparing Sample C versus their in house reference standard.

Estimates of relative potency Potency estimates relative to in house reference standards Potency estimates were calculated relative to in house (IH) reference standards where available, for each laboratory and assay method as summarised in tables 5-8 and figure 1. The combined potency data for each bioactivity is summarised in table 9. Intra-laboratory GCV values for Sample A (candidate) and Sample C CDC potencies relative to in house reference standards ranged between 3% and 13%. The GM potency estimates from the individual laboratories for Sample A ranged from 0.95 to 1.48 and Sample C varied between 0.93 and 1.47 (table 5). The dispersion of these potency estimates was measured by the inter-laboratory GCV and was 16% and 14% for Samples A and C, respectively (table 9). Laboratories 9, 10 and 12 showed relatively higher CDC potency estimates than the rest of the laboratories for both preparations when compared to their in house reference standards. Intra-laboratory variability was higher for the ADCC assays than for the CDC assay with GCV values ranging between 7% and 36% for the individual laboratories. The GM potency estimates ranged from 0.44 to 2.17 and from 0.64 to 2.42 for Sample A and C, respectively, when expressed relative to in house reference standards (table 6). The inter-laboratory variability of the calculated potency estimates was 62% and 51% for Samples A and C, respectively. The highest potency estimates were calculated for laboratory 12 and the lowest for laboratory 10.

WHO/BS/2017.2309 Page 12 For cell-binding assays the intra-laboratory variability ranged between 7% and 17% (GCV) and the potency values relative to in house reference standards ranged between 0.57 and 1.09 and from 0.57 to 1.04 for Samples A and C, respectively (table 7). The inter-laboratory GCV of the potency values was 33% and 29% for Samples A and C, respectively. Laboratory 10 showed the lowest potency. The apoptosis assay was performed only by laboratory 6 and the GM potency was estimated as 1.09 for Sample A and 1.01 for Sample C when expressed relative to their in house reference standard with intra-laboratory GCVs of 13% and 11% respectively (table 8). Table 5. Laboratory geometric mean relative potency estimates for CDC assays. Versus IH Ref Candidate Sample C (Sample A) GM %GCV n GM %GCV 1.01 1.00 1.15 1.01 1.08 0.99 1.26 1.48 1.11 1.42 0.96 0.99 0.95 10 7 7 n/a 12 9 4 n/a 12 n/a 7 4 6 9 0.98 Not available 9 0.98 Not available 9 1.12 2 1.07 14 1.01 8 0.93 8 1.22 2 n/a 9 1.13 2 1.47 9 0.95 Not available 9 0.99 8 0.94 12 8 9 7 7 9 8 n/a 13 n/a 4 3 7

Method CDC CDC CDC CDC CDC CDC CDC CDC CDC CDC CDC CDC CDC CDC CDC CDC

Lab 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

n 8 9 9 4 8 8 8 0 8 1 9 9 8

Versus candidate (Sample A) Coded duplicate Sample C (Sample B) GM %GCV n GM %GCV 0.99 1.03 1.00 1.03 0.96 0.95 0.99 0.98 0.99 1.05 1.02 1.00 1.00 0.98 1.00 1.00 16 11 5 10 7 n/a 6 9 4 6 10 5 9 4 3 7 8 3 9 9 9 2 7 8 9 5 8 4 9 8 9 7 0.99 0.95 0.98 0.97 0.98 1.02 0.92 0.95 0.96 1.16 1.03 0.97 0.99 0.98 1.00 0.99 17 19 5 7 10 2 13 9 8 n/a 10 6 6 6 3 4

n 8 3 9 9 9 3 7 7 9 2 8 3 9 7 9 8

Table 6. Laboratory geometric mean relative potency estimates for ADCC assays. ADCC-R and ADCC-DR indicate labs using a single or dual gene reporter assay, respectively and end-point killing based assays are shown as ADCC-NK and ADCC-P for assays using NK cell lines or primary cells, respectively. Versus IH Ref Candidate Sample C (Sample A) % % GM n GM n GCV GCV 0.85 1.28 1.46 0.94 0.44 2.17 0.75 0.78 26 21 27 7 21 n/a 10 32 5 0.95 6 1.39 15 1.73 5 1.17 9 0.64 Not available 2 2.42 9 1.13 Not available 8 0.92 Not available 25 20 16 7 15 13 14 36 6 9 8 6 7 7 9 9

Method ADCC-R ADCC-NK ADCC-R ADCC-R ADCC-NK ADCC-P ADCC-R ADCC-NK ADCC-P ADCC-P ADCC-DR

Lab 1 6 7 8 10 11 12 13 14 15 16

Versus candidate (Sample A) Coded duplicate Sample C (Sample B) % % GM n GM GCV GCV 0.96 1.00 0.99 1.01 0.91 1.03 1.00 0.99 0.92 0.86 0.93 30 13 17 7 18 33 20 10 25 20 12 9 8 9 9 9 7 6 9 3 7 7 1.06 1.11 1.32 1.25 1.44 1.60 1.06 1.50 1.48 1.17 1.26 23 15 29 7 25 21 22 8 13 35 15

n 7 7 6 9 9 7 5 9 4 7 6

WHO/BS/2017.2309 Page 13 Table 7. Laboratory geometric mean relative potency estimates for cell-binding assays. Versus IH Ref Candidate Sample C (Sample A) % % GM n GM n GCV GCV 1.09 0.91 0.57 0.97 7 17 12 9 7 1.04 3 0.92 5 0.57 Not available 3 0.86 8 11 9 8 6 3 6 3

Method Binding Binding Binding Binding Binding

Lab 6 8 10 12 16

Versus candidate (Sample A) Coded duplicate Sample C (Sample B) % % GM n GM n GCV GCV 0.99 1.02 0.96 0.99 0.95 7 19 16 10 4 9 3 5 3 3 0.93 1.01 1.01 1.02 0.88 11 11 5 n/a 1 8 3 4 1 3

Table 8. Laboratory geometric mean relative potency estimates for apoptosis assays. Versus IH Ref Candidate (Sample A) Method Apoptosis Lab 6 GM 1.09 % GCV 13 n 9 Sample C GM 1.01 % GCV 11 n 7 Versus candidate (Sample A) Coded duplicate Sample C (Sample B) % % GM n GM n GCV GCV 1.00 12 7 0.89 12 7

Table 9. Potency summary data relative to in house reference standards Versus IH Ref Method GM CDC ADCC Binding Apoptosis 1.10 0.98 0.86 1.09

Candidate (Sample A) LCL UCL % GCV 1.00 0.65 0.55 n/a 1.20 1.47 1.36 n/a 16 62 33 n/a

n 13 8 4 1

GM 1.06 1.20 0.83 1.01

LCL 0.97 0.85 0.55 n/a

Sample C UCL % GCV 1.15 1.69 1.25 n/a 14 51 29 n/a

n 12 8 4 1

Potency estimates relative to the candidate preparation (Sample A) Potency estimates were calculated relative to the candidate preparation (Sample A) used as a reference standard for each laboratory and assay. These are summarised in tables 5-8 and figure 1. The combined potency data for each of the bioactivity tested is shown in table 10. Intra-laboratory GCV values for CDC assays ranged between 2% and 19% similarly to the GCV for the potency estimates when expressed relative to in house references as shown before. The GM potency estimates ranged from 0.95 to 1.05 for Sample B and 0.92 to 1.16 for Sample C (table 5). The potency values for the different laboratories expressed relative to the candidate preparation (Sample A) were in good agreement with inter-laboratory variabilities (GCV) of 3% and 5% for Sample B (coded duplicate) and Sample C, respectively. These GCVs were considerably lower than those calculated relative to in house

WHO/BS/2017.2309 Page 14 reference standards. When combining the CDC data from all 16 participants the potency estimates for the coded duplicate and for Sample C were 1.00 and 0.99, respectively. Intra-laboratory variability for the ADCC assays ranged between 7% and 35% for the individual laboratories, which is comparable to the variability observed for potency values relative to in house references. However, the GM potency estimates relative to candidate preparation (Sample A) ranged from 0.86 to 1.03 and from 1.06 to 1.60 for Samples B and C, respectively, values that were less dispersed than those calculated relative to in house reference standards (table 6). This was also illustrated by the low inter-laboratory GCVs of the combined laboratory ADCC potencies of 6% and 16% for Samples B and C, respectively. The combined data showed an ADCC potency of 0.96 for Sample B (coded duplicate) and 1.28 for Sample C. For cell-binding assays the intra-laboratory variability was between 1% and 19% and also comparable to that calculated with in house reference standards. The calculated GM potencies ranged between 0.95 and 1.02 for Sample B (coded duplicate) and from 0.88 to 1.02 for Sample C respectively (table 7). The overall calculated potency values for both preparations were 0.98 with a GCV of 3% for Sample B (coded duplicate) and 0.97 with a GCV of 7% Sample C. The apoptosis assay, performed only by laboratory 6 showed a GM potency of 1.00 for the coded duplicate (Sample B) and 0.89 for Sample C when expressed relative to the candidate preparation (Sample A) with an intra-laboratory GCV of 12% for both estimates (table 8). The data show that the use of the candidate preparation (Sample A) as reference standard to calculate the relative potency of Samples B and C allows a close agreement between laboratories for each of the bioactivities tested (CDC, ADCC or cell binding) (table 10). Table 10. Potency summary data relative to candidate preparation (Sample A) Method CDC ADCC Binding Apoptosis Versus candidate (Sample A) Coded duplicate (Sample B) GM LCL UCL % GCV n GM LCL 1.00 0.96 0.98 1.00 0.98 0.93 0.95 n/a 1.01 1.00 1.02 n/a 3 6 3 n/a 16 11 5 1 0.99 1.28 0.97 0.89 0.96 1.16 0.90 n/a

Sample C UCL % GCV 1.01 1.41 1.05 n/a 5 16 7 n/a

n 16 11 5 1

WHO/BS/2017.2309 Page 15 Figure 1. Box-plot summary of laboratory geometric mean relative potency estimates

Stability prediction and monitoring Accelerated degradation studies Stability studies to predict yearly loss of bioactivity require acceleration degradation experiments conducted through a period of time and at elevated temperatures. To date, bioactivity of the candidate preparation was measured after 10, 20 and 27 months and no detectable loss of CDC and ADCC bioactivity was only found even at the higher temperatures tested. Stability prediction wascannot yet be calculateddrawn from the isrelative bioactivities of the accelerated degradation samples at the three single time data points available to date, as shown in the ECBS 2017 meeting. The results predict no loss of CDC activity and very marginal loss of ADCC activity (0.01%) upon storage at the recommended storage temperature of -20 ºC with Chi-squared P-values of 0.308 and 0.302 (goodness of fit) respectively. andLonger-term accelerated degradation studies for NIBSC 14/210 are still ongoing. Additional data accrued and forward plans will be shown in the ECBS 2017 meeting. Currently available Stability prediction data at 10 months is shown in table 11.

“Real time” stability studies “Real time” stability studies are also ongoing for stability monitoring. Data available from the three a single time points up to date, after 10, 20 and 27 months of storage at the recommended temperature of -20°C and compared to baseline temperature of -70°C, is

WHO/BS/2017.2309 Page 16 shown in table 12 1as discussed. Additional data available and forward plans will be shown in the ECBS 2017 meeting. The data to date shows no loss of CDC or ADCC activity at the recommended temperature of -20 ºC. Further no loss of activity was recorded at 4 ºC (Table 12) or +20 ºC (not shown). Table 11. Accelerated degradation studies: Predicted percentage of annual loss of activity for the candidate preparation (14/210) at the recommended storage temperature of -20 ºC, as per data available to date (up to 27 months). The 95% upper confidence limit and goodness of fit parameter (Chi-squared P-value) are also shown. Bioactivitya CDC ADCC a

Storage Temperature -20 ºC -20 ºC

Predicted % Loss per Year 0 0.01

95% UCL % Loss 0 0.06

Chi-squared P-value 0.308 0.302

The studies were performed at NIBSC using WIL-2 S cells as target cells and human serum complement for the CDC assay and a reporter effector cell line for ADCC assay (Promega, UK) (Cheng, Garvin et al. 2014). Accelerated thermal degradation samples were assayed for CDC and ADCC bioactivity, and potency was calculated relative to the baseline temperature (-70 °C) for each temperature and time point and used to feed the model. For each time point, data from one assay with two independent plates per temperature and two independent dilution series per plate was used with the expection of ADCC at 10 months where only ampoules stored at 37 °C and -20 °C were tested in one plate.

Table 12. Stability monitoring: Potency estimates for the candidate preparation (14/210) stored at the recommended storage temperature of -20 ºC and at 4 ºC for 10, 20 and 27 months relative to ampoules stored at baseline temperature of -70 °C. The 95% upper and lower confidence limits are also shown. Time (months) 10 -20 °C 4 °C CDC 20 -20 °C 4 °C 27 -20 °C 10 20 ADCC 27 -20 °C -20 °C 4 °C -20 °C 4 °C Storage Temperature 4 °C Using -70 °C as reference a a Relative 95% LCL 95% UCL potency 0.97 0.95 0.95 0.98 0.99 0.99 1.01 1.11 1.05 0.85 0.97 0.92 0.91 0.87 0.90 0.92 0.94 0.94 0.98 0.88 0.62 0.87 1.02 1.00 1.03 1.08 1.06 1.04 1.08 1.26 1.26 1.16 1.07

Bioactivity

a

WHO/BS/2017.2309 Page 17 a

The studies were performed at NIBSC using WIL-2 S cells as target cells and human serum complement for CDC assay and a reporter effector cell line for ADCC assay (Promega, UK) (Cheng, Garvin et al. 2014). At each time point, potency was calculated relative to the baseline temperature (-70 °C) for CDC and ADCC bioactivities. One assay with two independent plates per temperature and two independent dilution series per plate was used with the expection of ADCC at 10 months where only ampoules stored at -20 °C were tested in one plate.

Discussion Monoclonal antibodies are complex molecules and changes in critical quality attributes (CQA) and critical process parameters (CPP) may have a significant impact on their clinical safety and potency. It is recognised that both unidentified deviations and intended process changes, although tightly regulated, may lead to product drift and evolution (Ramanan and Grampp 2014). Further, the introduction of biosimilars may result in various marketed products becoming different over time. An example of changes in the quality attributes of batches of the innovator rituximab product (Rituxan® and Mabthera®) and in particular on the glycosylation profile, has been reported and associated with an increase in ADCC activity suggesting changes in the manufacturing process (Schiestl, Stangler et al. 2011). Bioassays are key tools for product characterisation and depend on the use of reference standards. The reference product (innovator) is neccessary to demonstrate biosimilarity (Nikolov and Shapiro 2017) (FDA guidelines, 2016; EMA Guidelines, 2010). However, the reference product and laboratory proprietary in house reference standards, are on their own unable to ensure continuity of potency of post-marketed products, especially when products by various manufacturers are approved. Furthermore the situation may rise when the innovator product may cease to exist. Rituximab approved products (Mabthera® in Europe or Rituxan® in the USA) are marketed and dosed in mass units, with no biological activity information on their label. Products comply with CDC bioactivity specifications in units per volume as per the proprietary manufacturer’s reference standard. This constitutes a challenge for biosimilar manufacturers that have no independent reference standard to calibrate their bioassays and rely solely on the reference product and their own in house reference standards to define bioactivity. It has been acknowldeged that the establishment of an international standard for the bioactivities of rituximab would contribute to the calibration of potency assays and that way support product consistency amongst manufacturers and other stakeholders over time (Thorpe and Wadhwa 2011). With this view, the suitability of a candidate preparation of rituximab to serve as the 1 st WHO International Standard for the bioactivities of rituximab was assessed in a multi-centre collaborative study. For the first time data has become available to illustrate the role for this new class of reagents, a role that has been very much discussed and raised polarised opinions in the last few years (WHO Technical Report Series, 66th Report, 999:13-15, 2016). The study was used to assess the bioactivity of the rituximab candidate preparation using the participants qualified bioassays and in house reference standards. Participant laboratories included manufacturers and regulatory institutions. Thus, in house reference standards used are assumed to be those routinely used in supporting product development, system suitability and/or preclinical and clinical studies for the manufacturer labs. Typically, a clinical batch of the innovator product was used as in house reference for the study by other laboratories. In general, data for the CDC, ADCC and cell-based binding activities showed very poor agreement between laboratories when the potency estimates of the study preparations (Sample A, B and C) were expressed relative to the in house reference standards, with the greatest variability between laboratories observed for the ADCC assay (GCV of 62% and

WHO/BS/2017.2309 Page 18 51% for Sample A and C, respectively). These results reflect the differences between the in house reference standards used by each laboratory and highlight the lack of potency bioassay harmonisation amongst laboratories and the need for a higher order reference standard. Different factors may contribute to the greater inter-laboratory variability in the potency for ADCC, including the intrinsic complexity of the ADCC assay, the different assay platforms used and the sensitivity of the assay to potential differences in the quality attributes of the rituximab in house reference standards that may not have such an impact in other bioactivities (i.e. different glycosylation profile may affect ADCC activity as described by Schiestl et al., (Schiestl, Stangler et al. 2011) (Cheng, Garvin et al. 2014) but is less likely to impact on CDC activity (Kellner, Derer et al. 2014)). In general a higher intra-laboratory variability (736%) for the ADCC assay as compared to the CDC, cell-based binding and apoptosis assays (3-17%) may also reflect specific challenges of the ADCC bioassay that affect performance. However no association with the ADCC assay platform (reporter gene assays or end-point killing assays) and intra-laboratory assay variability was found in this study. Neither could the inter-laboratory variability in ADCC activity be linked to the assay platform performance suggesting that the reported differences mainly reflect the differences between the in house reference standards used by each laboratory, differences that are more pronounced in terms of ADCC activity. For this study, assay validity criteria was set based on data from coded duplicates (blinded identical samples). As expected, the lowest number of invalid assays was found when the coded duplicates (Samples A and B) were compared which can only be attributed to assay performance. In general, the percentage of invalid assays for other sample comparisons was similar to that for the coded duplicate (i.e. high or low percentage of invalid assays for all comparisons). This was in line with the study preparations and in house reference having a comparable behaviour in the bioassays, albeit some differences in potency may be observed. Thus, a low percentage of invalid assays for the coded duplicates whilst a consistent high percentage of invalid assays for other comparisons, may reflect dissimilar behaviour of the samples in the bioassay. This may be the case, to some extent, for example in the reporter ADCC bioassay for laboratories 1, 8 and 12 (Appendix, table A3). International reference materials are intended to be long-lasting stable preparations suitable for global distribution, thus product-specific formulation and process development is optimised to fulfil this requirement whilst preserving bioactivity required for the standard’s intended use (ECBS guidelines –WHO Technical Reports, 55th Report, 932:73-130, 2006). Although a number of assays were considered invalid using the criteria defined for this study, all of the 16 laboratories reported data that was used to estimate the relative potency of the preparations in each of the bioassays. Under the criteria used, 70-85% of the assays were valid assays, depending on the preparations compared, and contributed to the overall potency calculations. Note that due to the methods applied to determine equivalence criteria, an invalidity rate of at least 10% is to be expected using this approach. Preliminary assays were carried out by the participants in order to establish suitable dilution schemes for the study preparations using their in house qualified methods; consequently sample concentration range was optimised and differed for the different laboratories. One of the participants, laboratory 11, reported ADCC data using a concentration range optimised for the candidate preparations that was different to that used in their in house validated method for their in house reference standard. Equivalent doses of in house rituximab reference standard failed their assay system suitability criteria and the lab only returned data for the study preparations (Samples A, B and C). It was noted that single gene reporter ADCC assays showed a decrease in bioactivity at the higher antibody concentration(s) and

WHO/BS/2017.2309 Page 19 one or two doses were excluded in the analysis; however, this trend was not evident for laboratory 16 that performed a dual reporter assay or for ADCC killing assays with the exception of laboratory 10. Furthermore laboratory 15 reported that their in house validated ADCC reporter assay was unsuitable to assess the study preparations due to reduced activity at the highest concentration, an effect that was not observed with their in house reference standard. This laboratory however successfully performed an end-point killing assay using PBMC with both, the study preparations and their in house reference standard, and returned data that is included in the study analysis and contributes to the potency results reported. It can be speculated that the inhibition observed for the reporter assays at the highest concentrations could be due to direct effects of the study preparations on the surrogate effector cells. It has been shown that excess endogenous human IgG in serum inhibits therapeutic antibody ADCC activity (Iida, Misaka et al. 2006, Preithner, Elm et al. 2006). The lyophilised preparations contain 1% human serum albumin (HSA ≥ 95% albumin)- that is approximately 100 times more HSA protein in mass than rituximab protein- therefore traces of IgG in the HSA may be significant at high concentrations of the rituximab study preparations and interfere with the reporter assay by competing for the FcR on the surrogate effector cells or affecting the NFAT-RE signalling pathway thus damping the reporter signal at high concentrations. Usually, low concentration of low IgG serum (i.e. 4%) is used in ADCC reporter assays to avoid IgG interference (Promega, ADCC reporter bioassay). Inhibitory effects due to competition for CD16 binding could also be expected on ADCC killing assays, although for these assays often 10% FBS supplemented media is reported (Chung, Lin et al. 2014). For primary cell-based killing assays, higher expression of CD16 receptors as compared with the reporter effector cells has been shown to positively impact on rituximab ADCC activity (Hatjiharissi, Xu et al. 2007) and also for PBMC assays, other cell types such as monocytes may also mediate ADCC activity in the coculture (Chung, Lin et al. 2014). Further direct cytotoxicity and/or steric hindrance may also contribute at the highest concentrations of rituximab as the total protein increases, a phenomenon that is frequently observed in many immunoassays. When a dual reporter assay was used (laboratory 16), the direct effects of the preparation on the effector cells may be normalised, the potential effect on the surrogate effector cells was not observable and/or the dual reporter cell line may be more resilient to the effects of the HSA concentration. However, it should be emphasised that although some of these factors may need consideration and may contribute to the effects observed, the antibody concentration range used varies amongst labs and therefore it is unclear whether these effects remained undetected or were not present for some of the assays and labs. The present study shows that the potency estimates for rituximab bioactivities in vitro using participant in house methodologies were in good agreement when the different laboratories used the candidate standard preparation as opposed to the great variability observed when the potency was calculated relative to their in house reference standards. The CDC assay is the current lot release assay for rituximab products but the study was extended to other relevant bioactivities such as ADCC, cell-based binding and apoptosis that are also assessed during product development and characterisation. The results from the study substantiate the same conclusion for CDC and ADCC activity and although the data was limited, cell-binding and apoptosis results further support the findings. The data suggest that the candidate preparation is suitable to serve as an international standard to help harmonise potency data between laboratories in line with the intended role of this new type of reagent. The GM relative potency for Sample C in CDC assays was 0.99 (GCV 5%) showing a similar potency to that of the candidate standard, but the ADCC potency estimate of Sample C was significantly greater than 1 (GM= 1.28) and had higher associated variability (GCV 16%). This variability

WHO/BS/2017.2309 Page 20 was greater than that observed for the coded duplicate (Sample B) (GCV 6%) and was not related to the ADCC assay platform used. We can hypothesise that these results may reflect differences between the two rituximab preparations (Samples A and C) that affect specifically their ADCC potency and were identified by the different labs and ADCC assays to a different extent. Published reports (Hsieh, Aggarwal et al. 2016) (Chung, Lin et al. 2014) on trastuzumab and on an anti-CD20 monoclonal antibody respectively, describe differences between the ADCC potency and efficacy of these Mabs when different effector cells- primary PBMC or primary NK cells, engineered NK cell lines or engineered reporter cell lines- are used to asses ADCC activity. In this work the differences are credited to differences in the biological properties and expression profiles of different activating and inhibitory receptors in the effector cells used, i.e. differences in the number of CD16 cell surface molecules (FcɣIII receptors), receptor polymorphism, molecular interactions between the Fc domain of the Mab and the FcɣIII receptor, the involvement of specific FcɣIII receptor subunits and transduction signals in the primary cells, transfected NK cells and reporter cells. Further ADCC activity using primary cells is also affected by cryopreservation and overnight PBMC resting as this has an effect on CD16 expression and NK activation (Mata, Mahmood et al. 2014). Taken together, these factors could explain the source of the variability for the ADCC potency observed for Sample C by the different laboratories. Although limited data was gathered, cell binding activity results show similar and consistent data between the labs for both preparations (Samples A and C). A slightly lower activity was observed for Sample C relative to the candidate preparation (Sample A) by the one laboratory performing an apoptosis assay. Stability studies over 10 months indicated that the candidate preparation has not suffered any significant loss of activity supporting its suitability as a standard for bioassays. However, further accelerated degradation and real time stability studies are under way and will be used to predict and monitor potential loss of activity in time. No reference method to test the bioactivity of rituximab exists, thus typically, a variety of methods are used to assign a value to an international standard, and therefore the definition of unit is not dependent on a specific method of determination (WHO Technical report series, 932, 2006). In the light of the present collaborative study, the arbitrary values of 1,000 IU of CDC activity per ampoule, 1,000 IU of ADCC activity per ampoule, 1,000 IU of cell-binding activity per ampoule and 1,000 IU of apoptotic activity per ampoule are proposed for the candidate preparation (NIBSC code 14/210) to serve as the 1st International Standard for performance characterisation, calibration and validation of bioassays for rituximab. It is recognised that different rituximab preparations might differ in their bioactivities ratio. Therefore, in view of the need for replacement of the 1st IS in the future, it is proposed that independent unitage values are assigned for the 4 independent bioactivities assessed, with the caveat that only a limited number of laboratories performed the bioassay in the case of cellbinding and apoptosis in the current collaborative study. It is important to stress that the content of rituximab per ampoule is nominal and no declared mass content for the proposed preparation is given. Thus, the proposed unitage does not define the specific activity of the preparation and it is not intended to be used to derive an specific activity or to infer an specific activity for regulatory purposes. Further, the properties and characteristics of the material used to make this reference standard, such as purity and specific biological activity (U/mg) are not intended to serve any regulatory role in defining biosimilarity, and should not be extrapolated, calculated or inferred as serving this purpose in anyway whatsoever. Neither the unitage of the proposed IS is intended to form basis of revised labelling nor dosing requirements as these are the remit and decision of the competent authorities only. Therefore, the proposed 1st International Standard for the bioactivities of

WHO/BS/2017.2309 Page 21 rituximab serves a different role to that of the reference product. The IS provides a higher order reference standard for bioassays facilitating rituximab bioactivity data harmonisation amongst stakeholders, globally and across the life-cycle of rituximab products.

Conclusions Based on the results of the multi-centre collaborative study, the rituximab candidate preparation NIBSC Code 14/210 is deemed suitable to serve as the 1st WHO International Standard for the in vitro biological activities of rituximab and as such the establishment of this standard will be proposed to the ECBS in the next annual meeting in October 2017. The assignment of the following values will be proposed: 1,000 IU of CDC activity per ampoule 1,000 IU of ADCC activity per ampoule 1,000 IU of cell-binding activity per ampoule 1,000 IU of apoptotic activity per ampoule The data suggests that the use of the candidate rituximab IS for bioassays may assist stakeholders in the calibration and validation of bioactivity assays. In addition, the IS will help to harmonise reported rituximab potency values between laboratories and consequently we expect this will have a positive impact on new product development and on the consistency of rituximab products pre- and post-marketing.

Acknowledgements We are grateful to Sandoz GmbH (Austria) for kindly donating the material to develop the candidate preparation. We are also thankful to the sixteen participating laboratories that contributed to the study performing the bioassays and providing helpful discussions and comments. We also thank Paul Matejtschuk and Kiran Malik for their assistance in the pilot fills and process optimisation and staff in Standards Processing Division (SPD) for the preparation of the candidate fill and in the dispatching the study preparations. Finally we would like to thank Adrian Bristow, Chris Burns, Susan Thorpe and Meenu Wadhwa for their support and helpful discussions.

References European Medicines Agency, MabThera: EPAR- Product Information. 2016. Available at: http://www.ema.europa.eu/docs/en_GB/document_library/EPAR__Product_Information/human/000165/WC500025821.pdf European Medicines Agency, 2005. MabThera: EPAR—Scientific Discussion. Available at: http://www.ema.europa.eu/docs/en GB/document library/EPAR ScientificDiscussion/human/000165/WC500025817.pdf European Medicines Agency, 2012. Guideline on similar biological medicinal products containing biotechnology-derived proteins as active substance: quality issues (revision 1).

WHO/BS/2017.2309 Page 22 Available at: http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2012/05/WC5 00127960.pdf DrugBank DB00073. Available at: https://www.drugbank.ca/drugs/DB00073 WHO Model List of Essential Medicines, 19th List, April 2015. Available at: http://www.who.int/medicines/publications/essentialmedicines/EML_2015_FINAL_amended _NOV2015.pdf?ua=1 GaBI Online - Generics and Biosimilars Initiative. Rituximab Biosimilars [cited 2016 Sep 9]. Available at: http://www.gabionline.net/Biosimilars/General/Biosimilars-of-rituximab Derbyshire M. Patent expiry dates for best-selling biologicals. Generics and Biosimilars Initiative Journal (GaBI Journal). 2015;4(4):178-9. doi:10.5639/gabij.2015.0404.040 WHO Expert Committee on Biological Standarisation, 2007. Fixty sixth Report. Global needs in standardization of products derived by biotechnology. WHO Technical Report Series 941: 12-13. WHO Expert Committee on Biological Standarisation, 2016. Sixty sixth Report. Report of a WHO informal consulation on international standards for biotherapeutic products. WHO Technical Report Series 999:13-15. WHO Expert Committee on Biological Standarisation, 2006. Fifty fifth Report. Recommendations for the preparation, characterization and establishment of international and other biological reference standards. WHO Technical Report Series 932:73-130. FDA, 2016. Guidance for Industry. Biosimilars: Clinical Pharmacology Data to Support a Demonstration of Biosimilarity to a Reference Product. European Medicines Agency, 2012. Guideline on similar biological medicinal products containing monoclonal antibodies – non-clinical and clinical issues. EMA/CHMP/BMWP/403543/2010 R Core Team (2015). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. Cheng, Z. J., D. Garvin, A. Paguio, R. Moravec, L. Engel, F. Fan and T. Surowy (2014). "Development of a robust reporter-based ADCC assay with frozen, thaw-and-use cells to measure Fc effector function of therapeutic antibodies." J Immunol Methods 414: 69-81. Chung, S., Y. L. Lin, C. Reed, C. Ng, Z. J. Cheng, F. Malavasi, J. Yang, V. Quarmby and A. Song (2014). "Characterization of in vitro antibody-dependent cell-mediated cytotoxicity activity of therapeutic antibodies - impact of effector cells." J Immunol Methods 407: 63-75. Dotan, E., C. Aggarwal and M. R. Smith (2010). "Impact of Rituximab (Rituxan) on the Treatment of B-Cell Non-Hodgkin's Lymphoma." P T 35(3): 148-157. Forstpointner, R., M. Unterhalt, M. Dreyling, H. P. Bock, R. Repp, H. Wandt, C. Pott, J. F. Seymour, B. Metzner, A. Hanel, T. Lehmann, F. Hartmann, H. Einsele, W. Hiddemann and G. German Low Grade Lymphoma Study (2006). "Maintenance therapy with rituximab leads

WHO/BS/2017.2309 Page 23 to a significant prolongation of response duration after salvage therapy with a combination of rituximab, fludarabine, cyclophosphamide, and mitoxantrone (R-FCM) in patients with recurring and refractory follicular and mantle cell lymphomas: Results of a prospective randomized study of the German Low Grade Lymphoma Study Group (GLSG)." Blood 108(13): 4003-4008. Glennie, M. J., R. R. French, M. S. Cragg and R. P. Taylor (2007). "Mechanisms of killing by anti-CD20 monoclonal antibodies." Mol. Immunol 44(16): 3823-3837. Hatjiharissi, E., L. Xu, D. D. Santos, Z. R. Hunter, B. T. Ciccarelli, S. Verselis, M. Modica, Y. Cao, R. J. Manning, X. Leleu, E. A. Dimmock, A. Kortsaris, C. Mitsiades, K. C. Anderson, E. A. Fox and S. P. Treon (2007). "Increased natural killer cell expression of CD16, augmented binding and ADCC activity to rituximab among individuals expressing the Fc{gamma}RIIIa-158 V/V and V/F polymorphism." Blood 110(7): 2561-2564. Hiddemann, W., M. Kneba, M. Dreyling, N. Schmitz, E. Lengfelder, R. Schmits, M. Reiser, B. Metzner, H. Harder, S. Hegewisch-Becker, T. Fischer, M. Kropff, H. E. Reis, M. Freund, B. Wormann, R. Fuchs, M. Planker, J. Schimke, H. Eimermacher, L. Trumper, A. Aldaoud, R. Parwaresch and M. Unterhalt (2005). "Frontline therapy with rituximab added to the combination of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) significantly improves the outcome for patients with advanced-stage follicular lymphoma compared with therapy with CHOP alone: results of a prospective randomized study of the German Low-Grade Lymphoma Study Group." Blood 106(12): 3725-3732. Hsieh, Y. T., P. Aggarwal, D. Cirelli, L. Gu, T. Surowy and N. M. Mozier (2016). "Characterization of FcgammaRIIIA effector cells used in in vitro ADCC bioassay: Comparison of primary NK cells with engineered NK-92 and Jurkat T cells." J Immunol Methods. Iida, S., H. Misaka, M. Inoue, M. Shibata, R. Nakano, N. Yamane-Ohnuki, M. Wakitani, K. Yano, K. Shitara and M. Satoh (2006). "Nonfucosylated therapeutic IgG1 antibody can evade the inhibitory effect of serum immunoglobulin G on antibody-dependent cellular cytotoxicity through its high binding to FcgammaRIIIa." Clin Cancer Res 12(9): 2879-2887. Jefferis, R. (2009). "Recombinant antibody therapeutics: the impact of glycosylation on mechanisms of action." Trends Pharmacol Sci 30(7): 356-362. Kellner, C., S. Derer, T. Valerius and M. Peipp (2014). "Boosting ADCC and CDC activity by Fc engineering and evaluation of antibody effector functions." Methods 65(1): 105-113. Kirkwood, T. B. (1977). "Predicting the stability of biological standards and products." Biometrics 33(4): 736-742. Mata, M. M., F. Mahmood, R. T. Sowell and L. L. Baum (2014). "Effects of cryopreservation on effector cells for antibody dependent cell-mediated cytotoxicity (ADCC) and natural killer (NK) cell activity in (51)Cr-release and CD107a assays." J Immunol Methods 406: 1-9. Nikolov, N. P. and M. A. Shapiro (2017). "An FDA perspective on the assessment of proposed biosimilar therapeutic proteins in rheumatology." Nat Rev Rheumatol.

WHO/BS/2017.2309 Page 24 Preithner, S., S. Elm, S. Lippold, M. Locher, A. Wolf, A. J. da Silva, P. A. Baeuerle and N. S. Prang (2006). "High concentrations of therapeutic IgG1 antibodies are needed to compensate for inhibition of antibody-dependent cellular cytotoxicity by excess endogenous immunoglobulin G." Mol Immunol 43(8): 1183-1193. Ramanan, S. and G. Grampp (2014). "Drift, evolution, and divergence in biologics and biosimilars manufacturing." BioDrugs 28(4): 363-372. Sanz, I. (2009). "Indications of rituximab in autoimmune diseases." Drug Discov Today Ther Strateg 6(1): 13-19. Schiestl, M., T. Stangler, C. Torella, T. Cepeljnik, H. Toll and R. Grau (2011). "Acceptable changes in quality attributes of glycosylated biopharmaceuticals." Nat Biotechnol 29(4): 310312. Smith, M. R. (2003). "Rituximab (monoclonal anti-CD20 antibody): mechanisms of action and resistance." Oncogene 22(47): 7359-7368. Thorpe, R. and M. Wadhwa (2011). "Intended use of reference products & WHO International Standards/Reference Reagents in the development of similar biological products (biosimilars)." Biologicals 39(5): 262-265. Udpa, N. and R. P. Million (2016). "Monoclonal antibody biosimilars." Nat Rev Drug Discov 15(1): 13-14. Wadhwa, M., H. N. Kang, I. Knezevic, R. Thorpe and E. Griffiths (2011). "WHO/KFDA joint workshop on implementing WHO guidelines on evaluating similar biotherapeutic products, Seoul, Republic of Korea 24-26 August, 2010." Biologicals 39(5): 349-357.

WHO/BS/2017.2309 Page 25

Appendix 1 Table A1. Equivalence bounds for α, β and δ parameters used as assay validity criteria in this study and overall percentage of invalid assays when ratios were calculated relative to the candidate preparation (Sample A) or in house reference preparation (IH Ref). Note that an assay is invalid for a given sample comparison (ratio) if it does not meet the upper equivalence bounds for one or more of the parameters (α, β and δ). % of invalid assays vs Candidate (Sample A) Coded duplicate (Sample B) % of invalid assays vs IH Ref Sample C

Upper equivalence bound Bioactivity Number of laboratories Upper asymptote (α) Asymptote difference (δ) Slope factor ()

Sample C

IH Ref

CDC ADCC Binding

16 11 5

1.12 1.22 1.15

1.18 1.36 1.22

1.74 2.25 1.74

17 (n=137) 18 (n=101) 15 (n=27)

20 (n=137) 25 (n=101) 30 (n=27)

20 (n=122) 27 (n=81) 25 (n=24)

21 15 25

Table A2. Percentage of invalid CDC assays when ratios were calculated relative to candidate preparation (Sample A) or IH reference % of invalid assays vs candidate (Sample A) Coded IH Ref duplicate Sample C (Sample B) 0 11 11 Not available 0 Not available 0 67 22 11 11 78 0 78 0 Not available 0 0 67 0 0 0 33 22 11 0 44 11 56 0 11 0 13 67 0 0 0 0 22 22 0 78 11 67 0 22 0 0 0 0 0 67 22 11 11 78 0 78 0 0 % of invalid assays vs IH Ref Coded Candidate duplicate Sample C (Sample A) (Sample B) 0 11 11 Not available 0 Not available 0 83 0 22 0 89 0 78 0 Not available 0 0 0 0 0 33 11 11 11 100 11 89 0 0

Laboratory 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

WHO/BS/2017.2309 Page 26 Table A3. Percentage of invalid ADCC assays when ratios were calculated relative to candidate preparation (Sample A) or IH reference % of invalid assays vs candidate (Sample A) Laboratory IH Ref 1 6 7 8 10 11 12 13 14 15 16 44 33 17 44 0 Not available 78 0 Not available 11 Not available Coded duplicate (Sample B) 0 11 0 0 0 22 33 0 63 22 42 Sample C 22 22 33 0 0 22 44 0 50 22 50 % of invalid assays vs IH ref Coded duplicate Sample C (Sample B) 44 33 67 22 0 22 44 0 11 0 Not available 89 33 0 0 Not available 0 11 Not available

Table A4. Percentage of invalid cell-binding assays when ratios were calculated relative to candidate preparation (Sample A) or IH reference % of invalid assays vs candidate (Sample A) Coded IH Ref duplicate Sample C (Sample B) 22 0 11 0 0 0 44 44 56 Not available 0 67 0 0 0 % of invalid assays vs IH ref Coded Candidate duplicate Sample C (Sample A) (Sample B) 22 33 33 0 33 0 44 67 33 Not available 0 0 0

Laboratory 6 8 10 12 16

WHO/BS/2017.2309 Page 27 Table A5. Percentage of invalid assays per parameter and bioactivity % Invalid due to Upper Asymptote Bioactivity CDC ADCC Binding Coded Duplicate (Sample B) 10 11 11 Sample C 12 13 19 IH Ref 9 22 17 % Invalid due to Asymptote difference Coded IH Duplicate Sample C Ref (Sample B) 10 13 14 11 11 20 19 16 21 % Invalid due to Slope factor Coded Duplicate (Sample B) 10 11 11 Sample C 10 4 4 IH Ref 13 6 21 Coded Duplicate (Sample B) 137 101 27 n Sample C 137 101 27 IH Ref 122 81 24

WHO/BS/2016.2309 Page 28 Figure A1. Equivalence of model parameters relative to the candidate preparation (Sample A) in CDC assays.

WHO/BS/2016.2309 Page 29 Figure A2. Equivalence of model parameters relative to the candidate preparation (Sample A) in ADCC assays.

WHO/BS/2016.2309 Page 30 Figure A3. Equivalence of model parameters relative to the candidate preparation (Sample A) in binding assays.

WHO/BS/2016.2309 Page 31 Figure A4. Extreme values (maximum of upper 90% confidence limit and reciprocal of lower 90% confidence limit) for the ratio of parameter estimates between sample preparations (IH reference versus candidate (Sample A), Coded duplicate (Sample B) versus candidate and Sample C versus candidate) in CDC assays. The line represents the calculated upper equivalence bound at 1.18 (δ), 1.74 (β) and 1.12 (α). Note that same colour symbol per sample comparison (in rows) denotes same assay plate. CDC IH r e f vs C a n d id a te (S a m p le A ) D iffe r e n c e in A s y m p to te (  )

CDC IH r e f vs C a n d id a te (S a m p le A ) S lo p e fa c to r (  ) 1 .5

CDC IH r e f vs C a n d id a te (S a m p le A ) U p p e r lim it (  )

2

3

E x t r e m e v a lu e

E x t r e m e v a lu e

1 .5

2

E x t r e m e v a lu e 1 5 6 8 0 1 3 7 9 2 3 5 6

1 .2 5

1 .2 5

1 .5 1 .2 5

1

1

1

1

6

7

8

9

0

1

2

3

5

3

5

6

1

5

6

7

0

3

8

9

1

2

3

5 1 4 1 1 5

0

0

0

0

0

0

0

1

1

1

1

1

1

0

0

0

0

0

0

0

1

1

1

1

1

1

0

0

0

0

0

0

0

1

1

1

1

L a b o ra to ry

L a b o ra to ry

L a b o ra to ry

CDC C a n d id a te (S a m p le A ) vs C o d e d d u p lic a te (S a m p le B ) D iffe r e n c e in A s y m p to te (  )

CDC C a n d id a te (S a m p le A ) vs C o d e d d u p lic a te (S a m p le B ) S lo p e fa c to r (  ) 1 .5

CDC C a n d id a te (S a m p le A ) vs C o d e d d u p lic a te (S a m p le B ) U p p e r lim it (  )

2

3

E x t r e m e v a lu e

E x t r e m e v a lu e

2

E x t r e m e v a lu e

1 .5

1 .2 5

1 .5 1 .2 5

1 .2 5

1

1

1

1

2

4

6

7

8

9

0

1

2

4

3

5

3

1

2

4

6

7

8

9

0

1

2

4

6

5

6

1

2

4

6

7

8

9

0

1

2

1

3

5

3

5

3

5

0

0

0

0

0

0

0

0

0

1

1

1

1

1

1

0

0

0

0

0

0

0

0

0

1

1

1

1

1

1

1

1

0

0

0

0

0

0

0

0

0

1

1

1

1

3

L a b o ra to ry

L a b o ra to ry

L a b o ra to ry

CDC S a m p le C vs C a n d id a te (S a m p le A ) D iffe r e n c e in A s y m p to te (  )

CDC S a m p le C vs C a n d id a te (S a m p le A ) S lo p e fa c to r (  ) 1 .5

CDC S a m p le C vs C a n d id a te (S a m p le A ) U p p e r lim it (  )

2

3

E x t r e m e v a lu e

E x t r e m e v a lu e

1 .5

2

1 .2 5

1 .5 1 .2 5

1

1

E x t r e m e v a lu e

1 .2 5

1

1

2

4

5

6

8

9

0

1

2

3

4

3

7

5

6

1

2

3

4

5

7

8

9

0

1

2

4

6

2

3

4

5

7

8

9

0

1

2

4

6

3

5

1

6

3

0

0

0

0

0

0

0

0

0

1

1

1

1

1

1

1

0

0

0

0

0

0

0

0

0

1

1

1

1

1

1

1

0

0

0

0

0

0

0

0

0

1

1

1

1

1

1

5

L a b o ra to ry

L a b o ra to ry

L a b o ra to ry

1

6

1

6

6

WHO/BS/2016.2309 Page 32 Figure A5. Extreme values (maximum of upper 90% confidence limit and reciprocal of lower 90% confidence limit) for the ratio of parameter estimates between sample preparations (IH reference versus candidate (Sample A), Coded duplicate (Sample B) versus candidate and Sample C versus candidate) in ADCC assays. The line represents the calculated upper equivalence bound at 1.36 (δ), 2.25 (β) and 1.22 (α). Note that same colour symbol per sample comparison (in rows) denotes same assay plate. ADCC IH r e f vs C a n d id a te (S a m p le A ) D iffe r e n c e in A s y m p to te (  ) 6 5 4

ADCC IH r e f vs C a n d id a te (S a m p le A ) S lo p e fa c to r (  ) 5 4

ADCC IH r e f vs C a n d id a te (S a m p le A ) U p p e r lim it (  ) 4

3

E x t r e m e v a lu e

E x t r e m e v a lu e

3

3

E x t r e m e v a lu e 7 8 6 1 2 0 3 5

2

2

2

1 .5

1 .5 1 .2 5 1

1 .2 5 1

1

1

7

8

6

2

0

3

5

7

8

6

1

2

0

3

0

0

0

1

0

1

1

1

0

0

0

1

0

1

1

1

0

0

0

1

0

1

1

R e p o rte r

K illin g a s s a y ( N K c e ll lin e )

K illin g a s s a y (P B M C )

R e p o rte r

K illin g a s s a y ( N K c e ll lin e )

K illin g a s s a y (P B M C )

R e p o rte r

K illin g a s s a y ( N K c e ll lin e )

K illin g a s s a y (P B M C )

L a b o ra to ry

L a b o ra to ry

L a b o ra to ry

ADCC C a n d id a te (S a m p le A ) vs C o d e d d u p lic a te (S a m p le B ) D iffe r e n c e in A s y m p to te (  ) 6 5

ADCC C a n d id a te (S a m p le A ) vs C o d e d d u p lic a te (S a m p le B ) S lo p e fa c to r (  ) 5 4 4

ADCC C a n d id a te (S a m p le A ) vs C o d e d d u p lic a te (S a m p le B ) U p p e r lim it (  )

4

3

E x t r e m e v a lu e

E x t r e m e v a lu e

3

3

E x t r e m e v a lu e

2

2

2

1 .5

1 .5 1 .2 5 1

1

1

1

7

8

*

6

0

1

2

3

4

5

*

7

8

2

0

3

1

4

7

8

2

*

0

3

1

1

6

5

1

6

4 1

1 (P B M C )

5

6

6

0

0

0

1

0

1

1

1

1

1

0

0

0

1

0

1

1

1

1

1

0

0

0

1

6

0

1

1

1

1

1

R e p o rte r

K illin g a s s a y ( N K c e ll lin e )

K illin g a s s a y (P B M C )

R e p o rte r

K illin g a s s a y ( N K c e ll lin e )

K illin g a s s a y (P B M C )

R e p o rte r

1

K illin g a s s a y ( N K c e ll lin e )

K illin g a s s a y

L a b o ra to ry

L a b o ra to ry

L a b o ra to ry

ADCC S a m p le C vs C a n d id a te (S a m p le A ) D iffe r e n c e in A s y m p to te (  ) 6 5

ADCC S a m p le C vs C a n d id a te (S a m p le A ) S lo p e fa c to r (  ) 5 4 4

ADCC S a m p le C vs C a n d id a te (S a m p le A ) U p p e r lim it (  )

4

3

E x t r e m e v a lu e

E x t r e m e v a lu e

3

3

E x t r e m e v a lu e

2

2

2

1 .5

1 .5 1 .2 5 1

1 .2 5 1 1

1

7

8

2

*

0

3

6

1

4

5

*

1

7

8

2

6

0

3

4

1

7

8

2

*

0

3

1

5

6

1

4

6

6

0

0

0

1

0

1

1

1

1

1

0

0

0

1

0

1

1

1

1

1

0

0

0

1

6

0

1

1

1

1

1

1

R e p o rte r

K illin g a s s a y ( N K c e ll lin e )

K illin g a s s a y (P B M C )

R e p o rte r

K illin g a s s a y ( N K c e ll lin e )

K illin g a s s a y (P B M C )

R e p o rte r

1

K illin g a s s a y ( N K c e ll lin e )

K illin g a s s a y (P B M C )

L a b o ra to ry

L a b o ra to ry

L a b o ra to ry

1

5

1

5

WHO/BS/2016.2309 Page 33 Figure A6. Extreme values (maximum of upper 90% confidence limit and reciprocal of lower 90% confidence limit) for the ratio of parameter estimates between sample preparations (IH reference versus candidate (Sample A), Coded duplicate (Sample B) versus candidate and Sample C versus candidate) in binding assays. The line represents the calculated upper equivalence bound at 1.22 (δ), 1.74 (β) and 1.15 (α).Note that same colour symbol per sample comparison (in rows) denotes same assay plate. C e ll B in d in g IH r e f vs C a n d id a te (S a m p le A ) D iffe r e n c e in A s y m p to te (  ) 2

C e ll B in d in g IH r e f vs C a n d id a te (S a m p le A ) S lo p e fa c to r (  ) 3

C e ll B in d in g IH r e f vs C a n d id a te (S a m p le A ) U p p e r lim it (  ) 2

E x t r e m e v a lu e

E x t r e m e v a lu e

1 .5

2

E x t r e m e v a lu e 6 8 0 6

1 .5

1 .2 5

1 .5

1 .2 5

1 .2 5

1

1 6 8 0 6 0 0 1 1

1

6

8

0

0

0

1

1

0

0

1

L a b o ra to ry

L a b o ra to ry

L a b o ra to ry

C e ll B in d in g C a n d id a te (S a m p le A ) vs C o d e d d u p lic a te (S a m p le B ) D iffe r e n c e in A s y m p to te (  ) 2 3

C e ll B in d in g C a n d id a te (S a m p le A ) vs C o d e d d u p lic a te (S a m p le B ) S lo p e fa c to r (  ) 2

C e ll B in d in g C a n d id a te (S a m p le A ) vs C o d e d d u p lic a te (S a m p le B ) U p p e r lim it (  )

E x t r e m e v a lu e

E x t r e m e v a lu e

2

1 .5

E x t r e m e v a lu e

1 .5

1 .5

1 .2 5

1 .2 5

1 .2 5

1

1 6 8 0 2

1 6

6

8

2

6

8

0

6

0

2

1

6

0

0

1

1

0

0

1

1

1

1

0

0

1

1

L a b o ra to ry

L a b o ra to ry

L a b o ra to ry

C e ll B in d in g S a m p le C vs C a n d id a te (S a m p le A ) D iffe r e n c e in A s y m p to te (  ) 2 3

C e ll B in d in g S a m p le C vs C a n d id a te (S a m p le A ) S lo p e fa c to r (  ) 2

C e ll B in d in g S a m p le C vs C a n d id a te (S a m p le A ) U p p e r lim it (  )

E x t r e m e v a lu e

E x t r e m e v a lu e

2

1 .5

E x t r e m e v a lu e

1 .5

1 .5

1 .2 5

1 .2 5

1 .2 5

1

1

1

6

8

2

6

8

0

6

0

2

6

8

0

2

0

0

1

1

1

6

0

0

1

1

0

0

1

1

L a b o ra to ry

1

L a b o ra to ry

L a b o ra to ry

1

6

1

6

WHO/BS/2016.2309 Page 34

Appendix 3 Below comments by one participant. NIBSC: Do you agree that your data has been correctly reported? Participant’s comment: With the current description of the calculations, we are not able to verify the data from DRL. Hence, kindly provide the exact calculations performed to arrive to Upper equivalence bound and % of invalid assays. NIBSC’s discussion: In this study we assess parallelism between dose-response curves by demonstrating the equivalence of three parameters that define the curves (α, β and δ). The upper equivalence bounds for the 3 parameters (α, β and δ) were calculated using the complete data set from all participants for each assay method (including yours). Figure A1, A2 and A3 shows the equivalence of model parameters relative to the candidate preparation (Sample A) using the complete data set. These parameters are expected to be equivalent when the same sample is compared (Sample A and B) and the equivalence bounds correspond to values where 10% of assays (data from all labs) were above these values. The calculations to determine the equivalence bounds require the data from all participating laboratories. Furthermore it should be noted that the assay validity criteria defined by these calculated equivalence bounds are only applicable to the analysis of data from this study.

NIBSC: Do you agree with the proposed unitage for the candidate preparation NIBSC 14/210 (1000 IU/ampoule of CDC activity, 1000 IU/ampoule of ADCC activity, 1000 IU/ampoule of cell-binding activity and 1000 IU/ampoule of apoptotic activity)? Participant’s comment: It is very arbitrary and no rationale is provided for it. Moreover, it is same for all the assays. Rituximab is sold as 100 mg and 500 mg presentation, so does the IU would change with each presentation accounting for different protein amount per ampoule? NIBSC’s discussion: International Units for a 1st WHO International Standard are always set as arbitrary units (WHO Technical report series 932: 73-130). Unitage for replacement standards upon depletion (i.e. 2nd IS) will be calibrated against the 1st IS. For simplicity we chose 1000 IU for each of the 4 independent bioactivities tested for the 1st WHO RTX IS, but it is likely that different preparations may have different bioactivity ratios. The IS is not intended to change labelling or dosing of Rituximab products. Labelling and dosing requirements are set by the competent authorities and labelling of Rituximab products is expressed in mass of Rituximab protein per vial as you pointed out.

NIBSC: Please include additional comments where necessary: Participant’s comment: Page 6: Nominal concentration will impact potency, it is ideal to determine mass content to use as reference standard

WHO/BS/2016.2309 Page 35 NIBSC’s discussion: It is not the intended use of the IS to define specific activity (IU/mg) but IU of bioactivity per ampoule. The nominal value shown is not a declared mass content for the candidate preparation. Participant’s comment:In Table 6, 7 8, it is good to include sample B data against in-house reference standard, as data with sample A is limited. NIBSC’s discussion: We provided Samples A, B and C, however, not all of the laboratories included an in-house reference. Thus “not available” is shown in the tables when that is the case. The data for sample B would also be as limited as for sample A. Participant’s comment :Appendix 1 –Table A1. Upper equivalence bound limits are derived from average of 16 labs. Question: what is the variability observed between labs, what is the number of analyses considered from each lab? How can we verify the data? NIBSC’s discussion: All of the data for all of the labs for each independent assay (16 labs for CDC, 11 labs for ADCC and 5 labs for Binding assay) was included (see “n” in table for number of assays included). The variability of the data for each individual parameter, lab, assay and comparison is shown in Figures A4, A5 and A6. Participant’s comment: Appendix 1 Table A1. % invalids for sample A &B against IH Ref data is not shown NIBSC’s discussion: The table shows % of invalid assays for IH Ref vs Sample A (second column from the right), which is the same comparison as A against IH Ref. Sample B is the coded duplicate for Sample A and thus the % of invalid assays for IH Ref vs Sample B is not shown. Participant’s comment: Appendix 1 Table A1, A2,A3. % invalids are high for individual labs compared to total invalids, upper equivalence bound limits calculation is not clear. NIBSC’s discussion: Yes, in some cases a high number of invalid assays were observed for some labs, assays and given comparison. Please refer to Figures A4, A5 and A6 for detailed plots. Explanations regarding upper equivalence bound see above. Participant’s comment: Table 9 and 10 what is ‘n’ (number of labs or number of potency estimates)? Why is the ‘n’ different in each Table? NIBSC’s discussion: ‘n’ is the number of labs used to determine the overall GM potency estimate. Note that not all of the labs included an IH ref and thus the different “n” in Table 9 and 10. (Table 4 shows whether IH ref was available per lab and assay)

WHO/BS/2016.2309 Page 36 Participant’s comment: Percentage of invalid assays for ratios in CDC and ADCC assays were observed to be high (Table A2 & A3). In discussion section (page 18), it was concluded 70-85% of the assays are valid. It is not correlating. NIBSC’s discussion: Please refer to Table A5. The percentage “70-85 % of the assays” corresponds to overall data i.e. all labs. As you point out, individually, as shown in tables A2 and A3, some labs showed a higher number of invalid assays than others in the study. Overall 7085% of the assays reported were valid. Participant’s comment :Based on the Figure 1 there are certain unanswered questions: (a) The relative potency of CDC for Sample A and Sample C against in-house reference has positive bias whereas binding of sample A has negative bias. Why would this happen? This indicates structural determinants play an important role. (2) The ADCC data using sample A as a reference has a positive bias ranging up to 150% with Sample C. Does this mean that we are accepting this level of variability? Overall this sentence “the candidate preparation is suitable to serve as an international standard to help harmonise potency data between laboratories in line with the intended role” in Discussion (on page number 19) is not supporting the data that have been obtained. At most it is only applicable for CDC assays and not for ADCC assays, where the different laboratories have shown much higher potency values and variability. NIBSC’s discussion: a) The Figure summarises the potency estimates calculated using IH reference as standard or the Candidate Sample A as Standard, also shown numerically in Tables 9 and 10. Yes, the GM potency estimates relative to IH ref are skewed >1 or <1. These may be because potential structural differences between Samples A and C that have an impact on its bioactivity but also because of the use of different IH ref standards. Using a common ref standard (Candidate IS, Sample A) helps to harmonise the report of bioactivity data and thus may help to elucidate potential differences in sample bioactivity (as the report of the bioactivity is not biased by the use of different IH ref amongst labs). b) The GM potency estimate of Sample C versus Sample A is 1.28 (1.16, 1.41) for ADCC. This data shows that Sample C has a higher ADCC bioactivity than Sample A, and the GCV shows the associated variability to this estimate is wider than for other assays (i.e. CDC). Find discussion of this observation on pages 19 and 20: “The GM relative potency for Sample C in CDC assays was 0.99 (GCV 5%) showing a similar potency to that of the candidate standard, but the ADCC potency estimate of Sample C was significantly greater than 1 (GM= 1.28) and had higher associated variability (GCV 16%). This variability was greater than that observed for the coded duplicate (Sample B) (GCV 6%) and was not related to the ADCC assay platform used. We can hypothesise that these results may reflect differences between the two rituximab preparations (Samples A and C) that affect specifically their ADCC potency and were identified by the different labs and ADCC assays to a different extent. Published reports ….” Despite the variability of the potency estimate for ADCC for Sample C vs Sample A being 16% (GCV), the use of a common preparation (Candidate, Sample A) reduces significantly the variability of the data as compared to using IH ref as standard (GCV 51%). Despite the complexity of the ADCC assay and the variability observed, as acknowledged through the

WHO/BS/2016.2309 Page 37 discussion, the data from the collaborative study shows that the Candidate helps harmonising the reporting of potency data not only for CDC but also for ADCC. Participant’s comment: The Conclusion on page 21, “In addition, the IS will help to harmonise reported rituximab potency values between laboratories and consequently we expect this will have a positive impact on new product development and the consistency of rituximab products pre- and post-marketing.” What are the glycan values of sample A and sample C with respect to total afucosylation (High Mannose and Afucose)? When sample A is used as a reference in the ADCC assay, the potency values for sample C are upto 150%. So does this mean it is acceptable? If not, then how does it become the reference standard? This acceptance may be misleading when the glycan values are different resulting in different ADCC values. NIBSC’s discussion: The candidate WHO IS is intended as a potency standard for bioassay and is not suitable nor intended to be used for physicochemical characterisation. We can hypothesise as mentioned in the discussion, that the differences between Samples A and C found in the ADCC assay could be due to potential differences in the glycan profile of these samples. The study does not aim to evaluate the differences between the samples provided, thus the differences between Sample A and Sample C are irrelevant to the suitability of the candidate. However the variability of the potency estimates between labs when using sample A as a Standard is relevant to evaluate whether the Candidate helps harmonising the reporting of bioactivity data. Here we suggest that the variability observed may be associated to the different ability of the ADCC assays to discriminate the differences between Sample A and Sample C. In the report this variability is acknowledged and discussed together with the complexity of the ADCC assay and different platforms used, but despite of the bigger variability in the ADCC assay, from the data we can conclude that the use of a common RTX Candidate preparation very much improved data harmonisation supporting its suitability.

WHO/BS/2016.2309 Page 38

Appendix 4 1st WHO reference reagent for the biological activities of Rituximab-like monoclonal antibodies COLLABORATIVE STUDY PROTOCOL 1. BACKGROUND Monoclonal antibodies, derived by recombinant DNA technology, comprise the fastest growing group of innovative biotherapeutics with over 35 already approved for use in humans and hundreds more in clinical development. The expiration of patents on many recombinant therapeutic products, including monoclonal antibodies, has resulted in the development of so-called ‘biosimilars’, thereby widening the market for such products and increasing patient accessibility to these biotherapeutics. The WHO have recognised a global need for standardisation of biotechnology products following requests for advice on appropriate control measures to ensure safety, quality and efficacy (WHO Technical Report Series, 56 th Report, 941: 12-13, 2007). Such issues were further discussed in a recent WHO/KFDA workshop and published in a special issue of Biologicals (Biologicals 39, 5:349-357, 2011). The WHO reference reagents for biological activity of therapeutic monoclonal antibodies are intended for the evaluation of bioassay performance including calibration and validation of potency assays ( Biologicals 39, 5:262-265, 2011). Rituximab is one of the first monoclonal antibodies to be approved for human use. Its wide range of indications across both oncology and rheumatology is reflected by Rituximab having the largest market for any monoclonal antibody. Rituximab was one of the first therapeutic monoclonal antibody to lose market exclusivity and several manufacturers have already produced or are developing Rituximab biosimilars that aim to match the innovator Rituximab as closely as possible. Rituximab is a chimeric anti-CD20 IgG1κ monoclonal antibody and its mechanism of action and clinical efficacy in depleting B cells involves a combination of Fc effector functions that include complement dependent cytotoxicity (CDC), antibody dependent cellular cytotoxicity (ADCC) and induction of apoptosis. CDC assays are generally used for potency testing and lot release of Rituximab final products. 2. AIM The aim of the study is to evaluate the suitability of a candidate lyophilised Rituximab monoclonal antibody as the 1st WHO reference reagent for biological activity of Rituximab-like monoclonal antibodies intended for the characterisation, calibration and validation of potency assays. To support this, the bioactivity of the candidate reference reagent and two comparator product preparations will be assessed and compared with participant laboratories in-house standards. The preparations will be evaluated for their CDC activity and, where possible, for its ADCC activity. The assignment of a value in units of biological activity per ampoule of the candidate reference material will be investigated. 3. MATERIALS PROVIDED    6 or 11 ampoules of preparation code A 6 or 11 ampoules of preparation code B 6 or 11 ampoules of preparation code C

WHO/BS/2016.2309 Page 39 The preparations have been lyophilised from a 1 ml solution containing 100μg Rituximab protein in formulation buffer (25mM Tri-Sodium citrate dehydrate, 150mM Sodium Chloride, 1% HSA v/v, pH 6.5). Note that 6 ampoules for each material (Code A, Code B and Code C) are provided for laboratories determining CDC activity only. From those, 2 ampoules are intended for establishing the dilution working range (preliminary studies), 3 ampoules are intended for performing 3 assay runs to test CDC activity on 3 experimental days, and 1 extra ampoule is provided in case of accidental loss. Note that 11 ampoules for each material (Code A, Code B and Code C) are provided for laboratories determining both CDC and ADCC activities. From those, 2 ampoules are intended for establishing the dilution working range for CDC assays (preliminary studies), 3 ampoules are intended for performing 3 assay runs to test CDC activity on 3 experimental days, 2 ampoules are intended for establishing the dilution working range for ADCC assays (preliminary studies), 3 ampoules are intended for performing 3 assay runs to test ADCC activity on 3 experimental days, and 1 extra ampoule is provided in case of accidental loss. All ampoules are stored at -20oC until reconstitution and use. 4. CAUTION These preparations are not for administration to humans. As with all materials of biological origin they should be regarded as potentially hazardous to health. Detailed MSDS can be found in the “Instructions for Use”. The preparations should be used and discarded according to your own laboratories safety procedures. Such safety procedures should include the use of adequate personal protective equipment and avoiding the generation of aerosols. 5. RECONSTITUTION Do not reconstitute until the day of the assay. Tap the ampoule gently to collect the material at the bottom end. Ensure that the disposable ampoule safety breaker provided is pushed down on the stem of the ampoule and against the shoulder of the ampoule body. Hold the body of the ampoule in one hand and the disposable ampoule breaker covering the ampoule stem between the thumb and first finger of the other hand. Apply a bending force to open the ampoule. Care should be taken to avoid cuts and projectile glass fragments that might enter one’s eyes. Take care that no material is lost from the ampoule and no glass falls into the ampoule. Within the ampoule is dry nitrogen gas at slightly less than atmospheric pressure. Reconstitute the ampoule contents with 1.0 ml of sterile distilled/deionized water only on the day of the assay. Allow several minutes for reconstitution with occasional gentle mixing (avoid vortexing). Transfer the reconstituted contents to a low-protein binding sterile polypropylene capped tube. The reconstituted contents of each ampoule are equivalent to a nominal concentration of 100 μg/ml Rituximab.

WHO/BS/2016.2309 Page 40 6. TESTS SUMMARY A test assay should be able to measure Rituximab complement dependent cytotoxicity (CDC) activity. When possible, an additional test assay should be able to measure Rituximab antibody dependent cellular cytotoxicity (ADCC). For each bioactivity test (CDC or ADCC), reconstitute one ampoule of each preparation (A, B and C) per assay run. An assay run will be defined here as the combination of 3 plates tested on one day using at least two independent dilution series of each preparation per plate and adequate randomisation of the plate layout. Note that fresh independent dilution series for each plate should be prepared. It is requested from the participating laboratories to perform 3 assay runs in total on 3 days, one run per day, using fresh ampoules for each day and for each bioactivity assay. Include in-house assay controls and references if available. Assay method(s) and plate layouts are provided only as examples in Appendix 1 and 2 and within data sheet files under “Examples”. Labs are encouraged to use their in house established and validated assays. 6.1 Complement dependent cytotoxicity (CDC) activity

PROCEDURE A test assay should be able to measure the complement dependent cytotoxicity (CDC) activity of Rituximab in a suitable CD20 expressing cell line. A test assay should have been validated or have a defined level of qualification to measure Rituximab CDC activity. A test should include in-house assay controls (i.e. positive and negative controls) and references (in-house Rituximab reference material) if available. All testing materials should be assayed simultaneously within each plate and assay run. Note that each plate represents “one assay” and should include at least two independent dilution sets of each of the reconstituted material. Three plates combined represent an assay run and should be performed per day where preparations are randomised as much as reasonably possible. The assay should be repeated in 3 experimental days (3 assay runs) and include further sample randomisation between days. An example of CDC assay protocol is included in Appendix 1 (outlined only). Labs are encouraged to use their in-house established and validated assays. SAMPLE PREPARATION Note: Aseptic technique is required at all times A suitable dilution scheme for the samples should be established by the collaborator laboratory during preliminary studies using the provided ampoules (2 ampoules). The working dilution range should achieve asymptotes for maximal and minimal activity and two to three-fold dilutions are assumed to be appropriate. A nominal protein concentration targeting the midrange for 0.5-0.05 μg/mL (final concentration in the plate) may be a suitable starting point; however laboratories should establish the appropriate concentration working range in their CDC assay. A suggested dilution scheme for two-fold dilution series and instructions are shown below: Dilute the reconstituted contents of each preparation in assay media using sterile polypropylene containers (i.e. tubes or plate) to prepare the working serial dilutions as follows: First, prepare a ten-fold dilution of the reconstituted ampoule (for example by adding 100 μL of net sample into 900 L of Assay media, label as Dilution #1) and mix gently avoiding vortexing. Similarly, prepare additional serial twofold dilutions of Dilution #1 taking care of using new pipette tips for each dilution step (for example add 500 μL of Dilution #1 into 500 uL of Assay media to prepare Dilution #2 and mix gently avoiding

WHO/BS/2016.2309 Page 41 vortexing; subsequently using a new pipette tip transfer 500 μL of Dilution #2 into 500 L of Assay media to prepare Dilution #3 and mix gently and so on). Make the dilutions in Assay Media using a calibrated pipette. Use a new tip for each dilution step and mix gently avoiding vortexing. Two independent replicate dilutions per plate are prepared. Three plates per day are performed randomising the samples in each plate. The combined 3 plates tested on one day represent one assay run. Include “no antibody” control wells (Assay media only). Please note that an additional ampoule of each preparation has been provided in case of losses. DAY 1 (First run) Prepare adequate dilutions of the reconstituted content of each preparation as established during the preliminary studies. Repeat the testing in 3 independent plates using at least two dilution replicates per plate and randomising the test samples in the plates preferably in a non-clustered fashion and as reasonably possible to account for potential plate and well effects. Prepare each independent assay plate with a fresh dilution series of the reconstituted ampoule. DAY 2 (Second run) Similarly to day 1, prepare adequate dilutions of the reconstituted content of each preparation and repeat the testing in 3 independent plates. Plate layout is randomised in relation to day 1 (first run) to further minimise well positional effects. Prepare each independent assay plate with a fresh dilution series of the reconstituted ampoule. DAY 3 (Third run) Similarly to days 1 and 2, prepare adequate dilutions of the reconstituted content of each preparation and repeat the testing in 3 independent plates. Plate layout is randomised in relation to day 1 and day 2 (first and second runs) to further minimise well positional effects. Prepare each independent assay plate with a fresh dilution series of the reconstituted ampoule. DATA ANALYSIS Analysis of the data should be performed as per the collaborating laboratory’s own protocol with the relative potency of the three preparations, Code A, Code B and Code C calculated relative to Rituximab in-house reference material or relative to preparation Code A if no in-house reference material is available. 6.2 Antibody dependent cytotoxicity (ADCC) activity

PROCEDURE A test assay should be able to measure the antibody dependent cytotoxicity (ADCC) activity of Rituximab using a suitable CD20 expressing target cell line and effector cell assay system. A test assay should have been validated or have a defined level of qualification to measure Rituximab ADCC activity. A test should include in-house assay controls (i.e. positive and negative controls) and references (in-house Rituximab reference material) if available. All testing materials should be assayed simultaneously within each plate and assay run. Note that each plate represents “one assay” and should include at least two independent dilution sets of the reconstituted material. Three plates combined represent an assay run and should be performed per day where preparations are randomised as much as is reasonably possible. The

WHO/BS/2016.2309 Page 42 assay should be repeated in 3 experimental days (3 assay runs) and include further sample randomisation between days. Examples of ADCC assays (reporter and a classical assay using NK effector cells) are included in Appendix 2 (outlined only). Labs are encouraged to use their in-house established and validated assays. SAMPLE PREPARATION Note: Aseptic technique is required at all times A suitable dilution scheme for the samples should be established by the collaborator laboratory during preliminary studies using the provided ampoules (2 ampoules). The working dilution range should achieve asymptotes for maximal and minimal activity and three to five-fold dilutions are assumed to be appropriate. A nominal protein concentration targeting the midrange for 0.1-1 ng/mL (final concentration in the plate) may be a suitable starting point; however laboratories should establish the appropriate concentration working range in their ADCC assay. A suggested dilution scheme for four-fold dilution series and instructions are shown below as an example: Dilute the reconstituted contents of each preparation in assay media using sterile polypropylene containers (i.e. tubes or plate) to prepare the working serial dilutions as follows: First, prepare a ten-fold dilution of the reconstituted ampoule (for example by adding 100 μL of net sample into 900 L of Assay media, label as pre-dilution and mix gently avoiding vortexing. Then, prepare a ten-fold dilution of the pre-dilution (for example by adding 100 μL of pre-dilution into 900 uL of Assay media, label as Dilution #1) and mix gently avoiding vortexing. Similarly, prepare additional serial four-fold dilutions of Dilution #1 taking care of using new pipette tips for each dilution step (for example add 250 μL of Dilution #1 into 750 uL of Assay media to prepare Dilution #2 and mix gently avoiding vortexing; subsequently using a new pipette tip transfer 250 μL of Dilution #2 into 750 uL of Assay media to prepare Dilution #3 and mix gently and so on). Make the dilutions in Assay Media using a calibrated pipette. Use a new tip for each dilution step and mix gently avoiding vortexing. Two independent replicate dilutions per run (plate) are prepared. Three runs per day are performed randomising the samples in each run. Include “no antibody” control wells (Assay media only). Cell-free assay media may be also included. Target cells only control (background spontaneous cell death in the absence of effector cells) may be also included for classical ADCC assays. Maximum target cell killing control (i.e. using lysis buffer) may be also included for classical ADCC assays. Please note that an additional ampoule of each preparation has been provided in case of losses. DAY 1 (First run) Prepare adequate dilutions of the reconstituted content of each preparation as established during the preliminary studies. Repeat the testing in 3 independent plates using at least two dilution replicates per plate and randomising the test samples in the plates preferably in a non-clustered fashion and as reasonably possible to account for potential plate and well effects. Prepare each independent assay plate with a fresh dilution series of the reconstituted ampoule.

WHO/BS/2016.2309 Page 43 DAY 2 (Second run) Similarly to day 1, prepare adequate dilutions of the reconstituted content of each preparation and repeat the testing in 3 independent plates. Plate layout is randomised in relation to day 1 (first run) to further minimise well positional effects. Prepare each independent assay plate with a fresh dilution series of the reconstituted ampoule. DAY 3 (Third run) Similarly to days 1 and 2, prepare adequate dilutions of the reconstituted content of each preparation and repeat the testing in 3 independent plates. Plate layout is randomised in relation to day 1 and day 2 (first and second runs) to further minimise well positional effects. Prepare each independent assay plate with a fresh dilution series of the reconstituted ampoule. DATA ANALYSIS Analysis of the data should be performed as per the collaborator laboratory’s own protocol with the relative potency of the three preparations, Code A, Code B and Code C calculated relative to Rituximab in-house reference material or relative to preparation Code A if no in-house reference material is available. 7. RECORDING RESULTS Data should be recorded using the Excel sheet templates provided. Please provide details of the bioassay procedure, plate design, validity criteria, data analysis and results. Complete accompanying methodology questionnaire (in excel data sheet templates provided). Please supply all raw data to NIBSC for statistical evaluation.

Please return electronic copies of the completed data sheets and questionnaire to Dr Simon Hufton (simon.hufton@nibsc.org) and Dr Sandra Prior (sandra.prior@nibsc.org).

Please return your results as soon as possible by the 6th May 2016 and no later than the 13th May 2016.

Participants in a WHO study do so under the following conditions:  The data should not be published or cited before the formal establishment of the standard by WHO, without the expression permission of the NIBSC Study organizer  It is normal practice to acknowledge participants as contributors of data rather than co-authors in publications describing the establishment of the standard  Individual participants’ data will be coded and reported “blind” to other participants during the reparation of the study report, and also in subsequent publications.  Participants will receive a copy of the report of the study and proposed conclusions and recommendations for comment before it is further distributed.

WHO/BS/2016.2309 Page 44 APPENDIX 1 CDC ASSAY EXAMPLE –Outlined only Equipment and Reagents:

     

Cell Line: WIL2-S (ATCC® CRL-8885) or other suitable target cell line expressing CD20 antigen. Assay Medium for WIL2-S: RPMI (95.7%), 1M HEPES (2%), BSA (1.3%), PenStrep (1%) Complement source: Human Complement (e.g. Quidel Cat No. A112 or equivalent) Assay plate: Sterile white 96-flat bottom cell culture plates (e.g. Nunc Cat No. 136101 or equivalent) Detection reagent: Alamar Blue Cell Viability Reagent (e.g. Thermo Scientific Cat No. 88952 or equivalent) Fluorescence Plate Reader e.g., Spectra Max M5.

Procedure: WIL2-S cells are maintained as per ATCC instructions and kept in culture for at least 2 days prior to the assay. Cells at log growth phase are harvested, washed by centrifugation and resuspended in assay medium at 1x106 cells/ml. Cell counts and viability are measured by Trypan blue exclusion in an automated cell counter (Countess™, Thermo Fisher). Nine independent concentrations of test antibody and in-house reference material in assay media are tested in two independent replicate dilution series per plate. Fifty µl of antibody are added per well into the flat bottom tissue culture 96-well plates and then 50 µl human complement are added. For the assay, complement is thawed and two-fold diluted in cold assay media and stored on ice or 2-8 ˚C until use within 1 hour. No antibody control wells are also prepared (assay media only). Subsequently, 50µl of the cell suspension is added to all wells and the plates are gently agitated on an orbital shaker at room temperature for 2 min. Plates are incubated for 1-2 hours at 37°C, 5% CO2 under a humid atmosphere followed by the addition of 50µl/well of Alamar Blue cell viability reagent and brief agitation as described before. The plates are then incubated for an additional 18 hours at 37°C, 5% CO2 under a humid atmosphere. Fluorescence of the wells is determined at 530-535nm excitation and 590nm emission in a fluorescence plate reader. Bioactivity in the absence of complement is also assayed in a separate plate by adding 50 μL of assay media, alternative a heat inactivated complement control can be used. Note that different plate layouts are possible.

WHO/BS/2016.2309 Page 45 Day 1-First Run Plate 1 (duplicate dilution series) 1 A B C D E F G H 2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12

Code A Replicate 1 Code B Replicate 1 Code C Replicate 1 In-house REF Replicate 1 Code A Replicate 2 Code B Replicate 2 Code C Replicate 2 In-house REF Replicate 2

Plate 2 (fresh duplicate dilution series)

1 A B C D E F G H

2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12

In-house REF Replicate 1 Code A Replicate 1 Code B Replicate 1 Code C Replicate 1 In-house REF Replicate 2 Code A Replicate 2 Code B Replicate 2 Code C Replicate 2

Plate 3 (fresh duplicate dilution series) 1 A B C D E F G H 2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12

Code C Replicate 1 In-house REF Replicate 1 Code A Replicate 1 Code B Replicate 1 Code C Replicate 2 In-house REF Replicate 2 Code A Replicate 2 Code B Replicate 2

Note that samples are further randomised in the plates for day 2 (plates 4, 5 and 6) and day 3 (plates 7, 8 and 9). Plate 10 (In the absence of complement- control plate) 1 A B C D E F G H 2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12

Code A Replicate 1 Code B Replicate 1 Code C Replicate 1 In-house REF Replicate 1 Code A Replicate 2 Code B Replicate 2 Code C Replicate 2 In-house REF Replicate 2

WHO/BS/2016.2309 Page 46 APPENDIX 2 Note that effector:target (E:T) cell ratio, antibody concentration range, assay buffer and incubation time may differ with different ADCC assay protocols and whether a reporter or a classical ADCC assay with PBMC or NK cells is used. ADCC reporter bioassay EXAMPLE –Outlined only Equipment and Reagents:

     

Target Cell Line: WIL2-S (ATCC® CRL-8885) or other suitable target cell line expressing CD20 antigen. Effector Cell line: Jurkat cells expressing high affinity variant V158 of FcγRIIIa receptor and transfected with NFAT-luciferase reporter gene or suitable ADCC reporter cell line (e.g. Promega Cat. No. G7010 or equivalent) Assay Medium: RPMI 1640, L-glutamine (1%), low IgG serum (4%) Assay plate: Sterile white 96-flat bottom cell culture plates (e.g. Nunc Cat No. 136101 or equivalent) Detection reagent: Luciferase substrate and buffer (e.g. Promega Bio-Glo™ Luciferase Assay System Cat. No. G7940 or equivalent) Luminescence Plate Reader e.g., Spectra Max M5.

Procedure: WIL2-S cells are maintained as per ATCC instructions and passaged at least two times prior to use in the bioassay. Cells at log growth phase are harvested, washed by centrifugation and resuspended in assay medium at 0.5x106 cells/ml. Effector cells are maintained as per manufacturer’s instructions and cell density is adjusted to 3x106 cells/ml. Cell counts and viability are measured by Trypan blue exclusion in an automated cell counter (Countess™, Thermo Fisher). Nine independent concentrations of test antibody and in-house reference material in assay media are tested in two independent replicate dilution series per plate. Twenty five µl of target cells are added per well into the flat bottom tissue culture 96-well plates and then 25 µl per well of antibody dilution series are added. Subsequently, 25µl of the effector cell suspension is added to the wells to allow an effector to target cell ratio of 6:1 and the plates are incubated for 6 hours at 37°C, 5% CO2 under a humid atmosphere. No antibody control (assay media only in the presence of target and effector cells) and cell-free wells (assay media only) are also included. Plates are removed from the incubator and equilibrated to room temperature for 15 min followed by the addition of 75 μL of the luciferase assay reagent and incubation at room temperature for 5 -30 min. Luminescence of the wells is determined in a luminescence plate reader. Note that different plate layouts are possible.

WHO/BS/2016.2309 Page 47 Day 1-First Run Plate 1 (duplicate dilution series) 1 A B C D E F G H 2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12

Plate 2 (fresh duplicate dilution series) 1 A B C D E F G H 2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12

In-house REF Replicate 1 Code A Replicate 1 Code B Replicate 1 Code C Replicate 1 In-house REF Replicate 2 Code A Replicate 2 Code B Replicate 2 Code C Replicate 2

Assay media

Plate 3 (fresh duplicate dilution series) 1 A B C D E F G H 2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12

Code C Replicate 1 In-house REF Replicate 1 Code A Replicate 1 Code B Replicate 1 Code C Replicate 2 In-house REF Replicate 2 Code A Replicate 2 Code B Replicate 2

Assay media

Note that samples are further randomised in the plates for day 2 (plates 4, 5 and 6) and for day 3 (plates 7, 8 and 9).

Assay media

Assay media

Assay media

Code A Replicate 1 Code B Replicate 1 Code C Replicate 1 In-house REF Replicate 1 Code A Replicate 2 Code B Replicate 2 Code C Replicate 2 In-house REF Replicate 2

Assay media

WHO/BS/2016.2309 Page 48 ADCC Classical bioassay with NK effector cells EXAMPLE –Outlined only (Schnueriger et al., 2011 Mol Immunol) Equipment and Reagents:

        

Target Cell Line: WIL2-S (ATCC® CRL-8885) or other suitable target cell line expressing CD20 antigen. Effector Cell line: CD16.NK92 expressing high affinity variant V158 of FcγRIIIa receptor or other suitable NK cells (CD16 transfected NK cell lines, primary PBMC or NK isolated from PBMC) Assay Medium: WIL2-S culture media (i.e. RPMI (95.7%), L-glutamine (1%), 1M HEPES (2%), heat inactivated FBS (10%), PenStrep (1%)) Assay plate: Sterile 96-U bottom cell culture plates (e.g. Falcon Cat No. 353077 equivalent) Read out plate: White 96-flat bottom plates (e.g. Nunc Cat No. 136101 or equivalent) Target cell labelling: DELFIA® BATDA labelling reagent (e.g. Perkin Elmer Cat No. C136-100 or equivalent) DELFIA® Lysis Buffer (e.g. Perkin Elmer Cat No. 4005-0010 or equivalent) Detection reagent: Europium Solution (e.g. Perkin Elmer Cat No. C135-100 or equivalent) Fluorescence Plate Reader e.g., Spectra Max M5.

Procedure: WIL2-S cells are maintained as per ATCC instructions and passaged at least two times prior to use in the bioassay. Cells at log growth phase are harvested, washed by centrifugation and resuspended in assay medium at 1x106 cells/ml. The cells are labelled with DELFIA® BATDA labelling reagent as per manufacturer instructions. The labelled target cells were washed twice in PBS and resuspended at a cell density of 0.2x106 cells/ml. A CD16 transfected NK cell line used as effector cells is maintained as per manufacturer’s instructions and cell density is adjusted to 0.6x106 cells/ml. Cell counts and viability are measured by Trypan blue exclusion in an automated cell counter (Countess™, Thermo Fisher). Nin e independent concentrations of test antibody and in-house reference material in assay media are tested in two independent replicate dilution series per plate. One hundred µl of a 1:1 pre-mixed target:effector cell suspension that allows a final effector to target ratio of 3:1 are added per well into the U-bottom tissue culture 96-well plates and then 100 µl per well of antibody dilution series is transferred into the wells. Twenty µl of DELFIA® Lysis Buffer as maximum release (killing) control is added to designated wells and the plates are incubated for 2 hours at 37°C, 5% CO2 under a humid atmosphere. No antibody control (assay media only in the presence of target and effector cells) and target cells alone (as spontaneous release control) are also included. Plates are removed from the incubator and centrifuged at 1000g for 5 min and 40 μL of cell-free supernatant is transferred from each well into a white flat-bottom 96-well plate. Then 160 μL per well of Europium solution is added and incubated for approx. 15 min with gentle shaking. Time-resolved fluorescence is measured in a fluorescence plate reader using excitation at 345 n and emission at 615 nm. Note that different plate layouts are possible.

WHO/BS/2016.2309 Page 49 Day 1-First Run Plate 1 (duplicate dilution series) 1 No Ab

2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12 No Ab

A B C D E F G H

Code A Replicate 1 Code B Replicate 1 Code C Replicate 1 In-house REF Replicate 1 Code A Replicate 2 Code B Replicate 2 Code C Replicate 2 In-house REF Replicate 2

Target cells only

Plate 2 (fresh duplicate dilution series) 1 No Ab

2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12 No Ab

A B C D E F G H

In-house REF Replicate 1 Code A Replicate 1 Code B Replicate 1 Code C Replicate 1 In-house REF Replicate 2 Code A Replicate 2 Code B Replicate 2 Code C Replicate 2

Target cells only

Plate 3 (fresh duplicate dilution series) 1 No Ab

2 Dilution#1

3 Dilution#2

4 Dilution#3

5 Dilution#4

6 Dilution#5

7 Dilution#6

8 Dilution#7

9 Dilution#8

10 Dilution#9

11 No Ab

12 No Ab

A B C D E F G H

Code C Replicate 1 In-house REF Replicate 1 Code A Replicate 1 Code B Replicate 1 Code C Replicate 2 In-house REF Replicate 2 Code A Replicate 2 Code B Replicate 2

Target cells only

Note that samples are further randomised in the plates for day 2 (plates 4, 5 and 6) and for day 3 (plates 7, 8 and 9).

Lysis Buffer

Lysis buffer

Lysis buffer

WHO/BS/2016.2309 Page 50

Appendix 5 Draft Instructions for use

WHO/BS/2016.2309 Page 51