Anouk Donners

Innovative drug monitoring; of factor VIII and emicizumab in haemophilia A An<>uk Donners

Innovative drug monitoring of factor VIII and emicizumab in haemophilia A Anouk Donners

COLOPHON The work presented in this thesis was performed at the Department of Clinical Pharmacy in collaboration with the Van Creveldkliniek at the University Medical Center Utrecht, Utrecht, the Netherlands. Provided by thesis specialist Ridderprint, ridderprint.nl Printing: Ridderprint Cover design: David Veldhuizen Layout and design: Eduard Boxem, persoonlijkproefschrift.nl ISBN: 978-94-6483-066-8 © 2023 A.A.M.T. Donners All rights reserved. No parts of this thesis may be reproduced or transmitted in any form or by any means without permission in writing by the author, or when appropriate, by the publishers of the publications.

Innovative drug monitoring of factor VIII and emicizumab in haemophilia A Innovatieve geneesmiddelmonitoring van factor VIII en emicizumab in hemofilie A (met een samenvatting in het Nederlands) Proefschrift ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof. dr. H.R.B.M. Kummeling, ingevolge het besluit van het college voor promoties in het openbaar te verdedigen op donderdag 1 juni 2023 des middags te 12.15 uur door Anouk Anna Marie Therese Donners geboren op 16 februari 1991 te Heerlen

Promotoren: Prof. dr. A.C.G. Egberts Prof. dr. R.E.G. Schutgens Copromotoren: Dr. C.M.A. Rademaker Dr. K. Fischer Beoordelingscommissie: Prof. dr. A. de Boer Dr. M.H. Cnossen Prof. dr. H.C.J. Eikenboom Prof. dr. R. Gehring Prof. dr. G. Pasterkamp (voorzitter)

Ars longa, vita brevis, occasio praeceps, experimentum periculosum, iudicium difficile. The art is long, and life short, opportunity fleeting, experiment treacherous, judgment difficult. ‒ Hippocrates

Contents Chapter 1 General Introduction 12 Chapter 2 Six-step workflow for the quantification of therapeutic monoclonal antibodies in biological matrices with liquid chromatography mass spectrometry - A tutorial Anal Chim Acta 2019; 1080: 22‒34 24 Chapter 3 Quantification of coagulation factor VIII in human plasma with liquid chromatography tandem mass spectrometry using a selective sample purification with camelid nanobodies J Pharm Biomed Anal 2019; 175: 112781 50 Chapter 4 Comparison between coagulation factor VIII quantified with one-stage activity assay and with mass spectrometry in haemophilia A patients: Proof of principle Int J Lab Hematol 2020; 42: 819‒826 68 Chapter 5 Quantification of emicizumab by mass spectrometry in plasma of people with haemophilia A: A method validation study Res Pract Thromb Haemost 2022; 6: e12725 84 Chapter 6 Pharmacokinetics and Associated Efficacy of Emicizumab in Humans: A Systematic Review Clin Pharmacokinet 2021; 60: 1395‒1406 102 Chapter 7 The efficacy of the entire-vial dosing of emicizumab: real-world evidence on plasma concentrations, bleeds and drug waste Res Pract Thromb Haemost 2023; 7: e100074 124 Chapter 8 DosEmi study protocol: a phase IV, multicenter, open-label, crossover study to evaluate noninferiority of pharmacokineticguided reduced dosing compared with conventional dosing of emicizumab in persons with haemophilia A Submitted 142 Chapter 9 General Discussion 158 Chapter 10 Summary 180 Chapter 11 Nederlandse lekensamenvatting 186 Appendices About the author List of coauthors List of publications Dankwoord 194 195 198 200

CHAPTER General Introduction Anouk A.M.T. Donners Carin M.A. Rademaker Roger E.G. Schutgens Toine C.G. Egberts Kathelijn Fischer

12 Chapter 1 Haemophilia A Haemophilia A is an X-linked congenital bleeding disorder caused by a deficiency of coagulation factor VIII (FVIII). The diagnosis is based on a person’s endogenous FVIII activity and is classified into severe (<1%), moderate (1 to 5%) or mild (>5 to 40%) haemophilia A [1]. Approximately eight out of 100,000 people have haemophilia A (PwHA) in the Netherlands [2]. These PwHA receive care at six dedicated Haemophilia Treatment Centers, of which Van Creveldkliniek, established in 1964, is the oldest. PwHA experience bleeding, predominantly into major joints, such as ankles, knees and elbows, leading to painful and disabling arthropathy. Intracranial bleeds and bleeding into internal organs may also occur among PwHA, which can be life-threatening. When left untreated, the bleeds occur spontaneously among those suffering from severe haemophilia. In contrast, persons with moderate and mild haemophilia present a milder phenotype in which trauma and surgical interventions can provoke uncontrolled bleeding. The life expectancy of PwHA has been extended from 10–15 years a century ago to a nearly normal life expectancy and quality of life today [3]. Pharmacotherapy for people with haemophilia A Haemophilia therapy has progressed remarkably throughout the twentieth century, progressing from no available therapy to complete blood and plasma infusions. The cornerstone of therapy for PwHA is the substitution of missing FVIII [4, 5]. In the 1960s, the discovery of cryoprecipitate, which involves concentrating FVIII in a pellet, led to the industrial manufacturing of plasma-derived FVIII (pdFVIII) in the 1970s [6]. This breakthrough provided the first efficacious treatment of bleeds and marked the beginning of home care and self-infusion. The success was overshadowed in the 1980s by the outbreak of serious and fatal blood-borne viral infections such as hepatitis and HIV/ AIDS in a large proportion of PwHA due to a lack of screening methods [7, 8]. Fortunately, safer products became available in the 1990s due to new DNA technology that provided recombinant FVIII (rFVIII) and manufacturing advancements in viral inactivation and virus removal, which provided safer pdFVIII [6, 9]. PwHA were burdened by frequent weekly intravenous injections due to the short half-lives (SHL) of rFVIII products. This drove the development of the extended half-life (EHL) rFVIII products in the 2010s, although maximum half-lives of only 20 hours were achieved [10, 11]. The improvements of EHL products included Fc-infusion, conjugation of polyethylene glycol or shortening of the protein sequence to increase stability [12]. Currently, 15 FVIII products are available on the Dutch market [13]. Substitution therapy with these FVIII products can be given prophylactically to prevent bleeds (especially for severe PwHA) or on-demand to treat bleeds. Prophylactic FVIII replacement therapy has effectively reduced the average treated bleeds from 20–30 to 1–4 per person per year [1, 14]. Moreover, prophylaxis can convert a severe phenotype into a moderate phenotype

13 General Introduction when maintaining FVIII activity above 1% [15, 16]. The main challenges of FVIII therapy are the FVIII-inhibitor development in 30% of PwHA, which renders the treatment ineffective, the burden of intravenous administration and the costs. Financial restraints lead to poor access and health inequalities worldwide, as the cost of prophylaxis therapy is estimated at ~150,000 euros per person per year [17, 18]. A new therapeutic strategy became available with the introduction of non-factor replacement therapies. Emicizumab, which was approved in 2018, was the first therapy offering prophylaxis in PwHA with and without FVIII inhibitors. This recombinant, bispecific, monoclonal antibody (mab) effectively mimics the FVIII function to an equivalent of 10–20% FVIII activity [19-21]. Because there is no sequence homology to FVIII, the PwHA with FVIII inhibitors have an effective prophylactic option for the first time. Compared to FVIII products, the other revolutionising benefits of emicizumab are the subcutaneous instead of intravenous administration and the monthly instead multiple-weekly dosing frequency. Side effects, which rarely occur, include thrombotic events with concomitant use of activated prothrombin complex concentrate [22] and <0.6% immunogenicity with declining plasma concentrations [23]. While most PwHA in the Netherlands wish to start emicizumab prophylaxis, patient access is limited due to the financial impact on healthcare budget. Public pricing of emicizumab therapy is higher than FVIII therapy and is set at ~400,000 euros per person per year, although non-public pricing is probably close to the costs of the current FVIII therapies [24-26]. Another disadvantage of emicizumab is that it does not fully correct coagulation, and is therefore not suitable for the management of acute bleeds or major surgery [5]. Several other non-factor replacement products are currently under clinical investigation and are expected to be approved in the near future [27]. Their mode of action is either mimicking the FVIII function (i.e., similar to emicizumab) or targeting the natural coagulation inhibitors (i.e., antithrombin, tissue factor pathway inhibitor, or activated protein C). Rare thromboembolic events have been reported and warrant continuous post-marketing surveillance, although the overall safety profile looks promising [28]. The ultimate goal of haemophilia treatment is a phenotypical cure, which is achievable today, as Roctavian® valoctocogene roxaparvovec, the first gene therapy product of its kind, was approved in June 2022 [29]. Gene therapy allows PwHA to avoid the fears and obligations of haemophilia A treatment for a number of years [30, 31]. Monitoring Prescribing drugs involves more than writing a prescription and following the drug label [32, 33]. It also includes monitoring of the individual by continuous evaluation of the benefit-risk balance of pharmacotherapy and optimisation. Monitoring can involve the clinical observation of the individual or actually measuring markers that indicate a disease status. Many of these markers are biomarkers measured in a laboratory or by the 1

14 Chapter 1 individual at home (point-of-care testing). In fact, it has been shown that for a substantial part of the commonly used drugs, the summary of product characteristics (SmPC) (i.e., the drug label) or the clinical guidelines advise (“prescribe”) the measurement of laboratory markers [34]. The aim of drug dosing is to reach a so-called therapeutic window. The balance between benefit and risk (i.e., desired effect and toxicity) is considered optimal for an individual in this window [35] (see Figure 1). Subtherapeutic effects are expected under the window and supratherapeutic due to maximum effect resulting in toxic or financial toxicity. For some drugs (e.g., aminoglycosides, lithium, digoxin), the therapeutic window is narrow, and strict monitoring is of the essence. Monitoring can involve the measurement of either an endogenic biomarker (e.g., glucose, INR, anti-Xa) or the measurement of the drug in plasma, serum, blood or other specimens. The latter is called therapeutic drug monitoring (TDM). Figure 1. Dose-response curve with a therapeutic window. Monitoring of FVIII The standard monitoring approach in haemophilia management does not involve the measurement of a drug concentration; instead, the biomarker ‘FVIII activity’ is used. This FVIII activity is monitored to diagnose haemophilia A, to dose the FVIII products and to provide clinical support to PwHA (e.g., during surgery or bleeding) (see Figure 2). FVIII activity is used for diagnosing purposes to classify the severity of haemophilia A. The frequency of monitoring and treatment strategy are based on the severity classification.

15 General Introduction People with severe haemophilia A are monitored closely and receive prophylaxis early in life, while people with moderate and mild haemophilia A are monitored less often and mostly receive on-demand treatment. For dose monitoring, a minimal effective FVIII activity is the goal, and this goal differs between clinical situations. For instance, ≥1% of FVIII activity is required for prophylaxis, 30% for treating a mild bleed, or around 100% during surgery. High FVIII activity is not necessarily toxic but merely unattractive from a financial perspective, and high variability for FVIII products has been reported [4, 10]. Thus, monitoring for dose purposes is important to ensure sufficient efficacy. FVIII activity is currently measured with different clotting assays, mainly by the onestage clotting assay (OSA) or the chromogenic-substrate assay (CSA). These assays are routinely available in many clinical laboratories worldwide and are relatively fast and cheap to perform using an automated coagulation analyser. These assays demonstrate discrepancies in some cases, for instance, between the mild and moderate severity, between different FVIII-product types or between different laboratories [36]. Assay variability has been related to the use of different reagents, assay settings, assay interference (e.g., anti-drug antibodies [ADA] or non-factor products) and some lack of standardisation [36-38]. Ascertaining the FVIII activity is, however, critical in haemophilia A, particularly in the lower area of 0−6%. Measurement errors may worsen a person’s bleeding prognosis due to misdiagnosis of the severity classification, which results in postponed prophylaxis, or by suboptimal dosing of the FVIII products, which results in insufficient efficacy or financial “toxicity”. Monitoring of emicizumab Emicizumab is a so-called biopharmaceutical. The standard monitoring approach for biopharmaceuticals is TDM and the determination of anti-drug antibodies (ADA). The general purpose of monitoring is to guide treatment decisions to optimise treatment and cost-effectiveness. The dose selection of biopharmaceuticals is often set at the upper end of the dose–response curve, given the absence of a maximum tolerated dose. Moreover, the most important toxicity, immunogenicity, is not dose related [35, 39, 40]. For emicizumab specifically, its drug label does not mention laboratory monitoring or measurement of concentrations during treatment. However, guidelines on emicizumab therapy do recommend monitoring its concentration when suspecting ADAs [41]. Moreover, the monitoring of emicizumab might be useful in order for clinicians to recognise therapy non-adherence and for researchers to investigate its clinical pharmacology. Lastly, cost-efficient monitoring might be appealing when efficacy is maintained with lower doses of emicizumab, leading to healthcare savings and improved patient access [42]. Thus, the role of TDM for emicizumab is helpful in clinical support, research and potentially in cost-efficient dosing (see Figure 2). 1

16 Chapter 1 The standard method for measuring the concentration of emicizumab in clinical practice is based on a calibrated, modified clotting assay. This assay is not commonly adopted in clinical laboratories, and moreover, assay interference by high concentrations of FVIII concentrates or anti-emicizumab antibodies has been reported [43]. Figure 2. Monitoring purposes of factor VIII and emicizumab. Mass spectrometry: A novel technique in haemophilia A In the last decennium, liquid chromatography (LC) coupled with tandem mass spectrometry (MS/MS) has rapidly become the state-of-the-art technique for TDM in the laboratory landscape. Advantages of this technique over other techniques, such as clotting assays or enzyme-linked immunosorbent assays, include high sensitivity, speed, resolution, accuracy and reproducibility. In addition to these analytical benefits, LC-MS/ MS methods offer the potential of multiplexing, leading to more efficient measurement runs. Especially advantageous is the low sampling volume, which offers minimal blood drawing for children and the potential of dried-blood-spot measurements. Additionally, LC-MS/MS analysis will likely continue to grow as one of the leading techniques for the quantification of proteins (i.e., proteomics) in a clinical setting.

17 General Introduction Knowledge gaps Measuring the concentrations of FVIII and emicizumab with LC-MS/MS has not been done before. This LC-MS/MS technique may be of interest to treating clinicians because of its many advantages in comparison to other assays. The potential role of bioanalysis with LC-MS/MS in haemophilia A is schematically presented in Figure 3. For FVIII, the dose, the FVIII activity and the bleed outcomes are well correlated, but the role of the FVIII concentration measured with LC-MS/MS in the dose–concentration–response relation is unknown. For emicizumab, no biomarker has yet been identified, and the entire dose–concentration–outcome relation, established in the pre-approval studies, is unclear. Bioanalysis with the LC-MS/MS technique could be applied to start filling these knowledge gaps and to further optimise the monitoring of FVIII and emicizumab in PwHA in the future. Two questions arise from these knowledge gaps: - How is the relationship between FVIII concentration and FVIII activity best described? - How low can we dose emicizumab without sacrificing the desired clinical response? Figure 3. Dose–concentration–response relationship with bioanalysis of FVIII and emicizumab in PwHA. 1

18 Chapter 1 Thesis objectives This thesis is designed to optimise the drug monitoring of FVIII and emicizumab in PwHA. To achieve this objective, I will: 1) Develop and validate the LC-MS/MS methods that quantify FVIII and emicizumab in human plasma; 2) investigate the dose–biomarker relationship of FVIII and the dose–concentration– response relationships of emicizumab; and 3) propose and evaluate a cost-efficient dosing strategy for emicizumab. Outline In Chapters 2, 3 and 5, I provide a framework for measuring therapeutic proteins, such as FVIII and emicizumab, in human plasma using LC-MS/MS bioanalysis. In Chapter 4, I investigate the concentration–biomarker relationship of FVIII, and in Chapters 5−7, I investigate the dose–concentration–response relationship for emicizumab. In Chapters 6−8, I conduct studies to support a cost-efficient approach to emicizumab treatment, which may inspire others to dose biopharmaceuticals more affordable. An overall discussion in Chapter 9 is presented in the last chapter of this thesis to put the lessons learned into a broader perspective. I also deliberate on the potential future applications of monitoring FVIII concentrations in PwHA and a framework for the cost-efficient dosing of biopharmaceuticals. This thesis can be summarised as the groundwork for measuring and monitoring FVIII and emicizumab with LC-MS/MS bioanalysis in PwHA. Author’s contribution AD conceived the idea and set-up the general introduction. AD conducted literature review, outlined and wrote the general introduction. Throughout the process, AD asked and implemented input and feedback from the supervision team.

19 General Introduction References 1. Mannucci PM, Tuddenham EG. The hemophilias--from royal genes to gene therapy. N Engl J Med. 2001; 344: 1773-1779. 2. Stichting_HemoNED. Jaarrapportage 2021. Data en trends vanuit het Nederlands Hemofilie Register en het digitale logboek VastePrik. https://hemoned.nl/publicaties/jaarrapportages/. Accessed on 2010-2022. 3. Berntorp E, Fischer K, Hart DP, Mancuso ME, Stephensen D, Shapiro AD, Blanchette V. Haemophilia. Nat Rev Dis Primers. 2021; 7: 45. 4. Srivastava A, Santagostino E, Dougall A, Kitchen S, Sutherland M, Pipe SW, Carcao M, Mahlangu J, Ragni MV, Windyga J, Llinas A, Goddard NJ, Mohan R, Poonnoose PM, Feldman BM, Lewis SZ, van den Berg HM, Pierce GF, panelists WFHGftMoH, co a. WFH Guidelines for the Management of Hemophilia, 3rd edition. Haemophilia. 2020; 26 Suppl 6: 1-158. 5. Aledort L, Mannucci PM, Schramm W, Tarantino M. Factor VIII replacement is still the standard of care in haemophilia A. Blood Transfus. 2019; 17: 479-486. 6. Marchesini E, Morfini M, Valentino L. Recent Advances in the Treatment of Hemophilia: A Review. Biologics. 2021; 15: 221-235. 7. Farrugia A, Smit C, Buzzi A. The legacy of haemophilia: Memories and reflections from three survivors. Haemophilia. 2022; 28: 872-884. 8. O’Mahony B. Haemophilia care in Europe: Past progress and future promise. Haemophilia. 2020; 26: 752-758. 9. Mannucci PM. Hemophilia therapy: the future has begun. Haematologica. 2020; 105: 545-553. 10. Versloot O, Iserman E, Chelle P, Germini F, Edginton AN, Schutgens REG, Iorio A, Fischer K, Pharmacokinetic Expert Working Group of the International Prophylaxis Study G. Terminal half-life of FVIII and FIX according to age, blood group and concentrate type: Data from the WAPPS database. J Thromb Haemost. 2021; 19: 1896-1906. 11. Franchini M, Mannucci PM. The More Recent History of Hemophilia Treatment. Semin Thromb Hemost. 2022; 48: 904-910. 12. Graf L. Extended Half-Life Factor VIII and Factor IX Preparations. Transfus Med Hemother. 2018; 45: 86-91. 13. CBG-MEB. Geneesmiddeleninformatiebank ATC code B02BD02. https://www. geneesmiddeleninformatiebank.nl. Accessed on 20-10-2022. 14. Ay C, Perschy L, Rejto J, Kaider A, Pabinger I. Treatment patterns and bleeding outcomes in persons with severe hemophilia A and B in a real-world setting. Ann Hematol. 2020; 99: 2763-2771. 15. Den Uijl IE, Mauser Bunschoten EP, Roosendaal G, Schutgens RE, Biesma DH, Grobbee DE, Fischer K. Clinical severity of haemophilia A: does the classification of the 1950s still stand? Haemophilia. 2011; 17: 849-853. 16. Tiede A, Abdul Karim F, Jimenez-Yuste V, Klamroth R, Lejniece S, Suzuki T, Groth A, Santagostino E. Factor VIII activity and bleeding risk during prophylaxis for severe hemophilia A: a population pharmacokinetic model. Haematologica. 2021; 106: 1902-1909. 1

20 Chapter 1 17. Fischer K, Steen Carlsson K, Petrini P, Holmstrom M, Ljung R, van den Berg HM, Berntorp E. Intermediatedose versus high-dose prophylaxis for severe hemophilia: comparing outcome and costs since the 1970s. Blood. 2013; 122: 1129-1136. 18. Pierce GF, Adediran M, Diop S, Dunn AL, El Ekiaby M, Kaczmarek R, Konkle BA, Pipe SW, Skinner MW, Valentino LA, Robinson F, Ampartzidis G, Martin J, Haffar A. Achieving access to haemophilia care in low-income and lower-middle-income countries: expanded Humanitarian Aid Program of the World Federation of Hemophilia after 5 years. Lancet Haematol. 2022; 9: e689-e697. 19. Kizilocak H, Marquez-Casas E, Malvar J, Carmona R, Young G. Determining the approximate factor VIII level of patients with severe haemophilia A on emicizumab using in vivo global haemostasis assays. Haemophilia. 2021; 27: 730-735. 20. Lenting PJ. Laboratory monitoring of hemophilia A treatments: new challenges. Blood Adv. 2020; 4: 2111-2118. 21. Ferriere S, Peyron I, Christophe OD, Kawecki C, Casari C, Muczynski V, Nathwani A, Kauskot A, Lenting PJ, Denis CV. A hemophilia A mouse model for the in vivo assessment of emicizumab function. Blood. 2020; 136: 740-748. 22. Kizilocak H, Marquez-Casas E, Malvar J, Young G. Safety of FEIBA and emicizumab (SAFE): Dose escalation study evaluating the safety of in vivo administration of activated prothrombin complex concentrate in haemophilia A patients on emicizumab. Haemophilia. 2023; 29: 100-105. 23. Schmitt C, Emrich T, Chebon S, Fernandez E, Petry C, Yoneyama K, Kiialainen A, Howard M, Niggli M, Paz-Priel I, Chang T. Low immunogenicity of emicizumab in persons with haemophilia A. Haemophilia. 2021; 27: 984-992. 24. ZIN. Horizonscan - emicizumab voor routinematige profylaxe van bloedingen bij patiënten met hemofilie A zonder remmers tegen factor VIII, versie 6. https://www.horizonscangeneesmiddelen.nl/ geneesmiddelen/emicizumab-cardiovasculaire-aandoeningen-hemostase_bevorderende_medicatie/ versie6. Accessed on 20-10-2022. 25. ZIN. Horizonscan - emicizumab for routine prophylaxis of bleeding episodes in patients with haemophilia A (congenital factor VIII deficiency): mild or moderate disease for whom prophylaxis is clinically indicated, version 2. https://www.horizonscangeneesmiddelen.nl/geneesmiddelen/emicizumab-cardiovasculaireaandoeningen-hemostase_bevorderende_medicatie%5B2%5D/versie2. Accessed on 10-01-2023. 2022; online ahead of print. 26. Oka G, Pieragostini R, Roussel-Robert V, Paubel P, Degrassat-Theas A, Lopez I. [Assessment of the budgetary impact of an emicizumab therapy introduction for patients with severe haemophilia A without inhibitor]. Ann Pharm Fr. 2022; online ahead of print. 27. Swan D, Mahlangu J, Thachil J. Non-factor therapies for bleeding disorders: A primer for the general haematologist. EJHaem. 2022; 3: 584-595. 28. Gualtierotti R, Pasca S, Ciavarella A, Arcudi S, Giachi A, Garagiola I, Suffritti C, Siboni SM, Peyvandi F. Updates on Novel Non-Replacement Drugs for Hemophilia. Pharmaceuticals (Basel). 2022; 15. 29. RIVM. De ziekten die de hielprik opspoort. https://www.pns.nl/hielprik/ziekten-die-hielprik-opspoort. Accessed on 01-06-2022. 30. Spadarella G, Di Minno A, Brunetti-Pierri N, Mahlangu J, Di Minno G. The evolving landscape of gene therapy for congenital haemophilia: An unprecedented, problematic but promising opportunity for worldwide clinical studies. Blood Rev. 2021; 46: 100737.

21 General Introduction 31. Di Minno G, Castaman G, De Cristofaro R, Brunetti-Pierri N, Pastore L, Castaldo G, Trama U, Di Minno M. Progress, and prospects in the therapeutic armamentarium of persons with congenital hemophilia. Defining the place for liver-directed gene therapy. Blood Rev. 2022: 101011. 32. De Vries T, Henning R, Hogerzeil H, Fresle D. Guide to good prescribing. World Health Organisation, Geneva. 1994. 33. Tichelaar J, Richir MC, Garner S, Hogerzeil H, de Vries T. WHO guide to good prescribing is 25 years old: quo vadis? Eur J Clin Pharmacol. 2020; 76: 507-513. 34. Geerts AF, De Koning FH, Van Solinge WW, De Smet PA, Egberts TC. Instructions on laboratory monitoring in 200 drug labels. Clin Chem Lab Med. 2012; 50: 1351-1358. 35. Papamichael K, Vogelzang EH, Lambert J, Wolbink G, Cheifetz AS. Therapeutic drug monitoring with biologic agents in immune mediated inflammatory diseases. Expert Rev Clin Immunol. 2019; 15: 837-848. 36. Peyvandi F, Oldenburg J, Friedman KD. A critical appraisal of one-stage and chromogenic assays of factor VIII activity. J Thromb Haemost. 2016; 14: 248-261. 37. Muller J, Miesbach W, Pruller F, Siegemund T, Scholz U, Sachs UJ, Standing Commission Labor of the Society of T, Haemostasis R. An Update on Laboratory Diagnostics in Haemophilia A and B. Hamostaseologie. 2022; 42: 248-260. 38. Kitchen S, Jennings I, Makris M, Kitchen DP, Woods TA, Walker ID. Factor VIII assay variability in postinfusion samples containing full length and B-domain deleted FVIII. Haemophilia. 2016; 22: 806-812. 39. Lee SY. Therapeutic Drug Monitoring of Biologic Agents in the Era of Precision Medicine. Ann Lab Med. 2020; 40: 95-96. 40. Perry M, Abdullah A, Frleta M, MacDonald J, McGucken A. The potential value of blood monitoring of biologic drugs used in the treatment of rheumatoid arthritis. Ther Adv Musculoskelet Dis. 2020; 12: 1759720X20904850. 41. Jenkins PV, Bowyer A, Burgess C, Gray E, Kitchen S, Murphy P, Platton S, Riddell A, Chowdary P, Lester W. Laboratory coagulation tests and emicizumab treatment A United Kingdom Haemophilia Center Doctors’ Organisation guideline. Haemophilia. 2020; 26: 151-155. 42. Lehtinen AE, Lassila R. Do we need all that emicizumab? Haemophilia. 2022; 28: e53-e55. 43. Coppola A, Castaman G, Santoro RC, Mancuso ME, Franchini M, Marino R, Rivolta GF, Santoro C, Zanon E, Sciacovelli L, Manca S, Lubrano R, Golato M, Tripodi A, Rocino A, ad hoc Working G. Management of patients with severe haemophilia a without inhibitors on prophylaxis with emicizumab: AICE recommendations with focus on emergency in collaboration with SIBioC, SIMEU, SIMEUP, SIPMeL and SISET. Haemophilia. 2020; 26: 937-945. 1

CHAPTER Six-step workflow for the quantification of therapeutic monoclonal antibodies in biological matrices with liquid chromatography mass spectrometry: A tutorial Mohsin El Amrani Anouk A.M.T. Donners C. Erik Hack Alwin D.R. Huitema Erik M. van Maarseveen Anal Chim Acta 2019; 1080: 22-34

24 Chapter 2 ABSTRACT The promising pipeline of therapeutic monoclonal antibodies (mAbs) demands robust bioanalytical methods with swift development times for pharmacokinetic studies. Over the past decades ligand binding assays were the methods of choice for absolute quantification. However, the production of the required anti-idiotypic antibodies and ligands limits high-throughput method development for sensitive, accurate, and reproducible quantification of therapeutic mAbs. In recent years, high-resolution liquid chromatography-tandem mass spectrometry (LC-MS) systems have enabled absolute quantification of therapeutic mAbs with short method development times. These systems have additional benefits, such as a large linear dynamic range, a high specificity and the option of multiplexing. Here, we briefly discuss the current strategies for the quantification of therapeutic mAbs in biological matrices using LC-MS analysis based on top-down and middle-down quantitative proteomics. Then, we present the widely used bottom-up method in a six-step workflow, which can be used as guidance for quantitative LC-MS/MS method development of mAbs. Finally, strengths and weaknesses of the bottom-up method, which currently provides the most benefits, are discussed in detail.

25 Tutorial on LC-MS/MS methods quantifying mAbs INTRODUCTION Therapeutic monoclonal antibodies (mAbs) nowadays are widely accepted as valuable treatment options for patients suffering from a variety of diseases, particularly in the areas of oncology and immune diseases. At present, 76 mAbs have been granted market authorization by the Food and Drug Administration (FDA) and European Medicines Agency (EMA) and are now commercially available for therapeutic use [1]. Judging from drug pipelines this number is set to grow considerably in the near future [2]. Therapeutic mAbs target pathological processes with high specificity and concomitantly lead to fewer side effects compared to conventional small molecule based therapies [3]. Furthermore, due to the high molecular weight of mAbs, the clearance pathway is not by renal elimination after hepatic enzyme metabolism but rather by proteolytic catabolism, receptor-mediated uptake and degradation, and sometimes by the catabolic pathway of their molecular target. Two thirds of mAbs are salvaged from degradation by binding to the protective neonatal Fc-receptor (FcRn) particularly on endothelial cells, which extends their elimination half-life to ~18-21 days [4]. MAb production and design has made great strides from the early discovery in 1975 by Kohler and Milstein [5]. The progression from murine mAbs (1975) using hybridoma technology to chimeric mAbs (1984) using recombinant DNA techniques to humanized mAbs (1988) using complementary determining region (CDR) grafting and finally to fully human mAbs (1994) using phage display or transgenic mice took less than 20 years [6-9]. These steps were essential to reduce the risk of anti-drug antibodies (ADA) development and allergic reactions associated with first generation mAbs [10-12]. In fact, additional requirements from the EMA, Food and Drug Administration (FDA) and World Health Organization (WHO) for the evaluation and monitoring of immunogenicity of new biopharmaceuticals were mandated as part of regulatory approval, together with a rigorous post-authorization pharmacovigilance with product-level traceability for of all biopharmaceuticals [13-17]. The discovery of new therapeutic targets and the high treatment efficacy of biopharmaceuticals accelerated the development of novel mAbbased therapies [16, 18]. For this purpose, bioanalytical methods were necessary to facilitate the required preclinical pharmacokinetic (PK) studies. In addition, therapeutic drug monitoring of mAbs concentrations can highlight accelerated drug clearance in patients which is indicative of ADA development and loss of drug response. Traditional bioanalytical methods such as ligand-binding assays rely on an anti-idiotypic antibody or a ligand with high avidity towards the therapeutic protein of interest. However, the development of such antibodies is notoriously difficult and time consuming [19-21]. Therefore, advances in analytical techniques were essential to attain shorter method development times, which is why liquid chromatography tandem mass-spectrometry (LC-MS/MS) has received increasing interest as an alternative method for quantification over the last decade. Following strength and weaknesses analysis of ligand binding assays, this tutorial systematically addresses bioanalytical methods to quantify therapeutic mAbs in biological matrices using LC-MS. Three main branches of quantitative proteomics 2

26 Chapter 2 using top-down (intact), middle-down (semi-intact) and bottom-up (signature peptide) strategies are briefly explored. Finally, the most widely used bottom-up quantification strategy via signature peptide is chronologically discussed in a general workflow where strengths and weaknesses of each step are extensively explained. Ligand binding assay Enzyme-linked immunosorbent assay (ELISA) is arguably the most widely used form of ligand binding assay (LBA), because of its high sensitivity, ease of use, and low instrumental costs [22, 23]. Moreover, due to the spectrophotometric detection principle, ELISA offers a safer alternative with a high ease of use compared to historically used radio immunoassays which require special facilities and operators trained to handle radioactive material [24]. In general, ELISA methods are based on the quantification of the target protein (antigen) using strategies, such as direct, indirect or sandwich type ELISA (Figure 1) [25-27]. In the final step, quantification is performed by adding a substrate solution, often tetramethylbenzidine (TMB), which is gradually oxidized by horseradish peroxidase (HRP) to a colored product. To improve sensitivity the use of a polyclonal secondary antibody in indirect or in sandwich type ELISA can help boost the signal intensity [28]. Alternatively, polymerized HRP can be used to increase sensitivity. Background staining is a very common challenge in these assays, and is often caused by non-specific interactions of either the primary or the secondary polyclonal antibody used [29]. The choice of binding strategy can help minimize the risk of non-specific interactions. For example, in sandwich type ELISA’s, an additional anti-idiotypic antibody binding the antigen at a different epitope provides higher specificity [30]. The reliance of ELISA methods on these specific anti-idiotypic antibodies or ligands can lead to lengthy method development times [31] and higher consumables costs compared to LC-MS/MS methods. Furthermore, ELISA’s allow for one component analysis in one specific bio-matrix, while LC-MS/MS methods allows for multiplexed measurement of multiple therapeutic and endogenous proteins in various bio-matrices [32-34] and due to the narrow linear dynamic range in ELISA, accurate quantification may require multiple sample dilutions thus limiting sample throughput. Importantly, the validation acceptance criteria for ELISAs are less stringent in comparison to LC-MS/MS methods [35, 36], mainly because ELISA methods cannot incorporate internal standards to correct for binding efficiencies influenced by sample matrix or component loss due to binding and washing steps. Finally, different results can be obtained with different ELISA assays as was demonstrated by Vande Casteele by comparison of three commercially available ELISA kits for infliximab and anti-infliximab quantification, stating that comparison of drug levels and ADA monitoring is hampered by lack of standardization [37]. Inter-assay variability makes it difficult for clinicians to compare results from other centers and invalid measurements affects decisions made in patient’s diagnosis and treatment.

27 Tutorial on LC-MS/MS methods quantifying mAbs Figure 1. A) Direct ELISA, B) Indirect ELISA and C) Sandwich ELISA. Top-down, middle-down and bottom-up quantitative proteomics Quantification with LC-MS offers various advantages over LBA and over the years various LC-MS strategies have been explored and reported. Three main strategies are discussed in detail namely topdown, middle-down and bottom-up proteomics. The flow chart in Figure 2 depicts the decision making process for the preferred strategy for quantification based on factors, such as the intended assay specifications in terms of sensitivity and selectivity and the applied instrumentation and materials. For therapeutic drug monitoring of trough levels, in most cases a LLOQ of 1 mg/L would suffice [38]. This would allow for the quantification of the peptide using LC-MS/MS or intact after selective purification using various instruments such as LC-HRMS, LBA or LC-FLD. However, for pharmacokinetic applications more sensitive assays are regularly required (e.g., 100 µg/L or less) to characterize the terminal elimination phase. Therefore, in most cases only LC-MS/MS using bottom-up proteomics to quantify the signature peptide or LBA after selective purification with anti-idiotypic antibodies would be suitable (Figure 2). 2

28 Chapter 2 Figure 2. General recommendation for therapeutic mAb quantification depending on instruments and materials available and on assay requirements; Quantification on the basis of peptide (bottom-up) using LC-MS/MS or LC-HRMS and intact on the basis of middle-down and top-down strategies with LC-HRMS or intact on the basis of LBA, HPLC-FLD. Bottom-up quantitative proteomics is preceded by denaturation and enzymatic digestion of the therapeutic mAb which releases numerous peptides of different chain lengths. Peptides that are unique for the mAb (signature peptides) are selected for measurement. These peptides are easily and efficiently separated using a standard reverse-phase HPLC system and thereafter quantified on a standard triple quadrupole mass spectrometer. In general, the chromatographic peak shape of the peptides is more symmetrical compared to the peak shape of proteins due to fewer secondary interactions on the stationary phase. Top-down and middle-down quantitative proteomics is based on the measurement of intact or semi-intact proteins. Large proteins such as mAbs can shift to high charge states during electrospray ionization yielding a mass to charge (m/z) ratio within the working range ∼1800 − 4000 of a high resolution mass spectrometers (HRMS) such as Orbitrap or Time-Of-Flight (TOF). These methods do not require protein unfolding and enzymatic digestion which can be challenging and time-consuming to optimize.

29 Tutorial on LC-MS/MS methods quantifying mAbs However, top-down and middle-down methods do have some limitations. Firstly, targeted sample purification with anti-idiotypic antibodies or ligands is necessary due to structural similarities between the therapeutic mAb and endogenous IgG. Secondly, the required HRMS apparatus is expensive compared to the triple quadrupole mass spectrometer. Finally, attaining the required lower limit of quantifications (LLOQ) can be challenging because of wider precursor charge distribution, broadened chromatographic peak shape of large proteins, and mAb glycoform heterogeneity [39-41]. Some additional steps can be implemented to gain higher signals and thus lower LLOQ’s. For example, human immunoglobulin G1 in rat serum was successfully quantified with LC-HRMS after target specific purification with anti-idiotypic antibodies followed by deglycosylation [32]. Here, a remarkable LLOQ of 0.1 mg/mL was achieved through this strategy by utilizing a high sample volume (50 mL) in combination with high volume injection (60 mL) and 1mm diameter analytical column. Top-down quantification can also be performed by HPLC coupled to fluorescence detector (FLD). For example, intact trastuzumab, bevacizumab and infliximab in human serum were successfully quantified by HPLC-FLD after targeted purification [42, 43]. Unlike HRMS methods, FLD methods are not affected by signal dilution caused by charge distributions. However, fluorescents measurements have a low specificity. Most proteins have similar excitation and emission spectra which result in noisy and overlapping chromatographic peaks. Furthermore, because of the lower sensitivity, a higher sample volume (~100 mL) is required which limits its applicability. Middle-down strategies can also be used to reduce precursor charge distribution found in intact analysis. Here, only a portion of the mAb is measured such as the light chains after dithiothreitol (DTT) reduction, or Fab regions after limited Lys-C digestion [44, 45]. In contrast to intact mAb measurement, these regions are smaller and are usually free from glycan chains, leading to fewer precursor ions resulting in an increase in signal intensity of the mAb. As can be seen from Figure 2, peptide level quantification via LC-MS/MS or LC-HRMS instruments have low requirement and are thus frequently employed. This approach has been extensively used for numerous mAbs with great success. However, for some fully human/humanized therapeutic mAbs, quantification via signature peptide can be challenging because of the human polyclonal serum background [45]. In these situations, a targeted purification followed by intact or peptide level quantification might be preferred. Bottom-up quantification Currently, bottom-up quantification of mAbs in biological matrices using signature peptides is the most common approach. This principle offers fast, easy and flexible method development with high detection sensitivity using standard triple quadrupole mass spectrometers. 2

30 Chapter 2 Table 1. Published methods for peptide level quantification of therapeutic mAbs with LC-MS/MS. Sample type Internal standard used Rodent Monkey Human Analogue protein SIL peptide SIL protein Extended SIL peptide Dimethyl labeled [33, 47-53] [21, 54-64] [34, 65-85] [47, 63, 76, 79-81, 83] [34, 49, 53, 54, 57, 60, 64, 67-71, 73, 75, 78, 85] [21, 33, 48, 51, 52, 56, 59, 62, 72, 74, 77, 82] [58, 61, 65, 73] [55] Sample purification None (whole digest) Albumin depletion Pellet Digestion Protein A Protein G Anti-human FC antibody Anti-Idiotypic antibody Ligand [21, 47, 55, 60, 73, 78] [33, 54] [49-52, 57, 58, 61, 63, 64, 66, 67, 75, 85] [45, 53, 68-71, 80, 84] [65, 76, 81] [48, 56, 59, 60, 62] [34, 65] [72, 74, 77, 79, 82, 83] Digestion by denaturation Digestion by reduction & alkylation Guanidine Urea Surfactant TFE None TCEP and IAA DTT and IAA DTT [47] [21, 33, 48, 54-56, 59, 74, 82] [34, 50, 60, 62, 73, 78, 85] [66] [49, 51, 57, 64, 67, 71, 75, 82, 83] [33, 48, 60, 65, 77] [21, 34, 47, 50, 52-56, 58, 59, 61-63, 66, 73, 76, 78, 80, 81, 85] [72, 79] Post digest SPE cleanup Mass spectrometer LLOQ method performance RP Ion exchange Triple Quad Q-Trap HRMS ≤0.5mg/L ≤1 mg/L >1mg/L [33, 47, 48, 75, 78] [47, 52, 60, 61, 63, 73] [21, 34, 47, 50, 51, 53, 55, 56, 58-62, 64-68, 71-76, 79-81, 85] [33, 48, 49, 54, 57, 63, 78, 82] [52, 77, 83] [34, 47, 48, 50, 53, 56, 59, 62, 65, 67, 71, 77, 84, 86] [49, 52, 55, 57, 65, 68, 74, 75, 81-83, 85] [21, 33, 51, 54, 58, 60, 63, 64, 66, 73, 76, 78, 80] Abbreviations: SIL: stable isotopically labeled; TFE: trifluoroethanol; TCEP: tris(2-carboxyethyl)phosphine; IAA: iodoacetamide; DTT: dithiothreitol; SPE: solid phase extraction; RP: reversed phase.

31 Tutorial on LC-MS/MS methods quantifying mAbs Generally, these published methods share similarities as they all include steps depicted in Figure 3. The major differences between these methods are the selection of internal standard, sample processing and digestion conditions (Table 1). The merits and drawbacks of various options in each steps of method development will be discussed in next sections. Figure 3. General workflow to develop quantitative LC-MS/MS method to measure therapeutic mAbs. Signature peptide selection The first step in method development is selection of unique signature peptides to serve as surrogate for quantification. The peptide sequence of the therapeutic mAb is essential for this step. Sequences of approved therapeutic mAb can be found in the Immunogenetics Information system® (http://www.imgt.org/) or in Drugbank (http://www.drugbank.ca). For the quantification of chimeric mAbs in human serum, tryptic peptides from the entire variable region can be chosen and targeted. However, for human or humanized mAbs the choice is limited to the complementarity-determining regions (CDRs) of which there are six in the variable light and heavy chains. In silico tryptic digestion can be performed manually or by using an online tool ‘Protein Prospector’ (http://prospector. ucsf.edu). The peptides generated can be screened online with protein Blast® software (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The peptide sequence is compared against the target organism (biological matrix of the analyte) from an appropriate database such as UniProtKB/Swiss-prot. Any peptide scoring under 100% for either the query cover or the positive identification, represents a unique peptide and as such a potential quantifier. Thereafter, a list of potential signature peptides can be screened for amino acid stability. Amino acids such as cysteine and methionine are susceptible to oxidation leading to mass shifts of +16, +32, +48 Da depending on the number of oxidation reactions. Asparagine followed by a smaller amino acid such as glycine can readily be deamidated into aspartic acid, isoaspartic acid and succinimide during tryptic digestion, causing a mass shift of +0.98, +0.98, and -17.03 Da respectively. Also N terminal glutamine cyclization can occur at alleviated pH and prolonged digestion time. Here, the loss of ammonia leads to a mass shift of -17.03 Da [87]. Preferably, these amino acids should be avoided. However, when stable alternatives are unavailable, these peptides can still be used if a stable isotopically labeled (SIL) internal standard is included. For example, infliximab, somatropin and nivolumab quantification with signature peptides containing a methionine were previously reported [47, 66, 68]. Also, endogenous insulin-like growth factor 1 was successfully quantified in serum with a signature peptide containing two cysteines that were protected from oxidation by attaching a carboxymethyl group on the thiol moiety through iodoacetamide alkylation and acid hydrolysis following disulfide bond 2

32 Chapter 2 reduction [88]. Nevertheless, care must be taken when these peptides are used since in vivo degradation of the therapeutic mAb via the above mentioned pathways can lead to an underestimation of circulating mAb as was demonstrated by Bults and colleagues in their research into deamidation of trastuzumab in human plasma [67]. After a list of potential signature peptide candidates is composed, a tryptic digest of the stock standard is preformed and run on LC-MS/MS to identify the various eluting peptides. An example of this is illustrated in Figure 4 where a number of infliximab peptides were identified using a shallow gradient (5-50% acetonitrile 0.1% formic acid in 20 min). It is important to note that due to the slow scan speed of the triple quadrupole mass spectrometer a high stock standards concentration (~100 mg/mL) should be digested. Figure 4. Signature peptides identification of infliximab tryptic digest with a triple quadrupole operating in full scan MS (300-1500 m/z) (A), GLEWVAEIR precursor elucidation though comparison with theoretical mass (B) and finally, conformation of the precursor sequence through fragmentation (20 eV) of [Mþ2H]2þ and product ion scan (150-1500 m/z) (C). Internal standard selection Methods that include an IS are able to correct for various factors causing variability during sample analysis, such as component losses during sample preparation as well as

33 Tutorial on LC-MS/MS methods quantifying mAbs instrument related factors such as injection, ionization and fragmentation. Depending on the IS used, different levels of corrections can be obtained (Table 2). Also in ELISA there have been attempts to incorporate IS in the assays [89]. However, the correction ability here is limited to dilution corrections only. In LC-MS/MS, SIL proteins can correct for the entire sample pretreatment and analysis because of their matching amino acid sequence and conformational folding and are considered the gold standard in quantitative proteomics [21, 90]. Unfortunately, these SIL variants of therapeutic proteins are often very expensive and only a limited number are commercially available. As an alternative, Nouri-Nigjeh and colleagues have shown that hybrid calibration, which use the therapeutic mAb of interest as ‘calibrator’ in combination with a SIL peptide or extended-SIL peptide as IS, can obtain accurate and precise results in whole sample digestion methods [90]. This observation was also supported by Prasad and Unadkat stating that SIL peptides can be used when maximum trypsin digestion is ensured [91]. The largest source of variability in this type of work-up originates from ionization suppression due to sample complexity. In contrast to the calibrator protein, flanking SIL peptides and extended peptides are easily dissolved in the sample matrix and due to the lack of structural folding and S-S bonds, provide easier access to the cleavage site. Therefore, the correction for digestion efficiency is expected to be very low using this approach. Furthermore, experiments performed in house using a regular SIL peptide and a SIL extended peptide showed that SIL peptide performed better than SIL extended peptide since the SIL extended peptide produced additional variability during digestion that was not correlated to variability found in calibrator protein digestion. Dimethyl labeling was used by Ji and coworkers and was found to be a cheap way of generating multiple labeled peptides from the protein calibrator [55]. However, reaction conditions need to be carefully optimized to obtain maximum labeling efficiency. This principle has not gained a lot of ground since cheap SIL labeled peptides with high purity can easily be obtained. Analogue proteins are also frequently used to correct for sample purification and digestion (Table 1). However, the peptides generated from these analogue proteins are not identical to the signature peptides of the target mAb. Therefore, differences in charge and or hydrophobicity could lead to suboptimal correction for clean-up and enrichment steps. Moreover, differences in protein folding, solubility and disulfide bond location between the calibrator protein and the analogue may only result in moderate correction for digestion if preceding protein reduction, alkylation and denaturation was suboptimal. Matrix effect correction for ionization relies on the elution order of the signature peptide and the IS. So, depending on the elution similarities between the signature peptide and the analogue peptide, varying levels of corrections can be achieved. This is also true for fragmentation correction, here similarities in amino acid sequences between the signature peptide and the analogue determine the levels of correction. Nevertheless, Li and colleagues have shown that when a selective purification is used, analogue proteins can perform better than SIL peptide or a SIL flanking peptide [56]. Here, sample recovery was the major contributor to the method error and therefore, by including an analogue protein that can experience the same losses as the calibrator protein, correction was achieved. Furthermore, variability in LC-MS/MS analysis is 2

Made with FlippingBook

RkJQdWJsaXNoZXIy MTk4NDMw