Pharmaco-epidemiological assessment of the clinical use of biologic therapies in the management of rheumatoid arthritis

Transcription

1 Pharmaco-epidemiological assessment of the clinical use of biologic therapies in the management of rheumatoid arthritis A thesis submitted to The University of Manchester for the degree of Doctor of Philosophy in the Faculty of Medical and Human Sciences 2011 Moetaza Mahmoud Hassab Elsayed Soliman School of Pharmacy and Pharmaceutical Sciences Drug Usage and Pharmacy Practice Group

2 List of Contents List of Contents... 2 List of Tables List of Figures Abstract Declaration Copyright statement Acknowledgments The author Dedication List of publications List of Abbreviations Glossary Introduction Role of clinical pharmacy in evaluating drug therapies Measuring efficacy and safety of new drugs Clinical trials Pharmaco-epidemiological studies Rheumatoid arthritis Clinical features Diagnosis Epidemiology Aetiology

3 1.3.5 Pathogenesis Outcome measures in RA in clinical trials and clinical care Disease activity measures Disease activity score (DAS) The European League Against Rheumatism (EULAR) The American College of Rheumatology (ACR) response Patient s reported outcomes Health Assessment Questionnaire (HAQ) European Quality of life 5 Dimensions (EQ-5D) Short Form 36 (SF-36) Management of RA Analgesics and non-steroidal anti-inflammatory drugs (NSAIDs) Glucocorticoids Non-biologic disease-modifying anti-rheumatic drugs (nbdmards) Biologic therapies Rituximab (RTX) Structure Mechanism of action of RTX License and approval of RTX for RA Literature review Assessment of drug persistence with anti-tnf therapies in RA Introduction Method

4 2.1.3 Anti-TNF therapies overall drug persistence A comparison of drug persistence between the three anti-tnf therapies The effect of nbdmard co-therapy on drug persistence comparing across the three anti-tnf therapies Other predictors of anti-tnf therapies discontinuation Persistence with second alternative anti-tnf therapies after switching from one anti- TNF to another Discussion Assessment of the use of RTX in RA from both randomised controlled trials (RCTs) and observational studies: literature review Randomised controlled clinical trials (RCTs) of RTX in RA Efficacy of RTX in RA patients from RCTs Efficacy of RTX mono-therapy and with concomitant nbdmards Efficacy of RTX in anti-tnf naïve and in anti-tnf failures Efficacy of RTX in RF positive versus negative patients Efficacy of different doses of RTX Efficacy of repeated doses of RTX On-going RCTs of RTX in RA Summary Observational studies of RTX in RA Effectiveness of RTX in RA patients from the observational studies Effectiveness of RTX mono-therapy and with concomitant nbdmard Effectiveness of RTX in nbdmards failures and in anti-tnfs failures Effectiveness of different doses of RTX

5 Effectiveness in RF positive versus RF negative patients Predictors of response to RTX Effectiveness of repeated doses of RTX Discussion and conclusions Findings from RCTs Findings from observational studies PhD aim and objectives Introduction Aim General Objectives Methodology Study design Sample size Ethical approval Inclusion criteria Patient recruitment Baseline characteristics Follow-up Data management BSRBR recruitment rates BSRBR RTX cohort Switching therapy within the BSRBR Baseline characteristics

6 5. Drug persistence with anti-tnf therapies in patients with rheumatoid arthritis Aim and objectives Methods Inclusion and exclusion criteria The cohorts used in the analysis Baseline characteristics Definition of persistence Statistical analysis Results Patients included Identifying the seven cohorts of the baseline concomitant nbdmards The cohorts used in the analysis Overall anti-tnf therapies cohort Baseline characteristics Percentage of patients who stopped the medication Overall persistence with anti-tnf therapies Predictors of anti-tnf therapy discontinuation Comparing drug persistence across the three anti-tnf therapies Baseline characteristics Percentage of patients who stopped the medication Persistence with the three cohorts of the three anti-tnf therapies Effect of baseline concomitant nbdmard therapy on persistence with anti -TNF therapies Baseline characteristics

7 Percentage of patients who stopped the medication Persistence with the anti-tnf therapies across the seven cohorts of baseline concomitant nbdmards therapy Summary Discussion Choice of treatment following failure a first anti-tnf therapy: rituximab or a second alternative anti-tnf therapy Aim and objectives Methods Inclusion and exclusion criteria Identification of patients Outcome measures Baseline and follow-up data collection Baseline characteristics Statistical analysis Results Patients included Assessment of clinical effectiveness Baseline characteristics Comparing crude clinical effectiveness Adjusted comparison of clinical effectiveness Assessment of patient reported physical function Baseline characteristics Comparing crude improvements in physical function

8 Adjusted comparison of improvements in physical function Summary Discussion Assessment of effectiveness of rituximab in rheumatoid arthritis Aim and objectives Methods Inclusion and exclusion criteria Outcome measures Baseline characteristics Statistical analysis Results Patients cohorts Assessment of clinical effectiveness Baseline characteristics Overall clinical effectiveness to RTX therapy Comparing clinical effectiveness of RTX according to previous anti-tnf history Comparing clinical effectiveness of RTX according to RF status Comparing clinical effectiveness of RTX according to baseline concomitant nbdmards Predictors of clinical effectiveness of RTX Assessment of patient reported physical function and quality of life after RTX therapy Baseline characteristics Overall patient reported physical function and quality of life at six months

9 Comparing patient reported physical function and quality of life according to previous anti-tnf history Comparing patient reported physical function and quality of life according to RF status Comparing patient reported physical function and quality of life according to baseline concomitant nbdmards Predictors of patient reported physical function in anti-tnf failure patients Summary Discussion Discussion and conclusions Persistence with anti-tnf therapies in RA patients Comparative effectiveness of RTX versus a second alternative anti-tnf Effectiveness of RTX in RA patients Originality of the research Strengths Broad inclusion criteria of the patients Large sample size Limitations Lack of randomisation Time of data collection Limitations related to statistical analysis Other limitations Final conclusions Recommendations for clinical practice

10 8.9 Recommendations for future work References Appendices Appendix 2.1: Search strategy used to identify studies that looked at the persistence with the anti-tnf in RA Appendix 2.2: Search strategy used to identify RCTs of RTX in RA Appendix 2.3: Search strategy used to identify observational studies of RTX in RA Appendix 4.1: The original ethical approval for the BSRBR Appendix 4.2: Extension of ethical approval for recruitment of RTX patients Appendix 4.3: Consultant baseline questionnaire Appendix 4.4: Prior biologic exposure form Appendix 4.5: Short baseline questionnaire Appendix 4.6: Patient baseline questionnaire Appendix 4.7: Health assessment questionnaire (HAQ) Appendix 4.8: European quality of life 5 dimension (EQ-5D) form Appendix 4.9: The consultant follow-up questionnaire Appendix 4.10: The patient follow-up questionnaire Appendix 7.1: EQ-5D utility score algorithm (34) Word count: 57,203 10

11 List of Tables Table 1.1: The 1987 revised criteria for RA classification Table 1.2: The 2010 American College of Rheumatology/European League Against Rheumatism classification criteria for RA Table 1.3 : Formula for calculation of DAS, DAS28 and conversion from DAS to DAS Table 1.4 : The EULAR response criteria classified as good, moderate or no response Table 1.5: SF-36 taxonomy Table 1.6: Licensing details of the biologic therapies for use in RA (from FDA and EMA websites) Table 1.7: NICE guidance for the use of adalimumab, etanercept and infliximab therapies in RA 48 Table 1.8: NICE guidance for the use of biologic drugs after failing anti-tnf therapy in RA Table 2.1: Published studies on persistence with anti-tnf therapies in RA Table 2.2: Overall persistence rates of anti-tnf therapies first course in RA Table 2.3: Persistence rates of the three anti-tnf therapies in RA with no consideration of the presence or absence of the concomitant nbdmards therapy (from the literature) Table 2.4: Persistence rates with anti-tnf therapies when taken alone, in combination with MTX or in combination with other nbdmards Table 2.5: Other predictors of anti-tnf therapies discontinuation (from the literature) Table 2.6: Design characteristics of the RCTs of first course of RTX in RA Table 2.7: Baseline characteristics of patients included in the RCTs of RTX in RA Table 2.8: Change from baseline in response rates at week Table 2.9: Change from baseline in patients quality of life outcomes reported in the RCTs (24 week endpoint) Table 2.10: General design characteristics of the RCTs of second course of RTX in RA Table 2.11: General design characteristics of observational studies of RTX in RA Table 4.1: Baseline characteristics of the patients receiving biologic therapies in the BSRBR Table 5.1: Patients included in the analysis

12 Table 5.2: Baseline concomitant nbdmards Table 5.3: The single, double and triple use of concomitant nbdmards at baseline Table 5.4: The seven cohorts of baseline concomitant nbdmards considered for the analysis.125 Table 5.5: Patient characteristics of the total anti-tnf cohort Table 5.6: Number of patients who stopped the medication in the entire anti-tnf cohort Table 5.7: Survivor function (95% CIs) for the overall anti-tnf therapy persistence Table 5.8: Hazard ratios (95 % confidence intervals) in the univariate and multivariate Cox proportional hazards analysis of anti-tnf therapies discontinuation Table 5.9: Patient characteristics of the three cohorts of the three anti-tnf therapies Table 5.10: Number of patients who stopped the medication in the three anti-tnf cohorts Table 5.11: Survivor function (95% CIs) for patients continuing anti-tnf therapies Table 5.12: Hazard ratios (95 % confidence intervals) in the unadjusted and adjusted analyses for anti-tnf therapy discontinuation Table 5.13: Baseline demographic characteristics for the seven cohorts according to the baseline nbdmards Table 5.14: Baseline disease characteristics for the seven cohorts according to the baseline nbdmards Table 5.15: Number of patients who stopped the medication in the seven cohorts of the baseline concomitant nbdmards Table 5.16: Survivor function (95% CIs) for patients who continued anti -TNF therapies medication in the seven cohorts of the baseline nbdmards Table 5.17: Hazard ratios (95 % confidence intervals) in the unadjusted and adjusted analysis of anti-tnf therapies discontinuation Table 5.18: Adjusted hazard ratios (95 % confidence interval) for anti-tnf therapy discontinuation (within each anti-tnf agent) Table 6.1: Baseline characteristics (at time of switching) for patients with DAS28 data Table 6.2: Six months clinical effectiveness

13 Table 6.3: Undjusted and adjusted regression analyses results for the six months EULAR response Table 6.4: Baseline characteristics at time of switching for patients with HAQ scores Table 6.5: Six month patient reported physical function Table 6.6: Adjusted and unadjusted regression of six months patient s reported physical functio n Table 7.1: Patients included in the analysis Table 7.2: Baseline characteristics of RTX patients with baseline and six month DAS28 scores..180 Table 7.3: Clinical effectiveness of rituximab six months after starting the therapy Table 7.4: Previous anti-tnf history of the anti-tnf failure patients Table 7.5: Baseline characteristics of RTX patients according to previous anti-tnf history Table 7.6: Clinical effectiveness of rituximab according to previous anti-tnf history Table 7.7: Clinical effectiveness of rituximab according to rheumatoid factor status Table 7.8: Response to rituximab according to the baseline concurrent nbdmards Table 7.9: Beta-coefficients (95% CI) for linear regression models of change in disease activity (change in DAS28) Table 7.10: Relative risk ratios (95% CI) for multinomial regression analysis of EULAR response Table 7.11: Odds ratios (95% CI) of logistic regression analysis of achieving disease remission..190 Table 7.12: Undjusted and adjusted regression models of six months response in anti-tnf failures versus anti-tnf naïve patients Table 7.13: Baseline characteristics of the patients Table 7.14: Improvements in quality of life and physical function six months after starting RTX Table 7.15: Changes in the five domains of EQ-5D profile six months after starting RTX Table 7.16: Baseline characteristics of the patients Table 7.17: Improvements in quality of life and physical function according to previous anti-tnf history

14 Table 7.18: Changes in the five domains of EQ-5D profile according to previous anti-tnf history Table 7.19: Improvements in quality of life and physical function according to RF status Table 7.20: Improvements in quality of life and physical function according to baseline concomitant nbdmards Table 7.21: Univariate and multivariate regression models of predictors of physical function at six months (change in HAQ) Table 7.22: Univariate and Multivariate regression models of predictors of six months physical function (Achieving MCID in HAQ) Table 7.23: Adjusted and unadjusted regression results for changes in physical function in anti- TNF failures versus anti-tnf naïve patients

15 List of Figures Figure 1.1: Schematic diagram showing the structure of RTX Figure 1.2: Diagram showing the binding of RTX to CD20 on the surface of the B cell Figure 1.3: Schematic diagram showing the mechanism of action of RTX Figure 2.1: Patients achieving ACR20 response at 24 weeks in the RCTs of RTX in RA. Results from the RCTs of RTX in RA Figure 2.2: Moderate or good EULAR response at 24 weeks in the RCTs of RTX in RA. Results from RCTs of RTX in RA Figure 2.3: Patients achieving ACR20, 50 and 70 responses in the DANCER trial and SERENE trials Figure 4.1: Key timelines in the BSRBR Figure 4.2: Baseline data collection within the BSRBR Figure 4.3: Recruitment rates of anti-tnf cohorts in the BSRBR over time Figure 4.4: Recruitment rates of RTX cohort stratified by type of registration Figure 4.5: Flow chart of RTX patients in the BSRBR (by 21 st December 2010) Figure 5.1: Kaplan-Meier plot of persistence with anti-tnf therapies in total (green) and when the reason for discontinuation was inefficacy (blue), or adverse events (red) Figure 5.2: Kaplan-Meier plot of persistence with each of the three anti-tnf therapies; etanercept (red), adalimumab (blue) and infliximab (green) Figure 5.3: Kaplan-Meier plot for persistence with anti-tnf therapy in the seven cohorts of the baseline concomitant nbdmards Figure 6.1: Flow diagram of patients included in the comparative analysis Figure 7.1: EQ-5D profile at baseline and six months after starting RTX

16 - The University of Manchester - Moetaza Mahmoud Hassab Elsayed Soliman - Doctor of Philosophy - Pharmaco-epidemiological assessment of the clinical use of biologic therapies in the management of rheumatoid arthritis - 08/12/2011 Abstract Objectives: The aim of this thesis was to evaluate the clinical use of rituximab (RTX) in rheumatoid arthritis (RA) patients treated in routine clinical practice in the UK, taking account of the previous therapies (anti-tumour necrosis factor (anti-tnf) therapies). Methods: The analysis involved RA patients registered with the British Society for Rheumatology Biologics Register. Kaplan-Meier survival analysis was used to study the persistence with anti- TNF therapies. Drug persistence was compared across the anti-tnf therapies and according to the most common concomitant non-biological disease modifying anti-rheumatic drugs (nbdmards) for up to five years. Adjusted multivariate Cox proportional hazard models were used to compare drug persistence across the subgroups. Change in Disease Activity Score (DAS28) and European League Against Rheumatism (EULAR) response were used to assess the clinical effectiveness of RTX while change in Health Assessment Questionnaire (HAQ) score was used to assess the physical function of the patients six months after starting RTX. Logistic regression was used to compare EULAR response and achieving a minimal clinically important difference (MCID) in HAQ (at least 0.22) between patients who started RTX and those who switched to a second alternative anti-tnf after failing a first anti-tnf therapy. Multivariate regression analyses were used to identify factors associated with the clinical effectiveness and the improvements in physical function six months after starting RTX. The models included baseline demographics, disease characteristics, baseline quality of life and previous drug history. Results: Among 10,396 RA patients receiving their first anti-tnf therapy, five-year drug persistence (95% CI) was 42% (41%: 43%). Infliximab was the most likely discontinued anti-tnf therapy. Compared to methotrexate (MTX), patients receiving no nbdmard, leflunomide or sulfasalazine were more likely to discontinue their first anti-tnf therapy while patients receiving MTX in combination with other nbdmards showed superior persistence with their anti -TNF therapy. After failing the first anti-tnf therapy, patients who switched to RTX were significantly more likely to achieve EULAR response and MCID in HAQ; odds ratio (95% CI) 1.31 (1.02: 1.69) and 1.49 (1.07: 2.08) respectively compared to those who switched to an alternative anti-tnf therapy. In a cohort of 646 RA patients who started RTX, the mean DAS28 scores significantly improved six months after starting RTX with mean (95% CI) change of (-1.53: -1.30). 17% of the patients achieved a good EULAR response and 43% achieved a moderate response. Higher baseline DAS28 score and positive rheumatoid factor (RF) status were significantly associated with a decrease in disease activity, while females and patients with higher baseline HAQ score were less likely to respond to RTX. In a cohort of 484 patients receiving RTX, the mean HAQ scores significantly improved by (-0.16: -0.09) units. High baseline HAQ score was significantly associated with six months improvement in HAQ. Older patients, females, current smokers and patients receiving concurrent steroids were less likely to show an improvement in their HAQ scores. Conclusions: Compared to MTX, concomitant use of two or three nbdmards combinations including MTX with anti-tnf therapies resulted in better persistence with the anti-tnf therapies. After failing the first anti-tnf therapy, starting RTX may be of more benefit than switching to an alternative anti-tnf therapy. RTX has proven to be effective in improving both the clinical and patients reported outcomes in routine clinical practice in the UK. Response to RTX was influenced by baseline DAS28, RF status, baseline HAQ, age, gender, smoking, and concurrent use of steroids. 16

17 Declaration No portion of the work referred to in this thesis has been submitted in support for an application for another degree or qualification of this or any other university or other institute of learning. Copyright statement 1. The author of this thesis (including any appendices and/or schedules to this thesis) owns certain copyright or related rights in it (the Copyright ) and s/he has given The University of Manchester certain rights to use such Copyright, including for administrative purposes. 2. Copies of this thesis, either in full or in extracts and whether in hard or electronic copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as amended) and regulations issued under it or, where appropriate, in accordance with licensing agreements which the University has from time to time. This page must form part of any such copies made. 3. The ownership of certain Copyright, patents, designs, trademarks and other intellectual property (the Intellectual Property ) and any reproductions of copyright works in the thesis, for example graphs and tables ( Reproductions ), which may be described in this thesis, may not be owned by the author and may be owned by third parties. Such Intellectual Property and Reproductions cannot and must not be made available for use without the prior written permission of the owner(s) of the relevant Intellectual Property and/or Reproductions. 4. Further information on the conditions under which disclosure, publication and commercialisation of this thesis, the Copyright and any Intellectual Property and/or Reproductions described in it may take place is available in the University IP Policy (see in any relevant Thesis restriction declarations deposited in the University Library, The University Library s regulations (see and in The University s policy on presentation of Theses. 17

18 Acknowledgments In the first place, I would like to acknowledge my deep thanks to ALLAH the merciful for the endless giving that helped me to carry out this work starting from the opportunity itself and ending with all the abilities to finish it. My deep gratitude also goes to the Egyptian government for funding my PhD in the UK. I would like to thank Professor Darren Ashcroft and Dr Kimme Hyrich for their supervision of this PhD, giving me all their support, guidance and advice. Our continuous regular meetings helped this PhD to develop in steady way achieving progress by each meeting. They always encouraged me to think about presenting my work in conferences and guided me throughout the PhD. They also encouraged me to draft my results in the form of academic papers which were then submitted for publication. I would like to thank all study members of the British Society of Rheumatology Biologics Register (BSRBR). I am grateful to Professor Deborah Symmons for help and support. She kindly examined my first year continuation report. My thanks also go to Dr Kath Watson for offering access to the BSRBR that helped me in conducting my research. I want to thank Dr Mark Lunt for his helpful statistical guidance. I would like also to thank all the patients and rheumatologists who participated in the BSRBR. Special thanks to PhD students who were using data from the BSRBR; James Galloway and Louise Mercer. I am grateful to the Faculty of Medical and Human Sciences training team for postgraduate training programmes which helped me so much in improving my academic skills such as academic writing and presentations. My thanks go also to the School of Pharmacy and Pharmaceutical Sciences for their school seminars and postgraduate away days which were really useful. Special thanks to all PhD students within the Drug Usage and Pharmacy Practice Group for answering my questions, listening to my presentations, sharing PhD forums, discussions and cooperation. Finally, I would like to thank my parents, my husband and my children for understanding that sometimes I just disappear working on my PhD. 18

19 The author I am an Egyptian citizen who was born in March I graduated from the Faculty of Pharmacy, Mansoura University in May 1997 and was awarded a BSc in Pharmaceutical Sciences (with honours). After graduation I worked as a demonstrator in the Faculty of Pharmacy, Mansoura University. In October 2006 I was awarded MSc degree in Pharmaceutical Sciences from the same university. During my Master s degree, I studied the preparation and evaluation (in-vitro and in-vivo) of different pharmaceutical formulations containing different drugs. In 2007, I applied for a grant to undertake an international PhD in clinical pharmacy. In September 2008, I started my PhD at the School of Pharmacy and Pharmaceutical Sciences at the University of Manchester. Dedication To my respectful parents, my beloved husband, and my dear children 19

20 List of publications Published articles Soliman M., Ashcroft D., Watson K., Lunt M, Symmons D., and Hyrich K., on behalf of the British Society for Rheumatology Biologics Register. Impact of concomitant use of DMARDs on the persistence with anti-tnf therapies in patients with rheumatoid arthritis: Results from the British Society of Rheumatology Biologics Register. Ann Rheum Dis 2011;70 (4): Permission for the re-use of this article in the current PhD was granted by the journal in 30 th August Soliman M., Hyrich K., Lunt M, Watson K., Symmons D., and Ashcroft D., on behalf of the British Society for Rheumatology Biologics Register. Effectiveness of rituximab in patients with rheumatoid arthritis: observational study from the British Society of Rheumatology Biologics Register. Journal of Rheumatology (accepted for publication on 16 th August 2011). Submitted articles Soliman M., Hyrich K., Lunt M, Watson K., Symmons D., and Ashcroft D. on behalf of the British Society for Rheumatology Biologics Register. Rituximab or a second anti -TNF therapy for rheumatoid arthritis patients who have failed their first anti-tnf? Comparative analysis from the British Society for Rheumatology Biologics Register. Submitted to Arthritis Care and Research (a revised version was resubmitted in December 2011) Soliman M., Hyrich K., Lunt M, Watson K., Symmons D., and Ashcroft D. on behalf of the British Society for Rheumatology Biologics Register. Impact of Rituximab on quality of life and physical function of patients with rheumatoid arthritis: prospective observational study from the British Society for Rheumatology Biologics Register. Submitted to Rheumatology. 20

21 Abstracts presented at conferences Soliman M., Ashcroft D., Watson K., Lunt M, Symmons D., and Hyrich K. on behalf of the British Society for Rheumatology Biologics Register. Benefit of Concomitant DMARD Use on the Persistence with Anti-TNF Therapies in Rheumatoid Arthritis: Results from the British Society of Rheumatology Biologics Register (BSRBR). Rheumatology (2010) 49(suppl 1): i3-i5 Soliman M., Ashcroft D., Watson K., Lunt M, Symmons D., and Hyrich K. Effectiveness of rituximab in rheumatoid arthritis: Results from the British Society of Rheumatology Biol ogics Register (BSRBR). Rheumatology (2011) 50 (suppl 3): iii31-iii34. Soliman M., Ashcroft D., Watson K., Lunt M, Symmons D., and Hyrich K. Predictors of response to rituximab in patients with rheumatoid arthritis: Results from the British Society of Rheumatology Biologics Register (BSRBR). Rheumatology (2011) 50(suppl 3): iii111-iii129 Soliman M., Hyrich K., Lunt M, Watson K., Symmons D., and Ashcroft D. Which is more effective, Rituximab or a second anti-tnf therapy for rheumatoid arthritis patients who have failed their first anti-tnf? Analysis from the British Society for Rheumatology Biologics Register (BSRBR). Pharmacoepidemiology and Drug Safety 2011; 20: S13. Soliman M., Hyrich K., Lunt M, Watson K., Symmons D., and Ashcroft D. Impact of rituximab on the quality of life and physical function of patients with rheumatoid arthritis: results from the British Society for Rheumatology Biologics Register. Arthritis & Rheumatism (2011); 63 (10 Suppl). 21

22 List of Abbreviations ACPA ACR ADA ANOVA Anti-TNF AS BSR BSRBR CI CRP DAS28 EMA EQ-5D ESR ETN EULAR FDA HAQ HCQ HR Anti-citrullinated protein antibody American College of Rheumatology Adalimumab One-way analysis of variance Anti-tumour necrosis factor Ankylosing spondylitis British Society for Rheumatology British Society for Rheumatology Biologics Register Confidence interval C-reactive protein Disease activity score 28 joints European Medicines Agency European Quality of Life 5 Dimensions Erythrocyte sedimentation rate Etanercept European League Against Rheumatism Response Criteria Food and Drug Administration Health Assessment Questionnaire Hydroxychloroquine Hazard ratios IL-1 Interleukin 1 IL-6 Interleukin 6 INF IQR LEF Infliximab Inter quartile Range Leflunomide 22

23 MCID MCP MREC MTP MTX nbdmards NHS NICE NSAIDs OR PIP PsA RA RCTs RF RTX SD SF-36 SJC SSZ TJC UK VAS Minimum clinically important difference Metacarpophalangeal joint Multicentre Research Ethics Committee Metatarsophalangeal joint Methotrexate Non-biologic disease modifying anti-rheumatic drugs National Health Service National Institute for Health and Clinical Excellence Non-steroidal anti-inflammatory drugs Odds ratio Proximal interphalangeal joint Psoriatic arthritis Rheumatoid arthritis Randomised controlled trials Rheumatoid factor Rituximab Standard deviation Short Form-36 Swollen joint count Sulfasalazine Tender joint count United Kingdom Visual Analogue Scale 23

24 Glossary Betacoefficients Confounder DAS28 Disease remission EQ-5D profile EQ-5D utility score EULAR response HAQ In a linear model, a Beta- coefficient is the amount by which an outcome (dependent variable) increases if a predictor (independent variable) increases by one unit, provided that none of the other independent variables change. A variable that influences the strength of an association between an exposure and an outcome because it is associated with the exposure and the outcome. Disease activity score that includes 28 joints count (DAS28) is a disease activity measure in rheumatoid arthritis. A rheumatoid arthritis patient is considered in a state of disease remission if he/she achieved DAS28 score < 2.6. A descriptive system that includes 5 dimensions that define health (mobility, self-care, activities, pain/discomfort and anxiety/depression). A weighted index on a continuous scale (ranges from to 1.00) that describe general health. It was developed from weights for the different health states of the EQ-5D profile. The EULAR response criteria classify patients as good, moderate or nonresponders according to the change in their disease activity scores over time as well as the final disease activity score reached at the time of assessment. Health Assessment Questionnaire (HAQ) is an arthritis specific measure that evaluates patients physical function on a continuous scale (from 0 to 3) with lower scores representing less disability. Hazard ratio In a survival analysis, a hazard function is the risk of an event to occur at time t. the hazard ratio is the ratio of the hazard function of an event to occur in one group to the hazard function of the event in another group. MCID in HAQ Achieving minimum clinically important difference (MCID) in HAQ is defined as achieving at least a 0.22 unit improvements in HAQ score. nbdmards Odds ratio Persistence Non-biologic disease modifying anti-rheumatic drugs are the main group of treatments used for treating rheumatoid arthritis. The most commonly used nbdmards are methotrexate, hydroxychloroquine, sulphasalazine, and leflunomide. The odds ratio is the ratio of the odds of an event to occur in one group to the odds of it to occur in another group. It is used to describe the strength of association when the outcome (dependant variable) is a binary outcome. Persistence with a therapy is defined as the length of time the patients stayed on the therapy. 24

25 Propensity score Quality of life A propensity score is the probability of a patient being assigned to a particular treatment given the set of known baseline covariates. Propensity scores are used to overcome selection bias by equating the two groups of patients based on the baseline covariates. Quality of life is a term used to evaluate the general health and well-being of individuals. The measures of quality of life can be generic or disease specific. RTX Rituximab (RTX) is a chimeric monoclonal antibody which acts by depleting B- cells. It was licensed for use in rheumatoid arthritis in 2006 in both Europe and USA. 25

26 1. Introduction 26

27 This chapter provides a brief description of the methods used for the assessment of the efficacy of new drugs, and then focuses on the target disease considered in this PhD, namely Rheumatoid Arthritis (RA). The clinical presentation of the disease covers the definition, clinical symptoms, diagnosis, and management of RA. The chapter describes the outcome measures used to assess therapies used for the management of RA. Finally, more detailed information is provided on the target therapy of this PhD, Rituximab (RTX) including details about its market authorisation, chemical structure, use, and mechanism of action. 1.1 Role of clinical pharmacy in evaluating drug therapies Clinical pharmacy is defined as that area of pharmacy concerned with the science and practice of rational medication use. (1) Reviewing drug use which includes drug information, utilization, evaluation and selection, is one of the most important components of clinical pharmacy practice. The importance of reviewing drug use is particularly important when patients are receiving newly licensed drugs when limited information is known about their long term safety and effectiveness in routine clinical practice. In this case, the role of clinical pharmacy research in generation and application of new knowledge that contributes to improve health and quality of life comes to take place. (1) In the case of assessing the use of drugs in large populations, it becomes a pharmaco-epidemiological assessment (2) which comes under the epidemiological studies because it uses the techniques of epidemiology to study the use and the effects of drugs. 1.2 Measuring efficacy and safety of new drugs Drug discovery is a very long, difficult and expensive process. Measuring efficacy and safety of new drugs is the final step that comes after this long drug discovery process. Preclinical examination of safety and efficacy of new drugs starts in the laboratory either in-vitro or in-vivo with animals; a very wide range of dosing is used to collect preliminary information about the safety and efficacy of the tested drug. The clinical research process which examines safety and 27

28 efficacy of the medications intended for human use usually starts after the preclinical examination Clinical trials Before a new drug is licensed for use in humans, it must go through clinical trials which usually aim to study the efficacy and safety of the medication. Clinical trials are usually conducted after the preclinical confirmation of safety and efficacy of the tested drug. Comparisons to a placebo or other active medication may be considered. Phase 0 clinical trials are the first in human trials, carried out in one person at a time, which are known as micro-dosing studies which test if the drug behaves in humans as it was expected from preclinical studies. These trials give very preliminary data on the pharmacokinetics and pharmacodynamics of the tested drug. Phase I trials are carried out in a small number of healthy human subjects to investigate the safety, pharmacokinetics, pharmacodynamics, and metabolism of a new drug in the human body. (3) Ascending dosing of the drug is used to determine the most tolerable therapeutic dose in the human body. Phase I clinical trials are usually open label. Phase II trials are the trials in humans with a target disease. These are also the first to assess the efficacy and short term safety of a new drug. Different doses of the drug may be tested. (4) These trials may be randomised and may be blinded trials. Phase III randomised controlled trials (RCTs) are randomised double blinded placebo controlled trials which aim to assess the efficacy and safety of a new drug in patients with the target disease. These trials are considered the gold standard measure of drug e fficacy and short term safety and are required for the licensing and approval of use of new drugs. The license is usually accompanied by a plan for post marketing assessment of safety (risk management plan). 28

29 Randomisation should ensure that every patient has an equal chance to receive the medication while blinding means the patients do not know the medication they receive during the trial. Double blinding means both the patients and the assessors do not know the medication examined and for achieving this, the placebo and the drug should have the same appearance and dosing frequencies. Randomisation and blinding are used to decrease bias in the assessment of the drug therapy. Limitations of RCTs in the assessment of new drugs can be discussed in terms of e ffectiveness, safety, cost-effectiveness and convenience of the new drug to the patients. When a new drug is introduced to the market after license, often very little is known about its effectiveness compared with existing treatments. The RCTs undertaken for licensing purposes usually compare the new drug against placebo. Moreover, these RCTs usually assess the short term efficacy which may not reflect the long term use of the drug after license. Regarding the safety, rare and/or long term adverse events are usually not well studied in RCTs because at this stage the trials tend to be of short duration and usually a relatively small number of patients are enrolled. In addition, the patients enrolled in these trials are usually selected very carefully with exclusion of some categories of patients such as elderly patients, patients with other co-morbidities and patients receiving other medications. These excluded categories of patients are likely to be more susceptible to adverse events and excluding them may lead to an incomplete picture of the effectiveness and safety of the new medication in routine clinical practice (poor external validity). Cost-effectiveness of the new medication compared to other drugs present in the market including the cost of monitoring and the cost of treating adverse events caused by the medication is usually not assessed in RCTs conducted before license of a medication. As a result of all these limitations, post marketing evaluation of the new drugs is essential and has an important role in the overall assessment of new drugs. 29

30 Phase IV trials (Post marketing surveillance) often include the subgroups of patients that have been excluded from the RCTs. (5) The main aim of these trials is to assess the longer term safety of drugs and usually they can detect rare and long term adverse events. (6) Phase IV studies are considered important sources for post marketing surveillance. (5) Failure of a drug in terms of its long term safety in phase IV clinical trials or in post marketing surveillance in general may lead to withdrawal from the market. (7) Pharmaco-epidemiological studies Pharmaco-epidemiology is the study of the use of and the effects of drugs in large numbers of people. (2) These studies are post marketing studies that are usually non-experimental and observational which follow patients receiving the tested drug in routine clinical practice. It involves large cohort studies as registries and claims databases. The contribution of pharmacoepidemiology is mainly reporting in the effectiveness and safety of the drug(s) under evaluation. These studies usually add to what was found in premarketing studies, including the effectiveness of the drug in combination of other drugs or in patients with other diseases, the effectiveness in unstudied patients as elderly population, the discovery of uncommon or delayed effects of the drug, possible indications in new illnesses or new population of patients with the same targeted disease, the discovery of previously not reported adverse events or beneficial effects, economic implications, the effect of overdose, and the effectiveness compared to the already available possible alternatives to the drug. (2) Contributions to the general health improvements and patient quality of life beside the optimal use of the drug are also studied. The main aim of this PhD is to conduct and analyse a pharmaco-epidemiological study of the post marketing clinical effectiveness and use of rituximab (RTX); a drug recently licensed (2006) for use in the management of rheumatoid arthritis (RA). 30

31 1.3 Rheumatoid arthritis Clinical features RA is a systemic autoimmune disease which causes inflammation and damage of joints leading to patient disability. It can affect most joints in the body. The inflammatory reaction in the affected joint leads to joint swelling, stiffness (especially after a time of inactivity as in the morning) and pain. The pain is linked to the disease severity which varies accordingly. As the disease progress, this inflammation then invades and damages the cartilage and bone of the joint leading to deformity of the joint and patient disability. Weakness in the surrounding muscles, tendons and ligaments can also develop. Other symptoms of RA include fatigue, malaise, fever, weight loss and depression. (8) Other organs of the body may be affected (for example lung disease and skin nodules may also develop). RA can also cause more generalized bone loss that may lead to osteoporosis Diagnosis The diagnosis of RA cannot be confirmed by one single diagnostic test; however, laboratory measures such as serum rheumatoid factor (RF), erythrocyte sedimentation rate (ESR), C- reactive protein (CRP) and blood cell count can give some objective data for the diagnosis of RA as recommended by the American College of Rheumatology Subcommittee on RA. (9) The diagnosis of RA is derived from a cluster of clinical, laboratory and radiographic confirmations. Classification criteria of RA were initially developed in 1956 by the American Rheumatism Association (ARA). (10) In 1987, the revised criteria for classification of RA were developed based on an analysis of 262 patients with RA and 262 control subjects with no RA (but with other rheumatic diseases) (Table 1.1). (11) After the criticism of lack of sensitivity of the 1987 revised criteria in early disease, the 2010 RA classification criteria were developed in a collaboration of 31

32 both the American College of Rheumatology (ACR) and the European League Against Rheumatism (EULAR) (Table 1.2). (12) Table 1.1: The 1987 revised criteria for RA classification (11) Criterion Definition 1. Morning stiffness Morning stiffness in and around the joints, lasting at least 1 hour before 2. Arthritis of 3 or more joint areas 3. Arthritis of hand joints 4. Symmetric arthritis 5. Rheumatoid nodules 6. Serum rheumatoid factor 7. Radiographic changes maximal improvement At least 3 joint areas simultaneously have had soft tissue swelling or fluid (not bony overgrowth alone) observed by a physician. The 14 possible areas are right or left PIP, MCP, wrist, elbow, knee, ankle, and MTP joints At least 1 area swollen (as defined above) in a wrist, MCP, or PIP joint Simultaneous involvement of the same joint areas (as defined in 2) on both sides of the body (bilateral involvement of PIPs, MCPs, or MTPs is acceptable without absolute symmetry) Subcutaneous nodules, over bony prominences, or extensor surfaces, or in juxtarticular regions, observed by a physician Demonstration of abnormal amounts of serum rheumatoid factor by any method for which the result has been positive in <5% of normal control subjects Radiographic changes typical of RA on posteroanterior hand and wrist radiographs, which must include erosions or unequivocal bony decalcification localised in or most marked adjacent to the involved joints (osteoarthritis changes alone do not qualify) MCP metacarpophalangeal joint, PIP proximal interphalangeal joint and MTP metatarsophalangeal joint. For classification purposes, a patient shall be said to have rheumatoid arthritis if he/she has satisfied at least 4 or these 7 criteria. Criteria 1 through 4 must have been present for at least 6 weeks. Patients with 2 clinical diagnoses are not excluded. Designation as classic, definite, or probable rheumatoid arthritis is not to be made. (11) 32

33 Table 1.2: The 2010 American College of Rheumatology/European League Against Rheumatism classification criteria for RA (12) Target population (Who should be tested?): Patients who 1) have at least 1 joint with definite clinical synovitis (swelling)* 2) with the synovitis not better explained by another disease Classification criteria for RA (score-based algorithm: add score of categories A-D; a score of 6/10 is needed for classification of a patient as having definite RA) A. Joint involvement 1 large joint 2 10 large joints 1 3 small joints (with or without involvement of large joints)** 4 10 small joints (with or without involvement of large joints) > 10 joints (at least 1 small joint) B. Serology (at least 1 test result is needed for classification) Negative RF and negative ACPA Low-positive RF or low-positive ACPA High-positive RF or high-positive ACPA C. Acute-phase reactants (at least 1 test result is needed for classification) Normal CRP and normal ESR Abnormal CRP or normal ESR D. Duration of symptoms < 6 weeks 6 weeks 33 Score *The criteria are aimed at classification of newly presenting patients. In addition, patients with erosive disease typical of rheumatoid arthritis (RA) with a history compatible with prior fulfilment of the 2010 criteria should be classified as having RA. Patients with long-standing disease, including those whose disease is inactive (with or without treatment) who, based on retrospectively available data, have previously fulfilled the 2010 criteria should be classified as having RA. Differential diagnoses differ in patients with different presentations, but may include conditions such as systemic lupus erythematosus, psoriatic arthritis and gout. If it is unclear about the relevant differential diagnoses to consider, an expert rheumatologist should be consulted. Although patients with a score of less than 6/10 are not classifi able as having RA, their status can be reassessed and the criteria might be fulfi lled cumulatively over time. Joint involvement refers to any swollen or tender joint on examination, which may be confirmed by imaging evidence of synovitis. Distal interphalangeal joints, first carpometacarpal joints and first metatarsophalangeal joints are excluded from assessment. Categories of joint distribution are classified according to the location and number of involved joints, with placement into the highest category possible based on the pattern of joint involvement. Large joints refers to shoulders, elbows, hips, knees and ankles. ** Small joints refers to the metacarpophalangeal joints, proximal interphalangeal joints, second to fifth metatarsophalangeal joints, thumb interphalangeal joints and wrists. In this category, at least one of the involved joints must be a small joint; the other joints can include any combination of large and additional small joints, as well as other joints not specifically listed elsewhere (eg, temporomandibular, acromioclavicular, sternoclavicular, etc.). Negative refers to international unit (IU) values that are less than or equal to the upper limit of normal (ULN) for the laboratory and assay; low-positive refers to IU values that are higher than the ULN but three of less times the ULN for the laboratory and assay; high-positive refers to IU values that are more than three times the ULN for the laboratory and assay. When rheumatoid factor (RF) information is only available as positive or negative, a positive result should be scored as low-positive for RF. Normal/abnormal is determined by local laboratory standards. Duration of symptoms refers to patient self-report of the duration of signs or symptoms of synovitis (eg, pain, swelling, tenderness) of joints that are clinically involved at the time of assessment, regardless of treatment status. ACPA: anti-citrullinated protein antibody; CRP: C-reactive protein; ESR: erythrocyte sedimentation rate. (12)

34 1.3.3 Epidemiology About 1% of the adult population is affected by RA, in a male to female ratio of 1/ 2.5. The disease is most common among people aged years and its incidence increases with age. (8) Lack of incidence and prevalence studies in developing countries limits understanding of the worldwide prevalence of RA, however available prevalence studies suggest that there are variations in the occurrence of RA from country to country, or from world continent to another. (13) In the UK, based on the 1987 revised criteria for RA classification, a study in 2002 estimated that the prevalence of RA was 1.16% in females and 0.44% in males. (14) Aetiology The actual cause of the disease is unknown. Worldwide variation in the occurrence of RA may suggest the involvement of some environmental and genetic contributing factors. (15) Likewise, the higher prevalence of RA in females may suggest sex hormones as a contributing factor. (16) Infectious agents have also been suspected but not fully understood Pathogenesis Due to unknown stimuli, an inflammatory reaction starts in the synovial membrane. The reaction starts by the activation of immune cells including T-lymphocytes, macrophages and B- lymphocytes. The activated T-lymphocytes also directly activate macrophages and B cells. The activated macrophages start to produce pro-inflammatory cytokines such as tumour necrosis factor (TNF), interleukin 1 (IL-1), and interleukin 6 (IL-6). These inflammatory cytokines further promote the stimulation of more macrophages, involvement of more inflammatory cells, and the production of destructive enzymes that start to destroy the cartilage of the joint and even the bone leading to the state of deformities associated with RA. (17;18) 34

35 1.4 Outcome measures in RA in clinical trials and clinical care Evaluation of the management of RA in clinical trials or in daily clinical practice is based on the use of outcome measures that either measure the activity of the disease itself or report on patient outcomes during and after treatment. The clinical manifestations of RA differ from patient to patient, and this heterogeneity makes the evaluation of RA disease activity a complex process and increases the need for reliable and validated outcome measures. Individual outcome measures such as the tender joint count, swollen joint count, morning stiffness, erythrocyte sedimentation rate or C-reactive protein have traditionally been used to measure disease activity. The use of these different individual measures in different studies made comparison between studies difficult so the development of a composite measure that includes more than one individual measure was useful in terms of facilitating comparisons across different studies and describing multiple aspects of the disease. As a result, the disease activity score (DAS) was developed in the Netherlands, in 1990, (19) along with the European League against Rheumatism (EULAR) (20) response criteria. The American College of Rheumatology (ACR) have also established a core set of outcome measures and a definition of response. (21;22) Disease activity measures Disease activity score (DAS) The core components of the DAS are the swollen joint count, tender joint count, acute phase reactant (erythrocyte sedimentation rate or C-reactive protein) and patient global assessment. Depending on which definition of DAS is used (Table 1.3); different formulas are used to calculate the score. The original DAS is a continuous scale up to ten. (23) It is calculated using a 44-swollen joint count (score 0-44), the Ritchie articular index (RAI, range 0-78) as a count of tender joints, the 35

36 erythrocyte sedimentation rate (ESR) and a general health assessment on a visual analogue scale of (19) Prevoo et al. found that a DAS < 1.6 corresponds to a state of remission according to the American Rheumatism Association (ARA). (24) Other variations of the score include substituting the ESR with the C-reactive protein (CRP) and omitting the general health score where it is not available. The DAS28 is similar to the original DAS but includes a simplified joint count. The DAS28 is a continuous scale up to 9.4, (23) consisting of a 28 swollen joint count (range 0-28), a 28 tender joint count (range 0-28), erythrocyte sedimentation rate (or C-reactive protein), and an optional general health visual analogue scale (range 0-100). A patient is considered to be in remission state if DAS28 is lower than 2.6. (25) DAS28 was developed and validated by Prevoo et al. (26) using a cohort of 227 RA patients. The results showed that DAS28 was successfully able to discriminate between high and low disease activity as well as the DAS score that includes 44 joints count. It is difficult to directly compare DAS and DAS28 as they have different scales, but a transformation formula is available to convert DAS28 to DAS which facilitates comparisons across different studies (Table 1.3). (27) 36

37 Table 1.3 : Formula for calculation of DAS, DAS28 and conversion from DAS to DAS28 (23;26;27) ( Formula DAS = 0.54 x (RAI) x (SJC 44) x ln (ESR) Description DAS formula using ESR x GH DAS = 0.54 x (RAI) x (SJC 44) x ln (ESR) DAS = 0.54 x (RAI) x SJC x ln DAS formula using ESR without the patient global assessment. DAS formula using CRP (CRP+1) x GH DAS= 0.54 x (RAI) x SJC x ln (CRP+1) DAS28 = 0.56 x (TJC 28) x (SJC 28) x ln DAS formula using CRP without the patient global assessment. DAS28 formula using ESR (ESR) x GH DAS28 = [0.56 x (TJC 28) x (SJC 28) x Ln (ESR)] x DAS28 = 0.56 x (TJC 28) x (SJC 28) x ln DAS28 formula using ESR without the patient global assessment. DAS28 formula using CRP (CRP+1) x GH DAS28 = [0.56 x (TJC 28) x (SJC 28) x ln (CRP+1)] x DAS28 = ((1.072) x DAS) DAS28 formula using CRP without the patient global assessment. Conversion of DAS28 to DAS : square root, RAI: Ritchie Articular Index, SJC: swollen joint count, CRP: C-reactive protein, GH: general health (patient global assessment), TJC: tender joint count, ESR: erythrocyte sedimentation rate. 37

38 The European League Against Rheumatism (EULAR) The use of the mean change in DAS score in clinical trials did not reflect the number of responders or the individual patient s disease response. As a result, the EULAR response criteria were developed to categorise responders. The EULAR response criteria classify patients as good, moderate or non-responders according to the change in their DAS (20) or DAS28 (27) scores over time as well as the final DAS or DAS28 score reached (Table 1.4). In a trial comparing sulfasalazine and hydroxychlroquine in a cohort of 142 RA patients, the level of RA activity was described as low (DAS 2.4), moderate (2.4 < DAS 3.7), or high (DAS > 3.7). (20) Validation of EULAR response that included 28 joints versus those including more comprehensive joint count was assessed using 105 RA patients receiving sulfasalazine, methotrexate or both, disease activity was describes as high (DAS28 score > 5.1), moderate (3.2 < DAS28 5.1), and low (DAS28 3.2). (27) The results showed no differences among the three groups of patients as measured by all tested response criteria which suggested a valid use of EULAR response criteria including 28 joint counts. Table 1.4 : The EULAR response criteria classified as good, moderate or no response. (20;27) (Adapted from Fransen et al. (23) ). Final DAS Final DAS28 Improvement in DAS or DAS28 from baseline >1.2 > 0.6 and Good Moderate Non responders > 2.4and 3.7 > 3.2 and 5.1 Moderate Moderate Non responders > 3.7 > 5.1 Moderate Non responders Non responders 38

39 The American College of Rheumatology (ACR) response In 1993, the common single disease activity measures used in RA clinical trials were evaluated, using structured literature review along with analysis of data from clinical trials, to assess the validity of these measures and develop a core set of disease activity measures (the ACR) to be used specifically in clinical trials of RA to differentiate between patients receiving treatment versus placebo. The ACR core data set consists of seven measures; a swollen joint count, a tender joint count, acute phase reactant (erythrocyte sedimentation rate or C-reactive protein), physician global status, physical function, patient global assessment, and pain. Radiographs of hands and feet are used also, if trial include more than 1 year of follow-up. (22) Further analysis led to developing percentage of improvement ACR responses as ACR20 (21) ACR50 and ACR70. (28) ACR20 is defined as improvement of at least 20% in tender joint count and swollen joint count, as well as in any three of the remaining five measures in the ACR core data. ACR20 is the main outcome in clinical trials of RA. The ACR50 and ACR70 are defined as 50% and 70% improvement (at least) respectively. The ACR response describes response of individual patients and does not categorise patients. Unlike the DAS, the ACR response is not a continuous scale i.e it is either achieved or not. Validation of ACR response criteria that include 28 joints versus those including more comprehensive joint count was assessed using 105 RA patients. (27) The results suggested a valid use of ACR response criteria including 28 joint counts. The validity of the EULAR and ACR response sets were compared in a study involving seven randomised trials and the results showed that they have comparable validity in discrimination between the treatment groups and there was also good agreement in results between the two sets of response criteria. (29) 39

40 1.4.2 Patient s reported outcomes These outcomes give information about patients health, physical function, patient satisfaction and symptoms. Patient s reported outcomes include the Health Assessment Questionnaire (HAQ) (which is an arthritis specific measure), European Quality of life 5 Dimensions (EQ-5D) and Short Form 36 (SF-36) survey (which are generic quality of life measures) Health Assessment Questionnaire (HAQ) The full HAQ was developed in 1980 and it was based on five dimensions: pain, disability, medication effects, costs of care, and mortality. (30) The short HAQ (3 pages) involves only 2 dimensions: HAQ s patient global and pain visual analogue scales (VAS) and the HAQ disability index (HAQ-DI) which is the most commonly used and referred to as the HAQ (31) The HAQ includes 20 items that represent functional activities and assess the movements and motor activities of the upper extremity, lower extremity and both upper and lower extremities. The 20 questions are in 8 categories about the usual abilities (dressing and grooming, rinsing, eating, waking, hygiene, reach, grip and general activities). The patient is asked to respond to each question by one of four choices (0 to 3); either without any difficulty (0), with some difficulty (1), with much difficulty (2), or unable to do over the past week (3). The questionnaire also ask the patient if he/she uses any aids/devices or helps from other persons in these abilities, if yes then scores of 0 and 1 are increased into 2. The responses to the 20 questions are then combined together and averaged to produce a continuous scale from zero to three with lower scores representing less disability. (31) Patients can be classified as responders or non-responders according to whether or not they achieved the minimum clinically important differences (MCID) in HAQ (improvement of 0.22 units at least). (32) 40

41 European Quality of life 5 Dimensions (EQ-5D) EQ-5D is a measure of health status that was developed by the European quality of life group (Euro QoL Group). It consists of two parts; the first part is a descriptive system that includes 5 dimensions that define health (mobility, self-care, activities, pain/discomfort and anxiety/depression). (33) Each dimension is expressed in 3 categories (no problem, some problem or extreme problem). This generates 243 health states to which 2 states were added (dead and unconscious). The second part is the EQ visual analogue scale (EQ-VAS) which describes the participant self-rated health on a continuous scale from 0 to 100. The descriptive system of the EQ-5D is not effective in comparing improvements in quality of life among different patients groups; hence there was a need to develop a weighted index on a continuous scale to facilitate these comparisons. Weights for the different health states of the EQ-5D were developed from a large UK population. (34) These weights can be used to calculate the EQ-5D utility which is a weighted index on a continuous scale that ranges from to 1.00 where 0 represents death, 1 represents full health, and negative values represent health states that are valued as worse than death. (34) Short Form 36 (SF-36) The SF-36 is a survey of 36 questions. The taxonomy of items and concepts underlying the construction of the SF-36 scales has three levels: items, scales and summary measures; eight scales (physical functioning, bodily pain, general health, role physical, mental health, vitality, role emotional and social functioning) that aggregate two to ten items each, and two summary measures (physical health and mental health) that aggregate scales (Table 1.5). (35) Response to these questions are summed and converted to a scale of 0 to 100 where higher score represent better quality of life. 41

42 Table 1.5: SF-36 taxonomy (35) Summary measures Scales Items Physical health Physical functioning (limitation in activities) Role physical (physical health problems) Bodily pain General health Vigorous activities Moderate activities Lift or carry groceries Climb several flights of stairs Climb one flight of stairs Bend, knee or stoop Walk more than one mile Walk several hundred yards Walk 100 yard Bathing or dressing Cut down time Accomplished less Limited in kind Had difficulty Pain magnitude Pain interfere with work General health now General health now compared to 1 year ago Get ill more easily than others As health as any body Expect health to get worse Health is excellent Mental health Vitality Mental health Social functionality Role emotional (mental health problems) Feel full of life Have a lot of energy Feel worn out Feel tired Very nervous Down in the dumps Calm and peaceful Downhearted and depressed Happy Social extent Social time Cut down time Accomplished less Less carefully 42

43 1.5 Management of RA Analgesics and non-steroidal anti-inflammatory drugs (NSAIDs) are used for control of pain in RA, with selective use of glucocorticoids, and initiation of a non-biologic disease-modifying antirheumatic drug (nbdmard) as the mainstay of treatment in RA Analgesics and non-steroidal anti-inflammatory drugs (NSAIDs) Paracetamol (alone or in combination with codeine) is commonly used as analgesic in RA. NSAIDs are also used for treatment of RA to reduce pain and swelling of joints. Since this group of drugs do not affect disease progression, they should not be used alone except for very mild disease. (9) Glucocorticoids Steroids (7.5 mg of prednisone daily) can relieve the symptoms of RA very effectively (9) and they can delay the damage of the joints. (36) Minimum steroid dosages should be used to avoid the risk of serious side events such as osteoporosis, weight gain, muscle wasting and an increased risk of serious infection. Injection of glucocorticoids can offer temporary effect when a single joint is inflamed and disabled. (9) Withdrawal of steroids should be slow to avoid withdrawal symptoms and Addisonian crisis. Systemic steroids can also be used when treatment with disease-modifying anti-rheumatic drugs is first initiated but the effect of treatment has not yet started. (9) Non-biologic disease-modifying anti-rheumatic drugs (nbdmards) The main goal of treating RA is to control the disease progress and stop joint destruction. The main group of treatments used for this purpose are disease modifying anti-rheumatic drugs (DMARDs). With the recent introduction of biologic therapy for RA, these drugs are now referred to as non-biologic DMARDs (nbdmards). The most commonly used nbdmards are methotrexate (MTX), hydroxychloroquine (HCQ), sulphasalazine (SSZ), and leflunomide (LEF). 43

44 Other less commonly used nbdmards include gold, azathioprine, cyclosporin and D- penicillamine. First-line treatment in patients with recent onset of RA is usually MTX monotherapy or in combination with another nbdmard such as SSZ or HCQ. Recent survey evidence suggested that MTX is the first preference nbdmard of most UK rheumatologists. (37) Combinations of nbdmards are more effective than single-drug treatments. (38) When patients fail to respond or tolerate traditional nbdmards, the next treatment option will generally involve the use of newer biologic therapies Biologic therapies The management of RA has dramatically changed after the introduction of biologic therapies. Biological products include a wide range of products such as vaccines, blood and blood components, allergenics, somatic cells, gene therapy, tissues, and recombinant therapeutic proteins. Biologics can be composed of sugars, proteins, or nucleic acids or complex combinations of these substances, or may be living entities such as cells and tissues. Biologics are isolated from a variety of natural sources - human, animal, or microorganism - and may be produced by biotechnology methods and other cutting-edge technologies. Gene-based and cellular biologics, for example, often are at the forefront of biomedical research, and may be used to treat a variety of medical conditions for which no other treatments are available. (39) Table 1.6 summarises the currently licensed (June 2011) biologic therapies for RA. The current biologic therapies include anti-tumour necrosis factor (anti-tnf) therapies (etanercept, adalimumab, infliximab, certolizumab pegol and golimumab), the IL-1 receptor antagonist, (anakinra), IL-6 antogonist (tocilizumab), T cell activation blocker (abatacept), and the anti- CD20 marked B cells, (rituximab). 44

45 Table 1.6: Licensing details of the biologic therapies for use in RA (from FDA and EMA websites) Etanercept (Enbrel) Infliximab (Remicade) Adalimumab (Humira) Certolizumab pegol (Cimzia) Golimumab (Simponi) Anakinra (Kineret) Biologic target Tumour necrosis factor (TNF) Interleukin- 1 Type Receptor Antibody Antibody Antibody Antibody Receptor antagonist Approval Date 2 Nov. 10 Nov. 31 Dec. 22 April 24 April 14 Nov. by FDA for use in RA EMA authorisation date * Licensed dose Route of administration Approved by the NHS 3 Feb mg twice weekly or, 50 mg once weekly 13 Aug Sep mg/kg at weeks 0, 2 and 6, and thereafter every 8 weeks 40 mg every other week or equivalent. 1 Oct mg at weeks 0, 2 and 4, then and thereafter 200 mg every 2 weeks. 1 Oct mg once a month (Given on the same date each month) 8 Mar mg once daily Tocilizumab (RoActemra) Abatacept (Orencia) Rituximab (Mabthera) (Rituxan) Interleukin-6 T cells B cells Antibody Protein Antibody 08 Jan Jan mg/kg every 4 weeks 23 Dec May 2007 Weight dependent dose every 2 weeks for 3 doses, then every 4 weeks 28 Feb Jun (RA in 2006) Two 1000 mg infusions 2 weeks apart SC injection IV infusion SC injection SC injection Injection SC injection IV infusion IV infusion IV infusion Yes Yes Yes Yes Yes No Yes No Yes Combination with MTX No Yes No Yes Yes No Yes Yes Yes FDA: Food and Drug Administration, EMA: European Medicines Agency, RA: rheumatoid arthritis, SC: subcutaneous, IV: intravenous, NHS: National Health services, MTX: methotrexate. * The exact date of license for use in RA could not be found at EMA web site. 45

46 Infections, injection-site reactions, and allergic reactions are considered the main potential side effects of biologic therapies in RA. Biologic therapies are contra-indicated in patients with hypersensitivity to the active substance and in those with active infections. Anti -TNF therapies are contraindicated in patients with moderate or severe heart failure and in malignancies. (40-42) Anti-Tumour necrosis factor (anti-tnf) therapies Anti-TNF therapies decrease the levels of circulating TNF-alpha, which is one of the key cytokines driving the inflammatory process in RA. Now, there are five anti-tnf therapies licensed for the use in RA; etanercept (ETN), infliximab (INF), adalimumab (ADA), certolizumab pegol, and golimumab. ETN, INF and ADA have been routinely used in the last decade. However, recently, certolizumab pegol and glomumab were licensed for use in RA. Etanercept (ETN) is a dimer of a chimeric fusion protein which competes with TNF for the TNFreceptor. ETN in combination with MTX is indicated for moderate to severe active RA when the response to nbdmards, including MTX has been inadequate. ETN can be given as mono-therapy in case of contraindication or intolerance to MTX. ETN, alone or in combination with MTX, has been shown to improve physical function and reduce the rate of progression of joint damage. (40) Infliximab (INF) is a chimeric immunoglobulin 1 (IgG1) antibody which binds to TNF-alpha and hence blocks its action. INF in combination with MTX is indicated for moderate to severe active RA when the response to nbdmards, including MTX has been inadequate. Reduction in the rate of the progression of joint damage, as measured by X-ray, has been demonstrated in patients treated with INF. (41) Adalimumab (ADA) is also a human IgG1 monoclonal antibody administered as subcutaneous injection with demonstrated effectiveness in clinical practice. (43) The summary of product 46

47 characteristics (SPC) states that ADA should be given in combination with MTX, except where MTX is not tolerated or is considered inappropriate. (42) Certolizumab pegol is a recombinant, humanised antibody that is conjugated to polyethylene glycol. The starting dose for adult RA patients is 400 mg at weeks 0, 2 and 4, then a maintenance dose of 200 mg every two weeks. MTX should be used with certolizumab pegol where appropriate. (44) Golimumab is a human monoclonal antibody produced by recombinant DNA technology using a murine hybridoma cell line. It is licensed for moderate to severe, active RA patients who have failed nbdmards therapy including MTX, and for severe, active and progressive RA patients who are MTX naïve. (45) The anti-tnf therapies are recommended for RA patients who have failed conventional nbdmards. It is recommended that INF be prescribed in combination with MTX to reduce the production of human anti-chimeric anti-bodies and improve response. Although it is not absolutely required that ETN and ADA are co-prescribed with MTX there is now growing evidence that this combination is more effective than mono-therapy. (46-48) Certolizumab pegol and golimumab are licensed to be used in combination with MTX. Treatment guidelines suggest that anti-tnf treatment should be withdrawn if an improvement in DAS28 of 1.2 points or more is not maintained at six months of initiation of treatment. (49) The National Institute of Health and Clinical Excellence (NICE) have developed clinical guidelines on the use of ETN, INF, and ADA therapies in RA in the UK (Table 1.7). The NICE guidelines recommended that certolizumab and golimumab should follow those guidelines for ETN, INF, and ADA therapies. (50;51) The Nice guidelines did not approve the use of golimumab in MTX naïve patients within the National Health Services (NHS) as no evidence was provided by manufacturers. (52) 47

48 Table 1.7: NICE guidance for the use of adalimumab, etanercept and infliximab therapies in RA (49) Guidance 1 The tumour necrosis factor alpha (TNF-α) inhibitors adalimumab, etanercept and infliximab are recommended as options for the treatment of adults who have both of the following characteristics. Active rheumatoid arthritis as measured by disease activity score (DAS28) greater than 5.1 confirmed on at least two occasions, 1 month apart. Have undergone trials of two disease-modifying anti-rheumatic drugs (DMARDs), including methotrexate (unless contraindicated). A trial of a DMARD is defined as being normally of 6 months, with 2 months at standard dose, unless significant toxicity has limited the dose or duration of treatment. 2 TNF-α inhibitors should normally be used in combination with methotrexate. Where a patient is intolerant of methotrexate or where methotrexate treatment is considered to be inappropriate, adalimumab and etanercept may be given as monotherapy. 3 Treatment with TNF-α inhibitors should be continued only if there is an adequate response at 6 months following initiation of therapy. An adequate response is defined as an improvement in DAS28 of 1.2 points or more 4 After initial response, treatment should be monitored no less frequently than 6-monthly intervals with assessment of DAS28. Treatment should be withdrawn if an adequate response (as defined in 1.3) is not maintained. 5 An alternative TNF-α inhibitor may be considered for patients in whom treatment is withdrawn due to an adverse event before the initial 6-month assessment of efficacy, provided the risks and benefits have been fully discussed with the patient and documented. 6 Escalation of dose of the TNF-α inhibitors above their licensed starting dose is not recommended. 7 Treatment should normally be initiated with the least expensive drug (taking into account administration costs, required dose and product price per dose). This may need to be varied in individual cases due to differences in the mode of administration and treatment schedules. 8 Use of the TNF-α inhibitors for the treatment of severe, active and progressive rheumatoid arthritis in adults not previously treated with methotrexate or other DMARDs is not recommended. 9 Initiation of TNF-α inhibitors and follow-up of treatment response and adverse events should be undertaken only by a specialist rheumatological team with experience in the use of these agents. 48

49 The efficacy of anti-tnf therapies in the treatment of RA against placebo or MTX has been shown in several RCTs. However, no direct comparisons between the three anti-tnf therapies (ETN, INF, and ADA) have been undertaken. A meta-analysis and indirect comparison between the three anti-tnf therapies were performed by Nixon et al. (53) using data from 13 RCTs. No differences in efficacy between the three anti-tnf therapies were identified. A meta-analysis of 13 RCTs of biologic therapies (4 for ETN, 4 for INF and 5 for ADA) was also undertaken in 2008 (54) and the results showed that the three anti-tnf therapies had superior efficacy to placebo or MTX alone in RA patients. In addition, another meta-analysis of 12 randomised clinical trials comparing INF+MTX versus MTX mono-therapy (including both open labelled and blinded trials) was performed by Zintzaras et al. (55) The results showed that INF (3 and 10 mg/kg) +MTX were more effective therapies than MTX alone; odds ratio (OR) = 3.52 (2.14:5.79) for 3 mg/kg dose (the licensed dose) and 5.06 (3.88:6.59) for 10 mg/kg dose. Golimumab and certolizumab have shown efficacy in RA patients in a number of RCTs. (56-60) Interleukin receptor antagonists Anakinra blocks interleukin-1 (IL-1) which is an important cytokine in developing damage associated with RA. Anakinra is licensed for the treatment of the signs and symptoms of RA in combination with MTX, in patients with an inadequate response to MTX alone. (61) Several randomised controlled trials have found anakinra to be effective when used alone or in combination with MTX when compared against placebo. (62;63) In the UK, anakinra was appraised by NICE and was not approved for use in the NHS on the grounds of lack of cost-effectiveness. (64) Tocilizumab is a humanised monoclonal antibody that blocks the interleukin-6 (IL-6) receptor. It is produced from Chinese hamster ovary cells by recombinant DNA technology. It is licensed for RA patients who do not respond or have intolerance to nbdmards or anti-tnf therapies. It is 49

50 licensed to be used in combination with MTX; however it can be used alone in case of intolerance to MTX. (65) The NICE guidelines approved the use of tocilizumab within the NHS in RA patients who had failed to respond to one or more anti-tnf therapy and had failed rituximab (RTX) or had discontinued RTX due to an adverse event. (66) T cell activation blocking Abatacept is a fusion protein that is produced by recombinant DNA technology involving ovary cells of the Chinese hamster. Abatacept prevents the full activation of T cells through binding to B7 protein on antigen presenting cells which is an important step in the activation of T cells. In combination with MTX, abatacept is licensed for RA patient who have failed traditional nbdmards including MTX or anti-tnf therapy. (67) However, the NICE did not recommend the use of abatacebt within the NHS in the UK based on the lower efficacy and higher cost compared to other subcutaneous biologic therapies for RA. (68) B-cell depletion Rituximab (MabThera (according to the European Medicines Agency EMA); Rituxan, (according to Food and Drug Administration FDA)) is a monoclonal antibody which acts by depleting B-cells via targeting cells bearing the CD20 surface marker. Due to the role of B cells in the pathogenesis of autoimmune diseases including RA, RTX was introduced for treatment of RA and was licensed to be used in RA in Only two-thirds of MTX resistant RA patients respond to anti-tnf and about 30% of patients fail to respond to anti-tnf. (69) When patients fail a TNF blocker, they are very unlikely to respond to anakinra. (70) Switching to another TNF antagonist may be effective in 50 % of patients only. (71) This increases the importance and hope associated with recently introduced biological therapies especially when they have a different mechanism of action such as RTX. As RTX is the main target drug in this PhD, the following paragraphs describe this drug in further details. 50

51 1.6 Rituximab (RTX) Structure RTX is a human/mouse chimeric monoclonal antibody that was developed by Biogen Idec pharmaceuticals by recombinant technology. RTX consists of two heavy and two light chains of amino acids (Figure 1.1) with a molecular weight of g/mol and a formula of C 6416 H 9874 N 1688 O 1987 S 44. The binding affinity of RTX (antibody) to the CD20 (antigen) on the B cell is about 8 nm. (72;73) The binding to the CD20 antigen occurs through the murine region while the human region is responsible for the binding to the macrophages and the complement activation. RTX is currently co-marketed by Biogen Idec and Genentech in the U.S.A and by Roche in Europe. RTX is marketed as a clear colourless sterile liquid concentrate for intravenous infusion. Light chain Heavy chain Antigen binding Murine region Human region Complement activation Macrophage binding Figure 1.1: Schematic diagram showing the structure of RTX. (72;73) 51

52 CD20 RTX Cell membrane B cell Figure 1.2: Diagram showing the binding of RTX to CD20 on the surface of the B cell. Adapted from Pescovitz et al. (74) Mechanism of action of RTX B cells play an important role in the pathogenesis of RA and they are the source of rheumatoid factor which is present in about 80% of RA patients. CD20 is a protein that is present on the membrane surface of the B cell (Figure 1.2) in all stages of developing the B cell except the final plasma cell stage. CD20 crosses the cell membrane four times. The function of the CD20 surface marker is not clear but it is thought to play a role in calcium influx and hence the activation of the cell. The mechanism of action of RTX in depleting B cells can be described by three actions on the cell; antibody-dependent cell-mediated cytotoxicity, complement-dependent cytotoxicity and apoptosis (Figure 1.3). (74;75) 52

53 Macrophages RTX 1 CD20 Phagocytosis of the B cell B cell Membrane attack complex 2 Cell lysis 3 Cell lysis 1. Antibody-dependent cell-mediated cytotoxicity: antibody labeled B cells become recognizable by natural killer cells and phagocytes leading to cell killing or phagocytosis. 2. Complement-dependent cytotoxicity binding of RTX to B cells activates the membrane attack complex (MAC) which causes holes in the cell membrane followed by cell lysis. 3. Apoptosis: binding of RTX to B cells causes cell self-destruction and lysis. Figure 1.3: Schematic diagram showing the mechanism of action of RTX. (74;75) 53

54 1.6.3 License and approval of RTX for RA The European Medicines Agency (EMA) state that Rituximab in combination with MTX is indicated for the treatment of adult patients with severe active rheumatoid arthritis who have had an inadequate response or intolerance to other disease-modifying anti-rheumatic drugs (DMARD) including one or more tumour necrosis factor (TNF) inhibitor therapies. (76) Likewise, the US FDA license states that The use of RITUXAN (rituximab) in combination with MTX to reduce the signs and symptoms in adult patients with moderately- to severely-active rheumatoid arthritis who have had an inadequate response to one or more TNF antagonist therapies has been approved. (77) The licensed recommended dose of RTX consists of two 1000-mg infusions given intravenously 2 weeks apart (a single course of RTX). Re-treatment with another course of RTX may be used but not before 6 months from the first course. The NICE had developed clinical guidelines on the use of RTX therapy in RA in the UK in 2007 (78) which was replaced by guidelines for the use of biologics after failure of anti -TNF therapy in 2010 (Table 1.8). (79) The next chapter will review the published experience with RTX in more detail, including a review of its efficacy. As the majority of patients who receive RTX will also have failed anti-tnf, treatment persistence with anti-tnf therapy will also be reviewed in order to better understand the number and types of patients who will be eligible for RTX. 54

55 Table 1.8: NICE guidance for the use of biologic drugs after failing anti-tnf therapy in RA (79) Guidance 1 Rituximab in combination with methotrexate is recommended as an option for the treatment of adults with severe active rheumatoid arthritis who have had an inadequate response to, or are intolerant of, other disease-modifying anti-rheumatic drugs (DMARDs), including at least one tumour necrosis factor (TNF) inhibitor. Treatment with rituximab should be given no more frequently than every 6 months. 2 Treatment with rituximab in combination with methotrexate should be continued only if there is an adequate response following initiation of therapy and if an adequate response is maintained following retreatment with a dosing interval of at least 6 months. An adequate response is defined as an improvement in disease activity score (DAS28) of 1.2 points or more. 3 Adalimumab, etanercept, infliximab and abatacept, each in combination with methotrexate, are recommended as treatment options only for adults with severe active rheumatoid arthritis who have had an inadequate response to, or have an intolerance of, other DMARDs, including at least one TNF inhibitor, and who cannot receive rituximab therapy because they have a contraindication to rituximab, or when rituximab is withdrawn because of an adverse event. 4 Adalimumab monotherapy and etanercept monotherapy are recommended as treatment options for adults with severe active rheumatoid arthritis who have had an inadequate response to, or have an intolerance of, other DMARDs, including at least one TNF inhibitor, and who cannot receive rituximab therapy because they have a contraindication to methotrexate, or when methotrexate is withdrawn because of an adverse event. 5 Treatment with adalimumab, etanercept, infliximab and abatacept should be continued only if there is an adequate response (as defined in 1.2) 6 months after initiation of therapy. Treatment should be monitored, with assessment of DAS28, at least every 6 months and continued only if an adequate response is maintained. 6 When using DAS28, healthcare professionals should take into account any physical, sensory or learning disabilities, communication difficulties, or disease characteristics that could adversely affect patient assessment and make any adjustments they consider appropriate. 7 A team experienced in the diagnosis and treatment of rheumatoid arthritis and working under the supervision of a rheumatologist should initiate, supervise and assess response to treatment with rituximab, adalimumab, etanercept, infliximab or abatacept. 55

56 2. Literature review 56

57 The aim of this chapter was to provide the results of a literature search which went in two directions; the first was to have a view of the patients who are eligible to receive RTX, the target drug in this PhD, in terms of their persistence with their previous therapies, namely anti- TNF therapies and the reasons for its discontinuations. The second was to search the literature to find out what we know and what we do not know so far about the performance of RTX in RA patients who were treated under ideal conditions or in routine clinical practice. 2.1 Assessment of drug persistence with anti-tnf therapies in RA Persistence with a therapy is defined as the length of time the patients stayed on the therapy Introduction Clinical trials have found that approximately 30% of patients treated with anti-tnf therapies fail to respond to the therapy. (69) This fact emphasises the importance of alternative treatment options for RA such as RTX which is licensed for patients who have failed one or more anti -TNF therapies. In addition to non-responders to anti-tnf therapies, patients who discontinue the anti-tnf therapy for reasons other than inefficacy, such as the occurrence of adverse events, should be taken into consideration as they may also need to initiate alternative treatments to manage RA. In summary, patients who discontinue anti-tnf therapies are eligible for starting RTX therapy. As a result, it was necessary to study the persistence with anti-tnf therapies because the patients who do not continue with anti-tnf therapies are most likely to receive RTX, the target drug under investigation in this PhD. 57

58 2.1.2 Method Ovid Medline and Embase databases were searched to identify studies that examined drug persistence with the anti-tnf therapies in RA. The search strategy (Appendix 2.1) inclusion criteria were: inclusion of RA patients, the use of one or more anti-tnf therapy, assessment of drug persistence with the anti-tnf therapy, and the study should be an observational study. The search excluded articles that were not written in English, did not involve human use and articles published before 1995, prior to the use of anti-tnf therapies in RA. The search was also limited to the first three licensed anti-tnf therapies (ETN, INF, and ADA) since they have been routinely used in the last decade. In total, 26 studies examining persistence with anti-tnf therapies in RA were identified, as shown in Table 2.1. The identified studies included 13 studies involving large national registers with a mean sample size of 1,863 patients and 13 studies undertaken in one or more centres with a mean sample size of 221 patients. The study design of the identified studies was either prospective or retrospective. The patients included in the studies were receiving anti -TNF therapies between 1995 and The identified studies considered the persistence with anti -TNF therapies with different objectives; some studies looked at the overall persistence, persistence with the three anti-tnf therapies, persistence with anti-tnf mono-therapy or with MTX, persistence according to the reason for discontinuation, persistence with the second course of anti-tnf therapies, persistence in different rheumatic diseases, predictors of anti-tnf therapy discontinuation, and some studies considered two or more of these objectives together. This literature review will analyse and compare the findings from the identified studies. 58

59 Table 2.1: Published studies on persistence with anti-tnf therapies in RA Study Study design Data used Enrolment period Persistence with the first course Number of patients Country Flendrie, 2003 (80) Cohort study One centre Netherlands Chevillotte- Maillard, 2005 (81) Cohort study 9 hospitals Before Nov RA=60 AS=23 France Zink, 2005 (82) Cohort study National register May Sep Germany Duclos, 2006 (83) Retrospective observational study 1 centre RA=440 PsA=290 Other=40 France Kristensen, 2006 (84) Prospective cohort study. National register March Dec ,161 Sweden Carmona, 2006 (85) Prospective cohort study. National register RA=4,006 SpA,=1,524 Spain Brocq, 2007 (86) Grijalva, 2007 (87) Retrospective study Retrospective cohort study. One hospital Tennessee Medicaid databases Aug June 2005 RA=304 AS=92 PsA=46 France ,036 USA Tang, 2008 (88) Retrospective study Claims database Jan Jun ,242 USA Bertoli, 2008 (89) Heiberg, 2008 (90) Retrospective study. Prospective cohort study. Out-patient private centre National register Aug Dec Dec Aug Argentina RA=847 PsA=172 AS=249 Norway RA: rheumatoid arthritis, SpA: spondyloarthropathies, PsA: psoriatic arthritis, AS: ankylosing spondylitis, TNF: tumour necrosis factor. * The patients are RA patients otherwise mentioned. 59

60 Table 2.1 (continued): Published studies on persistence with anti-tnf therapies in RA Enrolment Number of Study Study design Data used Country period patients Ducoulombier, Observational Multicentre - 50 France 2007 (91) study Geborek, 2002 (92) Cohort study 7 centres March Nov Sweden Flendrie, 2005 (93) Prospective cohort One centre From Jan Netherlands Ibanez, 2005 (94) Vander, 2006 (95) Voulgari, 2005 (96) Observational study Observational study Observational study Ostergaard, 2007 (97) Prospective cohort study. Hetland, 2008 (98) Finckh, 2009 (99) Tanaka, 2008 (100) Prospective cohort study Observational study Retrospective study. Gomez-Reino, 2006 (101) Prospective cohort study. Hyrich, 2007 (102) Hjardem, 2007 (103) Laas, 2008 (104) Prospective cohort study. Prospective cohort study. Retrospective cohort study. Markenson, Retrospective 2011 (105) study. - Dec Sep Spain 7 centres Belgum - National register National register National register Three centres Sep to June Greece Before Denmark ,813 Denmark March Dec.2006 Persistence with the second course National register National register National register 1,218 Switzerland Till Dec Japan Feb Sep April st course =2,235 2 nd course =194 Spain 6,739 UK Denmark One hospital Finland National register Oct Jan USA 60

61 2.1.3 Anti-TNF therapies overall drug persistence The overall drug persistence of the three anti-tnf therapies due to all reasons of discontinuation was addressed in eight studies (Table 2.2). The one year drug persistence was not comparable throughout the eight studies. This may be due to the difference in the characteristics of patients included in the different studies such as the difference of the percent of patients receiving anti-tnf therapies as mono-therapies or with concomitant nbdmards, co-morbidities, and disease severity. Table 2.2: Overall persistence rates of anti-tnf therapies first course in RA (from the literature) Study Overall persistence rates (%) First year Second year Third year Carmona et al. (85) Brocq et al. (86) Tang et al. (88) Heiberg et al. (90) Ostergaard et al. (97) Hetland et al. (98) Gomez-Reino et al. (101) Hjardem et al. (103)

62 2.1.4 A comparison of drug persistence between the three anti-tnf therapies Ten studies compared the persistence with the three anti-tnf therapies. From those ten studies, four studies reported no differences between the three anti-tnf therapies, one study reported superior persistence with INF and five studies reported inferior persistence with INF than the other two anti-tnf therapies. From those ten studies, seven studies reported on persistence without consideration of the presence or absence of concomitant nbdmards (Table 2.3) and three studies reported on persistence of the three anti-tnf therapies when received as mono-therapy or with concomitant nbdmards (Table 2.4). Studies that reported no significant differences between the three anti-tnf therapies In the study of Flendrie et al. (80) the one year persistence with ETN, INF and ADA was 74%, 66% and 73% respectively; with no significant difference between the three anti-tnf therapies. Similar results were reported by Duclos et al (83) where the two year persistence was 51%, 48% and 60% for ETN, INF and ADA respectively (p=0.48), however when the analysis was undertaken according to the reason of stopping, INF showed inferior persistence (p=0.06) than ETA and ADA when the reason of stopping treatment was intolerance. Only 57% of the patients in the later study were RA patients. Similar one year drug persistence (95% CI) with ETN (69% (62: 75%)) and INF (65% (58: 73%)) was also reported by Zink et al. (82) 62

63 Table 2.3: Persistence rates of the three anti-tnf therapies in RA with no consideration of the presence or absence of the concomitant nbdmards therapy (from the literature) Study Persistence rates (%) Flendrie et al. 1 st year Etanercept Infliximab Adalimumab 2 nd year 3 rd year 4 th year 1 st year 2 nd year 3 rd year 4 th year 1 st year 2 nd year Studies that assessed the persistence with the three anti-tnf therapies 3 rd year (80) Duclos et al. (83) Zink et al. (82) Brocq et al. (86) Gomez-Reino et al. (101) Ostergaard et al. (97) Markenson, 4 th year 2011 (105) Bertoli et al. (89) Tanaka et al. (100) Studies that assessed the persistence with infliximab Vander, Cruyssen et al (95) Ducoulombier et al. (91) Voulgari et al. (96)

64 Studies that reported inferior persistence with INF than the other two anti-tnf therapies Ostergaard et al. (97) reported inferior one year persistence with INF (71%) when compared with ETN (73%) (p=0.04). Likewise, Heiberg et al. (90) reported significant difference between the three anti- TNF therapies persistence rates in favour of ETN (the persistence rates were not reported); however, this difference disappeared when the analysis was limited to patients who had never received previous anti-tnf therapies. INF also showed a lower two year persistence rate (46%) compared to ETN (68%) (p=0.0001) and ADA (66%) (p=0.01) in another study involving 304 patients (86) Similar performance was reported by Gomez-Reino et al. (101) where the one year persistence (95%CIs) was significantly lower for INF (0.81 (0.79: 0.82)) than for ETN (0.88 (0.85: 0.91)) and ADA (0.87 (0.82: 0.91)). Studies that reported on persistence with one anti-tnf therapy (INF) The literature search identified eight studies that examined the persistence with one anti -TNF therapy only (INF), one of them compared persistence in older and younger populations (in both RA and ankylosing spondylitis (AS) population) and no differences were identified. (81) The persistence rates were reported in five studies (Table 2.3). The remaining two studies examined INF persistence in combination with LEF or MTX The effect of nbdmard co-therapy on drug persistence comparing across the three anti-tnf therapies Only one study reported superior persistence with INF where the one year persistence was 78% compared to 72.8% in ETN and 70.8% in ADA (p=0.005) (Table 2.4). (88) This analysis was limited to anti-tnf therapy when taken in combination with MTX and suggests better persistence of INF when taken with MTX (as licensed). The exclusion of ETN and ADA patients receiving no or other nbdmards may be the reason of this superior persistence with INF. 64

65 A threefold higher risk of drug discontinuation was reported by Kristensen et al. (84) in patients treated with INF compared to ETN (p<0.001). The difference between the two drugs was still significant when the analysis was undertaken in subgroups of patients treated with MTX, other nbdmards or anti-tnf mono-therapy. Table 2.4: Persistence rates with anti-tnf therapies when taken alone, in combination with MTX or in combination with other nbdmards Study Year Persistence rates (%) Anti-TNF + MTX Anti-TNF monotherapy Anti-TNF + other nbdmards ETN INF ADA ETN INF ETN INF Kristensen et al. (84) 1 st year nd year th year th year Zink et al. (82) 1 st year Tang et al. (88) 1 st year ETN: Etanercept, INF: Infliximab, ADA: Adalimumab. Zink et al. (82) reported that the use of other nbdmards with anti-tnf therapies was superior to MTX and no nbdmards therapy with more apparent differences in the INF cohort than in the ETN cohort; the corresponding one year persistence rate with MTX, other nbdmards and no nbdmards were 71.6%, 74.9 % and 64.4% for ETN; and 66.2%, 72.1% and 44.2% for INF). 65

66 Comparison of drug persistence with and without nbdmards The effect of the concomitant use of nbdmards with the anti-tnf therapies on the persistence with these anti-tnf therapies was addressed in nine studies. Of these, eight studies reported higher anti - TNF therapy persistence rates when taken in combination with nbdmards. Duclos et al. (83) was the first study to compare anti-tnf persistence rates with and without nbdmards. The persistence rates were higher without concomitant nbdmards (p=0.001) but when the analysis was limited to MTX the significant difference disappeared (p=0.59). Role of concomitant use of MTX The role of concomitant MTX therapy compared to other nbdmards or no nbdmards was studied by Kristensen et al. (84) and the results reported that patients receiving concomitant MTX were more likely to continue the anti-tnf therapy than those receiving other nbdmards (p<0.01) or those receiving no nbdmards (p<0.001). Similar results were found by Heiberg et al. (90) in which concomitant MTX therapy was associated with better drug persistence compared to mono-therapy (p<0.001). In addition, Ostergaard et al. (97) reported a tendency towards better drug persistence with concomitant MTX than without MTX (p=0.08). Zinc et al. (82) reported that other nbdmards therapy was superior to MTX and no nbdmards therapy. Role of concomitant use of LEF and other nbdmards The role of concomitant LEF and other nbdmards (compared to MTX) on anti -TNF therapy persistence was assessed by Finckh et al. (99) and the results showed no significant differences in the overall persistence rates between the three anti-tnf therapies (p=0.69). The authors suggested that the use of LEF and other nbdmards is as successful as the use of MTX which is the most commonly used concomitant nbdmards with the anti-tnf therapies. 66

67 Flendrie et al. (93) compared the persistence with INF when taken with LEF (LEF group) and when taken alone or with other nbdmard (non-lef group); no differences between the two groups were detected. When the comparison involved only MTX versus LEF in the study by Ibanez et al. (94), the persistence was superior in the INF+MTX cohort than in the INF+LEF cohort and the main reason for treatment discontinuation in the INF+LEF cohort was due to adverse events. Grijalva et al. (87) compared the persistence with treatment with twelve different therapies including anti-tnf therapies, nbdmards or combination treatment. INF mono-therapy was more likely to be discontinued compared to ETN alone and ADA+MTX. Patients receiving INF alone were more likely to discontinue the medication than INF+MTX (p=0.01) while the opposite occurred with ETN as ETN+MTX was more likely to be discontinued than ETN alone (p=0.0222) Other predictors of anti-tnf therapies discontinuation Seven studies looked at predictors of anti-tnf therapy discontinuation (Table 2.5) other than concomitant use of nbdmard therapy; all of them used a Cox regression model to identify the predictors of discontinuation. It is difficult to compare the results from these studies because different covariates were often considered in the different studies. However, a few common general areas were studied. Demographic variables Older age (HR (95% CIs) 1.11 (1.01: 1.22)) was found to be a significant predictor of drug discontinuation. (84) Older age was also associated with the discontinuation of INF due to adverse events. (100) In contrast, older age (more than 60) was not associated with treatment discontinuation in the study by Gomez-Reino et al. (101) 67

68 Female sex was significantly associated with treatment discontinuation in the study by Heiberg et al. (90) while it was not associated with treatment discontinuation in the study by Gomez-Reino et al. (101) Male sex was associated with INF discontinuation due to adverse events in the stratified Cox regression carried out by Tanaka et al. (100) Disease features High HAQ score (HR (95% CIs) 1.17 (1.06: 1.30)) and a high number of previous nbdmards (HR (95% CIs) 1.13 (1.02: 1.25)) were significant predictors of anti-tnf therapy discontinuation while high baseline CRP level (HR (95% CIs) 0.90 (0.81: 0.98)) was a significant predictor of high treatment persistence rate. (84) In contrast, the number of previous nbdmards was not associated with treatment discontinuation in the study by Heiberg et al. (90) Rheumatoid factor negative disease was associated with the discontinuation of INF due to adverse events. (100) High disease activity was significantly associated with treatment discontinuation while disease duration was not a significant predictor of discontinuation in the one year performance study by Heiberg et al. (90) TJC (HR (95% CIs) 1.17 (1.05: 1.31)) was associated with higher INF discontinuation in the multivariate analysis carried out by Bertoli et al. (89) 68

69 Table 2.5: Other predictors of anti-tnf therapies discontinuation (from the literature) Covariates Studies involved the three anti-tnf therapies Studies involved INF only Duclos et al. (83) Kristensen et al. (84) Tang et al. (88) Heiberg et al. (90) Demographic variables Gomez- Reino et al. (101) Tanaka et al. (100) Bertoli et al. (89) Old age - Yes No - No Yes - Female sex - - No Yes No No - Disease duration Number of previous nbdmards RF negative disease Disease activity Disease features No - Yes No Yes Yes - No Yes Yes HAQ score - Yes CRP level - No TJC Yes - Not studied. TNF: tumour necrosis factors, nbdmards: non-biologic disease modifying antirheumatic drugs, RF: rheumatoid factor, HAQ: health assessment questionnaire, CRP: C reactive protein, TJC: tender joint count. 69

70 2.1.7 Persistence with second alternative anti-tnf therapies after switching from one anti-tnf to another The literature search identified four studies that compared treatment persistence with a second course of anti-tnf therapy with a first course. Gomez-Reino et al. (101) reported that drug persistence was lower when the anti-tnf therapies were used as a second course (0.79 (0.74: 0.83)) than the first course (0.83 (0.82: 0.84)). However, Hyrich et al. (102) reported that the overall persistence (due to all reasons of discontinuation) of the second course was similar to the first course; but when the analysis was restricted to the reason for switching (adverse events or inefficacy) a significant difference appeared. The study reported that the reason for discontinuation of the second course was related to the reason for discontinuation of the first anti-tnf therapy. When the reason for switching was ineffectiveness the rate of discontinuation of the second course for ineffectiveness was higher than the first course. Similarly, patients who stopped their first course for an adverse event were also more likely to stop their second course for an adverse event as well. In contrast, Hjardem et al. (103) reported higher persistence with the second course compared to the first course after switching from anti-tnf therapy to another Discussion The one year overall persistence with anti-tnf therapies ranged from 65.4% to 83.0%. Inferior persistence with INF compared to ETN and ADA was more common in the studies reporting on the comparison of drug persistence between the three anti-tnf therapies. The persistence with INF was better when taken with concomitant MTX therapy. The persistence rates in different studies were difficult to compare directly due to the differences in the baseline characteristics of the included population. The effect of these characteristics increases especially when these characteristics represented a predictor of drug discontinuation. 70

71 In clinical practice the use of the anti-tnf therapies may differ from the approved use in terms of the use of concomitant nbdmards. Although ETN and ADA are approved to be used alone, in clinical practice ADA and ETN are often prescribed with one or more nbdmards. Likewise, INF which is approved to be used with MTX is sometimes used alone or with other nbdmards. This may be due to the fact that when patients are prescribed an anti-tnf therapy, they tend to continue, stop, add and/or replace their nbdmards. Little is known about the role of these concomitantly used nbdmards on the persistence with the anti-tnf therapies. Most of the studies that addressed the role of concomitant nbdmards concentrated on MTX an d combined all the other nbdmards in one group mainly due to small sample size, making it difficult to appreciate any differences between biologic/nbdmard combinations. 71

72 2.2 Assessment of the use of RTX in RA from both randomised controlled trials (RCTs) and observational studies: literature review The efficacy of RTX in RA patients was evaluated via a critical review of the published literature. A search of electronic databases was developed to identify RCTs and observational studies that assessed the efficacy and clinical effectiveness of a first course of RTX in the treatment of RA. EMBASE, MEDLINE and the Cochrane library databases were used for the search (see search strategies appendices 2.2 and 2.3). The search was limited to publications from 1995 onwards since RTX is very recently introduced in the management of RA. Articles written in languages other than English were excluded. Appendix 2.2 includes the search strategy for the randomised controlled trials, while Appendix 2.3 includes the search strategy for the observational studies of RTX in RA. Data from relevant studies were extracted, tabulated and compared. The inclusion criteria for the randomised controlled trials were: inclusion of RA patients, the use of RTX, assessment of efficacy of a first course of RTX, presence of a comparator arm (placebo or active drug comparison) and randomisation of patients. While the inclusion criteria for the observational studies were inclusion of a cohort of RA patients (minimum sample size of ten pati ents), the use of RTX, assessment of effectiveness of a first course of RTX. The observational study should identify, observe, record, classify and analyse information. If two publications reported on the same group of patients, only the most recent publication was included. If the assessment did not include any clinical or patient reported outcomes, the study was excluded. 72

73 2.2.1 Randomised controlled clinical trials (RCTs) of RTX in RA According to the search strategy, five RCTs were identified. Table 2.6 shows a summary of the design characteristics of these trials, while Table 2.7 reports on the demographic characteristics of patients included in these trials. The patients included in the IMAGE trial were younger and had shorter disease duration and 70% of them were nbdmard naïve. The mean age of the other studies participants was 52.3 years and the percentage of females ranged from 73% to 86%. The disease duration was in the range from 6.6 years to 12.1 years and the patients had failed a mean of 2.2 previous nbdmards. The mean disease activity (DAS28) in all studies was 2.8 with the highest disease activity shown in IMAGE trial patients. Where HAQ was recorded, the mean HAQ score was 1.8, indicating moderate to severe levels of disability (Table 2.7). 73

74 Table 2.6: Design characteristics of the RCTs of first course of RTX in RA Controlled trial Study design Duration Number of patients (ITT) Comparisons Early RA MTX naïve patients IMAGE (106;107) Edwards et al (108;109) REFLEX (110;111) SERENE (112) DANCER (113;114) Randomised, double-blind, placebo-controlled phase III international dose ranging study Multi-centre randomised, double-blind, placebocontrolled study Multi-centre, randomised, double-blind, placebo-controlled phase III study Randomised, double-blind, placebo-controlled phase III international dose ranging study Randomised, double-blind, placebocontrolled, international, dose-ranging, phase II b trial 52 weeks 755 (748) (from 169 centres in 27 countries) nbdmard Failures (2 years) Primary end point was 24 weeks (2 years) Primary end point was 24 weeks 48 weeks (Primary end point was 24 weeks ) 161 (161) 11 countries International Anti-TNF Failures 520 (499) US, Europe, Canada and Israel 511 (509) (from 102 centres in 11 countries) nbdmard failures and/or anti-tnf Failures 24 weeks 465 (367) International 1.placebo + MTX (n=249) 2. RTX (500 mg) + MTX (n=249) 3. RTX (1000 mg) + MTX (n=250) 1.MTX (n=40) 2.RTX(n=41) 3.RTX+CTX (n=40) 4. RTX + MTX (n=40) 1.placebo + MTX (n=209) 2. RTX + MTX (n=311) 1.placebo + MTX (n=172) 2. RTX (500 mg) + MTX (n=167) 3. RTX (1000 mg) + MTX (n=170) 1.placebo + MTX (n=149) 2. RTX (500 mg) + MTX (n=124) 3. RTX (1000 mg) + MTX (n=192) With either : 1. No glucocorticoids, 2. Intravenous methylprednisolone 3. Intravenous methylprednisolone + oral prednisone. Outcome measures DAS28 HAQ ACR50* ACR 20, ACR 70 EULAR ACR20 * ACR 50,70 DAS28 EULAR HAQ SF-36 ACR20 * ACR 50,70 DAS28 EULAR HAQ SF-36 ACR20 * ACR 50, ACR 70 DAS28 EULAR SF-36 HAQ * Primary outcome measure. Placebos of RTX, MTX and CTX were used when the corresponding drug was not administered. RA: rheumatoid arthritis, MTX: methotrexate, CTX: cyclophosphamide, RTX: rituximab, ACR: American College of Rheumatology, EULAR: the European League against Rheumatism, HAQ: Health Assessment Questionnaire, SF-36: Short Form 36, VAS-pain: visual analogue scale, PGA: Patient Global Assessment of Disease Activity, ITT: intent to treat population, nbdmard: non-biologic disease-modifying antirheumatic drugs. 74

75 Table 2.7: Baseline characteristics of patients included in the RCTs of RTX in RA Study arms Study Edwards et al (108) REFLEX (110) DANCER (113) SERENE (112) IMAGE (107) MTX RTX RTX+ CTX RTX+ MTX MTX* RTX+ MTX MTX* mg RTX + MTX 1000mg RTX + MTX MTX* 500mg RTX + MTX 1000mg RTX + MTX ITT Age (yr) Female sex (%) Disease duration (yr) RF positive (%) Previous nbdmard (other than MTX) Previous anti-tnfs (%) MTX* 500mg RTX + MTX 1000mg RTX + MTX NR (70% of the patients were nbdmard naïve) SJC (no.) TJC (no.) $ CRP (mg/dl) NR ESR (mm/hr) NR DAS HAQ NR Values are mean otherwise specified. MTX: methotrexate, CTX: cyclophosphamide, RTX: rituximab, SJC: swollen joints count, TJC: tender joints count, ESR: e rythrocyte sedimentation rate, DAS: diseaseactivity score, CRP: C-reactive protein, nbdmard: non-biologic disease-modifying anti-rheumatic drugs, NR: not reported, ITT: intent to treat population, RF: rheumatoid factor, HAQ: health assessment questionnaire. SJC was assessed in all joints in the Edward and the DANCER trials and in 66 joint in the REFLEX, SERENE, and IMAGE trials. $ TJC was assessed in all joints in the Edward and the DANCER trials and in 68 joint in the REFLEX, SERENE, and IMAGE trials. Placebos of RTX, MTX and CTX were used when the corresponding drug was not administered. * Plus placebo RTX

76 2.2.2 Efficacy of RTX in RA patients from RCTs The efficacy results of the RCTs are represented in Tables 2.8 and 2.9. Table 2.8 presents the change in different disease activity measures from baseline, while Table 2.9 reports on patients reported outcomes. The two tables show results taken at 24 weeks from baseline. The REFLEX and DANCER trials used the ACR20 as the primary outcome measure which is not routinely used in clinical practice, which increases the importance of the other used outcomes measures in these trials. From Table 2.8, it is clear that the response in the RTX treated groups is higher than the placebo group Efficacy of RTX mono-therapy and with concomitant nbdmards According to the licensed indications, RTX should be used in combination with MTX ex cept in those patients who have intolerance or contraindications to MTX when RTX mono-therapy is recommended. In Edwards et al clinical trial of RTX in RA, (108) RTX mono-therapy was reported to have higher response than MTX but the difference was not statistically significant. This finding is not surprising because those patients were MTX failures. The study did not compare statistically the efficacy of RTX mono-therapy to the combined therapy of RTX with cyclophosphamide and MTX which would have been helpful in order to provide a clearer picture of the results and to determine which treatment produced the best clinical response. The Edwards et al. (2004) clinical trial of RTX in RA (108) showed that the RTX combination either with cyclophosphamide or MTX resulted in significantly higher improvement than the MTX plus placebo group. When patients were followed up to 2 years by Strand et al. (2006) (109) without further RTX treatment (patients who received MTX continued receiving it), at week 104, the percentage of patients who remained in the follow-up were 15, 10, 21.9 and 45% in the placebo, RTX, RTX plus cyclophosphamide and RTX plus MTX groups respectively. This suggests that combination MTX therapy can lead to a more sustained response following RTX treatment in RA. 76

77 This may be the reason why further trials used RTX in combination with MTX. In addition RTX mono-therapy seemed to perform worse than MTX with a low percentage of patients. In the REFLEX and DANCER, SERENE, and IMAGE trials, patients received concomitant MTX with RTX. In these all trials the RTX group showed significantly better clinical efficacy than MTX plus placebo group. In the REFLEX trial, at 24 weeks the ACR 20, 50, 70 and EULAR response were significantly greater in the RTX group than the placebo group (p < in each comparison). Also the RTX treated group showed a significant decrease in DAS28 score compared to the placebo treated group. (110) Similar results were observed in DANCER trial where the proportions of patients who achieved ACR20, ACR50 and ACR70 response were significantly greater in both RTX groups compared to the placebo group (p< ) (Figure 2.1). (113) The DANCER trial also examined the use of concomitant glucocorticoids (intravenous methylprednisolone premedication, intravenous methylprednisolone premedication plus oral prednisone for 2 weeks or placebo glucocorticoids) and the results showed that glucocorticoids had no significant effect on ACR20 response. 77

78 Table 2.8: Change from baseline in response rates at week 24. Study Edwards et al (108) REFLEX (110) DANCER (113) SERENE (112) IMAGE (107) Study arms MTX RTX RTX+ CTX RTX+ MTX MTX * RTX+ MTX MTX * 500mg RTX + MTX 1000mg RTX + MTX MTX * 500mg RTX + MTX 1000mg RTX + MTX ITT MTX * 500mg RTX + MTX 1000mg RTX + MTX Change in DAS28 (mean ± SD) -1.3 ± ± ± ± Patient achieving disease remission (%) NR 0 9 NR Moderate or good EULAR Response (%) Patients achieving ACR (%) NR Patients achieving 33 ACR50 13 P=0.059 (%) Patients achieving ACR70 (%) MTX: methotrxate, CTX: cyclophosphamide, RTX: rituximab, ACR: American College of Rheumatology, EULAR: the European League Against Rheumatism, NR: not reported. P values are compared to control group< Obtained from the published figure. Placebos of RTX, MTX and CTX were use d when the corresponding drug was not administered. * Plus placebo RTX 78

79 MTX* MTX RTX RTX+CTX RTX+MTX RTX+MTX MTX * RTX 500mg+MTX RTX 1000mg+MTX MTX * Percentage of patients RTX 500mg+MTX RTX 1000mg+MTX Edward REFLEX DANCER SERENE Figure 2.1: Patients achieving ACR20 response at 24 weeks in the RCTs of RTX in RA. Results from the RCTs of RTX in RA. (108;110;112;113). Placebos of RTX, MTX and CTX were used when the corresponding drug was not administered. * Plus placebo RTX 79

80 Table 2.9: Change from baseline in patients quality of life outcomes reported in the RCTs (24 week endpoint). Study Strand et al (109) REFLEX (111) DANCER (114) SERENE (112) IMAGE (106) Study arms MTX RTX Change in HAQ Patients achieving MCID in HAQ (%) Change in SF-36 PCS Change in SF-36 MCS RTX+ CTX RTX+ MTX MTX * RTX+ MTX MTX * RTX (500mg) + MTX RTX (1000mg) + MTX MTX * RTX (500mg) + MTX RTX (1000mg) + MTX MTX * RTX (500mg) + MTX Mean Mean ± DS Mean (SE) Mean Mean ± ± 0.6 RTX (1000mg) + MTX NR NR 1.48 ± ± ± ± (0.78) 2.15 (0.89) 7.08 (0.77) 7.40 (0.78) (0.80) 4.52 (0.87) MTX: methotrexate, CTX: cyclophosphamide, RTX: rituximab, SF-36: short form 36, HAQ: Health Assessment Questionnaire, MCID: minimal clinically important difference, PCS: physical component summary, MCS: mental component summary, NR: not reported, SE: standard error. Placebos of RTX, MTX and CTX were used when the corresponding drug was not administered. * Plus placebo RTX. P values are compared to control group < NR 80

81 Efficacy of RTX in anti-tnf naïve and in anti-tnf failures Patients included in the SERENE trial and the Edwards et al clinical trials of RTX were nbdmard failures. On the other hand, patients enrolled in the REFLEX trial were anti -TNF failures. The DANCER trial enrolled patients who failed (at least one but not more than 5) nbdmard (other than MTX) and/or biologic therapies. The licensed indications for RTX in patients who have had an inadequate response or in tolerance to nbdmard including one o r more TNF inhibitor therapies was granted on evidence from the REFLEX trial as it enrolled only the anti-tnf therapy failures population. The results of the DANCER trial would be strengthened if the two groups (nbdmard failures and biologic failures) had been separated and compared. None of the RCTs compared both populations of patients (nbdmard failures and anti -TNF failures), which makes it difficult to compare the efficacy of RTX in these 2 different populations. However, comparing results between studies would be more beneficial as it would help to study the clinical efficacy of RTX in different patient population groups. Figure 2.2 shows the moderate and good EULAR response in the different arms in the RCTs and it is noticed from the Figure that the patients of the Edwards et al. trial showed the highest response followed by the SERENE trial, DANCER trial then the REFLEX trial. This might result from the included population, as the Edwards et al., 2004 and SERENE trials included nbdmard failures (showed the highest response) while the REFLEX trial included anti-tnf failures (the lowest response) and the DANCER trial included both populations (intermediate response). When the placebo responses in the three trials were compared, it is found that the REFLEX trial had the lowest placebo response and the number of responders is low. This may explain why it is difficult to compare results between studies. 81

82 MTX* MTX MTX * RTX RTX+CTX RTX+MTX RTX+MTX RTX 500mg+MTX RTX 1000mg+MTX MTX * Percentage of patients RTX 500mg+MTX RTX 1000mg+MTX Edward REFLEX DANCER SERENE Figure 2.2: Moderate or good EULAR response at 24 weeks in the RCTs of RTX in RA, results from RCTs of RTX in RA. (108;110;112;113) Placebos of RTX, MTX and CTX were used when the corresponding drug was not administered. * Plus placebo RTX Efficacy of RTX in RF positive versus negative patients B cells are believed to be the source of RF which may explain the very early thoughts that RTX will act only on RF positive patients. Edward et al. trial included only RF positive patients. The study population in the DANCER trial were mainly RF positive however very limited number of RF negative were included to assess the safety in this group of patients. In the REFLEX trial, the authors reported that fewer RF negative patients achieved ACR20 compared to RF positive patient. SERENE trial authors also reported greater improvements in DAS28 in RF positive (n=144) patients than in RF negative patients (n=24). Similar findings were reported in the IMAGE trial where enhanced response was noticed in RF positive and/or anti-citrullinated protein antibody (ACPA) positive (n=218) compared to RF negative and/or ACPA negative (n=24); at week 52 the proportions of patients achieving ACR20 were 81% and 71% respectively. It should be pointed out that the small sample size in these subgroups may limit the findings. 82

83 Percentage of patients Larger sample size is required to assess this question of response within these subgroups of patients Efficacy of different doses of RTX The DANCER, SERENE and IMAGE trials used two different doses of RTX (1000 and 500 mg). Within the three trials, the two doses of RTX showed significant superior clinical and patient reported responses compared to the placebo group. However, in the DANCER trial the ACR20 and ACR50 responses were similar but the ACR 70 responses were more frequent in RTX 1000 mg dose group than in the 500 mg dose group (see Figure 2.3) but no statistical comparison between the responses in the two doses arms was performed. Moreover, the authors of the IMAGE trial found that only the 1000mg dose can produce significant inhibition of joint damage compared to MTX. Since the inhibition of joint damage is an important achieveme nt in the management of the disease, the 1000mg dose may be the optimal dose of RTX in RA patients MTX * RTX 500mg+MTX RTX 1000mg+MTX ACR20 ACR50 ACR70 ACR20 ACR50 ACR70 DANCER SERENE Figure 2.3: Patients achieving ACR20, 50 and 70 responses in the DANCER trial and SERENE trials (112;113). * Plus placebo RTX. 83

84 Efficacy of repeated doses of RTX Currently there are two published RCTs that looked at the efficacy of a second course of RTX in RA patients (Table 2.10). (115;116) In both trials a second course of RTX resulted in sustained improvements of the disease with similar safety profile. The response of different doses after retreatment was tested and both RTX 500 mg and 100mg doses were effective with better improvements in 1000mg dose in some efficacy outcomes at week 48. (116) However, there might be a problem in the randomisation in this trial as stated by the authors; Some patients received a treatment regimen other than that to which they were randomly assigned Table 2.10: General design characteristics of the RCTs of second course of RTX in RA Number of Patient Study Study design patients Study arms population (ITT) SUNRISE trial, 2010 (115) MIRROR trial, 2010 (116) Open label first course of RTX followed by randomised placebo controlled second course of RTX at 24 weeks International multicenter randomised placebo controlled phase III trial 559 (475) (from 143 centres in the USA) 378 (346) (from 81 centre in 18 countries) 1.placebo+MTX (n=157) 2.RTX+MTX (n=318) 1. Two courses of RTX 500mg+MTX (n=134) 2. Two courses of RTX 1000mg+MTX (n=93) 3. First course of RTX 500mg+MTX, and second course of RTX 1000mg+MTX (n=119) Patients had failed at least one anti-tnf therapy ~26% of the patient had failed previous anti- TNF therapy Primary end point ACR20 at week 48 ACR20 at week 48 MTX: methotrexate, RTX: rituximab, RA: rheumatoid arthritis, ACR: American college of rheumatology, and ITT: intent to treat. 84

85 On-going RCTs of RTX in RA An internet search identified a number of on-going phase III RCTs of RTX in RA. Most of them are studying the performance of RTX in different populations of patients such as MTX naïve patients, MTX failures, single anti-tnf therapy failures, and multiple anti-tnf therapy failures. Others are investigating repeated doses of RTX. The efficacy and safety of RTX in RA associated lung disease is also being investigated. Some RCTs are looking at the efficacy of RTX in combination with nbdmard other than MTX such as leflunomide or in combination with other biologics such as tocilizumab. Comparative efficacy of RTX versus other biologics after failing anti-tnf therapy is also been investigated Summary This review of the efficacy of RTX presented in three RCTs has shown that RTX is an effective treatment in RA patients. However, these trials did not cover certain important information about RTX efficacy (such as the comparative efficacy versus other available biologic options of treatment). This literature review also examines the observational studies of RTX in RA in order to understand more about RTX in RA. The observational studies use real life populations and the inclusion criteria of patients are usually wider than in the RCTs which make the results more externally valid. Moreover, observational studies can provide further data on the effectiveness of RTX therapy Observational studies of RTX in RA The literature search found 20 observational studies (Table 2.11). Overall, the individual studies vary in sample size and the majority did not include a comparator group. Half of the identified observational studies were of sample size less than 50 patients. 85

86 2.2.4 Effectiveness of RTX in RA patients from the observational studies Effectiveness of RTX mono-therapy and with concomitant nbdmard One of the identified observational studies reported on the use of RTX mono-therapy. (117) However, six other studies used RTX mono-therapy in some patients but no comparisons were performed between those patients and rest of the population. Owczarczyk K. et al (117) enrolled in total 40 patients; 20 of them received RTX mono-therapy and the other 20 patients received RTX with concomitant MTX. No statistically significant differences in the DAS28 score and EULAR response were identified between these two groups at week 16. However, the concomitant MTX group showed a higher percentage of patients achieving ACR50 and ACR70 or maintaining the response after week 48. This suggests that this combination is preferred especially in patients tolerating MTX. A retrospective study of 57 patients reported more frequent EULAR responders in patient receiving RTX in combination with other nbdmards (85%) compared to those receiving it with MTX (70%). (118) Another retrospective study of ten patients had found that LEF is a possible successful concomitant nbdmard with RTX in case of intolerance to MTX. (119) No difference in response of RTX mono-therapy versus in combination with MTX was reported in a recent study of 97 patients. (120) All the other observational studies used concomitant MTX with RTX (some studies used it in some patients only) and some studies used other concomitant nbdmard such as cyclophosphamide, leflunomide and hydroxychloroquine. (117; ) However, none of these studies reported on the difference in response between concomitant nbdmard users versus non nbdmard users. 86

87 Table 2.11: General design characteristics of observational studies of RTX in RA Observational Study Duration Patients Number (nationality) Treatment protocol nbdmard failures McGonagle et al. 12 months (124) (UK) Brulhart et al (125) 6 months (only 3 patients were followed up to 1 year) 10 (Switzerland) Scheinberg et al. 3 months (126) (Brazil) Bokarewa M. et al months (121) (Sweden) Finckh et al (122) Finckh et al (127) Up to 9 months 116 (Switzerland) Up to 9 months 318 (Switzerland) Jois et al (123) 6 months 20 (UK) 2 RTX infusions 2 weeks apart. (either 1000mg dose or 500mg were received) Anti-TNF failures 2 infusions of 1000 mg RTX at 2 weeks intervals. 375 mg/m 2 weekly for 4 weeks, or 2 infusions of 1000 mg, 2 weeks apart. 87 Outcome measures EULAR DAS28 DAS28 CRP ESR DAS mg/m 2 weekly for 4 weeks DAS28 -its components 50 patients received RTX and 66 patients received a second or a third anti-tnf alternative 155 patients received RTX and 163 patients received a second or a third anti-tnf alternative 2 infusions of 1000 mg RTX two weeks apart. DAS28 - its components DAS28 EULAR DAS28-HAQ Authors conclusions RTX is effective as first line biological therapy and can be given to patients with anti-tnfs contraindications RTX is effective in RA patients who have failed anti - TNF RTX seems to be safe and effective in RA patients. RTX is effective and well tolerated in RA patients who have failed anti- TNF RTX is more effective than switching to alternative anti - TNF therapies in patients who had failed anti-tnf therapy When the reason for stopping the previous anti -TNF is ineffectiveness, switching to RTX is more effective than alternative anti-tnf RTX is effective in patients resistant to anti -TNF therapies and DMARD Tsiakalos et al (128) Up to 24 weeks 17 (Greece) 2 infusions of 1000 mg RTX two weeks apart. ACR20,50,70 DAS28-HAQ RTX is effective. Similar response in RF positive versus negative patients. Anti-TNF and anakinra failures Henes JC et al (129) 6 months 15 (Germany) 2 infusions of 1000 mg RTX two weeks apart DAS28 RTX is effective in this very resistant population of patients. RA: rheumatoid arthritis, MTX: methotrexate, RTX: rituximab, ESR: erythrocytes sedimentation rate, DAS: disease-activity score, nbdmard: non-biologic disease-modifying antirheumatic drugs, CRP: C-reactive protein, HAQ: Health Assessment Questionnaire, EULAR: European League Against Rheumatism, TNF: tumour necrosis factor, NR: not reported.

88 Table 2.11 (continued): General design characteristics of observational studies of RTX in RA Observational study Duration Patients Number (nationality) Treatment protocol Outcome measures Authors conclusions Mixed DMARD failures and anti-tnf failures Yin HL et al (130) 24 weeks 10 2 infusions of 1000 mg EULAR RTX is found to be effective and well tolerated. (China) RTX DAS28, HAQ Assous et al (131) 24 weeks 50 (France) 2 infusions of 1000 mg RTX two weeks apart EULAR DAS28 Single cycle of RTX+MTX provide significant improvements and good tolerance Galarza C. et al (132) NR 21 (Colombia and Ecuador) Owczarczyk K. et al. 16 weeks (117) (Germany) Valleala, et al Median of 4 months 57 (118) (retrospective) (Finland) Vital et al (133) 6 months 103 (UK) Henes et al (119) 6 months 10 (Germany) Vander,Cruyssen et al (134) Up to median of 70 weeks 401 (Belgium) Quartuccio et al months 110 (135) (Italy) Chatzidionysiou et al (136) Up to 6 months 2011 (10 European registries) Solau-Gervais et al. 20 weeks (120) (France) 375 mg/m 2 weekly for 4 weeks, or 2 infusions of 1000 mg NR 2 infusions of 1000 mg RTX 2 infusions of 1000 mg RTX 375 mg/m 2 weekly for 4 weeks, or 2 infusions of 1000 mg 2 infusions of 1000 mg RTX 375 mg/m 2 weekly for 4 weeks, or 2 infusions of 1000 mg NR 2 infusions of 1000 mg RTX 88 DAS28 HAQ DAS28 EULAR DAS28, EULAR DAS28 EULAR DAS28 EULAR DAS28 EULAR DAS28 EULAR ACR50 DAS28 EULAR DAS28 EULAR RTX might be used as the first-choice treatment for patients with RA. RTX mono-therapy is effective in patients with active RA (combination with MTx is preferred). RTX is effective in combination with MTX and other nbdmards Low DAS28 and RTX mono-therapy were predictors of low response rates to RTX RTX can be used in combination with Leflunomide when MTX is not tolerated DAS28 decreases after treatment. Patients failing higher number of anti-tnf were less likely to respond RF positivity, lower HAQ, and lower number of previous anti -TNF therapies were associated with ACR50 Seropositive patients and those who failed no or not more than one anti-tnf showed superior response Similar response in anti-tnf naïve versus failures, and in RTX mono therapy versus with MTX and higher response in RF positive subjects RA: rheumatoid arthritis, MTX: methotrexate, RTX: rituximab, DAS28: disease-activity score 28 joints, nbdmard: non-biologic disease-modifying anti-rheumatic drugs, HAQ: Health Assessment Questionnaire, EULAR: European League Against Rheumatism, TNF: tumour necrosis factor, NR: not reported.

89 Effectiveness of RTX in nbdmards failures and in anti-tnfs failures McGonagle et al was the only identified study that looked at RTX as a first biologic in nbdmard failures and the results of this study showed that RTX may be used as first-line biological therapy in the clinical settings and can be safely used in patients with contraindications to anti-tnfs. Eight observational studies included anti-tnf failures; one of them also enrolled patients with contraindications to anti-tnf (131) and another one included patients that had also failed anakinra (129) (see Table 2.11). Assous et al (131) examined patients who had failed anti-tnfs, ten of them had failed all three available anti-tnfs, and patients who had contraindications to anti-tnfs. These two groups had no significant difference in the base-line characteristics and the results of the study showed no significant difference in response of the two groups. No significant difference was also found between rheumatoid factor (RF) positive and RF negative patients. (131) Henes JC. et al included very resistant RA patients who failed nbdmard, anti-tnfs and anakinra; the results showed that the mean DAS28 improved from 5.9 to 3.9 at the third month and then stayed at 3.9 till the ninth month; in addition, 80% of patients showed at least a moderate EULAR response with 20% of the cohort reaching a good response after three months. RTX was safe and effective in this very resistant population. (129) Eleven observational studies included mixed populations of nbdmards failures and anti-tnf failures but not all of them had compared the response across the two groups. Owczarczyk K. et al. (117) compared the response of these two populations and the results showed no significant difference in response between them as measured by DAS28 (these two populations in this study were subgroups of the RTX mono-therapy group). Similar findings were reported by Solau-Gervais et al. in 95 patients. (120) All the identified studies demonstrated the effectiveness of RTX without comparing it to other therapies except, one interesting study of Finckh et al (122) compared the effectiveness of 89

90 RTX with alternative anti-tnf therapy in patients who failed at least one anti-tnfs. At 6 months the RTX treated population showed a statistically significant decrease in mean DAS28 compared to the alternative anti-tnf treated population (p = 0.01); the mean decrease in the DAS28 was among patients treated with RTX and among those treated with alternative anti-tnf therapy. An update of this study with larger population concentrated on subgroups according to the number of previous biologic therapies, reason for discontinuation of the last biologic therapy, type of last therapy and the use of concomitant nbdmards. (127) Effectiveness of different doses of RTX It is noticed that different doses and different dosing protocols were used in the identified observational studies and this makes the comparison and combining of the results of these studies difficult. The reason why those studies used these different dosing protoco ls may be that some of them started before the results of the RCTs especially DANCER and REFLEX trials were published or before the drug received its license for RA. Ten studies used two 1000 mg infusions of RTX given two weeks apart. Three studies used f our weekly infusions of 375 mg RTX each. Four studies combined these two protocols of dosing but none of them compared the effectiveness of these protocols. One study used 2 doses 500 mg and 1000 mg taken as two infusions two weeks apart. (124) The results of this study showed that patients who received 500 mg RTX showed less response, measured as EULAR response at 3, 6, 9 and 12 months, than those receiving 1000 mg dose of RTX but the difference in response was not examined statistically in the study. A recent study by Vital et al. using highly sensitive flow cytometry assays had clarified that the response to RTX is dependent on the degree of B dell depletion rather than the dose used (500 or 1000mg) and this can be used to predict the patients who will respond to lower dose of RTX. (137) 90

91 Effectiveness in RF positive versus RF negative patients It is noticed that some studies were limited to RF positive patients. However most of the studies had included mixed population of RF positive and negative patients. RF negative patients seem to respond to RTX however some studies reported superior response in RF positive patients. A study of 17 patients reported similar response rates in RF positive and negative patients. (128) Three other large studies reported that RF positive patients were more likely to respond to RTX. (120;135;136) Predictors of response to RTX The literature search identified only three studies that looked at predictors of response to RTX. The three of them were recent (2009, 2010, and 2011) and looked at possible predictors of clinical outcome measures such as EULAR or ACR50 response criteria rather than the patient reported outcomes such as the physical function. Two studies used multivariate regression models while the third one used only univariate models. A study of 110 patients (32 anti -TNF naive and 78 anti-tnf failures) reported that patients with positive RF, lower baseline HAQ score, or those who had failed fewer anti-tnf agents were more likely to achieve an ACR 50 response. (135) Another recent study of 103 patients (40 anti-tnf naive and 63 anti-tnf failures) identified high DAS28 and concomitant use of nbdmards as predictors of response to first course of RTX in univariate logistic regression. (133) Very recent multiple registers study (674 anti-tnf naive and 1170 anti-tnf failures) has identified low DAS28 score, lower number of previous biologics and anti-citrullinated protein antibody (ACPA) as predictors of good EULAR response. (136) Effectiveness of repeated doses of RTX 91

92 Assous et al reported that the a single cycle of treatment may lead to sustained effectiveness without re-treatment and this is economically and clinically interesting, but longterm data are required to judge this. (138) Some observational studies considered re-treatment with RTX, although most of these studies included very few patients. In addition, the follow-up time was very short and in some studies the results of the retreated population was not discussed. However, some studies focused on the effectiveness of repeated doses of RTX. (139;140) and the results showed that retreated patients responded without major side effects. Thurlings et al (140) studied the effect of re-treatment (allowing 6 months between repeated courses i.e. when the B cells recover from the previous course) on non-responders (seven patients) versus responders (seventeen patients) to the first course of RTX and the results suggested that RTX re-treatment was not effective in patients who did not respond to the first course of RTX. Popa C. et al presented long term follow-up of repeated courses of RTX over more than 5 years; patients received 1, 2, 3, 4 or 5 cycles of RTX. The results showed that all patients exhibited B cell depletion (except one) and the average duration of effectiveness from a single cycle was 15 months and re-treatment time was average of 20 months. (139) Discussion and conclusions Findings from RCTs The results of the five trials need to be unified to get an overall conclusion. The literature search identified a meta-analysis of the serious infections that arose in three of the randomised trials. Although the meta-analysis will not exclude the bias that may be associated with the published trials, it still unifies the results and gives a clearer conclusion. However, the inclusion of different populations in the three RCTs makes it inappropriate to combine results for efficacy of these studies. 92

93 In the RCTs, comparison of RTX mono-therapy and combined therapy with MTX or other nbdmard needs to be further studied to assess the group of patients who have contraindications to MTX. Moreover inclusion and comparison of different population of patients such as nbdmard failures and anti-tnf failures in the same study is suggested. The sample size within those trials did not allow subgroups analysis such as the response across RF positive and negative patients. None of the RCTs compared response of RTX to response of other alternative biologic therapies after failing one anti-tnf therapy. The SERENE trial showed that RTX can be used as first line biologic treatment in early RA however head to head trial of RTX versus other biologics is important before confirming this. In an indirect comparative analysis across nine biologics in RA patients who are biologic naïve from published RCTs, RTX was ranked third in 6 months model and second in 12 month s model. (141) The RCTs involved a range of 161 to 748 patients with a range of 40 to 298 patients in each RTX arm. Primary end points were taken at week 24, and patients enrolled had certain inclusion criteria and they did not receive other medications for other diseases. In each trial there was a list of exclusion criteria as diagnosis with other rheumatic disorders. In real life, the number of RA patients is larger, the period for their treatment lasts for a longer time and the patients may suffer from other diseases than RA. This comparison between the RCTs and real life settings suggests that these trials may not be reflecting real life setting where patients may also receive other medications for different illnesses. This finding may lead us to think about performing a comparison between RCTs and clinical practice data of RTX in RA Findings from observational studies Although RCTs are the gold standard for measuring the efficacy and short term safety of new medications, this literature review did not limit the search to RCTs. Observational studies were 93

94 also included. RCTs cannot answer all questions about the studied intervention and observational studies are more appropriate and likely to reflect daily medical practice. (142) Moreover observational studies may be more likely to run subgroups analysis and may look at possible predictors of response. This strengthens the role of observational studies in evaluating the use of RTX in RA. Moreover, observational studies can explore things not studied in RCTs such as the comparison against different available treatment options, the effect of co - morbidities and the role of different concomitant therapies. The observational studies of RTX in RA published to date answered some questions that were not examined in the RCTs, such as the response compared to other biologic alternative in a head to head comparison. An earlier prospective cohort study of 116 patients has suggested that RTX may be more effective in terms of a change in disease activity score (DAS28) and inflammation markers. (122;127) However, to date, there are no comparative studies that have reported on improvements in physical function. The appropriate dose and the strategy of administration of RTX need more investigation, as the study of Galarza C. et al (132) reported that patients who withdrew due to adverse events improved despite the uncompleted dose regimen (did not receive the second 1000 mg dose). Some studies used a different dosing protocol such as the study of Bokarewa M. et al (121) and reported good effectiveness results. It was also noted that some identified observational studies started before the results of all the RCTs were published. For example some studies such as the study of Popa C. et al (139) enrolled only seropositive population because the results of REFLEX trial, which showed no difference between seropositive and negative patients in response to RTX, was not published. Also, some studies used certain dosing protocols just because they started before the results of the RCTs were published. However, observational studies examining response in seropositive and negative patients may be helpful as the RCTs were not powered to detect differe nce 94

95 between these populations of patients. Some of the observational studies were extremely small and were not of a high power. To achieve the wanted power, an observational study should involve hundreds of patients. In particular, observational studies that reported on the effectiveness of RTX in improving quality of life and physical ability were of very small sample size (range 10 to 21 patients). The literature search identified only three observational studies that examined potential predictors of response to RTX. To date, there are no published studies that have examined potential predictors of better improvement of physical function after starting RTX. The data and results of the observational studies of RTX in RA are incomparable. The different design, population, duration and outcome measures of each study make it difficult to compare and more difficult to answer simple questions that arise in clinical practice such as how many patients responded? An overall answer to this question, including all the identified studies cannot be simply developed which makes it essential to call for a large, better designed observational study with a large number of patients and a long enough duration and study patients from real life clinical practice to optimize the clinical use of RTX in the management of RA. 95

96 3. PhD aim and objectives 96

97 3.1 Introduction Building on the conclusions made in the last chapter, it is important to study the effectiveness of RTX in RA when used in routine clinical practice. Given that RTX is licensed in the UK to be used for RA patients who have failed at least one anti-tnf therapy, understanding persistence with anti-tnf therapies was considered an important study point for this programme of research. In clinical practice, after failing the first anti-tnf therapy; the patient may either start RTX or switch to an alternative anti-tnf therapy. Comparing the response after commencing each of these two treatment options is of significant interest to the treating physician when deciding which treatment option will be of most benefit to the patient. 3.2 Aim The aim of this thesis was to evaluate the clinical effectiveness and use of RTX in RA patients treated in routine clinical practice in the UK, taking account of the patients who were eligible for receiving RTX. 3.3 General Objectives The effectiveness of RTX was assessed by addressing the following four key areas: (i) To study the RA patients who were eligible for starting RTX therapy by examining the persistence with anti-tnf therapies and impact of the use of different concurrent nbdmards on the persistence with anti-tnf therapies. (ii) To compare the clinical effectiveness and changes in RA patients physical function after switching either to an alternative anti-tnf therapy or RTX after failing the first anti-tnf therapy. 97

98 (iii) To examine the clinical effectiveness of RTX in a broad inclusion criteria cohort of RA patients that reflects routine clinical practice in the UK, and to identify factors that predict effectiveness. (iv) To examine the improvements in the quality of life and physical function of RA patients after receiving RTX, and to identify factors that predict improvements. To achieve these objectives, data from a large national prospective observational cohort study, the British Society for Rheumatology Biologics Register (BSRBR), was used. The BSRBR is recruiting and following RA patients treated with biologic therapies in the UK. Chapter four describes the general methodology of the BSRBR. 98

99 4. Methodology 99

100 4.1 Study design The BSRBR is a prospective longitudinal observational multi-centre national study in the UK. It was established in 2001 by the British Society for Rheumatology (BSR) with a primary aim of assessing the long term safety of the biologic therapies in RA. ( ) In 2002, the National Institute for Health and Clinical Excellence (NICE) approved the use of the anti -TNF therapies for RA through the National Health Services (NHS) with a recommendation to register all patients receiving anti-tnf therapies in the BSRBR (146) in order to monitor the long-term safety of these therapies. At the same time, a cohort of patients receiving nbdmards/conventional therapy was recruited and followed in the same manner as the biologic cohort. In 2007, the NICE recommended the use of RTX in combination with MTX as an option for the management of RA patients who have failed at least one anti-tnf therapy. (78) Subsequently, an extension to the BSRBR to recruit RA patients receiving RTX was approved and the registration for RTX therapy started in May The funding for the BSRBR was provided by the relevant pharmaceutical companies (that are marketing these therapies). The BSR receives restricted income from UK pharmaceutical companies, who are the market authorisation holders for these therapies, to run and maintain the study. This income finances a separate contract between the BSR and the Arthritis Research UK Epidemiology Unit at the University of Manchester who provide and run the BSRBR data collection, management and analysis services. The principal investigators and their team have full academic freedom and are able to work independently of pharmaceutical industry influence. 4.2 Sample size The sample size for the anti-tnf cohort, which included adalimumab (ADA), etanercept (ETN) and infliximab (INF), was calculated to allow detecting a doubling in the risk of lymphoma in patients receiving anti-tnf therapies (exposed cohort) compared to the nbdmard cohort (unexposed cohort). As a result, a target sample size of 4000 patients fol lowed for a minimum 100

101 of five years (total 20,000 person years per drug) in each of the ETN, INF and ADA cohorts was recruited. A target of 4000 patients not exposed to biologic therapy were also recruited in the comparison cohort and followed for a minimum of five years. Given that these three anti-tnf therapies were licensed for use at different times, the recruitment period was not consistent among the different therapies. The recruitment targets of 4000 patients per anti -TNF therapy were subsequently achieved for ETN, INF and ADA in 2005, 2007 and 2008, respectively (Figure 4.1). Given that RTX acts by depleting B cells, patients receiving RTX may be at an increased risk of infection. The sample size for the RTX cohort was calculated on the ability to de tect a doubling in the risk of infection in patients receiving RTX versus patients receiving traditional nbdmards (RTX is used in the treatment of lymphoma and therefore lymphoma could not be considered as an endpoint for sample size calculation). Consequently, a target sample size of 1100 patients followed for a minimum of five years was recruited. The recruitment target was achieved in June 2011 (Figure 4.1). 4.3 Ethical approval The original ethical approval for the BSRBR was provided by the North West Multicentre Research Ethics Committee (MREC) in December 2000 (Appendix 4.1). Extension of ethical approval for recruitment of RTX patients was accepted by the same MREC on January 2007 (Appendix 4.2). 101

102 2000 UK license of ETN and INF 2001 Start of the BSRBR - start of ETN and INF recruitments 2002 NICE guidelines on anti-tnf therapies 2003 License of ADA - start of ADA recruitments 2005 Target for ETN recruitment achieved 2006 License of RTX 2007 Target for INF recruitment achieved 2008 Start of RTX recruitments Target for ADA recruitment achieved 2011 Target for RTX recruitment achieved Figure 4.1: Key timelines in the BSRBR 102

103 4.4 Inclusion criteria To be eligible for recruitment to the BSRBR anti-tnf cohort, patients should have RA diagnosed by a consultant or if the 1987 ACR criteria were satisfied, aged 16 or more, had previously failed at least two nbdmards including MTX and started biologic therapy not longer than six months prior to the registration date. The patients were given informed consent to participate in the study. Those participants who were initially registered with the BSRBR as anti-tnf patients and subsequently switched to RTX were re-registered as RTX patients, with a new identification number and their follow-up period starting again, if they fulfilled the above inclusion criteria at the time of re-registration. 4.5 Patient recruitment To register a patient with the BSRBR, first the patient was asked to read the patient information sheet and if in agreement to participate, sign the consent form which includes (i) the patient agreeing to complete the questionnaires, (ii) an understanding that the rheumatologist will provide information about his/her health records, and (iii) data will be available to the research team at The University of Manchester. Patients who were re - registered as RTX patients were asked to re-sign a new consent form at the time of reregistration in order to allow the follow-up clock to be re-started. 4.6 Baseline characteristics When a patient agreed to participate, the consultant was asked to complete the consultant baseline questionnaire (Appendix 4.3) which included recording data about the ACR criteria of the disease, the individual components of the DAS28 score at the time the therapy was started, current and previous disease therapies, co-morbidities if present and the history of specific co- 103

104 morbidities (including cancer and TB). Additional information was also collected at this time point including weight, height, blood pressure and smoking status (Figure 4.2). For those patients who were registered for starting RTX therapy, the previous biologic history was collected. Thus, if the patient was new to the BSRBR when registered as a RTX patient, the consultant was asked to complete both a consultant baseline questionnaire and a prior biologic exposure form (Appendix 4.4) which collected data on the type and start/stop dates of previous biologic therapies and the reason for discontinuation (inefficacy, remission, adverse events, or other reasons). If the patient had previously been registered for anti-tnf therapy, then switched to RTX and was re-registered for RTX therapy, the consultant was asked to complete a short baseline questionnaire (one page) (Appendix 4.5) rather than the full consultant baseline questionnaire. The short baseline questionnaire included data on DAS28, current drug therapy and comorbidities. The other patient demographic characteristics were recalled from the original consultant baseline form which was completed when the patient was first registered with the BSRBR for anti-tnf treatment. The previous biologics received and the reasons for discontinuation in this group of patients were obtained from the follow-up forms which were collected whilst the patient was registered for anti-tnf therapy. At time of registration, the patient was also asked to complete the patient baseline questionnaire (Appendix 4.6) which collected data on working status, date of birth, ethnicity and smoking history. The baseline HAQ score (Appendix 4.7) and EQ-5D (starting from August 2006) (Appendix 4.8) were also collected from the patient at this stage. 104

105 Registration ALL patients completed HAQ and EQ-5D Previously registered RTX patient New RTX patient Anti-TNF patient Consultant completed Short baseline form Other baseline data collected from the records of the patient when he was on anti-tnf Prior biologic exposure form Consultant baseline questionnaire - DAS28 - Co-morbidities - Current drug therapy - Demographics - RF status - Smoking status - Previous DMARD therapies - Previous biologic therapies - Reason for stopping last anti- TNF therapy - Previous biologic therapies - Reason for stopping last anti- TNF therapy - Demographics - DAS28 - Co-morbidities - RF status - Smoking status - Previous and current DMARD therapies Figure 4.2: Baseline data collection within the BSRBR. 105

106 4.7 Follow-up The BSRBR aimed to follow-up all the registered patients for a minimum of five years, even if the patient stopped the therapy or switched to another treatment. Follow-up was assessed every six months for the first three years then every year thereafter. The follow-up included three approaches; consultant follow-up questionnaires, patient follow-up questionnaires (for three years) and flagging with the Office of National Statistics (ONS) to capture information on any deaths or malignancies occurring prior or post-registration. The consultant follow-up questionnaire (Appendix 4.9) examined whether there were any changes in the patient s biologic therapy and whether or not the patient discontinued the biologic therapy. If the patient discontinued the therapy, then the start date, the stop date and the reason for discontinuation (classified as inefficacy, remission, adverse event or other reasons) were recorded. If the patient experienced an adverse event or other reasons leading to withdrawal, details of such events were then recorded. Data on any adverse events or any new illnesses and the components of the DAS28 score are also included in this questionnaire. The consultant follow-up questionnaire also included a request to supply any missing data from the previous consultant follow-up questionnaire. The patient follow-up questionnaire (Appendix 4.10) included data on medical problems such as any admission to hospital, any newly prescribed drugs or any referrals to hospital clinics in the last six months. In addition, the HAQ, and EQ-5D data were recorded. 4.8 Data management All the data collected for each patient on the paper forms were posted to the researchers at the University of Manchester and the data were entered into a Microsoft Access database by trained staff. DAS28 scores were calculated using the formula: DAS28 = 0.56 x (tender joint 106

107 count 28) x (swollen joint count 28) x ln (erythrocyte sedimentation rate) x general health. Each patient was assigned a specific unique identification number. All efforts were made to obtain missing forms/data and these were followed up via reminders on two occasions at specified time periods. The data of the BSRBR have been archived at certain time points and saved periodically. At the beginning of this PhD, the author received training by the BSRBR data entering team to get familiar with different forms and where every form is stored within the data window. The BSRBR data was stored in the form of tables in a Microsoft Access database. There was coding schedules that describe each variable name in each table within the Microsoft access data. Using those coding schedules the author was able to know where each question in the forms is stored in the Microsoft access data. The BSRBR dataset is stored in a university computer network drive. The author s computer, at the author s work station, was mapped to have access to this drive. The Office of the Data Protection Commissioner (ODBC) data source administrator in the author computer had been set to open the BSRBR dataset given the author s login details. STATA 10.1 software was used to load the data from ODBC sources. In STATA a table from Microsoft Access database can be loaded from ODBC sources then the needed variables can be selected and extracted so that a STATA dataset, which can be used for the analysis, can be developed. The developed data set was then sorted, cleaned and sometimes merged with other data from other table from the BSRBR Microsoft Access database. The author carried out random checks for accuracy of data entering, using the hard copies of the forms returned to the BSRBR where data were compared to the stored data in the BSRBR database. 107

108 Cumulative number of patients recruited For each of the analyses carried out in this PhD, the author developed an algorithm of series of STATA commands (STATA do file) that first prepared the data for each specific analysis; this included data cleaning, shaping, identification of the patients through the inclusion and exclusion criteria specified for each analysis, generation of some variables out of the existing variables (such as the EULAR response which was generated from the 6 months DAS28 scores and its improvement from baseline DAS28 scores). The cleaned prepared datasets were then used to describe the baseline characteristics of the included patients and stored for statistical analysis which suits the study questions raised in each of the analyses carried out in this PhD. The statistical analysis methodology will be described in details for each analysis in the following three chapters. 4.9 BSRBR recruitment rates Given that the three anti-tnf therapies were available for use at different times, recruitment rates for the three therapies were not constant over time (Figure 4.3) ETN INF ADA Figure 4.3: Recruitment rates of anti-tnf cohorts in the BSRBR over time. 108 ETN INF ADA

109 Cumulative number of patients recruited The rate of RTX recruitment was slow in 2008, as it was the initial use of RTX in clinical practice, but a rise in the rate appeared by the beginning of 2009 and remained high thereafter (Figure 4.4). By 2010 the recruitment rate of RTX patients who were new to the register was slightly higher than those who were previously registered with the BSRBR Dec-08 Jun-09 Dec-09 Jun-10 Dec-10 Time (6 months) All RTX Patients New patients Previously registered patients Figure 4.4: Recruitment rates of RTX cohort stratified by type of registration. Since there is no information on those patients who refused to take part in the study, the response rate of recruitment cannot be specified. Given that the patient baseline questionnaire was sent out by the researchers once the consultant baseline form was received, the response rate of the patient baseline questionnaire can be calculated. Ninety-nine percent of the anti-tnf patient had returned their baseline questionnaires compared to 98% in RTX patients. 109

110 By December 2010, all the anti-tnf patients had reached six months of follow-up, for 90% of them the six months consultant questionnaires were returned. For RTX, 870 patients had reached six months of follow-up, 88% of them having the six months consultant questionnaires returned. The six months patient follow-up questionnaires were returned for 72% of the anti- TNF patients and 65% of the RTX patients who had reached six months of follow-up BSRBR RTX cohort By December 2010, 1138 RA patients starting RTX therapy were registered with the BSRBR as RTX patients. Four hundred and eighty two patients of them were previously registered with the BSRBR as anti-tnf patients while 656 patients were completely new to the register (Figure 4.5). Six hundred and forty six patients had both the baseline and six months consultant questionnaires (with a recorded DAS28 score) returned. Four hundred and forty eight patients had returned both the baseline and six months patient questionnaires along with the HAQ forms. Within the register there were 1,239 patients who were registered as anti-tnf patients then switched to RTX but were not reregistered for RTX therapy Switching therapy within the BSRBR Within the BSRBR some anti-tnf patients had switched their anti-tnf therapy to another biologic therapy. However, due to the study design follow-up continued regardless whether the patient had switched therapy or not. No additional data were collected at the time of switching; however the dates of switching were possible to be identified using the start and stop dates of the biologic therapies collected at each follow-up. Some of the patients who switched from anti-tnf therapy to RTX were re-registered as RTX patients if they fulfilled the inclusion criteria for RTX registration. After re-registration the subsequent follow-up commenced from the date of re-registration. Patients who were receiving RTX but not reregistered (predominantly those who switched to RTX prior to May 2008 when the RTX cohort was opened for recruitment or those patients who did not agree to be re-registered and 110

111 followed up for another five years) were not eligible to be included in the effectiveness analysis of RTX therapy as no baseline data were collected at the time of switching to RTX. However, for the comparative effectiveness analysis with patients who switched to another anti-tnf, those RTX patients (who were not reregistered for RTX therapy) were included since there was no baseline data collected for the comparison group of patients who switched to a second anti - TNF therapy as the analysis depended on the start and stop dates of the biologic therapies. 111

112 Patients registered with the BSRBR and Receiving RTX N= 2,377 Registered as RTX patients N= 1,138 Registered as anti-tnf patients and switched to RTX but not reregistered as RTX patients N= 1,239 New patients 656 Previously registered patients 482 No baseline data at the time of switching to RTX. The follow-up clock starts at the date of registration for the anti-tnf. RTX Cohort Consultant forms Patient forms Time (Months) N= 1,138 (1,133 had a recorded DAS28 N= 1,113 (999 had a recorded HAQ score) Baseline data 0 N= 870 N= 870 Reached 6 months since registration N= 766 (648 had a recorded DAS28 score) N= 565 (490 had a recorded HAQ score) First follow-up 6 N= 646 N= 448 Both baseline and six months data with recorded DAS28 or HAQ Figure 4.5: Flow chart of RTX patients in the BSRBR (by 21 st December 2010) 112

113 4.12 Baseline characteristics By 21 st December 2010, a total of 13,122 patients receiving anti-tnf therapies (ETN, INF, or ADA) and 1,138 patients receiving RTX were registered with the BSRBR. The baseline characteristics of the two cohorts of patients are shown in Table 4.1. In general, the mean age of the patients was in the sixth decade of life and 76% of the patients were females. The patients were of high disease activity (mean DAS28 over 6.1) and had experienced over 13 years of disease. Some differences were noticed between the patients registered for anti-tnf therapies and those registered for RTX therapy. The latter group of patients were generally older (mean (SD) of 59 (12) years versus 56 (12) years) with longer disease duration (15 (10) years versus 13 (10) years) and lower disease activity (DAS28 of 6.1 (1.2) versus 6.6 (1.0)). RTX patients were also more RF positive patients (68% versus 65%), and larger proportions of them had one or more co-morbidity (74% versus 62%). RTX patients used a higher number of previous nbdmards (4.2 (1.8) versus (3.9 (1.7)). Larger proportions of RTX patients were also receiving concurrent MTX compared to the anti-tnf cohort (50% versus 40%). 113

114 Table 4.1: Baseline characteristics of the patients receiving biologic therapies in the BSRBR* Characteristics RTX cohort Anti-TNF cohort Number of patients 1,138 13,122 Demographics Age (year) 59.5 ± ± 12.3 Female, n (%) 874 (76.8) 10,010 (76.3) Current smoker, n (%) 246 (21.6) 2,358 (21.2) Co-morbidities, n (%) No co-morbidities One co-morbidity More than one co-morbidity Disease characteristics 298 (26.2) 344 (30.2) 496 (43.6) 4,990 (38.0) 4,186 (31.9) 3,946 (30.0) Disease duration(year) 14.6 ± ± 9.6 Rheumatoid factor positive, n (%) 703 (68.0) 8,493 (64.7) Swollen joint count (no.) 8.8 ± ± 6.2 Tender joint count (no.) 14.1 ± ± 7.4 Global health visual analogue 69.0 ± ± 19.9 scale C-reactive protein (mg/dl) 33.4 ± ± 42.6 Erythrocyte sedimentation rate (ESR) (mm/hr) 44.0 ± ± 28.6 Disease Activity Score ± ± 1.0 Quality of life and physical function EQ-5D visual analogue scale 45.8 ± ± 22.0 EQ-5D utility score 0.28 ± ± 0.33 Health assessment questionnaire 1.97 ± ± 0.58 Current and previous drug history Number of previous nbdmards 4.2 ± ±1.7 Current nbdmards, n (%) None MTX Other 270 (23.7) 570 (50.1) 298 (26.2) 3,779 (28.8) 5,216 (39.8) 4,125 (31.4) Previous steroids, n (%) 871 (76.5) 9,357 (71.3) Current steroids, n (%) 537 (47.2) 5,867 (44.7) * Based on BSRBR data up to 21 st December Values are mean (SD) otherwise specified. TNF: tumour necrosis factors, RTX: rituximab, MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, EQ-5D: European Quality of life 5 Dimensions, Co-morbidities include angina, hypertension, stroke, epilepsy, myocardial infarction, asthma, chronic obstructive pulmonary disease,liver disease, renal disorder, peptic ulcers, diabetes, demyelination, hyperthyroidism, depression or a history of tuberculosis or cancer. The EQ-5D data were available for 1794 anti -TNF patients. 114

115 5. Drug persistence with anti-tnf therapies in patients with rheumatoid arthritis 115

116 RA patients who discontinue anti-tnf therapies are eligible for starting RTX therapy, the target drug under investigation in this PhD. This chapter studies the persistence with anti-tnf therapies in RA patients. 5.1 Aim and objectives This chapter aims to evaluate drug persistence of the first course of anti -TNF therapies in RA patients treated in routine clinical practice. The following specific objectives were considered. To study the overall persistence with anti-tnf therapy. To identify predictors of anti-tnf therapy discontinuation. To examine whether there are differences between drug persistence across the three anti - TNF therapies (ETN, INF, and ADA). To examine the effect of concomitant nbdmards therapy on persistence with the anti-tnf therapies. To assess the above four objectives according to the reason for discontinuation (all reasons, ineffectiveness of the therapy, and development of adverse events). To achieve these objectives, a cohort of patients registered with the BSRBR was used. 5.2 Methods BSRBR data frozen at 28 th June 2009 were used for the data analysis presented in this chapter Inclusion and exclusion criteria Patients inclusion criteria (performed in order) i. All patients registered with the BSRBR ii. Patients with RA indication iii. Patients starting treatment with their first course of ETN, INF or ADA. iv. Patients receiving the most common nbdmards or no concomitant nbdmard. 116

117 Patients exclusion criteria i. Patients who started treatment six months or more before registration (to decrease potential bias due to left censoring). ii. Patients who had received previous biologic therapies (including anakinra and RTX) iii. Patients for whom the consultants did not return any consultant follow-up questionnaires The cohorts used in the analysis All the included patients were considered as the total cohort in this analysis. This total cohort was divided into sub-cohorts in two different ways. Firstly, based on which anti-tnf and secondly based on which concomitant nbdmard(s) were used with the anti -TNF therapy at baseline. In order to make this later division of the cohort, the most commonly used nbdmards with the biologic treatment at baseline were identified and then the use of single versus combination nbdmards was considered Baseline characteristics The baseline characteristics of the total cohort, the three cohorts of the biologic therapy and the sub-cohorts of the baseline concomitant nbdmards were studied. The baseline characteristics described the patients demographics including age, gender, disease duration, smoking status, current and previous nbdmards therapies and co-morbidities (if present at time of registration or if a patient had a history of co-morbidities such as TB or cancer). In addition the disease characteristics such as the DAS28 and HAQ scores were also described Definition of persistence Persistence with anti-tnf therapies was defined as the length of time the patients stayed on the therapy. The start of the therapy was identified from the date the patient started the therapy. The stopping of the therapy was identified from the date the patient stopped the 117

118 therapy. Patients were censored on the date of the last follow-up questionnaire if the patient did not stop the therapy or the date of death, if the patient died (whichever came first) Statistical analysis STATA 10.1 software was used for performing all statistical calculations and related graphs. Differences in baseline characteristics between the three cohorts of biologic therapies and differences between the seven cohorts of the baseline concomitant nbdmards were tested using ANOVA test if the data were continuous or by Pearson Chi square test if the data w ere categorical. Statistically significant differences were considered when p<0.05. The survival analysis considers both the occurrence of the event and the time taken for the event to occur. The Kaplan-Meier survival model is the default model for survival analysis. Censoring occur when the exact time taken for the event to occur is unknown. Left censoring occurs if the event occurs before the study begins. The study design should avoid the bias of left censoring. Right censoring occurs if a patient died, a patient left the register or if the study ended. Right censoring is fine provided that it is not associated with the event. The Kaplan- Meier survival function avoids the bias of right censoring as patients are removed from the analysis at the last known point of contact. As a result Kaplan-Meier survival plots were used to study persistence with anti-tnf therapies in this study. Survival analysis is usually developed for events that occur once, such as death for example. In this analysis the event was stopping the medication, so the first time the medication was stopped was identified as the event. If a patient stopped a medication then reused it, the first stop was the only considered stop. However, if the period of stopping and reusing the medication was three months or less, the stopping in this case was not considered as a true discontinuation. These very short stopping periods were common in patients who developed 118

119 adverse events that required temporary discontinuation of the medication or in patients who required surgery or who were abroad on holiday for a short period. The analysis was developed to study the persistence with anti -TNF therapies in the total cohort, in the three anti-tnf therapies cohorts and according to the sub-cohorts of baseline concomitant nbdmards (seven sub-cohorts). Kaplan-Meier survival plots were used to study the persistence in each cohort. The log rank test was used to compare the survivor functions to test differences between the persistence in the three cohorts and across the seven cohorts. Univariate and multivariate Cox proportional hazard models were used to identify the factors associated with anti-tnf therapy discontinuation in total and according to the reason for discontinuation (inefficacy or adverse events). The models examined (i) demographic characteristics (age (10 year age-bands), gender, presence of co-morbidities, current smoking, number of previous nbdmards, current steroids) and (ii) baseline disease characteristics (disease duration (10 year age bands), RF status, baseline DAS28 score, and baseline HAQ score). For the multivariate models, forward stepwise selection method was used to select the variables using significance level of The results were presented as hazard ratios and 95% confidence intervals. The models were stratified by the anti-tnf therapies, the baseline concomitant nbdmards, and the year of treatment start. Both unadjusted and adjusted models were used to compare the persistence rates (i) across the three anti-tnf therapies using ETN as reference and (ii) across the seven concomitant nbdmard therapies groups using MTX as reference. The models were adjusted for baseline demographic and disease characteristics and stratified by the year of anti-tnf treatment start. The models that compared the three anti-tnf therapies were stratified by the concomitant nbdmards. The models that compared the persistence across the seven cohorts according the 119

120 baseline nbdmards were stratified by the anti-tnf therapy. In each case, three models were developed according to the reason for discontinuation (overall, inefficacy, and adverse event). Finally, models that compared the persistence across the seven cohorts of baseline concomitant nbdmards in each of the three anti-tnf therapies separately were developed. These models were also adjusted for baseline demographic and disease characteristics and were stratified by the year of anti-tnf treatment start. 5.3 Results Patients included By 28 th June 2009, a total of 19,146 patients were registered with the BSRBR (Table 5.1). Only patients with RA (n=17,409) were used in this analysis while patients with psoriatic arthritis (PsA) (n=709), ankylosing spondylitis (AS) (n=633) and other rheumatic diseases (n=395) were excluded. From the 17,409 patients with RA, patients treated with ETN (n=4,453), INF (n=4,196) and ADA (n=4,466) were kept while patients treated with anakinra (n=143), RTX (n=376) and no biologic drugs (BSRBR control cohort) (n=3,775) were excluded. From those included patients, patients who started treatment six months or more before registration with the BSRBR (n=1,027) were excluded to decrease the bias of left censoring. Finally, the analysis was limited to the patients for whom at least one consultant follow-up questionnaire was returned (n=11,515). 120

121 Table 5.1: Patients included in the analysis Patients Excluded Included All patients registered with the BSRBR 19,146 Patients with RA 1,737 17,409 Patients receiving first course of ETN, INF or ADA 4,294 13,115 Patients started therapy within 6 months of registration 1,027 12,088 Patient for whom at least one consultant follow-up questionnaire was returned Patients receiving the most common nbdmards or no concomitant nbdmard Excluded due to not meeting the criterion ,515 1,119 10,

122 5.3.2 Identifying the seven cohorts of the baseline concomitant nbdmards Table 5.2 shows the concomitantly used nbdmards at baseline. From the Table, it was found that the five most commonly used nbdmards at baseline were MTX, sulfasalazine (SSZ), hydroxychloroquine (HCQ), leflunomide (LEF) and azathioprine (AZA). However, this Table did not consider single or multiple uses of these nbdmards which are often prescribed in clinical practice. As a result, these five top nbdmards were considered and Table 5.3 is presented showing the single and multiple uses of concomitant nbdmards. Table 5.3 shows that the largest groups were single MTX (38.4%), no nbdmards (29.0%), MTX+ SSZ (7.8%), single LEF (5.3%), MTX+SSZ+HCQ (3.6%), MTX+HCQ (3.5%) and single SSZ (2.7%). The remaining nine groups were small in number (less than 2.6%) and hence were excluded (n=1,119). For consistency, those patients were excluded from the entire analysis. Table 5.4 shows the seven cohorts which were considered in this analysis. 122

123 Table 5.2: Baseline concomitant nbdmards Baseline concomitant nbdmards Patients, n (%) * Methotrexate 6,501 (56.46) Sulfasalazine 1,850 (16.07) Hydroxychloroquine 1,135 (9.86) Leflunomide 945 (8.21) Azathioprine 243 (2.11) Gold 172 (1.49) Cyclosporine 79 (0.69) Penicillamine 79 (0.69) Cyclophosphamide 7 (0.06) Auranofin 2 (0.02) Other 31 (0.27) No nbdmards 3,339 (29.00) nbdmards: non-biologic disease modifying anti-rheumatic drugs. * The patients may be in one or more categories. 123

124 Table 5.3: The single, double and triple use of concomitant nbdmards at baseline Baseline concomitant nbdmards Patients, n (%) None 3,339 (29.00) MTX 4,418 (38.37) SSZ 308 (2.67) Single nbdmard therapy LEF 610 (5.30) HCQ 146 (1.27) AZA 165 (1.43) Other single nbdmard 177 (1.54) MTX-SSZ 902 (7.83) MTX-LEF 177 (1.54) MTX-HCQ 401 (3.48) Double nbdmard therapy Triple nbdmard therapy MTX-AZA 23 (0.20) Double nbdmard combination therapy 248 (2.15) excluding MTX Other double nbdmard combination 93 (0.81) therapy including MTX MTX-SSZ-HCQ 418 (3.63) Other triple nbdmard combination 69 (0.60) therapy including MTX Other triple nbdmard combination 21 (0.18) therapy excluding MTX Total 11,515 (100.00) MTX: methotrexate, SSZ: sulfasalazine, HCQ: hydroxychloroquine, LEF: leflunomide AZA: azathioprine, nbdmards: non-biologic disease modifying anti-rheumatic drugs, 124

125 Table 5.4: The seven cohorts of baseline concomitant nbdmards considered for the analysis Baseline concomitant nbdmards Patients, n (%) None 3,339 (32.12) MTX* 4,418 (42.50) SSZ* 308 (2.96) LEF* 610 (5.87) MTX-SSZ 902 (8.68) MTX-HCQ 401 (3.86) MTX-SSZ-HCQ 418 (4.02) Total 10,396 (100.00) * The concomitant nbdmard was used alone (with no other nbdmards). MTX: methotrexate, SSZ: sulfasalazine, HCQ: hydroxychloroquine, LEF: leflunomide, nbdmards: non-biologic disease modifying anti-rheumatic drugs, TNF: tumour necrosis factors The cohorts used in the analysis The total cohort (n=10,396) Separate cohorts for the three anti-tnf therapies ETN= 3,605 (34.68%) INF= 3,136 (30.17%) ADA= 3,655 (35.16%) Seven cohorts taking account of the baseline concomitant nbdmards therapy (Table 5.4). 125

126 5.3.4 Overall anti-tnf therapies cohort Baseline characteristics The total cohort included 10,396 patients, 76% of which were female and the mean age was fifty six years (Table 5.5). Twenty two percent of the patients were current smokers, while 62% of them had co-morbidities. These co-morbidities included angina, hypertension, myocardial infarction, stroke, epilepsy, asthma, chronic obstructive pulmonary disease, peptic ulcers, liver disease, renal disorder, demyelination, diabetes, hyperthyroidism, depression or history of TB and cancer. The patients had a mean duration of 13 years with RA. Sixty four percent of the patients were found to have a positive RF status. Disease activity was found to be high as measured by DAS28 (mean of 6.6) and disability of the patients was also high as indicated by the HAQ score (mean of 2.0). The patients had previously received a mean of four nbdmards Percentage of patients who stopped the medication By the end of five years of follow-up, 49.6% of the patients were still on their medication while 21.6% of the patients stopped their medication due to inefficacy of the biologic therapy, 20.7% of them stopped the medication due to development of adverse events and the remainder (8.1%) of the patients stopped the medication due to other/missing reasons. From the 846 patients who stopped the medication due to other/missing reasons, 135 stopped due to missing reasons (which may include inefficacy or adverse events), 23 patients stopped due to remission and 688 stopped due to other reasons which are not related to the medication including being on holiday, undergoing elective surgery, or travelling outside the UK for longer than 3 months. 126

127 Table 5.5: Patient characteristics of the total anti-tnf cohort Patient characteristics Total cohort Number of patients 10,396 Demographic characteristics Age (year) 56.1 ±12.3 Female sex, n (%) 7,915 (76.1) Number of previous nbdmard (mean no.) 3.9 ±1.6 Current smokers, n (%) 2,271 (22.0) Co-morbidities, n (%) 6,424 (61.8) Disease characteristics RF positive, n (%) 6,714 (64.6) Disease duration (year) 13.3 ± 9.7 S JC (no.) 11.3 ± 6.2 TJC (no.) 15.6 ± 7.4 Global health VAS 72.7 ±19.8 CRP (mg/l) 45.3 ± 42.5 ESR (mm/hr) 46.1 ± 28.7 DAS28 score 6.6 ±1.0 HAQ score 2.0 ± 0.6 Anti-TNF therapy ETN, n (%) 3,605 (34.7) INF, n (%) 3,136 (30.2) ADA, n (%) 3,655 (35.1) Values are mean ± SD unless otherwise specified, nbdmards: non-biologic disease modifying anti-rheumatic drugs, RF: rheumatoid factor, SJC: swollen joint count, TJC: tender joint count, VAS: visual analogue scale, ESR: erythrocyte sedimentation rate, DAS28: disease activity score 28-joints and HAQ: health assessment questionnaire, ETN: etanercept, INF: infliximab, ADA: adalimumab..*co-morbidities included one or more of angina, hypertension, myocardial infarction, stroke, epilepsy, asthma, chronic obstructive pulmonary disease, peptic ulcer, liver disease, renal disorder, demyelination, diabetes, hyperthyroidism, depression or a history of TB or cancer. 127

128 Table 5.6: Number of patients who stopped the medication in the entire anti -TNF cohort Outcome Total cohort Number of patients 10,396 Follow-up by person-years 2.4 Mean (SD) /max (1.8) /8.1 Still on medication, n (%) 5,157 (49.6) Stopped due to inefficacy, n (%) 2,240 (21.6) Stopped due to adverse event, n (%) 2,153 (20.7) 128

129 Overall persistence with anti-tnf therapies The overall persistence with the anti-tnf therapies (Table 5.7) decreased from 72% in the first year to 42% in the fifth year. The number of patients contributing follow-up in this last year (1,030 patients) was less than in the first year (7,078 patients) and this should be taken in consideration in this comparison. A Kaplan-Meier plot of the overall persistence is shown in Figure 5.1. When the analysis was split by the reason for discontinuation, stopping due to inefficacy and due to adverse events showed similar persistence with the anti -TNF therapies (log-rank p=0.30). Table 5.7: Survivor function (95% CIs) for the overall anti-tnf therapy persistence Years of follow-up Survivor function (95% CIs) All reasons Due to inefficacy Due to adverse events Year (0.71: 0.73) 0.87 (0.86:0.87) 0.86 (0.86: 0.87) Year (0.58: 0.60) 0.79 (0.87: 0.80) 0.80 (0.79: 0.81) Year (0.51: 0.53) 0.75 (0.74: 0.76) 0.76 (0.75: 0.77) Year (0.45: 0.47) 0.71(0.70: 0.73) 0.73 (0.72: 0.74) Year (0.41: 0.43) 0.69 (0.68: 0.70) 0.71 (0.70: 0.73) 129

130 Time (year) Number of subjects at risk Figure 5.1: Kaplan-Meier plot of persistence with anti-tnf therapies in total (green) and when the reason for discontinuation was inefficacy (blue), or adverse events (red). 130

131 Predictors of anti-tnf therapy discontinuation Three separate models were developed to identify the predictors of anti -TNF therapy discontinuation, namely when the stopping was due to inefficacy, adverse eve nts, or for any reason. Each model involved both univariate and multivariate Cox proportional hazards analysis (Table 5.8). High baseline HAQ score was the strongest predictor of treatment discontinuation due to all reasons in both the univariate and multivariate models. Female sex, having co-morbidities, being a current smoker, and failing 4 or more previous nbdmards were also significant predictors of overall anti-tnf therapy discontinuation in both the univariate and the multivariate Cox proportional hazards models. When the analysis was limited to inefficacy as the reason for discontinuation, previous failure of 4 or more nbdmards was the strongest predictor of treatment discontinuation, followed by high baseline HAQ. High baseline DAS28 score and female gender were also significant predictors of anti-tnf therapies discontinuation due to inefficacy in the multivariate Cox proportional hazards model. Older patients and those with longer disease duration were less likely to discontinue their therapy due to inefficacy. When the analysis was limited to adverse events as the reason for discontinuation, high baseline HAQ score was the strongest predictor of treatment discontinuation. Failing 4 or more previous nbdmards, having co-morbidities, positive RF status, longer disease duration and older age were significant predictors of discontinuation due to adverse events in the multivariate Cox proportional hazards model. 131

132 Table 5.8: Hazard ratios (95 % confidence intervals) in the univariate and multivariate Cox proportional hazards analysis of anti-tnf therapies discontinuation Covariates Hazard ratios (95 % confidence intervals) Overall stopping Stopping due to inefficacy Stopping due to adverse events Univariate analysis Multivariate analysis Univariate analysis Multivariate analysis Univariate analysis Multivariate analysis Demographic variables Baseline age (per 10 years) 1.03* (1.01:1.05) * (0.92:0.99) 0.95* (0.91:0.99) 1.15* (1.11:1.19) 1.09* (1.04:1.13) Female gender (yes/no) 1.13* (1.06:1.21) 1.09* (1.01:1.16) 1.24* (1.11:1.37) 1.16* (1.04:1.30) 1.04 (0.94:1.15) - Co-morbidities (yes/no) 1.13* (1.07:1.20) 1.10* (1.03:1.17) 1.07 (0.98:1.17) * (1.14:1.37) 1.14* (1.04:1.26) Current smoker (yes/no) 1.10* (1.03:1.18) 1.10* (1.03:1.18) 1.14* (1.03:1.25) (0.92:1.13) - Disease features 4 or more previous nbdmards 1.21* (1.14:1.28) 1.17* (1.10:1.24) 1.19* (1.09:1.30) 1.30* (1.17:1.43) 1.29* (1.18:1.42) 1.17* (1.06:1.29) RF positive (yes/no) 1.03 (0.97:1.09) (0.88:1.05) * (1.04:1.25) 1.11* (1.01:1.22) Current steroids (yes/no) 1.07* (1.01:1.13) (0.93:1.10) * (1.03:1.23) - Disease duration (per 10 years) 1.03 (1.00:1.06) * (0.85:0.93) 0.84* (0.80:0.88) 1.17* (1.12:1.22) 1.09* (1.04:1.14) Baseline DAS28 (unit) 1.07* (1.04:1.10) * (1.08:1.18) 1.09* (1.04:1.14) 1.05* (1.01:1.09) - Baseline HAQ (unit) 1.32* (1.25:1.39) 1.27* (1.21:1.34) 1.27* (1.18:1.38) 1.25* (1.14:1.36) 1.41* (1.30:1.53) 1.30* (1.19:1.42) nbdmards: non-biologic disease modifying anti-rheumatic drugs, RF: rheumatoid factor, DAS: disease activity score; HAQ: health assessment questionnaire. *P < Co-morbidities include angina, hypertension, myocardial infarction, stroke, epilepsy, asthma, chronic obstructive pulmonary disease, peptic ulcer, liver disease, renal disorder, demyelination, diabetes, hyperthyroidism, depression or history of TB or cancer. All models were stratified by year of treatment start, anti-tnf therapy, and concomitant nbdmards therapies. - Variable was not included. 132

133 5.3.5 Comparing drug persistence across the three anti-tnf therapies Baseline characteristics The baseline characteristics for the three cohorts of the anti-tnf therapies are shown in Table 5.9. The three cohorts were similar in the demographic characteristics with slightly shorter mean (SD) disease duration of 12.8 (9.9) years in ADA cohort compared to 13.5 (9.5) years in ETN and 13.5 (9.5) in INF cohorts (p-value=0.0001). A slightly lower mean CRP of 39.5 mg/l in ADA cohort compared to 48.2 mg/l in INF cohort and ETN cohorts was also noted. It was noticed that the p-values showed significant differences between the three cohorts; however the observed values seem to be very similar. This is likely explained by the high power of the study to detect very small differences due to the large sample size Percentage of patients who stopped the medication During five years of follow-up, patients receiving INF were more likely to stop treatment than those receiving ADA or ETN (Table 5.10). When the reasons for discontinuation were considered, INF patients showed higher percent of discontinuation due to inefficacy and adverse events compared to those receiving ETN and ADA. In the three cohorts, the percentages of patients stopping the medication due to inefficacy and due to adverse events were similar. 133

134 Table 5.9: Patient characteristics of the three cohorts of the three anti-tnf therapies Characteristics Etanercept Infliximab Adalimumab p-value* Number of patients 3,605 3,136 3,655 - Demographic characteristics Age (yr) 56.0 ± ± ± Female sex, n (%) 2,777 (77.0) 2,370 (75.6) 2,768 (75.7) 0.29 Number of previous nbdmard (mean no.) 4.1 ± ± ± Current smokers, n (%) 741 (20.7) 698 (22.4) 832 (22.9) 0.06 Co-morbidities, n (%) 2,301 (63.8) 1,910 (60.9) 2,213 (60.6) Disease characteristics RF positive, n (%) 2,336 (64.8) 2,094 (66.8) 2,284 (62.5) Disease duration (yr) 13.5 ± ± ± S JC (no.) 11.4 ± ± ± TJC (no.) 15.8 ± ± ± Global health VAS, 72.8 ± ± ± CRP (mg/l) 48.2 ± ± ± ESR (mm/hr) 47.0 ± ± ± DAS28 score 6.6 ± ± ± HAQ score 2.1 ± ± ± Values are mean ± SD unless otherwise specified nbdmards: non-biologic disease modifying anti-rheumatic drugs, RF: rheumatoid factor, SJC: swollen joint count, TJC: tender joint count, VAS: visual analogue scale, ESR: erythrocyte sedimentation rate, DAS: disease activity score and HAQ: health assessment questionnaire. * P values are comparing the three biologic cohorts. 134

135 Table 5.10: Number of patients who stopped the medication in the three anti-tnf cohorts Outcome Etanercept Infliximab Adalimumab Number of patients 3,605 3,136 3,655 Follow-up by person-years 2.8(1.9)/ (1.9)/ (1.4)/6.3 Mean(SD)/max Still on medication, n (%) 1,919 (53.2) 1,167 (37.21) 2,071 (56.66) Stopped due to inefficacy, n (%) 696 (19.31) 815 (25.99) 729 (19.95) Stopped due to adverse event, n (%) 743 (20.61) 786 (25.06) 624 (20.71) Persistence with the three cohorts of the three anti-tnf therapies A Kaplan-Meier plot showing the persistence with the three anti-tnf therapies demonstrated that INF was more likely to be discontinued compared to ETN and ADA (Figure 5.2). The Log rank test carried out to identify differences between the survivor functions of the three anti -TNF therapies showed a significant difference (p<0.0001) between them in favour of ETN. Log rank test also showed that INF was inferior to ADA (p<0.0001). When the analysis was split by the reason of discontinuation, INF remained inferior to ETN and ADA whether the reason of stopping was due to inefficacy or adverse events (Table 5.11). After adjustments for baseline characteristics, patients were significantly more likely to discontinue INF and ADA compared to ETN (Table 5.12). Patients were significantly more likely to discontinue INF compared to ADA. 135

136 Number of subjects at risk Anti-TNF Time (year) therapy Etanercept 3, Infliximab 3, Adalimumab 3, Figure 5.2: Kaplan-Meier plot of persistence with each of the three anti-tnf therapies; etanercept (red), adalimumab (blue) and infliximab (green). 136

137 Table 5.11: Survivor function (95% CIs) for patients continuing anti-tnf therapies Years of follow-up Survivor function (95% CIs) Etanercept Infliximab Adalimumab Stopping due all reasons Year (0.73: 0.76) 0.70 (0.68: 0.71) 0.70 (0.69: 0.72) Year (0.63: 0.66) 0.52 (0.50: 0.54) 0.59 (0.57: 0.61) Year (0.57: 0.60) 0.43 (0.41: 0.44) 0.53 (0.51: 0.55) Year (0.52: 0.55) 0.37 (0.35: 0.38) 0.47 (0.44: 0.49) Year (0.47: 0.51) 0.33 (0.31: 0.35) 0.42 (0.38: 0.45) Stopping due to inefficacy Year (0.87: 0.89) 0.86 (0.85: 0.88) 0.84 (0.83: 0.86) Year (0.81: 0.84) 0.76 (0.74: 0.77) 0.78 (0.77: 0.80) Year (0.78: 0.81) 0.69 (0.67: 0.71) 0.75 (0.74: 0.78) Year (0.76: 0.79) 0.65 (0.63: 0.66) 0.73 (0.70: 0.74) Year (0.71: 0.76) 0.63 (0.61: 0.65) 0.69 (0.66: 0.72) Stopping due to adverse events Year (0.86: 0.88) 0.85 (0.84: 0.86) 0.87 (0.86: 0.88) Year (0.81: 0.84) 0.77 (0.75: 0.78) 0.81 (0.79: 0.82) Year (0.78: 0.80) 0.72 (0.70: 0.73) 0.79 (0.77: 0.80) Year (0.75: 0.78) 0.68 (0.66: 0.70) 0.76 (0.73: 0.78) Year (0.71: 0.76) 0.65 (0.63: 0.68) 0.73 (0.70: 0.75) 137

138 Table 5.12: Hazard ratios (95 % confidence intervals) in the unadjusted and adjusted analyses for anti-tnf therapy discontinuation Anti-TNF Therapy Hazard ratios (95 % confidence intervals) (i) Etanercept (reference category) Overall stopping Stopping due to inefficacy Stopping due to adverse events Unadjusted Adjusted Unadjusted Adjusted Unadjusted Adjusted Infliximab 1.79* (1.65:1.95) 1.75* (1.60:1.91) 1.70* (1.49:1.94) 1.70* (1.48:1.94) 1.72* (1.51:1.97) 1.61* (1.40:1.85) Adalimumab 1.30 * (1.20:1.42) 1.32* (1.21:1.44) 1.39* (1.22:1.58) 1.41* (1.29:1.61) 1.18* (1.03:1.33) 1.18* (1.04:1.35) (ii) Adalimumab (reference category) Etanercept 0.77* (0.71: 0.83) 0.76* (0.70: 0.82) 0.72* (0.63: 0.81) 0.71* (0.62: 0.81) 0.85* (0.75: 0.83) 0.85* (0.74: 0.97) Infliximab 1.38* (1.26:1.50) 1.32* (1.21:1.45) 1.22* (1.07:1.39) 1.20* (1.05:1.38) 1.46* (1.28:1.79) 1.37* (1.19:1.57) Adjusted for baseline age, female sex, co-morbidities, current smoking, number of previous nbdmards, current steroids, disease duration, baseline DAS28, baseline HAQ. All models were stratified by year of biologic start and the concomitant nbdmards. *p<

139 5.3.6 Effect of baseline concomitant nbdmard therapy on persistence with anti-tnf therapies Baseline characteristics The baseline demographic characteristics (Table 5.13) and disease characteristics (Table 5.14) were described for the seven cohorts of the baseline nbdmards. The seven cohorts were different in size which ranged from 4,418 patients (MTX cohort) to 308 patients (SSZ cohort). Differences in the baseline characteristics among the seven cohorts were observed. The seven groups were different in the proportion of the patients on each biologic. Fifty seven percent of the patients receiving no nbdmards were ETN patients, while forty nine percent of patients receiving single MTX were INF patients. The seven cohorts were similar in disease activity and patient disability. Patients receiving MTX-SSZ, MTX-SSZ-HCQ and MTX-HCQ were younger and showed shorter disease duration and lower percent of co -morbidities compared to the rest of the cohorts. The MTX-SSZ cohort showed the lowest proportion of females and the lowest number of previous nbdmards Percentage of patients who stopped the medication During five years of follow-up, patients receiving no concomitant nbdmards showed the highest percent of discontinuation followed by the patients receiving LEF as a concomitant therapy (Table 5.15). When the reason for discontinuation was considered, the LEF cohort showed the highest percent of discontinuation due to inefficacy, while patients receiving no concomitant nbdmards showed the highest percentage of discontinuation due to adverse events. 139

140 Table 5.13: Baseline demographic characteristics for the seven cohorts according to the baseline nbdmards Characteristics Baseline concomitant nbdmards therapy P-value None MTX SSZ LEF MTX- SSZ MTX- HCQ MTX- SSZ- HCQ - Number of patients 3,339 4, Age (yr) 58.2 ± ± ± ± ± ± ±12.5 < Female gender, n (%) 2,605 (78.0) 3,358 (76.0) 228 (74.0) 469 (76.9) 628 (69.6) 326 (81.3) 301 (72.0) <0.001 Number of previous nbdmards 4.4 ± ± ± ± ± ± ±1.1 < (Mean no.) Current smokers, n (%) 696 (21.0) 947 (21.6) 62 (20.3) 139 (23.0) 212 (23.6) 99 (24.8) 116 (27.8) Co-morbidities, n (%) 2,212 (66.2) 2,557 (57.9) 190 (61.7) 371 (60.8) 478 (53,0) 214 (53.4) 231 (55.3) Anti-TNF therapy Receiving ETN, n (%) 1,921 (57.5) 925 (20.9) 121 (39.3) 244 (40.0) 221 (24.5) 91 (22.7) 82 (19.6) Receiving INF, n (%) 255 (7.6) 2,188 (49.5) 28 (9.1) 105 (17.2) 295 (32.7) 126 (31.4) 139 (33.3) <0.001 Receiving ADA, n (%) 1,163 (34.8) 1,305 (29.5) 159 (51.6) 261 (42.8) 386 (42.8) 184 (45.9) 197 (47.1) Values are mean ± SD unless otherwise specified. MTX: methotrexate, SSZ: sulfasalazine, HCQ: hydroxychloroquine, LEF: leflunomide, nbdmard: non-biologic disease modifying anti-rheumatic drugs, TNF: tumour necrosis factors, ETN: etanercept, INF: infliximab, and ADA: adalimumab. 140

141 Table 5.14: Baseline disease characteristics for the seven cohorts according to the baseline nbdmards Disease characteristics Baseline concomitant nbdmards therapy P-value None MTX SSZ LEF MTX- SSZ MTX- HCQ MTX- SSZ- Disease duration (year) 14.4 ± ± ± ± ± ± ± 8.3 < RF positive, n (%) 2,152 (64.5) 2,858 (64.7) 200 (64.9) 398 (65.3) 546 (60.5) 272 (67.8) 288 (68.9) 0.02 S JC (no.) 11.4 ± ± ± ± ± ± ± TJC (no.) 15.9 ± ± ± ± ± ± ± 7.5 < Global health VAS 74.1 ± ± ± ± ± ± ± CRP (mg/l) 50.9 ± ± ± ± ± ± ± ESR (mm/hr) 49.6± ± ± ± ± ± ± DAS28 score 6.7 ± ± ± ± ± ± ± 0.9 < HAQ score * 2.1 ± ± ± ± ± ± ± Values are mean ± SD unless otherwise specified. MTX: methotrexate, SSZ: sulfasalazine, HCQ: hydroxychloroquine, LEF: lef lunomide, nbdmards: non-biologic disease modifying anti-rheumatic drugs, RF: rheumatoid factor, SJC: swollen joint count, TJC: tender joint count, VAS: visual analogue scale, ESR: e rythrocyte sedimentation rate, DAS: disease activity score, and HAQ: health assessment questionnaire. * HAQ was not available for 5% of the patients. HCQ - 141

142 Table 5.15: Number of patients who stopped the medication in the seven cohorts of the baseline concomitant nbdmards Outcome None MTX SSZ LEF MTX- SSZ MTX- HCQ MTX- SSZ- HCQ Number of patients 3,339 4, Follow-up by person-years Mean(SD)/max (1.8) /7.7 (1.8) /7.7 (1.6) /5.5 (1.6) /8.1 (1.6) /7.1 (1.6) /7.5 (1.7) /7.5 Still on medication, n (%) 1,475 (44.1) 2,165 (49.0) 163 (52.9) 272 (44.6) 579 (64.2) 247 (61.6) 256 (61.2) Stopped due to inefficacy, n (%) 763 (22.8) 959 (21.7) 70 (22.7) 160 (26.2) 148 (16.4) 71 (17.7) 69 (16.5) Stopped due to adverse event, n (%) 833 (25.0) 895 (20.3) 55 (17.9) 135 (22.1) 108 (12.0) 62 (15.5) 65 (15.6) MTX: methotrexate, SSZ: sulfasalazine, HCQ: hydroxychloroquine, LEF: leflunomide, nbdmard: non- biologic disease modifying anti-rheumatic drugs, TNF: tumour necrosis factor. 142

143 Persistence with the anti-tnf therapies across the seven cohorts of baseline concomitant nbdmards therapy Persistence with anti-tnf therapies in the seven cohorts (with all reasons for stopping combined) is shown in Figure 5.3 and the survivor functions (95% CIs) according to the reason of stopping are shown in Table The curves in the figure overlap, and a Log-rank test for equality of survivor functions was performed to test the significance of the difference between different cohorts. Patients receiving anti-tnf therapy in combination with MTX+SSZ showed the highest persistence rates of their anti-tnf therapy with no significant difference when compared with patients receiving MTX+HCQ (p=0.19) and patients receiving MTX+SSZ+HCQ (p=0.06) but a significant difference when compared to MTX (p<0.0001) and all the rest of the cohorts. In other words, concomitant double and triple combination of nbdmards including MTX resulted in equivalent drug persistence which was superior to the rest of the cohorts. No significant difference in persistence rates were found between patients receiving single MTX and those receiving single SSZ (p= 0.83). In addition patients receiving single MTX or single SSZ showed higher persistence rates than those receiving either no nbdmards (p< and p=0.03 respectively) or those receiving single LEF (p= and p< and respectively). No significant difference in persistence rates with anti-tnf therapies was found between patients receiving no concomitant nbdmards and those receiving LEF as the only nbdmard (p=0.56). 143

144 Number of subjects at risk Concomitant nbdmard Time (year) None 3, MTX 4, SSZ LEF MTX-SSZ MTX-HCQ MTX-SSZ- HCQ Figure 5.3: Kaplan-Meier plot for persistence with anti-tnf therapy in the seven cohorts of the baseline concomitant nbdmards MTX: methotrexate, SSZ: sulfasalazine, HCQ: hydroxychloroquine, LEF: leflunomide 144

145 Table 5.16: Survivor function (95% CIs) for patients who continued anti-tnf therapies medication in the seven cohorts of the baseline nbdmards Years of follow-up Survivor function (95% CIs) None MTX SSZ LEF MTX-SSZ MTX-HCQ MTX- SSZ- HCQ Stopping due to all reason Year (0.65:0.68) 0.73 (0.72:0.74) 0.72 (0.68:0.78) 0.64 (0.60:0.69) 0.82 (0.80:0.85) 0.82 (0.77:0.85) 0.77 (0.72:0.81) Year (0.52:0.56) 0.59 (0.58:0.61) 0.57 (0.52:0.64) 0.52 (0.48:0.57) 0.73 (0.70:0.76) 0.69 (0.62:0.73) 0.67 (0.62:0.72) Year (0.46:0.49) 0.52 (0.50:0.53) 0.52 (0.46:0.58) 0.46 (0.42:0.51) 0.65 (0.62:0.69) 0.63 (0.56:0.67) 0.62 (0.56:0.67) Year (0.41:0.45) 0.46 (0.44:0.47) 0.48 (0.41:0.54) 0.40 (0.35:0.44) 0.61 (0.56:0.64) 0.56 (0.50:0.61) 0.56 (0.49:0.62) Year (0.36:0.40) 0.42 (0.40:0.44) 0.46 (0.39:0.53) 0.36 (0.31:0.41) 0.55 (0.51:0.59) 0.51 (0.45,0.58) 0.53 (0.46,0.59) Stopping due to inefficacy Year (0.83:0.86) 0.87 (0.86:0.88) 0.83 (0.80:0.88) 0.82 (0.78:0.85) 0.92 (0.90:0.94) 0.92 (0.89:0.95) 0.88 (0.84:0.91) Year (0.76:0.79) 0.79 (0.78:0.81) 0.76 (0.70:0.81) 0.72 (0.67:0.76) 0.86 (0.83:0.89) 0.85 (0.80:0.89) 0.85 (0.81:0.89) Year (0.72:0.76) 0.75 (0.74:0.77) 0.73 (0.69:0.78) 0.68 (0.63:0.72) 0.81 (0.78:0.84) 0.81 (0.74:0.84) 0.80 (0.76:0.85) Year (0.69:0.73) 0.71 (0.69:0.66) 0.72 (0.66:0.78) 0.64 (0.59:0.69) 0.80 (0.76:0.83) 0.75 (0.69:0.80) 0.78 (0.72:0.82) Year (0.64:0.69) 0.69 (0.67:0.71) 0.70 (0.62:0.76) 0.62 (0.56:0.67) 0.77 (0.74:0.81) 0.74 (0.67,0.79) 0.75 (0.67,0.81) Stopping due to adverse event Year (0.81:0.84) 0.87 (0.86:0.88) 0.90 (0.87:0.94) 0.84 (0.80:0.87) 0.93 (0.91:0.94) 0.91 (0.87:0.94) 0.90 (0.86:0.92) Year (0.74:0.77) 0.81 (0.80:0.82) 0.81 (0.75:0.85) 0.79 (0.75:0.82) 0.91 (0.88:0.92) 0.85 (0.79:0.88) 0.84 (0.78:0.87) Year (0.69:0.73) 0.77 (0.75:0.78) 0.78 (0.71:0.83) 0.75 (0.71:0.79) 0.87 (0.84:0.90) 0.84 (0.78:0.87) 0.82 (0.76:0.87) Year (0.67:0.71) 0.74 (0.72:0.75) 0.74 (0.67:0.80) 0.71 (0.65:0.76) 0.84 (0.80:0.87) 0.80 (0.72:0.84) 0.80 (0.73:0.84) Year (0.65:0.69) 0.72 (0.70:0.73) 0.74 (0.67:0.80) 0.69 (0.63:0.74) 0.82 (0.78:0.85) 0.77 (0.70:0.83) 0.79 (0.73:0.84) MTX: methotrexate, SSZ: sulfasalazine, HCQ: hydroxychloroquine, LEF: leflunomide, nbdmards: non-biologic disease modifying antirheumatic drugs, TNF: tumour necrosis factors. 145

146 When the patients stopping their anti-tnf therapy for inefficacy were considered, LEF concomitant therapy remained in the inferior position and MTX+SSZ remained in the superior position. However, when the analysis was limited to adverse events as the reason for withdrawal, patients receiving no concomitant nbdmard therapies showed the lowest persistence, while MTX+SSZ remained in the superior position. The results of the multivariate Cox proportional analysis (Table 5.17) showed that all cohorts (compared to the MTX mono-therapy cohort, as MTX is the most commonly used nbdmard) were significantly associated with anti-tnf therapies discontinuation; either negatively or positively. Subjects receiving MTX-SSZ, MTX-HCQ or MTX-SSZ-HCQ concomitant therapies were found to be more likely to continue their anti-tnf therapy compared to those receiving single MTX. In contrast, subjects receiving single SSZ, single LEF or no concomitant nbdmards therapies were found to be less likely to continue their anti-tnf therapy. When the analysis was limited to inefficacy or adverse events as reasons for discontinuation, the results remained consistent. When the analysis was stratified by each anti -TNF therapy, the concomitant nbdmards showed similar performance compared to the total anti-tnf cohort except for in the ETN cohort in which patients receiving single SSZ showed similar persistence rates to those receiving single MTX (Table 5.18). 146

147 Table 5.17: Hazard ratios (95 % confidence intervals) in the unadjusted and adjusted analysis of anti-tnf therapies discontinuation nbdmards Co-therapy Hazard ratios (95 % confidence intervals) Overall stopping Stopping due to inefficacy Stopping due to adverse events Unadjusted Adjusted* Unadjusted Adjusted* Unadjusted Adjusted* MTX The reference category None 1.54 (1.43:1.65) 1.40 (1.30:1.51) 1.45 (1.30:1.62) 1.34 (1.20:1.51) 1.69 (1.52:1.89) 1.47 (1.30:1.65) SSZ 1.22 (1.03:1.45) 1.23 (1.03:1.47) 1.34 (1.05:1.71) 1.34 (1.04:1.74) 1.16 (0.88:1.52) 1.21 (0.91:1.60) LEF 1.50 (1.34:1.69) 1.41 (1.25:1.59) 1.64 (1.38:1.94) 1.58 (1.32:1.88) 1.47 (1.22:1.76) 1.34 (1.10:1.62) MTX-SSZ 0.70 (0.62:0.79) 0.76 (0.67:0.86) 0.74 (0.62:0.88) 0.77 (0.64:0.92) 0.59 (0.49:0.73) 0.70 (0.57:0.86) MTX-HCQ 0.78 (0.66:0.92) 0.81 (0.68:0.96) 0.82 (0.65:1.05) 0.83 (0.64:1.06) 0.80 (0.62:1.04) 0.85 (0.66:1.11) MTX-SSZ-HCQ 0.82 (0.70:0.96) 0.80 (0.68:0.95) 0.80 (0.62:1.02) 0.75 (0.58:0.96) 0.83 (0.64:1.07) 0.87 (0.67:1.13) MTX: methotrexate, SSZ: sulfasalazine, HCQ: hydroxychloroquine, LEF: leflunomide, nbdmards: non-biologic disease modifying anti-rheumatic drugs, TNF: tumour necrosis factors. * Adjusted for baseline age, female sex, number of previous nbdmards, current steroids, co-morbidities, current smoking, disease duration, baseline DAS28, baseline HAQ. All models were stratified by year of anti-tnf start and anti-tnf therapy. 147

148 Table 5.18: Adjusted hazard ratios (95 % confidence interval) for anti-tnf therapy discontinuation (within each anti-tnf agent) nbdmard Adjusted hazard ratios (95 % confidence interval) co-therapy ETN INF ADA MTX N Overall stopping Stopping due to inefficacy Stopping due to adverse events N Overall stopping Stopping due to inefficacy Reference Stopping due to adverse events N Overall stopping Stopping due to inefficacy Stopping due to adverse events None (1.12:1.45) 1.13 (0.92:1.38) 1.51 (1.23:1.86) (1.17:1.63) 1.17 (0.88:1.55) 1.50 (1.18:1.90) (1.25:1.61) 1.50 (1.25:1.80) 1.36 (1.11:1.67) SSZ (0.64:1.24) :1.36) 1.04 (0.63:1.70) (0.92:2.19) 1.87 (1.02:3.42) 0.79 (0.32:1.90) (1.11:1.80) 1.57 (1.11:2.23) 1.50 (1.02:2.19) LEF (1.02:1.57) 1.41 (1.03:1.92) 1.34 (0.96:1.87) (1.43:2.26) 2.17 (1.56:3.02) 1.47 (1.00:2.15) (1.07:1.60) 1.41 (1.06:1.90) 1.25 (0.91:1.72) MTX-SSZ (0.49:0.85) 0.69 (0.46:1.05) 0.55 (0.34:0.91) (0.73:1.04) 0.99 (0.76:1.28) 0.70 (0.51:0.95) (0.57:0.90) 0.61 (0.43:0.85) 0.82 (0.58:1.17) MTX-HCQ (0.68:1.37) 1.08 (0.66:1.76) 1.07 (0.62:1.86) (0.58:0.98) 0.82 (0.56:1.20) 0.70 (0.46:1.08) (0.58:1.03) 0.69 (0.45:1.07) 0.94 (0.60:1.45) MTX-SSZ- HCQ (0.64:1.35) 0.86 (0.49:1.52) 0.92 (0.49:1.71) (0.59:0.96) 0.64 (0.43:0.96) 0.88 (0.61:1.28) (0.60:1.05) 0.81 (0.55:1.21) 0.78 (0.49:1.24) N: number of patients, INF: infliximab, ETN: etanercept, ADA: adalimumab, MTX: methotrexate, SSZ: sulfasalazine, HCQ: hydroxychloroquine, LEF: leflunomide, nbdmards: non-biologic disease modifying anti-rheumatic drugs, TNF: tumour necrosis factors Adjusted for baseline age, gender, number of previous nbdmards, presence of co-morbidities, current smoking, current steroids, disease duration, baseline DAS28, and baseline HAQ. All models were stratified by start year. 148

149 5.3 Summary In this cohort of patients who were treated in routine clinical practice in the UK and who were treated with their first anti-tnf therapy, the following results were found. The overall persistence with the anti-tnf therapies was 72% in the first year which decreased to 42% in the fifth year. High baseline HAQ score, female gender, having co-morbidities, current smoking, and failing 4 or more previous nbdmards were significant predictors of overall anti -TNF therapies discontinuation. Patients were significantly more likely to discontinue INF and ADA compared to ETN. Subjects receiving double and triple concomitant nbdmards including MTX w ere more likely to continue their anti-tnf therapy compared to those receiving single MTX. In contrast, subjects receiving single SSZ, single LEF or no concomitant nbdmards therapies were less likely to continue their anti-tnf therapy. The results were largely consistent when the analyses were undertaken according to the reason for discontinuation (all reasons, ineffectiveness of the therapy, and developing adverse event). 5.4 Discussion Drug persistence with anti-tnf therapies has previously been assessed in a number of studies involving national registers. These studies have reported variable results. One year drug persistence has been assessed in four national studies with persistence rates of 83%, 74%, 65.4% and 73% in the Spanish register, American database, Norwegian register and Danish register respectively compared to 72% in the present study. (88;90;101;103) The Spanish register assessed the overall persistence with 149

150 anti-tnf therapies up to three years, with a survival function (95% CIs) of 0.83 (0.81: 0.84), 0.72 (0.71: 0.74) and 0.65 (0.63: 0.67) in the first, second and third years respectively compared to 0.72 (0.71: 0.73), 0.59 (0.58: 0.60) and 0.52 (0.51: 0.53) in this PhD. (101) Differences in baseline characteristics, enrolment period, and study design across patients from different registries may explain the variation in results. In addition, differences in the guidelines applied in each country of those studies may also explain the variation in the results. In the present study INF showed inferior persistence to ETN and ADA, this comes in agreement with a number of previous studies. (82;84;97;101) The reasons for this are not clear. However it may be explained by the fact that INF was first approved by the EMA in 1999 followed by ETN in 2000 and finally ADA in So INF was used early, before new medications (ETN and ADA) became available. There is a possibility that some patients have switched to the new medications when they became available. However, all the regression models were stratified by the year of starting the anti - TNF therapies to adjust for this possible confounding and INF was still inferi or to ETN and ADA. The route of administration of the drug and the need for monitoring during and after administration may affect patient and the physician preferences for the available anti-tnf therapies. Subcutaneous injection (in the case of ETN and ADA) is more straightforward and convenient to the patients and the physicians compared to intravenous infusion (in case of INF) especially when there is a possibility of poor vein access in some patients. High baseline HAQ score was the strongest predictor of treatment discontinuation due to all reasons. High HAQ score was previously identified as a predictor of ETN and INF treatment termination in a cohort of 1,161 RA patients in Sweden. (84) High baseline HAQ may be a marker of 150

151 irreversible resistant disease. Previous studies had shown that patients with high HAQ scores were less likely to respond to anti-tnf therapies. (147) The present study agrees with the study by Heiberg et al. (90) who reported female sex as a significant predictor of treatment discontinuation. High number of previous nbdmards was reported to be a significant predictor of overall treatment discontinuation. (83;84) The present study confirmed this finding in the overall discontinuation and in the discontinuation due to inefficacy. Females may have some biological differences compared to males that might have affected the effectiveness and tolerance of the therapy. In addition RA is reported to be more common in females and the reasons are not fully understood. Higher number of previous nbdmards may reflect that the patients had drug-resistant disease or they could not tolerate drugs in general. Having co-morbidities and current smoking were found to be significant predictors of anti -TNF therapy discontinuation. Smoking put the patient under risk of developing adverse events that might not be related to the drug but may lead to rapid drug termination. Similarly patients with comorbidities may be more likely to develop adverse events. (147) The results have confirmed the benefits of combining MTX with anti-tnf therapies. In the case of INF, which is recommended to be used in combination with MTX, (148) the benefit may be due to the reduction of human anti-chimeric antibodies. (46;149) These finding come in agreement with both clinical trials and observational data. (47;150;151) However, the mechanism of this benefit is not well understood with the other anti-tnf therapies. In the literature, the studies that explored the role of the concomitant use of nbdmards on the persistence with anti-tnf therapies were mostly limited to MTX and combined all the other nbdmards in one group usually due to small sample size. In this PhD, the role of the most 151

152 commonly used nbdmards including double and triple combinations of nbdmards was examined. Current results showed that patients who received anti-tnf therapies in combination with two or three nbdmards including MTX were more likely to continue their anti-tnf therapy compared to those who were receiving their anti-tnf with only MTX. This may be explained by the ability of some patients to tolerate drugs in general. Patients who were able to tolerate double and triple nbdmards combinations may be more likely to tolerate the anti-tnf therapy. This explanation may be supported by the high rate of adverse events among those patients receiving no concomitant nbdmards. In contrast patients who received an anti-tnf alone or in combination with one nbdmard (SSZ or LEF) were more likely to discontinue their anti-tnf therapy. Those patients may have been unable to tolerate nbdmards and subsequently they could not tolerate the anti -TNF therapy. This may be supported by the higher number of previous nbdmards in those patients. Previous reports on the benefits and risks of combining anti-tnf therapy with LEF have provided mixed results. Two studies have found similar efficacy, whether measured using DAS28, ACR scores or treatment persistence, between anti-tnf in combination with either MTX or LEF. (99;152) A further Dutch study (n=162), which compared anti-tnf persistence between those receiving LEF and those receiving any other nbdmard(s), also found similar outcomes. (93) In contrast Strangfeld et al. (153) reported lower persistence rates of anti-tnf therapies (especially with INF) (n=1769) when combined with LEF than when combined with MTX, although this did not reach statistical significance. Other studies, the majority limited to the combination of LEF with INF, have reported acceptable efficacy although there has been a concern about the safety and longer-term tolerability of anti-tnf therapy when combined with LEF. ( ) In particular, there has been a concern about dermatological and autoimmune adverse events. A small study from Leeds, UK, reported a high incidence of induction of antinuclear antibodies and double stranded DNA antibodies in patients receiving INF with LEF. (159) Although there was no control group in this report, there was a high 152

153 incidence of immune-related adverse events leading to anti-tnf discontinuation. In the current analysis, combination of LEF with INF was less common than with either ETN or ADA, although a stratified analysis found similar results across all three agents. Similar results to LEF were found with the combination of SSZ with anti -TNF therapy, particularly when SSZ was combined with the monoclonal antibodies, INF and ADA. Less has been published on this combination. A clinical trial found a similar efficacy between ETN mono-therapy and ETN + SSZ in SSZ failures. (160) In contrast, in our study, ETN + SSZ appeared to result in better outcomes than ETN mono-therapy, although the difference in persistence between the three anti-tnf therapies did not reach statistical significance. 153

154 6. Choice of treatment following failure a first anti-tnf therapy: rituximab or a second alternative anti-tnf therapy 154

155 The findings presented in the last chapter showed that the persistence with the first anti -TNF therapy was 72% after the first year and then the persistence rates decreased each year reaching 42% after five years. In clinical practice, those patients who discontinue their anti -TNF therapy either due to inefficacy or due to developing an adverse event may consider starting another biologic therapy. One option for them would be to start another alternative anti-tnf therapy. Alternatively, with the introduction of RTX, those patients could switch biologic drug class to RTX. Comparing the response after each of those two options is of interest, as there are currently no head -to-head randomised controlled trials to guide clinical decision-making on this issue. This chapter will assess this comparison using data from the BSRBR. 6.1 Aim and objectives This chapter aims to compare the response of RTX versus a second alternative anti -TNF therapy in RA patients treated in routine clinical practice who had failed their first anti -TNF therapy. The following two objectives were considered. To compare at six months the clinical effectiveness of RTX versus a second alternative anti- TNF. To compare at six months the patient s reported physical function after switching to RTX or a second alternative anti-tnf. To achieve these objectives, a cohort of patients registered with the BSRBR was used. 155

156 6.2 Methods BSRBR data frozen at 21 st December 2010 were used for the analysis presented in this chapter Inclusion and exclusion criteria Patient inclusion criteria (performed in order) i. All patients registered with the BSRBR ii. Patients with RA indication iii. Patients who had failed their first course of anti-tnf therapy (ETN, INF or ADA). iv. Patients who had switched to either RTX or a second alternative anti-tnf therapy. Patient exclusion criteria i. Patient who switched to an alternative anti-tnf therapy or RTX after failing more than one anti- TNF therapy. ii. Patients who were anti-tnf therapy naïve. iii. Patients for whom no data (DAS28/ HAQ) were available at the time they switched or at the time of assessment of response Identification of patients Switchers were identified if the biologic therapy the patient was on in a follow-up was different from the patient s baseline biologic therapy or from the biologic therapy reported in the previous followup. Patients who were registered with the BSRBR when they started RTX (after failing only one anti- TNF therapy) and were not in the register while they were on the anti -TNF therapy were also counted as switchers. The date of switching was identified from the date of start of the new biologic therapy. The date of assessment of response was identified six months after the date of switching. The reason for switching was identified from the reason for stopping the first anti -TNF therapy. 156

157 6.2.3 Outcome measures Clinical outcomes included improvements in disease activity score 28 joints (DAS28), European League Against Rheumatism (EULAR) criteria (good, moderate, and non-responders), and proportions of patients achieving remission (DAS28<2.6). Patient reported outcomes included improvements in Health Assessment Questionnaire (HAQ) scores and proportion of patients achieving minimal clinically important difference in HAQ (0.22 units or more). The time end point of assessment of response was six months from starting RTX or the second anti -TNF therapy Baseline and follow-up data collection The BSRBR was not designed to collect disease activity or disability data at the time a patient switched his therapy. Instead, if a patient switched his therapy, the BSRBR follow-up continued as scheduled regardless of whether the patient had stopped or switched his therapy. However, within the BSRBR, the collected data either at baseline or at follow-ups were recorded along with the dates it was measured and therefore, corresponding data could be identified where recorded. For the purpose of this analysis the patients were included if they had recorded data (DAS28 and/or HAQ) at the time they switched their therapy and six months after they switched. As the BSRBR was not intentionally designed to collect data at the time a patient switched therapy, a time window of three months was used to match the dates of switching with the dates of the BSRBR collected data. The patient was included if he/she had recorded data within three months before the switching date and corresponding recorded data within three months before or after the six-month follow-up date. 157

158 6.2.5 Baseline characteristics The baseline (switching-time) age and the disease duration of the patients were calculated from the patients age recorded in their BSRBR consultant baseline questionnaire and the time of switching. Switching-time DAS28 (and its components) and HAQ score were identified if they were dated within three months before switching (reported on a BSRBR baseline form or a follow-up form). Other demographic characteristics (that would not change by time) were obtained from the patients BSRBR consultant baseline forms; this included gender, and rheumatoid factor status. As co-morbidities could potentially change with time, data from the BSRBR were used to test this change. For patients who were re-registered for RTX therapy, the co-morbidities were collected at both the first registration (for anti-tnf therapy) and the second registration (for RTX). The two lists of co-morbidities for those patients were compared and no differences were observed. As a result, the use of the co-morbidities listed at the time of the first registration was valid and carried forward in this analysis Statistical analysis STATA 10.1 software was used for performing all statistical analyses. The statistical analysis involved two separate analyses. The first included patients who had DAS28 scores recorded at the time they switched to either RTX or a second anti -TNF therapy and six months after switching. The second included patients who had HAQ scores recorded at the time they switched to either RTX or a second anti-tnf therapy and six months after switching. Separate comparisons between the patients who switched to RTX and those who switched to a second anti - TNF therapy were performed in each of these two analyses. 158

159 Differences in baseline characteristics between the two cohorts of patients (patients who switched to RTX and those who switched to a second anti-tnf therapy) were tested by the two independent samples t-test if the data were continuous or by Pearson Chi square test if the data was categorical. The response rates were compared between patients who switched to RTX and those who switched to a second anti-tnf using two independent samples t-test for continuous outcomes and Pearson Chi square tests for categorical outcomes. If a patient discontinued therapy or started a further biologic within the six months before assessment of response, the patient was counted as a nonresponder in the categorical outcomes. Unadjusted and adjusted regression models were used to compare the response between patients who switched to RTX or to a second alternative anti-tnf therapy. Ordinal regression was used to compare EULAR response (the odds ratio represented being in a higher EULAR category). Logistic regression was used to compare (i) EULAR responders versus non responders and (ii) achieving MCID in HAQ versus not achieving it. In order to adjust for any differences between RTX and anti -TNF patients and overcome absence of randomisation to the therapy, the models were adjusted for propensity scores. Logistic regression models were used to calculate propensity scores. (161) Two separate models were developed, the first for patients with DAS28 data and the second for patients with HAQ scores data. The models included age, disease duration, DAS28 at time of therapy start, co-morbidities, the first anti-tnf therapy (previously failed anti-tnf) and the reason for stopping the first anti-tnf. A propensity score can be defined as the probability of a patient being assigned to a particular treatment (RTX or a second alternative anti-tnf) given the set of known baseline covariates. 159

160 Propensity scores are used to overcome selection bias by equating the two groups of patients based on the baseline covariates. The propensity score was firstly introduced in 1983 (162) to estimate treatment effects when no randomisation is considered in treatment assignment with the aim to reduce the effect of confounding. The most common way to calculate propensity scores is to perform logistic regression to calculate the odds of any given patient receiving treatment based upon the baseline characteristics. Then a balancing method is used to balance the covariates. (163) The inverse probability of treatment weighting method (161) was used as a balancing method within this analysis to give each patient a weight which was then used to adjust the regression models. 6.3 Results Patients included The included patients were either registered as anti-tnf patients and switched to another anti-tnf, registered as anti-tnf patient then switched to RTX patients and not reregistered for RTX, or were registered as RTX patient (both who were on the register during the first anti-tnf or were new to the register). Figure 6.1 follows those patients. By 21 st December 2010, there were 4,158 patients in the BSRBR who had switched their first anti - TNF to a second alternative anti-tnf and 1,180 patients (660 registered as anti-tnf patient registered as RTX patient) who had switched their first anti-tnf to RTX. DAS28 data at time of switching and six months later were available for total of 1,328 patients while, HAQ data at time of switching and six months later were available for total of 937 patients. 160

161 Total RA patients registered with the BSRBR 18,194 Registered as anti-tnf 13,122 Receiving RTX 2,377 Had switched from their first anti-tnf to a second alternative anti-tnf 4,158 Registered as RTX patient 1,138 Had failed only one previous anti-tnf 520 Not registered as RTX patient 1,239 Had failed only one previous anti-tnf 660 Had baseline HAQ 930 Had baseline DAS28 1,564 Had baseline HAQ 453 Had baseline DAS Had baseline HAQ 79 Had baseline DAS All RTX patients who had baseline HAQ data 532 All RTX patients who had baseline DAS28 data 749 Had both baseline and 6 months HAQ 693 Had both baseline and 6 months DAS Had both baseline and 6 months HAQ 244 Had both baseline and 6 months DAS Patients with HAQ data 937 Patients with DAS28 data 1328 Figure 6.1: Flow diagram of patients included in the comparative analysis. 161

162 6.3.2 Assessment of clinical effectiveness Baseline characteristics From a total of 1,328 patients included in this assessment with DAS28 scores, 941 patients switched to a second anti-tnf therapy and 387 patients switched to RTX. Baseline characteristics at the time the patients switched their therapy are shown in Table 6.1. Patients who switched to a second anti-tnf were generally younger (55.6 years versus 58.7 years; p- value <0.0001) and a lower proportion of them had co-morbidities (58.0% versus 66.9%; p-value <0.003) compared to patients who switched to RTX. Anti-TNF patients also had a lower mean (SD) DAS28 score (5.9 (1.4) compared to 6.2 (1.2) in RTX patients). Inefficacy as a reason of discontinuation of the first anti-tnf was more common in patients who switched to a second alternative anti-tnf (59.5%) than in patients who switched to RTX (49.9%) Comparing crude clinical effectiveness Patients who switched to a second alternative anti-tnf showed a mean (95% CI) improvements in their DAS28 score of (-1.28: -1.07) units compared to (-1.48: -1.17) in patients who switched to RTX. Statistical tests of significance of difference between the two groups showed no significant difference (p=0.12) (Table 6.2). Within six months from switching, 283 patients had stopped their therapy or switched again and were counted as EULAR non-responders. Patients who switched to a second alternative anti-tnf showed lower EULAR response rates compared to those who switched to RTX (p=0.04). A good EULAR response was achieved in 17.1% of the patients who switched to RTX compared to 13.5% of those who switched to a second alternative 162

163 anti-tnf. A moderate EULAR response was achieved in 37.7% of the patients who switched to RTX compared to 33.8% of those who switched to a second alternative anti-tnf. However, a greater proportion of patients who switched to a second anti-tnf achieved disease remission. Disease remission was achieved in 7.2% of the patients who switched RTX compared to 10.4% of those who switched to a second anti-tnf therapy (p=0.07) Adjusted comparison of clinical effectiveness After adjustment for propensity scores that considered baseline DAS28 score, co-morbidities, type of failed anti-tnf therapy, and interaction between age and reason for switching; patients who switched to RTX after failing their first anti-tnf were significantly more likely to improve as measured by EULAR response criteria, compared to those who switched to a second alternative anti- TNF therapy with OR and 95%CI of 1.34 (1.05: 1.70) (Table 6.3). 163

164 Table 6.1: Baseline characteristics (at time of switching) for patients with DAS28 data Characteristics Switched to Switched to P value anti-tnf RTX Number of patients Mean age ± SD (year) 55.6 ± ± 11.2 < Female, n (%) 757 (80.5) 300 (77.5) 0.23 Mean disease duration ± SD (year) 14.0 ± ± RF positive, n (%) 604 (64.2) 242 (64.9) 0.06 Co-morbidities*, n (%) 546 (58.0) 259 (66.9) TJC (no.) ± SD 12.5 ± ± 8.1 <0.001 S JC (no.) ± SD 8.2 ± ± ESR (mm/hr) ± SD 43.0 ± ± CRP (mg/l) ± SD 39.7 ± ± Global health VAS ± SD 63.6 ± ± 20.8 <0.001 Mean DAS28 ± SD 5.9 ± ± 1.2 <0.001 Reason for stopping the first anti-tnf: Inefficacy, n (%) 547 (58.1) Adverse events, n (%) 257 (27.3) Other/missing, n (%) 137 (14.6) The first anti-tnf therapy: ETN, n (%) 266 (28.3) Monoclonal antibody, n (%) 675 (71.7) Therapy switched to, n (%) ETN: 494 (52.5) INF: 103 (11.0) ADA: 344 (36.5) 190 (49.1) (24.8) 101 (26.1) 159 (41.1) < (58.9) RTX - RTX: rituximab, anti-tnf: anti-tumour necrosis factor, RF: rheumatoid factor, TJC: tender joint count, SJC: swollen joint count, ESR: erythrocytes sedimentation rate, CRP: C-reactive protein, VAS: visual analogue scale, DAS: disease activity score, ETN: etanercept, INF: infliximab, ADA: adalimumab. Test of significance between the two groups of patients. *Co-morbidities included one or more of angina, hypertension, myocardial infarction, stroke, epilepsy, asthma, chronic obstructive pulmonary disease, peptic ulcer, liver disease, renal disorder, demyelination, diabetes, hyperthyroidism, depression or history of TB and cancer. 164

165 Table 6.2: Six months clinical effectiveness Outcomes Switched to anti-tnf Switched to RTX P value* Number of patients Baseline DAS (5.8: 6.0) 6.2 (6.1: 6.3) <0.001 Six months DAS (4.6: 4.8) 4.9 (4.7: 5.0) 0.15 Change in DAS (-1.3: -1.1) -1.3 (-1.5: -1.2) 0.12 EULAR response, n (%): Good 127 (13.5) 66 (17.1) Moderate 318 (33.8) 146 (37.7) 0.04 None 496 (52.7) 175 ( 45.2) Achieving disease remission, n (%) 98 (10.4) 28 (7.24) 0.07 Values are in mean (95% confidence intervals) otherwise specified. RTX: rituximab, anti-tnf: antitumour necrosis factor, EULAR: European League Against Rheumatism, DAS: disease activity score. * Test of significance between the two groups of patients. 165

166 Table 6.3: Undjusted and adjusted regression analyses results for the six months EULAR response Therapy Logistic regression of EULAR responders versus non responders Ordinal regression of EULAR response Anti-TNF : Reference Odds ratios (95% CI) Unadjusted Adjusted* Unadjusted Adjusted* RTX (1.06: 1.71) (1.02: 1.69) (1.07: 1.68) (1.05: 1.70) * Adjusted for propensity scores that included DAS28 score, co-morbidities, type of the failed anti -TNF therapy, and interaction between age and reason for switching using the logistic regression propensity score model. RTX: rituximab, TNF: tumour necrosis factor, CI: confidence intervals, and EULAR: European League Against Rheumatism Assessment of patient reported physical function Baseline characteristics From a total of 937 patients included in this assessment, 693 patients switched to a second anti-tnf therapy and 244 patients switched to RTX. Baseline characteristics at the time the patients switched their therapy are shown in Table 6.4. Patients who switched to a second anti-tnf were generally younger and a lower proportion of them had co-morbidities compared to patients who switched to RTX. The two groups of patients also had similar mean (SD) HAQ score Comparing crude improvements in physical function Patients who switched to a second alternative anti-tnf and those who switched to RTX showed similar mean (95% CI) improvements in their HAQ scores (-0.11 (-0.13: -0.08) and (-0.17: -0.08) units respectively) (p=0.12) (Table 6.5). 166

167 Within six months from switching, 109 patients had stopped their therapy or switched it again and were counted as not achieving MCID in HAQ. A lower proportion of patients who switched to a second alternative anti -TNF (29.6%) achieved a MCID in HAQ compared to those who switched to RTX (38.4%) (p=0.01). The MCID in HAQ was achieved in 38.4% of the patients who switched RTX compared to 29.6% of those who switched to a second anti-tnf therapy Adjusted comparison of improvements in physical function After adjustment for propensity scores that considered baseline age, co-morbidities, the type of the first failed anti-tnf, interaction between disease duration and the reason for switching; patients who switched to RTX after failing their first anti-tnf were significantly more likely to improve as measured by achieving a MCID in HAQ compared to those who switched to a second alternative anti-tnf therapy with OR and 95%CI of 1.49 (1.07: 2.08) (Table 6.6). 167

168 Table 6.4: Baseline characteristics at time of switching for patients with HAQ scores Characteristics Switched to Switched to RTX P value anti-tnf Number of patients Mean age ± SD (yr) 57.7 ± ± 10.8 <0.01 Female, n (%) 558 (80.5) 189 (77.5) 0.31 Mean disease duration ± SD (yr) 14.6 ± ± RF positive, n (%) 444 (64.1) 166 (68.0) 0.15 Co-morbidities*, n (%) 420 (60.6) 166 (68.0) 0.04 Mean HAQ ± SD 1.96 ± ± Reason for stopping the first anti-tnf Inefficacy, n (%) Adverse events, n (%) 383 (55.2) 208 (30.1) 113 (46.5) 75 (30.6) Other/missing, n (%) 102 (14.7) 56 (22.9) The first anti-tnf therapy: ETN, n (%) 185 (26.7) 103 (42.2) <0.001 Monoclonal antibody, n (%) 508 (73.3) 141 (57.8) Therapy switched to, n (%) ETN: 380 (54.8) INF: 58 (8.4) RTX - ADA: 255 (36.8) RTX: rituximab, TNF: tumour necrosis factor, RF: rheumatoid factor, HAQ: health assessment questionnaire, ETN: etanercept, INF: infliximab, ADA: adalimumab. Test of significance between the two groups of patients. *Co-morbidities included one or more of angina, hypertension, myocardial infarction, stroke, epilepsy, asthma, chronic obstructive pulmonary disease, peptic ulcer, liver disease, renal disorder, demyelination, diabetes, hyperthyroidism, depression, or history of TB or cancer. 168

169 Table 6.5: Six month patient reported physical function Outcomes Switched to anti-tnf Switched to RTX P value* Number of patients Baseline HAQ 1.96 (1.92: 2.00) 1.99 (1.91: 2.07) 0.49 Six months HAQ 1.85 (1.80: 1.90) 1.86 (1.78: 1.95) 0.81 Change in HAQ (-0.13: -0.08) (-0.17: -0.08) 0.51 Patients achieving MCID improvement in HAQ, n (%) 205 (29.6) 94 (38.4) 0.01 Values are in mean (95% confidence intervals) otherwise specified. RTX: rituximab, anti-tnf: antitumour necrosis factor, MCID: minimal clinically important difference, HAQ: health assessment questionnaire. * Test of significance between the two groups of patients. Table 6.6: Adjusted and unadjusted regression of six months patient s reported physical function Therapy Logistic regression of achieving MCID in HAQ Odds ratios (95% CI) Unadjusted Adjusted* Anti-TNF Reference RTX 1.48 (1.09: 2.01) 1.49 (1.07: 2.08) * Adjusted for propensity scores that included baseline age, co-morbidities, the type of the first anti- TNF, interaction between disease duration and the reason for switching; using the logistic regression propensity score model. RTX: rituximab, anti-tnf: anti-tumour necrosis factor, CI: confidence intervals, MCID: minimal clinically important difference, and HAQ: health assessment questionnaire. 169

170 6.3.4 Summary In this cohort of patients who were treated in routine clinical practice in the UK and who had failed their first anti-tnf therapy, the following results were found. Switching to RTX was found to be more effective than switching to a second anti-tnf therapy. The superior effectiveness of RTX was seen in both improving the clinical outcomes as measured by the EULAR response criteria and improving the patient reported outcomes as measured by achieving a MCID in HAQ scores. 6.4 Discussion To my knowledge, this is the first study to compare physical function after switching to either RTX or a second alternative anti-tnf therapy in RA patients who had failed their first anti-tnf. A previous observational study of Finckh et al. had compared the change in DAS28 after switching to either RTX or alternative anti-tnf. (122) However the sample size was small; 50 patients switched to RTX and 66 patients switched to an alternative anti-tnf compared to 387 and 941 patients respectively in the current analysis. A recent update of this study was of larger sample size (155 and 163 patients respectively). (127) However, the focus on this update was on subgroups of the patients according to the reason for switching or the number of failed anti-tnf therapies. The results of the current analysis came in agreement with the results of those previous observational studies that reported superior response in patients who switched to RTX. (122;127) This superior response to RTX may be explained by the different mechanism of action offered by RTX. 170

171 There was a higher rate of achieving disease remission in the patients who switched to an alternative anti-tnf therapy; however it was not statistically significant. This might be due to the lower DAS28 score at start of therapy in this group of patients. Previous analysis from the BSRBR had shown that, six months after initiating the first anti-tnf therapy in RA patients, the mean change in DAS28 was -2.1 with 18% of the patients achieved good EULAR response and 50% were moderate responders. (147) In comparison to the results of the current analysis, it was noticed that the overall mean response to the second alternative anti -TNF (-1.17 mean change in DAS28, 13.5% good responders and 33.8% moderate responders) was relatively lower than that of the first anti-tnf. After failing the first anti-tnf, the patients in both arms were of similar physical function. The improvements in HAQ score six months after switching was small and did not reach the MCID in both treatments which may suggest an irreversible physical disability in those patients who already have failed one anti-tnf. 171

172 7. Assessment of effectiveness of rituximab in rheumatoid arthritis 172

173 Findings from the last chapter showed that RTX was a more effective treatment option for RA patients who had failed their first anti-tnf therapy compared to an alternative anti-tnf therapy. This emphasises the importance of RTX as a treatment option for RA patients and supports the further examination of the performance of RTX in routine clinical practice in a very wide inclusion criteria cohort of patients in which all patients receiving RTX can be included, irrespective of their previous biologic history. This chapter examines the performance of RTX in RA patients treated in routine clinical practice in the UK. 7.1 Aim and objectives This study aimed to assess both the clinical effectiveness of RTX and patient reported outcomes after receiving RTX, in RA patients, in routine clinical practice in the UK. The following specific objectives were considered. To study the overall six month clinical effectiveness of RTX. To study the clinical effectiveness of RTX in subgroups of patients according to previous anti - TNF history, RF status, and the concomitant nbdmards. To identify predictors of clinical effectiveness of RTX six months after starting the the rapy. To examine whether being an anti-tnf naïve patient or having previously failed one or more anti-tnf therapies would predict the clinical response to RTX. To study the overall six month impact on quality of life and physical function of patients after receiving RTX. To study the quality of life and physical function of patients treated with RTX in subgroups of patients according to previous anti-tnf history, RF status, and concomitant nbdmards. To identify predictors of six month improvements in physical functions of patients after receiving RTX. 173

174 To examine whether being an anti-tnf naïve patient or having previously failed one or more anti-tnf therapies would predict six month physical function of patients after receiving RTX. To achieve these objectives, a cohort of patients registered with the BSRBR was used. 7.2 Methods BSRBR data frozen at 21 st December 2010 were used for the analysis of this chapter Inclusion and exclusion criteria Patients inclusion criteria (performed in order) i. All patients registered with the BSRBR ii. Patients with RA indication iii. Patients who were registered as RTX patients. iv. Patients who had completed a minimum of six months of follow-up. Patients exclusion criteria i. Patient who received RTX but were not registered as RTX patients (no data were available at the time they started RTX). ii. In case of studying the overall six month clinical effectiveness of RTX, patients for whom no DAS28 data were recorded at the time RTX was started or at the time of assessment. iii. In case of study the overall six month physical function of patients after receiving RTX, patients for whom no HAQ data were recorded at the time RTX was started or at the time of assessment Outcome measures 174

175 The end point time of assessment of clinical response to RTX was six months from starting the therapy. The outcome measures of clinical effectiveness included the change in DAS28 scores, EULAR response criteria, and percentage of patients achieving disease remission which is defined as DAS28 score < 2.6. The outcome measures of patients physical function included the change in HAQ scores and the percentage of patients achieving MCID in HAQ which is defined as achieving at least a 0.22 unit improvements in HAQ score. The outcome measures of the health related quality of life of the patients included (i) the EQ-5D profile, (ii) EQ-5D utility score (range to 1.00 where 0 = death, 1 = full health and negative values represent health states considered = worse than death) and (iii) EQ-5D VAS (range 0 to 100) (for further details see Chapter 1 section ). The EQ-5D utility score was calculated using the algorithm shown in Appendix 7.1. (34) Baseline characteristics The baseline characteristics of the total RTX patients cohort and according to the sub-cohorts of patients who were anti-tnf naïve versus those who were anti-tnf failures were studied. The baseline characteristics described the patients demographics including age, gender; smoking, current and previous nbdmards therapies and co-morbidities. In addition the disease characteristics were also described. This information of baseline data were obtained from different forms within the BSRBR depending on whether the patient was new to the register or had previously been registered with the BSRBR during previous anti-tnf therapy (see Figure 4.2, Chapter 4). 175

176 7.2.4 Statistical analysis STATA 10.1 software was used for all statistical analyses. The statistical analysis involved two separate analyses. The first included patients who had DAS28 scores recorded at the time they started RTX and six months after starting. The second analysis considered patients who had HAQ scores recorded at the time they started RTX and six months after starting. Paired t-test was used to test the significance of the improvements in six months outcomes compared to baseline (DAS28 scores, HAQ scores, EQ-5D utility scores, EQ-5D VAS). McNemar's chisquare test was used in case of binary outcomes (proportions of patients reporting any problems in EQ-5D profile) at six months compared to baseline. Two independent samples t-tests were used to compare the change in DAS28 scores, change in HAQ scores, change in EQ-5D utility scores, and change in EQ-5D VAS between (i) RF positive versus negative patients and (ii) anti-tnf naïve versus anti-tnf failure patients. One-way analysis of variance (ANOVA) was used to compare the change in the same outcome measures across patients who were receiving RTX as (i) mono-therapy, (ii) in combination with MTX, and (iii) in combination with other nbdmards (other than MTX). Pearson Chi squared tests were used to examine differences in proportions of patients achieving different EULAR response rates, disease remission, and proportions of patients achieving a MCID in HAQ between the different cohorts of the patients. Predictors of response to RTX at six-months were identified using both univariate and multivariate regression models. Different regression models were used for different outcome measure. Predictors of improvements in DAS28 scores and HAQ scores were identified using linear regression 176

177 models while predictors of achieving disease remission and achieving a MCID in HAQ were identified using logistic regression models. Multinomial logistic regression models were used to identify predictors of achieving a moderate EULAR response and a good EULAR response (compared to EULAR non-responders). Given that some of the patients were anti-tnf naïve and the previous biologic history is likely to be important in terms of probability to predict the response to RTX, anti-tnf naïve patients were not included in the regression models that examined predictors of response to RTX. The regression models examined a number of baseline variables that included: age (by 10 year age - bands), gender, presence of co-morbidities, disease duration (10 year bands), RF status, baseline DAS28 score, baseline HAQ score (0.1 unit), baseline concurrent steroid usage, baseline concurrent nbdmards (none/mtx/other), previous steroid usage, number of the previous nbdmards (4 or more), number of the previous anti-tnf therapies (more than one), and the reason for discontinuation of the last anti-tnf therapy. Given that there is a degree of correlation among some of these covariates, they cannot be included in one multivariate model all together. As a result, a selection method was essential to choose covariates that should be included in the multivariate models. Specifically, a forward stepwise selection multivariate regression models were used. A significance level of 0.05 was appl ied to select the variables. Given that anti-tnf naïve patients were not considered in the regression models that examined potential predictors of response to RTX, it still remained to be determined whether being anti-tnf naïve predicted response to RTX. To address this question, separate unadjusted and adjusted 177

178 regression models were used. Propensity scores were used to adjust for any differences at baseline between anti-tnf naïve and failure patients and for other predictors of response to RTX. Propensi ty scores were calculated using a logistic regression model. Beta- coefficients and their corresponding 95% CIs were used to present results from the linear regression models. Odds ratios and their corresponding 95% CIs were used to present results from the logistic models while the relative risk ratios (95% CIs) were used to represent results from the multinomial logistic models. 7.3 Results Patients cohorts By December 2010, a total of 1,138 RA patients were registered with the BSRBR as receiving RTX therapy (Table 7.1). Of them, 766 patients had completed six months of follow-up. Six hundred and forty six patients had DAS28 data recorded on both the baseline and the six months follow-up forms. Table 7.1: Patients included in the analysis Patients Included All patients registered with the BSRBR 19,944 Patients with RA 18,194 Patients registered as RTX patients 1,138 Patient who completed at least six months of follow-up 766 Patients with a recorded DAS28 / HAQ scores at treatment start and six months after 646/

179 7.3.2 Assessment of clinical effectiveness A total of 646 patients had a recorded DAS28 score at baseline and six months follow-up and were included in this assessment Baseline characteristics The patients had a mean (SD) age of 58.7 (11.6) years and had experienced long disease duration with a mean (SD) of 14.3 (10.1) years. About 70% of the patients were females. Twenty one percent of the patients were current smokers and seventy one percent of them had additional comorbidities (Table 7.2). The disease activity was high with a mean (SD) DAS28 score of 6.2 (1.1) and the mean (SD) HAQ score of 2.0 (0.6) Overall clinical effectiveness to RTX therapy The mean change in DAS28 scores over 6 months was -1.4 (95% CI -1.5: -1.3). In this cohort of patients, 17.2% of the patients achieved a good EULAR response, 43.2% achieved a moderate EULAR response while 39.6% of patients were classified as EULAR non-responders (Table 7.3). Disease remission was achieved in 8.1% of the patients. 179

180 Table 7.2: Baseline characteristics of RTX patients with baseline and six month DAS28 scores Characteristics All RTX patients Number of patients 646 Demographics Age (year) 58.7 ± 11.6 Female, n (%) 497 (76.9) Current smoker, n (%) 136 (21.1) Co-morbidities, n (%) 460 (71.2) Disease characteristics Disease duration(year) 14.3 ± 10.1 Rheumatoid factor positive, n (%) 424 (65.6) Swollen joint count (no.) 8.9 ± 5.8 Tender joint count (no.) 14.3 ± 7.8 Global health visual analogue scale 70.0 ± 20.2 C-reactive protein (mg/l) 34.2 ± 38.0 Erythrocyte sedimentation rate (mm/hr) 44.6 ± 29.9 Disease activity score (DAS28) 6.2 ± 1.1 Quality of life and physical function EQ-5D visual analogue scale 45.9 ± 22.6 EQ-5D utility score 0.3 ± 0.3 HAQ score 2.0 ± 0.6 Current and previous drug history Number of previous nbdmards 4.2 ± 1.8 Current nbdmards, n (%) None MTX Other 143 (22.1) 343 (53.1) 160 (24.8) Anti-TNF naïve, n (%) 106 (16.4) Previous steroids, n (%) 485 (75.1) Current steroids, n (%) 300 (46.4) Values are mean (SD) otherwise specified. TNF: tumour necrosis factors, RTX: rituximab, MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, and EQ-5D: European Quality of life 5 Dimensions, HAQ: health assessment questionnaire. 180

181 Table 7.3: Clinical effectiveness of rituximab six months after starting the therapy Outcome Entire cohort Number of patients 646 Mean change in DAS28 (95% CI) -1.4 (-1.5: -1.3) EULAR response, n (%) Good 111 (17.2) Moderate 279 (43.2) None 256 (39.6) Achieving disease remission, n (%) Yes 52 (8.1) No 594 (91.9) EULAR: European League Against Rheumatism Comparing clinical effectiveness of RTX according to previous anti-tnf history Within the 646 patients included in this assessment, 540 patients (83.6%) had previously failed at least one anti-tnf therapy while 106 patients (16.4%) were anti-tnf therapy naïve. Within the patients who were anti-tnf failures, 310 patients (57.4%) had previously failed only one anti-tnf therapy while 182 (33.7%) had previously failed two anti-tnf therapies. Twenty one patients (3.9%) had failed all three anti-tnf therapies (Table 7.4). The most recently stopped anti- TNF therapy was ETN, ADA and INF in 43.0%, 38.0%, and 14.4% of the patients respectively. Inefficacy of the last anti-tnf was the reason for discontinuation in 43.2% of the patients. 181

182 Significant differences in baseline characteristics were found between patients who were anti-tnf naive and those who were anti-tnf failures. Anti-TNF failures patients were generally younger, but with longer disease duration, more females, had tried more previous nbdmards, had less co - morbidity and were receiving RTX more commonly in combination with MTX (Table 7.5). Table 7.4: Previous anti-tnf history of the anti-tnf failure patients Characteristic Anti-TNF failure patients Number of patients 540 Ever had ETN, n (%) 238 (44.1) Ever had INF, n (%) 138 (25.6) Ever had ADA, n (%) 218 (40.4) Number of previous anti-tnf therapies Only one anti-tnf therapy, n (%) 310 (57.4) Two anti-tnf therapies, n (%) 182 (33.7) Three anti-tnf therapies, n (%) 21 (3.9) Number not recorded, n (%) 27 (5.0) Most recently discontinued anti-tnf therapy ETN, n (%) 232 (43.0) INF, n (%) 78 (14.4) ADA, n (%) 205 (38.0) Not recorded, n (%) 25 (04.6) Reason for discontinuing last biologic therapy Inefficacy, n (%) 233 (43.1) Adverse events, n (%) 117 (21.7) Other reason, n (%) 42 (7.8) Not recorded, n (%) 148 (27.4) TNF: tumour necrosis factors, ETN: etanercept, INF: infliximab, ADA: adalimumab 182

183 Table 7.5: Baseline characteristics of RTX patients according to previous anti-tnf history Characteristics Anti-TNF failure Anti-TNF naïve patients patients Number of patients Demographics 183 P value* Age (year) 58.1 ± ± Female, n (%) 427 (79.1) 70 (66.0) 0.01 Current smoker, n (%) 114 (21.1) 22 (20.8) 0.93 Co-morbidities, n (%) 376 (69.6) 84 (79.3) 0.05 Disease characteristics Disease duration(year) 14.8 ± ± Rheumatoid factor positive, n (%) 346 (64.1) 78 (73.6) 0.15 Swollen joint count (no.) 8.9 ± ± Tender joint count (no.) 14.3 ± ± Global health visual analogue scale 70.1 ± ± C-reactive protein (mg/l) 33.7 ± ± ESR (mm/hr) 44.4 ± ± Disease activity score (DAS28) 6.2 ± ± Quality of life and physical function EQ-5D visual analogue scale 45.8 ± ± EQ-5D utility score 0.3 ± ± HAQ score 2.0 ± ± Current and previous drug history Number of previous nbdmards 4.3 ± ± Current nbdmards, n (%) None MTX Other 108 (20.0) 316 (58.5) 116 (21.5) 35 (33.0) 27 (25.5) 44 (41.5) <0.001 Previous steroids, n (%) 407 (75.4) 78 (73.6) 0.70 Current steroids, n (%) 250 (46.3) 50 (47.2) 0.87 Values are mean (SD) otherwise specified. TNF: tumour necrosis factors, RTX: rituximab, MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, and EQ-5D: European Quality of life 5 Dimensions, HAQ: health assessment questionnaire, ESR: Erythrocyte sedimentation rate. *P-value tests for significant differences between the two groups of patients.

184 Anti-TNF naïve patients showed significantly better crude improvements in DAS28 score s (mean of (95% CIs -1.98: -1.44)) compared to patients who were anti-tnf failures (mean of (95% CIs -1.49: -1.23)) (Table 7.6). Better EULAR response rates were also observed with patients who were anti-tnf naïve compared to those who were anti-tnf failures, however it did not reach statistical significance (p=0.06). Anti- TNF naïve patients were also more likely to achieve disease remission (14.2%) compared to those who were anti-tnf failures (6.9%) (p=0.01). Table 7.6: Clinical effectiveness of rituximab according to previous anti-tnf history Outcome Anti-TNF failures Anti-TNF naïve P value* Number of patients Mean change in DAS28 (95% CI) (-1.49: -1.23) (-1.98: -1.44) 0.03 EULAR response, n (%) Good 87 (16.1) 24 (22.6) Moderate 229 (42.4) 50 (47.2) 0.06 None 224 (41.5) 32 (30.2) Achieving disease remission, n (%) Yes 37 (6.9) 15 (14.1) No 503 (93.2) 91 (85.9) 0.01 TNF: tumour necrosis factors, and EULAR: European League Against Rheumatism. * Test of significance between the two groups of patients. 184

185 Comparing clinical effectiveness of RTX according to RF status Within the 646 patients included in this assessment, 424 patients (65.6%) were RF positive while 203 patients (31.4%) were RF negative. RF status was not recorded for 19 patients. At baseline, there were no significant differences in baseline characteristics between RF positive and negative patients. Six months after receiving RTX, RF positive patients showed slightly better improvements in DAS28 scores compared to RF negative patients, however the difference did not reach statistically significant (p=0.09) (Table 7.7). RF positive and RF negative patients showed similar EULAR response rates (p=0.20) and similar proportions also achieved disease remission (p=0.38). Table 7.7: Clinical effectiveness of rituximab according to rheumatoid factor status Outcome RF positive RF negative P value* Number of patients, (%) 424 (65.63) 203 (31.42) - Mean change in DAS28 (95% CI) (-1.63: -1.34) (-1.46: -1.07) 0.09 EULAR response, n (%) Good 77 (18.2) 28 (13.8) Moderate 187 (44.1) 85 (41.9) 0.20 None 160 (37.7) 90 (44.3) Achieving disease remission, n (%) Yes 38 (9.0) 14 (6.9) No 386 (91.0) 189 (93.1) 0.38 EULAR: European League Against Rheumatism, RF: rheumatoid factor. * Test of significance between the two groups of patients. 185

186 Comparing clinical effectiveness of RTX according to baseline concomitant nbdmards Within the 646 patients included in this assessment, 343 patients (53.1%) were receiving RTX in combination with MTX, 143 patients (22.1%) were receiving RTX as mono-therapy, and 160 patients (24.8%) were receiving RTX in combination with other nbdmards (other than MTX). The three subgroups of patients showed similar improvements in DAS28 scores (p=0.86), similar EULAR response rates (p=0.82), and similar proportions of them achieved disease remission (p=0.37) (Table 7.8). Table 7.8: Response to rituximab according to the baseline concurrent nbdmards Outcome RTX+MTX RTX monotherapy RTX+ Other nbdmards P value* Number of patients Mean change in DAS28 (95% CI) (-1.59: -1.28) (-1.69: -1.18) (-1.60: -1.12) 0.86 EULAR response, n (%) Good 62 (18.1) 20 (14.0) 29 (18.1) Moderate 148 (43.1) 65 (45.4) 66 (41.3) 0.82 None 133 (38.8) 58 (40.6) 65 (40.6) Achieving disease remission, n (%) Yes 24 (7.0) 11 (7.7) 17 (10.6) No 319 (93.0) 132 (92.3) 143 (89.4) 0.37 TNF: tumour necrosis factors, RTX: rituximab, MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, and EULAR: European League Against Rheumatism.* Test of significance across the three groups of patients. 186

187 Predictors of clinical effectiveness of RTX In the univariate analyses, higher baseline DAS28 was significantly associated with a linear reduction in disease activity (Table 7.9). Higher baseline DAS28 was also significantly associated with achieving a moderate EULAR response (Table 7.10). RF positive patients were more likely to achieve a good EULAR response while females and patients with higher baseline HAQ were significantly less likely to achieve a good EULAR response. Older patients, patients with higher baseline HAQ and those with high baseline DAS28 were less likely to achieve disease remission (Table 7.11). In the multivariate linear regression analysis, higher baseline DAS28 scores and RF positive status were significantly associated with a decrease in disease activity while higher baseline HAQ and female gender were associated with an increase in disease activity (a poorer response to treatment). For each one unit increase in baseline DAS28, there was a 0.57 (95% CI 0.69: 0.46) unit decrease in the disease activity while for each one unit increase in baseline HAQ, there was 0.29 (95% CI 0.06: 0.52) unit increase in disease activity (Table 7.9). RF positive status was significantly associated with a decrease in disease activity by 0.30 (95% CI 0.58: 0.03) units. Similar results were found in the multivariate multinomial regression analysis examining factors associated with EULAR response; in which higher baseline DAS28 score was significantly associated with achieving a moderate EULAR response (relative risk ratio (95% CI) 1.62 (1.34: 1.97)). Higher baseline HAQ score was significantly associated with not achieving a good EULAR response with relative risk ratio (95% CI) of 0.92 (0.88: 0.97). Neither RF positive status nor female gender reached statistical significance (Table 7.10). 187

188 Table 7.9: Beta-coefficients (95% CI) for linear regression models of change in disease activity (change in DAS28) Variables Univariate model Multivariate model Age (10 years) 0.01 (-0.10: 0.12) - Female gender 0.25 (-0.06: 0.57) 0.45 (0.12: 0.78) * Having at least one co-morbidity 0.22 (-0.06: 0.50) - Disease duration (10 year) (-0.14: 0.12) - Current smoking 0.24 (-0.08: 0.55) - Rheumatoid factor positive (-0.54: 0.01) (-0.58: -0.03) * Baseline DAS (-0.61: -0.42) * (-0.69: -0.46) * Baseline HAQ (-0.26: 0.18) 0.29 (0.06: 0.52) * Previous steroids (-0.36: 0.24) - On steroids at baseline 0.20 (-0.06: 0.47) - 4 or more previous nbdmards (-0.32: 0.21) - Concurrent use of nbdmards: MTX is reference No nbdmards Other nbdmards 0.04 (-0.29: 0.37) 0.17 (-0.16: 0.49) - Number of previous anti-tnfs: one anti-tnf is reference More than one previous anti-tnf (-0.35: 0.18) - Reason for stopping of the last biologic therapy: Inefficacy is reference Other reasons 0.11 (-0.15: 0.37) - TNF: tumour necrosis factors, MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, DAS: disease activity score, and HAQ: health assessment questionnaire. *P < Other reasons include adverse events, remission, other or missing reasons. Comorbidities include: angina, hypertension, stroke, epilepsy, myocardial infarction, asthma, chronic obstructive pulmonary disease, liver disease, renal disorder, peptic ulcers, diabetes, demyelination, hyperthyroidism, depression or a history of tuberculosis or cancer. 188

189 Table 7.10: Relative risk ratios (95% CI) for multinomial regression analysis of EULAR response Univariate models Multivariate model Variables Moderate Good Moderate Good Age (10 years) 1.06 (0.90: 1.25) Female gender 0.82 (0.51: 1.31) Having at least one comorbidity 0.68 (0.45: 1.02) 0.85 (0.69: 1.04) 0.54 (0.30: 0.96)* 0.84 (0.49: 1.45) (0.41: 1.23) 0.54 (0.28: 1.03) - - Disease duration (10 year) 0.98 (0.81: 1.18) Current smoking 0.63 (0.40: 1.00) Rheumatoid factor positive 1.11(0.75: 1.65) Baseline DAS (1.34: 1.89)* Baseline HAQ (0.1 unit) 1.03 (1.00: 1.07) Previous steroids 1.13 (0.74: 1.74) On steroids at baseline 0.78 (0.54: 1.12) 4 or more previous nbdmards 1.46 (1.00: 2.15) 1.09 (0.85: 1.40) 0.78 (0.43: 1.43) 1.89 (1.06: 3.39)* 0.92 (0.75: 1.12) 0.93 (0.90: 0.97)* 1.10 (0.62: 1.95) 0.78 (0.47: 1.28) 1.15 (0.69: 1.92) (0.74: 1.75) 1.62 (1.34: 1.97)* 1.01 (0.97: 1.05) 1.69 (0.92: 3.11) 1.13 (0.89: 1.43) 0.92 (0.88: 0.97)* Concurrent use of nbdmards: MTX is reference No nbdmards Other nbdmards 1.00 (0.62: 1.60) 0.81 (0.51: 1.29) 0.76 (0.39: 1.48) 0.73 (0.39: 1.38) Number of previous anti-tnfs: one anti-tnf is reference More than one previous anti-tnf 1.40 (0.95: 2.06) 0.76 (0.44: 1.30) Reason for stopping of the last biologic therapy: Inefficacy is reference Other reasons 1.16 (0.80: 1.68) 0.99 (0.60: 1.63) TNF: tumour necrosis factors, MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, DAS: disease activity score, HAQ: health assessment questionnaire and EULAR: European League Against Rheumatism. *P < Other reasons include adverse events, remission, other or missing reasons. Comorbidities include: angina, hypertension, stroke, epilepsy, myocardial infarction, asthma, chronic obstructive pulmonary disease, liver disease, renal disorder, peptic ulcer, diabetes, demyelination, hyperthyroidism, depression or a history of tuberculosis or cancer. 189

190 Table 7.11: Odds ratios (95% CI) of logistic regression analysis of achieving disease remission Variables Univariate models Multivariate model Age (10 years) 0.65 (0.49: 0.85) * 0.72 * (0.53: 0.97) Female gender 0.60 (0.29: 1.26) - Having at least one co-morbidity 0.79 (0.39: 1.60) - Disease duration (10 year) 1.01 (0.73: 1.42) - Current smoking 0.71(0.29: 1.74) - Rheumatoid factor positive 1.91 (0.85: 4.27) - Baseline DAS (0.48: 0.81) * - Baseline HAQ (0.1 unit) 0.88 (0.83: 0.92) * 0.87 * (0.83: 0.92) Previous steroids 1.43 (0.61: 3.34) - On steroids at baseline 1.24 (0.64: 2.42) - 4 or more previous nbdmards 0.81 (0.41: 1.60) - Concurrent use of nbdmards: MTX is reference No nbdmards Other nbdmards 1.18 (0.51: 2.77) 1.24 (0.55: 2.82) - Number of previous anti-tnfs: one anti-tnf is reference More than one previous anti-tnf 0.92 (0.46: 1.84) - Reason for stopping of the last biologic therapy: Inefficacy is reference Other reasons 0.70 (0.36: 1.37) - TNF: tumour necrosis factors, MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, DAS: disease activity score, and HAQ: health assessment questionnaire. *P < Other reasons include adverse events, remission, other or missing reasons. Comorbidities include: angina, hypertension, stroke, epilepsy, myocardial infarction, asthma, chronic obstructive pulmonary disease, liver disease, renal disorder, peptic ulcer, diabetes, demyelination, hyperthyroidism, depression or a history of tuberculosis or cancer. 190

191 In the multivariate logistic regression model looking at predictors of disease remission (Table 7.11), higher baseline HAQ remained significantly associated with failure to achieve remission (OR (95% CI) 0.87 (0.83: 0.92)). In addition, older patients were less likely to achieve disease remission (OR (95% CI) 0.72 (0.53: 0.97)). All patients who had failed all three of the anti-tnf therapies did not achieve disease remission (n=21). After adjustment for propensity scores, no significant differences in the change in DAS28, EULAR response rates or proportions of patients achieving disease remission were found between patients who were anti-tnf naïve and those patients who had previously failed at least one anti - TNF therapy (Table 7.12). The logistic model for propensity score included RF status, baseline DAS28, baseline HAQ score, age, gender, disease duration, co-morbidities, number of previous nbdmards, concurrent nbdmards, and interactions between baseline HAQ and gender, and between baseline HAQ and disease duration. 191

192 Table 7.12: Undjusted and adjusted regression models of six months response in anti-tnf failures versus anti-tnf naïve patients Variable Change in DAS28 EULAR response Achieving disease remission Anti-TNF failures Anti-TNF naïve patients Linear regression Multinomial regression Logistic regression Coefficient (95% CI) Odds ratios (95% CI) Odds ratios (95% CI) Unadjusted Adjusted Unadjusted Adjusted Unadjusted Adjusted * (-0.67: -0.04) (-0.54: 0.13) Moderate Good Moderate Good 1.53 (0.95: 2.47) Reference 1.93 * (1.08: 3.46) 1.84 (0.98: 3.47) 1.66 (0.77: 3.56) 2.24 * (1.18: 4.25) 1.35 (0.60: 3.06) TNF: tumour necrosis factors, DAS: disease activity score, EULAR: European League Against Rheumatism.* P < adjusted for propensity score that included age, disease duration, gender, co-morbidities, number of previous nbdmards, concurrent nbdmards, RF status, baseline DAS28, baseline HAQ score, interaction between baseline HAQ and disease duration, and interaction between baseline HAQ and gender. 192

193 7.3.3 Assessment of patient reported physical function and quality of life after RTX therapy A total of 448 patients had a recorded HAQ score at baseline and at six months follow-up and were included in this assessment Baseline characteristics The patients had a mean (SD) age of 59.8 (11.4) years and had experienced mean disease duration of 14.8 years. Seventy eight percent of the patients were females. Eighteen percent of the patients were current smokers and seventy percent of them had additional co-morbidities (Table 7.13). Disease activity scores were high with a mean (SD) DAS28 score of 6.1 (1.2) and the mean (SD) HAQ score of 1.96 (0.59). The patients also had baseline mean (SD) utility score of 0.29 (0.32) and baseline mean (SD) EQ-5D VAS of 45.6 (22.2) Overall patient reported physical function and quality of life at six months The mean (95% CI) HAQ score at baseline was 1.96 (1.91: 2.02) which significantly improved to 1.84 (1.78: 1.90) six months after receiving RTX (p<0.0001, paired t-test), this represented difference of (-0.16: -0.09). The MCID in HAQ was achieved by 36.6% of the patients (Table 7.14). Likewise, the EQ-5D VAS improved significantly from 45.6 (43.5: 47.7) to 53.0 (51.1: 54.9) (p<0.0001, paired t- test). EQ-5D utility scores improved significantly from 0.29 (0.26: 0.32) to 0.39 (0.36: 0.42) (p<0.0001, paired t-test). Six months after receiving RTX, the proportion of patients rating themselves with a utility score that would be classified as representing a health state worse than death (as measured by a EQ-5D utility score less than zero), decreased significantly from 32.3% (145 patients) at baseline to 20.0% (90 patients). 193

194 Table 7.13: Baseline characteristics of the patients Patient characteristics 194 All RTX patients Number of patients 448 Age (year) 59.8 ± 11.4 Female gender, n (%) 351 (78.4) Current smoking, n (%) 81 (18.1) Having co-morbidities, n (%) 316 (70.5) Disease duration (year) 14.8 ± 10.4 RF positive, n (%) 302 (67.4) DAS ± 1.2 EQ-5D visual analogue scale 45.6 ± 22.2 EQ-5D utility score 0.29 ± 0.32 HAQ score 1.96 ± 0.59 Number of previous nbdmards 4.2 ± 1.8 Current nbdmards, n (%) None MTX Other 108 (24.1) 221 (49.3) 119 (26.6) Previous steroids, n (%) 344 (76.8) Current steroids, n (%) 204 (45.5) Number of previous anti-tnf therapies Only one anti-tnf therapy,, n (%) Two anti-tnf therapies, n (%) Three anti-tnf therapies, n (%) Number is not recorded, n (%) Last discontinued anti-tnf therapy ETN, n (%) INF, n (%) ADA, n (%) Not recorded, n (%) Reason for stopping last anti-tnf therapy Inefficacy, n (%) Adverse events, n (%) Other, n (%) Not recorded, n (%) 207 (46.3) 116 (25.8) 20 (4.45) 21 (4.7) 151 (33.7) 60 (13.4) 134 (29.9) 19 (4.2) 148 (33.0) 92 (20.5) 29 (6.5) 95 (21.2) Values are mean (SD) otherwise specified. TNF: tumour necrosis factors, RTX: rituximab, MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, RF: rheumatoid factor, DAS28: disease activity score 28 joints, EQ-5D: European Quality of life 5 Dimensions, HAQ: health assessment questionnaire, ETN: etanercept, INF: infiximab, and ADA: adalimumab.

195 Table 7.14: Improvements in quality of life and physical function six months after starting RTX Outcome RTX patients Number of patients 448 Six month outcomes HAQ score (mean (95% CI)) 1.84 (1.78: 1.90) * Patients achieving MCID in HAQ, n (%) 164 (36.6) EQ VAS (mean (95% CI)) (51.12: 54.88) * EQ-5D utility score (mean (95% CI)) 0.39 (0.36: 0.42) * Change from baseline Change in HAQ (mean (95% CI)) (-0.16: -0.09) Change in EQ VAS (mean (95% CI)) 7.03 (4.61: 9.46) Change in EQ-5D utility score (mean (95% CI)) 0.10 (0.07: 0.13) * P < 0.05 (compared to baseline, paired t-test). EQ-5D: European Quality of life 5 Dimensions, RTX: rituximab, VAS: visual analogue score, MCID: minimum clinically important difference and HAQ: health assessment questionnaire. The five domains of the EQ-5D for RTX patients at baseline and six months after receiving the therapy are shown in Table Six months after receiving RTX, significant improvements were seen in self-care, activities and pain as lower proportions of patients reported any problems (some or extreme problems) in those three domains compared to baseline (Figure 7.1). The proportion of patients who reported extreme problems in pain decreased from 44.8% (196 patients) to 26.7% (119 patients) six months after receiving RTX treatment (p<0.001). 195

196 Table 7.15: Changes in the five domains of EQ-5D profile six months after starting RTX Domains EQ-5D profile, n (%) Extreme problems Some problems No problems At baseline Mobility 4 (0.91) 398 (90.87) 36 (8.22) Self-care 22 ( 5.02) 350 (79.91) 66 (15.07) Activities 79 (18.04) 347 (79.22) 12 (2.74) Pain 196 (44.75) 235 (53.65) 7 (1.60) Anxiety 24 (5.53) 229 (52.76) 181 (41.71) At six months Mobility 1 (0.22) 399 (89.46) 46 (10.31) Self-care 13 (2.95) 338 (76.82) 89 (20.23) Activities 65 (14.57) 344 (77.13) 37 (8.30) Pain 119 (26.74) 308 (69.21) 18 (4.04) Anxiety 29 (6.55) 223 (50.34) 190 (43.12) EQ-5D: European Quality of life 5 Dimensions. 196

197 Proportions of patients reporting any problems 100 * * 80 * Mobility Self care Activities Pain Anxiety Baseline months Figure 7.1: EQ-5D profile at baseline and six months after starting RTX. * P value < 0.05 (compared to baseline, McNemar's chi-square test) 197

198 Comparing patient reported physical function and quality of life according to previous anti- TNF history Of the 448 patients included in the assessment of patient reported outcomes after receiving RTX, 364 patients (81.2%) had previously failed at least one anti-tnf therapy and 84 patients (18.8%) were naïve to anti-tnf therapies. Within the patients who were anti-tnf failures, 207 patients (56.9%) had previously failed only one anti-tnf therapy while 116 (31.9%) had previously failed two anti-tnf therapies. Twenty patients (5.5%) had failed all three anti-tnf therapies (Table 7.16). The most recently stopped anti-tnf therapy was ETN, ADA and INF in 41.5%, 36.8%, and 16.5% of the patients respectively. Inefficacy of the last stopped anti-tnf was the reason for discontinuation in 40.6% of the patients and adverse events in 25.3% of the patients. Significant differences in baseline characteristics were found between patients who were anti -TNF naive and those who were anti-tnf failures. Anti-TNF failure patients were generally younger, but with longer disease duration, a greater proportion were female, had tried more previous nbdmards, had fewer co-morbidities and were receiving RTX more commonly in combination with MTX (Table 7.16). However the two groups of patients has similar scores for baseline disease activity, quality of life and physical function. The improvements in physical function and quality of life for the two groups of patients are shown in Table The two groups of patients showed similar improvements in HAQ, EQ-5D utility, and EQ- 5D VAS scores. The five domains of the EQ-5D for the two groups of patients at baseline and six months after receiving the therapy are shown in Table

199 Table 7.16: Baseline characteristics of the patients Patient characteristics Anti-TNF naïve Anti-TNF failures P value Number of patients Age (year) 63.0 ± ± Female gender, n (%) 57 (67.9) 294 (80.8) 0.01 Current smoker, n (%) 18 (21.4) 63 (17.3) 0.38 Co-morbidities, n (%) 68 (81.0) 248 (68.1) 0.02 Disease duration(year) 11.7 ± ± RF positive, n (%) 59 (70.2) 243 (66.8) 0.68 DAS28 score 6.1 ± ± EQ-5D VAS 45.6 ± ± EQ-5D utility score 0.30 ± ± HAQ score 1.87 ± ± Number of previous nbdmards 3.6 ± ± 1.8 <0.001 Current nbdmards, n (%) None MTX Other 28 (33.3) 17 (20.3) 39 (46.4) 80 (22.0) 204 (56.0) 80 (22.0) <0.001 Previous steroids, n (%) 65 (77.4) 279 (76.7) 0.88 Current steroids, n (%) 37 (44.1) 167 (45.9) 0.76 Number of previous anti-tnf therapies Only one anti-tnf therapy, n (%) Two anti-tnf therapies, n (%) Three anti-tnf therapies, n (%) Number not recorded, n (%) Last discontinued anti-tnf therapy ETN, n (%) INF, n (%) ADA, n (%) Not recorded, n (%) Reason for stopping last anti-tnf therapy Inefficacy, n (%) Adverse events, n (%) Other reason, n (%) Not recorded, n (%) (56.9) 116 (31.9) 20 (5.5) 21 (5.7) 151 (41.5) 60 (16.5) 134 (36.8) 19 (5.2) 148 (40.6) 92 (25.3) 29 (8.0) 95 (26.1) Values are mean ± SD otherwise specified. TNF: tumour necrosis factors, RTX: rituximab, MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, RF: rheumatoid factor, DAS28: disease activity score-28 joints. EQ-5D: European Quality of life 5 Dimensions, HAQ: health assessment questionnaire, ETN: etanercept, INF: infiximab, and ADA: adalimumab

200 Table 7.17: Improvements in quality of life and physical function according to previous anti -TNF history Outcome Anti-TNF failures Anti-TNF naïve P value Number of patients Six months outcomes HAQ (mean (95% CI)) 1.87 (1.80: 1.94) 1.71 (1.57: 1.85) 0.04 Patients achieving MCID in HAQ, n (%) 130 (35.71) 34 (40.48) 0.41 EQ VAS (mean (95% CI)) (50.50: 54.70) (50.62: 59.03) 0.38 EQ-5D utility score (mean (95% CI)) 0.38 (0.35: 0.41) 0.45 (0.38: 0.52) 0.07 Change from baseline Change in HAQ (mean (95% CI)) (-0.15: -0.07) (-0.24: -0.09) 0.27 Change in EQ VAS (mean (95% CI)) 6.64 (3.90: 9.38) 8.84 (3.71: 13.98) 0.49 Change in EQ-5D utility score (mean (95% CI)) 0.10 (0.06: 0.13) 0.14 (0.07: 0.21) 0.35 TNF: tumour necrosis factors, EQ-5D: European Quality of life 5 Dimensions, VAS: visual analogue score, MCID: minimum clinically important difference, and HAQ: health assessment questionnaire. 200

201 Table 7.18: Changes in the five domains of EQ-5D profile according to previous anti-tnf history Domains EQ-5D profile, n (%) At baseline Extreme problems Anti-TNF failures (n=364) Anti-TNF naïve (84) Some problems No problems Extreme problems Some problems No problems Mobility 4 (1.12) 324 (90.76) 29 (8.12) 0 74 (91.36) 7 (8.64) Self-care 20 (5.60) 286 (80.11) 51 (14.29) 2 ( 2.47) 64 (79.01) 15 (18.52) Activities 66 (18.49) 282 (78.99) 9 (2.52) 13 (16.05) 65 (80.25) 3 (3.70) Pain 160 (44.82) 191 (53.50) 6 (1.68) 36 (44.44) 44 (54.32) 1 (1.23) Anxiety 21 (5.93) 188 (53.11) 145 (40.96) 3 ( 3.75) 41 (51.25) 36 (45.00) At six months Mobility 1 (0.28) 327 (90.08) 35 (9.64) 0 72 (86.75) 11 (13.25) Self-care 11 (3.06) 286 (79.67) 62 (17.27) 2 ( 2.47) 52 (64.20) 27 (33.33) Activities 56 (15.43) 279 (76.86) 28 (7.71) 9 (10.84) 65 (78.31) 9 (10.84) Pain 98 (27.07) 249 (68.78) 15 (4.14) 21 (25.30) 59 (71.08) 3 (3.61) Anxiety 28 (7.73) 183 (50.55) 151 (41.71) 1 (1.23) 40 (49.38) 40 (49.38) EQ-5D: European Quality of life 5 Dimensions, and TNF: tumour necrosis factors. 201

202 Comparing patient reported physical function and quality of life according to RF status Within the 448 patients included in this assessment, 302 patients (67.41%) were RF positive while 140 patients (31.25%) were RF negative. RF status was not recorded for six patients. Six months after receiving RTX, RF positive and negative patients showed similar improvements in HAQ scores, EQ-5D utility scores, and EQ-5D VAS (Table 7.19). Similar proportions of the two groups of patients achieved a MCID in HAQ (p=0.78). Table 7.19: Improvements in quality of life and physical function according to RF status Outcome RF positive RF negative P value Number of patients Six months outcomes Mean HAQ (mean (95% CI)) 1.83 (1.76: 1.91) 1.84 (1.74: 1.95) 0.87 Patients achieving MCID in HAQ, n (%) 108 (35.76) 52 (37.14) 0.78 EQ VAS (mean (95% CI)) (50.69: 55.27) (49.66: 56.56) 0.95 EQ-5D utility score (mean (95% CI)) 0.40 (0.37: 0.44) 0.38 (0.33: 0.44) 0.54 Change from baseline Change in HAQ (mean (95% CI)) (-0.16: -0.08) (-0.18: -0.06) 0.96 Change in EQ VAS (mean (95% CI)) 7.29 (4.40: 10.18) 6.66 (2.01: 11.32) 0.82 Change in EQ-5D utility score (mean (95% CI)) 0.10 (0.07: 0.14) 0.11 (0.05: 0.17) 0.93 RF rheumatoid factor, EQ-5D: European Quality of life 5 Dimensions, VAS: visual analogue score, MCID: minimum clinically important difference and HAQ: health assessment questionnaire. 202

203 Comparing patient reported physical function and quality of life according to baseline concomitant nbdmards Within the 448 patients included in this assessment, 221 patients (49.3%) were receiving RTX in combination with MTX, 108 patients (24.1%) were receiving RTX as mono-therapy, and 119 patients (26.6%) were receiving RTX in combination with other nbdmards (other than MTX). The three subgroups of patients showed similar improvements in HAQ scores (p=0.89), similar improvements in EQ-5D utility score (p=0.45), and similar proportions of patients achieved a MCID in HAQ (p=0.18) (Table 7.20). Table 7.20: Improvements in quality of life and physical function according to baseline concomitant nbdmards Outcome RTX+ Other P RTX+MTX RTX mono-therapy nbdmards value Number of patients Six months outcomes HAQ (mean (95% CI)) 1.84 (1.75: 1.92) 1.95 (1.83: 2.07) 1.74 (1.62: 1.86) 0.05 Patients achieving MCID in HAQ, n (%) EQ VAS (mean (95% CI)) EQ-5D utility score (mean (95% CI)) Change from baseline Change in HAQ (mean (95% CI)) Change in EQ VAS (mean (95% CI)) 79 (35.75) 47 (43.52) 38 (31.93) (52.05: 57.49) (47.29: 54.76) (47.93: 55.30) (0.36: 0.44) 0.34 (0.28: 0.40) 0.42 (0.37: 0.48) (-0.17: -0.07) 8.12 (4.68: 11.55) (-0.21: -0.07) 6.61 (1.35: ) (-0.19: -0.05) 5.49 (3.88: 10.09) Change in EQ-5D utility score (mean (95% CI)) 0.09 (0.05: 0.14) 0.14 (0.07: 0.21) 0.09 (0.03: 0.15) 0.45 MTX: methotrexate, nbdmards: non-biologic disease modifying anti-rheumatic drugs, EQ-5D: European Quality of life 5 Dimensions, VAS: visual analogue score, MCID: minimum clinically important difference and HAQ: health assessment questionnaire

204 Predictors of patient reported physical function in anti-tnf failure patients In the univariate linear regression analysis examining possible predictors of improvements in HAQ score, higher baseline HAQ score and higher baseline DAS28 were significantly associated with improvements in six months HAQ score. An improvement of (95% CI -0.16: -0.03) units in HAQ was seen for each one unit increase in baseline HAQ. Likewise, an improvement of 0.03 (95% CI : ) units in HAQ was seen for each one unit increase in baseline DAS28 score. In the forward stepwise multivariate linear regression analysis examining possible predictors of improvements in HAQ score, older age, female gender, current smoking and concurrently receiving steroid therapy were significantly associated with lower HAQ scores (Table 7.21). In contrast, high baseline HAQ scores were significantly associated with linear improvements in six months HAQ scores. An improvement of (95% CI -0.20: -0.07) units in HAQ was seen for each one unit increase in baseline HAQ. In both the univariate and the forward stepwise multivariate logistic model examining possible predictors of achieving a MCID in HAQ (Table 7.22), current smokers were significantly less likely to achieve a MCID in HAQ with an odds ratio of 0.46 (95% CI) (0.24: 0.9). 204

205 Table 7.21: Univariate and multivariate regression models of predictors of physical function at six months (change in HAQ) Variables Linear change in HAQ Linear regression (Coefficient (95% CI)) Univariate analysis Forward stepwise multivariate analysis Age (10 years) 0.03 (-0.01: 0.06) 0.06 (0.02: 0.09)* Female gender 0.07 (-0.03: 0.17) 0.14 (0.04: 0.24)* Disease duration (10 year) 0.02 (-0.02: 0.06) - Co-morbidities (-0.14: 0.02) - Current smoking 0.09 (-0.02: 0.19) 0.12 (0.02: 0.22)* RF positive status 0.01 (-0.09: 0.09) - Baseline DAS (-0.06: )* - Baseline HAQ score (-0.16: -0.03)* (-0.20: -0.07)* Baseline EQ-5D VAS 0.00 (0.00: 0.00) - Baseline EQ-5D utility score 0.11 (-0.01: 0.23) - Had previous steroids 0.04 (-0.05: 0.13) - Concurrent steroids at baseline 0.06 (-0.02: 0.14) 0.08 (0.01: 0.16)* Number of previous nbdmards (had 4 or more) Concurrent nbdmards: MTX is reference 0.02 (-0.07: 0.10) - No nbdmards (-0.13: 0.07) - Other nbdmards (-0.10: 0.10) - Reason for stopping last anti-tnf: Inefficacy is the reference Other reasons (-0.09: 0.07) - Number of failed anti-tnfs: One anti-tnf is reference More than one anti-tnf 0.04 (-0.04: 0.12) - TNF: tumour necrosis factors, nbdmards: non-biologic disease modifying anti-rheumatic drugs, MTX: methotrexate, DAS disease activity score, EQ-5D: European Quality of life 5 Dimensions, VAS: visual analogue score, MCID: minimum clinically important difference and HAQ: health assessment questionnaire. *P < Other reasons include adverse events, remission, and other or missing reasons. 205

206 Table 7.22: Univariate and Multivariate regression models of predictors of six months physical function (Achieving MCID in HAQ) Variables Achieving MCID in HAQ Logistic regression (Odds ratio (95% CI)) Univariate analysis Forward stepwise multivariate analysis Age (10 years) 0.93 (0.77: 1.13) - Female gender 0.61 (0.36: 1.03) - Disease duration (10 year) 0.87 (0.70: 1.08) - Co-morbidities 1.61 (1.00: 2.59) - Current smoking 0.46 (0.24: 0.87)* 0.46 (0.24: 0.88)* RF positive status 0.83 (0.53: 1.32) - Baseline DAS (0.96: 1.36) - Baseline HAQ score 1.39 (0.95: 2.02) - Baseline EQ-5D VAS 0.99 (0.98: 1.00) - Baseline EQ-5D utility score 0.71 (0.36: 1.39) - Had previous steroids 0.73 (0.45: 1.21) - Concurrent steroids at baseline 0.76 (0.49: 1.19) - Number of previous nbdmards (had 4 or more) Concurrent nbdmards: MTX is reference 1.05 (0.67: 1.65) - No nbdmards 1.60 (0.94: 2.71) - Other nbdmards 0.87 (0.50: 1.52) - Reason for stopping last anti-tnf: Inefficacy is the reference Other reasons 0.94 (0.61: 1.45) - Number of failed anti-tnfs: One anti-tnf is reference More than one anti-tnf 0.79 (0.50: 1.24) - TNF: tumour necrosis factors, nbdmards: non-biologic disease modifying anti-rheumatic drugs, MTX: methotrexate, DAS disease activity score, EQ-5D: European Quality of life 5 Dimensions, VAS: visual analogue score, MCID: minimum clinically important difference and HAQ: health assessment questionnaire. *P < Other reasons include adverse events, remission, and other or missing reasons. 206

207 Patients who had previously failed at least one anti-tnf therapy and those who were anti-tnf naïve showed similar changes in HAQ scores, and similar proportions of patients achieved a MCID in HAQ in both unadjusted and adjusted models using propensity scores (Table 7.23). The propensity score logistic regression model included baseline HAQ score, age, gender, disease duration, co-morbidities, number of previous nbdmards, concurrent steroids, current smoking and an interaction between disease duration and gender. Table 7.23: Adjusted and unadjusted regression results for changes in physical function in anti - TNF failures versus anti-tnf naïve patients Variable Linear change in HAQ Linear regression Beta-coefficient (95% CI) Achieving a MCID in HAQ Logistic regression Odds ratio (95% CI) Unadjusted Adjusted * Unadjusted Adjusted * Anti-TNF failures Reference Anti-TNF naïve patients (-0.14: 0.04) (-0.18: 0.03) (0.76: 2.00) (0.68: 2.18) TNF: tumour necrosis factors, HAQ: health assessment questionnaire, MCID: minimal clinically important difference. * Adjusted for propensity score that included baseline HAQ score, age, gender, disease duration, co-morbidities, number of previous nbdmards, concurrent steroids, current smoking and interaction between disease duration and gender. 207

208 7.3.4 Summary In this cohort of RA patients who were receiving RTX in routine clinical practice in the UK, the following key results were found. After six months from starting RTX therapy, RTX was found to be effective in improving both the clinical and patient reported outcomes. High baseline DAS28 score and RF positive status were identified as significant predictors of clinical effectiveness of RTX. High baseline HAQ score and female gender were identified as significant predictors of lower response to RTX six months after starting RTX. High baseline HAQ score was identified as a significant predictor of improvements in six months HAQ scores. Older age, female gender, being a current smoker and concurrently receiving steroid therapy were identified as predictors of lower improvements in HAQ scores. Being anti-tnf naïve was not found to predict the response to RTX. 7.4 Discussion The response to RTX in the anti-tnf failures was comparable to the results of a small observational study (n=20) which reported six month EULAR response of 60% (123) compared to 58.5% in the results presented in this thesis. However, the response in the REFLEX trial was slightly higher (65%). (110) The higher mean ± SD baseline disease activity of patients in the REFLEX trial (6.9 ± 1.0 versus 6.2 ± 1.2 in our cohort) may explain the higher response rates. In general comparing the response of the current analysis with the response reported in clinical trials might be inappropriate due to the different populations include d in those trials with different baseline characteristics especially for those characteristics that have been found to predict response. 208

209 After adjustment for propensity scores, the response to RTX in anti -TNF failures versus naïve patients was similar. The unadjusted differences in response in anti-tnf naïve patients may possibly be explained by the fact that the anti-tnf failures had longer disease duration and had failed a larger number of previous nbdmards suggesting more resistant disease. To date, no RCTs have compared the response in anti-tnf failures versus anti-tnf naïve patients. However, one small observational study has previously compared the response of these two populations (n=20); suggesting no significant difference in response as measured by change in DAS28 (1.8 versus 1.4). However, these patients were receiving RTX as mono-therapy. (117) Another recent study of 40 anti-tnf naive patients and 63 anti-tnf failures showed similar responses in univariate analysis (p=0.118). (133) Similar responses were also reported in a study of 14 anti-tnf naive patients and 81 anti-tnf failures (p=1). (120) Patients receiving MTX with RTX showed a similar response to those receiving RTX as monotherapy or with other nbdmards in both the univariate and multivariate analyses. These findings come in agreement with the RCT of Edward et al. (2006). (108) Similar finding were reported in a retrospective observational study of 57 patients. (118) A very recent retrospective study of 95 patients also reported similar EULAR response in patients treated with RTX + MTX and those treated with RTX alone. (120) Given that sustained benefits over 2 years of follow-up were shown only in RTX+ MTX therapy rather than RTX mono-therapy, (109) further observational studies are needed to confirm if patients receiving RTX mono-therapy will experience sustainability of response after successive treatments. As previously discussed, RA patients who were receiving combination of two or three nbdmards with anti-tnf therapies were less likely to discontinue their anti-tnf therapy, due to inefficacy, compared to those who were receiving no concomitant nbdmards. In contrast, concomitant nbdmards did not seem to affect the initial response to RTX. Different mechanism of action of RTX compared to anti-tnf therapies may explain this finding. However, the six 209

210 months time follow-up may not be enough to reflect the impact of nbdmards on the response to RTX and again further study may confirm this impact after repeated treatment courses of RTX. RF positive status was found to be a significant predictor of the decrease in disease activity ( (95% CI -0.58: -0.03)) but this was not found to translate into neither a moderate nor a good EULAR response. This may be explained by the low coefficient of the decrease in DAS28 (0.30) (which is less than 0.6 which translates to non EULAR response) (see Chapter 1, Table 1.4). RTX acts by depleting B cells which are believed to be the source of RF, consequently there was a thought that RTX may not act in RF negative patients and this may be the reason why the very first RCT of RTX included only RF positive patients. (108) However, subsequent trials showed that RTX can act in both RF negative and positive patients (110;113) and this was the case in our cohort of patients. In this situation, it is worth mentioning that the mechanism of action of B cell depletion in RA remains not fully understood. In a previous study RF positivity was found to be a significant predictor of EULAR moderate/good response to RTX (OR (95% CI) of 7.5 (2.2: 25.4)). (135) RF positive patients also showed a superior EULAR response to RTX than RF negative patients in a very recent retrospective observational study. (120) The current results come in agreement with those studies however, RF positivity was only associated with a decrease in DAS28 but not achieving EULAR response. The presence of other auto-antibodies may be affecting the results of this analysis and lack of data on anti - citrullinated protein antibody (ACPA) may limit these findings as some of the RF negative patients may be ACPA positive and hence responding. ACPA has been reported to be a significant predictor of EULAR good response. (136) Higher baseline DAS28 was found to be a significant predictor of a reduction in DAS28 and achieving a moderate EULAR response in anti-tnf failures. Regression to the mean may have played a role in the linear reduction of DAS28 however its role may have been minimized in 210

211 achieving moderate EULAR response as it consider both the change in DAS28 and the final achieved DAS28. As was the case for anti-tnf therapies discontinuation, female gender was a significant predictor of lower response to RTX as measured by the increase in disease activity. Again, this may suggest that there may be genetic factors or certain hormones that may explain this finding. In this situation it should be mentioned that RA is more common in females and there is no clear understanding of the reasons behind that. Higher baseline HAQ score was a significant predictor of non-response to RTX as measured by both the decrease in disease activity and achieving good EULAR response. Previous results from the BSRBR identified high baseline HAQ as a significant predictor of EULAR non-response in RA patients treated with anti-tnf therapies, (147) suggesting that patients with high HAQ score are more likely to have irreversible disease. Lower baseline HAQ was found to be significantly associated with achieving ACR response 50 in 110 RA patients treated with RTX, however in the same study lower baseline HAQ was not associated with EULAR response criteria. (135) In addition the limited external validity in clinical trials in routine clinical practice may explain those differences. In the multivariate analysis for predictors of remission, higher baseline HAQ, older patients and those who had failed all three anti-tnf agents were less likely to achieve disease remission. The model did not identify any positive predictors of remission. Overall, the proportion of patients achieving remission was low (8.1%). In comparison with the published RCTs that looked at the patient-reported outcomes six months after starting RTX in RA patients, the improvements reported in those trials were higher than the results of the current analysis; the mean change in HAQ was -0.44, and in the REFLEX (111) and DANCER (113;114) trials respectively compared to in the current study. Likewise, the 211

212 proportions of the patients achieving MCID in HAQ were higher in the RCTs compared to the current results; 63.8%, 67%, and 36.6% respectively. Differences in the baseline demographic characteristics, especially those associated with response, may possibl y explain the higher response rates in RCTs. The patients in the REFLEX (111) and DANCER (113;114) trials were younger than the current cohort of patients (mean ± SD age of 52 ± 12, and 51 ± 11) respectively, compared to 60 ± 11 in the current analysis, recognising that we found that older age was significantly associated with a poorer HAQ response. Current smoking was identified as the only significant predictor of not achieving MCID in HAQ. Smoking and other environmental factors may be associated with pain and physical disability in RA patients. (164) Previous analysis from the BSRBR found that current smoking was significantly associated with non-response to infliximab. (147) 212

213 8. Discussion and conclusions 213

214 The previous three chapters addressed the general aim and the objectives of this thesis. This chapter aims to interpret these findings and discuss their contribution to current knowledge. This chapter will also reflect on the main strengths and potential limitations of the research, and present recommendations for future research. 8.1 Persistence with anti-tnf therapies in RA patients The thesis evaluated the persistence with the first course of anti-tnf therapies in patients with RA for up to five years, in which persistence was compared across three anti -TNF therapies (ETN, INF, and ADA). The large sample size of the BSRBR has allowed this thesis to study a number of anti-tnf and nbdmard combinations which included double and triple combinations of nbdmards including MTX. These combinations are used in clinical practice however their association with the persistence with anti-tnf therapies has not previously been investigated. The results showed that patients were more likely to discontinue INF and ADA compared to ETN. The results confirmed the superior persistence with all anti-tnf therapies when combined with MTX compared to mono-therapy. Moreover the results showed that the patients who received anti-tnf therapies in combination with two or three nbdmards including MTX were more likely to continue their anti-tnf therapy compared to those who were receiving their anti-tnf with only MTX. In contrast patients who received an anti-tnf in combination with SSZ or LEF were more likely to discontinue their anti-tnf therapy compared to those who received their anti-tnf with MTX. High baseline HAQ score, female sex, having co-morbidities, being a current smoker, and failing four or more previous nbdmards were significant predictors of overall anti-tnf therapy discontinuation in both the univariate and the multivariate Cox proportional hazards model. 8.2 Comparative effectiveness of RTX versus a second alternative anti-tnf 214

215 The findings showed that about thirty percent of the patients who were treated with their first anti-tnf had discontinued therapy within one year. In this situation, the thesis addressed an important clinical question about what would be the most clinically effective therapeutic option for those patients; switching to RTX or to another alternative anti-tnf therapy? The results showed that, six months after switching, patients who switched to RTX were significantly more likely to improve as measured by both achieving EULAR response and achieving a MCID in HAQ compared to those who switched to another anti-tnf therapy. 8.3 Effectiveness of RTX in RA patients The thesis includes an evaluation of the effectiveness of RTX in a wide inclusion criteria cohort of patients treated in routine clinical practice. The results showed that, six months after starting the therapy, RTX was found to be effective in the management of RA as measured by both the clinical and patient reported outcomes. Interestingly, the results came in agreement with the published clinical trials that examined the efficacy or RTX in RA patients. The patients showed significant decrease in both the mean DAS28 (-1.42 (95% CI -1.53, -1.30)) and HAQ scores (-0.12 (95% CI -0.16: -0.09)). Moderate or good EULAR response was achieved in 60% of patients, and a MCID in HAQ was achieved in 36.6% of the patients. The quality of life of the patients improved significantly after receiving RTX as measured by the EQ-5D utility score 0.10 (95% CI 0.07: 0.13) with a significant decrease in the proportions of patients reporting any problems in self-care, daily activities and pain. A high baseline DAS28 score and RF positive status were identified as significant predictors of clinical effectiveness of RTX. In contrast, high baseline HAQ score and female gender were identified as significant predictors of lower response to RTX six months after starting the therapy. Likewise, high baseline HAQ score was identified as a significant predictor of improvements in six months HAQ scores. Older age, female gender, current smoking and concurrent steroid therapy were identified as predictors of lower improvements in HAQ scores. Patients who had not previously received anti-tnf therapies and those who had previously failed 215

216 one or more anti-tnf therapy showed similar response rates to RTX as measured by both clinical and patient reported outcomes. 8.4 Originality of the research The originality of this research rose from the fact that many of the questions addressed in this research have never been addressed in the literature before. This includes addressing the impact of different double and triple nbdmards combinations on drug persistence with anti -TNF therapies in RA patients. This research was also the first to compare the patient reported outcomes after switching to either RTX or a second anti-tnf in patients who had failed their first anti-tnf therapy. In addition, to date, this was the largest observational study that reported on patient reported outcomes after receiving a first course of RTX. This is also the first study to report on EQ-5D-measured health utility in those patients. These results would facilitate costutility analysis in subsequent research in the future. This research was also the first to identify predictors of improvements in patients physical function after recei ving a first course of RTX. 8.5 Strengths Broad inclusion criteria of the patients One of the key strengths of the thesis is the broad inclusion criteria of the patients studied which in turn reflected routine clinical practice in the UK. Patients who started RTX were included irrespective of their RF status, co-morbidities, anti-tnf history, and concurrent nbdmards. Although INF is licensed to be used in combination with MTX, within this real life cohort of patients INF was found to be used alone or with other nbdmards. Likewise, although RTX is licensed to be used in combination with MTX for patients who had failed one or more anti -TNF therapy, 16.4% of the patients were found to be anti-tnf naïve patients and only 53.1% of RTX patients were receiving RTX in combination with MTX. This can be explained by the fact that some patients had contraindications to anti-tnf therapies (such as prior malignancy) so they 216

217 were prescribed RTX as the first-line biologic therapy. Likewise, patients who had contraindications or intolerance to MTX were prescribed RTX alone or in combination with other nbdmards. It was observed that patients who were anti-tnf naïve had more co-morbidities compared to those who had failed previous anti-tnf therapies, these co-morbidities may be the reason why those patients were prescribed RTX rather than anti-tnf therapies. It is also possible that some of the patients may have been included in a clinical trial of response of RTX in anti - TNF naïve patients in the UK and hence were receiving RTX without trying anti-tnf therapies first Large sample size With a very large cohort of 10,963 patients who were starting their first anti -TNF therapy, the thesis analysed the persistence with anti-tnf therapy in sub-cohorts of patients who were receiving different double and triple combinations of nbdmards. Despite this very large sample size, some sub-cohorts of the patients were still very small when the analysis was stratified by both the type of anti-tnf and the type of the concomitant nbdmards (for example INF-SSZ combination was found in 28 patients). However, this remains the first analysis to run such comparisons with this sample size. Although the BSRBR was not designed to collect data at the time patients switched their biologic therapy, the large sample size of the BSRBR has allowed this thesis to perform a comparative effectiveness analysis after switching to either RTX or alternative anti-tnf therapy by allowing a time window of 3 months within which the patient was included if he/she had data recorded; ending with analysis of 387 patients switched to RTX and 941 patients switched to anti -TNF therapy which is the largest in nature to date. 217

218 Very recently, an international multi-register observational study examined the effectiveness of RTX in RA patients. (136) The work presented in this thesis provides additional insight by reporting on the effectiveness of RTX in a large national register considering both clinical effectiveness and patient reported physical function and quality of life. 8.6 Limitations The limitations related to the analyses presented in this thesis are inherent to the fact that this is an observational study in which patients were not randomised to their therapies. Unmeasured confounders may partly explain some of the results. Moreover, the BSRBR was established primarily to study the safety of biologic therapies and it was not developed to specifically address questions of treatment effectiveness especially in the case of switching from biologic therapy to another Lack of randomisation To overcome the lack of randomisation, the analyses within the thesis were adjusted for a number of potential confounders. The Cox-proportional models that compared the persistence with anti-tnf therapies across the three anti-tnfs and across the sub-groups according to concurrent nbdmards, were adjusted for differences in baseline demographic and disease characteristics. However, there may still be unmeasured confounders such as the ability of some patients to tolerate drugs in general. As discussed, patients who were able to tolerate double and triple nbdmards combinations may be more likely to tolerate the anti -TNF therapy. Likewise, patients on no concurrent nbdmards or single LEF may have been unable to tolerate other nbdmards and subsequently they could not tolerate the anti-tnf therapy. It was also noticed that patients who were receiving double or triple nbdmards were of shorter disease duration and might have had less severe disease in general. Moreover those patients 218

219 were more likely to be starting ADA rather than INF or ETN which may be related to the time they started the therapy as ADA registrations were later than the other two anti -TNF therapies; it is possible that at this later time the possibility of prescribing two or three nbdmards might have been more widespread. Although, the statistical models were stratified by the choice of anti-tnf therapy and the year of anti-tnf start, and adjusted for disease duration, these differences may persist. Taking these limitations in mind, it cannot be concluded that combination of nbdmard therapy including MTX should be started when anti -TNF therapy is started, but it can be concluded that it is recommended to continue combination of nbdmards if tolerated when a patient is starting ant-tnf therapy. The limitation of lack of randomisation was seen also in the comparative effectiveness analysis where the patients were not randomised to start RTX or to switch to a second alte rnative anti- TNF therapy after failing the first anti-tnf. It was a decision of the treating consultant who may have preferred one option over another based on certain baseline characteristics of the patients. Moreover, in this cohort of patients there might be no choice at all, as some patients who switched to another anti-tnf may have done this at a time RTX was not even available. In an attempt to overcome this potential source of bias all the comparative analyses were adjusted for propensity scores which considered any differences in the patients baseline characteristics. The same issue was seen when response to RTX was compared between patients who were anti- TNF naïve and those who have failed at least one anti-tnf therapy. Therefore the analysis was adjusted for propensity scores which considered any differences in the patients baseline characteristics. As the adjustments were necessary within these analyses, the conclusions were drawn from the adjusted models not the crude comparisons Time of data collection 219

220 Data on concurrent nbdmards were collected at the time the biologic therapy was started. It may be likely that some patients may have either decreased or indeed started new nbdmards following the start of biologic therapy based on the initial response. A future analysis that follows the concurrently used nbdmards to take into account whether they were dropped or replaced during the follow-up time may produce different results but will require an even larger population as the number of different treatment combinations will increase further. As previously discussed, the BSRBR was designed primarily to study the safety of biologic therapies in RA rather than comparative effectiveness analyses across different treatment options. Subsequently the BSRBR did not collect data at the time patients switch one anti-tnf therapy by another. However, the regular collection of DAS28 and HAQ scores and the very large sample size allowed running such analyses. Since DAS28 and HAQ score were not routi nely collected at the time a biologic therapy was switched, with the exception of those patients who were reregistered in the RTX cohort, a time window had to be used to include patients if they had data collected within this time window. This was applied for both baseline and follow-up data. Therefore, not all the patients who switched their therapy were included. However, the final sample size of this analysis was still the largest to date in addressing this specific clinical question. As there was no baseline data re-recorded at the time of switch, baseline data which do not change by time were directly imported from the patient baseline forms at the time of registration with the BSRBR. This included data on co-morbidities. It may be possible that some patients may have developed new co-morbidities during their first anti-tnf therapy time. However, comparing co-morbidities for RTX patients who had co-morbidity data collected at both the start of anti-tnf and RTX therapies showed the same mean number of co-morbidities. Although there were more patients with one or more co-morbidities in the arm of patients who switched to RTX rather than those who switched to a second anti-tnf, the response was higher 220

221 in RTX patients. This may suggest that the co-morbidities were not necessarily acting as a confounder. It should be mentioned that having co-morbidities was identified as a predictor of anti-tnf therapy discontinuation but at the same time co-morbidity was not found to affect the response after switching; reflecting that co-morbidities did not necessarily play a role in the response but it may had affected the choice of therapy the patient switched to. The data of the BSRBR were collected in the routine clinical setting and the timing of clinic appointments may not fall exactly at proposed times of the study protocol, meaning there were some allowances needed in the timing of data collection. For RTX patients who were reregistered with the BSRBR, there was a time period the data management team contacting the consultant to ask for completion of the appropriate forms and the team in some patients, would not had got a quick response. However, within these forms the BSRBR asked for the DAS28 measured at the time RTX was started irrespective of the date the form was filled Limitations related to statistical analysis The comparative effectiveness analysis of RTX versus a second alternative anti -TNF was limited to the patients who had failed only one anti-tnf therapy. Although this was the a priori objective of the analysis, in clinical practice, patients may switch their therapy after failing more than one anti-tnf therapy. Within the patients included in the analysis it was noticed that some patients switched their therapy for a second time. Future analysis that considers multiple switching may be of interest, although the sample size will be smaller. However, the thesis addressed the question which is most relevant to current clinical practice. In this situation, it is worth mentioning that a very recent study from the Dutch biologic register (September 2011) has reported better improvements in patients who switched to RTX compared to in those who switched to a third anti-tnf after failing two anti-tnf therapies. (165) 221

222 In the same analysis, to maximise the power, the patients who switched to a second anti -TNF were combined together in one group rather than analysing by each subgroup according to the anti-tnf therapy (ETN, INF, and ADA). Likewise, subgroups of patients according to the particular reason why they switched their first anti-tnf therapy were not studied. However, our analysis allowed for adjustment for the reason for switching in the propensity models Other limitations As expected in most observational studies, missing data for some variables may be a possible limitation of the current analyses in this PhD. However, the missing data in the variables did not exceed 10%. In line with previous data collected for the patients receiving anti -TNF therapies in the BSRBR, it was noticed that there were fewer patients returning follow-up HAQ forms compared to consultant forms in the patients who started RTX. There is a possible limitation if the non-return of the HAQ form was associated with patient disability and hence the results will be influenced. But this observation came in line with all the patients registered with the BSRBR rather than in a specific treatment. This may be related to the fact that the patient may be less keen to send back the form in comparison to the treating consultant. There might be a concern about the accuracy of individual data item; different consultants may report different issues given this very large sample size cohort of patients from different centres within the UK. Different consultant may prescribe different therapies based on different characteristics, however this is the case within the routine clinical practice and it remains a potential unmeasured confounder. The decision to start or stop the therapy including the reason for discontinuation was left to the treating consultant. The reasons for discontinuation may not fall into one single category. The treating consultant was asked to choose one reason for discontinuation. 222

223 8.7 Final conclusions In conclusion, the thesis has shown that compared to ETN and ADA, RA patients were more likely to discontinue INF. Compared to MTX as a concomitant nbdmard therapy with anti -TNF therapies, MTX in combination with SSZ, HCQ or both resulted in better long term persistence with the anti-tnf therapies. However, LEF and SSZ mono-therapies in combination with anti-tnf therapies resulted in lower rates of persistence for both inefficacy and developing adverse event. The results of the comparative effectiveness analysis that compared RTX versus a second alternative anti-tnf therapy in RA patients who discontinued their first anti -TNF therapy, suggest that switching to RTX is of more benefit than switching to a second alternative anti -TNF therapy. The benefits of RTX were superior in both achieving EULAR response criteria and achieving MCID in HAQ. Six months after starting the therapy, first course of RTX was found to be an effective treatment in RA patients who were treated in routine clinical practice in the UK. The response was similar in anti-tnf failures and anti-tnf naïve patients. High baseline disease activity and positive RF status were associated with a significant decrease in disease activity while females and patients with high baseline HAQ scores were less likely to improve. RTX was also found to be effective in improving both patients physical function as measured by HAQ score and the patients health related quality of life as measured by EQ-5D. Older patients, females, current smokers, and those receiving concurrent corticosteroids were significantly less likely to improve their HAQ score. Higher baseline HAQ score was associated with better improvements in HAQ which did not reach the MCID in HAQ. 8.8 Recommendations for clinical practice 223

224 For clinical practice it is recommended to continue concomitant nbdmards when anti -TNF therapy is started. An exception is to consider discontinuation of LEF and SSZ mono-therapies when anti-tnf therapy is commenced (except for SSZ-ETN combination). These results also suggest that in clinical practice, for patients who had failed their first anti-tnf therapy, in line with NICE recommendations, it is better to start RTX at this point rather than switching to a second anti-tnf therapy. RTX can be used effectively in anti-tnf naïve patients. Although the license indication of the use of RTX in RA restricted RTX for patients who had previously failed at least one anti -TNF therapy, in routine clinical practice when patients have contraindication or intolerance to anti -TNF therapies RTX was used effectively and showed similar performance as in patients who were anti-tnf failures. However, cost-effectiveness analysis and the safety profile of RTX therapy are necessary to confirm whether RTX should become first line biologic therapy in RA or not. Both RF positive and negative patients can be effectively treated with RTX. However, RF positive patients showed superior improvements which was significant but was not enough to be translated into EULAR response. In case of contraindication or intolerance to MTX, RTX can be used alone or in combination with other nbdmard if tolerated. There is no indication to stop MTX when tolerated. It should be noticed that this recommendation may be limited by the fact that the impact of MTX may not become apparent within the first six months of follow-up. It is recommended for RTX to be used in patients with high disease activity as those patients will show an improvement but it should be noticed that those patients will be less likely to achieve either good EULAR response or disease remission. Patients receiving RTX should be advised to stop smoking. Finally, it is recommended to start RTX therapy in younger age and in patients with less physical disability before a time when physical disability might be irreversible. Older patients and those 224

225 with high physical disability will be less likely to achieve disease remission. Patients with high physical disability will show initial improvements in their physical function but it will not reach clinically important improvements. 8.9 Recommendations for future work Due to the possible effect of unmeasured confounders in the analysis examining the impact of different concomitant nbdmards on persistence with anti-tnf therapies, it could not be concluded that patients should be starting double or triple nbdmards combinations including MTX or should be substituting SSZ or LEF mono-therapies with other nbdmards such as MTX. This is because the patients in this study were already tolerating double and triple nbdmards at anti-tnf therapy start. A clinical trial in which patients are randomised to start or stop different nbdmards combinations at anti-tnf therapy start would be suggested to study this issue further. The comparative effectiveness analysis of RTX versus a second anti-tnf therapy was limited to patients who had failed only one anti-tnf therapy. Future analysis that considers the response in multiple anti-tnf switchers, or specifically looking at the reason for the switching is likely to be of interest. After the recent approvals of abatacept and tocilizumab in RA patients, the options of available therapies to treat patients who have failed their first anti -TNF have increased. The time of this PhD did not allow including such treatment comparisons as these drugs are very recently introduced into clinical practice. A future comparison of effectiveness after switching to abatacept, tocilizumab, RTX, or a second alternative anti-tnf is likely to be of interest. Newly licensed anti-tnf therapies could also be included in the comparison. A clinical trial that randomise patients to either treatment options is also likely to be of interest. An important recommendation for future work lies in the fact that it is very important to study the effectiveness after a second course of RTX. Retreatment with a second course of RTX is 225

226 allowed after a minimum of six months from the first course. As recruitment of RA patients receiving RTX in the BSRBR started at May 2008, the number of patients who were retreated (minimum six months from recruitment) and followed up for another six months from retreatment was too small to be analysed within the time of this PhD. Examining both clinical effectiveness and patient reported outcomes after retreatment with a second course of RTX is recommended for future research. After RTX has showed effectiveness in routine clinical practice, studying the safety profile of the patients after receiving RTX is a very important area of research. The results of the published RCTs showed that treatment with RTX is safe in RA patients in the short term. (108;110;113) However, long term safety needs to be studied. Moreover the tight inclusion criteria of the RCTs might not reflect the use in routine clinical practice. The safety of repeated doses of RTX might be questioned because of the fear that repeated B cells depletion may lead to immunodeficiency and increased risk of infections. However, recent RCTs found that retreated patients showed sustained improvements of the disease with a similar safety profile over six months of follow-up. (115;116) 226

227 References Reference List (1) The definition of clinical pharmacy. Pharmacotherapy 2008;28(6): (2) Morrato EH. Textbook of Pharmacoepidemiology Edited by Brian L. Strom and Stephen E. Kimmel. American Journal of Epidemiology 2007;166(7): (3) Raven A. Clinical trials : an introduction. 2nd ed. Oxford : Radcliffe Medical, (4) Di Giovanna I, Hayes G, Institute of Clinical Research. Principles of clinical research. Petersfield : Wrightson Biomedical Pub., for the Institute of Clinical Research, (5) Jadad AR. Randomised controlled trials : a user's guide. London : BMJ Books, (6) Talbot JC, Nilsson BS. Pharmacovigilance in the pharmaceutical industry. Br J Clin Pharmacol 1998;45(5): (7) Furberg CD, Pitt B. Withdrawal of cerivastatin from the world market. Curr Control Trials Cardiovasc Med 2001;2(5): (8) Lee DM, Weinblatt ME. Rheumatoid arthritis. Lancet 2001;358(9285): (9) Guidelines for the management of rheumatoid arthritis: 2002 Update. Arthritis Rheum 2002;46(2): (10) Bennet GA, Cobb S, Jacos R, Jessr RA, Ropes MW. Proposed diagnostic criteria for rheumatoid arthritis. Bull Rheum Dis 1956;7(4): (11) Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31(3): (12) Aletaha D, Neogi T, Silman AJ, Funovits J, Felson DT, Bingham CO, III, et al rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Ann Rheum Dis 2010;69(9): (13) Alamanos Y, Voulgari PV, Drosos AA. Incidence and prevalence of rheumatoid arthritis, based on the 1987 American College of Rheumatology criteria: a systematic review. Semin Arthritis Rheum 2006;36(3): (14) Symmons D, Turner G, Webb R, Asten P, Barrett E, Lunt M, et al. The prevalence of rheumatoid arthritis in the United Kingdom: new estimates for a new century. Rheumatology (Oxford) 2002;41(7):

228 (15) Barton A, Worthington J. Genetic susceptibility to rheumatoid arthritis: an emerging picture. Arthritis Rheum 2009;61(10): (16) Spector TD, Roman E, Silman AJ. The pill, parity, and rheumatoid arthritis. Arthritis Rheum 1990;33(6): (17) Brennan FM, McInnes IB. Evidence that cytokines play a role in rheumatoid arthritis. J Clin Invest 2008;118(11): (18) Choy EH, Panayi GS. Cytokine pathways and joint inflammation in rheumatoid arthritis. N Engl J Med 2001;344(12): (19) van der Heijde DM, van 't Hof MA, van Riel PL, Theunisse LA, Lubberts EW, van Leeuwen MA, et al. Judging disease activity in clinical practice in rheumatoid arthritis: first step in the development of a disease activity score. Ann Rheum Dis 1990;49(11): (20) van Gestel AM, Prevoo ML, van 't Hof MA, van Rijswijk MH, van de Putte LB, van Riel PL. Development and validation of the European League Against Rheumatism response criteria for rheumatoid arthritis. Comparison with the preliminary American College of Rheumatology and the World Health Organization/International League Against Rheumatism Criteria. Arthritis Rheum 1996;39(1): (21) Felson DT, Anderson JJ, Boers M, Bombardier C, Furst D, Goldsmith C, et al. American College of Rheumatology. Preliminary definition of improvement in rheumatoid arthritis. Arthritis Rheum 1995;38(6): (22) Felson DT, Anderson JJ, Boers M, Bombardier C, Chernoff M, Fried B, et al. The American College of Rheumatology preliminary core set of disease activity measures for rheumatoid arthritis clinical trials. The Committee on Outcome Measures in Rheumatoid Arthritis Clinical Trials. Arthritis Rheum 1993;36(6): (23) Fransen J, van Riel PL. The Disease Activity Score and the EULAR response criteria. Clin Exp Rheumatol 2005;23(5 Suppl 39):S93-S99. (24) Prevoo ML, van Gestel AM, van THM, van Rijswijk MH, van de Putte LB, van Riel PL. Remission in a prospective study of patients with rheumatoid arthritis. American Rheumatism Association preliminary remission criteria in relation to the disease activity score. Br J Rheumatol 1996;35(11): (25) Fransen J, Creemers MC, van Riel PL. Remission in rheumatoid arthritis: agreement of the disease activity score (DAS28) with the ARA preliminary remission criteria. Rheumatology (Oxford) 2004;43(10): (26) Prevoo ML, van 't Hof MA, Kuper HH, van Leeuwen MA, van de Putte LB, van Riel PL. Modified disease activity scores that include twenty-eight-joint counts. Development and validation in a prospective longitudinal study of patients with rheumatoid arthritis. Arthritis Rheum 1995;38(1):

229 (27) van Gestel AM, Haagsma CJ, van Riel PL. Validation of rheumatoid arthritis improvement criteria that include simplified joint counts. Arthritis Rheum 1998;41(10): (28) Felson DT, Anderson JJ, Lange ML, Wells G, LaValley MP. Should improvement in rheumatoid arthritis clinical trials be defined as fifty percent or seventy percent improvement in core set measures, rather than twenty percent? Arthritis Rheum 1998;41(9): (29) van Gestel AM, Anderson JJ, van Riel PL, Boers M, Haagsma CJ, Rich B, et al. ACR and EULAR improvement criteria have comparable validity in rheumatoid arthritis trials. American College of Rheumatology European League of Associations for Rheumatology. J Rheumatol 1999;26(3): (30) Fries JF, Spitz P, Kraines RG, Holman HR. Measurement of patient outcome in arthritis. Arthritis Rheum 1980;23(2): (31) Bruce B, Fries JF. The Health Assessment Questionnaire (HAQ). Clin Exp Rheumatol 2005;23(5 Suppl 39):S14-S18. (32) Wells GA, Tugwell P, Kraag GR, Baker PR, Groh J, Redelmeier DA. Minimum important difference between patients with rheumatoid arthritis: the patient's perspective. J Rheumatol 1993;20(3): (33) The EuroQol Group. EuroQol--a new facility for the measurement of healthrelated quality of life. Health Policy 1990;16(3): (34) Dolan P, Gudex C, Kind P, Williams A. A social tariff for EuroQol: results from a UK general population survey. York: Centre for Health Economics, University of York [cited 2010 Nov 18];Available from: URL: (35) Ware JE, Jr., Sherbourne CD. The MOS 36-item short-form health survey (SF- 36). I. Conceptual framework and item selection. Med Care 1992;30(6): (36) Kirwan JR. The effect of glucocorticoids on joint destruction in rheumatoid arthritis. The Arthritis and Rheumatism Council Low-Dose Glucocorticoid Study Group. N Engl J Med 1995;333(3): (37) Jobanputra P, Wilson J, Douglas K, Burls A. A survey of British rheumatologists' DMARD preferences for rheumatoid arthritis. Rheumatology (Oxford) 2004;43(2): (38) Calguneri M, Pay S, Caliskaner Z, Apras S, Kiraz S, Ertenli I, et al. Combination therapy versus monotherapy for the treatment of patients with rheumatoid arthritis. Clin Exp Rheumatol 1999;17(6): (39) Food and Drug Administration. What Are "Biologics" Questions and Answers [cited 2011 Jul 26];Available from: URL: (40) European Medicines Agency. Enbrel Summary of product characteristics, ANNEX I [cited 2011 Jul 27];Available from: URL: 229

230 _Product_Information/human/000262/WC pdf (41) European Medicines Agency. Remicade Summary of product characteristics, ANNEX I [cited 2011 Jul 27];Available from: URL: _Product_Information/human/000240/WC pdf (42) European Medicines Agency. Humira Summary of product characteristics, ANNEX I [cited 2011 Jul 27];Available from: URL: _Product_Information/human/000481/WC pdf (43) Burmester GR, Mariette X, Montecucco C, Monteagudo-Saez I, Malaise M, Tzioufas AG, et al. Adalimumab alone and in combination with diseasemodifying antirheumatic drugs for the treatment of rheumatoid arthritis in clinical practice: the Research in Active Rheumatoid Arthritis (ReAct) trial. Ann Rheum Dis 2007;66(6): (44) European Medicines Agency. Cimzia Summary of product Characteristics [cited 2011 Jul 27];Available from: URL: _Product_Information/human/001037/WC pdf (45) European Medicines Agency. Simponi Summary of product Characteristics [cited 2011 Jul 27];Available from: URL: _Product_Information/human/000992/WC pdf (46) Maini R, St Clair EW, Breedveld F, Furst D, Kalden J, Weisman M, et al. Infliximab (chimeric anti-tumour necrosis factor alpha monoclonal antibody) versus placebo in rheumatoid arthritis patients receiving concomitant methotrexate: a randomised phase III trial. ATTRACT Study Group. Lancet 1999;354(9194): (47) Breedveld FC, Weisman MH, Kavanaugh AF, Cohen SB, Pavelka K, Van Vollenhoven R., et al. The PREMIER study: A multicenter, randomized, doubleblind clinical trial of combination therapy with adalimumab plus methotrexate versus methotrexate alone or adalimumab alone in patients with early, aggressive rheumatoid arthritis who had not had previous methotrexate treatment. Arthritis Rheum 2006;54(1): (48) Weinblatt ME, Kremer JM, Bankhurst AD, Bulpitt KJ, Fleischmann RM, Fox RI, et al. A trial of etanercept, a recombinant tumor necrosis factor receptor:fc fusion protein, in patients with rheumatoid arthritis receiving methotrexate. N Engl J Med 1999;340(4): (49) National Institute for Health and Clinical Effectiveness Technology Appraisal Guidance 130. Adalimumab, etanercept and infliximab for the treatment of rheumatoid arthritis [cited 2011 Jul 23];Available from: URL: 230

231 (50) National Institute for Health and Clinical Effectiveness. Certolizumab pegol for the treatment of rheumatoid arthritis [cited 2011 Sep 20];Available from: URL: (51) National Institute for Health and Clinical Effectiveness. Golimumab for the treatment of rheumatoid arthritis after the failure of previous diseasemodifying anti-rheumatic drugs [cited 2011 Sep 20];Available from: URL: (52) National Institute for Health and Clinical Effectiveness. Golimumab for the treatment of methotrexate-naive rheumatoid arthritis [cited 2011 Sep 20];Available from: URL: (53) Nixon RM, Bansback N, Brennan A. Using mixed treatment comparisons and meta-regression to perform indirect comparisons to estimate the efficacy of biologic treatments in rheumatoid arthritis. Stat Med 2007;26(6): (54) onso-ruiz A, Pijoan JI, Ansuategui E, Urkaregi A, Calabozo M, Quintana A. Tumor necrosis factor alpha drugs in rheumatoid arthritis: systematic review and metaanalysis of efficacy and safety. BMC Musculoskelet Disord 2008;9:52. (55) Zintzaras E, Dahabreh IJ, Giannouli S, Voulgarelis M, Moutsopoulos HM. Infliximab and methotrexate in the treatment of rheumatoid arthritis: a systematic review and meta-analysis of dosage regimens. Clin Ther 2008;30(11): (56) Keystone E, Heijde D, Mason D, Jr., Landewe R, Vollenhoven RV, Combe B, et al. Certolizumab pegol plus methotrexate is significantly more effective than placebo plus methotrexate in active rheumatoid arthritis: findings of a fiftytwo-week, phase III, multicenter, randomized, double-blind, placebocontrolled, parallel-group study. Arthritis Rheum 2008;58(11): (57) Fleischmann R, Vencovsky J, van Vollenhoven RF, Borenstein D, Box J, Coteur G, et al. Efficacy and safety of certolizumab pegol monotherapy every 4 weeks in patients with rheumatoid arthritis failing previous disease-modifying antirheumatic therapy: the FAST4WARD study. Ann Rheum Dis 2009;68(6): (58) Smolen J, Landewe RB, Mease P, Brzezicki J, Mason D, Luijtens K, et al. Efficacy and safety of certolizumab pegol plus methotrexate in active rheumatoid arthritis: the RAPID 2 study. A randomised controlled trial. Ann Rheum Dis 2009;68(6): (59) Emery P, Fleischmann RM, Moreland LW, Hsia EC, Strusberg I, Durez P, et al. Golimumab, a human anti-tumor necrosis factor alpha monoclonal antibody, injected subcutaneously every four weeks in methotrexate-naive patients with active rheumatoid arthritis: twenty-four-week results of a phase III, multicenter, randomized, double-blind, placebo-controlled study of golimumab before methotrexate as first-line therapy for early-onset rheumatoid arthritis. Arthritis Rheum 2009;60(8):

232 (60) Smolen JS, Kay J, Doyle MK, Landewe R, Matteson EL, Wollenhaupt J, et al. Golimumab in patients with active rheumatoid arthritis after treatment with tumour necrosis factor alpha inhibitors (GO-AFTER study): a multicentre, randomised, double-blind, placebo-controlled, phase III trial. Lancet 2009;374(9685): (61) European Medicines Agency. Kineret Summary of product characteristics, ANNEX I [cited 2011 Jul 17];Available from: URL: _Product_Information/human/000363/WC pdf (62) Cohen S, Hurd E, Cush J, Schiff M, Weinblatt ME, Moreland LW, et al. Treatment of rheumatoid arthritis with anakinra, a recombinant human interleukin-1 receptor antagonist, in combination with methotrexate: results of a twenty-four-week, multicenter, randomized, double-blind, placebocontrolled trial. Arthritis Rheum 2002;46(3): (63) Cohen SB, Moreland LW, Cush JJ, Greenwald MW, Block S, Shergy WJ, et al. A multicentre, double blind, randomised, placebo controlled trial of anakinra (Kineret), a recombinant interleukin 1 receptor antagonist, in patients with rheumatoid arthritis treated with background methotrexate. Ann Rheum Dis 2004;63(9): (64) National Institute for Health and Clinical Effectiveness. Anakinra for rheumatoid Arhritis [cited 2011 Jul 27];Available from: URL: (65) European Medicines Agency. RoActemra Summary of product Characteristics [cited 2011 Jul 27];Available from: URL: _Product_Information/human/000955/WC pdf (66) National Institute for Health and Clinical Effectiveness. Tocilizumab for the treatment of rheumatoid arthritis [cited 2011 Sep 20];Available from: URL: (67) European Medicines Agency. Orencia Summary of product Characteristics [cited 2011 Jul 27];Available from: URL: _Product_Information/human/000701/WC pdf (68) National Institute for Health and Clinical Effectiveness. Abatacept for the treatment of rheumatoid arthritis after the failure of conventional diseasemodifying anti-rheumatic drugs [cited 2011 Sep 20];Available from: URL: (69) Criscione LG, St Clair EW. Tumor necrosis factor-alpha antagonists for the treatment of rheumatic diseases. Curr Opin Rheumatol 2002;14(3): (70) Genovese MC, Cohen S, Moreland L, Lium D, Robbins S, Newmark R, et al. Combination therapy with etanercept and anakinra in the treatment of 232

233 patients with rheumatoid arthritis who have been treated unsuccessfully with methotrexate. Arthritis Rheum 2004;50(5): (71) Hansen KE, Hildebrand JP, Genovese MC, Cush JJ, Patel S, Cooley DA, et al. The efficacy of switching from etanercept to infliximab in patients with rheumatoid arthritis. J Rheumatol 2004;31(6): (72) Reff ME, Carner K, Chambers KS, Chinn PC, Leonard JE, Raab R, et al. Depletion of B cells in vivo by a chimeric mouse human monoclonal antibody to CD20. Blood 1994;83(2): (73) Boye J, Elter T, Engert A. An overview of the current clinical use of the anti- CD20 monoclonal antibody rituximab. Ann Oncol 2003;14(4): (74) Pescovitz MD. Rituximab, an anti-cd20 monoclonal antibody: history and mechanism of action. Am J Transplant 2006;6(5 Pt 1): (75) Taylor RP, Lindorfer MA. Drug insight: the mechanism of action of rituximab in autoimmune disease--the immune complex decoy hypothesis. Nat Clin Pract Rheumatol 2007;3(2): (76) European Medicines Agency. EPARs for authorised medicinal products for human use June 1 [cited 2010 Nov 18];Available from: URL: _Assessment_Report_-_Variation/human/000165/WC pdf (77) Food and Drug Administration. Department of Health and Human Services February 28 [cited 2010 Nov 18];Available from: URL: 1_ltr.pdf (78) National Institute for Health and Clinical Excellence. Rituximab for the treatment of rheumatoid arthritis [cited 2011 Jul 23];Available from: URL: (79) National Institute for Health and Clinical Effectiveness. Adalimumab, etanercept, infliximab, rituximab and abatacept for the treatment of rheumatoid arthritis after the failure of a TNF inhibitor [cited 2011 Sep 20];Available from: URL: (80) Flendrie M, Creemers MC, Welsing PM, den Broeder AA, van Riel PL. Survival during treatment with tumour necrosis factor blocking agents in rheumatoid arthritis. Ann Rheum Dis 2003;62 Suppl 2:ii30-ii33. (81) Chevillott lard H, Ornetti P, Mistrih R, Sidot C, Dupuis J, Dellas JA, et al. Survival and safety of treatment with infliximab in the elderly population. Rheumatology (Oxford) 2005;44(5): (82) Zink A, Listing J, Kary S, Ramlau P, Stoyanova-Scholz M, Babinsky K, et al. Treatment continuation in patients receiving biological agents or conventional DMARD therapy. Ann Rheum Dis 2005;64(9):

234 (83) Duclos M, Gossec L, Ruyssen-Witrand A, Salliot C, Luc M, Guignard S, et al. Retention rates of tumor necrosis factor blockers in daily practice in 770 rheumatic patients. J Rheumatol 2006;33(12): (84) Kristensen LE, Saxne T, Nilsson JA, Geborek P. Impact of concomitant DMARD therapy on adherence to treatment with etanercept and infliximab in rheumatoid arthritis. Results from a six-year observational study in southern Sweden. Arthritis Res Ther 2006;8(6):R174. (85) Carmona L, Gomez-Reino JJ. Survival of TNF antagonists in spondylarthritis is better than in rheumatoid arthritis. Data from the Spanish registry BIOBADASER. Arthritis Res Ther 2006;8(3):R72. (86) Brocq O, Roux CH, Albert C, Breuil V, Aknouche N, Ruitord S, et al. TNFalpha antagonist continuation rates in 442 patients with inflammatory joint disease. Joint Bone Spine 2007;74(2): (87) Grijalva CG, Chung CP, Arbogast PG, Stein CM, Mitchel EF, Jr., Griffin MR. Assessment of adherence to and persistence on disease-modifying antirheumatic drugs (DMARDs) in patients with rheumatoid arthritis. Med Care 2007;45(10 Supl 2):S66-S76. (88) Tang B, Rahman M, Waters HC, Callegari P. Treatment persistence with adalimumab, etanercept, or infliximab in combination with methotrexate and the effects on health care costs in patients with rheumatoid arthritis. Clin Ther 2008;30(7): (89) Bertoli AM, Strusberg I, Baravalle M, Betelu Z, Calas D, Morales L, et al. Rate and causes of infliximab discontinuation in patients with rheumatoid arthritis in a private clinical practice. J Clin Rheumatol 2008;14(6): (90) Heiberg MS, Koldingsnes W, Mikkelsen K, Rodevand E, Kaufmann C, Mowinckel P, et al. The comparative one-year performance of anti-tumor necrosis factor alpha drugs in patients with rheumatoid arthritis, psoriatic arthritis, and ankylosing spondylitis: results from a longitudinal, observational, multicenter study. Arthritis Rheum 2008;59(2): (91) Ducoulombier V, Solau E, Coquerelle P, Houvenagel E, Siame JL, Desprez X, et al. Long-term results of infliximab therapy in rheumatoid arthritis: experience acquired by the North-Pas-de-Calais hospital network. Joint Bone Spine 2007;74(1):56-9. (92) Geborek P, Crnkic M, Petersson IF, Saxne T. Etanercept, infliximab, and leflunomide in established rheumatoid arthritis: clinical experience using a structured follow up programme in southern Sweden. Ann Rheum Dis 2002;61(9): (93) Flendrie M, Creemers MC, Welsing PM, van Riel PL. The influence of previous and concomitant leflunomide on the efficacy and safety of infliximab therapy in patients with rheumatoid arthritis; a longitudinal observational study. Rheumatology (Oxford) 2005;44(4):

235 (94) Cobo Ibanez T., Yehia TM, Balsa CA, Hernandez SA, Martin ME. Safety and efficacy of leflunomide and infliximab versus methotrexate and infliximab combination therapy in rheumatoid arthritis. Rheumatology (Oxford) 2005;44(11): (95) Vander CB, Van LS, Wyns B, Westhovens R, Durez P, Van den BF, et al. Fouryear follow-up of infliximab therapy in rheumatoid arthritis patients with longstanding refractory disease: attrition and long-term evolution of disease activity. Arthritis Res Ther 2006;8(4):R112. (96) Voulgari PV, Alamanos Y, Nikas SN, Bougias DV, Temekonidis TI, Drosos AA. Infliximab therapy in established rheumatoid arthritis: an observational study. Am J Med 2005;118(5): (97) Ostergaard M, Unkerskov J, Linde L, Krogh NS, Ravn T, Ringsdal VS, et al. Low remission rates but long drug survival in rheumatoid arthritis patients treated with infliximab or etanercept: results from the nationwide Danish DANBIO database. Scand J Rheumatol 2007;36(2): (98) Hetland ML, Lindegaard HM, Hansen A, Podenphant J, Unkerskov J, Ringsdal VS, et al. Do changes in prescription practice in patients with rheumatoid arthritis treated with biological agents affect treatment response and adherence to therapy? Results from the nationwide Danish DANBIO Registry. Ann Rheum Dis 2008;67(7): (99) Finckh A, Dehler S, Gabay C. The effectiveness of leflunomide as a co-therapy of tumour necrosis factor inhibitors in rheumatoid arthritis: a population-based study. Ann Rheum Dis 2009;68(1):33-9. (100) Tanaka Y, Takeuchi T, Inoue E, Saito K, Sekiguchi N, Sato E, et al. Retrospective clinical study on the notable efficacy and related factors of infliximab therapy in a rheumatoid arthritis management group in Japan: one-year clinical outcomes (RECONFIRM-2). Mod Rheumatol 2008;18(2): (101) Gomez-Reino JJ, Carmona L. Switching TNF antagonists in patients with chronic arthritis: an observational study of 488 patients over a four-year period. Arthritis Res Ther 2006;8(1):R29. (102) Hyrich KL, Lunt M, Watson KD, Symmons DP, Silman AJ. Outcomes after switching from one anti-tumor necrosis factor alpha agent to a second antitumor necrosis factor alpha agent in patients with rheumatoid arthritis: results from a large UK national cohort study. Arthritis Rheum 2007;56(1): (103) Hjardem E, Ostergaard M, Podenphant J, Tarp U, Andersen LS, Bing J, et al. Do rheumatoid arthritis patients in clinical practice benefit from switching from infliximab to a second tumor necrosis factor alpha inhibitor? Ann Rheum Dis 2007;66(9): (104) Laas K, Peltomaa R, Kautiainen H, Leirisalo-Repo M. Clinical impact of switching from infliximab to etanercept in patients with rheumatoid arthritis. Clin Rheumatol 2008;27(7):

236 (105) Markenson JA, Gibofsky A, Palmer WR, Keystone EC, Schiff MH, Feng J, et al. Persistence with Anti-Tumor Necrosis Factor Therapies in Patients with Rheumatoid Arthritis: Observations from the RADIUS Registry. J Rheumatol 2011;38(7): (106) Rigby W, Ferraccioli G, Greenwald M, Zazueta-Montiel B, Fleischmann R, Wassenberg S, et al. Effect of rituximab on physical function and quality of life in patients with rheumatoid arthritis previously untreated with methotrexate. Arthritis Care Res (Hoboken ) 2011;63(5): (107) Tak PP, Rigby WF, Rubbert-Roth A, Peterfy CG, van Vollenhoven RF, Stohl W, et al. Inhibition of joint damage and improved clinical outcomes with rituximab plus methotrexate in early active rheumatoid arthritis: the IMAGE trial. Ann Rheum Dis 2010;70(1): (108) Edwards JC, Szczepanski L, Szechinski J, Filipowicz-Sosnowska A, Emery P, Close DR, et al. Efficacy of B-cell-targeted therapy with rituximab in patients with rheumatoid arthritis. N Engl J Med 2004;350(25): (109) Strand V, Balbir-Gurman A, Pavelka K, Emery P, Li N, Yin M, et al. Sustained benefit in rheumatoid arthritis following one course of rituximab: improvements in physical function over 2 years. Rheumatology (Oxford) 2006;45(12): (110) Cohen SB, Emery P, Greenwald MW, Dougados M, Furie RA, Genovese MC, et al. Rituximab for rheumatoid arthritis refractory to anti-tumor necrosis factor therapy: Results of a multicenter, randomized, double-blind, placebocontrolled, phase III trial evaluating primary efficacy and safety at twenty-four weeks. Arthritis Rheum 2006;54(9): (111) Keystone E, Burmester GR, Furie R, Loveless JE, Emery P, Kremer J, et al. Improvement in patient-reported outcomes in a rituximab trial in patients with severe rheumatoid arthritis refractory to anti-tumor necrosis factor therapy. Arthritis Rheum 2008;59(6): (112) Emery P, Deodhar A, Rigby WF, Isaacs JD, Combe B, Racewicz AJ, et al. Efficacy and safety of different doses and retreatment of rituximab: a randomised, placebo-controlled trial in patients who are biological naive with active rheumatoid arthritis and an inadequate response to methotrexate (Study Evaluating Rituximab's Efficacy in MTX inadequate responders (SERENE)). Ann Rheum Dis 2010;69(9): (113) Emery P, Fleischmann R, Filipowicz-Sosnowska A, Schechtman J, Szczepanski L, Kavanaugh A, et al. The efficacy and safety of rituximab in patients with active rheumatoid arthritis despite methotrexate treatment: results of a phase IIB randomized, double-blind, placebo-controlled, dose-ranging trial. Arthritis Rheum 2006;54(5): (114) Mease PJ, Revicki DA, Szechinski J, Greenwald M, Kivitz A, Barile-Fabris L, et al. Improved health-related quality of life for patients with active rheumatoid arthritis receiving rituximab: Results of the Dose-Ranging Assessment: 236

237 International Clinical Evaluation of Rituximab in Rheumatoid Arthritis (DANCER) Trial. J Rheumatol 2008;35(1): (115) Mease PJ, Cohen S, Gaylis NB, Chubick A, Kaell AT, Greenwald M, et al. Efficacy and safety of retreatment in patients with rheumatoid arthritis with previous inadequate response to tumor necrosis factor inhibitors: results from the SUNRISE trial. J Rheumatol 2010;37(5): (116) Rubbert-Roth A, Tak PP, Zerbini C, Tremblay JL, Carreno L, Armstrong G, et al. Efficacy and safety of various repeat treatment dosing regimens of rituximab in patients with active rheumatoid arthritis: results of a Phase III randomized study (MIRROR). Rheumatology (Oxford) 2010;49(9): (117) Owczarczyk K, Hellmann M, Fliedner G, Rohrs T, Maizus K, Passon D, et al. Clinical outcome and B cell depletion in patients with rheumatoid arthritis receiving rituximab monotherapy in comparison with patients receiving concomitant methotrexate. Ann Rheum Dis 2008;67(11): (118) Valleala H, Korpela M, Mottonen T, Hienonen-Kempas T, Kauppi M, Hannonen P, et al. Rituximab therapy in patients with rheumatoid arthritis refractory or with contraindication to anti-tumour necrosis factor drugs: real-life experience in Finnish patients. Scand J Rheumatol 2009;38(5): (119) Henes JC, Schedel J, Kanz L, Koetter I. Rituximab and concomitant leflunomide for the treatment of rheumatoid arthritis. Rheumatol Int 2010;30(5): (120) Solau-Gervais E, Prudhomme C, Philippe P, Duhamel A, Dupont-Creteur C, Legrand JL, et al. Efficacy of rituximab in the treatment of rheumatoid arthritis. Influence of serologic status, coprescription of methotrexate and prior TNFalpha inhibitors exposure. Joint Bone Spine (121) Bokarewa M, Lindholm C, Zendjanchi K, Nadali M, Tarkowski A. Efficacy of anti- CD20 treatment in patients with rheumatoid arthritis resistant to a combination of methotrexate/anti-tnf therapy. Scand J Immunol 2007;66(4): (122) Finckh A, Ciurea A, Brulhart L, Kyburz D, Moller B, Dehler S, et al. B cell depletion may be more effective than switching to an alternative anti-tumor necrosis factor agent in rheumatoid arthritis patients with inadequate response to anti-tumor necrosis factor agents. Arthritis Rheum 2007;56(5): (123) Jois RN, Masding A, Somerville M, Gaffney K, Scott DG. Rituximab therapy in patients with resistant rheumatoid arthritis: real-life experience. Rheumatology (Oxford) 2007;46(6): (124) McGonagle D, Tan AL, Madden J, Taylor L, Emery P. Rituximab use in everyday clinical practice as a first-line biologic therapy for the treatment of DMARDresistant rheumatoid arthritis. Rheumatology (Oxford) 2008;47(6): (125) Brulhart L, Ciurea A, Finckh A, Notter A, Waldburger JM, Kyburz D, et al. Efficacy of B cell depletion in patients with rheumatoid arthritis refractory to anti- 237

238 tumour necrosis factor alpha agents: an open-label observational study. Ann Rheum Dis 2006;65(9): (126) Scheinberg M, Hamerschlak N, Kutner JM, Ribeiro AA, Ferreira E, Goldenberg J, et al. Rituximab in refractory autoimmune diseases: Brazilian experience with 29 patients ( ). Clin Exp Rheumatol 2006;24(1):65-9. (127) Finckh A, Ciurea A, Brulhart L, Moller B, Walker UA, Courvoisier D, et al. Which subgroup of patients with rheumatoid arthritis benefits from switching to rituximab versus alternative anti-tumour necrosis factor (TNF) agents after previous failure of an anti-tnf agent? Ann Rheum Dis 2010;69(2): (128) Tsiakalos AP, Avgoustidis NK, Moutsopoulos HM. Rituximab therapy in Greek patients with rheumatoid arthritis. Biologics 2008;2(4): (129) Henes JC, Richter C, Kanz L, Koetter I. B-cell depletion in patients with rheumatoid arthritis refractant to multiple TNF blockers and the interleukin 1 receptor-antagonist anakinra : Good responses in an extreme negative selection. Rheumatol Int 2007;28(1):33-7. (130) Yin HL, Chiu MC, Hung TC, Mak A, Yuk CM, Lung YK. Rituximab for refractory rheumatoid arthritis: a 24-week open-label prospective study. Open Rheumatol J 2007;1:1-4. (131) Assous N, Gossec L, Dieude P, Meyer O, Dougados M, Kahan A, et al. Rituximab therapy in rheumatoid arthritis in daily practice. J Rheumatol 2008;35(1):31-4. (132) Galarza C, Valencia D, Tobon GJ, Zurita L, Mantilla RD, Pineda-Tamayo R, et al. Should Rituximab be Considered as the First-Choice Treatment for Severe Autoimmune Rheumatic Diseases? Clin Rev Allergy Immunol 2008;34(1): (133) Vital EM, Dass S, Rawstron AC, Buch MH, Goeb V, Henshaw K, et al. Management of nonresponse to rituximab in rheumatoid arthritis: predictors and outcome of re-treatment. Arthritis Rheum 2010;62(5): (134) Vander CB, Durez P, Westhovens R, Kaiser MJ, Hoffman I, De KF. The Belgian MIRA (MabThera In Rheumatoid Arthritis) registry: clues for the optimization of rituximab treatment strategies. Arthritis Res Ther 2010;12(5):R169. (135) Quartuccio L, Fabris M, Salvin S, Atzeni F, Saracco M, Benucci M, et al. Rheumatoid factor positivity rather than anti-ccp positivity, a lower disability and a lower number of anti-tnf agents failed are associated with response to rituximab in rheumatoid arthritis. Rheumatology (Oxford) 2009;48(12): (136) Chatzidionysiou K, Lie E, Nasonov E, Lukina G, Hetland ML, Tarp U, et al. Highest clinical effectiveness of rituximab in autoantibody-positive patients with rheumatoid arthritis and in those for whom no more than one previous TNF antagonist has failed: pooled data from 10 European registries. Ann Rheum Dis 2011;70(9):

239 (137) Vital EM, Rawstron AC, Dass S, Henshaw K, Madden J, Emery P, et al. Reduceddose rituximab in rheumatoid arthritis: efficacy depends on degree of B cell depletion. Arthritis Rheum 2011;63(3): (138) Assous N, Gossec L, Dougados M, Kahan A, Allanore Y. Efficacy of rituximab in patients with rheumatoid arthritis refractory or with contra-indication to antitumor necrosis factor-alpha drugs in daily practice: an open label observational study. Clin Exp Rheumatol 2007;25(3):504. (139) Popa C, Leandro MJ, Cambridge G, Edwards JC. Repeated B lymphocyte depletion with rituximab in rheumatoid arthritis over 7 yrs. Rheumatology (Oxford) 2007;46(4): (140) Thurlings RM, Vos K, Gerlag DM, Tak PP. Disease activity-guided rituximab therapy in rheumatoid arthritis: The effects of re-treatment in initial nonresponders versus initial responders. Arthritis Rheum 2008;58(12): (141) Devine EB, fonso-cristancho R, Sullivan SD. Effectiveness of biologic therapies for rheumatoid arthritis: an indirect comparisons approach. Pharmacotherapy 2011;31(1): (142) Papanikolaou PN, Christidi GD, Ioannidis JP. Comparison of evidence on harms of medical interventions in randomized and nonrandomized studies. CMAJ 2006;174(5): (143) Hyrich KL, Watson KD, Isenberg DA, Symmons DP. The British Society for Rheumatology Biologics Register: 6 years on. Rheumatology (Oxford) 2008;47(10): (144) Silman A, Symmons D, Scott DG, Griffiths I. British Society for Rheumatology Biologics Register. Ann Rheum Dis 2003;62 Suppl 2:ii28-ii29. (145) Watson K, Symmons D, Griffiths I, Silman A. The British Society for Rheumatology biologics register. Ann Rheum Dis 2005;64 Suppl 4:iv42-iv43. (146) National Institute for Health and Clinical Effectiveness. Guidance on the use of etanercept and infliximab for the treatment of rheumatoid arthritis [cited 2011 Jul 23];Available from: URL: PDF.pdf (147) Hyrich KL, Watson KD, Silman AJ, Symmons DP. Predictors of response to anti- TNF-alpha therapy among patients with rheumatoid arthritis: results from the British Society for Rheumatology Biologics Register. Rheumatology (Oxford) 2006;45(12): (148) National Institute for Health and Clinical Effectiveness. Adalimumab, etanercept and infliximab for the treatment of rheumatoid arthritis [cited 2011 Jul 23];Available from: URL: (149) Maini RN, Breedveld FC, Kalden JR, Smolen JS, Davis D, Macfarlane JD, et al. Therapeutic efficacy of multiple intravenous infusions of anti-tumor necrosis 239

240 factor alpha monoclonal antibody combined with low-dose weekly methotrexate in rheumatoid arthritis. Arthritis Rheum 1998;41(9): (150) Klareskog L, Van Der Heijde D., De Jager JP, Gough A, Kalden J, Malaise M, et al. Therapeutic effect of the combination of etanercept and methotrexate compared with each treatment alone in patients with rheumatoid arthritis: double-blind randomised controlled trial. Lancet 2004;363(9410): (151) Hyrich KL, Symmons DP, Watson KD, Silman AJ. Comparison of the response to infliximab or etanercept monotherapy with the response to cotherapy with methotrexate or another disease-modifying antirheumatic drug in patients with rheumatoid arthritis: results from the British Society for Rheumatology Biologics Register. Arthritis Rheum 2006;54(6): (152) De Stefano R., Frati E, Nargi F, Baldi C, Menza L, Hammoud M, et al. Comparison of combination therapies in the treatment of rheumatoid arthritis: leflunomide-anti-tnf-alpha versus methotrexate-anti-tnf-alpha. Clin Rheumatol 2010;29(5): (153) Strangfeld A, Hierse F, Kekow J, von Hinueber U., Tony HP, Dockhorn R, et al. Comparative effectiveness of tumour necrosis factor alpha inhibitors in combination with either methotrexate or leflunomide. Ann Rheum Dis 2009;68(12): (154) Godinho F, Godfrin B, El Mahou S., Navaux F, Zabraniecki L, Cantagrel A. Safety of leflunomide plus infliximab combination therapy in rheumatoid arthritis. Clin Exp Rheumatol 2004;22(3): (155) Hansen KE, Cush J, Singhal A, Cooley DA, Cohen S, Patel SR, et al. The safety and efficacy of leflunomide in combination with infliximab in rheumatoid arthritis. Arthritis Rheum 2004;51(2): (156) Kalden JR, Nusslein HG, Wollenhaupt J, Burmester GR, Kruger K, Antoni C. Combination treatment with infliximab and leflunomide in patients with active rheumatoid arthritis: safety and efficacy in an open-label clinical trial. Clin Exp Rheumatol 2008;26(5): (157) Kiely PD, Johnson DM. Infliximab and leflunomide combination therapy in rheumatoid arthritis: an open-label study. Rheumatology (Oxford) 2002;41(6): (158) Alcorn N, Saunders S, Madhok R. Benefit-risk assessment of leflunomide: an appraisal of leflunomide in rheumatoid arthritis 10 years after licensing. Drug Saf 2009;32(12): (159) Bingham SJ, Buch MH, Kerr MA, Emery P, Valadao Barcelos AT. Induction of antinuclear antibodies in patients with rheumatoid arthritis treated with infliximab and leflunomide. Arthritis Rheum 2004;50(12): (160) Combe B, Codreanu C, Fiocco U, Gaubitz M, Geusens PP, Kvien TK, et al. Etanercept and sulfasalazine, alone and combined, in patients with active 240

241 rheumatoid arthritis despite receiving sulfasalazine: a double-blind comparison. Ann Rheum Dis 2006;65(10): (161) Sato T, Matsuyama Y. Marginal structural models as a tool for standardization. Epidemiology 2003;14(6): (162) Little RJ, Rubin DB. Causal effects in clinical and epidemiological studies via potential outcomes: concepts and analytical approaches. Annu Rev Public Health 2000;21: (163) Lunt M, Solomon D, Rothman K, Glynn R, Hyrich K, Symmons DP, et al. Different methods of balancing covariates leading to different effect estimates in the presence of effect modification. Am J Epidemiol 2009;169(7): (164) Symmons DP. Environmental factors and the outcome of rheumatoid arthritis. Best Pract Res Clin Rheumatol 2003;17(5): (165) Blom M, Kievit W, Donders AR, den Broeder AA, Straten VH, Kuper I, et al. Effectiveness of a Third Tumor Necrosis Factor-{alpha}-blocking Agent Compared with Rituximab After Failure of 2 TNF-blocking Agents in Rheumatoid Arthritis. J Rheumatol

242 Appendices Appendix 2.1: Search strategy used to identify studies that looked at the persistence with the anti-tnf in RA Search 1 rheumatoid arthritis.mp. or Arthritis, Rheumatoid/ 2 TNF.mp 3 persistence.mp 4 Survival Analysis/ or Survival/ or survival.mp. 5 Tumour Necrosis Factor-alpha/ or anti-tnf$.mp. 6 infliximab.mp. 7 etanercept.mp. 8 adalimumab.mp. 9 8 or 6 or 7 or 2 or or and 10 and 9 12 limit 11 to english language 13 limit 12 to human 14 limit 13 to yr=" " 15 remove duplicates from

243 Appendix 2.2: Search strategy used to identify RCTs of RTX in RA Search Cochrane library #1 MeSH descriptor Arthritis, Rheumatoid explode all trees #2 (rheumatoid arthritis) #3 (#1 OR #2) #4 (rituximab):ti,ab,kw #5 (rituxan):ti,ab,kw #6 (mabthera):ti,ab,kw #7 (#4 OR #5 OR #6) #8 (#3 AND #7) Ovid Medline 1 exp Arthritis, Rheumatoid/ 2 rituximab.mp. 3 rituxan.mp. 4 mabthera.mp. 5 random$.mp. 6 4 or 3 or and and 5 9 limit 8 to (english language and humans and yr=" ") Embase 1 exp Rheumatoid Arthritis/ 2 rituximab.ti,tw,ab. 3 rituxan.ti,tw,ab. 4 mabthera.ti,tw,ab. 5 2 or 4 or and 5 7 random$.mp. 8 6 and 7 9 limit 8 to (human and english language and yr=" ") 243

244 Appendix 2.3: Search strategy used to identify observational studies of RTX in RA Search Ovid Medline 1 exp Arthritis, Rheumatoid/ 2 rituximab.mp. 3 rituxan.mp. 4 mabthera.mp. 5 4 or 3 or and 5 7 observational study.mp. 8 Cohort Studies/ or cohort.mp. 9 Prospective Studies/ or prospective.mp. 10 Retrospective Studies/ or retrospective.mp. 11 follow up.mp. 12 clinical practice.mp or 11 or 7 or 10 or 9 or and limit 14 to (english language and humans and yr=" ") Embase 1 exp Rheumatoid Arthritis/ 2 rituximab.ti,tw,ab. 3 rituxan.ti,tw,ab. 4 mabthera.ti,tw,ab. 5 2 or 4 or and 5 7 observational.mp. 8 cohort.mp. 9 prospective.mp. 10 retrospective.mp. 11 follow up.mp. 12 clinical practice.mp or 11 or 7 or 10 or 9 or and limit 14 to (human and english language and yr=" ") 244

245 Appendix 4.1: The original ethical approval for the BSRBR 245

246 246

247 247

248 Appendix 4.2: Extension of ethical approval for recruitment of RTX patients 248

249 249

250 Appendix 4.3: Consultant baseline questionnaire 250

251 251

252 252

253 253

254 254

255 255

256 256

257 Appendix 4.4: Prior biologic exposure form 257

258 Appendix 4.5: Short baseline questionnaire 258

259 Appendix 4.6: Patient baseline questionnaire 259

260 260

261 261

262 Appendix 4.7: Health assessment questionnaire (HAQ) 262

263 263

264 Appendix 4.8: European quality of life 5 dimension (EQ-5D) form 264

265 265

266 Appendix 4.9: The consultant follow-up questionnaire 266

267 267

268 268