MabThera (rituximab) NICE STA Submission

Transcription

1 MabThera (rituximab) NICE STA Submission ACHIEVING CLINICAL EXCELLENCE IN THE TREATMENT OF RHEUMATOID ARTHRITIS Roche Submission to the National Institute for Health and Clinical Excellence 21 st November 2006 Page 1 of 181

2 Contents 1. Description of technology under assessment 3 2. Statement of the decision problem 8 3. Executive summary Context Clinical evidence Cost effectiveness Assessment of factors relevant to the NHS and other parties References Appendices 147 Page 2 of 181

3 Section A 1. Description of technology under assessment 1.1. Give the brand name, approved name and, where appropriate, therapeutic class. For devices please provide details of any different versions of the same device. Brand name: MabThera Approved name: rituximab Therapeutic class: antineoplastic agents 1.2. Does the technology have a UK marketing authorisation/ce marking for the indications detailed in this submission? If so, please give the date on which authorisation was received. If not, please state current UK regulatory status, with relevant dates (for example, date of application and/or expected approval dates). Rituximab was granted marketing authorisation by the European Medicines Evaluation Agency (EMEA) on 6th July 2006, stating that: rituximab in combination with methotrexate is indicated for the treatment of adult patients with severe active RA who have had an inadequate response or intolerance to other disease-modifying antirheumatic drugs including one or more anti-tumour necrosis factor inhibitor therapies What are the (anticipated) indication(s) in the UK? For devices, please provide the (anticipated) CE marking, including the indication for use. The licensed indication for rituximab is in combination with methotrexate for the treatment of adult patients with severe active RA who have had an inadequate response or intolerance to other disease-modifying anti-rheumatic drugs including one or more anti-tumour necrosis factor inhibitor therapies. The above indication is the result of the first stage of the development programme. The second stage of the phase III programme is currently underway assessing the safety and efficacy of rituximab in the management of signs and symptoms and disease progression in patients who are Disease-Modifying Anti-Rheumatic Drugs (DMARD) naïve and in those patients who have failed to respond to DMARDs such as methotrexate, sulphasalazine and leflunomide. Page 3 of 181

4 The overall aim is to deliver a file for consideration by the regulatory authorities in It is anticipated, subject to the product profile that the file will request the following licensed indications for rituximab in combination with methotrexate For the treatment of moderate to severe, active rheumatoid arthritis in adult patients when the response to disease-modifying anti-rheumatic drugs including methotrexate has been inadequate For the treatment of severe, active and progressive rheumatoid arthritis in adults not previously treated with methotrexate To reduce the rate of progression of joint damage as measured by X-ray and to improve physical function, when given in combination with methotrexate To what extent is the technology currently being used in the NHS for the proposed indication? Include details of use in ongoing clinical trials. If the technology has not been launched, please supply the anticipated date of availability in the UK. For the current licence (TNF inhibitor failures), patient numbers are approximately 500. This however is rising due to the licensed availability of rituximab and is expected to be in the region of 800 by the end of 2006 The number of patients currently in clinical trials within the phase III program described above (DMARD naïve and DMARD IR) is ~60. As discussed previously it is anticipated that this trial programme will deliver a file in 2008 with licence from the EU in early Does the technology have regulatory approval outside the UK? If so, please provide details. The marketing authorisation for rituximab was granted by the EMEA and covers all European Countries. In addition, rituximab is approved in this indication in a large number of other countries throughout the world Is the technology subject to any other form of health technology assessment in the UK? If so, what is the timescale for completion? Rituximab in RA has been assessed by the Scottish Medicines Consortium. Their final SMC Detailed Advice Document was made available to the public via the SMC website Page 4 of 181

5 on Monday 13th November Advice was scheduled to be issued to NHS Scotland on October 5 th For pharmaceuticals, what formulation(s) (for example, ampoule, vial, sustained-release tablet, strength(s) and pack size(s) will be available? Mabthera 100mg concentrate for solution for infusion, pack of 2 vials Mabthera 500mg concentrate for solution for infusion, pack of 1 vial 1.8. What is the proposed course of treatment? For pharmaceuticals, list the dose, dosing frequency, length of course and anticipated frequency of repeat courses of treatment. A course of rituximab consists of two 1000mg IV infusions. The recommended dosage of rituximab is 1000mg by IV infusion followed by a second 1000mg IV infusion two weeks later. Repeat treatment with subsequent courses of rituximab is a clinical decision based on signs and symptoms of RA. On average, the repeat treatment interval is 6-12 months in the majority of patients, but more than a third do not require re-treatment for more than a year (Data on File MabRA010). The treatment interval between first and second treatment courses is similar to that between second and third treatment courses. First to second course interval: mean 33.2 weeks, median 30.9 weeks (min, max: ); second to third treatment interval: mean 32.2 weeks, median 30.1 weeks (min, max: ) (Data on File MabRA011). Rituximab should not be given at an interval of less than 16 weeks What is the acquisition cost of the technology (excluding VAT)? For devices, provide the list price and average selling price. If the unit cost of the technology is not yet known, please provide details of the anticipated unit cost, including the range of possible unit costs. 1 course of MabThera (2x 1000mg) = 3, Therefore, assuming a repeat treatment every 9 months the average annual drug cost = 4,657. TNF inhibitor drug costs are also listed below for comparison. Page 5 of 181

6 Comparison of Drug Costs over 5 years Treatment Assumptions Summary Reference rituximab Dose 1000mg on day 1 and day 15 SPC 1 Dosing Schedule Average time to repeat treatment is 9 months SPC 1 Cost 50ml vial contains MIMS *500mg vials are required to get 1000mg dose on day 1 and day 15 Day 1: 1,746.30, Day 15: 1, Cost per cycle = 3, Annual cost, 5 year average = 4,890 Treatment Assumptions Summary Reference Etanercept Dose 25mg twice weekly OR 50mg MIMS 2 once weekly sub-cut injection Dosing Schedule Twice weekly OR Once weekly MIMS 2 Cost 50mg MIMS 2 Weekly Monthly cost = Annual cost, 5 year average = 9,295 Treatment Assumptions Summary Reference Adalimumab Dose 40mg sub-cut injection MIMS 2 Dosing Schedule Every other week MIMS 2 Cost 40mg MIMS 2 Monthly cost = 715 Annual cost, 5 year average = 9,295 Treatment Assumptions Summary Reference Infliximab Dose 3mg/kg by IV infusion, MIMS 2 Assume average patient weight of 75kg Average patient dose = 225mg Dosing Schedule Repeated 2 weeks and 6 weeks after first infusion, then every 8 weeks MIMS 2 Cost 100mg Require 3 vials per 225mg dose Cost per dose = 1, Monthly cost = 839 Annual cost, 5 year average = 8,560 MIMS 2 Page 6 of 181

7 The annual costs above are averaged over a 5 year period to account for the different treatment schedules of each treatment and loading doses at the beginning of the treatment What is the setting for the use of the technology? In hospital as a day case procedure, under the guidance of a clinician trained in rheumatology For patients being treated with this technology, are there any other aspects that need to be taken into account? For example, are there additional tests or investigations needed for selection, or particular administration requirements, or is there a need for monitoring of patients over and above usual clinical practice for this condition? What other therapies, if any, are likely to be administered at the same time as the intervention as part of a course of treatment? There are no additional tests or investigations needed (other than those required in usual clinical practice for infusion of a biologic) prior to initiation of rituximab. In clinical trials, patients received subsequent courses of rituximab as needed according to the treating clinician s assessment of disease activity and irrespective of the peripheral B lymphocyte count.) Rituximab is given in combination with methotrexate (dose determined by treating clinician). Patients should receive treatment with 100mg IV methylprednisolone 30 minutes prior to rituximab infusion along with an anti histamine and anti pyretic. Background therapy with glucocorticoids, salicylates, nonsteroidal anti-inflammatory drugs or analgesics can be continued during rituximab treatment. Page 7 of 181

8 2. Statement of the decision problem Population Intervention Comparator(s) Final scope issued by NICE Adults with severe active rheumatoid arthritis who have had an inadequate response to or intolerance of other DMARDS including one or more TNF alpha inhibitors. rituximab in combination with methotrexate Management strategies without rituximab, for example: Alternative DMARDs, including one or more TNF alpha inhibitors (such as adalimumab, etanercept and infliximab) Decision problem addressed in the submission As per final scope As per final scope rituximab is licensed for the management of severe active RA in patients who have had an inadequate response or intolerance to one or more TNF inhibitors. There are therefore two potential groups of pharmacological comparators depending on whether current NICE guidance on the use of infliximab and etanercept is considered. 1. Return to DMARDs The first potential comparator is where NICE guidance on sequential use is adhered to. Patients who have failed to respond to an anti TNFdue to inefficacy are severely limited in their therapeutic options and in these situations patients are returned on to non biological DMARDs such as leflunomide or combinations of DMARDs. This occurs in a considerable population of patients. Possible treatment sequence following TNF inhibitor failure (adalimumab assumed tobe used as first TNF inhibitor): 1. Leflunomide 2. Gold Page 8 of 181

9 Outcomes The outcome measures to be considered include: physical function joint damage pain fatigue adverse effects of treatment 3. Cyclosporine 4. Palliative care (MTX) The use of non disease modifying interventions (e.g NSAIDs, COX IIs and surgery) occurs primarily in tandem with disease modifying pharmacological interventions and as such Roche do not deem these to be a valid comparator when considered alone. 2. TNF inhibitors used sequentially The second potential comparator is where TNF inhibitors are used sequentially. It is recognised that a large number of rheumatologists are sequentially using TNF inhibitors in patients whose response to the first anti TNF has been inadequate due to inefficacy. This of course contradicts current NICE guidance. Therefore, this current practice suggests that one potential comparator for consideration when reviewing rituximab is the use of a second-line anti TNF. Many rheumatologists believe that this is the best alternative care. Possible treatment sequence following TNF inhibitor failure (adalimumab assumed to be used as first TNF inhibitor): 1. Adalimumab 2. Infliximab 3. Leflunomide 4. Gold 5. Cyclosporine 6. Palliative care (MTX) As well as the stated outcome measures, the inhibition of disease progression will be considered and evaluted in its own right. Specific outcome measures highlighted will be American College of Rheumatology (ACR) scores, Disease Activity Scores (DAS), EULAR scores, Health Assessment Questionaire (HAQ) score, Fatigue (FACIT-F) score, Short Page 9 of 181

10 Other considerations mortality health-related quality of life. If the evidence allows, the appraisal will attempt to identify criteria for selecting patients for whom rituximab would be particularly appropriate. The intervention will be appraised according to its marketing authorisation. Form (SF-36) scores and the Sharp radiographic assessment scores. Currently no predictors of response are known Page 10 of 181

11 Section B 3. Executive summary Background Rheumatoid Arthritis (RA) is the most common inflammatory polyarthropathy in the UK, and affects between 0.5% and 1% of the population. Estimates in western populations indicate an annual incidence of 0.5 per 1000 population and a prevalence of 8 per 1000 population. 3 RA is a chronic, progressive, destructive and disabling condition with significant morbidity and mortality, impacts severely on quality of life and represents a considerable economic burden. 4, RA affects all aspects of patients lives from education and employment through to family and social life. Life expectancy in RA patients can be reduced by around 3-18 years 5. The mainstay of RA treatment intervention to date involves the use of disease modifying anti-rheumatic drugs (DMARDs). Additionally in 1999, the first TNF inhibitor therapies were introduced, however their effectiveness is often lost after a period of time or patients suffer treatment induced side effects resulting in discontinuation. This can happen in approximately 25-50% of all patients treated. A significant patient population therefore exists with severe disease which has failed TNF inhibitor therapy. Currently, NICE does not recommend the sequential use of TNF inhibitors due to lack of evidence and consequently there is a significant unmet need within this patient population for a clinically and cost effective treatment. Rituximab Brand Name: MabThera, Approved name: rituximab. Rituximab is a monoclonal antibody that depletes the CD20+ B cells implicated in the immunopathogenesis of RA. Indication: rituximab was granted marketing authorisation on 6 th July 2006, with a licence stating that: rituximab in combination with methotrexate is indicated for the treatment of adult patients with severe active RA who have had an inadequate response or intolerance to other disease-modifying anti-rheumatic drugs including one or more anti-tumour necrosis factor inhibitor therapies. Rituximab is available in the following formulations and pack sizes: Rituximab 100mg concentrate for solution for infusion, pack of 2 vials Page 11 of 181

12 Rituximab 500mg concentrate for solution for infusion, pack of 1 vial A course of rituximab consists of two 1000mg IV infusions. The recommended dosage of rituximab is 1000mg by IV infusion followed by a second 1000mg IV infusion two weeks later. Repeat treatment with subsequent courses of rituximab is a clinical decision based on signs and symptoms of RA. On average, the repeat treatment interval is 6-12 months in the majority of patients, but more than a third do not require re-treatment for more than a year. 6 The mean time to repeat treatment observed in the phase III randomised control trial was 301 days. Rituximab should not be given at an interval of less than 16 weeks. 1 course of rituximab (2x 1000mg) = 3, Therefore, assuming a repeat treatment every 9 months (mid point of SPC) the average annual drug cost = 4,657, pro rata over 3 years. Currently NICE guidance prohibits the further use of anti-tnfs following an inadequate response to a prior anti-tnf. Consequently, a traditional DMARD such as leflunomide or methotrexate may be considered an appropriate comparator for rituximab. Randomised head to head data exists for such a comparison of methotrexate with rituximab. However it may be argued that current UK practice involves the sequential use of anti-tnf therapy, in which case such treatment may be considered an appropriate comparator. No head to head data exists for such a comparison with anti- TNFs; however published methods of indirect comparisons are presented in our submission, providing strong evidence of comparable efficacy with existing anti-tnf therapy at a substantially lower cost to the NHS. Clinical Effectiveness Randomised controlled clinical trials of rituximab with methotrexate monotherapy as a comparator have clearly demonstrated a significant impact on both disease activity measures and on patient quality of life. The key results at 24 weeks from the rituximab clinical development programme show a 51% improvement in ACR20 with rituximab versus placebo. This is translated into a patient benefit with a significant drop in HAQ- DI and fatigue which is maintained out to 48 weeks. This is the first time this order of Page 12 of 181

13 magnitude of benefit has been observed in randomised control trials in such a severe RA population. Furthermore, these trials have been performed in a population of patients who have had an inadequate response or intolerance to one or more TNF inhibitor therapies. The evidence base generated therefore represents a landmark in the management of this patient group. The efficacy of rituximab has been shown to be maintained with repeated courses with no evidence of loss of response. Although the use of rituximab in RA is new, the drug has also been used highly effectively in the management of non-hodgkin s Lymphoma (NHL) since 1998 and over 730,000 patients have been treated to date. The safety profile of rituximab in RA is comparable to that seen with the TNF inhibitors. The majority of adverse events are related to infusion reactions which tend to be mild to moderate. Cost Effectiveness A micro simulation markov model was constructed based upon the phase III randomised control trial of rituximab to estimate the incremental costs and QALYs from the introduction of rituximab in its licensed indication compared to current standard of care. The model structure and design reflected existing published examples of economic evaluations in RA (Bansback 2004, Kobelt 1999, Barton 2004) as utilised in previous NICE appraisals. 7,8,9 Two separate cost effectiveness scenarios were presented, one reflecting current NICE guidance and a second scenario assuming sequential use of anti-tnf therapies: Scenario 1 (NICE recommended treatment) Scenario 2 (Sequential use of anti-tnfs) Intervention sequence rituximab + MTX Leflunomide + MTX Gold Ciclosporine MTX / Palliative Care Comparator sequence Leflunomide + MTX Gold Ciclosporine MTX / Palliative Care Intervention sequence rituximab + MTX Adalimumab + MTX Infliximab + MTX Leflunomide + MTX Gold Ciclosporine MTX / Palliative Care Comparator sequence Adalimumab + MTX Infliximab + MTX Leflunomide + MTX Gold Ciclosporine MTX / Palliative Care Compared to current NICE recommended treatment (scenario 1) the economic model estimated that the addition of rituximab to the existing treatment strategy generates an additional 10,675 in NHS costs and provides an additional QALYS per patient. In scenario 2 rituximab generates an additional 6,103 and QALYs. Consequently, the incremental cost effectiveness ratios (ICER) for rituximab are Page 13 of 181

14 14,690 and 11,601 in scenarios 1 and 2 respectively. Probabilistic sensitivity analysis illustrated that rituximab had an ICER below 30,000 in over 88% of model iterations performed. Budget impact for the NHS The budget impact estimates the cost to the NHS in England and Wales of adding rituximab to the current anti-tnf treatment sequence. The cost to the NHS of implementing rituximab is estimated by modelling patients sequencing through the currently available anti-tnf treatments. This cost is then offset against the cost of treating the same patient pool following the addition of rituximab to this pre-defined treatment sequence. The introduction of rituximab to a treatment sequence including anti-tnf therapy generates cost savings to the NHS over the first five years. Assuming all eligible patients with an inadequate response to an anti-tnf receive rituximab these savings in drug acquisition costs range from 5,468,108 in the first year to 11,865,311 in the fifth year. It is also estimated that resource service impact of the addition of rituximab to the anti- TNF treatment sequence may also results in net resource savings in terms of hospital nurse time, pharmacist time and community nurse time. This is due to the existing IV infusion biologic DMARD, infliximab, being more resource intensive due to its frequency of administration compared to rituximab. Page 14 of 181

15 4. Context 4.1. Please provide a brief overview of the disease/condition for which the technology is being used. Provide details of the treatment pathway and current treatment options at each stage. Disease background Rheumatoid Arthritis (RA) is the most common inflammatory polyarthropathy in the UK, and affects between 0.5% and 1% of the population. Estimates in Western populations indicate an annual incidence of 0.5 per 1000 population and a prevalence of 8 per 1000 population. 3 It is a chronic, progressive, destructive and disabling condition that carries a significant morbidity and mortality rate, impacts severely on quality of life, and represents a considerable economic burden. RA affects all aspects of life from education and employment through to family and social life. It is estimated that 40% of people with RA stop working within 5 years of diagnosis million working days were lost in Great Britain because of RA in This represents a loss in productivity of 833 million. 11 There were 45,887 hospital admissions in England for inflammatory arthritis in RA typically causes chronic joint inflammation that causes pain, stiffness and fatigue. Chronic synovial inflammation is associated with progressive joint damage that causes disability and reduced quality of life, and approximately 25 % of RA patients require major joint replacement surgery within 20 years of diagnosis. 4 Moreover, RA is a systemic disease with many extra-articular manifestations including vasculitis, pericarditis, pulmonary fibrosis, corneal melt and leg ulcers. In the past, it has not always been appreciated that RA is an important cause of morbidity and mortality. The mortality of patients with RA is increased when compared to the general population, 5 and it is estimated that the life expectancy of patients with RA is reduced by between 3-18 years. The causes of death mirror those observed in the general population, with the greatest excess of deaths in patients with RA being caused by cardiovascular disease, although there is a significantly increased risk of malignancy (particularly lymphoma) and respiratory disease compared to the general population. 12,13,14,15 Page 15 of 181

16 Management of RA is holistic and multidisciplinary, with physical therapy and surgical interventions running in parallel with drug treatment. Key aims of treatment include disease remission, controlling joint pain and inflammation, reducing joint damage, disability and loss of function, and maintaining and improving quality of life. BSR guidelines BSR guidelines for the management of RA have been published and recommend that the care of RA patients is delivered by a multidisciplinary team comprising the general practitioner (GP), rheumatologist, orthopaedic surgeon, nurse specialist, physiotherapist, occupational therapist, dietician, podiatrist, orthotist, pharmacist and social worker. 16 Early diagnosis and referral to secondary care services allows the prompt introduction of appropriate drug management with analgesics, non steroidal anti inflammatory drugs (NSAIDs) and disease modifying anti rheumatoid drugs (DMARDs). Current BSR guidelines state that patients with RA should be established on diseasemodifying therapy as soon as possible after a diagnosis of RA is confirmed and that disease-modifying therapy should be part of an aggressive package of care, incorporating escalating doses, intra-articular steroid injections, parenteral methotrexate and combination therapy, rather than sequential monotherapy, progressing to biologic therapy when required 17. Methotrexate has displaced sulphasalazine as the most commonly used first DMARD agent in the UK, and other DMARDs such as hydroxychloroquine, and leflunomide are also widely prescribed. Despite the early use of DMARDs, many patients have evidence of persisting disease activity, or encounter significant drug toxicity. In these patients, alternative DMARD therapy is used but a proportion of patients are intolerant of, or resistant to multiple conventional DMARDs. Since 2001, patients with active RA despite the use of at least two conventional DMARDs (including methotrexate) have been assessed for eligibility for treatment with TNF inhibitor therapy according to the British Society of Rheumatology guidelines. 16,18 These guidelines include eligibility criteria, contraindications to TNF inhibitor therapy (relative and absolute) and guidance on stopping therapy in the event of non-response. Briefly, they indicate that patients will be eligible for TNF inhibitor therapy if they have active RA (with a DAS28 > 5.1) following inadequate response to at least two DMARDs (including methotrexate) in the absence of any contra-indication. Treatment should be normally stopped if patients DAS28 fails to improve by > 1.2. Most patients respond very well to TNF inhibitor therapy initially, but approximately 25-50% fail to respond to treatment, lose response or suffer drug-related toxicity. These patients have the most Page 16 of 181

17 severe disease and are at the greatest risk of progressive joint damage, need for joint replacement surgery, progressive disability and increased mortality. Once patients have failed TNF inhibitor therapy there are very few therapeutic options that have been proven to be effective in randomised controlled trials. Some patients will be treated with a further TNF inhibitor drug, contrary to current NICE guidance; others will be given conventional DMARDs, often in unproven combinations. rituximab is the first licensed treatment option for patients who have failed TNF inhibitor therapy What was the rationale for the development of the new technology? Conventional drug therapy for RA relies on various combinations of non-steroidal antiinflammatory drugs (NSAIDs), analgesics, corticosteroids and disease-modifying antirheumatic drugs (DMARDs). Evidence suggests that patients with RA should be treated with DMARDs soon after diagnosis, as delaying treatment leads to worse outcomes, including progression of joint damage 6. The introduction of tumour necrosis factor alpha inhibitors (TNF inhibitors) represented a huge step forward in the treatment of RA. Etanercept and infliximab are currently recommended by NICE as options for treatment of patients who have continuing clinically active RA that has not responded adequately to at least 2 DMARDs including methotrexate (unless contraindicated). The guidance sets out criteria for use, including exclusion criteria and criteria for withdrawal of therapy. In particular, treatment should be withdrawn in the event of serious drug-related toxicity or because of lack of response at 3 months. However, approximately 25 50% of patients fail to respond to TNF inhibitors 4,5,6,12,13 thus creating a need for additional treatment options. Although this is a relatively small number, the clinical unmet need for this patient group is significant and as RA is a lifelong disease, its impact on work and health is immense. rituximab is one such additional option which targets B cells. This unique mechanism of action can elicit a clinically meaningful response in patients who have failed on multiple other treatments. 14, What is the principal mechanism of action of the technology? Although the precise pathogenesis of RA remains unclear, it has been postulated that multiple exogenous and/or endogenous antigenic triggers act in the presence of a background genetic predisposition to initiate a self-perpetuating series of autoimmune responses in the synovial compartment, leading to inflammation and eventual destruction of the joint. Many cell populations, including monocytes, macrophages, B Page 17 of 181

18 cells, T cells, endothelial cells and fibroblasts participate in the ongoing inflammatory process. 16 The precise contribution of B cells to the immunopathogenesis of RA is not fully understood but strong evidence for a critical role of B cells in RA initially came from a small study of rituximab in combination with cyclophosphamide and glucocorticoids. 17 Subsequent studies have reported results further supporting the critical role of B cells in the immunopathogenesis of RA. 14,15,16 rituximab selectively targets a subset of B cells that express CD20, leaving stem, pro-b and plasma cells unaffected. rituximab depletes peripheral B cells from the circulation by three distinct mechanisms: complement-mediated B cell lysis, cell-mediated cytotoxicity and induction of apoptosis. As expected for a human IgG1 antibody, rituximab directly binds the C1q complement component, initiating complement-mediated lysis of circulating B cells 19 rituximab binds strongly to Fc receptors on human effector cells, such as macrophages and natural killer cells, inducing antibody-dependent cellular cytotoxicity (ADCC) 20 rituximab has also been shown to induce apoptosis, or programmed cell death, in vitro 20 The engineered human component of the antibody is an essential feature in the efficacy of MabThera, with in vitro studies confirming that rituximab is significantly more effective at inducing ADCC and complement-mediated lysis than the parent murine antibody from which it is derived What is the suggested place for this technology with respect to treatments currently available for managing the disease/condition? As stated in Section 4.1, once patients have failed TNF inhibitor therapy there are very few therapeutic options that have been proven to be effective in randomised controlled trials. Some patients will be treated with a further TNF inhibitor drug, contrary to current NICE guidance; others will be given conventional DMARDs, often in combinations that have not been assessed in formal randomised controlled trials. A significant patient Page 18 of 181

19 population therefore exists with severe disease, who have failed anti TNF therapy and consequently there is a significant unmet clinical need within this patient population. rituximab (in combination with methotrexate) is the first licensed treatment option for patients who have failed TNF inhibitor therapy. Its place in therapy should be as per its licence: adult patients with severe, active RA who have had an inadequate response or intolerance to other disease modifying anti-rheumatic drugs including one or more tumour necrosis factor (TNF) inhibitor therapies. Clinical data show that the efficacy is greatest in this population of hard-to-treat patients if rituximab is used after the first TNF inhibitor failure Describe any issues relating to current clinical practice, including any variations or uncertainty about best practice. The main aim of management of RA is to obtain remission, control symptoms and signs of the disease and to maintain function. The BSR now recommend more aggressive therapy of early RA, with DMARD therapy as a mainstay of early intervention in RA. As part of this aggressive package of care, combination therapy, rather than sequential monotherapy, along with biologic (TNF inhibitor) therapy, (particularly in combination with methotrexate) is recommended, in accordance with NICE guidelines. Not all individuals respond to a given biologic agent and the data supporting clinical response from switching biologics is heterogeneous at best. It has been hypothesised that the biological differences between the three TNF inhibitors may be important with respect to efficacy, and may be a rationale for sequential use. The sequential use of TNF inhibitors is widespread although not supported by current NICE guidance, where failure to respond has been due to inefficacy. NICE have recently published an Appraisal Consultation Document on the use of adalimumab, etanercept and infliximab for the treatment of rheumatoid arthritis, which states that: The sequential use of TNFa inhibitors is only recommended in the event of toxicity developing to an individual agent. Sequential use on the basis of lack of response to an individual agent is not recommended outside the context of a study designed to generate robust and relevant outcome data, including the direct measurement of health-related quality of life and disease progression over long-term follow-up 211 At the time of writing however, the Final Appraisal Document has not been published. Page 19 of 181

20 However, in the public domain there is a significant lack of robust and relevant outcome data to support sequential use following failure due to inefficacy. As stated above, in cases of toxicity a second line TNF inhibitor may be considered, however with the advent of newer alternative agents with RCTs within this patient population, then this rationale may need to be updated. For example looking at the BSRBR data on the outcomes with second line TNF inhibitors, then if a patient fails a first line TNF inhibitor due to intolerance, the likelihood that patient will again fail because of recurrent toxicity is about two or three fold.5 Assuming current NICE guidance is adhered to, then leflunomide may also be an alternative therapy in the event of anti TNF failure. In respect to leflunomide s efficacy in TNF inhibitor failure patients this has not to our knowledge been investigated Provide details of any relevant guidelines or protocols. rituximab in RA has been assessed by the Scottish Medicines Consortium. Their final SMC Detailed Advice Document was made available to the public via the SMC website on Monday 13th November The SMC recommended that the use of rituximab be as per licence in the management of severe RA. The BSR will be reviewing the clinical application of rituximab in due course and has issued an interim statement recognising that there is clear evidence to support the use of rituximab as a treatment for adult patients with refractory RA, who have had an inadequate response to previous treatments including TNF alpha inhibitor. Whilst there have not been any direct comparisons with TNF inhibitor drugs in RA, rituximab appears to be as effective as other licensed biologic agents. Page 20 of 181

21 5. Clinical evidence 5.1. Identification of studies The following databases were used to identify relevant studies: Medline via Dialog DataStar. Medline 1993 to date (MEYY) and Medline-In process-latest eight weeks (MEIP) were searched on 17 October Medline was searched from 1993 to the present, and also the eight weeks prior to 17 October 2006 Embase via Dialog DataStar. Embase 1993 to date (EMYY) and Embase latest eight weeks (EMBA) were searched on 17 October Embase was searched from 1993 to the present, and also the eight weeks prior to 17 October 2006 The Cochrane Library, accessed via Wiley Interscience at www3.interscience.wiley.com. Searched on 17 October The Cochrane Library was searched with unrestricted dates up to 17 October 2006 EULAR abstracts (the European League against Rheumatism) (annual meetings ) searched via the website Searched on 19 October ACR (American College of Rheumatology) abstracts searched (annual meetings ) via the website Searched on 19 October and 6 th November 2006 Searches used index and text words which included rituximab and rheumatoid arthritis as major descriptors. The search was restricted to include only documents relating to humans and clinical trials. The search was further restricted manually according to inclusion/exclusion criteria in There were no restrictions by language. Details of the search strategies used are provided in Appendix 2, section Study selection Complete list of RCTs The main source of data in this submission is the REFLEX (Randomised Evaluation of Long-term Efficacy of rituximab in RA) pivotal Phase III study (WA17042), as published by Cohen et al. (2006), which evaluated the efficacy and safety of rituximab in patients Page 21 of 181

22 who had an inadequate response or were intolerant to TNF inhibitors (ie the licensed indication population). A further six abstracts have also been included, which include data from the open-label extension to WA RCT results 1. rituximab for rheumatoid arthritis refractory to anti-tumor necrosis factor therapy: Results of a multicenter, randomized, double-blind, placebo-controlled, phase III trial evaluating primary efficacy and safety at twenty-four weeks. Cohen SB, Emery P, Greenwald MW et al., Arthritis and Rheumatism 2006; 54 (9): Repeat treatment with rituximab improves physical function and quality of life in patients with rheumatoid arthritis and an inadequate response to TNF inhibitors. Tak PP, Mease P, Bombardieri S et al. Ann Rheum Dis 2006; 65(Suppl II): Long-term efficacy and safety of a repeat treatment course of rituximab in rheumatoid arthritis patients with an inadequate response to one or more TNF inhibitors. Keystone E, Fleischmann R, Emery P et al. Ann Rheum Dis 2006; 65 (Suppl II): Prolonged efficacy of rituximab in rheumatoid arthritis patients with inadequate response to one or more TNF inhibitors: 1-year follow-up of a subset of patients receiving a single course in a controlled trial (Cohen study). Cohen S, Emery P, Greenwald M et al. Ann Rheum Dis 2006; 65 (Suppl II): Prevention of joint structural damage at 1 year with rituximab in rheumatoid arthritis patients with an inadequate response to one or more TNF inhibitors (Cohen study). Keystone E, Emery P, Peterfy CG et al. Ann Rheum Dis 2006; 65 (Suppl II): Efficacy of rituximab in active RA patients with an inadequate response to one or more TNF inhibitors. Kremer JM, Tony H, Tak PP et al. Ann Rheum Dis 2006; 65 (Suppl II): rituximab improves health-related quality of life as measured by the SF-36: domain score results from the reflex study. Kielhorn S, Bombardieri R, Aultman F et al.. Ann Rheum Dis 2006;65(Suppl II):324 Page 22 of 181

23 Inclusion and exclusion criteria Inclusion criteria Published papers or abstracts which evaluated the following were included: rituximab had to be the major focus of the paper, in order to eliminate papers which mentioned rituximab as part of a discussion of treatments for rheumatoid arthritis Rheumatoid arthritis had to be a major focus of the paper, in order to eliminate papers covering the use of rituximab in other autoimmune diseases, NHL or other haematological malignancies RA efficacy endpoints should be used as the focus for the data Patient population should consist of those patients who had an inadequate response or intolerance to one or more TNF inhibitor therapies, to be consistent with the rituximab licence Clinical trial data rather than case reports etc Controlled studies Documents relating to humans since work in animal models is not relevant to this application Exclusion criteria Published papers or abstracts which evaluated the following were excluded: Any papers providing a review, update or commentary on data published elsewhere were excluded, as only current clinical trial data are required Animal studies or in vitro research only human data are required A further study, DANCER (Phase II, Dose-ranging Assessment: international Clinical Evaluation of MabThera in RA), protocol number WA17043, Emery et al, 2006, 22 has not been included in the main analysis. This was a Phase IIb, randomised, doubleblind, double-dummy, controlled, international, multifactorial study of nine different treatment regimens in a 3 x 3 configuration, that comprised three different dose levels Page 23 of 181

24 of rituximab (including placebo) and three different corticosteroid regimens (including placebo). Although 465 patients were randomized in the study, only 29% of patients had received prior TNF inhibitors and, due to the study design, only one third of these received the licensed dose of rituximab. However the ~10% of subjects in the WA17043 who had prior TNF inhibitor therapy and the licensed dose of rituximab were included in the safety analysis presented in Section 5.7 of this submission (Regulatory Summary data requested) and are also included in the pooled analysis of repeat treatment efficacy List of relevant RCTs There are no RCTs that compare rituximab directly with the appropriate comparators (second or third TNF inhibitor, or leflunomide) List of relevant non-randomised controlled trials None included Ongoing studies No further evidence is anticipated from the Roche sponsored clinical development program for rituximab in autoimmune diseases within the next 12 months. Page 24 of 181

25 Flow chart for study analysis 69 citations identified and retrieved by electronic database search 23 excluded on basis of title 36 citations obtained Citations obtained from conference abstracts n=56 85 citations excluded after examination of full text publications/ abstracts 1 citation 6 conference abstracts met inclusion criteria Reasons for exclusion Not RCT No RA endpoints Not controlled Not relevant patient population Repeated data Page 25 of 181

26 5.3. Summary of methodology of relevant RCTs Methods The clinical efficacy data for rituximab (2 x 1000 mg infusions) in combination with methotrexate (MTX) and pre-infusion corticosteroid therapy in rheumatoid arthritis (RA) derives from one pivotal, double-blind, randomised, phase III global study (WA17042) in RA patients who have had an inadequate response to: 1) at least one anti tumour necrosis factor (TNF inhibitor) therapy and 2) no more than five disease modifying antirheumatic drugs (DMARDs) 23. This details the initial 24-week data from an ongoing 2 year study. Additional information, including efficacy of repeated treatment with rituximab, is provided from the interim analyses of data from the following sources: patients remaining in the original study beyond week 24 open-label treatment under the extension protocol WA17531 providing the opportunity to administer further courses of rituximab treatment (2 x 1000 mg). A data cut-off of 14 th October 2005 has been used for all long-term data analyses. Pivotal Phase III Study (WA17042), Cohen et al, Study design (see Figure 1 below) WA17042 was a double-blind, placebo-controlled, randomised multinational study in patients with active RA who experienced an inadequate response to at least one TNF inhibitor therapy, for either lack of efficacy or lack of tolerability. Randomisation was stratified by region (US; non-us) and rheumatoid factor (RF) status (RF positive 20 IU/mL; RF negative, < 20 IU/mL) in a 3:2 (active:placebo) ratio. Patients received study drug on Days 1 and 15 (placebo or rituximab 1000 mg) with methylprednisolone 100 mg IV (pre-infusion at days 1 and 15) and oral prednisone (60 mg/day on days 2 through 7; 30 mg/day on days 8 through 14), for a total dose of 820 mg prednisoneequivalent in both arms. All patients received stable weekly doses of MTX (10-25 mg) and folate ( 5 mg/wk in single or divided dose). From weeks 16 to 24 inclusive, patients requiring additional treatment who showed less than a 20% improvement from baseline in both swollen joint count (SJC) and tender joint count (TJC) could enter an Page 26 of 181

27 open-label rescue arm and receive rituximab 2 x 1000 mg + MTX. Patients who did not fulfil the criteria for rescue could discontinue from the study and receive standard of care at the investigator s discretion. Patients who entered the rescue arm or who withdrew to receive standard of care for RA were considered non-responders for the primary analysis. After week 24 (primary time endpoint), patients who had achieved 20% improvement in both SJC and TJC at any visit after week 16, had completed the week 24 visit but who had active disease ( 8 SJC and 8 TJC) were eligible to receive further courses of treatment with rituximab in a dedicated open-label study (WA17531). Figure Participants Patients. Eligible adult patients were recruited from 114 rheumatology centres in the U.S., Europe, Canada, and Israel. Patients had RA for at least 6 months, according to the 1987 American College of Rheumatology (ACR) Criteria 24 ; and had active disease, defined as 8 swollen joints (66 joint count) and 8 tender joints (68 joint count), C- reactive protein (CRP) level 1.5 mg/dl or erythrocyte sedimentation rate (ESR) 28 mm/h, and radiographic evidence of at least one joint with a definite erosion attributable to RA, as determined by a central reading site. Eligible patients had received MTX (10-25 mg/week for at least 12 weeks) prior to screening, with the last 4 weeks at a stable dose. Page 27 of 181

28 Patients were excluded for a history of rheumatic autoimmune disease other than RA (except secondary Sjögren s syndrome); significant systemic involvement secondary to RA (vasculitis, pulmonary fibrosis, or Felty s syndrome); or ACR functional class IV disease. Enrolled patients had experienced an inadequate response to previous or current TNF inhibitor treatment: infliximab ( 3 mg/kg, at least 4 infusions), adalimumab (40 mg every other week for 3 months), or etanercept (25 mg twice weekly for 3 months), or were intolerant to at least one administration of these therapies. Patients discontinued etanercept for 4 weeks and infliximab or adalimumab for 8 weeks prior to randomization. Baseline characteristics: patients baseline characteristics are described in the table below (Table 1). Both study groups were balanced for age, gender, disease activity, and previous and concomitant treatments for RA. Patients had long-standing (~12 years) and active disease as determined by elevated inflammatory markers, disease activity (DAS28), disability, radiographic scores, and numbers of swollen and tender joints. A majority of patients (79%) in each group were RF positive at baseline. Most patients in both groups (89%) had received DMARDs other than MTX (defined in Table 1), and all patients had received TNF inhibitor agents. In both treatment groups, over 90% of all patients had previously taken 1 or 2 TNF inhibitor therapies, with 9% of patients receiving 3 TNF inhibitors. At entry, 91% of all patients demonstrated an inadequate response to TNF inhibitor therapy because of a lack of efficacy. Glucocorticoids, most commonly prednisone and triamcinolone, were used by 61% of placebo patients and 65% of rituximab patients at baseline. Page 28 of 181

29 Page 29 of 181

30 Patient numbers Of 751 patients screened, 520 were enrolled: 209 were randomized to placebo, and 311 to rituximab. Three patients were randomized but did not receive treatment. One patient randomized to rituximab received placebo instead, and was analyzed in the placebo group for safety but the rituximab group for efficacy. Patients receiving a portion of at least one infusion of placebo or rituximab were included in the safety analysis (209 patients in the placebo group and 308 in the rituximab group). A total of 21 patients were excluded from the ITT population, including those for whom treatment was unblinded due to rituximab vial breakage, those who never received treatment, those treated prior to randomization, and those enrolled at a centre where the blinding of the efficacy assessor was potentially compromised. Thus, the ITT population included 499 patients (201 in the placebo group and 298 in the rituximab group). Sensitivity analyses that included these patients demonstrated no change in the significance of the results. A total of 54% of the patients randomized to placebo and 82% of the patients randomized to rituximab completed the 24 weeks of study: 96 patients in the placebo group who received study medication withdrew prematurely, with 80 going on to receive rituximab rescue therapy and 16 entering the safety follow-up predominantly because of inadequate response. Of the 309 treated patients in the rituximab group, 53 Page 30 of 181

31 who received study medication withdrew early and entered the safety follow-up prior to week 24, 1 patient required rescue therapy, and 1 patient withdrew due for unknown reasons Outcomes The trials utilised internationally recognised outcome measures (OMERACT) 25. Using the American College of Rheumatology 20% (ACR20) response criterion, patients were categorised as a responder if they exhibit at least 20% improvement in swollen joint count and tender joint count and at least three of five other variables, namely pain score, physician global assessment, patient global assessment, physical function and acute phase reactant (usually C-reactive peptide). An ACR50 or ACR70 response requires at least 50% or 70% improvement in these variables, respectively. Physical function is measured using the Health Assessment Questionnaire (HAQ) score, which ranges between zero (no disability) and three (very severe disability). Improvements in HAQ score of >0.22 are significant. The disease activity score (DAS28) is a composite score that provides an absolute measure of disease activity; it is derived from the swollen joint score, tender joint score, patient global assessment of disease activity and ESR. DAS > 5.1 represents active disease, DAS > 3.2 moderate disease, DAS > 2.6 low disease activity. EULAR good response is defined as a fall in DAS of >1.2 to a level of < 2.6. EULAR moderate response is defined by a fall in DAS > 1.2, to a level > 2.6 but < 5.1. Fatigue is measured using the FACIT-Fatigue scale (Functional Assessment of Chronic Illness Therapy-Fatigue). Health-related quality of life was measured using the Short Form 36 (SF-36) questionnaire. The primary endpoint was the proportion of patients with an ACR20 response at week 24. Secondary endpoints included ACR50 and ACR70 responses; changes from baseline to week 24 in disease activity scores (DAS28); EULAR response criteria; changes from baseline in the ACR core set of SJC, TJC, patient and physician global assessments, patient pain assessment, patients assessment of disability (health assessment questionnaire-disability index [HAQ-DI]), CRP, and ESR responses over time to week 24. Additional endpoints included changes from baseline at week 24 in FACIT-F score and SF-36. Genant-modified Sharp radiographic scores were also measured as an exploratory endpoint. Regardless of their status in the study, all patients in this study were to return for scheduled radiographs. Page 31 of 181

32 Statistical analysis and definition of study groups Sample size calculations were based on the assumption of detecting a difference between rituximab-mtx (0.45) and placebo-mtx (0.30) in the proportion of patients with an ACR20 response. On the basis of these assumptions and using a conservative exact test (Fisher s) with a two-sided 5% significance level, a sample size of 500 patients randomized 3:2 (rituximab:placebo) provided the study with 91% power to detect a difference between the treatments. The efficacy analyses were made on the intent-to-treat (ITT) population, defined as all randomized patients who received any part of an infusion of study medication, excluding some patients as described earlier. Patients who prematurely withdrew from the study or who entered rescue were included in the ITT population as non-responders. The primary analysis was a test of the difference in proportions of patients with an ACR20 response at week 24 between the placebo and rituximab groups using a Cochran-Mantel-Haenszel (CMH) analysis, stratifying by baseline rheumatoid factor (positive vs. negative) and region (US vs. non-us). Results were expressed as proportions, corresponding to 95% confidence intervals of the difference between response rates and p values. The null hypothesis assumed that the proportion of patients with an ACR20 response at week 24 for rituximab was the same as that for placebo. rituximab treatment was judged effective compared with placebo if there was sufficient statistical evidence to reject the null hypothesis in favour of rituximab. Secondary endpoints of ACR50 and ACR70 responses were analyzed using the same statistical methodology as described for ACR20. Changes in DAS28 were assessed using an ANOVA model with baseline DAS28, RF, region, and treatment as terms in the model. Categorical DAS28 responders (EULAR response) were assessed using a CMH test stratified for RF and region. Changes from baseline in the individual ACR core set parameters of SJC, TJC, patient and physician global assessments, HAQ, pain, CRP, and ESR were summarized by descriptive statistics. The difference between the rituximab group and the placebo group was assessed using ANOVA models with baseline parameter of interest, RF, and region as terms in the model. ACR 20/50/70 and DAS28 changes over weeks 8, 12, 16, 20 and 24 were assessed using descriptive, graphical, and ANOVA techniques, as appropriate. For the radiographic endpoints, the primary population was a modified intention-to-treat (ITT) population, defined as all patients randomized who received at least a part of the Page 32 of 181

33 first treatment (any part of two study infusions), and who had both a screening / baseline and at least 1 post-treatment radiograph. The primary radiographic analysis compared the change in the total Sharp-Genant score between screening / baseline and week 56. The change in total scores for patients with missing data at week 56 (excluding those patients in whom the week 56 radiograph was obtained outside a pre-specified time window days 350 to 434) were imputed using linear extrapolation from their change in total score observed between screening and week 24. The use of linear extrapolation from an earlier post treatment radiograph is an accepted methodology and has been used as the primary method of imputation in previous radiographic studies in RA Critical appraisal of relevan RCTs WA17042 was a parallel group, multinational, multicentre, randomised, placebocontrolled, double-blind study. The trial was conducted in the UK, US, Europe, Canada and Israel. US clinical practice is different from that in Europe/UK, as physicians in the US tend to initiate patients on to biologic therapy earlier than in Europe/UK. This manifests as a regional difference in the proportions of patients who received previous TNF inhibitor agents and biological response modifying treatments, with more patients in the US receiving earlier TNF inhibitor treatment compared with patients in centres outside of the US. Therefore randomization into the study was stratified by region (a binary variable of US or Non-US). This was a blinded study, with the study sponsor, investigators, and patients unaware of the treatment assignment of each patient. Patients were randomized at a ratio of 3:2 to receive intravenous infusions of rituximab or placebo on days 1 and 15. The aim of the unequal allocation was to reduce the number of patients allocated to placebo and increase safety information from the active rituximab group. The dose of rituximab used was 1000mg on day 1 and day 15, in combination with methotrexate, as per licence. Randomization was administered by a central randomization centre, and stratified by rheumatoid factor (RF positive, defined as a screening value of RF 20 IU/mL, or negative, RF screening value of < 20 IU/mL) and region (US or non-us). Eligible patients were randomized via an interactive voice response system (IVRS) and assigned a unique medication number. As confirmation, the investigator was provided a written verification of each patient s registration. Treatment was to occur within 24 hours of the patient being randomized into the study. Page 33 of 181

34 A dual assessor approach was used to evaluate efficacy and safety. Only the Safety Assessor had access to all clinical and laboratory (safety) data and was able to make any necessary changes to the patient s medical therapy, thus minimizing the chance of unblinding of the efficacy assessor who only had access to efficacy data. Sample size calculations were determined as per section Analyses used were appropriate for rheumatoid arthritis trials (see section 5.3.4). Main analyses were performed on the ITT population. Patients were required to have active RA and failed at least one TNF inhibitor therapy. Patients generally had severe RA (mean baseline DAS 6.8, 6.9, placebo and active, respectively) with a mean disease duration of years. This is in accordance with the licensed indication and reflects the population likely to receive rituximab in UK. Primary efficacy and safety were evaluated at 24 weeks; after 24 weeks, patients entered a post-treatment period and were followed up every 2 months for 18 months, for an overall study duration of 24 months. Demographics show that the treatment groups were well balanced with respect to demographic and baseline disease characteristics. Seventy-nine percent of patients in both groups were seropositive for rheumatoid factor (RF 20 IU/mL). The mean RA disease duration was approximately 12 years, with patients in both groups having similar mean scores for the key measures of disease activity (swollen and tender joint counts, CRP and ESR). The mean TJC (approximately 33 in each group) was higher than the mean SJC count (approximately 23 in each group), reflecting the severity of disease in this patient population. Consistent with this, the mean DAS28 at baseline was high, being 6.8 and 6.9, respectively, for the placebo + MTX and rituximab + MTX groups. The mean HAQ-DI score in both groups was 1.9, which is consistent with the severity and duration of RA in this patient population. Similar mean total Sharp-Genant radiographic scores were recorded by patients in each group (47.9 for placebo + MTX patients and 48.3 for rituximab + MTX patients). Patients entering the study had received a median of 2 prior DMARDs and one previous TNF inhibitor therapy. The maximum number of previous DMARDs was 9 in the placebo + MTX group and 8 in the rituximab + MTX group, and some patients in each group had received three prior TNF inhibitor therapies. Page 34 of 181

35 50% of the patients were from the US. As discussed earlier, there are different treatment strategies in the US with regard to timing of biologic therapies. However, regional data have been presented to address this issue Results of the relevant comparative RCTs In the following section, the results have been structured as follows, to capture short and longer-term time horizons and primary, secondary and exploratory endpoints: a) efficacy data at 24 weeks b) long-term efficacy data after one course of rituximab (48 weeks) c) radiographic data d) repeat treatment efficacy data Within each section, primary efficacy endpoints (and any subgroup analyses) have been presented first (ACR scores), followed by secondary and exploratory endpoints where available: DAS, EULAR, HAQ-DI, SF-36 and FACIT-F. WA17042 Of the 520 patients who were enrolled, 517 received at least part of a rituximab or placebo infusion and comprised the safety population. The intent-to-treat population (ITT) included 499 patients: 394 who were RF positive, and 105 who were RF negative. Baseline characteristics and reasons for exclusion of patients are included earlier in this text. a) Clinical efficacy data at 24 weeks Primary endpoint The proportion of ACR20 responders at week 24 (primary endpoint) was significantly higher in the rituximab + MTX group than in the placebo + MTX group (51% vs 18%, respectively; (p < ), Figure 2. Logistic regression analysis showed that rituximab treated patients were 5 times more likely to achieve an ACR20 response than placebotreated patients. Page 35 of 181

36 Figure 2 Subgroup analyses When the ACR results presented above were split by the number of prior TNF inhibitor therapies used by the patients, results were as follows 26 : Table 2 1 prior anti-tnf 2 prior anti-tnfs rituximab + MTX (n=179) Placebo + MTX (n=121) ACR20 58% 21% ACR50 30% 7% ACR70 14% 1% (n=119) (n=80) ACR20 42% 14% ACR50 22% 3% ACR70 10% 3% The proportion of patients achieving an ACR20 response at week 24 in the placebo + MTX group was similar in both US and non-us regions [COMMERCIAL-IN- CONFIDENCE REMOVED] Analysis of the interaction between treatment and geographic region showed that while ACR20 response rates for rituximab patients were statistically significant (P < ) for both US and non-us patients, the treatment effect was greater for non-us patients Page 36 of 181

37 at the 10% level (P = 0.08). Nonetheless, the treatment effect in favour of rituximab + MTX was statistically significant for both regions (p < 0.001); see Table 3 below. Table 3: [COMMERCIAL-IN-CONFIDENCE REMOVED] In addition, fewer RF-negative patients than RF-positive patients achieved an ACR20 response at week 24, both in patients treated with placebo (12% versus 19%) and in patients treated with rituximab (41% versus 54%); differences in the ACR20 response rates between the placebo and rituximab groups were significant for both RF-positive patients (P < ) and RF-negative patients (P < ). There was no significant interaction between treatment and RF status (P = 0.9), indicating that the effect of treatment on the ACR response was not dependent on baseline RF status. (Table 4). Table 4 Secondary endpoints All secondary efficacy endpoints at week 24 based on clinical measures (ACR50, ACR70, change in DAS28, EULAR response, change in ACR core set parameters, changes in both physical and mental health SF-36 domains, and fatigue [FACIT-F]) showed statistically greater improvement (p ) following therapy with rituximab + MTX compared with placebo + MTX (Table 5). Page 37 of 181

38 Table 5 ACR 50, ACR70 responses The proportion of patients achieving ACR50 and ACR70 responses at week 24 were significantly higher for rituximab-treated patients than placebo-treated patients (27% versus 5% for ACR50 and 12% versus 1% for ACR70; P < for each comparison). ACR20 response rates over time showed a statistically significant (P < ) separation between rituximab and placebo treatment by week 8 (Figure 3); ACR50 and ACR70 responses over time showed a statistically significant separation by week 12 and week 16 of treatment, respectively. Page 38 of 181

39 Figure 3 ACR20 Response Rates Over Time (ITT Population) EULAR/DAS Significantly more rituximab-treated patients achieved good or moderate EULAR responses as compared with placebo treated patients (65% versus 22%; P < ), (Figure 4), with significant reductions in mean DAS28 scores from baseline ( 0.4 versus 1.9; P < ). In addition, a greater proportion of patients receiving rituximab achieved EULAR low disease activity and remission responses compared with patients receiving placebo (Figure 3). DAS28 scores were statistically significant (P < ) between treatment groups at every time point from week 8 through week 24. Page 39 of 181

40 Figure 4 ACR core set The mean reductions from baseline in all parameters of the ACR improvement criteria (used to calculate the ACR responses) were significantly greater for patients receiving rituximab (Table 2). Notably, at week 24, the swollen and tender joint counts, patient s and physician s global assessments of disease activity, pain scores, HAQ DI scores, CRP levels, and ESRs were significantly decreased in rituximab-treated patients; placebo-treated patients reported little or no changes in the swollen or tender joint counts, HAQ DI scores, pain scores, CRP levels, or ESRs at week 24. The CRP level was essentially unchanged in placebo-treated patients, but decreased by a mean of 2.1 mg/dl in rituximab-treated patients (P < ). Of the 298 rituximab-treated patients in the ITT population at week 24, 18 (6%) had a HAQ DI score of 0, 29 (10%) had a swollen joint count of 0, and 21 (7%) had a tender joint count of 0. Of the 201 placebotreated patients in the ITT population at week 24, 1 (0.5%) had a HAQ DI score of 0, 2 (1%) had a swollen joint count of 0, and 3 (1.5%) a tender joint count of 0. A total of 80 of the 281 rituximab-treated patients (28%) and 18 of the 180 placebotreated patients (10%) with elevated CRP levels at baseline had CRP levels in the Page 40 of 181

41 normal range by week 24. ESR decreased by a mean of 18.5 mm/hour in rituximabtreated patients, compared with a decrease of 4.1 mm/hour in placebo-treated patients (P < ). Patient-reported outcomes Patients receiving rituximab reported a significant reduction in fatigue, as measured by the FACIT-F scale, which decreased by a mean of 9.1 points, representing a 29.6% improvement from baseline (P < ); patients receiving placebo reported a mean decrease of 0.5 points in the FACIT-F score. At week 24, mean SF-36 scores for the mental and physical health domains increased by 4.7 and 5.8, respectively, in rituximab-treated patients, and by 1.3 and 0.9, respectively, in placebo-treated patients; both differences were statistically significant (P < ). The rituximab group showed in all 8 domains higher scores than the placebo group with greatest changes in bodily pain, role physical, role emotional and vitality (Table 6 below) 27. Table 6 Change from baseline to week 24 for SF-36 subscales and component scores by treatment SF-36 Domain rituximab + MTX Placebo + MTX p-value* MCID n=308 n=206 Physical Functioning Yes Role Physical Yes Bodily Pain < Yes General Health < Yes Vitality < Yes Role Emotional Yes Social Functioning Yes Mental Health Yes Physical Component Score < n.a. Mental Component Score n.a. *Unadjusted T-test (Satterthwaite) of differences in week 24. Page 41 of 181

42 Changes in radiographic end points will be reviewed in the long-term efficacy section of WA b) Long-term efficacy from a single course of rituximab (48 week data) 28 After reaching the primary endpoint at week 24, patients were followed every 8 weeks for an additional 18 months, to give an overall study duration of 24 months. Patients requiring further courses of rituximab treatment could enter extension protocols from week 24 onwards if they met repeat treatment criteria. Longer term efficacy after the initial, randomised treatment has been analysed based on all treated patients, according to the treatment to which they were randomised. Patients continuing to respond to initial treatment were analysed at 48 weeks. These patients did not receive any additional treatment courses other than the initial rituximab infusions. Background MTX was continued throughout this period. Patients and investigators remained blinded to treatment group. ACR Response over Time After First Treatment Course The following analyses summarize response after the initial randomized treatment course. The data are based on the all treated population (as randomized). This differs from the primary ITT analysis which excluded a small number of patients (see details in WA17042). Data are therefore presented for response after first treatment course (24 and 48 weeks). Relative to the placebo + MTX group, a higher proportion of patients in the rituximab + MTX group had at least an ACR 20 response at week 24, week 48 (Table 4) and at all other time points in the pivotal study. These results demonstrate that rituximab + MTX treated patients achieved sustained responses compared with placebo + MTX. A relatively small number of rituximab + MTX patients withdrew before or at week 24 for lack of efficacy. After week 24, patients with at least 8/8 active joints and who had experienced at least 20% improvement in joint count could withdraw to receive further treatment in the open-label study. The observed withdrawal rate post week 24 (see observed data in Table 7) can also be attributed to the fact that according to the above criteria, the investigator could administer further courses of treatment in patients who were still responding relative to baseline but who had residual active disease. Nonetheless, a significant proportion of patients (114/308; 37%) randomized to Page 42 of 181

43 rituximab + MTX remained in follow up at week 48 after a single course of treatment compared with only 11% of patients initially randomized to placebo. Table 7 Change in Disease Activity (Mean DAS28) after First Treatment Course Comparisons following Initial Treatment in Randomised Study Patients in the rituximab + MTX group remaining on study post week 24 in WA17042 were still showing improvement in DAS28 relative to their original baseline (mean change -2.1 at week 48) (Table 8 below) Table 8 Page 43 of 181

44 Proportion of Patients with EULAR Response, Low Disease Activity and Remission over Time after First Treatment Course Consistent with ACR response data, the proportion of patients achieving EULAR responses, low disease activity and remission was higher at post week 24 time points in rituximab + MTX-treated patients than in placebo + MTX-treated patients. For example, the proportion of patients in remission at week 48 was 14/308 (5%) in the rituximab + MTX arm and 1/209 (<1%) in the placebo + MTX arm using non responder imputation (Table 9). Table 9 Proportion of Patients with Improvement in HAQ-DI after a Single Course of Treatment A higher proportion of patients in the rituximab + MTX treatment arm than in the placebo + MTX arm had clinically significant improvements in HAQ-DI scores at post week 24 time points in the WA17042 study. A 0.22 decrease in HAQ-DI represents the minimal clinically important difference in HAQ. 29 Thus 58% of the originally randomised patients had a clinically significant improvement in HAQ-DI 24 weeks after the initial treatment with rituximab + MTX (compared with 23% in the placebo + MTX arm, Table 10). At 48 weeks, 23% of the originally randomized patients had a clinically significant improvement in HAQ-DI 48 weeks after the initial treatment with rituximab + MTX (compared with 6% in the placebo + MTX arm, Table 10). Page 44 of 181

45 Table 10 Changes in SF-36 after a Single Course of Treatment Improvements in physical and mental health scores were greater in the rituximab + MTX treatment arms than in the placebo + MTX treatment arms. Table 11 presents the mean changes from baseline to week 56 in the aggregated physical and mental component scores for each study. Table 11 Patients in follow-up beyond a year after the first course of rituximab were still showing improvement relative to baseline in both physical and mental health component scores. Page 45 of 181

46 c) Radiographic endpoints 30 These data were generated from WA17042 and its open-label extension, WA This section details the outcomes of radiographic analyses of the reduction in the progression of joint damage in adult patients with active rheumatoid arthritis who have had an inadequate response or intolerance to other disease-modifying anti-rheumatic drugs including one or more TNF inhibitor therapies. Overview of Efficacy Of the 520 patients randomized into WA17042, 463 were included in the radiographic analysis. The excluded patients were those who missed a screening radiograph or who had no post-treatment radiographic assessment. Radiographic assessments were taken at screening, prior to entry into WA17042 and at 24 and 56 weeks after randomization into the study. Post randomization radiographs were also sought in patients who had withdrawn from either study, again at the times relative to their randomization in WA Consequently, at the time of the 56-week radiograph, patients could have the following study dispositions: Patients post week 24 and remaining in WA17042 Placebo patients rescued with rituximab and remaining in WA17042 Patients transferring to a repeat treatment extension protocol, either after 24 weeks in WA17042 or following rescue with rituximab Patients who withdrew at any point and entered safety follow-up either in WA17042 or the extension study In total, 311 patients were randomized to receive rituximab 2 x 1000 mg in combination with MTX (weekly dose 10 to 25 mg), of whom 277 actually received randomized treatment and had a screening radiographic assessment plus at least one radiographic assessment post screening / baseline. A total of 209 patients received MTX (weekly dose 10 to 25 mg), with placebo infusions, of whom 186 actually received randomized treatment and had a screening / baseline radiographic assessment plus at least one radiographic assessment after screening / baseline. Of these 186 patients, 150 (81%) had received at least one Page 46 of 181

47 course of rituximab during the 56-week observation period, either via rescue therapy or through completing 24 weeks in WA17042 and transferring to the extension study. Primary radiographic information was derived from comparative analyses performed at week 56, with the original placebo + MTX group, irrespective of whether or not patient had subsequently received rituximab. Disease Activity Measures Radiological Endpoints Radiographic assessments of the hands and feet were collected and scored using the Sharp criteria, as modified by Genant 31 and were scored by two independent readers blinded to treatment assignment and time point. Statistical Analysis and Data Handling See section Efficacy Results Radiographic data obtained at week 24 showed a significant reduction in joint space narrowing (JSN) score with rituximab + MTX compared to placebo + MTX. Changes in total Sharp-Genant score and erosion score suggested less progression in patients treated with rituximab; however, statistically significant effects on these endpoints were not achieved at the 24 week time point. However, as the study was possibly underpowered to demonstrate a difference in the change of the total Sharp-Genant score over a short-term period (24 weeks or less) this finding was not unexpected. Radiographic data obtained at week 56 demonstrated a significant reduction in structural joint damage with rituximab in combination with MTX compared to placebo + MTX therapy. The primary radiographic endpoint (change from screening / baseline in the total Sharp-Genant score at week 56) showed a highly significant lower mean change from baseline for patients treated with rituximab + MTX than for patients who received placebo + MTX (1.00 vs. 2.31, p = ). This finding was supported by the analyses of changes from screening / baseline at week 56 in the two components of the total Sharp-Genant score: the total erosion score (0.59 and 1.32, for rituximab + MTX vs placebo + MTX respectively; p = ) and the JSN score (0.41 and 0.99, respectively; p = ; Table 12). Page 47 of 181

48 Table 12 Summary of Change from Baseline to Week 56 in Sharp-Genant Total Score (Primary Analysis), Erosion Score, and Joint Space Narrowing Score (ITT Population) Placebo + MTX N = 186 rituximab + MTX N = 277 Primary Radiographic Endpoint Change from screening / baseline in total Sharp-Genant score N Mean (SD) 2.31 (5.283) 1.00 (2.751) Median Min, Max -7.4, , 17.4 p-value Secondary Radiographic Endpoints Change from screening / baseline in total erosion score N Mean (SD) 1.32 (3.155) 0.59 (1.843) Median Min, Max -5.9, , 11.8 p-value Change from screening / baseline in joint space narrowing score N Mean (SD) 0.99 (2.571) 0.41 (1.323) Median Min, Max -5.5, , 7.6 p-value Proportion of patients with no erosive progression over 56 weeks N No Erosive progression 95 (52%) 166 (61%) Difference in proportions na % CI na 0.00, 0.18 p-value d) Long Term Efficacy following repeated courses of rituximab 32, 33 In the primary study, after reaching the primary endpoint at week 24, patients were followed every 8 weeks for an additional 18 months, to give an overall study duration of 24 months. Page 48 of 181

49 To demonstrate efficacy of repeated treatment with rituximab, patients requiring further courses of rituximab treatment could enter an extension protocol from week 24 onwards if they met the repeat treatment criteria outlined below. Longer term efficacy after the initial, randomised treatment has been analysed based on all treated patients, according to the treatment to which they were randomised. In addition, for the long-term assessment of efficacy following repeated treatment courses, to give a better indication of the response pattern, an additional, small number of patients have been included from WA17043; data have been pooled for patients who only received the expected licensed dose of rituximab 2 x 1000 mg + MTX regimen for first and subsequent courses and who received prior TNF inhibitor therapy. This allows examination of efficacy in the context of the licensed indication for the use of rituximab in those patients who have failed TNF inhibitor therapy. Disposition of Patients Figure 5 gives an overview of the patients entering the extension study and the total number of patients providing efficacy and safety data. As shown, a total of 281 patients entered the open label extension study as of 14 October Page 49 of 181

50 Figure 5 WA17042 n=517 Double blind treatment n=164 Rituximab 1000mg x2 + MTX n=308 Placebo + MTX n=209 n=92 n=72 Open label treatment with rituximab 2x1000mg + MTX Rescue treatment with rituximab 1000mg x2 + MTX n=45 WA17531 n=281 Pooled analysis includes additional patients from WA17043 who had failedtnf inhibitor therapy and had 2x 1000mg rituximab Efficacy of Repeated Treatment In the following sections, responses are calculated both relative to the original baseline and to each treatment course baseline. Repeat treatment criteria The initial eligibility for repeat treatment within WA17043 only allowed the repeat treatment of patients who had achieved an ACR50 (the primary end point for that study) at 6 months and had less than 20% change from baseline in joint counts at the time of repeated treatment. This criterion was developed to ensure patients who were Page 50 of 181

51 retreated were B cell replete as this was the first study in the patient population to be carried out. The problems associated with this are that only the initial good responders were repeat treated and patients had to flare significantly before receiving an additional course. However, once the optimum regime of 2 x 1000 mg rituximab plus methotrexate had been identified, the criteria were adjusted so patients had to achieve at least 20% improvement in swollen and tender joints and have active disease ( 8 swollen and tender joints) to receive additional rituximab and methotrexate (as criteria for WA17531 outlined below). The open label study WA17531 accepted patients from WA17042 protocol who had completed their first treatment course. The eligibility criteria for these patients allowed patients who had an initial 20% improvement in swollen and tender joints following the first course but had a return of disease activity based on swollen and tender joint counts 8. The issues highlighted above with the original open label extension still remain in part but to a lesser degree. Not all patients for example were B cell replete as measured on a normal flow cytometry assay. It should be noted that the meeting of the eligibility criteria for the extension study did not mandate repeat treatment and essentially the decision was based on the clinical judgement of the investigator. The data relating to those patients who had previously been exposed to TNF inhibitor therapies were retreated within these new less severe protocol criteria. ACR Response Following Repeated Treatments In general, patients show better responses from the original baseline following their second course of treatment (eg, 12% and 21% ACR70 response at week 24 after first and second course, respectively, Figure 6, Table 13). These 155 patients were the first to complete two courses so this positively selects out those patients who needed repeat treatment earlier (for example one third of patients went beyond a year before needing to be retreated) and thus are not the patients who have achieved the greatest benefit from rituximab. There is no evidence of decreasing responses with the third treatment course, although the number of patients is small. Page 51 of 181

52 Table 13 Page 52 of 181

53 Figure 6 ACR responses measured relative to the treatment course baseline, show similar response rates across treatment courses. Since disease activity prior to each additional course is generally lower than the original baseline, these data suggest that additional courses of treatment with rituximab not only maintain a response relative to the original baseline but also cause a further decrease in disease activity. This observation is also supported when the disease activity is measured by DAS28 (Figure 7). Change in DAS28 after Repeated Treatments Mean DAS28 scores decreased from the original baseline after the first treatment course and were generally still lower than this baseline prior to the second treatment course. In patients in the prior TNF inhibitor population who had received three courses of rituximab (N = 81), the original DAS28 was 7.07, and prior to the second treatment course this was lower (6.17), and lower still prior to the third (6.01). This indicates that, on average, patients disease activity had not returned to the original baseline levels prior to receiving subsequent treatment courses with rituximab + MTX. This reflects the retreatment criteria used in the development programme. Additional courses provide additional benefit (Figure 7). Page 53 of 181