Adalimumab, etanercept, infliximab,certolizumab pegol, golimumab, tocilizumab and abatacept for the treatment of rheumatoid arthritis not previously treated with disease-modifying antirheumatic drugs and after the failure of conventional disease-modifying antirheumatic drugs only: systematic review and economic evaluation

Aetiology

Rheumatoid arthritis (RA) is a chronic inflammatory disease characterised by progressive, irreversible, joint damage, impaired joint function, and pain and tenderness caused by swelling of the synovial lining of joints and is manifested with increasing disability and reduced quality of life. 1 The primary symptoms are pain, morning stiffness, swelling, tenderness, loss of movement, fatigue and redness of the peripheral joints. 2,3 RA is associated with substantial costs, both direct (associated with drug acquisition and hospitalisation) and indirect (owing to reduced productivity). 4 RA has long been reported as being associated with increased mortality,5,6 particularly due to cardiovascular events. 7

Epidemiology

The initial classification criteria for RA were produced in 1987 by the American College of Rheumatology (ACR). 8 The National Institute for Health and Care Excellence (NICE) Clinical Guideline (CG) 799 provides a summary of the ACR criteria, namely that patients must have at least four of the seven criteria (morning stiffness lasting at least 1 hour; swelling in three or more joints; swelling in hand joints; symmetrical joint swelling; erosions or decalcification on radiograph of hand; rheumatoid nodules; and abnormal serum rheumatoid factor). The first four criteria must have been present for at least a period of 6 weeks. However, in the CG the guideline development group preferred a clinical diagnosis of RA rather than the ACR criteria because ‘an early persistent synovitis where other pathologies have been ruled out needs to treated as if it is RA to try to prevent damage to joints. Identification of persistent synovitis and appropriate early management is more important than whether the disease satisfies classification criteria’, referencing the European League Against Rheumatism (EULAR) recommendations. 10

In 2010 the ACR and EULAR jointly published Rheumatoid Arthritis Classification Criteria, which focused on features at earlier stages of disease that are associated with persistent and/or erosive disease rather than defining the disease by its late-stage features. 11 The classification criteria allocate scores to characteristics of joint involvement, serology, acute-phase reactants and duration of symptoms, to produce a score between 0 and 10, inclusive, with those scoring ≥ 6 and with obvious clinical synovitis being defined as having ‘definite RA’ in the absence of an alternative diagnosis that better explains the synovitis.

Two classifications have dominated the measurement of improvement in RA symptoms: (1) ACR responses;12 and (2) EULAR responses. 13

The initial ACR response was denoted as an ACR20, which required a 20% improvement in tender joint counts; a 20% improvement in swollen joint counts; and a 20% improvement in at least three of the following five ‘core set items’: physician global assessment; patient global assessment; patient pain; self-reported disability (using a validated instrument); and erythrocyte sedimentation rate (ESR)/C-reactive protein (CRP).

The ACR response has been widely adopted in randomised controlled trials (RCTs), although studies have shown that the value can vary between trials owing to the timing of the response. 14 Since the inception of the ACR20 two other response criteria (ACR50 and ACR70) have become more widely used, which are similar to ACR20 and differing only in the level of improvements required to be denoted a responder.

In the UK, monitoring the progression of RA is often undertaken using the Disease Activity Score 28 joints (DAS28). This assesses 28 joints in terms of swelling (SW28) and of tenderness to the touch (TEN28) and also incorporates measures of the ESR and a subjective assessment (SA) on a scale of 0–100 made by the patient regarding disease activity in the previous week.

The equation for calculating DAS28 is as follows:15

The DAS28 can be used to classify both the disease activity of the patient and the level of improvement estimated within the patient.

The EULAR response criteria use the individual change in DAS28 and the level of DAS28 reached to classify trial participants as good, moderate or non-responders. 13 The EULAR response criteria and the ACR20 improvement criteria were found to have reasonable agreement in the same set of clinical trials,16 although van Gestel et al. 16 state that the EULAR response criteria showed better construct and discriminant validity than did ACR20. EULAR response has been reported less frequently in RCTs than ACR responses, although EULAR is much more closely aligned to the treatment continuation rules stipulated by NICE, which require a DAS28 improvement of more than 1.2 to continue treatment. The relationship between change in DAS28 and the level of DAS28 reached with EULAR response is shown in Table 1. Dependent on the initial Disease Activity Score (DAS) score of the patient, this would equate to either a good or moderate EULAR response, as shown in the second column of Table 1.

Patients with a DAS28 of ≤ 3.2 are stated as having inactive disease, those with a DAS28 of > 3.2 and ≤ 5.1 are stated as having moderate disease and those with a DAS28 of > 5.1 are stated as having very active disease. 15

A widely used measure of patient disability is the Health Assessment Questionnaire (HAQ). The HAQ is a patient-completed disability assessment17 which has established reliability and validity and has been used in many published RCTs in RA. HAQ scores range from 0 to 3, with higher scores indicating greater disability. The HAQ is a discrete scale with step values of 0.125, resulting in 25 points on the HAQ scale.

Incidence and prevalence

There are an estimated 400,000 people in England and Wales with RA,18 with approximately 10,000 incident cases per year. 19 The disease is more common in females (1.16%) than in males (0.44%),19 with the majority of cases being diagnosed when patients are aged between 40 and 80 years20 and with peak incidence in patients in their seventies. 19 Traditionally, patients have been treated with conventional disease-modifying antirheumatic drugs (cDMARDs), which include methotrexate (MTX), sulfasalazine (SSZ), hydroxychloroquine (HCQ), leflunomide (LEF), and gold injections (GLDs) as well as corticosteroids, analgesics and non-steroidal anti-inflammatory drugs (NSAIDs). However, more recently, a group of drugs have been developed consisting of monoclonal antibodies and soluble receptors that specifically modify the disease process by blocking key protein messenger molecules (such as cytokines) or cells (such as B-lymphocytes). 9 Such drugs have been labelled as biologic disease-modifying antirheumatic drugs (bDMARDs) and form the focus of this report.

Significance for the NHS

Owing to previous NICE technology appraisals (TAs) recommending a number of bDMARDs (see Current service provision), with a potential sequence of three bDMARDs, there has been a considerable increase in expenditure on RA interventions. Given the remit of this research to establish the clinical effectiveness and cost-effectiveness of bDMARDs in advance of cDMARDs for patients with less severe disease (assumed to be those with a DAS28 of between > 3.2 and ≤ 5.1), there is potential for the expenditure to increase further should NICE guidance on these populations be positive. The majority of interventions are provided subcutaneously and would therefore require little additional staff time should there be positive guidance, although this would increase for those drugs which are given intravenously.

Further detailed information on the background of RA can be found within the relatively recent NICE CG. 9 Additional information can also be located in the British Society for Rheumatology guidelines. 21

Clinical guidelines

For people with newly diagnosed RA, NICE CG799 recommends a combination of cDMARDs [including MTX and at least one other disease-modifying anti-rheumatic drug (DMARD) plus short-term glucocorticoids] as first-line treatment, ideally beginning within 3 months of the onset of persistent symptoms. Where combination therapies are not appropriate (e.g. where there are comorbidities or pregnancy), DMARD monotherapy is recommended. Where DMARD monotherapy is used emphasis should be on increasing the dose quickly to obtain best disease control. For the purposes of this assessment the term intensive DMARDs has been used to denote that this is treatment with multiple cDMARDs simultaneously.

Current National Institute for Health and Care Excellence technology appraisal guidance

National Institute for Health and Care Excellence guidance (TA130,22 TA18623 and TA22524) recommends the use of the tumour necrosis factor (TNF) inhibitors etanercept (ETN; Enbrel^®, Pfizer), infliximab [IFX; Remicade^®, Merck Sharp & Dohme Corp. (MSD)], adalimumab (ADA; Humira^®, AbbVie), certolizumab pegol (CTZ; Cimzia^®, UCB Pharma) and golimumab (GOL; Simponi^®, MSD) in people with RA after the failure of two cDMARDs, including MTX, and who have a DAS28 > 5.1. Terminated NICE guidance (TA224) was unable to issue recommendations for the use of GOL in people with RA that have not been treated with MTX. 25

Technology Appraisal 24726 recommends tocilizumab (TCZ; RoActemra^®, Roche) as an alternative to TNF inhibitors in the same circumstances as in TA130,22 that is in patients with a DAS28 > 5.1 after trying two cDMARDs. NICE guidance TA28027 recommends the use of intravenous (i.v.) abatacept (ABT; Orencia^®, Bristol-Myers Squibb) in people with RA after the failure of cDMARDs in the same circumstances as TA130; the subcutaneous (s.c.) formulation has not been appraised.

A simplified summary of NICE-recommend bDMARDs is shown in Figure 1. This defines the sequence of treatments that have received positive guidance for patients with a DAS28 of > 5.1. In summary, the typical route would be intensive cDMARDs followed by a bDMARD, followed by rituximab (RTX) plus MTX, then TCZ before returning to cDMARDs.

FIGURE 1.

Summary of the position of bDMARDs within NICE TA recommendations for sequence of treatments for patients with RA and a DAS28 > 5.1. a, If RTX and MTX is contraindicated or withdrawn owing to adverse events then the following can be used: ADA or ETN or IFX or ABT in combination with MTX; ADA or ETN monotherapy TA195,28 TCZ in combination with MTX TA247,26 assuming these have not been used previously in the sequence; b, would not be used if TCZ has been used previously in the sequence.

It is noted that NICE CG79 recommends the use of intensive cDMARDs which have been assumed to be used rather than two cDMARDs used in monotherapy, although this latter option is acceptable.

The National Institute for Health and Care Excellence has also issued guidance (TA195,28 TA22524 and TA24726) on the treatment of RA after the failure of a TNF inhibitor, but such guidance falls outside the scope of this appraisal.

National Institute for Health and Care Excellence criteria for continuing treatment

Each of the NICE TAs states that for patients to continue treatment with a bDMARD there must have been an improvement in DAS28 of at least 1.2 points at 6 months. If this criterion has not been met then treatment should be stopped and the next intervention in the sequence initiated.

Data were provided by the British Society for Rheumatology Biologics Register (BSRBR) to the Assessment Group and were used to assess the time on first biologic conditional on EULAR response. These indicate that over 25% of patients who had no EULAR response at 6 months were still on treatment at 4.5 years, with the median treatment time being 319 days. This shows that there is not strict adherence to the NICE criteria for continuation of treatment. The majority of patients (94%) had a DAS28 of > 5.1, indicating that the severity criteria stated by NICE were reasonably well adhered to.

Interventions considered in the scope of this report

The scope of the work is to ascertain the clinical effectiveness and cost-effectiveness of seven interventions within three populations that will be detailed subsequently. These interventions are ABT, ADA, CTZ, ETN, GOL, IFX and TCZ. It is noted that ABT can be delivered in two formulations, intravenously and subcutaneously, and that both have been modelled separately. Owing to the large number of interventions these have been initially summarised by mode of action. There then follows a summary of the UK marketing authorisation for each intervention along with a description of administration method. This text is similar to that within the protocol. 29

Mode of action

Adalimumab, ETN, IFX, CTZ and GOL all inhibit the activity of tumour necrosis factor alpha (TNF-α), a pro-inflammatory mediator that is partly responsible for damage to the joints in RA.

Abatacept is a selective modulator of the T-lymphocyte activation pathway. It binds to molecules on the surface of antigen-presenting cells, preventing full activation of the T lymphocytes and interrupting the inflammatory process.

Tocilizumab inhibits the activity of the cytokine interleukin 6, a pro-inflammatory molecule that is also partly responsible for damage to the joints in RA.

Marketing licence and administration method

Abatacept, in combination with MTX, has a UK marketing authorisation for the treatment of moderate to severe active RA in adult patients who responded inadequately to previous therapy with one or more cDMARDs, including MTX or a TNF-α inhibitor. It can be administered by i.v. infusion or by s.c. injection.

Adalimumab, in combination with MTX, has a UK marketing authorisation for the treatment of moderate to severe active RA in adults when the response to cDMARDs, including MTX, has been inadequate and for the treatment of severe, active and progressive RA in adults not previously treated with MTX. ADA can be given as monotherapy in case of intolerance to MTX or when continued treatment with MTX is inappropriate. It is administered subcutaneously.

Certolizumab pegol, in combination with MTX, has a UK marketing authorisation for the treatment of moderate to severe active RA in adult patients when the response to cDMARDs, including MTX, has been inadequate. CTZ can be given as monotherapy in case of intolerance to MTX or when continued treatment with MTX is inappropriate. It is administered subcutaneously.

Etanercept, in combination with MTX, has a UK marketing authorisation for the treatment of moderate to severe active RA in adults when the response to cDMARDs, including MTX (unless contraindicated), has been inadequate, and for the treatment of severe, active and progressive RA in adults not previously treated with MTX. ETN can be given as monotherapy in case of intolerance to MTX or when continued treatment with MTX is inappropriate. It is administered subcutaneously.

Golimumab, in combination with MTX, has a UK marketing authorisation for the treatment of moderate to severe active RA in adult patients when the response to cDMARD therapy, including MTX, has been inadequate, and for the treatment of severe, active and progressive RA in adults not previously treated with MTX. It is administered subcutaneously.

Infliximab, in combination with MTX, has a UK marketing authorisation for the reduction of signs and symptoms as well as the improvement in physical function in adults with active disease when the response to DMARDs, including MTX, has been inadequate. It is also licensed for the treatment of severe, active and progressive RA in adults not previously treated with MTX or other cDMARDs. It is administered by i.v. infusion.

Tocilizumab, in combination with MTX, has a UK marketing authorisation for the treatment of moderate to severe active RA in adult patients who have either responded inadequately, or who were intolerant, to previous therapy with one or more DMARDs or TNF antagonists. In these patients, TCZ can be given as monotherapy in case of intolerance to MTX or where continued treatment with MTX is inappropriate. TCZ is administered by i.v. infusion.

Current usage in the NHS

There is widespread use of the interventions within the NHS. Robust values of the exact breakdown by intervention are not known.

Identification of important subgroups

The current NICE guidance has already identified a subgroup by stating that to receive a bDMARD the patient must have received two cDMARDs and have active RA with a DAS28 in excess of 5.1. The research questions within this report include estimating the cost-effectiveness if the severity criteria were lessened to include patients with a DAS28 of > 3.2; and estimating the cost-effectiveness of using bDMARDs in advance of cDMARDs.

An important clinical subgroup encompasses those patients in whom bDMARDs cannot be given in combination with MTX. The clinical effectiveness and cost-effectiveness of licensed bDMARDs in this population will be estimated in this assessment.

The anticipated costs associated with the interventions

The costs associated with each intervention need to take into account factors, including the acquisition cost of the drug [incorporating any Patient Access Scheme (PAS)]; the average weight of patients with RA for those interventions that are weight based; the administration costs associated with infusions and of district nurses performing s.c. injections; and any loading doses required in the first year.

The acquisition costs and dosing regimens were taken from the British National Formulary (www.bnf.org; accessed June 201330) with details of PASs taken from the manufacturers’ submissions.

The average weights of patients with RA were estimated using data (n = 12,176) from the BSRBR. To be able to be used with all of the weight-based dosing regimens, a large number of categories were required, as detailed in Table 2. From these categories the average cost per dose for those with a weight-based dose can be calculated.

Additional loading doses in the first year were calculated based on the relevant regimen and the administration cost. Table 3 provides a simplified summary of the assumed mean acquisition costs per intervention and can be used to provide indicative rather than exact values. Within the mathematical model described later, timings of costs are explicitly incorporated and also the fact that in some subgroups the distribution of weights may differ from that of the full BSRBR database, a factor also considered within the Assessment Group model.

Additional treatments in a sequenced strategy

The nature of RA treatment being sequenced meant that it was necessary for the Assessment Group and the manufacturers to incorporate the costs and effectiveness of RTX into the model as this has positive NICE guidance following the withdrawal of a bDMARD. These will be discussed as applicable.

Interventions

A detailed description of each of the interventions is provided in Chapter 1, Description of the technologies under assessment. Table 4 summarises the relationship between the market authorisation and the decision problem detailed in Overall aims and objectives of assessment: that is, whether or not the intervention is licensed to be used prior to the initiation of MTX intervention; as a monotherapy (i.e. without needing to be given in combination with MTX); for patients with severe RA; and for patients with moderate to severe RA.

Populations (including subgroups)

The scope issued by NICE defines three distinct populations with RA and includes (1) adults with severe active RA not previously treated with cDMARDs; (2) adults with severe active RA who have been previously treated with cDMARDs but not bDMARDs; and (3) adults with moderate to severe active RA who have been previously treated with cDMARDs only, including MTX (unless contraindicated or inappropriate). Henceforth, these will be referred to as population 1, population 2 and population 3, respectively.

Although the NICE scope did not specify the definition of severe active RA and moderate to severe active RA, the following definition (based on expert clinical advice to the Assessment Group) has been adopted: severe active RA will be defined by a DAS28 of ≥ 5.1 and moderate to severe active RA will be defined as a DAS28 of between 3.2 and 5.1.

As the scope issued by NICE explicitly defined subgroups, no further subgroups will be assessed, with the exception of those patients in whom bDMARD treatment needs to be given as monotherapy. Separate analyses will be conducted for those in whom MTX can be tolerated and in those who can only receive bDMARD monotherapy.

The Assessment Group has chosen to deviate from the scope for population 1 as the definition in the scope stated that MTX needed to have been used previously. Given this definition, the populations were mutually exclusive but not exhaustive, as patients without prior bDMARD treatment who had not received MTX but had instead received an alternative cDMARD would not be allocated to any of the populations. In consultation with NICE and our clinical experts the Assessment Group broadened its interpretation of population 1 to allow previous treatment with any cDMARD.

It is noted that the number of interventions considered in population 1 is fewer than for populations 2 or 3, as only four interventions (ADA, ETN, GOL and IFX) are licensed in this population.

Populations outside the scope of the research

The following groups were explicitly excluded from the research by the scope issued by NICE:

• the initiation of treatment in patients without active RA

• patients with a DAS of < 3.2 who had received previous treatment with cDMARDs

• patients with a DAS of < 5.1 who had not been previously treated with cDMARDs

• patients who had been previously treated with one or more bDMARDs.

Relevant comparators

The relevant comparators within the final scope differ according to the population considered. The scope stated that tofacitinib (TOF; Xeljanz^®, Pfizer; Jakvinus^®, Pfizer) would be included if NICE had issued positive guidance prior to the report’s completion, but this did not occur and therefore TOF was not evaluated.

1. For severe active RA not previously treated with MTX or other DMARDs:

   1. combination therapy with cDMARDs (including MTX and at least one other DMARD, such as SSZ and LEF as recommended in NICE CG799)

   2. the interventions will be compared with each other.

2. For severe active RA that has been previously treated with cDMARDs only:

   1. management strategies involving further cDMARDs (e.g. SSZ, LEF), NSAIDs and corticosteroids

   2. the interventions will be compared with each other.

3. For moderate to severe active arthritis that has been previously treated with cDMARDs only:

   1. management strategies involving further cDMARDs (e.g. SSZ, LEF), NSAIDs and corticosteroids

   2. the interventions will be compared with each other.

Outcomes

The outcome measures to be considered include:

• disease activity

• physical function

• joint damage

• pain

• mortality

• fatigue

• radiological progression

• extra-articular manifestations of disease

• adverse effects of treatment

• health-related quality of life.

Data were also collected on other outcome measures, including disease duration, number of previous cDMARDs and percentage of patients who had received bDMARDs, in case there was sufficient variation in baseline measurements that these could be investigated as treatment effect modifiers within data synthesis.

Identification of studies

The aims of the search were to provide as comprehensive a retrieval as possible of clinical effectiveness evidence relating to ABT, ADA, CTZ, ETN, GOL, IFX and TCZ and to identify additional relevant treatments for potential inclusion in the NMA.

Inclusion and exclusion criteria

Inclusion and exclusion criteria for the selection of clinical effectiveness and safety evidence were defined according to the decision problem outlined in the NICE scope. 40

The inclusion of potentially relevant articles was undertaken using a two-step process. First, all titles and abstracts were examined for inclusion by one reviewer. Any citations that clearly did not meet the inclusion criteria (e.g. animal studies, studies unrelated to RA) were excluded. Second, full-text articles were initially examined by one reviewer. It was intended in the original protocol that a second reviewer would check approximately 10% of citations. However, because of the very large number of citations identified in the clinical effectiveness searches, this was not possible in the time scales available for this appraisal process. Any uncertainty in the inclusion and exclusion of potential full-text articles was resolved through discussion with the review team. Where agreement could not be reached, expert clinical advice was sought for a final decision.

The relevance of each article for the systematic review was assessed according to the following criteria.

Data abstraction and critical appraisal strategy

Data relevant to the decision problem were extracted by one reviewer. Data were extracted without blinding to authors or journal. Study arms where intervention treatments were administered in line with licensed indications were extracted; where there was a slight divergence between the regimen used in the RCT and the licensed regimen, this was explicitly highlighted. It was proposed in the original protocol29 that at least 10% of data extraction forms be checked by a reviewer. However, the Assessment Group ensured that all data included in the NMA were double checked by a second reviewer. For data not contributing to the NMA, data were extracted for the following time points: primary end point (for selected efficacy data); latest available controlled RCT end point (for efficacy and safety data); and latest available LTE study end point (for safety data only). The safety data extracted were informed by the Summary of Product Characteristics [available at www.medicines.org.uk/emc/ (accessed 1 April 2014)] and FDA prescribing information for each intervention. 41–47 Graphical data contributing to the NMA were estimated using Engauge software [version 4.1; Mark Mitchell, Los Angeles, CA, USA (2011)] and graphical data not contributing to the NMA were estimated manually by a reviewer. Where multiple publications of the same study were identified, data extraction was undertaken on all relevant associated publications and findings were presented as a single study. Discrepancies were resolved by discussion, with involvement of a third reviewer when necessary.

The methodological quality of each included study was assessed by one reviewer. It was originally intended in the protocol29 that quality assessment would be checked by a second reviewer, but this was not feasible within the time scales available for the appraisal process. The quality assessment of included studies was informed by selected items listed in the NHS Centre for Reviews and Dissemination report48 and Cochrane Risk of Bias tool. 49 Additional quality issues specific to the assessment of RA RCTs (as described by Karsh et al. 50) were also considered during the evaluation of studies.

Methods of data synthesis

The extracted data and quality assessment variables were presented for each study, both in structured tables and as a narrative description.

As the identified evidence base permitted the undertaking of NMAs for the estimation of treatment effects, supplementary meta-analyses were not undertaken. NMAs were conducted to determine efficacy using two different disease activity measures (ACR and EULAR responses).

Methods for the estimation of efficacy using network meta-analysis

Quantity and quality of research available

Quantity of research available

As a result of the searches described in Methods for reviewing effectiveness, a total of 43,764 citations were identified for the review of clinical effectiveness and safety. This was reduced to 27,464 following deletion of duplicate citations. The study selection process is represented as a PRISMA diagram (Figure 2). A total of 27,334 citations were excluded at title and abstract levels (1606 being non-English-language records). Of the remaining records, a total of 60 studies were included in the review. Studies excluded at the full-text stage are presented (with rationale for exclusion) in Appendix 3.

FIGURE 2.

Flow diagram of study inclusion (adapted from PRISMA).

Randomised controlled trials included in the systematic review of clinical effectiveness and NMAs of ACR and EULAR responses are presented in Table 5 (with MTX-naive and cDMARD-experienced labels denoting trials included in population 1 and populations 2 and 3 respectively).

TABLE 5 - Trials included in the systematic review and NMAs

ACT-RAY, ACTemra (tocilizumab) RAdiographic studY; ADACTA, ADalimumab ACTemrA (tocilizumab) head-to-head-study; ADORE, ADjuvant Oxaliplatin in REctal Cancer; AiC, academic-in-confidence; AIM, Abatacept in Inadequate responders to Methotrexate; AMPLE, abatacept vs. adalimumab in biologic naive RA patients with background MTX; APPEAL, Asia-Pacific Study in Patients to be Treated With Etanercept or an Alternative Listed; ARMADA, Anti-TNF factor Research study program of the Monoclonal antibody ADalimumab (D2E7) in rheumatoid Arthritis; ASPIRE, Active controlled Study of Patients receiving Infliximab for the treatment of Rheumatoid arthritis of Early onset; ASSET, Abatacept Systemic SclErosis Trial; ASSURE, Abatacept Study of Safety in Use with other RA ThErapies; ATTEST, A Trial for Tolerability, Efficacy, and Safety in Treating rheumatoid arthritis (infliximab); ATTRACT, Anti-TNF trial in Rheumatoid Arthritis with Concomitant Therapy; AUGUST II, A Phase II Dose-finding Study of Atacicept in Rheumatoid Arthritis (RA); BeST, BEhandelings STrategie; CERTAIN, efficacy and safety of CERTolizumab pegol After INcomplete response to DMARDS in RA patients with low to moderate disease activity; CHANGE, Clinical investigation in Highly disease-affected rheumatoid Arthritis patients in Japan with Adalimumab applying staNdard and General Evaluation study; COMET, Combination Of METhotrexate and etabercept in early rheumatoid arthritis; CREATE IIb, A 6-month Randomised, Double-blind, Open Arm Comparator, Phase IIb, With AZD9056, in Patients With Rheumatoid Arthritis (RA); DE019, Efficacy and Safety of Adalimumab in Patients With Active Rheumatoid Arthritis Treated Concomitantly With Methotrexate; ERA, Early Rheumatoid Arthritis (etanercept); GO-BEFORE, GOlimumab in active rheumatoid arthritis BEFORE methotrexate therapy; GO-FORTH, golimumab in combination with methotrexate in Japanese patients with active rheumatoid arthritis; GO-FORWARD, golimumab in active rheumatoid arthritis despite methotrexate therapy; GUEPARD, GUÉrir la PolyARthrite rheumotoide Débutante (adalimumab); HIT HARD, High Induction THerapy with Anti-Rheumatic Drugs (adalimumab and methotrexate); IDEA, Remission induction comparing infliximab and high-dose intravenous steroid, followed by treat-to-target: a double-blind, randomised, controlled trial in new-onset, treatment-naive, rheumatoid arthritis; JESMR, Japanese Efficacy and Safety of Etanercept on Active Rheumatoid Arthritis Despite Methotrexate Therapy; LARA, Latin American Rheumatoid Arthritis study; MEASURE, secukinumab in ankylosing spondylitis; OPERA, OPtimized treatment algorithm for patients with Early Rheumatoid Arthritis; OPTIMA, OPTimal protocol for treatment Initiation with Methotrexate and Adalimumab; PBO, placebo; PREMIER, Patients REceiving Methotrexate and Infliximab for the treatment of Early Rheumatoid arthritis; RACAT, Rheumatoid Arthritis Comparison of Active Therapies in Methotrexate Suboptimal Responders study; REALISTIC, RA EvALuation In Subjects receiving TNF Inhibitor Certolizumab pegol; RED-SEA, a Randomised Efficacy and Discontinuation Study of Etanercept versus Adalimumab; SAMURAI, Study of Active controlled Monotherapy Used for Rheumatoid Arthritis, an IL-6 Inhibitor; SATORI, Study of Active controlled TOcilizumab for Rheumatoid arthritis patients with an Inadequate response to methotrexate; STAR, Safety Trial of Adalimumab in Rheumatoid arthritis; START, Safety Trial for rheumatoid Arthritis with Remicade Therapy; Swefot, Swedish pharmacotherapy; TOWARD, TOcilizumab in combination With traditional DMARD therapy.

Trial name/study	Intervention	Population	Included in NMA?
Abe et al., 200656	IFX	cDMARD experienced	Not in NMA (14-week RCT)
ACT-RAY57	TCZ	cDMARD experienced	Yes
ADACTA58	ADA, TCZ	cDMARD experienced	Yes
ADORE59,60	ETN	cDMARD experienced	Not in NMA (16-week study)
AIM61–65	ABT	cDMARD experienced	Yes
AMPLE66	ADA, ABT	cDMARD experienced	Yes
APPEAL67,68	ETN	cDMARD experienced	Not in NMA (16-week study)
ARMADA69,70	ADA	cDMARD experienced	Yes
ASPIRE71	IFX	MTX naive	Not in NMA (no ACR/EULAR data at 22–30 weeks)
ASSET72	ABT	cDMARD experienced	Not in NMA (4-month RCT)
ASSURE73	ABT	cDMARD experienced	Not in NMA (no ACR/EULAR data at 22–30 weeks)
ATTEST74	IFX, ABT	cDMARD experienced	Yes
ATTRACT75	IFX	cDMARD experienced	Yes
AUGUST II76	ADA	cDMARD experienced	Yes
Bejarano et al., 200877	ADA	MTX naive	Not in NMA (no ACR/EULAR data at 22–30 weeks)
BeST78	IFX	MTX naive	Yes
CERTAIN79	CTZ	cDMARD experienced	Yes
CHANGE80	ADA	cDMARD experienced	Yes
COMET81–83	ETN	MTX naive	Yes
DE01984	ADA	cDMARD experienced	Yes
deFilippis et al., 200685	ETN, IFX	cDMARD experienced	Yes
Durez et al., 200486	IFX	cDMARD experienced	Not in NMA (14-week study, no valid comparator arm)
Durez et al., 2007120	IFX	MTX naive	Yes
ERA87	ETN	MTX naive	Yes
ETN Study 30988,89	ETN	cDMARD experienced	Yes
GO-BEFORE90	GOL	MTX naive	Yes
GO-FORTH91	GOL	cDMARD experienced	Yes
GO-FORWARD92	GOL	cDMARD experienced	Yes
GUEPARD93	ADA	MTX naive	Not in NMA (no ACR/EULAR data at 22–30 weeks)
HIT HARD94	ADA	MTX naive	Yes
IDEA95	IFX	MTX naive	Not in NMA (no ACR/EULAR data at 22–30 weeks)
CREATE IIb96	ETN	cDMARD experienced	Yes
JESMR97	ETN	cDMARD experienced	Yes
Kay et al., 200898	GOL	cDMARD experienced	Not in NMA [no eligible ACR/EULAR data at 22–30 weeks (owing to PBO group crossover)]
Kim et al., 200799	ADA	cDMARD experienced	Yes
Kume et al., 2011100	ADA, ETN	MTX naive	Not in NMA (early escape at 12 weeks with no imputation for missing data)
Lan et al., 2004101	ETN	cDMARD experienced	Not in NMA (12-week study)
LARA102	ETN	cDMARD experienced	Yes
MEASURE103	TCZ	cDMARD experienced	Not in NMA (no ACR/EULAR data at 22–30 weeks)
Moreland et al., 1999104/Mathias et al., 2000105	ETN	cDMARD experienced	Yes
Nishimoto et al., 2004106	TCZ	cDMARD experienced	Not in NMA (no ACR/EULAR data at 22–30 weeks)
OPERA107	ADA	MTX naive	Not in NMA (no ACR/EULAR data at 22–30 weeks)
OPTIMA108	ADA	MTX naive	Yes
PREMIER109	ADA	MTX naive	Yes
Quinn et al., 2005110	IFX	MTX naive	Not in NMA (no ACR/EULAR data at 22–30 weeks)
RACAT111/O’Dell et al., 2013112	ETN	cDMARD experienced	Yes
REALISTIC113	CTZ	cDMARD experienced	Not in NMA (no biologic-naive ACR/EULAR data at 22–30 weeks)
RED-SEA114	ADA, ETN	cDMARD experienced	Not in NMA (no ACR/EULAR data at 22–30 weeks)
SAMURAI115	TCZ	cDMARD experienced	Yes
SATORI116	TCZ	cDMARD experienced	Yes
STAR117	ADA	cDMARD experienced	Yes
START118	IFX	cDMARD experienced	Yes
Swefot119	IFX	cDMARD experienced	Yes
AiC information has been removed	AiC information has been removed	AiC information has been removed	AiC information has been removed
TOWARD121	TCZ	cDMARD experienced	Yes
Van De Putte et al., 2004122	ADA	cDMARD experienced	Yes
Wajdula 2000 (reported in Chen et al., 2006123)	ETN	cDMARD experienced	Not in NMA (12-week study)
Weinblatt et al., 1999124	ETN	cDMARD experienced	Yes
Wong et al., 2009125	IFX	cDMARD experienced	Not in NMA (no ACR/EULAR data at 22–30 weeks)
Zhang et al., 2006126	IFX	cDMARD experienced	Not in NMA (18-week study)

Sixty RCTs were included in the systematic review of clinical effectiveness. These comprised six trials with head-to-head comparisons of included biologic interventions, [academic-in-confidence (AiC) information has been removed], and 53 trials of biologic interventions compared with placebo (PBO) or cDMARDs.

Methotrexate-naive trial populations are considered separately in the following results section as population 1. For population 1 there were a total of 15 RCTs included in the systematic review (ABT n = 0, ADA n = 6, CTZ n = 0, ETN n = 2, GOL n = 1, IFX n = 5, TCZ n = 0 and head-to-head biologics n = 1). Eight of the MTX-naive trials had data available for the NMA. All these seven trials provided ACR data; however, only one90 contributed EULAR data for analysis. A head-to-head trial of ADA versus ETN was identified but this trial was not eligible for the NMA (due to early escape at 12 weeks with no imputation for missing data). 100

There were 45 trials with cDMARD-experienced populations (considered as populations 2 and 3) (ABT n = 3, ADA n = 7, CTZ n = 2, ETN n = 11, GOL n = 3, IFX n = 7, TCZ n = 6, head-to-head biologics n = 5 and grouped antiTNFs n = 1). Of these, 30 trials had data available for the NMA.

Twelve trials that did not satisfy the inclusion criteria for the systematic review (as outlined in Methods for reviewing effectiveness) were excluded from the systematic review but were used as additional evidence and explored in sensitivity analyses in the NMA (Table 6). These trials contributed ACR and/or EULAR data to sensitivity analyses only. Of these, 10 trials had populations with a small proportion that had received prior biologics (≤ 20%). The other remaining trials were not in the base case because they had populations in which some patients were MTX naive or cDMARD and others were not, or patients were responding to MTX.

TABLE 6 - Trials not eligible for the systematic review but providing additional evidence for NMA sensitivity analyses

ACQUIRE, subcutaneous abatacept versus intravenous abatacept; LITHE, tocilizumab safety and the prevention of structural joint damage methotrexate and sulfasalazine combination trial; OPTION, tOcilimumab Pivotal Trial In methotrexate inadequate respONders; ORAL Standard, Tofacitinib or Adalimumab versus Placebo in Rheumatoid Arthritis; RAPID, Rheumatoid Arthritis Prevention of structural Damage.

Trial name/study	Intervention	Allocated population	Rationale for ineligibility in systematic review
ACQUIRE127	ABT	cDMARD experienced	3.4–6% prior biologics
AMBITION55,128	TCZ	cDMARD experienced	5–9% prior biologics, mix of MTX naive and prior MTX
Yamamoto et al., 2011129	CTZ	cDMARD experienced	16% prior biologics
LITHE130	TCZ	cDMARD experienced	11% prior biologics
NCT00254293131	ABT	cDMARD experienced	2.6% prior biologics
OPTION132	TCZ	cDMARD experienced	5–9% prior biologics
ORAL Standard133	ADA, TOF	cDMARD experienced	10% prior biologics
RA0025134	CTZ	cDMARD experienced	15% prior biologics
RAPID1135	CTZ	cDMARD experienced	4% prior biologics
RAPID2136	CTZ	cDMARD experienced	1.6% prior biologics
TEAR53	ETN	cDMARD experienced and MTX naive	Mix of MTX-naive and prior MTX, some patients (less than 30%) had any prior cDMARD use
TEMPO54	ETN	cDMARD experienced and MTX naive	Mix of MTX-naive, and prior MTX but not inadequate response
Kremer et al., 2012137	TOF	cDMARD experienced	Did not include any bDMARD within the NICE scope
van der Heijde et al., 2013138	TOF	cDMARD experienced	Did not include any bDMARD within the NICE scope

In addition, two trials providing supplementary network linkages were included in the NMA. These RCTs did not include any of the included interventions as specified in the decision problem, but evaluated TOF versus PBO. 137,138 Both of these trial populations had some prior biologic use (and therefore these trials were considered within the NMA sensitivity analyses).

Quality of research available

The quality of the included RCTs is presented in Table 345 (see Appendix 4) and summarised in Figure 3. There is a reasonably low risk of bias overall among studies included in this review. Items where risk of bias was greatest were those that assessed comparability of groups, blinding and selective reporting. Items generating a large proportion of ‘unclear’ responses (indicating a lack of clarity in reporting) were those relating to generation of allocation sequence, allocation concealment and selective reporting of outcomes. Items with a low risk of bias in a large proportion of trials were comparability at baseline, blinding, analysis by allocated treatment group and most (≥ 80%) participants randomised included in the final analysis. A modified intention-to-treat population was used in around half of trials for efficacy and safety analyses (which was typically based on all randomised patients who received at least one dose of study drug being included in analyses).

FIGURE 3.

Risk-of-bias graph. mITT, modified intention to treat.

Summary of trials and population characteristics

There were some differences between trials in population characteristics, treatment and trial duration. For some trials, intervention and control arms differed in terms of numbers/combinations of concomitant cDMARDs. Some trials allowed physician discretion in other therapies. There was some variation between trials in prior treatment history and disease duration. There was some variation in how early withdrawals were decided, with variation in length of time on allocated treatment.

Assessment of effectiveness

Additional evidence (trial data not eligible for full systematic review but included to inform network meta-analysis sensitivity analyses for populations 2 and 3)

Study and population characteristics for the trials ineligible for the full systematic review but provided as additional evidence to inform sensitivity analyses are presented in Table 344 (see Appendix 4). Two RCTs137,138 in which TOF was evaluated were included as evidence to supplement the network.

The ACR and EULAR responses in populations 2 and 3 RCTs used in the sensitivity analyses are presented in Tables 17 and 18 respectively.