Baricitinib for moderate to severe rheumatoid arthritis

Baricitinib's marketing authorisation covers its use at 4 points in the treatment pathway, specifically in adults with:

• moderate, active rheumatoid arthritis that has not responded adequately to conventional disease-modifying antirheumatic drugs (DMARDs)

• severe, active rheumatoid arthritis that has not responded adequately to conventional DMARDs

• severe, active rheumatoid arthritis that has not responded adequately to biological DMARDs, including at least 1 tumour necrosis factor-alpha (TNF‑alpha) inhibitor

• severe, active rheumatoid arthritis that has not responded adequately to biological DMARDs, including at least 1 TNF‑alpha inhibitor and when rituximab is contraindicated or withdrawn because of adverse events.
  
  The committee also noted that the marketing authorisation includes the use of baricitinib alone or with methotrexate.

NICE currently recommends the use of the biological DMARDs in NICE's technology appraisal guidance on adalimumab, etanercept, infliximab, certolizumab pegol, golimumab, tocilizumab and abatacept (of which adalimumab, certolizumab pegol, etanercept, golimumab and infliximab are TNF‑alpha inhibitors), in combination with methotrexate, in people with severe rheumatoid arthritis that has not responded to intensive treatment with combinations of conventional DMARDs. Disease severity is assessed using the disease activity score (DAS28). A DAS28 of more than 5.1 indicates severe disease (between 3.2 and 5.1 indicates moderate disease, less than 3.2 but more than 2.6 indicates mild disease and less than 2.6 indicates disease remission). For people who meet these criteria but cannot take methotrexate, the guidance recommends that adalimumab, certolizumab pegol, etanercept or tocilizumab may be used as monotherapy.

For people with severe rheumatoid arthritis who have already had at least 1 TNF‑alpha inhibitor that hasn't worked, NICE's technology appraisal guidance on adalimumab, etanercept, infliximab, rituximab and abatacept recommends the biological DMARD rituximab in combination with methotrexate for treating severe active rheumatoid arthritis. If rituximab is contraindicated or withdrawn because of an adverse event, the guidance recommends abatacept, adalimumab, etanercept or infliximab in combination with methotrexate. NICE also recommends golimumab under the same circumstances in NICE's technology appraisal guidance on golimumab. If methotrexate is contraindicated or withdrawn because of an adverse event, NICE's technology appraisal guidance on adalimumab, etanercept, infliximab, rituximab and abatacept recommends adalimumab or etanercept as monotherapy. NICE's technology appraisal guidance on tocilizumab and certolizumab pegol recommend both treatments as alternatives to TNF‑alpha inhibitors in the same circumstances (that is, for people with severe rheumatoid arthritis who have already had at least 1 TNF‑alpha inhibitor that hasn't worked, in combination with methotrexate when rituximab is contraindicated or withdrawn. Certolizumab pegol is also recommended as monotherapy if methotrexate is contraindicated or withdrawn). NICE technology appraisal guidance also recommends tocilizumab in combination with methotrexate when neither TNF‑alpha inhibitors nor rituximab have worked.

The committee heard from the patient experts that rheumatoid arthritis is a lifetime condition that can severely reduce quality of life. The clinical experts stated that conventional DMARDs such as methotrexate are inadequate for many people. They added that the disease sometimes does not respond adequately to the first biological DMARD prescribed, and that there are few tools available to predict response to help decide which treatment to use. Both the clinical and patient experts said it would be helpful to have new treatments that can be used at various points in the treatment pathway, alongside biological DMARDs after failure of conventional DMARDs. The clinical and patient experts agreed that methotrexate is often not well tolerated; the clinical experts noted that up to a third of people who are prescribed methotrexate with biological DMARDs do not take methotrexate because of side effects. The clinical experts emphasised that baricitinib is a novel treatment with a different mode of action to the biological DMARDs. They noted that the selective inhibition of Janus kinase 1 and 2 will affect a broad range of cytokines involved in the pathogenesis of rheumatoid arthritis. The clinical experts also noted the fast kinetic action of baricitinib compared with biological DMARDs. Both the clinical and patient experts also highlighted that baricitinib is given orally, which has major benefits for both patients and the health system. The patient experts emphasised that this is an important factor for people who have difficulty injecting themselves because of the disease affecting their hands. The patient experts also noted that some current treatments have to be stopped if the person gets an infection, and that some treatments may cause injection site reactions. The committee recognised that rheumatoid arthritis significantly affects quality of life. It concluded that there is a need for new treatment options, particularly when there is an inadequate response to conventional or biological DMARDs.

The committee was aware that the company had analysed 4 distinct subgroups in which baricitinib could be used:

• People with moderate rheumatoid arthritis whose disease has responded inadequately to conventional DMARDs.

• People with severe rheumatoid arthritis whose disease has responded inadequately to conventional DMARDs.

• People with severe rheumatoid arthritis whose disease has responded inadequately to biological DMARDs and for whom rituximab is a treatment option.

• People with severe rheumatoid arthritis whose disease has responded inadequately to biological DMARDs and for whom rituximab is contraindicated or not tolerated.
  
  The relevant comparators varied by subgroup. The committee concluded that it was appropriate to consider the 4 groups separately and that the company had broadly included the appropriate comparators.

The company's clinical evidence came from 4 phase 3 randomised controlled trials and 1 long-term safety and tolerability study. The trials included people with moderate to severe rheumatoid arthritis, as defined in section 3.2. The trials were:

• RA‑BEAM, which included people whose disease responded inadequately to methotrexate and who had not had biological DMARDs. Baricitinib 4 mg was given once daily and the comparators were placebo and adalimumab. Background methotrexate was given to all the groups.

• RA‑BUILD, which included people whose disease responded inadequately to conventional DMARDs and who had not had biological DMARDs. Baricitinib 2 mg or 4 mg was given once daily and the comparator was placebo. People taking conventional DMARDs with or without methotrexate before the study continued to take background therapy.

• RA‑BEACON, which included people whose disease responded inadequately to TNF‑alpha inhibitors. Baricitinib 2 mg or 4 mg was given once daily and the comparator was placebo. Background conventional DMARDs were given to all the groups.

• RA-BEGIN, which included people who had not had any conventional or biological DMARDs. Baricitinib 4 mg was given once daily, with or without methotrexate. The comparator was methotrexate. The committee was aware that the marketing authorisation for baricitinib does not include the treatment of rheumatoid arthritis in people who have not had any conventional or biological DMARDs (that is, this subgroup).
  
  The long-term safety and tolerability study, RA-BEYOND, included people with moderate or severe rheumatoid arthritis who took part in a separate phase 2b study or 1 of the 4 trials described above. Baricitinib 2 mg or 4 mg was given once daily, with or without conventional DMARDs.

The primary outcome of all the randomised controlled trials was the proportion of people achieving a 20% improvement in the American College of Rheumatology response criteria (ACR20) at week 12 or 24. Secondary outcomes included the proportion of people achieving a 50% or 70% improvement in the response criteria (ACR50 and ACR70 respectively), and the proportion of people meeting the European League Against Rheumatism (EULAR) response criteria. The committee concluded that the trials were relevant and adequate for its decision-making.

The committee considered RA‑BEAM and RA‑BUILD, which included people with moderate to severe rheumatoid arthritis which responded inadequately to conventional DMARDs. In RA‑BEAM, there was a significant increase in the proportion of people meeting the ACR20 criteria at 12 weeks with 4 mg baricitinib plus conventional DMARDs compared with conventional DMARDs alone (odds ratio [OR] 3.6; 95% confidence interval [CI] 2.7 to 4.7, p=0.001). A smaller response was seen with 4 mg baricitinib plus conventional DMARDs compared with adalimumab plus conventional DMARDs (OR 1.5; 95% CI 1.1 to 2.0, p=0.014 for ACR20). Significant improvements in ACR20 and EULAR good and moderate responses were also seen in RA‑BUILD for 4 mg baricitinib plus conventional DMARDs compared with conventional DMARDs alone (OR 2.5; 95% CI 1.7 to 3.7, p=0.001 for ACR20 and OR 3.5; 95% CI 2.3 to 5.4, p=0.001 for EULAR). The committee also noted that in RA‑BUILD, 2 mg baricitinib plus conventional DMARDs improved ACR20 and EULAR responses in this population compared with conventional DMARDs alone (OR 3.0; 95% CI 2.0 to 4.4, p=0.001 for ACR20 and OR 3.3; 95% CI 2.2 to 5.0, p=0.001 for EULAR good and moderate response). The committee concluded that 4 mg baricitinib plus conventional DMARDs has similar efficacy to adalimumab plus conventional DMARDs, and is more effective than conventional DMARDs alone in people with moderate to severe rheumatoid arthritis which has responded inadequately to conventional DMARDs.

The committee considered RA‑BEACON, which included people with moderate to severe rheumatoid arthritis which responded inadequately to TNF‑alpha inhibitors. There was a significant increase in the proportion of people meeting the ACR20 criteria and having a EULAR moderate or good response rate at 12 weeks for 4 mg baricitinib plus conventional DMARDs compared with conventional DMARDs alone (OR 3.4; 95% CI 2.2 to 5.4, p=0.001 for ACR20 and OR 3.6; 95% CI 2.3 to 5.7, p=0.001 for EULAR moderate and good response). The committee also noted that 2 mg baricitinib plus conventional DMARDs also improved ACR20 and EULAR response rates in this population compared with conventional DMARDs alone (OR 2.7; 95% CI 1.7 to 4.2, p=0.001 for ACR20 OR 2.7; 95% CI 1.8 to 4.2, p=0.001 for EULAR moderate and good response). The committee concluded that both dosages of baricitinib, when given with conventional DMARDs, are more effective than conventional DMARDs alone in people with moderate to severe rheumatoid arthritis which has responded inadequately to TNF‑alpha inhibitors.

The committee noted that across all 3 randomised controlled trials in which patients had previously had conventional or biological DMARDs (RA‑BEAM, RA‑BUILD and RA‑BEACON), the safety profile of baricitinib was similar to that of the conventional DMARDs. In addition, it noted that the safety profiles were found to be similar in the head-to-head comparison of baricitinib and adalimumab (RA‑BEAM).

The committee was aware that other than the direct comparison with adalimumab, the only evidence available on the comparative effectiveness of baricitinib and the biological DMARDs was from the company's network meta-analyses. The company did separate analyses for patients whose disease inadequately responded to either conventional or biological DMARDs, using ACR and EULAR outcome measures.

At 24 weeks' follow-up, for patients whose disease inadequately responded to conventional DMARDs, the network meta-analysis showed:

• Baricitinib plus conventional DMARDs gave better EULAR response rates than conventional DMARDs alone.

• Baricitinib plus conventional DMARDs gave similar EULAR response rates to the biological DMARDs plus conventional DMARDs.
  
  The exception to this was tocilizumab plus conventional DMARDs, which gave better EULAR result than all the other treatments. However, the clinical experts noted that the trials of tocilizumab had slightly different characteristics than the trials of the other technologies, and they considered tocilizumab to have similar effectiveness to the other biological DMARDs. Tocilizumab monotherapy showed similar results to baricitinib and the biological DMARDs when used with conventional DMARDs.
  
  At 24 weeks' follow-up, for patients whose disease inadequately responded to biological DMARDs, the network meta-analysis showed:

• Baricitinib (2 mg and 4 mg) plus conventional DMARDs gave better EULAR response rates than conventional DMARDs alone. A dose response was seen, with 4 mg baricitinib having a better EULAR response than 2 mg baricitinib.

• Rituximab plus conventional DMARDs gave a better point estimate of the EULAR response rate than baricitinib (2 mg and 4 mg) plus conventional DMARDs.

The committee heard from the ERG that there were problems with the methods used in the company's network meta-analysis. These included the conversion of ACR data to EULAR data before synthesis, the use of simultaneous models for baseline and treatment effects, the use of a random effects model for 1 population and a fixed effects model for the other, and poor model fit. In addition, the company had pooled the control data inappropriately. The ERG corrected the errors in the company's network meta-analysis. Having reviewed both analyses, the committee concluded that the results of the corrected network meta-analysis and the company's network meta-analysis were broadly comparable.

The company identified 9 UK-based cost-effectiveness studies. The committee was aware that 8 of these were associated with previous NICE technology appraisals guidance; 1 was an independent published review. The company did not identify any studies that included baricitinib, but the committee noted that the studies were nonetheless relevant and appropriate.

The company used an individual patient-based discrete event simulation model for its economic evaluation. The model simulates patients' disease progression through the sequences of treatments being compared. It was based on the model used by the assessment group during the production of NICE's technology appraisal guidance on adalimumab, etanercept, infliximab, certolizumab pegol, golimumab, tocilizumab and abatacept for rheumatoid arthritis. The model categorised patients based on their EULAR response (good, moderate or no response) at 6 months. Response rates were based on the company's network meta-analysis. The company analysed cost effectiveness for each of the subgroups described in section 3.5. The committee concluded that the model structure was appropriate for its decision-making.

The company's model included costs associated with drug acquisition, drug administration and monitoring, and hospitalisation. The committee was aware that baricitinib and several of the biological DMARDs have patient access schemes. It noted that the company had incorporated the patient access scheme prices for baricitinib, certolizumab pegol and golimumab in the model, but not the confidential patient access schemes for abatacept and tocilizumab. The incremental cost-effectiveness ratios (ICERs) that incorporated these confidential patient access schemes cannot be reported here, but the range of ICERs usually considered to be cost effective is from £20,000 to £30,000 per quality-adjusted life year (QALY) gained. The company had also calculated the average cost of drug doses using the average weight, rather than the distribution of the weight of the modelled patient population. The committee was also aware that the company overestimated the number of doses and therefore the costs of infliximab.

The company did not identify any evidence on the effectiveness of adalimumab, certolizumab pegol, etanercept or infliximab plus conventional DMARDs in patients with severe active rheumatoid arthritis which has responded inadequately to biological DMARDs when rituximab is contraindicated or not tolerated. In the absence of these data, the company used the same efficacy estimates for these treatments as those in patients with severe active rheumatoid arthritis which has responded inadequately to conventional DMARDs. The EULAR responses for all treatments were higher in these patients than in those with an inadequate response to biological DMARDs. The committee heard from the ERG that because of this, the company's base case is likely to have overestimated the efficacy of adalimumab, certolizumab pegol, etanercept and infliximab plus conventional DMARDs in patients with active rheumatoid arthritis which has responded inadequately to biological DMARDs when rituximab is contraindicated or not tolerated. The committee accepted this in the absence of any other evidence.

Health-related quality of life data were collected using a health assessment questionnaire (HAQ) in RA‑BEAM, RA‑BUILD and RA‑BEACON. Patient-level responses were converted to utility index-based EQ-5D-5L scores using the UK-specific scoring algorithm reported by Hernandez Alava et al. (2012). The committee was aware this was not in line with the analysis done during NICE's technology appraisal on adalimumab, etanercept, infliximab, certolizumab pegol, golimumab, tocilizumab and abatacept, but it heard from the ERG that this approach was unlikely to change the overall conclusions.

The ERG identified several issues with the company's economic analyses including:

• Limitations with the company's network meta-analysis because an inappropriate random effects model was assumed for the baselines. In addition, simultaneous baseline and treatment effect models were used without ensuring that information in the baseline model did not propagate to the relative treatment effect model. Furthermore, studies that reported EULAR responses were synthesised along with converted EULAR response outcomes from studies that only reported ACR responses.

• A lack of face validity in several of the scenario analyses, partly because of transcription and programming errors.

• Limitations with the probabilistic sensitivity analyses because of programming errors, including an error which resulted in patients having some biological DMARDs never achieving a good or moderate EULAR response.

• Using the efficacy of treatments in the population with an inadequate response to conventional DMARDs for all biological DMARDs in the treatment sequence, regardless of their position in the sequence.

• Rounding HAQ scores to the nearest valid HAQ score, rather than allowing HAQ scores to be sampled based on a continuous HAQ value.

• Incorrect implementation of the HAQ trajectory classes by assigning each patient to a single class based on the probability of class membership, instead of using an average weighted by the probability of class membership.

• Assuming that patients who achieve a moderate or good EULAR response at 24 weeks had an instant reduction in HAQ score when starting treatment.

• Averaging HAQ across large time periods, which may lead to inaccurate results because the relationships between HAQ score and EQ-5D and between HAQ score and hospitalisation costs are not linear.

• Excluding intravenous abatacept and subcutaneous tocilizumab from the list of comparators.

• Using a less accurate method to map HAQ scores to EQ-5D (Malottki et al. 2011) than that used during the NICE technology appraisal guidance on adalimumab, etanercept, infliximab, certolizumab pegol, golimumab, tocilizumab and abatacept (Hernandez Alava et al. 2013).

• Assuming that baricitinib would be used before intensive therapy with conventional DMARDs for patients with moderate rheumatoid arthritis (this was not supported by the clinical experts).

• Re-estimating the age of death at every event, which resulted in slightly different expected life years, which would be exacerbated by sequences of different lengths.

• Using the average weight of the population in the relevant trials to calculate average dose, which assumes there is a linear relationship between weight and dose costing and does not take into account drug wastage, for example.

• Overestimating the average number of doses, and thereby the cost, of infliximab that would be given in a year.
  
  The ERG stated that these errors were unlikely to change the broad conclusions of the company's model. The committee concluded that although there were several errors in the company's economic model, it was adequate for its decision-making.

The company carried out several scenario analyses. In one, the company assumed that patients having conventional DMARDs or palliative care had a linear increase in their HAQ scores at a yearly rate of 0.045 and 0.060 respectively (based on Malottki et al. 2011), instead of using the latent class approach. For the moderate rheumatoid arthritis population, the ICER for baricitinib plus conventional DMARDs compared with intensive conventional DMARDs alone decreased from £37,420 to £20,965 per QALY gained. In the severe rheumatoid arthritis population, the ICERs were slightly lower for the most effective drugs. The company also:

• adjusted the HAQ score for baricitinib plus methotrexate so that it deteriorated at half of the rate assumed for conventional DMARDs

• used HAQ score improvements for baricitinib calculated from trial data rather than the UK rheumatoid arthritis database

• used a different time to treatment discontinuation for patients on baricitinib

• used alternative methods to map HAQ scores to the EQ-5D

• accounted for serious adverse events

• tapered baricitinib from 4 mg once daily to 2 mg once daily.
  
  The committee heard from the ERG that these scenarios were unlikely to change the results of the cost-effectiveness analysis. The committee agreed that any exploratory analyses would not change its conclusion that baricitinib is broadly comparable to the other biological DMARDs recommended by NICE.

In the moderate active rheumatoid arthritis population whose disease has responded inadequately to conventional DMARDs, the company's base-case ICER for the baricitinib sequence compared with the conventional DMARD sequence was £37,420 per QALY. The committee noted that the company used a different sequence for this population to that used in NICE's technology appraisal guidance on adalimumab, etanercept, infliximab, certolizumab pegol, golimumab, tocilizumab and abatacept. The ERG did not correct this or the errors in the model because they were unlikely to change the conclusions, and it could use the model from the other appraisal as a reference. The ERG noted that the median ICER of biological DMARDs in the NICE appraisal of adalimumab, etanercept, infliximab, certolizumab pegol, golimumab, tocilizumab and abatacept was around £50,000 per QALY gained. Taking into account the cost-effectiveness evidence for baricitinib in patients with moderate active rheumatoid arthritis whose disease has responded inadequately to conventional DMARDs, the committee considered that baricitinib plus conventional DMARDs did not have plausible potential to be cost effective in this population.

In the company's base-case analysis for the severe rheumatoid arthritis population whose disease has responded inadequately to conventional DMARDs, baricitinib plus conventional DMARDs dominated all its comparators (that is, it was both less costly and more effective). The exception to this was certolizumab pegol plus conventional DMARDs, which had an ICER of £18,400 per QALY compared with baricitinib plus conventional DMARDs. The committee noted that there are confidential patient access schemes in place for subcutaneous abatacept and intravenous tocilizumab, which the company did not include in its analysis. The ERG calculated new ICERs using the confidential comparator prices. The committee noted that all the comparisons produced very similar estimates of clinical and cost effectiveness, and concluded to recommend baricitinib plus conventional DMARDs as a cost-effective use of NHS resources for people with severe rheumatoid arthritis whose disease has responded inadequately to conventional DMARDs.

In the company's base-case analysis for the severe rheumatoid arthritis population whose disease has responded inadequately to biological DMARDs and for whom rituximab is a treatment option, baricitinib plus conventional DMARDs was dominated by rituximab plus conventional DMARDs (that is, it was more costly and less effective). The committee concluded that baricitinib plus conventional DMARDs was not a cost-effective use of NHS resources for people with severe rheumatoid arthritis whose disease has responded inadequately to biological DMARDs if rituximab is a treatment option.

In the pairwise analysis for the severe rheumatoid arthritis population whose disease has responded inadequately to biological DMARDs and for whom rituximab is contraindicated or not tolerated, baricitinib plus conventional DMARDs was dominated by golimumab plus conventional DMARDs. Compared with all other comparators, the ICERs ranged from £16,201 to £484,782. In the full incremental analysis, baricitinib plus conventional DMARDs dominated or extendedly dominated all comparators except for certolizumab pegol plus conventional DMARDs, which had an ICER of £16,201 per QALY gained. The ICERs for biosimilar etanercept plus conventional DMARDs compared with baricitinib plus conventional DMARDs, and adalimumab plus conventional DMARDs compared with baricitinib plus conventional DMARDs, were also less than £30,000 per QALY gained. The committee again noted the confidential patient access schemes in place for subcutaneous abatacept and intravenous tocilizumab, which the company did not include in its analysis. The ERG calculated new ICERs using the confidential comparator prices. The committee noted that all the comparisons produced very similar estimates of clinical and cost effectiveness, and concluded to recommend baricitinib plus conventional DMARDs as a cost-effective use of NHS resources for people with severe rheumatoid arthritis whose disease has responded inadequately to biological DMARDs and for whom rituximab is not a treatment option.

The committee was aware that the marketing authorisation for baricitinib includes its use as a monotherapy, but that the company did not present an economic analysis for baricitinib alone for patients who cannot have methotrexate. The committee noted that the only available evidence for baricitinib alone is in people who have not had conventional DMARDs, which is outside of its marketing authorisation. The committee recognised the considerable uncertainty about the effectiveness of baricitinib alone in people whose rheumatoid arthritis has had an inadequate response to conventional or biological DMARDs. The committee heard from the ERG that data from RA-BEGIN showed that the addition of methotrexate to 4 mg baricitinib produced similar ACR scores compared with baricitinib alone. The committee agreed that baricitinib monotherapy provides similar clinical efficacy to baricitinib plus conventional DMARDs. It concluded that its recommendations for baricitinib plus conventional DMARDs should also apply to baricitinib alone.