Golimumab for the treatment of rheumatoid arthritis after the failure of previous disease-modifying anti-rheumatic drugs

Clinical effectiveness

4.5 The Committee considered the evidence on the clinical effectiveness of golimumab in combination with methotrexate and noted that the manufacturer's submission considered golimumab at two positions in the treatment pathway – after treatment with conventional DMARDs only, and after treatment with both conventional DMARDs and a TNF inhibitor. The Committee heard from clinical specialists how golimumab would fit into the current treatment pathway. It heard that golimumab may be used either as a first TNF inhibitor therapy in people whose disease has not responded to conventional DMARD therapy, or as a second TNF inhibitor therapy in people who have had previous therapy with a TNF inhibitor. Following comments received during consultation regarding the marketing authorisation for golimumab, the manufacturer was asked to confirm whether the marketing authorisation for golimumab includes people who have had previous therapy with a TNF inhibitor. The manufacturer stated that golimumab was approved on the basis of the GO‑FORWARD and GO‑AFTER studies and that its use in people who have had previous therapy with a TNF inhibitor is consistent with the marketing authorisation and the evidence. The Committee concluded that the two positions in the treatment pathway as included in the manufacturer's submission were appropriate to be considered in this appraisal.

4.6 For people who have previously had only conventional DMARDs, the Committee considered the evidence from the two placebo-controlled trials of golimumab in combination with methotrexate (GO‑FORWARD and Kay et al. 2008). It noted that golimumab in combination with methotrexate had greater clinical effectiveness than placebo in combination with methotrexate. The Committee then discussed the mixed treatment comparison presented by the manufacturer in the absence of head‑to‑head trials comparing the efficacy of golimumab with that of the other available TNF inhibitors. The Committee noted that the mixed treatment comparison suggested that there were no statistically significant differences in ACR20, ACR50 and ACR70 response rates between golimumab and the other TNF inhibitors, and that the credibility intervals around the estimates were wide. The Committee heard from clinical specialists that they considered the different TNF inhibitors to have broadly similar efficacy. The Committee discussed the potential heterogeneity between the studies included in the comparison, recognising concerns about the comparability of the certolizumab pegol studies. It further noted comments received in consultation that it was inappropriate to include the TEMPO study and the TNF inhibitor monotherapy studies in the mixed treatment comparison. However, the Committee noted the sensitivity analyses performed by the ERG, which showed that the exclusion of these studies did not significantly alter the estimates of cost effectiveness. The Committee concluded that, based on the ACR response rates, golimumab had been demonstrated to be more effective than placebo and that there was no convincing evidence that golimumab was either more or less effective than the other TNF inhibitors.

4.7 The Committee considered the evidence on the clinical effectiveness of golimumab compared with placebo for the people who had had previous treatment with both conventional DMARDs and a TNF inhibitor. It noted that there was a single trial (GO‑AFTER) comparing golimumab with placebo and this trial showed that golimumab had greater clinical effectiveness than placebo. The Committee discussed the indirect comparison of golimumab and rituximab performed in the absence of head-to-head trials comparing the efficacy of golimumab with that of rituximab. It agreed that rituximab is an appropriate comparator for this population, although it was aware that since the manufacturer's submission NICE has published technology appraisal guidance recommending the use of tocilizumab, abatacept and a second TNF inhibitor in certain people who have had previous treatment with a TNF inhibitor. The Committee concluded that although the point estimates favoured rituximab, the indirect comparison did not demonstrate any statistically significant differences in clinical efficacy between golimumab and rituximab. The Committee noted that the additional analyses provided by the manufacturer included cost-effectiveness results for the comparison of golimumab, tocilizumab and abatacept in people who have had previous treatment with a TNF inhibitor. However, the Committee noted that separate data on the clinical effectiveness of golimumab compared with tocilizumab and abatacept were not provided.

4.8 The Committee discussed the long-term data for ACR response and proportion of people maintaining a HAQ improvement equal to or greater than 0.25 in the DMARD-experienced population in the GO‑FORWARD trial. The Committee noted limitations to the data, specifically that the trial had a placebo-controlled phase only up to 24 weeks, and included participants in the placebo arm who had crossed over to golimumab at week 14 because their disease was inadequately controlled. Despite these limitations the Committee agreed that the data suggested that the efficacy of golimumab was maintained over the long term. The Committee also discussed the long-term SF‑36 data submitted by the manufacturer and accepted that golimumab in combination with methotrexate had been shown to have a positive benefit on health-related quality of life compared with placebo. The Committee concluded that these data suggested that efficacy of golimumab was maintained.

4.9 The Committee considered clinical-effectiveness evidence for subgroups of people in the GO‑FORWARD trial who had either moderately or severely active rheumatoid arthritis as defined by their baseline DAS28 score. It noted that the analysis was in a small number of people, particularly the subgroup with moderately active rheumatoid arthritis. It further noted that the analysis was post hoc although it had been provided in line with the scope for the appraisal. For these reasons the Committee concluded that there was uncertainty surrounding the results. It noted that the majority of clinical evidence is in people with severely active rheumatoid arthritis.

4.10 The Committee discussed the 52- and 104‑week radiographic progression data (measured by the vdH‑S) from the GO‑FORWARD study submitted by the manufacturer following consultation on the appraisal consultation document. It noted that these data showed no statistically significant difference from baseline in vdH‑S score between golimumab 50 mg and placebo. The Committee heard from the manufacturer that both groups in the trial had shown minimal radiographic progression, which meant that golimumab could not improve on the results seen in the placebo group. The Committee noted a number of explanations provided for the minimal progression in both groups, including the short placebo-controlled period, the use of radiographic outcomes as secondary endpoints in relation to the size of the study and lower baseline disease activity levels in the golimumab trials compared with trials of other biological treatments. The Committee then discussed the 52‑week radiographic progression data from the GO‑BEFORE study provided as supporting evidence for the GO‑FORWARD data. It noted that these data had been used to support the licence extension for golimumab. The Committee was not persuaded that the data had demonstrated an absence of underlying radiographic progression while on treatment with golimumab, but it concluded that the data demonstrated that the combination of golimumab and methotrexate reduced the rate of radiographic progression.

4.11 The Committee discussed the adverse events seen in the golimumab RCTs and the results from the mixed treatment comparison and the indirect comparison of golimumab and the comparators in both populations. It noted that the data from the mixed treatment and indirect comparisons suggested few statistically significant differences in relative risk between the treatments but that these were associated with considerable uncertainty. It heard from the clinical specialists that there are no long-term adverse event data for golimumab but that they expected the adverse event profile of golimumab to be no different from that of other TNF inhibitors. The clinical specialists suggested that since golimumab is administered once a month, there might be fewer adverse events compared with other TNF inhibitors as a result of the reduced frequency of administration. The Committee concluded that there was uncertainty surrounding the adverse event profile of golimumab because of the limited long-term data, but that golimumab's adverse event profile had not been shown to be different from that of other TNF inhibitors.

Cost effectiveness

4.12 The Committee considered the economic model that evaluated golimumab as part of a sequence of treatments in people who had had previous treatment with conventional DMARDs only and who had not had a previous TNF inhibitor. It noted that, on the whole, the model used similar assumptions to other models submitted in previous appraisals of TNF inhibitors in rheumatoid arthritis, but that there were some differences from the other models, for example the exclusion of ACR70 response data, alternative rates of disease progression while on treatment and alternative methods for deriving estimates of utility. The Committee noted that the ACR70 response data and rates of underlying disease progression, similar to those used in other NICE technology appraisals, had subsequently been appropriately included in a revised economic model.

4.13 The Committee considered the utility estimates incorporated in the original model, and noted that the utility was derived from the ACR response, which was converted to a change in HAQ score and then mapped to EQ‑5D. The Committee recognised that a similar approach to mapping had been used in previous NICE technology appraisals of biological treatments for rheumatoid arthritis. However, the Committee noted that this was different from the NICE reference case, which recommends inclusion of directly collected utility data. The Committee then discussed the sensitivity analysis submitted by the manufacturer following the consultation on the appraisal consultation document, using the SF‑36 data from the GO‑FORWARD study. It noted comments from the ERG about the method that the manufacturer used to generate the SF‑6D for the methotrexate group and that the ERG was unclear why this approach had been taken. The Committee considered that a more appropriate method for the analysis would have been to use the data from the placebo group directly. However, it concluded that the sensitivity analysis suggested that the methodology to derive the utility in the base-case analysis had not been shown to be unreasonable.

4.14 The Committee discussed the 100 mg dose, which is indicated for people who weigh more than 100 kg and whose rheumatoid arthritis has not responded after three or four doses of golimumab. It noted that evidence for the clinical and cost effectiveness of this dose was not included in the original submission. The Committee understood that even though the proportion of people who received this dose might be quite small, if the acquisition cost was included in the model the ICER for golimumab would be expected to be higher than that estimated in the base case presented by the manufacturer. The Committee noted that following consultation on the appraisal consultation document, the manufacturer did not submit any additional data regarding the 100 mg dose, but instead proposed a patient access scheme that would provide the 100 mg dose at the same cost as the 50 mg dose in people for whom the higher dose is suitable. The Committee recognised that the patient access scheme has been accepted by the Department of Health. The Committee considered that analyses should have been presented both with and without the proposed patient access scheme, but concluded that with the patient access scheme, the manufacturer's analysis including only the costs of the 50 mg dose could be used as a basis for decision making.

4.15 The Committee noted that the economic analysis from the manufacturer had assumed that there was no progression of disease while on treatment with a TNF inhibitor, but that there was progression while on treatment with conventional DMARDs and on palliative treatment. The Committee discussed the progression of disease while on treatment with TNF inhibitors for people who have had therapy with conventional DMARDs only. The Committee considered that an assumption of no progression while on treatment with a TNF inhibitor could be an overestimate of the benefits of treatment. However, it heard from clinical specialists that although no progression on treatment may appear optimistic, findings from long-term studies suggest that it is a reasonable assumption for people whose rheumatoid arthritis responds to treatment. The Committee recognised that similar assumptions had been made in other NICE technology appraisals of TNF inhibitor treatments for rheumatoid arthritis. The Committee was aware of the long-term radiographic progression data submitted by the manufacturer following consultation on the appraisal consultation document (see 4.9). It noted that these data showed that golimumab reduced the rate of radiographic progression, albeit in a different population. The Committee concluded that in line with NICE technology appraisals of other TNF inhibitors, it would be appropriate to consider the estimates of cost effectiveness that assumed no disease progression while on treatment with a TNF inhibitor. However, it considered that this assumption was uncertain and may overestimate the benefits of treatment.

4.16 The Committee then discussed the revised version of the economic model and sensitivity analyses submitted by the manufacturer that included ACR70 response data and a rate of disease progression while on palliative treatment of 0.06 HAQ score units per year. The Committee discussed the ERG's review of the revised model, noting that the ERG considered that the errors in the previous model had been corrected and changes implemented appropriately. The Committee noted that the ICERs for golimumab were at the upper end of the range of £25,000–£28,000 per QALY gained produced by other drugs in the class; however, the frequency of administration may generate additional health-related benefits. The Committee noted that the 100 mg dose of golimumab was not considered in the economic model, but that because of the patient access scheme (as described in 2.4), the cost of the 100 mg dose would be equal to that of the 50 mg dose. The Committee was persuaded that, on balance, with the patient access scheme golimumab could be considered a cost-effective option for the treatment of rheumatoid arthritis if used in the same way as other TNF inhibitors, as recommended in 'Adalimumab, etanercept and infliximab for the treatment of rheumatoid arthritis' (NICE technology appraisal guidance 130) and 'Certolizumab pegol for the treatment of rheumatoid arthritis' (NICE technology appraisal guidance 186).

4.17 The Committee considered the economic model for the group of people who have had previous treatment with both conventional DMARDs and a TNF inhibitor. It noted that the base-case analysis showed that rituximab was dominated by golimumab because golimumab was less costly and more effective. It was aware that in this analysis it was assumed that rituximab is re-administered every 6 months. The Committee heard that the ERG considered re-administration of rituximab every 9 months to be more reflective of clinical practice. The Committee further heard from clinical specialists that for people responding to rituximab treatment the re-treatment intervals would be greater than 6 months. The Committee heard from the ERG about the costs for the first year of rituximab treatment that are included in the model. For the first 6 months of treatment, 1.5 courses of rituximab are included and 1 course of rituximab is included for the second 6 months. The Committee heard that it is unclear why a greater number of courses are required in the first 6 months than in subsequent 6-month periods. The Committee concluded that the rituximab costs had been overestimated in the original economic model, and that a re-treatment interval of 9 months is more appropriate.

4.18 The Committee discussed the progression of disease while on treatment for people who have had previous treatment with conventional DMARDs and a TNF inhibitor. It noted that the manufacturer had assumed that the TNF inhibitors all stop progression of disease while on treatment, but that for rituximab it was assumed that the disease continues to worsen while on treatment by an increase of 0.045 per year in HAQ score. It noted that this is the same as the rate used for conventional DMARDs. The Committee heard from the ERG and clinical specialists that this underestimates the benefits of rituximab, and that it would have been more appropriate to assume that, for people whose disease responds to treatment, rituximab reduces the progression of disease to the same extent as the TNF inhibitors. The Committee was not persuaded that it is appropriate to assume a differential rate of underlying progression of disease between rituximab and golimumab, and concluded that this assumption overestimates the cost effectiveness of golimumab compared with rituximab.

4.19 The Committee discussed the results of the manufacturer's revised version of the economic model and the ERG's exploratory analyses for the group of people who have had previous treatment with both conventional DMARDs and a TNF inhibitor, which compared golimumab with rituximab. It agreed that the ERG's amendments to increase the time between treatment intervals for rituximab and remove the assumption of a differential rate of underlying progression of disease were appropriate. The Committee noted that when these assumptions were changed rituximab was associated with lower costs and more QALYs than golimumab. The Committee therefore concluded that golimumab would not be a cost-effective use of NHS resources in people who have had previous treatment with conventional DMARDs and a TNF inhibitor and for whom rituximab is an appropriate treatment option.

4.20 The Committee recognised that in August 2010 NICE published technology appraisal guidance recommending the TNF inhibitors adalimumab, etanercept, infliximab and abatacept, as well as tocilizumab, for people with rheumatoid arthritis who are unable to have rituximab therapy because of contraindications or if rituximab is withdrawn because of an adverse event ('Adalimumab, etanercept, infliximab, rituximab and abatacept for the treatment of rheumatoid arthritis after the failure of a TNF inhibitor' [NICE technology appraisal guidance 195] and 'Tocilizumab for the treatment of rheumatoid arthritis' [NICE technology appraisal guidance 198]). The Committee agreed that it was appropriate to consider this group of people and the treatment options now available to them. The Committee discussed the revised analyses submitted by the manufacturer and noted that these did not include the other TNF inhibitors (that is, adalimumab, etanercept and infliximab), but did include abatacept and tocilizumab. It further noted the manufacturer's rationale that the other TNF inhibitors could not be included because there were no data from RCTs for these agents in this position in the treatment pathway. The Committee noted that the ICERs for golimumab in comparison with methotrexate were similar to those for abatacept and tocilizumab. The Committee understood that both abatacept and tocilizumab had been recommended for this patient group in NICE technology appraisal guidance (NICE technology appraisal guidance 195 and NICE technology appraisal guidance 198), with most plausible ICERs of between £20,000–30,000 per QALY gained, and that the TNF inhibitors: adalimumab, etanercept and infliximab had also been recommended in this way with ICERs in this range. On balance the Committee considered that the evidence before it indicated that golimumab would be no less cost effective than the other TNF inhibitors when used in this population. Therefore the Committee concluded that with the patient access scheme, golimumab is an appropriate use of NHS resources in people with rheumatoid arthritis who are unable to have rituximab therapy because of contraindications or if rituximab is withdrawn because of an adverse event, if used in the same way as other TNF inhibitors, as recommended in 'Adalimumab, etanercept, infliximab, rituximab and abatacept for the treatment of rheumatoid arthritis after the failure of a TNF inhibitor' (NICE technology appraisal guidance 195).

Summary of Appraisal Committee's key conclusions