3 The manufacturer's submission
3.1 The main clinical effectiveness data were derived from a single phase III randomised controlled trial (RCT) – GO-REVEAL. The trial compared golimumab with placebo for the treatment of active and progressive psoriatic arthritis in people who had symptoms despite the use of current or previous DMARDs or non-steroidal anti-inflammatory drugs. Of the 405 trial participants, 113 were randomised to placebo, 146 were randomised to a 50 mg dose of golimumab and 146 were randomised to a 100 mg dose of golimumab. Randomisation was maintained for 24 weeks. Upward titration was allowed at week 16, such that the participants in the placebo group could switch to 50 mg golimumab and those in the 50 mg golimumab group could have their dose increased to 100 mg if their disease had failed to respond. In the placebo group 50% of participants crossed over to golimumab 50 mg treatment and in the golimumab 50 mg group 20% crossed over to golimumab 100 mg treatment. Outcomes were assessed at 14 and 24 weeks.
3.2 The primary outcomes in GO-REVEAL were American of Rheumatology (ACR) 20 response at week 14 and the change from baseline in the psoriatic arthritis modified van der Heijde-Sharp (vdH-S) score at week 24. Secondary outcomes included ACR 20 response at week 24, Psoriatic Arthritis Response Criteria (PsARC) response at weeks 14 and 24, and Psoriasis Area and Severity Index (PASI) 75 improvement at week 14 in participants with psoriasis that affected 3% or more of their body surface area at baseline. Physical functional status was measured by Health Assessment Questionnaire (HAQ) score at week 24. Health-related quality of life was measured by the Short Form 36 Health Survey (SF-36) at week 14.
3.3 The week 14 results of GO-REVEAL indicated that, compared with placebo, golimumab showed a statistically significant improvement in joint disease. An ACR 20 response was seen in 50.7% of participants in the 50 mg treatment arm compared with 8.8% in the placebo arm (relative risk [RR] 5.727, 95% confidence interval [CI] 3.24 to 10.56). A PsARC response was seen in 73.3% of participants in the 50 mg treatment arm compared with 21.2% in the placebo arm (RR 3.451, 95% CI 2.46 to 4.87). Golimumab also showed a statistically significant improvement in skin disease as measured by PASI 75 at both 14 and 24 weeks. A PASI 75 response was seen in 40.4% of participants in the 50 mg treatment arm compared with 2.5% in the placebo arm (RR 15.945, 95% CI 4.62 to 59.11) at 14 weeks, and in 55.9% of participants in the 50 mg treatment arm compared with 1.4% in the placebo arm (RR 40.794, 95% CI 7.86 to 232.88) at 24 weeks. There was also a statistically significant improvement in functional status (HAQ) at 24 weeks. A mean HAQ score change from baseline of 0.33 (standard deviation [SD] 0.55, p < 0.001) was observed in the golimumab 50 mg arm compared with −0.01 (SD 0.49) in the placebo arm. Data on HAQ score change from baseline were not available for the 14-week time point.
3.4 The manufacturer reported that golimumab 50 mg produced a statistically significant reduction from baseline in vdH-S score of 0.16 (p = 0.01) at 24 weeks compared with placebo. The reduction from baseline in vdH-S score was not statistically significant in the golimumab 100 mg group (p = 0.09). The manufacturer did not report vdH-S scores at the 14 week time point.
3.5 The Evidence Review Group (ERG) reported that the main limitation of the efficacy evaluation of golimumab was that the analyses of efficacy outcomes were restricted to the GO-REVEAL trial, which had a limited sample size and was of limited duration (see section 3.1).
3.6 The manufacturer stated that the most frequently reported adverse events associated with golimumab therapy were infections and infestations, including upper respiratory tract infections and nasopharyngitis. The manufacturer reported that the safety profile of golimumab was comparable to that of the other TNF inhibitors adalimumab, etanercept and infliximab.
3.7 The ERG reported concerns about the adverse event data presented for golimumab. It noted that no long-term adverse event data had been presented, and that in its original submission the manufacturer had not included adverse event data on golimumab from controlled studies of its use in other conditions such as rheumatoid arthritis or ankylosing spondylitis. The ERG reported that the manufacturer's conclusion that golimumab has a safety profile comparable to that of the other TNF inhibitors may be premature.
3.8 Following consultation on the Appraisal Consultation Document, the manufacturer submitted evidence on the long-term safety of golimumab. These data included 104-week results from the GO-REVEAL extension study in addition to 52- and 104-week safety data in trial participants with psoriatic arthritis, rheumatoid arthritis and ankylosing spondylitis who had received treatment with golimumab across all of the original phase III studies. These data were marked as confidential and therefore cannot be reported.
3.9 In the absence of head-to-head comparisons between golimumab and the other TNF inhibitors, the manufacturer conducted a mixed treatment comparison. The mixed treatment comparison included seven trials: the GO-REVEAL trial (golimumab versus placebo); two RCTs comparing etanercept with placebo (Mease 2000 and Mease 2004); two RCTs comparing infliximab with placebo (IMPACT and IMPACT 2); and two RCTs comparing adalimumab with placebo (ADEPT and Genovese 2007). All of the TNF inhibitors have marketing authorisations for the treatment of active and progressive psoriatic arthritis that has responded inadequately to previous DMARDs.
3.10 The trials included in the mixed treatment comparison were similar in terms of joint disease severity at baseline (for example, mean tender joint count and mean swollen joint count). There were differences, however, in the proportions of trial participants who could be evaluated for psoriasis endpoints at baseline. Most participants had received treatment with one prior DMARD, although no trial specified non-response to at least two DMARDs.
3.11 The outcomes included in the mixed treatment comparison analyses were PsARC response, change in HAQ score given PsARC response to treatment, change in HAQ score given no PsARC response, and change in PASI in people with psoriasis that affected 3% or more of their body surface area at baseline. The manufacturer selected absolute changes as the main outcomes, stating that these were the most appropriate outcomes for economic modelling. No analysis of the ACR outcomes was included in the mixed treatment comparison.
3.12 The results of the mixed treatment comparison indicated that of the four TNF inhibitors, golimumab was associated with the third highest PsARC response and absolute change in PASI from baseline. Of the four TNF inhibitors, golimumab had the lowest HAQ score change from baseline, both in people whose disease responded to treatment based on PsARC score and in those whose disease did not respond. The numerical values for each outcome derived from the mixed treatment comparison were marked as confidential and therefore cannot be reported.
3.13 The ERG reported that the network of trials included in the mixed treatment comparison was appropriately constructed, but that there were differences among the trial populations in disease severity and number of previously tried DMARDs (with many participants having received only one previous DMARD). The ERG commented that the trial populations were not precisely representative of the population with active and progressive psoriatic arthritis for whom TNF inhibitors are recommended in current British Society for Rheumatology guidelines and in Etanercept, infliximab and adalimumab for the treatment of psoriatic arthritis (NICE technology appraisal guidance 199 [TA199]).
3.14 The manufacturer developed its own economic evaluation, which comprised a patient cohort model. The model compared the effects of treatment with golimumab (50 mg) in adults with active and progressive psoriatic arthritis whose disease had responded inadequately to DMARDs with the effects of treatment with infliximab, adalimumab and etanercept and with palliative care. All people entered the model with the same baseline characteristics as participants in the GO-REVEAL trial and left the model at death, irrespective of the treatment regimen. The model used a 12-week cycle for the first two cycles and annual cycles thereafter. The model captured response to treatment using HAQ score (conditional on PsARC response) as the arthritis measure and PASI score as the psoriasis measure. If there was no response to treatment at 12 weeks (according to PsARC), treatment was discontinued. The price year used for costs was not reported in the manufacturer's submission. Costs and benefits were discounted at 3.5% per annum over 40 years.
3.15 The manufacturer reported that estimates of treatment effectiveness – including PsARC response, HAQ score changes from baseline for people whose disease had responded to treatment according to PsARC at 12 weeks, HAQ score changes from baseline for those whose disease had not responded to treatment according to PsARC at 12 weeks, and PASI change from baseline in people with measurable psoriasis – were derived from the mixed treatment comparison.
3.16 The model assumed that people who continue treatment with a TNF inhibitor maintain their initial improvement in HAQ score. The same ongoing rate of withdrawal from treatment was used for all the TNF inhibitors (16.5% per annum) and represented withdrawal because of treatment failure or adverse events.
3.17 The manufacturer combined data from IMPACT2 (a study of infliximab) and GO-REVEAL using the 'Gray' algorithm to estimate utility values. The Gray algorithm converts Short Form 36 (SF-36) data to EuroQol (EQ-5D) estimates and then to utilities. The disutility of adverse events was not modelled.
3.18 The manufacturer reported that resource use associated with treatment, administration and monitoring of infliximab, etanercept and adalimumab was taken from the Assessment Group's model for TA199. In the patient access scheme (as described in section 2.4) the manufacturer provides the 100 mg dose of golimumab at the same cost as the 50 mg dose. Therefore, only the cost of the 50 mg dose of golimumab (£774.58) was included in the model. The model contained an additional 4 hours of staff nurse costs for training people to self-administer subcutaneous TNF inhibitors. The costs of infliximab were initially calculated on the assumption that vial sharing was allowed (using an average of 3.5 vials per infusion, although this assumption was later removed following a request for clarification from the ERG). The costs associated with adverse events were not included.
3.19 The manufacturer revised its original base-case estimates in response to a request from the ERG for clarification about the way utilities were calculated and for the removal of the infliximab vial sharing assumption. The revised base-case results produced total costs, total quality-adjusted life years (QALYs) and incremental cost-effectiveness ratios (ICERs, pairwise comparisons with palliative care) as follows:
palliative care: total costs of £62,224 and total QALYs of 6.61
adalimumab: total costs of £86,410 and total QALYs of 7.89, resulting in an ICER of £18,824 per QALY gained
golimumab: total costs of £94,151 and total QALYs of 8.21, resulting in an ICER of £19,993 per QALY gained
etanercept: total costs of £94,578 and total QALYs of 8.49, resulting in an ICER of £17,177 per QALY gained
infliximab: total costs of £106,620 and total QALYs of 8.49, resulting in an ICER of £23,578 per QALY gained.
3.20 The ERG reported that the manufacturer had not provided an incremental analysis in which dominated and extendedly dominated options were excluded. An option is 'dominated' if there is another option that is less costly and more effective. An option is 'extendedly dominated' when its ICER is higher than that of the next, more effective, option when compared with a common baseline (that is, it is dominated by a combination of two other alternatives). The ERG recalculated the manufacturer's base-case results by incrementally comparing each treatment with the next, more effective, option and excluding those that were extendedly dominated. The recalculated base-case results showed that both adalimumab and golimumab were extendedly dominated by a combination of etanercept and palliative care. Etanercept in comparison with palliative care was associated with an incremental cost of £32,354 and an incremental QALY gain of 1.88, resulting in an ICER of £17,209. Infliximab was dominated by etanercept.
3.21 The manufacturer conducted two subgroup analyses: one of the population with 'predominantly' rheumatic disease and one of the population with 'significant' psoriasis. The ERG recalculated the results of these analyses as described in section 3.20. The results of the recalculated subgroup analyses show adalimumab and golimumab to be extendedly dominated by a combination of etanercept and palliative care. Etanercept in comparison with palliative care was associated with an incremental cost of £34,492 and an incremental QALY gain of 2.21, resulting in an ICER of £15,607 per QALY gained in the rheumatic disease subgroup. In the psoriasis subgroup, etanercept in comparison with palliative care was associated with an incremental cost of £31,564 and an incremental QALY gain of 2.25, resulting in an ICER of £14,028 per QALY gained. Infliximab was dominated by etanercept in the rheumatic disease subgroup, and was associated with an incremental cost of £5702 and an incremental QALY gain of 0.01, resulting in an ICER of £570,200 per QALY gained in comparison with etanercept in the psoriasis subgroup.
3.22 The ERG commented that the model structure was reasonable. The ERG stated that the inclusion of costs to cover time for training in self-injection may have been unnecessary, but reported that all other included costs were appropriate. The ERG considered that it may have been appropriate to account for the possibility of dose escalation to 100 mg (as per the marketing authorisation; see section 2.3). The ERG reported that the subgroup analyses were appropriate.
3.23 Full details of all the evidence are in the manufacturer's submission and the ERG report.