Upadacitinib for treating active psoriatic arthritis after inadequate response to DMARDs

3.1 The patient and clinical experts explained that psoriatic arthritis is a lifelong condition that seriously affects people's quality of life. It can develop at a young age and affects all aspects of a person's life, including education, career, relationships and family life. The patient experts explained that symptoms such as fatigue, pain and associated comorbidities such as inflammatory bowel disorders, cardiovascular disease and diabetes, can have a substantial physical and psychological effect. The clinical and patient experts explained that people with psoriatic arthritis have symptoms ranging from mild, non-destructive disease to erosive and deforming arthritis with substantial effects on physical functioning. Symptoms can include swollen fingers and toes through to inflammation of larger joints such as elbows and knees, joints in the back, and tendonitis. Skin and nail psoriasis also affect quality of life. The committee concluded that psoriatic arthritis substantially affects health-related quality of life.

3.2 The clinical experts explained that the main aim of treatment for active psoriatic arthritis is to control joint and connective tissue inflammation. This prevents joint damage progressing and the associated pain and disability. People will usually have treatment with non-steroidal anti-inflammatory drugs, corticosteroids and conventional disease-modifying antirheumatic drugs (DMARDs), such as methotrexate. In line with NICE's technology appraisal guidance on etanercept, infliximab and adalimumab, people are eligible for biological or small-molecule treatments if their disease is poorly controlled after 2 conventional DMARDs. Biological or small-molecule treatments include:

3.3 The company submission identified 4 subpopulations and presented analyses for 3 of these:

3.4 The SELECT-PsA 1 trial provides direct evidence of the efficacy of upadacitinib compared with adalimumab and upadacitinib compared with placebo (assumed to represent best supportive care). No direct evidence is available to allow comparison of upadacitinib with 8 out of the 9 comparators. The SELECT-PsA 2 trial provides direct evidence for a comparison of upadacitinib compared with placebo. No direct evidence is available to allow comparison of upadacitinib with 5 of the 6 comparators. Clinical advice to the ERG was that adalimumab is commonly the first biological DMARD prescribed after at least 2 conventional DMARDs. The ERG explained that the lack of direct clinical evidence for all other active comparators meant that the company did network meta-analyses (NMAs). In the company submission, the NMA results for those who have not previously had treatment with biological DMARDs (subpopulation 2) were also assumed to apply to people for whom TNFi treatments are contraindicated or not tolerated (subpopulation 4). But the ERG explained that the company presented no clinical evidence to suggest that the effectiveness of biological DMARDs for subpopulation 2 is the same as for subpopulation 4. The ERG also explained that there is no evidence from the NMAs to support the use of upadacitinib to treat the biological-experienced population who have received prior treatment with apremilast or tofacitinib. The committee noted that the company's approach was broadly similar to recent appraisals in the disease area, and concluded that the results from the NMAs should be taken into account for decision making.

3.5 To evaluate upadacitinib compared with comparator treatments, the company did NMAs for all main outcomes, for:

3.6 The committee noted that the company's model was based on that used in NICE's technology appraisal guidance on certolizumab pegol and secukinumab for the treatment of psoriatic arthritis. Using a Markov structure to capture all costs and outcomes associated with upadacitinib and the comparators, the model included up to 2 lines of active treatment before best supportive care. The company assumed an assessment for response to treatment after 12 weeks for upadacitinib. The committee noted this was consistent with the approach taken in previous NICE technology appraisal models. The ERG confirmed that the model structure was consistent with previous models used in NICE technology appraisals for psoriatic arthritis. But, using a limited number of active treatment lines may not represent NHS clinical practice. The number of treatment options (including best supportive care) that are available for the biological-naive, biological-experienced and TNFi-contraindicated populations are 9, 5 and 5, respectively. The committee recalled that because of the range of treatments and because the disease is varied and unpredictable there is no standard treatment sequence in the NHS (see section 3.2). People will almost always start treatment with conventional DMARDs such as methotrexate, and then move onto biological DMARDs if their disease is not adequately controlled. But the exact sequence of treatments is determined by the course of the disease for each person. The company agreed with the ERG that these are inherent limitations in the economic model, but emphasised that this was a standard model accepted in previous technology appraisals for psoriatic arthritis. The committee agreed that the model is limited in how accurately it represents the number and sequence of treatments used in clinical practice. But, the committee noted that it was consistent with previous NICE technology appraisals for psoriatic arthritis. The committee concluded that the model structure does not reflect NHS clinical practice but is appropriate for decision making. The committee also concluded that technology appraisals in the future should take into account the number and sequence of treatments used in clinical practice.

3.7 The ERG was aware that in NHS clinical practice, the TNFi-contraindicated population generally have more than 1 line of treatment, and best supportive care is generally not an appropriate first-line treatment option for this population. The cost-effectiveness results for the TNFi-contraindicated population should therefore be identical to the biological-naive population who had ustekinumab as a second-line treatment (after excluding TNFi treatments as first-line options). The ERG explored this as a scenario, but this did not alter the overall conclusions on the cost effectiveness of upadacitinib in the TNFi-contraindicated population. The company agreed that the scenario analysis done by the ERG was appropriate. The committee agreed with the company and ERG, and concluded that the model does not accurately reflect the treatment sequence for the TNFi-contraindicated population that would be seen in NHS clinical practice. The committee also concluded that the ERG's approach for modelling the treatment sequence for the TNFi-contraindicated population was more appropriate.

3.8 The health assessment questionnaire-disability index (HAQ-DI) is the quality-of-life measure used in the company's economic model. The company's model cost-effectiveness results are driven by PsARC response and HAQ-DI reduction conditional on PsARC. Therefore, the strength of the clinical-effectiveness evidence for these outcomes is central to the credibility of the cost-effectiveness results. The ERG identified several limitations with the implementation of HAQ-DI in the company's economic model. First, results for HAQ-DI conditional on PsARC were not available from the company NMAs for several comparators, and so these were sourced from previous technology appraisals in psoriatic arthritis. For the biological-naive population, these were results for certolizumab pegol, ixekizumab, secukinumab, and tofacitinib. For the biological-experienced population, these were ixekizumab, secukinumab (300 mg only), and tofacitinib. While the ERG recognises that no other sources of HAQ-DI change conditional on PsARC response are available, it explained that using results from different sources without appropriate adjustments adds uncertainty to the company's cost-effectiveness results. The company agreed with the ERG that no alternative data sources or approaches were possible. It explained that while this introduces some uncertainty, this represents a pragmatic approach that enabled reasonable effectiveness estimates and incremental cost-effectiveness ratios (ICERs) to be generated for all comparators. It noted that the same approach was accepted by the committee in the recent NICE technology appraisal guidance on tofacitinib for treating active psoriatic arthritis after inadequate response to DMARDs. The committee concluded that the identified limitations in the NMAs are common to previous technology appraisals for psoriatic arthritis and cannot be resolved.

3.9 The ERG identified a discrepancy in how the company describes the modelling of HAQ-DI over time in its evidence submission, and the way in which it has been implemented in the company's economic model. The company stated that, in line with the model on which all recent technology appraisals for psoriatic arthritis have been based (see section 3.5), people whose disease responds to a first- or second-line biological DMARD have a HAQ-DI score that is constant until the treatment is stopped. At that point it increases (deterioration in function) instantly to their baseline score. HAQ-DI then increases in line with the natural history (the rate of increase for people who did not respond to treatment). But this does not align with how it has been implemented in the company's model. In the company's model, when a responder to a biological DMARD stops treatment, their HAQ-DI score increases instantly to a value between their baseline value and the HAQ-DI score for non-responders to a biological DMARD. Those who are classed as non-responders to a biological DMARD have a HAQ-DI score that has already been increasing in line with natural history since the start of the model. The HAQ-DI score then converges with the rate of increase for non-responders, rather than progressing in parallel to that rate. In part, this is because HAQ-DI increases slowly over time for everybody, whether or not they have psoriatic arthritis. But, it is also possible that a person whose psoriatic arthritis is responding to active treatment may still develop some joint damage. Or, that existing joint damage is getting worse because of use of that joint over time. In addition, HAQ-DI might also be expected to increase over time because of the presence of comorbidities. However, the clinical expert further explained that when active treatment is stopped, it would be expected that the HAQ-DI gradient should be better than natural history, rather than worse. The committee agreed that the model should rebound to baseline, as previous models in this area do. Following the first appraisal committee meeting, the company explained that they considered the implementation to be appropriate and consistent with Markov models used in previous technology appraisals for psoriatic arthritis. NICE sought informal advice from an assessment group who developed the original model and who were the ERG for recent technology appraisals for psoriatic arthritis (The Centre for Reviews and Dissemination [CRD] and Centre for Health Economic [CHE] Technology Assessment Group [TAG] at the University of York). They and the ERG for this appraisal agreed that the company's approach was inconsistent with the most recent accepted approach used in the NICE technology appraisal guidance on guselkumab for treating active psoriatic arthritis after inadequate response to DMARDs (from now on referred to as TA711). NICE provided the company with a technical solution to resolve this issue in a similar way to the approach used in TA711 provided by CRD and CHE TAG at the University of York. The company implemented this solution and provided a revised model. The ERG for this appraisal confirmed that the company's revised model using the technical solution reflected the approach used in TA711. The committee noted that the company's revised model results had a negligible impact on the cost-effectiveness results. The committee concluded that HAQ-DI progression in the model is implemented in a similar way to recent appraisals for psoriatic arthritis and was acceptable for decision making.

3.10 At submission, the company did not present a scenario in which the effect of HAQ-DI increases for people whose disease responds to a biological DMARD while having treatment. The ERG had suggested to the company that results from such a scenario would have been informative. The company responded that this situation was not clinically plausible. Clinical expert opinion to the company suggested that it was implausible because people experiencing disease progression at the natural history rate would quickly be swapped to an alternative treatment because of a lack of response. Clinical expert opinion provided at the technical engagement stage confirmed that increasing HAQ-DI in a person responding to treatment would be an unusual scenario, and usually related to a parallel comorbidity rather than to psoriatic arthritis disease progression. Following the first appraisal committee meeting, the company provided this scenario analysis. The ERG was satisfied that it did not meaningfully affect the cost-effectiveness results. The committee concluded that given the lack of effect on the results, this did not need be considered further.

3.11 The company did not present evidence of clinical effectiveness by presence of concomitant psoriasis. The committee recalled that in previous psoriatic arthritis appraisals results were presented by psoriasis subgroup. But the company and ERG did present the cost-effectiveness results by presence of concomitant psoriasis (no psoriasis, mild to moderate, moderate to severe) by using a combination of body surface area and psoriasis area and severity index (PASI) scores. The ERG considered that this approach was appropriate. The committee concluded that results based on psoriasis severity were relevant for consideration.

3.12 Because upadacitinib and the comparators have commercial arrangements, the exact ICERs are confidential and cannot be reported here. The company provided cost-effectiveness results for each subpopulation and by psoriasis severity:

3.13 The committee was aware that clinicians and people with psoriatic arthritis would welcome an additional treatment option. It was aware there were already several treatment options available and recommended by NICE for each of subpopulations (see section 3.2). The committee therefore agreed that it was important to consider fully incremental analyses to assess value for money, as outlined in NICE's guide to the methods of technology appraisal 2013. The committee noted that for the people who have not had a biological DMARD, the fully incremental analysis showed that upadacitinib would not be the most cost-effective option when taking into account what it considered an acceptable ICER range representing value for money for the NHS. Pairwise analysis showed that upadacitinib might be cost effective against some but not all options currently available in the NHS. The committee considered that the pairwise analysis results had limited relevance in this appraisal because all the comparators were relevant and therefore the fully incremental analysis was the most appropriate use. The committee acknowledged that previous technology appraisals for psoriatic arthritis may have considered the differences in benefits and costs to be small. However, this is no longer the case because of the availability of biosimilars, and changes to confidential arrangements. The committee concluded that upadacitinib was not a cost-effective use of NHS resources for people who have not had a biological DMARD and so was not recommended in this subpopulation.

3.14 The cost-effectiveness results for people who have had a biological DMARD or when TNFi treatments are contraindicated were within what NICE considers a cost-effective use of NHS resources. Therefore, it concluded that it could recommend upadacitinib, alone or with methotrexate, for treating active psoriatic arthritis in adults whose disease has not responded well enough to DMARDs or who cannot tolerate them, only if:

3.15 The committee considered that the recommendation to stop treatment based on an inadequate PsARC response (in NICE's technology appraisal guidance on etanercept, infliximab and adalimumab for the treatment of psoriatic arthritis) was also appropriate for upadacitinib. It noted that some people may have physical, sensory or learning disabilities or communication difficulties that could affect their responses to the PsARC, and concluded that this should be taken into account when using the PsARC. The committee was also aware that the PASI might underestimate disease severity in people with darker skin. The committee concluded that, when using the PASI, healthcare professionals should take into account skin colour and how this could affect the PASI score and make the clinical adjustments they consider appropriate.