Adalimumab and dexamethasone for treating non-infectious uveitis

Clinical evidence

4.4 The committee was aware that the comparators in the appraisal scope included corticosteroid injections and implants, systemic immunosuppressive therapies, tumour necrosis factor-alpha inhibitors (such as infliximab), intravitreal methotrexate and best supportive care. The scope also stated that the interventions should be compared with each other. However, direct clinical evidence was only available for the interventions compared with either placebo or a sham procedure. The evidence for adalimumab came from the VISUAL I and VISUAL II trials and the evidence for dexamethasone intravitreal implant came from the HURON trial. The VISUAL trials compared adalimumab plus background therapy (that is, initial steroids tapered to zero with or without 1 immunosuppressant) with placebo plus background therapy. The HURON trial compared dexamethasone plus background therapy with a sham procedure plus background therapy. The committee noted that there was no clinical evidence which directly compared adalimumab with dexamethasone and the assessment group did not do an indirect comparison using HURON and the VISUAL trials. The assessment group advised that an indirect comparison was not appropriate because patient characteristics in VISUAL I, VISUAL II and HURON differed and there was a lack of common comparators and outcomes. Furthermore, adalimumab and dexamethasone could be used at different points in the treatment pathway. The committee agreed that there was a lack of relevant evidence on therapy for non-infectious uveitis, with varied and often limited current treatments available. However, the available clinical evidence was adequate for decision-making.

Patients included in the trials

4.5 Although the marketing authorisations for adalimumab and dexamethasone did not distinguish between active and inactive disease, the committee understood that VISUAL I and HURON included patients with active disease but VISUAL II included patients with inactive disease. The committee heard from the clinical experts that the distinction between active and inactive disease was clinically relevant because they were different populations, which would have different treatments. It heard that because maintenance treatment with immunosuppressants and corticosteroids may control inactive disease, the next line of treatment, such as adalimumab or dexamethasone intravitreal implant may not be needed. The committee was aware of an ongoing trial, ASTUTE, that examines the effectiveness of adalimumab compared with placebo as an add-on to standard care. It understood there were no data from this study at the time of this appraisal. The committee concluded that it would take these different populations into account when making its final recommendations.

4.6 The committee noted that the inclusion criteria for the VISUAL trials did not specify patients with macular oedema. It heard from clinical experts that people with macular oedema have a high risk of blindness. Because patients with macular oedema were not specifically included in the VISUAL trials, the clinical effectiveness of adalimumab may be underestimated in this group. It also heard from clinical experts that people with bilateral disease or systemic disease are likely to have a higher risk of blindness compared with people with unilateral or localised disease. The committee noted that most patients (over 90%) in the VISUAL trials had bilateral or systemic non-infectious uveitis. It also noted that the proportion of people with bilateral uveitis in HURON was unclear but patients had only 1 dexamethasone implant, and current clinical practice preferred dexamethasone for unilateral disease. The committee concluded that it would be useful to distinguish unilateral from systemic and bilateral disease and that people with a higher risk of blindness formed a clinically important subgroup.

Clinical-effectiveness results

4.7 The committee noted that the primary outcome in the VISUAL trials was a composite measure of time to treatment failure. The committee understood that the VISUAL trials showed that adalimumab had improved outcomes, such as time to treatment failure and visual acuity, compared with placebo. The committee noted that the primary outcome in the HURON trial was the proportion of people with a vitreous haze score of 0. It understood that HURON showed that dexamethasone had improved outcomes, such as vitreous haze score and visual acuity (in the affected eye), compared with the sham procedure. The committee concluded that there is evidence to show that both adalimumab and dexamethasone are clinically effective treatments for improving visual acuity, anterior chamber cell grade and vitreous haze.

Table 1 Summary of clinical-effectiveness results

Model structure

4.8 The committee noted the assessment group had developed a Markov model with 4 health states in the base case (on treatment, treatment failure, permanent blindness and death). The assessment group gave 3 separate base cases, based on the underpinning trial evidence: adalimumab for active disease, adalimumab for inactive disease, and dexamethasone intravitreal implant for active disease. The assessment group noted it had not been possible to distinguish between unilateral and bilateral disease in the model; however, over 90% of patients in the VISUAL trials on adalimumab had bilateral disease. The committee concluded that the assessment group's decision to separate its analyses into 3 separate base cases was appropriate and supported by clinical evidence and that it would need to consider all 3 base-case analyses when making its recommendations.

4.9 The committee was aware that the assessment group did an exploratory analysis with a 'remission' health state for the adalimumab model only. This was based on clinical advice that some people who have treatment with adalimumab will have disease which is in remission (the assessment group excluded this health state from its base case because of a lack of evidence). The committee understood that the exploratory analysis assumed that after around 2 years of stable disease, treatment is no longer needed. This is because the disease is in remission but patients will have the same health-related quality of life as when they were having treatment. The committee heard from the clinical experts that the disease could be expected to go into remission in at least some of the people who have adalimumab. The committee concluded that although there is no evidence for remission, it was reasonable to assume that at least some people's disease could be in remission after treatment with adalimumab.

Modelling the rate of blindness

4.10 The committee noted that the follow-up time in the HURON trial was 26 weeks, but a maximum follow-up of 80 weeks was included in the VISUAL trials. None of the trials reported patients with permanent legal blindness. The assessment group advised that the rate of blindness and the relative risk of blindness associated with adalimumab and dexamethasone had a large effect on the incremental cost-effectiveness ratio (ICER) and did scenario analyses to model this potential effect. The committee acknowledged there was a lack of robust, long-term studies for the rate of blindness but concluded that the scenario analyses including blindness were appropriate for decision-making because it is likely that both adalimumab and dexamethasone had an effect on the rate of blindness, although the extent of this effect was uncertain.

4.11 The committee understood that in its base case, the assessment group preferred to use a constant annual rate of blindness of 0.0066 from Dick et al. (2016), a retrospective analysis of 1,769 insurance claims of adults with non-infectious intermediate uveitis, posterior uveitis or panuveitis in the US. In contrast, the company for adalimumab preferred the higher rate of blindness (0.0374) that was used in a scenario analysis by the assessment group. This rate was reported in a retrospective review of 315 medical records in the UK (Durrani et al. 2004). However, the assessment group advised that this study included a wider population compared with the scope of this appraisal (only 61% of patients had posterior, intermediate or panuveitis and age ranged from 7 to 86 years) and was done in a tertiary centre in which patients are more likely to have severe, and often bilateral uveitis. The committee was aware that a higher rate of blindness would lead to more favourable cost-effectiveness results for the interventions. The committee agreed that for people at higher risk of blindness (for example, people with macular oedema and bilateral disease) the background rate of blindness is likely to be higher than in the base case. However, it noted that macular oedema was not the only precursor to blindness. The committee also agreed that the risk of permanent legal blindness in people having dexamethasone was probably lower compared with the risk of blindness in people having adalimumab. This is because dexamethasone is often used to treat unilateral disease, whereas adalimumab is more often used to treat bilateral disease later in the treatment pathway. The committee concluded that the rate of blindness in people at high risk was uncertain but likely to be higher than in the base case (0.0066). The committee also concluded the base-case rate of 0.0066 was acceptable for unilateral disease, because although this group might include people with macular oedema and at a higher risk of unilateral blindness, this was not the same as permanent legal blindness.

4.12 The committee noted that the assessment group's base case used a relative risk of blindness of 0.5 for dexamethasone (that is, a 50% lower rate of blindness in the dexamethasone group compared with the comparator group). For adalimumab, the base case did not allow blindness in either the adalimumab or comparator arms while on treatment, and used the same overall rate of blindness after treatment failure, which was strictly defined in the VISUAL trials. However, there was a lack of evidence to support a relative risk of 0.5 for dexamethasone. The clinical experts agreed that there was a lack of evidence to support this assumption, but considered a value of 0.5 to be clinically plausible for the affected eye. For adalimumab, the committee accepted the base-case assumptions. The committee concluded that the assessment group's approach to modelling was appropriate for decision-making provided that the dexamethasone rate is only applied to the affected eye, unless both eyes were affected and treated.

4.13 The committee noted that the assessment group did additional scenario analyses for adalimumab for active non-infectious uveitis using the committee's preferred utility for blindness of 0.57. The scenarios combined varying the relative risk of blindness with treatment and the rate at which treatment is stopped because of remission. The committee recalled its earlier conclusion that at least some people having adalimumab are likely to go into remission. The committee concluded that these scenarios were appropriate for decision-making because they accounted for both the possible effect on blindness and the additional benefit of remission.

Health-related quality of life

4.14 The committee was aware that the quality-of-life data from the clinical trials were assumed to include the effects of adverse events during the treatment period. Utility values for blindness were also taken from the literature. The committee heard from the patient and clinical experts that they were unsure whether this approach to modelling utility included the effect of uveitis on the whole person. This was because uveitis substantially affected quality of life, with visual disability having significant consequences for people (including depression and stress, for example, because of a loss of ability to support self and family), and their families and carers. However, the assessment group stated that the trial included holistic treatment benefit, as well as the main costs of adverse events and blindness. The committee agreed that uveitis had a significant effect on quality of life, and that there were limited data to inform the utility assumptions. However, it was aware that the utility values are designed to represent whole person health. The committee concluded that the utility values used were appropriate for decision-making.

4.15 The committee considered the approach to modelling the utility of the blindness health state. The committee noted that the quality-of-life value used in the base case (0.38 from Czoski-Murray et al. 2009) was low, and agreed that scenario analyses using the higher utility of 0.57 (from Brown et al. 1999) were more plausible. It was aware that in the model people were either permanently legally blind or not blind. The committee was aware this omitted the effect of worsening visual acuity and that level of vision was likely to be a continuous variable. The committee further discussed the effect of blindness depending on whether disease was unilateral or bilateral, which was not captured in the model. In its experience of previous appraisals for eye diseases, the utility loss of blindness in both eyes was likely to be much higher than in unilateral blindness.

4.16 The committee noted that EQ‑5D data were reported at baseline and follow-up in the VISUAL trials, but only at baseline in HURON. It was aware that all 3 trials also assessed health-related quality of life using the Visual-Functioning Questionnaire (VFQ‑25) and that this measure is more specific to visual function. The committee understood that to model utilities for adalimumab over time, the assessment group used EQ‑5D data directly from the VISUAL trials. To estimate utility over time for dexamethasone, the assessment group used individual patient-level VFQ‑25 data and mapped these to EQ‑5D using a regression analysis. Individual patient-level data for adalimumab were not made available in time for the assessment group to use them. The committee heard from the assessment group that using VFQ‑25 data instead of EQ‑5D data had only a small effect on the ICERs. The committee concluded that the methodology used to derive this utility was acceptable for decision-making.

Resource use

4.17 The committee noted that to calculate the cost of blindness, the assessment group assumed that 30% of patients would have residential care. It understood that this cost was based on a health technology assessment (Colquitt et al. 2008) on treating age-related macular degeneration, which is likely to affect people who are older than those with uveitis. It recalled that people with non-infectious uveitis are between 20 to 50 years and only a small proportion would need residential care. The committee concluded that the proportion of people needing residential care is likely to be overestimated in the assessment group's model and, if this proportion were smaller, the ICERs would increase (that is, the treatments would become less cost effective) for the scenario analyses involving blindness.

4.18 The committee heard from the clinical experts that using multiple dexamethasone implants consecutively was associated with adverse events, including increased intraocular pressure and cataracts. It heard that for this reason, the clinical experts would use at most 3 implants consecutively although this may vary in clinical practice. It also heard from the assessment group that using multiple implants consecutively was likely to produce similar cost-effectiveness results because the model assumed that dexamethasone would only provide a treatment benefit for around 6 months. The committee concluded that consecutive use of dexamethasone was unlikely to have a large effect on the cost-effectiveness analyses.

Most plausible incremental cost-effectiveness ratios

4.19 The committee recalled its earlier concern that the cost of blindness had been overestimated in the model. This meant that the base-case and scenario analyses for both interventions favoured the interventions that were more effective in reducing blindness. However, it also noted that there was uncertainty around the background rate of blindness, which it considered had been underestimated for the high-risk groups for whom treatment with adalimumab or dexamethasone is possible. This was likely to make the cost-effectiveness results more conservative. The committee concluded that, although there was uncertainty, the reduction (that is, improvement) in ICERs resulting from overestimated costs of blindness were likely to be offset by the low rate of background blindness in this high-risk group.

4.20 The committee noted that the assessment group's base-case ICER for adalimumab in patients with inactive non-infectious uveitis was £321,405 per quality-adjusted life year (QALY) gained. It noted that all the ICERs in all the scenario analyses were above £80,000 per QALY gained. The committee agreed that these ICERs were substantially above the range normally considered a cost-effective use of NHS resources. The committee also noted that people with inactive disease would be unlikely to have treatment with adalimumab in clinical practice (see section 4.5). Therefore the committee did not recommend adalimumab for treating inactive non-infectious uveitis.

4.21 The committee noted that the assessment group's base-case ICER for adalimumab in patients with active non-infectious uveitis was £95,506 per QALY gained. However, this base case did not take account of its reasoning that there would be a relatively severely affected subgroup of patients (see section 4.6). The committee noted that disease was likely to be more severe in people with bilateral disease later in the treatment pathway. It agreed that the treatment would be more cost effective in those at higher risk of permanent legal blindness, and bilateral disease with macular oedema was a useful proxy for this. However, the committee also recognised that people with unilateral disease in the better seeing eye were also at a high risk of permanent blindness if they have poor visual acuity in the other eye. Using its preferred assumptions for severe disease (see section 4.6), high rates of blindness (see sections 4.11 and 4.12), utility values (see section 4.15) and occasional remission (see section 4.9), adalimumab resulted in ICERs that ranged from £23,688 to £37,279 per QALY gained, and these were probably lower because the rate of blindness was likely to have been underestimated for patients with progressive loss of visual acuity (see section 4.11). The committee noted that the VISUAL trials included some patients taking only 1 immunosuppressant but understood that in clinical practice, 2 immunosuppressants may be taken as second-line treatment (see section 4.2). Therefore it agreed that either 1 or 2 immunosuppressants may be used as part of second-line treatment before adalimumab is started. The committee also took into account the lack of available treatment options for this subgroup, the evidence from the patient and clinical experts about the adverse effects associated with current treatment options and comments from consultation. Taking all of this into account, it recommended adalimumab as a cost-effective use of NHS resources for treating non-infectious uveitis in the posterior segment of the eye in adults, if there is:

4.22 The committee noted that the assessment group's base-case ICER for dexamethasone in patients with active uveitis was £20,058 per QALY gained. The committee recalled that the clinical experts stated that people who have dexamethasone in current clinical practice are likely to have disease affecting only 1 eye (see section 4.2), but the proportion of unilateral uveitis in the HURON trial was unclear (see section 4.6). Using the committee's preferred assumptions for unilateral disease (see section 4.6), low rates of blindness (see section 4.11 and section 4.12) and utility values (see section 4.15) resulted in ICERs that ranged between £25,000 and £49,000 per QALY gained. The committee considered that the lower ICER would apply to patients whose better seeing eye needed treatment, because this reflected a risk of bilateral blindness; the higher ICER applied to patients at no risk of bilateral blindness, and was likely to be a significant overestimate because the disutility of monocular blindness was not modelled. The committee decided that dexamethasone for monocular disease with worsening vision and a risk of blindness was an acceptable use of NHS resources and that patient need is high. The committee concluded that the ICER for unilateral disease with a risk of blindness was likely to be in the range normally considered cost effective, and recommended dexamethasone for treating active non-infectious uveitis with worsening vision and a risk of blindness.