Ruxolitinib cream for treating non-segmental vitiligo in people 12 years and over

The key clinical evidence came from TRuE‑V1 (n=330) and TRuE‑V2 (n=344), which were phase 3, double-blind, randomised controlled trials. Both trials were multinational with no UK sites. They included a double-blind phase (24 weeks) in which people were randomised to either ruxolitinib or vehicle cream (no active treatment) twice a day. This was followed by an open-label extension (28 weeks) in which everyone had ruxolitinib cream. The population was people 12 years and over with non-segmental vitiligo affecting at least 0.5% of body surface area on the face, and at least 3% of body surface area on non-facial areas. The total body vitiligo area (facial and non-facial) could not exceed 10% of body surface area. Assessment of the extent of the condition in the trials was measured using the Vitiligo Area Scoring Index (VASI). People in the trials had a facial VASI (F‑VASI) score of at least 0.5 and a total body VASI (T‑VASI) score of at least 3. Most people (74%) had 'stable vitiligo' at baseline and generally represented people that had vitiligo for a long time (mean 14.8 years since diagnosis). The primary outcome from the TRuE‑V trials was repigmentation, defined as the proportion of people with an improvement of at least 75% from baseline in the F-VASI score (F‑VASI 75) at week 24. People in the TRuE-V open-label extension were assigned to one of 2 cohorts (A or B) based on their F‑VASI responses at the time of enrolment. Cohort A had a complete or almost complete facial repigmentation by year 1 of TRuE‑V (F‑VASI 90 or more), but cohort B did not have F‑VASI 90 by year 1. The company presented pooled results from TRuE‑V1 and TRuE‑V2 because the trial designs were identical. In the intention-to-treat population, the proportion of people with F‑VASI 75 at week 24 was statistically significantly higher in the ruxolitinib group compared with the vehicle-cream group (odds ratio 4.17, 95% confidence interval 2.43 to 7.14, p<0.0001). The committee noted that the company had positioned ruxolitinib cream for use for non-segmental vitiligo with facial involvement, but the primary repigmentation outcome focused on improvements in vitiligo on the face only. It considered that improvements in F‑VASI did not necessarily correspond directly to the quality of life of people with vitiligo, because changes in F‑VASI did not always correlate with changes to T‑VASI. The clinical experts considered that the Vitiligo Noticeability Scale score is clinically relevant and may be a more accurate measure of the efficacy of treatment because it is a patient-reported outcome. In the intention-to-treat population, the proportion of people with a Vitiligo Noticeability Scale score of 4 or 5 (which indicates that a person's vitiligo is a lot less noticeable or no longer noticeable) was significantly higher in the ruxolitinib group than the vehicle-cream group (odds ratio 6.52, 95% confidence interval 3.11 to 13.67, p<0.0001) at week 24. The committee concluded that ruxolitinib cream increases repigmentation and reduces the noticeability of vitiligo patches compared with vehicle cream. It considered phototherapy (with or without topical treatments) to be a relevant comparator (see section 3.6) and noted it had not been presented with any clinical evidence for this comparison. It understood the company had explored the feasibility of an indirect treatment comparison but considered that there was insufficient evidence to robustly compare the efficacy of ruxolitinib cream with phototherapy. The committee acknowledged there may be limitations in doing this comparison. But it concluded that the company should provide comparative evidence for ruxolitinib cream with all relevant comparators, including phototherapy. This was provided at consultation by the company (see section 3.8).

After consultation, the company provided an indirect treatment comparison with phototherapy, with or without topical corticosteroids. Both a naive and matching-adjusted indirect comparison were provided. Data from HI‑Light trial (a randomised, pragmatic, 3‑arm placebo-controlled trial) informed the clinical effectiveness of phototherapy. The population was people aged 5 years and over, with non-segmental vitiligo affecting less than 10% of body surface area. Assessment of the extent of the condition was measured using repigmentation scores according to the Vitiligo Noticeability Scale. Data from the pooled TRuE‑V trials informed the clinical effectiveness of ruxolitinib cream. Results of the indirect treatment comparison showed that people who had ruxolitinib cream were statistically significantly more likely to experience an overall response (25% to 100% repigmentation) than people having phototherapy, after 9 months of treatment. The EAG concluded that neither approach taken by the company would provide reliable effect estimates of the comparison with ruxolitinib cream and phototherapy, and it had no confidence in the results of the comparison. It had concerns about variation in baseline characteristics between HI‑Light and the TRuE‑V trials, and between the arms of the HI‑Light trial. It also noted discrepancies between baseline characteristics reported for each trial that could be used in matching and meaningful differences in the outcomes measured in each trial. The EAG noted that given the trial design and the evidence presented, the company had made significant effort to produce the analyses, and that the limitations in the comparability of evidence for ruxolitinib cream and phototherapy was beyond the control of the company at this stage of the appraisal. The committee considered that the analysis justified the clinical opinion that ruxolitinib cream would be used before phototherapy because of the increased clinical efficacy, but did not provide a robust enough comparison to inform cost–utility analyses. This confirmed that the comparison of no active treatment followed by phototherapy is most representative of the positioning of ruxolitinib cream.

The clinical-effectiveness evidence in the company's submission was based on the pooled full-trial populations from TRuE‑V1 and TRuE‑V2. The EAG advised that the clinical evidence was not consistent with the target population (people who have had topical first-line treatments or when these treatments are unsuitable) or the prior-therapy subgroup used in the model (people who have had any previous treatment). The committee noted that the company submitted evidence for the prior-therapy subgroup in response to clarification. But the EAG advised that this was not submitted in a format that could be fully appraised. The committee understood there was a slightly higher response rate to ruxolitinib cream for people who had previous treatment than for the full trial population. It noted the EAG's critique that, without complete data for the prior-therapy subgroup, it was not possible to determine whether this was evidence of a true difference in treatment effect between treatment lines. The committee decided it was unclear how generalisable the full-trial populations from the TRuE‑V trials were to the target population who would be eligible for ruxolitinib cream. It concluded that the company should provide a full submission of evidence for the prior-therapy and target-population subgroups that can be appraised by the EAG. At consultation, the company provided all available evidence from the pooled TRuE‑V trials relating to the efficacy of ruxolitinib cream in the overall population and prior-therapy populations. The prior-therapy population from the pooled TRuE‑V trials had a slightly higher response rate with ruxolitinib cream than the overall trial population (odds ratio 4.6 [p<0.0001] compared with 4.17 [p<0.0001]). The EAG explained that although the results indicated ruxolitinib cream is highly effective in reducing vitiligo, it was unable to determine if the effects were comparable between the overall population and subgroup populations because no statistical comparison was provided by the company. The committee concluded it was unable to consider the prior-therapy subgroup separately, so did not review clinical and cost-effectiveness evidence for this subgroup.

In its original submission, the company presented a subgroup analysis on the outcome of repigmentation (assessed by the proportion of patients achieving F-VASI 75 at week 24) by Fitzpatrick score, based on pooled data from TRuE-V1 and TRuE-V2 trials. At the first committee meeting, the company explained that there was no significant difference in repigmentation between people with brown and black skin tones (defined then as having a Fitzpatrick score of 3 to 6) and those with white skin tones (defined as having a Fitzpatrick score of 1 or 2). The clinical and patient experts explained that the impact of vitiligo patches varies individually and does not necessarily depend on a person's skin colour or Fitzpatrick score. They also described how a vitiligo patch on the face could be as distressing for a person with a Fitzpatrick score of 1 as for someone with a Fitzpatrick score of 6.

After the appeal, the company submitted a revised subgroup analysis by Fitzpatrick score on several quality-of-life outcomes. In these revised analyses, the company changed the definition of the subgroups defining black or brown skin tones as Fitzpatrick score 4 to 6 and white skin tones as 1 to 3. The company did not perform tests of interaction. Results suggested that ruxolitinib appeared to be associated with an increased treatment effect in people with black and brown skin tones on several quality-of-life outcomes, including the Dermatology Life Quality Index (DLQI). But, as stated by both the company and the EAG, confidence intervals were wide and overlapping for all quality-of-life outcomes, indicating a high level of uncertainty or lack of statistical significance, particularly for the vitiligo-specific quality-of-life (VitiQoL) score and the Vitiligo Noticeability Score. The EAG noted that although there was not a minimal clinically important difference in these quality-of-life outcomes defined or identified, the magnitude of treatment effect on DLQI suggested that people with black and brown skin tones may experience greater quality-of-life benefit with ruxolitinib than those with white skin tones. The EAG also noted that the company did not present subgroup analysis by Fitzpatrick score for F-VASI outcomes on which the company's model was based. But, the committee was aware that a recent publication (Seneschal et al. 2025), which used the pooled data of TRuE-V1 and TRuE-V2, reported that improved repigmentation (assessed using F-VASI 75) was seen in both people with black and brown skin tones (defined as Fitzpatrick score 4 to 6) and people with white skin tones (Fitzpatrick score 1 to 3).

At the third committee meeting, the company explained that it submitted the revised subgroup analyses on quality-of-life outcomes because they were related to the upheld appeal points about people with black and brown skin tones. The company highlighted that it does not see ruxolitinib being recommended for specific subgroups defined by skin colour. Both the clinical and patient experts echoed this. The committee noted that the company's revised subgroup analyses were post hoc and exploratory, so it questioned the company's rationale for redefining the subgroups. The committee expressed its concern about poor statistical practice and queried whether the company had done a test for interaction to determine if there was a difference in the treatment effect between skin types. The company stated there is no clear cut-off when using the Fitzpatrick score to define skin types. It defined black and brown skin tone by Fitzpatrick score 4 to 6 because this is widely accepted in the literature, and the recategorised subgroups suggested a better treatment effect on quality-of-life outcomes that favour ruxolitinib than in the previous categorisation. A patient expert confirmed that defining black and brown skin tones as a Fitzpatrick score of 4 to 6 was acceptable.

A clinical expert noted that DLQI mainly measures physical symptoms. This may be appropriate for other skin conditions, but not for vitiligo because of the related social and cultural stigma and increased stress. The committee considered the company's revised post-hoc subgroup analyses to be poor statistical practice. Yet, it noted that it was plausible that a greater quality-of-life benefit may be seen in people with black and brown skin tones, and the company's trials were not designed to capture this. The committee concluded that ruxolitinib might be associated with greater improvement in quality of life in people with black and brown skin tones (as defined by a Fitzpatrick score of 4 to 6) than in people with white skin tones (Fitzpatrick score of 1 to 3), but this was highly uncertain. The committee also noted that its remit for this third meeting, after the appeal, was not to conduct a new appraisal of the technology, but to consider the upheld appeal points and changes to the final draft guidance where needed. The committee took this into account in its decision making.

The company initially presented a Markov state-transition model comparing ruxolitinib cream with vehicle cream. This used 7 mutually exclusive health states based on response status and including the opportunity for retreatment. At the first committee meeting the committee concluded that the company's model did not reflect clinical practice, significantly biased the cost-effectiveness results in favour of ruxolitinib cream and was not suitable for decision making. The committee decided the company should provide a revised model to correct the structural flaws, including:

• The definition of who would continue treatment with ruxolitinib cream did not reflect expected clinical practice. The company's model assumed that people who reach F‑VASI 50 to 75 (a 50% to 75% improvement in F‑VASI score from baseline) at week 24 have not had a response. The committee considered that the company's continuation rule underestimated the proportion of people who would continue ruxolitinib cream after 24 weeks. The model should reflect anticipated continuation of ruxolitinib cream in clinical practice.

• People in the non-response health state could not have any improvement in their vitiligo. The committee considered that this structural assumption did not reflect clinical practice, in which another treatment option would usually be offered.

• The maintenance period health state in the model included people who had an F‑VASI 75 at week 24. These people continued using ruxolitinib or vehicle cream. The EAG stated that it was structurally impossible for people reaching F‑VASI 75 to 89 in the maintenance period health state to transition to the stable health state, in which they stopped treatment.

• People who had an F‑VASI 90 response and stopped treatment had the same topical treatment used previously (either ruxolitinib or vehicle cream) if their vitiligo subsequently relapsed (defined as response dropping below F‑VASI 75). The committee agreed with the EAG that retreatment with vehicle cream did not reflect NHS clinical practice.
  
  After consultation the company made substantial changes to the structure, input parameters and assumptions underlying its original model as follows:

• People with an initial response of F‑VASI 90 by year 1 transition straight into the stable health state.

• People with an initial response of less than F‑VASI 25 by year 1 transition directly to the non-response health state.

• People with an initial response between F‑VASI 25 and F‑VASI 89 transition to the maintenance/retreatment health state for an upper limit of 1 year. Response is reassessed at year 2 and is linked to the response achieved at year 1. If F‑VASI 90 is reached by year 2, people transition to the stable health state. If not, they transition to the non-response health state.

• Everyone is eligible to transition to the stable health state, correcting the structural error relating to people reaching F‑VASI 75 to 89 in the maintenance period health state.
  
  The company applied the updated model structure to 4 comparisons, with no active treatment (either followed by phototherapy or not) and phototherapy (either as monotherapy or in combination with topical corticosteroid). The committee decided the most appropriate comparison was no active treatment followed by phototherapy, but noted that the efficacy of phototherapy was not included in this analysis. The EAG considered the company's revised model to be an improvement on its previous model and more closely matched how it would expect ruxolitinib cream to be used in clinical practice. The model also had more realistic expectations of assessing and monitoring response from the available clinical data. But the EAG still had concerns about the revised model, including:

• uncertainty and reliability of assumptions about retreatment after loss of response (see section 3.13)

• validity relating to the proportion of people reaching F‑VASI 90 (see section 3.14)

• discrepancy between baseline and 'no response' health state utility values and other concerns with these values (see section 3.17).
  
  The committee concluded that the updated model more closely matched the expected use of ruxolitinib cream in clinical practice. But some issues had not been resolved, particularly how the model is based around a response from a baseline for facial vitiligo only. This may not reflect distinct health states representing the course of the disease and how it affects people's quality of life and resource uses. The committee concluded that there remained some structural uncertainty, but it did not think it would be resolved with any further changes to the model. So, it agreed the model structure was adequate for decision making. But it paid close attention to the structural limitations of the model and any potential biases these created.

The SmPC for ruxolitinib cream recommends applying a thin layer of cream twice daily to the depigmented skin areas up to a maximum of 10% of body surface area. No more than 2 individual tubes (100 g each) of ruxolitinib cream should be used each month. The committee understood that the dose of ruxolitinib cream is likely to vary for each person depending on the size of the area of vitiligo and will depend on a person's adherence to the SmPC. The patient expert explained that healthcare professionals would need to provide detailed information to support people in managing how much cream they apply to their vitiligo patches. The company stated that the patient information leaflet would provide information on how much people should apply. The company's model assumed that the pooled median daily dose of treatment in the TRuEV‑ trials (across the ruxolitinib and vehicle cream arms, week 1 to week 24) reflected the expected daily dose of ruxolitinib cream in NHS clinical practice. This was less than the 200 g per month limit in the SmPC. The EAG advised it is more appropriate to use the mean dose of topical ruxolitinib alone, rather than the median dose across trial arms. It noted that the mean dose of ruxolitinib cream in the pooled TRuEV‑ trials was larger than both the median and the dose limit of ruxolitinib cream specified in the SmPC. The committee noted this implied that some people in the TRuEV‑ trials used significantly more ruxolitinib cream than recommended. The company outlined how it had assessed individual patient-level body surface area and dosing data from the TRuEV‑ trials stratified by trial and treatment arm. It explained that the treatment duration for a small number of people in the trials had been miscalculated as 1 day, because the treatment duration for these people had not been recorded in the trials. The company explained that excluding the results for these outliers reduced the mean dose of ruxolitinib cream to a value similar to the median. The committee noted that the EAG had presented 2 alternative base cases using either the mean dose of ruxolitinib cream from the TRuE‑V trials (week 1 to week 52) or the maximum recommended dose in the SmPC. It understood that changing the ruxolitinib cream dosing assumptions had a large impact on the incremental cost-effectiveness ratio (ICER). The committee concluded that the mean dose of topical ruxolitinib alone from the pooled TRuEV trials should be used in the model, using appropriate methods‑ to account for any missing data.

At consultation, the company updated its base-case dosing assumption with an estimated mean daily dose of ruxolitinib cream. This was calculated by applying a lognormal distribution to the entire TRuE‑V trial dosing data. It also provided an alternative scenario in which the mean daily dose was estimated by excluding the outliers that had missing treatment-duration data. The EAG noted that it had not received any numerical or graphical validation for the lognormal distribution, or any assessment of statistical goodness-of-fit. So, it could not validate the appropriateness of the company's analysis. It advised that the only evidence-based approach was to use the revised estimate that excluded the outliers. The company noted that real-world evidence from Europe and the US suggested that the mean use of ruxolitinib cream in clinical practice is expected to range between values lower than those seen in TRuE‑V. It suggested that the difference seen between TRuE‑V and real-world evidence could be related to the body surface area restriction within TRuE‑V's exclusion criteria. So, the trial population had a higher mean body surface area (7.4%) than those in real-world studies (1.4% to 3.8%). The EAG explained that determining the true cost of ruxolitinib cream to the NHS would be difficult because several factors influence this, including dispensing practices, overall intended use and retreatment. The committee noted the uncertainty inherent in the trial design because the primary outcome of facial VASI score only may not be explicitly linked to dose used, which would be linked to the entire body surface area affected. The patient expert noted that dose would likely be significantly reduced with higher responses to maintain response on smaller patches. The committee concluded it was most appropriate to use the dosing estimate from TRuE‑V that excluded the outliers rather than the estimate calculated using the lognormal distribution.

The economic model structure contains an optional retreatment component for people who have had a stable (F‑VASI above 90 from baseline) response to treatment, containing the retreated and stable retreated health states. People can enter these states after relapse from the stable state, defined as a loss of response equivalent to less than F‑VASI 75 from trial baseline. The EAG noted that the response required to enter the retreated state differs markedly from the initial treatment period, where people with a response between F‑VASI 25 and F‑VASI 89 transition to the maintenance/retreatment state. It explained that people who enter the retreated state would instantly leave according to their F‑VASI response; people with a response less than F‑VASI 90 would transition to the non-response state, and those with a response of F‑VASI 90 would enter the stable retreated state. In reality, people in the retreated state would have treatment for a given period of time, followed by an assessment of their vitiligo's response to treatment. The EAG explained that once people experience a reduction in F‑VASI response to below F‑VASI 75 from baseline, they transition to the non-response health state. People entering this state remain there for the lifetime of the model and are unable to return to any previous states within the model. This issue was also present in the company's initial model. As a result of this, people were permitted to have only 1 course of retreatment with ruxolitinib cream. The EAG thought this did not reflect expected use in clinical practice because people may have multiple rounds of ruxolitinib cream if their vitiligo responds to it. The patient expert explained that people would not necessarily continue to apply ruxolitinib cream continuously once repigmentation had occurred, and the frequency of application would be expected to reduce after an initial period of treatment. The clinical experts agreed it would be unusual for people to continuously apply creams for multiple years, and they would likely have maintenance treatment rather than stopping treatment completely. The committee was disappointed that the company did not revise the model to address these issues with retreatment and the permanency of the non-response state. It agreed with the clinical and patient experts that people with a high response to treatment would likely have maintenance treatment (or a reduced dose) rather than the highly controlled stopping and reinitiation rules in the economic model. So, the committee concluded that the model likely underestimated both the costs and benefits of treatment in this section of the model structure. The EAG noted concerns with the benefits accrued in the ruxolitinib retreatment phase in proportion to the costs when compared with the same ratio in the initiation phase. The committee agreed with concerns about the costs and benefits associated with retreatment, because the costs were approximately equivalent to 1 month of treatment (at the initiation phase), but benefits were modelled to potentially last for multiple years. It concluded that the model did not accurately capture the likely reality of retreatment in clinical practice or ultimate disease course. It considered this structure likely biased in favour of ruxolitinib cream because of the minimal costs, but this was unclear because the clinical evidence of an appropriate maintenance dose is not available.

When assessing the revised model, the EAG noted that the average additional time people spent with a response of F‑VASI 90 had approximately doubled in comparison to the initial version. It provided Markov model traces from both versions of the model. It attempted to validate the proportions of people with a response of F‑VASI 90 at year 1 and 2 by comparing them to time spent at F‑VASI 90 seen in the TRuE‑V trials (30.3% [106/350] at year 1 and 18.7% [61.8% of 30.3%] at year 2). Analysis done by the EAG indicated that proportion of people achieving F‑VASI 90 at year 2 was not consistent between models and was higher than seen in the TRuE-V trials. The EAG also noted that in the revised model, the proportion of people achieving F‑VASI 90 increased between years 1 and 2. This implied that more people who did not achieve F‑VASI 90 by year 1 did so by year 2, whereas a smaller proportion of people who achieved F‑VASI 90 at year 1 had lost their response by year 2. The company disagreed with the premise of the EAG's validation exercise because the proportion of people reaching F‑VASI 90 at year 2 only included those that initially had an F‑VASI 90 response at year 1 from the TRuE‑V trials (cohort A). Also, it did not include people with a potentially slower response to treatment that resulted in an F‑VASI 90 response by year 2 (cohort B). The committee considered the proportion of people having a F‑VASI 90 response or higher was a key driver in the economic model and should be informed and validated against results from the trial using an intention-to-treat style analysis of everyone randomised at the start of treatment, because this level of response could take more than 1 year to realise. It recognised that this analysis was complicated in using the available evidence because of the trials' 6‑month open-label extension and crossover, re-randomisation at 1 year of each of the cohorts and analysis of attrition. The company considered the total number who reached a F‑VASI 90 response was greater than what the model predicted, which validated the model outcomes. The EAG commented that it was unable to validate the proportion of people who would have this response quoted by the company and it was unclear how this was derived. The committee agreed with the company that more people may have an F‑VASI 90 response with further treatment after 1 year, and the revised model structure allows for this. So, it did not question the validity of the output on this basis. But it agreed that for the purposes of validation of the trial outcomes in the appropriate population, it was unclear what proportion of people in the TRuE‑V trials had a F‑VASI 90 response to ruxolitinib cream at each time point. This is because the design of the studies did not allow for this to easily be established and there was a lack of clear reporting from the company on how the estimates from the trials were derived. The committee considered that the company's estimate of the proportion of people achieving F‑VASI 90 at year 2 was plausible, but it was uncertain how this estimate was derived. It concluded that the uncertainty behind this key driver of the model interacts with the issue of retreatment (see section 3.13), because it describes the proportion of people who would potentially benefit from retreatment/maintenance of treatment.

The non-response health state included phototherapy costs as part of best supportive care, every 4‑week cycle for 10 years from baseline. The company assumed that a large proportion of people in the non-response health state (the company considers the exact figure to be confidential so it cannot be reported here) have a course of hospital-based phototherapy for 9 months every year. The EAG advised that the company overestimated the proportion of people who have phototherapy and the expected costs of such treatment in the non-response health state. It considered the company's assumption of near-continuous phototherapy was not plausible given current NHS dermatology capacity constraints. The committee noted that the company's estimate of the proportion of people who would have phototherapy was much higher than the estimate provided by clinical experts, who considered that about 25% of people would have phototherapy (see section 3.4). The clinical expert explained that a course of hospital-based phototherapy for vitiligo would be for no longer than 12 months, because it would not be realistic to expect people to attend hospital appointments beyond this period. They explained that a person could potentially have another course of phototherapy in their lifetime, but it would not be possible to have continuous phototherapy each year. The committee decided that the company's assumptions about the use of phototherapy in the non-response health state likely biased the cost-effectiveness results in favour of ruxolitinib cream. It concluded that the company should revise its phototherapy treatment-duration assumptions and the proportion of people who have phototherapy in line with clinical practice for people with vitiligo. At consultation, the company updated its cost-effectiveness analyses to align with the committee's preferences on phototherapy in the non-response state. When comparing ruxolitinib cream with phototherapy (with or without topical corticosteroids) or no active treatment only, people in the non-response health state do not have phototherapy. When comparing ruxolitinib cream with no active treatment followed by phototherapy, 25% of people are assumed to have phototherapy in the non-response health state in line with the clinical opinion received at the first committee meeting. The committee concluded that the company's updated assumptions on phototherapy were reflective of clinical practice for people with vitiligo.

In the model, the number of appointments for NHS psychological support varied depending on the health state. The EAG advised that the company overestimated the proportion of people having psychological support in its base-case analysis. The EAG noted that in the TRuE‑V trials at baseline, the mean scores on the Hospital Anxiety and Depression Scale (HADS) were within normal range. It considered that there was no difference in HADS score between those having ruxolitinib and vehicle cream at 24 weeks. The committee considered that this suggested that a lower proportion of people would be expected to have psychological support than modelled by the company and that this would not largely differ based on response to treatment. The EAG reduced the proportion of people having psychological support and applied this value to all health states in its base-case analyses. This was based on clinical advice to the EAG, which suggested that about 15% of people with vitiligo are referred to psychological support resources. The committee noted the company's model also assumed that people in the non-response health state would have NHS dermatology appointments about every 2 months for 10 years after baseline. The clinical experts explained that this did not reflect clinical practice given current NHS dermatology resource constraints. The committee decided that the company's dermatology attendance and psychological support assumptions overestimated resource use, which likely biased the cost-effectiveness results in favour of ruxolitinib cream. It noted that changing these assumptions had a large impact on the ICER. The committee preferred the EAG's approach for modelling the proportion of people having psychological support. It concluded that the company should revise its assumptions on NHS dermatology attendance in line with expected clinical practice for people with vitiligo. At consultation, the company updated its disease management assumptions in line with the committee's preferences. Monitoring in secondary care by a dermatologist for people with vitiligo that did not respond to treatment was reduced to 15%, and all health states included 15% of people accessing psychological support services. The committee concluded that these disease management assumptions were more aligned with expected clinical practice for people with vitiligo.

NICE's health technology evaluations manual notes that judgements about the acceptability of a technology as an effective use of NHS resources will take into account the degree of certainty around the ICER. The committee will be more cautious about recommending a technology if it is less certain about the ICERs presented. The committee considered the most appropriate comparison was with no active treatment followed by phototherapy. Including the confidential patient access scheme for ruxolitinib cream, the company's deterministic base-case ICER for this comparison was £18,103 per QALY gained. The EAG presented analyses that included minor corrections to the company's base case and its preferred modelling assumptions. These included:

• capping the utility values at general population values, and setting the F‑VASI 25 to 49 value equal to the F‑VASI 50 to 74 value

• assuming the dose of ruxolitinib cream was the mean value from the trials, excluding the outliers whose duration of treatment was imputed as 1 day.
  
  Including the confidential patient access scheme for ruxolitinib cream, the EAG's probabilistic base-case results for this comparison was £25,856 per QALY gained at the second committee meeting. The company proposed to revise its confidential patient access scheme after the appeal, which slightly improved the ICER (the data is confidential so cannot be reported here). The committee decided that neither of these estimates captured the underlying structural uncertainty inherent in the model and key inputs. The committee noted that the ICER was sensitive to applying a different modelling approach to the 'no-response' utility state and capping the utility values to that of the general population as proposed by the EAG. Applying these approaches in the model and considering a number of plausible scenarios resulted in ICERs that ranged from £33,065 per QALY gained to £167,585 per QALY gained at the second committee meeting, which slightly improved with the company's proposed revised confidential patient access scheme at the third committee meeting (the data is confidential so cannot be reported here). The committee thought this probably captured the range of uncertainty in the quality-of-life estimates. But it still had concerns about how retreatment was operationalised and whether it reflects how ruxolitinib cream would be used in clinical practice, which it concluded could not be captured in the ICER calculation.

The committee noted potential equality issues raised at scoping, in the stakeholder and expert submissions and during the appeal. It recalled that the impact and risk of sunburn exists for all skin tones, but vitiligo is more noticeable in people with brown and black skin tones. Also, the burden of vitiligo and its physical, social and psychosocial implications may be disproportionately greater in people with black and brown skin tones and young people (see section 3.3). As well as greater disease burden, there may be an additional cultural burden and social stigma for people with black and brown skin tones, which may lead them to more discrimination and social exclusion (see section 3.3). The committee also noted the risk of depression and anxiety with vitiligo, which may be greater among people from ethnic minority backgrounds. It discussed comments that if ruxolitinib cream were recommended, it should be offered to all people with vitiligo irrespective of their ethnicity or any other protected characteristic. At the third committee meeting, the patient experts also stated that ruxolitinib should not be recommended for individual subgroups. But the greater impact of vitiligo on people with black and brown skin tones and the associated physical, social and psychosocial implications for them should be considered in the committee's decision making.

The committee also noted the comments highlighting how vitiligo may disproportionately impact young people (see section 3.3), disrupting their education, daily life, friendships, social relationships and early careers. It discussed whether, if ruxolitinib cream were recommended, it should be available only to people 12 years and over. At the third committee meeting, the company explained that it is looking into gaining marketing authorisation for younger age groups.

The committee heard that people with black or brown skin tones with vitiligo, compared with people with white skin tones and vitiligo, are more likely to be stigmatised and socially excluded (see section 3.3). The submissions from patient groups after appeal explained that access to secondary care takes a long time and may be available in only some areas. Stakeholder and expert submissions highlighted the personal and financial burden associated with a course of phototherapy; this may mean it is not suitable for some people who are eligible for treatment (see section 3.4). Patient submissions and the patient expert explained how a lack of access to phototherapy is worse for people who have a lower income and who may not have flexible work hours or cannot travel to phototherapy sessions.

The committee understood its obligations in relation to the Equality Act 2010. It was also aware that aiming to reduce health inequalities is one of the principles guiding the development of NICE recommendations. This may mean, where appropriate and there is relevant evidence, recommendations may be made for specific groups of people. It recalled that the company, the patient and clinical experts, and stakeholders in response to draft guidance, did not want ruxolitinib to be recommended for specific subgroups defined by skin colour (see section 3.10). It also noted that the company's revised subgroup analysis by Fitzpatrick score after appeal reflected poor statistical practice. But it was aware that the social and cultural implications constitute a large part of the disease burden for people with black and brown skin tones. Questionnaires, such as DLQI, do not capture psychological distress, stigma or social burden that are particularly relevant to vitiligo. So, it is likely that some benefits associated with ruxolitinib were further underestimated. The committee was not presented with evidence on children, who are outside of the marketing authorisation, or young people (aged 12 to 17 years). But it thought there may be some benefits associated with ruxolitinib in this population group that were not fully captured in the model. The committee took this into account in its decision making.

The committee noted that ruxolitinib cream is the first licensed treatment for non-segmental vitiligo involving the face in people 12 years and over (see section 3.4). It recognised that because ruxolitinib cream is a topical treatment it may have advantages over phototherapy that requires someone with vitiligo to visit the hospital to complete a course. The committee noted the company's statement that the utility estimates derived from condition-specific outcome measures mapped to EQ‑5D (see section 3.17) may not fully capture the magnitude of the health-related quality of life impairment of living with vitiligo. It also noted that some quality-of-life benefits associated with ruxolitinib might be further underestimated for people with black or brown skin tones, or not fully captured for young people (see section 3.3). It considered these factors when deciding if ruxolitinib cream was innovative. Otherwise, the committee did not identify additional benefits of ruxolitinib cream not captured in the economic modelling. So, it concluded that all additional benefits of ruxolitinib cream had already been considered.