The appraisal committee (section 5) considered evidence submitted by Gilead Sciences and a review of this submission by the evidence review group (ERG). See the committee papers for full details of the evidence.
Two trials recruited people who had previous DAA treatment (genotypes 1–6, with or without compensated cirrhosis) and compared 12-week sofosbuvir–velpatasvir–voxilaprevir treatment with placebo (POLARIS-1) or sofosbuvir–velpatasvir (POLARIS-4).
Two trials recruited people who had no previous DAA treatment and compared 8-week sofosbuvir–velpatasvir–voxilaprevir treatment with sofosbuvir–velpatasvir in people with genotypes 1–6, with or without compensated cirrhosis (POLARIS-2) or in people with genotype 3 and compensated cirrhosis (POLARIS-3).
The ERG considered that the trials were generally well conducted, although there was a higher risk of bias in POLARIS-2, -3 and -4 because they were open-label studies, and because only POLARIS-1 randomised all patients. The trial results showed high sustained virological response at 12 weeks, ranging from 80% to 100%, irrespective of hepatitis C virus genotype, cirrhosis stage or treatment history. The committee concluded that sofosbuvir–velpatasvir–voxilaprevir is effective for treating chronic hepatitis C across all subgroups and for genotypes 1–6.
3.3 The most commonly reported adverse events with sofosbuvir–velpatasvir–voxilaprevir were headache and fatigue. The patient expert noted that sofosbuvir–velpatasvir–voxilaprevir is a 3-agent treatment and therefore may result in more adverse events than a 2-agent treatment, but it is still generally tolerable. The committee agreed that sofosbuvir–velpatasvir–voxilaprevir has a relatively favourable safety and tolerability profile (irrespective of cirrhosis stage or treatment history) and concluded that the adverse events associated with sofosbuvir–velpatasvir–voxilaprevir were generally tolerable.
3.4 The structure of the model and its assumptions about the natural history of the disease are similar to models submitted for other NICE technology appraisals for chronic hepatitis C. The ERG noted that treatment-related mortality and background mortality are related to treatment duration and can lead to counterintuitive results when comparing treatments of unequal durations. This is because the mortality in the model starts earlier for the shorter treatment. The company explained that this is a conservative assumption for sofosbuvir–velpatasvir–voxilaprevir because of the short treatment duration of 8 to 12 weeks. The committee was aware that the company had grouped people with mild and moderate fibrosis into a single health state (non-cirrhotic), and agreed that this was consistent with how people are diagnosed in current practice. The committee concluded that the structure of the model was acceptable for decision-making.
Reinfection and future transmission of hepatitis C is modelled as a scenario analysis for people with genotype 3 who have not had DAA
3.6 The company did a separate scenario analysis using dynamic transition modelling, which investigated the impact of onward transmission and reinfection for people with genotype 3 who have not had DAA. In the analysis, the company assumed that only people who inject drugs transmit hepatitis C or become reinfected after being cured. The results were similar to the results of the company's base case. The ERG explained that the scenario analysis made simplifying assumptions and was done only for people with genotype 3 who have not had DAA, with no results by cirrhosis status provided. In previous NICE chronic hepatitis C appraisals, the committee stated that it would have preferred to see a model including both reinfection and transmission. Having seen that the company's results were similar when both types of model structures were used, the committee concluded that the company's base-case model (excluding reinfection and transmission) was acceptable for decision-making.
The company's naive indirect comparison of sustained virological response rates leads to uncertainty in the model results
3.7 The company used a naive indirect comparison to compare sofosbuvir–velpatasvir–voxilaprevir with the relevant comparators in people with genotype 3 who have not had DAA. Because of the lack of comparative trial data for some of the comparators, an indirect treatment comparison was not feasible. The rates of sustained virological response for the comparators in the company's model were selected from individual arms of randomised controlled trials. The company used some of the same rates of sustained virological response for comparator technologies as those used in NICE's technology appraisal guidance on sofosbuvir–velpatasvir. The committee noted that this approach meant that the results were at risk of the kind of bias normally associated with observational studies. The ERG noted that the company combined the rates for people who had, and people who did not have, a previous treatment for the sofosbuvir plus ribavirin treatment. But for the sofosbuvir plus peginterferon alfa and ribavirin treatment, it used only the rates for people who did not have a treatment before. The ERG combined the rates for people who had, and people who did not have, a previous treatment for sofosbuvir plus peginterferon alfa and ribavirin in its exploratory analyses; this had only marginal impact on the company's results. The committee agreed that for consistency, it preferred the ERG's approach to estimating the rates for sofosbuvir plus peginterferon alfa and ribavirin. It concluded that overall, the company's method of estimating efficacy in the model introduced some uncertainty in the results.
3.8 The company used the same sources for non-treatment-specific transition probabilities as those used in NICE's technology appraisal guidance on sofosbuvir–velpatasvir. This included Kanwal et al. (2014) for genotype-specific fibrosis progression and Cardoso et al. (2010) for non-fibrosis progression (not genotype-specific). In its revision to the company's base case, the ERG applied a transition probability from a more recent source (Hepatitis C Trust, 2017) for progression from liver transplant to death. Also, the ERG explored using transition probabilities for compensated cirrhosis (to decompensated cirrhosis and hepatocellular carcinoma), decompensated cirrhosis (to hepatocellular carcinoma and death), and hepatocellular carcinoma (to death) from Fattovich et al. (1997) instead of Cardoso et al. in a scenario analysis, as recommended in the sofosbuvir–velpatasvir guidance. The ERG's analyses had only marginal impact on the company's results. The committee was generally satisfied with the company's approach, but preferred the ERG's approach of using more recent data sources to estimate transition probabilities for progression from liver transplant to death.
The committee prefers utility values from clinical trials, but it accepted the company's utility values
3.9 In its base-case analyses, the company used utility data from the literature (Wright et al. 2006 and Vera-Llonch et al. 2013) in line with NICE's previous technology appraisal guidance on hepatitis C. This was to inform the difference in utility of a health state with or without sustained virological response. The ERG stated that although EQ-5D was not an outcome in the POLARIS trials, health-related quality-of-life data were collected (for example, SF-36 data) and a review of utilities (especially for severe heath states) is needed. The committee noted that in the sofosbuvir–velpatasvir guidance it emphasised that, when available, utility values from the clinical trials are preferred. The committee accepted the company's base-case utility estimates, but stressed that in future hepatitis C appraisals, utility values from the literature will no longer be considered acceptable if there are utility values collected in clinical trials.
3.11 The committee's preferred assumptions were based on the ERG's revisions to the company's base case, including:
the more consistent estimation of sustained virological response for sofosbuvir plus peginterferon alfa and ribavirin in people with genotype 3 who have not had DAA (95.1% and 87.9% for non-cirrhotic and cirrhotic disease respectively; see section 3.7)
the more recent transition probability from liver transplant to death (16% and 5.2% in year 1 and subsequent years respectively; see section 3.8)
increasing the proportion of mild compared with moderate fibrosis in the non-cirrhotic state (from a 83:17 split to a 50:50 split) to better reflect clinical experience and
decreasing the length of follow-up for people without cirrhosis who had a sustained virological response from 2 years to 1 year to better reflect clinical practice.
Using the committee's preferred assumptions and the confidential price discounts for sofosbuvir–velpatasvir–voxilaprevir and its comparators, the ICERs for sofosbuvir–velpatasvir–voxilaprevir were below £20,000 per quality-adjusted life year (QALY) gained, except for the scenario with 12-week treatment for sofosbuvir–velpatasvir–voxilaprevir for people with genotype 3 and compensated cirrhosis who have not had DAA. In this scenario, the ICER was considerably above the range normally considered a cost-effective use of NHS resources (that is, between £20,000 and £30,000 per QALY gained). So 12-week treatment for sofosbuvir–velpatasvir–voxilaprevir in this subgroup could not be recommended. The committee concluded that sofosbuvir–velpatasvir–voxilaprevir was cost effective for the populations in the company's base-case analysis. It can be recommended for treating chronic hepatitis C:
as a 12-week treatment for people with genotypes 1–6 (with or without compensated cirrhosis) who have had DAA and
as an 8-week treatment for people with genotype 3 (with or without compensated cirrhosis) who have not had DAA before.
3.12 Previous NICE technology appraisal guidance on hepatitis C included recommendations on treatment and prescribing decisions because of capacity constraints within the NHS. The clinical experts stated that many people eligible for treatment, particularly people with cirrhosis, have now had treatment creating additional capacity to treat more. The clinical experts also stated that having more affordable drugs with shorter treatment durations also creates additional capacity. However, NHS England commented that there is considerable value in the existing NICE recommendation for multidisciplinary teams in the operational delivery networks to prioritise treatment for people with the highest unmet clinical need and the need for its continuation. NHS England considers that removing this wording would create major challenges and that the capacity constraints within the NHS have not changed sufficiently for the recommendation to be removed at this present time. On balance, after considering arguments both for and against, the committee accepted it was appropriate to continue to include the recommendation on this aspect (see section 1.2) as in previous NICE guidance for the oral hepatitis C treatments.
3.13 The committee considered whether sofosbuvir–velpatasvir–voxilaprevir could be considered innovative, and whether the company's economic analysis had captured all associated health-related benefits. The committee agreed with the company that there is an unmet need for people who have had unsuccessful treatment with DAA. However, the committee concluded that it had taken these potential benefits into account when considering the cost effectiveness of sofosbuvir–velpatasvir–voxilaprevir.
3.14 The committee noted potential equality issues raised during the NICE scoping process. Chronic hepatitis C disproportionately affects some populations such as certain immigrant populations, prison populations, and drug users, in terms of accessing the healthcare system and having access to innovative new treatments. In addition, the Haemophilia Society suggested that this treatment should be a priority for people with a bleeding disorder. Having decided that sofosbuvir–velpatasvir–voxilaprevir should be recommended for all the groups for whom there was evidence presented, the committee agreed that its recommendations were fair. It concluded that no further consideration of potential equality issues was needed to meet NICE's obligation to promote equality of access to treatment.