3 The company's submission

The Appraisal Committee (section 8) considered evidence submitted by Gilead Sciences and a review of the submissions by the Evidence Review Group (ERG; section 9).

3.1 The company provided clinical‑effectiveness evidence, identified by systematic review, consisting of 13 studies investigating the effect of sofosbuvir plus ribavirin alone or ribavirin and peginterferon alfa in adults with chronic hepatitis C. These included:

  • Studies in people who have not had treatment for hepatitis C virus (HCV) before (described as 'treatment‑naive' in this document) with genotype 1, 4, 5 or 6 HCV:

    • 1 phase III open‑label, single arm study (NEUTRINO, n=327)

    • 3 phase II randomised controlled trials (ATOMIC, n=332; QUANTUM, n=50; SPARE, genotype 1 only, n=60).

  • Studies in people with treatment‑naive HCV or in people who have had treatment before (described as 'treatment‑experienced' in this document) with genotype 2 and 3 HCV:

    • 4 phase III randomised controlled trials (FISSION, n=499 treatment‑naive; FUSION, n=201 treatment‑experienced; POSITRON, n=278 treatment‑naive and ‑experienced, people who had treatment before were considered to be intolerant to interferon, ineligible for interferon or unwilling to take it; VALENCE, n=419 treatment‑naive and ‑experienced)

    • 1 phase II open‑label, randomised controlled trial (ELECTRON, n=95 treatment‑naive)

    • 1 phase II single‑arm open‑label study (LONESTAR‑2, n=47 treatment‑experienced)

    • 1 phase II 2‑arm open‑label single cohort study (PROTON, n=25 treatment‑naive).

  • 1 phase II open‑label 4 cohort study in people with genotype 1, 2 and 3 HCV and HIV co‑infection (PHOTON‑1, n=223).

  • 1 phase II open‑label single‑arm study in people with HCV waiting for a liver transplant (P7977‑2025, n=61).

    People in the sofosbuvir trials were tested for cirrhosis using Fibrotest (a biomarker test that uses the results of 6 blood serum tests to generate a score correlating to the degree of liver damage) and Fibroscan (a non‑invasive scan allowing the measurement of liver fibrosis based on its elasticity). No liver biopsies were performed at study entry and therefore liver fibrosis according to METAVIR score (which is based on liver biopsy histology) was not available for the sofosbuvir trials.

Evidence in people with genotype 1, 4, 5 or 6 HCV

Treatment-naive population

3.2 NEUTRINO compared the efficacy and safety of sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks in people with genotype 1, 4, 5 or 6 treatment‑naive chronic HCV with a historical control. The primary outcome was sustained virological response 12 weeks after the end of treatment. The study did not include sites in the UK. A historical control rate of 60% for sustained virological response was used for peginterferon alfa‑2a and ribavirin, taken from the phase III telaprevir (ADVANCE) and boceprevir (SPRINT‑2) trials. The people in the study had a median age of 54 years (age range 19 to 70 years); 64% were men; 78% had baseline HCV RNA greater than 6 log10 IU/ml (viral load, or the number of virus particles in the blood; a viral load less than 6 log10 IU/ml has been linked to better response to treatment); 17% had cirrhosis; 89% had genotype 1 HCV and 11% had genotype 4, 5 or 6 HCV.

3.3 Results from the NEUTRINO study showed that 12 weeks after the end of treatment with sofosbuvir plus peginterferon alfa and ribavirin, 90% (95% confidence interval [CI] 87 to 93%, p<0.001) of people with genotype 1, 4, 5 or 6 treatment‑naive HCV had a sustained virological response. Cirrhosis and non‑CC IL28B polymorphism were both associated with a reduced sustained virological response at 12 weeks: 92% (95% CI 89 to 95%) for people without cirrhosis, 80% (95% CI 67 to 89%) for those with cirrhosis (p=0.0018), and 98% (95% CI 93 to 100%) for people with the IL28B CC genotype polymorphism compared with 87% (95% CI 82 to 91%) for those with the non‑CC IL28B polymorphism (p=0.006). Sustained virological response at 12 weeks was 90% for people with genotype 1 HCV. Overall, the sustained virological response at 12 weeks was 97% for people with genotype 4, 5 or 6 HCV (100% in the 33 people without cirrhosis, and 50% in the 2 people with cirrhosis). No patient had a relapse during treatment. Relapse after virological response at the end of treatment occurred in 28 of 327 people after stopping treatment; 25 completed 12 weeks of treatment and 3 did not complete the treatment course. The company did not provide any information about the family origin of people in the NEUTRINO study.

3.4 ATOMIC compared the efficacy and safety of sofosbuvir plus peginterferon alfa and ribavirin in people with genotype 1, 4, 5 or 6 treatment‑naive chronic HCV. The study included 3 treatment arms: 1 arm had sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks, the second arm had the same treatment for 24 weeks, and the third arm had the same treatment for 12 weeks, followed by 12 weeks of sofosbuvir monotherapy (sofosbuvir monotherapy is outside of the marketing authorisation for sofosbuvir, but the results were included by the company for information). The primary outcome was sustained virological response 24 weeks after the end of treatment. Most people had genotype 1 HCV, and no one with genotype 5 HCV was enrolled in the study. Results showed that after treatment with sofosbuvir, sustained virological responses of 96–98% were achieved in each treatment arm. This suggests that for sofosbuvir plus peginterferon alfa and ribavirin treatment there is no increase in sustained virological response when treatment is extended beyond 12 weeks.

3.5 QUANTUM compared the efficacy and safety of sofosbuvir plus ribavirin for 12 weeks with 24 weeks of treatment in people with genotype 1, 4, 5 or 6 treatment‑naive chronic HCV. The primary outcome was sustained virological response 12 weeks after the end of treatment, which was 56% (n=25) for people who had 12 weeks of sofosbuvir plus ribavirin and 52% (n=25) for people who had 24 weeks of sofosbuvir plus ribavirin. Results showed no statistically significant difference between treatment arms.

3.6 SPARE was a 2 part study that investigated the efficacy and safety of sofosbuvir plus ribavirin treatment for 24 weeks in people with genotype 1 treatment‑naive chronic HCV. The first part was a 1 arm open‑label study of the efficacy and safety of sofosbuvir plus ribavirin treatment for 24 weeks. The second part investigated 24 weeks of sofosbuvir plus ribavirin (using the licensed weight‑based dose) compared with sofosbuvir plus a low, unlicensed dose of ribavirin. The primary outcome was sustained virological response 24 weeks after the end of treatment. In part 1 of the study, sustained virological response was achieved in 90% (n=9) of people. In part 2, 24 people in each group (96%) had viral suppression by week 4 of treatment. However after completing treatment, 7 people in the weight‑based ribavirin group and 10 people in the low‑dose ribavirin group had a relapse. The sustained virological response 12 weeks after the end of treatment in the intention‑to‑treat population was 68% (n=25) for people having 24 weeks of sofosbuvir plus weight‑based ribavirin and 48% (n=25) for people having 24 weeks of sofosbuvir plus low‑dose ribavirin. The sustained virological response 24 weeks after the end of treatment was the same as the response 12 weeks after the end of treatment.

Treatment-experienced population

3.7 The company did not provide any evidence for the efficacy of sofosbuvir plus ribavirin, or sofosbuvir plus peginterferon alfa and ribavirin in people with genotype 1, 4, 5 or 6 treatment‑experienced chronic HCV.

Evidence in people with genotype 2 or 3 HCV

Treatment-naive population

3.8 FISSION compared sofosbuvir plus ribavirin for 12 weeks with peginterferon alfa‑2a plus ribavirin treatment for 24 weeks in people with genotype 2 or 3 treatment‑naive chronic HCV. The primary outcome was sustained virological response 12 weeks after the end of treatment. The non‑inferiority of sofosbuvir plus ribavirin compared with peginterferon alfa‑2a plus ribavirin for sustained virological response at 12 weeks (primary end point) was tested. People in the study were randomised in a 1:1 ratio and stratified by the presence or absence of cirrhosis, HCV genotype (2 or 3) and baseline HCV RNA level (<6 log10 IU/ml or ≥6 log10 IU/ml). The people in the study had a median age of 50 years (range from 19 to 77 years); 66% were men; 57% had baseline HCV RNA levels greater than 6 log10 IU/ml; 20% had cirrhosis; 72% had genotype 3 HCV.

3.9 Results from FISSION showed that at 12 weeks after the end of treatment, sustained virological response was 67% in both treatment groups. Sofosbuvir plus ribavirin was non‑inferior to peginterferon alfa‑2a plus ribavirin for the primary end point. The absolute difference between treatment groups after adjusting for stratification was 0.3% (95% CI −7.5% to 8.0%, non‑inferiority p<0.001). HCV genotype and cirrhosis were associated with differences in sustained virological response (see table 2).

Table 2 Sustained virological response 12 weeks after the end of treatment from FISSION

Percentage of people with a sustained virological response 12 weeks after end of treatment (95% CI)

HCV genotype

Sofosbuvir plus ribavirin

Peginterferon alfa‑2a plus ribavirin

2

97% (90 to 100%), (n=70)

78% (66 to 87%), (n=67)

3

56% (48 to 63%), (n=183)

63% (55 to 70%), (n=176)

2 or 3 without cirrhosis

72% (65 to 78%), (n=206)

74% (67 to 80%), (n=193)

2 or 3 with cirrhosis

47% (33 to 62%), (n=50)

38% (25 to 53%), (n=50)

2 without cirrhosis

98.3%*, (n=59)

81.5 %*, (n=54)

2 with cirrhosis

90.9%*, (n=11)

61.5 %*, (n=13)

3 without cirrhosis

61%*, (n=145)

71%*, (n=139)

3, with cirrhosis

34%*, (n=38)

30%*, (n=37)

*Confidence intervals not reported, HCV; hepatitis C virus

3.10 ELECTRON was carried out in 2 centres in New Zealand. The study included 8 treatment arms, only 5 of which were used by the company to inform its submission. The treatment arms presented included people with genotype 1, 2 and 3 treatment‑naive chronic HCV who had sofosbuvir plus ribavirin with or without peginterferon alfa‑2a. In 4 of the treatment arms, people with genotype 2 or 3 HCV had sofosbuvir plus ribavirin for 12 weeks and either 0, 4, 8 or 12 weeks of peginterferon alfa‑2a. In another treatment arm, added as a protocol amendment after the 4 previous dose‑ranging treatment arms were completed, people with genotype 2 or 3 HCV had sofosbuvir plus peginterferon alfa‑2a and ribavirin for 8 weeks. Across the 5 study arms that were included in the company's submission, 100% of people with genotype 2 or 3 HCV had a sustained virological response 12 weeks after the end of treatment.

3.11 PROTON was a study in 22 centres in the USA in which people with genotype 1, 2 and 3 treatment‑naive chronic HCV had sofosbuvir plus peginterferon alfa‑2a and ribavirin. The company only presented results from the study arm that included people with genotype 2 or 3 HCV, who had sofosbuvir plus peginterferon alfa‑2a and ribavirin for 12 weeks, because the other study arm was not used to inform its regulatory submission. Results from PROTON showed that sustained virological response 12 weeks after the end of treatment was 92% (no confidence interval reported in company's submission) across both genotypes, and 93% in people with genotype 2 HCV and 90% in people with genotype 3 HCV.

Treatment-experienced population

3.12 FUSION compared the efficacy and safety of sofosbuvir plus ribavirin for either 12 or 16 weeks in people with genotype 2 or 3 chronic HCV, whose disease had no response to previous HCV treatment (25%), or had lost its initial response during or after previous HCV treatment (75%). The study did not include sites in the UK. The people in the study had a median age of 56 years (range 24 to 70 years); 70% were men; 73% had baseline HCV RNA levels greater than 6 log10 IU/ml; 34% had cirrhosis; 63% had genotype 3 HCV.

3.13 Results from FUSION showed that HCV genotype and cirrhosis were associated with differences in sustained virological response (see table 3).

Table 3 Sustained virological response 12 weeks after the end of treatment from FUSION

Percentage of people with a sustained virological response 12 weeks after end of treatment

HCV genotype

Sofosbuvir plus ribavirin for 12 weeks

Sofosbuvir plus ribavirin for 16 weeks

2 or 3

50% (95%CI 40 to 60%), (n=100)

73% (95% CI 63 to 81%), (n=95)

2

86%*, (n=36)

94%*, (n=32)

3

30%*, (n=64)

62%*, (n=63)

2 without cirrhosis

96%*, (n=26)

100%*, (n=23)

2 with cirrhosis

60%*, (n=10)

78%*, (n=9)

3 without cirrhosis

37%*, (n=38)

63%*, (n=40)

3 with cirrhosis

19%*, (n=26)

61%*, (n=23)

*Confidence intervals not reported, HCV; hepatitis C virus

3.14 LONESTAR‑2 evaluated the efficacy and safety of sofosbuvir plus peginterferon alfa‑2a and ribavirin for 12 weeks in people with genotype 2 or 3 chronic HCV, whose disease had no response to previous HCV treatment (15%), or had lost its initial response during or after previous HCV treatment (85%). The study included 1 site in the USA. The people in the study had a median age of 56 years (age range 39 to 72 years); 68% were men; the mean baseline HCV RNA level was 6.2 log10 IU/ml (range from 4.0 to 7.2 log10 IU/ml); 55% had cirrhosis; 51% had genotype 3 HCV.

3.15 Results from LONESTAR‑2 showed that sustained virological response 12 weeks after the end of treatment was 89% (no confidence intervals were reported in the company's submission) in people with genotype 2 or 3 HCV. HCV genotype and cirrhosis were not associated with statistically significant differences in sustained virological response. At 12 weeks after the end of treatment, sustained virological response was 96% and 83% in people with genotype 2 and genotype 3 HCV respectively. In people with genotype 2 HCV without cirrhosis, sustained virological response at 12 weeks after the end of treatment was 100% and in people with cirrhosis it was 93%. In people with genotype 3 HCV, the sustained virological response was 83% for people with and without cirrhosis.

Treatment-naive or treatment-experienced

3.16 VALENCE was an unblinded study in which all people with genotype 2 HCV had sofosbuvir plus ribavirin for 12 weeks, and those with genotype 3 HCV had sofosbuvir plus ribavirin for 24 weeks. Because of changes made during the study, 11 people with genotype 3 HCV had a 12 week course of therapy. As a result of emerging data showing that people with genotype 3 HCV had higher sustained virological responses when they were treated for longer, treatment for all people with genotype 3 in the study was extended to 24 weeks and the goals of the study were redefined to be descriptive and not include hypothesis testing. People in the study had a median age of 51 years (range 19 to 74 years); 60% were men; the mean baseline HCV RNA level was 6.4 log10 IU/ml; 21% had cirrhosis; 78% had genotype 3 HCV. In around 65% of people with treatment‑experienced HCV, initial response was lost during or after previous treatment, 30% had no response to interferon‑based treatment, and 5% were intolerant to interferon.

3.17 Results from VALENCE showed that sustained virological response 12 weeks after the end of treatment for people with genotype 2 HCV having sofosbuvir plus ribavirin for 12 weeks was 93% (no confidence intervals were reported in company's submission). In people with genotype 3 HCV who were treated for 24 weeks, the sustained virological response 12 weeks after the end of treatment was 84% overall, 93% in people who had never been treated and 77% in people who had been previously treated (no confidence intervals for this study were reported in the company's submission).

Population for whom interferon treatment was unsuitable (treatment-naive and treatment-experienced)

3.18 POSITRON evaluated the efficacy and safety of sofosbuvir plus ribavirin compared with placebo for 12 weeks in people with genotype 2 or 3 HCV. People in the study had previously discontinued interferon therapy owing to unacceptable adverse events (the company referred to this group as interferon intolerant), or had a concurrent medical condition preventing therapy with an interferon‑containing regimen (the company referred to this group as interferon ineligible), or were unwilling to have interferon treatment. Similar proportions of people with genotype 2 and 3 HCV were enrolled (51% and 49% respectively) in the study. People were randomised in a 3:1 ratio to receive sofosbuvir plus ribavirin or placebo, and were stratified (grouped) by the presence or absence of cirrhosis. The difference in sustained virological response 12 weeks after the end of treatment was assessed for superiority, which was demonstrated if the p‑value was less than 0.05. People treated in the study had a median age of 54 years (range 21 to 75 years); 54% were men; 70% had baseline HCV RNA levels greater than 6 log10 IU/ml; 16% had cirrhosis. The proportions of people who were intolerant to interferon, ineligible for interferon, or unwilling to have it were 9%, 44%, and 47% respectively. Most people had not had previous treatment for chronic hepatitis C (81.3%).

3.19 Results from POSITRON showed that HCV genotype and cirrhosis were associated with differences in sustained virological response in people treated with sofosbuvir plus ribavirin (see table 4). The difference in sustained virological response between the sofosbuvir plus ribavirin and the placebo group was statistically significant (p<0.001) for people with genotype 2 or 3 chronic HCV.

Table 4 Sustained virological response 12 weeks after the end of treatment from POSITRON

Percentage of people with a sustained virological response 12 weeks after end of treatment

HCV genotype

Sofosbuvir plus ribavirin for 12 weeks

Placebo

2 or 3

78% (95%CI 72 to 83%), (n=207)

0%, (n=71)

2

93%*, (n=109)

3

61%*, (n=98)

2 without cirrhosis

92%*, (n=92)

2 with cirrhosis

94%*, (n=17)

3 without cirrhosis

68%*, (n=84)

3 with cirrhosis

21%* (n=14)

*Confidence intervals not reported, HCV; hepatitis C virus

People with HIV and HCV co-infection

3.20 The safety and efficacy of 12 or 24 weeks of treatment with sofosbuvir plus ribavirin in people with genotype 1, 2 or 3 chronic hepatitis C who were co‑infected with HIV was evaluated in an open‑label clinical study (PHOTON‑1). People in the study had genotype 2 or 3 treatment‑naive or treatment‑experienced HCV or genotype 1 treatment‑naive HCV. People with genotype 2 or 3 treatment‑naive HCV had sofosbuvir plus ribavirin for 12 weeks. People with genotype 2 or 3 treatment‑experienced HCV and people with genotype 1 treatment‑naive HCV had sofosbuvir plus ribavirin for 24 weeks. Participants were either not on antiretroviral therapy with a CD4+ cell count above 500 cells/mm3 or had virologically suppressed HIV‑1 with a CD4+ cell count above 200 cells/mm3. At the time of enrolment, 95% of participants had antiretroviral therapy. Preliminary sustained virological response at 12 weeks was available for 210 people.

3.21 Results from PHOTON‑1 showed that treatment with sofosbuvir plus ribavirin for 12 weeks in people with genotype 1, 2 or 3 HCV and HIV co‑infection and 24 weeks in people with genotype 1, 2 or 3 HCV and HIV co‑infection resulted in sustained virological response at 12 weeks after treatment irrespective of HCV genotype (≥93%; interim analysis). Similar safety and tolerability profiles were reported in people with HIV and HCV co‑infection and in people with HCV only.

People awaiting liver transplant

3.22 An open‑label clinical study (P7977‑2025) was carried out in people with chronic hepatitis C awaiting a liver transplant. It evaluated the safety and efficacy of sofosbuvir plus ribavirin administered before transplant to prevent post‑transplant HCV reinfection. The primary end point of the study was post‑transplant virological response (HCV RNA undetectable at 12 weeks after transplant). People with HCV regardless of genotype, with hepatocellular carcinoma suitable for liver transplant, had 400 mg sofosbuvir and 1000–1200 mg ribavirin daily for a maximum of 24 weeks. This was subsequently amended to 48 weeks or until the time of liver transplant, whichever was first. An interim analysis of results for 61 people who had sofosbuvir and ribavirin, most of whom had genotype 1 HCV, showed that 44 had a liver transplant up to 48 weeks after treatment with sofosbuvir and ribavirin and 41 had no detectable HCV RNA at the time of their transplant. Results suggested that treatment with sofosbuvir and ribavirin prevented HCV recurrence in 64% of people compared with a 100% historical risk of reinfection without prophylaxis. During treatment, HCV RNA suppression in people with well‑compensated cirrhosis awaiting a liver transplant for hepatocellular carcinoma was rapid and similar to that seen in other patient populations treated with sofosbuvir regimens.

Adverse effects of treatment

3.23 The company presented data on adverse events for NEUTRINO, FISSION, FUSION, POSITRON and VALENCE. The most common adverse events among people receiving sofosbuvir and ribavirin therapy (with or without peginterferon alfa) were fatigue, headache, anaemia, nausea, insomnia, irritability, rash, pruritis, myalgia, decreased appetite, influenza‑like illness, chills, pyrexia, and neutropenia. Of these events, fatigue and headache were usually the most frequent, affecting more than 40% of the people in some studies.

Health-related quality of life

3.24 The company assessed health‑related quality of life during the phase II and III trials using the Chronic Liver Disease Questionnaire – Hepatitis C (CLDQ‑HCV), the Functional Assessment of Chronic Illness Therapy‑Fatigue measurement system (FACIT‑F), the Work Productivity and ActivityImpairment questionnaire (WPAI), or the Short Form‑36 items survey (SF‑36). People in the phase III trials were not aware of their sustained virological response when completing the quality‑of‑life questionnaires. The results from NEUTRINO showed that there were differences in health‑related quality of life scores between baseline and the end‑of‑treatment and that scores returned to baseline values by the post‑treatment week 12 visit. The results from the FISSION study indicated that health‑related quality of life during treatment in people who had peginterferon alfa‑2a plus ribavirin was statistically significantly lower than for people in the sofosbuvir plus ribavirin arm. No difference was seen between the arms 12 weeks after the end of each treatment. The CLDQ‑HCV results from FUSION indicated that health‑related quality of life scores did not decrease significantly in either treatment group and there were no statistically significant differences in overall scores between the groups. The health‑related quality of life data obtained from POSITRON showed decreases (worsening) in all SF‑36 scales and the Mental Component and Physical Component scores in both treatment groups during treatment (baseline through to week 12). In the sofosbuvir plus ribavirin group the differences were statistically significant (p<0.001) from baseline in the Physical Function scale, Role Physical, Vitality, Social Functioning, Role Emotional, and Mental Health scales; however, there were no statistically significant differences from placebo at any time point.

Mixed treatment comparison

3.25 The company carried out a mixed treatment comparison to explore the comparative data for sofosbuvir and other relevant comparators. Because of limited data, a mixed treatment comparison network could not be formed for all the relevant populations in the decision problem and the comparison was done only for people with genotype 1, 2 or 3 treatment‑naive HCV for whom interferon therapy was suitable. In addition, the company's economic model required that efficacy data were split by cirrhosis status and these data were not available for all trials. In people with genotype 1 HCV, a network including sofosbuvir was possible only by linking 2 small phase II trials (ATOMIC and PROTON) which included only people without cirrhosis. In people with genotype 2 or 3 HCV, the mixed treatment comparison results were based on people with and without cirrhosis combined. The results of the mixed treatment comparison showed that for people with genotype 1 treatment‑naive HCV regardless of cirrhosis status, 84.9% of those who had sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks had a sustained virological response at 12 weeks after treatment, compared with 46.2% of people who had peginterferon alfa and ribavirin for 48 weeks, 76.5% of people who had telaprevir plus peginterferon alfa and ribavirin and 69.7% of people who had boceprevir plus peginterferon alfa and ribavirin. For people with genotype 2 treatment‑naive HCV without cirrhosis, 98.6% of those who had sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks had a sustained virological response at 12 weeks after treatment, compared with 85.6% of people who had peginterferon alfa and ribavirin for 24 weeks. For people with genotype 2 treatment‑naive HCV with cirrhosis, sustained virological response was achieved in 97.5% of people who had sofosbuvir and ribavirin for 12 weeks, and in 67.5% of those who had peginterferon alfa and ribavirin for 24 weeks. For people with genotype 3 treatment‑naive HCV without cirrhosis for whom interferon therapy was suitable, 62% of those who had sofosbuvir plus ribavirin for 12 weeks had a sustained virological response at 12 weeks after treatment, compared with 68.3% of people who had peginterferon alfa and ribavirin for 24 weeks. For people with genotype 3 treatment‑naive HCV with cirrhosis who had sofosbuvir plus ribavirin for 12 weeks or peginterferon alfa and ribavirin for 24 weeks, the sustained virological response was similar (47.8% and 42.8% respectively) between the 2 treatment groups. The company highlighted several limitations to its mixed treatment comparison including the fact that the 12‑week sofosbuvir plus ribavirin regimen used to treat genotype 3 HCV is not licensed. Therefore, the company stated that the results of the mixed treatment comparison could not be considered robust.

Evidence Review Group comments

3.26 The ERG reviewed the clinical evidence in the company's submission. It considered that the company's interpretation of the clinical evidence was overall justified and unbiased. However, the ERG cautioned that most of the evidence provided did not directly address the decision problem, because of the lack of head‑to‑head studies against current standard of care comparators. In addition, it highlighted that no studies were included that examined the efficacy of sofosbuvir within its marketing authorisation for people with genotype 1 treatment‑experienced HCV. The ERG also noted that some of the evidence from VALENCE supporting the treatment regimens licensed for use in people with genotype 3 HCV should be interpreted with caution because randomisation was broken during the study, and some people were switched from 12 to 24 weeks of treatment with sofosbuvir plus ribavirin.

3.27 The ERG noted that the company only included adverse events from the 5 phase III studies (NEUTRINO, FISSION, FUSION, POSITRON and VALENCE). However, the ERG confirmed that the adverse events in the phase II studies were similar to those in the phase III studies. Overall, the ERG was satisfied that the evidence showed that treatment with sofosbuvir‑based regimens was generally well tolerated and led to fewer adverse events than treatment with peginterferon alfa and ribavirin.

Cost effectiveness

3.28 The company identified 112 cost‑effectiveness studies of chronic hepatitis C treatments. No studies were identified that compared sofosbuvir with alternative treatments.

3.29 To assess the cost effectiveness of sofosbuvir the company submitted a multi‑state Markov model, which compared sofosbuvir plus ribavirin and sofosbuvir plus peginterferon alfa and ribavirin with the comparators defined in the decision problem (that is, boceprevir or telaprevir plus peginterferon alfa and ribavirin for people with genotype 1 HCV, and peginterferon alfa and ribavirin or placebo for people with other HCV genotypes). The structure of the model was based on published health economic models, but was amended by the company to reflect the data available from its pivotal clinical trials and only distinguished between people with and without cirrhosis. The company used patient characteristics from the HCV UK research database to inform the population entering the model, including mean age at start of treatment, disease severity distribution and weight. The model had a total of 9 health states according to disease stage and treatment response. People entered the model in either the non‑cirrhotic or compensated cirrhosis stages of disease. People who had antiviral treatment could move into the non‑cirrhotic or sustained virological response cirrhotic health states. Those who did not clear the virus after treatment remained in their respective health states, or progressed to more severe stages of chronic HCV. All patients in the decompensated cirrhosis health state were assumed to be candidates for liver transplant. The model assumed that people who have a sustained virological response will not progress to more severe health states during or after therapy. Reversion to less severe health states was not permitted if treatment was unsuccessful.

3.30 The company applied age‑specific general population mortality rates to each health state in the model. The same model structure was used for all patients irrespective of HCV genotype or treatment experience. For the first 2 years a 3‑month cycle was used in the model, then the remaining cycles each lasted 1 year. A half‑cycle correction was applied, which is consistent with previous hepatitis C appraisals. An NHS and personal and social services perspective was taken and a lifetime horizon was used, with costs and outcomes discounted at 3.5%.

3.31 Data from clinical trials were used to inform model inputs for treatment effects, health‑related quality of life and adverse events. Treatment effect data were based on the sustained virological responses taken from the sofosbuvir clinical trials. If data for comparators were not available in these trials, they were taken from other published studies identified by the company. The company collected quality‑of‑life scores at baseline, week 12 during treatment, and at 4, 12 and 24 weeks after treatment. The SF‑36 quality of life data were converted to SF‑6D utility data and used in the company's base case. The company also converted SF‑36 data to the EQ‑5D and incorporated these in a deterministic sensitivity analysis. Adverse event rates were obtained from the sofosbuvir clinical trials and published studies. The company incorporated the rates of grade 3 and 4 pruritus, diarrhoea and nausea, vomiting, rash, anaemia, thrombocytopenia, neutropenia, and depression from the trials into the model so that drug acquisition costs could be included for interventions associated with managing these adverse events.

3.32 The company used transition probabilities for disease progression from 2 published UK health technology assessments and 1 UK study: Hartwell et al. (2011), Shepherd et al. (2007), and Grishchenko et al. (2009), which used estimates from the Trent database (a large sample of people with HCV who attended only non‑tertiary centres in the UK).

3.33 Utility values estimated from the sofosbuvir clinical studies were not used to inform the model. Instead, the company used utility values from previous technology appraisals for hepatitis C treatments that were based on the UK trial of mild chronic hepatitis C by Wright et al. (2006). The company calculated treatment‑related utilities by applying treatment‑related utility decrements to the baseline utility estimates.

3.34 The company compared sofosbuvir plus ribavirin (with or without peginterferon alfa), telaprevir plus peginterferon alfa and ribavirin, boceprevir plus peginterferon alfa and ribavirin, peginterferon alfa plus ribavirin, and best supportive care. Sofosbuvir, telaprevir, boceprevir, peginterferon alfa‑2a and ribavirin were used in the model according to their marketing authorisations. The company applied no stopping rules, lead‑in phase, or option for sofosbuvir retreatment, in line with the sofosbuvir clinical trials.

3.35 The company used costs in the model that reflected the UK NHS perspective, comprising treatment‑related costs (drug acquisition and patient monitoring), health‑state costs and adverse event costs. Drug costs were based on the list price in the BNF (June 2013). Costs for drugs used to treat adverse events in the model were taken from the BNF. The company used the BNF price of ribavirin (Copegus 400 mg, 56 tablet packs) at a cost of £246.65 in the model. Costs for the health states in the model were identified using published sources taken from the resource and costs systematic review done by the company. The costs for the non‑cirrhotic health state were based on a calculation of the costs associated with mild and moderate cirrhosis (Wright et al.) using an assumed 77%: 23% split between mild and moderate cirrhosis. Costs were inflated to 2011/12 prices (using the Hospital and Community Health Service pay and prices index).

3.36 Monitoring costs included resource unit costs of outpatient appointments, inpatient care, tests and investigations (virology, pathology, haematology, immunology, radiology) and procedures (liver biopsy). The source for monitoring costs was the National Schedule of Reference Costs, published studies or expert opinion.

Results

3.37 The company presented base‑case analyses for sofosbuvir plus ribavirin with or without peginterferon alfa compared with current standard of care, based on HCV genotype, interferon eligibility and treatment history. The company's results show that sofosbuvir treatment regimens increased the cost of treatment, but were associated with more quality‑adjusted life years (QALYs) gained, a greater probability of sustained virological response (cure), and a reduction in end‑stage liver disease and death.

Genotype 1

3.38 For people with genotype 1 treatment‑naive HCV for whom interferon therapy was suitable, the company's base‑case analysis showed that the incremental cost‑effectiveness ratio (ICER) for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £14,930 per QALY gained (incremental cost £19,129; incremental QALYs 1.3). Boceprevir plus peginterferon alfa and ribavirin was dominated (that is, sofosbuvir plus peginterferon alfa and ribavirin was less expensive and more effective) and telaprevir plus peginterferon alfa and ribavirin was extendedly dominated (that is, its ICER is higher than that of the next, more effective option when compared with a common baseline). For people with genotype 1 treatment‑naive HCV for whom interferon therapy was not suitable, the company's base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £49,249 per QALY gained (incremental cost £63,903; incremental QALYs 1.3).

3.39 The company did not do an economic analysis for genotype 1 treatment‑experienced HCV because there was no clinical evidence available to populate the economic model for this population. However, an economic analysis for genotype 1 treatment‑experienced HCV was later provided by the company (see section 3.69).

Genotype 2

3.40 In people with genotype 2 treatment‑naive HCV for whom interferon therapy was suitable, the company's base‑case ICER for sofosbuvir plus ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 24 weeks was £46,324 per QALY gained (incremental cost £27,779; incremental QALYs 0.6). In people with genotype 2 treatment‑naive HCV for whom interferon therapy was not suitable, the company's base‑case ICER for sofosbuvir plus ribavirin for 12 weeks compared with no treatment was £8154 per QALY gained (incremental cost £20,051; incremental QALYs 2.5).

3.41 In people with genotype 2 treatment‑experienced HCV for whom interferon therapy was suitable, the company's base‑case ICER for sofosbuvir plus ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £9274 per QALY gained (incremental cost £21,498; incremental QALYs 2.3). In people with genotype 2 treatment‑experienced HCV for whom interferon therapy was not suitable, the company's base‑case ICER for sofosbuvir and ribavirin treatment for 12 weeks compared with no treatment was £8591 per QALY gained (incremental cost £20,697; incremental QALYs 2.4).

Genotype 3

3.42 In people with genotype 3 treatment‑naive HCV for whom interferon therapy was suitable, the company's base‑case ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 24 weeks was £20,613 per QALY gained (incremental cost £24,970; incremental QALYs 1.2). In people with genotype 3 treatment‑naive HCV for whom interferon therapy was not suitable, the company's base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £21,478 per QALY gained (incremental cost £55,137; incremental QALYs 2.6).

3.43 In people with genotype 3 treatment‑experienced HCV for whom interferon therapy was suitable, the company's base‑case ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £8557 per QALY gained (incremental cost £19,634; incremental QALYs 2.3). In people with genotype 3 treatment‑experienced HCV for whom interferon therapy was not suitable, the company's base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £28,569 per QALY gained (incremental cost £58,828; incremental QALYs 2.1).

Genotypes 4, 5 or 6

3.44 In people with genotype 4, 5 or 6 treatment‑naive HCV for whom interferon therapy was suitable, the company's original base‑case ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £26,797 per QALY gained (incremental cost £23,942; incremental QALYs 0.9). In sensitivity analyses, the sustained virological response for peginterferon alfa plus ribavirin for 48 weeks (which varied between 26.6% and 73.4%) had a large impact on the cost‑effectiveness results.

3.45 The company tested the robustness of the model using deterministic sensitivity analyses. Results showed that the ICERs were most sensitive to changes in the discount rate (varied between 0% and 6% for costs and outcomes simultaneously) and the utility increment after achieving a sustained virological response. The company concluded that the results of the deterministic sensitivity analyses showed that the ICERs for sofosbuvir remained below £20,000 per QALY gained in the following subgroups:

  • people with genotype 2 treatment‑naive and treatment‑experienced HCV, for whom interferon treatment is unsuitable, compared with no treatment

  • people with genotype 2 and 3 treatment‑experienced HCV, for whom interferon therapy is suitable, compared with no treatment.

    The results of the deterministic sensitivity analyses showed that the ICERs for sofosbuvir were between £20,000 and £30,000 per QALY gained in the following subgroups:

  • people with genotype 1 treatment‑naive HCV, for whom interferon therapy is suitable, compared with peginterferon alfa and ribavirin and boceprevir plus peginterferon alfa and ribavirin

  • people with genotype 2 treatment‑experienced HCV, for whom interferon therapy is suitable, compared with peginterferon alfa and ribavirin

  • people with genotype 3 treatment‑experienced HCV, for whom interferon therapy is suitable, compared with peginterferon alfa and ribavirin.

    The results of the deterministic sensitivity analyses showed that the ICERs for sofosbuvir were above £30,000 per QALY gained in the following subgroups:

  • people with genotype 1 treatment‑naive HCV, for whom interferon therapy is suitable, compared with telaprevir plus peginterferon alfa and ribavirin

  • people with genotype 3 treatment‑naive HCV, for whom interferon is unsuitable, compared with no treatment

  • people with genotype 3 treatment‑naive HCV, for whom interferon therapy is suitable, compared with peginterferon alfa and ribavirin.

3.46 The company also carried out probabilistic sensitivity analyses to explore parameter uncertainty. Although it did not draw any specific conclusions from the analyses, results suggested that sofosbuvir had less than a 50% probability of being cost effective in 6 of the base‑case comparisons (if the maximum acceptable ICER was £20,000 per QALY gained) and greater than a 50% probability of being cost effective in 9 of the base‑case comparisons (if the maximum acceptable ICER was £30,000 per QALY gained).

3.47 During clarification, the company explored the effect of including a transition probability from Cardoso et al. (2010) (0.005; 95% CI 0.013 to 0.002) from the sustained virological response cirrhotic health state to the hepatocellular carcinoma health state in the economic model, and varied the probability in line with the upper and lower limits of the 95% confidence interval (see section 3.52). The ICERs from the company's exploratory analyses were slightly higher than those estimated in the company's base case (ranging from £7507 per QALY gained [for sofosbuvir plus peginterferon alfa and ribavirin compared with boceprevir plus peginterferon alfa and ribavirin in people who are eligible for interferon, with genotype 1 treatment‑naive HCV] to £54,957 per QALY gained [for sofosbuvir plus ribavirin compared with 24 weeks of peginterferon alfa‑2a and ribavirin treatment in people with genotype 2 treatment‑naive HCV, for whom interferon therapy is suitable]).

HCV and HIV co-infected populations

3.48 The company provided a separate economic analysis for people co‑infected with HIV and HCV. In people with HIV and genotype 1 treatment‑naive HCV for whom interferon therapy is suitable, the company's base‑case ICER for sofosbuvir plus ribavirin treatment for 24 weeks compared with no treatment was £28,504 per QALY gained, or £43,836 per QALY gained compared with peginterferon alfa‑2a and ribavirin treatment for 48 weeks. The company's base‑case ICER for sofosbuvir plus ribavirin for 12 weeks compared with peginterferon alfa 2a and ribavirin treatment for 48 weeks in people with genotype 2 treatment‑naive HCV and HIV‑co‑infection was £55,867 per QALY gained. The company's base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment in people with genotype 2 treatment‑experienced HCV and HIV co‑infection was £10,572 per QALY gained, or £128,248 per QALY gained compared with peginterferon alfa‑2a and ribavirin treatment for 48 weeks. For people with genotype 3 treatment‑naive HCV and HIV co‑infection, 12 weeks of sofosbuvir plus ribavirin dominated treatment with peginterferon alfa and ribavirin. The ICERs for 24 weeks of sofosbuvir plus ribavirin were £10,646 per QALY gained compared with no treatment and £90,822 per QALY gained compared with peginterferon alfa‑2a and ribavirin for 48 weeks in people with genotype 3 treatment‑experienced HCV and HIV co‑infection. The company did not provide an economic analysis for people co‑infected with HIV and genotype 4, 5 or 6 HCV.

Evidence Review Group comments

3.49 The ERG reviewed the company's model and economic systematic review. The ERG considered that the company's methods of economic evaluation and the model produced were acceptable. It validated the company's model by comparing the total costs and QALYs predicted by the model for the treatment‑naive interferon‑eligible genotype 1 HCV population with the corresponding figures for treatment with peginterferon alfa plus ribavirin with and without boceprevir or telaprevir obtained from the previous NICE technology appraisals for boceprevir for the treatment of genotype 1 chronic hepatitis C and telaprevir for the treatment of genotype 1 chronic hepatitis C. The ERG found that the company's model for sofosbuvir was broadly consistent with previous models considered in NICE technology appraisals for hepatitis C, in terms of total costs and QALYs assumed for peginterferon alfa and ribavirin, and for telaprevir. However, the ERG noted that there was a discrepancy between models in the total costs estimated for boceprevir, but it was not able to account for the differences without reviewing the data used in the boceprevir submission. The ERG also considered that the company's economic model captured most of the important aspects of the disease pathway, but noted that it did not include a transition from the sustained virological response cirrhotic health state to the hepatocellular carcinoma health state, which had been previously included in other hepatitis C models. Despite this omission from the company's model, the ERG showed that it did not affect the base‑case ICERs substantially. The ERG considered that the company's model extrapolated intermediate outcomes to final outcomes in a consistent way, drawing on standard sources from the literature.

3.50 The ERG noted that the transition probabilities used by the company in its economic model for the HCV and HIV co‑infected population from the non‑cirrhotic to compensated cirrhosis health states were higher than those assumed for the mono‑infected population. The ERG further noted that people with HCV and HIV co‑infection are likely to have a higher mortality than the population with HCV only, regardless of sustained virological response, and this was not taken into account in the company's model. The ERG noted that a study by Van Der Helm et al. (2013) concluded that the effects of HCV treatment on HIV progression needed to be evaluated further. Therefore, in the ERG's opinion, the evidence needed to accurately evaluate the cost effectiveness of sofosbuvir in the HCV and HIV co‑infected population was not currently available.

Additional exploratory ERG analyses

3.51 The ERG carried out several exploratory analyses, which included:

  • adding a transition probability from the sustained virological response cirrhotic health state to the hepatocellular carcinoma health state and exploring the effect of using different transition probabilities between these 2 health states

  • assessing the effect on the ICERs of variations to all‑cause mortality probabilities

  • evaluating the effect of changing the average age of entry into the model

  • using a range of alternative sustained virological response estimates from studies of comparator treatments

  • assessing the effect of an alternative distribution of people with cirrhosis

  • exploring the effect of the number of people having 24 weeks of sofosbuvir compared with those having 12 weeks of sofosbuvir

  • assessing the effect of using alternative utility increments after sustained virological response.

3.52 The ERG heard from its clinical advisers that a transition from the sustained virological response cirrhotic health state to the hepatocellular carcinoma health state should be included in models for chronic hepatitis C to reflect the clinical course of the disease. The ERG noted that this transition was not included in the company's economic model. In addition, the ERG was unable to calculate the transition probability between these 2 health states used by the company (0.005) in its response to clarification based on the study by Cardoso et al. (see section 3.47). The ERG recalculated the transition probability using the Cardoso et al. study, to produce a value of 0.0123 (95% CI 0.028 to 0.0218). The effect of including this value (and also the upper and lower confidence interval values) was explored by the ERG in a sensitivity analysis that produced ICERs ranging from £7593 per QALY gained [for sofosbuvir plus peginterferon alfa and ribavirin compared with boceprevir plus peginterferon alfa and ribavirin in people with genotype 1 treatment‑naive HCV, for whom interferon therapy is suitable] to £60,887 per QALY gained [for sofosbuvir plus ribavirin compared with 24 weeks of peginterferon alfa and ribavirin treatment in people with genotype 2 treatment‑naive HCV for whom interferon therapy is suitable]).

3.53 The ERG noted that the company used the simple average of mortality from men and women to calculate the age‑specific mortality used in the model. The ERG commented that more men are treated for HCV in clinical practice in England, therefore a weighted average should have been used. During clarification, the company re‑ran their economic model with weighted average mortality probabilities, but did not indicate what weights were used to obtain its results. The ERG carried out an exploratory analysis using a weighting of 61% men and 39% women as used in Wright et al. The ICERs from the ERG's exploratory analyses were slightly higher than those estimated in the company's base case (ranging from £7453 per QALY gained [for sofosbuvir compared with boceprevir plus peginterferon alfa and ribavirin in people with genotype 1 treatment‑naive HCV, for whom interferon therapy is suitable] to £50,083 per QALY gained [for sofosbuvir compared with no treatment in people with genotype 1 treatment‑naive HCV, for whom interferon is unsuitable]).

3.54 The ERG noted that the company's model used an efficacy estimate drawn from a single source when multiple efficacy estimates were available for the same treatment and indication. NICE asked the ERG to carry out further exploratory analyses to inform the Committee's understanding of the effect on the company's ICERs of using a range of alternative sustained virological response estimates from studies of comparator treatments. For people with genotype 1 treatment‑naive HCV, the ERG used alternative estimates of sustained virological response for boceprevir plus peginterferon alfa‑2b and ribavirin from the SPRINT‑2 study, in line with estimates used in the NICE technology appraisal of boceprevir (that is, 68.2% for people with no cirrhosis and 41.7% for people with cirrhosis, compared with the company's base‑case values of 64.1% for people with no cirrhosis and 55.0% for people with cirrhosis). Applying the alternative sustained virological response estimates for boceprevir gave a lower ICER than the company's base case (the ICER was not reported by the ERG in its additional analyses).

3.55 The ERG also used alternative sustained virological response estimates for people with genotype 3 treatment‑naive HCV, who are eligible for interferon. First, the ERG modelled the effect of an alternative sustained virological response of 90.7% for sofosbuvir plus peginterferon alfa‑2a and ribavirin. This response was at the lower end of the 95% confidence interval for the population with no cirrhosis from the sofosbuvir trials (an estimate of 97.4% was used in the company's base case, an average of the sustained virological response from ELECTRON and PROTON). The resulting ICER for sofosbuvir plus peginterferon alfa‑2a and ribavirin compared with peginterferon alfa‑2a and ribavirin for this subgroup was £23,772 per QALY gained (compared with the company's base‑case ICER of £20,613 per QALY gained). The ERG also explored the effect of an alternative sustained virological response of 92.3% for sofosbuvir plus peginterferon alfa‑2a and ribavirin in people with genotype 3 treatment‑naive HCV and cirrhosis. Applying this alternative response lowered the ICER to £18,187 per QALY gained.

3.56 The ERG also explored the effect of alternative assumptions about the natural history of chronic hepatitis C infection on the company's ICERs. In particular, it investigated the effect of assuming an alternative distribution of cirrhosis. The ERG used a distribution for new and existing cirrhosis obtained from Hartwell et al. (2011) based on data from a London teaching hospital where 32% of existing patients with HCV and 10% of new patients with HCV had cirrhosis. The results of the exploratory analysis suggested that using this distribution increased the company's base‑case ICERs for people with treatment‑naive HCV and reduced the ICERs for people with treatment‑experienced HCV across all genotypes. In people with genotype 3 treatment‑naive HCV for whom interferon therapy was suitable, the ICER for sofosbuvir plus peginterferon alfa‑2a and ribavirin (12 weeks) compared with peginterferon alfa and ribavirin (24 weeks) increased from £20,613 per QALY gained in the company's base case to £30,175 per QALY gained. The ICERs for treatment for people with genotype 1 treatment‑naive HCV remained above £30,000 per QALY gained (irrespective of interferon eligibility). Results for all other subgroups remained below £30,000 per QALY gained.

3.57 The ERG explored the effect of using a lower transition probability from the non‑cirrhotic health state to the compensated cirrhosis health state, at age 40 years in the company's model. The resulting ICERs increased across all subgroups and were higher than those estimated in the company's base case (ranging from £9458 per QALY gained [for sofosbuvir compared with boceprevir plus peginterferon alfa and ribavirin in people with genotype 1 treatment‑naive HCV, for whom interferon therapy is suitable] to £61,077 per QALY gained [for sofosbuvir compared with no treatment in people with genotype 1 treatment‑naive HCV, for whom interferon treatment is unsuitable]).

3.58 The ERG explored the effect on the company's base‑case ICERs of varying the percentage of people having 24 weeks of sofosbuvir treatment compared with those having 12 weeks of treatment. The 3 subgroups who might have 12 or 24 weeks of sofosbuvir treatment according to the marketing authorisation are people with:

  • genotype 1 HCV, having sofosbuvir plus peginterferon alfa and ribavirin

  • genotype 2 HCV, having sofosbuvir and ribavirin

  • genotype 3 HCV, having sofosbuvir plus peginterferon alfa and ribavirin.

    The ERG's clinical advisers differed in their opinions about how long these groups would have treatment. One clinical expert stated that it would be unlikely that more than 1–2% of people would be considered better off with longer therapy and that this group are identified in the summary of product characteristics as needing consideration for longer treatment periods. Another clinical expert stated that at least 20% of people might need 24 weeks of therapy, especially those who are intolerant to interferon or have severe cirrhosis.

3.59 The ERG pointed out that the company's economic model allows for a 12 week regimen of sofosbuvir plus peginterferon alfa and ribavirin for people with genotype 1 treatment‑naive HCV for whom interferon therapy is suitable and a 12 week regimen of sofosbuvir and ribavirin for people with genotype 2 treatment‑naive HCV (regardless of interferon eligibility). The economic model did allow for a 24 week regimen of sofosbuvir and ribavirin for various genotype 3 HCV subgroups. The ERG therefore compared sofosbuvir plus ribavirin for 24 weeks with either peginterferon alfa and ribavirin treatment for 24 weeks in people with genotype 3 treatment‑naive HCV and either no treatment or 48 weeks of peginterferon alfa and ribavirin treatment for people with genotype 3 treatment‑experienced HCV. The resulting ICERs were more than double the company's base‑case results (which assumed that these patient groups only have 12 weeks of sofosbuvir and ribavirin).

3.60 The ERG carried out sensitivity analyses to evaluate the effect of changing the average age of entry into the model. The resulting ICERs generally decreased when a lower average age of 35 years was selected for entry into the model, and were higher when the average age selected was 55 years. The lowest ICERs ranged from £6717 per QALY gained (using 35 years) to £9170 per QALY gained (using 55 years) for sofosbuvir compared with boceprevir plus peginterferon alfa and ribavirin in people with genotype 1 treatment‑naive HCV, for whom interferon therapy is suitable. The highest ICERs ranged from £47,254 per QALY gained (using 35 years) to £60,976 per QALY gained (using 55 years) for sofosbuvir compared with 24 weeks of peginterferon alfa‑2a and ribavirin treatment in people with genotype 2 treatment‑naive HCV, for whom interferon therapy is suitable.

3.61 The ERG also carried out sensitivity analyses to explore the effect on the company's ICERs of using different utility increments (0 and 0.04 [taken from Vera‑Llonch et al. 2013], compared with the company's estimate of 0.05) after sustained virological response. The resulting ICERs in the ERG's sensitivity analysis were consistently higher than the company's base‑case results. When using a utility increment of 0.04, the ICERs ranged from £7899 per QALY gained for sofosbuvir plus peginterferon alfa and ribavirin compared with boceprevir plus peginterferon alfa and ribavirin in people with genotype 1 treatment‑naive HCV, for whom interferon therapy is suitable, to £53,793 per QALY gained for sofosbuvir plus peginterferon alfa and ribavirin compared with no treatment in people with genotype 1 treatment‑naive HCV, for whom interferon treatment is unsuitable. When using a utility increment of 0, the ICERs ranged from £12,732 per QALY gained for sofosbuvir plus peginterferon alfa and ribavirin compared with boceprevir plus peginterferon alfa and ribavirin in people with genotype 1 treatment‑naive HCV, for whom interferon therapy is suitable, to £92,795 per QALY gained for sofosbuvir plus peginterferon alfa and ribavirin compared with no treatment in people with genotype 1 treatment‑naive HCV, for whom interferon treatment is unsuitable.

3.62 The ERG also explored the effect on the company's base‑case ICERs when alternative estimates of sustained virological response for peginterferon alfa plus ribavirin to those used by the company (from McHutchison et al. 2009) were applied, for people with genotype 1 treatment‑naive HCV for whom interferon therapy is suitable. The ERG used estimates from Roberts et al. (2009), which reported a sustained virological response of 51% for people without cirrhosis and 6% for people with cirrhosis, and estimates from Hadziyannis et al. (2004), which were 56% and 38% respectively. Using the estimates from Roberts et al. the company's ICER for sofosbuvir plus peginterferon alfa and ribavirin compared with peginterferon alfa and ribavirin increased to £18,209 per QALY gained. Similarly, the base‑case ICER increased to £21,848 per QALY gained for the same comparison when estimates from Hadziyannis et al. were used.

3.63 The ERG also carried out a scenario analysis that considered the combined impact on the company's ICERs of including a transition from the sustained virological response cirrhotic health state to the hepatocellular carcinoma health state, alternative utility increments after a sustained virological response, and an alternative estimate of efficacy for peginterferon alfa and ribavirin in the population with genotype 1 treatment‑naive HCV who are eligible for interferon (using values described in sections 3.52, 3.61 and 3.62). The ICERs from the ERG's exploratory analyses were higher than those estimated in the company's base case (ranging from £9415 per QALY gained [for sofosbuvir plus peginterferon alfa and ribavirin compared with boceprevir plus peginterferon alfa and ribavirin in people with genotype 1 treatment‑naive HCV, for whom interferon therapy is suitable] to £109,526 per QALY gained [for sofosbuvir plus ribavirin compared with no treatment in people with genotype 1 treatment‑naive HCV, for whom interferon is unsuitable]).

Additional analyses for genotypes 1 and 3 HCV

3.64 During consultation the Committee asked that the company provide additional evidence, which the Committee believed would permit it to come to a better informed conclusion. Specifically, the Committee requested that the company carry out several exploratory analyses for sofosbuvir plus ribavirin, with or without peginterferon alfa, compared with peginterferon alfa and ribavirin in people with genotype 1 and genotype 3 chronic hepatitis C, because these genotypes represent 89% of HCV infections in England. This included revised cost‑effectiveness analyses presented separately for people with and without cirrhosis, with and without HIV‑co‑infection, and by treatment history. The Committee asked that the analyses should incorporate:

  • the transition from the sustained virological response cirrhotic health state to the hepatocellular carcinoma health state, using the transition probability estimates from Cardoso et al.

  • alternative sustained virological response estimates for peginterferon alfa and ribavirin (for example from Hadziyannis et al.)

  • alternative utility increments after sustained virological response (for example SF‑36 values from the trials collected at 24 weeks post‑treatment, and Vera‑Llonch et al.) and

  • alternative costs for ribavirin (for example, the cost of generic ribavirin as calculated by the Commercial Medicines Unit Electronic Market Information Tool) in the model.

    The Committee also asked that sensitivity analyses, including the Committee's assumptions, should be explored:

  • assuming that up to 100% of people with genotype 3 HCV have sofosbuvir plus ribavirin for 24 weeks

  • assuming that an increased proportion of people for whom interferon therapy is suitable may be unwilling to have interferon treatment and therefore have sofosbuvir plus ribavirin for 24 weeks

  • varying the age of entry into the model from 35 and 55 years

  • varying all‑cause mortality by assuming the population entering the model comprises 61% men and 39% women, in line with estimates from Wright et al.

3.65 The company provided the additional analyses requested, although the revised base‑case assumptions were slightly different to those requested by the Committee. The company justified each change to the revised base‑case assumptions, which included a transition from the sustained virological response cirrhotic health state and the non‑sustained virological response cirrhotic health state to the hepatocellular carcinoma health state, using the transition probability estimates from Cardoso et al. (2010); alternative utility increments from Vera‑Llonch et al., alternative costs for ribavirin and all‑cause mortality assuming the population entering the model comprises 61% men and 39% women, in line with estimates from Wright et al. The company chose to use the sustained virological response rates for peginterferon alfa and ribavirin from McHutchison et al. for its revised base case, and provided a sensitivity analysis incorporating the sustained virological response rates from Hadziyannis et al. (see section 3.74).

3.66 The company presented ICERs to the Committee for subgroups stratified by genotype, treatment history, interferon eligibility and cirrhosis status. The company provided the Committee with a 'global' ICER based on all patients mono‑infected with HCV for genotypes 1 to 6, which was £16,199 per QALY gained. The company also provided the Committee with 'global' ICERs by genotype, weighted by treatment history and cirrhosis status, which ranged from £10,753 per QALY gained in people with genotype 1 HCV to £31,361 per QALY gained in people with genotype 2 HCV.

Genotype 1

3.67 For people with treatment‑naive genotype 1HCV, for whom interferon therapy is suitable, the company's revised base‑case analysis showed that the ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £17,476 per QALY gained. The ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with boceprevir plus peginterferon alfa and ribavirin was £10,335 per QALY gained and £15,396 per QALY gained compared with telaprevir plus peginterferon alfa and ribavirin. For people with genotype 1 treatment‑naive HCV, for whom interferon therapy is not suitable, the company's base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £47,611 per QALY gained.

3.68 When stratified by the presence or absence of cirrhosis, the company's revised base‑case analysis showed that the ICER for sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks in people with treatment‑naive HCV who are eligible for interferon was £25,237 per QALY gained for people without cirrhosis, and £5352 for people with cirrhosis. The stratified ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with boceprevir plus peginterferon alfa and ribavirin was £14,280 per QALY gained for people without cirrhosis and £2819 per QALY gained for people with cirrhosis. The stratified ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with telaprevir plus peginterferon alfa and ribavirin was £22,304 per QALY gained for people without cirrhosis and £4253 per QALY gained for people with cirrhosis. For people with genotype 1 treatment‑naive HCV for whom interferon therapy is not suitable, the company's base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment, stratified by cirrhosis status, was £51,478 per QALY gained for people without cirrhosis and £35,754 per QALY gained for people with cirrhosis.

3.69 Because of the lack of clinical trial evidence in people with genotype 1 treatment‑experienced HCV, the company provided the Committee with an estimated cost‑effectiveness calculation for this subgroup. The company explained that historically, approximately 50% of people with genotype 1 treatment‑naive HCV had disease that responded to treatment with peginterferon alfa and ribavirin, but in the interim analysis provided to the US Food and Drug Administration (FDA), 89% of people with genotype 1 treatment‑naive HCV in NEUTRINO had disease that responded to sofosbuvir plus peginterferon alfa and ribavirin. The FDA accepted that the higher rate of overall sustained virological response seen in NEUTRINO was likely driven by those patients who were virus‑free 12 weeks after the end of treatment, but who would not have had this response if treated with peginterferon alfa and ribavirin alone. Assuming that people who would have had a sustained virological response with peginterferon alfa and ribavirin alone had a sustained virological response with sofosbuvir plus peginterferon alfa and ribavirin, the FDA assumed that the increase in sustained virological response from 50% to 89% represented the efficacy of sofosbuvir plus peginterferon alfa and ribavirin. The FDA calculated that given the high sustained virological responses in NEUTRINO, an approximate sustained virological response was 78% for people with treatment‑experienced genotype 1 HCV. Additionally, the company presented interim evidence from Pol et al. (2013), a study of the efficacy of sofosbuvir in people with genotype 1 treatment‑experienced HCV, which suggested that sustained virological responses in this group were 74% after treatment with sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks. Using the estimated sustained virological response of 78% calculated by the FDA, the company's ICER for sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks compared with peginterferon alfa and ribavirin alone for 48 weeks was £12,641 per QALY gained. The company's ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with boceprevir plus peginterferon alfa and ribavirin was £683 per QALY gained and £8203 per QALY gained when compared with telaprevir plus peginterferon alfa and ribavirin in people with genotype 1 treatment‑experienced HCV.

Genotype 3

3.70 In people with genotype 3 treatment‑naive HCV for whom interferon therapy is suitable, the company's revised base‑case ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 24 weeks was £21,860 per QALY gained. In people with genotype 3 treatment‑naive HCV for whom interferon therapy is unsuitable, the company's base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £21,049 per QALY gained.

3.71 In people with genotype 3 treatment‑experienced HCV for whom interferon therapy is suitable, the company's base‑case ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £13,883 per QALY gained. In people with genotype 3 treatment‑experienced HCV, for whom interferon therapy is unsuitable, the company's base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £27,483 per QALY gained.

3.72 When stratified by the absence or presence of cirrhosis, in people with genotype 3 treatment‑naive HCV for whom interferon therapy is suitable, the company's revised base‑case ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 24 weeks was £40,623 per QALY gained for people without cirrhosis and £6556 per QALY gained for people with cirrhosis. In people with genotype 3 treatment‑naive HCV for whom interferon therapy is unsuitable, the company's revised base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £28,044 per QALY gained in people with no cirrhosis, and £10,505 per QALY gained in people with cirrhosis. In people with treatment‑experienced genotype 3 HCV for whom interferon therapy is suitable, the company's revised base‑case ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £18,592 per QALY gained in people without cirrhosis and £6260 per QALY gained in people with cirrhosis. In people with genotype 3 treatment‑experienced HCV for whom interferon therapy is unsuitable, the company's revised base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £31,416 per QALY gained in people without cirrhosis and £19,179 per QALY gained in people with cirrhosis.

HCV and HIV co-infected populations

3.73 The company provided a separate economic analysis for people co‑infected with HIV and HCV. The company reported results from the 1910 study (Rodriguez‑Torres et al. [2013]), which included (F0‑F3) patients with HCV and HIV co‑infection and no cirrhosis. The sustained virological response of 90% seen in the 1910 study was also seen in people with HCV mono‑infection and no cirrhosis in NEUTRINO, which suggested that similar response rates are seen in people with genotype 1 HCV having sofosbuvir plus peginterferon and ribavirin for 12 weeks, regardless of HCV and HIV co‑infection status. The company presented revised base‑case ICERs in people co‑infected with HIV and genotype 1 or 3 HCV, which ranged from £10,376 per QALY gained in people with genotype 3 treatment‑experienced HCV and HIV having sofosbuvir and ribavirin for 24 weeks compared with no treatment, to £27,059 per QALY gained in people with genotype 1 treatment‑naive HCV having sofosbuvir and ribavirin for 24 weeks compared with no treatment.

Additional sensitivity analyses

3.74 The Committee asked that the revised base case include the sustained virological responses from Hadziyannis et al. rather than from McHutchison et al. for peginterferon alfa and ribavirin in people with genotype 1 treatment‑naive HCV. The company expressed concern about this approach because the sustained virological response in Hadziyannis et al. assumed that a METAVIR score from F0 to F2 represented people without cirrhosis and a score from F3 to F4 represented people with cirrhosis, whereas in NEUTRINO, METAVIR scores of F4 for cirrhosis and F0‑F3 for non‑cirrhosis were defined. The company also commented that people in the McHutchison et al. study were more representative of people in the NEUTRINO study. Therefore, the company used the sustained virological responses from McHutchison et al. in its revised base‑case analysis. However, it provided the Committee with the results of a scenario analysis in which it used the sustained virological responses for peginterferon alfa and ribavirin from Hadziyannis et al., which suggested that for people with genotype 1 treatment‑naive HCV, the ICER for sofosbuvir plus peginterferon alfa and ribavirin compared with peginterferon alfa and ribavirin alone was £25,014 per QALY gained, whereas its revised base‑case ICER using the sustained virological responses for peginterferon alfa and ribavirin from McHutchison et al. was £17,476 per QALY gained.

3.75 The company presented a scenario analysis assuming that 100% of people with genotype 3 HCV for whom interferon therapy is suitable would have 24 weeks of sofosbuvir and ribavirin. In people with genotype 3 treatment‑naive HCV for whom interferon therapy is suitable, assuming that 100% of people would have 24 weeks of sofosbuvir and ribavirin treatment compared with 24 weeks of peginterferon alfa and ribavirin treatment, the ICER was £46,956 per QALY gained. In people with genotype 3 treatment‑experienced HCV for whom interferon therapy is suitable, assuming that 100% of people would have 24 weeks of sofosbuvir and ribavirin treatment compared with 48 weeks of peginterferon alfa and ribavirin treatment, the ICER was £48,306 per QALY gained. The company also presented the results of a scenario analysis using the upper (20%) and lower (2%) proportion of people with genotype 3 HCV for whom interferon therapy is suitable and who were expected to have sofosbuvir and ribavirin for 24 weeks as suggested by the ERG's clinical advisers. In people with genotype 3 treatment‑naive HCV for whom interferon therapy is suitable, assuming that 2% of people would have 24 weeks of sofosbuvir and ribavirin treatment compared with 24 weeks of peginterferon alfa and ribavirin treatment, the ICER was £22,385 per QALY gained, and £27,062 per QALY gained when 20% was used. In people with genotype 3 treatment‑experienced HCV for whom interferon therapy is suitable, assuming that 2% of people would have 24 weeks of sofosbuvir and ribavirin treatment compared with 24 weeks of peginterferon alfa and ribavirin treatment, the ICER was £14,467 per QALY gained and £19,890 per QALY gained when 20% was used.

3.76 The ERG reviewed the company's additional evidence. The ERG noted that the company had used most of the Committee's preferred base‑case assumptions (see section 3.64). It further noted that the company did not provide a sensitivity analysis exploring the impact on the ICER of using utility data collected in the clinical trials. The ERG carried out an exploratory analysis in which it used all of the Committee's preferred assumptions to calculate the ICERs for treatment for people with genotype 1 and 3 HCV and also carried out the exploratory scenario analyses requested by the Committee (see section 3.64). The ERG's exploratory analyses resulted in an ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks of £30,993 per QALY gained for people with treatment‑naive genotype 1 HCV, for whom interferon therapy is suitable. The ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with boceprevir plus peginterferon alfa and ribavirin was £12,172 per QALY gained and £18,704 per QALY gained compared with telaprevir plus peginterferon alfa and ribavirin. For people with genotype 1 treatment‑naive HCV, for whom interferon therapy is unsuitable, the base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £58,113 per QALY gained.

3.77 The ERG stratified the exploratory ICERs by the presence or absence of cirrhosis. The ICER for sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £38,460 per QALY gained for people without cirrhosis, and £12,891 for people with cirrhosis. The ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with boceprevir plus peginterferon alfa and ribavirin was £15,653 per QALY gained for people without cirrhosis and £2274 per QALY gained for people with cirrhosis. The ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with telaprevir plus peginterferon alfa and ribavirin was £24,509 per QALY gained for people without cirrhosis and £4680 per QALY gained for people with cirrhosis compared with telaprevir plus peginterferon alfa and ribavirin. For people with genotype 1 treatment‑naive HCV for whom interferon therapy is unsuitable, the base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £58,118 per QALY gained for people without cirrhosis and £58,093 per QALY gained for people with cirrhosis.

3.78 In people for whom interferon therapy is suitable, with genotype 3 treatment‑naive HCV, the ERG's exploratory analyses resulted in an ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 24 weeks of £28,666 per QALY gained. In people with genotype 3 treatment‑naive HCV for whom interferon therapy is unsuitable, the base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £26,611 per QALY gained. In people with genotype 3 treatment‑experienced HCV for whom interferon therapy is suitable, the ERG's exploratory analyses showed that the ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £16,979 per QALY gained. In people with genotype 3 treatment‑experienced HCV, for whom interferon therapy is unsuitable, the company's base‑case ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £34,261 per QALY gained.

3.79 When stratified by the presence or absence of cirrhosis, in people with genotype 3 treatment‑naive HCV for whom interferon therapy is suitable, the ERG's exploratory analyses resulted in an ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 24 weeks that was £46,036 per QALY gained for people without cirrhosis and £8318 per QALY gained for people with cirrhosis. In people with genotype 3 treatment‑naive HCV for whom interferon therapy is suitable, the ERG's exploratory analyses showed that the ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £31,851 per QALY gained in people with no cirrhosis, and £15,133 per QALY gained in people with cirrhosis. In people with treatment‑experienced genotype 3 HCV for whom interferon therapy is suitable, the ERG's exploratory analyses showed that the ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks was £20,694 per QALY gained in people without cirrhosis and £8093 per QALY gained in people with cirrhosis. In people with genotype 3 treatment‑experienced HCV for whom interferon therapy is unsuitable, the ERG's exploratory analyses showed that the ICER for sofosbuvir plus ribavirin for 24 weeks compared with no treatment was £35,744 per QALY gained in people without cirrhosis and £29,704 per QALY gained in people with cirrhosis.

Additional analyses in genotype 4, 5 and 6 HCV

3.80 During consultation several consultees, including the company, asked that the Committee reconsider the clinical and cost‑effectiveness evidence for people with genotype 4 HCV. The consultees indicated that people with genotype 4 HCV represent a group with a particularly high unmet need and minority ethnic groups represent a higher proportion of people who have genotype 4 HCV in the UK. Whereas genotype 4 HCV accounts for 5% of all HCV genotypes in the UK (but is more prevalent in the Middle East and Africa), the prevalence of genotype 4 in the UK is increasing because of migration, HIV co‑infection and intravenous drug use. Additionally, a consultee stated that the proportion of people with genotype 4 HCV who have haemophilia is higher than in other genotypes because of infection with blood products imported from abroad.

3.81 In order to fully explore the potential equality issues raised during consultation, additional evidence was requested from the company, which included HCV prevalence data by genotype and family origin, HIV co‑infection and haemophilia in the UK. The company provided NICE with evidence from a HCV genotype surveillance report commissioned by the company to be produced by Public Health England, which showed the proportion of people with genotype 1 or 3 HCV who were of white or white British family origin was 81% and 72% respectively, whereas minority ethnic groups represented 8% and 18% respectively. The proportion of people with genotypes 4, 5 and 6 HCV who were of white or white British family origin was 44%, 53% and 19%, respectively, whereas minority ethnic groups represented 39%, 28% and 74%, respectively (see table 5).

Table 5 Genotype by family origin

HCV genotype

White or white British (%)

Asian or Asian British (%)

Black or black British (%)

Other or mixed origin (%)

Unknown (%)

1

13675 (81)

875 (5)

116 (1)

265(2)

2023 (12)

2

1883 (84)

75 (3)

12 (1)

35 (2)

239 (11)

3

12001 (72)

2894 (17)

37 (0.22)

146 (0.88)

1532 (9)

4

593 (44)

378 (28)

48 (4)

90 (7)

239 (18)

5

25 (53)

7 (15)

6 (13)

0 (0)

9 (19)

6

12 (19)

4 (6)

1 (2)

42 (66)

5 (8)

Non‑1

46 (70)

3 (5)

0 (0)

1 (2)

16 (24)

Dual

13 (59)

6 (27)

0 (0)

0 (0)

3 (14)

HCV, hepatitis C virus

3.82 The company also provided genotype distribution data from the 2012 United Kingdom Haemophilia Centre Doctors' Organisation look‑back exercise, which reported a prevalence of 3% and 1% of HCV genotypes 4 and 5 respectively among UK HCV‑infected people with haemophilia. The proportion of people with haemophilia with genotypes 1, 2 and 3 HCV is 96%. Additionally, the company presented commercial‑in‑confidence evidence that a disproportionate number of people with HIV co‑infection have genotype 4 HCV compared with people without HIV co‑infection.

3.83 The Committee requested that the company submit a literature review of a range of sustained virological responses for peginterferon alfa and ribavirin to address uncertainty in the sustained virological responses in the comparator arm of the economic model. The company identified 7 studies reporting efficacy data for patients with genotype 4, 5 or 6 HCV in its systematic literature review. The company excluded all studies that recruited solely from Egypt, Africa, Asia or the Middle East because of documented differences in baseline characteristics and response to treatment. According to the company, only Manns et al. (2001) and Lindsay et al. (2001) provided combined data on genotypes 4, 5 and 6 HCV. Of these studies, Manns et al. included more patients (n=16) compared with Lindsay et al. (n=8). The company supported the use of Manns et al. with unpublished data from Imperial College London, which provides treatment for the largest numbers of people with genotype 4 HCV in England, suggesting that the sustained virological response of 50% from Manns et al. in people without cirrhosis was similar to the rate observed at Imperial College (49.5% in patients with genotype 4 having peginterferon alfa and ribavirin for 48 weeks). The company also calculated the sustained virological response for people with genotype 4, 5 or 6 HCV with cirrhosis to be 38.6% based on the relative difference in sustained virological response between people without cirrhosis and people with cirrhosis in the studies in genotype 1, 2 and 3 HCV.

3.84 The ERG did not agree that only studies that included combined genotypes 4, 5 and 6 HCV should be analysed, because this approach excluded some trials of genotype 4 HCV that have larger sample sizes. The ERG did its own rapid systematic literature review that identified 17 studies with sample sizes of 10 or more people with genotype 4 HCV. Ten of the studies took place in the Middle East (particularly Egypt) and 7 were from Europe (3 were randomised controlled studies, 4 were non‑randomised studies). The ERG calculated the overall weighted mean sustained virological response for the 4 well‑reported Middle Eastern studies out of the 10 and the 3 randomised controlled studies from the 7 European studies separately, resulting in a sustained virological response of 65.6% and 55.7%, respectively. When all 17 studies were combined with Manns et al. and Lindsay et al., the resulting weighted sustained virological response was 54.8%. The ERG's clinical advisers provided comment on the relevance of the Middle Eastern studies to NHS practice, stating that there is a significant difference between patients of Egyptian family origin in Europe and those in Egypt, because the average age and number of comorbidities is higher in most European studies, which translates into a reduced response rate.

3.85 The Committee also asked the company to provide a revised base‑case ICER for sofosbuvir plus peginterferon alfa and ribavirin treatment for 12 weeks compared with peginterferon alfa and ribavirin treatment for 48 weeks for people with genotype 4, 5 or 6 treatment‑naive HCV, using the Committee's preferred assumptions, which were using:

  • the Cardoso et al. transition probabilities from the sustained virological response with cirrhosis health state and the cirrhosis health state without a sustained virological response to the hepatocellular carcinoma health state

  • the utility increment after a sustained virological response from Vera‑Llonch et al. and

  • a men:women distribution ratio from Wright et al.

    The Committee also requested that the company show the impact of alternative sustained virological responses for peginterferon alfa and ribavirin alone for 48 weeks from the systematic review of published sources included in its initial submission to the Committee, together with a rationale for the preference of Manns et al. in the revised base‑case model over other sources.

3.86 The company used the sustained virological responses observed in NEUTRINO for the sofosbuvir plus peginterferon alfa and ribavirin arm of the model. The revised base‑case ICER for sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks compared with peginterferon alfa and ribavirin for 48 weeks (using the sustained virological response of 50% for peginterferon alfa and ribavirin from Manns et al.) increased from £26,797 per QALY gained in the original model to £27,505 per QALY gained in the revised base‑case model. The company had identified 7 studies, which included peginterferon alfa and ribavirin for 48 weeks in a treatment arm with sustained virological responses ranging from 33% (Zeuzem et al. 2005) to 77% (Fried et al. 2002). These sustained virological responses were not stratified by cirrhosis status, therefore the company applied the same sustained virological response for people without cirrhosis as for people with cirrhosis in the model. Sustained virological responses for sofosbuvir plus peginterferon alfa and ribavirin from the NEUTRINO study were stratified by cirrhosis status (100% for people without cirrhosis and 50% for people with cirrhosis). The resulting ICERs for sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks compared with peginterferon alfa and ribavirin for 48 weeks using the sustained virological response from Zeuzem et al. (33%) was £19,148 per QALY gained, and £244,387 per QALY gained when Fried et al. (77%) was used. The company also calculated the ICER for sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks compared with peginterferon alfa and ribavirin for 48 weeks using the sustained virological response of 97% for the combined cohort of people with and without cirrhosis in NEUTRINO. The resulting ICER was £47,394 per QALY gained.

3.87 The company provided a sensitivity analysis that used the transition probability from the compensated cirrhosis without a sustained virological response health state to hepatocellular carcinoma from Fattovich et al. (1997) rather than Cardoso et al., which increased the ICERs from £27,505 to £31,713 per QALY gained in the revised base case. When the sustained virological response for peginterferon alfa and ribavirin for 48 weeks from Zeuzem et al. and Fried et al. were used, the ICERs were £22,096 and £151,837 per QALY gained, respectively.

3.88 The ERG used the same revised base‑case assumptions as the company and applied the weighted sustained virological response from the 17 studies it identified, which resulted in a ICER of £37,820 per QALY gained and increased to £40,761 per QALY gained when the transition probability from Fattovich et al. was used for the transition from compensated cirrhosis without a sustained virological response to hepatocellular carcinoma. When the ERG used the weighted sustained virological response from the 10 Middle Eastern studies, the ICER for sofosbuvir plus peginterferon alfa and ribavirin for 12 weeks compared with peginterferon alfa and ribavirin for 48 weeks was £69,181 per QALY gained, whereas the ICER using the 7 European studies in people with genotype 4 HCV only was £40,664 per QALY gained, and £39,109 per QALY gained when the 7 European studies were supplemented with the results from Manns et al. and Lindsay et al.

3.89 Full details of all the evidence are in the committee papers.