Protease inhibitors have improved sustained virological response (SVR) rates for subjects with genotype 1 hepatitis C virus infection (HCV). There is however uncertainty regarding how, and in whom, these agents should be used. In previously treated subjects, prior response to interferon has a major effect on SVR rates with protease inhibitor therapy.
To assess the benefits of treatment and to understand the utility of a stopping rule for subjects with a poor interferon response following a 4-week lead-in with pegylated interferon and ribavirin.
Treatment responses and long-term outcomes were modelled using hypothetical 1000 subject cohorts with 5 years of follow-up. Treatment strategies were compared with number needed to treat (NNT) and comparative effectiveness approaches.
Over 5 years of follow-up the NNT to prevent liver-related mortality for subjects with advanced fibrosis was substantially lower than that for subjects with all fibrosis stages (18 vs. 60) indicating particular benefit in this high-risk population. The use of a stopping rule for subjects with advanced fibrosis and a poor interferon response after a 4-week lead-in reduces the number of subjects exposed to a protease inhibitor by 55%. However, 33% fewer liver-related deaths are prevented using this strategy, indicating that there is unacceptable harm associated with this approach over a 5-year follow-up period.
Subjects with advanced fibrosis should be prioritised for triple therapy on the basis of need. Treatment should be continued regardless of initial interferon response to maximise the early prevention of hepatitis C virus-related mortality.
Hepatitis C virus (HCV) infection affects up to 200 million individuals worldwide and is an important cause of both morbidity and mortality. Indeed up to 30% will develop cirrhosis with the attendant risks of liver failure and the development of hepatocellular cancer (HCC).[2, 3] Antiviral treatment has improved during the last two decades, but sustained virological response (SVR) rates remained below 50% for individuals infected with genotype 1 HCV who were treated with pegylated interferon and ribavirin. Since these were until recently the only available treatments many patients have been treated and not cured. Many of these individuals have cirrhosis, are at significant risk from liver related mortality and are prime candidates for more efficacious treatments that will reduce this mortality risk. To estimate this risk reduction it is crucial to understand the impact of the surrogate outcome of SVR on liver-related morbidity and mortality. This understanding permits well-informed discussion with patients who are considering retreatment.
In the last year, the first generation of directly acting antiviral (DAA) agents has been licensed for treatment of patients with genotype 1 HCV infection. These are NS3 serine protease inhibitors that, when used with pegylated interferon and ribavirin, substantially improve response rates and have the potential to cure many individuals who would not have been cured with the previous standard of care.[5, 6] This increased cure rate comes at the expense of increased adverse events and an increased pill burden. Furthermore, DAAs have the potential to cause drug resistance, analogous to that seen with antiviral treatment for human immunodeficiency virus and hepatitis B virus (reviewed in). Resistant species are associated with antiviral treatment failure, but the long-term ramifications of resistance are not known. It is possible that resistance to these agents will compromise the chance of treatment success with regimens that contain similar agents in the future. To limit the development of resistance, all of the phase II and III studies employed strict stopping rules to prevent futile drug exposure. Despite this up to 50% of those treated and who were not cured still developed drug resistant variants although these were frequently short-lived in the plasma.
In the boceprevir development programme and in the phase III study of previously treated subjects with telaprevir, a 4-week lead-in phase with pegylated interferon and ribavirin was employed.[8-10] This strategy confirmed that interferon responsiveness was a key determinant of successful treatment. It has been suggested that the lead-in could be used to identify individuals with poor interferon response where treatment with first-generation protease inhibitors should be avoided, thus eliminating the risk of drug resistance and reducing the frequency and severity of treatment-associated adverse events.[11-13] Indeed, some experts would use the 4-week lead-in with both telaprevir and boceprevir where poor interferon response might be expected to aid decision making. This approach is the subject of intense debate as some patients with poor response during the lead-in phase are subsequently cured with DAA containing treatment.
The aims of this study were therefore twofold: first to quantify the benefit associated with protease inhibitor treatment in previously treated subjects to aid patient selection for treatment, and second to evaluate the benefits and risks of using a lead-in phase with interferon and ribavirin dual therapy to identify patients who will benefit from addition of a protease inhibitor. The comparative effectiveness of treatment strategies was assessed using hypothetical patient cohorts over a clinically relevant 5-year follow-up period.
All patients who had previously failed treatment with interferon and ribavirin were considered suitable for treatment. The proportions of subjects with previous null response (<2log10 decline in plasma HCV RNA during 12 weeks of dual therapy), partial response (>2log10 decline in plasma HCV RNA but without achieving plasma HCV RNA PCR negativity) and relapse (PCR negativity at end of antiviral treatment, but subsequent relapse) were estimated at 45%, 20%, 35% respectively.[15-17] There is no head-to-head comparison of boceprevir and telaprevir, and treatment responses appear comparable.[8-10, 18] As patients with prior null response to dual therapy were not included in the registration studies of boceprevir we planned treatment using a prototypic protease inhibitor modelled on telaprevir. Sustained virological response rates were estimated from studies in previously treated subjects and their subgroup analyses (Table 1, and supplementary information).[9, 19] The utility of treating subjects with advanced fibrosis (defined as METAVIR F3/F4) was first determined using these SVR estimates and by comparison with treatment of subjects with all stages of fibrosis.
Table 1. Calculation of estimated rates of sustained virological response (SVR) for previously treated subjects
Prior treatment response
Advanced fibrosis only
Measured SVR (%)
Number of subjects with SVR per 100 subjects
Measured SVR (%)
Number of subjects with SVR per 100 subjects
Pooled measured SVR rates of protease inhibitor treated subjects were used to estimate SVR rates from an unselected, previously treated, HCV-infected population. Frequency of prior responses was determined as above: relapse 35%, partial response 20% and null response 45%.
Treatment regimens with and without a stopping rule after the 4-week pegylated interferon and ribavirin lead-in phase (<1log10 reduction in HCV RNA) in subjects with advanced fibrosis were then assessed. The variables included in this analysis are shown in Table 2 (and are summarised in Supplementary Figure S1). Standard stopping rules to avoid futile treatment were left in place (i.e. HCV RNA greater than 1000 IU/mL at either 4 weeks or 12 weeks after the start of triple therapy) but were not explicitly modelled. In each case, analyses were done containing only those showing previous partial or null response since it was considered likely that previous relapsers would receive full treatment regardless of response to lead-in pegylated interferon and ribavirin (SVR rates in this population are very high regardless of baseline fibrosis stage). Frequency of poor interferon response was extracted from a subgroup analysis of the REALIZE trial.[9, 19]
Table 2. Variables included in the analysis of a stopping rule after a 4-week lead-in with pegylated interferon and ribavirin
Estimations of the rates of liver-related mortality, hepatic decompensation, and for the development of HCC were reported in a meta-analysis of previously treated patients, and more recently in a prospective study of patients entered into clinical trials in a single centre. These data allow the calculation of the absolute risk reduction (ARR) for each of these clinical events with successful treatment. For instance in the meta-analysis, in previously treated subjects with advanced fibrosis and who do not achieve SVR the annual risk of liver-related mortality is estimated at 2.7%. For individuals with SVR the annual risk reduction is estimated at 0.19, or an annual mortality risk of 0.5% thus giving an ARR of liver-related mortality of 2.2% in those with SVR. The ARR for each outcome is given in Table 2. Adverse outcomes were calculated considering a follow-up duration of 5 years as it is unlikely that additional classes of DAAs will be licensed in that timeframe, and since it is interferon failure in this group that governs poor treatment response this is likely the minimum time until interferon-free regimens are licensed.
Comparison of treatment regimens
Calculations were based on hypothetical cohorts of 1000 subjects, and were compared using a decision analysis and comparative effectiveness approach.[23-25] These cohorts were assigned to receive treatment according to the regimens defined above. Treatment benefit was estimated using a number needed to treat (NNT) approach. The NNT was calculated using the formula below accounting both for the ARR associated with successful treatment (i.e. the SVR) and also the relative probability of that success.
Since the ARR is expressed per annum, to account for 5-year follow-up, the NNT was divided by 5 to give a 5-year NNT for each outcome.
The rate of clinical events in each cohort was then calculated using the NNT and compared by treatment regimen. The rate of the treatment related serious adverse events of hepatic decompensation (2%) and death (1%)[27-29] were also compared using this approach.
Selected parameters in the base case analysis were varied within plausible limits (Table 2). For instance, rates of ARR for liver-related mortality and HCC varied from 2% to 4%, and 2% to 5%, respectively, in line with recently published data.[21, 22] As relatively few subjects with advanced fibrosis were included in the phase III studies, and those that were had well-preserved liver function, SVR rates varied from 10% to 20% in previous null responders and from 40% to 60% in partial responders. The proportion of subjects not achieving a 1log10 reduction at 4 weeks also varied from 32% to 45% in partial responders and from 55% to 70% in null responders. The relative proportion of null and partial responders included varied from 80/20 to 60/40 splits. Finally, the rate of treatment emergent adverse events varied between 1% and 3% for hepatic decompensation and 0.5% and 2% for treatment-related death.
Mortality reduction after treatment
The effect of antiviral treatment on mortality has been assessed in relatively few studies due to the slowly progressive nature of HCV infection and the consequent use of SVR as a surrogate endpoint. To assess the treatment benefit of protease inhibitor containing triple therapy in previously treated subjects, we calculated the 5-year NNT for two groups: those with any stage of liver disease, and those with advanced fibrosis. For those treated with a protease inhibitor, the 5-year NNT for those with any stage of liver disease is 60 to prevent one liver-related death whereas it is 18 when considering those with advanced fibrosis, despite a reduced SVR (Table 3). The impact of the NNT is highlighted when the number of deaths prevented in each treatment strategy is calculated in hypothetical patient cohorts: more than threefold more deaths are prevented when treating only subjects with advanced disease. These data highlight the importance of considering treatment in this high-risk group and indicate that this group should be prioritised for treatment on the basis of need.
Table 3. Comparison of liver-related mortality benefit in treatment of previously treated subjects with HCV infection stratified by fibrosis stage
HCV deaths prevented per 1000 patient cohort over 5 years
ARR, annual risk reduction; NNT, number needed to treat; SVR, sustained virological response.
Calculated SVR based on estimated proportion of prior null and partial responders as described.
Advanced fibrosis only
Comparative effectiveness of including a stopping rule after the 4-week lead-in
The outcomes of subjects treated with protease inhibitor containing triple therapy are shown in Figure 1. The strategy of treating all patients regardless of response after the 4-week lead-in with pegylated interferon and ribavirin results in the maximal prevention of death (Figure 1a). Employing the virological response to a 4-week lead-in with dual therapy as a stopping rule (Figure 1b) allows treatment to be stopped in the majority of this difficult-to-treat population thus reducing protease inhibitor exposure, and reducing the total number of treatment emergent adverse events. However, there is a notable increase in the efficacy of protease inhibitor containing therapy if treatment is stopped in subjects not achieving a >1log10 reduction in HCV RNA. As those with unfavourable responses are excluded from protease inhibitor containing treatment SVR rates for subjects treated with triple therapy are increased from 37.9% in the treat all strategy to 56.1% when the 4-week stopping rule is applied (Figure 1a,b).
There is however a significant reduction in overall clinical effectiveness associated with using the 4-week stopping rule. The number of patients achieving SVR is reduced by 37% as a result of excluding those subjects who would otherwise have been cured by continued treatment (SVR ‘missed’, Figure 1b). As this is a high-risk group and subjects who are not cured remain at significant risk of liver related morbidity and mortality we calculated the effects of the 4-week stopping rule on these parameters (Table 4). For subjects treated with triple therapy but without a stopping rule at week 4, 42 deaths are prevented over 5-year follow-up. However, only 28 deaths are prevented when the 4-week stopping rule is applied, a reduction of 33%. This indicates that by using the 4-week stopping rule 14 premature deaths occur that would have otherwise have been prevented by treating in the absence of the stopping rule.
Table 4. Effect of treatment emergent adverse events on clinical outcomes after antiviral treatment in previously treated HCV infected subjects with advanced fibrosis
HCC, hepatocellular carcinoma; HCV, hepatitis C virus.
Calculations are based on a hypothetical 1000 subject cohort with 5-year follow-up after antiviral treatment.
Negative values indicate events that are reduced when the stopping rule is applied.
Prevented HCV death
On treatment death
Prevented HCV decompensation
On treatment decompensation
On treatment HCC
The impact of treatment emergent adverse effects on outcome
This difficult to treat population is at risk of treatment emergent adverse effects that can negatively impact on the outcome of treatment. In similar studies of dual therapy hepatic decompensation occurs in approximately 2% of those treated and death occurs in approximately 1%.[28, 29] These data are supported by the initial safety data reported in the early expanded access to protease inhibitor programmes in France. The impact on these episodes is summarised in Table 4. For instance, treatment emergent hepatic decompensation (i.e. decompensation precipitated by treatment) decreases overall treatment effectiveness. In effect there is an approximately 40% reduction in the net number of decopmensation episodes prevented by treatment in the treat-all strategy. When the 4-week stopping rule is employed there is a reduction in treatment-related decompensation episodes, but the advantage of treating all subjects regardless of response at week 4 of treatment remains. The impact on death is similar and these data indicate that the development of treatment emergent adverse effects does not significantly reduce the comparative benefit that is achieved by treating subjects regardless of virological response at treatment week 4.
In sensitivity analyses, the overall benefit of continued treatment despite poor virological response at treatment week 4 was maintained despite variation in the proportion of null and partial responders, low SVR rates to protease inhibitor containing treatment, and also in cohorts containing an increased frequency of subjects not achieving >1log10 reductions at treatment week 4. However, in scenarios where there was greater mortality associated with treatment, and particularly where this was associated with low treatment efficacy, there was no difference in overall mortality between those treated regardless of virological response and those treated according to the 4-week stopping rule (Table 5). These analyses support not using the 4-week lead-in as a universal stopping rule but rather individualising treatment where subjects at high risk of on-treatment mortality due to the presence of advanced cirrhosis and portal hypertension[28, 29] should have interferon sensitivity considered in treatment planning.
Table 5. Outcomes of sensitivity analyses
Net deaths prevented by treatment strategy
Deaths incurred by using 4-week stopping rule
Stop if <1log10 reduction after 4 weeks
SVR, sustained virological response.
Calculations are based on a hypothetical 1000 subject cohort with 5-year follow-up after antiviral treatment. Outcomes presented here are based on the extremes of the sensitivity parameters provided in Table 2.
High ARR death
Low ARR death
Low SVR rates
Increased prior null response
Increased poor interferon response
High on-treatment mortality
Low SVR AND high on-treatment mortality
The development of novel DAA agents targeting HCV has the potential to significantly improve outcomes for subjects with HCV infection.[5, 6] There are however a number of questions regarding how, and in whom, these agents should be used.[13, 14, 31] This analysis provides a clear rationale for prioritising subjects with advanced fibrosis who are at significant risk of liver-related morbidity and mortality for treatment. This group represents at least 25% of infected subjects at current estimates and targeted treatment offers the potential to reduce liver-related mortality in a relevant timeframe. Our modelling cautions against the use of the 4-week lead-in phase as a universal decision point in subjects with advanced fibrosis as this strategy will disadvantage as many as 37% of subjects who would otherwise ultimately have derived benefit from continued treatment.
Recent licensing of DAA agents has brought further complexity to the management of genotype 1 HCV infection. In addition, the rapid development of new agents has led to uncertainty regarding which subjects to treat now, and which patients might be deferred from treatment. Using the analyses presented here we have illuminated this difficult topic, particularly highlighting the benefits of treating subjects with advanced fibrosis. This population is in need of effective treatment and the benefit of this analysis is that it clarifies the clinical outcome that results as a consequence of treatment vs. no treatment in a clinically relevant timeframe. There is clinical heterogeneity in a group containing all subjects with advanced fibrosis (METAVIR F3/F4), however, the data included in the model are supported by clinical data drawn directly from this population. For instance, treatment responses are from randomised controlled data, and long-term follow-up data are from a meta-analysis of more than 1400 subjects.[9, 21] Using these analyses to support clinical decision making and the prioritisation of subjects for treatment should allow clinicians to maximise treatment benefit through mortality reduction in the next 5 years despite relatively low SVR rates and whilst further developments in therapy occur. Furthermore, the analyses provide critical information in the explanation of the potential benefits (and risks) of treatment for clinicians to share with subjects with advanced fibrosis who are considering treatment.
The rationale for using the 4-week lead-in phase for decision making has been based on concerns regarding increased toxicity in subjects with advanced disease, and also concerns about the evolution and persistence of drug resistant variants.[11-13] The registration studies for both boceprevir and telaprevir included patients with advanced fibrosis, albeit at a relatively low frequency, and these patients had well compensated liver disease. In subjects treated with pegylated interferon and ribavirin dual therapy there is evidence that the treatment emergent adverse effects of hepatic decompensation and death are more common in those with advanced cirrhosis and portal hypertension.[28, 29] Our analysis indicates that allowing for treatment-associated decompensation at a rate of 1 in 50 treated patients and mortality at 1 in 100 (in keeping with reports from early access programs), there remains a benefit in continuing treatment regardless of treatment response at treatment week 4. The development of resistant variants is more difficult to model since the consequences of these frequently short-lived variants are uncertain. Whilst there is the potential for these variants to persist and to impact on future protease inhibitor containing treatment and candidacy for future clinical trials of perhaps more effective therapies, the likelihood is that for many of the subjects included in this analysis this would be the last opportunity for treatment (prior to liver transplantation or death). This is especially apparent as it is interferon sensitivity that governs the poor responses to treatment in this group and that pegylated interferon is likely to remain a component of standard of care therapy for the next 5 years at least.
This analysis is limited by several factors. The data are extracted from several studies, none of which were intended to address this particular question. Furthermore, none of the studies contained large numbers of patients with advanced fibrosis and indeed the boceprevir development programme did not address treatment responses in prior null responders. The treatment responses for boceprevir and telaprevir appear broadly similar and we therefore considered a prototypic protease inhibitor based on the published data that were available. In sensitivity analyses, we varied the proportions of prior null and partial responders, the SVR rates and the rates of treatment emergent adverse effects to address the areas where there were uncertainty in the primary reports. These analyses indicated that there was benefit of continued treatment regardless of virological response at treatment week 4. There were however important exceptions to this, and that is in scenarios where there are high rates of treatment emergent mortality. In these situation there was no benefit to continued treatment following a <1log10 reduction in HCV RNA when considering 5-year follow-up. These scenarios might be represented by subjects with advanced cirrhosis and significant portal hypertension where treatment is already risky and where a failure to respond to the 4-week lead-in could be considered an indication to stop treatment. The analyses presented here therefore provide important evidence to support individualised treatment decision making in this difficult population. Further studies addressing this population should be carefully designed to ensure that stopping rules are implemented both for virological failure and to protect those included from excessive treatment related morbidity and mortality.
This analysis contains a relatively crude assessment of benefit and harm. Whilst this is based on important clinical outcomes of death, hepatic decompensation and the development of HCC rather than the surrogate measure of SVR it is likely that the lasting benefits of SVR are underestimated. Whilst several investigators have studied the effects of SVR on more relevant outcomes, such as serum biochemistry and fibrosis stage,[35, 36] the overall clinical outcomes of SVR are seldom reported. We therefore utilised a meta-analysis of clinical outcomes after SVR. A more recent prospective report suggests that both the annual risk of liver-related mortality, and the ARR in mortality might be greater than we considered suggesting that the mortality reductions in the base case analysis might underestimate the true effect of treatment. Furthermore considering follow-up of only 5 years duration will underestimate the overall benefit of treatment. This parameter was however set at 5 years to permit decisions to be made in this clinically relevant timeframe. Additional data indicating that productivity is decreased and absenteeism is increased in HCV infected subjects suggests significant additional benefit from successful treatment.[37, 38] Although these outcomes are difficult to model the benefit we have highlighted in continuing treatment regardless of response at treatment week 4 is likely increased when all factors are considered. Of course, additional treatment emergent adverse events will also be incurred through continued treatment of this population regardless of virological response at week 4, and some of these will be severe. In those subjects with advanced disease these are likely to be more frequent, and to contribute to treatment emergent decompensation and death. These events, however, do not negate the benefit of treatment. A decision to wait for further treatment developments in this high-risk group therefore has the potential to cause the greatest harm when compared with any of the treatment strategies presented here and the mortality reduction noted with treatment strongly cautions against that approach.
These analyses describe previously treated subjects with prior null and partial responses. As the absolute risk reduction in mortality associated with successful treatment in the untreated population is likely to be similar these findings are generalisable to all subjects with advanced fibrosis undergoing treatment with protease inhibitor containing treatment. The findings are however not applicable to subjects with early fibrosis where the harms of treatment manifest through treatment emergent adverse effects, or the development of resistant variants, may outweigh the benefits of continued treatment in poor interferon responders. Whilst information on the significance of resistance is emerging there are concerns that the harms relating to resistant variants may be significant and individualised decision making is appropriate until such data are available.
In summary, this analysis indicates that subjects with advanced disease should be prioritised on the basis of need. Furthermore, when considering protease inhibitor treatment of previously treated subjects with advanced fibrosis, this should be done without consideration of interferon responsiveness following the 4-week lead-in. This analysis provides critical information regarding both selection and on-treatment decision making for previously treated subjects that should be included in guidance for physicians using protease inhibitors to treat subjects with HCV infection.
The authors thank Dr Matthew Armstrong and Dr Richard Parker for critical reading of the manuscript. Declaration of personal interests: David Mutimer has served as an advisory board member for Janssen and MSD. Ian Rowe has served as an advisory board member for Janssen. Declaration of funding interests: Ian Rowe and Diarmaid Houlihan are supported by Medical Research Council Clinical Research Training Fellowships.