Variable short duration treatment versus standard treatment, with and without adjunctive ribavirin, for chronic hepatitis C: the STOP-HCV-1 non-inferiority, factorial RCT

Article history

The research reported in this issue of the journal was funded by the EME programme as project number 14/02/17. The contractual start date was in February 2016. The final report began editorial review in February 2020 and was accepted for publication in June 2021. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The EME editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2021 Cooke et al. This work was produced by Cooke et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2021 Cooke et al.

Background

Hepatitis C virus (HCV) is a major challenge in the UK both for the individual with the virus and for public health. In 2013, an estimated 215,000–265,000 individuals were living with HCV infection in the UK. 1,2 Those who are chronically infected are at risk of severe liver diseases (e.g. cirrhosis, liver failure and hepatocellular carcinoma). Progression to end-stage liver disease is more rapid in those who have other medical conditions, particularly co-infection with human immunodeficiency virus (HIV). Treatment of infected individuals has the additional potential to reduce ongoing transmission through needle use and sexual contact, and from mother to child.

Morbidity and mortality from HCV have been an increasing challenge to the NHS. In 2013, health-care costs related to HCV were estimated to be £82.7M per year and productivity losses £184M–367M per year. 3 HCV-related hospital admissions rose from 612 in 1998 to 2268 in 2011. HCV-related deaths rose from 98 in 1996 to 381 in 2011. The proportion of liver transplants undertaken because of HCV rose steadily from 10% in 1996 to 18% in 2011.

Viral genotype remains a key factor in determining the preferred treatment options for HCV in the NHS. Globally, genotype 1 is the most common, accounting for approximately 46% of all infections, which is very similar to the estimated rate in the UK. 4 As the most common genotype in most well-resourced health economies, particularly the USA, genotype 1 was the greatest focus of the initial development of new oral drugs. However, other genotypes also make a substantial contribution to the HCV burden in the UK.

Curative treatments have been available for HCV for some time. However, in early 2015, standard treatment for HCV infection still involved long courses (i.e. 24–48 weeks) of relatively toxic therapy (pegylated interferon alpha plus ribavirin), with a modest chance of cure (40–50%). The nature of therapy remained a major barrier to the uptake of treatment and, hence, control of the epidemic. Since 2015, a new generation of well-tolerated oral direct-acting antivirals (DAAs) has transformed HCV treatment, with the potential to cure HCV in most patients after 8–12 weeks of therapy. All HCV-infected adults with mild disease could, in theory, be cured with these regimens, substantially reducing future morbidity and mortality. For genotype 1, the first two interferon-free combination regimens approved in the NHS were:

1. a ritonavir-boosted triple combination of paritaprevir/ritonavir, ombitasvir (Viekirax^®; AbbVie, Chicago, IL, USA) and dasabuvir (Exviera^®; AbbVie, Chicago, IL, USA)

2. sofosbuvir and ledipasvir (Harvoni^®; Gilead Sciences, Inc., Foster City, CA, USA).

The ritonavir-boosted combination of paritaprevir, ritonavir and ombitasvir (without dasabuvir) has also been approved for genotype 4 infection. Other regimens that are active against more, or even all, viral genotypes have also been approved or have been submitted for approval.

The new treatments for HCV offer the potential for curative therapy for the individual and the opportunity to break transmission pathways, leading to the real possibility of eliminating the HCV epidemic in the UK. A recent systematic review showed a clear benefit of HCV cure in improving health outcomes across a range of clinical settings,5 and there is no evidence to suggest that this differs according to the means used to achieve cure. However, initial costs for treatment were very high, at approximately £3000 per week, placing strain on limited health budgets. 6 Beyond costs, licensed durations of therapy with DAAs, at 8–12 weeks, although significantly shorter, and therefore more tolerable than those of previous interferon-based therapies7, remain challenging for many patients with HCV, whose chaotic lifestyles are a barrier to adhering to treatment. Shorter courses of treatment would potentially increase access to treatment for difficult-to-reach groups and could have an impact on onward transmission.

The rapid development of treatment options for HCV has led to an ambitious World Health Organization strategy to eliminate viral hepatitis as a global public health threat by 2030. 8 This includes a target of treating 80% of those chronically infected with HCV, many of whom will be patients who will still find it challenging to complete a full treatment course of 8–12 weeks. Such patients will become an increasingly important part of clinical practice as treatment coverage expands to reach marginalised groups.

Shorter treatment courses of licensed therapies are one clear mechanism to increase coverage into these groups, including those with active illicit drug use, as they should increase adherence. 9,10 Shortened courses of licensed therapy may have sufficiently high efficacy in acute or recent HCV infection for them to be recommended routinely. 11 However, limited data are available in chronic infection to identify which patients might be able to achieve high cure rates with shorter durations of therapy. In particular, in two small Phase II studies, short-duration treatment in all patients, without risk stratification, achieved cure rates of only 20–40% after a fixed 4-week treatment course and of 57–95% after a fixed 6-week treatment course. 12,13 Furthermore, few of the combinations or durations that have been trialled have subsequently been licensed for use. In the case of licensed therapies, recommendations to shorten therapy to 8 weeks (rather than 12 weeks) are based on baseline viral load (VL) (< 6,000,000 IU/ml)14 and subgenotype,15 but no criteria for recommending < 8 weeks’ therapy have been tested or validated in those with chronic infection. Nevertheless, the very high cure rates (i.e. > 95%) on standard 8- to 12-week treatment courses make it clear that many patients are being prescribed much more medication than they require to be cured, resulting in unnecessary inconvenience and costs.

When a clinician initiates treatment in the knowledge that there is a high risk that the patient may not complete therapy, or aims to use an ultrashort course of therapy for the same reason, an important concern is that virological failure may be accompanied by emerging resistance, which could then compromise future treatment options. However, less resistance could also theoretically emerge with shorter courses of treatment. Ribavirin, a generically available guanosine analogue, was widely used to increase cure rates with previous pegylated interferon-based therapies, and there is some evidence that it may improve rates of virological cure with shorter treatment courses. 16 There is also the possibility that it might reduce the rate at which resistance emerges in those failing treatment when added to short-course therapy. 17 However, these hypotheses have not been tested in a randomised trial.

Rationale

The very high cure rates achieved with 12- and 8-week DAA regimens raise the question ‘What is the minimum duration of treatment that can achieve cure in the majority of patients?’. 18 As above, minimising (effective) treatment duration is important for ensuring the widest and most equitable access to curative therapy across all patients (particularly those who will struggle to take medicine and are likely to require support) for the same fixed budget, and for minimising toxicity. However, as observed with previous interferon-based therapies, it is likely that response to DAA treatment will depend on individual-level characteristics, offering the opportunity to stratify short-course treatment. The best-studied biomarker for stratifying treatment duration is plasma HCV ribonucleic acid (RNA) VL. Based on mean baseline VL levels and decline from VL baseline levels in studies to date, and assuming, at most, a modest negative correlation between initial values and rates of decline, a ‘sliding scale’ of 4–7 weeks’ combination DAA treatment (where the precise duration of treatment depends on the individual’s pre-therapy HCV VL) should reduce virus levels to < 1 copy in the whole person. The question addressed in this trial is whether or not such an HCV VL-stratified DAA duration (13–50% shorter, i.e. 50–83% of the original length) followed by retreatment for those failing initial treatment gives similar cure rates to longer fixed-duration (8-week) therapy followed by retreatment in individuals with mild chronic HCV disease. Biomarker-stratified variable-duration ultrashort-course treatment would enable more patients to be cured within the same overall budget, and would also have benefits in terms of less potential toxicity and regimens that are easier to adhere to for patients. This is particularly important for HCV, as a substantial minority of those infected come from disadvantaged populations (e.g. drug users, homeless persons and prisoners).

In addition, although ribavirin was an essential component of previous interferon-based treatments, its role in DAA regimens is less clear, potentially providing minimal additional benefit when added to more potent regimens. However, it is cheap and has less toxicity when given for short duration, and modest benefits could allow shorter DAA regimens to be used more effectively. Therefore, the trial also tested whether or not the addition of ribavirin is beneficial in short-course treatment, using a partial factorial design in those randomised to therapy that is shorter than the full licensed duration. First-line treatment choice was in line with recommended NHS options.

Objectives

The overarching aim is to evaluate the efficacy of biomarker-stratified treatment of HCV infection and of adjunctive ribavirin with combination DAAs. This would allow the identification of patients with minimal fibrosis and chronic HCV infection who can be offered a high probability of cure with shortened courses of interferon-free all-oral DAA regimens. Such stratification will reduce the cost per cure and improve access for those unable to adhere to 8–12 weeks of treatment.

The primary objectives of the STOP-HCV-1 trial were to test:

• whether or not a biomarker-stratified short-course first-line treatment (with variable duration of between 4 and 7 weeks determined by patient baseline VL) followed by 12 weeks of retreatment for those failing therapy is non-inferior to a fixed-duration, 8-week first-line treatment followed by 12 weeks of retreatment for those failing therapy, in terms of overall HCV cure in patients with minimal fibrosis and chronic genotype 1 or 4 HCV infection

• the benefits and risks of adding adjunctive ribavirin to 4–8 weeks’ first-line therapy for HCV infection.

The different strategies above will be tested using ombitasvir/paritaprevir/dasabuvir/ritonavir (genotype 1), ombitasvir/paritaprevir/ritonavir (genotype 4) or glecaprevir/pibrentasvir (genotypes 1 and 4) as first-line treatment, and sofosbuvir/ledipasvir/ribavirin as retreatment (genotypes 1 and 4).

The secondary objectives were as follows:

• To test whether or not 4–7 weeks’ first-line biomarker-stratified treatment is non-inferior to 8 weeks’ fixed-duration first-line treatment in mild HCV infection (i.e. excluding retreatment responses).

• To test whether or not retreatment with 12 weeks of an alternative combination regimen, given after detecting virological failure on first-line treatment, still achieves cure in the majority of the small proportion of patients failing short-course first-line treatment.

• To explore whether or not factors other than baseline HCV VL influence and, therefore, could better predict the response to (1) short-course DAA treatment and (2) retreatment. Factors explored will include viral factors (such as minority resistance variants and viral diversity, and initial virological response), host factors {such as age, body mass index (BMI) and human genetic variation [notably, interleukin (IL) 28 polymorphisms]} and immune factors (such as immune phenotyping before and after treatment initiation). Mechanistic work will be embedded in the Medical Research Council (MRC) HCV Stratified Medicine Consortium (to be reported separately).

• To validate the performance of a novel point-of-care device for detecting IL-28B polymorphism (to be reported separately).

Trial design

The STOP-HCV-1 trial was an open-label randomised controlled trial that tested biomarker-stratified short-course first-line and retreatment DAA oral treatment regimens to cure mild chronic HCV disease. Patients were allocated 1 : 1 using a factorial design to each of:

• open-label variable ultrashort treatment compared with fixed-duration first-line treatment (1 : 1)

• open-label adjunctive ribavirin or not (1 : 1). (Note that patients receiving glecaprevir/pibrentasvir and randomised to fixed 8 weeks’ first-line treatment were excluded.)

All patients received first-line ombitasvir/paritaprevir/(dasabuvir)/ritonavir or glecaprevir/pibrentasvir (based on genotype and local availability of the different regimens) and sofosbuvir/ledipasvir/ribavirin retreatment as necessary.

Participants

All participants met the trial-specified inclusion and exclusion criteria detailed below. Written informed consent was obtained from all participants after they received an explanation of the aims, methods, benefits and potential hazards of the trial and before any trial-specific procedures were performed or any blood was taken for the trial.

Trial setting

Participants were recruited from 14 UK NHS hospital trusts:

1. Singleton Hospital, Swansea Bay University Health Board

2. University Hospitals of Leicester NHS Trust

3. Imperial College NHS Trust

4. St George’s Healthcare NHS Foundation Trust

5. Royal Free London NHS Foundation Trust

6. Nottingham University Hospitals NHS Trust

7. Royal Surrey County Hospital NHS Foundation Trust

8. Brighton and Sussex University Hospitals NHS Trust

9. John Radcliffe Hospital, Oxford University Hospitals NHS Foundation Trust

10. Glasgow Royal Infirmary, NHS Greater Glasgow and Clyde

11. Newcastle Freeman Hospital, Newcastle Upon Tyne Hospitals NHS Foundation Trust

12. Chelsea and Westminster Hospital NHS Foundation Trust

13. Central and North West London NHS Foundation Trust

14. Sheffield Teaching Hospitals NHS Foundation Trust.

The main criterion for selecting participating hospitals was that they had the potential to recruit the required number of chronic (> 6 months) HCV genotype 1a/1b- and 4-infected participants within the agreed recruitment period. This was established by the use of a trial-specific site survey. Sites also needed to meet the following criteria:

• no competing studies that would have an impact on the ability to enrol quickly to the trial

• turnaround of no more than 7 days for HCV VL test results

• ability to provide 24-hour cover for trial patients

• local governance approval likely to take < 3 months.

The overall trial design is summarised in Figure 1.

FIGURE 1.

Trial schema. Note that, as above, the ribavirin randomisation was a partial factorial in those randomised to a course shorter than the full licensed duration of therapy (i.e. the vast majority of patients recruited to the trial; see Chapter 3).

Patient and public involvement

The trial was developed with the Hepatitis C Trust (London, UK) and, in particular, Rachel Halford (who succeeded Charles Gore), who was one of two patient and public involvement representatives on the Trial Steering Committee. The Hepatitis C Trust advised on the design of the interventions, in particular the determination of the duration of variable-course therapy by baseline HCV VL, the acceptability of the follow-up schedule and assessments and the information provided to patients. The Hepatitis C Trust is helping to disseminate the trial’s results beyond the academic and health-care professional community to other patient groups.

Trial intervention: duration of treatment

All patients were randomised to VUS (intervention) or fixed 8-week (control) initial treatment. In protocol versions 1.0–4.0 inclusive, the intervention duration was between 4 and 7 weeks’ first-line treatment, on a sliding scale determined by the screening HCV VL. The proposed stratification rule was determined from the mean and standard deviation (SD) baseline VL, and the mean estimated declines, from previous trials (mean screening VL ≈ 6.25 log₁₀ IU/ml, SD 0.4 log₁₀ IU/ml; mean estimated decline 2.15 log₁₀ IU/ml per week). Together, these could be used to estimate the duration of treatment needed to reduce levels to ≈ 1 copy in the whole body at end of treatment (< 0.0001 IU/ml), including a conservative assumption of a moderate negative correlation between baseline and decline in VL, as no data are available on this parameter.

This biomarker-stratified treatment duration was implemented as a specific number of days of first-line treatment based on the screening VL, as shown in Table 1. (Note that the declines are linear on a log-scale and so the absolute value in IU/ml does not increase linearly across the categories in this table.) Based on recent trials, it was expected that ≈ 15% of recruited patients (with screening HCV VL of < 10,000,000 IU/ml) would receive the minimum treatment and ≈ 5% of the maximum treatment.

Protocol versions 1.0–4.0 prespecified that the biomarker-stratified duration would be adapted if the upper limit of the 99.9% confidence interval (CI) for the risk difference between variable and fixed duration was < 65%. The adaptation criterion was met at the second meeting of the Data Monitoring Committee (DMC) and the decision was taken by the DMC to change the potential DAA treatment length for variable-duration participants from 4–6 weeks (VUS1) to 4–7 weeks (VUS2), implemented in protocol version 5.0, and illustrated as the solid blue line on Figure 2. An alternative ‘cut-off point’ (shown as dashed blue line in Figure 2) would require a single threshold HCV VL to be chosen and reflects biological variation less well. All participants randomised from 1 April 2017 were treated under VUS2.

FIGURE 2.

Biomarker-stratified variable ultrashort first-line treatment duration from protocol version 5.0 onwards (VUS2).

All patients were prescribed the first 4 weeks of first-line therapy at randomisation and the remaining first-line treatment (as per their randomised group) was provided at the week 4 visit. All patients were offered an optional patient diary card personalised with their specific combination regimen [tablets once daily (OD)/bis in die (twice a day) (b.i.d.)] and treatment duration to help them record pill taking. Any doses missed during the treatment course were to be taken at the end of the prescribed course.

Trial intervention: drug regimens

The trial allowed three possible first-line drug combinations with which participants could be treated, depending on their genotype and local availability:

1. Viekirax (ombitasvir/paritaprevir/ritonavir) and Exviera (dasabuvir) for genotype 1a/1b

2. Viekirax (ombitasvir/paritaprevir/ritonavir) for genotype 4

3. Maviret^® (AbbVie, Chicago, IL, USA) (glecaprevir/pibrentasvir) for genotypes 1a/1b and 4.

The licensed duration of Viekirax, with or without Exviera, was a 12-week treatment course and so the fixed-duration arm represented a shorter than standard course. The licensed duration of Maviret was an 8-week treatment course and so the fixed-duration arm represented the standard course. Viekirax without Exviera was added in protocol amendment 5 and Maviret in protocol amendment 6.0 (see Appendix 2). In practice, very few patients in the trial received these regimens (see Chapter 3).

With all three possible first-line treatments, participants randomised to the VUS arm were also randomised to receive or not receive ribavirin. In the case of participants randomised to the 8-week fixed-duration arm, those taking Viekirax, with or without Exviera, were also randomised to receive or not to receive ribavirin. Participants taking Maviret who were randomised to 8 weeks’ fixed-duration treatment did not receive ribavirin because this 8-week course was already the licensed duration.

Participants in whom first-line treatment failed were offered 12 weeks’ Harvoni (sofosbuvir and ledipasvir) with ribavirin, regardless of their initial DAA regimen.

Randomisation

Randomisation was performed via a computer-generated program at the STOP-HCV-1 Co-ordinating Centre [MRC Clinical Trials Unit (CTU) at University College London (UCL), London, UK]. Patients were allocated 1 : 1 using a factorial design to each of:

• biomarker-stratified VUS compared with fixed-duration treatment

• adjunctive ribavirin or not (a partial factorial in those randomised to a shorter course than the full licensed duration of therapy).

Randomisation was stratified by study centre, HCV genotype and study drug regimen using a minimisation algorithm, incorporating a probabilistic element securely into the online trial database. Randomisation determined the duration of first-line therapy rather than the choice of DAAs, which was prespecified by the investigator before randomisation based on local availability.

Allocation concealment mechanism

Each allocation was generated within the trial database only at the point of randomisation and after it was confirmed that the participant was eligible and was to be randomised. Allocations were generated using minimisation with a probabilistic element and so there was no predetermined allocation sequence to conceal. To further conceal the potential allocation, study centres were not informed of the randomisation strata.

Implementation

On the day of randomisation, participant eligibility was checked at sites and the data confirming eligibility were entered onto a case report form and sent to MRC CTU. The data were entered into the database at MRC CTU and, again, checked for eligibility. Once the participant had been confirmed as eligible, the database would perform randomisation using the computer-generated program. Sites were then informed of the allocation and length of DAA treatment required for the participant.

Blinding

All randomisations were open label and, therefore, there were no unblinding procedures. It would have been infeasible to blind the durations of five different drugs for variable durations from 4 to 8 weeks. Given the lack of blinding, the primary outcome was based on an objective laboratory biomarker (HCV VL).

Assessment and follow-up

All participants were followed by the site teams for 24 weeks after the end of first-line treatment or retreatment (where applicable) for evaluation of virological response and toxicity. Participants on first-line therapy had clinical assessments on days 3, 7, 10, 14 and 28 and at the end of therapy (EOT) (where EOT was not day 28), followed by weeks 2, 4, 8, 12 and 24 after EOT. All outcome measures below were assessed at these clinic visits. All patients failing treatment were retreated as soon as practicable after failure was identified. The schedule of assessment is in Appendix 3.

The trial closed in August 2018, after which time no further recruitment was possible.

Outcomes

Sample size

The trial was originally powered to demonstrate:

• non-inferiority of biomarker-stratified variable ultrashort 4- to 7-week first-line treatment followed by 12 weeks’ retreatment compared with fixed 8-week first-line treatment with the same retreatment

• superiority of adjunctive ribavirin in first-line treatment.

The primary end point for the non-inferiority comparison is overall SVR12 after first-line treatment and retreatment (where necessary), which was estimated at 98% for the control group, regardless of first-line combination, given the very high cure rates achieved with the 12-week ribavirin-containing regimens that will be used for retreatment and the limited impact of prior DAAs treatment on response to subsequent regimens.

As an example, from previous trials,23 we can assume 96% and 84% SVR12 with 8 weeks’ ombitasvir/paritaprevir/dasabuvir/ritonavir in patients with genotypes 1b and 1a, respectively. With a 1 : 2 ratio of presenting cases (reflecting UK prevalence24), the cure rate in the control first-line group would be 88% prior to retreatment. Conservatively assuming a cure rate of retreatment of 85% to allow for potential role of mutations, particularly in the NS5A gene, would lead to an overall 98% cure rate in the control group. However, similar overall 98% cure rates could be achieved with lower first-line treatment and higher retreatment cure rates (e.g. 65% first-line treatment and 94% retreatment) or higher first-line treatment and lower retreatment cure rates (e.g. 95% first-line treatment and 60% retreatment). Although first-line cure rates may be slightly lower or higher with 8 weeks of different first-line combinations (in particular, first-line cure rates might be expected to be higher with 8 weeks’ glecaprevir/pibrentasvir, as this is the licensed indication, albeit without much real-world experience to date), in practice, it is unlikely that an overall cure rate of 98% from first-line treatment plus retreatment can be exceeded, and so the control group rate of 98% is reasonable across different first-line regimens.

Assuming a 98% cure rate overall for the control group, for a 4% non-inferiority margin, 80% power, a one-sided test and an alpha of 0.025, the required sample size for the biomarker-stratified duration comparison is 408 patients, allowing for 5% early withdrawal.

The 4% non-inferiority margin is arbitrary, but ensures that overall cure rates in the biomarker-stratified short-course group would be well over 90% if the trial were to declare non-inferiority. Furthermore, even small genuine reductions in overall cure rate with short-course treatment substantially decrease the trial’s power to demonstrate non-inferiority [i.e. a reduction of 52%, 25% and 11% of short-course treatment (from 80%) genuinely achieves overall cure rates that are 1%, 2% or 3% lower, respectively].

If non-inferiority is not demonstrated, a total of 408 patients is likely to provide reasonable power to investigate other predictors of cure, such as presence of viral quasispecies, including resistance, age (related to immune health), IL-28 polymorphisms and BMI.

The calculation of sample size for the fixed and duration non-inferiority comparison is conducted under the null hypothesis for the ribavirin superiority comparison (i.e. no effect). Given its partial factorial nature, estimates of the effect of adjunctive ribavirin are determined from generalised linear models, which include terms to reflect the randomisations and the specific first-line DAA regimen received. Therefore, patients randomised to 8 weeks’ glecaprevir/pibrentasvir effectively do not contribute to this comparison. When protocol version 6.0 was approved, we estimated that this would be approximately 25% of patients (n = 102). In practice, only two patients received this regimen in the trial (see Chapter 3). A total of 306 patients randomised to adjunctive ribavirin (or not) provides 75–85% power to identify a 10% improvement in first-line cure rate associated with adjunctive ribavirin for first-line cure rates of 83–86% without ribavirin and of 93–96% with ribavirin (two-sided alpha = 0.05), and > 80% power to identify a 15% improvement in first-line cure rate associated with adjunctive ribavirin for first-line cure rates of 60–80%, allowing for 5% early withdrawal, as above.

Interim monitoring and analyses

The protocol prespecified that the DMC would meet approximately 6-monthly, and four 6-monthly meetings took place. The protocol prespecified an adaptation in the case that the upper limit of the 99.9% CI for the risk difference between variable and fixed duration was < 65%. The adaptation criterion was met at the second DMC meeting on 10 April 2017 and the decision was taken by the DMC to change the potential DAA treatment length for variable-duration participants from 4–6 weeks (VUS1) to 4–7 weeks (VUS2).

Statistical methods

Analyses followed the principle of intention to treat, including all follow-up, regardless of changes to treatment. The statistical analysis plan (SAP) prespecified that any patient who was randomised in error (defined as the realisation that the patient should not have been randomised before taking study drug and not ever taking study drug) and, hence, not followed up would be excluded.

A patient was formally considered as lost to follow-up (LTFU) if they had not been seen at the final EOT plus 24 weeks visit within a –6- to +12-week window. If a patient was LTFU before the visit at which an outcome was measured (EOT plus 12 weeks for SVR12 and EOT plus 24 weeks for SVR24), then the following methods were prespecified in the SAP, but not in the protocol, to be used to determine the patient’s outcome:

• If the missing HCV VL was between two undetectable measurements, then it was assumed to be undetectable.

• HCV VL results from local practice were sought for patients with missing SVR12 and SVR24 outcomes. If the local result was undetectable, then the patient was assumed to be undetectable at EOT plus 12 weeks/EOT plus 24 weeks. (One patient had a detectable local VL, but as they had no confirmatory subsequent VL they were not considered to have failed and were also not counted as a cure.)

After following these methods, the percentage of patients without an outcome was < 10% and so the analysis was restricted to complete cases (as prespecified in the SAP).

Primary analyses of outcomes restricted to first-line therapy were stratified by first-line DAA strategy in place [VUS1 (before 1 April 2017) or VUS2 (after 1 April 2017)] as a main effect, and as an interaction with randomised group (fixed duration vs. variable duration), where the p-value for the interaction term was < 0.05. For analyses of SVR12 after first-line therapy only, analysis was also performed separately within the VUS1/VUS2 strata. Primary analyses of outcomes, including retreatment, were unstratified, reflecting the overall strategy comparison and because no patients failed after receiving retreatment. Primary analyses were not stratified by centre, given the large number of centres with small numbers of patients recruited.

Primary analysis of the primary end point included all randomised participants other than those considered randomised in error (following the SAP) and for whom no VL data could be obtained. A per-protocol analysis (prespecified in the protocol) included patients receiving > 90% and < 110% of the prescribed duration of first-line treatment and where the difference between screening and enrolment HCV VL values would have led to a difference of ≤ 2 days in allocated duration of DAAs had they been allocated to the variable-duration group. Secondary analyses were conducted considering all LTFU patients as failures and all LTFU patients as cured. An additional secondary analysis excluding reinfections identified by genome sequencing, which was prespecified in the protocol, was not performed as no reinfections were identified.

For the primary analysis, a risk difference and 95% CI were obtained from a binomial regression on the risk difference scale using a generalised linear model. Kaplan–Meier plots and Cox proportional hazard models were used for analyses of time until failure (any type). Secondary analyses of primary treatment failure and VL rebound (i.e. the components of overall treatment failure) used competing risks methods (e.g. cumulative incidence plots and subhazard ratios) to account for the possibility that the patient would experience the other type of failure. Binomial generalised estimating equations with an independent working correlation were used to analyse the percentage of patients with undetectable HCV VL at each time point.

Safety outcomes were analysed using chi-squared p-values, and Cox proportional hazard models were also used. To assess the change in laboratory values over time (other than for HCV VL), generalised estimating equations (normal distribution) with an independent correlation structure adjusted for baseline values were used. Sensitivity analyses of changes in laboratory values used alternative error structures and mixed-effects models, but these provided results similar to those of the primary analysis.

Baseline values of laboratory test results were those taken closest to randomisation. No laboratory test results taken after randomisation were used for baseline. HCV VL was log₁₀ transformed for analysis as a continuous variable. Other continuous measures were transformed using Box–Cox transformations when there were gross (p < 0.0001) deviations from normality, as assessed using the Shapiro–Wilk test. Analyses of measurements at a given point in follow-up used the closest available measurement to that time point in evenly spaced windows. If a visit fell in two visit windows, then it was classed as belonging to the later window, except where this led to no visits within in the first window and two within the second, in which case it was classed as belonging to the first visit window.

Subgroup analyses were conducted to assess the consistency of effects across different participant characteristics. All subgroup analyses were adjusted for the interaction between VUS strategy and duration randomisation because it was highly significant in the primary analysis. For the duration comparison, interaction tests within binomial models on the risk difference scale were used for subgroup analyses. For the ribavirin comparison, owing to non-convergence of the models, p-values were obtained from marginal effects after logistic regression for subgroups. Heterogeneity p-values for IL-28B polymorphisms considering CC/CT/TT genotype as an ordinal factor were obtained from ordered logistic regression. Continuous factors were categorised into terciles, as well as using fractional polynomial models. Heterogeneity p-values could not be estimated for all subgroups because of small numbers or perfect prediction. No formal adjustment for multiple testing was made for subgroup analyses.

Protocol changes

The trial was approved by Cambridge South Research Ethics Committee (reference 15/EE/0435) and the Medicines and Healthcare products Regulatory Agency. See Appendix 2 for changes to the protocol.

Recruitment and participant allocation

Recruitment opened on 17 March 2016, with the first participant randomised on 18 March 2016, and closed on 31 August 2018, with the last participant randomised on 28 August 2018. In total, 217 individuals were screened for entry to the trial and 204 participants were randomised to variable ultrashort treatment of 4–7 weeks (n = 102) or DAA treatment for a fixed duration of 8 weeks (n = 102), and to receive adjunctive ribavirin (n = 101) or not (n = 103) (Figure 3). The most common reason for not randomising a participant was that their screening HCV VL was too high (n = 4). Two participants were not eligible because they had not been infected with HCV for ≥ 6 months. Other reasons for ineligibility were infection with HCV of a genotype other than 1a/1b or 4, a FibroScan result that was too high, having been previously exposed to DAAs for the current infection, having low estimated CrCl, receiving contraindicated medication, not using effective contraception, not attending randomisation visit and moving away during screening (n = 1 for each).

FIGURE 3.

Participant flow diagram. Note that individuals can have more than one reason for exclusion.

Two participants were randomised in error, defined as never intended to be randomised (e.g. data entry error, not infected with HCV, study drug never being dispensed), with the error realised and notified immediately. Therefore, these two participants were excluded from all analyses, as prespecified in the SAP. One of these participants was receiving contraindicated medication. The other was confirmed to not be infected with HCV after randomisation but prior to drugs being dispensed.

Recruitment

The trial stopped recruiting after randomising 204 participants because of slow recruitment and the lack of available patients at participating centres. Follow-up continued until the last participant’s last visit (24 weeks after EOT) on 4 April 2019.

Baseline characteristics

Sixty-two (31%) participants were female; participants’ median age was 45 [interquartile range (IQR) 37–53] years and their median FibroScan score was 4.9 (IQR 4.2–5.8) kPa (Table 5). Sixty-eight (34%) participants were co-infected with HIV.

Median screening VL was 711,423 (IQR 218,776–1,995,262) IU/ml and median enrolment VL was slightly higher, at 741,946 (IQR 249,097–1,872,136) IU/ml (Lin’s concordance coefficient 0.84), in samples taken a median of 19 (IQR 13–33) days apart. A total of 166 (82%) patients were infected with genotype 1a, 34 (17%) were infected with genotype 1b and two (1%) were infected with genotype 4. Sixty (30%) patients had the CC genotype of the IL-28B gene, 106 (52%) had the CT genotype and 27 (13%) had the TT genotype, with the genotype unknown in nine (4%) participants. Of the 188 participants with baseline sequencing available for post-trial analysis, 27 (14%) had a RAS to any prescribed first-line drug.

Twenty-four (12%) participants had previously been treated unsuccessfully with interferon and ribavirin for their current infection, 10 (5%) had been successfully treated for a previous infection and three (2%) had spontaneously cleared a previous infection. The most common causes of HCV infection were injecting drug use (n = 99, 50%) and having a high-risk sexual partner (n = 71, 35%). Sixty-four (32%) participants had current or recent illicit substance abuse and 13 (6%) had current or recent alcohol abuse.

All but four (2%) participants received ombitasvir/paritaprevir/ritonavir plus dasabuvir as their first-line treatment (see Table 5).

Follow-up and treatment received

Overall, first-line follow-up was very good, with only a small number of visits missed [four (2%) at day 3, five (2%) at day 7, six (3%) at day 14, four (2%; n = 167 expected) at day 28, none at EOT, six (3%) at EOT plus 4 weeks, eight (4%) at EOT plus 8 weeks and three (1%) at EOT plus 12 weeks] (Figures 4 and 5). In total, 13 (6%) participants were LTFU and one participant withdrew consent (Table 6). Eleven (5%) participants were LTFU during first-line treatment, but eight of these participants missed the last visit only (i.e. EOT plus 24 weeks) and attended the visit at which the primary outcome was measured (i.e. EOT plus 12 weeks). In the case of the other three participants who were LTFU while on first-line treatment, the last visit was in one case on day 28, in one case at EOT plus 4 weeks and in one case at EOT plus 8 weeks. Follow-up for retreatment was similarly good, with three (5%) of the visits missed at week 2, one (2%) visit missed at week 4, two (3%) visits missed at week 8, two (3%) visits missed at EOT, four visits (6%) visits missed at EOT plus 4 weeks, 10 (16%) visits missed at EOT plus 8 weeks and three (5%) visits missed at EOT plus 12 weeks (Figure 6). During the retreatment phase, one participant withdrew consent (last visit at week 4) and two participants became LTFU (last visits at week 2 and EOT plus 12 weeks).

FIGURE 4.

First-line follow-up by duration randomisation.

FIGURE 5.

First-line follow-up by ribavirin randomisation.

FIGURE 6.

Retreatment follow-up.

The mean number of days of first-line DAA treatment was 56 (SD 4.2) in the fixed-duration arm and 35 (SD 5.7) in the VUS arm, with those randomised to VUS1 taking 32 (SD 4.2) days and those randomised to VUS2 taking 39 (SD 5.6) days (Figure 7). One participant was LTFU before completing first-line treatment, seven stopped DAAs early, one stopped ribavirin early and seven stopped late (Table 7). Those participants stopping late were not prescribed more than their allocated duration (the difference resulted from taking any missed doses at the end of treatment, as instructed) (see Chapter 2). Among those participants stopping early, two chose to do so, a further two stopped because of AEs, two missed doses and did not take these at the end of treatment, one lost 4 days’ worth of drugs and one had drug supply issues. Fifty-five (28%) participants reported missing at least one first-line dose, with 40 (20%) participants reporting missing a dose only once. The percentage of participants reporting a missed dose increased with time on treatment, rising to 29 (14%) participants reporting a missed dose at their EOT visit (Figures 8 and 9).

FIGURE 7.

Randomised and actual treatment duration by failure at SVR24.

FIGURE 8.

Reported first-line missed doses by duration randomisation.

FIGURE 9.

Reported first-line missed doses by ribavirin randomisation.

The mean days of first-line plus retreatment DAAs was 64 (SD 24.3) in the fixed-duration arm and 77 (SD 42.8) in the VUS arm, with those randomised to VUS1 taking 85 (SD 43.2) days and those randomised to VUS2 taking 63 (SD 36.8) days. One participant was LTFU, one withdrew consent before completing retreatment, two stopped all retreatment early, three stopped ribavirin early and three stopped late (Table 8). Of those participants stopping early, two did so because of AEs and three missed doses and did not take these at the end of treatment. Twenty-four (39%) participants reported missing at least one retreatment dose, with 17 (27%) participants reporting missing a dose only once (Figures 10 and 11).

FIGURE 10.

Reported retreatment missed doses by duration randomisation.

FIGURE 11.

Reported retreatment missed doses by ribavirin randomisation.

Numbers analysed

Although some participants were LTFU before the time point at which some outcomes were measured, it was possible to ascertain many of these outcomes through routine medical records when participants had returned to clinic for their usual care. Specifically, in total, one participant withdrew consent (at retreatment week 4) and a further 13 (6%) participants were LTFU (first-line treatment, n = 1; post first-line EOT, n = 9; on retreatment, n = 2; post retreatment EOT, n = 1). However, HCV VL results were available from medical notes for most of those who did not withdraw consent, meaning that first-line SVR12 and SVR24 could not be ascertained for only three (1%) and six (3%) participants, respectively, and overall (i.e. first-line treatment plus retreatment) SVR12 and SVR24 for only five (2%) and eight (4%) participants, respectively (excluded from the corresponding analyses following the SAP; see Chapter 2).

All analysis was by original assignment groups. The primary analysis was by intention to treat, with a secondary per-protocol analysis for SVR12. The per-protocol population was defined as those receiving first-line treatment for > 90% and < 110% of the prescribed duration based on prescription and temporary/permanent discontinuation and in whom the difference in HCV VL values between screening and enrolment would have led to a difference in allocated duration of DAAs of ≤ 2 days had participants been allocated to the VUS group. Although the median difference in VL between screening and enrolment was 0.01 (IQR –0.19–0.21) log₁₀ IU/ml, the absolute differences were greater (Figure 12), leading to 57 (28%) participants being excluded from the per-protocol analysis because the difference in duration of treatment with DAAs would have been ≥ 3 days had duration been determined by the enrolment rather than the screening VL [31 (23%) VUS1 participants vs. 26 (39%) VUS2 participants because the second strategy received more drug overall] (see Figure 7). In total, 70 participants (70%) randomised to VUS compared with 72 (71%) participants randomised to fixed duration, and 68 (68%) participants randomised to ribavirin compared with 74 (73%) participants not randomised to ribavirin, were included in the per-protocol population (Table 9).

FIGURE 12.

Hepatitis C virus VL at screening and enrolment (all patients) (a) by assay and (b) by duration randomisation and failure status.

Outcomes and estimation

SVR12

All participants (197/197) achieved SVR12 after first-line treatment and any retreatment (Table 10), with a difference of 0% (95% CI –3.8% to 3.7%), within the prespecified 4% non-inferiority margin. Ninety-one per cent (91/101; 95% CI 86% to 97%) of those randomised to the 8 weeks’ fixed-duration treatment achieved SVR12 after first-line treatment, compared with 48% (47/98; 95% CI 39% to 57%) of those randomised to VUS (a difference of –43%, 95% CI –54% to –32%; p < 0.001) (Figure 13). The proportion of participants who achieved SVR12 after first-line treatment was significantly higher in the group randomised to VUS2 (72%, 23/43) than in the group randomised to VUS1 (36%, 24/66; p = 0.001) (interaction between duration randomisation and strategy p = 0.001). There was evidence of differences in rates of SVR12 after first-line treatment between those randomised to ribavirin (68%, 68/98) and those not (72%, 70/101; p = 0.48 adjusting for interaction between duration randomisation and strategy). There was no evidence of and interaction between ribavirin and duration randomisations overall (heterogeneity p = 0.16, adjusted for the interaction between duration randomisation and strategy) or in the variable-duration group, where SVR12 was 52% (25/48; 95% CI 37% to 67%) with ribavirin compared with 44% (22/50; 95% CI 30% to 59%) without ribavirin (difference 8%, 95% CI –10% to 27%; p = 0.38).

TABLE 10 - SVR12 outcomes by duration and ribavirin randomisation (intention to treat)

Estimate is an average over both DAA strategies and taken from a model that includes an interaction between randomisation and DAA strategy (p = 0.001).

Heterogeneity p-value for the interaction between duration randomisation and DAA strategy (p = 0.001).

Outcome	Treatment arm, n (%, 95% CI)		Total, n (%, 95% CI)	Risk difference (95% CI); p-value
	Varying duration	Fixed duration
Randomised	100	102	202
Primary outcome evaluable	97	100	197
Primary outcome: SVR12 – first-line treatment or retreatment	97 (100, 96 to 100)	100 (100, 96 to 100)	197 (100, 98 to 100)	0 (–0.038 to 0.037)
SVR12: first-line treatment evaluable	98	101	199
SVR12: first-line treatment only	47 (48, 39 to 57)	92 (91, 86 to 97)	139 (70, 64 to 75)	–0.43 (–0.54 to –0.32); p < 0.001a
Received VUS1	66	67	133
SVR12: first-line treatment only	24 (36, 25 to 48)	62 (93, 86 to 99)	86 (65, 58 to 71)	–0.56 (–0.69 to –0.43); p < 0.001
Received VUS2	32	34	66
SVR12: first-line treatment only	23 (72, 56 to 87)	30 (88, 77 to 99)	53 (80, 71 to 90)	–0.16 (–0.35 to 0.03); p = 0.09
	Ribavirin	No ribavirin
Randomised	100	102	202
Primary outcome evaluable	97	100	197
SVR12: first-line treatment or retreatment	97 (100, 96 to 100)	100 (100, 96 to 100)	197 (100, 98 to 100)	0 (–0.038 to 0.037)
SVR12: first-line treatment evaluable	98	101	199
Primary outcome: SVR12 – first-line treatment only	69 (68, 61 to 76)	70 (72, 65 to 78)	139 (70, 65 to 75)	–0.03 (–0.13 to 0.06); p = 0.48b

FIGURE 13.

Overall SVR12 and first-line SVR12 by randomised groups and strategy (VUS1/VUS2). Fixed, overall SVR12 for 8-week therapy; fixed1, fixed duration when VUS duration received VUS1; fixed2, fixed duration when VUS duration received VUS2; FL, first line; RT, retreatment.

Subgroup analysis (prespecified)

Sixteen subgroups were prespecified in the protocol and SAP. Differences between duration randomisation groups in rates of first-line SVR12 were significantly smaller among participants in whom the virus was suppressed at days 7 and 14 than among those in whom the virus was not suppressed (heterogeneity p = 0.02 and p = 0.03, respectively) (Figures 14 and 15). The effect of the virus being suppressed at day 3 could not be formally tested because all participants in whom VL was already suppressed by day 3 achieved SVR12 on first-line treatment, but the difference in SVR12 among participants in whom VL was not suppressed was substantially larger (0% VUS vs. 40% fixed duration). There was also a trend for difference in SVR12 on first-line treatment between the fixed- and variable-duration groups to be larger for those with baseline RAS than for those without (heterogeneity p = 0.051). Although the evidence was weak, the difference in rate of SVR12 on first-line treatment between fixed- and variable-duration groups was numerically much larger in the case of those who had previously been unsuccessfully treated with interferon than among those who had not (heterogeneity p = 0.13). No subgroup favoured variable duration.

FIGURE 14.

Subgroup analysis by duration comparison. D, day.

FIGURE 15.

Variation in the difference in the rate of first-line SVR12 between fixed- and variable-duration groups by key subgroups for fixed duration vs. varying duration. Note that solid bars represent the first subgroup (detectable VL on the various days shown, no previous unsuccessful treatment and no baseline resistance to drugs taken as first-line treatment) and empty bars the second subgroup (undetectable VL on the various days shown, previous unsuccessful treatment and baseline resistance to drugs taken as first-line treatment). p-values are heterogeneity p-values comparing the difference between fixed and VUS1/2 strategies across the two subgroups. A heterogeneity p-value for day 3 VL cannot be estimated because of perfect prediction. D, day.

There was no evidence of any variation in the (lack of) effect of ribavirin across these subgroups (heterogeneity p ≥ 0.16) (Figure 16).

FIGURE 16.

Subgroup analysis by ribavirin comparison. D, day.

Considering the time when individuals first became undetectable (rather than the prespecified subgroup analyses according to whether they were detectable or undetectable at specific time points), all 10 individuals who became undetectable at day 3 of treatment achieved first-line SVR12 regardless of treatment duration [as did 31 of 38 (82%) participants who were first undetectable at day 7] (Figures 15 and 17).

FIGURE 17.

First-line SVR12 by time first suppressed. Note that figure does not include eight patients who never had VL suppression. In addition, for patients allocated 28–31 days, their EOT visit is also their 28-day visit and they are included in each group. D, day.

SVR24

All participants achieved SVR24 after first-line treatment or retreatment (194/194; risk difference 0%, 95% CI –3.8% to 3.8%) (Table 11). After first-line treatment only, 89% (88/99; 95% CI 83% to 95%) of participants randomised to fixed duration achieved SVR24, compared with 47% (46/97; 95% CI 38% to 56%) of participants randomised to VUS1/2, that is a difference of –42% (95% CI –53% to 31%; p < 0.001). There was no evidence of differences between the groups randomised to or not randomised to ribavirin for SVR24 after first-line treatment (p = 0.87).

TABLE 11 - SVR24 outcomes by duration and ribavirin randomisation (intention to treat)

Estimate is an average over both DAA strategies and taken from a model that includes an interaction between randomisation and DAA strategy (p = 0.001).

Heterogeneity p-value for the interaction between duration randomisation and DAA strategy (p = 0.001).

Outcome	Treatment arm, n (%, 95% CI)		Total, n (%, 95% CI)	Risk difference (95% CI); p-value
	Varying duration	Fixed duration
Randomised	100	102	202
Outcome evaluable	96	98	194
SVR24: first-line treatment or retreatment	96 (100, 96 to 100)	98 (100, 96 to 100)	194 (100, 98 to 100)	0 (–0.038 to 0.038)
Reached EOT plus 24 weeks	97	99	196
SVR24: first-line treatment only	46 (47, 38 to 56)	88 (89, 83 to 95)	134 (68, 63 to 74)	–0.42 (–0.53 to –0.31); p < 0.001a
	Ribavirin	No ribavirin
Randomised	100	102	202
Outcome evaluable	96	98	194
SVR24: first-line treatment or retreatment	96 (100, 96 to 100)	98 (100, 96 to 100)	194 (100, 98 to 100)	0 (–0.038 to 0.038)
Reached EOT plus 24 weeks	97	99	196
Primary outcome: SVR24 – first-line treatment only	68 (69, 61 to 76)	66 (68, 60 to 76)	134 (68, 63 to 74)	0.01 (–0.09 to 0.11); p = 0.87b

Hepatitis C virus viral load rebound and primary first-line treatment failure

In total, 41 (20%) participants had HCV VL rebound, defined as having a confirmed HCV VL greater than the LLOQ after two consecutive visits with HCV VL less than the LLOQ, with the latter confirmatory result also > 2000 IU/ml. Six (6%) participants in the fixed-duration group and 35 (35%) participants in the variable-duration group experienced rebound (a difference of 29%, 95% CI 18% to 39%; p < 0.001) (Figure 18). There was no evidence of a difference in the percentage of participants experiencing rebound between those randomised to and those not randomised to ribavirin (19% vs. 22%, respectively; p = 0.59) (Figure 19).

FIGURE 18.

Cumulative incidence plot of HCV VL rebound by duration randomisation.

FIGURE 19.

Cumulative incidence plot of HCV VL rebound by ribavirin randomisation.

Twenty-one (10%) participants had primary first-line treatment failure, defined as having a confirmed > 1 log₁₀ increase from HCV VL nadir on treatment and > 2000 IU/ml (Figure 20). Participants randomised to variable duration were significantly more likely to experience primary failure than those randomised to fixed duration, with an estimated difference of 10% (95% CI 2% to 18%; p = 0.008). There was no evidence of differences in percentages with primary first-line treatment failure between those randomised to receive or not receive ribavirin (10% vs. 11%, respectively; p = 0.83) (Figures 21 and 22).

FIGURE 20.

Cumulative incidence plot of primary first-line treatment failure by duration randomisation.

FIGURE 21.

Cumulative incidence plot of primary first-line treatment failure by ribavirin randomisation.

FIGURE 22.

Resistance at baseline (all tested participants) and at first-line treatment failure (failures only) classified by (a) first-line drugs and (b) retreatment drugs. p-values are relate to comparisons of resistance to (a) first-line drugs and (b) retreatment drugs with resistance to any other DAA between ribavirin groups. Dark-blue bars represent resistance to drugs received as (a) first-line treatment and (b) retreatment and light-blue bars represent resistance to all DAAs, including those not used in the trial.

Detectable viral load 4 weeks after randomisation

Overall, 21% (41/197) of participants had a detectable VL at day 28. There was no evidence of differences between fixed- and variable-duration groups or between the ribavirin group and no-ribavirin group (p = 0.08 and p = 0.26, respectively) (Table 12).

TABLE 12 - Detectable VL 4 weeks after randomisation by duration and ribavirin randomisation (intention to treat)

Outcome	Treatment arm, n (%)		Total, n (%)	Risk difference (95% CI); p-value
	Varying duration	Fixed duration
Randomised	100	102	202
HCV VL 4 weeks after randomisation	96	101	197
Detectable VL	15 (16)	26 (26)	41 (21)	–0.10 (–0.21 to 0.01); p = 0.08
	Ribavirin	No ribavirin
Randomised	100	102	202
HCV VL 4 weeks after randomisation	97	100	197
Detectable VL	17 (18)	24 (24)	41 (21)	–0.06 (–0.18 to 0.05); p = 0.26

Emergent resistance-associated substitutions to first-line treatment

Sixty-two (31%) participants met the criteria for failure on first-line treatment [fixed-duration group, 11/102 (11%); variable-duration group, 51/100 (51%), p < 0.0001; ribavirin group, 29 (29%); no-ribavirin group, 33 (32%), p = 0.68]. More failures occurred with VUS1 (62%, 42/68) than with VUS2 (28%, 9/32) (p = 0.002; interaction with fixed vs. variable strategy p = 0.003). Twenty-five per cent (14/56) of participants developed a new RAS (not present at baseline) to at least one of their prescribed first-line drugs (Table 13). Within paired samples available at baseline and failure, there was no evidence of difference in development of emergent RASs between those randomised to variable and fixed durations (24% vs. 30%, respectively; p = 0.77). However, those randomised to ribavirin were significantly less likely to develop new RASs (11% vs. 38% of those not randomised to ribavirin, a difference of –27%, 95% CI –48% to –6%; p = 0.01), with significantly lower rates of resistance to any DAA and to non-structural protein 5A (NS5A) inhibitors (both p = 0.01).

TABLE 13 - Emergent RASs by duration and ribavirin randomisation (intention to treat)

Outcome	Treatment arm		Total	Risk difference (95% CI); p-value
	Varying duration	Fixed duration
Randomised, n	100	102	202
Sequenced data after failure/total failures, n/N	46/51	10/11	56/62
Emergent resistance-associated variant to first-line drugs, n (%)	11 (24)	3 (30)	14 (25)	–0.05 (–0.36 to 0.27); p = 0.78
	Ribavirin	No ribavirin
Randomised, n	100	102	202
Sequenced data after failure, n/N	27/29	29/33	56/62
Emergent resistance-associated variant to first-line drugs, n (%)	3 (11)	11 (38)	14 (25)	–0.27 (–0.48 to –0.06); p = 0.01

Ancillary analyses

Failures

Overall, 62 participants failed on first-line treatment (Table 17), all of whom started retreatment a median 2.9 weeks after meeting the failure criteria. Only one participant failed while on treatment (at EOT, 28 days DAA) (Figure 23). The majority of VUS1 participants failed at EOT plus 4 weeks, with smaller numbers failing at EOT plus 8 weeks and EOT plus 12 weeks. By contrast, more VUS2 participants failed at EOT plus 8 weeks than at EOT plus 4 weeks (comparing timing of failure between VUS1 and VUS2, p = 0.08; variable duration vs. fixed duration, p = 0.07; VUS1 vs. fixed duration, p = 0.03). Two participants, both randomised to 8 weeks’ fixed-duration treatment without ribavirin, failed after achieving SVR12. However, there was no evidence of differences in failure VLs (VUS1, median 158,073 IU/ml; VUS2, median 89,125 IU/ml; fixed-duration group, median 346,737 IU/ml; VUS2 vs. VUS1, p = 0.41; VUS2 vs. fixed-duration group, p = 0.21) (see Figure 23).

TABLE 17 - Summary of failures

Outcome	Treatment arm				Total (N = 202)
	Varying duration, ribavirin (N = 49)	Varying duration, no ribavirin (N = 51)	Fixed duration, ribavirin (N = 51)	Fixed duration, no ribavirin (N = 51)
Meeting failure criteria, n (%)	23 (47)	28 (55)	6 (12)	5 (10)	62 (31)
Started retreatment, n (%)	23 (100)	28 (100)	6 (100)	5 (100)	62 (100)
Median (IQR) [range] weeks from meeting criteria to retreatment	3.3 (2.0–6.9) [0.9–32.7]	2.6 (2.1–3.2) [1.3–10.1]	4.6 (2.6–8.1) [2.6–14.0]	3.1 (2.1–4.9) [1.9–7.0]	2.9 (2.0–4.4) [0.9–32.7]

FIGURE 23.

First-line failure. (a) Timing of first-line treatment failure and (b) HCV VL at first-line treatment failure. Note that HCV VL shown is the first, not the confirmatory, VL. No failures were considered reinfections after genome sequencing.

Changes in viral load and viral load suppression (prespecified analysis)

There was no evidence of differences in the mean HCV VL according to either type of randomisation (duration or ribavirin) from screening up to day 14 (Figures 24 and 25). There was also no evidence of difference in the percentage of patients with known undetectable VLs between those randomised to variable-duration treatment and those randomised to fixed-duration treatment up to day 28 (p = 0.13) (Figure 26). After EOT, the percentage of patients with an undetectable VL was significantly lower in the group randomised to VUS1 than in those randomised to fixed-duration treatment, but there was no evidence of a difference between those randomised to VUS2 and those randomised to fixed-duration treatment (p < 0.001 and p = 0.53, respectively). There was no evidence of difference between the percentages with undetectable VL at any time point between ribavirin randomisation groups (global p = 0.82) (Figure 27).

FIGURE 24.

Mean HCV VL on first-line treatment by duration randomisation.

FIGURE 25.

Mean HCV VL on first-line treatment by ribavirin randomisation.

FIGURE 26.

Hepatitis C virus VL suppression by duration randomisation.

FIGURE 27.

Hepatitis C virus VL suppression on first-line treatment by ribavirin randomisation.

Adverse events (harms)

Ten SAEs occurred during the trial, five in each of the duration groups (Table 18) and five in each of the ribavirin groups (Table 19). None was related to study drug (Table 20). Two SAEs were life-threatening, nine necessitated or prolonged hospitalisation and one was another important medication condition. There were 21 grade 3 or 4 AEs in 14 participants, four of which were probably or definitely related to first-line drugs and eight of which were probably or definitely related to retreatment drugs. Sixteen AEs led to changes in study drugs. There was a trend towards more first-line drug changes in those randomised to ribavirin (p = 0.06) and more retreatment drug changes in those randomised to VUS treatment (p = 0.06), but there was no evidence of differences in the number of AEs between groups according to either randomisation. Three grade 3 or 4 anaemias were observed and all occurred in participants randomised to VUS treatment and ribavirin (p = 0.12).

There was no evidence of differences in levels of haemoglobin, alkaline phosphatase (ALP) or bilirubin or in CrCl between the duration randomisation groups at any time point (p > 0.10); however, there were differences at EOT and EOT plus 12 weeks in ALT level (each p = 0.02) and from day 28 in aspartate aminotransferase (AST) level (p < 0.01) (Figure 28). In the case of the ribavirin randomisation, there was no evidence of a difference between randomised groups in ALT, AST or ALP level or in CrCl at any time point (p > 0.11) (Figure 29). For those randomised to ribavirin, there was a significant decrease in haemoglobin and a significant increase in bilirubin while on treatment (p < 0.001); however, levels had returned to normal by EOT plus 12 weeks.

FIGURE 28.

Change in laboratory parameters by duration randomisation: (a) haemoglobin; (b) ALT; (c) AST; (d) ALP; (e) CrCl; and (f) bilirubin. RT, retreatment; W, week.

FIGURE 29.

Change in laboratory parameters by ribavirin randomisation: (a) haemoglobin; (b) ALT; (c) AST; (d) ALP; (e) CrCl; and (f) bilirubin. RT, retreatment; W, week.

Interpretation

The main question addressed by the trial was whether or not a strategy of using shorter, and personalised (based on screening HCV VL), first-line DAA treatment followed by immediate retreatment with standard 12-week treatment courses in everyone not achieving initial cure, overall, can achieve the same rates of SVR12 as standard fixed-duration first-line treatment courses, potentially while using less drug overall. To the best of our knowledge, this is one of the very few, large, strategic, post-licensing trials in this disease area.

Overall, we found non-inferiority of strategies using first-line ultrashort treatment durations, with both groups achieving 100% SVR12 rate after retreatment. Although the initial shortening strategy (i.e. VUS1) was able to cure only 36% of participants after first-line treatment, strikingly, a relatively small increase in ultrashort treatment duration (from a mean of 32 days to 39 days) led to SVR12 rates doubling from 36% to 72%. Interestingly, the rate of SVR12 achieved by the 8-week fixed-duration strategy (91%) was not higher than in previous Phase II trials of shorter treatment courses,23 even though the trial inclusion criteria limited VL to under 6,000,000 IU/ml, which would have been expected to reduce the risk of failure. The first-line cure rates achieved even with VUS2 are not sufficiently high to routinely recommend such a strategy for stable patients able to adhere to 8–12 weeks’ therapy, particularly as overall mean treatment duration with VUS2 was very similar to the overall mean treatment duration with the fixed 8-week duration. However, the findings do suggest that a high proportion of patients can be cured with, on average, approximately 60% of the licensed duration of first-line therapy with the agents used in the trial. Further work is ongoing to identify predictors of cure on VUS1 and VUS2.

Declining adherence to DAA therapy as treatment continues has been demonstrated previously, with patients suggesting ‘feeling as if the treatment is working’ as a reason for decreasing adherence. 10 Adherence is often considered to be better in trials than in real-world situations, but we found similar decreases in adherence with time on first-line treatment, with 28% of participants reporting any missed first-line doses. Adherence to retreatment was even poorer than adherence to first-line treatment. However, the fact that retreatment resulted in 100% SVR12 despite this again illustrates the likely overtreatment of the majority of individuals by standard courses. SVR12 was 72% with VUS2 of mean duration 39 days (despite 28% of participants reporting missing any first-line doses), suggesting that intermittent non-adherence may be less important than overall adherence during weeks 4–8 of first-line treatment. These findings emphasise the importance of supporting adherence after week 4 of therapy to ensure good cure rates in hard-to-reach populations. 14,25

In practice, one important concern about starting treatment in a patient considered unlikely to complete an 8- to 12-week course is the risk of virological failure with emergent resistance that might compromise retreatment options. This is particularly the case when, as in this trial, retreatment does not include a protease inhibitor (in contrast to licensed retreatment options). In our population, shorter first-line treatment strategies did not reduce participants’ ultimate ability to achieve SVR12. The trial used a first-line regimen that is active only against specific genotypes; however, there is no obvious reason why this would not also be the case with pangenotypic first-line treatment regimens. Retreatment with a 12-week course of sofosbuvir, ledipasvir and ribavirin achieved 100% cure rates, a finding that is reassuring from an ethics perspective in terms of future trials of short-course or personalised therapy, but also suggests that it may be a reasonable retreatment option for patients failing therapy more generally, at least in some situations, for example where access to licensed retreatment options, such as sofosbuvir/velpatasvir/voxilaprevir, remains limited.

Ribavirin is generally considered to have a relatively poor side-effect profile, with anaemia and fatigue, in particular, being common. The very high efficacy and low AE rates achieved with DAAs7,16 have, therefore, limited ribavirin’s role in treatment recommendations in the DAA era. However, some patients may still benefit from ribavirin,17 with some preliminary work suggesting that adding ribavirin to short-course therapy does increase SVR12 rates26 and possibly also reduces the rate of emergent resistance in patients failing therapy. 27 Here, adjunctive ribavirin was actually well tolerated in first-line treatment and retreatment, with only 2–4% of participants experiencing AEs related to trial drugs or causing changes to trial drugs, including ribavirin, despite careful elicitation of AEs at scheduled visits.

However, across both the fixed 8-week and the variable ultrashort randomised groups, there was no evidence of improvements in SVR12 with adjunctive ribavirin. In participants randomised to 4–7 weeks’ first-line therapy, SVR12 rates were 8% higher in those randomised to ribavirin, but this result did not reach statistical significance. However, for the first time in a randomised comparison, we found that the emergence of resistance was significantly lower in those failing therapy (12% with ribavirin vs. 38% without). Further work to characterise the impact of ribavirin on different resistance variants and the mechanism of action of ribavirin is ongoing. However, it is important to note that the success of 12 weeks’ retreatment was unaffected by emergent resistance. The main role for adjunctive ribavirin may, therefore, be to reduce the potential for compromise of subsequent retreatment in patients considered at high risk of not completing standard courses of therapy.

Tailoring the duration of therapy based on initial treatment response, so-called response-guided therapy, was standard for interferon-based treatment courses that were prolonged and associated with substantial toxicity. 28 The very high viral suppression rates after 2 weeks of DAA therapy mean that response-guided approaches are not currently recommended, and there are limited data supporting such a strategy in routine practice. 29,30 For the first time, we show that very early response to treatment, namely achieving VL suppression by day 3, or even day 7, may help predict success of shortened courses of licensed therapy. Relatively few patients achieved this early suppression (approximately 5% and 20%, respectively) and early monitoring will be impractical for many outpatient settings. However, where patients are treated in a supervised setting, for example prisoners or as inpatients, such an approach may help guide management and should be investigated further. 30,31

In settings where treatment costs are proportional to duration of therapy, shortened therapy and retreatment could theoretically offer a more cost-effective approach to treatment than standard duration therapy for all. A prior modelling study comparing fixed-duration (8-week) therapy with shortened therapy (4 or 6 weeks, fixed duration) found that the 8-week fixed strategy was most likely to be cost-effective unless SVR12 rates reached 77% for 6 weeks’ therapy. 32 In our trial, a strategy with VUS1 actually used more drug (a mean of 85 days, including retreatment) and so is very unlikely to be cost-effective, although VUS2 (a mean of 63 days, including retreatment) may be in some circumstances, depending on local costs for treatment. Whether or not such an approach is viable will depend on the local health system context.

Limitations

The main trial limitation is that the trial fell short of its original recruitment target for complex reasons relating to public sector commissioning for drug treatment. However, the higher than anticipated success of retreatment (predicted to be 85% but was actually 100%) meant that the trial was still able to demonstrate non-inferiority according to its prespecified margin. Most patients were randomised to a treatment that is now not widely used and so the relevance to other treatment courses will need to be established prospectively.

Generalisability

A major strength of the trial is that it did recruit a range of participants, including around one-third of participants who were HIV/HCV co-infected and in whom we found no evidence of differential response to variable-duration strategies or to ribavirin. Around one-third of participants were also female.

This trial was designed to test strategies for treatment, rather than specific regimens. Almost all recruitment took place when the combination of ombitasvir, paritaprevir, dasabuvir and ritonavir (Viekirax) was a preferred first-line treatment in the UK NHS. This combination still remains a recommended option in the NHS, but the degree to which our findings can be generalised to other combinations with broader genotype coverage is unknown. The similar declines in HCV VL and increases in suppression seen with this and with other DAA combinations suggest that it is plausible that the relationship between duration of therapy and SVR12 is similar for other combinations approved for 12-week treatment courses in patients with mild disease.

The trial did require a stable population able to adhere to a follow-up schedule with significantly more visits than regular standard of care. Despite the fact that around one-third of participants reported actively using recreational drugs, visit non-attendance was low and self-reported adherence reasonably high.

Overall evidence

Overall, we found that unsuccessful short-course first-line therapy did not compromise retreatment with sofosbuvir/ledipasvir/ribavirin, with 100% SVR12 achieved overall and evidence of non-inferiority compared with a fixed 8-week treatment duration plus retreatment. SVR12 rates were significantly increased when ultrashort treatment varied between 4 and 7 weeks rather than between 4 and 6 weeks. We found no evidence of ribavirin significantly improving first-line SVR12, but it significantly reduced resistance emergence in those failing first-line treatment.

Implications for health care

Currently, clinical decision-making is complex when there are concerns that a patient will be unable to complete a full course of treatment. Our findings suggest that shortened courses of treatment (of 4–7 weeks) are able to achieve cure in a majority of patients in these settings. Furthermore, where it is feasible to measure early virological response, this may have a role in identifying a group of participants who will be cured with shorter courses of therapy. More importantly, with the combination used (which included a protease inhibitor in first-line therapy), any resistance that developed did not appear to compromise retreatment and the addition of ribavirin significantly reduced the emergence of resistance in treatment failure.

Recommendations for future research

STOP-HCV-1 co-applicants

Professor Graham Cooke, Professor Ann Sarah Walker, Professor Ellie Barnes, Professor Sarah Pett, Dr Stephen Ryder, Dr Daniel Forton, Professor Graham Foster, Kosh Agarwal, Ashley Brown, Dr Richard Riley and Mr Charles Gore (patient representative).

STOP-HCV-1 Trial Steering Committee

Professor Sir Ian Weller (chairperson), Dr Collette Smith, Dr Patrick Ingiliz, Mr Robert James, Ms Rachel Halford, Professor Graham Cooke, Professor Ann Sarah Walker and Professor Ellie Barnes.

STOP-HCV-1 Trial Management Group

Professor Graham Cooke, Professor Ann Sarah Walker, Fleur Hudson, Ellie Barnes, Dr Stephen Ryder, Katie Topping, Emma Hudson, Leanne McCabe, Aminata Sy, Emily Dennis, Cara Purvis, Dr Christopher Jones, Professor Sarah Pett, Helen Ainscough, Ania Spurdens, Charlotte Russell and Deborah Alawode.

Data Monitoring Committee

Professor David Lalloo (chairperson), Professor John Dillon, Professor Tim Peto, Professor Katherine Fielding and Dr Oscar Bortolami.

Contributions of authors

Graham S Cooke (https://orcid.org/0000-0001-6475-5056) was the chief investigator, he conceived, designed and managed the trial, was a site principal investigator and was responsible for participant recruitment and data collection, and co-wrote the first draft of the report.

Sarah Pett (https://orcid.org/0000-0002-4851-4727) conceived, designed and managed the trial, and reviewed and provided comment on the report.

Leanne McCabe (https://orcid.org/0000-0002-6509-8999) performed statistical analysis, and reviewed and provided comment on the report.

Christopher Jones (https://orcid.org/0000-0002-6301-2282) reviewed and provided comment on the report.

Richard Gilson (https://orcid.org/0000-0002-9572-193X) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Sumita Verma (https://orcid.org/0000-0001-7021-8409) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Stephen D Ryder (https://orcid.org/0000-0001-9649-4444) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Jane D Collier (https://orcid.org/0000-0002-7046-1713) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Stephen T Barclay (https://orcid.org/0000-0003-0415-785X) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Aftab Ala (https://orcid.org/0000-0001-6257-3160) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Sanjay Bhagani (https://orcid.org/0000-0003-2557-4337) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Mark Nelson (https://orcid.org/0000-0002-5944-6096) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Chin Lye Ch’Ng (https://orcid.org/0000-0003-3179-9313) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Benjamin Stone (https://orcid.org/0000-0003-4999-2250) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Martin Wiselka (https://orcid.org/0000-0002-4587-5826) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Daniel Forton (https://orcid.org/0000-0002-0903-9169) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Stuart McPherson (https://orcid.org/0000-0002-5638-2453) was a site principal investigator and was responsible for participant recruitment and data collection, and reviewed and provided comment on the report.

Rachel Halford (https://orcid.org/0000-0003-4569-2824) reviewed and provided comment on the report.

Dung Nguyen (https://orcid.org/0000-0001-9390-9990) performed sequencing and bioinformatic analyses, and reviewed and provided comment on the report.

David Smith (https://orcid.org/0000-0001-7778-7137) performed sequencing and bioinformatic analyses, and reviewed and provided comment on the report.

M Azim Ansari (https://orcid.org/0000-0003-2790-8353) performed sequencing and bioinformatic analyses, and reviewed and provided comment on the report.

Helen Ainscough (https://orcid.org/0000-0003-2912-2661) reviewed and provided comment on the report.

Emily Dennis (https://orcid.org/0000-0002-9634-7767) managed the trial, and reviewed and provided comment on the report.

Fleur Hudson (https://orcid.org/0000-0001-5891-7514) managed the trial, and reviewed and provided comment on the report.

Eleanor J Barnes (https://orcid.org/0000-0002-0860-0831) conceived and designed the trial, and reviewed and provided comment on the report.

Ann Sarah Walker (https://orcid.org/0000-0002-0412-8509) conceived, designed and managed the trial, and co-wrote the first draft of the report.

Data-sharing statement

The MRC CTU at UCL supports data-sharing where appropriate. Requests should be made to the corresponding author.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research. The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the MRC, NETSCC, the EME programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the EME programme or the Department of Health and Social Care.

Participating centres

Trial co-ordination and oversight