Cost-effectiveness of telaprevir in combination with pegylated interferon alpha and ribavirin in previously untreated chronic hepatitis C genotype 1 patients

Abstract

Background:

Telaprevir (T, TVR) is a direct-acting antiviral (DAA) used for the treatment of genotype 1 chronic hepatitis C virus (HCV) infection. The sustained virological response (SVR) rates, i.e., undetectable HCV RNA levels 24 weeks after the end of treatment, is what differentiate treatments. This analysis evaluated the cost-effectiveness of TVR combined with pegylated interferon (Peg-IFN) alfa-2a plus ribavirin (RBV), with Peg-IFN and RBV (PR) alone or with boceprevir (B, BOC) plus Peg-IFN alfa-2b and RBV, in naïve patients.

Methods:

A Markov cohort model of chronic HCV disease progression reflected the pathway of naïve patients initiating anti-HCV therapy. SVR rates were derived from a mixed-treatment comparison including results from Phase II and III trials of TVR and BOC, and trials comparing both PR regimens. SVR has significant impact on survival, quality-of-life, and costs. Incremental cost per life year (LY) gained and quality-adjusted-life-year (QALY) gained were computed at lifetime, adopting the (National Health Service) NHS perspective. Cost and health outcomes were discounted at 3.5%. Uncertainty was assessed using deterministic and probabilistic sensitivity analyses. Sub-group analyses were also performed by interleukin (IL)-28B genotype and fibrosis stage.

Results:

Higher costs and improved outcomes were associated with T/PR relative to PR alone, resulting in an ICER of £12,733 per QALY gained. T/PR retained a significant SVR advantage over PR alone and was cost-effective regardless of IL-28B genotype and fibrosis stages. T/PR regimen ‘dominated’ B/PR, generating 0.2 additional QALYs and reducing lifetime cost by £2758. Sensitivity analyses consistently resulted in ICERs less than £30,000/QALY for the T/PR regimen over PR alone.

Limitations:

No head-to-head trial provides direct evidence of better efficacy of T/PR vs B/PR.

Conclusion:

The introduction of TVR-based therapy for genotype 1 HCV patients is cost-effective for naïve patients at the £30,000 willingness-to-pay threshold, regardless of IL-28B genotype or fibrosis stage.

Keywords::

• Cost-effectiveness analysis
• Chronic hepatitis C infection
• Treatment-naïve patients
• Telaprevir
• Boceprevir

Introduction

Infection with the hepatitis C virus (HCV) is a major public health problem and a leading cause of chronic liver disease, which may progress in a proportion of patients to cirrhosis and hepatocellular carcinoma (HCC)1. The goals of anti-HCV treatment include eradication of the virus (i.e., achieving a sustained virologic response [SVR], defined as undetectable HCV RNA levels 24 weeks after the end of treatment), to prevent viral transmission and disease progression1.

The combination of pegylated interferon (Peg-IFN) and ribavirin (RBV) for 48 weeks was the standard of care therapy for treating genotype 1 chronic HCV infection for many years as they were the only licensed agents in the UK until 2011. However, the success rate for Peg-IFN and RBV (PR) treatment is less than 50% among HCV genotype 1-infected patients; adverse events are common. Response rates are even lower in patients with advanced fibrosis. Response rates are also sub-optimal in patients with prior treatment failure2–4. These factors highlight the need for more effective therapy.

The direct-acting antivirals (DAAs)—telaprevir (TVR) and boceprevir (BOC)—were approved by the US Food and Drug Administration (FDA; May 2011) and by the European Medicines Agency (EMA; September 2011) for the treatment of chronic genotype 1 HCV infection in adults who are either treatment-naïve or were previously treated with interferon-based therapy, and who have compensated liver disease including cirrhosis5–8. Both TVR and BOC inhibit the non-structural (NS) 3/4A serine protease, which is essential for the replication of HCV.

The significant improvements in SVR rates associated with the addition of HCV protease inhibitors to PR therapy, as well as the potential for shortening treatment duration in rapid responders, represents an important benefit for people with chronic genotype 1 HCV9,10. According to the current 2011 practice guideline of the American Association for the Study of Liver disease (AASLD), the optimal treatment strategy for HCV is now the combination of PR with one of the two new DAAs, TVR and BOC. Therefore, the landscape of HCV management has significantly changed11.

The objective of this analysis was to evaluate the cost-effectiveness of TVR-based regimens, compared with PR alone or BOC-based therapy, in treatment-naïve patients with chronic HCV genotype 1 infection. This was the first economic evaluation of DAAs in HCV. The long-term natural history of chronic genotype 1 HCV infection necessitates the use of decision modeling to appreciate the benefits of each treatment, and to account for the off-setting of direct costs on a long-time horizon. We considered cost-effectiveness in terms of incremental cost per life-year (LY) gained and quality-adjusted life-year (QALY) gained from the perspective of the National Health Service (NHS) in England and Wales.

Methods

The Markov model

A Markov model was developed to reflect the natural history of chronic genotype 1 HCV, simulating the pathway a naïve patient may experience from the time of diagnosis onwards. This model was based on previous economic assessments and UK National Institute for Health and Clinical Excellence (NICE) appraisals12–15. The model structure is shown in Figure 1.

Figure 1. Markov model diagram. SVR, sustained virologic response; HCV, hepatitis C virus.

Patients entered the model in the ‘mild fibrosis’ (defined as METAVIR F0-F1), ‘moderate fibrosis’ (defined as METAVIR F2), or ‘compensated cirrhosis’ (defined as METAVIR F3-F4) health states, and received treatment with either TVR plus PR (T/PR), PR alone, or BOC plus PR (B/PR; see below for details). Consistent with previous economic models, patients achieving an SVR 24 weeks after the end of treatment were assumed to have cleared the virus and to be cured, without further disease progression. Unsuccessfully treated patients could either remain in their original health state for a period of time, or progress to more advanced stages of liver disease such as cirrhosis, decompensated cirrhosis, and/or HCC. Patients with advanced liver disease (decompensated cirrhosis or HCC) were assumed to be potential candidates for liver transplantation. For the early stages of disease, we assumed that progression was correlated with aging and used age-dependent transition probabilities (TPs; Table 1), which were derived from an economic evaluation where TP inputs were stratified by age at treatment16. The remaining TPs were taken from published articles and previous economic evaluations12–15,17–19.

Table 1. Annual transition probabilities (TPs) used within the Markov model.

Initial health state transitioned from:	To:	Probability of transitioning (DSA ranges)
Mild fibrosis16	Moderate fibrosis	<35 years old: 0.015 (0.011–0.019); 36–45 years old: 0.023 (0.017–0.029); >45 years old: 0.053 (0.026–0.044)
Moderate fibrosis16	Cirrhosis	<35 years old: 0.021 (0.016–0.026); 36–45 years old: 0.032 (0.024–0.040); >45 years old: 0.048 (0.036–0.060)
Cirrhosis15	Decompensated cirrhosis	0.039 (0.029–0.049)
Cirrhosis15	HCC	0.014 (0.011–0.018)
Cirrhosis with SVR	HCC	0.014 (0.011–0.018)
Decompensated cirrhosis15	HCC	0.014 (0.011–0.018)
Decompensated cirrhosis15	Liver transplant	0.020 (0.015–0.025)
Decompensated cirrhosis15	Death	0.130 (0.098–0.163)
HCC	Liver transplant	0.040 (0.030–0.050)
HCC15	Death	0.43 (0.323–0.538)
Liver transplant15	Death	0.15 (0.113–0.188)
Post-liver transplant15	Death	0.057 (0.043–0.071)

HCC, hepatocellular carcinoma; SVR, sustained virologic response.

All health states carried a risk of death (age-dependent general mortality and specific mortality associated with fibrosis/disease stage). Background mortality was taken from the UK Office for National Statistics (ONS) and disease-specific mortality from published articles14,15. A cycle length of 1 year was used and a half-cycle correction implemented assuming that state transition occurred halfway through each cycle16. A lifetime time horizon was considered, i.e., up to 100 years of age, and the perspective utilized was that of the NHS in the UK.

Model inputs

Patient characteristics and treatment regimens

The model followed three cohorts of adult patients from the ADVANCE and SPRINT-2 trials who had chronic genotype 1 HCV infection and had not received any previous anti-HCV therapy9,10. Each cohort was defined by their mean age at treatment (30 years: 15.2%, 40 years: 24.2%, and 50 years: 60.6%) and by the fibrosis stage. The split between mild fibrosis (METAVIR F0-F1), moderate fibrosis (METAVIR F2), and cirrhosis (METAVIR F3-F4) was based on data from the ADVANCE study9. Key characteristics of the patients are shown in Table 2.

Table 2. Model inputs.

Base case inputs (DSA ranges)	Mild fibrosis	Moderate fibrosis	Cirrhosis	Decompensated cirrhosis	HCC	Liver transplant	Post-liver transplant
Patient characteristics9
Age at treatment: 30 years, %	63.6	32.7	3.7
Age at treatment: 40 years, %	43.4	40.6	16.0
Age at treatment: 50 years, %	30.8	43.3	25.9
Drug costs
PR alone	£7292	£7292	£7292
T/PR	£25,455	£25,455	£25,455
B/PR	£26,342	£26,342	£26,342
Treatment duration9
PR alone, weeks	36.2	36.2	36.2
T/PR, weeks
PR, weeks	26.9	26.9	26.9
TVR, weeks	10.7	10.7	10.7
B/PR
PR, weeks	31.1	31.1	31.1
BOC, weeks	27.1	27.1	27.1
Monitoring costs
PR alone	£972	£972	£1033
T/PR	£882	£882	£928
B/PR	£928	£928	£989
Adverse management costs
EPO	£1650*	£1650*	£1650*
Health state costs
Without treatment15	£175 (88–263)	£911 (455–1366)	£1445 (723–2169)
SVR following treatment15,16	£220 (110–329)	£268 (134–403)	£475 (237–712)
Advanced disease stages15	–	–	–	£11,583 (5791–17,374)	£10,321 (5161–15,482)	£46,720 (23,360–70,080)	£12,016 (6008–18,024)
Utilities9,15
Without treatment	0.77 (0.578–0.963)	0.66 (0.495–0.825)	0.55 (0.413 – 0.688)
PR alone	0.675	0.579	0.482
T/PR	0.682	0.585	0.487
B/PR	0.682	0.585	0.487
SVR following treatment	0.82 (0.615–1)	0.72 (0.540–0.900)	0.61 (0.458 – 0.763)
Advanced disease stages	–	–	–	0.45 (0.338–0.563)	0.45 (0.338–0.563)	0.45 (0.338–0.563)	0.67 (0.503–0.838)

PR, peginterferon/ribavirin; T/PR, telaprevir plus PR; B/PR, boceprevir plus PR; SVR, sustained virologic response; PI, protease inhibitor; HCC, hepatocellular carcinoma; TVR, telaprevir; BOC, boceprevir.

*183 units per week for 159 days.

The model compared three therapeutic regimens. The first regimen was T/PR, where patients received triple therapy with TVR plus Peg-IFN alfa-2a and RBV for 12 weeks followed by Peg-IFN alfa-2a and RBV alone for another 12 weeks (if HCV RNA undetectable at weeks 4 and 12; extended rapid virological response [eRVR]) or 36 weeks (if eRVR not achieved). The second regimen was Peg-IFN alfa-2a and RBV alone for 48 weeks. The third regimen was B/PR, where patients received Peg-IFN alfa-2b plus RBV for 4 weeks, followed by triple therapy with BOC plus Peg-IFN alfa-2b and RBV for another 24 weeks (if HCV RNA was undetectable at week 8 through 24) or 44 weeks (if not achieving these criteria).

Resource use and healthcare costs

The frequency and intensity of tests and monitoring associated with anti-viral therapy as well as unit costs were taken from a previous health technology assessment (HTA) report14. Any alteration to the initial protocol to include further monitoring, either due to the treatment itself or the presence of cirrhosis, was discussed and agreed with a clinical expert. The monitoring costs were computed for each treatment regimen (T/PR, B/PR, and PR alone) based on the treatment duration.

Each treatment group reported some adverse events, the most common being rash, pruritus, nausea, diarrhoea, and anemia9,10. The costs of managing those adverse events were determined according to the clinical trial practice and taken from the Personal Social Services Research Unit (PSSRU)20. The possibility of receiving erythropoietin (EPO) for patients suffering from anemia was restricted to patients receiving BOC, since this was disallowed in the phase III TVR trials. The mean exposure (number of days of EPO) was taken from the SPRINT-2 trial10, and cost assumption was based on the British National Formulary (BNF) recommendation21. The cost of blood transfusion to manage anemia was not considered for both. The health state costs for SVR, mild and moderate fibrosis, compensated and decompensated cirrhosis, HCC, liver transplantation, and post-liver transplantation were taken from previous economic evaluations and updated to 2009/2010 values using the Hospital and Community Health Services (HCHS) pay and prices index20. Costs used in the model are shown in Table 2.

Drug costs were included and used unit costs from the BNF (2011)21. Drug costs for Peg-IFN alfa-2a assumed a dose of 180 μg/0.5 mL administered once a week, which corresponded to a weekly cost of £124.40. In comparison, the dosage of Peg-IFN alfa-2b was 120 μg per week, resulting in a weekly cost of £159.60. Since the RBV dose varied with the patient’s weight, we assumed a daily dose of 1000 mg. This corresponded to a weekly cost of £77.08. TVR costs were estimated based on a 750 mg dose administered 3-times daily, corresponding to a weekly cost of £1866.50, and totalling £22,398 for a full 12-week course. The cost of the PR and T/PR regimens was calculated based on the expected number of weeks that patients would receive double or triple therapies. For TVR, the expected duration of therapy was contingent on virological responses (e.g., eRVR positive/negative), stopping rules, and premature discontinuations, and was based on the ADVANCE trial9. BOC costs were inputted as £2800 for a 28-day, 336-tablet pack (excluding VAT)21,22, representing £30,800 for a 44-week course. The expected duration of therapy also took into account the stopping rules and premature discontinuations as reported in the SPRINT-2 trial10.

Health-related quality-of-life (HRQoL)

EQ-5D provides a generic measure of quality-of-life. The EQ-5D valuation index summarizes the information of the five dimensions (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) of the descriptive system into one index that attaches weight to each of the levels in each dimension23,24. EQ-5D valuation index data captured from the trials were applied to disutilities during the year of treatment. This enabled a differential impact on HRQoL to be appropriately implemented in the model, reflecting both the type of treatment and the duration of therapy that patients received.

The relative decrements of utility between baseline and SVR for the T/PR and PR regimens were directly derived from the ADVANCE trial9, in which the EQ-5D instrument was administered at several points in time. It is expected that the longer the treatment duration, the more adverse events are likely to be experienced by the patients25,26; however, in the absence of data, the decrement associated with the B/PR regimen was assumed to be the same as that associated with the T/PR regimen. Utilities associated with the other health states were the same as the ones used by Hartwell et al.15. All utilities are presented in Table 2.

Efficacy: SVR rates

The relative efficacy of T/PR compared with PR alone or B/PR was estimated based on the results of mixed-treatment comparison (Table 3)27. This analysis was run on eight studies (i.e., PROVE1 and 2, ADVANCE, SPRINT-1 and 2, IDEAL, Rumi et al. and Yenice et al.)9,10,28–33 investigating various T/PR, B/PR, and PR regimens. A fixed-effect model was performed using WinBugs version 1.4 (MRC Biostatistics unit, Cambridge, UK), employing Markov chain Monte Carlo (MCMC) simulation, based on 50,000 iterations after 50,000 burn-in iterations. This analysis was carried out on the trial results and differs from the ones reported in the respective label for TVR (79% for T/PR vs 46 % for PR) and BOC (66% for B/PR and 38% for PR)7,8.

Table 3. Sustained virological response (SVR) rates inputted into the model.

Patients with SVR, %^a (DSA ranges) Treatment regimens	Overall population	Disease severity			IL-28B genotype
		Mild fibrosis	Moderate fibrosis	Cirrhosis	CC	CT	TT
PR	42.9 (39.23–46.59)	45.6	47.5	32.9	63.6	25.0	23.1
T/PR	73.9 (67.6–79.7%)	81.3	75.0	61.6	90.0	70.6	72.7
B/PR^b	66.5 (59.1–73.9)	NA	NA	NA	NA	NA	NA

^aSVR rates were estimated by a meta-analysis9,10^,28–33.

^bSVR rates according to disease severity were not available for B/PR.

PR, peginterferon/ribavirin; T/PR, telaprevir plus PR; B/PR, boceprevir plus PR; IL-28B, interleukin-28B; NA, not applicable.

Cost-effectiveness analyses

The model analyses were performed using Excel 2007 and Visual Basic Application (VBA) and estimated the total lifetime costs, LYs, and QALYs per patient for the T/PR, B/PR, and PR alone regimens. Future costs and outcomes were discounted at the recommended rate of 3.5% per year34. The results of the analyses are presented as incremental cost-effectiveness ratios (ICERs), expressed as cost per LY gained and cost per QALY gained. A £30,000 willingness-to-pay threshold was used to conclude on the cost-effectiveness of the different anti-HCV regimens. As secondary outcomes, the model also reported the long-term complications that were avoided (i.e., cirrhosis, liver transplant, death).

Cost-effectiveness sub-group analyses

IL-28B genotype: Single nucleotide polymorphisms (SNPs) in the region of the interleukin (IL)-28B gene have been strongly associated with treatment success in HCV genotype 1-infected patients treated with PR alone; patients with the TT or CT IL-28B genotypes showed lower SVR rates than patients with the CC genotype35–38. Post-hoc sub-group analyses explored the incremental efficacy of T/PR over PR alone or in patients with different IL-28B genotypes using data from patients in ADVANCE (T/PR)9. The post-hoc analysis carried out on the ADVANCE trial showed that 72% of the CC patients qualified for shorter treatment duration (24 weeks in total) compared to 16% with PR alone.

Disease severity: The cost-effectiveness of the T/PR regimen vs the PR regimen was also assessed by disease severity, since SVR rates were available separately for patients with mild fibrosis (METAVIR F0 to F1), moderate fibrosis (METAVIR F2), or cirrhosis (METAVIR F3 to F4). At the time of the analysis, it was not possible to conduct a similar comparison between the T/PR and B/PR regimens because detailed SVR rates with B/PR were not available according to fibrosis stage.

Sensitivity analyses

One-way sensitivity analyses were conducted on parameter values to examine their effects on ICERs, including discount rate (0% and 6% for both costs and benefits), SVR rates, TPs, utility values, treatment duration, and health state costs. SVR rates and utility decrements were varied within their respective 95% confidence intervals, the TPs within 25% of their initial values, the utilities and costs of health state costs within 10% and 50% of their initial values, respectively.

Probabilistic sensitivity analyses (PSAs) were also carried out by varying key model parameters randomly across their potential distributions over 1000 iterations using values drawn at random from the probability distributions. The distributions of the main parameters were derived from the ones used by NICE14,15. Beta distributions were applied to transition probabilities and utilities, Gamma distributions to costs, and decrement of utilities and posterior distributions computed from the meta-analysis were used for the efficacy data. Each resulting output was plotted as a point on the cost-effectiveness plane, yielding a scatter plot. Cost-effectiveness acceptability curves (CEACs) were derived from these outputs and indicated the probability of each regimen being cost-effective, over a range of possible values of willingness-to-pay.

Results

Cost-effectiveness analyses: base case

In the base case, the T/PR regimen was found to be highly cost-effective within the current NICE thresholds when compared with PR alone, with an ICER of £12,733 per QALY gained, below the £20,000 threshold; the cost per life-year (LY) gained was £20,945. The T/PR regimen was associated with higher costs than the PR regimen at lifetime. This could be explained by the incremental drug cost in the TVR arm during the first year of £18,170. On the other hand, the number of discounted QALYs favored the T/PR regimen, i.e., 13.89 vs 13.03 years for PR alone, resulting in an incremental gain of 0.86 years (Table 4). The discounted life expectancy of patients who received T/PR was estimated to be 19.49 years compared with 18.96 years in the patients who received PR alone. The T/PR regimen dominated the B/PR regimen as it was £2758 less expensive and resulted in 0.203 more QALYs, as shown in Table 4.

Table 4. Cost-effectiveness results: base case.

	T/PR	PR	B/PR	T/PR vs PR		T/PR vs B/PR
				Incremental costs and QALYS	ICER (cost per QALY gained)	Incremental costs and QALYS	ICER (cost per QALY gained)
Total costs	£35,347	£24,420	£38,105	£10,927	£12,733	−£2758	T/PR dominates
Total QALYs	13.89	13.03	13.68	0.86		0.20

PR, peginterferon/ribavirin; T/PR, telaprevir plus PR; B/PR, boceprevir plus PR; QALY, quality-adjusted life-year; ICER, incremental cost-effectiveness ratio.

Regarding clinical outcomes, the base case analysis estimated that treating 1000 patients with TVR would avoid 180 instances of cirrhosis, 12 liver transplants, and seven deaths at lifetime compared with PR, and 43 instances of cirrhosis, three liver transplants, and two deaths compared with B/PR.

Sub-group analyses: IL-28B genotype and disease severity

IL-28B genotype

The T/PR regimen remained a more cost-effective treatment than PR alone, regardless of IL-28B genotype (Table 5), with ICERs below £20,000 per QALY for TT, CT, and CC patients. Greater QALY gains and smallest incremental costs and ICERs for T/PR vs PR alone were seen for patients with the CT or TT genotype compared with the CC genotype. The cost per QALY gained for CT and TT was, respectively, estimated at £5986 and £4820.

Table 5. Cost-effectiveness results: sub-group analyses.

	T/PR	PR	B/PR	T/PR vs PR
				Incremental costs and QALYS	ICER (cost per QALY gained)
IL-28B genotype
CC
Total costs	£31,628	£19,631	£34,564	£11,997	£16,419
Total QALYs	14.33	13.60	14.11	0.73
CT
Total costs	£36,102	£28,554	£38,436	£7548	£5986
Total QALYs	13.80	12.54	13.64	1.26
TT
Total costs	£35,612	£29,998	£40,812	£6615	£4820
Total QALYs	13.86	12.48	13.36	1.37
Disease stage
Mild CHC
Total costs	£29,415	£15,444	NA	£13,971	£17,595
Total QALYs	16.17	15.38	NA	0.79
Moderate CHC
Total costs	£34,821	£24,638	NA	£10,182	£11,446
Total QALYs	13.74	12.85	NA	0.89
Cirrhosis
Total costs	£50,050	£41,843	NA	£8207	£9607
Total QALYs	9.45	8.59	NA	0.85

PR, peginterferon/ribavirin; T/PR, telaprevir plus PR; B/PR, boceprevir plus PR; IL-28B, interleukin-28B; QALY, quality-adjusted life-year; ICER, incremental cost-effectiveness ratio; CHC, chronic hepatitis C.

Disease severity

The T/PR regimen remained a more cost-effective treatment than PR alone regardless of disease severity (Table 5), with ICERs below £20,000, irrespective of fibrosis stage. The T/PR regimen was notably more cost-effective than PR alone for patients with more severe disease (Table 5). The cost per QALY gained was estimated at £17,595 for mild patients, with the ICER decreasing when treating moderate (METAVIR F2) and cirrhotic (METAVIR F3-F4) patients, to reach £11,446 and £9607, respectively. Data limitations at the time of this analysis prevented similar analyses vs B/PR.

Sensitivity analysis: deterministic sensitivity analyses (DSAs) and PSAs

The ICERs were found to be sensitive to changes in key parameters including discount rates, the cost of TVR (as well as the treatment duration), utilities associated with SVR among patients with mild and moderate fibrosis, the other health state utilities, and SVR rates associated with the T/PR regimen. Figure 2 displays the 10 DSAs that had the biggest impact on the model outputs, i.e., the cost per QALY gained compared to PR alone. Given the lifetime horizon considered in this study, ICERs when comparing T/PR with PR alone were the most sensitive to a change in the discount rates applied on both costs and outcomes. The parameters with the next most substantial effect on the results were the utilities associated with reaching SVR among patients with mild and moderate fibrosis, followed by the drug cost for TVR.

Figure 2. Tornado chart of cost per QALY gained for the T/PR regimen vs PR alone. PR, peginterferon/ribavirin; T/PR, telaprevir plus PR; SVR, sustained virologic response; SOC, standard of care; QALY, quality-adjusted life-year; ICER, incremental cost-effectiveness ratio.

All of the PSA simulations for T/PR compared with PR alone fell in the north-east quadrant of the cost-effectiveness plane (Figure 3), and demonstrated that the T/PR regimen is cost-effective depending on the willingness-to-pay threshold per QALY gained. This is confirmed by the cost-effectiveness acceptability curve, which showed that 85% of the simulations had an ICER per QALY gained below £20,000 for T/PR vs PR alone (97% are below £30,000)(Figure 4).

Figure 3. Cost-effectiveness plane: cost per QALY (£/QALY) gained for the T/PR regimen vs PR alone. PR, peginterferon/ribavirin; T/PR, telaprevir plus PR; QALY, quality-adjusted life-year.

Figure 4. Cost-effectiveness acceptability curve based on the cost per QALY (£/QALY) gained for the T/PR regimen vs PR alone and vs B/PR. PR, peginterferon/ribavirin; B/PR, boceprevir plus PR; T/PR, telaprevir plus PR; QALY, quality-adjusted life-year.

Sensitivity analyses for the T/PR vs B/PR regimens showed that T/PR remained dominant when varying the key parameters, except when the unit cost of TVR was increased by 25%. The PSA revealed that, with the current assumptions, the T/PR regimen had approximately a probability of 72% to dominate B/PR (i.e., to be both more effective and less expensive).

Discussion

Successful HCV treatment results in an SVR and is thought to be tantamount to a cure of the disease. It is inferred, but not yet robustly proven, that a SVR improves the disease outcome. Unfortunately, less than 50% of genotype 1 HCV patients achieve an SVR on PR alone2–4,31,39–41.

TVR, a DAA, has been approved for the treatment of chronic genotype 1 HCV patients. The clinical efficacy of TVR in combination with PR has been demonstrated in Phase II and III randomized clinical trials9,28,29,42,43. Improved response rates have been observed in both naïve and in previously-treated patients who have failed PR treatment. Despite the side-effects of DAAs, and their continued necessity for PR backbone therapy, the high rates of early viral response translate into higher rates of cure than with PR alone in genotype 1 HCV. Treatment with T/PR can be shortened in 58–65% of treatment-naïve patients without cirrhosis. Hence, this study analysed the cost-effectiveness of TVR-based triple regimen compared with the current PR alone therapy and another recently approved DAA, i.e., BOC-based triple regimen, for the treatment of chronic hepatitis C genotype 1 treatment-naïve patients.

Based on available data for clinical efficacy, TPs, and costs, we found that treating genotype 1 HCV-infected patients with the TVR-based regimen resulted in an ICER of £12,733 per QALY gained when compared with PR alone. The cost of Peg-IFN alfa-2a was used in the base case analysis; when selecting Peg-IFN alfa-2b, the CE results were better suggesting that the choice of Peg-IFN alfa-2a was conservative. Use of the T/PR response-guided therapy (i.e., 24 weeks of T/PR therapy in those with an eRVR) regimen was estimated to increase life expectancy by 6 months over PR alone. This benefit in favor of T/PR was also demonstrated by the number of discounted QALYs gained; 10 quality-adjusted-months were gained over a lifetime with T/PR vs PR alone.

The cost-effectiveness of TVR-based therapy within currently acceptable thresholds, compared with PR alone, can be explained by its greater efficacy that offsets the higher cost of treatment in the long term. SVR were observed in 74–79% naïve T/PR treated patients. Shorter treatment duration (24 weeks in total) was possible in 58–65% of patients7. Previous HTAs have demonstrated the cost-effectiveness of currently available PR therapy for the treatment of naïve chronic HCV patients14,15. The findings from our analysis are consistent with the guidance from NICE to recommend TVR-based therapy as a clinically-effective and cost-effective treatment option relative to PR therapy for treatment-naive chronic HCV patients1.

The greatest gains with the T/PR regimen vs PR were in patients with characteristics associated with a poor response to PR, such as patients with CT/TT IL-28B genotypes and patients with advanced liver disease. TVR-based therapy improved the SVR rates across all IL-28B genotypes compared with PR alone, with substantially increased rates in those with CC genotypes and a doubling of the SVR rate in CT/TT. When carrying out a sub-group analysis according to the IL-28B polymorphism, TVR-based therapy compared with PR alone was shown to be cost-effective across all genotypes (CC, CT, and TT). TVR-based therapy demonstrated cost-effectiveness in CC genotype patients with the ICER of £16,419. In addition, 72% of the CC patients qualified for shorter treatment duration (24 weeks in total) compared with 16% of those taking PR alone. The ICER was estimated to be the lowest among TT patients, reaching £4820 per QALY gained against PR alone.

Sub-group analyses results demonstrated that TVR-based therapy was more cost-effective compared with PR alone in patients with higher unmet needs. In moderate (METAVIR, F2) patients, the ICER was lower (£11,446) compared with all treatment-naive patients (£12,733). As expected the ICERs were lowest in the population with severe fibrosis. The incremental cost of £8207 and incremental QALY of 0.854 resulted in an ICER of £9607 per QALY gained among cirrhotic patients. However, it should be pointed out that the model extrapolated data from a small number of patients, since cirrhotic patients represented only 20% of the ADVANCE population9.

Although the ICERs further improved in patients with severe fibrosis, there is a need to treat patients early to prevent them from progressing to advanced disease. The ability to treat patients early with a highly efficacious treatment that has the promise of shorter treatment duration has the potential to improve treatment tolerance associated with long duration of therapy, and ultimately avoid potential complications associated with chronic hepatitis C. Since HCV is a potentially curable disease, treating patients and achieving SVR early as a one-time investment could result in major long-term cost savings from the reduction of the incidence, prevalence, and transmission of HCV disease, fibrosis, cirrhosis, and HCC, as well as reduced demand for liver transplantation. These additional benefits have not been quantified to be incorporated in the current model.

The increased cost of recently approved first generation protease inhibitors to treat chronic HCV necessitates a cost-effectiveness analysis to compare these triple regimens. We demonstrated in this analysis that TVR was cost-effective compared with BOC-based therapy. This comparison must be interpreted with caution. In the absence of comparative data based on direct head-to-head evidence from a prospective, randomized controlled trial, we used the results of a mixed-treatment comparison that demonstrated better efficacy for T/PR vs B/PR treatment patients27. The BOC and TVR trials could only be linked in the network through the comparison of PRα-2a vs PRα-2b. As most comparisons are based on only one study, the estimate of the between-study variance had poor precision. As a result, it was hard to differentiate heterogeneity from differences caused by the different treatments compared. Additionally, a major assumption in any indirect comparison analysis and network meta-analysis is that the relative treatment effect is constant across variability in trial characteristics. Some differences in trial populations were detected between the trials (i.e., prevalence of advanced liver disease, black race, and baseline viral load); the higher proportion of patients with low baseline viral load in ADVANCE (TVR) compared to SPRINT-2 (BOC) suggests that the estimate of the OR between TVR and BOC may be conservative. Future studies should further examine the cost-effectiveness of T/PR compared to B/PR when such data are available.

One-way sensitivity analysis of the base case showed that the model outputs were sensitive to discount rates used on both costs and benefits. This is not unexpected given the lifetime perspective considered in this analysis and the fact that treatment costs are concentrated within the first year. The model was also shown to be robust in response to changes in key parameters such as SVR rates and TPs. Overall, these results demonstrate a high probability for the T/PR regimen to be cost-effective compared with PR alone. This was confirmed by the PSA and the cost-effectiveness acceptability curve, where 85% of the simulations had an ICER below £20,000 and 97% below £30,000 per QALY gained for T/PR vs PR alone. These results could be explained by the higher efficacy of the T/PR regimen among treatment-naïve patients, and that the increased costs were largely offset by a reduction in progression to more severe and costly health states. The T/PR regimen also remained dominant over the B/PR regimen when varying the key parameters, except when the unit cost of TVR was increased by 25%. The PSA results indicated that the T/PR regimen had approximately a probability of 72% to be both more effective and less expensive than B/PR. Both the first generation protease inhibitors were cost-effective compared with PR alone.

The extensive sensitivity analyses showed that the model provided robust estimates of the high likelihood for TVR to be cost-effective. However, this model has some limitations. First, even though our model is based on the example that is most frequently used in the published literature, there is a lack of detailed data on the natural history of the disease. In addition, model parameters, and assumptions including TPs, utility values, and health state costs, were not derived from the population of interest (i.e., patients within the UK). As a result, if UK-specific data became available, a corresponding change in the model parameters beyond the ranges used in the DSAs and PSAs could affect the cost-efficacy analysis results. Comparative data based on direct head-to-head evidence from a prospective, randomized, controlled, clinical trial could be utilized in the future modeling. Factors such as alcohol abuse contribute significantly to HCV disease progression but were not incorporated in this model since we focused on the populations enrolled in the clinical trials. The management of anemia will differ between centers and countries; the uncommon use of blood transfusion in the clinical trial (∼0.04% of the patients) was associated with treatment discontinuation and was not included in the model. The use of EPO was only included for patients receiving the BOC regimen in the model as it was not allowed to manage patients receiving TVR in Phase III studies. The difference in cost of anemia management between BOC and TPV strategies may be over-estimated in real life. The rational of this differentiation was to maintain consistency between trial efficacy and trial protocol.

This economic model evaluates the cost-effectiveness of new DAAs in the treatment-naïve population and does not address the evolution of potential drug resistance in patients with treatment failure with a DAA. From a population sequence analysis from the TVR clinical trials, it appears that resistant HCV variants are replaced by wild-type virus over time44,45, and additional data suggest that T/PR may be effective if used later as part of an attempt at re-treatment46. Although these findings are encouraging, the full implications of resistance in the real-life practice setting and associated outcomes are to be further evaluated. A separate cost-effective model has been developed to examine the cost-effectiveness of T/PR compared to PR or B/PR in a treatment-experienced patient population.

HCV infection represents a significant economic burden. The number of patients with advanced liver disease is expected to rise in the next few decades as a large cohort of baby boomers with chronic hepatitis C age47. Indeed, it is anticipated that the economic burden of hepatitis C will more than double over the next two decades. This is mainly due to the slow clinical progression of the disease, meaning that patients who were infected 20–40 years ago will only now start to progress to the later stages of the disease such as cirrhosis, liver disease, and HCC48. Although long-term data are emerging in support of the theory that early treatment of chronic HCV infection may reduce morbidity from the disease, the hypothesis remains to be robustly tested3–5,49–51.

Conclusions

In this analysis, we have demonstrated the cost-effectiveness of T/PR therapy in treatment-naïve patients. The results from our model suggest that response-guided T/PR therapy is cost-effective compared with PR alone, regardless of disease severity and IL-28B genotypes, and cost saving compared with PR. The study does not answer though the question of which is the most relevant and important group to treat now. Many factors determine this paradigm. Nevertheless, the model does demonstrate the candidacy of patients with both mild and advanced fibrosis. Results from this study can help decision-makers set priorities in terms of budget allocation in management of patients with chronic hepatitis C. However, we recognize economic evaluations do not examine the effectiveness of penetration of treatment into key cohorts of infected persons, and that higher prices and complex treatments limit the incentive to implement wider programs of ascertainment and treatment.

In conclusion, our cost-effectiveness model demonstrated that T/PR would be the preferred option compared with PR alone or B/PR, based on the current assumptions. If the higher SVR rates associated with TVR-based therapy can be shown to reduce the risk of disease progression, then it has the potential to further significantly alter the overall cost of managing this population.

Transparency

Declaration of funding

This analysis was funded by Janssen Pharmaceuticals. The authors confirm that the paper is an accurate representation of the study results. Several authors are employees of the sponsor and were involved in the study design, data collection and analysis, interpretation of data, writing of the report, and the decision to submit the paper for publication.

Declaration of interest

SC and FB are employees of OptumInsight, who received funding from Janssen Pharmaceuticals. SG, SC, and SL are employees of Janssen Pharmaceuticals. GD has received consulting fees from Roche, Merck, and Janssen pharmaceuticals

Acknowledgments

We thank Joanne Williams and Catherine Elliott (Gardiner-Caldwell Communications) for general editing/styling and co-ordination support (funded by Janssen Pharmaceuticals).

Author contributions

CS, FB, SG, and GD were involved in designing these analyses; SC and SL were involved in the collection, analysis, and interpretation of data; SC and SL wrote the first draft of the report, which was subsequently critically reviewed by all authors; all authors were involved in the decision to submit the paper for publication.