Chronic hepatitis C represents a major public health burden due to the development of cirrhosis, which is associated with an increased risk of end-stage liver disease (ESLD) related morbidity and mortality [1]. In the UK, it is estimated that 214,000 people are currently chronically infected with the hepatitis C virus (HCV) [2]; of these, between 10% and 20% are likely to develop cirrhosis over the 20–30 year period following infection [3].UK hospital admissions for HCV-related ESLD complications have increased from 608 in 1998 to 2,390 in 2012, and HCV-related deaths have increased from 98 in 1996 to 428 in 2012 [2]. The slow progressive nature of HCV infection means that whilst the incidence of new infections has decreased over time [4], the incidence of late stage complications will continue to rise unless those infected are successfully treated. The accepted clinical endpoint of HCV treatment is the achievement of a sustained virologic response (SVR), defined as an undetectable level of HCV RNA at a certain time point after end of treatment (e.g. 12 [SVR12] or 24 weeks [SVR24]). EASL guidelines state that the primary goal of HCV therapy is to cure the infection, which is generally associated with resolution of liver disease in patients without cirrhosis [5]. Long-term follow-up studies have shown that an SVR corresponds to a definitive cure of HCV infection in more than 99% of cases [6]. Patients who achieve SVR are at significantly decreased risk of hepatic morbidity and mortality [7]. HCV genotype, genetic polymorphisms (IL28B) and the stage of liver fibrosis upon initiating treatment are the strongest predictors of the likely achievement of SVR [5].

Current standard of care (SoC) in the UK for HCV genotype 1 infection is triple therapy comprising peg-interferon alfa, ribavirin and a protease inhibitor (either boceprevir or telaprevir) [8,9]. For non-genotype 1 HCV infection, SoC is combination therapy with peg-interferon alfa plus ribavirin [10].

Rates of SVR vary between 40% and 80% depending on the treatment used, HCV genotype, disease stage and other patient characteristics [11]. The therapy landscape for treating HCV is rapidly expanding [11]. 2014 UK guidelines identify roles for the new direct acting antivirals (DAAs) daclatasvir, sofosbuvir and simeprevir in the management of chronic hepatitis C [12]. While these therapies are associated with higher SVRs, reduced treatment durations and fewer adverse events compared to SoC, concerns have been raised that the significant differences in the cost-per-cure for newer regimens may result in providers potentially favouring cheaper, less well tolerated regimens [12]. The emergence of new treatments adds additional choice and regimen complexity for clinicians and payers. Estimating the health economic consequences of competing therapies is generally undertaken using cost-effectiveness analysis. In the current HCV landscape, the possible permutation of such analysis is considerable. Cost-effectiveness is fundamentally influenced by treatment response, which is dependent upon factors such as genotype, fibrosis stage and prior treatment status, but also age, duration of therapy (due to either discontinuation or response-guided therapy) and treatment disutility. Consequently, aligning therapy- and patient-specific characteristics to ensure optimal value for money will become increasingly challenging. In order to offer insight into the potential clinical and economic benefit associated with treatment success in HCV, the principle objective of this study was to quantify the benefit that increasing levels of SVR would be expected to realise in terms of predicted life-years (LYs) gained, quality-adjusted life-years (QALYs) gained and costs of managing ESLD-related complications. As the credibility of these projections are dependent upon the robustness of the underlying model used, a secondary objective was to assess the validity and generalisability of the results, and therefore the model used, from a UK perspective.

Fig. 2 reports the per-patient discounted and non-discounted LYs for patients aged 40, 50 and 60 years, stratified by fibrosis stage (F0–F4) and rates of SVR (0%–100%). In patients with fibrosis stage F4, aged 40 years, discounted LYs ranged from 14.0 years for no treatment (0% SVR), to 19.5 with a therapy that achieved 100% SVR. Assuming no discounting, the range was 21.9–35.5 LYs. This figure also enabled the calculation of incremental LYs gained; for example, patients aged 50 in fibrosis stage F3 were estimated to gain an additional 1.8 discounted LYs when comparing treatments with an SVR rate of 100% versus 50% (19.2 versus 17.4 LYs), and 3.6 discounted LYs when comparing an SVR of 100% to no treatment (0% SVR) (19.2 versus 13.3 LYs).

Fig. 3 depicts the graphical representation of the per-patient QALYs gained for the same scenarios (patients aged 40, 50 and 60 years, stratified by fibrosis stage [F0–F4] and rates of SVR [0%–100%]). Predicted gain in QALYs per 10% improvement in SVR varied by fibrosis stage and age, and were obtained by subtracting relevant absolute levels of QALYs reported, Fig. 3. In patients aged 40, QALY gains per 10% improvement in SVR ranged from 0.23 (F0) to 0.64 (F4) (discounted) and 0.58 (F0) to 1.35 (F4) (non-discounted). QALY gains decreased with advancing age; for example, a 10% improvement in SVR in F4 was associated with a QALY gain of 0.40 (discounted) and 0.63 (non-discounted) for 60 year old patients compared to 0.64 (discounted) and 1.35 (non-discounted) for 40 year old patients. In those aged 40, QALY were maximised when SVR reached 100%, at 18.0 years (discounted) and 34.3 years (non-discounted). QALYs decreased to 7.5 years (discounted) and 11.8 years (non-discounted) for patients in F4 who remain untreated (0% SVR). Fig. 3 also enabled the incremental QALY to be calculated for differing rates of SVR. For example, patients aged 50 in fibrosis stage F3 had an expected discounted QALY gain of 1.9 QALYs when comparing an SVR rate of 90% (3.3 QALYs) to 50% (1.9 QALYs).

Fig. 4 presents the total expected per-patient lifetime discounted and undiscounted costs (£000s) associated with HCV-related complications. Projected expenditure was minimised with 100% SVR at approximately £300 when considering both discounted and non-discounted values in those aged 40, in fibrosis stage F0 and F1. Expenditure rises to £49,300 (discounted) and £81,500 (non-discounted) when considering a 40 year old patient in F4 who remains untreated (0% SVR). The incremental cost of disease-related complications for varying rates of SVR can also be calculated by subtracting absolute values; for example, within the discounted estimates, there is an expected incremental complication-related cost of £13,100 when comparing an SVR rate of 40% (total cost £20,600) with 80% (total cost £7,500) in patients aged 50 in fibrosis stage F2.

Projecting the long-term clinical and cost consequences associated with HCV disease progression is a necessary requirement for the evaluation of new technologies for the treatment of chronic HCV. The principle analysis herein demonstrates the significant impact on expected costs, LYs and QALYs associated with increasing levels of SVR. The SVR rates presented in this analysis are a realistic representation of those relevant to contemporary HCV-related clinical practice. A noteworthy observation relates to the costs and life expectancy associated with no treatment (SVR 0%), in which discounted per-patient lifetime costs approach £50,000 in those with more advanced disease and discounted life expectancy is reduced by around 8 years. With an estimated prevalence of chronic HCV infection in the UK of approximately 215,000 and around 3% treated each year [2], the potential total cost impact and life years lost is significant.

In line with reported progression rates, patients treated successfully from F4 were allowed to progress to HCC and DC [17]. This significantly impacted predicted costs and resulted in lower quality-adjusted life expectancy. Consequently it is likely that health benefit is maximised and costs minimised by treating patients prior to reaching fibrosis stage F4.

The estimated costs and health outcomes presented in this study are based on long-term modelled projections and therefore demonstrating the validity of the model used is crucial. Model validation has been described as the “process of determining whether a simulation model is an accurate representation of the system for the particular objectives of the study” [40]. Consequently, this validation exercise has focused on demonstrating that the model predicts cost-effectiveness and incidence of ESLD endpoints consistent with those previously reported in UK-specific cost-effectiveness analyses. Costs and health-related quality of life are key drivers of cost-effectiveness. Most data for costs and utilities from identified studies were transparent and readily accessible for extraction. In most studies, no cost was reported for the SVR health state beyond the drug acquisition costs and sometimes monitoring costs directly associated with treatment. Although costs and utilities for most disease complications were well documented for the first year, there were differences of approach for subsequent years spent in the same health state, with the common assumption that each year with a particular complication is associated with the same costs and utilities as the previous year(s), while other studies presented different cost and/or utility for subsequent years. The greatest variation in cost was associated with the cost of transplantation in the first year, with costs ranging from £27,330 to £58,736.

Large variability in values employed for disease transition progression rates, utilities and costs can impact cost-effectiveness analyses. This was most notable when validating ICER values using MONARCH transition rates: ICER values were over- and under-estimated when validated to studies by Grieve et al. [39] and Shepherd et al. [41]. The study by Grieve et al. [39] is an economic evaluation of interferon-alfa in mild chronic hepatitis C patients. The likely cause of over-estimation is the effect of slower disease progression with MONARCH transition rates compared to rates reported in Grieve et al. [39]. Assuming no treatment intervention, the 20-year incidence rate of ESLD complications for patients in F4 (compensated cirrhosis) observed with MONARCH was 0.120 compared with an incidence of 0.309 for the Grieve study. This absolute difference caused an underestimation of the QALY gain between treatment arms. The second study, by Shepard et al. [41], was a HTA for the use of interferon-based dual therapy in the treatment of chronic hepatitis C. The study progressed patients from a single state of chronic hepatitis C to ELSD complications. In order to replicate the study using MONARCH transition rates, there was a need to distribute patients across fibrosis stages, since no information was available regarding fibrosis severity; a uniform distribution across fibrosis stages (F0–F3) was assumed. This assumption led to a significantly increased rate of progression to ESLD complications; the 20-year incidence of F4 (compensated cirrhosis), assuming no treatment, was 0.368 when utilising MONARCH rates and 0.171 when utilising study-specific rates. This under-estimation highlights the importance of understanding baseline characteristics amongst patient cohorts.

Validation to clinical endpoints observed in UK cost-effectiveness studies provides insight into the predictive consistency of the MONARCH model to previously published analysis. Of note, however, is the relative paucity of robust epidemiological evidence suitable for informing and validating HCV disease models in general. A review of key model parameters utilised in HCV cost-effectiveness analyses identified 34 models that drew their disease progression inputs predominantly from nine studies (published between 1989–1997) [32]. Consequently, while this study demonstrates consistency with other published UK cost-effectiveness analysis, there is a need for more contemporary UK HCV epidemiological analysis to be undertaken and published.

In reviewing published UK cost-effectiveness analysis, a common pathway of six disease states was identified: chronic hepatitis C, compensated cirrhosis, decompensated cirrhosis, hepatocellular carcinoma, liver transplant and death. In some studies, the early disease states and/or the decompensated cirrhosis state were further subdivided e.g. decompensated cirrhosis was further subdivided into ascites, variceal haemorrhage and hepatic encephalopathy [41,42]. Some studies only describe chronic hepatitis C disease states as mild, moderate or severe [35,38,39,43,44]. In this instance, assumptions were made that equated mild, moderate or severe health states to fibrosis stages (mild = F0 and F1, moderate = F2 and F3 and severe = F4). These assumptions and variations between studies affected the R² goodness of fit statistics.

Validating the transition rates embedded within the MONARCH model to the limited UK-specific epidemiological studies available provides reassurance that the MONARCH cost-effectiveness model has the ability to consistently model the natural history of HCV in line with other studies. An important component of model validation within the context of decision-making is the ability to demonstrate whether the choice of model used might influence the decisions made. The analysis presented here shows that use of the MONARCH model would not have significantly altered the ICERs to such an extent that it would have resulted in a different decision. This study has however highlighted those structural assumptions that can substantially influence cost-effectiveness ratios.

As the treatment landscape continues to evolve, we anticipate that the research questions of the future will begin to inform on aspects of patient heterogeneity currently unaccounted for in HCV clinical and epidemiological studies. This may necessitate a move away from a Markov modelling approach and introduce the use of patient-level simulation. In addition, greater emphasis should be placed on justifying the choice of underlying data and ensuring that data used are the most relevant to the research question. While acknowledging these data limitations, the model produces consistent estimates of life expectancy and quality-adjusted life expectancy to published UK studies and consistent with those used to inform prior HTA decisions.

In this study, expected disease costs, life expectancy and quality-adjusted life expectancy for various ages, fibrosis stages and SVR rates have been tabulated. As new treatments for HCV emerge, it is useful to gauge the expected health economic consequences associated with SVR rates, without the need to employ a particular model. These tables are intended to provide a general overview of the likely benefits associated with treatment and as a consequence of improving SVR rates. The validation analysis undertaken helps to demonstrate the structural validity of the MONARCH cost-effectiveness model in projecting the long-term clinical and cost consequences associated with HCV disease progression, against a range of cost-effectiveness and clinical endpoints. The aim of the validation analysis was to provide reassurance that the projections from the MONARCH were consistent with those previously published in the UK.