Sovaldi Cost-Effectiveness/Disease Outcome Analysis
HCV Treatment Cost-Effectiveness/ SOF/PR SVR Reduces Liver Disease vs PI/PR Treatments - (08/08/14)
A decision-analytic Markov model evaluated the health outcomes of SOF compared with current treatment options....."the number of cases of advanced liver disease were 77-81% lower for SOF+PR vs PR, 7078% lower for SOF+PR vs BOC+PR, and 60-67% lower for SOF+PR vs Telaprevir+PR"
AASLD/EASL NY Conference: Outcome analysis of ledipasvir/sofosbuvir (LDV/SOF) for the treatment of chronic hepatitis C infected genotype 1 patients in the UK. - (09/26/14)......"ledipasvir/sofosbuvir is highly effective in preventing advanced liver disease (compensated cirrhosis/HCC/decompensated cirrhosis/liver transplant), & saving lives & QALYs"
Cost-effectiveness of sofosbuvir-based triple therapy for untreated patients with genotype 1 chronic hepatitis C - (08/08/14)
Cost-effectiveness analysis of sofosbuvir plus peginterferon/ribavirin in the treatment of chronic hepatitis C virus genotype 1 infection........"Reductions in the incidence of new cases of liver-disease complications with sofosbuvir + pegIFN/RBV compared with pegIFN/RBV, boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV were 64-82%, 50-68%, 43-58% and 33-56%, respectively. Sofosbuvir + pegIFN/RBV was typically associated with the lowest 1-year cost per SVR. When considering the lifetime incremental cost per quality-adjusted life-year gained, sofosbuvir + pegIFN/RBV was the most cost-effective treatment option assessed. Sofosbuvir + pegIFN/RBV generally dominated (less costly and more effective than) boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV."
.......Conclusion: Sofosbuvir + pegIFN/RBV yields more favourable future health and economic outcomes than current treatment regimens for patients across all levels of treatment experience and cirrhosis stage, as well as for individuals with or without HIV co-infection."
Alimentary Pharmacology & Therapeutics
Sofosbuvir, an oral NS5B nucleotide polymerase inhibitor, is indicated for the treatment of patients infected with hepatitis C virus (HCV).
To evaluate the long-term health economic outcomes of sofosbuvir + pegylated interferon alfa/ribavirin (pegIFN/RBV) compared with current treatments in patients infected with HCV genotype 1 in the US.
A decision-analytic Markov model was developed to estimate health outcomes, number needed to treat and short-term and long-term economic outcomes, including incremental cost-effectiveness ratios and cost per sustained virological response (SVR), for several sofosbuvir-comparator regimen pairings for a cohort of 10 000 patients. It considered three patient cohorts: treatment-naïve, treatment-experienced and treatment-naïve human immunodeficiency virus (HIV) co-infected. Subgroup analyses were conducted for treatment-naïve patients with and without cirrhosis.
Reductions in the incidence of new cases of liver-disease complications with sofosbuvir + pegIFN/RBV compared with pegIFN/RBV, boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV were 64-82%, 50-68%, 43-58% and 33-56%, respectively. Sofosbuvir + pegIFN/RBV was typically associated with the lowest 1-year cost per SVR. When considering the lifetime incremental cost per quality-adjusted life-year gained, sofosbuvir + pegIFN/RBV was the most cost-effective treatment option assessed. Sofosbuvir + pegIFN/RBV generally dominated (less costly and more effective than) boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV.
Sofosbuvir + pegIFN/RBV yields more favourable future health and economic outcomes than current treatment regimens for patients across all levels of treatment experience and cirrhosis stage, as well as for individuals with or without HIV co-infection.
Download the PDF here
Hepatitis C virus (HCV) infection imposes a substantial economic burden on the United States (US) healthcare resources each year.[1-3] An evidence-based review found that numerous studies indicate associations of HCV with clinical, quality of life and economic burden, with chronic HCV being independently and significantly associated with increased healthcare costs. Early intervention with HCV therapy has been shown to be associated with lower downstream direct medical costs. The treatment of HCV infection has evolved rapidly in recent years and the introduction of novel, all-oral, interferon-free regimens is expected to increase treatment uptake and cure rates. It is therefore becoming increasingly important to evaluate the cost-effectiveness of emerging treatments across the general HCV infected population and across various subpopulations, including those that are difficult to treat such as patients co-infected with human immunodeficiency virus (HIV).
The prevalence of HCV in patients infected with HIV is approximately 30%. Patients co-infected with HIV/HCV have a worse prognosis and lower responsiveness to current standard therapies than those with HCV mono-infection.[7, 8] Other subpopulations of individuals with HCV that are particularly difficult to treat include patients with cirrhosis, especially those with clinically decompensated liver disease, because the current standard treatment is complicated by the occurrence of haematological adverse events. Reducing the rate of liver-disease progression in treatment-experienced patients using currently available therapies has also been shown to be difficult.
Hepatitis C virus genotype 1 accounts for 79% of all chronic cases, making it the most common type of HCV infection in the US. Until recently, the standard of care for patients infected with this genotype was an oral protease inhibitor (boceprevir or telaprevir) combined with pegylated interferon alfa (pegIFN) plus ribavirin (RBV).Although 60-80% of previously untreated patients who received boceprevir or telaprevir + pegIFN/RBV showed sustained virological response (SVR) in clinical trials,[13-16] and had a shorter overall duration of treatment than patients receiving pegIFN/RBV alone, these regimens were associated with many adverse events and high discontinuation rates, particularly in the HCV cirrhotic patient population. The proportion of patients progressing to cirrhosis in the US has been projected to increase rapidly over the course of this decade because of the longer-term persistence of HCV infection. Real-world effectiveness studies of these regimens in the general HCV patient population have shown SVR rates of 45-49% and resultant costs of $183 428 and $195 000 per SVR.
Simeprevir is a second-generation protease inhibitor that was approved at the end of 2013 for the treatment of patients infected with HCV genotype 1 and/or 4, in combination with pegIFN/RBV. Clinical trials have shown that it is associated with higher SVR rates than pegIFN/RBV alone, is well tolerated and is associated with mild and reversible adverse events.[24-26] Sofosbuvir is a nucleotide analogue HCV NS5B polymerase inhibitor shown to be effective in an interferon-limiting regimen in patients infected with HCV genotype 1, 4, 5 or 6[27, 28] and in an interferon-free regimen for those infected with HCV genotype 2 or 3[27-29] in phase III clinical trials. In patients infected with HCV genotype 1, the addition of sofosbuvir to pegIFN/RBV regimens yielded SVR rates of approximately 90% 12 weeks after completion of treatment.
Sofosbuvir-based oral regimens have an acceptable safety profile and are generally well tolerated, with no reported premature discontinuations of therapy due to adverse events.[28, 30] The American Association for the Study of Liver Diseases and the Infectious Diseases Society of America (AASLD/IDSA) guidelines were recently updated with simeprevir- and sofosbuvir-based regimens.
Earlier studies have demonstrated that protease inhibitors in combination with pegIFN/RBV are more cost-effective than pegIFN/RBV alone in patients previously treated for chronic HCV genotype 1 infection.[32, 33] More recent cost analyses have indicated that all-oral, interferon-free regimens are cost-effective compared with the standard of care for patients infected with genotype 1.[34, 35] However, no studies to date have evaluated the cost-effectiveness of sofosbuvir in combination with pegIFN/RBV against currently available HCV treatment regimens. The aim of this study was to develop a model to assess the health outcomes and cost-effectiveness of sofosbuvir + pegIFN/RBV compared with the currently available therapies (pegIFN/RBV, boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV) for the treatment of patients infected with HCV genotype 1 from a third-party payer perspective in the US. The model included patients with different levels of treatment experience and cirrhosis stages, as well as individuals with and without HIV co-infection.
This study comprises a health economic analysis of sofosbuvir + pegIFN/RBV for the treatment of patients with HCV genotype 1. Outcomes from the cost-effectiveness model show that sofosbuvir + pegIFN/RBV yields the most favourable future health economic outcomes compared with other currently available regimens across a broad spectrum of patients, including those with different treatment experiences and severity of fibrosis as well as individuals with and without HIV co-infection.
Compared with other current treatment regimens, sofosbuvir + pegIFN/RBV was associated with the lowest incidence of liver-disease complications such as decompensated cirrhosis, HCC, liver transplant and HCV-related deaths. In all patient subgroups, the estimated percentage reduction in these sequelae was greater than 43% for sofosbuvir vs. the comparator regimens, except for simeprevir + pegIFN/RBV in the treatment-experienced patient subgroup (a reduction of 33%). The lower 1-year cost per SVR associated with sofosbuvir + pegIFN/RBV indicated that more patients can be successfully treated under a fixed budget with this regimen than with boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV. The greater increase in SVR rate and the shorter treatment duration with sofosbuvir than with other regimens resulted in significant risk reductions in morbidity (i.e. liver-disease complications), death and costs. The 1-year SVR costs associated with sofosbuvir + pegIFN/RBV were higher than those associated with pegIFN/RBV alone in treatment-naïve patients and those without cirrhosis. However, it should be noted that pegIFN/RBV alone or with boceprevir or telaprevir is not recommended for treatment of HCV genotype 1 in either of these patient populations.
Previous cost analyses have revealed that HCV treatment regimens including protease inhibitors (boceprevir and telaprevir) are cost-effective.[32, 33, 79] The findings from the present analysis, relating to the lifetime incremental cost per QALY gained for patients infected with HCV genotype 1, indicate that sofosbuvir + pegIFN/RBV is currently the most cost-effective option available, even including treatment-experienced patients and those with cirrhosis.
Sofosbuvir + pegIFN/RBV was both less costly and more effective than boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV regimens. Results from the sensitivity analysis indicated that sofosbuvir + pegIFN/RBV continued to be a cost-effective strategy (compared with both simeprevir + pegIFN/RBV and no treatment) even when key clinical and cost parameters (such as SVR rates, adverse event incidence, costs for treatment monitoring and management of adverse events and transition probability estimates) were adjusted across wide, yet plausible, ranges. The outcomes from this model were also robust when sensitivity analyses were performed to account for lower cirrhosis prevalence and fibrosis rates, various transition probabilities to recurrence in patients who achieve SVR (with and without cirrhosis), and the rate of HCC and decompensation in patients with cirrhosis. These findings support those from other recently published cost-effectiveness models, which indicated that oral anti- viral regimens are cost-effective compared with the current standard of care for HCV genotype 1.[34, 35]
The outcomes from the model revealed that, for all treatment regimens analysed, the 1-year costs per SVR were lower for treatment-naïve patients than for treatment-experienced patients; similarly, they were also lower for patients without cirrhosis than cirrhotic patients. These findings are comparable to those reported in a recent cost analysis of protease inhibitor-based therapy. Cases of liver-disease complications are lower when therapy is initiated early, because of higher SVR rates among patients without cirrhosis than among patients with cirrhosis, and the resulting costs avoided via averted cases of liver-disease complications.
Another important finding from this study is that the benefits of treating HCV are considerable, with over a 69% reduction in advanced liver disease cases when treating with an efficacious regimen such as sofosbuvir + pegIFN/RBV compared with no treatment. Gordon et al. demonstrated that the annual costs of end-stage liver disease in patients with untreated HCV were five times higher than the annual costs of noncirrhotic disease in treated patients. Hagan et al. found that the ICER for receiving HCV treatment vs. no treatment was $10 920/QALY. Our cost-effectiveness model found similar ICERs. The ICER for the treatment of HCV genotype 1 treatment-naïve patients with sofosbuvir was around $6000/QALY compared with no treatment. This ICER is well below the commonly accepted willingness-to-pay threshold of $50 000/QALY and less than the ICERs for Papanicolaou test and biennial mammography screening ($17 000/QALY and $27 000/QALY, respectively, in 2013 US dollars).[81, 82]
There are a few potential limitations of this model. First, it is predominantly populated with clinical trial data, which represents a controlled rather than a real-world environment. Numerous studies have suggested that real-world therapy completion rates and SVR rates associated with existing treatment regimens are substantially lower, and adverse event incidence is higher, than previously reported in clinical trial settings.[21, 42, 52, 83, 84] The cost per SVR achieved may be substantially higher than those reported in this analysis when real-world adverse event incidence, discontinuation and virological failures are considered. Sufficient real-world data to populate the model fully are not available; therefore, issues related to differences between clinical trial and real-world efficacy and safety could not be accounted for in this analysis. In addition, the SVR rates for each comparator were obtained from separate clinical trial arms, as no head-to-head clinical trials or meta-analyses including sofosbuvir and its comparators were available at the time of this analysis.
Furthermore, some of the SVR rates for the treatment options in the cirrhosis patient subgroup were based on clinical trials with small sample sizes; as a result, F3 and F4 data were combined in some instances. Patient demographical and clinical characteristics may be different across clinical trials, which can affect patients' SVR rates. However, we took stage of cirrhosis at baseline into consideration, which is the most significant predictor of SVR rate in patients with HCV. In addition, the deterministic sensitivity analysis demonstrates that the findings from this study are robust over a range of SVR rates. Overall, adverse event management accounted for less than 3% of costs for all comparators. When the relevant data become available, additional analyses will be necessary to determine the potential impact of the greater expected real-world differences between the sofosbuvir-based regimens and the other available therapies.
Another potential limitation to consider is that in order to estimate the long-term impact of health and economic outcomes in a clinical trial setting, the model predicted the course of liver disease for each individual over their remaining lifetime and mortality based on literature estimates of natural disease progression data. It is possible that disease progression may vary depending on a patient's gender, race, other comorbidities (e.g. co-infection with hepatitis B virus) or alcohol consumption, which were not accounted for in this model. However, to estimate transition probabilities, we used nationally representative, recent data that controlled for covariates or previously used data for HCV models. The model is also limited with regard to the fact that it does not account for the potential risk of HCC to develop in patients who achieve SVR (specifically if treated at F4 cirrhosis stage). An additional limitation is that this study only included US data; costs of HCV treatments may vary considerably across other countries which may influence cost-effectiveness analysis outcomes. In the future, it would therefore be beneficial for this analysis to be repeated for other countries. Furthermore, other benefits of efficacious and tolerable regimens, such as lower indirect costs, reduction in the incidence of extrahepatic manifestations, impact on viral transmission and improved work productivity were not considered in this model. The model does, however, reflect the best available evidence consistent with the opinions of a panel of experts in hepatology and serves as a valuable tool for evaluating the health economic outcomes associated with HCV infection therapies.
Clinical trials have demonstrated that sofosbuvir-based regimens for patients infected with HCV genotypes 1-6 resulted in high SVR rates and a favourable safety profile and tolerability as substantiated by the recommended status on the AASLD/IDSA guidelines.[27, 29, 31] In combination with the outcomes from this evaluation, the clinical attributes of sofosbuvir and its broad utility across patient populations can be projected to result in advantageous real-world economic outcomes. Further analyses need to incorporate real-world data, additional genotypes and future comparators.
In summary, these findings demonstrate that long- term sofosbuvir + pegIFN/RBV is the most cost-effective treatment option for patients infected with HCV genotype 1 because of averted liver-disease costs. Earlier initiation of the more effective sofosbuvir-based regimen yields improved health economic outcomes than later initiation, reducing advanced liver-disease complications and the downstream costs associated with advancing disease.
Materials and methods
Model overview and assumptions
A decision-analytic Markov model was developed in Microsoft Excel 2010 to estimate health and economic outcomes for sofosbuvir and comparators in patients with HCV genotype 1. The model consisted of an initial decision tree, in which patients were eligible to receive treatment, and a state-transition model to project patients' outcomes. Figure 1 shows the health-state transitions included in the model following the treatment phase. In the decision tree, patients received sofosbuvir + pegIFN/RBV or one of four comparator treatments: pegIFN/RBV, boceprevir + pegIFN2b/RBV, telaprevir + pegIFN2a/RBV or simeprevir + pegIFN2a/RBV. A comparison was also made with receiving no treatment. The decision tree divided a hypothetical cohort of 10 000 patients into two health states at the end of each treatment period on the basis of response to the therapy: SVR or chronic HCV infection. The Markov state-transition model included the health-state transitions following the decision-analytic phase. For each 1-year model cycle, patients remained in or transitioned between the following health states: SVR without cirrhosis; SVR with compensated cirrhosis; without cirrhosis; compensated cirrhosis; decompensated cirrhosis; hepatocellular carcinoma (HCC); liver transplant and death. Those patients who did not receive treatment continued to progress to other health states.
Decompensated cirrhosis encompassed a variety of serious symptomatic manifestations including ascites, variceal bleeding and hepatic encephalopathy. Patients who achieved SVR were assumed to maintain SVR and experience no further disease progression until their death. It was assumed that individuals who did not achieve SVR were at risk of progressive liver disease. Patients were assumed to complete only one course of treatment before achieving SVR; retreatment was not evaluated. In a sensitivity analysis, it was assumed that there was progression of liver disease in patients who achieved SVR but at a slower rate of deterioration than those who did not achieve SVR, or that they could be re-infected or have recurrence. The model structure, inputs and assumptions were validated by a panel of clinical hepatologists and an infectious disease specialist (Sammy Saab, Stuart Gordon, Aijaz Ahmed, Mark Sulkowski and Zobair Younossi).
We considered three representative patient cohorts with HCV genotype 1 infection: (i) patients with HCV genotype 1 infection who have never been treated (treatment-naïve) and are not HIV co-infected, (ii) patients with HCV genotype 1 infection who have been previously treated (treatment-experienced) and are not HIV co-infected and (iii) patients co-infected with HIV and HCV genotype 1 infection who have never been treated (treatment-naïve HIV co-infected). Characteristics of the modelled patient population were based on baseline characteristics of participants in sofosbuvir clinical trials.[27, 36] Eighty-three per cent of the population had no cirrhosis (METAVIR fibrosis score F0-F3) and 17% had cirrhosis (METAVIR fibrosis score F4). This is in alignment with estimates for the US: the proportion of patients infected with HCV who have cirrhosis was projected to be 25% in 2010. The mean age was 52 years and mean body weight was 79 kg. For the treatment-naïve cohort, subgroup analysis was conducted for patients with cirrhosis and without cirrhosis at the initiation of treatment.
The model incorporated clinical inputs for treatment efficacy, duration and adverse events for the sofosbuvir-based and comparator regimens. These were derived from three key sofosbuvir clinical trials[27, 38, 39] and published literature on the four comparator regimens.[13, 14, 24, 25, 27, 38-55] To date, there have been no clinical trials conducted on treatment-experienced patients infected with HCV genotype 1 and treated with sofosbuvir + pegIFN/RBV.
Therefore, the related clinical inputs were based on US Food and Drug Administration modelling, which correlated SVR rates in treatment-experienced patients with those in treatment-naïve patients, adjusting for multiple negative predictors of SVR, including high viral load, cirrhosis status and IL28B genotype (CT and TT).
The primary efficacy measure used in the model was the SVR, 12 weeks after the conclusion of treatment in the case of the sofosbuvir regimen and 24 weeks after the conclusion of treatment in the case of each of the comparators (Table 1). The measurement of SVR rates at 12 weeks and at 24 weeks has been demonstrated to be concordant for a large population of patients with HCV infection across different treatment regimens and durations.[57, 58] SVR rates were stratified according to whether patients did or did not have cirrhosis at the initiation of treatment, because SVR rates are different at various stages of cirrhosis. SVR rates for patients with and without cirrhosis at the initiation of treatment were calculated for each regimen using relative risk and weighted distribution based on clinical trials. SVR rates for genotypes 1a and 1b were derived separately and the relative risks for noncirrhotic vs. cirrhotic SVR rates were applied. The relative prevalences of genotypes 1a and 1b in the US were derived from published literature and the proportion of patients with genotype 1a and Q80K polymorphism was assumed to be 48%. Mean treatment duration for each regimen was taken from clinical trials (Table 1); it was determined using the percentage of the cohort that discontinued early for various reasons, such as adverse events, stopping rules, response-guided therapy and the time points at which these discontinuations occurred. The SVR rates and treatment duration for pegIFN2a and 2b were assumed to be the same.
The model accounted for clinically relevant grade 3/4 adverse events using data from clinical trials; reported events included nausea, vomiting, diarrhoea, pruritus, rash, thrombocytopenia, neutropenia, depression and central nervous system events. Grades 2-4 were considered for anaemia because management with epoetin-alfa and blood transfusion begins at grade 2 for some patients. For each comparator, the proportion of patients experiencing each adverse event was stratified by level of treatment experience.
The annual transition probabilities for the Markov model are presented in Figure 1 and Table 2. Transition probabilities for patients without SVR were derived from recently published literature.[37, 62-64] The transition probability for patients without cirrhosis (F0-F3) going to compensated cirrhosis (0.043) was based on a weighted average of transition rates/probabilities for a representative population. Patients (both cirrhotic and noncirrhotic) who did not achieve SVR (either treatment failures or those who did not receive treatment) were at risk of death related to liver disease once decompensation occurred. In every health state, patients were at risk of death not related to liver disease. Probabilities of HCV-related death were taken from published literature on liver-related mortality.[63, 64] For causes of death not related to HCV, inputs were based on US general population probabilities by age. For the HIV/HCV co-infected population, the transition probabilities for the compensated cirrhosis health state were estimated to be twice the values used for the mono-infected population.[66-68] Patients with HIV/HCV co-infection were not considered eligible for liver transplant. Liver transplantation for patients co-infected with HIV and HCV is not performed routinely because the 5-year survival rate is less than 50%, which is substantially below that of the acceptable threshold.[69-71]
The model accounted for four types of HCV-related cost: drug regimen, treatment monitoring, adverse events and health state (downstream liver-disease complications). All costs were in US dollars and were inflated to 2013 values using the medical care component of the consumer price index. Calculated drug regimen costs were based on indicated drug dosing, mean clinical trial therapy duration and unit drug costs (Table 2). Monitoring costs varied by treatment regimen, level of treatment experience, and cirrhosis status. Weekly costs were calculated according to specified monitoring resource use, and unit costs using the Resource-Based Relative Values Scale. Total monitoring costs over the treatment period were then aggregated from the weekly totals.
Adverse event costs were estimated based on the incidence of each event and pharmacy costs associated with their management. Pharmacy costs for the management of each adverse event in the model were based on drug-treatment algorithms and unit wholesale acquisition costs from Red Book Online. Anaemia was assumed to be treated via blood transfusion and/or erythropoietin. Table 2 presents the aggregate costs for each health state. For each health state, in-patient, out-patient, emergency department, ambulatory and pharmacy costs were accounted for. For the treatment-naïve HIV co-infected patient cohort, additional HIV-related health state costs ($18 715) were included, assuming that patients remained in the 351-500 CD4 cell count range and were on highly active anti-retroviral therapy.
The utility values assigned in the model are presented in Table 2. Each health state in the model was assigned a utility score between 0 (death) and 1 (perfect health) to quantify patient utilities for residing in that state. These were estimated from published studies reporting 36-item Short-Form Health Survey (SF-36) health-related quality of life (HRQoL) scores, which were converted to 6-dimension Short-Form Health Survey (SF-6D) utility scores.[76, 77] Utility decrements, which were specific to each treatment regimen, were assigned during treatment to account for reduced HRQoL associated with treatment-related adverse events. A utility increment of 0.05 was assigned to patients who achieved SVR.
Time horizon and discounting
Short-term (1-year) and long-term (lifetime) time horizons were modelled. Both costs and outcomes were discounted at 3.0% per year. The Markov model was terminated when the cohort reached 100 years of age.
Comparisons of health and economic outcomes were made across treatment regimens for all three cohorts. The following long-term health outcomes were estimated: new cases of compensated cirrhosis; cases of decompensated liver disease, HCC and liver transplant; and number of HCV-related deaths. The number needed to treat (NNT) was also calculated, which indicates the number of patients who would need to be treated with a specified regimen, rather than remain untreated, to avoid one negative outcome (e.g. one case of liver-disease sequelae).
The short-term health economic outcome of the model was 1-year costs per SVR. This was calculated by dividing the total per-patient costs accrued in year 1 by the SVR rate outcome to provide costs per successfully treated patient. The long-term health economic outcomes comprised: life-years (life expectancy); quality-adjusted life-years (QALYs); total lifetime costs (including costs associated with drug regimen, monitoring, adverse events and health state); and incremental lifetime cost per QALY gained [which is an incremental cost-effectiveness ratio (ICER); it indicates the cost-effectiveness of one treatment regimen over another (sofosbuvir + pegIFN/RBV vs. its five comparators)]. The incremental lifetime cost per QALY gained for sofosbuvir + pegIFN/RBV were compared against the conventional willingness-to-pay threshold of $50 000 per QALY.
For treatment-naïve cohorts, a subgroup analysis was conducted for patients with and without cirrhosis at the initiation of treatment to evaluate the impact of early treatment for HCV genotype 1 infection on health economic outcomes.
To test the robustness of the base-case results, a comprehensive deterministic sensitivity analysis was conducted for the lifetime incremental cost per QALY gained for sofosbuvir + pegIFN/RBV compared with simeprevir + pegIFN/RBV. This one-way sensitivity analysis assessed the impact of individual changes to model inputs and assumptions on the model results. Key parameters varied included SVR rates (±10% to maximum of 99%), transition probabilities (95% confidence intervals), costs for adverse event management and health states (±20%), incidence of adverse events (±50%), and utility estimates (±20%). The deterministic sensitivity analysis which compared sofosbuvir + pegIFN/RBV with simeprevir + pegIFN/RBV was also manipulated to investigate the effect of lower cirrhosis prevalence (10%), lower fibrosis progression rates (0.034 for mono-infected patients and 0.068 for HIV co-infected patients) and a range of transition probabilities for recurrence in patients who achieve SVR, with and without cirrhosis (0.01, 0.02 and 0.03). For patients with cirrhosis who achieve SVR, the costs of HCC screening (e.g. alpha-fetoprotein and liver ultrasonography) were included. A probabilistic sensitivity analysis was also conducted varying each parameter by the same ranges as used in the deterministic sensitivity analysis by the appropriate distributions: costs by gamma distribution and transition probabilities, SVR rate, and utilities by beta distribution. Drug costs in the probabilistic sensitivity analysis were varied by ±10% and SVR rates for patients with cirrhosis by ±20%. In addition, deterministic and probabilistic sensitivity analyses were performed for sofosbuvir + pegIFN/RBV compared with no treatment.
Long-term health outcomes
The projected number of cases of advanced liver-disease complications (decompensated cirrhosis, HCC, liver transplant, and HCV-related death) per treated cohort of 10 000 patients is presented in Table 3. Of all the treatment regimens assessed, sofosbuvir + pegIFN/RBV was associated with the lowest number of all types of advanced liver-disease complications, even in difficult-to-treat patients. For example, in the treatment-experienced patient subgroup, the incidence of HCC was 469 per 10 000 patients in the cohort treated with sofosbuvir + pegIFN/RBV, but ranged from 700 to 1299 across the four treatment comparators. Table 3 also presents the percentage reduction in advanced liver-disease complications for patients receiving sofosbuvir + pegIFN/RBV compared with the other regimens. Reductions in the incidence of liver-disease complications with sofosbuvir + pegIFN/RBV ranged from 64% to 82%, 50% to 68%, 43% to 58% and 33% to 56% compared with pegIFN/RBV, boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV, respectively. In difficult-to-treat patients, reductions in liver-disease complications with sofosbuvir + pegIFN/RBV compared with current treatment were substantial: 33-64% in treatment-experienced patients and 50-82% in patients co-infected with HIV. Compared with no treatment, the reductions in advanced liver-disease complications ranged from 69% to 87% when treating with sofosbuvir + pegIFN/RBV.
The NNT outcomes can be found in Table S1. The NNT to avoid a liver transplant in treatment-naïve and treatment-experienced patients was 84 and 106, respectively, for the sofosbuvir + pegIFN/RBV regimen, but ranged from 145 to 547, 120 to 194, 105 to 162 and 104 to 137 for pegIFN/RBV, boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV, respectively.
Short-term health economic outcomes
The 1-year cost per SVR was lower for sofosbuvir + pegIFN/RBV than telaprevir-based, boceprevir-based, or simeprevir-based regimens in all the patient subpopulations, including difficult-to-treat patients (Figure 2). The cost per SVR for sofosbuvir + pegIFN/RBV was 8-15% ($9882-$20 576) lower in treatment-naïve patients, 13-30% ($20 038-$57 266) lower in patients co-infected with HIV, and 26-38% ($50 734-$88 964) lower in treatment-experienced patients than the cost per SVR for boceprevir + pegIFN/RB, telaprevir + pegIFN/RBV, and simeprevir + pegIFN/RBV, respectively. When compared with pegIFN/RBV alone, sofosbuvir + pegIFN/RBV was associated with lower 1-year SVR costs in treatment-experienced patients (55% lower than pegIFN2a/RBV and 52% lower than pegIFN2b/RBV), treatment-experienced patients co-infected with HIV (43% lower than pegIFN2a/RBV and 40% lower than pegIFN2b/RBV) and patients with cirrhosis (14% lower than pegIFN2a/RBV and 9% lower than pegIFN2b/RBV). However, the 1-year costs per SVR were lower with pegIFN/RBV compared with sofosbuvir in treatment-naïve patients (22% lower with pegIFN2a/RBV and 27% lower with PegIFN2b/RBV) and in patients without cirrhosis (27% lower with pegIFN2a/RBV and 32% lower with pegIFN2b/RBV).
In the subgroup analysis of outcomes in patients with and without cirrhosis, 1-year cost per SVR, across all treatments analysed in the model, was 15-47% ($19 444-$72 313) lower in patients who initiated treatment at the noncirrhotic stage compared with those whose treatment was initiated at the cirrhotic stage.
Long-term health economic outcomes
Table 4 presents the lifetime health economic outcomes for patients infected with HCV genotype 1 by patient population. In the model, treatment-naïve patients treated with sofosbuvir + pegIFN/RBV lived 0.92, 0.63, 0.41 and 0.41 years longer than those treated with pegIFN/RBV alone, boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV, respectively. Treatment-experienced patients receiving sofosbuvir + pegIFN/RBV lived 1.35, 0.74, 0.61 and 0.37 years longer than patients receiving pegIFN/RBV alone, boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV, respectively.
Sofosbuvir + pegIFN/RBV treatment also provided patients with the most QALYs compared with the other regimens in all patient populations analysed (Table 4). Treatment-naïve patients treated with sofosbuvir + pegIFN/RBV gained 1.10, 0.82, 0.53 and 0.50 more QALYs than patients treated with pegIFN/RBV alone, boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV, and simeprevir + pegIFN/RBV, respectively. Treatment-experienced patients treated with sofosbuvir + pegIFN/RBV gained 1.67, 0.96, 0.73 and 0.47 more QALYs than patients treated with pegIFN/RBV alone, boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV, respectively.
For treatment-naïve patients, the total lifetime healthcare costs of patients treated with the sofosbuvir + pegIFN/RBV regimen ($132 560) were lower than those of patients treated with the boceprevir + pegIFN/RBV ($136 973), telaprevir + pegIFN/RBV ($143 009), and simeprevir + pegIFN/RBV ($135 183) regimens. The lowest total costs, however, were associated with the pegIFN/RBV regimens ($105 299-$108 504). In treatment-experienced patients, the total costs of the sofosbuvir + pegIFN/RBV regimen ($148 812) were also lower than those of the boceprevir + pegIFN/RBV ($165 983), telaprevir + pegIFN/RBV ($165 428) and simeprevir + pegIFN/RBV ($168 251) regimens. In treatment-experienced patients, the lowest total costs were associated with the pegIFN/RBV regimens ($141 647-$145 009; Table 4).
Sofosbuvir + pegIFN/RBV dominated (i.e. was less costly and more effective than) the boceprevir + pegIFN/RBV, telaprevir + pegIFN/RBV and simeprevir + pegIFN/RBV regimens in most of the cases analysed. Compared with pegIFN/RBV, sofosbuvir + pegIFN/RBV had incremental costs per QALY gained ranging from $2277 to $4290 in treatment-experienced patients, and from $21 869 to $24 783 in treatment-naïve patients, which compare favourably to the conventional $50 000 per QALY willingness-to-pay threshold. Treatment with sofosbuvir was also cost-effective compared with no treatment, with ICERs ranging from dominance to $16 617 per QALY (also below the $50 000 per QALY willingness-to-pay threshold), depending on the patient subgroup (Table 4).
In the subgroup analysis, which compared patients by disease severity at the initiation of treatment, total costs were 38-46% ($74 963-$93 208) lower among patients initiating treatment before the development of cirrhosis than among those with compensated cirrhosis. This result stems from higher SVR rates among patients without cirrhosis and the resulting costs avoided via averted cases of liver-disease complications (Table 4).
Figure 3 presents results of the deterministic sensitivity analysis for sofosbuvir + pegIFN/RBV compared with simeprevir + pegIFN/RBV among treatment-naïve patients. It displays the 20 factors that most heavily influenced the ICER. The ICERs ranged from -$13 093 to $16 970. The ICERs were negative for all values of key parameters that were varied for sofosbuvir compared with simeprevir (i.e. sofosbuvir + pegIFN/RBV dominated simeprevir + pegIFN/RBV). In general, reductions in cirrhosis prevalence and fibrosis rates, and varying transition probabilities to recurrence in patients who achieve SVR, with and without cirrhosis (0.01, 0.02 and 0.03), increased the incremental cost per QALY gained (Table S2). However, these outcomes still supported sofosbuvir being the most cost-effective option overall. The probabilistic sensitivity analysis results (Figure 4) show that the probability of sofosbuvir + pegIFN/RBV being cost-effective vs. simeprevir + pegIFN/RBV was 96.8% at the $50 000 willingness-to-pay level.
Deterministic and probabilistic sensitivity analyses for sofosbuvir + pegIFN/RBV compared with no treatment are presented in Figures S1 and S2, respectively. The outcomes from these analyses also support sofosbuvir + pegIFN/RBV being the most cost-effective option. The probability of sofosbuvir + pegIFN/RBV being cost-effective vs. no treatment was 100% at the $50 000 willingness-to-pay level (Figure S2).