Cost-Effectiveness of Novel Regimens for the Treatment of Hepatitis C Virus
Download the PDF here
17 March 2015 Ann Intern Med.
Mehdi Najafzadeh, PhD; Karin Andersson, MD; William H. Shrank, MD, MSHS; Alexis A. Krumme, MS; Olga S. Matlin, PhD; Troyen Brennan, MD, JD, MPH; Jerry Avorn, MD; and Niteesh K. Choudhry, MD, PhD
Background: New regimens for hepatitis C virus (HCV) have shorter treatment durations and increased rates of sustained virologic response compared with existing therapies but are extremely expensive.
Objective: To evaluate the cost-effectiveness of these treatments under different assumptions about their price and efficacy.
Design: Discrete-event simulation.
Data Sources: Published literature.
Target Population: Treatment-naive patients infected with chronic HCV genotype 1, 2, or 3.
Time Horizon: Lifetime.
Intervention: Usual care (boceprevir-ribavirin-pegylated interferon [PEG]) was compared with sofosbuvir-ribavirin-PEG and 3 PEG-free regimens:
sofosbuvir-simeprevir, sofosbuvir-daclatasvir, and sofosbuvir-ledipasvir. For genotypes 2 and 3, usual care (ribavirin-PEG) was compared with sofosbuvir-ribavirin, sofosbuvir-daclatasvir, and sofosbuvir-ledipasvir-ribavirin (genotype 3 only).
Outcome Measures: Discounted costs (in 2014 U.S. dollars), quality-adjusted life-years (QALYs), and incremental cost-effectiveness ratios.
Results of Base-Case Analysis: Assuming sofosbuvir, simeprevir, daclatasvir, and ledipasvir cost $7000, $5500, $5500, and $875 per week, respectively, sofosbuvir-ledipasvir was cost-effective for genotype 1 and cost $12 825 more per QALY than usual care. For genotype 2, sofosbuvir-ribavirin and sofosbuvir-daclatasvir cost $110 000 and $691 000 per QALY, respectively. For genotype 3, sofosbuvir-ledipasvir-ribavirin cost $73 000 per QALY, sofosbuvir-ribavirin was more costly and less effective than usual care, and sofosbuvir-daclatasvir cost more than $396 000 per QALY at assumed prices.
Results of Sensitivity Analysis: Sofosbuvir-ledipasvir was the optimal strategy in most simulations for genotype 1 and would be cost-saving if sofosbuvir cost less than $5500. For genotype 2, sofosbuvir-ribavirin-PEG would be cost-saving if sofosbuvir cost less than $2250 per week. For genotype 3, sofosbuvir-ledipasvir-ribavirin would be cost-saving if sofosbuvir cost less than $1500 per week.
Limitation: Data are lacking on real-world effectiveness of new treatments and some prices.
Conclusion: From a societal perspective, novel treatments for HCV are cost-effective compared with usual care for genotype 1 and probably genotype 3 but not for genotype 2.
Primary Funding Source: CVS Health.
· Newer regimens to treat hepatitis C virus (HCV) seem efficacious but are extremely expensive.
· The cost-effectiveness of standard HCV regimens was compared with newer regimens containing sofosbuvir for each HCV genotype. Newer regimens were cost-effective for genotype 1 and probably genotype 3 but were not cost-effective for genotype 2. Some regimens were cost-saving with sufficient reduction in the cost of sofosbuvir.
· Data came from clinical trials rather than usual practice.
· If prices of newer hepatitis C drugs were reduced, new HCV treatments may not only be cost-effective but may also reduce the cost of HCV treatment over the long term.
Drug therapies for hepatitis C virus (HCV) infection have been available for more than a decade. Despite approval of the protease inhibitors boceprevir and telaprevir in 2011, which have substantially increased rates of sustained virologic response (SVR) for patients infected with HCV genotype 1, many patients do not complete recommended treatment owing to shortcomings of pegylated interferon (PEG) (1).
Several new regimens may represent important improvements over current HCV treatments. The once-daily nucleotide polymerase inhibitor sofosbuvir (Sovaldi, Gilead Sciences) was approved in December 2013 (2) to be used in combination with ribavirin and PEG in treatment-naive patients infected with HCV genotypes 1 and 4 and with ribavirin alone in patients infected with HCV genotypes 2 and 3. Sofosbuvir can achieve higher SVR rates in substantially shorter treatment times than existing regimens (3-6). Shorter treatment durations and higher SVR rates, even among nonresponders, also seem possible with other PEG-free regimens consisting of sofosbuvir in combination with simeprevir (7-8), daclatasvir (9), or ledipasvir (10-13).
Despite their promise, these novel therapies are very expensive and, considering that more than 3 million patients (14) may be eligible for these therapies, the budgetary implications have generated widespread concern (15-16). Little is known about the relative societal health benefit and value of the new treatments for hepatitis C compared with current options. Therefore, we conducted a cost-effectiveness analysis to evaluate the balance between health benefit and health care expenditures for these treatments under different assumptions about their price and efficacy.
Novel treatments for HCV substantially shorten treatment length, achieve substantially higher SVR rates, and offer interferon-free regimens for patients who cannot tolerate interferon or do not adequately respond to current medications. However, little is known about their economic value compared with the current care. Our study investigated the assumptions under which the new regimens would or would not be considered cost-effective strategies.
From a societal perspective, the newly approved PEG-free regimen of sofosbuvir-ledipasvir for 12 weeks could be very cost-effective relative to usual care (costing $12 825 per QALY gained) for patients with HCV genotype 1. In probabilistic sensitivity analysis, sofosbuvir-ledipasvir seems to be the optimal treatment strategy in the greatest number of simulations. Other treatment regimens for patients with HCV genotype 1 also provide relatively good value in our base-case models but were less effective and more costly than sofosbuvir-ledipasvir and rarely optimal in probabilistic analysis. For patients with genotype 2, the newly approved regimen of sofosbuvir-ribavirin cost approximately $110 168 per QALY gained compared with usual care. For patients with genotype 3, sofosbuvir-ledipasvir-ribavirin for 12 weeks cost approximately $73 000 per QALY gained compared with usual care, which represents relatively good value.
Although we found the sofosbuvir-ribavirin-PEG and sofosbuvir-ledipasvir regimens for genotype 1 and potentially sofosbuvir-ledipasvir-ribavirin for genotype 3 to be cost-effective at their currently assumed prices ($7000 per week for sofosbuvir and an additional $875 per week for ledipasvir), the offset savings from avoiding complications related to HCV do not outweigh the cost of the drugs themselves; thus, these strategies do not seem to actually reduce overall spending. This is an exceptionally high bar, however, that is generally not expected when evaluating whether a new strategy represents good value for money. Nevertheless, in our sensitivity analysis, we found that if the price of sofosbuvir was less than $5500 per week, a regimen of sofosbuvir-ledipasvir could actually be cost-saving for genotype 1. Similarly, if the cost of sofosbuvir was less than $4500 per week, sofosbuvir-ribavirin-PEG could be cost-saving. In contrast, for genotype 2, the newly approved regimen of sofosbuvir-ribavirin would be cost-saving only if sofosbuvir cost less than $2250 per week. For genotype 3, sofosbuvir-ledipasvir-ribavirin would be cost-saving if sofosbuvir cost less than $1500 per week. This reflects the relative limitations of new therapies for treatment of genotypes 2 and 3. When compared with no treatment, sofosbuvir-ribavirin for genotype 2 and sofosbuvir-ledipasvir-ribavirin for genotype 3 resulted in incremental cost-effectiveness ratios of $45 344 and $27 950 per QALY gained, respectively. This is particularly relevant to patients for whom a usual PEG-based regimen is not an option.
An analysis such as ours can measure the additional cost required to achieve an incremental QALY from a societal perspective, but it does not account for the effect of such expenditures on near-term health care budgets nor the fact that the organizations that will pay for these drugs in the near term may not be the ones to primarily benefit from their downstream effects. Regardless of the cost-effectiveness of novel hepatitis C treatments, there is considerable concern that their very high prices could substantially increase short-term overall drug spending for many public and private payers. Further, unpleasant PEG-based regimens are not prescribed to all patients infected with early-stage HCV and HCV screening has not been a routine practice. The far greater tolerability of the newer drugs could mean that HCV screening will become more common, with resultant increases in the demand for these agents.
In the face of this economic reality, payers may give priority to patients for whom treatment is most cost-effective. For example, our analysis suggests that all treatment strategies for genotype 1 were more economically attractive in patients with higher fibrosis stages and in those who were younger at the time of treatment initiation. However, the ethics of resource allocation based on such criteria are far from clear.
Our analysis has several limitations. Our model focused on treatment-naive patients; however, new regimens have also been shown to be effective in nonresponders and patients with relapse after initial SVR. We did not directly include parameters for nonadherence in our model. But overall SVR rates implicitly reflect patients' adherence to response-guided treatments, at least in the clinical trial setting. More detailed modeling of adherence would probably bias the results against usual care and therefore indicate that the new drugs are more cost-effective than is reflected in our trial-based analyses. We did not include other factors, such as insurance coverage and geographic variations, that could influence access to health care. We based our analysis on wholesale acquisition costs, but in reality, prices negotiated between manufacturers and large insurers are often lower than these prices. Because this information is not publicly available, the results of our sensitivity analyses can serve as a tool for understanding the financial impact of lower prices on the economic value of new treatments.
We restricted our analysis to direct medical costs and did not consider the effect of treatment on indirect costs (that is, productivity loss), nonmedical costs (for example, resources spent by patients to seek medical care), or costs accrued from prolonged life expectancy. Our model cannot incorporate all elements of clinical decision making, such as patient preference for oral therapy or risk for decompensation with PEG. We have not accounted for the effect of SVR on transmission rates, although inclusion would probably make the cost-effectiveness results more attractive for treatments with higher rates of SVR. Finally, more regimens for HCV treatment are expected to be approved (53); greater competition may lead to reduced therapy costs, which would alter our findings and probably increase the value of treating hepatitis C over time.
In summary, our analysis suggests that from a societal perspective, sofosbuvir-based treatment regimens seem to represent good long-term economic value in treatment-naive patients with HCV genotypes 1 and potentially genotype 3 but not for those with genotype 2. If these drugs became available at lower prices, they could not only improve health outcomes but also reduce long-term health care costs.
We developed a discrete-event simulation (DES) model using Arena, version 12.00 (Rockwell Automation), to simulate the natural history and progression of liver disease among treatment-naive patients infected with chronic HCV genotype 1, 2, or 3 and compare clinical and economic outcomes of treatment strategies (Figure 1) (17-20). Our modeling approach and assumptions have been explained in the Appendix.
Treatment Strategies and Efficacy Assumptions
Treatment strategies for each HCV genotype were defined in accordance with current clinical guidelines and proposed indications for new drugs (Supplement Figure 1). On the basis of recent clinical trials, we considered 5 treatment strategies for patients infected with HCV genotype 1: usual care consisting of response-guided triple therapy using boceprevir-ribavirin-PEG for 28 to 48 weeks; newly approved triple therapy using sofosbuvir-ribavirin-PEG for 12 weeks; and 12-week PEG-free regimens using sofosbuvir-simeprevir, sofosbuvir-daclatasvir, or sofosbuvir-ledipasvir.
For patients infected with HCV genotype 2, we evaluated 3 treatment strategies: usual care consisting of dual therapy with ribavirin-PEG for 24 weeks, the newly approved PEG-free regimen using sofosbuvir-ribavirin for 12 weeks, and a PEG-free regimen using sofosbuvir-daclatasvir for 12 weeks.
For patients infected with HCV genotype 3, we evaluated 4 treatment strategies: usual care consisting of dual therapy with ribavirin-PEG for 24 weeks and PEG-free regimens using sofosbuvir-ribavirin for 24 weeks, sofosbuvir-daclatasvir for 12 weeks, and sofosbuvir-ledipasvir-ribavirin for 12 weeks.
We modeled treatment efficacy based on SVR, which was defined as an HCV RNA level below the lower limit of quantification measured at 12 weeks after the end of treatment (3).
The SVR rates from different treatment strategies were derived from the results of published clinical trials (Table 1) (3, 6-13, 21-25). We also assumed that alcohol use negatively affects SVR rates in the base-case analysis and varied this in the sensitivity analysis (Table 1) (27-28).
The assumptions that defined the natural history of the disease in our model are presented in Table 1. We assumed progression of liver disease to be a function of patient-level variables. In accordance with the observed disease progression rates in 3 population-based cohorts of patients with HCV, we assumed that the change in the Meta-analysis of Histologic Data in Viral Hepatitis (METAVIR) score (42) was larger in patients with higher levels of daily alcohol consumption and men (30). We calibrated the model to emulate observed progression rates in community-based cohorts (Appendix Figure 1) (31-32). Annual transition rates from compensated to decompensated cirrhosis or hepatocellular carcinoma were based on natural history models that have been empirically calibrated to epidemiologic data on HCV infection seroprevalence and liver cancer mortality (34-35).
We modeled increased rates of mortality not related to liver disease among patients with HCV using sex- and race-dependent hazard ratios based on the results of NHANES III (Third National Health and Nutrition Examination Survey) (33). Background mortality rates stratified by sex and race were derived from U.S. life tables published as part of National Vital Statistics Reports (36). We assumed that a proportion of patients with decompensated cirrhosis or hepatocellular carcinoma received a liver transplant based on Model for End-Stage Liver Disease criteria (43). A summary of data sources has been provided in the Supplement Table.
Health-related quality-of-life weights associated with each health state in the model were derived from the peer-reviewed literature (Table 1) (33, 38). Age-specific baseline quality-of-life estimates were based on results from the Medical Expenditure Panel Survey (37). We assumed that there was no HCV-related residual effect on quality of life after achieving SVR, and the negative effect of PEG on quality of life was proportional to the length of interferon treatment.
Annual costs of HCV treatment at different stages of disease, cost of treatment drugs, and cost of treatment-related side effects were included in the model.
Only direct medical costs were included; indirect costs due to productivity loss or nonmedical costs (for example, cost of seeking medical care) were not included. Annual costs associated with different stages of HCV were derived from previous cost-effectiveness and observational studies comparing the medical costs of patients with HCV at different stages of disease with control patients without HCV (33, 39-40, 44-48). The costs of sofosbuvir ($7000 per week), simeprevir ($5500 per week), and ledipasvir ($875 per week) are based on the wholesale acquisition costs (2, 41). Because daclatasvir has not yet been approved for use in the United States and no pricing information is available, we assumed it had the same cost as simeprevir. Whenever necessary, we adjusted unit costs for inflation by using the U.S. Consumer Price Index to reflect 2014 U.S. dollars (49).
We conducted 1-way sensitivity analyses by changing all of our input parameters one at a time across their possible ranges (Table 1). We then performed a probabilistic sensitivity analysis (50) by varying all of our model parameters simultaneously. For this purpose, we sampled 10 000 independent sets of input parameters from their probability distributions; for each set of parameters, we modeled a cohort of 10 000 hypothetical patients per treatment strategy (51-52). The results of these probabilistic sensitivity analyses are reported using incremental cost-effectiveness acceptability curves that reflect the probability of each treatment strategy having the highest net monetary benefit at various willingness-to-pay thresholds. We also compared the novel regimens with no treatment, as well as dual therapy with RBV and PEG alone.
Role of the Funding Source
The funding source had no role in the design, conduct, or reporting of this analysis, or in the decision to submit the manuscript for publication.
HCV Genotype 1
The results of our base-case analysis comparing usual care (boceprevir-ribavirin-PEG) with the new regimens are presented in Table 2, and cost-effectiveness frontiers are shown in Figure 2. With usual care, the average quality-adjusted life expectancy was 11.28 QALYs, and patients incurred average lifetime costs of $100 926. Treatment with sofosbuvir-ribavirin-PEG increased quality-adjusted life expectancy to 12.19 QALYs and was more costly, with average lifetime costs of $120 648 (Table 2). As a result, the incremental cost-effectiveness ratio of sofosbuvir-ribavirin-PEG compared with usual care was $21 528 per QALY gained.
All 3 PEG-free regimens achieved higher quality-adjusted survival than usual care (12.26, 12.36, and 12.40 QALYs for sofosbuvir-simeprevir, sofosbuvir-daclatasvir, and sofosbuvir-ledipasvir, respectively). But they were also more costly, with lifetime expenditures of $171 023, $169 747, and $115 358, respectively (Table 2). Assuming sofosbuvir-ledipasvir cost $7875 per week, this regimen results in the largest QALY gain and smallest incremental cost compared with usual care, with an incremental cost-effectiveness ratio of $12 825 per QALY gained.
Compared with no treatment, the costs per additional QALY gained were $30 001, $50 951, $48 206, and $25 291 for sofosbuvir-ribavirin-PEG, sofosbuvir-simeprevir, sofosbuvir-daclatasvir, and sofosbuvir-ledipasvir, respectively (Appendix Tables 1 and 2). The costs per additional QALY gained were $35 836 for usual care (boceprevir-ribavirin-PEG) and $24 833 for dual therapy (ribavirin-PEG) compared with no treatment.
Because the pricing of daclatasvir has not been determined and that of sofosbuvir, simeprevir, and ledipasvir could change as new drugs enter the marketplace, we varied the weekly cost of all 4 drugs between $500 and $9500 and examined the thresholds at which each treatment strategy offered the largest net monetary benefit (Figure 3, top). At a conventional willingness-to-pay threshold of $50 000 per QALY gained, sofosbuvir-ledipasvir was the optimal strategy, but if ledipasvir cost more than $2400 per week, sofosbuvir-ribavirin-PEG would be optimal (Figure 3, top). Changing the cost of simeprevir across a range from $500 to $9500 did not affect which regimen was optimal, but if daclatasvir had a weekly price lower than $700, sofosbuvir-daclatasvir would become optimal. Sofosbuvir-ledipasvir would be cost-saving if sofosbuvir cost less than $5500 per week (Supplement Figure 2). The sofosbuvir-ribavirin-PEG regimen would be cost-saving if sofosbuvir cost less than $4500 per week.
The assumed SVR rates from different treatments also influenced their value. Sofosbuvir-ledipasvir was the optimal strategy unless its SVR rate was less than 87% or if the SVR rate of sofosbuvir-ribavirin-PEG exceeded 99% (Figure 4, top). In those cases, sofosbuvir-ribavirin-PEG was the optimal strategy (Appendix Figure 2).
Incremental cost-effectiveness ratios were also sensitive to patient characteristics, including fibrosis stage and age. Overall, all treatment strategies were more economically attractive in patients with higher fibrosis stages (Appendix Figure 3) and in those who were younger at the time of treatment initiation (Appendix Figure 4). With the exception of annual discount rate, utility weight for stage F4 fibrosis, hazard ratio of non-liver-related death, and the costs associated with stages F0 to F3 fibrosis, other parameters had very small effects on our results (Supplement Figure 3).
The results of probabilistic sensitivity analysis suggested that at a willingness-to-pay threshold of $13 000 per QALY, usual care and sofosbuvir-ledipasvir were equally optimal strategies (Appendix Figures 5 and 6). Above this threshold, sofosbuvir-ledipasvir had a higher likelihood of being the optimal strategy.
HCV Genotype 2
With usual care (RBV-PEG), quality-adjusted survival was 11.86 QALYs and lifetime expenditures were $54 005 (Table 2). Sofosbuvir-ribavirin increased expected survival to 12.37 QALYs, with costs of $109 958, which resulted in an incremental cost-effectiveness ratio of $110 168 per QALY gained (Figure 2, middle). Treatment with sofosbuvir-daclatasvir did not offer any advantage over sofosbuvir-ribavirin because it resulted in smaller quality-adjusted survival (12.24 QALYs) and was more expensive ($316 845).
Compared with no treatment, sofosbuvir-ribavirin and sofosbuvir-daclatasvir resulted in incremental cost-effectiveness ratios of $45 344 and $137 973 per QALY gained, respectively (Appendix Table 3).
At a willingness-to-pay threshold of $50 000 per QALY, usual care was the optimal strategy for genotype 2, but if sofosbuvir cost less than $4500 per week (Figure 3, middle), sofosbuvir-ribavirin would be optimal. Sofosbuvir-ribavirin would be cost-saving if sofosbuvir cost less than $2250 per week. Sofosbuvir-daclatasvir was not cost-saving at any prices of sofosbuvir and daclatasvir analyzed (Supplement Figure 4).
Usual care also remained the optimal strategy in these patients when SVR rates of PEG-free regimens varied between 80% and 100% (Figure 4, middle). Treatment of patients at more severe stages of fibrosis (Appendix Figure 3) and at younger ages (Appendix Figure 4) was relatively more cost-effective. Annual discount rate, utility weight for fibrosis stages, the risk for non-liver-related death, and disutility due to PEG-based regimens had larger effects on incremental cost-effectiveness ratios than other model parameters (Supplement Figure 5).
For genotype 2, our probabilistic sensitivity analysis found that sofosbuvir-ribavirin is most likely to be the optimal strategy at a willingness-to-pay threshold of $110 000 per QALY or higher (Appendix Figures 5 and 6).
HCV Genotype 3
Usual care (RBV-PEG) produced quality-adjusted survival of 11.50 QALYs, with lifetime costs of $58 323 (Table 2). Sofosbuvir-ribavirin would lead to lower quality-adjusted survival (11.37 QALYs), with very large costs ($207 872) (Figure 2, bottom). Although sofosbuvir-daclatasvir increased expected survival to 12.16 QALYs, it was still very expensive ($317 830) and resulted in an incremental cost-effectiveness ratio of $396 229 per QALY gained. The sofosbuvir-ledipasvir-ribavirin regimen increased expected survival to 12.35 QALYs, with a lifetime cost of $120 464, and resulted in an incremental cost-effectiveness ratio of $72 236 per QALY gained.
Compared with no treatment, sofosbuvir-ribavirin costs $108 443, sofosbuvir-daclatasvir costs $119 664, and sofosbuvir-ledipasvir-ribavirin costs $27 950 for each additional QALY gained (Appendix Table 3).
At a willingness-to-pay threshold of $50 000 per QALY, usual care was the optimal strategy for patients with genotype 3, but if sofosbuvir cost less than $5500 per week (Figure 3, bottom), sofosbuvir-ledipasvir-ribavirin would be optimal. Usual care remained the optimal strategy for SVR rates between 50% and 100% (Figure 4, bottom). Sofosbuvir-ledipasvir-ribavirin would be cost-saving if sofosbuvir cost less than $1500 per week (Supplement Figure 6).
Sofosbuvir-ledipasvir-ribavirin and sofosbuvir-daclatasvir were more cost-effective in patients at more severe stages of fibrosis (Appendix Figure 3) and in those who were younger (Appendix Figure 4). Compared with usual care, however, the incremental cost-effectiveness ratio of sofosbuvir-daclatasvir remained above approximately $200 000 per QALY and the newly approved sofosbuvir-ribavirin resulted in fewer QALYs and higher costs than usual care in all situations. As with genotype 2, annual discount rate, utility weight for fibrosis stages, hazard ratio of non-liver-related death, and disutility associated with PEG-based regimens had a larger effect on incremental cost-effectiveness ratios (Supplement Figure 7). In the probabilistic sensitivity analysis, the willingness-to-pay threshold had to be greater than $75 000 per additional QALY gained for sofosbuvir-ledipasvir-ribavirin to be the optimal strategy over usual care in most simulations (Appendix Figures 5 and 6).