Clinical Infectious Diseases Advance Access published March 16, 2015 1  The Cost-effectiveness, Health Benefits, and Financial Costs of New Antiviral Treatments for Hepatitis C Virus t p David B. Rein1, John S. Wittenborn1, Bryce D. Smith2, Danielle K. Liffmann1, John W. Ward2 i r 1NORC at the University of Chicago, Public Health Department. 3520 Piedmont Rd. N.E., Suite c 225, Atlanta, GA 30305 s 2U.S. Centers for Disease Control and Prevention, Division of Viral Hepatitis, 1600 Clifton Road, Atlanta, GA 30333 u Corresponding Author: David B. Rein, NORC at the University of Chicago, Public Health D o w Department, 3520 Piedmont Rd. N.E., Suite 225, Atlanta, GAn 30305, 404-240-8402, rein‐ nlo a [email protected]  de d a fro m h Alternate Corresponding Author: John S Wittenborn, NORC at the University of Chicago, Public ttp Health Department, 55 East Monroe Street, 30th FlMoor, Chicago, IL 60603, 312-519-5718, ://cid .o wittenborn‐[email protected] xfo rdjo u rn a Summary: Compared to previous treatments for hepatitis C, new treatments provide much ls.o d rg higher cure rates with far fewer side-effects and contraindications. This paper estimates the cost- b/ y J effectiveness and financial impact of new hepatitis C medications as compared to treatments of u le the past. e s L e v in o t n M a p rc h Source of support: National Foundation for the Centers for Disease Control and Prevention, Inc. 1 6 , 2 MOU # 527-11 SC 0 e 15 c c A © The Author 2015. Published by Oxford University Press on behalf of the Infectious Diseases Society of  America. All rights reserved. For Permissions, please e‐mail: [email protected]. 2  Abstract  Background: New hepatitis C virus (HCV) treatments deliver higher cure rates with t fewer contraindications increasing demand for treatment and health care costs. The copst- effectiveness of new treatments is unknown. i r Methods: We conducted a microsimulation of guideline testing followed by alternative c treatment regimens for HCV among the U.S. population aged 20 and older to estimate cases s identified, treated, sustained viral response (SVR); deaths; medical costs; quality-adjusted life u years (QALYs); and the incremental cost-effectiveness ratio (ICER) of different treatment D o w n nlo options expressed as discounted lifetime costs and benefits from the healthcare perspective. ad e d a fro m Results: Compared to treatment with pegylated interferon, ribavirin (PR), and a protease h ttp inhibitor (PI) for HCV genotype (G) 1 and PR alonMe for G2/3, treatment with PR and Sofosbuvir ://cid .o x fo (PRS) for G1/4 and treatment with Sofosbuvir and ribavirin (SR) for G2/3 increased QALYs by rdjo u rn a ls 555,226, reduced deaths by 80,682, at an incremental cost of $26.2 billion, and an ICER of .o d rg b/ y $47,304 per QALY gained. As compared to PRS/SR, treating with an all oral regimen of Ju le e s L e Sofosbuvir and Simeprevir (SS) for G1/4 and SR for G2/3, increased QALYs by 1,110,451 and vin o n t M reduced deaths by an additiponal 164,540 at an incremental cost of $80.1 billion and an ICER of arc h 1 6 $72,169. In sensitivity analysis, where treatment with SS effectiveness was set to the list price of , 2 0 e 15 Viekira Pak™ and then Harvoni™, treatment cost $24,921 and $25,405 per QALY gained as c compared to SS/SR. c Conclusions: New treatments are cost-effectiveness per person treated but pent up A demand for treatment may create challenges for financing. 3  In 2012, the US Centers for Disease Control and Prevention (CDC) recommended that Americans born during 1945-1965 receive a one-time antibody test to identify hepatitis C virus t p (HCV) infection (birth-cohort testing). [1-3] In 2013, this recommendation was affirmed by the United States Preventive Services Task Force (USPSTF) citing the large health beinefits of birth- r cohort testing predicted by modeling studies.[2-6] From 2011-2013, at least six published studies c found HCV testing and treatment to be cost-effectiveness, using different parameters and s assumption.[2, 4, 6-9] Adjusting the aggregate results from these studies into per person u D incremental costs and quality adjusted life years (QALYs) allows for the visual comparison of o w n nlo a their results (Figure 1). de d a fro m h Since publication of the birth-cohort testing recommendations, new highly effective ttp M ://cid .o drugs have been released and clinical treatment recommendations have been updated to x fo rd jo u incorporate their use.[10] In this paper, we modified a previously published model of the cost- rn a ls .o d rg effectiveness of birth-cohort testing to assess the cost-effectiveness, financial impacts, and health b/ y J u le benefits of birth-cohort testing usineg new treatments under the assumption of broad population- s L e v in based implementation.[2] on t M a p rc h 1 6 METHODS , 2 0 e 15 Decision Analytic Model c We programmed (Microsoft Visual Studio 2010, Redmond, WA) a Monte Carlo simulation c model of the natural history of hepatitis C with antibody prevalence estimates stratified by age, A gender, race/ethnicity, and history of injecting drugs. The model’s natural history, validation, and economic parameters have been previously described, and revisions to the model’s parameters are included in Tables 1a and 1b and technical documentation.[2, 11] Compared to previous 4  versions, the model’s structure now assumes that a sustained viral response (SVR) results in a reduced risk of hepatocellular carcinoma (HCC) instead of risk elimination. t p Model Cohorts We modeled the U.S. population aged 20 or older, totaling 229,185,985 ini 2012.[57] We r stratified the population based on age, sex, and lifetime risk of injecting drugs.[58] We further c stratified these cohorts into those with and without antibody to HCV (based on year of birth), and s those with antibodies into those with chronic (78%) and cleared (22%) infections.[59] We u D assumed 25% of chronically infected patients were not interested in treatment or were not o w n nlo a reachable by the health care system, and assumed the remainder would be offered testing.[60-63] de d a fro m We estimated starting fibrosis rates using data from biopsy results of newly diagnosed h ttp patients observed in the retrospective component oMf the Birth-cohort Evaluation to Advance ://cid .o x fo rd Screening and Testing for Hepatitis C (BEST-C) study.[64] We used census life tables to jo u rn a ls calculate the annual probability of mortality from non-hepatic causes and assigned a relative risk .o d rg b/ y J of mortality of 1.42 for individuals who reported ever injecting drugs.[2, 65] u le e s L e v Screening and Treatment Scenarios in o n t M For the purpose of opur simulation, we assumed that 18.5% of those outside the 1945 to arc h 1 6 1965 birth-cohort would be offered testing and that 100% of those in the birth-cohort would be , 20 e 15 offered testing if they could be reached through the health system. Of those who accepted testing c and tested positive for HCV RNA, we compared the cost-effectiveness and health impacts of 5 c treatment alternatives; (1) No treatment (NT); (2) Pegylated interferon and ribavirin (PR) for 48 A weeks for genotypes 1 and 4, and for 24 weeks for genotypes 2 and 3; (3) PR for 24 weeks plus an additional protease inhibitor (PRPI) for 12 weeks for genotypes 1 and 4 or PR for 24 weeks for genotypes 2 and 3; (4) PR plus Sofosbuvir (PRS) for 12 weeks for genotypes 1 and 4, 5  Sofosbuvir plus ribavirin (SR) for 12 weeks for genotype 2, and SR for 24 weeks for genotype 3; or (5) Simeprevir and Sofosbuvir (SS) for 12 weeks for genotypes 1 and 4, SR for 12 weeks for t p genotype 2, and SR for 24 weeks for genotype 3. We assumed all treatments occurred in the first year of the simulation. These treatments are consistent with those evaluated by miajor medical r societies in creating their HCV treatment guidelines.[10] Although guidelines discourage the use c of older line treatments, we include them to facilitate comparisons with other studies. We also s separately report preliminary results for interferon-free combination of ledipasvir and u D Sofosbuvir, which was approved after initial submission of this paper. o w n nlo a Screening, Contraindication, and Antiviral Initiation de d a fro m We assumed that 91% of those offered testing would accept and 90% of those who tested h ttp positive would receive those results and be evaluatMed for treatment.[66] To estimate the ://cid .o x fo rd proportion of patients who would receive treatment we conducted a meta-analysis of rates of jo u rn a ls treatment found across 12 published studies of community treatment of patients with HCV .o d rg b/ y J infection only.[15-26] We estimated the proportion who would be treated with pegylated u le e s L e v interferon based treatments (0.242) and its credible interval (0.229-0.251) using Monte Carlo in o n t M Markov Chain (MCMC) sipmulation methods programmed with Proc MCMC of the SAS 9.2 arc h 1 6 Software (SAS Institute, Cary, NC).[67] We also estimated the proportion of persons who would , 20 e 15 be treated (0.719) with non-pegylated interferon-based treatments and its credible interval c (0.689-0.747).   c Effectiveness, Cost, and Benefit of Antiviral Therapy A Older forms of treatment have exhibited lower rates of real world effectiveness and cost than in clinical trial data, but real-world data are not yet available for newer treatments. To enable equivalent comparisons we used clinical trial estimates of efficacy and published package 6  estimates of cost for all treatments. The benefit of successful treatment was an SVR which varied with treatment type and virus genotype. For pegylated interferon based treatments, we also t p assumed a quality adjusted life year decrement that varied with the duration of treatment. We assumed an SVR eliminated fibrosis progression associated with chronic HCV inifection. For r patients with cirrhosis, we assumed an SVR was also associated with a relative risk of HCC of c 0.24.[39]   s Testing and Medical Treatment Costs  u D o w We set the cost of testing via routine risk-based assessnments to $25.65 per person tested, nloa d e d equal to the incremental costs of testing using an electroanic health record prompt system in an from h ttp unpublished CDC study. Diagnosed patients who dMid not undergo antiviral therapy or achieve an ://cid .o x fo SVR were assumed to receive HCV-related medical management, with costs per stage estimated rd jo u rn a as the average costs used across seven previously published cost-effectiveness studies.[2, 4, 6-9, ls .o d rg b/ 52] Patients who achieved an SVR accrued annual monitoring costs. Non-treatment clinical y J u le e s L management increased costs without increasing benefits. ev in o n Utility Losses t M a p rc h 1 6 Uninfected persons were assigned annual QALY values that decreased with age to , 2 0 e 15 account for other health conditions.[68] For persons with HCV, we collected utility losses from 5 c studies across 7 HCV states: SVR, METAVIR 0–1, METAVIR 2–3, compensated cirrhosis, c DCC, HCC, and post-liver transplant then summarized the scores as reported elsewhere.[2, 69- A 73] Annual QALYs for patients on pegylated interferon-based therapy were multiplied by 0.85 adjusting for treatment duration.[14] 7  Simulation, Outcomes, and Sensitivity Analysis We estimated medical outcomes, costs, and QALYs associated with each scenario t p accounting for uncertainty in each of the model’s key parameters using probabilistic sensitivity analysis (PSA), reporting the mean and the empirical 95% credible interval for eaich outcome. r We estimated the incremental cost-effectiveness ratio (ICER) for routine and birth-cohort testing c combined followed by each treatment scenario as compared to the next most costly alternative. s For PRS/SR and for SS/SR, we estimated the ICER of immediate treatment compared to no u D treatment (NT; scenario 1) for people in METAVIR stages F0, F1, F2, F3, and F4. For PRS/SR o w n nlo a compared to PRPI and for SS/SR compared to PRS/SR we tested the univariate sensitivity of the de d a fro m ICER to uncertainty in the model’s key parameters by evaluating results based on the upper and h ttp lower bounds of the 95% confidence interval of eaMch parameter included in tables 1a and 1b. ://cid .o x fo rd We estimated the cost of treatment for SS/SR at which the ICER was equal to $50,000 jo u rn a ls per QALY gained compared to PRS/SR and compared to NT. Compared to NT, we estimated .o d rg b/ y J the treatment cost at which the ICER of PRS/SR and SS/SR was equal to $50,000 per QALY u le e s L e v gained for patients treated at stages F0 and F1. in o n t M For all patients, we pestimated the cost-effectiveness of SS/SR compared to PRS/SR and arc h 1 6 to NT when the cost of SS was set to the list price of Viekira Pak™ ($83,319) and the list price , 20 e 15 of Harvoni™ ($94,500). We provide only limited results for these scenarios, because these c treatments were released during this manuscript’s review process. c RESULTS A Of the 229.2 million Americans aged ≥20 years in 2012, we estimated 3.7 million were antibody positive for HCV, 2.9 million were chronically infected, and 1.5 million were identified through testing prior to the development of end stage liver disease or death from other causes. 8  With no testing or treatment (scenario 1), we estimated that 1.18 millionof those chronically infected (41.1%) would develop DCC or HCC and die in those states prior to model termination t p at age 100 (Table 2). For comparison to other studies, the model’s 45-year mortality rate was 18.7% assuming age of infection of 25 years and a starting fibrosis state of F0. Wiith no testing r or treatment, currently infected patients were expected to generate $100.3 billion in discounted c incremental hepatitis C medical costs during their lifetimes. s The health benefits and cost impacts of treatment scenarios u D With testing and PR treatment (Scenario 2), 356,657 patients were treated of whom o w n nlo a 156,880 achieved an SVR reducing the number of HCV-associated deaths from 1,181,554 to de d a fro m 1,131,638, a reduction of 49,916 deaths compared to NT. Compared to NT, testing followed by h ttp PR treatment increased QALYs by 306,537 and meMdical costs by $18.3 billion. With the same ://cid .o x fo rd number of patients treated with PR, as compared to NT, PRPI (Scenario 3) increased patients jo u rn a ls achieving an SVR by 237,618 and reduced the number of deaths from HCV to 1,106,130, a .o d rg b/ y J reduction of 75,424 deaths. Compared to NT, PRPI increased QALYs by 477,066 and increased u le e s L e v medical costs by $20.8 billion. With testing and PRS/SR treatment (Scenario 4), 541,136 in o n t M patients were treated of whpich 489,573 achieved an SVR reducing the number of deaths from arc h 1 6 HCV by 156,106 compared to NT. Compared to NT, PRS/SR increased QALYs by 1,032,292 , 20 e 15 and increased and medical costs by $47.0 billion. Finally, with testing and SS/SR treatment c (Scenario 5), 1,057,148 patients were treated of which 1,010,225 achieved an SVR reducing the c number of deaths from HCV by 320,646 compared to NT. Compared to NT, SS/SR increased A QALYs by 2,142,743 and medical costs by $127.1 billion. 9  Incremental Cost-effectiveness The ICER of PR versus NT was $59,792 per QALY gained (Table 2). PR was t p extendedly dominated by PRPI. Compared to NT, the ICER of PRPI was $43,530 per QALY gained, PRS/SR cost $47,237 per QALY gained compared to PRPI, and SS/SR coist $72,169 per r QALY gained compared to PRS/SR. Compared to no treatment, the incremental cost per QALY c gained was $59,792 for PR, $43,530 for PRPI, $45,524 for PRS/SR, and $59,333 for SS/SR. s SENSITIVITY ANALYSES u D Compared to NT, the ICER of both PRS/SR and SS/SR was sensitive to the fibrosis stage o w n nlo a at the time of treatment, from $173,800 per QALY gained for SS/SR at stage F0 to $13,000 per de d a fro m QALY gained for PRS/SR for patients with cirrhosis (Figure 2). The ICER of PRS/SR h ttp compared to PRPI was most sensitive to the cost oMf PRS/SR treatment, QALY improvements ://cid .o x fo rd assumed to occur after an SVR, the speed of fibrosis progression, QALY losses associated with jo u rn a ls moderate fibrosis (F2, F3) and cirrhosis (F4), the medical cost of DCC, the probability of an .o d rg b/ y J SVR for PRS/SR, and the risk reduction of HCC among people with cirrhosis who had achieved u le e s L e v an SVR (Figure 3a-b). No other parameter in the model changed the ICER by more than 5% in o n t M when set to the bounds of ipts 95% confidence interval. The ICER of SS/SR compared to PRS/SR arc h 1 6 was sensitive to similar variables (cost of treatment, QALY losses associated with infection prior , 20 e 15 to end stage disease, the probability of an SVR, and the impact of an SVR on reducing HCC). c The ICER of SS/SR compared to PRS/SR fell to $50,000 per QALY gained at a c treatment cost of $136,000. Compared to NT, the ICER of SS/SR was equal to $50,000 per A QALY gained at a treatment cost of $139,000. Assuming the same level of effectiveness, SS/SR cost $24,921 per QALY gained compared to PRS/SR and $31,828 compared to NT at the price 10  of Viekira Pak™, and $25,405 per QALY gained compared to PRS/SR and $35,100 compared to NT at the list price of Harvoni™. t p Compared to NT, treating patients at stage F0 with PRS/SR would need to cost $37,600 to achieve an ICER of $50,000 per QALY; $47,000 for treatment with SS/SR. Ailso as compared r to NT, treating patients at stage F1 with PRS/SR would need to cost $73,000 to achieve an ICER c of $50,000 per QALY; $82,000 for treatment with SS/SR. s CONCLUSIONS u D Our estimates indicate that the treatment alternatives for HCV of pegylated interferon o w n nlo a combined with ribavirin and Sofosbuvir, and the all-oral combinations of Sofosbuvir and de d a fro m Simeprevir increase QALYs compared to their alternatives at a cost of $47,237 per QALY h ttp gained for PRS/SR and $72,169 per QALY gainedM for SS/SR. During review of this article, two ://cid .o x fo rd interferon-free combination treatments for genotype 1 HCV patients (Harvoni™ and Viekira jo u rn a ls Pak™) with lower list prices ($94,500 and $83,319) compared to SS/SR. Assuming an equal .o d rg b/ y J effectiveness for these combinations as for SS, the lower prices would result in cost-effectiveness u le e s L e v of approximately $25,000 per QALY gained for new treatments compared to PRS/SR, and of in o n t M approximately $32,000 to $p35,000 per QALY gained compared to NT. Potentially lower prices arc h 1 6 would improve treatment cost-effectiveness further. , 20 e 15 However, financing the treatment of all Americans who could benefit from antiviral therapy c will be a continuing challenge given the number of individuals who are undiagnosed, untreated, c or failed to respond to older treatment regimens. Simply linking diagnosed patients to clinical A settings in which they can be evaluated for treatment remains an ongoing challenge which is likely to reduce the potential benefits and costs of new treatments for the foreseeable future.[74, 75]