Economic Implications of Entecavir Treatment in Suppressing Viral Replication in Chronic Hepatitis B (CHB) Patients in China from a Perspective of the Chinese Social Security Program


Yong Yuan, Bristol-Myers Squibb Company, PO Box 4500, Princeton, NJ 08543-4000 USA. E-mail:


Objectives:  Of estimated 112 million persons infected with chronic hepatitis B (CHB) in China, 15% to 40% will eventually develop liver complications. Most patients do not actively seek antiviral agents for treatment due in part to lack of good understanding of the disease. Entecavir is a new therapeutic option for CHB patients and the purpose of this study was to evaluate the cost-effectiveness of entecavir treatment in China, based on projected clinical benefits from its superior viral suppression efficacy.

Methods:  The analysis was based on the perspective of the Chinese Social Security program. Adjusted relative risks on the association between viral load (VL) and clinical end points (liver cirrhosis/hepatocellular carcinoma) were derived from a publication of a Taiwan CHB prospective cohort with 42,115 person-years of follow-up, and applied to patients enrolled in a randomized phase III trial in China. In this trial, hepatitis B virus (HBV) DNA (by polymerase chain reaction assay) was the key efficacy end point after 48 weeks of treatment with either entecavir or lamivudine monotherapy. Entecavir and lamivudine daily prices were assumed to be Renminbi Yuan (RMB) 40 and 16.71, respectively. Life expectancy tables were based on China vital statistics. Direct medical cost and utility scores for different phases of CHB were estimated from published China specific data, and costs were adjusted to 2006 values using the Chinese Consumer Price Index. Probabilistic sensitivity analyses were conducted to evaluate parameter uncertainty on event distribution and treatment failure rates beyond the trial period.

Results:  A total of 519 subjects were enrolled in the study, comprising of 82% males, 87% HBeAg+, and a mean age of 30 years. Based on the efficacy measurement of the percentage of patients achieving HBV DNA <300 copies/ml at week 48, entecavir was superior to lamivudine (78.7% vs. 46.7%, respectively [P < 0.05]). In the base case, compared with lamivudine, 1 year of entecavir therapy gained 0.305 quality-adjusted life year (QALY) at an incremental cost of 5368 RMB, with a 3% annual discount. Compared with lamivudine, using entecavir cost an incremental 17,590 RMB per QALY gained (95% CI 6333–56,407).

Conclusions:  Based on the results of this study, entecavir is likely to be cost-effective in treating hepatitis B patients in China based on the World Health Organization's recommended maximum willingness to pay threshold.


Chronic hepatitis B virus (HBV) infection is a progressive liver disease which affects about 350 million people worldwide. Chinese national surveys of the epidemiology of chronic hepatitis B (CHB) indicate that approximately 112 million Chinese citizens are chronically infected with HBV, representing 32% of the global disease burden. China is the country with the single largest number of CHB affected persons [1]. HBV infection imposes a massive socioeconomic burden in China because 30% of chronically infected individuals subsequently develop progressive and potentially fatal liver disease, and an estimated 20% of these patients die of the complications of CHB infection [2]. It is estimated the annual medical cost for a patient with CHB or compensated cirrhosis (CC) is around 10% of the average Beijing resident's annual income. Moreover, the annual medical cost for a patient with decompensate cirrhosis (DC) or hepatocellular carcinoma (HCC) far exceeds their annual income [3].

Unfortunately, only a small minority of patients who might benefit from the antiviral therapy have received treatments. The potential reasons for this include (but are not limited to) an evolving understanding of different treatment approaches, side effects of therapies such as inteferons, relatively high rate of emergent drug resistance with older oral antivirals such as lamivudine (LVD), uncertainty of therapy's long-term clinical and economic outcomes, and high treatment costs over a short period of time.

Currently, the most prescribed antiviral agent for chronic HBV infection in China is LVD, which demonstrates anti-HBV efficacy and provides clinical benefits. Nevertheless, the occurrence of drug resistant HBV mutations has generated concern among physicians prescribing LVD [4–7]. A new deoxyguanine nucleoside analog, named entecavir (ETV), has demonstrated excellent suppression of HBV replication without significant side effects or evidence of mitochondrial toxicity. In addition, long-term clinical trial data have not shown evidence of viral resistance [8].

This study aims to evaluate the cost-effectiveness of ETV treatment compared to LVD, by translating the HBV DNA reduction observed with ETV treatment in a China phase III trial into estimates of gains in life expectancy (LE) and medical cost offsets, to arrive at an estimate of the incremental cost per qualify-adjusted life year (QALY) saved. Use of such results would be helpful to facilitate decision-making of formulary listing of the government health insurance program, and to provide perspective to physicians and patients on new intervention that may offer both short and long-term clinical and economic benefits.


Theoretical Basis for Economic Modeling

Because CHB-related liver complications usually take years to become apparent, a typical clinical trial with 1-year follow-up is not adequate for capturing the treatment impact on the clinical events. Thus, it is often necessary to use surrogate end points as substitutes for projecting disease progression. The main surrogate end points traditionally used to assess treatment response include undetectable serum HBV DNA level, HBeAg status, normalization of alanine aminotransferase (ALT) level, and improvement in liver histology. Nevertheless, a growing body of evidence suggests that both ALT level and HBeAg seroconversion are poor predictors of long-term disease outcome and their associations with a reduction of risk of cirrhosis and HCC is still under debate [3,9]. Evidence suggests that HBV DNA is a more appropriate prognostic marker of liver disease progression [9–11]. Therefore, therapies which impact this marker may provide a useful index of the effectiveness of treatments.

The study data on risk of disease progression based on the level of viral load were derived from the Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer-Hepatitis B Virus (REVEAL-HBV) study [10,11]. This was a prospective cohort study which showed that, hepatitis B viral load was a strong independent predictor of liver cirrhosis and HCC events (Table 2) [11–13].

Table 2.  Model parameters and data inputs
ParametersReference case valueRangeSource
  • *

    The results are slightly different from the final clinical study report for the trial, as patients with missing HBV DNA data at either baseline or week 48 were treated as missing and excluded in this economic analysis.

  • All costs are expressed in 2006 RMB.

  • CHB, chronic hepatitis B; DC, decompensate cirrhosis; HCC, hepatocellular carcinoma; REVEAL-HBV, Risk Evaluation of Viral Load Elevation and Associated Liver Disease/Cancer-Hepatitis B Virus.

Age of patients (years) 30 16–64BMS China trial—AI463023
Gender Both sexes Male, femaleBMS China trial—AI463023
Discount rate (annual %)
 Cost 3% 0–5%Reference [14]
 Life expectancy 3% 0–5%Reference [14]
Estimated relative risk by serumAll CirrhosisDCHCC  
HBV DNA levels (copies/ml)
 <300 (undetectable)111 REVEAL-HBV [11–13]
 300–99991.42.71.295% CIs are available from referencesREVEAL-HBV [11–13]
 10,000–99,9992.52.72.995% CIs are available from referencesREVEAL-HBV [11–13]
 100,000–999,9996.95.99.595% CIs are available from referencesREVEAL-HBV [11–13]
 ≥1 million9.819.315.295% CIs are available from referencesREVEAL-HBV [11–13]
Annual incidence rates with undetectable HBV DNA
 Compensated cirrhosis 0.34% 0.30–0.40%REVEAL-HBV [11]
 Decompensate cirrhosis 0.02%  0–0.05%REVEAL-HBV [11]
 HCC 0.11% 0.10–0.12%REVEAL-HBV [10]
Annual morality rate
 Decompensate cirrhosis 14.4% 10%-20%Reference [20]
 HCC 23.3% 20%-25%Reference [19]
Average time to event from study entry (years)
 Compensated cirrhosis 8 1–14REVEAL-HBV [11]
 Decompensate cirrhosis 9 1–14REVEAL-HBV [11]
 HCC 7 1–14REVEAL-HBV [10]
Annual medical costs
 Compensated cirrhosis 1,500 1,000–2,000Reference [3]
 Decompensate cirrhosis 13,803 10,000–16,000Reference [3]
 HCC 38,450 30,000–50,000Reference [3]
 CHB 0.71 0.99Reference [21,23]
 Compensated] cirrhosis 0.76 0.80Reference [21,23]
 Decompensate cirrhosis 0.266 0.60Reference [21,23]
 HCC 0.30 0.73Reference [21,23]
Study drugsEntecavir 0.5 mg Lamivudine 100 mg  
Daily cost/patient40 16.71±10%Official prices approved by Shanghai municipal bureau of pricing
Actual days of use/patient/year354 353303–405BMS China trial—AI463023
Adefovir 10 mg daily cost/patient (as a salvage therapy) 21 ±10%Official prices approved by Shanghai municipal bureau of pricing
Proportion of patients with HBV DNA levels (copies/ml)*Entecavir 0.5 mg Lamivudine 100 mg  
<300 (undetectable)78.7% 46.7% BMS China trial—AI463023
300–9,99914.7% 16.9% BMS China trial—AI463023
10,000–99,9994.7% 9.2% BMS China trial—AI463023
100,000–999,9990.8% 6.5% BMS China trial—AI463023
≥1 million1.1% 20.7% BMS China trial—AI463023

Description of Economic Model Structure

Following the clinical trial design, a previously published decision tree model was used to evaluate the cost-effectiveness of ETV compared to lamuvidine in suppressing HBV DNA to an undetectable level [14]. An illustrative conceptual model framework is shown in Figure 1. In brief, a hypothetical cohort of 1000 CHB patients either received LVD or ETV at the model entry; these patients might develop viral resistance with continued drug use or experience viral rebound after the treatment cessation. HBV DNA values were updated annually to incorporate the first year trial efficacy results and the impacts from subsequent development of viral resistance or viral rebound after treatment cessation. CC, DC, and HCC events were projected within 10 years, separately, at a population level, using the published REVEAL-HBV risk predicting models. The annual incidence rates for subjects with hepatitis B viral load <300 copies/ml were 0.34% for CC, 0.02% for DC, and 0.11% for HCC [10,11].

Figure 1.

Illustrative model diagram of 10-year disease progression. This is a modified tree model projecting disease risks based on current HBV DNA levels. CC, compensated cirrhosis; CHB, chronic hepatitis B; DC, decompensate cirrhosis; HBV, hepatitis B virus; HCC, hepatocellular carcinoma.

The study perspective was from that of a China social security program responsible for all direct health-care expenditures. Costs and life years were discounted at an annual rate of 3%. All model parameters and data input values are provided in Table 2.

Statistical software (SAS, SAS Institute, Cary, NC, USA) was used for all data analyses, and a spreadsheet-based model using Visual Basic for Applications (Microsoft Corporation, Redmond, WA, USA) was created for all modeling and simulation executions.

Clinical Efficacy Data

Efficacy and safety.  Efficacy and safety data came from BMS-AI463023 [15], a randomized (1:1), double-blind, double-dummy phase III study enrolled exclusively in China, comparing the safety and efficacy of ETV 0.5 mg once daily (QD) to LVD 100 mg QD administered for 1 year to subjects with chronic HBV infection. Randomization was stratified by hepatitis B e antigen (HBeAg) status and investigative site. The primary objective was to determine the proportion of subjects treated with ETV who achieved a response for the composite end point (week 48 virological and biochemical response), which included our end point of interest—week 48 serum HBV DNA by polymerase chain reaction (PCR) assay.

After 1 year of treatment, ETV resulted in superior clinical benefit over LVD in treating nucleoside-naïve patients with chronic HBV infection, measured by reduction in HBV DNA and normalization of ALT. ETV was well tolerated, with a safety profile comparable to that of LVD. Because of the equivalent safety results between therapies, the risk and costs of drug-related adverse events were not considered in the analysis.

Study patient and time.  An intent-to-treat analysis was conducted. Patients with missing HBV DNA data at either baseline or week 48 were treated as missing and excluded in the primary analyses. An alternative imputation method using Last Observation Carried Forward was also implemented, but had minimal impact on our findings. The study period was from a patient's randomization to the end of blinded treatment or week 48, whichever came first.

Cost Estimates

Two cost components were considered in the analyses: study drug costs and medical costs of treating liver complications. Indirect medical costs and lost productivity were not included in the analyses. Costs are expressed in year 2006 Renminbi Yuan (RMB) (US $ 1 = 7.55 RMB, as of September 12, 2007).

All drug costs were calculated based on the actual usage of study drugs (97% compliance) recorded on the clinical trial Case Report Forms. The daily acquisition cost of ETV 0.5 mg was 40 RMB, and 16.71 RMB for LVD 100 mg and 21 RMB for adefovir as a salvage therapy, based on current official prices approved by Shanghai municipal bureau of pricing. The annual costs of study drugs were estimated by the daily costs multiplied by number of days of drug use in the trial.

Annual medical costs including inpatient, outpatient, and laboratory tests were derived from the published Chinese cost data: 1500 RMB for CC, 13,803 RMB for DC, and 38,450 RMB for HCC [3]. These costs were then adjusted using the Consumer Price Index in China [16], and applied to patients at the time when events occurred and until they died. The actual clinical trial outpatient physician visits and those of laboratory tests recorded in the trial were excluded from the cost estimates because there were largely driven by the study protocol and therefore were nearly identical in both treatment arms.

Life Year Gain and Utility Estimates

Age- and gender-specific LE estimate for CHB and CC were based on the published vital data for the general population [17]. LE for DC or HCC was estimated based on the declining exponential approximation of LE method: inverse of annual event mortality [18]. Annual mortality was assumed to be 23.3% for HCC and 14.4% for DC [19,20].

All CHB patients were assigned a baseline utility of 0.71 at entry—meaning that 1 year of life in a person with CHB would be equivalent to 0.71 year of healthy life, and an estimated utility weight of 0.76 for CC, and 0.266 for DC, and 0.30 for HCC. Ratings of CHB-associated health states were elicited from a representative sample of 100 uninfected individuals in the China using a visual analog scale and weighted using the standard gamble method [21].

Other Modeling Assumptions

To appropriately discount the subsequent medical costs and life year lost after these events had occurred, in the base case analysis, we deterministically assigned average time to events, about 8 years based on the observed event time in the REVEAL-HBV cohort, to the trial patients who later developed a CC, DC or HCC event, although we also modeled time to event probabilistically later in the sensitivity analyses.

In the reference case analysis, the distributions of subjects with different HBV DNA levels beyond the first year after the ETV or LVD treatment cessation were assumed based on a published study range showing the cumulative reappearance rates for serum HBV-DNA following cessation of LVD therapy at 1 year was 48% [22]. For those responders who failed to reach the undetectable viral level in the next year, they would be evenly allocated into four higher HBV DNA categories. The number of events were first projected based on the first-year observed HBV DNA data, and then repeated for each year based on viral rebound-adjusted viral data from years 2 to 10, and an average of these projections were used for our final estimate of number of liver cirrhosis and HCC events.

Sensitivity Analyses

To investigate whether our findings are robust and sensitive to any input parameter variation, we first ran the univariate sensitivity analyses on the following input variables based on their possible ranges (Table 2): age, gender, drug price, treatment duration, viral resistance rate for ETV after 2 years, event medical costs, and time to event. We also ran alternative sensitivity analyses on the following key input parameters:

First, utility scores have rarely been measured for CHB patients, the utilities for liver complications used in this study relied on a single survey study for CHB patients. To test the impact of varying utility scores for the liver complications on our cost-effectiveness results, a different set of published utility tariff values was also used [23], but were not chosen for the base case analyses as their utility scores were primarily derived from the surveys on hepatitis C patients.

To evaluate whether ETV for a longer term use is also cost-effective, we extended duration of ETV treatment for 3, 5, and 10 years, assuming the trial efficacy observed for ETV in the first year would sustain beyond the trial period. Because the viral resistance problem has been well documented for patients taking LVD, we assumed that, as long as the patients were taking LVD, there were additional patients who would develop treatment resistance each year, the cumulative LVD viral resistance rate from years 1 to 5 were: 14%, 38%, 49%, 66%, 69% [24], and stayed 69% beyond 5 years. Once they developed LVD resistance, they had to switch to use adefovir as a salvage therapy, although assuming that their viral load would not worsen. In a separate sensitivity analysis, we also assumed that patients developing LVD resistance would be treated with the addition of adefovir to LVD therapy to reflect recent clinical practice in the management of lamvudine-resistance patients although conservatively assuming the treatment efficacy would be maintained. We assumed no treatment resistance for ETV therapy based on the recent long-term trial data [7].

To evaluate uncertainty with respect to model parameters, probabilistic sensitivity analyses with 1000 iterations were conducted for the following two key parameters with the biggest uncertainty concern: 1) viral rebound rates after the treatment cessation: a beta distribution with values for shape parameters alpha (38.88) and beta (42.12) derived from the reported mean and standard deviation of viral rebound rates; and 2) time to the first event: a gamma distribution with the sample mean (survival years, CC 7.6; DC 8.6; HCC 7.1) and standard deviation (CC 3.95; DC 2.77; HCC 3.53) observed from the REVEAL-HBV cohort. A gamma distribution was deemed appropriate because time to event was constrained to be positive and the gamma distribution is only defined for positive values.

Incremental Cost-Effectiveness Ratio (ICER) Benchmark in China

Certainly, the lower the cost-effectiveness ratio, the better, but what should determine an upper limit? In the United States, $100,000 to $150,000 per QALY has become an accepted benchmark for policymakers and insurance agencies [25]. But that level is unlikely to be sustainable in developing countries like China. The World Health Organization's Commission on Macroeconomics and Health [26] recommended choosing interventions that had cost-effectiveness ratios less than three times the Gross National Income (GNI) per capita. With a GNI per capita of $2228.9 in China in 2005, based on the official report from the World Bank, assuming a 10% growth rate in 2006, we estimated that 55,533 RMB could be an acceptable maximum willingness to pay threshold. To put the estimated cost-effectiveness ratios in context from a Chinese social security program perspective, we used this benchmark to determine whether ETV treatment would fit within their maximum willingness to pay for the values.


Demographic characteristics were balanced between treatment groups (Table 1). The majority of treated subjects were male (82%) and Asian (100% Chinese), with a mean age of 30 years. HBV disease characteristics were comparable between the two treatment groups. All subjects had detectable HBsAg at baseline. Most subjects were HBeAg-positive at baseline (ETV 87%, LVD 85%). The mean baseline ALT was comparable for the two treatment groups (ETV 195.55 U/L, LVD 197.56 U/L).

Table 1.  Baseline demographics and HBV characteristics (BMS-AI463023)
CharacteristicETV 0.5 mg (n = 258)LVD 100 mg (n = 261)
  1. ALT, alanine aminotransferase; ETV, entecavir; HBV, hepatitis B virus; LVD, lamivudine; PCR, polymerase chain reaction.

Age, years
 Mean (SE)30 (0.5)30 (0.6)
Gender, n (%)
 Male211 (82)217 (83)
 Female47 (18)44 (17)
Race, n (%)
 Asian/Pacific Islanders258 (100)261 (100)
HBV DNA by PCR, log10 copies/ml
 Mean (SE)8.64 (0.062)8.48 (0.069)
HBV surface antigen: n (%)
 Postive258 (100)261 (100)
HBV E antigen: n (%)
 Postive220 (87)212 (85)
 Negative38 (13)49 (15)
 Mean (SE)195.55 (8.716)197.56 (11.15)

Mean baseline HBV DNA levels were comparable for the 2 treatment groups using the bDNA assay (ETV 2.56 log10 MEq/ml, LVD 2.42 log10 MEq/ml) and PCR assays (ETV 8.64 log10 copies/ml, LVD 8.48 log10 copies/ml).

The estimated clinical and economic outcomes are reported in Table 2 and Table 3. ETV was superior to LVD for the proportion of subjects who achieved HBV DNA <300 copies/ml by PCR assay at week 48. At the end of 48-week trial period, there were 79% patients in the ETV arm, and 47% patients in LVD arm, reaching the predefined undetectable HBV DNA goal, <300 copies/ml, costing 16,807 RMB per patient to reach this goal.

Table 3.  Cost-effectiveness results for entecavir versus lamivudine
 Entecavir 0.5 mgLamivudine 100 mgDifference
  1. All costs are expressed in 2006 RMB.

  2. CHB, chronic hepatitis B; HCC, hepatocellular carcinoma; QALY, quality-adjusted life year.

Number of CHB patients at entry1,0001,000 
Duration of treatment (years)11 
Total discounted drug costs14,159,1675,893,2698,265,898
Projected liver complication in 10 years
Compensated cirrhosis11514732
Decompensate cirrhosis10155
Total discounted medical costs8,646,27911,544,164−2,897,884
Discounted life year lost1,0751,460385
Discounted QALY lost8531,158305
Cost per life year saved13,958 
Cost per QALY saved17,590 

Among a hypothetical cohort of 1000 patients, we projected that there would be 32, 5, and 19 fewer cases of CC, DC, and HCC events, respectively, in the ETV arm, compared to the LVD arm. These translated into about 2.9 million RMB medical cost offsets, and gain an aggregate 385 life years in 10 years, with ETV treatment.

With a daily drug cost of 40 RMB for ETV, 1 year of ETV therapy gained 0.305 QALY at an incremental cost of 5368 RMB, with a 3% annual discount. Compared with LVD, using ETV cost an incremental 17,590 RMB per QALY gained (95% CI 6333–56,407).

Univariate sensitivity analyses showed that our findings are most sensitive to drug prices, treatment duration, efficacy, discount rate, age, and cost, but not sensitive to gender. ICER, estimated using alternative utility scores [23]: 0.99 for CHB, 0.80 for CC, 0.60 for DC, and 0.73 for (HCC), was 13,330 RMB, a more favorable result. Although LVD does not have a generic version available yet in China, to evaluate the possible impact of generic LVD on our research results, we assumed 20% discount for general LVD price (based on the current adefovir prices in China as a analog: brand 25 RMB, generic 21 RMB), this yielded an ICER estimate of 21,452 RMB, still well below 55,533 RMB threshold.

A plot of the probabilistic sensitivity analysis (PSA)—derived joint distribution of cost and effectiveness is shown in Figure 2, most points lie below the cost-effectiveness threshold, indicating that there is high possibility that the ETV therapy is cost-effective. Cost-effectiveness acceptability curves are shown in Figure 3. The favorable cost-effective results are very robust, with 97.6% of PSA-derived ICER estimates below a hypothetic 55,533 RMB/QALY threshold. Longer term modeling also showed that 3, 5, or 10 years of ETV treatment would still be cost-effective, yielding an incremental cost per QALY saved of 44,511, 54,358, and 67,874, in RMB, respectively (Fig. 3), assuming that LVD-resistant patients would switch to adefovir monotherapy. Alternatively, in a separate sensitivity analysis assuming that the treatment-resistant patients would use add-on adefovir, ICERs were lower, ranging from 37,496 (3 years) to 36,447 (10 years), primarily due to the relatively higher cost of adefovir adding-on therapy.

Figure 2.

Monte Carlo simulations on the CE plane. CE, cost effectiveness; QALY, quality-adjusted life year; RMB, Renminbi Yuan.

Figure 3.

CE acceptability curves by entecavir treatment duration. CE, cost effectiveness; RMB, Renminbi Yuan.


The major health risks and economic impacts associated with CHB infection seem to be primarily driven by the later development of HCC and complications of DC. In China, CHB infection is an important cause of HCC, particularly among individuals during the most productive decades of life span. Current management options for HCC remain limited partly because there is a paucity of sensitive methods for early diagnosis, resulting in most patients late in the disease course. Liver transplantation remains the only effective therapy for late complications like DC and HCC; unfortunately, this treatment is relatively too expensive and unavailable to the overwhelming majority of patients in China.

Although availability of HBV vaccine world wide including China, the high prevalence of HBV remains one of the significant public health issues in China due to its burden to the society. One major reason is that there is an enormous human reservoir for the virus given the very high rate of maternal-neonatal and horizontal transmission among children, and this is compounded by the fact that most infections early in life are asymptomatic; thus, most cases are not detected clinically. Therefore, the benefits of early treatment for CHB in suppressing the viral and reducing consequent risk of cirrhosis and HCC are substantive. The added demonstration of cost-effectiveness of such treatment from a China social security program perspective would inform decision-making in the light of limited health-care resources in the country.

The disease progression data used in this model were derived from the recently published REVEAL-HBV study [10,11] which showed that the incidence rates of HCC and cirrhosis increased across a biological gradient of serum HBV DNA levels in a dose–response relationship. These findings implicate viral replication in the progression of chronic HBV infection and contribute to the rationale for antiviral therapy to arrest progression of liver disease [27]. Although the REVEAL-HBV study provides the most robust information on the relationships between viral load and disease progression and mortality, it is not the only source of evidence to show this. In another prospective cohort study [28] conducted in Haimen city, China, with 11 years of follow-up, it was shown that cohort entry HBV viral load was a significant predictor of mortality and morbidity from HCC and chronic liver disease. Several studies have demonstrated similar findings [29–31]. Additionally, in studies of patients with HCC who have undergone surgery, one of the strongest predictors of tumor recurrence is the amount of circulating virus as measured by the HBV DNA [32,33].

In this model, we have assumed that risk of hepatic complications seen in patients with low viral loads can be simulated by reducing the viral load with antiviral therapy. Whereas the magnitude of the risk reduction may vary and is currently not known, the principle of this assumption is not only logical but biologically plausible. In a large meta-analyses of 26 prospective mostly registration studies with a total of 3428 study subjects (2524 were HBeAg-positive at baseline), the level of viral replication and the change in viral replication were significantly correlated with histological grading and change in histological grading, serological and biochemical response [34]. An Italian study published in 2004 showed that in about 656 HBeAg-negative CHB subjects treated with LVD, the presence of resistance as determined by viral breakthrough and rebound had a worse outcome in all parameters including mortality when compared to those that were effectively suppressed by LVD. This study was in patients with eAntigen-negative disease where immune-induced seroconversion is not an option and therefore the benefit of therapy is from direct antiviral effect of the drug [35]. Indeed, decrease in viral load through antiviral therapy has been associated with histological improvement [36–42], increased survival of patients with decompensate liver disease from hepatitis B [43–51], and improved clinical outcome in patients with HBV reactivation following chemotherapy [29,31,52–62]. Finally, a blinded placebo controlled randomized clinical trial has demonstrated that in CHB patients with advanced fibrosis, effective suppression of viral replication with an antiviral drug resulted in dramatic reductions in progression to more severe liver disease. This study proves the principle that informed the current model by demonstrating that the level of suppression achieved by an antiviral drug has a causal relationship with the risk of disease progression [63].

Although clinical benefits of reducing viral load have been demonstrated in recent literature, to the best of our knowledge only one other economic analyses has been conducted based on the end point of antiviral therapy-induced suppression of viral replication [64]. In that analysis, data from the REVEAL-HBV study was used to inform the disease state progression probabilities and the economic value of effectively suppressing viral replication in immune tolerant patients (who currently do not qualify for therapy under current guidelines) was shown in that analysis. Ours is therefore the first analyses applying this end point in a population of patients that meet all current treatment guidelines for therapy.

HBeAg seroconversion has traditionally been thought of as uniformly good outcome for patients, and for some it still is, but recent data have shown that many patients who have undergone HBeAg seroconversion still go on to develop severe complications (including HCC) [65]. Seroconversion and the absence of HBeAg are probably still important clinical end points, but the risk of disease progression even with these end points being met still depends on the degree of ongoing viral replication in the host. The ultimate goal of therapy for patients with chronic HBV infection is to prevent progression of liver disease to cirrhosis and HCC. Because HBV replication is implicated in the outcome of chronic HBV infection, the primary aim of therapy is durable suppression of serum HBV DNA to the lowest levels possible. This, in turn, will lead to the other aims of therapy, including histological improvement and ALT normalization [27]. Historically, the presence (i.e., levels greater than assay limit of detection) or absence (i.e., levels less than assay limit) of HBV DNA by hybridization techniques was a major determinant of treatment candidacy. Undetectable HBV DNA levels by hybridization techniques have in the past been considered clinically insignificant. With the advent of sensitive real-time PCR, serum HBV DNA has become the most useful measurement, and enables the accurate monitoring of HBV DNA at levels as low as 10 IU/ml, which has been recommended to be used to establish a patient's baseline HBV DNA level before treatment and to monitor response to antiviral therapy or viral rebound associated with resistance [27].

Our analyses are conservative in many aspects which have been described in details elsewhere [14]. In addition, we did not include indirect medical costs, caregiver support cost, and lost productivity in this study because national representative data of indirect costs are not reported in China; however, we believe that these indirect disease burdens could be very significant and costly to the China society and individual Chinese patients. Inclusion of such indirect costs would make the cost-effectiveness of ETV even more favorable.

Our study results still need to be viewed with caution. The longer-term treatment rebound rates for both drugs are so far not available. The optimal duration of ETV or LVD administration in CHB patients remains to be determined due to frequent relapse after cessation the therapy. Further studies are also needed to confirm these modeling results and determine whether a reduction in liver cirrhosis and HCC with ETV therapy actually occurs. Probabilistic sensitivity analysis needs to consider more model parameter uncertainty.

The potential cost-effectiveness for ETV compared with other drugs such as adefovir or interferon was not evaluated due to lack of head-to-head randomized trial data. Nevertheless, some recent published studies have shown that ETV is even more superior to adefvior in suppressing viral load [66–68].

Although we accounted for the actual compliance rate in the trial, we would expect the compliance rate in the real world practice would be much lower. Nevertheless, we still yet need to determine the potential impact of compliance on treatment efficacy.

In conclusion, our model consistently showed that both the short-term and long-term use of ETV is likely to be not only clinically effective but also economically attractive for the China Social Security program. Although eradication of hepatitis B is rarely achieved with the current treatments, the availability of ETV as part of an early treatment strategy for the current CHB patients may be perceived as economically justifiable to the China Social Security program. The treatment target can be applied for both HBeAg-positive and -negative patients and should result in prolonged viral suppression of HBV DNA to levels at which liver disease progression is unlikely to develop. Highly sensitive testing methods based on polymerase chain reaction (PCR) have become available for measuring serum level of HBV DNA. A new guideline recommending a simple treatment goal of achieving the undetectable viral load goal level such as <300 copies/ML would be more beneficial to both physicians and patients when rendering treatment decision.

The authors would like to thank our BMS colleagues Anne Cross, Eskinder Tafasse, Jun Su for their assistance for providing clinical trials data and publication references. We are grateful to Gilbert L'Italien for his helpful editorial suggestions.

Source of financial support: This research was funded by Bristol-Myers Squibb. Three of the authors (Yong Yuan, Uchenna H. Iloeje, and Hong Li) are employees of Bristol-Myers Squibb. Joel Hay received a grant from Bristol-Myers Squibb.