Prophylactic strategies for hepatitis B patients undergoing liver transplant: A cost-effectiveness analysis



Hepatitis B immunoglobulin with lamivudine prophylaxis (LAM/HBIG) is effective in preventing Hepatitis B (HBV) recurrence posttransplant but is expensive and inconvenient. Lamivudine-resistant HBV, which has limited the usefulness of lamivudine monoprophylaxis in transplant, can now be effectively controlled with adefovir dipivoxil. We performed a cost-effectiveness analysis on the strategies of lamivudine prophylaxis with adefovir rescue(LAM/ADV) compared to combination LAM/intravenous fixed high-dose HBIG prophylaxis(LAM/ivHBIG) or LAM/intramuscular HBIG prophylaxis(LAM/imHBIG). Markov modeling was performed with analysis from societal perspective. Probability rates were derived from systematic review of the literature and cost taken from MEDICARE database. Outcome measures were incremental cost-effectiveness ratio(ICER) and cost to prevent each HBV recurrence and death. Analysis was performed at 5 years posttransplant as well as at end of life expectancy (15 years). Combination LAM/ivHBIG cost an additional USD562,000 at 15 years, while LAM/imHBIG cost an additional USD139,000 per patient compared to LAM/ADV. Although there is an estimated increase in recurrence of 53% with LAM/ADV and 7.6% increased mortality at the end of life expectancy (15 years), the ICER of LAM/ivHBIG over LAM/ADV treatment is USD760,000 per quality-adjusted life-years and for LAM/imHBIG, USD188,000. Cost-effectiveness is most sensitive to cost of HBIG. Lamivudine prophylaxis with adefovir dipivoxil salvage offers the more cost-effective option for HBV patients undergoing liver transplant but with higher recurrence and death rate using a model that favors LAM/HBIG. Lowering the cost of HBIG maintenance will improve cost-effectiveness of LAM/HBIG strategy. In conclusion, a tailored approach based on individual risks will optimize the cost-benefit of HBV transplant prophylaxis. Liver Transpl 12:736–746, 2006. © 2006 AASLD.

Hepatitis B virus (HBV) infection had previously been considered a relative contraindication to orthotopic liver transplantation as it was associated with viral recurrence rate of greater than 80% and dismal mortality rate of 50% at 2 years.1 However, in recent years, the use of prophylaxis against HBV reinfection with nucleoside analogues such as lamivudine, Hepatitis B immunoglobulin (HBIG), or its combination has reduced the recurrence rate significantly.2

Lamivudine is a potent inhibitor of HBV deoxyribonucleic acid (DNA) synthesis. In the transplant setting, lamivudine monoprophylaxis is effective in reducing HBV recurrence, with 1 year recurrence rates reported as 10%,3 23%,4 and up to 32%.5 Unfortunately, its efficacy is reduced by emergence of breakthrough resistant strains with prolonged therapy. Although most patients with HBV recurrence posttransplant remain stable,6 presumably owing to the reduced replication fitness of mutant virus, recurrence has been associated with poor clinical outcomes, including accelerated liver cirrhosis, graft loss, and death.7

Long-term HBIG monoprophylaxis posttransplant was used in Europe and its efficacy in preventing HBV recurrence, 35% recurrence at 1 year posttransplant,8 is probably similar to lamivudine. It is less effective in patients with high pretransplant HBV DNA and, like lamivudine, results in selection of escape mutant viruses.9, 10 Combination prophylaxis with lamivudine and long-term high-dose HBIG appears to be the most effective strategy for preventing HBV recurrence with rates of 0–10%11 1 year after transplant with the longest follow-up up to 60 months post transplant.12 Five-year survival for HBV patients transplanted in the era of combination prophylaxis is also now equivalent to those transplanted for other etiologies.2, 13 However, the cost of HBIG is prohibitively high (first year cost >United States Dollars (USD)100,000; subsequent yearly cost > USD50,000).11 This has limited its universal use, especially in many developing countries where HBV is endemic. Even in developed countries, many centers have used modified strategies of intramuscular HBIG or HBIG based on target levels of antibody to HBV surface antigen (anti-HBsAb) to limit the cost, with the hope of achieving similar if not equivalent efficacy.

Adefovir dipivoxil is a new nucleotide analogue and has been recently shown to be effective in posttransplant patients with lamivudine resistant HBV. A 48-week treatment of adefovir dipivoxil in post-transplant (median 200 weeks) patients with lamivudine resistant HBV infection resulted in reduction of HBV DNA by 4.3 log10 copies/mL, normalization of serum alanine aminotransferase (ALT), albumin, bilirubin, and prothrombin time, improvement of child pugh score, and posttransplant survival of 93%.14

Patients who thus develop recurrent disease due to viral resistance to lamivudine can be effectively controlled with adefovir dipivoxil. In Asian countries where cost of lifelong HBIG is prohibitive, lamivudine monotherapy with adefovir rescue has become a practical option with fairly good results.15 This raises the question of how this transplant strategy might compare in terms of cost-effectiveness with the combination of lamivudine and indefinite HBIG prophylaxis, either as classical intravenous fixed high-dose formula or modified intramuscular “on demand” strategy.

We present here a Markov model comparing the outcomes and cost-effectiveness of these treatment strategies.


HBV, hepatitis B virus; HBIG, hepatitis B immunoglobulin; DNA, deoxyribonucleic acid; USD, United States Dollars; ALT, alanine aminotransferase; CEA, cost-effectiveness analysis; LAM, lamivudine; ADV, adefovir dipivoxil; PCR, polymerase chain reaction; HBsAg, hepatitis B surface antigen; QALY, quality adjusted life years; UNOS, United Network for Organ Sharing; ICER, incremental cost-effectiveness ratio; OPTN, Organ Procurement and Transplantation Network.; GNI, gross net income.


Model Structure (Fig. 1)

A Cost-effectiveness analysis (CEA)16 was performed using a Markov Model17 comparing the three prophylactic strategies for chronic HBV patients undergoing liver transplant.

Figure 1.

Strategies of LAM/ADV vs. LAM/HBIG in HBV liver transplantation. Schema of Hepatitis B prophylaxis strategies compared by Markov Modeling. Patients are channeled to one of the Markov States: No Recurrence, Stable Recurrence (HBV surface Antigen (HBsAg) positive but HBV DNA negative by non-PCR assay),Viral Resistance (HBsAg positive and HBV DNA positive by non-PCR assay), and Death. Patients who develop viral resistance are further divided into those who continue to remain stable, those requiring retransplant, or resulting in death. In LAM/ivHBIG: HBIG administered intravenously as fixed high-dose.10,000IU intraoperatively, and then daily X7 days and monthly thereafter indefinitely. In LAM/imHBIG: HBIG administered intravenously perioperatively, 10,000IU intraoperatively and daily ×7 days then converted to intramuscular injection 1,560U 2-4 weekly based on target anti-HBs levels. Supplementation by intravenous administration 10,000IU when antiHBS levels fall below target despite intensified intramuscular injection.

The three prophylactic strategies were the following:

  • 1Lamivudine monoprophylaxis pretransplant with adefovir dipivoxil rescue when lamivudine resistance develops,
  • 2Combination prophylaxis with lamivudine pretransplant and addition of indefinite high-dose HBIG at transplant, with adefovir added when lamivudine resistance develops, and
  • 3Combination prophylaxis with lamivudine pre-transplant, intravenous HBIG peritransplant, and conversion to intramuscular HBIG for long-term maintenance, with adefovir added when lamivudine resistance develops.

Details of the three strategies used in the model are as follows. In the Lamivudine prophylaxis/Adefovir rescue strategy (LAM/ADV), patients are started on lamivudine pretransplant. Adefovir dipivoxil (10 mg daily or titrated dose according to renal function) is added when lamivudine resistance develops, either before or after transplant (defined as breakthrough of positive HBV DNA by non-PCR assay).

In the Lamivudine/ivHBIG prophylaxis/Adefovir salvage strategy (LAM/ivHBIG), lamivudine is similarly started for patients before transplant. Intravenous HBIG is started intraoperatively and is continued indefinitely posttransplant with lamivudine. HBIG is administered using the high-dose regimen commonly used in United States (intravenous infusion of 10,000 IU of HBIG during the anhepatic phase, followed by 10,000 IU/day for 7 days, and then 10,000 IU/month lifelong).12 Adefovir is added when patients develop resistance to lamivudine either pretransplant or posttransplant.

The Lamivudine/imHBIG prophylaxis/Adefovir salvage strategy (LAM/imHBIG), modeled after that reported by Han et al.,18 is similar to the LAM/ivHBIG strategy except that 7 days after transplant, patients are converted and maintained on fortnightly or monthly intramuscular HBIG to keep antiHBs levels above target levels of > 500 IU/L in the 6 months posttransplant, > 250IU/L 6 to 12 months post transplant and > 150 IU/L after 1 year posttransplant. Supplemental doses of intravenous HBIG are used if antiHBs levels became subtherapeutic.

At any time point in the cycle, patients are channeled into the following Markov States: No Recurrence, Stable Recurrence (HBV surface Antigen (HBsAg) positive but HBV DNA negative by non-PCR assay),Viral Resistance (HBsAg positive and HBV DNA positive by non-PCR assay), and Death. Patients who develop viral resistance are further divided into those who continue to remain stable, those with accelerated disease resulting in death or those requiring retransplant.

The base population was a hypothetical simulated cohort of 1,207 patients identical to the 1,207 patients who underwent orthotopic liver transplant for HBV cirrhosis from 1996 to 2004 in the United Network for Organ Sharing (UNOS) database. Cost-effectiveness analysis was then compared between the three intervention strategies. Outcome measures were incremental cost-effectiveness ratio (ICER) and cost per incidence of HBV recurrence prevented and cost per life saved. Analysis was done from societal perspective.

Study Variables

Base Population (Table 1)

The base population characteristics and waiting time for transplant was obtained from UNOS database for time period January 1996 to July 2004. This cohort consists of 1,207 adult patients with a median age of 52 years and median waiting time of 2.2 years (based on OPTN data as of November 1, 2004). Projected survival expectancy (15 years) was derived from actual UNOS data for transplants performed for other causes.

Table 1. Input Variables Used To Construct Markov Chain Transition Probabilities for Base Case Analysis
 Variables used to model 0–5 years posttransplant (actual clinical data except where indicated)Variables used to model 5–15 years posttransplant (projected)Data source (reference) (up to 5 years posttransplant)
  1. Variables used in the Markov Modeling to determine cost and outcome. Variables for the first 5 years were extracted from actual best available clinical data reported in trials. Beyond 5 years, hypothetical incidence data were projected based on known existing trends.

Percent of replicative patients (HBV DNA +ve)55%55%5
Life expectancy15 years15 yearsBased on OPTN data (as of November 1, 2004)
Incidence of HBV recurrence(1:3:5 years)(10:15 years) 
Survival of HBV transplant(1:3:5 years)(10:15 years) 
Incidence of viral resistance(1:3:5 years)(10:15 years) 
 Adefovir0%:6%:10% (estimated)25%:40%14,19
Incidence of retransplant for HBV recurrence5% (estimated)35%5
Incidence of renal impairment   
 ADV stopped2% per annum (estimated)4% per annum14
 Renal replacement therapy1% per annum (estimated)2% per annum 

Incidence Rates and Variables for HBV Disease

The incidence of lamivudine resistance pretransplant and posttransplant, incidence of HBV recurrence, and patient survival up to 5 years after transplant were extracted from largest most representative clinical trials, identified by systematic PUBMED search for studies published from 1996 to 2004 reporting outcomes on the treatment described2, 5, 12, 14, 18, 19 (Table 2). In scenarios where there were no clinical data available, the best available data favoring LAM/HBIG was used. LAM/ADV data was extracted from studies by Perrillo et al.5 and Study 435 by Schiff et al.14, 19 We combined survival data from the lamivudine monotherapy trial by Perrillo et al. (for first 200 weeks) and from Study 435 thereafter (based on actual years posttransplant) for modeling LAM/ADV. Although Study 435 had excluded lamivudine resistant patients with decompensated liver disease, the patients in Study 435 have had resistance for some time before being started on adefovir (median 200 weeks posttransplant before adefovir was started although resistance occurred at median 56 weeks. Follow-up was 80 months post transplant with adefovir duration up to 144 weeks). We assumed that if adefovir had been started at onset of lamivudine resistance, the incidence of progressive liver disease would be low and survival outcome for LAM/ADV in the model should be no worse than LAM monotherapy patients or actual Study 435 patients, where adefovir was not started immediately at onset of resistance. This in effect, would bias against LAM/ADV and favor Lam/HBIG. For LAM/ivHBIG, variables from two studies (Han et al., 200012 and Steinmuller et al.2) with combined follow-up up to 60.3 months was pooled. Beyond 5 years, hypothetical incidence data was projected using computer algorithm to extrapolate the actual reported data beyond 5 years, based on reported trends and expected viral resistance patterns. LAM/imHBIG data was taken from study reported by Han et al.18 with outcomes reported at median of 1,051 days posttransplant. For the purpose of comparison, we assumed that this strategy would be as effective as the LAM/ivHBIG over the 15 years life span.

Table 2. Studies Used for Derivation of Actual Variables up to 5 Years
StudynProphylaxisFollow-up duration (median)No recurrenceSurvival
  • LAM/ADV variables and LAM/HBIG variables were pooled from the largest trials that had reported outcomes using the various treatments. Mth,month; n,number of subjects in trial.

  • *

    DNA undetectable (non-PCR assay).

Perrillo et al., 2001547LamivudineUp to 48 months60%80%
Schiff et al., 200314196ADV 200 weeks posttx48 weeks34%*93%
Schiff et al., 200419204ADV 195 weeks posttx144 weeks78%*88%
Han et al., 20001259LAM/HBIG1–61.8 (15.3) month100%98% (15 month)
Steinmuller et al251LAM/HBIG31–40 (35) month92%93% (48 month)
Han et al., 20031859LAM/im HBIG with IV supplementation1–37 (35) month98%95% (35 month)

Incidence of Adefovir Resistance and Adverse Effects

Potential adverse effects from prolonged adefovir treatment with resultant additional costs for treating renal impairment and higher recurrence rate as a result of adefovir withdrawal were taken into account.14 Although the 2% renal failure documented in study 435 could have been contributed by a multitude of factors including underlying renal disease, use of nephrotoxic drugs, and intercurrent illness, we assumed a worse case scenario of 2% renal impairment per year due to adefovir and 1% requiring renal replacement therapy. Incidence of adefovir resistance in posttransplant patients was assumed to be similar to non-transplant patients but having an increasing trend similar to lamivudine.

Costs and Outcome (Table 3)

We calculated costs of prophylaxis and treatment based on current data obtained from US MEDICARE fee schedule.20 All indirect cost to society and individual including cost of follow-up, traveling costs, and cost of nursing care were incorporated. All costs accrued from time of entering the transplant program until death were included and adjusted for inflation (estimated 2% annual inflation), discounted 3% annually, and expressed in USD (year 2005). Estimation was also made for additional cost incurred in management of possible renal impairment with long-term adefovir, regular screening liver biopsy for patients with HBV recurrence, and also the cost of accelerated disease assumed with adefovir resistance, including the need for retransplant.

Table 3. Costs of Various Treatment
 Cost (USD)Data source (reference)
  1. All costs were calculated from societal perspective to include medical cost to individual as well as societal cost such as transport and caregiver's time and resources.

Treatment cost  
 first year post transplant108,30018
 subsequent years51,300 
*includes facility fees, nursing time, transport  
 first year post transplant80,00018
 subsequent years9,120–10,000 
(include average of 1 IV supplementation/patient)  
Renal replacement therapy62,000 
Miscellaneous cost/year  
Incremental cost with recurrence (include regular 3 monthly HBV DNA testing, protocol screening liver biopsy and close follow up)8,00020
Incremental cost with accelerated disease (includes hospitalization, treatment for decompensation and complication, caregiver time, and resources)15,00020
Cost of retransplant (first year)312,000OPTN data as of Nov 1 04
Annual discount3% 

Quality of life utilities scores were adapted from those previously reported for Hepatitis C,21 assuming that utility scores for the various disease states due to Hepatitis B would be similar. Utilility scores for patients with viral resistance was calculated by averaging utilities for chronic hepatitis and compensated cirrhosis, assuming equal representation using proportionate variance similar to that before transplant. We also factored in utility scores for decompensated liver cirrhosis for the last 2 years for those who died before the end of expected life span. All effects were discounted annually by 3%.


Analysis was performed using computer software Data Pro 2004 by Treeage Software, Inc., MA. Cost-effectiveness threshold was set at USD 50,000/QALY using conventional threshold for CEA in United States.22

Sensitivity analysis was performed for each parameter using both one-way and two-way analysis to test robustness of estimation for both costs and effect in the model as well as to identify factors that affected cost-effectiveness most. Range of variables used for sensitivity analysis encompassed all relevant ranges reported in the literature. Sensitivity analysis for cost of HBIG covered the expected cost in USA, Europe, and Asia. Multi-variate analysis was also performed using worst case and best case scenarios for various parameters.


Base Case Analysis (Table 4)

Using estimated 55% positive replicative status for the base population, the Markov model predicts that the strategy of LAM/ADV would result in recurrence rate of 42% at 5 years and 65% at 15 years (Table 4). This is in contrast to either LAM/HBIG strategy (assumed to have similar efficacy), which has recurrence of 10% at 5 years and 12% at 15 years. LAM/ADV, compared to LAM/HBIG at 5 years posttransplant, would result in an additional 385 cases (32%) of HBV recurrence and an additional 20 (1.7%) HBV related deaths. At end of life expectancy (15 years posttransplant), the incremental recurrence is 640 cases (53%) and incremental number of deaths due to Hepatitis B recurrence is 92 (7.6%).

Table 4. Results
 5 Years Post Transplant AnalysisFull Life Expectancy Analysis (15 Years)
LAM/ADVLAM/ivHBIGIncremental LAM/ivHBIG over LAM/ADVLAM/imHBIGIncremental LAM/imHBIG over LAM/ADVLAM/ADVLAM/ivHBIGIncremental LAM/ivHBIG over LAM/ADVLAM/imHBIGIncremental LAM/imHBIG over LAM/ADV
  1. Outcome of Markov modeling and cost-effectiveness indicators at 5 years posttransplant as well as at end of full life expectancy (15 years). Although LAM/ADV has higher recurrence rate and death compared to LAM/HBIG, ICER of LAM/ivHBIG over LAM/ADV is USD 2.8M per QALY at 5 years posttransplant and USD760K per QALY at end of life expectancy which makes it not cost-effective. LAM/imHBIG is cheaper with ICER over LAM/ADV of USD1.22M at 5 years and USD188K at end of life expectancy. M, million; K, thousand; LY, life years.

Total cost (USD)31.4M338M306.6M164M133M135M814M678M303M167M
Cost per patient (USD)26K280K254K136K110K112K674K562K251K139K
No (%) recurrence506 (42%)121 (10%)385 (32%)121 (10%)385 (32%)785 (65%)145 (12%)640 (53%)145 (12%)640 (53%)
No (%) death64 (5.3%)44 (3.7%)20 (1.7%)44 (3.7%)20 (1.7%)173 (14%)81 (7%)92 (7.6%)81 (7%)92 (7.6%)
Life years saved per patient4.
QALY saved per patient3.763.850.093.850.099.7610.50.7410.50.74
Incremental cost/recurrence prevented  USD796K USD345K  USD1.04M USD260K
Incremental cost/death prevented  USD15.3MUSD12.7K/patient USD6.64MUSD5.5 K/patient  USD7.4MUSD6.1K/patient USD1.81MUSD1.5K/patient
Incremental cost per LY saved  USD 1.58M 687.5K  USD624K USD154K
Incremental cost per QALY saved (ICER)  USD2.8M 1.22M  USD760K USD188K

LAM/ADV strategy cost USD26,000 per patient (total USD31.4 million) at 5 years and USD112,000/patient (total USD135 million) at end of life expectancy. In contrast, LAM/ivHBIG strategy would cost USD280,000/patient (total USD338 million) at 5 years post-transplant and USD674,000/patient (total USD814 million) at end of life expectancy with a resultant incremental cost over LAM/ADV of USD254,000/patient (total USD306 million) at 5 years and USD562,000/patient (total USD678 million) at end of life expectancy. LAM/imHBIG is much cheaper than LAM/ivHBIG, costing USD136,000/patient (total USD164 million) at 5 years and USD251,000/patient (total USD303 million) at 15 years with incremental cost over LAM/ADV of USD110,000/patient (total USD 133 million) at 5 years and USD139,000/patient (total USD167M) at 15 years.

The incremental QALY lost using LAM/ADV compared to either LAM/HBIG combinations is 0.09 QALY/patient and 0.74 QALY/patient at 5 and 15 years posttransplant, respectively (Fig. 2). This translates to incremental cost-effectiveness ratio of USD2.8 million/QALY at 5 years and USD760,000/QALY at 15 years for LAM/ivHBIG over LAM/ADV. LAM/imHBIG would have an ICER of USD1.22 million at 5 years and USD188,000 at 15 years posttransplant over LAM/ADV. At the end of life expectancy, the cost of preventing each recurrence is USD1.04 million and to prevent each death is USD7.4 million for LAM/ivHBIG and for LAM/imHBIG, USD260,000 and USD1.81 million, respectively.

Figure 2.

Cost-effectiveness of strategies used in HBV transplant (per patient at end of life expectancy,15 years). Both Lamivudine and HBIG monotherapy are more expensive and less effective than LAM/ADV (dominated). Both LAM/ivHBIG and LAM/imHBIG are more effective than LAM/ADV(Incremental effectiveness: 0.74 QALY) but are much more expensive. Incremental cost of LAM/ivHBIG is USD562,000 and thus less cost-effective, as can be seen by the steep gradient between LAM/HBIG and LAM/ADV. LAM/imHBIG is much cheaper with an incremental cost of USD139,000.

Sensitivity analysis (Table 5)

Only the cost of HBIG has a significant impact on ICER. The age at which transplant was performed has a moderate impact. None of the other parameters affect the outcome of the cost-effectiveness analysis.

Table 5. Sensitivity Analysis (End of Life Expectancy Analysis)
VariableRange for sensitivity analysisLAM/ivHBIG over LAM/ADVLAM/imHBIG over LAM/ADV
ICER (USD/QALY) (Best case scenario)ICER (USD/QALY) (Worst case scenario)ICER (USD/QALY) (Best case scenario)ICER (USD/QALY)(Worst case scenario)
  • One-way sensitivity analysis was performed to test validity of outcome conclusions for each of the variables estimated. Multiple-way analysis was also performed using worst case and best case scenarios.

  • *

    The single variable that could alter the cost-effectiveness of LAM/HBIG, making it more cost-effective than LAM/ADV, is cost of HBIG. (ICER<threshold of USD50 000).M,Million;K,Thousand.

Age at transplant30 to 70401K1.5M78K235K
Incidence of HBV recurrence(1:5:15 years)594K1.2M145K356K
 LAM/HBIG0%:8%:12% to 4%:14%:25%    
 LAM/ADV10%:28%:40% to 30%:50%:80%    
Survival post transplant(1:5:15 years)544K1.8M136K389K
 LAM/HBIG98%:90%:83% to 92%:85%:75%    
 LAM/ADV95%:87%:78% to 88%:80%:60%    
Incidence of Viral resistance(1:5:15 years)402K953K158K360K
 LAM/ADV0%:2%:20% to 2%:10%:70%    
Incidence of retransplant15% to 50%600K980K167K230K
Incidence of Renal impairment 312K1.5M125K412K
ADV stopped0% to 8% per annum    
Renal replacement0% to 5% per annum    
Life Expectancy8 to 20 years450K950K117K360K
Cost (USD)     
LAM0.8K to 1.8K700K812K151K245K
ADV3K to 10K 000736K824K159K261K
HBIG (average cost/year)4K to 60K38K*900K38K900K
Waiting time to transplant0 to 5 years640K845K167K211K
Discount1% to 5%450K900K147K290K
Inflation1% to 3%670K880K160K247K

Cost of HBIG

In both LAM/HBIG strategies, cost of HBIG has the highest impact on ICER (Fig. 3). To allow comparability between strategies, we factored in the peritransplant HBIG cost and used a common variable, the average annual cost of HBIG over 15 years to perform the sensitivity analysis. In the two LAM/HBIG strategies, varying the cost of annual HBIG (in either route of administration) from USD4000 to USD60,000 showed a swing of the ICER from USD38,000 to USD900,000/QALY. Using ICER of USD50,000 as cost-effectiveness threshold, LAM/HBIG becomes cost-effective over LAM/ADV when the annual cost of HBIG dips below USD6,000 per year. Cost of Lamivudine and adefovir have minimal impact on the ICER.

Figure 3.

Sensitivity analysis: Impact of cost of maintenance HBIG on ICER. Average annual cost of HBIG has a significant impact on the ICER of LAM/HBIG over LAM/ADV. As the cost of HBIG varies in different countries and also between the modes of injection, the ICER can vary widely. Intramuscular administration uses less HBIG doses and has lower overhead cost, such as day ward, methylprednisolone, and nursing time, and thus is generally more cost-effective than LAM/ivHBIG. Using ICER of USD50,000 as threshold for cost-effectiveness, LAM/HBIG becomes cost effective when average annual cost of HBIG is less than USD6,000. If annual HBIG cost drops below USD2,500, ICER becomes negative with LAM/HBIG dominating LAM/ADV. Average annual cost of HBIG = [Cost of HBIG peri-operative period + cost of HBIG for rest of year 1 + cost of HBIG for duration of expected life span (15 years)]/expected life span (15 years).

Age at Transplant

Age at transplant has a minimal effect in LAM/ivHBIG arm as any gain in life years is negated by the high HBIG maintenance cost. However in the LAM/imHBIG group, it has a moderate impact in that younger patients would benefit more from the decreased recurrence and increased life years. A 30-year-old patient undergoing transplant with LAM/imHBIG would bring down the ICER to a more reasonable USD78,000/QALY.

Recurrence of LAM/HBIG vs. LAM/ADV and Recurrence Survival

The incidence of HBV recurrence in LAM/ivHBIG strategy has minimal impact on the ICER. Even in the best case scenario of having cumulative recurrence of 80% for LAM/ADV and 12% for LAM/HBIG at 15 years, the ICER is still high at USD594,000/QALY for LAM/ivHBIG and USD145,000/QALY for LAM/imHBIG (Table 5). Similarly, survival rates for HBV recurrence also show a similar minor impact on ICER. Best case scenario using 83% survival for LAM/HBIG and 60% for LAM/ADV, yields an ICER of USD544,000 per QALY for LAM/ivHBIG and USD136,000/QALY for LAM/imHBIG.

Risk of Adefovir Resistance, Death and Complication From Adefovir Treatment

Risk of adefovir resistance at 15 years was varied widely from 30 to 70% and risk of death from resistance was varied from 5 to 20% per annum. The ICER for best case scenario in favor of LAM/HBIG is USD402,000/QALY for LAM/ivHBIG and USD158,000 for LAM/imHBIG.

Profile of Patients Undergoing Transplant: Incidence of Replicative State Pretransplant, Median Age at Transplant and Life Expectancy

The life expectancy, waiting time, and incidence of pretransplant replicative state have minimal impact on cost-effectiveness.


With the availability of adefovir dipivoxil and accumulating evidence of its efficacy in controlling lamivudine-resistant recurrence in posttransplant patients, the cost-benefit of LAM/HBIG over LAM/ADV therapy has been called into question especially in view of the prohibitive cost of HBIG. Our cost-effectiveness analysis, using a Markov model shows that lamivudine prophylaxis followed by adefovir rescue, is the most cost-effective strategy compared to either of the LAM/HBIG combination strategies. The overwhelming cost of HBIG prophylaxis is primarily responsible for driving cost-effectiveness in favor of LAM/ADV strategy. This, however, is at the possible price of higher recurrence and mortality rate (bearing in mind the model favors LAM/HBIG) although most of the resistance to ADV and resultant death occur much later nearer the end of life expectancy and hence limit the benefit in terms of absolute and quality-adjusted life-years saved from HBV recurrence. In addition, the incremental mortality rate of LAM/ADV, attributed to HBV recurrence (7.6% at 15 years), is likely to be lower in reality as LAM/HBIG trials, from which mortality rates were derived, were more recent than those of LAM/ADV and improvements in surgery and medical care may have accounted for some of the difference.

We presented both the 5-year posttransplant and full life expectancy analysis comparing the cost-effectiveness of the two strategies. While the 5-year posttransplant data is based on actual clinical data, we feel the 15-year full life cycles analysis, although based on extrapolated hypothetical data, is more representative of the actual cost-effectiveness as the 5-year analysis may underestimate life savings that will accumulate over time with every life saved.

Owing to the prohibitive cost of HBIG, many centers in US have adopted strategies to reduce HBIG doses. These include using lower doses of HBIG, “on demand” HBIG based on a target anti-HBsAb level, short-term HBIG or intramuscular HBIG18, 23, 24 (Fig. 4). However, reports supporting the use of these strategies have involved only small samples with short follow-up, and the efficacy needs to be further confirmed with large-scale long-term studies. In our model, we used the two arms of LAM/ivHBIG (fixed high-dose) and modified LAM/imHBIG (intramuscular on demand and intravenous supplementation) to illustrate the upper limit of cost-effectiveness and how this change with the modified lower HBIG doses, assuming that the modified regimen are as efficacious as LAM/ivHBIG. Shouval et al.23 recommended limiting the dose of HBIG using an “on demand” regimen with different target anti HBsAb levels for patients with different risks. Lok et al.11 proposed using HBIG only for the peri-transplant period in native non-replicators. Using our model, these two strategies compared to LAM/ADV, would reduce the ICER to USD270,000 and USD430,000/QALY, respectively. The Australian experience24 of using a very low dose of intramuscular HBIG without intraoperative HBIG have the lowest ICER of USD58,000 per QALY, just short of the ICER threshold for cost-effectiveness. However the concern of having no intraoperative protection when the theoretical risk of HBV recurrence is highest, requires further evaluation.

Figure 4.

ICER of various HBV prophylaxis strategies compared to LAM/ADV by cost-effectiveness analysis. The high cost of intravenous HBIG makes any combination strategy highly not cost-effective compared to LAM/ADV (A-D). Only use of low-dose intramuscular HBIG for maintenance without intraoperative HBIG (E) brings the ICER close to the cost-effectiveness threshold of USD50,000/QALY. IV, intravenous; IM, intramuscular; IU, International Unit; U, Unit.

In estimating variables for the analysis, we used all available published data to justify our estimation. Sensitivity analysis was performed on as wide a range of variable as possible, exceeding the range of variables reported in the literature. Estimations were deliberately conservative, favoring LAM/HBIG as the gold standard treatment.

Previous studies have looked at cost-effectiveness of HBV transplant in terms of HBV recurrence rather than by survival. However, the definition of HBV recurrence has not been standardized and accounts for differences in reported recurrence rates for lamivudine monoprophylaxis. Studies using positive HBsAg as recurrence resulted in reports of higher recurrence compared to those who use positive HBV DNA (non PCR assay) as recurrence criteria. Lo et al.4 reported six patients, who had positive HBsAg posttransplant but undetectable HBV DNA even by PCR assay. These patients showed no evidence of recurrent disease up to 37 months posttransplant. It is likely that positive HBV DNA with positive HBsAg reflect the highest risk of recurrent HBV disease. Data from our transplant center showed similar findings.25 In our model, we have used recurrence rates for lamivudine prophylaxis from studies where the criterion for recurrence was defined as positive HBsAg only. This definition leads to an overestimation of meaningful HBV recurrence as patients successfully controlled with adefovir but are HBsAg positive would be included. Thus the true incidence of clinically significant recurrence would likely be lower than that estimated in our model.

Adefovir has been used in treatment for HBV for up to 5 years but was only licensed in the USA in 2002. While reported incidence of side effects and resistance has been low,26 the lack of long-term safety data cautions against its generalized use as first line therapy. In our model, we projected long-term efficacy, side effects, and resistance based on current viral kinetics patterns and further factored in a conservative buffer to avoid a type I error. We estimated a generous resistance rate of 40% to adefovir after 15 years of transplant, 35% retransplant rate and 35% death rate from HBV resistance as well as 2% annual incidence of developing renal failure requiring renal replacement therapy. In reality, these rates are probably exaggerated and even in patients with adefovir resistance, the mutant viruses may possibly be susceptible to the other class of antiviral currently in development, and hence altering their outcome. To date, study 43514 is the only published study of adefovir use in liver transplant patients. However, this study is in design quite different to LAM/ADV strategy used in our cost effective analysis. Patients in study 435 did not receive adefovir immediately upon development of lamivudine resistance, and by calculation, the median time post-transplantation for starting lamivudine was 144 weeks after development of therapeutic failure to lamivudine. At this stage, 25% of patients had already developed cirrhosis as a result of uncontrolled lamivudine resistance.

There are several assumptions and limitations in our analysis. All patients were assumed to be 100% compliant to treatment strategy although in real life, non-compliance can be a problem and may result in life-threatening flares or higher incidence of mutation especially for patients on the LAM/ADV arm, resulting in treatment failure and higher cost. Viral kinetics to adefovir was assumed to follow a similar pattern to other nuclear analogues and the model may not adequately address sudden abrupt changes in recurrence and death rates in time. We also assumed that regular monitoring will timely identify all patients who develop lamivudine resistance and adefovir can be started before deterioration occurred. In reality, some patients who develop resistance may develop rapid progressive liver failure, too fast for adefovir to reverse. We excluded all confounding factors such as transplants for co-infection with HCV or HIV and hepatocellular carcinoma, as the outcomes for these scenarios are different. We have also not considered the option of combination lamivudine with adefovir pretransplant, adefovir as first line monoprophylaxis or use of entecavir as salvage drug, as there are no published long-term studies of such strategies in liver transplant patients. While these strategies would theoretically reduce HBV resistance and result in even more cost savings, more clinical data would be required to explore their efficacy in this setting.

In our model, although there is increase in recurrence of HBV of 54% at end of 15 years in the LAM/ADV group, these recurrences are defined technically by positive HBsAg, while true clinically important outcomes such as QALY only show a marginal benefit for LAM/HBIG despite the model being heavily biased against LAM/ADV. In other words, LAM/HBIG is highly effective at preventing recurrence but is expensive and comes with the inconvenience of monitoring anti-HBs levels, and intravenous or intramuscular therapy. LAM/ADV compromises by accepting a higher recurrence (based on existing data) but by controlling viral activity, a reasonable outcome by cost-effectiveness indicators is achieved. LAM/ADV is also considerably more convenient with once daily oral dosing. Successful prolonged suppression of HBV DNA by adefovir, with undetectable DNA (PCR negative) increasing to 78% of patients after 144 weeks of treatment,19 probably does translate into long-term survival, although long-term studies would be needed to confirm such an assumption.

How much is society willing to pay for the added benefit of a superior healthcare strategy? The cost of HBIG varies widely depending on location and includes not just the drug cost but the facility cost of administrating the drug. Rates quoted include USD750 for intramuscular dose of 1,500 IU in USA18 to EUR1,270 (USD1,552) for 2,000IU dose in Italy27 and USD1,130 for 2,000 IU dose in Asia (Drug cost in Singapore, Fig. 3). In first world countries where willingness-to-pay threshold may be higher or where HBIG prices are low, such as in United Kingdom where IV HBIG maintenance is estimated at USD18,000 per year,28 lamivudine with low-dose HBIG may be the preferred strategy, especially in young patients who would benefit most from the decreased HBV recurrence and recurrence related deaths. In contrast, in many countries in Asia where resources are limited such as China, where the GNI per capita in year 2003 was mere USD4,980 compared to USD 37,610 in USA,29, LAM/ADV offers the most cost-effective strategy for QALY saved. Several new formulations of HBIG are currently being developed with the potential of having longer duration of action and hence, overall lower dose requirement for prophylaxis. If the annual cost of HBIG can be brought down to USD6,000 annually, LAM/HBIG will become more cost-effective than LAM/ADV. In reality, it is likely that a tailored regimen, based on the risk of recurrence of each individual patient, local cost of HBIG, age of patient undergoing transplant, and the country's willingness-to-pay threshold, will improve the overall cost-effectiveness. We believe our study adds to the information needed when strategizing the optimal prophylactic strategy.

In conclusion, lamivudine prophylaxis followed by adefovir rescue when lamivudine resistance develops, using highly conservative estimates, is the most cost-effective strategy in terms of QALY saved. The cost of HBIG is the primary factor affecting cost-effectiveness. Modified regimens of low-dose HBIG with lamivudine may be the optimal strategy for countries with high willingness-to-pay threshold and where maximal efficacy in terms of recurrence prevention is desired, such as for high-risk groups. Prospective clinical trials are needed to determine the most optimal strategy based on individual risks and confirm the potential savings.


We thank Prof Li Shu-Cheun for critical review of the manuscript. The content is the responsibility of the authors alone and does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.