SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Author contributions
  8. Data Sharing
  9. References
  10. Supporting Information

Both liver resection (LR) and cadaveric liver transplantation (CLT) are potentially curative treatments for patients with hepatocellular carcinoma (HCC) within the Milan criteria and with adequate liver function. Adopting either as a first-line therapy carries major cost and resource implications. The objective of this study was to estimate the relative cost-effectiveness of LR against CLT for patients with HCC within the Milan criteria using a decision analytic model. A Markov cohort model was developed to simulate a cohort of patients aged 55 years with HCC within the Milan criteria and Child-Pugh A/B cirrhosis, undergoing LR or CLT, and followed up over their remaining life expectancy. Analysis was performed in different geographical cost settings: the USA, Switzerland and Singapore. Transition probabilities were obtained from systematic literature reviews, supplemented by databases from Singapore and the Organ Procurement and Transplantation Network (USA). Utility and cost data were obtained from open sources. LR produced 3.9 quality-adjusted life years (QALYs) while CLT had an additional 1.4 QALYs. The incremental cost-effectiveness ratio (ICER) of CLT versus LR ranged from $111,821/QALY in Singapore to $156,300/QALY in Switzerland, and was above thresholds for cost-effectiveness in all three countries. Sensitivity analysis revealed that CLT-related 5-year cumulative survival, one-time cost of CLT, and post-LR 5-year cumulative recurrence rates were the most sensitive parameters in all cost scenarios. ICERs were reduced below threshold when CLT-related 5-year cumulative survival exceeded 84.9% and 87.6% in Singapore and the USA, respectively. For Switzerland, the ICER remained above the cost-effectiveness threshold regardless of the variations. Conclusion: In patients with HCC within the Milan criteria and Child-Pugh A/B cirrhosis, LR is more cost-effective than CLT across three different costing scenarios: the USA, Switzerland, Singapore. (Hepatology 2015;61:227–237)

Abbreviations
CLT

cadaveric liver transplantation

DEALE

declining exponential approximation of life expectancy

HCC

hepatocellular carcinoma

ICER

incremental cost-effectiveness ratio

LR

liver resection

LT

liver transplantation

OPTN

Organ Procurement Transplant Network Database

QALYs

quality-adjusted life years

Hepatocellular carcinoma (HCC) is a common cancer with a higher disease burden in East Asia due to the prevalence of chronic viral hepatitis in the region.[1] For patients with HCC within the Milan criteria and adequate liver function for resection (Child-Pugh A/B cirrhosis), both liver resection (LR) and liver transplantation (LT) are potentially curative treatments. There is, however, ongoing debate about which treatment is more appropriate.[2, 3] While specialized centers demonstrate that LT for HCC within the Milan criteria can achieve 5-year survival rates of more than 80%,[4-7] reports from large registries such as the Organ Procurement Transplant Network Database (OPTN) show 5-year survival rates of between 60-65%.[8] Emerging data show improving trends in LR outcomes, possibly due to better patient selection, postoperative management, and multimodality treatment for recurrences.[9] While recurrence rates in LT are often lower than LR, LT faces considerable resource challenges in many countries, such as the limited supply of cadaveric transplant organs, leading to cancer progression during long waiting times that limit intention-to-treat survival for LT.[10] LT patients also require long-term immune suppression, with attendant risks and significant lifetime costs.

Randomized trials comparing the two treatments are neither ethical nor practical. Due to the large financial outlay and recurrent costs needed to run a liver transplant program for a large number of HCC patients,[11] the decision to adopt either therapy as a first-line option carries major implications with respect to costs, utility of scarce resources, and expectations of the population for quality healthcare. It is in this setting that a decision analytical model can help compare the two treatments from the perspective of the healthcare system. The aim of this study was to estimate the relative cost-effectiveness of LR versus LT for HCC patients within the Milan criteria and adequate liver function using a decision analytic model.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Author contributions
  8. Data Sharing
  9. References
  10. Supporting Information

A Markov cohort model[12] was developed to simulate a cohort of patients aged 55 years with early HCC (defined by the Milan criteria) and adequate liver function (defined as Child-Pugh A/B cirrhosis), who have undergone either LR or cadaveric liver transplantation (CLT) and followed over a time horizon of their remaining life expectancy (Fig. 1). Among the two alternatives, the baseline comparator is LR. Early HCC was defined as HCC meeting the Milan criteria (solitary nodule not exceeding 5 cm; not more than three nodules, none exceeding 3 cm; no evidence of macrovascular invasion or distant metastasis).[13] While living donor liver transplantation is increasing, especially in Asia,[14] this study is focused on CLT as it remains the primary form of transplantation internationally.[15]

image

Figure 1. Markov Cohort Model. Lines with arrows indicate transition from one health state to another (or itself) per cycle.

Download figure to PowerPoint

In the model, a 1-month cycle time was selected as the clinically most pertinent interval short enough to reflect the cost and quality of life impact from major surgical treatment episodes.[16] During each cycle, a patient occupied one health state and moved to another according to transition probabilities (Fig. 1). Each health state had associated costs and utilities, and these were accumulated over the lifetime of the simulated population. Monthly transition probabilities were derived from cumulative probabilities using the declining exponential approximation of life expectancy (DEALE) method.[17]

Literature reviews were conducted to derive values for the transition probabilities, utilities, and costs. A full list of parameters used in the model is presented in Tables 1 to 3. Details on parameters used to derive transition probabilities between two states and references used to derive values for these parameters are found in Supporting Material 1. When possible, transition probabilities for LR were obtained from studies from Lim et al.,[9] which is a systematic review of recent LR outcomes in early HCC patients. This is supplemented by additional analyses performed on the Singapore General Hospital (SGH) LR database comprising 241 patients (Supporting Material 2), and the OPTN database for LT comprising 4,381 patients (Supporting Material 3). In the absence of published data, several modeling assumptions listed below were made.

Table 1. Base Case Value and Sensitivity Range Extracted From Literature for Transition Probabilities
ParameterBase Case Value (Median of Literature Range Unless Separately Referenced)Literature Range Tested
  1. a

    Annual decompensation rate was converted to monthly probability using the formula: 1-e (-rate x time), where time is 1/12.

  2. b

    Using the assumption of a declining exponential approximation of life expectancy, the monthly probability was derived from 5-year cumulative survival by using the formula: 1-(prob)1/60, where prob is 5-year cumulative survival.

  3. c

    Using the assumption of a declining exponential approximation to disease- or progression- free survival, post-LR 5-year cumulative recurrence rate, annual mortality risk and 24 month dropout risk from tumor progression were converted to the monthly probability using the formula: 1-(1-prob)1/time, where prob refers to the parameter and time refers to 60, 12, 24 respectively.

  4. d

    Annual mortality risk of recurrent HCC is first calculated from post recurrence 5-year cumulative survival or time to median survival by using 1-(prob)1/time, where prob refers to the 5-year cumulative survival or 50% for median survival, and time refers to 5 or 1.42 years.

  5. e

    Monthly probability of getting CLT is calculated by using the formula 1-(1-prob)(365/12)/time, where prob is 50% and the time is median wait list time to CLT.

  6. #Also refers to Supporting Material 2.

  7. **Also refers to Supporting Material 3.

  8. ††Refer to Supporting Material 8 for detailed list of supporting references used.

Background (all-cause) mortality[40]Age-specific 
Cirrhosis annual decompensation rate (%)[18-21]11.8[18]3.9-12.5
Derived monthly decompensation rate (%)a0.980.32-1.04
Decompensated cirrhosis-related 5-year cumulative survival (%)[19, 20, 22-24]3514-50.8
Derived monthly mortality risk (%)b1.731.12-3.22
LR-related 30-day perioperative mortality (%)[25, 26, 31, 35-39]2.20-5.3
Post-LR 5-year cumulative recurrence rate (%)[25],[27-31],#6242-78
Derived monthly recurrence rate (%)c1.600.90-2.49
Annual mortality risk of recurrent HCC (%)[26],[28],d,#30.123.6-38.7
Derived monthly mortality risk (%)c2.942.22-4.00
Median waiting-list time to CLT (days)7,aa,bb110aa36-234
Derived monthly probability of obtaining CLT (%)e17.448.62-44.33
24 month dropout risk due to tumor progression (%)[41]53[41]-
Monthly dropout risk due to tumor progression (%)bb-2.3-5.6
Derived monthly dropout risk (%)c3.102.3-5.6
Annual mortality risk of HCC outside Milan (%)[44-47]44.233.8-51.0
Derived monthly mortality risk (%)c4.753.38-5.77
CLT-related 30-day perioperative mortality (%)[35],aa,bb3.2aa0-9.88
CLT-related 5-year cumulative survival (%)[3, 5, 7, 8, 25],[35],aa,bb67.8aa60-90
Derived monthly mortality risk (%)b0.650.18-0.85
Table 2. Base Case Value and Sensitivity Range for Costs
ParameterBase Case Value (Median or Point Estimate From Literature)Range Tested (50% to 200% of Base Case Value)
  1. a

    Refer to Supporting Material 8 for detailed list of supporting references used.

  2. b

    Baseline derived from sum of values obtained from the references.

  3. c

    Cost of 3 TACE sessions divided by X months, where X is the estimated average survival of the patient calculated from annual mortality rate using the assumption of a declining exponential approximation to survival.

  4. §Based on expert opinion.

  5. Abbreviation: TACE, transarterial-chemoembolization.

Costs in USA ($)  
One time cost of surgical treatments  
LR costa25,08612,543–50,172
CLT cost[16],a137,70168,851–275,402
TACE per sessiona25,96112,981–51,922
Monthly follow-up cost  
Compensated cirrhosis[16],a6131-122
Decompensated cirrhosisa1,519760-3,038
Post-CLT in Year 1a5,410b2,705-10,820
Post-CLT in Year 2 and onwardsa958344-2,215
HCC recurrence1,770c885-3,540
Contraindication to CLT3,894c1,947-7,788
Costs in Switzerland ($)  
One time cost of surgical treatments  
LR costa59,84829,924-119,696
CLT cost[57],a223,236111,618-446,472
Unit cost of TACE per session[57]6,8553,428-13,710
Monthly follow-up cost  
Compensated cirrhosisa4221-84
Decompensated cirrhosisa1,889945-3,778
Post-CLT in Year 1a2,6401,320-5,280
Post-CLT in Year 2 and onwardsa1,503752-3,006
HCC recurrence467c234-934
Contraindication to CLT1,028c514-2,056
Costs in Singapore ($)  
One time cost of surgical treatments  
LR cost§12,6986,349-25,396
CLT cost§158,73079,365-317,460
TACE per session§5,9522,976-11,904
Monthly cost of follow-up  
Compensated cirrhosis[63]7940-160
Decompensated cirrhosis[63]1,029515-2,058
Post-CLT in Year 1[63]1,159580-2,318
Post-CLT in Year 2 and onwards[63]713357-1,426
HCC recurrencec406c203-812
Contraindication to CLTc893c447-1,786
Table 3. Base Case Value and Sensitivity Range Extracted From Literature for Utilities
ParameterBase Case Value (Median of Literature Range)Literature Range Tested
  1. a

    Refer to Supporting Material 8 for detailed list of supporting references used.

Compensated cirrhosisa0.760.65-0.90
Decompensated cirrhosisa0.660.37-0.86
Post-CLT in Year 1a0.690.64-0.71
Post-CLT in Year 2 and onwardsa0.730.62-0.84
Contraindication to CLTa0.630.26-0.86
HCC recurrencea0.630.26-0.86
Base Case Clinical Estimates in the LR Arm

We assumed that patients in the LR arm underwent surgery within a cycle time of 1 month and were in a state of compensated cirrhosis. Here, they were subjected to the risks of decompensation, tumor recurrence, and age-related mortality. As there was a paucity of prospective data on risk of progression to decompensated cirrhosis in this group of patients, we obtained data from studies involving non-HCC patients with cirrhosis. Further, we assumed that the history of LR for early HCC does not alter these two risks. The annual decompensation rate of 11.8% was based on Fleming et al.,[18] which was the latest study with the largest sample size among the relevant studies.[18-21] The remaining studies provided the range for the sensitivity analysis. The decompensated cirrhosis 5-year cumulative survival of 35% was the median value derived from another series of studies.[19, 20, 22-24]

The 62% post-LR 5-year cumulative recurrence rate and the 30.1% annual mortality risk of recurrent HCC were the medians derived from studies included in Lim et al.[25-31] and data from the SGH LR database (Supporting Material 2). The modeling of the recurrence state was necessary, as LR patients generally suffer from frequent recurrence, and an increasing variety of therapies are used to manage recurrences in actual clinical settings, impacting overall cost-effectiveness of the treatment.[9] As there was no universally accepted protocol for treating tumor recurrences,[32-34] it was assumed that patients underwent a variety of therapies but had a consolidated postrecurrence outcome. We assumed the cost of these treatments to be equivalent to three transarterial-chemoembolization (TACE) sessions. The 30-day postoperative mortality of 2.2% was the median derived from studies within Lim et al.'s review.[25, 26, 31, 35-39] Age-related mortality risks were based on United States life tables.[40]

Base Case Clinical Estimates in the CLT Arm

Similarly, we assumed that patients waiting for CLT started in a state of compensated cirrhosis and were at risk of decompensation, tumor progression (beyond the Milan criteria), and age-related mortality. Patients with early HCC were assumed to carry no additional short-term cancer mortality risks, and were subjected to the same annual decompensation rate of 11.8% for non-HCC patients. The risk of tumor progression to beyond Milan criteria was assumed to be constant and independent from the state of cirrhosis. Very few studies evaluated the dropout rate from the LT waitlist due to tumor progression. Only one study reported a 53% cumulative dropout probability at 24 months among 103 HCC patients,[41] which was adopted as the base case estimate. In USA, the state of cirrhosis affects the LT waiting time through the Model for Endstage Liver Disease (MELD) score.[42] However, we opted to apply a constant transplantation probability irrespective of their state of cirrhosis to simplify the model, as not all countries adopt this prioritization scheme. The median waiting-list time to CLT of 110 days was taken from our OPTN analysis (Supporting Material 3). Likewise, the CLT-related 30-day perioperative mortality of 3.2% and 5-year cumulative survival of 67.8% were derived. Once tumor progressed, CLT was contraindicated and patients were assumed to receive only up to three sessions of TACE. We could not find any study reporting mortality risk specifically in patients who dropped out of the transplantation waiting list due to tumor progression. Hence, it was assumed to be similar to the mortality risk of unresectable HCC treated with TACE. Based on a recent review on TACE for unresectable HCC,[43] four randomized controlled trials involving patients with Child-Pugh A/B cirrhosis were selected to derive the median annual mortality risk of HCC beyond Milan at 44%.[44-47] We did not model post-CLT recurrence as a separate state as these recurrences usually have poor survival[48] and there is paucity of suitable data. Patients in decompensated cirrhosis state were assumed to carry only the mortality risk from decompensated cirrhosis and not from cancer.

Costs

Our analysis was performed from the perspective of the healthcare system, and hence only direct medical costs were accounted for. Rather than using hypothetical cost data, we chose to build cost scenarios using actual country-specific data reported from open sources. Medline was searched for primary cost studies in English and published between January 1, 2000 to December 31, 2012, with the following Medical subject heading (MESH) terms and their free text variants: “Carcinoma, Hepatocellular,” “Liver Transplantation,” “Hepatectomy,” “Liver Cirrhosis” in combination with “Costs and cost analysis.” Complete articles were short-listed from potentially relevant citations and retrieved. The reference lists of these articles were also examined. Articles that reported charges but not costs, overlapping datasets, and those without specific costs applicable to the health states in our model were excluded. Adopting this approach, the most relevant and complete data were available from Singapore, the USA, and Switzerland. In addition, the large OPTN liver transplantation registry was also from the USA. Incidentally, these countries represent different costing scenarios over three continents. All reported costs were converted to year 2012 U.S. dollars[49] using the U.S. Consumer Price Index for Medical Care Costs.[50] Costs obtained from the literature were used as our base case. If a range of data was available, the median cost was used as the base case value. Expert opinion on cost supplemented the information if no cost was available from the literature. Both utilities and costs were discounted by 3% yearly.[51]

The endpoint used to assess both health benefits and costs was the incremental cost-effectiveness ratio (ICER), which was the difference in costs divided by the corresponding difference in quality-adjusted life years (QALYs). For the USA, we adopted the commonly cited cost-effectiveness threshold of $50,000/QALY.[52] For Switzerland and Singapore, we adopted the value proposed by the WHO[53] for highly cost-effective interventions (less than GDP per-capita), as there was no established threshold, $51,507/QALY for Switzerland[54] and $50,123/QALY for Singapore.[55]

Sensitivity Analyses

One-way sensitivity analysis was performed for all transition probabilities, costs, and utilities. For transition probabilities and utilities, the sensitivity analysis was done by varying each parameter over the range of variations reported in the literature and evaluating the impact on the outcome. For costs, given the lack of range data for most parameters, a wider range of 50%-200% on the base case value was deemed sufficient for cost estimates to account for potential cost disparities in centers with different volumes of patients (Tables 1−3). The transition probabilities within the top five most sensitive parameters were then further analyzed using scenario analysis, which assumed the most pessimistic extremes of these parameters to evaluate their impact on model outcome.

Probabilistic sensitivity analysis using the Monte-Carlo simulation was also performed. Point estimates for key input parameters were replaced with distributions, from which random draws were made during 10,000 reiterations. The uncertainty in all binary probability parameters was assumed to have a beta distribution, cost parameters were assumed to have a log-normal distribution, and the parameter start age was assumed to have a normal distribution (Supporting Material 4). TreeAge Pro (TreeAge Software, Williamstown, MA) was used for modeling.

Results

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Author contributions
  8. Data Sharing
  9. References
  10. Supporting Information
Base Case Analysis

The model projected 1-year, 3-year, and 5-year overall survival of 94%, 74%, 53% for LR arm, and 89%, 69%, and 57% for CLT arm (Supporting Fig. S5). LR treatment provided an average gain of 5.5 years life expectancy, while CLT offered 1.8 years more. With quality of life adjustments, LR treatment resulted in 3.9 QALYs while CLT had an additional 1.4 QALYs. ICERs of CLT compared to LR ranged from $111,821/QALY in Singapore to $156,300/QALY in Switzerland (Table 4). The ICERs were above the country-specific thresholds for cost-effectiveness in all three countries, suggesting that CLT was consistently not a cost-effective option compared to LR in all three cost scenarios.

Table 4. Incremental Cost-Effectiveness Ratios Comparing Liver Resection Versus Cadaveric Liver Transplantation in the Three Countries at the Base Case
Various CostsUSASwitzerlandSingapore
LRCLTLRCLTLRCLT
QALYs (years)3.95.33.95.33.95.3
Incremental QALY gained (years)1.41.41.4
Lifetime cost (US$)81,870244,876100,256320,37939,097196,578
Incremental cost (US$)163,005220,122157,481
ICER (US$)115,743156,300111,821
Cost-effectiveness threshold (US$)50,00051,50750,123
Is CLT cost-effective?NoNoNo
Sensitivity Analysis

Figure 2 demonstrates the tornado diagrams of the various input parameters for the USA, Switzerland, and Singapore. CLT-related 5-year cumulative survival was the most sensitive parameter in all three scenarios. The next most sensitive parameters were one-time treatment cost for CLT, utility, and monthly cost for post-CLT in year 2 and onwards, and post-LR 5-year cumulative recurrence rate. For the USA, if the CLT-related 5-year cumulative survival was improved to 87.6% or above, the corresponding ICER would reduce below the threshold and be more cost-effective compared to LR. Likewise, a similar outcome was achieved in Singapore if CLT-related 5-year cumulative survival was 84.9% or above. All other parameters were not sufficiently sensitive to bring ICER below the cost-effectiveness threshold. For Switzerland, ICER was always above the cost-effectiveness threshold regardless of the variation in any single parameter.

image

Figure 2. Tornado diagrams of one-way sensitivity analyses for three countries. Blue color indicates low-range input values, while red color indicates high-range. Cost-effectiveness threshold is indicated by the broken blue line. Ranges are shown in Table 1. TACE: transarterial-chemoembolization. (A) USA: The ICER will be below the cost-effectiveness threshold when the CLT-related 5-year cumulative survival increases to 87.6% or above. (B) Switzerland: As long as parameters were varied within the ranges, the ICER was always above the cost-effectiveness threshold. (C) Singapore: The ICER will be below the cost-effectiveness threshold when the CLT-related 5-year cumulative survival increases to 84.9% or above.

Download figure to PowerPoint

Scenario Analysis

The transition probabilities within the top five most sensitive parameters were CLT-related 5-year cumulative survival and post-LR 5-year cumulative recurrence rate. These were included in the scenario analysis. Annual mortality risk post-LR recurrence, which represented the outcomes of different modalities used to treat post-LR recurrences such as TACE or radiofrequency ablation (RFA), were also included in the scenario analysis. This was because of uncertainties from the lack of consensus for treating post-LR recurrences and varied combinations employed by different centers tailored to local needs. Assuming the most pessimistic post-LR recurrence rate and annual mortality risk post-LR recurrence, and CLT-related 5-year cumulative survival of 80%, ICER would be $52,967 in the USA, $71,661 in Switzerland and $48,622 in Singapore (Fig. 3A). In the same scenario, one-time cost of CLT would need to be kept below $125,000, $134,000, $166,000 for CLT to be cost-effective in the USA, Switzerland, and Singapore, respectively (Fig. 3B-D). The corresponding one-time CLT cost would have to be less than $73,000 in the USA, $93,000 in Switzerland, and $107,000 in Singapore, if the 5-year survival rate for CLT was subsequently reduced to 70%. In the pessimistic scenario of LR, if the one-time CLT cost was kept at the base case estimate, the CLT-related 5-year cumulative survival would need to be above 83% and 79% for CLT to be cost-effective in the USA and Singapore, respectively. CLT was not cost-effective even at 90% CLT-related 5-year cumulative survival in Switzerland.

image

Figure 3. Scenario analyses: the most pessimistic scenario for liver resection. (A) Impact of CLT-related 5-year cumulative survival on ICER. (B) USA: Combined impact of CLT-related 5-year cumulative survival and one-time treatment cost for CLT on ICER. (C) Switzerland: Combined impact of CLT-related 5-year cumulative survival and one-time treatment cost for CLT on ICER. (D) Singapore: Combined impact of CLT-related 5-year cumulative survival and one-time treatment cost for CLT on ICER.

Download figure to PowerPoint

Probabilistic Sensitivity Analysis (PSA) and Willingness-to-Pay Analysis

For the three countries, CLT could gain more QALYs but ICER would exceed country-specific cost-effectiveness thresholds. For the USA, median ICER was $110,712/QALY ($72,400 to $315,114/QALY for 95% of the trials). The median was $151,599 ($99,020 to $437,213/QALY) and $108,889 ($69,139 to $303,741/QALY) for Switzerland and Singapore, respectively. Overall, CLT was shown to be less cost-effective as compared to LR consistently across all three countries (Supporting Fig. S6).

The willingness-to-pay analysis (Supporting Fig. S7) indicated the proportion of trials that attained cost-effectiveness of a given strategy for willingness-to-pay thresholds up to 3 times GDP per capita of a specific country. For the USA, all trials showed that LR is the optimal strategy at a willingness-to-pay threshold of $50,000/QALY. If decision makers were willing to pay $111,079/QALY or higher, CLT would likely be favored as supported by more than 50% of trials. Likewise, for Switzerland and Singapore, the willingness-to-pay thresholds have to be at least $152,658/QALY and $109,471/QALY, respectively, for CLT to be more favorable than LR.

Discussion

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Author contributions
  8. Data Sharing
  9. References
  10. Supporting Information

Our results indicated that although CLT offered improved life expectancy, it was not cost-effective compared to LR across all three countries at the base case. One-way sensitivity analysis also revealed LR to be more cost-effective unless CLT-related 5-year cumulative survival was greater than 87.6% for the USA or 84.9% for Singapore. LR was always more cost-effective regardless of the variation in any single parameter for Switzerland. When CLT-related 5-year cumulative survival was less than 83% in the USA or 79% in Singapore, LR was always more cost-effective even with the most pessimistic post-LR 5-year cumulative recurrence rate and annual mortality risk of recurrent HCC. Although it has been demonstrated that patients with HCC undergoing CLT could achieve 5-year survival rate greater than 80% in specialized centers,[4-7] emerging data from OPTN and the European Liver Transplant Registry reported 5-year survival of 65% and 66%, respectively.[8, 56] This study revealed that at CLT-related 5-year cumulative survival of 70%, the one-time cost of CLT needed to be kept below $73,000 for CLT to be more cost-effective than LR in the USA.

As it is widely accepted that the etiology and natural history of HCC varies with the geographic locations, our sensitivity analysis tested the entire range of outcomes that is available from the published literature, and included studies from Europe, the USA, and Asia. Our model and results can thus be applied to any country, provided clinical outcomes (and costs) in that country fall within the range we have tested.

To our knowledge, our model is the first to use a distinct health state to account for the impact of non-CLT treatments used in managing post-LR recurrence. Previous modeling studies had confined such treatments only to palliation,[57] which may not be reflective of clinical practice today, as many centers employ various modalities to manage recurrences.[28, 58, 59] Many of these treatments such as RFA and TACE could possibly achieve good outcomes of more than 50% on 5-year overall survival in selected patients.[27, 60] Emerging treatment such as selective internal radiation therapy using yittrium-90 has also shown promise.[61, 62] These efforts would likely continue to improve postrecurrence outcomes for LR patients and make it an increasingly effective option. The outcomes from these and newer modalities can be easily incorporated into our model by updating the relevant transition probabilities.

A key limitation of this study was the paucity of published data on LR and CLT outcomes that are specific for HCC with Child-Pugh A/B cirrhosis. While many studies had been conducted for this group of patients undergoing LR, the overwhelming majority were retrospective studies. Many of the studies on CLT did not specifically report outcomes for patients with Child-Pugh A/B cirrhosis, as CLT was mainly indicated for patients with poor liver function at most centers. However, as more specific data become available more precise estimations of model parameters by regions or countries can be made, and our decision analytic model will progressively improve and strengthen.

Another limitation involved the estimation of utility values for different health states. There was little data on direct patient-elicited utilities for different stages of liver diseases such as early HCC, advanced HCC, and post-LR with or without recurrence. In the absence of large longitudinal studies, utility estimates for chronic cirrhosis conditions were subjected to the high heterogeneity of using different estimation methods and different patient samples in available studies. There was little data available on country-specific cost estimates for each health state in the model.[63] The evidence synthesis was further challenged by two major factors: 1) different methodologies were used to assess costs within and across countries; 2) studies were completed in different time-periods with majority reporting costs of historical cohorts over a decade ago. Thus, it was difficult to make any intercountry comparison or direct generalization on the cost estimates. The uncertainties associated with these limitations, however, have been accounted for by testing wide ranges of input parameters in sensitivity analyses to come to a robust conclusion.

In conclusion, patients with HCC within the Milan criteria and Child-Pugh A/B cirrhosis, liver resection is more cost-effective than CLT, across three different costing scenarios: the USA, Switzerland, and Singapore.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Author contributions
  8. Data Sharing
  9. References
  10. Supporting Information

The authors thank Duke-NUS/SingHealth Academic Medicine Research Institute for support and the writing assistance of Taara Madhavan (Associate, Clinical Sciences, Duke-NUS Graduate Medical School). The authors also thank the Organ Procurement Transplant Network Database (OPTN) for providing the data dated 30 June 2010.

Author contributions

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Author contributions
  8. Data Sharing
  9. References
  10. Supporting Information

Study concept and design: All authors; Acquisition of data: Lim, Wang, Siddiqui; Analysis and interpretation: All authors; Draft of article: Lim, Wang; Critical evaluation of the article for important intellectual content: All authors; Final approval of the version to be published: All authors; Guarantor: Lim, Wang.

Data Sharing

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Author contributions
  8. Data Sharing
  9. References
  10. Supporting Information

Consent was not obtained but the presented data are anonymized and risk of identification is low.

References

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Author contributions
  8. Data Sharing
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Materials and Methods
  4. Results
  5. Discussion
  6. Acknowledgment
  7. Author contributions
  8. Data Sharing
  9. References
  10. Supporting Information

Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/10.1002/hep.27135/suppinfo.

FilenameFormatSizeDescription
hep27135-sup-0001-suppinfo01.pdf484KSupplementary Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.