Liver Transplant Program, Division of Gastroenterology and Hepatology, Department of Medicine, Stanford University Medical Center, Palo Alto, CA
Address reprint requests to Mindie H. Nguyen, M.D., M.A.S., Liver Transplant Program, Division of Gastroenterology and Hepatology, Department of Medicine, Stanford University Medical Center, 750 Welch Road, Suite 210, Palo Alto, CA 94304. Telephone: 650-498-5691; FAX: 650-498-5692; E-mail: email@example.com
Hepatocellular carcinoma (HCC) is a leading cause of cancer-related deaths worldwide.[1, 2] Despite advances in HCC screening and surveillance for earlier detection and treatment, recent studies still report an overall 5-year survival rate of only 16%. Although improvements in HCC screening among high-risk populations do allow the diagnosis of earlier stage, localized HCC in patients who would benefit more from curative therapies, the success of any screening program hinges on its ability to offer effective treatment options with good long-term outcomes, such as liver transplantation (LT).[1, 3, 4] Previous studies have demonstrated that HCC patients meeting the Milan criteria (a single tumor no more than 5 cm in diameter or fewer than 3 tumors, each no more than 3 cm in diameter) can achieve good long-term survival after LT with 5-year survival rates greater than 70% to 80%.[4-6] As a result, starting in 2002, the United Network for Organ Sharing (UNOS) allocated additional Model for End-Stage Liver Disease (MELD) exception points to HCC patients with tumors meeting the Milan criteria and thereby improved their priority on the LT wait list. Although the initial MELD exception system assigned 24 points to patients with stage 1 HCC (1 tumor < 2 cm in diameter) and 29 points to patients with stage 2 HCC (1 tumor 2-5 cm in diameter or 2 or 3 tumors, each ≤ 3 cm in diameter), this was subsequently modified in April 2003 to assign 20 points for stage 1 HCC and 24 points for stage 2 HCC. In March 2005, another revision was enacted, and it assigned 22 points for stage 2 HCC.
LT rates for HCC immediately rose in the years after the introduction of the MELD exception status.[8, 9] However, with the increasing shortage of donor livers available for transplantation, it is not known whether the rates of LT for HCC have continued to rise in more recent years. Furthermore, although the MELD exception status allotted to HCC patients meeting the Milan criteria in principle standardized the prioritization for LT, it is not clear whether sex-specific or race/ethnicity-specific disparities in the receipt of LT persist in this cohort. Using a large, ethnically diverse population-based cohort in the United States, the current study evaluated trends in LT among transplant-eligible HCC patients in the most recent time period.
PATIENTS AND METHODS
Study Design and Patient Population
The current retrospective cohort study used data from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) cancer registry. The most recent version of the SEER registry (SEER*Stat 8.0.4, November 2012 submission) includes data from 1973 to 2010 and from 18 registries (Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco–Oakland, Los Angeles, San Jose–Monterey, greater California, Seattle–Puget Sound, Utah, Alaska Native Tumor Registry, Kentucky, Louisiana, New Jersey, rural Georgia, and greater Georgia) and represents approximately 28% of the US population. Our analyses used data from the entire registry cohort. However, detailed site-specific treatment data for the liver (eg, local tumor destruction, surgical resection, and LT) were available only from 1998 onward. Before 1998, the detailed surgical treatment of liver cancers was not reported, and thus the current study used the 1998-2010 patient cohort.
SEER identifies HCC according to International Classification of Diseases for Oncology, 3rd edition.[11, 12] Expanded race and ethnicity classifications were used for the 1998-2010 cohort: non-Hispanic whites, blacks, Asians/Pacific Islanders (Asians), and white Hispanics (Hispanics). Smaller numbers of HCC patients in other race/ethnicity groups (American Indian/Alaskan Natives, black Hispanics, and Asian Hispanics) precluded a precise estimation of LT rates, and thus they were not included in the current study.
The 1998-2010 cohort was separated into 3 time periods so that we could evaluate trends in LT rates: the pre-MELD exception era (1998-2003) and 2 MELD exception eras (2004-2008 and 2009-2010). The year 2003 was chosen as the cutoff for the pre-MELD exception era to account for the lag time in the effect of the implementation of the MELD exception status in late 2002 due to the multiple revisions since the initial policy was introduced. The MELD exception period was separated into 2 periods (2004-2008 and 2009-2010) to allow a more representative estimation of LT rates in the more recent period. Preliminary analyses demonstrated that the 2009-2010 period included more than 13,000 patients with HCC; this was a sufficient sample size for statistical analyses of LT trends, and the time period was still sufficiently close to the present time.
The SEER registry provides data regarding the number and size of each tumor present in the liver. Using these data, we were able to determine whether each patient's HCC disease burden was within or outside the Milan criteria. HCC staging definitions were based on the SEER staging system, which is unique to the SEER registry and used primarily for describing the extent of disease. Localized stage describes tumors confined to 1 lobe of the liver with or without vascular invasion. Regional-stage tumors include the involvement of more than 1 lobe via the contiguous growth of a single lesion, an extension to local structures (the diaphragm, extrahepatic bile ducts, or gallbladder), or an extension to regional lymph nodes. Distant-stage tumors include metastatic disease, an extension of the cancer to nearby organs (pancreas, pleura, or stomach), or the involvement of distant lymph nodes.
Treatment categories were analyzed with SEER site-specific surgery definitions (no therapy, local tumor destruction, surgical resection, and LT). Local tumor destruction included radiofrequency ablation and percutaneous ethanol injection but not transarterial chemoembolization (TACE). According to communication with the SEER registrar, data on TACE treatment are categorized separately under radiation and are not distinguishable from other types of radiation-based therapy (eg, brachytherapy or beam radiation). Furthermore, the SEER registrar indicated that the inclusion of TACE is relatively new, and the data quality is hindered by the consistency of the reporting and significant underreporting. A preliminary analysis of the 1998-2010 HCC cohort indicated that 95.8% of the HCC patients were categorized as receiving no form of radiation therapy or having an unknown radiation therapy status. Because of concerns about the inconsistency of data reporting and the underreporting of therapy, radiation treatment was not included to preserve the accuracy of the analyses and conclusions.
Clinical and demographic characteristics were compared across the 3 time periods (1998-2003, 2004-2008, and 2009-2010). Chi-square tests were used to compare categorical variables, and an analysis of variance was used to compare continuous variables in each time period. Multivariate logistic regression models were used to evaluate independent predictors of undergoing LT. Forward stepwise logistic regression methods included variables that were biologically important (eg, age and sex) or demonstrated significant associations (P < 0.1) in the univariate models. The final model included sex, age, race/ethnicity, Milan criteria, tumor number and size, tumor stage, and time period. Statistical significance was met with a 2-tailed P value < 0.05. All statistical analyses were performed with the Stata 10 statistical package (StataCorp, College Station, TX). Review by the institutional review board was not required for this study because human subjects were not involved, as per U.S. Department of Health and Human Services guidelines and the SEER database is publicly available without individually identifiable private information.
In all, there were 60,772 HCC patients in the 1998-2010 cohort, and 30% (n = 18,215) were within the Milan criteria. Among HCC patients meeting the Milan criteria, only 13.4% underwent LT during this period.
HCC Trends Across Time Periods
Age-adjusted HCC incidence rates were highest for Asians and lowest for non-Hispanic whites (Supporting Fig. 1); however, HCC incidence rates for Asians appeared fairly stable over time, whereas the rates for other racial/ethnic groups increased. As of the year 2000, although Asians still had the highest HCC incidence rates, the rates of other groups were close behind: the HCC incidence rates per 100,000 person-years in 2010 were 12.4 for Asians, 11.8 for Hispanics, 8.9 for blacks, and 5.2 for non-Hispanic whites. The majority of the HCC patients were men, and the mean age at the diagnosis of HCC was slightly lower in the more recent time periods (Table 1). In addition, there were significant increases in the proportions of black HCC patients (11.4% in 1998-2003, 12.6% in 2004-2008, and 13.4% in 2009-2010; P < 0.001) and Hispanic HCC patients (15.8% in 1998-2003, 17.6% in 2004-2008, and 18.9% in 2009-2010; P < 0.001), but there was a decrease in the proportion of Asian HCC patients (Table 1).
Table 1. Clinical Characteristics of HCC Patients From 1998 to 2010
1998-2003 (n = 20,004)
2004-2008 (n = 27,571)
2009-2010 (n = 13,200)
The data are presented as means and standard deviations.
Localized HCC within Milan criteria and age < 70 years (%)
The median size of the largest tumor decreased, but there was a significant trend toward fewer solitary tumors and more multifocal tumors in more recent times (Table 1). The proportion of localized-stage tumors increased with each time period (45.0% in 1998-2003, 50.4% in 2004-2008, and 51.7% in 2009-2010; P < 0.001), whereas the proportion of distant/advanced tumors decreased. In addition, the proportion of tumors meeting the Milan criteria also significantly and consistently increased across the analyzed time periods (22.8% in 1998-2003, 31.8% in 2004-2008, and 37.1% in 2009-2010; P < 0.001).
Among all HCC patients, there was an increase in all forms of therapy received from 1998-2003 to 2004-2008, but there was a subsequent decrease in treatment in 2009-2010 (Table 2). A similar trend was seen in HCC patients within the Milan criteria (Table 2). However, what is most striking is the unacceptably high rate of no therapy among HCC patients within the Milan criteria (61.5% in 1998-2003, 56.1% in 2004-2008, and 65.3% in 2009-2010; P < 0.001). Even among patients with more favorable characteristics (localized HCC within the Milan criteria and age < 70 years), there remained unacceptably high rates of no therapy (55.0% in 2009-2010) and low rates of LT (15.3% in 2009-2010; Table 2).
Table 2. Treatment of HCC by Time Periods
NOTE: The data are presented as percentages.
(n = 20,004)
(n = 27,571)
(n = 13,200)
Local tumor destruction
Within Milan Criteria
(n = 4525)
(n = 8701)
(n = 4867)
Local tumor destruction
Localized Stage Within Milan Criteria
(n = 2803)
(n = 5781)
(n = 3282)
Local tumor destruction
Localized Stage Within Milan Criteria and Age < 70 Years
(n = 2015)
(n = 4359)
(n = 2560)
Local tumor destruction
In addition, across all HCC categories, LT rates increased from 1998-2003 to 2004-2008, but they declined in 2009-2010 (Table 2). To investigate whether this trend was reflective of a true decrease in LT for HCC or a result of an increasing imbalance between organ availability and wait-listed HCC patients, we compared the actual number of transplants with LT rates across the 3 time periods and stratified them by race/ethnicity, sex, and age groups (Fig. 1A-C). Among patients < 70 years old with localized HCC within the Milan criteria, the number of actual LT procedures increased from 75.5 per year in 1998-2003 to 208.2 per year in 2004-2008, and the number remained stable at 201.5 per year in 2009-2010. However, among all racial/ethnic groups, the proportion of patients < 70 years old with localized HCC within the Milan criteria undergoing LT increased from 1998-2003 to 2004-2008 but declined in 2009-2010, with the lowest rates of LT seen among Asians (Fig. 1A). Similar trends were seen when the patients were stratified by sex (Fig. 1B) and age groups (Fig. 1C), and this suggests that the decline in the rates of LT for HCC in the more recent time period may be largely a result of the increasing number of HCC patients on the LT wait list, which was compounded by the shortage of available donor organs.
Predictors of LT
In comparison with the 1998-2003 period, HCC patients in the 2009-2010 period were less likely to undergo LT [odds ratio (OR) = 0.63, 95% confidence interval (CI) = 0.57-0.71, P < 0.001], whereas patients in 2004-2008 were just as likely to undergo LT (OR = 1.03, 95% CI = 0.92-1.14, P = 0.62; Table 3). In comparison with non-Hispanic whites, blacks (OR = 0.48, 95% CI = 0.41-0.56, P < 0.001), Asians (OR = 0.65, 95% CI = 0.57-0.73, P < 0.001), and Hispanics (OR = 0.76, 95% CI = 0.68-0.85, P < 0.001) were all significantly less likely to undergo LT. As expected, HCC patients within the Milan criteria were significantly more likely to undergo LT in comparison with those outside the Milan criteria. Older patients and patients with more advanced tumor stages were less likely to undergo LT (Table 3).
Table 3. Univariate and Multivariate Logistic Regression Analyses of Predictors of LT
Within Milan criteria
Number of tumors
The current study evaluated trends in LT for HCC in a large population-based cohort in the United States from 1998 to 2010. In comparison with the pre-MELD exception era of 1998-2003, the MELD exception time periods of 2004-2008 and 2009-2010 demonstrated increasing proportions of localized HCC and decreasing proportions of distant/advanced HCC. In addition, the proportion of HCC patients who had tumors within the Milan criteria continued to rise. Although the actual number of LT cases increased significantly from the pre-MELD exception era of 1998-2003 (75.5 per year) to the MELD exception period of 2004-2008 (208.2 per year) and stabilized in 2009-2010 (201.5 per year), the proportion of HCC patients undergoing LT increased from 1998-2003 to 2004-2008 but declined in 2009-2010. HCC patients in the more recent time period (2009-2010) and patients of racial/ethnic minorities were significantly less likely to undergo LT.
With the implementation of the MELD exception policy for HCC patients within the Milan criteria, the rates of LT for HCC rose significantly. Using UNOS data from 1997 to 2007, Ioannou et al. reported a 6-fold increase in the proportion of LT recipients with HCC resulting from MELD exception policy implementation (4.6% in 1997-2002 versus 26.0% in 2002-2007). Similar trends were seen in data from the Scientific Registry of Transplant Recipients. Thuluvath et al. reported a steady increase in the number of LT recipients with HCC in similar time periods (999 in 2002 versus 1656 in 2008). Despite the increasing proportion of LT procedures for HCC, the number of HCC patients who remained on the LT wait list at the end of the year also continued to rise in that study (382 in 2002 versus 890 in 2008). In the current study, the number of HCC patients undergoing LT also rose from 75.5 per year in 1998-2003 to 208.2 per year in 2004-2008, but the trend did not continue into the 2009-2010 period with only 201.5 transplants per year. Correspondingly, the proportion of transplant-eligible HCC patients who underwent LT increased from 13.3% in 1998-2003 to 15.2% in 2004-2008 (P < 0.01) but then declined to 10.2% in 2009-2010 (P < 0.001; Table 2 and Fig. 1A-C). Furthermore, more than 65% of transplant-eligible HCC patients did not receive any form of surgical therapy in the most recent era (Table 2). Although these findings are limited by the exclusion of TACE from the treatment categories, it is still a concern that such a large proportion of transplant-eligible HCC patients did not undergo curative surgical resection or LT.
These findings clearly indicate that although LT has been offered more to HCC patients since the implementation of MELD exception policies, the shortage of available organs is contributing to an increasing imbalance between HCC patients eligible for LT and those who actually undergo LT.[1, 4, 13-16] As a result of this imbalance, longer wait-list times and, consequently, risk progression of disease beyond transplantation eligibility and subsequent dropout from the wait list are expected.[13, 15-17] The evolving changes in the number of MELD exception points allocated to transplant-eligible HCC patients may be another potential factor contributing to the decreasing proportion of HCC patients undergoing LT. When MELD exception points for HCC were initially implemented, patients with stage 1 HCC (1 tumor < 2 cm in diameter) were allocated 24 points, and patients with stage 2 HCC (1 tumor 2-5 cm in diameter or 2-3 tumors, each ≤ 3cm in diameter) were allocated 29 points. With time, the number of points allocated to HCC has been gradually reduced, so that patients with stage 2 HCC currently are allocated 22 points. Despite these modifications, transplant-eligible HCC patients may still continue to possess a disproportionate advantage over patients with other chronic liver diseases, and the exception point advantage allocated to HCC patients on LT wait lists may need to be further modified. Although the increasing utilization of TACE has been effective in bridging HCC patients to LT, TACE is neither a curative treatment nor a comparable treatment endpoint for transplant-eligible HCC patients.[18-22] Living donor liver transplantation (LDLT) is an alternative approach that has a potentially unlimited supply of available organs, but early studies reported significantly higher rates of HCC recurrence associated with LDLT cohorts (5-year HCC recurrence rate = 29%).[23, 24] However, more recent studies have demonstrated recurrence and survival comparable to those associated with deceased donor LT.[25, 26] A recent meta-analysis by Liang et al. demonstrated no significant difference in overall 5-year survival (OR = 0.64, 95% CI = 0.33-1.24) or 5-year HCC recurrence (OR = 1.21, 95% CI = 0.44-3.32) between LDLT recipients and deceased donor LT recipients. Nevertheless, risks to living liver donors are real, and the recent media emphasis on complications from LDLT have hampered the expansion of LDLT programs across the United States. Clearly, LDLT will not be the final solution to the current shortage of available organs, and future efforts should focus on enhancing donor recruitment through education and outreach as well as novel therapies that may eventually prevent the development of HCC and thus preclude the need for LT for HCC.
Before the implementation of the MELD allocation system for LT, significant racial disparities in the receipt of LT existed.[28-31] Siegel et al. demonstrated that blacks (OR = 0.43, 95% CI = 0.21-0.90) and Asians (OR = 0.57, 95% CI = 0.36-0.89) were significantly less likely to undergo LT in comparison with whites in an early cohort of transplant-eligible HCC patients from 1998 to 2002. Reid et al. reported similar observations based on an analysis of UNOS data from 1994 to 1998. According to standardized transplantation ratios that compared the racial distribution of LT recipients with the racial distribution of the US population, blacks were significantly less likely to undergo LT in comparison with whites (black/white ratio = 0.65/1.05). Whether the implementation of the MELD system, which was intended to be a more objective tool placing more emphasis on the severity of disease and less emphasis on the waiting time, has been successful in reducing racial disparities in LT is not clear. In a large, retrospective cohort study, Moylan et al. used the UNOS database to compare LT patterns in the pre-MELD period of 1996-2000 and the MELD period of 2002-2006. In comparison with whites in the pre-MELD era, blacks were significantly more likely to die or become too sick for LT (OR = 1.51, 95% CI = 1.15-1.98) and were less likely to undergo LT (OR = 0.75, 95% CI = 0.59-0.97). However, in the MELD era (2002-2006), blacks no longer had an increased likelihood of dying or becoming too sick for LT (OR = 0.96, 95% CI = 0.74-1.26) and had a similar likelihood of undergoing LT (OR = 1.04, 95% CI = 0.84-1.28). In the current study, which included only HCC patients, blacks, Asians, and Hispanics all had significantly lower rates of LT than non-Hispanic whites; this included those presenting in a fairly recent time period (2009-2010; Fig. 1A). Although this disparity was most prominent in the pre-MELD exception period of 1998-2003, the differences persisted in all 3 time periods studied. Even after correction for multiple demographic characteristics (age and sex), tumor characteristics (tumor stage and Milan criteria), and time periods, blacks, Asians, and Hispanics were all significantly less likely to undergo LT in comparison with non-Hispanic whites (Table 3). Clearly, racial disparities in LT for HCC patients persist. Whether these differences result from disparities in HCC screening among high-risk groups that subsequently lead to more advanced disease at presentation precluding curative therapy or from disparities in access to care that prevent transplant-eligible patients from an initial LT evaluation is not clear. Future research on the management of HCC should target resources to address such disparities.
The strengths of the current study are highlighted by the large sample, which represents a sizeable proportion of the US population. The SEER cancer registry is globally recognized as an authoritative source of cancer statistics. The focus on the 1998-2010 cohort ensured consistent reporting of tumor staging and treatment data and allowed an accurate comparison of pre-MELD exception and MELD exception eras, including a precise evaluation of the most recent time period (2009-2010). However, the current study is limited by factors typical of large registry-based studies. Most importantly, the etiology of HCC was not available for analysis; and although previous studies have demonstrated differences in posttransplant survival with different HCC etiologies, disparate rates of LT by HCC etiology have not been reported. In addition, locoregional therapy (specifically TACE) is increasingly used for treating HCC patients not eligible for resection or LT and for down-staging or bridging HCC patients to LT.[18-22] However, data for TACE are a relatively new addition to the SEER registry and suffer from significant underreporting and inconsistencies in reporting. In the SEER registry, the date of treatment was not readily available, and this prevented a time-to-event Cox regression model to evaluate predictors of undergoing LT. Thus, our study used a binary logistic regression model to evaluate predictors of LT.
In summary, the MELD exception policy significantly increased the number of LT procedures for HCC patients immediately after its implementation, but this did not continue into the more recent time period (2009-2010). The proportion of transplant-eligible HCC patients also increased in the more recent time period, but the proportion of those actually undergoing LT actually decreased, and this was most likely due to the donor organ shortage. In addition, substantial racial disparities in the receipt of LT persisted for HCC patients after the implementation of the MELD exception policy, even in the more recent time period (2009-2010). Blacks and Asians continued to have the lowest rates of LT. Most striking is that more than 65% of transplant-eligible HCC patients in the more recent time period did not receive any form of curative therapy. Further efforts are needed to promote education and outreach among potential organ donors to improve the availability of LT. In addition, quality improvement efforts should target research and education to address the persisting racial disparities in LT as well as the dismally low rates of curative therapy for transplant-eligible HCC patients in the United States.