Approximately one-half of patients with early-stage hepatocellular carcinoma meeting Milan criteria did not receive local tumor destructive or curative surgery in the post-MELD exception era




Since 2002, priority Model for End-stage Liver Disease (MELD) exception status has been given to patients with hepatocellular carcinoma (HCC) who meet the Milan criteria. Since then, the number of liver transplantations performed in patients with HCC has increased, but to the authors' knowledge, few studies to date have examined the effect of MELD exception recommendations on therapy use and survival rates in a nationwide sample. The current study examines therapy use and long-term survival rates among patients with HCC tumors meeting the Milan criteria in the post-MELD exception era.


The current study is a retrospective cohort study of 2179 patients with localized HCC meeting the Milan criteria who were registered between 2004 and 2007 in the Surveillance, Epidemiology, and End Results database.


A total of 43% of patients did not receive any specific therapy. Overall, the 5-year relative survival rate for patients receiving only supportive care was dismal at 24% (95% confidence interval [95% CI], 21%-27%), whereas that for patients undergoing liver transplantation was 77% (95% CI, 71%-82%). Long-term survival was found to be dependent on age, race/ethnicity, and type of therapy received. A multivariate Cox proportional hazards model adjusted for age, race/ethnicity, and type of therapy received demonstrated that, compared with white patients, black patients had significantly poorer survival outcomes (hazards ratio, 1.23; 95% CI, 1.03-1.47 [P = .02]), whereas Asian/Pacific Islander patients had significantly better survival rates when compared with white patients (HR, 0.66; 95% CI, 0.57-0.77 [P < .001]).


Despite having localized disease that met transplantation criteria, nearly 50% of the large nationwide cohort of patients with HCC in the current study received only supportive care and had dismal 5-year relative survival rates, especially among black patients. Cancer 2014;120:1725–1732. © 2014 American Cancer Society.


Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related death worldwide.[1, 2] It is the fifth leading cause of cancer in men and the seventh leading cause of cancer in women in United States.[3, 4] Although incidence rates for most cancers in United States are declining, incidence rates for HCC in both men and women continue to rise.[3] It is estimated that there will be 30,640 new HCC cases diagnosed in 2013 (22,720 in men and 7920 in women).[3] Similarly, mortality rates for HCC have also increased over the past 2 decades and it is estimated that there will be 21,670 HCC-related deaths in 2013 (14,890 in men and 6780 in women).[3] Despite advances in screening and early detection, HCC still has a poor overall 5-year survival rate of 16% according to the data published by the Surveillance, Epidemiology, and End Results (SEER) program between 2002 and 2008.[3]

Liver transplantation is considered to be the most effective therapeutic option for patients with HCC, with the best long-term outcome occurring when it is performed in patients with localized tumors meeting the Milan criteria (a single lesion measuring ≤ 5 cm or ≤ 3 lesions measuring ≤ 3 cm in largest diameter and the absence of vascular invasion or extrahepatic disease). Since 2000, the Model for End-stage Liver Disease (MELD) system has been implemented to predict 90-day pretransplant mortality in patients with chronic liver disease and to determine the priority of organ allocation for liver transplantation. However, this model was not equipped to accurately predict the mortality risk in patients with HCC because it did not take into account the effect of tumor progression with prolonged wait time. Beginning in 2002, a supplemental system was created by the United Network for Organ Sharing to allocate additional points to patients with HCC meeting the Milan criteria. The priority MELD system in HCC underwent several revisions and currently allocates 22 points for tumors meeting the Milan criteria. Although there is regional variability, this has been reported to result in shorter waiting periods for patients with tumors meeting the Milan criteria.[5] Studies have shown a 6-fold increase in the percentage of patients with HCC undergoing liver transplantations in the post-MELD exception era compared with earlier periods.[6]

The objective of the current study was to examine the impact of MELD exception on both treatment use and overall survival among patients with localized HCC meeting the Milan criteria as well as predictors of survival using a large nationwide cohort.


Study Design and Patient Population

The current study was a retrospective cohort study of patients with localized HCC that met the Milan criteria and who were identified from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) database between 2004 and 2007 (after the MELD exception era). The current study used the most recent version of the SEER registry (November 2012 submission), which includes data from 1973 through 2010. The SEER program is a principal source of information regarding cancer incidence, prevalence, and survival from specific geographic areas in United States and represents 28% of the US population.[7] Patients included in the study were recruited from 18 registries covering 9 states (Connecticut, Hawaii, Iowa, New Mexico, Utah, Alaska, Kentucky, Louisiana, and New Jersey) and 9 metropolitan areas (San Francisco/Oakland, Detroit, Seattle [Puget Sound], Atlanta [metropolitan], San Jose-Monterey, Los Angeles, rural Georgia, California excluding San Francisco/Oakland, and greater Georgia). The study population included adult patients aged ≥ 20 years with localized, histologically confirmed HCC tumors meeting the Milan criteria. Expanded race/ethnicity definitions that are included within the SEER registries are white, black, Asian/Pacific Islander, American Indian/Alaska Native, and unknown.

Statistical Analyses

Descriptive statistics were performed using the chi-square test to compare categorical variables. Survival analysis was performed using the Kaplan-Meier method[8] and survival curves were compared using the log-rank test[9] and stratified by sex, race/ethnicity (white vs black vs Asian/Pacific Islander), age categories (aged 20years-44 years vs aged 45 years-64 years vs aged ≥ 65 years), and type of therapy received (no surgical therapy/local tumor destruction vs local tumor destruction vs partial hepatectomy vs liver transplantation). Multivariate Cox proportional hazards models[10] were used to estimate hazards ratios (HRs) relating various demographic and clinical factors to long-term survival in a forward stepwise regression. The final model included age, race/ethnicity, and type of therapy received. The proportional hazards analysis assumption was tested using a significance test approach. The “therapy” variable violated this assumption and therefore this was used as a time-dependent variable to fix the nonproportional hazards model. Statistical significance was met with a 2-tailed P value < .05. Statistical analysis was performed using the SEER*Stat software (version 8.0.1)[11] statistical package provided for the analysis of the SEER database (National Cancer Institute, Bethesda, Md) as well as Stata (version 12.1; StataCorp, College Station, Tex).[12]


Patient Population

During the period between 2004 and 2007, a total of 21,710 patients with primary liver cancer were identified in the SEER database. Among these patients, 21,449 were adult patients with primary liver cancer, 17,252 of whom had histologically confirmed HCC. Among these, 7885 patients had localized HCC and 2179 met the Milan criteria. Patients with unclear/unspecified treatment data (56 patients) were eliminated from the treatment use analysis. A total of 166 patients whose tumors met the Milan criteria had incomplete data and therefore were also excluded from the survival analysis. Due to the small sample size of the American Indian/Alaska Native (16 individuals) and “other” (10 individuals) groups, these ethnic groups were also excluded from the survival analysis. This yielded a final sample of 1988 patients whose tumors met the Milan criteria for the survival analysis.

Patient Characteristics

As shown in Table 1, the vast majority of the cohort were aged ≥ 45 years (96%), with a median age of 59 years (range, 22 years-92 years). The majority of patients were also male (72%) and white (68%).

Table 1. Characteristics of Patients With Localized HCC Meeting the Milan Criteria Between 2004 and 2007
  1. Abbreviations: HCC, hepatocellular carcinoma;IQR, interquartile range; LTD, local tumor destruction.

Median age at diagnosis(IQR), y 
Age, y59 (22–92)
20–4489 (4%)
45–641322 (61%)
≥65768 (35%)
Male1578 (72%)
Female601 (28%)
White1462 (68%)
Black226 (10%)
Asian/Pacific Islander465 (22%)
Type of therapy 
No surgery/LTD-primary site911 (43%)
LTD545 (26%)
Partial hepatectomy264 (12%)
Liver transplantation403 (19%)

HCC Treatment

In the cohort of patients in the current study, between the years 2004 and 2007, no surgical therapy or local tumor destruction of the primary site was performed in 43% of the patients (911 patients), whereas 26% (545 patients) underwent local tumor destruction, 12% (264 patients) underwent partial hepatectomy, and the remaining 19% of patients (403 patients) underwent liver transplantation (Fig. 1). Therapies included under local tumor destruction were electrocautery, fulguration, photodynamic therapy, cryosurgery, laser surgery, percutaneous ethanol injection, heat-radiofrequency ablation, and others including acetic acid injection.

Figure 1.

Rates of utilization of the different types of therapy are shown.

Therapy use stratified by age demonstrated that older patients were more likely not to receive any form of surgical or local tumor destructive therapy compared with younger patients (P < .001) (Fig. 2). However, when stratified by therapy type, among various age groups, a similar percentage of patients underwent local tumor destruction (P = .13). The percentage of patients undergoing partial hepatectomy was significantly higher among patients aged 20 years to 44 years compared with older patients (P < .001). Finally, the percentage of patients undergoing liver transplantation was significantly different among the age groups, with patients aged ≥ 65 years being less likely to undergo liver transplantation compared with younger patients (those aged 20 years-44 years and those aged 45 years-64 years [7% vs 21% vs 25%]; P < .001).

Figure 2.

Therapy utilization stratified by age, gender, and race/ethnicity.

With regard to sex, females were less likely to receive any form of surgery or local tumor destructive therapy compared with males (P = .02); however, similar percentages of males and females underwent local tumor destruction (P = .94) or partial hepatectomy (P = .68). It is interesting to note that male patients were also more likely to undergo liver transplantation compared with female patients (20% vs 15%; P = .002).

Therapy use when stratified by race indicated that there were statistically significant differences in the percentages of white, black, and Asian/Pacific Islander patients who did not receive any curative or local tumor destructive therapy (P = .003), with white patients (43%) and black patients (45%) found to have a higher percentage of patients who did not undergo any curative or local tumor destructive therapy compared with Asian/Pacific Islander patients (35%). There were also statistically significant differences noted in the percentages of patients who underwent local tumor destruction (P = .004) or partial hepatectomy (P < .001), with a higher percentage of Asian/Pacific Islander patients receiving these therapies. However, a lower percentage of Asian/Pacific Islander patients (11%) underwent liver transplantation compared with white (21%) and black (16%) patients, and this difference was also found to be statistically significant (P < .001).

Overall Survival

The overall median survival of the study cohort was 39 months (range, 36 months-44 months) and the 5-year cumulative survival rate was 43% (95% CI, 40%-45%). Survival was found to be significantly poorer in older patients (P < .001) (Figs. 3A and 4A), whereas survival rates were similar in male and female patients (P = .33) (Figs. 3B and 4B). Large racial disparities were also observed, with black patients having the lowest survival (P < .001) (Fig. 3C). The 5-year survival rate in black individuals was only 29% (95% CI, 21%-36%) compared with 42% (95% CI, 39%-45%) in white individuals and 53% (95% CI, 47%-58%) in Asian/Pacific Islanders (P < .001). Patients receiving no specific local/surgical therapy fared worst, whereas patients who underwent liver transplantation had the best survival rates (P < .001) (Figs. 3D and 4D)

Figure 3.

Patients' 5-year relative survival rates are shown with 95% confidence intervals stratified by (A) age, (B), gender, (C) race/ethnicity, and (D) type of therapy received. LTD indicates local tumor destruction.

Figure 4.

Survival differences are shown stratified by (A) age, (B), gender, (C) race/ethnicity, and (D) type of therapy received. LTD indicates local tumor destruction.

Predictors of Survival

On multivariate analysis (Table 2), significant independent predictors of poor survival were age > 65 years (HR, 1.85; 95% CI, 1.30-2.63 [P < .001]) and black race (HR, 1.23; 95% CI, 1.03-1.47 [P < .02]). Significant independent predictors of improved survival were Asian ethnicity (HR, 0.66; 95% CI, 0.57-0.77 [P  < .001]) and having received palliative (HR, 0.54; 95% CI, 0.47-0.63 [P < .001]) or curative therapy with liver transplantation (HR, 0.12; 95% CI, 0.08-0.17 [P  < .001]).

Table 2. Predictors of Poorer Survival in Patients With Localized HCC Meeting the Milan Criteria Between 2004 and 2007
Risk FactorsHR (95% CI)PAdjusted HR (95% CI)P
  1. Abbreviations: 95% CI, 95% confidence interval; HCC, hepatocellular carcinoma;HR, hazards ratio; LTD, local tumor destruction.

Age, y    
20–44Reference Reference 
WhiteReference Reference 
Asian/Pacific Islander0.71(0.61–0.83)<.0010.66(0.57–0.77)<.001
Type of therapy    
No surgery/LTD -primary siteReference Reference 
Partial hepatectomy0.30(0.24–0.38)<.0010.25(0.19–0.33)<.001
Liver transplantation0.19(0.15–0.23)<.0010.12(0.08–0.17)<.001


In the post-MELD exception era, several studies have focused on rates of liver transplantation,[6] but to our knowledge few have focused on the impact on therapy utilization and overall survival. The current study focused on a subgroup of patients with HCC with tumors that met Milan criteria who were diagnosed after implementation of the MELD exception system for HCC. We found that 43% of this highly selected group of patients with early-stage tumors received no specific treatment at all. Indeed, only 57% of patients with HCC who were eligible for transplant received any form of curative or local tumor destructive therapy, suggesting major underuse of anti-HCC therapies. Prior studies included in a recent meta-analysis[13] also found lower treatment rates for HCC overall; however, to the best of our knowledge, the majority of these were studies based either at smaller tertiary care hospitals or were larger population-based studies that included patients with HCC of all stages. Tertiary care hospital-based studies can demonstrate a bias toward higher rates of treatment use when compared with what may be observed in the general population. The inclusion of patients with HCC of all stages as noted in prior population-based studies[14-17] does not allow for the examination of use rates in a more focused population that generally should be eligible for some type of therapy.

In the current study, the study population was limited to those patients presenting in the more recent time period, when more treatment modalities and MELD priority status were available, and in those patients with localized disease meeting the Milan criteria, who generally should not have been disqualified from therapy due to tumor factors. Nevertheless, even in this well-selected patient population that was eligible for transplant, treatment rates were still alarmingly low, with only approximately 50% of these individuals receiving any curative or local tumor destructive therapies.

Significant differences in the receipt of therapy were also observed by demographic features including age, sex, and race/ethnicity. Elderly patients were found to have lower overall rates of therapy use. One prior study demonstrated that older patients are less likely to be recommended for curative therapies and also to have higher rates of refusal for therapies.[18] In the end, they tend to receive less optimal therapy compared with younger patients.[19] The current study demonstrated a lower number of older patients undergoing liver transplantation. These differences may be attributed to higher rates of comorbidities in older patients as well as physician and patient perceptions of high perioperative risks and a lack of long-term benefits in older individuals. Earlier studies have shown that the aggressive treatment of patients with good liver function and performance status improved survival rates even in the very old,[20, 21] and age alone should not be considered a contraindication to the receipt of therapy. In a study of elderly patients with HCC, shorter survival was only secondary to undertreatment and survival differences by age disappeared in various treatment subgroups.[22] Performance status and physiological age are considered to be more important than chronological age when evaluating patients for the treatment of localized HCC and their associated potential survival benefits.[18]

With regard to sex, a larger percentage of female patients did not undergo any curative or local tumor destructive therapy, and there was a higher rate of liver transplantation among men compared with women. The reason for sex disparities with liver transplantation is unclear. One prior study demonstrated that women are more likely to die or become too ill for transplantation compared with men.[23] Part of the reason for this is believed to be due to a bias against women in the current MELD calculation, which include creatinine levels, because women generally have lower muscle mass and thus lower creatinine levels.[24] Women are also smaller in size overall, thereby contributing to difficulty in matching organ size to recipient body size in women, and leading to a longer wait time.[25]

However, the most consistent disparities found in the current study were racial disparities, which were statistically significant in both the overall use rates as well as within each category of therapy. One of what to our knowledge are the few prior studies to examine the effect of MELD exception on racial disparities in treatment use among patients with HCC demonstrated that black patients were less likely to receive any form of therapy compared with white patients, even among individuals with potentially curable localized HCC tumors.[26] Black patients were 50% less likely to be recommended for surgical therapy[26] and had higher rates of refusal of surgery.[18] Lower rates of therapy in black individuals were also attributed to reduced access to care.[18, 26] It is interesting to note that, in the current study, the percentage of white and black patients who did not receive any surgery or local tumor destructive therapy was similar, but another ethnic minority, Asian/Pacific Islanders, had lower rates of no treatment (43%, 45%, and 35%, respectively; P = .003). Specifically, Asian/Pacific Islander patients had higher rates of local tumor destruction and partial hepatectomy. However, compared with white patients, Asian/Pacific Islander patients were least likely to undergo liver transplantation, followed by black patients. The results of one prior study demonstrated that black patients reached their insurance spending limits more frequently and also are more likely to lose medical insurance coverage after a cancer diagnosis.[26, 27] Racial inequalities in the use of therapy may be partially mitigated by improving access to health care, as well as by providing education to alter personal beliefs, perceptions of cancer therapy, and distrust of health care systems to decrease refusal rates in black individuals.

As has been noted in prior studies, older patients[28] and those who did not receive any HCC-specific therapy were found to fare more poorly. What is remarkable among the results of the current study is that even in a very well-defined cohort of transplant-eligible patients with HCC, racial disparities in survival remain pronounced in this recent era (post-2004).

Even after multivariate analysis, racial disparities in survival continue to exist. As noted in other studies,[28, 29] when compared with white patients, black patients had lower rates of survival whereas Asian/Pacific Islander patients were found to have a higher survival rate. Asians tend to have a better awareness of HCC secondary to screening programs targeting this population and to present with HCC secondary to hepatitis B virus infection with compensated liver function, resulting in improved survival.[29] Even among patents with hepatitis C-related HCC, Asian individuals have been reported to have better hepatic function compared with non-Asian patients.[30]

The current study has several limitations. First, due to the retrospective study design, selection bias due to unrecorded/unavailable factors cannot be eliminated. The SEER database is also lacking in terms of detailed clinical information for the patients such as etiology of HCC, liver function, performance status, and comorbidities. Overall survival data do not reflect mortality risk from noncancer determinants. A decision regarding the type of treatment modality used is usually determined by various patient and physician factors, and reasons for the selection or refusal of therapy are not available in the database.

Despite its limitations, the current study has certain advantages. To our knowledge to date, the SEER database is the most comprehensive national database, providing information regarding most treatment strategies across various demographics. It represents a nationwide sample that provides us with the opportunity to assess treatment strategies and their outcomes in a comprehensive and diverse cohort of patients. The current study examined a more “modern ” cohort and focused on a very specific subset of patients with less advanced HCC who should be more likely to be eligible candidates for various treatments.


In the post-MELD exception era in patients with localized HCC who met the Milan criteria, nearly one-half of the patients did not receive any HCC-specific surgery or local tumor destructive therapy. Major racial disparities also remained prominent, with black patients having the lowest rates of therapy use and the lowest survival and Asian/Pacific Islander patients having the lowest rates of liver transplantation. Further studies are needed to help elucidate factors that contribute to low overall rates of treatment use (eg, access to care, treatment refusal, lack of specialized care centers, physician awareness of newly available therapies) and the continued significant racial disparities in both treatment use and long-term survival rates.


No specific funding was disclosed.


The authors made no disclosures.