The incidence of hepatocellular carcinoma (HCC), a frequent and incurable complication of cirrhosis, continues to rise. Orthotopic liver transplantation (OLT) has been proposed as a treatment for unresectable, intrahepatic HCC limited in extent to the Milan criteria adopted by the United Network of Organ Sharing (UNOS) in 1998. More recently, somewhat less restrictive University of California, San Francisco (UCSF)10, criteria were proposed. To examine the long-term outcomes of OLT for HCC patients and to assess the UNOS policy of assigning weighted allocation points to patients with HCC, we retrospectively analyzed 144 patients (113 after 1998) with HCC who underwent OLT over an 11-year period at 3 institutions from UNOS Region 1. We compared their outcomes with 525 patients (272 after 1998) who underwent OLT for nonmalignant liver disease. The 1- and 5-year survival rates were 80.3% and 46.7%, respectively, for patients with HCC and 81.5% and 70.6%, respectively, for patients without HCC (P = .020). However, there was no difference in survival between HCC and non-HCC patients after implementation of disease-specific allocation for HCC in 1998. A higher proportion of the HCC cohort was older and male and had chronic HCV infection and alcoholic liver disease. In univariate analysis, having alpha-fetoprotein (AFP) levels of 10 ng/mL or less and meeting clinical and pathologic UCSF criteria were each significant predictors of improved survival (P = .005, P = .02, and P = .03, respectively). AFP greater than 10 ng/mL and exceeding pathologic UCSF criteria were also significant predictors of recurrence (P = .003 and P = .02, respectively). In conclusion, taken together, our data suggest that OLT is an acceptable option for patients with early HCC and that UCSF criteria predict outcome better than Milan or UNOS criteria. Regardless of which criteria are adopted to define eligibility, strict adherence to the criteria is important to achieve acceptable outcomes. (Liver Transpl 2004;10:1343–1354.)
Approximately 19,000 new cases of hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma are projected to occur in the United States in 2004.1 The incidence of HCC doubled in the United States during the period 1975 to 1995 and continued to rise through 1998.2, 3 The majority of HCC occurs in patients with liver cirrhosis. In the United States, the major risk factors for HCC include chronic hepatitis C virus (HCV), hepatitis B virus (HBV), and alcoholic liver disease. Of these risk factors, chronic HCV infection is the leading risk factor that contributes to the rising incidence of HCC. This trend is likely to continue with the evolution of the large cohort infected with HCV during the 1970s and 1980s and the known latency of HCC development.2, 4
Orthotopic liver transplantation (OLT) has been proposed as effective therapy for those with limited intrahepatic HCC, and it is the only treatment that definitively addresses both the metachronous occurrence risk of HCC and the underlying liver disease. Current United Network of Organ Sharing (UNOS) accepted criteria for patients with early HCC who had acceptable outcomes after transplant were first proposed by Mazzaferro et al. in 1996.5 Under these criteria, OLT was recommended for potential recipients who had 1 lesion 5 cm or smaller or up to 3 lesions each 3 cm or smaller (Milan criteria). Several reviews of relatively small patient cohorts have suggested that OLT for patients with Milan criteria–defined HCC provides comparable survival to that achieved in patients undergoing OLT for nonmalignant liver disease.6–8 Yoo et al.9 retrospectively analyzed the outcomes of the UNOS data and found a significant improvement in survival over time for HCC patients undergoing OLT. The current 1- and 5-year survival rates for HCC patients undergoing OLT are 77.0% and 61.1%, respectively,9 which contrast with the 5-year survival rate of only 25.3% in 1987. These results suggested to some investigators that the Milan criteria might be too restrictive. After analyzing 70 consecutive patients with HCC who underwent OLT over a 12-year period at University of California, San Francisco (UCSF), Yao et al. proposed expanding the criteria (1 lesion ≤ 6.5 cm or 3 lesions, largest lesion ≤ 4.5 cm with a total tumor diameter ≤ 8 cm); they concluded these new criteria did not significantly compromise the excellent post-OLT outcomes achieved using the Milan criteria (Table 1).10
Table 1. Milan and University of California, San Francisco Criteria for Orthotopic Liver Transplantation
1 tumor less than or equal to 5 cm
1 tumor less than or equal to 6.5 cm
Up to 3 tumors, none larger than 3 cm
Multiple tumors, none larger than 4.5 cm and total tumor diameter less than or equal to 8 cm
We retrospectively analyzed the outcomes of 144 patients with HCC transplanted during an 11-year period at 3 institutions in UNOS Region 1 (New England) to assess the appropriateness of the region's policy of assigning weighted allocation points to those patients with early HCC. Previously, Haug et al. described 33 OLT recipients transplanted for primary hepatic malignancy between 1983 and 1990 from the Boston Center for Liver Transplantation.11 Our study is a follow-up to the Haug study and draws from the same 3 institutions that formerly made up the Boston Center for Liver Transplantation. We report the recurrence-free and overall survival outcomes in the HCC OLT patients and the key clinical and pathologic predictors of the outcomes.
HCC, hepatocellular carcinoma; OLT, orthotopic liver transplantation; UNOS, United Network for Organ Sharing; UCSF, University of California, San Francisco; AFP, alpha-fetoprotein; HCV, hepatitis C virus; HBV, hepatitis B virus; ALD, alcoholic liver disease.
Patients and Methods
One hundred and forty-four adult patients with end-stage liver disease and HCC underwent OLT between September 1992 and March 2003 at Massachusetts General Hospital (40), New England Medical Center (30), and Lahey Clinic program (74); together, these institutions represent the 3 largest liver transplant centers in UNOS New England Region 1. The 525 patients with nonmalignant liver disease who underwent OLT at Massachusetts General Hospital and Lahey Clinic program during the same timeframe served as a comparison cohort. Data on patients transplanted for nonmalignant disease were not available from New England Medical Center because of personnel turnover. Of the 144 patients with HCC, 38 had tumors found incidentally at transplant, 105 had HCC diagnosed before OLT, and pretransplant records were not available for the 1 remaining patient. For all patients, age at transplant, etiology of liver disease, and survival data were available. The last date of follow-up for the HCC and non-HCC patient populations from all 3 sites was March 2003.
Before 1998, among centers in UNOS Region 1, patients with confirmed intrahepatic HCC without size restrictions and no overt vascular involvement were considered “listable” according to their Childs-Turcotte-Pugh status. Disease-specific allocation was implemented by UNOS in November 1998, at which time patients with confirmed intrahepatic HCC who met Milan criteria were deemed listable and were assigned UNOS status 2B. With implementation of the model for end-stage liver disease in 2002, HCC was assigned a disease-specific score.
Pretransplant peak AFP levels were available for 117 of the 144 HCC patients; serum creatinine levels on the day of transplantation were available for all patients. Data on pretransplant neoadjuvant therapy for HCC, including radiofrequency ablation, transcatheter arterial chemoembolization, resection, cryoablation, and percutaneous ethanol injection, also were available for all patients with known HCC before OLT.
For patients with known HCC before OLT, preoperative tumor characteristics, including tumor numbers and size, were estimated from the ultrasound, computed tomography, or magnetic resonance imaging studies performed most recent to the date of transplantation. Clinical staging under the Milan and UCSF criteria was performed retrospectively using tumor number and sizes from the imaging study closest to the time of transplantation.5, 10 For those patients who received pre-OLT HCC resection, pre-OLT tumor characteristics were based on pathologic examination of the resected specimens in addition to any new lesions seen on imaging, which developed during the interval between resection and OLT (n = 7). For all other patients who received preoperative neoadjuvant HCC therapy, tumor characteristics before therapy were used for this analysis because of the difficulty of obtaining accurate assessment of residual tumor dimensions from posttreatment areas of necrosis and scar.
Data regarding receipt and type of post-OLT adjuvant chemotherapy were collected for 127 patients who had clear documentation. Patients who either had completed or had begun a course of adjuvant chemotherapy by the time of follow-up were counted as having chemotherapy in this analysis.
At Massachusetts General Hospital, patients with HCV received post-OLT immunosuppression according to protocol, which varied according to the date of transplantation.12 Before July 1998, all patients received conventional steroid induction with methylprednisolone 200 mg tapered to 20 mg by posttransplant day 5 in addition to tacrolimus and azathioprine. Subsequently, non-HCV–infected patients were continued on this protocol. The subset of HCV-infected patients received methylprednisolone at lower doses of 20 mg tapered to 10 mg by post-OLT day 2–3 but with mycophenolate mofetil in place of azathioprine.
At New England Medical Center, post-OLT immunosuppression varied over the 10-year timeframe and initially consisted of conventional steroid induction with taper in addition to cyclosporine and azathioprine. By 1995, immunosuppression predominantly consisted of steroid induction and taper and tacrolimus in combination with either azathioprine or mycophenolate mofetil. By the late 1990s, to prevent HCV recurrence, attempts were made to minimize immunosuppression for HCV-infected patients, and early steroid withdrawal in combination with tacrolimus monotherapy was used.
The post-OLT immunosuppression also varied at the Lahey Clinic program. All patients received a conventional steroid bolus and taper to 20 mg by posttransplant day 6. Between 1992 and March 2002, steroids were weaned to 10 mg per day by month 6, and patients were off steroids by month 18. From March 2002 through March 2003, patients were weaned to 10 mg steroids per day by posttransplant day 7 and were off steroids by week 6. The primary immunosuppressive agent used throughout the study period was tacrolimus. Azathioprine was used regularly until the late 1990s, and mycophenolate mofetil was used thereafter. During the final year of the study, mycophenolate mofetil was stopped after 6 months unless acute rejection or renal dysfunction occurred.
Histopathologic Analysis of Explants
The explanted livers were examined by experienced histopathologists at the respective institutions. The number of tumors, tumor dimensions, histologic grade (grade 1: well-differentiated; grade 2: moderately differentiated; grade 3: poorly differentiated), and presence or absence of microscopic vascular invasion was recorded. Fourteen patients did not have complete pathologic data regarding total number of tumors or tumor dimensions, including 1 patient with incidental HCC. Sixty-eight percent (98/144) of patients had data regarding degree of differentiation of their HCC. For tumors with a range of histologic differentiation, the tumors were categorized under the less differentiated category, for example, well to moderately differentiated tumors were classified as moderately differentiated (grade 2). Three patients did not have data regarding microvascular invasion. Each case was retrospectively staged according to the Milan and UCSF criteria on the basis of the histopathologic reports. Pathologic Milan staging was assessable in 133 patients (92%), and UCSF staging was assessable in 130 patients (90%).
Statistical and Survival Analysis
The HCC and non-HCC cohorts were compared by age at transplant using the 2-tailed independent samples Student t test and by gender and etiology of liver disease using the Fisher exact test. Kaplan-Meier survival data for both the 144 HCC patients and the 525 patients without HCC were computed using SPSS 11.0 for Windows (SPSS Inc, Chicago, Ill). Date of recurrence was defined as the first date of imaging that documented recurrent HCC or the first AFP rise from posttransplant baseline, after which imaging studies documented recurrent cancer. Time to recurrence was censored at the date of the last follow-up. To assess potential predictors of mortality or recurrence among the HCC cohort, overall and recurrence-free survival were compared using the following preoperative variables: etiology of liver disease (HCV vs. non-HCV), pretransplant peak AFP levels (AFP ≤ 10 ng/mL vs. AFP > 10 ng/mL and AFP ≤ 100 ng/mL vs. AFP > 100 ng/mL), tumor number, largest preoperative tumor size (≤ 3 cm vs. > 3 cm and ≤ 5 cm vs. > 5 cm), and preoperative neoadjuvant HCC therapy. Overall survival and recurrence-free survival also were compared using the following postoperative variables: known versus incidental HCC, largest pathologic tumor size (≤ 3 cm vs. > 3 cm and ≤ 5 cm vs. > 5 cm), total tumor diameter, vascular invasion, degree of histologic differentiation (well, moderate, poor), and receipt of adjuvant chemotherapy. In addition, overall and recurrence-free survivals were analyzed on the basis of fulfillment of Milan and UCSF criteria preoperatively and pathologically. Patients with incidental HCC were excluded from the preoperative Milan and UCSF criteria and the largest preoperative tumor size analyses.
The log-rank test was used to test the equality of survival distributions using the variables stated previously. A P value of less than .05 was considered statistically significant. All variables reaching statistical significance as predictors of mortality or recurrence under the Kaplan-Meier analysis were entered into a multivariate Cox proportional hazard regression analysis. Eight patients without HCC were transplanted twice; all were in the non-HCC cohort. They were censored at the time of retransplantation and treated as a second case for their second transplant.
Baseline characteristics of the 144 patients with HCC and the 525 patients without HCC are summarized in Table 2. The HCC population had a larger proportion of men (P < .001), who were on average older (P < .001) and who fell into a narrower age range. When compared with the non-HCC populations, the HCC patients had a higher proportion of liver disease caused by HCV and alcoholic liver disease compared with the nonmalignant disease OLT population (60.4% vs. 35.4% and 41.7% vs. 25.9% respectively, P < .001 for both).
Table 2. Baseline Characteristics of 144 OLT patients with HCC and 525 Patients with Nonmalignant Liver Disease*
HCC (n = 144)
non-HCC (n = 525)
Several patients had more than 1 etiology of liver disease.
Abbreviations: OLT, orthotopic liver transplantation; HCC, hepatocellular carcinoma; ALD, alcoholic liver disease; HCV, hepatitis C virus; HBV, hepatitis B virus.
M:F (% male)
Mean age (yr)
Age range (yr)
Etiology of liver disease
HCV + ALD (%)
Preoperative and clinical staging characteristics and postoperative variables for the 144 HCC patients are summarized in Table 3. Of the 144 patients with HCC, 117 HCC patients had available pretransplant AFP levels with a median of 16 ng/mL. Only 105 had HCC diagnosed before OLT, and 33 received therapy before transplant. Eighty-eight patients with known HCC before OLT had sufficient imaging data to retrospectively determine clinical (preoperative) Milan or USCF criteria. Of these patients, 74 met the Milan and 81 met the UCSF criteria.
Table 3. Preoperative and Postoperative Variables Among 144 HCC Patients*
Abbreviations: pre-OLT, preorthotopic liver transplantation; HCC, hepatocellular carcinoma; UCSF, University of California San Francisco; AFP, alpha-fetoprotein.
Thirty-eight patients with incidental HCC were excluded from the analysis of the following preoperative variables: median number of tumors, median primary tumor size, preorthotopic liver transplantation HCC therapy, Milan criteria, and UCSF criteria.
Median no. tumors
Median primary tumor size (cm)
Median AFP (n = 117) (ng/ml)
Pre-OLT HCC therapy for patients with HCC diagnosed before OLT (n = 105)
Patients with known HCC before OLT with sufficient imaging data (n = 88)
Patients with documented adjuvant therapy (n = 127) receiving chemotherapy
Median no. tumors
Median primary tumor size (cm)
Milan criteria (n = 133)
UCSF criteria (n = 130)
Vascular invasion (n = 141)
Degree of differentiation (n = 98)
For analysis of postoperative variables, 38 patients had HCC found incidentally at transplant. Of the 127 patients who could be assessed for receipt of adjuvant treatment, 30 patients received systemic chemotherapy. The number of patients meeting the Milan and USCF criteria was 86 (64.7%) and 96 (73.8%) respectively.
Forty-six patients died within the follow-up period; only 14 of these deaths resulted from recurrent HCC. Median follow-up time for survivors within the HCC cohort was 20.6 months (range 0.8 to 126 months). Twelve patients died of a non-HBV or non-HCV infectious process, and the remaining died of multiorgan failure, cardiac arrest, hemorrhage, other cancers, or recurrent HBV or HCV.
Overall and Recurrence-Free Survival Analysis
Survival analysis revealed that HCC patients had reduced survival rates compared with non-HCC OLT patients among the 3 institutions (Fig. 1). One, 3-, and 5-year survival rates for HCC and non-HCC patients were 80.3%, 59.3%, 46.7% and 81.5%, 75.3%, and 70.6%, respectively (P = .02). When examining the survival with regard to the timing of OLT, the 5-year survival rate was 38.7% before and 56.3% after 1998 (P = .11), the year that implementation of disease-specific allocation criteria for HCC were initiated. After institution of the Milan criteria in 1998, significant differences in survival between HCC and non-HCC patients no longer occurred (P = .19) (Fig. 2).
When comparing the HCV-infected subsets of HCC and non-HCC patients, survival was not significantly different (1-, 3-, 5-year survivals 83.6%, 55.4%, 48.5% vs. 82.9%, 74.6%, 68.7%, respectively, P = .13) (Fig. 3). Complementary survival analysis of HCC and non-HCC patients without HCV as an etiology of liver disease showed significantly different survival outcomes between the two populations (P = .04) (Fig. 4). When comparing survival of HCC and non-HCC OLT patients among patients with viral etiology of liver disease (HBV, HCV) there was no significant difference (P = .12). Complementary survival analysis of HCC and non-HCC patients with non-viral liver disease, showed a significant difference in survival (P = .05, data not shown).
Twenty-two patients developed recurrent HCC by the end of the follow-up period (15.3%). Of these patients, 21 had available explant pathology, and 12 patients exceeded both the Milan and UCSF criteria. The median time to recurrence was 12.3 months. The median and mean time to death after recurrence was 6.65 and 8.9 months, respectively, for the 16 recurrent patients who had died (14 of whom died of recurrent HCC) within the follow-up period (range 2–23.6 months). Six patients who developed recurrent HCC were still alive by the last date of follow-up and had a median follow-up time of 9.1 months after recurrence. One patient has lived for 39 months since recurrence. This patient received adjuvant chemotherapy.
Analysis of recurrence-free survival rates for the 144 patient HCC cohort revealed 1-, 3-, and 5-year recurrence-free survival rates of 79.6%, 56.3%, and 42.8%, respectively.
Univariate Analysis of Predictors of Survival
Univariate analysis of 117 HCC patients with available pretransplant AFP levels revealed that an AFP of 10 ng/mL or less was associated with an overall 1- and 5-year survival rate of 91.1% and 64.1% when compared with 1- and 5-year survival of patients with an AFP greater than 10 ng/mL (79.7% and 31.5%, P = .005) (Table 4a). When stratifying patients under a more rigorous AFP cutoff level of AFP of 100 or less compared with AFP greater than 100, the difference in survival rates was also statistically significant; however, these levels were somewhat less predictive (P = .01). It should be acknowledged that this may be because of the small numbers of patients with AFP greater than 100 (n = 27).
Table 4a. Univariate Analysis of Statistically Significant Preoperative and Postoperative Variables as Predictors of Overall and Recurrence-Free Survival*
Overall survival (%)
Recurrence-free survival (%)
One-, 3-, and 5-year survivals are shown.
Clinical UCSF staging
Overall survival (%)
Recurrence-free survival (%)
Pathological UCSF staging
MILAN and UCSF Criteria
Eighty-eight of 144 patients with known HCC before OLT had sufficient imaging data to retrospectively determine clinical (preoperative) Milan criteria. The 74 of 88 patients who met Milan criteria did not have significantly better survival than those 14 of 88 patients who did not (P = .12, Fig. 5a). Interestingly, however, when comparing patients using clinical UCSF criteria, those patients meeting clinical UCSF criteria had a better survival rate than those whose tumor burden exceeded the UCSF criteria (P = .02, Fig. 5b). Seven patients had tumor burdens that exceeded Milan criteria but met UCSF criteria preoperatively. Among these patients, 1 died on the day of transplant immediately postoperatively because of myocardial infarction. The remaining 6 were alive at the last date of follow-up, with a median follow-up time of 14.6 months (range 1–47.5 months).
Survival between groups meeting or exceeding pathologic (postoperative) Milan criteria was not significantly different (P = .19). However, meeting pathologic UCSF criteria was associated with statistically significantly improved survival (P = .04). Ten patients who exceeded pathologic Milan criteria met pathologic UCSF criteria. This group of 10 patients had a 2-year survival rate of 77.1%. Two patients died, 1 at 3 months and the other at 22 months after OLT. Of the remaining 8 patients, median follow-up time was 32 months (range 6.6–73.5 months).
When only those patients with HCC meeting preoperative Milan criteria (74 patients) were compared with non-HCC patients, a significant difference in survival between the two groups was no longer evident (P = .77). The same was true when patients with HCC meeting UCSF criteria (67 patients) were compared with non-HCC patients (P = .82). The 1-, 3-, and 5-year survival rates for HCC patients meeting clinical Milan and UCSF criteria were 85.9%, 63.7%, 50.9% and 85.8%, 64.8%, 51.9%, respectively.
The following variables were analyzed and were not significant predictors of overall survival: pre-OLT neoadjuvant HCC therapy, largest preoperative tumor size (≤ 3 cm vs. > 3 cm or ≤ 5 cm vs. > 5 cm), incidental versus known HCC, microvascular invasion, tumor differentiation, or adjuvant post-OLT chemotherapy.
Univariate Analysis of Predictors of Recurrence-Free Survival
Analysis of predictors of recurrence-free survival revealed no difference in recurrence-free survival between subsets of patients based on the variables of tumor size, incidental versus known tumor at time of OLT, neoadjuvant HCC therapy, clinical or pathologic Milan staging, microscopic vascular invasion, or degree of differentiation. Pathologic UCSF staging, however, reached statistical significance in predicting recurrence-free survival (P = .008) although clinical UCSF staging showed a trend toward significance (P = .07). In univariate analysis, AFP of 10 ng/mL or less reached significance as a predictor of recurrence-free survival (P = .003) as did AFP of 100 ng/mL or less (P = .004).
Of these 22 patients with recurrent disease, 11 patients received adjuvant chemotherapy. Receipt of adjuvant chemotherapy was not associated with a difference in recurrence-free survival among these patients and was not associated with a longer survival time after recurrence (P = .31).
Cox regression analysis was used to analyze the association of the following variables with overall and recurrence-free survival: AFP of 10 ng/mL or less and clinical UCSF criteria and pathologic UCSF criteria. Because the variables of clinical and pathologic UCSF criteria are not independent of each other, separate analyses were performed combining AFP of 10 ng/mL or less and clinical UCSF criteria and combining AFP of 10 ng/mL or less and pathologic UCSF criteria. In multivariate analysis of AFP of 10 ng/mL or less and clinical UCSF criteria, both AFP of 10 ng/mL or less and clinical UCSF criteria were independent predictors of overall survival (P = .01, P = .01 respectively; Table 4b). Both variables were also independent predictors of recurrence-free survival (P = .008, P = .007 respectively). When AFP of 10 ng/mL or less was analyzed with pathologic UCSF criteria, AFP of 10 ng/mL or less was statistically significant, and pathologic UCSF criteria trended toward significance as independent predictors of survival (P = .04, P = .09 respectively). Both variables were significant independent predictors, however, of recurrence-free survival (P = .03, P = .03, respectively).
Table 4b. Multivariate Analysis of Statistically Significant Univariate Variables as Predictors of Overall and Recurrence-Free Survival†
Survival at Mean of Covariates (%)
Recurrence-Free Survival at Mean of Covariates (%)
One-, 3-, and available 5-year survivals at the mean of the covariates are shown. P values for individual covariates are shown.
Abbreviations: AFP, alpha-fetoprotein; UCSF, University of California San Francisco.
AFP ≤10 ng/mL
Clinical UCSF staging
AFP ≤10 ng/mL
Pathologic UCSF staging
Over the 11-year timeframe of our study, HCC patients undergoing OLT appeared to have a significantly reduced survival compared with their counterparts without HCC. However, since the implementation of the UNOS policy for organ allocation defined by the Milan criteria in 1998, patients transplanted with HCC and nonmalignant disease have had no statistically significant difference in survival outcomes. Furthermore, HCV-infected HCC patients had a survival comparable with their non-HCC HCV-infected counterparts, consistent with the study by Min et al.13 We also found that survival between HCV-infected and uninfected patients both within the HCC cohort and within the entire OLT population was not significantly different. Thus, the early disparity in survival between the HCC and non-HCC patients is mainly attributable to differences in survival among HCV-uninfected individuals. The basis for this difference remains unclear.
A significant portion of patients underwent transplantation despite exceeding clinical Milan criteria (14/88); 10 received OLT after UNOS criteria for HCC were implemented. For these patients, transplantation was justified at the time of listing either because of receipt of preoperative neoadjuvant HCC therapy (7/14), which successfully decreased the preoperative tumor burden, or because of ambiguity with respect to interpretation of the criteria (7/14). The inaccuracy of preoperative imaging studies may have contributed as well. Tumor characteristics before preoperative neoadjuvant therapy were used in this analysis because of the difficulty in accurate assessment of residual tumor dimensions from postablation or postchemoembolization scars. Thus, many of those patients who exceeded clinical Milan and UCSF criteria were patients with unacceptable tumor burdens whose ultimate outcome was not altered by objective response to preoperative neoadjuvant therapy, as shown by the fact that neoadjuvant therapy did not affect survival.
The less restrictive clinical UCSF criteria better separated survivors from nonsurvivors in our study. Furthermore, comparison of survival between HCC patients meeting Milan criteria and non-HCC patients revealed no significant difference. The same was true for comparison of survival between HCC patients meeting UCSF criteria and non-HCC patients. The small number of patients who failed to meet either Milan or UCSF criteria may limit the generalizability of these findings. Taken together, these findings suggest that, regardless of the particular criteria adopted, adherence should be rigid and that neoadjuvant HCC therapy in patients with tumors exceeding these criteria does not appear to justify OLT.
Several prior studies have shown similar survival rates for OLT patients with and without HCC based on the Milan criteria.5–8 Mazzaferro et al.5 reported a 4-year overall survival rate of 85% among HCC patients, and Figueras et al.7 reported a 5-year overall survival rate of 75%, which was essentially identical to the survival rate in their patients without HCC. In our cohort of patients, the 5-year survival rate was 70.6% in non-HCC patients and 46.7% in HCC patients. These HCC survival rates are nevertheless significantly improved over those previously observed in our institutions as described in Haug et al. more than a decade ago, in which actuarial survival rates for patients with HCC were 71%, 56%, and 42% at 1, 2, and 3 years, respectively.11 Interestingly, the survival outcomes between HCC and non-HCC patients were similar after 1998, when implementation of disease-specific allocation for HCC was initiated. Although not statistically significant, the 5-year survival rate for all outpatients was higher after implementation of Milan criteria in 1998, suggesting an improved OLT outcome for both HCC and non-HCC patients during the recent years. We speculate that these patients benefited from continued technical advances in transplantation, improvements in immunosuppression, and postoperative management.
In our study, AFP levels of 10 ng/mL or less and AFP levels of 100 ng/mL or less were significant predictors of improved overall and recurrence-free survival. In previous studies, pre-OLT AFP levels have shown variable significance as a predictor of overall and recurrence-free survival, and different cutoff levels have been analyzed. The UCSF group showed that an AFP level of more than 1,000 ng/mL was a predictor of poor overall mortality. Figueras et al. showed that an AFP level of 300 ng/mL or more was a predictor of poor overall and recurrence-free survival.7 Herrero et al., however, did not show the variable of AFP 10 ng/mL or less or 100 ng/mL or less to be clinically meaningful.14 In our study, AFP levels of greater than 1,000 ng/mL approached statistical significance as a predictor of survival (P = .06) and were statistically significant as a predictor of HCC recurrence (P = .02). However, AFP levels of greater than 300 ng/mL trended toward significance as a predictor of poorer survival and HCC recurrence (P = .08, and P = .10 respectively).
Aside from clinical UCSF criteria and AFP levels, none of the other variables studied were significant predictors of mortality or recurrence-free survival in univariate analysis. In other studies, several variables have been shown to be significant predictors of outcome, including vascular invasion, bilobar disease, pathologic T4 (vs. pT2), total tumor diameter greater than 8 cm, pathological Tumor Node Metastasis stage IV, and age at transplant greater than 55, but none of these have been consistently shown across multiple studies to be significant. Age at transplant, pT4 versus pT2, total tumor diameter greater than 8 cm, and vascular invasion were not significant predictors in our study. Interestingly, the recent UCSF study,10 which justifies expansion of OLT selection criteria by demonstrating statistically improved survival among patients meeting these expanded clinical and pathologic criteria, was confirmed by our study for both the clinical and pathologic criteria. In the study reported by Mazzaferro et al., fulfillment of pathologic Milan criteria was important in predicting the outcome.5
Thirty-eight (26.4%) patients in our cohort had incidental HCC, which is less than the 33% incidental rate reported by the group from UCSF, the 41% reported by Klintmalm et al. for the International Registry of Hepatic Tumors, and the 36% reported by the group from the University of Pittsburgh.10, 15, 16 Klintmalm et al. reported that incidental tumors were more likely to be smaller (< 5 cm) and solitary and less likely to have vascular invasion, less likely to be bilobar, and less likely to be poorly differentiated.15 The UCSF group reported the worrisome finding that the main lesion in 30.4% and 13% of incidental tumors were at least 3 cm and 5 cm respectively. In contrast, the Pittsburgh study found that only 5% of incidental HCCs were larger than 4 cm. In our study, all the tumors found at OLT were solitary with a median diameter of 3 cm. Unfortunately, 29% (11/38) and 11% (4/38) of patients with incidental HCC had primary tumors at least 3 cm and 5 cm, respectively. These frequencies are comparable with the UCSF data. Incidental tumors tended to be smaller (P = .08) and to have a lesser degree of vascular invasion, but neither of these variables reached significance. Forty-one percent of these patients had well-differentiated pathology, and 59% had moderately differentiated pathology. Interestingly, this group did not have a better survival than those with known HCC, consistent with previous studies.10, 15, 16 This may be because of the unusually high frequency of large incidental tumors.
The receipt of postoperative adjuvant chemotherapy did not confer improved overall or recurrence-free survival, which calls into question the utility of this modality. In fact, no other study analyzing HCC OLT outcomes has demonstrated a difference in overall or recurrence-free survival on the basis of the receipt of adjuvant chemotherapy. The fact that adjuvant chemotherapy does not significantly improve survival in this analysis may in part be related to the finding that those receiving chemotherapy had significantly larger postoperative tumor diameters (P = .010) and total tumor diameters (P = .006), although neither of these variables was associated with decreased survival. Because of the small number of patients receiving adjuvant chemotherapy, the retrospective nature of the analysis, and the lack of randomization of placebo control arm, our analysis cannot definitively assess the potential value of adjuvant chemotherapy in the post OLT setting. A randomized trial of postoperative adjuvant therapies for HCC is warranted to address this important question.
Our cumulative experience transplanting patients with and without HCC demonstrates that patients transplanted with HCC, whether discovered incidentally or before OLT, have a less favorable survival in comparison to patients with benign disease. Since the institution of the Milan criteria for OLT in 1998, survival outcomes for patients transplanted for HCC and for those with nonmalignant disease are no longer statistically different.
Of the preoperative and postoperative variables analyzed in our study, AFP levels of 10 ng/mL or less and clinical and pathologic UCSF criteria were predictors of survival in univariate analysis and AFP levels of 10 ng/mL or less and pathologic UCSF criteria were predictors of recurrence-free survival. These data justify a policy of assigning weighted allocation points to those patients with early intrahepatic HCC, which may be defined by either the Milan or UCSF preoperative criteria. In fact, the UCSF criteria may better distinguish more appropriate liver transplant candidates. Regardless of the criteria used for OLT selection, however, adherence to these criteria should be maintained for optimal outcome. Furthermore, because these criteria are a surrogate for extrahepatic tumor spread, our results suggest that patients with preoperative tumor burdens exceeding these criteria will likely have poorer transplant outcomes regardless of whether they receive preoperative intrahepatic neoadjuvant therapy to reduce tumor sizes to acceptable dimensions.
Raymond T. Chung was supported by NIH DK57857, the Hepatitis Foundation International and the Christopher Tripoli Hepatitis C Research Fund. Jessica Y. Leung was supported by the Pasteur Clinical Research Program and the Office of Enrichment Programs, Harvard Medical School. We thank Dr. David Schoenfeld for his help with statistical analysis and Ms. Dottie Monahan for her assistance in preparing this manuscript.