Appropriate patient selection is crucial in ensuring acceptable outcomes from orthotopic liver transplantation (OLT) for hepatocellular carcinoma (HCC). The United Network for Organ Sharing (UNOS) has elected to prioritize HCC patients for OLT based on criteria of tumor burden. However, it is unclear whether these criteria correlate with outcome, or with the pathobiological features associated with tumor recurrence. Therefore, we analyzed 109 consecutive patients undergoing OLT for HCC at our center, to determine the utility of present selection criteria in predicting outcome. Pathologic tumor staging of the explanted liver was based on the American Tumor Study Group modified tumor node metastases (pTNM) classification system. Multifocality was defined as >4 tumor nodules on explant. Survival analysis was performed using Kaplan-Meier and Cox proportional hazards regression methods. At a median follow-up of 18.9 months, the overall mortality was 19% with 15 patients (14%) dying of recurrent HCC. Kaplan-Meier 1, 3 and 5-year survival rates were 89.5%, 68%, and 65%, respectively. Recurrence-free rates of 1, 3, and 5 years were 89%, 75%, and 65%, respectively. On univariate analysis, the factors found to be significantly associated with recurrence of HCC were explant features of macrovascular invasion, tumor size (per centimeter increase), pTNM stage (per 1-stage increase), and pre-transplant serum alphafetoprotein (AFP) >300 ng/mL. In defining a threshold level, we found that explant tumor diameter ≥3 cm, and those tumors classified as at least pT3 on pathological examination, were significantly associated with recurrence (P = .01 and .03, respectively). Tumor size on explant was found to be strongly correlated with multifocality (P = .017) and vascular invasion (P = .02). Patients exceeding pathological UNOS criteria were 3.1 times more likely to have recurrence of HCC (P = .03). In conclusion, we found that tumor size appears to be a surrogate marker for negative pathobiological predictors of outcome, i.e., vascular invasion and multifocality. Present UNOS selection criteria for HCC based on tumor burden appear to provide adequate discriminatory power in predicting outcome of OLT. (Liver Transpl 2004;10:911–918.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The role of orthotopic liver transplantation (OLT) in the management of hepatocellular carcinoma (HCC) has evolved significantly over the past decade. The initial experience with OLT was limited to those with extensive tumors and was marked by uniformly dismal outcomes.1, 2 The first suggestion that favorable tumor characteristics may be associated with decreased recurrence came from the University of Pittsburgh experience wherein a series of unexpected or “incidental” tumors were found to have surprisingly good survival rates.3 This led to the recognition that early cancer detection and meticulous pretransplant staging were the key to successful outcomes. An important study in 1996 by Mazzaferro and his colleagues from Milan confirmed this view.4 By carefully selecting patients who satisfied specific tumor criteria, the Milan group was able to achieve excellent recurrence-free survivals of 92% at 3 years.
Against this encouraging backdrop, HCC once again became an important indication for transplantation. Recognizing that these patients were disadvantaged by an organ allocation system based solely on decompensated cirrhosis, the United Network of Organ Sharing (UNOS) decided to prioritize allocation of organs to those HCC patients who met the tumor criteria recognized in the Milan experience to have the best outcomes.4 The Milan criteria are based on tumor burden and limit prioritization for OLT to those who have either a single tumor under 5 cm or those with 3 or less tumors each under 3 cm, without evidence of metastatic disease or vascular invasion.
However, several questions persist regarding the general application and continued utilization of these relatively arbitrarily chosen criteria. Independent data validating their utility are limited, and it is unclear whether these criteria truly correlate with the pathobiological features predictive of tumor behavior. Additional concerns exist that these guidelines may be too restrictive, potentially excluding some patients who might benefit from OLT.
Tumor characteristics predicting post-OLT recurrence remain unsettled. Several different studies have reported that tumor size,5, 6 positive lymph nodes,5 poorly-differentiated histological features,5, 6 and bilobar disease7 show varying correlations with recurrence risk. Only vascular invasion predominated as a clear predictor of poor outcome.
In addition, consensus on the ideal tumor staging system is lacking. The traditional pathological pTNM classification does not appear to have adequate discriminatory power to predict HCC behavior.8 To address this, the American Liver Tumor Study Group recently incorporated the Milan criteria into its staging system, by using tumor size and number as factors in assessing tumor stage (Table 1).9 However, the main drawback of this system is its reliance on explant characteristics, which precludes its use as a tool in determining candidacy for OLT.
Table 1. Modified Tumor Node Metastases (pTNM) Classification of HCC
Tumor not found
1 nodule ≤1.9 cm
1 nodule 2.0–5.0 cm; 2 or 3 nodules, all ≤3.0 cm
1 nodule >5.0 cm; 2 or 3 nodules, at least one >3.0 cm
4 or more nodules, any size
pT2, pT3 or pT4a plus gross intrahepatic portal or hepatic vein involvement as indicated by CT, MRI, or US
Regional (portal hepatitis) nodes, involved
Metastatic disease, including extrahepatic portal or hepatic vein involvement
Any N1, any M1
These issues gained significance with the adoption of the model for end-stage liver disease (MELD) score in 2002.10 Recognizing that in early-stage HCC the risk of tumor progression supercedes that of death,11 patients with “early” HCC (i.e., satisfying Milan criteria) were allotted additional or “MELD exception” scores, thereby increasing their likelihood of transplantation. Many transplant centers, including ours, witnessed an unprecedented increase in the number of patients undergoing OLT for HCC. The increasing incidence of HCC in the United States12, 13 ensures that HCC will continue to be an important and growing indication for OLT. Therefore, it becomes crucial to re-examine the validity of our present selection criteria and to assess their impact on post-OLT outcomes.
Our study was aimed at assessing the utility of the Milan criteria in predicting recurrence-free survival following OLT, as well as to ascertain the correlation between these criteria and the pathobiological features impacting on tumor recurrence. Also, we attempted to define the correlation between pre-OLT imaging studies and HCC explant characteristics.
OLT, orthotopic liver transplantation; HCC, hepatocellular carcinoma; UNOS, United Network for Organ Sharing; pTNM, modified tumor node metastases classification; AFP, alphafetoprotein; CT, computed tomography; UCSF, University of California at San Francisco; MELD, Model for End-stage Liver Disease; PEI, percutaneous ethanol injection; TACE, transarterial chemoembolization; RFA, radiofrequency ablation; MRI, magnetic resonance imaging.
This was a cohort study utilizing the computerized transplant database of the Hospital of the University of Pennsylvania, Philadelphia, PA. All patients with HCC undergoing OLT between February 28,1996 and January 13,2003 were included in the study. Demographic and clinical information analyzed included age, sex, race, primary liver diagnosis, performance status, and presence of decompensations related to cirrhosis (variceal hemorrhage, ascites, encephalopathy). Laboratory investigations recorded most proximal to OLT included platelet count, albumin, bilirubin, creatinine, prothrombin time, and alphafetoprotein (AFP). Imaging examinations were classified according to type, presence of contrast enhancement and whether they were performed at our institution or elsewhere. Institutional review board approval was obtained before data collection.
The explanted liver was examined by an experienced histopathologist. All explants were sliced at 5 mm intervals, and detected tumors were carefully noted. Tumor location, size, number, presence of satellite lesions, and micro- or macrovascular invasion, if any, were recorded. Tumor grading was performed according to the modified Edmondson criteria14 (grade 1: well-differentiated; grade 2: moderately-differentiated; grade 3: poorly differentiated). Enlarged lymph nodes were sectioned and examined for metastatic disease.
Immunosuppressive therapy consisted of a calcineurin inhibitor (cyclosporine or tacrolimus) in combination with corticosteroids. Mycophenolate or azathioprine were not used routinely. Steroids were tapered within 6 months of OLT in the majority of patients. Tumor recurrence was assessed by measurements of AFP and computed tomography (CT) scan/magnetic resonance imaging of the chest/abdomen according to a well-defined protocol (i.e., every 3 months for the first year, every 6 months for the second year, and yearly thereafter).
Patients were followed until death, transplantation, or study closure (arbitrarily denoted as February 14, 2003). The outcome measures of interest were survival and recurrence-free survival. We report, therefore, all-cause mortality and recurrence of HCC as our primary end-points of interest.
Univariate analysis was performed on multiple clinical and pathological criteria. The clinical criteria studied included age, gender, race, ascites, encephalopathy, functional status, hepatitis serologies (including positivity for hepatitis B core antibody), and creatinine. The tumor criteria analyzed were the tumor location, size, multifocality (defined as more than 4 separate tumors), and vascular invasion (either micro- or macrovascular). Each of these criteria was categorized separately as to whether it was described on pre-transplant imaging or on explant analysis. Correlation between radiological studies and pathological analysis was then measured using Spearman correlation coefficients. Patients were classified as exceeding UNOS criteria on either radiological or pathological criteria. Patients exceeding UNOS criteria on either basis were then compared to those patients satisfying these criteria in terms of survival and tumor recurrence. Survival analysis was performed using Kaplan-Meier and Cox proportional hazards regression methods. Comparison of relevant variables used the Wilcoxon rank sum test and the Kruskal-Wallis test.
A total of 689 patients underwent OLT during the study period. Of these, 109 had HCC (16% of total transplants), either noted on pre-OLT imaging studies, or discovered only on explant examination (denoted as “incidental” tumors). After the MELD allocation system came into effect, the percentage of patients transplanted for HCC increased markedly (i.e., 40 of a total of 116 OLTs, or 34%). Demographic characteristics of patients transplanted for HCC are summarized in Table 2. Three patients underwent living donor liver transplant.
Table 2. Baseline Characteristics of 109 Patients With HCC Undergoing Liver Transplantation
A total of 95.5% of patients had an imaging study recorded before OLT. Contrast-enhanced magnetic resonance imaging and CT were performed in 72 (69%) and 17 (16%) of patients, respectively, with the remainder undergoing non-contrast examinations. The overall sensitivity of imaging studies in detecting the primary lesion was 77%. If performed within 6 months of OLT, the sensitivity was in the 82–88% range; however, when this interval exceeded 6 months, the sensitivity fell to 44%. Of note, the proximity of imaging to transplant was much greater in the pre-MELD era, after which we instituted a more rigid screening protocol for waiting patients. This probably explains the larger number of significant size “incidental” tumors and understaged tumors.
When analyzed by type of imaging study, the respective sensitivities were as follows: 65% using CT scan with contrast, 88% using magnetic resonance imaging with contrast, and 33% using CT scan without contrast.
After excluding incidental tumors and those with missing data, we were able to correlate radiological and explant data in 72 patients. Accurate HCC stage estimation was achieved in 66.6% of patients; stage underestimation was observed in 20 cases (27.8%). Among these, 9 patients had pT3 HCC underestimated as T2: as a result of underestimating the size of the main lesion in 2 patients, missing additional nodules in 5 patients, and a combination of both these factors in 2 patients. Five and 4 patients, respectively, had pT4 tumors underestimated as T2 and T3 tumors, always attributable to the nondetection of satellite nodules. Overall, inaccurate underestimation of the tumor occurred in 14 of the 20 patients that converted them from a “non-transplantable” (i.e., pT3/pT4) to a transplantable stage (T2). Overestimation of the tumor occurred in 4 patients (5.6%). The majority of these (3 patients) were due to inaccuracies in measuring tumor diameter; in 1 case, macroregenerative nodules were read as a tumor. This data is summarized in Table 3.
Table 3. Accuracy of Pre-transplant Imaging Procedures in the Diagnosis and Staging of Nonincidental HCC
Abbreviations: CT, computed tomography; MRI, magnetic resonance imaging.
Overall sensitivity of detecting main lesion (n = 104)
CT w/contrast (n = 17)
MRI w/contrast (n = 72)
CT w/o contrast (n = 12)
MRI w/o contrast (n = 2)
Ultrasound (n = 1)
Overall accuracy of imaging (within 1 cm of explant size)
Overall sensitivity of detecting satellite lesions (any imaging)
Pre-operative tumor staging (n = 72)
On stratification by tumor grade, 33, 44, and 11 patients had well-, moderately, and poorly differentiated tumors, respectively. A total of 17 patients had a multifocal tumor (i.e., >4 tumors in number). Microvascular invasion was noted in 27 patients, and macrovascular invasion in 6 patients. Missing data accounted for 10% of the total.
Follow-up ranged from 1.2 to 71.2 months. Median and mean duration of follow-up was 18.9 and 25.6 months, respectively. A total of 21 patients died in the course of follow-up (overall mortality 19%). Of these, 15 patients had recurrent HCC (overall HCC-related mortality 14%). Three other patients with documented recurrence were alive 6 months, 4 months, and 1 month from the time of diagnosed recurrence. The overall recurrence rate was 16.5%. The mean time from OLT to recurrence and death were 45.8 and 46.3 months, respectively. The mean period in months between documented recurrence and death was 186 days (range: 24 to 330 days).
Kaplan-Meier 1-, 3-, and 5-year survival rates were 89.5%, 68%, and 65%, respectively (Fig. 1). Recurrence-free rates of 1, 3, and 5 years were 89%, 75%, and 65%, respectively. Six patients died of causes other than HCC, hence explaining this discrepancy.
Predictors of Mortality
On univariate analysis, 2 parameters were found to correlate with both mortality and HCC recurrence. The first of these was the serum AFP level on the laboratory panel drawn most proximal to OLT. Using a cut-off value of 300 ng/mL, we found that patients with levels above this were significantly more likely to die and recur (HR = 12.08, P < .0001 for recurrence). However, only 10 patients in our series had an AFP level >300 ng/mL; hence this should be interpreted with caution.
The second parameter which significantly correlated with the risk of both death and recurrence was that of pathological vascular involvement; sub-analysis found that only macrovascular invasion was a significant predictor of outcome.
In addition to these 2 parameters, the following correlated significantly with HCC recurrence but not with mortality: explant tumor size (per centimeter increase); explant pTNM stage; and exceeding UNOS criteria on pathology. In attempting to detect a threshold level, we found that tumor size on pre-operative imaging studies of 4 cm and greater was associated with marginally worse recurrence rates (P =.06); a tumor size of ≥3 cm on explant was associated with significantly worse recurrence rates (P = .01). Those tumors classified by the pTNM system as being pT3 or greater (i.e., 1 nodule >5 cm; 2 or 3 nodules, at least one being >3 cm) were at higher risk of recurrence (P = .03). The grade of the tumor did not predict either recurrence or survival. This data is summarized in Table 4.
Table 4. Univariate Analysis of Clinical, Demographic and Pathological Characteristics as Predictors of Mortality and HCC Recurrence After OLT†
Time from OLT to Death
Time from OLT to Recurrence
HR (95% CI)
HR (95% CI)
Abbreviations: rad, radiology; HR, hazards ratio; UCSF, University of California at San Francisco; pTNM, modified tumor node metastases classification; OLT, orthotopic liver transplantation; HCC, hepatocellular carcinoma; UNOS, United Network for Organ Sharing.
After adjusting for age at OLT, we found the following correlations using multivariable Cox regression analysis: 1) both micro- and macrovascular involvement correlated with recurrence, but only macrovascular involvement correlated with death; and 2) AFP >300 ng/mL correlated with both recurrence and death. The risk of death for patients having macrovascular involvement was 5.47 times that of patients not having no vascular involvement. Patients having AFP ≥300 had 8.9 times greater risk of death compared to patients having AFP <300.
Outcome Based on Pretransplantation Selection Criteria
On the basis of pre-transplant imaging studies, 15 patients (14.4%) exceeded UNOS criteria, and were either transplanted from expanded criteria donors or underwent living donation. On pathological analysis, a total of 27 patients (26%) were noted to exceed UNOS criteria. Those exceeding UNOS criteria on the basis of radiological imaging did no worse than those satisfying these criteria (P value = .8). However, when UNOS criteria were applied to explant examination, we found a significant effect on time to recurrence (P = .03), but not to death (P = .3). In fact, those patients who exceeded UNOS criteria on pathological analysis were 3.1 times more likely to have recurrence (P = .03) than those who satisfied these criteria.
Also, we classified patients according to whether they satisfied the University of California at San Francisco (UCSF) criteria (i.e., solitary tumor ≤6.5 cm, or 3 or fewer nodules with the largest lesion <4.5 cm and total tumor diameter ≤8 cm), as defined by explant analysis. A total of 24 patients (21%) exceeded UCSF criteria as well as UNOS criteria; hence, only 3 of our patients who exceeded UNOS criteria satisfied UCSF criteria. We did not find that exceeding UCSF criteria was any more predictive of outcome than exceeding UNOS criteria.
Correlation Between Tumor Size and Other Parameters
On analyzing the association of tumor size on explant with other pathological variables, we found a strong correlation between tumor size and multifocality (P = .017), as well as between tumor size and either micro- or macrovascular invasion (P = .02). This suggests that tumor size may be a surrogate marker for biological parameters thought to be associated with high recurrence risk.
A total of 26 tumors (24%) were detected only at the time of explant examination, and hence were denoted as being “incidental”. The mean explant diameter of incidental tumors was 2.16 cm, significantly less than that of non-incidental tumors (mean 3.77 cm, P < .001), with a range from 0.6 to 4cm. The mean number of incidental tumors was 2, compared to 2.15 amongst non-incidental tumors (P = .51). While the majority of incidental tumors were early (i.e., in pTNM stages I and II), 3 tumors were found to be multifocal, of which 2 demonstrated macrovascular invasion. One of these patients had not been imaged in the year before OLT. The other 2 patients had undergone recommended screening with either a contrast-enhanced CT scan or magnetic resonance imaging within 4 months of OLT. Despite their smaller size and favorable stage, incidental tumors had similar survival/recurrence rates compared to non-incidental tumors (P = .17).
The efficacy of OLT in achieving excellent tumor-free survivals in HCC has been well-established. Our series confirms its utility in that regard. Our 4-year survival of 65% is comparable to that described in most series.4, 6, 15 Issues that remain unresolved include the variables impacting on tumor recurrence and ideal organ allocation criteria prioritizing HCC patients for OLT.
Our analyses indicate that large tumor size (specifically, if >4 cm on pretransplant imaging or >3 cm on explant examination) and advanced tumors (specifically ≥pTNM Stage III) correlate with higher recurrence rates. Macrovascular invasion was also strongly associated with poor outcomes. Furthermore, tumor size was shown to correlate with vascular invasion and multifocality, and hence may provide a surrogate marker for entities that are often difficult to detect before OLT. Therefore, our data provides strong validation of present UNOS selection criteria, which exclude large tumors, and specifically those that exceed pTNM stage III. The histological grade of the tumor appeared irrelevant to the outcome, which is at variance with some studies.16
Our results are similar to those reported in several large transplant series. In their pivotal study, Mazzaferro et al.4 demonstrated a 92% 4-year tumor-free survival rate in patients with pT2 HCC, compared with 59% for more advanced tumors. Figueras et al.17 in a smaller study demonstrated that tumor size >5 cm markedly reduced 1-year patient survivals (i.e., 25% compared to 88% in those with tumors <5cm). Similar results were demonstrated by Bismuth et al.18 who found that in 109 HCC patients, the 5-year survival rate in those with tumors <3cm in diameter was 66%, compared with 42% in those with tumors >5 cm.
Published data by Yao et al.15, 19 from UCSF examined their experience with 70 patients transplanted for HCC. Their data is similar to ours in that excellent 5-year survival rates of 72% were achieved in those with early (i.e., pT1 or pT2) HCC. However, their study suggested that survival was not compromised by application of the UCSF criteria, as described earlier. Of their group, 66% met Milan criteria. A total of 21% of those exceeding Milan criteria met UCSF criteria. Tumor recurrence was detected in 11% of patients at a mean of 8 months following OLT. The 2-year survival rate in those exceeding UCSF criteria was 50%, compared with 82% in the group satisfying these criteria. This data seems to argue that the UCSF criteria is a good discriminatory tool, and that expansion of standard selection criteria could lead to acceptable outcomes. However, this reported 2-year survival rate was associated with a wide 95% confidence interval, ranging from 54 to 96%, and results were based on pathological data rather than preoperative imaging. In our patients, this modest expansion of selection criteria did not appear to significantly impact on eligibility for transplantation. Only 3% of our patients exceeding UNOS criteria were within UCSF criteria. Exceeding UCSF criteria, while showing a trend toward worse survival, was less powerful a predictor of outcome than standard UNOS criteria.
In our series, 24% of tumors were incidental. The analysis of these incidental tumors may not be completely relevant in terms of defining the utility of pre-OLT radiological imaging and, by inference, UNOS criteria for selection. However, this patient subgroup offers valuable insights into the limitations of imaging studies in this context, and provides useful pathological data that we may use in correlating with outcome and in future survival analyses. Also, it further clarifies the outcome in this group of patients. In our series, even though these tumors were significantly smaller in diameter, they were not associated with better survival outcomes.
The literature on incidental hepatomas is somewhat contradictory. While data from both UCSF15 and the University of Pittsburgh7 show results similar to ours, a recently updated report from the International Tumor Registry20 somewhat disagrees. A total of 41% of those tumors were incidental, and the majority of these (93%) were <5 cm in diameter without evidence of vascular invasion. The overall survival in the incidental group was better than those with known HCC, but recurrence was noted in at least 15% of these patients, suggesting far from insignificant disease.
Our study also confirms the limitations of pre-operative imaging studies in staging tumors as well as in the detection of small satellite nodules. Our overall sensitivity in detecting the primary lesion was 75%. In concordance with other studies,21–26 we found that magnetic resonance imaging and contrast-enhanced CT scans provided the most accurate delineation of liver masses, but their utility was limited in smaller tumors. Most striking was the inability of any imaging study to detect the presence of satellite nodules–a limitation that led to understaging of a significant proportion of cases. As a result, OLT was undertaken in almost 20% of patients who on explant examination were found to have a degree of disease that should have precluded transplantation.
Present UNOS criteria have been criticized for their stringency and their potential to unfairly exclude selected patients with pTNM stage III disease who might fare well following OLT. While there is some evidence that modest expansion of tumor limits does not negatively impact outcome,15, 27 our data would argue that present UNOS criteria do justice to the majority of patients with HCC. In the absence of well-defined alternatives, these criteria serve as a good discriminant of those who would benefit most from an ever-scarcer resource. Further analysis of data gathered since the implementation of the MELD system with regard to prognostic significance of AFP levels, and more frequent imaging, will provide valuable insights. The fact remains that tumor behavior in HCC is poorly understood and incompletely elucidated. Ongoing research efforts utilizing molecular diagnostic techniques to identify precancerous changes such as loss of heterozygosity show promise, and may provide more accurate methods to stratify HCC patients according to the risk of recurrent disease. Until such time that these techniques become accepted into standard practice, present selection criteria, although not perfect and still in evolution, appear to be an appropriate means of determining organ allocation in those with HCC.