Moving past “One Size (and Number) Fits All” in the selection of candidates with hepatocellular carcinoma for liver transplantation


  • See Article on Page 1108

Address reprint requests to Francis Y. Yao, M.D., Division of Gastroenterology, Department of Medicine, University of California San Francisco, 513 Parnassus Avenue, Room S-357, San Francisco, CA 94143-0538. Telephone: 415-514-0332; FAX: 415-476-0659; E-mail:




hepatocellular carcinoma


locoregional therapy


liver transplantation


modified Response Evaluation Criteria in Solid Tumors


University of California San Francisco

Liver transplantation (LT) is a cornerstone in the management of early-stage hepatocellular carcinoma (HCC). Despite favorable overall posttransplant survival with the restrictive Milan criteria,[1] HCC recurrence still occurs in 10% to 15% of these patients.[1-4] On the other hand, some patients with tumors beyond the Milan criteria may achieve survival similar to the survival of patients with tumors within the Milan criteria.[2-5] Consequently, there are clearly factors beyond tumor size and number that are associated with worse tumor biology and thus a greater risk of HCC recurrence after LT. Microvascular invasion has been consistently shown to predict a poor prognosis after LT,[2-4] but the presence of microvascular invasion cannot be ascertained before LT.[6] In our ongoing efforts to refine current selection criteria and to further improve outcomes after LT, we are still searching for pretransplant tumor characteristics beyond size and number that can serve as surrogates for tumor aggressiveness and predict the likelihood of HCC recurrence after LT.

Although a serum tumor marker that reliably predicts the presence of microvascular invasion or a high risk of HCC recurrence after LT remains elusive, there is a growing body of evidence indicating that a high alpha-fetoprotein (AFP) level is associated with worse outcomes after LT.[7-13] However, the optimal AFP cutoff for predicting an unfavorable outcome has not yet been elucidated.[13] In recent years, there has also been growing interest in further defining the role of locoregional therapy (LRT) in the selection of candidates with HCC for LT.[14-16] Preliminary studies have suggested that a response to LRT based on the Response Evaluation Criteria in Solid Tumors may serve as a selection criterion for patients with HCC within or beyond the Milan criteria.[15, 16] The same principle has been applied to the down-staging of tumors by LRT within acceptable criteria based on tumor size and number, with the goal of achieving post-transplant survival similar to the survival of patients not requiring down-staging before LT.[2, 17]

In this issue of Liver Transplantation, Lai and colleagues from 6 centers throughout Europe further explore the impact of AFP and responses to LRT on outcomes after LT for patients with HCC within or beyond the Milan criteria.[18] In this large, multicenter cohort of 422 patients, the majority (73%) had HCC initially meeting the Milan criteria according to preoperative imaging, and the remaining patients were down-staged within the University of California San Francisco (UCSF) criteria.[3] When the modified Response Evaluation Criteria in Solid Tumors (mRECIST) were used to assess responses to LRT,[19] 29% exhibited a complete response, and 7% had progressive disease. HCC recurrence was observed in 14% of the study population. In a multivariate analysis of the entire population, patients with progressive disease had a 3.5-fold higher risk of HCC recurrence in comparison with patients with a complete response. The only stronger predictor of HCC recurrence was an AFP slope > 15 ng/mL/month, which was associated with a 5.4-fold increased risk of HCC recurrence. A majority of the patients who exceeded this AFP slope cutoff (14 of 26 patients or 54%) developed HCC recurrence, and they had an almost 4-fold increased risk of death after transplantation. The AFP slope was calculated in this study with 2 data points: at the time of listing and just before LT.

When the cohort was classified according to tumors within or beyond the Milan criteria, both an AFP slope > 15 ng/mL/month and mRECIST progression remained significant predictors of HCC recurrence in each group in a multivariate analysis. The authors also found that patients with HCCs initially outside the Milan criteria that were successfully down-staged had an excellent 5-year posttransplant tumor-free survival rate of 87% as long as they did not have an elevated AFP slope > 15 ng/mL/month or progressive disease according to mRECIST on imaging. This was similar to the 90% 5-year tumor free survival rate for those within the Milan criteria and without either risk factor. A drop in survival came only when a patient had at least 1 of these 2 risk factors: the 5-year tumor-free survival rate was 68% for those initially within the Milan criteria but only 42% for those who required down-staging.

Lai et al.[18] should be commended for their efforts in undertaking a large-scale, multicenter study to address an important and timely topic. The most intriguing aspect of this study is the observation that progressive disease according to mRECIST predicts poor posttransplant outcomes across all groups within and beyond the Milan criteria. The authors also address the complex, paradoxical situation in which tumors can be successfully down-staged to meet criteria for LT and yet still exhibit tumor progression according to mRECIST, often as a result of the development of new lesions. Although tumor progression despite LRT is a cause for concern before one proceeds with LT, there are more questions than answers. The evaluation of tumor responses after LRT was based only on the initial imaging and the last imaging before LT. This approach did not take into consideration the need for the reclassification of a tumor response if more than 1 LRT was performed between these time points, and this makes the interpretation of their results difficult. For example, if a tumor showed a complete response after the first LRT but recurred and was partially treated, this would have been classified as a partial response on the basis of the reassessment after the second LRT, but the patient might have been classified as having progressive disease in comparison with the initial imaging. The study by Lai et al. is retrospective and spans more than 10 years. It is unclear whether the radiographic assessments of responses to LRT were independently validated. The authors did not report wait-list times or wait-list outcomes in an intention-to-treat analysis. This information is important when one considers the effects of wait-list dynamics on the outcomes of patients exhibiting progressive disease after LRT. In a recent study from our institution involving 398 patients whose initial HCC met the Milan criteria at listing,[20] 71 patients (18%) had progressive disease according to mRECIST after their first LRT, and their cumulative incidence of dropout according to a competing risk analysis was 85% at 1 year. Consequently, a substantial proportion of patients with progressive disease after LRT might have already been selected out at centers with prolonged wait-list times.

The idea of examining the impact of the AFP slope on posttransplant outcomes is not new, and different methods and variable numbers of data points have been used in calculating the AFP slope.[21, 22] In a previous study by Vibert et al.[21] from Villejuif, France, a pretransplant AFP slope >15 ng/mL/month was associated with reduced overall survival as well as recurrence-free survival after LT. Lai et al.[18] used the same AFP slope of >15 ng/mL/month, which predicted both posttransplant HCC recurrence and mortality in the group with tumors meeting the Milan criteria but only HCC recurrence and not mortality in the subgroup exceeding the Milan criteria. Both studies used only 2 data points in calculating the AFP slope. This seems imprecise in assessing the trend of AFP levels over time and does not adequately account for the random fluctuations in the AFP levels that are often seen in patients with chronic liver disease even in the absence of HCC. Thus, a positive AFP slope relying on only 2 data points does not necessarily reflect a consistent rise in AFP levels before LT. The use of the initial AFP value at the time of listing in this study by Lai et al. also does not account for the initial effects of LRT on AFP. We would argue that a rising AFP value after LRT is a more useful marker for tumor progression. Another concern is the low sensitivity of an AFP slope > 15 ng/mL/month in predicting HCC recurrence (only 14% for the group within the Milan criteria and 31% for the group outside the Milan criteria): this raises doubts about its application in clinical practice as an exclusion criterion for LT.

Whether AFP progression is more relevant than static AFP levels as a prognostic factor requires further study. This is an important issue because data have continued to emerge supporting AFP as an important addition to tumor size and number in the selection of candidates for LT. In a large, multicenter French study, Duvoux et al.[7] developed and validated a predictive model using AFP combined with tumor size and number that performed significantly better than the Milan criteria alone in predicting HCC recurrence and overall survival. They identified a subset of patients with HCC exceeding the Milan criteria who had 5-year survival rates close to 70% as long as their AFP level was <100 ng/mL. Among patients with HCC within the Milan criteria, an AFP level > 1000 ng/mL was associated with a significantly higher risk of HCC recurrence and worse posttransplant survival. A recent study from our center found an AFP level > 1000 ng/mL at the time of LT to be strongly associated with the presence of microvascular invasion and tumor recurrence in patients with HCC meeting the Milan criteria before LT.[12] The 5-year recurrence free survival rate was only 52.7% for those with an AFP level >1000 ng/mL but 80.3% for all other patients. Applying an AFP cutoff >1000 ng/mL would have resulted in the exclusion of 5% of candidates from LT and a 20% reduction in the rate of posttransplant HCC recurrence. At our institution, patients with an AFP level >1000 ng/mL are now required to show a decrease in the AFP level to <500 ng/mL with LRT before they undergo LT.

Lai et al.[18] are the first to report down-staging to within the UCSF criteria as an indication for LT beyond the conventional Milan criteria. They have not provided additional information on the upper limits for tumor size and number in these patients before down-staging or the rate of successful down-staging in an intention-to-treat analysis. Nevertheless, successful tumor down-staging resulted in excellent posttransplant outcomes similar to those of patients meeting the Milan criteria after adjustments for the AFP slope and mRECIST progression after LRT. These same factors also significantly affected the prognosis for those with HCC meeting the Milan criteria. These findings therefore highlight the limitations of current selection criteria based on tumor size and number alone. Because patients in this study by Lai et al. fulfilled either the Milan or UCSF criteria (after down-staging) before LT, we should interpret AFP and mRECIST progression after LRT as possible ways for further refining, rather than replacing, tumor size and number in the selection of candidates for LT.

In summary, the intriguing data presented by Lai et al.[18] build nicely on the growing evidence that tumor size and number tell only a partial tale of the tumor characteristics that predict posttransplant HCC recurrence and that one size (or number) does not always fit all. Their study also provides valuable insight into an expanded role for LRT and AFP in the management of patients with HCC on the waiting list for LT. Observing tumor behavior after LRT, or to ablate and wait, is an important step in management.[14] The application of disease progression after LRT as an exclusion criterion for LT still requires prospective validation with vigorous radiographic assessments of the tumor diagnosis and response after LRT as well as reclassification of the tumor response after each LRT. Although there are shortcomings and uncertainties regarding tumor progression after LRT as an exclusion criterion for LT, the presence of tumor progression after LRT should nevertheless raise enough concerns to justify deferring (rather than excluding) LT in clinical practice until tumor control is achieved with additional LRT. The AFP slope is an interesting concept, but further refinements of its methodology are required for it to be used in assessing AFP progression. Further investigations are also needed to determine whether a progressive rise in AFP levels over time is better than static AFP levels in discriminating prognosis after LT.