Article first published online: 13 NOV 2008
Copyright © 2009 American Association for the Study of Liver Diseases
Volume 49, Issue 3, pages 832–838, March 2009
How to Cite
Toso, C., Asthana, S., Bigam, D. L., Shapiro, A. M. J. and Kneteman, N. M. (2009), Reassessing selection criteria prior to liver transplantation for hepatocellular carcinoma utilizing the scientific registry of transplant recipients database. Hepatology, 49: 832–838. doi: 10.1002/hep.22693
Potential conflict of interest: Nothing to report.
The data reported here were supplied by the Arbor Research Collaborative for Health as the contractor for the Scientific Registry of Transplant Recipients. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy of or interpretation by the Scientific Registry of Transplant Recipients or the US Government.
- Issue published online: 24 FEB 2009
- Article first published online: 13 NOV 2008
- Accepted manuscript online: 13 NOV 2008 12:00AM EST
- Manuscript Accepted: 13 OCT 2008
- Manuscript Received: 28 AUG 2008
- Swiss National Science Foundation
- FS Chia award
- Alberta Heritage Foundation for Medical Research (AHFMR)
- AHFMR scholar
- Canadian Institutes for Health Research/Wyeth Research Chair in Transplantation
The current model of liver graft allocation in place in the United States favors transplantation of patients with small hepatocellular carcinomas (HCCs) within the Milan criteria (a single tumor up to 5 cm in diameter or up to three lesions, none larger than 3 cm). Although several reports have suggested that these criteria could be extended, there is currently no agreement on new selection tools. In this study, we performed an overview of 6478 adult recipients of an isolated first liver transplant registered in the Scientific Registry of Transplant Recipients (SRTR) database. From March 2002 to January 2008, increasing numbers of patients outside Milan criteria (P ≤ 0.001) have been registered for a transplant, but they still represent less than 5% of the transplants performed for HCC. Of all the tested variables (tumor number, largest tumor size, and Milan and University of California San Francisco criteria), only total tumor volume (TTV; P ≤ 0.05) and alpha fetoprotein (AFP; P ≤ 0.001) could predict patient survival. While these two parameters demonstrated independent behaviors (no patient demonstrated an increase in both values), a composite score was defined, with patients with a TTV > 115 cm3 or an AFP > 400 ng/mL being outside criteria. The combined TTV/AFP score efficiently predicted posttransplant survival (hazard ratio = 2, 95% confidence interval = 1.7-2.4, P ≤ 0.001); patients not meeting these criteria had a survival below 50% at 3 years. Conclusion: According to the present SRTR data, Milan criteria are too restrictive, and patients with larger TTV can enjoy satisfactory posttransplant survivals. A composite patient selection score combining TTV and AFP was the most effective of all tested staging criteria for the prediction of posttransplant patient survival for candidates with HCC. (HEPATOLOGY 2009.)
Since the introduction of the Milan criteria in 1996, liver transplantation has progressively developed recognition as the best treatment option for patients with small unresectable hepatocellular carcinomas (HCCs).1 These criteria allow transplantation for patients with a single HCC up to 5 cm in diameter or up to three HCCs, none larger than 3 cm. Results in patient cohorts within the Milan criteria have now been reproduced by several centers,2–7 and these criteria have been adopted by the United Network for Organ Sharing for allocation of organs in the United States.
Although excellent results can be achieved when Milan criteria are respected, many recent data have suggested that they are too restrictive and that similar acceptable outcomes can be achieved with more liberal selection policies.8 The first well-structured evidence was reported by Yao et al.3 from the University of California San Francisco (UCSF), who demonstrated maintenance of good outcomes with the transplantation of patients with a single HCC up to 6.5 cm in diameter or with up to three HCCs, none larger than 4.5 cm, with a cumulative diameter up to 8 cm. Although their first study was based on explant tumor characteristics, the same group demonstrated more recently that similar excellent outcomes (93.6% of patients free of recurrence at 5 years) can be achieved on the basis of prospective radiological data.9
Despite validation by several other groups,4, 10 UCSF criteria have not been unanimously accepted yet. This may be related to the fact that a large multicenter study suggested decreased outcomes in patients between Milan and UCSF criteria.11 In addition, UCSF criteria allow transplantation for large HCCs (6.5 cm, representing a tumor volume of 144 cm3), which have been associated with decreased outcomes in several studies.12–14 UCSF criteria (like Milan criteria) also exclude all patients with more than three lesions, some of whom may have the potential to enjoy good outcomes if tumors remain at a reasonable size.5, 6, 8, 15
We have recently demonstrated in a three-center study including 288 patients that the total tumor volume (TTV) is the most accurate morphological criterion for selecting HCC patients prior to transplantation. Those with a TTV up to 115 cm3 demonstrated 79% 5-year survival, without tumor number restriction.15
The present large registry-based study was intended to perform an overview of the liver transplants reported in the United States and to assess morphological (tumor size, number, and TTV) and biological [alpha fetoprotein (AFP)] criteria that could help with the selection of the most appropriate transplant recipients from the population with HCC.
Patients and Methods
This study used data from the Scientific Registry of Transplant Recipients (SRTR). The SRTR data system includes data on all donors, wait-listed candidates, and transplant recipients in the United States submitted by the members of the Organ Procurement and Transplantation Network (OPTN), and it has been described elsewhere.16 The Health Resources and Services Administration (US Department of Health and Human Services) provides oversight of the activities of the OPTN and SRTR contractors. The study was reviewed and approved by the Health Research Ethics Board at the University of Alberta. The study population included all adult (≥18 years) patients who received a first isolated liver transplant from March 2002 to January 2008 for a diagnosis of HCC.
The following variables were used to predict patient survival: number of HCC, size of the largest tumor, Milan and UCSF criteria, TTV, and AFP. Data obtained on the date closest to transplant were used. The mean time between AFP measurement and transplant was 3 ± 2.7 months. Number and size variables were based on pretransplant radiological assessment and were available for all patients. The type of diagnosis modality used was reported for 5220 patients (80.5%). They included computed tomography (58.7%), magnetic resonance imaging (42.5%), biopsy (10.6%), ultrasound (8.3%), and/or angiogram (4.7%) performed 3.6 ± 3.7 months prior to transplant.
Milan criteria included patients with a single tumor up to 5 cm in diameter or up to three tumors, none larger than 3 cm.1 UCSF criteria included patients with a single tumor up to 6.5 cm in diameter or up to three tumors, none larger than 4.5 cm, with a cumulative diameter up to 8 cm.3 TTV was calculated as the sum of the volume of each tumor [(4/3)πr3] based on the maximum radius of each tumor.15
The occurrence and date of death were obtained from data reported to the SRTR by the transplanting centers and were completed by data from the US Social Security Administration and from the OPTN. All deaths were taken into account in the analysis, including those associated with HCC or not. Data related to HCC recurrence were not consistent (there was low granularity, and they were not previously validated) and were not included in the analysis. As a result, some patients may have been alive with an HCC recurrence and were not considered an event in the survival analyses.
A preliminary univariate analysis was performed with the Kaplan-Meier method, and groups were compared with the log-rank test. Cox regression was used to estimate covariate adjusted hazard ratios (HRs) of variables with a significance of at least 0.05 on the univariate analysis. Covariates included the Model for End-Stage Liver Disease (MELD), date of transplant, age at transplant, gender, race, primary underlying liver disease, and pretransplant tumor treatment (yes versus no). Statistical processing was performed in patients with a complete set of variables for each specific analysis.
Further analyses used a chi-square test to assess the ratio of patients who underwent transplantation outside Milan criteria according to the year of transplant. Results were provided as mean ± standard deviation. A standard alpha level of 0.05 indicated statistical significance. Analyses were conducted with SPSS 15.0 (SPSS, Chicago, IL).
During the study period, 6478 adult patients received an isolated first liver transplant and were registered in the SRTR database. Their characteristics are reported in Table 1. The population included mainly men (77%) and had a mean age of 56 ± 8 years. The most frequent underlying liver disease was related to hepatitis C virus infection (49.5%). Median follow-up (till death or last data report) was 13.4 (0-67.9) months.
|Mean age (years)||56 ± 8|
|Gender||Female: 1477/male: 5001|
|African American||531 (8)|
|Cause of liver disease (%)|
|HCV (±alcohol, ±HBV)||3221 (49.5)|
|Primary biliary cirrhosis||70 (1)|
|Primary sclerosing cholangitis||57 (1)|
|Alpha 1-antitrypsin deficiency||25 (0.5)|
|Metabolic disorder||15 (0.5)|
|Acute liver necrosis||228 (3.5)|
|Pretransplant HCC treatment (%)||2163 (33)|
|Outside Milan criteria (%)||205 (3.2)|
|Outside UCSF criteria (%)||51 (0.8)|
|Total tumor volume > 115 cm3 (%)||18 (0.3)|
|Mean serum alpha fetoprotein level (ng/mL)||262 ± 1473|
|Serum alpha fetoprotein level > 400 (%)||465 (8.9)|
Complete study variables were available for 80.4% of the patients. The most frequently missing variable was AFP, which was not reported for 1260 patients (19.5%).
The mean diameter of the tumors was 2.3 ± 1.1 cm (largest diameter: 8 cm). The mean number of tumors was 1.4 ±0.7 (largest number: nine tumors). Most of the patients were reported to fulfill Milan criteria (96.8%). However, an increasing number of patients outside Milan criteria underwent transplantation over time (P < 0.001; Fig. 1A). As expected, the TTV followed the same trend, and more patients with high cumulative tumor volumes underwent transplantation over time from 2003 to 2007 (Fig. 1B).
The number of tumors, size of the largest tumor, and Milan and UCSF criteria were not predictive of survival in the univariate analysis (Table 2). TTV was studied with a cutoff of 115 cm3, which was established in a previous study.15 This value was based on the risk of recurrence in our own recipient population at the University of Alberta and was obtained with the use of a receiver operating characteristic curve (area under the curve: 0.8). This cutoff was validated in a multicenter study assessing the risk of recurrence and survival.15 In the present SRTR data, TTV again was predictive of patient survival (P ≤ 0.05; Table 2), as was AFP (P ≤ 0.001).
|Three-Year Patient Survival (%)||P|
|Number of tumors|
|Total tumor volume|
|Alpha fetoprotein level|
The survival of patients within Milan criteria was similar to that of patients beyond Milan criteria but within UCSF criteria (P = 0.6) and those beyond UCSF criteria but with a TTV ≤ 115 cm3 (P = 0.6; Fig. 2). Only patients with a TTV > 115 cm3 demonstrated significantly lower survival (P ≤ 0.05).
When a multivariate analysis, corrected for MELD, date of transplant, age at transplant, gender, race, primary liver disease, and use of a pretransplant tumor treatment, was performed, both TTV (HR = 1.4 for TTV/100, P ≤ 0.05) and AFP (HR = 1.1 for AFP/1000, P ≤ 0.001) remained significant in predicting patient survival (Table 3).
|Total tumor volume/100||1.4||1.03–2||≤0.05|
|Alpha fetoprotein level/1000||1.1||1.1–1.19||≤0.001|
After plotting TTV and AFP values, we could identify three groups of patients: those with both a low TTV and a low AFP and those with either a high TTV or a high AFP (Fig. 3A). None of the reported patients had both a high TTV and a high AFP. We therefore performed the subsequent analysis using a combined TTV/AFP score, defining patients with a TTV > 115 cm3 or an AFP > 400 ng/mL as outside criteria. The TTV cutoff was reported from our previous report and corresponds to a single tumor with a maximum diameter of approximately 6 cm or three lesions with a maximum diameter of approximately 4.2 cm.15 The AFP cutoff of 400 ng/mL was selected from several published data7, 17, 18 and from our own series (the receiver operating characteristic curve area under the curve was 0.7 when we looked at recurrence).15 Among patients within Milan criteria, 8.8% were over the AFP cutoff. The TTV/AFP score was most accurate for predicting patient survival (HR = 2, 95% confidence interval = 1.7-2.4, P ≤ 0.001). Patients with a high TTV or a high AFP had a corrected 3-year survival lower than 50% (Fig. 3B). Four hundred seventy-nine recipients (7.4%) were included in this group.
According to the present SRTR data, Milan criteria are too restrictive, and patients with larger TTV can enjoy satisfactory posttransplant survivals. A combined patient selection score based on TTV and AFP was the most predictive of all tested criteria sets when applied to the SRTR data.
Over 95% of the liver transplants performed for HCC and registered in the SRTR database were done for patients fulfilling Milan criteria. We observed, however, a significant increase over time in the rate of recipients with large TTV exceeding Milan criteria. This observation may reflect a true increase in the number of such transplants and/or more accurate reporting of them by the centers. Both explanations can be viewed as an overall agreement that Milan criteria are too restrictive and that liver transplants can be performed safely in selected patients with larger and/or more numerous HCCs. This point of view is shared by many reports.3–6, 8–10, 15 It is also supported by the present data, as patients with a TTV up to 115 cm3 demonstrated survival similar to that of those respecting Milan or UCSF criteria (Fig. 2).
In the present registry-based study, both Milan and UCSF criteria failed to predict patient survival. This observation needs to be considered with caution, as tumor size and number were entered by the centers in order to obtain exception MELD points, which may introduce reporting bias. That said, although Milan, UCSF, and TTV were all based on the same reported numbers, only TTV remained powerful in selecting patients for transplant. This suggests the superiority of this score and supports our previously reported observations.15 Of note, this was achieved despite the proportionally low number of patients with large TTV who underwent transplantation.
The main advantage of TTV is that it does not include a rigid tumor number limit, unlike Milan and UCSF criteria. Such a limit prevents the transplantation of patients with more than three lesions, even if they are of very small size and would be associated with good posttransplant outcome.5, 6, 8, 15 In addition, the accuracy of the pretransplant radiological assessment is substantially improved by the use of TTV in comparison with Milan or UCSF, and thus better patient selection is achieved with pretransplant imaging studies.15 This is related to the fact that TTV gives more power to larger tumors as the calculated volume increases much faster than the related diameter [(4/3)πr3], and the sensitivity and specificity of radiology are higher for larger tumors.19
AFP was the other variable highly predictive of patient survival. This observation confirms several previous studies.7, 12, 20, 21 An AFP value was available for more than 80% of the studied transplants, reflecting the good granularity of the database (80.4% of cases had complete study variables). In addition, as AFP does not enter into patients' selection for transplantation at the present time, the registered data were likely very accurate.
Because of the lack of accurate data on HCC recurrence or on the presence of HCC at the time of death, we could not determine cutoff values for TTV or AFP from the present registry. We therefore used cutoffs of 115 cm3 for TTV and 400 ng/mL for AFP, which were determined from our own patient population with receiver operating characteristic curves based on the risk of HCC recurrence.15 The areas under the curve were 0.8 for TTV and 0.7 for AFP.15 The same AFP cutoff has also been selected by several other groups.7, 17, 18
A combination of TTV and AFP appears logical when we look at their distribution in Fig. 3. There is no linear correlation between them in the present population of transplant recipients, and this suggests that they behave independently. The absence of reported cases with both a high tumor volume and a high AFP may be linked to the fact that these patients have particularly aggressive tumors that are rapidly progressive, preventing listing and/or transplantation. Although TTV provides a morphological limit to HCC selection, AFP brings a simple assessment of tumor biology. The combined score would thus exclude patients with large HCC or with small HCC with potentially aggressive behavior, who have a corrected patient survival lower than 50% at 3 years.
Several potential variables were not assessed by the present study. Macrovascular tumor invasion has been recognized by several previous studies as being associated with poor posttransplant outcome.12, 15 The TTV/AFP selection score should therefore be applied only to patients without sign of vascular invasion by HCC on radiology. Biopsy-based factors, such as tumor grade, would also be of interest,3, 7, 15, 22 but they have several negative points that challenge their applicability. An HCC biopsy bears a risk of tumor seeding. In addition, large HCCs are not homogeneous, and the risk of needle biopsy sampling error is real. Finally, these risks would be multiplied in the case of multiple lesions, rendering the practicability of the biopsy difficult (patients with up to nine lesions were included in the present study).
It is difficult to estimate what impact the TTV/AFP score would have on the number of patients recruited for transplantation because the studied registry population was highly preselected. As demonstrated in a previous study, TTV (cutoff: 115 cm3) alone would increase the number of candidates by 28% to 53% in comparison with Milan criteria and by 16% to 26% in comparison with UCSF criteria.15 On the other hand, AFP (cutoff: 400 ng/mL) would have excluded 8.8% of the patients meeting Milan criteria in the present study. We can therefore speculate that the combined score would lead to a marginal net increase in the number of candidates.
Such an increase in the requirement of liver grafts would certainly represent a major challenge because of the limited organ resources.23 On the other hand, not offering a transplant to patients who have the potential to have good outcomes is ethically disturbing.
Taken together, the presented registry-based data suggest that Milan criteria are too restrictive and that a combined patient selection score based on TTV and AFP would be more appropriate for patient selection of liver transplantation in the presence of HCC.
- 18The value of serum alpha-fetoprotein in predicting tumor recurrence after liver transplantation for hepatocellular carcinoma. Dig Dis Sci. doi:10.1007/s10620-008-0349-0., , , , , , et al.