The adoption of the restrictive Milan criteria [a solitary hepatocellular carcinoma (HCC) tumor up to 5 cm in diameter or as many as 3 HCC nodules with a maximum diameter of 3 cm without macrovascular infiltration] has significantly improved the prognosis of patients with HCC after liver transplantation (LT).1, 2 The number of LT procedures performed for HCC has been steadily increasing. In recent years, however, concerns have been raised: the Milan criteria may be too restrictive, and many patients with advanced HCC who have an acceptable prognosis are being excluded from curative LT.3, 4 This has prompted several proposals for more liberal selection criteria, and most are based on macromorphological imaging results.5-7 There is increasing evidence that a tumor's biology, rather than its macromorphology, is associated with outcomes after LT for patients with advanced HCC. However, biological features of tumor aggressiveness, including a poor tumor grade and microscopic vascular invasion (MVI), are not routinely available before transplantation and may change in the context of interventional bridging therapies.8, 9
The uptake of [18F]fludeoxyglucose ([18F]FDG) on positron emission tomography (PET) scans, which is based on enhanced glucose metabolism in cancer cells, is a sensitive and noninvasive biological marker for assessing metabolic tumor viability. [18F]FDG PET is nowadays a well-established diagnostic tool for the evaluation and treatment monitoring of a variety of malignant tumors.10, 11 It is less suitable for the primary detection of HCC but has some prognostic value after liver resection.12-14 Recently, it has been demonstrated that enhanced preoperative [18F]FDG uptake correlates with histological parameters of tumor aggressiveness in the transplant setting.15-17 There are, however, no data about the value of pretransplant PET for identifying appropriate LT candidates with advanced HCC on clinical staging.
We, therefore, conducted a retrospective analysis to evaluate the prognostic significance of enhanced glucose metabolism on PET scans in a series of LT candidates with HCC beyond the Milan criteria on medical imaging.
PATIENTS AND METHODS
From January 1995 to December 2008, 111 patients with HCC were listed for LT at our center (Table 1). Data were collected in a prospective database and retrospectively analyzed.
Table 1. Characteristics of All Patients Listed for LT (n = 111)
Pretransplant clinical staging included preoperative and intraoperative ultrasonography, dynamic computed tomography (CT), and contrast-enhanced magnetic resonance tomography. Furthermore, whole-body [18F]FDG PET was performed for all patients at least once before LT. We did not routinely perform preoperative tumor biopsy.
At the time of listing, the tumors of all study patients met the Milan criteria according to radiographic staging.1 Pretransplant clinical tumor progression (with respect to the size and number of HCC tumors on radiography) did not result in patient dropout per protocol. Patients were removed from the waiting list for relevant biological tumor progression (ie, tumor invasion into a major vascular branch, lymph node metastases, extrahepatic tumor spread, or tumor-related symptoms).
Alpha-fetoprotein (AFP) measurements and abdominal ultrasound examinations were performed every 3 months, whereas cross-sectional imaging was performed at least every 6 months and in the case of significantly increased AFP levels.
Since 1999, LT candidates underwent transarterial chemoembolization (TACE) or percutaneous radiofrequency ablation (RFA) for their tumors if they were eligible for technical and/or functional reasons. Surgical resection was not considered standard bridging therapy before LT. The response to neoadjuvant therapy was determined with contrast CT scanning. The viable tumor volume was defined by contrast agent uptake in the arterial phase and washout in the portal venous and late venous phases.18, 19 Only the viable tumor volume was used for macromorphological staging.
All staging results were based on pretransplant medical imaging (not explant histopathology).
The immunosuppressive therapy consisted of a calcineurin inhibitor–based induction regimen that was augmented by mycophenolate mofetil or azathioprine. Except for patients with autoimmune hepatitis, steroids were tapered off at most 6 months after LT.
The posttransplant tumor surveillance consisted of AFP measurements and abdominal ultrasound examinations every 3 months and CT scans of the chest and abdomen every 6 months.
All explanted livers were histopathologically examined by 2 experienced pathologists who determined the size, number, total tumor diameter, stage, and vascular invasion status. Tumor differentiation was determined according to Edmondson and Steiner's grading system. Surgical and pathology staff cooperated on histopathological tumor staging, which was based on clinical and pathological data and was determined according to the 1997 edition of the tumor-node-metastasis criteria of the International Union Against Cancer.20 Furthermore, tumors were classified according to the Milan and University of California San Francisco (UCSF) criteria.1, 7
Pretransplant [18F]FDG PET was performed with an ECAT EXACT 47 scanner (Siemens, Erlangen, Germany); the whole-body mode was implemented as standard software. Patients fasted for at least 8 hours before PET scanning. A static emission scan was performed 60 to 90 minutes after the injection of approximately 360 MBq of [18F]FDG from the neck to the knee level in a 2-dimensional mode. The PET scans were correlated with the corresponding CT scans for accurate tumor localization. In recent years, an integrated PET/CT scan was performed. Two specialists in nuclear medicine evaluated the coronal, sagittal, and axial PET images. The [18F]FDG tumor uptake was assessed semiquantitatively through the determination of the tumor/background (T/B) ratio, as previously described.14-17 A T/B ratio > 1 indicated enhanced [18F]FDG uptake in the tumor ([18F]FDG-avid or PET+; Fig. 1). In contrast, a T/B ratio of 1 identified HCCs without suspicious glucose metabolism on a PET scan (non–[18F]FDG-avid or PET−) in comparison with surrounding noncancerous hepatic tissue.
Pretransplant Prognostic Variables
In order to identify pretransplant prognostic parameters, the following variables were included in univariate and multivariate analyses: donor and recipient ages, sex, AFP levels, number and size of tumor nodules on final pretransplant radiographic staging, responses to interventional bridging therapies on CT scans, Milan/UCSF status according to the final pre-LT clinical staging, and PET status for HCC.
All analyses were performed with the statistical software SPSS 17.0 for Windows. Correlations of clinical and histopathological variables with frequencies of tumor recurrence and tumor-related dropout from the waiting list were analyzed with χ2 tests and logistic regression. The analysis of recurrence-free survival was performed according to the Kaplan-Meier method. Patient survival rates in different groups were compared with the log-rank test. Variables with a significant prognostic impact in a univariate analysis (P < 0.05) were entered into a stepwise multivariate analysis that used a Cox multiple stepwise regression model.
[18F]FDG, [18F]fludeoxyglucose; AFP, alpha-fetoprotein; CT, computed tomography; HCC, hepatocellular carcinoma; LT, liver transplantation; MVI, microscopic vascular invasion; PET, positron emission tomography; RFA, radiofrequency ablation; TACE, transarterial chemoembolization; T/B, tumor/background; UCSF, University of California San Francisco.
Patient and Tumor Characteristics
Finally, 91 patients underwent LT. There were 60 male patients and 31 female patients. The mean recipient age was 58.3 ± 7 years at LT. Seventy-eight patients underwent LT with deceased donors, whereas 13 patients underwent LT with living donors. Both procedures were performed with standard techniques as previously described.21
According to the final preoperative clinical staging, 57 patients satisfied the Milan criteria (62.6%), whereas the tumor burden exceeded the Milan criteria in 34 patients (37.4%; Table 2). In 27 patients (29.7%), the tumors were determined to be beyond the UCSF criteria (a tumor lesion ≤ 6.5 cm or 2-3 lesions ≤ 4.5 cm with a total tumor diameter ≤ 8 cm and an absence of gross portal vein invasion) at LT.7
Table 2. Clinical Parameters According to Pretransplant Staging and Histopathological Parameters at Explantation for Liver Recipients (n = 91)
Patients Within the Milan Criteria
Patients Outside the Milan Criteria
Total patients (n)
AFP level (n)
Pretransplant treatment (n)
Mean number of tumor nodules
Mean maximum diameter of largest tumor nodule (cm)
Total tumor diameter (cm)
Response to bridging therapy
PET staging before LT
Patients Within the Milan Criteria
Patients Outside the Milan Criteria
Total patients (n)
Major vascular invasion
Lymphatic vascular invasion
As expected, patients within the Milan criteria had fewer HCCs, smaller dominant HCC nodules, and lower total tumor diameters on pretransplant clinical staging (P < 0.05; Table 2). In addition, in comparison with patients within the Milan criteria (15/57 or 26.3%), more patients with tumors outside the Milan criteria (20/34 or 58.8%) demonstrated enhanced glucose metabolism on pretransplant PET scans (P = 0.002; Table 2).
Responses to interventional bridging therapies were noted in 27 patients within the Milan criteria (73%) and in 8 recipients outside the Milan criteria (40%, P = 0.015; Table 2).
According to explant histopathology, there were more tumors with MVI in recipients outside the Milan criteria versus patients within the Milan criteria (P < 0.001). Furthermore, we noticed a trend of more high-grade HCCs in patients outside the Milan criteria (P = 0.141; Table 2).
PET Characteristics and Parameters of Tumor Biology
[18F]FDG PET scans were performed for all patients 1.5 to 15 months (median = 6 months) before LT and before the initiation of interventional bridging therapies.
Thirty-five liver recipients had positive preoperative PET scans (38.5%; Fig. 1), whereas the pretransplant PET findings were negative for 56 LT patients (61.5%).
The PET findings were associated with several parameters of pretransplant tumor staging, including the Milan/UCSF status on radiography, responses to bridging therapies on CT scans, and (by tendency) AFP levels at LT (Table 3). Furthermore, a PET+ status was significantly associated with explant histopathology parameters of aggressive tumor behavior, including MVI, grading, and lymphovascular invasion (P < 0.05; Table 3). Only 2 of 56 PET− patients (3.6%) developed posttransplant tumor recurrence, but 19 of 35 PET+ recipients (54.3%) did (P < 0.001; Table 3).
Table 3. PET Status and Tumor Characteristics of Liver Recipients (n = 91)
The mean posttransplant follow-up was 65.1 ± 44.9 months (range = 5-165 months). The overall and 5-year recurrence-free survival rates for the entire study population were 75.5% and 72.7%, respectively.
The 5-year recurrence-free survival rate was 86.2% for patients within the Milan criteria and 47.4% for patients outside the Milan criteria (P < 0.001).
There was no survival difference between patients receiving neoadjuvant interventions and those not receiving pre-LT treatments (76% versus 68%, P = 0.31). However, the outcomes were significantly better for progression-free responders (93%) versus nonresponders to interventional bridging therapy (49%, P = 0.004).
PET− patients outside the Milan criteria had a 5-year recurrence-free survival rate (81%) that was comparable to the rate for recipients within the Milan criteria (86.2%), but it was significantly higher than the rate for PET+ patients outside the Milan criteria (21%, P = 0.002; Fig. 2).
There was no significant difference in the 5-year recurrence-free survival rates for patients meeting the UCSF criteria [85% (n = 64)] and PET− patients exceeding the UCSF criteria [85.7% (n = 10)], whereas the outcomes were significantly worse for patients with [18F]FDG-avid HCC beyond the UCSF criteria [19.2% (n = 17), P = 0.001; Fig. 3].
In the univariate analysis, an AFP level <400 IU/mL, fewer than 3 tumor nodules, a dominant nodule diameter <5 cm, a total tumor diameter <10 cm, a tumor stage meeting the Milan criteria, and negative PET findings were associated with posttransplant recurrence-free survival (P < 0.05; Table 4). In the multivariate analysis, a PET− status, an AFP level <400 IU/mL, and a total tumor diameter <10 cm were identified as independent predictors of recurrence-free survival (P < 0.05; Table 4). Notably, the parameters defining the Milan criteria were not significantly associated with outcomes in the multivariate analysis.
Table 4. Univariate and Multivariate Analyses of Pretransplant Predictive Parameters for Liver Recipients (n = 91)
Log-Rank P Value
AFP level ≤ 400 U/L at LT
Number of tumor nodules ≤ 3
Tumor diameter of dominant nodule ≤ 5 cm
Total tumor diameter ≤ 10 cm
Within the Milan criteria
95% Confidence Interval
AFP level ≤ 400 U/L at LT
Total tumor diameter ≤ 10 cm
When neoadjuvant therapy was included in the multivariate analysis (n = 57), a PET+ status remained the only independent variable for tumor recurrence.
PET Status and Risk of Patient Dropout
Twenty LT candidates were dropped from the waiting list (18%): 13 for relevant tumor progression and 7 for other complications. The tumor-related dropout rate differed significantly between patients with [18F]FDG-avid HCC (10/46 or 21.7%) and patients with non–[18F]FDG-avid HCC (3/65 or 4.6%, P = 0.006). Among clinical parameters, only a PET+ status (hazard ratio = 5.7, confidence interval = 1.5-22.2, P = 0.01) was identified as an independent predictor of tumor-related patient dropout from the waiting list (Table 5).
Table 5. Univariate and Multivariate Analyses of Pretransplant Parameters Predicting Tumor-Related Dropout for All Listed Patients (n = 111)
AFP level > 400 U/L at LT
Number of tumor nodules > 3
Tumor diameter of dominant nodule > 5 cm
Total tumor diameter > 10 cm
Outside the Milan criteria
95% Confidence Interval
To our knowledge, we are presenting the largest study conducted so far for the evaluation of the prognostic significance of [18F]FDG PET in the LT setting. We have demonstrated that patients with non–[18F]FDG-avid HCC exceeding the Milan or UCSF criteria on pretransplant clinical staging may achieve an extraordinary 5-year recurrence-free survival rate greater than 80% after LT. A notable observation of our trial is that radiographic tumor progression beyond the Milan criteria while a patient is waiting for LT does not need to instantly result in dropout from the waiting list. Furthermore, we have been able to show that a PET+ status is an independent predictor of tumor-related dropout from the waiting list; this is another important result of our study.
[18F]FDG is a glucose analogue that competes with glucose at transport sites on the cell membrane and in a variety of intracellular enzymatic pathways. Because of different enzymatic activities, well-differentiated HCCs may resemble normal liver tissue, whereas poorly differentiated HCCs may be visualized as hot spots on PET scans.10-13, 22 The sensitivity of [18F]FDG PET is, therefore, low for the primary detection of HCC in comparison with many other cancers.12, 13, 22 In recent years, it has been increasingly used for the evaluation of biological tumor behavior in the setting of conventional surgery and palliative treatment.13, 14, 23 Apart from that, PET seems to have some value for the detection of extrahepatic metastases and posttransplant HCC recurrence.24, 25
Yang et al.15 from the Seoul transplant center were the first to propose PET imaging as a useful preoperative tool for estimating the risk of posttransplant tumor recurrence in patients with HCC. Recently, they were able to confirm these early data with quantitative fludeoxyglucose uptake measurements.17 However, in comparison with our study, the overall number of patients in this follow-up trial was considerably smaller (n = 59), and the mean posttransplant follow-up was significantly shorter (29 ± 17 months). Furthermore, we have analyzed the prognostic value of PET in the context of pretransplant clinical staging and not in the context of parameters available after transplantation (eg, histopathological features).
The results of our trial have to be discussed with caution. On the one hand, they confirm the importance of a restrictive selection process based on macromorphological tumor staging.1, 2 The outcomes were excellent for patients within the Milan criteria and for patients within the UCSF criteria in our study (Figs. 2 and 3). However, the adoption of the Milan/UCSF criteria was recently shown to not completely eliminate the risk of aggressive tumor features and, therefore, to increase the risk of posttransplant tumor recurrence.1, 26 Among our patients within the Milan criteria (n = 57), none of the 42 PET− patients (0%) and 7 of the 15 PET+ recipients (46.7%, P < 0.001) developed post-LT HCC recurrence. This interesting result illustrates the prognostic power of PET for identifying early-stage HCC with aggressive biological behavior. However, whether enhanced glucose metabolism on PET scans justifies patient dropout (even though standard selection criteria are fulfilled) remains a matter of critical debate. This controversial issue has to be further evaluated with a larger number of patients.
On the other hand, the results of our study clearly demonstrate that a stringent selection policy will significantly increase the risk of excluding appropriate LT candidates with advanced HCC. Approximately 20% of our study patients would have been excluded from a curative treatment if macromorphological tumor progression beyond the Milan criteria had been a dropout criterion by protocol in our series. Furthermore, the inaccuracy of clinical staging remains a critical issue in this context.26, 27 In our series, 33% of the patients were understaged by final pretransplant medical imaging. Other centers have reported similar findings; clinical imaging alone inaccurately stages 30% to 40% of patients undergoing LT for HCC.26-28 This may be a specific problem for patients who are presenting with borderline tumor stages, especially when both listing and dropout are associated with the Milan criteria.26 Hence, our waiting-list policy is mainly based on the tumor biology rather than the static macromorphology of HCC. Similarly to other major centers,29 we have adopted the Milan criteria for patient listing, but patient dropout is related to relevant biological tumor progression, including extrahepatic tumor spread, invasion into a major vascular branch, and tumor-related symptoms. It seems obvious that the number of patients benefiting from LT will rise in this context; however, the price is an increasing risk of tumor recurrence in others. Thus, the responsible application of such an innovative waiting-list policy urgently requires the implementation of powerful clinical parameters of tumor biology.
The current study provides some good evidence for the ability of PET to describe the biological activity of HCC in the transplant setting. First, the predictive value of a PET− status for excluding aggressive histopathological features was very high in our series (Table 3). It has been recently shown that the posttransplant outcomes of patients with advanced HCC may be significantly improved by the exclusion of patients with poor tumor grades, MVI, and high DNA.20-34 However, these data are not routinely available before LT because preoperative fine-needle biopsy results have been shown to poorly correlate with explant pathology results.35 Furthermore, tumor sampling errors and the theoretical risk of tumor spread are additional critical issues of preoperative tumor biopsy.35 In contrast, PET scanning is a noninvasive tool available before transplantation without dangerous adverse effects. Second, the rate of progression-free interventional therapy was significantly higher for patients with non–[18F]FDG-avid HCC (75.8%) versus patients with PET+ tumors (29.4%; Table 3). Because pretransplant tumor down-staging by bridging treatments has been shown to describe a favorable biology for advanced HCC,37-41 this interesting result seems to be another indication of the prognostic value of PET in the transplant setting. Furthermore, we have identified the PET status as the only pretransplant parameter able to independently predict tumor-related patient dropout from the waiting list. To our knowledge, this interesting correlation has not been described before. It has been recently demonstrated that enhanced glucose metabolism on PET scans is closely related to genetic markers of proliferative activity in HCC, and this may be a relevant trigger for tumor-related dropout in patients with [18F]FDG-avid HCC.42, 43
There are some limitations to our study. First, the trial was nonrandomized and retrospective. Second, the prognostic significance of a single pretransplant PET scan in the context of different waiting times and locoregional therapies before LT could be limited. Successful interventional therapies probably alter the glucose metabolism on PET scans. However, whether the conversion from a PET+ status to a PET− status in the context of interventional bridging treatments indicates a favorable tumor biology and can, therefore, prevent patient dropout cannot be answered by our data because we did not perform repeat PET evaluations by protocol. In order to address this relevant issue, a prospective trial implementing follow-up PET scans in the context of locoregional therapies seems to be mandatory. Furthermore, the PET results of our study may not be representative of the results for other transplant centers. As mentioned previously, tumor-related patient dropout was mainly based on major contraindications, and this may not be feasible for other institutions. However, the application of a waiting-list policy that is more restrictive than ours might limit the prognostic value of pretransplant PET assessments. This has to be further evaluated. Moreover, the sample size might be too small for identifying relevant trends in survival. In contrast, our trial is powered by a long and concise posttransplant follow-up period and the inclusion of a large number of patients with advanced HCC.
In conclusion, we have demonstrated that [18F]FDG PET is a useful metabolic imaging tool for the evaluation of biological tumor behavior in LT candidates with advanced HCC. Patients with non–[18F]FDG-avid HCC exceeding the Milan criteria or even the UCSF criteria on radiography may achieve excellent long-term recurrence-free survival after LT. LT candidates with [18F]FDG-avid HCC should be carefully re-evaluated during the waiting period for LT, and an aggressive down-staging treatment policy should be considered if it is possible. However, whether neoadjuvant bridging therapies can beneficially alter the glucose metabolism on PET scans and, therefore, prevent patient dropout from the waiting list has to be assessed in the future.