Vascular invasion of hepatocellular carcinoma (HCC) is a major risk factor for poor outcome after liver transplantation (LT). The aim of this retrospective analysis was to assess the value of preoperative positron emission tomography (PET) using 18F-fluorodeoxyglucose (18F-FDG) in liver transplant candidates with HCC for predicting microvascular tumor invasion (MVI) and posttransplant tumor recurrence.
Forty-two patients underwent LT for HCC after PET evaluation. Sixteen patients had an increased 18F-FDG tumor uptake on preoperative PET scans (PET +), while 26 recipients revealed negative PET findings (PET−) pre-LT. PET− recipients demonstrated a significantly better 3-year recurrence-free survival (93%) than PET + patients (35%, p < 0.001). HCC recurrence rate was 50% in the PET + group, and 3.8% in the PET—population (p < 0.001). PET + status was identified as independent predictor of MVI [hazard ratio: 13.4]. Patients with advanced PET negative tumors and patients with HCC meeting the Milan criteria had a comparable 3-year-recurrence-free survival (80% vs. 94%, p = 0.6).
Increased 18F-FDG uptake on PET is predictive for MVI and tumor recurrence after LT for HCC. Its application may identify eligible liver transplant candidates with tumors beyond the Milan criteria.
Hepatocellular carcinoma (HCC) in liver cirrhosis has become a major indication for liver transplantation (LT) in the last two decades (1). Early results of LT for HCC were associated with recurrence rates from 60% to 70%, and 5-year survival rates of less than 30% (2–4). More recently, transplant groups from Milan and Barcelona reported about a 4-year survival rate of 75% after LT in HCC patients. These extraordinary outcome data were a result of careful selection of patients with specific morphological tumor criteria. Only patients with HCC in cirrhosis who had a maximum of three tumor nodules that were ≤3 cm in maximum diameter, or a single tumor with a maximum diameter of 5 cm, both without macrovascular tumor invasion, were considered as being eligible for LT (5,6). The adoption of the so-called Milan criteria resulted in survival rates after LT that are similar to those after resection in noncirrhotic lesions with comparable tumor stage (1).
In recent years there is, however, an increasing concern that strict allocation of liver allografts based on these new morphological criteria may exclude a significant number of patients with HCC from potentially curative LT (7–10). Several groups have shown that parameters of tumor biology, such as tumor grade and vascular tumor invasion, may significantly influence outcome after LT (11–16). An accurate pretransplant prediction of these histopathological tumor features might lead to a reasonable expansion of the selection criteria, with regard to acceptable size and numbers of HCC nodules (17–20).
Specifically, microvascular tumor invasion (MVI) has recently been demonstrated to be a very strong predictor of tumor recurrence and poor survival after LT and liver resection for HCC (12,14,21,22). While gross vascular tumor invasion can be diagnosed pre-LT by high resolution imaging techniques (23), MVI cannot be determined preoperatively, since it is a histopathological parameter. It is, therefore, of great importance to identify clinical surrogate markers that are able to reliably indicate the presence of MVI already prior to LT.
Positron emission tomography (PET) using 18F-fluorodeoxyglucose (18F-FDG) has become a standard procedure within oncology during recent years. It is an important prognostic marker in various cancers, such as nonsmall lung cancer, pancreatic adenocarcinoma and neck cancers (24,25). Moreover, PET imaging has been established as a useful diagnostic tool for evaluating metastatic liver tumors (26,27). Preoperative 18F-FDG-PET has recently been shown to predict tumor differentiation and outcome after liver resection in patients with HCC (28,29). There is, however, only limited data about the value of 18F-FDG-PET as prognostic marker of biological behavior of HCC in the transplantation setting.
PET assessment has been implemented in our evaluation program for liver transplant candidates with liver malignancy in 1999.
We performed a retrospective analysis to determine the prognostic value of preoperative 18F-FDG-PET in liver transplant candidates with HCC, with special focus on its power for predicting aggressive tumor biology, such as MVI and posttransplant tumor recurrence.
Patients and Methods
Between April 1999 and December 2007, a total of 42 patients with HCC in liver cirrhosis underwent LT at our center. There were 35 male and 7 female recipients. Patients’ age was ranging between 38 and 68 years (median: 61 years).
Thirty-one patients received a full size transplant, while 11 patients underwent living-donor liver transplantation (n = 10) or received a left split liver graft (n = 1) (Table 1).
Table 1. Baseline characteristics of the study population (n = 42)
Pretransplant diagnosis of HCC was based on two abdominal imaging studies, including ultrasound and computed tomography (CT) or magnetic resonance imaging (MRI), showing consistent results. The course of ∝-fetoprotein-levels (AFP) was used for supporting suspected HCC diagnosis. AFP-levels within normal range, however, did not rule out the diagnosis. We did not perform preoperative tumor biopsy by protocol. Our inclusion criteria for LT in patients with unresectable HCC in liver cirrhosis was based on the Milan criteria, consisting of a 5 cm maximum tumor size for one single HCC nodule, or a maximum of three tumors each up to 3 cm, both without clinical evidence of vascular tumor invasion (5). Since January 1999, we used transarterial chemoembolization (TACE) as interventional bridging to LT in patients with HCC (30). Clinically relevant portal hypertension, as indicated by treatment-refractory ascites, low albumin level (<25 g/L), increased bilirubin level (>3 mg/dL) and severe coagulopathy, was a contraindication for TACE. If chemoembolization could not be realized due to technical problems, the possibility of percutaneous radiofrequency ablation of the tumor was investigated in recent years (30). Failure of HCC downstaging, however, was not used as a biological exclusion criterion for LT, as has been proposed by others (31,32).
Since January 1999, all liver transplant candidates with an intrahepatic malignancy underwent whole-body 18F-FDG-PET scanning during evaluation for LT. Currently, a total of 42 patients with HCC underwent LT after preoperative PET assessment.
PET was performed using an EACT EXACT 47 (Siemens, Erlangen, Germany) with the whole-body mode implemented as standard software. Patients were fasted for at least 8 h before undergoing PET and the plasma glucose concentration was measured before scanning. Static emission scanning was performed 60–90 min after injection of approximately 360 MBq 18F-FDG from the neck to the knee level in 2D mode. Images were reconstructed using an iterative reconstruction algorithm. The PET scans were compared with the corresponding CT or MRI for accurate tumor localization. The coronal, sagittal and axial images were evaluated visually to assess whether the 18F-FDG uptake in the tumor was (PET + status) or was not (PET—status) significantly higher (Figure 1) than in the surrounding noncancerous hepatic tissue (33). If PET identified an extrahepatic region of increased 18F-FDG uptake, HCC metastases or other malignant tissue had to be excluded by additional imaging studies or tissue biopsy.
The number of tumors, maximal tumor diameter and their total diameter in patients with multiple tumor nodules were determined. The histological grade according to the modified Edmondson criteria was assessed, with the final classification based on the area showing the poorest differentiation (34). MVI was defined as the presence of tumor emboli within the central vein, the portal vein or large capsular veins or involvement of the lobar or segmental branches of the portal vein or the hepatic veins (3). Pathological tumor staging was assigned by co-operation of the surgical and pathology staff based on clinical and pathological data according to the 5th edition of the Tumor, Node, Metastasis/International Union Against Cancer criteria of 1997 (36). In addition, tumors were classified as ‘Milan In’ versus ‘Milan Out’ according to the pathomorphological staging on explant livers.
Posttransplant management and follow-up surveillance
Immunosuppressive therapy after LT consisted of a quadruple induction regimen, cyclosporin A- or tacrolimus-based, augmented with mycophenolat mofetil or azathioprine, and prednisone (Table 1). Steroids were completely withdrawn 6 months from LT the latest, except for patients presenting with autoimmune hepatitis.
Our posttransplant tumor surveillance program consisted of abdominal ultrasound and AFP level measurement every 3 months during the entire follow-up period. In addition, CT scans (or MRI) of the chest and abdomen were performed every 6 months for the first posttransplant year, and minimum once yearly thereafter. Increasing AFP-levels alone without radiographic evidence of new tumor masses did not indicate HCC recurrence until becoming manifest in imaging studies. Patients with rising AFP levels and inconclusive imaging studies underwent additional whole body 18F-FDG-PET scanning.
Statistical analysis was performed using SPSS version 12.0 for Windows (SPSS Inc., Chicago, IL).
Comparison of categorial and continuous variables was performed using chi-square test and Mann–Whitney U-test, respectively. All data were reported as median with range, mean ± SD, as appropriate. Overall and recurrence-free survival was analyzed by the Kaplan–Meier method. Patient survival in different groups was compared using the log-rank test. Parameters being predictive for the presence of MVI and posttransplant HCC recurrence were assessed in a univariate analysis. All variables with a p-value less than 0.05 were then included in a multivariate analysis applying the Cox multiple backward stepwise model to identify parameters being independently predictive for MVI (p < 0.05).
PET scans were performed between 2 and 15.5 months (median: 6.75 months) pretransplantation and before initiating interventional therapy, where feasible. In one patient of this series (2.4%), a relevant extrahepatic 18F-FDG uptake in the submandibular region was assessed in preoperative PET imaging. Surgical resection of the affected tissue yielded an unspecific inflammatory lymph node specimen.
18F-FDG-PET findings and associated tumor factors
Tumor characteristics of the study group according to explant histopathology are summarized in Table 2.
Sixteen patients had positive preoperative PET scans (38.1%), while 26 recipients (61.9%) had no increased 18F-FDG uptake. Of 6 patients with poor tumor grade, 5 (83.3%) demonstrated a PET + status and of 17 recipients with MVI, 14 patients (82.3%) had a PET + finding. Moreover, PET + status was significantly associated with size, number and stage of tumors, as well as the pathomorphological Milan criteria (Table 3).
Table 3. Association between PET status and tumor variables
PET− (n = 26)
PET + (n = 16)
*By chi-square test.
Pre-LT AFP level
Number of tumor nodules
Size of largest tumor nodule
Overall tumor diameter
1 or 2
3 or 4
Milan status (pathomorphological)
18F-FDG-PET findings and risk of HCC recurrence
Current overall posttransplantation follow-up ranges from 4 to 108 months (median: 26 months). Calculated overall and recurrence-free 1- and 3-year survival rates were 93% and 79%, as well as 84.5% and 73%, respectively. Nine liver recipients (21.4%) developed HCC recurrence between 4 and 48 months (median: 12 months) post-LT. Of 16 patients with PET + status, 8 developed posttransplant HCC recurrence (50%), compared to 1 of 26 PET− patients (3.8%; p < 0.001). The positive and negative predictive values of positive PET findings for HCC recurrence were 50% and 96.2%, respectively. In univariate analysis, none of the clinical variables, but several pathomorphological variables and PET findings were significantly correlated with an increased risk of posttransplant HCC recurrence (Table 4). The 3-year recurrence-free survival rate of PET− patients was 93%, compared to 35% in the PET + population (Figure 2).
Table 4. Univariate analysis of clinical, morphological and histopathological variables influencing risk of HCC recurrence
(> versus ≤ 60 years)
(> versus ≤ 50 years)
(> versus ≤ 100 IU/mL)
(A versus B or C)
(DDT versus LDLT)
Cold ischemia time
(> versus ≤ 300 min.)
Warm ischemia time
(> versus ≤ 50 min.)
(CsA versus Tac)
Size of largest tumor nodule
(> versus ≤ 5 cm)
Number of tumor nodules
(> versus ≤ 3)
Total diameter of tumor nodules
(> versus ≤ 10 cm)
(3/4 versus 1/2)
(Milan Out versus Milan In)
(Poor versus moderate/well)
(Yes versus no)
(Pos. versus neg.)
(No versus yes)
(Yes versus no)
18F-FDG-PET findings and Milan criteria
Patients with HCC meeting the Milan criteria had a significantly better 3-year recurrence-free survival rate (94%), compared to those with HCC beyond the Milan criteria (63%; p = 0.008).
None of 17 Milan In recipients with PET− status have yet developed HCC recurrence (0%), compared to 1 of 3 Milan In patients with pre-LT PET + scans (33.3%, p = 0.004).
Of 9 PET− Milan Out-patients, only 1 was suffering from tumor recurrence (11.1%), compared to 7 of 13 Milan Out recipients with PET + status (53,8%, chi square, p = 0.004).
In consequence, Milan Out recipients with negative preoperative PET findings (n = 9) demonstrated a 3-year recurrence-free survival (80%) that was comparable to Milan In patients (n = 20; 94%; p = 0.6) and significantly better than in Milan Out recipients with positive pretransplant PET scans (n = 13; 29%) (Figure 3).
18F-FDG-PET findings and MVI
Of 17 recipients with presence of MVI, 8 patients (47%) demonstrated posttransplant tumor recurrence, while only 1 of 25 patients without MVI (4%) has developed recurrent HCC (p = 0.001).
Patients without the presence of MVI (n = 25) had a calculated 91% 3-year recurrence-free survival, compared to 28% in those with MVI (n = 17; p < 0.001).
Advanced tumor stage, tumors beyond the Milan criteria, poor tumor grade and pretransplant PET + status were univariately associated with the presence of MVI (Table 5). In multivariate analysis, only preoperative PET + status was identified as an independent predictor of MVI (Table 6). The positive and negative predictive values of positive PET imaging for MVI were 87.5% and 88.5%, respectively.
Table 5. Univariate analysis of predictive factors for microvascular tumor invasion
Size of largest tumor nodule
(> versus ≤5 cm)
Number of tumor nodules
(> versus ≤3)
Total diameter of tumor nodules
(> versus ≤10 cm)
(> versus ≤100 IU/mL)
(3/4 versus 1/2)
(Milan In versus Milan Out)
(Poor versus well / moderate)
(Pos. versus neg.)
(No versus yes)
(Yes versus no)
Table 6. Multivariate analysis of independent predictive factors for microvascular tumor invasion
PET (pos. versus neg.)
18F-FDG-PET findings and TACE
Overall 3-year recurrence-free survival posttransplantation was not different between patients with (n = 21, 70%) and without (n = 19.74%) pre-LT TACE.
In the PET− population, 3-year recurrence-free survival after LT was comparable in patients with (n = 12, 100%) and patients without pretreatment (n = 13.89%), since there was only one case of HCC recurrence in the nontreated group.
In the PET + population, posttransplant 3-year recurrence-free survival after preoperative TACE (n = 9.33%) was not superior to that of recipients without neoadjuvant therapy (n = 6.40%).
In none of the recipients undergoing pre-LT TACE complete tumor necrosis was confirmed at explant pathology. Partial tumor necrosis, however, was achieved in all of them (≥50% necrosis n = 15; <50% necrosis n = 6).
The results of our trial indicate that pretransplant 18F-FDG uptake on PET in liver transplant candidates with HCC is a useful additional parameter for the evaluation of tumor biology. This study shows that positive preoperative PET scans are significantly associated with an increased risk of posttransplant HCC recurrence and inferior outcome, which seems to be triggered by a high correlation of a PET + status and the presence of MVI in explant tumor pathology.
The outcome after LT in patients with HCC has markedly improved over the last two decades (1–6). This is a result of implementing strict morphological selection criteria, such as the Milan criteria by Mazzafero et al. (5) Five-year survival rates above 70% have since been reported, which are comparable to survival rates after LT in patients without HCC (5,6).
These promising results encouraged several transplant groups to carefully expand the selection criteria (7–11). Roayaie et al. reported about an overall survival rate of 55% after LT for patients with advanced HCC, when being enrolled in a multimodal preoperative chemotherapy concept (7). Yao et al. have extended the criteria to a single tumor ≤6.5 cm, or ≤3 tumors with the largest ≤4.5 cm and a total tumor burden ≤8 cm (UCSF criteria) and reported about a 75% 5-year survival rate, when using a neoadjuvant treatment protocol (13). In addition, the implementation of living donor liver transplantation was reported to further extend the selection criteria in HCC patients (37,38). Although some of the reported results were quite promising, the tumor recurrence rate per se has been significantly higher when the selection criteria were expanded beyond the Milan criteria (39,40). Moreover, extending number and size of the tumors was shown to increase the risk of patients’ drop out from the transplant waiting list, due to clinically relevant tumor progression (40). These facts illustrate the limitations of using tumor size and number as ultimate variables for selecting appropriate liver transplant candidates and point out the need for augmenting current selection criteria with biological parameters of tumor aggressiveness.
Posttransplant tumor recurrence is thought to result from cancer cells spread into the systemic circulation at the time of total hepatectomy (41,42). In fact, vascular invasion has been identified as the strongest predictor of poor outcome after LT for HCC (1,12,14,15), which is confirmed by our own experience (43). In contrast to gross vascular tumor invasion, which can be very frequently diagnosed prior to transplantation by complementary imaging techniques, MVI is a histopathological diagnosis that cannot be made prior to native hepatectomy (14,35,44,45). Therefore, other pretransplant available clinical surrogate markers for MVI have to be identified. Esnaola et al. and Pawlik et al. demonstrated that, apart from poor tumor grade, it is advanced tumor size that correlates strongly with presence of MVI at explant pathology (14,19). The unrestricted exclusion of patients with large and multiple HCC, however, may withhold a subset of liver transplant candidates from potentially curative LT. Vauthey et al. showed that size and number of tumor nodules have no significant impact on survival in the absence of MVI (35). These data are confirmed by our own experience, since posttransplant 3-year recurrence-free survival was not significantly different between Milan In recipients (94%) and patients with tumors beyond Milan criteria but without presence of MVI (75%, p = 0.64).
The value of pretransplant HCC grading by tumor biopsy for selecting adequate liver transplant candidates with advanced tumors is being discussed controversially (4,18). Pawlik et al. reported about poor diagnostic accuracy of pretransplant fine needle tumor biopsy to predict tumor grade in the explant pathology, due to a HCC-related heterogeneous pattern of differentiation (4). Furthermore, there seems to be a substantial risk of tumor seeding after HCC biopsy, which may be particularly relevant for immunocompromised liver transplant recipients (46).
Preoperative 18F-FDG tumor uptake on PET was identified as the only independent predictor of MVI in our trial (Table 6). Moreover, positive PET scans together with parameters of the Milan criteria and biological tumor aggressiveness were significantly associated with posttransplant tumor recurrence in a univariate analysis (Table 4). This is a very important result of our trial, since 18F-FDG-PET might, therefore, be proposed as an additional noninvasive tool for assessing biological tumor behavior in liver transplant candidates with HCC, without altering the tumor itself, as it may happen in preoperative tumor biopsy (46,47).
Several investigators have reported about a limited sensitivity of 18F-FDG-PET in the detection of HCC, when compared to other liver neoplasms (48,49). The difference in the accumulation of 18F-FDG in HCC and metastatic liver tumors is due to different activities of glucose-6-phosphatase, which is responsible for the conversion of FDG-6-phosphate to FDG. It is high in normal liver tissue and almost zero in metastatic liver tumors (48,49). In HCC, however, there is a wide variety of enzyme activities due to cell differentiation. It has been shown that well-differentiated HCC cells exhibit an 18F-FDG metabolism similar to that of normal liver tissue, whereas poorly differentiated tumor cells do not. Therefore, well-differentiated HCC regions tend to accumulate an amount of FDG that is comparable to normal surrounding liver tissue, which may propose PET for describing tumor aggressiveness rather than detection of HCC (48–51). In a clinical approach, Seo et al. demonstrated a significant correlation of high standardized uptake values (SUV) with high-grade tumor differentiation and poor outcome after liver resection for HCC (29). Based on the close relationship between MVI and poor tumor grade (12,14–16,28,29,51–53), we, therefore, hypothesized that PET scan patterns may have a clinically relevant correlation with the presence of MVI.
The results of our trial implicate that PET findings may optimize the selection process for LT in patients with HCC.
This might be particularly relevant for patients with tumors beyond the Milan criteria, who are rejected for LT in many centers. In our series, patients with advanced HCC and negative preoperative PET scans demonstrated a 3-year recurrence-free survival (80%) that was comparable with Milan In recipients (94%), and significantly better than for Milan Out patients with PET + status (35%, Figure 3). This result indicates that a subset of patients with advanced HCC may benefit from LT, due to the biological ‘low-aggression’ behavior of the tumor, and PET imaging may help with identifying these eligible liver transplant candidates. The future clinical follow-up of this special subpopulation of our series, however, has yet to be analyzed thoroughly.
The value of the Milan criteria for selecting appropriate liver transplant candidates is clearly confirmed by our trial, since recurrence-free survival rates for Milan In recipients were significantly higher than for Milan Out patients. Nevertheless, there are several problems associated with their implementation for patient selection, where pretransplant PET assessment may be helpful in the future.
First, using the same criteria for both, inclusion and exclusion, may result in a high drop-out rate, especially for patients, who are initially approaching to the upper criteria limits (54). This might exclude a significant number of patients with HCC exceeding the Milan criteria from curative LT, as proposed by our data and by other trials (7–11). Similar to other transplant groups with excellent survival results (e.g. Barcelona group), we, therefore, followed a policy of patients’ exclusion based on major events (macrovascular tumor invasion, lymph node metastases, tumor-related symptoms, extrahepatic tumor spread) rather than on the Milan criteria alone (6). Apart from that, tumor understaging by preoperative radiographic imaging is a well-known phenomenon in this context, resulting in a large fraction of patients with advanced tumors at explant pathology (6,39,55). In 17 patients of our series (40.5%), final pretransplant screening underestimated pathological tumor staging. Against this background, PET evaluation may be useful for further refining the decision making about drop-out of liver transplant candidates with tumor progression during the waiting period for transplantation. This, however, has yet to be analyzed by a prospective study.
Apart from being a retrospective analysis, there are some further limitations of our trial.
First, we did not use a quantitative uptake measurement in our study, since SUV levels were not available in all patients. Moreover, we did not perform repeat PET imaging by protocol during waiting time for transplantation. However, performing follow-up PET scans might be of special interest for evaluating the impact of interventional therapies on biological tumor viability. Torizuka et al. demonstrated that an increased or similar 18F-FDG uptake postchemoembolization suggested residual viable tumor tissue, whereas decreased or absent 18F-FDG uptake indicated more than 90% necrosis (56). These data seem to indicate that accuracy of a single pretransplant PET finding for predicting biological tumor characteristics in the explant specimen could be affected by additional TACE achieving complete tumor obliteration.
In our study population, however, there was no case of complete tumor necrosis following TACE. MVI was evident in the remaining viable tumor regions in 7 of 9 PET + patients undergoing pretransplant TACE (77.8%), and in 6 of 6 nontreated PET + recipients (100%). This result seems to explain the high accuracy of a single positive PET finding for predicting MVI in our trial, even though 21 patients underwent TACE pretransplantation. Nevertheless, the value of repeat PET evaluation for detecting changes in biological tumor aggressiveness during the waiting time for LT and, thereby, possibly altered patient prognosis post-LT should be assessed in a prospective framework.
In conclusion, our data suggest that preoperative 18F-FDG uptake on PET is a reliable predictor of MVI and tumor recurrence after LT for HCC. Patients with tumors beyond the Milan criteria and negative preoperative PET scans may achieve a posttransplant 3-year recurrence-free survival that is comparable to that of patients with tumors meeting the Milan criteria. In the context of neoadjuvant therapies, repeat PET evaluations by protocol might be useful for unfolding the full potential of PET to predict MVI.
Thereby, 18F-FDG-PET imaging could emerge as a useful diagnostic tool for improved decision making about listing or exclusion of liver transplant candidates with HCC.