Response to transarterial chemoembolization as a biological selection criterion for liver transplantation in hepatocellular carcinoma


  • See Editorial on Page 1189


Criteria to select patients with hepatocellular carcinoma (HCC) for liver transplantation (LT) are based on tumor size and number of nodules rather than on tumor biology. The present study was undertaken to assess the role of transarterial chemoembolization (TACE) in selecting patients with tumors suitable for LT. Ninety-six consecutive patients with HCC were treated by repeatedly performed TACE, 62 of them exceeding the Milan criteria. Patients meeting the Milan criteria were immediately listed, and patients beyond the listing criteria were listed upon downstaging of the tumor following successful TACE. Fifty patients were finally transplanted. Of these 50 patients, 34 exceeded the Milan criteria. In these 96 patients, overall 5-year survival was 51.9%. However, it was 80.9% for patients undergoing LT and 0% for patients without transplantation (P < 0.0001). Tumor recurrence was primarily influenced by the control of the disease through continued TACE during the waiting time. Freedom from recurrence after 5 years was 94.5% in patients (n = 39) with progress-free TACE during the waiting time. Tumor recurrence was significantly higher in patients (n = 11) who after initial response to TACE progressed again before LT (freedom from recurrence 35.4%; P = 0.0017). Progress-free course of TACE during the waiting time (P = 0.006; risk ratio, 8.95), and a limited number of tumor nodules as assessed in the surgical specimen (P = 0.025; risk ratio, 0.116) proved to be significant predictors for freedom from recurrence in the multivariate analysis. Milan criteria were without impact on recurrence. Our data suggest that sustained response to TACE is a better selection criterion for LT than the initial assessment of tumor size or number. Liver Transpl 12:1260-1267, 2006. © 2006 AASLD

Liver transplantation (LT) is the ultimate treatment of hepatocellular carcinoma (HCC). Compared to resection, LT bears 2 prinicipal advantages: The removal of the tumor is not restricted by the compromized functional capacity of the liver, and tumor recurrence caused by the precarcinomatous situation of the cirrhotic liver is eliminated. However, LT is justified only in early HCC, since the recurrence rate in advanced tumors is high.1, 2 Milan criteria reflecting tumor size and multifocality are widely used for selecting transplant candidates.1 Recently, extended criteria were proposed.3 In contrast, both the important prognostic markers vascular invasion4, 5 and tumor grading2, 4 are of limited clinical relevance, since vascular invasion and grading are difficult to assess before transplantation.

Pretreatment before transplantation has mainly focused on the issue of bridging the waiting time before transplantation.6–12 Only a few reports have addressed the question whether preoperative treatment, usually aiming at down-staging, is capable of influencing survival.5, 13, 14 Recently, excellent results have been reported after bridging by transarterial chemoembolization (TACE)9, 13, 15, 16 and by radiofrequency ablation.10, 11 Whereas radiofrequency ablation is feasible only in tumors with restricted diameter, TACE may also be used in large tumors exceeding the Milan criteria, but survival benefit after down-staging by TACE remains questionable.5, 13–15 It is generally believed that survival despite TACE pretreatment in patients with large tumors remains inferior to that of patients fulfilling the selection criteria.

From 1998 onward, we have included 96 patients with HCC in a pretreatment protocol by TACE among whom 50 underwent liver transplantation. In any patient, TACE was repeatedly performed in intervals of 6 weeks until transplantation. Herein we report the results in these 50 patients and focus on the influence of this pretreatment on outcome.


LT, liver transplantation; HCC, hepatocellular carcinoma; TACE, transarterial chemoembolization; CT, computed tomography.



Between May 1998 and May 2005, 96 patients younger than 65 years of age and without any prior therapy presented with HCC. These patients were included in a treatment protocol consisting of sequential TACE eventually followed by LT. Evidence of extrahepatic disease and other contraindications against transplantation were exclusion criteria, but not the size or number of nodules. In all patients, size and number of nodules were assessed by double-contrast spiral computed tomography (CT) scan. HCC was confirmed before TACE in 90 of the 96 patients by biopsy. In the other 6 patients diagnosis was based on 2 imaging procedures and elevated alpha-fetoprotein levels. Written informed consent was given by all patients. The study was approved by the local ethics committee. All data used in this study were retrieved from our prospectively collected database containing all patients with HCC.

After confirmation of diagnosis, patients supposed to be within the Milan criteria (n = 34) were immediately listed for LT at Eurotransplant. In patients exceeding the Milan criteria (n = 62), listing was performed only after the tumor had proved to be regressive during TACE pretreatment.

The characteristics of the patients transplanted following TACE are listed in Table 1. The tumor characteristics defining the Milan criteria are given in Table 2. In Table 2 Milan criteria are classified according to the maximal finding either in the CT scan or in the pathomorphological evaluation. Sixty-eight percent of the tumors exceeded the Milan criteria, and 88% of the patients had Child-Turcotte-Pugh class A or B disease.

Table 1. Characteristics of 50 Patients Following TACE Pretreatment Before LT
  • Abbreviations: HCV, hepatitis C virus; HBV, hepatitis B virus.

  • *

    Classification according to the maximal number of patients exceeding the Milan criteria either in the CT scan or in the pathomorphological evaluation.

Median follow-up (range)891 (38-2,585) days
Age58.4 ± 7.1 years
Milan criteria fulfilled/exceeded (initial CT)23/27
Milan criteria fulfilled/exceeded (specimen)31/19
Milan criteria fulfilled/exceeded*16/34
1-3 nodules/multiple nodules (initial CT)31/19
1-3 nodules/multiple nodules (specimen)41/9
Tumor diameter ≤ 3 cm/>3 cm (initial CT)19/31
Tumor diameter ≤ 3 cm/>3 cm (specimen)30/20
T classification: T1/T2/T3 (specimen)18/23/9
Grading: G1/G2/G32/35/13
Angioinvasion: yes/no12/38
Child-Turcotte-Pugh A/B/C33/11/6
Initial disease: HCV/HBV/alcoholic/others)16/8/19/7
Table 2. Tumor Characteristics Defining the Milan Criteria
Diameter of Largest LesionNumber of Lesions
  • NOTE: Patients in bold are by definition meeting the Milan criteria. (Milan criteria are classified according to the maximal finding either in the CT scan or in the pathomorphological evaluation.)

  • *

    The diameters of the largest lesion in each patient were 6.0 and 8.0 cm.

  • The diameters of the largest lesion in each patient were 6.0, 8.0, and 12.0 cm.

  • The diameters of the largest lesion in each patient were 6.0 and 6.4 cm.

  • §

    The diameters of the largest lesion in each patient were 6.0, 6.0, 7.0, 7.3, 10.5, and 12.0 cm.

 123> 3
≤3 cm3368
>3-≤5 cm4085
>5 cm2*326§

TACE Protocol and Monitoring of Tumor Growth

For embolization, lipiodol (20 mL) and mitomycin (10 mg in 20 mL water) were injected into the supplying artery until stop of flow. The amount of lipiodol was increased to a maximum of 30 mL in large tumors. TACE was repeated every 6 weeks (Fig. 1) until transplantation and every time a triphasic spiral CT scan (2.5-mm collimation, 2-mm reconstruction interval) was performed and local tumor growth was assessed by 2 independent investigators. The measurement of the tumor was based upon the arterial phase during the CT scan. Every 3 months, abdominal and thoracic CT scans were performed to identify extrahepatic spread. This TACE protocol and the monitoring of tumor growth were strictly followed in all patients meeting or exceeding the listing criteria.

Figure 1.

Clinical outcome of 96 patients enrolled in the pretreatment protocol using TACE meeting (n = 34) or exceeding (n = 62) the Milan criteria; reasons for treatment failure and final number of LT performed. *For the definition of progress during waiting time, see text. **According to pathomorphology of the surgical specimen only 16 of them met the Milan criteria. wks., weeks; WL, still on waiting list.

Definitions of Tumor Growth and Hepatic Decompensation

Tumor response was defined according to the Response Evaluation Criteria in Solid Tumors (RECIST).17 The following 2 definitions are of important for exclusion (patients within the Milan criteria) or inclusion (patients beyond the listing criteria).

Progress was defined as at least a 20% increase in the sum of the largest diameter of target lesions or the appearance of new nontarget lesions and/or nonmeasurable lesions. Patients meeting the Milan criteria, who were immediately listed, were delisted if they fulfilled this definition despite the initiation of the TACE pretreatment.

Partial response in patients with tumors exceeding the Milan criteria meant at least a 30% decrease in the sum of the largest diameter of 5 target lesions, taking as reference the baseline sum of the largest diameter. Patients beyond the Milan criteria were considered to be down-staged if they fulfilled this definition.

Progress During Waiting Time

After initial response to TACE, in some patients the tumor(s) started to progress again during the waiting time. Any increase in size or number of tumor nodules was considered as “progress during waiting time” when confirmed by 2 independent observers, even if the progress did not reach the RECIST definition. This stricter definition of progress was considered to be of crucial importance for tumor recurrence.

Functional decompensation, restricted to patients with Child-Turcotte-Pugh C cirrhosis, was defined as encephalopathy, increase in ascites and/or increase in bilirubin by more than 2 mg/dL. Functional decompensation was a reason to exclude the respective patient from the study.

Liver transplantation was performed using piggyback technique. Postoperative immunosuppression consisted of tacrolimus (n = 78) or cyclosporine (n = 9). Methylprednisolone was given in all patients beginning with 10 mg/kg intraoperatively and tapered down to 12 mg after 2 weeks. Target of rapamycin inhibitors or adjuvant chemotherapy were not used. All patients were followed at regular intervals after transplantation.

Statistical Analysis

All statistical analyses (chi-square test, Wilcoxon test, Kaplan-Meier estimate, log rank test and Cox regression) were performed using the SPSS program 11.5 for Windows (SPSS, Chicago, IL). Survival means actuarial survival according to Kaplan-Meier. In recurrence-free survival, both death and recurrence are counted as events. Freedom from recurrence means that only recurrence is counted as to be an event, whereas death from other reasons is censored. The time from the first TACE was used for survival calculations of all 96 patients. For all other calculations, the time of transplantation was the starting point.


Thirty-four patients meeting the Milan criteria were immediately listed. Of them, 6 had to be delisted again when their tumor immediately progressed according to the RECIST criteria despite TACE. One patient had to be delisted for hepatic decompensation. Twenty-three patients meeting the listing criteria remained on the waiting list and were finally transplanted. Four patients are still on the waiting list (Fig. 1).

Sixty-two of the 96 patients with no other contraindications against LT exceeded the Milan criteria as assessed by the initial CT scan. In 34 of these 62 patients, there was a partial response of the HCC after the initiation of the TACE pretreatment. These downstaged patients were listed and TACE was continued until transplantation. In 5 of the 28 patients who were not listed, TACE pretreatment had to be stopped due to hepatic decompensation, whereas 23 patients had progress or stable disease and systemic chemotherapy or supportive treatment were initiated. During the time on the waiting list, 6 of 34 patients had to be delisted due to extrahepatic disease (n = 5) and hepatic decompensation (n = 1). Twenty-seven patients were finally transplanted, and 1 patient is still on the waiting list.

Two patients and 9 others assumed to meet or to exceed, respectively, the Milan criteria were of particular interest during the waiting time. After initial partial response or stable disease (patients meeting the Milan criteria) or downstaging (patients exceeding the Milan criteria), the tumor(s) proceeded to progress before LT despite continued TACE. In 3 cases, an increase in size of target lesions (less than 20% of the sum of largest diameter) occurred, in 6 cases nonmeasurable de novo nodules were found, and in 2 cases both were encountered. These patients were defined as having progress during the waiting time. Nevertheless, all these patients were transplanted. The number of TACE cycles performed to downstage the patients exceeding the Milan criteria before listing, the numbers of TACE cycles during the waiting time, and the respective waiting time for each group of patients are shown in Table 3.

Table 3. Median Waiting Time and Numbers of TACE Performed in Patients With HCC
 Median WT, daysNo. of TACE median
  1. Abbreviations: WT, waiting time; MC, Milan criteria.

All patients218 (22-756)5 (2-14)
Within the MC250 (22-756)4 (2-10)
Beyond the MC (pretreatment)176 (57-579)3 (2-11)
Beyond the MC(after listing)102 (22-281)2 (0-5)
Beyond the MC (total)268 (129-818)5 (2-14)
Patients with progress during WT281 (72-756)7 (2-14)
Patients without progress during WT158 (22-716)5 (2-11)

According to the pathomorphological assessment of the surgical specimen, 7 patients initially assumed to meet the Milan criteria exceeded these criteria. Therefore, the final number of patients meeting and exceeding the Milan criteria amounted to 16 and 34, respectively. The figures are demonstrated in the Tables 1 and 2. According to this final classification of the Milan criteria only 1 patient within, but 10 patients beyond these criteria had progress during the waiting time (chi-square, P = 0.065).

The overall analysis of the 96 patients being candidates for LT resulted in a 5-year survival of 51.9%. As expected, the survival of patients not responding to down-staging (patients beyond the Milan criteria) or with initial progress (meeting the Milan criteria) was significantly inferior to transplanted patients (80.9% compared to 0%, respectively (P < 0.0001) (Fig. 2). Five-year survival (calculated from the time of transplantation) for transplanted patients was 74.4%. The 5-year recurrence-free survival and the 5-year freedom from recurrence are demonstrated in Figure 3. They were 68.3% and 77.8%, respectively.

Figure 2.

Survival of 96 patients enrolled in the TACE protocol. The difference in 5-year survival between downstaged (transplanted) patients and patients not responding to TACE is highly significant (P < 0.0001). (Survival is calculated from the beginning of TACE treatment.) Solid line denotes all patients included in the analysis. Dotted line denotes patients not responding to TACE. Dashed line denotes transplanted patients. Cross denotes censored.

Figure 3.

Recurrence-free survival (death and recurrence are counted as events in the Kaplan-Meier estimate) and freedom from recurrence (only recurrence is counted) in 50 patients following TACE pretreatment before LT. Solid line denotes freedom from recurrence. Dashed line denotes recurrence-free survival. Cross denotes censored.

To assess the oncologic capacity of TACE to influence the course following LT, only freedom from recurrence is going to be considered in the further evaluation. In total, there were 7 recurrences of HCC following LT. Five of them occurred in the 11 patients with tumor progress during the waiting time compared to only 2 in 39 patients with uneventful course of TACE before transplantation (chi-square, P = 0.001). This led to a 5-year freedom from recurrence of 35.4% in patients with progress during the waiting time vs. 94.5% in patients with uneventful TACE (Fig. 4; log rank: P = 0.0017). Whereas the number of TACE procedures did not differ between the 2 groups (Wilcoxon test, P = 0.10; compare Table 3), the difference in waiting time reached statistical significance (Wilcoxon test, P = 0.025). The degree of tumor necrosis in the explant in patients with progress during the waiting time and patients without progress was comparable (complete, partial, focal necrosis: 1, 5, 5, and 3, 20, 16, respectively). The follow-up of patients with and without progress during the waiting time was also comparable (836 ± 489 vs. 906 ± 715 days, respectively; Wilcoxon test, P = 0.916).

Figure 4.

Posttransplant freedom from recurrence. TACE without progress (denoted by solid line) during waiting time vs. TACE with progress (denoted by dashed line) during waiting time. Cross denotes censored.

In patients meeting (n = 16) or exceeding (n = 34) the Milan criteria, freedom from recurrence after 5 years was 93.8% and 74.5%, respectively (P = 0.4211) (Fig. 5). In the univariate analysis, grading and the number of tumor nodules were influential on the rate of tumor relapse. In addition, progress during the waiting time gained a highly significant influence on tumor recurrence. Remarkably, Milan criteria were not significant (Table 4). The results of the multivariate analysis for freedom from recurrence are summarized in Table 5.

Figure 5.

Five-year freedom from recurrence according to the Milan criteria (fulfilled vs. exceeded) as assessed by the initial CT or the surgical specimen (maximal Milan criteria). Solid line denotes fulfilled criteria. Dashed line denotes criteria exceeded. Cross denotes censored.

Table 4. Univariate Analysis for Freedom From Recurrence (50 Patients With TACE Pretreatment Before LT)
Predictorn5-Year Survival, %Log rank
  • Abbreviations: mult., multiple; AFP, alpha-fetoprotein; WT, waiting time.

  • *

    Maximal means that Milan criteria exceeded either in the initial CT scan or in the surgical specimen.

Milan criteria   
(fulfilled vs. exceeded)   
Initial CT assessment23/2790.9/73.20.5586
In specimen31/1979.5/72.90.3254
T1/2 vs. 341/980.3/74.10.5595
G1/2 vs. 337/1390.7/50.80.0096
1, 2, 3 vs. mult. (in specimen)41/980.5/56.30.0373
1, 2, 3 vs. mult.   
(initial CT assessment)31/1986.3/69.80.2833
≤3/>3 cm   
(in specimen)30/2081.0/74.20.4290
≤ 3/>3 cm   
(initial CT assessment)19/3188.5/75.00.8557
AFP </≥ 1024/2675.0/80.30.7783
WT </≥ 120 days19/3182.5/71.70.8327
Age </≥ 55 years14/3683.6/76.70.8015
Progress during waiting time39/1194.5/35.40.0017
Table 5. Posttransplant Freedom From Recurrence (Multivariate Analysis)
PredictornSignificanceRR95% confidence interval for RR
  1. Abbreviation: RR, risk ratio.

1,2,3 vs. multiple nodules (in specimen)41/90.0250.1660.034-0.796


The issue of the present study is not to assess the dropout rate of patients listed for LT, the rate of patients who may be down-staged by TACE, or even the methodology of this approach, but to assess the oncological result of this treatment in patients undergoing LT. Our program was primarily not initiated to bridge the waiting time but to transplant patients who responded to TACE even if their tumor initially and after downstaging exceeded the listing criteria. Thereby, waiting time per se and response to neoadjuvant treatment might select patients with biologically favorable tumors.

It is highly desirable to select patients with biologically favorable tumors for LT in order to avoid scarce organs being used for a questionable outcome. Moreover, immunosuppression may contribute to accelerated tumor growth and decreased survival. This is of particular interest in living donation LT, where tumor recurrence also gains importance for the donor.

As early as 1997, Bismuth's group reported on results of TACE before LT. If all pretreated patients were included in their analysis, the difference in survival between TACE and non-TACE patients was not significant. For larger tumors, however, the difference in survival was highly significant depending on whether or not TACE led to downstaging of the tumor.14 Remarkably, on the average only 1.96 courses of TACE were performed in this study. Recently, Roayaie reported acceptable results in 43 out of 80 patients with tumors larger than 5 cm transplanted following 1.53 (mean) cycles of TACE.5

In our study, the number of performed TACE cycles was higher than in most other studies,5, 9, 12, 14, 15, 20 and TACE was scheduled in short regular intervals. As a consequence, the time between the first TACE and transplantation was long (268 days in patients beyond the selection criteria).

The majority of TACE procedures (44 of 50 patients) has been performed in Child-Turcotte-Pugh class A and B cirrhosis. Poor liver function is a contraindication against TACE and other pretransplant approaches.10, 12, 16 In this study, hepatic decompensation occurred in 5 and 2 patients during downstaging and waiting time, respectively. Hydropic decompensation, encephalopathy, increase in bilirubin, and loss of synthetic capacity were the uniform features in these patients. Since the patient's condition is an important factor influencing survival following LT, those patients who tolerate TACE may expect good posttransplant survival. It should be underscored that this might be a reason why in our patients survival following LT is superior to the recently published data of large multicenter reports.21, 22 Regardless of the functional aspects in the management of patients with HCC, the main issue of this study remains the oncologic response to TACE. Therefore, freedom from recurrence was considered to be of particular interest for the present analysis.

Freedom from recurrence was not significantly different in patients meeting or exceeding the Milan criteria, regardless whether the Kaplan-Meier calculations were based on the initial CT criteria, on the specimen, or on the “maximal” Milan criteria (compare Table 4). The figures of these subgroups must be cautiously interpreted due to the limited number of patients. Nevertheless, it is to be stressed that the rate of patients exceeding the Milan criteria was high (68%). Considering the low sensitivity of imaging techniques in HCC,23, 24 it may be assumed that the initial number of HCC nodules in our patients might even be higher before TACE, and even after downstaging 19 patients did not meet the Milan criteria. This, however, did not result in an unacceptable rate of tumor recurrence. Therefore, Milan criteria seem not to be crucial in predicting the probability of tumor recurrence in this treatment protocol.

Liver transplantation was successfully performed in larger tumors after downstaging, and it has been assumed that biologically less-aggressive tumors had been selected by this approach.14 In contrast to this, the capacity of downstaging by TACE to select tumors excceeding the listing criteria for LT has recently been reported to be limited.13, 15 Our study suggests that oncological control reached by the scheduled TACE pretreatment during the waiting time is obviously of greater importance for the long-term prognosis than downstaging itself. According to our data, favorable tumors seem to be selected with a much higher degree of reliability, if the process of tumor control persists during the total waiting time.

In 11 patients, after an initial good response the tumor then proceeded to progress again during the waiting time. In 3 patients the sum of the diameters of the 5 target lesions increased to a limited degree, not reaching the definition of progress according to the RECIST criteria, in 6 patients de novo nonmeasurable lesions were detected, and in 2 patients both events occurred. Five of the 7 recurrences in our study were observed in these patients with minimal progress. This recurrence rate is significantly different from that of the 39 patients with progress-free TACE (2 of 39; chi-square, P = 0.0017). Therefore, to identify patients being at risk for tumor recurrence, any progress in the CT scans taken at any TACE cycle seems to be important.

Interestingly, median waiting time of patients with progress during the waiting time was significantly longer than that of patients without progression (281 vs. 158 days; P = 0.025). Therefore, waiting time itself seems not to be the main factor in selecting patients. The degree of necrosis reached by the repeatedly performed TACE before LT also did not reflect the different behavior of the 2 groups. It appears that less aggressive tumors respond to TACE; however, at any time point mutations within the tumor might lead to the outgrowth of more aggressive tumor cells, resulting in an increased risk of tumor recurrence after transplantation.

The hypothesis of tumor selection by TACE is corroborated by a comparison of the predictors of recurrence in the univariate analysis (Table 4). Progress-free TACE during the waiting time, number of nodules, and grading are significant predictors. Angioinvasion did not reach statistical significance. In the multivariate analysis progress-free TACE during the waiting time, and the number of nodules in the surgical specimen remained significant (compare Table 5). This observation as well as the results of the univariate analysis reflect the capacity of TACE to select different groups of patients.

In conclusion, our results suggest that large and even multifocal HCC can be successfully treated by LT if they show a good response and remain stable during the sequential pretreatment by TACE. This observation may form a basis of biologically more plausible selection criteria rather than simple tumor size or tumor number, provided that tumor progress is strictly defined. It may be of importance for future treatment concepts in patients with HCC, in particular for those in whom living donor liver transplantation is considered.