Analysis of risk factors for tumor recurrence after liver transplantation for hepatocellular carcinoma: Key role of immunosuppression


  • See Editorial on Page 494


To confirm recent observations about the relationship between immunosuppression and the recurrence of hepatocellular carcinoma (HCC) after liver transplantation (LT), we retrospectively analyzed 70 consecutive HCC patients who underwent LT and received cyclosporine (CsA)–based immunosuppression. CsA trough blood levels, measured with the same technique (fluorescence polarization immunoassay), were analyzed at different time points after transplantation. The exposure to the drug was calculated with the trapezoidal rule in each patient. CsA was associated with steroids in 26 patients and steroids and azathioprine in 44 patients. HCC recurred in 7 patients (10.0%). Different immunosuppressive schedules (CsA and steroids vs. CsA, steroids, and azathioprine) or the cumulative dosage of steroids and azathioprine did not influence HCC recurrence that was associated instead with CsA exposure (278.3 ± 86.4 ng/mL in recurrent vs. 169.9 ± 33.3 in tumor-free patients; P < 0.001); CsA exposure above 189.6 ng/mL was related to HCC recurrence at the receiver operating characteristic analysis (ROC). The relationship between CsA exposure; various clinical (sex, age, viral- vs. non–viral-related cirrhosis, preoperative vs. incidental diagnosis of HCC, alpha-fetoprotein [AFP] blood level), pathologic (pathologic tumor staging [pT] stage, presence of Milan criteria), and histologic (grading, presence of microvascular tumor invasion) parameters; and tumor recurrence were assessed. AFP (P = 0.032), microvascular tumor invasion (P = 0.044), and CsA exposure (P < 0.001) influenced recurrence-free survival at the univariate analysis; CsA exposure was the only independent prognostic determinant at multivariate analysis (P < 0.001). High CsA exposure favors tumor recurrence; CsA blood levels should be kept to the effective minimum in HCC patients. In the presence of pathologic and histologic risk factors, specific immunosuppressive protocols should be considered. (Liver Transpl 2005;11:497–503.)

Hepatocellular carcinoma (HCC) is a frequent finding in patients with chronic liver disease listed for liver transplantation (LT). Tumor recurrence after transplantation involves an ominous prognosis and strict selection criteria of transplant candidates on the basis of tumor features developed to minimize its incidence, among which the most widely adopted are the so-called Milan criteria.1 Although it is known that the pharmacologic immunosuppression required after transplantation can accelerate tumor growth, the possible influence of different immunosuppressive schedules on HCC recurrence after LT had been poorly investigated until recently.2, 3

In a previous report, we demonstrated a close relationship between the amount of cyclosporine (CsA), one of the most widely adopted immunosuppressant drugs, administered during the first postoperative year and tumor recurrence in patients who underwent LT for HCC.4 The aim of the current study was (1) to further investigate the possible relationship between the type and the degree of immunosuppression as expressed by exposure to the main immunosuppressant drug and tumor recurrence in patients transplanted for HCC and (2) to identify possible strategies to avoid tumor recurrence.

The influence of different schedules of immunosuppression and many clinical, pathologic, and histologic factors on HCC recurrence also were investigated with univariate and multivariate analysis.


HCC, hepatocellular carcinoma; LT, liver transplantation; CsA, cyclosporine; AUC, area under the curve; AFP, alpha-fetoprotein; ROC, receiver operating characteristic; pT, pathologic tumor staging; OR, odds ratio; 95% CI, 95% confidence interval.

Patients and Methods


Ninety-three consecutive patients with HCC who underwent transplantation between 1991 and 2002, who received CsA as the main immunosuppressant, and whose CsA trough whole blood levels were measured with the same technique (monoclonal fluorescence polarization immunoassay) were deemed suitable for comparison and were retrospectively evaluated. Patients who underwent transplantation before January 1991 had CsA blood levels measured with a radio immunoassay method that exhibited a high cross-reactivity toward CsA metabolites, which resulted in considerably higher values of blood CsA. Patients who underwent transplantation after August 2002 had CsA blood levels measured with the 2-hour postdose sampling. Both these groups of patients had values of CsA blood levels that were not comparable with those of the patients who underwent transplantation between 1991 and August 2002 and were not considered in the current study.

All patients underwent imaging techniques that always included ultrasonography and total body CT scan. None of the patients in the study had signs of extrahepatic malignancy at the time of transplantation. Further, none had signs of tumor spread at surgery.

The follow-up of these patients always included serum alpha-fetoprotein (AFP) measurement, abdominal ultrasonogram, and chest x-ray performed monthly up to the third postoperative month, then every 3 months up to the third postoperative year, then at least every 6 months. No patient received adjuvant therapy.

CsA trough whole blood levels were measured with monoclonal fluorescence polarization immunoassay (Cyclosporine Monoclonal Whole Blood, Abbott Laboratories, Abbott Park, IL). Fourteen of the 93 patients on CsA died postoperatively (mean survival 46.6 ± 25 days) without evidence of tumor recurrence and were excluded from the study group. Causes of death were postoperative multiorgan failure (7 cases), primary graft nonfunction (2 cases), cerebral hemorrhage (2 cases), myocardial infarction (1 case), sepsis (1 case), and intraoperative cardiac arrest (1 case). In addition, another 8 patients who were switched to tacrolimus and 1 who was lost to follow-up were excluded from the study (none of these patients had evidence of HCC recurrence). Therefore, 70 patients on CsA were available for the study: 62 men and 8 women with an average age of 51 years (range: 25–64 years) and a mean follow-up of 55.5 months (range: 9.9–147.8 months).

CsA Exposure

The CsA administration schedule was not aimed at reaching an established blood level of the drug but was determined by the clinician in charge on the basis of the clinical and biochemical picture of the patient. The area under the curve (AUC) of the CsA blood concentration versus the time course (CsA–AUC), until the recurrence or the end of follow-up, was calculated by using the trapezoidal rule. CsA blood levels tested at 7, 14, and 30 days and 3, 6, 9, 12, 18, 24, 30, 36, and 48 months after transplantation were available for all patients who reached those time points in the follow-up; because patients did not have the same length of follow-up and the analysis of those who had tumor recurrence ended at the time of the diagnosis of tumor relapse, the mean CsA exposure was defined as the CsA–AUC divided by the time of exposure to the drug. Further, the mean CsA blood levels of patients who did and did not experience HCC recurrence were assessed at the previously mentioned time points after LT and compared.

Other Risk Factors for Tumor Recurrence

Of the 70 patients in the study, 44 received a “triple” immunosuppressive regimen (CsA, steroids, azathioprine), and 26 received CsA in association with steroids. The cumulative dosage of steroids and azathioprine was assessed in relation to tumor recurrence. The dosage of steroids and azathioprine was not considered beyond the time of tumor recurrence. The two different schedules also were compared.

The influence of the following parameters on tumor recurrence-free survival rate were analyzed:

  • Recipient sex and age

  • Underlying liver disease (viral-related vs. non–viral-related disease)

  • Timing of the diagnosis of HCC (preoperative vs. incidental)

  • Preoperative AFP blood level

  • pT staging (pT1 vs. pT2-pT3)5

  • Fulfillment of the Milan criteria

  • Histologic grading (grades I-II vs. grades III-IV)

  • Microvascular tumor invasion (present vs. absent)

All the pathologic parameters considered in the current analysis were based on the findings of the removed native liver (Table 1).

Table 1. Characteristics of 70 Patients with HCC Who Underwent Liver Transplantation and Received CsA as Main Immunosuppressant Drug
 Number (%)Mean size (range)*
  • *

    Mean size of the main lesion in centimeters.

  • **

    Involving a major branch of the portal vein.

Recipient sex  
 Male62 (88.6) 
 Female8 (11.4) 
Child-Turcotte-Pugh class  
 A10 (14.3) 
 B33 (47.1) 
 C27 (38.6) 
Etiology of liver disease  
 HBV22 (31.4) 
 HBV-HCV4 (5.7) 
 HCV34 (48.6) 
 Other10 (14.3) 
Milan criteria  
 Unfulfilled13 (18.6) 
 Fulfilled57 (81.4) 
pT stage  
 pT124 (34.3)1.8 (0.6-4.8)
 pT244 (62.9) 
  Solitary133.0 (1.2-5.0)
  Multiple312.4 (0.5-4.5)
 pT32 (2.9) 
Type of tumor  
  pT116 (30.7)2.0 (1.2-4.8)
  pT234 (65.3)2.8 (0.8-5.0)
  pT32 (3.8)4.6 (4.0-5.2)
  pT18 (44.4)1.5 (0.6-2.5)
  pT210 (55.5)1.6 (0.5-3.3)
Histologic grading  
 I-II41 (58.6) 
 III-IV29 (41.4) 
Microvascular invasion  
 Absent41 (58.6) 
 Present29 (41.4) 

Statistical Analysis

CsA exposure, cumulative steroids and azathioprine dosage, CsA blood levels, recipient age, and size of the main tumor were expressed as continuous variables and reported as mean ± standard deviation. AFP was log-transformed before analysis (Kolmogorov-Smirnov test for normal distribution: P < 0.001 and P = 0.531 before and after transformation, respectively). Recipient sex, underlying liver disease, Milan criteria, pT stage, type of tumor, histologic grading, microvascular invasion, and immunosuppression schedule were expressed as categorical variables.

Univariate analysis of survival was performed using the Cox regression. The odds ratios (ORs) together with their 95% confidence intervals (95% CI) were also evaluated. To identify factors independently related to recurrence-free survival rate, the stepwise multivariate Cox regression was applied. Because of the low number of cases, only those factors that proved significant at the univariate analysis (P < 0.05) were taken into account.

To analyze the accuracy of CsA exposure in predicting tumor recurrence, the receiver operating characteristic (ROC) curve was used6; the optimal cutoff for differentiating patients with tumor recurrence from those who remained free of tumors was evaluated by means of the MedCalc software (MedCalc software, Mariakerke, Belgium).

Comparisons between CsA blood levels in the 2 groups at different time points were made using the Mann-Whitney U test. Two-tailed P values less than 0.05 were considered significant. All statistical evaluations were performed using the SPSS for Windows package (Version 8.0, SPSS Inc, Chicago, IL).


HCC recurred in 7 (10.0%) of 70 patients. Recurrence was diagnosed between 2 and 40 months after LT (median: 11 months); the patterns of recurrence are shown in Table 2. Recurrence-free survival was significantly related to CsA exposure (P < 0.001; OR: 1.05, 95% CI: 1.03-1.08) (Table 3). CsA exposure was higher in those patients who experienced tumor recurrence (mean: 278.3 ± 86.4 ng/mL) compared with patients who remained tumor free (mean: 169.9 ± 33.3 ng/mL).

Table 2. Patterns of Recurrence of HCC After Liver Transplantation in Patients Who Received CsA as Main Immunosuppressant Drug
PatientpT stageTime to recurrence (mos)Site of recurrence
  1. Abbreviation: LN, lymph nodes.

1pT232Spine, chest wall
2pT37Lung, multiple liver lesions
3pT211Multiple liver lesions
4pT211Neck mass, abdominal LN
5pT240Multiple liver lesions
6pT22Lung, abdominal LN
7pT224Spine, chest wall
Table 3. Univariate Assessment of Continuous Variables as Risk Factors for HCC Recurrence in Patients Who Received CsA as Main Immunosuppressant Drug
VariableMean ± SDP valueOR (95% CI)
CsA exposure (ng/mL)   
 Recurrence278.3 ± 86.4<0.0011.05 (1.02-1.08)
 No recurrence169.9 ± 33.3  
Cumulative steroids dosage (mg)   
 Recurrence3759.7 ± 1149.10.8871.00 (0.99-1.00)
 No recurrence3664.6 ± 1467.7  
Cumulative azathioprine dosage (mg)   
 Recurrence8675.0 ± 1237.40.8791.00 (0.99-1.00)
 No recurrence9387.5 ± 6678.9  
Recipient age   
 Recurrence51.1 ± 7.90.8030.99 (0.90-1.08)
 No recurrence50.6 ± 8.6  
AFP (Log10 ng/mL)   
 Recurrence1.62 ± 0.600.0323.19 (1.10-9.23)
 No recurrence1.06 ± 0.60  
Size of the main tumor (cm)   
 Recurrence2.8 ± 1.50.2551.42 (0.77-2.59)
 No recurrence2.3 ± 1.1  

The ROC analysis identified an optimal cutoff value of CsA exposure of 189.6 ng/mL to differentiate recurrent vs. nonrecurrent patients with a high accuracy (AUC : 0.937; 95% CI: 0.851-0.981). Patients with higher exposure to CsA (>189.6 ng/mL) had a 33.3% prevalence of recurrence (7 of 21), whereas no recurrence was observed in the 49 patients with lower CsA exposure (≤189.6 ng/mL). Sensitivity and specificity were 100% and 77.8%, respectively.

Mean CsA blood levels were higher in patients who developed HCC recurrence at each of the time points considered in the postoperative follow-up; in particular, these differences were statistically significant at days 14 (P = 0.006) and 30 (P = 0.030) and months 3 (P = 0.030), 6 (P = 0.017), and 12 (P = 0.026) after transplantation (Fig. 1). Mean CsA blood levels after 36 months were not comparable because there was only 1 patient in the recurrence group at this time point.

Figure 1.

Comparison between mean cyclosporine (CsA) blood levels in patients with and without hepatocellular carcinoma (HCC) recurrence after liver transplantation (LT).

Recurrence-free survival was significantly related to AFP blood levels (P = 0.032; OR: 3.19, 95% CI: 1.10-9.23). In fact, patients who experienced tumor recurrence showed higher AFP levels (mean log10 AFP: 1.62 ± 0.60 ng/mL) than those who did not (mean log10 AFP: 1.06 ± 0.60 ng/mL).

None of the other factors considered as continuous variables (age and size of the main tumor, cumulative dosage of steroids, and azathioprine administered in the first postoperative year) proved to be significantly related to recurrence-free survival rate (Table 3). Within the clinical and histologic characteristics considered as categorical variables (Table 4), only the presence of microscopic vascular invasion significantly correlated with a worse recurrence-free survival rate (P = 0.044; OR : 8.76, 95% CI: 1.05-72.8). Tumor recurrence was observed in 6 (20.7%) of 29 patients with microvascular invasion and in 1 (2.4%) of 41 patients without this finding.

Table 4. Univariate Assessment of Categorical Variables as Risk Factors for HCC Recurrence in Patients Who Received CsA as Main Immunosuppressant Drug
VariableNo. of recurrences/No. of patients (%)P valueOR (95% CI)
Recipient sex   
 Male6/62 (9.7)0.7280.69 (0.08-5.72)
 Female1/8 (12.5)  
Underlying liver disease   
 Viral-related7/60 (11.7)0.523
 Non–viral-related0/10 (—)  
Milan criteria   
 Unfulfilled1/13 (7.7)0.8321.25 (0.15-10.50)
 Fulfilled6/57 (10.5)  
pT stage   
 pT2-T37/46 (15.2)0.259
 pT10/24 (—)  
Type of tumor   
 Non-incidental7/52 (13.5)0.325
 Incidental0/18 (—)  
Histologic grading   
 III-IV4/29 (13.8)0.3671.99 (0.44-8.89)
 I-II3/41 (7.3)  
Microvascular invasion   
 Present6/29 (20.7)0.0448.76 (1.05-72.80)
 Absent1/41 (2.4)  
Immunosuppression schedule   
 CsA + steroids1/26 (3.8)0.2543.42 (0.41-28.49)
 CsA + steroids + azathioprine6/44 (13.6)  

None of the other factors considered as potential risks for tumor recurrence after LT (e.g., recipient sex, underlying liver disease, Milan criteria, pT stage, type of tumor and histologic grading, different immunosuppressive schedules—CsA, azathioprine, and steroids vs. CsA and steroids) showed a significant relationship with recurrence-free survival rate (Table 4). Stepwise multivariate analysis taking into account CsA exposure, AFP, and microvascular invasion showed that only CsA exposure was an independent risk factor for recurrence-free survival (P < 0.001), although AFP (P = 0.505) and microvascular invasion (P = 0.138) did not enter the picture.

The distribution of all the variables in patients with CsA exposure greater than 189.6 ng/mL (high exposure) and in patients with CsA exposure less than or equal to 189.6 ng/mL (low exposure) is reported in Table 5. In particular, no difference was found in the distribution of clinical, pathologic, and histologic factors in the 2 groups apart from a higher prevalence of pT2-T3 tumors in patients with higher exposure (18 of 46; 39.1%; P = 0.042); when matched with CsA exposure at the stepwise proportional hazard model, the pT as a risk factor for tumor recurrence showed no significance.

Table 5. Clinical and Histopathologic Features of Patients Who Underwent Transplantation According to Their Exposure to CsA
VariableNo. of patients with high CsA exposure*/All patients on CsA (%)P value
  • *

    CsA exposure above 189.6 ng/mL.

Milan criteria  
 Unfulfilled4/13 (30.8)0.947
 Fulfilled17/57 (29.8) 
pT stage  
 pT2-T318/46 (39.1)0.042
 pT13/24 (12.5) 
Type of tumor  
 Non-incidental17/52 (32.7)0.591
 Incidental4/18 (22.2) 
Histologic grading  
 III-IV9/29 (31.0)0.874
 I-II12/41 (29.3) 
Microvascular invasion  
 Present11/29 (37.9)0.341
 Absent10/41 (24.4) 
Immunosuppression schedule  
 CsA + steroids7/26 (26.9)0.871
 CsA + steroids + azathioprine14/44 (31.8) 

The incidence of acute and chronic rejection did not differ between patients who had higher (7 of 23 or 30%) or lower (8 of 49 or 16.3%) CsA exposure. Steroid boluses (1 g of methylprednisolone given intravenously) administered to treat acute rejection did not increase the risk of HCC recurrence (2 recurrences in 15 patients who received steroid boluses vs. 5 in 55 who did not; P = 1.0). No patient in this series was administered antibodies to lymphocytes in the treatment of rejection.


The selection of patients with HCC for liver transplantation has always represented a matter of debate. The frustrating experiences in the early days of liver transplantation with advanced tumors led physicians to outline strict selection criteria based on tumor macroscopic features that greatly improved the long-term results of transplantation in HCC patients.1, 7–9 However, the increasing number of potential candidates for liver transplantation who are denied a chance of cure because of even slight derangements from the established criteria requires a reconsideration of the selection of HCC patients and the definition of new prognostic parameters.10

Surprisingly, the possible role of a well-known risk factor for tumor growth, immunosuppression, in the outcome of liver transplantation for HCC has been poorly evaluated in the past. A previous study by our group showed that the amount of CsA administered in the first year after liver transplantation was an independent risk factor for HCC recurrence.4 However, it could be argued that the great variability in the absorption of the drug after oral administration might have influenced the results of that study; in fact, the intake of higher doses of CsA might not necessarily reflect a higher degree of immunosuppression once the blood level of the drug is tested. The measurement of the blood concentration of CsA with monoclonal fluorescent polarization immunoassay effectively reflects the degree of pharmacologic imunosuppression.11 Remarkably, the incidence of HCC recurrence in our series was much higher in those patients who had higher CsA exposure; this was the only independent prognostic determinant of tumor recurrence.

Unlike what we have previously reported, in the current study the multivariate analysis considered not only pathologic prognostic parameters such as the Milan criteria and TNM stage but also histologic features such as tumor grading and microvascular tumor invasion whose prognostic significance has been shown recently.12, 13

We provide further evidence on the key role of immunosuppression in tumor recurrence after liver transplantation; in particular, we recommend that in those patients undergoing transplantation for HCC who receive CsA-based immunosuppression, the exposure to the drug should not exceed the daily blood levels that identified here. On the basis of the increased clinical experience, the most commonly used immunosuppressant drugs can be handled safely with the aim of reducing their dosage to the effective minimum, which is of particular importance in patients who undergo transplantation for HCC. In our study, the incidence of acute and chronic rejection did not differ between patients with higher and lower CsA blood levels, which strengthens the importance of avoiding unnecessary overimmunosuppression.

Immunohistochemical markers detectable on the explanted liver were recently shown to allow a better definition of the aggressiveness of HCC and to correlate with the recurrence of disease14; immunosuppressive schedules might be tailored on individual liver transplant recipients on the basis of the evaluation of tumor biology. High-risk patients would probably benefit from keeping the calcineurin inhibitors-based immunosuppression as low as possible; for this purpose, it appears to be particularly promising the introduction in the immunosuppressive protocols of HCC patients of new immunosuppressant drugs that were shown to have anti-cancer effects.15, 16