Journal of Cancer Research and Clinical Oncology

, Volume 135, Issue 10, pp 1403–1412

Liver transplantation outcomes in 1,078 hepatocellular carcinoma patients: a multi-center experience in Shanghai, China


    • Liver Cancer Institute, Shanghai Medical School, Zhong Shan HospitalFudan University
  • Guang-Shun Yang
    • Dong Fang Gan Dan HospitalShanghai Second Military Medical University
  • Zhi-Ren Fu
    • Chang Zheng HospitalShanghai Second Military Medical University
  • Zhi-Hai Peng
    • Shanghai First HospitalMedical School of Jiaotong University
  • Qiang Xia
    • Ren Ji HospitalMedical School of Jiaotong University
  • Chen-Hong Peng
    • Rui Jing HospitalMedical School of Jiaotong University
  • Jian-Ming Qian
    • Hua Shan Hospital, Shanghai Medical SchoolFudan University
  • Jian Zhou
    • Liver Cancer Institute, Shanghai Medical School, Zhong Shan HospitalFudan University
  • Yang Xu
    • Liver Cancer Institute, Shanghai Medical School, Zhong Shan HospitalFudan University
  • Shuang-Jian Qiu
    • Liver Cancer Institute, Shanghai Medical School, Zhong Shan HospitalFudan University
  • Lin Zhong
    • Shanghai First HospitalMedical School of Jiaotong University
  • Guang-Wen Zhou
    • Rui Jing HospitalMedical School of Jiaotong University
  • Jian-Jun Zhang
    • Ren Ji HospitalMedical School of Jiaotong University
Original Paper

DOI: 10.1007/s00432-009-0584-6

Cite this article as:
Fan, J., Yang, G., Fu, Z. et al. J Cancer Res Clin Oncol (2009) 135: 1403. doi:10.1007/s00432-009-0584-6



To evaluate current selection criteria for patients undergoing liver transplantation (LT) in response to hepatocellular carcinoma (HCC), and to analyze the prognostic factors for successful transplantation.


We evaluated the outcome of 1,078 consecutive patients with HCC from the Shanghai Multi-Center Collaborative LT Group who underwent LT over a 6-year period. Clinicopathologic data for these patients were evaluated. The prognostic significance was assessed using Kaplan–Meier survival estimates and log-rank tests. Multivariate study with Cox’s proportional hazard model was used to evaluate the prognosis-relative aspects.


We determined that expansion of Milan criteria to include: a solitary lesion ≤9 cm in diameter, no more than three lesions with the largest ≤5 cm, a total tumor diameter ≤9 cm without macrovascular invasion, lymph node invasion and extrahepatic metastasis (referred to as the “Shanghai criteria”), resulted in overall survival (OS) and disease-free survival (DFS) rates that were similar to the Milan criteria. Multivariate analysis using the Cox proportional hazards regression model showed that the Child-Pugh-Turcotte classification (P = 0.010, 0.000), tumor differentiation (P = 0.001, 0.000), tumor size (P = 0.000, 0.000) and number (P = 0.014, 0.016), macrovascular invasion (P = 0.022, 0.000) and alpha-fetoprotein (AFP) levels (P = 0.031, 0.003) were independent predictors of OS and DFS, while post-LT chemotherapy (OS, P = 0.000) and tumor encapsulation (DFS, P = 0.038) were independent predictors of OS or DFS.


Shanghai criteria expanded the current criteria while maintaining similar survival.


Hepatocellular carcinomaLiver transplantationCriteriaPrognosis



Hepatocellular carcinoma


Liver transplantation






Computed tomography


Magnetic resonance imaging


World Health Organization


Mycophenolate mofetil




Transcatheter arterial chemoembolization


Percutaneous ethanol injection


Radio-frequency ablation


Overall survival


Disease-free survival






International Union against cancer


Hepatitis B virus


Hepatitis B surface antigen


Hepatitis C virus


Positron emission tomography


Hepatocellular carcinoma (HCC) is the third most common cause of death from cancer worldwide, and 55% of the new cases occur in China (Parkin et al. 2005). Most HCC patients in China have pre-existing HBV infection, which differs from the United States, Europe and Japan. Surgical resection is considered as the optimal treatment approach, but not more than 5% of patients qualify for surgery in China, and there is a high rate of recurrence even after “curative” resection (Tang 2001; Tang et al. 2004). Liver transplantation (LT) offers significant survival, but is limited by a paucity of donors (Said et al. 2007). During the past decade the use of LT has rapidly expanded in China (Rakela and Fung 2007) to include 30–50% of HCC patients. However, LT in China faces several challenges including debate over the brain-death law, a shortage of liver donors, and the absence of preventive and therapeutic strategies to prevent the recurrence of HCC (Huang 2007). The selection of patients with HCC for LT is based on prognostic systems that use tumor size as the main prognostic factor. The Milan system limits LT to patients with one lesion ≤5 cm in diameter or not more than three lesions, none of which exceeds 3 cm (Mazzaferro et al. 1996). The use of this system has resulted in a post-LT survival rate of 70% in 5 years. However, this system may be a little rigorous, because a large number of patients who would benefit from LT may be unnecessarily excluded. Researchers at UCSF expanded the Milan criteria and obtained survival rates similar to those using the Milan criteria (Yao et al. 2001, 2007). Based on our previous study of 251 HCC patients at the Liver Cancer Institute, Fudan University, we also found that the survival rate of patients with larger HCC tumor size who received LT was similar to groups that used the Milan criteria (Fan et al. 2006). To further evaluate the selection criteria and analyze prognostic factors, we analyzed the outcome of 1,078 consecutive patients with HCC in the Shanghai Multi-Center LT Collaborative Group who underwent LT over a 6-year period in seven hospitals. This population contains approximately 30% of all patients who undergo LT in China every year.

Patients and methods

A total of 1,078 HCC patients who underwent LT between April 2001 and May 2007 at seven Shanghai LT centers were included in this study. The inclusion and exclusion criteria of the patient cohort included distinctive pathological diagnosis of HCC with complete clinicopathological and follow-up data. Most of the patients underwent classic orthotropic LT, four patients underwent living donor LT, eight patients underwent combined LT and kidney transplantation, and 12 patients were retransplanted because of graft failure without tumor recurrence after LT. Eleven patients had incidental HCC, which was not apparent on imaging before LT. Follow up after transplantation was defined as the time between transplantation and death or the time of last follow up (as of May 2007). A total of 81 patients were lost to follow up within 2 years. Follow up ranged from 1 day to 73.2 months, with a mean of 35.3 months. The rate of complete follow up of all patients was 92.5%. Preoperative tumor stage was based on three abdominal imaging studies, including ultrasonography (US), computed tomography (CT), and/or magnetic resonance imaging (MRI), showing consistent results and performed within 3 months before LT. Positron emission tomography/CT (PET/CT) was used for preoperative tumor staging in 67 patients. Disparities between the imaging studies were resolved by assigning tumor stage using the procedure that showed the most characteristic HCC features with the highest technical quality based on thorough review by a transplantation hepatologist, transplantation surgeon, and a radiologist. Accuracy of the imaging procedures in the staging of known HCC before LT is summarized in Table 1. For patients with known solitary or multicentric HCC, the imaging sensitivity was recalculated using all tumor nodules identified in the liver explants. Individual tumor nodules were classified according to whether or not each imaging study performed on the patient within 3 months before LT identified the nodule. Extrahepatic metastasis was excluded based on chest and abdominal CT or MRI, and bone scintigraphy performed within 1 month before LT. Fine needle biopsy or aspiration of the tumor generally was avoided in the diagnostic evaluation of liver lesions because of the potential risk of bleeding and tumor seeding along the biopsy tract. Tumor recurrence was defined as: (1) CT scan, MRI, or plain film evidence of recurrent tumor (or metastatic disease) plus the appropriate clinical picture with biopsies (as described below); and/or (2) autopsy (Vivarelli et al. 2005; Perez-Saborido et al. 2007). Patients were followed for tumor recurrence using CT scans and α-fetoprotein levels (AFP) every 6 months for 5 years and yearly thereafter, or as the clinical situation dictated. Suspicious lesions in the liver or lung were resected or biopsied. Bone lesions were not routinely biopsied but were observed for bone pain and progression of growth. If the AFP level began to rise before the recurrence was documented, the date of recurrence was taken as the time that the AFP level began to increase once tumor recurrence was confirmed by imaging scans. HCC diagnosis was based on the World Health Organization (WHO) criteria (Wittekind 2006). Tumor differentiation was defined according to the Edmondson grading system (Wittekind 2006). The average waiting time for a liver donor was 4.7 months (ranging from 1 week to 23 months). During the waiting period, 164 patients (15.2%) had transcatheter arterial chemoembolization (TACE), percutaneous ethanol injection (PEI) or radio-frequency ablation (RFA), and 85 patients (7.9%) had a combination of TACE, PEI and/or RFA before LT. All of the cadaveric donors were obtained from brain death or no- heart-beating donors with consent for voluntary organ donation. All of the living organs were donated from spouses or lineal relatives with informed consent.
Table 1

Accuracy of radiologic procedures in the staging of known HCC before LT





Sensitivity of main lesion (%)

99.0 (n = 1064)

99.0 (n = 1070)

98.5 (n = 67)

Accuracy of the size of main lesion (%, error < 1 cm)




Sensitivity of satellite lesions(%)

94.6 (n = 481)

89.4 (n = 481)

94.7 (n = 38)

Accuracy of preoperative tumor staging (%)

82.1 (n = 1064)

80.4 (n = 1070)


Underestimate rate (%)




Overestimate rate (%)




HCC hepatocellular carcinoma, LT liver transplantation, CT computed tomography, MRI magnetic resonance imaging, US ultrasonography, PET positron emission tomography

Histopathologic studies

The explanted liver was examined by an experienced histopathologist. The number of tumors, maximal tumor diameter, the presence or absence of microvascular invasion, and histologic grade (based on the modified Edmondson criteria for tumor differentiation, grade 1: well-differentiated; grade 2: moderately differentiated; and grade 3: poorly differentiated) (Wittekind 2006) were recorded. The total tumor diameter for patients with multiple tumor nodules was calculated as the sum of the maximal diameter of each lesion in centimeters. If a suspicious perihilar lymph node was identified at the time of laparotomy, a frozen-section was examined histologically to exclude the presence of tumor. The pathologic tumor stage was initially reported using the pathological tumor-node-metastasis (pTNM) classification proposed by the International Union Against Cancer (UICC) (Wittekind and International Union Against Cancer 2005).

Immunosuppressive regimens

Immunosuppressive therapy after LT consisted of a triple drug regimen of cyclosporine or tacrolimus combined with corticosteroids and/or mycophenolate mofetil (MMF). Immunosuppression was started during surgery with 500 mg methylprednisolone, followed by taper, 240–40 mg per day over 6 days. Maintenance prednisone (5–20 mg daily) was used based on the perceived degree of patient debilitation. Tacrolimus was administered orally 12 h after LT at 0.05–0.1 mg/kg every 12 h. The dose was adjusted to reach target levels of 10–15 ng/mL plasma during the first 14 days. Levels were maintained within a range of 6–10 ng/mL during the second 14 days, and 5–8 ng/mL thereafter. MMF (0.25–0.5 mg, po, bid) was used in patients if tacrolimus did not reach the target level. A total of 214 patients were switched to sirolimus (SRL, Wyeth-Ayerst Research, Philadelphia, PA, USA)-based immunosuppression therapy. Tacrolimus and MMF were stopped once SRL therapy was initiated (Zhou et al. 2006). SRL was delayed until 1 month after LT to avoid affecting wound healing. Patients were given SRL at the initial loading dose of 3 mg/m2. SRL doses were then adjusted to achieve a steady-state whole-blood level of approximately 5–8 ng/mL. SRL oral solution was administered once daily in the morning, following dilution with water or orange juice. The dose was reduced or increased to 0.5 mg/day if the plasma level was higher than 8 ng/mL or lower than 5 ng/mL, respectively.

Postoperative chemotherapy

After operation, 299 patients accepted 1–6 cycles (mean: 3.1 cycles) of systemic chemotherapy (21–58 days after transplantation, mean 28 days). A 3-day chemotherapeutic regimen was prescribed with intravenous floxuridine 300 mg/m2 q d for 3 days, cisplatin 20 mg/m2 q d for 2 days, and doxorubicin 20 mg/m2 on the first and third days. Chemotherapy was repeated in 1 month intervals. If tumor recurrence or metastasis was confirmed, systemic chemotherapy was stopped and appropriate treatment such as TACE, PEI, RFA, and radiotherapy was provided.

Statistical analysis

Statistical analyses were performed by SPSS 15.0 for windows (SPSS, Chicago, IL). The Kaplan–Meier product-limit method with log rank test was used to evaluate survival rate. We report all-cause mortality for up to the last observation after LT. All the cause mortality was analyzed instead of only HCC-related mortality to avoid underestimating the impact of HCC on mortality. Overall survival (OS) was defined as the interval between LT and death or the last observation. The data was censored for patients who survived at the last observation. The data also was censored at the time of the last observation for patients who underwent re-LT. Disease-free survival (DFS) was calculated by considering any patients dead or living with recurrence as the event. The data were censored for living patients without signs of recurrence. Univariate or multivariate analysis using the Cox proportional hazards regression model was performed to reveal factors that might affect the survival rate. The results are reported as hazard ratios with 95% confidence intervals. P < 0.05 was considered statistically significant in all analyses.


Baseline characteristics of the patients

This study involved 970 men and 108 women, aged 15–82 years, with a median age of 49.0 years. Of those studies, 975 patients (90.5%) had liver cirrhosis. According to the Child-Pugh-Turcotte (CPT) classification of cirrhosis, 267 patients (24.8%) were Child’s class C (CPT 10–15), 304 (28.2%) were Child’s class B (CPT 7–9), and 507 (47.0%) were Child’s class A (CPT 5–6). Hepatitis B virus (HBV) was the most common etiology of cirrhosis, accounting for 915 (84.9%) cases. There were 30 patients (2.8%) with Hepatitis C virus (HCV), and four (0.4%) were coinfected with HBV and HCV. AFP levels were obtained within 3 months before LT. Normal AFP levels (≤20 ng/ml) were observed in 378 (35.1%) patients. One hundred and nine patients (10.1%) had previous hepatic resections for HCC and were known to have recurrent HCC. Baseline characteristics of the 1,078 patients are summarized in Table 2.
Table 2

Baseline characteristics of 1,078 HCC patients undergoing LT

Median age (year): 49.0 (range, 15–82)

 Age > 52


Sex (M/F): 970/108

 Etiology of liver disease

Number (%)




915 (84.9%)


30 (2.8%)

 HBV and HCV

4 (0.4%)


975 (90.4%)



109 (10.1%)


164 (15.2%)

 Combined (LR/TACE/RFA/PEI)

85 (7.9%)



CPT score


267 (24.8%)


304 (28.2%)


507 (47.0%)

AFP level (ng/ml)


381 (35.3%)


274 (25.4%)


423 (39.2%)

HCC Hepatocellular carcinoma, LT liver transplantation, HBV Hepatitis B virus, HCV Hepatitis C virus, LR liver resection, TACE transcatheter arterial chemoembolization, PEI percutaneous ethanol injection, CPT Child-Pugh-Turcotte, AFP alpha-fetoprotein

Histopathologic tumor characteristics in the liver explants

According to the sixth edition of TNM classification of UICC, 352 (32.7%) patients were classified as TNM-I, 300 (27.8%) were TNM-II, and 426 (39.5%) were TNM-III. Tumor differentiation was histologically graded I–II in 748 patients (69.4%), while 330 (30.6%) patients had poorly differentiated tumors of grade (III–IV). Vascular invasion was found in 339 patients (31.4%), including 213 (19.8%) with macrovascular invasion and 126 (11.7%) with microvascular invasion. All patients with macrovascular invasion had involvement of the portal vein or hepatic vein, either main trunk or first class branch. Evidence of lymph node invasion was found in 65 patients (6.0%). One hundred and fifty-nine patients (14.8%) had more than three tumor nodules, 140 (13.0%) patients had bilobe invasion, and 884 (82.0%) patients had incomplete tumor capsules. The explant histopathologic characteristics of HCC are summarized in Table 3.
Table 3

Histopathologic features of liver explants in 1,078 HCC patients undergoing LT


Number (%)

Modified pathologic stage


352 (32.7)


300 (27.8)


426 (39.5)

Size of main lesion (cm)


717 (66.5)


184 (17.1)


177 (16.4)

Total size of all lesions(cm)


771 (71.5)


48 (4.5)


259 (24.0)

Tumor number


597 (55.4)


322 (29.9)


159 (14.8)

Histologic grades


748 (69.4)


330 (30.6)

Vascular invasion


 Macro-vascular invasion

213 (19.8)

 Micro- vascular invasion

126 (11.7)


739 (68.6)

Tumor satellite





Lymph node metastasis


65 (6.0)


1013 (94.0)



194 (18.0)


884 (82.0)

Tumor site

 Left lobe

45 (4.2)

 Right lobe

893 (82.8)


140 (13.0)

Criteria (n = 969)


394 (40.7)


489 (50.5)


570 (58.8)

 Exceed Shanghai

399 (41.2)

HCC hepatocellular carcinoma, LT liver transplantation, TNM tumor-node-metastasis. The patients who had prior resection of HCC (n = 109) were excluded in the development of criteria for selection of patients with HCC for transplant

Survival and recurrence after LT according to different criteria

For the entire cohort of 1,078 patients, the OS and DFS rates were 79.9 and 69.3% in 1 year, and 50.2 and 43.9% in 5 years, respectively. HCC recurrence was observed in 265 patients (24.6%) after LT. The mean recurrence time was 10.0 months (range from 1–47 months). Death occurred in 344 patients (31.9%), of which 49 patients (4.6%) died within 1 month after LT and 153 patients (14.2%) died from tumor recurrence. Patients with previous hepatic resections for HCC before LT (n = 109) were excluded from the indication analysis thereafter. The 1 and 5-year OS were 86.6 and 78.8%, while the one and 5-year DFS were 65.2 and 55.7%, respectively (n = 394, 40.7%) using the Milan criteria (Mazzaferro et al. 1996). The 1 and 5-year OS were 86.2 and 79.2%, while the one and five-year DFS were 64.7 and 55.8%, respectively (n = 489, 50.5%) using the UCSF criteria (solitary tumor <6.5 cm, or three or fewer nodules with the largest lesion <4.5 cm and total tumor diameter <8 cm, without gross vascular invasion) (Yao et al. 2001). In our study (Shanghai criteria), the Milan criteria were expanded to include solitary lesions ≤9 cm in diameter, or no more than three lesions, the largest ≤5 cm with a total tumor diameter ≤9 cm, without macrovascular invasion, lymph node invasion and extrahepatic metastasis. In this case, the 1 and 5-year OS were 85.8 and 78.1%, while the 1 and 5-year DFS were 59.9 and 52.6%, respectively (n = 570, 58.8%), which did not differ significantly from the Milan criteria. One hundred and four patients (10.7%) of the Shanghai criteria group had recurrent HCC versus 64 patients (6.6%) in the Milan criteria group (P > 0.05). Of the patients who exceeded the Milan, but met the Shanghai criteria (n = 176, 18.2%), the one and five-year OS were 82.7 and 75.9%, while the 1 and 5-year DFS were 51.2 and 46.1%, respectively. There was no difference in OS and DFS between patients who met the Milan criteria and those who exceeded the Milan, but met the Shanghai criteria (P = 0.127, 0.062) (Fig. 1).
Fig. 1

Comparison of overall and disease-free survival between HCC patients undergoing liver transplantation divided by Milan or Shanghai criteria. The OS and DFS were significantly lower for the patients exceeded Milan (a, b) or Shanghai (c, d) criteria than for the patients who met the criteria (P = 0.000). There was no difference in OS and DFS between the patients met Milan criteria and the patients exceeded-Milan-met-Shanghai criteria (e, f, P > 0.05)

Predictors for survival

Univariate analysis was performed on 18 tumor-related parameters and additional clinical or demographic characteristics to determine predictors for OS and DFS. Significant predictors for OS and DFS included CPT classification (A/B vs. C), tumor differentiation (I–II vs. III–IV), total tumor size (≤9 vs. >9 cm), tumor number (≤3 vs. >3), tumor satellite, tumor encapsulation, lymph node invasion, macrovascular invasion, TNM stage (I–II vs. III) and AFP (≤300 vs. >300 ng/ml). Post-LT chemotherapy (no vs. yes) was a predictor for OS. Sex, age, liver cirrhosis, pretreatment, hepatitis (HBV or HCV), tumor site (single lobe vs. bilobe), microvascular invasion and SRL were not predictors for OS or DFS. Multivariate analysis indicated CPT classification (P = 0.010, 0.000), tumor differentiation (P = 0.001, 0.000), tumor size (P = 0.000, 0.000) and number (P = 0.014, 0.016), macrovascular invasion (P = 0.022, 0.000) and AFP level (P = 0.031, 0.003) were independent predictors for OS and DFS, while post-LT chemotherapy (OS, P = 0.000) and tumor encapsulation (DFS, P = 0.038) were independent predictors for OS or DFS, respectively. The results, including hazard ratio with 95% confidence intervals are summarized in Tables 4 and 5.
Table 4

Univariate analyses of factors associated with survival and recurrence




Hazard ratio (95%CI)

P value

Hazard ratio (95%CI)

P value

Sex (female vs. male)

0.73 (0.49–1.09)


0.62 (0.44–0.89)


Age, years (≤52 vs. >52)

0.88 (0.70–1.10)


0.88 (0.74–1.07)


Liver cirrhosis (no vs. yes)

1.09 (0.76–1.57)


1.02 (0.75–1.39)


Pretreatment (no vs. yes)

0.93 (0.74–1.17)


0.95 (0.78–1.16)


CPT (A/B vs. C)

1.61 (1.28–2.02)


1.60 (1.32–1.95)


Hepatitis (no vs. yes)

1.20 (0.87–1.66)


1.26 (0.95–1.68)


HBV (positive vs. negative)

1.11 (0.82–1.49)


1.21 (0.93–1.57)


HCV (positive vs negative)

1.25 (0.69–2.28)


0.99 (0.56–1.76)


Tumor differentiation (I–II vs. III–IV)

1.66 (1.33–2.06)


1.91 (1.58–2.30)


Total tumor size (cm) (≤9 vs. >9)

2.24 (1.82–3.00)


2.39 (1.91–2.98)


Tumor number (≤3 vs. >3)

1.83 (1.41–2.37)


1.88 (1.50–2.37)


Tumor satellite (no vs. yes)

1.31 (1.06–1.62)


1.31 (1.09–1.58)


Tumor site (single lobe vs. bilobe)

1.26 (0.92–1.73)


1.18 (0.90–1.56)


Tumor encapsulation (complete vs. none)

0.55 (0.39–0.78)


0.62 (0.47–0.82)


Lymph node invasion (no vs. yes)

1.76 (1.22–2.54)


1.81 (1.31–2.52)


Macrovascular invasion (no vs. yes)

1.81 (1.43–2.30)


2.16 (1.76–2.64)


Microvascular invasion (no vs. yes)

1.03 (0.74–1.44)


1.10 (0.83–1.45)


TNM stage (I–II vs. III)

1.56 (1.23–1.97)


1.66 (1.35–2.05)


AFP(ng/ml) (≤300 vs. >300)

1.41 (1.14–1.74)


1.49 (1.24–1.80)


Post-LT chemotherapy (no vs. yes)

0.67 (0.51–0.87)


0.81 (0.65–1.00)


SRL (no vs. yes)

1.09 (0.84–1.43)


0.96 (0.75–1.22)


OS overall survival, DFS disease-free survival, 95%CI 95% confidence interval, AFP alpha-fetoprotein, LT liver transplantation, TNM tumor-node-metastasis, SRL sirolimus, OS overall survival, DFS disease-free survival, AFP alpha-fetoprotein, CPT Child-Pugh-Turcotte, HBV Hepatitis B virus, HCV Hepatitis C virus; Univariate analysis, Cox proportional hazards regression model

Table 5

Multivariate analyses of factors associated with OS and DFS


Hazard ratio (95% CI)

P value


 CPT (A/B vs. C)

1.36 (1.08–1.72)


 Tumor differentiation (I–II vs. III–IV)

1.46 (1.16–1.84)


 Total tumor size (cm) (≤9 vs. >9)

1.74 (1.31–2.32)


 Tumor number (≤3 vs. >3)

1.50 (1.09–2.08)


 Tumor satellite (no vs. yes)

1.22 (0.93–1.58)


 Tumor encapsulation (complete vs. none)

0.77 (0.54–1.11)


 Lymph node invasion (no vs. yes)

1.05 (0.71–1.55)


 Macrovascular invasion (no vs. yes)

1.38 (1.05–1.83)


 TNM stage (I–II vs. III)

1.13 (0.86–1.49)


 AFP(ng/ml) (≤300 vs. >300)

1.27 (1.02–1.58)


 Post-LT chemotherapy (no vs. yes)

0.57 (0.42–0.76)



 CPT (A/B vs. C)

1.56 (1.27–1.91)


 Tumor differentiation (I–II vs. III–IV)

1.62 (1.33–1.98)


 Total tumor size (cm) (≤9 vs. >9)

1.71 (1.33–2.18)


 Tumor number (≤3 vs. >3)

1.41 (1.07–1.86)


 Tumor satellite (no vs. yes)

1.24 (0.99–1.56)


 Tumor encapsulation (complete vs. none)

0.74 (0.56–0.98)


 Lymph node invasion (no vs. yes)

1.02 (0.72–1.45)


 Macrovascular invasion (no vs. yes)

1.56 (1.23–1.96)


 TNM stage (I–II vs. III)

1.04 (0.82–1.31)


 AFP (ng/ml) (≤300 vs. >300)

1.33 (1.10–1.86)


OS overall survival, DFS disease-free survival, 95%CI 95% confidence interval, CPT Child-Pugh-Turcotte, TNM tumor-node-metastasis, AFP alpha-fetoprotein; Multivariate analysis, Cox proportional hazards regression model. Variables were adopted for their prognostic significance by univariate analysis (P < 0.05)


Chronic infection with HBV is the primary cause of HCC in high-risk areas including China and sub-Saharan Africa, whereas HCV plays a more prominent role in western countries and Japan (Chen et al. 2006). China has the biggest population of HBV carriers in the world; estimated to be between 20–30% of 1.3 billion people (Shi et al. 2005). As a result, liver cirrhosis and newly diagnosed HCC in China account for half of the total annual cases globally (Parkin et al. 2005). Of the approximately 300,000 new HCC patients diagnosed each year, <10% undergo curative liver resection. Most of the patients are non-resectable due to advanced tumor stage or poor liver preservation, which is why there are more HCC patients than patients with benign diseases on waiting lists for LT. In mainland China approximately half of the LTs were performed in patients with HCC, in sharp contrast with Europe and the United States (de Villa and Lo 2007). The main factor that limits the use of LT for HCC is the scarcity of donors. For this reason, we must reserve LT for patients with a good prognosis as well and those meeting the high requirements of HCC patients.

In 1996, Mazzaferro et al. reported excellent results using strict selection criteria based on tumor size (Milan criteria) (Mazzaferro et al. 1996). For the past decade, the Milan criteria for HCC have been the paradigm for selecting LT candidates. Although the Milan criteria have been widely accepted and adopted by UNOS, many authors argue that the criteria are too restrictive (Duffy et al. 2007; Onaca et al. 2007; Soejima et al. 2007; Todo et al. 2007). Yao et al. and others have provided evidence that it is possible to achieve satisfactory patient survival using expanded criteria, such as the UCSF criteria (Yao et al. 2001, 2002, 2007; Yao 2007). Our pervious retrospective study of the LT experience for 251 HCC patients also found that there was no significant difference in OS and DFS between the Milan and Shanghai criteria (solitary lesions ≤9 cm in diameter or not more than three lesions with the largest ≤5 cm, with a total tumor diameter ≤9 cm, without macrovascular invasion, lymph node invasion and extrahepatic metastasis) (Fan et al. 2006).

To further test the Shanghai criteria, we analyzed 1,078 valid cases of LT for HCC patients in Shanghai. Of these 1,078 patients, the OS and DFS rates were 80 and 69% in 1 year, 50 and 44% in 5 years, respectively, which seems lower than other centers in the Unite States and Europe (Duffy et al. 2007; Mourad et al. 2007; Rossi et al. 2007; Vitale et al. 2007; Zieniewicz et al. 2007). When analyzed more intensively, we found the overall recurrence rate was 24.6%, and around half of the deceased died of HCC recurrence. There are several possible reasons for the high recurrence rate post-LT in our patient cohort: (1) over 40% of the patients exceeded even the Shanghai criteria and around 20% had macrovascular invasion (either main trunk or first class branch of portal vein or hepatic vein). As we know, the early experience in mainland China mirrors that in Europe and North America. Most of the late stage HCC patients in this study were enrolled in 2001 and 2002, which was the very beginning stage of all LT centers in Shanghai. (2) The steep learning curve at the primary stage of LT. There were 49 (4.6%) patients who died within 1 month after LT. Many of these patients died from technical problems related to the surgery itself or from post-operative complications other than tumor recurrence. When stratified, we found that 394 patients fell within the Milan criteria, 489 within the UCSF criteria, and 570 within the Shanghai criteria, respectively. Differences in OS and DFS between these three groups did not achieve statistical significance. Furthermore, the 5-year cumulative recurrence rate did not differ significantly between the Milan and Shanghai groups (6.6 vs. 10.7%, P > 0.05). The 176 patients that met the Shanghai criteria and exceeded the Milan criteria had similar levels of OS and DFS as those that met the Milan criteria. Thus, use of the Shanghai criteria resulted in a good post-transplant outcome with a similar risk for tumor recurrence, and resulted in the treatment of more patients (44.7%) than did the Milan criteria. Although expanded criteria based on tumor size for LT has several problems, such as the dependency on radiologic imaging (Yao et al. 2001) that may underestimate tumor size or number, we still are inclined to use Shanghai criteria in our group for the following reasons. First, due to the increased numbers of HCC patients on the waiting list in China and the severe shortage of liver donors, use of the Shanghai criteria balances the needs of more HCC patients and the demands for excellent survival. Second, it has become apparent over the last 10–20 years that a non-selective approach to LT in patients with HCC resulted in sub-standard survival. In addition, macrovascular invasion and lymph node metastasis have been widely accepted as absolute contraindications to LT for many years (Schwartz et al. 1995; Bismuth et al. 1999; De Carlis et al. 2003; Hertl and Cosimi 2005; Duffy et al. 2007; Vitale et al. 2007; Zieniewicz et al. 2007; Mazzaferro et al. 2008). As a matter of fact, in China, regulations have lagged behind medical progress, with LT expanding in an underregulated manner at its very beginning (e.g., 5–8 years ago), there was a trend to perform LT for all HCC patients no matter how late the tumor stage, except for the patients with extrahepatic metastasis (Huang 2007; Huang et al. 2008). This explains the large numbers of late stage patients in this study. Although not officially adopted for LT, the Shanghai criteria have been accepted by many LT centers in China today. Third, tumor size and number can be easily obtained and standardized before LT. With the progress in imaging technology, the accuracy of preoperative HCC diagnosis has been greatly increased. In our group, since the 64-row multidetector CT and PET/CT were introduced for preoperative tumor staging, our ability to detect HCC nodules has increased considerably. Fourth, no other satisfactory prognostic factor is currently available.

In this study, we analyzed tumor-related parameters and clinical or demographic characteristics to determine predictors for OS and DFS. We found that CPT classification, tumor differentiation, tumor size, tumor number, tumor satellite, tumor encapsulation, lymph node invasion, macrovascular invasion, TNM stage and AFP level were all predictors for OS and DFS, in agreement with previous reports. Other than tumor size and number, those pathological factors cannot easily be obtained due to the need for a pre-LT biopsy, with the potential risk of abdominal bleeding and tumor seeding. In this series, most of the lymph node metastases were detected during or after operation. To enhance the survival rate, it is very important to improve the technology to exclude the patients with lymph node metastases before LT. Microvascular invasion was found in 126 patients, but not reached significant difference in OS and DFS comparisons. Although its prognostic value is not well accepted (Tang 2001; Marubashi et al. 2007), AFP showed good predictive value in this study, and was also a good indicator of tumor recurrence in AFP-positive patients. Recently, AFP has been accepted as an indicator to select the HCC patients for LT (Zheng et al. 2008). However, it is still arguable because most of HCV-related HCCs and around 40% HBV-related HCCs are AFP-negative, and the cut-off value of AFP is hard to decide for those AFP-positive HCCs (Debruyne and Delanghe 2008).

We also reviewed the effects of chemotherapy and SRL on tumor recurrence (Di Benedetto et al. 2007). Only prophylactic chemotherapy post-LT showed benefit with respect to OS. However, these results are still questionable due to defects in the retrospective study. To draw credible conclusions, well-designed prospective random clinical trials will be needed in the future.

To our knowledge, this is the largest group of HCC patients ever evaluated to compare different inclusion criteria for LT candidates in China. The Shanghai criteria offer the benefit of LT to an additional 45% of HCC patients who otherwise would have been excluded from LT under the more restrictive Milan criteria. Expansion of the inclusion criteria in the Shanghai criteria did not have a significant effect on the survival and recurrence rates. Of course, a majority of the cases analyzed in this study are HBV+, thus, the Shanghai criteria may only be specific to patients with hepatitis B-related HCC. In addition, we realize that this study was based on pathology findings rather than on preoperative imaging. Although major progress in imaging technology has been made in recent years, the likelihood of underestimating the tumor stage cannot be ignored. In future studies, it will be very important to investigate predictors other than tumor size and pathologic factors. More effective strategies to prevent tumor recurrence after LT for HCC patients must also be intensively explored.


This study was supported by combined grants from National Natural Science Foundation of China (No.30873039, 30571801), Shanghai Science and Technology Development Funds (No.06QA14012, No.054119530), Foundation of Shanghai Science Technology Commission (No. 07JC14010, 06xD14004, 044119608 and 07SP07003), the National Key Sci-Tech Special Project of China (No.2008ZX10002-022), and the Program for Excellent Disciplinary Leaders of Shanghai Health Bureau (No.LJ06004). We thank Roche for assistance in the organization of the Shanghai Multi-Center LT Collaborative Group.

Copyright information

© Springer-Verlag 2009