• Cirrhosis;
  • hepatocellular carcinoma;
  • liver transplantation;
  • outcomes


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. References

Liver transplantation (LT) for cirrhotic/Hepatocellular carcinoma (HCC) is associated with reduced survival in patients with poor histological features. Preoperative levels of alphafetoprotein (AFP) could predict negative biological features. AFP progression could be more relevant than static AFP levels in predicting LT outcomes. A total of 252 cirrhotic/HCC patients transplanted between 1985 and 2005 were reviewed. One hundred fifty-three patients were analyzed, 99 excluded (for nonsecreting tumors and/or salvage transplantation). Using receiver operating characteristics analysis for recurrence after LT, ‘progression’ of AFP was defined by >15 μg/L per month before LT. A total of 127 (83%) were transplanted under and 26(16%) over this threshold. After 45 months of follow-up (median), 5-year overall survival (OS) and recurrence free-survival (RFS) were 72% and 69%, respectively. Five-year survival in the progression group was lower than the nonprogression group (OS 54% vs. 77%; RFS 47% vs. 74%). Multivariate analysis showed progression of AFP >15 μg/L per month and preoperative nodules >3 were associated with decreased OS. Progression group and age >60 years were associated with decreased RFS. Male gender, progression of AFP and size of tumor >30 mm were associated with satellite nodules and/or vascular invasion. In conclusion, increasing AFP >15 μg/L/month while waiting for LT is the most relevant preoperative prognostic factor for low OS/DFS. AFP progression could be a pathological preoperative marker of tumor aggressiveness.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. References

Hepatocellular carcinoma (HCC) is the fifth most common cancer worldwide, and the third most common cause of cancer-related deaths (1). Underlying liver cirrhosis is present in a majority of patients (2). By treating both cirrhosis and cancer, liver transplantation (LT) may be curative and its results are superior to that of resection in selected patients (3). Indeed, the adoption of the restrictive selection criteria based on the number and on the size of HCC has produced a survival rate that is very similar to that of transplanted patients without cancer (3–7). Transplantation of patients with up to three nodules with a maximal diameter of 3 cm or a single nodule of a maximal diameter of 5 cm as defined by Milan Criteria represent the current selection standard (4). Regardless the criteria chosen to select patient for transplantation, the major aim of prelisting assessment is to detect extrahepatic metastases that are responsible of failure of the transplant strategy. Unfortunately, the best predictive factors of extrahepatic infraclinic metastases are only available after LT through the histological study of the whole liver specimen. In fact, most reports quote histological features (i.e. vascular invasion, tumoral differentiation, satellite nodules) as best predictors of the risk of recurrence following transplantation. This could explain why some transplanted patients within the Milan criteria have tumoral recurrence and poor survival and vice versa (5,8,9).

About half of patients with HCC have abnormal alphafoetoprotein (AFP) levels (10,11). In this specific population, observation of AFP levels is commonly used to monitor the tumor evolution. Interestingly, elevated AFP levels have been reported as a poor prognostic factor but no clear prognostic cut off level has been defined (10). Considering the AFP as an indicator of tumor doubling time in secreting HCC, we hypothesized that the progressive elevation of AFP could have more prognostic importance, compared to its static value before LT. The aim of this study was to evaluate the outcome of LT for AFP secreting HCC in cirrhotic patients according to the evolution of AFP before LT (a factor available at the time of decision for transplantation). Operative specimens were also analyzed to evaluate the relationship between the evolution of AFP and histological features.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. References

From November 1985 to October 2005, 290 consecutive cirrhotic patients with HCC were listed for LT in our department.

Diagnosis of HCC

In the pretransplant setting, the diagnosis was made either by protected biopsy of liver nodule (12) or according to Barcelona Criteria (13). All transplanted patients had histologically proven HCC from liver explants.

The severity of cirrhosis was classified according to Childs' classification. The number and size of HCC nodule(s) were assessed by abdominal computed tomography (CT) scan and liver ultrasonography. Levels of AFP were regularly measured.

Selection criteria for LT in HCC

The decision of listing patients for LT was made through a multidisciplinary meeting including liver surgeons, hepatologists and radiologists. Formal contraindications for LT listing included macroscopic portal tumoral thrombosis, extrahepatic tumor and history of other malignant tumors in the last 5 years. The selection of patients for transplant was initially based on our own criteria (3) (<3 nodules strictly inferior to 3 cm), and subsequently on Milan criteria (4) (≤3 nodules with a maximal diameter of 3 cm or one nodule with a maximal diameter of 5 cm). AFP levels were not considered for LT listing. From 2001, the policy of primary resection of HCC with secondary LT in cases of intrahepatic recurrence (salvage transplantation) was rarely applied in our department due to its poor results (14).

Therapeutical strategy

Except for child C patients, an arterial liver chemoembolization (TACE) was done, before patients were listed for LT (15). This step was performed, to treat the tumor and to eventually improve the mapping of HCC nodules through lipiodol uptake on control CT scan.

In patients without TACE, the decision for LT listing was made according to the result of initial morphological assessment. For patients treated with TACE, an eventual reappraisal of the number of nodule(s) was made, based on the number of lipiodol spots on the CT scans performed 1 month after the treatment. In child A–B patients with no or low lipiodol uptake and up to two subcapsular HCC lesions (<3 cm), a local treatment (radiofrequency or cryotherapy) was made. The policy of ‘bridge LT’ that implies surgical resection of HCC on waiting list was rarely performed (16).

All patients were followed every 3 months. A CT scan and/or liver ultrasonography and liver biological assessment were performed at each consultation. A dosage of AFP was done at least monthly during the waiting time period. In cases of disease progression, the follow-up was changed to monthly. If feasible, a repeat TACE or radiofrequency (RF) was performed for patients who were noted to have increasing AFP levels and/or radiologically proven disease progression. The patient was removed from the LT waiting list only if extrahepatic disease and/or tumoral portal thrombosis was demonstrated.

In these series, graft allocation was independent of the severity of the underlying liver disease and patients with HCC were given no priority as compared to other patients. Grafts were allocated ‘on first listed, first served’ basis.

Operative technique and postoperative management

In eligible patients, a deceased donor (3), domino (17) or living-donor LT (18) was performed using the described techniques.

Postoperative immunosuppression involved cycloscoporine or tacrolimus-based regimens. Corticoids were given to all patients initially and progressively decreased according to posttransplantation liver function tests. Following LT, all patients were monitored every 3 months with liver ultrasound, AFP levels and liver function tests. In addition, every 6 months, a thoracoabdominal CT scan was performed. Posttransplant HCC recurrences were defined with the same preoperative Barcelona criteria (13).

Design of the study and studied population

The aim of the study was to identify overall survival (OS) and recurrence-free survival (RFS) prognostic factors, available at the time of decision for transplantation in patients with AFP secreting HCC. This included AFP progression slope. Calculation of the slope of AFP progression before LT was done by including each static value of AFP of the patients in the waiting list. A dosage of AFP was made at least monthly during this period. The slope of AFP progression was calculated by considering the difference between the lowest and highest AFP values divided by the time between this two referral points. The slope of progression was calculated using continuously increasing AFP values observed during the last time interval before transplantation. The intraoperative and histological data were not included in the study due to lack of any clinical use on a transplant setting. All data used in this study were retrieved from our prospectively collected data base of patients transplanted for HCC.

Among 290 cirrhotic patients listed for HCC, 38 were not transplanted with a total drop-out rate of 13%. The drop-out was due to major deterioration of the underlying liver disease (coma and/or digestive hemorrhage) in 13 patients (4% of listed patients) and progression of tumoral disease (tumoral portal thrombosis and/or extrahepatic metastases) in 21 patients (7% of listed patients). Four patients (2% of listed patients) refused LT or developed other cancers.

Out of 252 transplanted patients, 92 (36%) who had normal AFP levels (<7 μg/L) were excluded from the study analysis. Local biochemical range value had been considered to define this cutoff. Among 160 patients (64%) with abnormal AFP (>7 μg/L), 7 patients were transplanted for HCC recurrence after surgical resection (salvage transplantation). These patients were excluded from analysis because of their reported poor prognosis compared to primarily liver transplanted patients in our series (14). Finally, 153 patients formed the study population.

The characteristics of the 153 transplanted patients are listed in Table 1. The studied patients were divided in two groups according to the slope of AFP evolution before LT, lower or upper to 15 μg/L per month. This threshold was chosen after a preliminary receiver operating characteristics (ROC) analysis on these 153 patients for tumor recurrence according to the slope of progression of AFP. The area under the curve (AUC) and the sensitivity and the specificity for obtained cutoff points were reported (19) (Figure 1). It was the 15 μg/L per month cut off level that allowed the best sensitivity (61%) and specificity (58%) to predict the incidence of recurrence. With this threshold, the positive predictive value and negative predictive value were 0.41 and 0.38, respectively. The threshold of 15 μg/L per month was the discriminating level used to separate the 153 patients into two groups. These two groups were named retrospectively ‘progression’ (>15 μg/L per month) and ‘no progression’ group (≤15 μg/L per month).

Table 1.  Comparison of patients in AFP progression group and AFP no progression group for preoperative parameters
 No progression AFP (n = 127)Progression AFP (n = 26)p
Age (years)55 ± 8 54 ± 5 0.45 
Gender no (%)   
 Female18 (14)3 (11)0.84 
 Male109 (85)23 (88) 
Delay between HCC diagnosis and LT (months)15 ± 1212 ± 8 0.44 
Duration on waiting list (months)3.4 ± 4  5.8 ± 8 0.06 
Cause of cirrhosis no (%)  0.38 
 Viral hepatitis B32 (25)9 (34) 
 Viral hepatitis C68 (53)11 (42) 
 Alcohol19 (15)5 (19) 
 Rare aetiologies or unknown11 (8)1 (4) 
Severity of cirrhosis no (%)  0.94 
 Child A34 (26)7 (26) 
 Child B53 (41)10 (38) 
 Child C40 (31)9 (34) 
Preoperative tumoral features   
 Number of nodule(s)2.3 ± 2.12.3 ± 2.00.99 
 Diameter of largest nodule (mm)32 ± 1737 ± 260.14 
 In Milan criteria (n ≤3/Ø≤3 cm or n = 1/Ø≤5 cm) no (%)85(67)14 (53)0.20 
Preoperative treatment no (%)100 (78%)19 (73%)0.52 
 Number of chemoembolisation (TACE)1.8 ± 1.12.1 ± 1.50.45 
 Local destruction (radiofrequency or cryotherapy) no (%)14 (11)1(3)0.30 
Biological data   
 AFP before LT 333 ± 20532609 ± 65280.001
 AFP >300 μg/L before LT no (%)8/127 (6)10/26 (38) 0.0001

Figure 1. The receiver operating characteristics (ROC) curve (area under curve was 0.557) used to determine the cut off value of alphafetoprotein (AFP) progression.

Download figure to PowerPoint

Of note, in the ‘no progression’ group (n = 127 patients), 29 patients in the waiting list exceeded the AFP slope threshold of 15 μg/L per month. These patients have been classified in the ‘no progression’ because a treatment while on the waiting list resulted in controlling the AFP progression, switching patients from the ‘progression’ to the ‘no progression’ group. Among the 127 patients of the ‘no progression’ group, 29 patients were in this situation.

Statistical analysis

Only items available preoperatively were analyzed, including recipients’ age, gender, pretransplantation treatment, etiology of liver cirrhosis and preoperative tumor characteristics. Additionally to the slope of AFP, different thresholds of AFP level before LT were chosen as preoperative static values and tested (100 μg/L, 200 μg/L, 300 μg/L, 400 μg/L and 1000 μg/L) (7,20).

Comparisons of the two groups previously defined according to AFP progression were made for all these parameters using respectively Mann–Whitney test and chi-square test for quantitative and qualitative variables. A univariate analysis was performed with OS and RFS as endpoints for all items available preoperatively including the variable of interest (Progression or not of AFP). OS and RFS probabilities were calculated using Kaplan–Meier method. A multivariate analysis using a Cox Model was performed for the factors identified as significant in the OS and RFS by the univariate analysis. The results of this study have been used to compare pathological features of the specimen.

Subsequently, by univariate analysis, we looked for preoperative factors linked to the presence of satellite nodules and/or vascular invasion on specimen, two parameters largely recognized as poor pathological factors associated with postoperative recurrence. The factors identified at univariate analysis were further analyzed by a logistic regression to find preoperative predictive factors that would better predict a poor postoperative prognostic feature. All the statistics were performed on Stat View 5.0 for Windows (SAS Institute, Cary, NC) except ROC curve that were calculated on SPSS (SPSS Inc., Chicago, IL).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. References

Among the 153 transplanted patients, 127 (83%) belonged to the ‘no progression’ group and 26 (16%) to the ‘progression’ group. Preoperatively, these two groups were similar except for the level of AFP before LT that was higher in the progression group (p = 0.001) (Table 1). These two groups were also similar in the choice of the surgical procedure (Table 2).

Table 2.  Comparison of patients in AFP progression group and AFP no progression group for surgical parameters
 No progression AFP (n = 127)Progression AFP (n = 26)p
Donor age (years)55 ± 8 54 ± 5 0.45
Liver-related donor transplantation no (/%)16 (12)5 (19)0.56
Domino liver transplantation no (%)25 (19)6 (23)0.62
Duration of cold ischemia8h10 ± 3h008h00 ± 3h500.97
Number of transfused red pack cells7.4 ± 7.47.0 ± 6.30.82
Postoperative mortality no (%)3 (2%)00.43

After a median follow-up of 45 months, 46 patients (30%) died. Two-month postoperative mortality was 2% (three patients). Altogether, 22 patients (14%) died without recurrence, 24 patients (15%) died with recurrence, 28 patients (18%) were alive with HCC recurrence and 79 patients (53%) without HCC recurrence.

At the last follow-up, overall mortality was higher in transplanted patients belonging to the ‘progression’ group as compared to those of the ‘no progression’ group (46% vs. 26%, p = 0.04). Recurrence was also significantly higher in the ‘progression’ group (34% vs. 14%, p = 0.01).

Overall survival

For the 153 patients, OS was 89%, 77% and 72% at 1, 3 and 5 years, respectively. OS was significantly lower for the patients of the ‘progression’ group compared to the ‘no progression’ group (84%, 60% and 54% vs. 90%, 80% and 77% at 1, 3 and 5 years, respectively; p = 0.02).

Recurrence-free survival

For the 153 patients, RFS was 85%, 74% and 69% at 1, 3 and 5 years, respectively. RFS was significantly lower for the patients of the ‘progression’ group compared to the ‘no progression’ group (76%, 52% and 47% vs. 86%, 78% and 74% at 1, 3 and 5 years, respectively; p = 0.01) (Figure 2). Cumulated recurrence rate was also significantly higher in the ‘progression’ group (Figure 3).


Figure 2. Overall survival and recurrence-free survival after liver transplantation (LT) in cirrhotic patients for alphafetoprotein (AFP) secreting hepatocellular carcinoma. Thin line: LT with increasing AFP under 15 μg/L per month before LT. Bold line: LT with increasing AFP over 15 μg/L per month before LT.

Download figure to PowerPoint


Figure 3. Recurrence rate related to time. Thin line: Liver transplantation (LT) with increasing AFP under 15 μg/L per month before LT. Bold line: LT with increasing alphafetoprotein (AFP) over 15 μg/L per month before LT.

Download figure to PowerPoint

Preoperative prognostic factors of survival

The univariate analysis demonstrated two preoperative factors (increasing AFP levels and >3 nodules) to be associated with a significant decrease of OS (Table 3). Transplanting patients outside Milan criteria (35% of patients) proved to have an impact on OS but this criterion was at the limit of statistical significance (p = 0.053). Four factors were associated with a significantly decrease of RFS: more than three nodules (p = 0.006), increasing AFP levels (p = 0.01), age greater than 60 years (p = 0.02) and patients outside Milan Criteria (p = 0.02). Interestingly, all static values of AFP tested before transplantation (AFP>100 μg/L, AFP >200 μg/L, AFP>300 μg/L, AFP>400 μg/L, AFP>1000 μg/L) were not correlated with OS or RFS after LT (Table 4).

Table 3.  Overall and recurrence-free survival univariate analysis for preoperative parameters
Preoperative parametersOverall survivalRecurrence-free survival
1 y3 y5 yp1 y3 y5 yp
 Male (n = 132)90%77%71%0.6885%74%69%0.84
 Female (n = 21)85%80%78% 80%75%75% 
 Age >60 years (n = 109)81%73%69%0.05888%77%74%0.02
 Age ≤60 years (n = 44)92%78%74% 79%65%58% 
Liver disease        
 HCV (n = 80)93%81%79%0.0680%71%63%0.09
 HBV (n = 41)95%78%74%0.4585%73%68%0.34
 Alcohol (n = 24)79%74%67%0.1475%75%62%0.10
 Other disease (n = 8)80%70%65%0.1268%70%61%0.13
Severity of liver disease        
 Child A (n = 41)92%85%78%0.3588%83%80%0.25
 Child B (n = 63)90%75%72%0.9284%73%68%0.76
 Child C (n = 49)85%72%69%0.3683%67%61%0.37
Tumor feature (on last preoperative assessment)        
 Largest nodule        
   ≤50 mm (n = 139)89%77%74%0.2385%75%71%0.27
   >50 mm (n = 14)91%75%54% 77%60%50% 
   ≤30 mm (n = 103)88%79%76%0.3285%78%74%0.13
   >30 mm (n = 50)92%73%65% 83%63%60% 
 Number of nodule(s)        
   ≤3 nodules (n = 132)88%78%76%0.0386%75%75%0.006
   >3 nodules (n = 21)95%69%45% 85%65%37% 
 Milan criteria        
   Inside Milan criteria (n = 99)87%79%79%0.05385%77%77%0.02
   Outside Milan criteria (n = 54)94%74%64% 85%70%58% 
 AFP feature at transplantation        
   AFP ≤300 μg/L (n = 134)91%78%74%0.0886%75%72%0.18
   AFP >300 μg/L (n = 19)83%58%58% 78%62%57% 
   No progression AFP (≤15 μg/L/month) (n = 127)90%80%76%0.0287%78%74%0.01
   Progression AFP (>15 μg/L/month) (n = 26)84%65%54% 76%47%47% 
Pretransplant treatment        
 TACE (n = 115)90%75%70%0.4585%73%68%0.4 
 No TACE (n = 38)86%82%75% 86%74%67% 
 Local destruction (n = 14)85%85%85%0.5086%71%53%0.54
 No local destruction (n = 139)90%76%72% 85%74%70% 
Type of transplantation        
 Deceased donor (n = 101)88%76%73%0.5584%73%72%0.57
 Domino (n = 31)93%74%59% 86%73%49% 
 LDLT (n = 21)90%85%85% 90%71%71% 
Table 4.  Overall and recurrence-free survival analysis for AFP features at transplantation
Preoperative parametersOverall survivalRecurrence-free survival
1 y3 y5 yp1 y3 y5 yp
AFP feature at transplantation        
 AFP ≤100 μg/L (n = 82)87%79%75%0.1784%78%75%0.19
 AFP >100 μg/L (n = 71)91%75%70% 87%74%70% 
 AFP ≤200 μg/L (n = 100)90%80%74%0.2 84%78%75%0.22
 AFP >200 μg/L (n = 53)90%71%60% 86%70%67% 
 AFP ≤300 μg/L (n = 134)91%78%74%0.0886%75%72%0.18
 AFP >300 μg/L (n = 19)83%58%58% 78%62%57% 
 AFP ≤400 μg/L (n = 133)90%78%74%0.0887%78%75%0.12
 AFP >400 μg/L (n = 19)77%58%58% 72%59%52% 
 AFP ≤1000 μg/L (n = 140)91%77%74%0.0687%77%75%0.06
 AFP >1000 μg/L (n = 12)74%55%55% 65%56%47% 
 No progression AFP (≤15 μg/L/month) (n = 127)90%80%76%0.0287%78%74%0.01
 Progression AFP (>15 μg/L/month) (n = 26)84%65%54% 76%47%47% 

At multivariate analysis, only two factors were associated with a significantly decreased OS: a slope of AFP superior to 15 μg/L per month (p = 0.03) and a preoperative number of tumor nodules >3 (p = 0.03). At multivariate analysis, a slope of AFP superior to 15 μg/L per month (p = 0.007) and an age >60 years old (p = 0.03) were factors that significantly decreased RFS (Table 5).

Table 5.  Comparison of patients in AFP progression group and AFP no progression group for pathological parameters
Preoperative variablesOverall survivalRecurrence-free survival
pRisk ratioCI 95%pRisk ratioCI 95%
  1. NS = not significant; CI = confidence interval..

Univariate analysis      
 Age >60 yearsNS  0.02- 
 Progression AFP (>15 μg/L/month)0.02  0.01  
 Number of nodules >30.03  0.006  
 Outside Milan criteriaNS  0.02  
Multivariate analysis      
 Number of nodules >30.032.132(1.064–4.366)-  
 Progression AFP (>15 μg/L/month)0.032.061(1.166–3.984)0.0072.450(1.267–4.716)
 Age >60 years-  0.031.919(1.038–3.546)

Pathological analysis

In patients of the ‘progression’ group, satellite nodules were more frequent (p = 0.02) and the rate of vascular invasion was higher (p = 0.01) than in patients of the ‘no progression’ group (Table 6). There was no other pathological difference according to AFP evolution on the waiting list. A comparison of histological features of the explants was done according to two other preoperative parameters that significantly influenced the survival at multivariate analysis. In addition to obvious differences (number of nodules and Milan criteria), patients with a number of preoperative tumor nodules >3 had a higher rate of vascular invasion (p = 0.04). No pathological differences were identified according to the patient's age (<60 or >60 years).

Table 6.  Comparison of patients of the ‘no progression group’ and ‘progression group’ for pathological parameters
 No progression AFP (n = 127)Progression AFP (n = 26)p
>3 nodules no (histo) (%)31 (24) 9 (36)0.29
>3 cm no (histo) (%)40 (31)10 (38)0.50
>5 cm no (histo) (%)10 (8)  4 (15)0.23
Presence of satellite nodules no (%)36 (28)13 (52)0.02
Vascular invasion no (%)41 (32)15 (57)0.01
Presence of capsula no (%)67 (52)13 (50)0.61
Well/Middle-poor differentiation no (%)17 (21)/64 (80)3 (14)/18 (85)0.16
Total necrosis (after TACE) no (%)35 (27) 3 (11)0.09
Outside of Milan criteria (histo) no (%)77 (61)12 (46)0.15

Preoperative predictive factors of histological aggressiveness

At univariate analysis, four preoperative factors were associated with the presence of satellite nodules and/or vascular invasion on the specimen: male gender (p = 0.004), disregard of Milan Criteria on last preoperative assessment (p = 0.002), the increasing AFP as defined (progression group) (p = 0.005) and a largest nodule >3 cm (p = 0.002).

At logistic regression, the increasing AFP (p = 0.04), the male gender (p = 0.03) and a largest nodule >3 cm emerged as predictive of satellite nodules and/or vascular invasion on the specimen (Table 7).

Table 7.  Logistic regression for histological aggressiveness
Preoperative variablePresence of satellite nodules and/or vascular invasion on specimen
pOdds ratioCI 95%
Univariate analysis   
 Outside Milan criteria0.002  
 Progression AFP (>15 μg/L/month)0.005  
 Largest nodule >30 mm0.002  
Logistic regression analysis   
 Male0.01 4.111.25–13.51
 Progression AFP (>15 μg/L/month)0.01 3.491.27–9.61 
 Largest nodule >30 mm0.04 2.281.01–5.15 

Impact of locoregional treatment on AFP progression and survival

As previously described, in 29 out of 127 patients in the ‘no progression’ group, TACE (n = 29) and local destruction (n = 4), were able to lower AFP progression under 15 μg/L per month before LT. The OS of the 29 patients was 89%, 73% and 73% at 1, 3 and 5 years, respectively better than that of patients of the ‘persistent’ progression group (73% vs. 54%, p = 0.14) although the difference was not statistically significant (Figure 4). Indeed, the survival of the 29 patients was similar to that of patients that have never presented an AFP increasing over 15 μg/L per month (n = 98 patients) (p = 0.28).


Figure 4. Recurrence-free survival after liver transplantation (LT) in cirrhotic patients for alphafetoprotein (AFP) secreting hepatocellular carcinoma in patients within Milan criteria on the specimen and outside Milan criteria. Thin line: LT with increasing AFP under 15 μg/L per month before LT. Bold line: LT with increasing AFP over 15 μg/L per month before LT.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. References

This study shows that AFP progression exceeding 15 μg/L per month was a major predictive factor of tumor recurrence and poor survival after LT for AFP secreting HCC patients.

This dynamic parameter was more accurate than any increased static value of AFP in patients with HCC awaiting liver transplant. In addition, a progression of AFP above the cut off level of 15 μg/L was more relevant than fulfilling Milan criteria to predict tumor recurrence (Figure 4). However, in patients outside Milan criteria, no differences in RFS using this cut off level were observed (Figure 4). Also, progression of AFP was associated with poor histological features (satellite nodules and/or vascular invasion) available only after LT is performed. Our results suggest that a progression of AFP over 15 μg/L per month could be a surrogate marker of adverse pathological findings, easily measurable preoperatively.

Patients with nonsecreting AFP HCC have been deliberately excluded from this study that aimed to find a cutoff of AFP progression that could predict tumor recurrence. In the same population of patients, we have confirmed that this cutoff, despite its relative low sensibility and specificity, was clinically relevant as proved by its significance in multivariate analysis. However, the application of our criteria in all patients of our series transplanted for HCC is also largely useful if we consider that patient with normal AFP were classified in ‘the nonprogression’ group.

Due to organ shortage, LT for malignancy should provide almost equivalent results as that of LT for benign disease. Indeed, some cutoffs concerning the number and the size of HCC have been defined to assure this objective (3–7). Nowadays, Milan criteria introduced first by Mazzafero et al. (4) as an extension of those initially published by our team (3) are widely used and successfully evaluated with an OS at 5 years between 61% and 70% (4–7). However, even when fulfilling the Milan criteria, significant survival differences exist between patients (5).

In our series, 34% of patients were transplanted outside Milan criteria. Interestingly, transplanting patients outside Milan Criteria did not influence the OS, probably because the extension of the indications was relatively restricted. Our rate of transplantation outside Milan criteria (34%) on preoperative data is still lower compared to the rate of 41% recently reported in a French multicentric study concerning 479 patients (21). Others have also suggested to transplant patients outside Milan criteria with acceptable results (22).

However, within Milan criteria, the poor result of some transplanted patients could be explained by the specific pathological features of the tumor (5–7,9,23). Poorly differentiated tumors, vascular invasion and presence of satellite nodules are the main factors associated with a poor prognosis (6,8). Hence, the major challenge of surgeons and hepatologists in LT for HCC is to assess accurately these characteristics before LT.

Assessment by a preoperative tumor biopsy has been proposed, however, along with the risk of tumor seeding and hemorrhage, the obtained results are far from accurate due to possible heterogeneity of differentiation in diverse areas of the tumor (24). Potentially, resection of HCC is the only way for a complete pathological analysis of the tumor, and could help to choose thereafter a LT ‘de principe’ in cases with poor prognostic factors (16,25). Although, this strategy is now proposed (26), favoring the LT for cases with poor prognostic factors may have a negative impact on the survival after LT.

Different cut off values of preoperative AFP, from 8.5 μg/L to 1000 μg/L, have been associated with a low survival after LT (7,20,23,27,28). In two series, this cut off value was 300 μg/L (7,20). In an intent-to-treat study, a high rate of AFP was also associated with a higher drop-out rate from LT waiting list (29,30). However, an AFP threshold was never used to preclude LT in cirrhotic patients with HCC, more so when the Milan criteria were fulfilled. To our knowledge, the evolution of AFP that independently correlates to OS and RFS has never been described before. Furthermore, this dynamic parameter (an increasing of AFP superior to 15 μg/L per month) is specific and could potentially be used to reappraise the indication of LT or the pretransplant strategy.

Also, our dynamic parameter was found to highly correlate with the presence of vascular invasion and satellite nodules on pathological analysis, two parameters reported to be strongly associated with OS and RFS (5–8,23). AFP expression by HCC has been linked with some adverse histological features (11). Even for small HCC, 15–33% remains associated with low differentiation, vascular invasion and intrahepatic metastases (11,31). The rate of LT outside Milan criteria on pathological examination was not significantly different according AFP evolution. This last point emphasized that a rapid AFP progression is a more reliable source of information of tumoral biology than a preoperative assessment of tumor load, meaning number and diameter, in this selected population.

TACE before LT for HCC aims to detect additional nodules missed by CT scan and ultrasonography, to downstage HCC within Milan criteria, to avoid drop-out on listing and to improve postoperative survival (15,32). If the benefit of TACE remains uncertain in term of survival because of conflicting results and lack of randomized controlled trials, to date no study has demonstrated additional risks brought about by this procedure at transplant surgery (32). Comparative survival analysis of AFP progressing patients with regards to the response or not to TACE is an important result of our study. Despite the absence of significant difference, there was a clear trend for better survival (73% vs. 54%) in patients responding to TACE suggesting that neoadjuvant treatment of HCC in patients waiting for LT is perhaps useful.

In summary, transplantation for HCC in cirrhotic patients having an AFP progression over 15 μg/L per month leads to a decreased 5-year OS of 54%, debatable in the critical situation of organ shortage. Progression of AFP could be a surrogate marker of poor pathological features useful to select patients for LT.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. References