Impact of UCSF criteria according to pre- and post-OLT tumor features: Analysis of 479 patients listed for HCC with a short waiting time

Authors


  • See Editorial on Page 1741

Abstract

Orthotopic liver transplantation (OLT) indication for hepatocellular carcinoma (HCC) is currently based on the Milan criteria. The University of California, San Francisco (UCSF) recently proposed an expansion of the selection criteria according to tumors characteristics on the explanted liver. This study: 1) assessed the validity of these criteria in an independent large series and 2) tested for the usefulness of these criteria when applied to pre-OLT tumor evaluation. Between 1985 and 1998, 479 patients were listed for liver transplantation (LT) for HCC and 467 were transplanted. According to pre-OLT (imaging at date of listing) or post-OLT (explanted liver) tumor characteristics, patients were retrospectively classified according to both the Milan and UCSF criteria. The 5-yr survival statistics were assessed by the Kaplan-Meier method and compared by the log-rank test. Pre-OLT UCSF criteria were analyzed according to an intention-to-treat principle. Based on the pre-OLT evaluation, 279 patients were Milan+, 44 patients were UCSF+ but Milan− (subgroup of patients that might benefit from the expansion), and 145 patients were UCSF− and Milan−. With a short median waiting time of 4 months, 5-yr survival was 60.1 ± 3.0%, 45.6 ± 7.8%, and 34.7 ± 4.0%, respectively (P < 0.001). The 5-yr survival was arithmetically lower in UCSF+ Milan− patients compared to Milan+ but this difference was not significant (P = 0.10). Based on pathological features of the explanted liver, 5-yr survival was 70.4 ± 3.4%, 63.6 ± 7.8%, and 34.1 ± 3.1%, in Milan+ patients (n = 184), UCSF+ Milan− patients (n = 39), and UCSF− Milan− patients (n = 238), respectively (P < 0.001). However, the 5-yr survival did not differ between Milan+ and UCSF+ Milan− patients (P = 0.33). In conclusion, these results show that when applied to pre-OLT evaluation, the UCSF criteria are associated with a 5-yr survival below 50%. Their applicability is therefore limited, despite similar survival rates compared to the Milan criteria, when the explanted liver is taken into account. Liver Transpl 12:1761-1769, 2006. © 2006 AASLD.

Orthotopic liver transplantation (OLT) has been proposed as an effective therapy for hepatocellular carcinoma (HCC) with tumors limited in size and number, and is the only treatment that definitively addresses both the metachronous occurrence risk of HCC and the underlying liver disease.1–4 The most widely used criteria for patient selection are those proposed by Mazzaferro et al.1 (1 lesion 5 cm or smaller or up to 3 lesions each 3 cm or smaller; i.e., the Milan criteria) (Table 1); these criteria have been adopted by the United Network for Organ Sharing. Due to these selection criteria, the retrospective analysis of United Network for Organ Sharing data5 found a significant improvement in survival over time for HCC patients undergoing OLT, with a 5-yr survival of 61.1% contrasting with the previously observed 5-yr survival rate of 25.3% in 1987.

Table 1. Tumor Characteristics According to Modified TNM (1998) and pTNM (2001) and Their Relation to OLT Criteria
Conventional criteria of OLTExpansion criteria of OLT
Modified TNM (1998)14Milan criteria (1996)1Modified pTNM (2001)7UCSF criteria (2001)7
  1. Abbreviations: TNM, tumor-node-metastasis; pTNM, pathological tumor-node-metastasis; CT, computed tomography; MRI, magnetic resonance imaging; US, ultrasound.

T1 = 1 tumor ≤1.9 cm;OLT if:T1 = 1 tumor ≤1.9 cm;OLT if:
T2 = 1 tumor 2.0-5.0 cm or 2 or 3 tumors, all ≤3.0 cm;● 1 tumor ≤5.0 cm; ● up to 3 tumors, none larger than 3.0 cmT2 = 1 tumor 2.0-5.0 cm or 2 or 3 tumors all ≤3.0 cm;● 1 tumor ≤6.5 cm
T3 = 1 tumor >5.0 cm or 2 or 3 tumors with at least 1 tumor >3.0 cm; T3a = 1 tumor 5.0-6.5 cm or 2 or 3 tumors between 3.0 and 4.5 cm with sum ≤8.0 cm;● up to 3 tumors, none larger than 4.5 cm and sum of diameter no larger than 8 cm
T4a = 4 or more tumors, any size; T3b = 1 tumor >6.5 cm or 2 or 3 tumors up to 4.5 cm, or sum up to 8.0 cm; 
T4b = any tumor + gross intrahepatic portal or hepatic vein involvement as indicated by CT, MRI, or US T4a = 4 or more tumors, any size;T4b = any T + gross intrahepatic portal or hepatic vein involvement as indicated by CT, MRI, or US

However, some investigators suggested that the Milan criteria might be too restrictive. Indeed, the proportion of HCC patients not fulfilling the Milan criteria and that could be cured by liver transplantation (LT) probably ranges from 27% to 49%.6 The retrospective analysis of 70 consecutive patients with cirrhosis and HCC who underwent OLT over a 12-yr period at the University of California, San Francisco (UCSF), prompted Yao et al.7 to propose an expansion of selection criteria for HCC, according to post-OLT tumor characteristics. In this study, the 5-yr survival rate of 60 patients with 1 lesion ≤6.5 cm or 3 lesions, largest lesion ≤4.5 cm with a total tumor diameter ≤8 cm on the explanted livers (UCSF criteria; Table 1) was 75.2%. When transposed to pre-OLT tumor evaluation of 39 patients with nonincidental HCC meeting the UCSF criteria, the 5-yr survival rate was 84%. Yao et al.7 concluded that these new criteria did not significantly compromise the excellent post-OLT outcomes achieved by the Milan criteria. It was therefore tempting to extend the Milan criteria for patient selection for cadaveric donation8 or for living donor LT.9 However, such criteria have been assessed a posteriori on the explanted livers, and their applicability to the pre-OLT evaluation deserves further studies.

The aim of the present work was to assess, on a large independent series of patients transplanted for HCC: 1) the validity of the UCSF criteria based on the pathological analysis of the explanted livers; and 2) the usefulness of these criteria when applied to pre-OLT evaluation.

Abbreviations

AP-HP, Assistance Publique-Hôpitaux de Paris; OLT, orthotopic liver transplantation; HCC, hepatocellular carcinoma; UCSF, University of California, San Francisco; LT, liver transplantation.

PATIENTS AND METHODS

Study Population

The study population consisted of 479 patients who had been listed for transplantation for HCC in 14 French LT centers between 1985 and 1998. Patients with incidental HCC, which were defined as tumors diagnosed on the explanted liver but not prior to transplant, as well as patients transplanted for fibrolamellar carcinoma were excluded.

According to preoperative characteristics and intention-to-treat principles, 468 patients, including 12 patients who had been dropped from the waiting list because of tumor progression and/or death (dropout rate = 2.5%), could be classified according to both the UCSF and Milan criteria. Pre-OLT evaluation was not available for 11 patients who were excluded from statistical analysis.

Based on pathological evaluation on the explanted liver, 461 patients were classified according to both the UCSF and Milan criteria (pathological data were not available for 7 patients who were excluded for statistical analysis).

The 468 patients were retrospectively classified in the following 3 groups:

  • Milan+ patients

  • UCSF+ but Milan− patients (patients who might benefit from the expended criteria)

  • UCSF− and Milan− patients.

Pretransplantation Data

The mean age of patients was 52.7 ± 9.2 yr. There were 82% of males with cirrhosis in 91% of cases. Cirrhosis was considered to be due to the hepatitis C virus in 38%, to alcohol abuse in 32%, to the hepatitis B virus in 20%, and other causes in 10%. A total of 55% of patients were classified as Child-Turcotte-Pugh10 class A, 31% as class B, and 14% as class C. The median alpha-fetoprotein level was 27.0 ng/mL (range 0.5-245,240). According to the Karnofsky index,11 a poor performance status was observed in 38% of patients at the time of OLT evaluation. Morphological features and preoperative staging of HCC were evaluated from abdominal imaging studies (ultrasound alone 20%, ultrasound and computed tomography 80%, and magnetic resonance imaging when available). The number, size, and location of the tumors as well as the presence of vascular involvement were determined from this prelisting evaluation. The median number of tumors was 1 (range 1-11), the median diameter of the largest tumor was 3.4 cm (range 0.6-25), and the median sum of the diameter of nodules was 4.0 cm (range 0.6-30). Tumors were located in the 2 liver lobes in 26% of the cases. A macrovascular involvement was noted in 8% of the cases.

Posttransplantation Tumor Characteristics

Pathological data of the explanted livers were obtained by reviewing the histopathological reports. Cirrhosis was present in 91% of the patients. When the liver was noted as noncirrhotic, the grade of fibrosis was quoted according to the METAVIR scoring system12 (F0 in 9 patients, F1 in 8 patients, F2 in 3 patients, F3 in 17 patients, and undetermined in 6 patients). The median number of tumors was 2 (range 1-51), the median diameter of the tumor was 3.5 cm (range 0.5-31), and the median sum of the largest diameter of nodules was 6.0 cm (range 0.5-56). Tumors were located in the 2 lobes in 45% of the cases. Microvascular and macrovascular involvements were noted in 26.4% and 13% of the cases, respectively. Tumor differentiation was quoted according to the World Health Organization classification and if multiple differentiation statuses were present in the same liver, the poorest was considered: 68.5% of tumors were well differentiated, 25.5% were moderately differentiated, and 6% were poorly differentiated. A nodal involvement was observed in 2.3% and an involvement of adjacent organs in 3.6%.

Immunosuppression and Follow-Up

The type of immunosuppressive drugs (calcineurin inhibitors vs. antilymphocyte antibodies) and the presence of a histologically proven acute rejection and its treatment were noted (methylprednisone pulses, antilymphocyte antibodies).

Screening for tumor recurrence was performed by serial measurements of alpha-fetoprotein and computed tomography scans of the thorax and abdomen every 3 months during the 2 first postoperative year and/or when clinically indicated. Additional imaging techniques (bone scan, magnetic resonance imaging) were done if necessary.

Postoperative death was defined as death during the third month post-OLT.

Causes of death were divided into 3 categories: HCC-related death, post-OLT death (including primary nonfunction of the graft, surgical complication, infectious death, and critical care mortality), and other (including recurrence of primary liver disease, cardiovascular complications, and other malignancy).

Tumor recurrence, length of follow-up from OLT to death, length of follow-up from OLT to HCC recurrence, and length of follow-up from OLT to last news, were also considered.

Statistics

Baseline characteristics of the patients were expressed as mean ± standard deviation and median and range when appropriate. Comparison between groups was done for continuous variables by the Kruskal-Wallis test and by the Mann-Whitney test if necessary. Categorical variables were compared by the χ2 test. Length of follow-up and survival were expressed as median and ranges. Kaplan-Meier estimates of survival were calculated using the intention-to-treat principle and compared by the log-rank test. A P value < 0.05 was considered statistically significant. The calculations were done with the BMDP package (Statistical Solutions, Saugus, MA).

RESULTS

Based on Pre-OLT Tumor Evaluation

Based on pre-OLT evaluation, there were 279 Milan+ patients (59.6%), 44 UCSF+ Milan− patients (9.4%), and 145 UCSF− Milan− patients (30.9%) (Fig. 1A).

Figure 1.

Distribution of patients according to Milan and UCSF criteria for liver transplantation. (A) Distribution according to imaging tumor features at the time of listing. (B) Distribution according to pathological tumor characteristics assessed on liver explant.

Characteristics of these 3 groups are summarized in Table 2. As expected, tumor characteristics differed between the 3 groups. Tumor stage rose progressively from Milan+ patients to UCSF− Milan− patients, with intermediate values in UCSF+ Milan− patients. Tumor characteristics also differed significantly between Milan+ and UCSF+ Milan− patients, with more nodules and larger diameters in UCSF+ Milan− patients, compared to Milan+ patients. Mean alpha-fetoprotein level rose progressively from 250 ng/mL in Milan+ patients, to 5,600 ng/mL in UCSF+ Milan− patients and 10,000ng/mL in UCSF− Milan− patients (P < 0.0001).

Table 2. Characteristics of Patients Listed for Transplantation According to UCSF and Milan Criteria on Pre-OLT Evaluation
 UCSF+ and Milan+ (n = 279)P**UCSF+ but Milan− (n = 44)UCSF− and Milan− (n = 145)P*
  1. Abbreviations: CTP, Child-Turcotte-Pugh; AFP, alpha-fetoprotein, P*, global comparison between the 3 groups; P**, intergroup comparison.

Age (yr)53.6 ± 8.2 51.2 ± 11.852.1 ± 9.70.6
Sex ratio (M/F)86%/14% 84%/16%86%/14%0.14
Presence of cirrhosis97%0.01488%80%<0.0001
CTP: A/B/C53%/30%/17% 62%/31%/7%55%/35%/10%0.14
AFP: mean (ng/mL)250 ± 6000.025,600 ± 30,00010,000 ± 37,2000.0004
Pre-OLT treatment of HCC75%0.2766%51%<0.0001
Number of nodules1.4 ± 0.6<0.00011.8 ± 0.72.7 ± 2.0<0.0001
Maximal diameter of the largest nodule (cm)2.8 ± 1.0<0.00014.6 ± 1.06.9 ± 4.0<0.0001
Sum of diameters (cm)3.2 ± 1.2<0.00016.1 ± 1.39.9 ± 4.8<0.0001
Bilobar location12%0.000533%52%<0.0001
Venous thrombosis0%0.990%26%<0.0001
Median waiting time: months (range)3.6 (0.1-32.3)0.354.5 (0.1-20.4)2.4 (0.3-21.7)<0.0001

In an intention-to-treat analysis and based on pre-OLT evaluation, 5-yr overall survival rates were 60.1 ± 3.0%, 45.6 ± 7.8%, and 34.7 ± 4.0%, respectively, in Milan+ patients (n = 279), UCSF+ Milan− patients (n = 44), and UCSF− Milan− patients (n = 145) (P < 0.001; Fig. 2). The 5-yr survival tended to be lower in UCSF+ Milan− patients than in Milan+ patients, but this difference did not reach significance (P = 0.10). Overall 5-yr survival in USCF+ patients (either Milan+ or negative) was 59%.

Figure 2.

The 5-yr overall survival of patients listed for transplantation (intention-to-treat principle) according to UCSF and Milan criteria assessed on preoperative tumor features.

Similar results were observed regarding 5-yr disease free survival (60.4 ± 3.0%; 47.8 ± 8.3%, and 32.3 ± 4.0%, in Milan+ patients, UCSF+ Milan− patients, and UCSF− Milan− patients, respectively; P < 0.001, but P = 0.14 between Milan+ and UCSF+ Milan− patients).

The 5-yr cumulative risk of HCC recurrence in transplanted patients (n = 467) differed significantly between Milan+ patients (20.2 ± 2.7%), UCSF+ Milan− patients (27.1 ± 8.6%), and UCSF− Milan− patients (52.6 ± 4.8%) (P = 0.0025; Fig. 3). However, the difference between Milan+ patients and UCSF+ Milan− patients did not reach significance (P = 0.15).

Figure 3.

The 5-yr cumulative risk of HCC recurrence after liver transplantation according to UCSF and Milan criteria assessed on preoperative tumor features.

Follow-up data of patients are summarized in Table 3. Despite a statistical difference between the three groups, there was no difference between Milan+ patients and UCSF+ Milan− patients.

Table 3. Follow-Up Data of Patients Listed for Transplantation According to UCSF and Milan Criteria on Pre-OLT Evaluation
 UCSF+ and Milan+ (n = 279)P**UCSF+ but Milan− (n = 44)UCSF− and Milan− (n = 145)P*
  1. P*, global comparison between the 3 groups; P**, intergroup comparison.

Dropout: n (%)5 (1.8%) 2 (4.5%)5 (3.4%)0.78
Postoperative death: n (%)27 (10%) 4 (10%)17 (12%)0.89
HCC recurrence: n (%)49 (18%)0.311 (26%)65 (46%)<0.0001
Median time between OLT and HCC recurrence: months (range)16.1 (0.9-99.3)0.5416.1 (10.0-31.9)7.2 (0.2-125.0)0.0013
Death: n (%)118 (43%)0.0624 (57%)102 (73%)<0.0001
Causes of death: HCC/post-OLT/other (%)45%/17%/37%0.4350%/0%/50%65%/12%/23%0.018
Median follow-up after OLT: months (range)52.4 (0.0-170.0)0.4348.8 (0.4-146.0)26.7 (0.0-186.0)0.0003

To assess the influence of radiological understaging, we evaluated the number of misclassified patients in the different groups of patients. A total of 34% Milan+ patients were underscored compared to 48% of UCSF+ Milan− patients (P = 0.09). Due to imaging techniques improvements, we also restricted the database on patients transplanted between 1995 and 1998. Over this period, 180 patients were classified as Milan+ and 24 patients as UCSF+ Milan−. A total of 28% of Milan+ patients were underscored, compared to 8.3% of UCSF+ Milan− patients. However, this decrease in underscored patients in the UCSF+ Milan− group was not associated with a change in 5-yr overall survival (51.9 ± 10.8% compared to 68.3 ± 3.9% for Milan+ patients; P = 0.18; data not shown).

Based on post-OLT tumor features, there were 184 Milan+ patients (39.9%), 39 UCSF+ but Milan− patients (8.6%), and 238 UCSF− and Milan− patients (51.6%) (Fig. 1B).

The features of these 3 groups are summarized in Table 4. As already observed for pre-OLT evaluation, tumor characteristics differed between the 3 groups: tumor stage rose progressively from Milan+ patients to UCSF− Milan−patients, with intermediate values for UCSF+ Milan− patients. Tumor characteristics also differed significantly, with more nodules and larger diameters, and more frequent microvascular invasion in UCSF+ Milan− patients, compared to Milan+ patients (30% vs. 13%; P = 0.009). Yet tumor differentiation showed only a trend toward more moderately- to poorly-differentiated tumors in UCSF+ Milan− patients compared to Milan+ patients (33% vs. 22%; P = 0.16). Mean alpha-fetoprotein level rose progressively from 400 ng/mL in Milan+ patients to 800 ng/mL in UCSF+ Milan− patients and 7,000 ng/mL in UCSF− Milan− patients (P < 0.0001). Prevalence of cirrhosis and pre-OLT HCC therapy as well as the mean waiting time differed significantly between the 3 groups but was not different between Milan+ and UCSF+ Milan− patients.

Table 4. Characteristics of Patients According to UCSF and Milan Criteria on Pathological Evaluation of the Explanted Liver
 UCSF+ and Milan+ (n = 184)P**UCSF+ but Milan− (n = 39)UCSF− and Milan− (n = 238)P*
  1. Abbreviations: CTP, Child-Turcotte-Pugh; AFP, alpha-fetoprotein; P*, global comparison between the 3 groups; P**, intergroup comparison.

Age (years)53.0 ± 9.0 54.2 ± 7.752.1 ± 9.60.48
Sex ratio (M/F)78%/22% 87%/13%85%/15%0.08
Presence of cirrhosis96%0.295%86%0.0008
CTP: A/B/C55%/30%/15% 68%/21%/11%52%/35%/13%0.46
AFP (mean) (ng/ml)400 ± 2,5000.95800 ± 2,8007,000 ± 31,400<0.0001
Pre-OLT treatment of HCC75%0.771%58%0.002
Number of nodules1.6 ± 1.10.0022.0 ± 0.87.4 ± 8.2<0.0001
Maximal diameter of the largest nodule (cm)2.4 ± 0.9<0.00014.4 ± 1.16.3 ± 4.7<0.0001
Sum of diameters (cm)3.1 ± 1.4<0.00015.7 ± 1.011.5 ± 6.5<0.0001
Bilobar location15%0.00637%70%<0.0001
Macrovascular invasion0%0.990%25%<0.0001
Microvascular invasion13%0.00930%35%<0.0001
Tumor differentiation: well/moderate-poor78%/22%0.1667%/33%62%/38%0.0003
Median waiting time: months (range)3.6 (0.1-32.3)0.224.0 (0.8-21.7)2.9 (0.1-30.3)0.01

The 5-yr overall survival rates were 70.4 ± 3.4%, 63.6 ± 7.8%, and 34.1 ± 3.1% in Milan+ patients (n = 184), UCSF+ Milan− patients (n = 39), and UCSF− Milan− patients (n = 238), respectively (P < 0.001) (Fig. 4). However, the 5-yr survival did not differ significantly between Milan+ patients and UCSF+ Milan− patients (P = 0.33) and the overall 5-yr survival in USCF+ patients (either Milan+ or negative) was 69%.

Figure 4.

The 5-yr overall survival of patients according to UCSF and Milan criteria assessed on pathological reports.

Similar results were observed for 5-yr disease-free survival (70.2 ± 3.5%, 62.7 ± 7.9%, and 32.9 ± 3.1%, in Milan+, UCSF+ Milan−, and UCSF− Milan− patients, respectively (P < 0.001). However, again, the difference did not reach statistical significance between Milan+ and UCSF+ Milan− patients (P = 0.26).

The 5-yr cumulative risk of HCC recurrence differed significantly between Milan+ patients, UCSF+ Milan− patients, and UCSF− Milan− patients (9.4 ± 2.4% vs. 16.5 ± 6.7% vs. 52.6 ± 3.8%, respectively; P < 0.001). However, the difference between Milan+ patients and UCSF+ Milan− patients did not reach significance (P = 0.13; Fig. 5).

Figure 5.

The 5-yr cumulative risk of HCC recurrence according to UCSF and Milan criteria assessed on pathological reports.

Follow-up data of these patients are summarized in Table 5. Despite a statistical difference between the 3 groups, there was no difference between Milan+ patients and UCSF+ Milan− patients except for the median time between OLT and HCC recurrence, which was significantly shorter in UCSF+ Milan− patients compared to Milan+ patients (7.6 vs. 22.0 months; P = 0.009).

Table 5. Follow-Up Data According to UCSF and Milan Criteria on Pathological Evaluation of the Explanted Liver
 UCSF+ and Milan+ (n = 184)P**UCSF+ but Milan− (n = 39)UCSF− and Milan− (n = 238)P*
  1. Abbreviations:P*, global comparison between the 3 groups; P**, intergroup comparison.

Postoperative death: n (%)17 (9%)0.185 (13%)33 (14%)0.02
HCC recurrence: n (%)17 (9%)0.255 (13%)107 (45%)<0.0001
Median time between OLT and HCC recurrence: months (range)22.0 (8.5-81.5)0.0127.6 (2.1-28.0)10.0 (0.2-125.0)0.002
Death: n (%)65 (35%)0.4616 (42%)71<0.0001
Causes of death: HCC/post-OLT/other28%/38%/34%0.4333%/40%/27%66%/22%/12%<0.0001
Median follow-up: months (range)58.5 (0.0-170)0.2255.4 (0.4-146)27.8 (0.0-186)<0.0001

Comparison of UCSF+ Milan− Patients According to Pre-OLT (n = 44) and Post-OLT (n = 39) Evaluation

To determine why 5-yr survival was poorer in UCSF+ Milan− patients selected on the basis of pre-OLT evaluation (group 1) than in UCSF+ Milan− patients selected on postoperative evaluation (group 2) (45.6% vs. 63.6%), the preoperative characteristics of the tumors as well as the features of the tumors on the explanted liver were compared in these 2 groups (Table 6).

Table 6. Characteristics of UCSF+ but Milan− Patients According to Pre- or Post-OLT Evaluation
 UCSF+ but Milan− on pre-OLT(n = 44)UCSF+ but Milan− on post-OLT (n = 39)P
Age (years)51.2 ± 11.854.2 ± 7.70.47
Sex ratio (M/F)84%/16%87%/13%0.14
Presence of cirrhosis88%95%0.17
CTP: A/B/C62%/31%/7%68%/21%/11%0.4
AFP mean (ng/ml)5,600 ± 30,000800 ± 2,8000.3
Pre-OLT treatment of HCC66%71%0.08
Pre-OLT tumor features
 Number of nodules1.8 ± 0.71.6 ± 0.80.27
 Maximal diameter of the largest nodule (cm)4.6 ± 1.04.2 ± 1.50.13
 Sum of diameters (cm)6.1 ± 1.35.0 ± 1.90.003
 Bilobar location33%20%0.08
 Venous thrombosis0%0%0.99
Post-OLT tumor features
 Number of nodules5.7 ± 9.22.0 ± 0.80.019
 Maximal diameter of the largest nodule (cm)4.6 ± 2.24.4 ± 1.10.34
 Sum of diameters (cm)8.3 ± 5.85.7 ± 1.00.004
 Bilobar location50%37%0.43
 Macro-vascular invasion7%0%0.02
 Micro-vascular invasion26%30%0.85
 Tumor differentiation: well/moderate to poor61%/39%67%/33%0.44
Follow-up
 Period of OLT: ≤93/94-96/97-9835%/33%/32%29%/34%/37%0.09
 Median waiting time (months) (range)4.5 (0.1-20.4)4.0 (0.8-21.7)0.56
 Postoperative death10%13%0.8
 HCC recurrence26%13%0.051
 Death57%42%0.19
 HCC related death50%40%0.17

Preoperative tumors features were rather similar in both groups. In particular, the number of nodules and the maximal diameter of the largest nodule did not differ significantly in the 2 groups. However, the sum of diameters of the tumors was significantly higher in group 1 than in group 2 (6.1 ± 1.3 cm vs. 5.0 ± 1.9 cm; P = 0.003).

On the explanted liver, the number of nodules was significantly higher (5.7 ± 9.2 vs. 2.0 ± 0.8; P = 0.019) and the sum of diameters of the nodules was significantly larger (8.3 ± 5.8 cm vs. 5.7 ± 1.0 cm; P = 0.004) in group 1 than in group 2.

Accordingly, the HCC recurrence rate was significantly higher in UCSF+ Milan− patients who had been classified pre-OLT (group 1) compared to patients who had been classified post-OLT (group 2) (26% vs. 13%; P = 0.05).

DISCUSSION

The Milan criteria, which are currently used in most LT centers for selection of LT candidates for HCC, are too restrictive, since 27 to 49% of the patients not fulfilling these criteria could be cured by LT without recurrence.6 In the present work we studied how the recently proposed extended UCSF criteria might be applied to LT selection of HCC patients.

Three main results can be drawn from this study.

First, this study shows that an expansion of HCC selection criteria based on the USCF criteria, defined either pre- or postoperatively, would account for only a 10% increase in OLT indications for HCC, compared to a Milan-based selection process.

Second, this study confirms that, based on the pathological features of the tumors, the UCSF criteria are associated with a good 5-yr survival rate of 69%, a figure in accordance with the previously reported survival rate of 75.2% by Yao et al.7 Such results were observed in patients meeting both the Milan and UCSF criteria (Milan+ patients) but also in the subgroup of patients that might benefit from the extension to the UCSF criteria (UCSF+ Milan−), with 5-yr survival rates of 70.4% and 63.6%, respectively (not significant).

Third, this study shows that when applied to the pretransplantation evaluation and despite a short waiting time of 4 months, the impact of the UCSF criteria might not be as high as expected. Indeed, in this setting, although we observed an overall 59% 5-yr survival rate in UCSF+ patients, the 5-yr survival rate in UCSF+ Milan− patients tended to be reduced compared to Milan+ patients (45.6% vs. 60.4%; P = 0.10). Although not significant and probably due to a lack of power, such a trend indicates that the UCSF selection criteria could be associated with a lower survival than the Milan criteria. The UCSF selection criteria should therefore be applied with caution to pre-LT evaluation.

The reason why UCSF criteria, when assessed preoperatively, were associated with a reduced survival deserves further analysis. First, the data presented herein shows that the preoperative morphological evaluation obviously underestimated tumor stage when compared to the pathological staging as assessed on the explanted liver. Indeed, comparison of tumor features in patients who had been classified UCSF+ Milan− either pre-OLT or post-OLT showed different characteristics. On the explanted liver, the number of nodules was significantly higher (5.7 ± 9.2 vs. 2.0 ± 0.8; P = 0.019) and the sum of diameters of the nodules was significantly larger (8.3 ± 5.8 vs. 5.7 ± 1.0 cm; P = 0.004) in pre-OLT-classified than in post-OLT-classified patients, reflecting the tumor underscoring by pre-OLT evaluation. Accordingly, the HCC recurrence rate was significantly higher in UCSF+ Milan− patients who had been classified pre-OLT compared to patients who had been classified post-OLT (26% vs. 13%; P = 0.05)

Second, microvascular invasion was significantly more frequent on the explanted liver in UCSF+ Milan− patients compared to Milan+ patients (30% vs. 13%; P = 0.009) and the median time between OLT and HCC recurrence was significantly shorter in UCSF+ Milan− patients compared to Milan+ patients (7.6 vs. 22 months; P = 0.009). These findings clearly suggest that UCSF+ Milan− patients (even when assessed on the explanted liver) probably have more aggressive tumors with a more aggressive recurrence course than Milan+ patients.

In general, our results show that the UCSF criteria should not be applied to preoperative evaluation. Indeed, preoperative UCSF criteria clearly identify a subgroup of LT candidates with an intermediate prognosis, with a 5-yr survival rate of 45.6%. This survival rate is about 14.5% less than the 5-yr survival associated with the use of the Milan selection criteria in our series and below the 50% 5-yr survival rate that the transplant community considers as the minimal survival rate consistent with a rational use of liver grafts. Moreover, these results were observed in a cohort with a short waiting time of 4 months and a dropout rate of 2.5%. Therefore, in the majority of scenarios with waiting time between 6 to 12 months and dropout rate of 15 to 25%,7, 13 the expected outcomes should be even worse.

In LT programs with major organ shortages, these results preclude the use of the pre-OLT UCSF criteria for selection of HCC patients, except in the setting of studies reassessing prospectively the accuracy of the UCSF criteria in the long term. On the other hand, in mild organ shortage OLT programs (patient-oriented programs) or in the case of living donation, such criteria could be proposed on an individual basis for selection of patients with an intermediate risk of recurrence, although this latter point deserves further ethical consideration.

In conclusion, this study shows that based on radiological evaluation, the UCSF criteria are associated with a 45.6% 5-yr survival rate in UCSF+ Milan− patients, despite a waiting time of 4 months and a dropout rate of 2.5%. Thus, these criteria cannot be applied nowadays, due to the shortage of donors and the agreement of the community to provide 5-yr survival rates of more than 50%.

Ancillary