LIVER TRANSPLANTATION 21:101–111, 2015
ORIGINAL ARTICLE
Prognostic Role of Plasma Vascular Endothelial Growth Factor in Patients With Hepatocellular Carcinoma Undergoing Liver Transplantation Wei Zhang,1,3 Richard Kim,4 Cristiano Quintini,1 Koji Hashimoto,1 Masato Fujiki,1 Teresa Diago,1 Bijan Eghtesad,1 Charles Miller,1 John Fung,1 Ann Tan,5 K. V. Narayanan Menon,2 and Federico Aucejo1 1 Hepatobiliary & Liver Transplant Surgery and 2Gastroenterology and Hepatology, Cleveland Clinic Foundation, Cleveland, OH; 3Hepatic Surgery Center, Department of Surgery, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China; 4Department of Gastrointestinal Oncology, H. Lee Moffitt Cancer Center, Tampa, FL; and 5BC Cancer Agency-Abbotsford, Abbotsford, Canada
Vascular endothelial growth factor (VEGF) is pivotal in the development of hepatocellular carcinoma (HCC). Studies have demonstrated the prognostic value of circulating VEGF levels in patients undergoing liver resection or locoregional therapy (LRT) for HCC. We investigated the significance of preoperative plasma VEGF levels in patients with HCC undergoing liver transplantation (LT) at a Western transplant center. Pre-LT plasma VEGF levels were measured with an enzyme-linked immunoassay for 164 patients with HCC undergoing LT. The preoperative plasma VEGF level was correlated with clinicopathological variables and overall and recurrence-free post-LT survival. A higher pre-LT plasma VEGF level was significantly associated with pre-LT LRT (P 5 0.01), multiple tumors (P 5 0.02), a total tumor diameter 5 cm (P 5 0.01), bilobar tumor distribution (P 5 0.03), tumor vascular invasion (VI; P < 0.001), and HCC beyond the Milan criteria (P < 0.001). Patients with a plasma VEGF level > 44 pg/mL had significantly worse overall and disease-free survival than those with VEGF levels 44 pg/mL (P 5 0.04 and P 5 0.02, respectively). In a multivariate analysis, a plasma VEGF level > 44 pg/mL was independently associated with tumor VI (P < 0.001) and recurrence-free survival (hazard ratio 5 2.12, 95% confidence interval 5 1.08-4.14, P 5 0.03). In conclusion, in patients with chronic end-stage liver disease and HCC, a pre-LT plasma VEGF level > 44 pg/mL C 2014 AASLD. may be a predictor of tumor VI and recurrence-free post-LT survival. Liver Transpl 21:101-111, 2015. V Received May 21, 2014; accepted September 29, 2014. Liver transplantation (LT) stands as the most important surgical therapy for treating patients with hepatocellular carcinoma (HCC) and end-stage liver disease and for treating patients for whom resection has failed because of tumor recurrence or liver insufficiency.1 In patients within the Milan criteria (single tumor 5 cm or up to 3 tumors, each < 3 cm), LT has been consistently associated with excellent long-term results.2 However, there are numerous reports recognizing groups of patients with extended criteria HCC able to
exhibit survival similar to that of patients within the Milan criteria.3 Therefore, the tumor biology, in addition to mere morphometric characteristics, has a critical impact in determining patient outcomes.4 Numerous studies have investigated various tissue and serum HCC biomarkers, including the presence of circulating cancerous cells and their correlation with patient survival, but further research is needed to incorporate these into clinical practice along with scoring systems capable of defining patient selection for LT
Abbreviations: AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; LRT, locoregional therapy; LT, liver transplantation; OS, overall survival; PIVKAII, protein induced by vitamin K absence; RFS, recurrence-free survival; TACE, transarterial chemoembolization; VEGF, vascular endothelial growth factor; VI, vascular invasion. Potential conflict of interest: Nothing to report. Address reprint requests to Federico Aucejo, M.D., Hepatobiliary & Liver Transplant Surgery, Cleveland Clinic Foundation, A100, 9500 Euclid Avenue, Cleveland, OH 44195. Telephone: 216-445-7159; FAX: 216-449-9375; E-mail: [email protected] DOI 10.1002/lt.24013 View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases
C 2014 American Association for the Study of Liver Diseases. V
102 ZHANG ET AL.
or other therapies.5 Vascular invasion (VI) is one of the strongest predictors of tumor relapse after LT.6 Unfortunately, the presence of VI can be determined only at the time of surgery via explant histopathological analysis; meanwhile, accurate pretransplant predictors of VI are lacking. Pretransplant HCC biopsy for assessing molecular biomarkers has been attempted,7 but HCC lesions in the same patient can be heterogeneous and are often not amenable to percutaneous biopsy; additionally, there is an associated risk of cancerous cell seeding, which makes this practice impractical.8 Reported serum biomarkers of a poor prognosis for patients with HCC include the following: alphafetoprotein (AFP), lens culinaris agglutinin–reactive AFP percentage of the total AFP concentration, and protein induced by vitamin K absence II (PIVKAII).9 However, many HCCs do not secrete these biomarkers or do so only in moderate amounts, and this undermines their clinical relevance. Additionally, these biomarkers can be elevated in patients with liver disease in the absence of HCC.10 Vascular endothelial growth factor (VEGF), a protein that promotes angiogenesis, has an important role in the development of HCC by fostering tumor cell survival, proliferation, and vessel formation.11 Previous studies have demonstrated that high serum VEGF levels can predict a poor prognosis after liver resection,12 radiofrequency ablation,13 transcatheter arterial chemoembolization,14,15 or hepatic arterial infusion chemotherapy.16 However, these experiences have been reported mostly from Asian centers, and the reproducibility of these results in the West and in the setting of LT is lacking. The aim of this study was to determine the association between plasma VEGF levels and the clinicopathological features and survival of patients with HCC undergoing LT at a single Western center.
LIVER TRANSPLANTATION, January 2015
Figure 1. Consolidated Standards of Reporting Trials diagram of the study population.
exhibit disease progression after locoregional therapy (LRT) were also eligible for LT. A histopathological examination of the explanted liver was performed by expert liver pathologists. The length of the longest axis of HCC was defined as the tumor diameter, and the number of lesions was determined via the counting of only viable ones. The degree of tumor necrosis and the differentiation and presence of lymphovascular invasion were described. The tumor burden was staged according to the pathological tumor-node-metastasis system and graded according to the Edmondson and Steiner system. For tumors with heterogeneous grades of differentiation, the most aggressive grade was considered.
PATIENTS AND METHODS From January 2007 to December 2013, 293 patients with HCC underwent LT at the Cleveland Clinic. With approval of the institutional review board, 164 of the 293 patients were analyzed in this study. Twenty-seven patients with incidental HCC that had not been diagnosed by pre-LT imaging, 5 patients undergoing transplantation for HCC recurrence after surgical resection, 5 patients undergoing transplantation for cholangiocarcinoma, and 92 patients without pre-LT plasma VEGF values were excluded, as outlined in Fig. 1. HCC was diagnosed preoperatively either by the characteristic appearance on cross-sectional imaging (computed tomography or magnetic resonance) and serum AFP levels in accordance with standard criteria17 or by biopsy in patients with atypical imaging characteristics. The diagnosis of HCC was confirmed by histopathological explant analysis. Patients with HCC within the Milan criteria were eligible for LT and granted exception Model for End-Stage Liver Disease points. Patients with HCC beyond the Milan criteria without evidence of vascular or extrahepatic spread who were successfully down-staged and did not
Assay of the Plasma VEGF Level Peripheral venous blood samples were taken from HCC patients every 3 months until the time of LT. Plasma was prepared from 3 to 5 mL of the peripheral venous blood collected in ethylenediaminetetraacetic acid–containing tubes. Samples were then centrifuged at 4 C for 15 minutes (3000 rpm), removed, aliquoted, and snapfrozen at 220 C. Plasma VEGF-A (VEGF165 isoform) was measured with an enzyme-linked immunoassay (Quantikine human VEGF enzyme-linked immunosorbent assay kit; R&D Systems, Minneapolis, MN). With this technique, a monoclonal antibody specific for VEGF was precoated onto a microplate. Samples and standards were pipetted into the wells of microtiter plates, and any VEGF present was bound by the immobilized antibody. After the unbound substances had been washed away, an enzyme-linked polyclonal antibody specific for VEGF was added to the wells. After a wash to remove any unbound antibodyenzyme reagent, a substrate solution was added to the wells, and color developed in proportion to the amount of VEGF bound in the initial step. The color
LIVER TRANSPLANTATION, Vol. 21, No. 1, 2015
ZHANG ET AL. 103
development was stopped, and the intensity of the color was measured. Most patients had 1 plasma VEGF level determination because of the short waiting time to LT from listing. For the small number of patients with more than 1 determination, the last level before LT was considered for analysis. For those patients who received pre-LT LRT, the plasma VEGF level obtained after LRT was considered for analysis.
level for the entire cohort was 15.5 ng/mL (3.02929.0 ng/mL). The median platelet count was 84.5 3 109/L (12.0-317.0 3 109/L). Among the 164 patients, 132 patients (80.5%) underwent LRT at least once before LT; this included transarterial chemoembolization (TACE), radiofrequency ablation, and radioembolization. The median waiting time for LT was 59.5 days (range 5 3-558 days).
Patient Follow-Up
Histopathological Analysis
The post-LT follow-up consisted of chest, abdominal, and pelvic contrast triphasic computed tomography and/or contrast magnetic resonance imaging at 3- to 6-month intervals for the first 3 years after transplantation and at 6- to 12-month intervals for years 4 and 5 after transplantation. Tumor markers including AFP, AFP-L3%, PIVKAII, and plasma VEGF were obtained at similar intervals. The post-LT tumor recurrence diagnosis was established on the basis of imaging and biopsy.
Necrosis of HCC was present in 121 patients (73.8%). Among them, 114 patients (94.2%) had received pre-LT LRT. Tumor necrosis was rated as 0% to 50%, 51% to 99%, and 100% in 88 (53.7%), 64 (39.0%), and 12 patients (7.3%), respectively. The median total tumor diameter was 3.8 cm (range 5 0.5-16.4 cm). The median number of lesions was 2 (range 5 1-9). Multifocal disease was present in 51.8% (n 5 85), and a bilobar distribution was present in 31.1% (n 5 51). Twenty-eight patients (17.1%) had well-differentiated HCC, 124 (75.6%) had moderately differentiated HCC, and 12 (7.3%) had poorly differentiated HCC. VI was present in 47 patients (28.7%). One hundred twenty-seven patients (77.4%) had HCC within the Milan criteria. Among the 37 patients with HCC beyond the Milan criteria, 22 patients (59.5%) had VI.
Statistical Analysis Continuous variables were expressed as medians and ranges and were compared with the Mann-Whitney U test. Categorical data were compared with the v2 test or Fisher’s exact test. The cutoff value for plasma VEGF was determined according to the best discrimination between patients with or without tumor VI with a receiver operating characteristic curve analysis. A logistic regression analysis was performed to identify independent variables for VI. Among 164 patients, the survival analysis was performed only for the 113 patients who reached a minimal post-LT follow-up of 24 months. Cumulative overall survival (OS) and recurrence-free survival (RFS) were calculated with Kaplan-Meier methods, and the difference between the curves was evaluated with log-rank testing. A multivariate analysis with a Cox model was performed for the factors identified as significant for OS and RFS in the univariate analysis. The differences were considered to be significant at P < 0.05. All statistical analyses were performed with SPSS 17.0 (SPSS, Inc., Chicago, IL).
RESULTS Patient Demographics The median age for the 164 patients (128 men and 36 women) was 62 years (range 5 41-78 years). Hepatitis C virus (HCV) infection was the most frequent etiology of cirrhosis [98 patients (59.8%)]. The Child-Pugh classification was A for 102 patients (62.2%), B for 49 patients (29.9%), and C for 13 patients (7.9%). The Barcelona Clinic Liver Cancer classification was A for 119 patients (72.6%) and B for 45 patients (27.4%). The median chemical Model for End-Stage Liver Disease score was 10.0 (range 5 6-19). The median plasma VEGF level for the entire cohort was 31.0 pg/ mL (30.0-272.0 pg/mL), and the median serum AFP
Association Between the Pre-LT Plasma VEGF Level and Clinicopathological Variables No significant association was observed between the plasma VEGF level and sex, age, HCV infection, Child-Pugh class, or ascites. A trend toward higher plasma VEGF levels was observed in patients with serum AFP levels > 100 ng/mL (median: 37.0 versus 30.5 pg/mL, P 5 0.07). The plasma VEGF level was significantly higher in patients who received LRT before transplantation versus those who did not (median: 34.0 versus 30.0 pg/mL, P 5 0.01). The median time between LRT and plasma VEGF measurement was 51 days. Significantly higher plasma VEGF levels were associated with multiple tumor lesions and a bilobar distribution (P 5 0.02 and P 5 0.03, respectively). The plasma VEGF level in patients with a total tumor diameter 5 cm was higher than the levels in patients with a total tumor diameter < 5 cm (median: 36.5 versus 30.0 pg/mL, P 5 0.01). Patients with HCC with VI had higher plasma VEGF levels than patients without VI (median: 48.0 versus 30.0 pg/mL, P < 0.001). A significantly higher plasma VEGF level was also associated with HCC beyond the Milan criteria (median: 48.0 versus 30.0 pg/mL, P < 0.001; Table 1). Predictors of tumor VI are shown in Table 2. The cutoff value of 44 pg/mL for VEGF obtained from the receiver operating characteristic curve analysis was used in the univariate analysis. The VI group included more patients under the age of 65 years old in comparison with the group without VI (P 5 0.02). The
104 ZHANG ET AL.
LIVER TRANSPLANTATION, January 2015
TABLE 1. Association Between Pre-LT Plasma VEGF Levels and Clinicopathological Variables Variable Sex Male Female Age 65 years >65years Etiology HCV-positive HCV-negative Child-Pugh class Child A Child B Child C Ascites Present Absent AFP 100 ng/mL >100 ng/mL Pre-LT LRT Present Absent Tumor number Solitary Multiple Total tumor diameter 44 pg/mL, a total tumor diameter 5 cm, and poor/moderate tumor differentiation (P < 0.001, P 5 0.001, and P 5 0.013, respectively; Table 3).
Post-LT OS and RFS For the 113 patients, the median post-LT follow-up was 45.8 months (range 5 3-86.3 months). The 1-, 3-, and 5-year OS rates were 93.8%, 81.2%, and 69.4%, respectively. The 1-, 3-, and 5-year RFS rates were 86.7%, 71.9%, and 64.5%, respectively. By the time of the data analysis, 27 patients (23.9%) had developed post-LT HCC recurrence. The median time to HCC recurrence was 22.5 months (range 5 3.1-54.6 months). Seventy-seven patients were alive without recurrence (68.1%), 7 patients were alive with recurrence (6.2%), and 29 patients had died (25.7%). Among these 29 patients, 20 died of tumor recurrence, 2 died of liver failure, 2 died of cardiac failure,
LIVER TRANSPLANTATION, Vol. 21, No. 1, 2015
ZHANG ET AL. 105
1 died of respiratory failure, 2 died of multiorgan failure from graft-versus-host disease, 1 died of throat cancer, and 1 died of an undetermined cause.
Prognostic Significance of Pre-LT Plasma VEGF Level Figure 2 shows the prognostic value of the plasma VEGF level for OS and RFS in 113 patients with the cutoff value of 44 pg/mL. Patients with a plasma VEGF level > 44 pg/mL had significantly worse
TABLE 2. Univariate Analysis of Clinicopathological Factors Related to VI
Variable
No VI
VI
P
(n 5 117)
(n 5 47)
Value
92 25
36 11
0.78
69 48
37 10
0.02
81 36
32 15
0.89
97 20
30 17
0.008
92 25
18 29
44 pg/mL Child-Pugh class A B C Waiting time 0-3 months >3 months Pre-LT LRT Present Absent Tumor number Solitary Multiple Total tumor diameter 44 pg/mL (P 5 0.04), a total tumor diameter 5 cm (P 5 0.004), tumor VI (P < 0.001), poor/moderate tumor differentiation (P 5 0.04), and a tumor beyond the Milan criteria (P 5 0.002) were significant risk factors that adversely affected OS. Significant prognostic factors for RFS included a plasma VEGF level > 44 pg/mL (P 5 0.02), a total tumor diameter 5 cm (P 5 0.002), tumor VI (P < 0.001), and a tumor beyond the Milan criteria (P 5 0.001; Table 4). In the multivariate analysis, VI was an independent predictor of OS (P < 0.001). A plasma VEGF level > 44 pg/mL and a total tumor diameter 5 cm were independent predictors of RFS (P 5 0.03 and P 5 0.005, respectively; Table 5). Eleven patients who had VI with a low plasma VEGF level (44 pg/mL, P 5 0.02; Fig. 3). With respect to Milan criteria classification, patients were stratified into 4 groups (Fig. 4A): (1) within the Milan criteria with a VEGF level 44 pg/mL (n 5 58), (2) within the Milan criteria with a VEGF level > 44 pg/mL (n 5 25), (3) beyond the Milan criteria with a VEGF level 44 pg/mL (n 5 11), and (4) beyond the Milan criteria with a VEGF level > 44 pg/mL (n 5 19). The 5-year RFS rates for patients who met the Milan criteria with a VEGF level 44 pg/mL and for patients who met the Milan criteria with a VEGF level > 44 pg/ mL were 93.9% and 68.0%, respectively (P 5 0.04). For patients beyond the Milan criteria with a VEGF level < 44 pg/mL, 5-year RFS was 72.7%; for patients beyond the Milan criteria with a VEGF level > 44 pg/ mL, it was 18.4% (P 5 0.003). Patients beyond the Milan criteria with a VEGF level 44 pg/mL had RFS similar to that of patients within the Milan criteria (P 5 0.31; Fig. 4B). Among the 113 patients for whom the survival analysis was performed, 88 patients received LRT. Patients with a VEGF level 44 pg/mL had better RFS than those with a VEGF level > 44 pg/mL (P 5 0.002). For the patients who did not receive LRT (n 5 25), better RFS was also observed when the VEGF level was 44 pg/mL (P 5 0.002; Fig. 5A,B).
Post-LT Plasma VEGF Level The patients who experienced post-LT HCC recurrence had higher median post-LT plasma VEGF levels than the patients without post-LT HCC recurrence (53.5 versus 33.0 pg/mL, P 5 0.001). Notably, in the group of patients with HCC recurrence, post-LT plasma VEGF levels dropped significantly in comparison with pre-LT levels (112.5 versus 45.0 pg/mL, P < 0.001) and increased significantly at the time of HCC recurrence (45.0 versus 71.0 pg/mL, P 5 0.001). With respect to the patients without post-LT HCC
106 ZHANG ET AL.
LIVER TRANSPLANTATION, January 2015
TABLE 3. Multivariate Analysis of Independent Predictors of VI Variable Plasma VEGF: >44 versus 44 pg/mL Total tumor diameter: 5 versus 100 ng/mL) Plasma VEGF (>44 pg/mL) Child-Pugh class (C) Ascites (absent) Waiting time (>3 months) Previous LRT (yes) Tumor number (multiple) Total tumor diameter (5 cm) Tumor location (bilobar) VI (present) Tumor differentiation (poor/moderate) Milan criteria (beyond)
RFS
Interval)
P Value
1.11 (0.47-2.62) 2.04 (0.89-4.80) 1.04 (0.46-2.34) 1.24 (0.55-2.82) 2.13 (1.00-3.98) 1.07 (0.61-1.85) 0.57 (0.28-1.19) 1.73 (0.70-4.25) 2.10 (0.72-6.10) 1.39 (0.66-2.94) 2.85 (1.34-6.06) 1.53 (0.73-3.21) 3.82 (1.80-8.10) 6.47 (0.88-47.65) 3.05 (1.47-6.31)
0.81 0.10 0.93 0.61 0.04 0.95 0.13 0.23 0.16 0.39 0.004 0.26
ORIGINAL ARTICLE
Prognostic Role of Plasma Vascular Endothelial Growth Factor in Patients With Hepatocellular Carcinoma Undergoing Liver Transplantation Wei Zhang,1,3 Richard Kim,4 Cristiano Quintini,1 Koji Hashimoto,1 Masato Fujiki,1 Teresa Diago,1 Bijan Eghtesad,1 Charles Miller,1 John Fung,1 Ann Tan,5 K. V. Narayanan Menon,2 and Federico Aucejo1 1 Hepatobiliary & Liver Transplant Surgery and 2Gastroenterology and Hepatology, Cleveland Clinic Foundation, Cleveland, OH; 3Hepatic Surgery Center, Department of Surgery, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China; 4Department of Gastrointestinal Oncology, H. Lee Moffitt Cancer Center, Tampa, FL; and 5BC Cancer Agency-Abbotsford, Abbotsford, Canada
Vascular endothelial growth factor (VEGF) is pivotal in the development of hepatocellular carcinoma (HCC). Studies have demonstrated the prognostic value of circulating VEGF levels in patients undergoing liver resection or locoregional therapy (LRT) for HCC. We investigated the significance of preoperative plasma VEGF levels in patients with HCC undergoing liver transplantation (LT) at a Western transplant center. Pre-LT plasma VEGF levels were measured with an enzyme-linked immunoassay for 164 patients with HCC undergoing LT. The preoperative plasma VEGF level was correlated with clinicopathological variables and overall and recurrence-free post-LT survival. A higher pre-LT plasma VEGF level was significantly associated with pre-LT LRT (P 5 0.01), multiple tumors (P 5 0.02), a total tumor diameter 5 cm (P 5 0.01), bilobar tumor distribution (P 5 0.03), tumor vascular invasion (VI; P < 0.001), and HCC beyond the Milan criteria (P < 0.001). Patients with a plasma VEGF level > 44 pg/mL had significantly worse overall and disease-free survival than those with VEGF levels 44 pg/mL (P 5 0.04 and P 5 0.02, respectively). In a multivariate analysis, a plasma VEGF level > 44 pg/mL was independently associated with tumor VI (P < 0.001) and recurrence-free survival (hazard ratio 5 2.12, 95% confidence interval 5 1.08-4.14, P 5 0.03). In conclusion, in patients with chronic end-stage liver disease and HCC, a pre-LT plasma VEGF level > 44 pg/mL C 2014 AASLD. may be a predictor of tumor VI and recurrence-free post-LT survival. Liver Transpl 21:101-111, 2015. V Received May 21, 2014; accepted September 29, 2014. Liver transplantation (LT) stands as the most important surgical therapy for treating patients with hepatocellular carcinoma (HCC) and end-stage liver disease and for treating patients for whom resection has failed because of tumor recurrence or liver insufficiency.1 In patients within the Milan criteria (single tumor 5 cm or up to 3 tumors, each < 3 cm), LT has been consistently associated with excellent long-term results.2 However, there are numerous reports recognizing groups of patients with extended criteria HCC able to
exhibit survival similar to that of patients within the Milan criteria.3 Therefore, the tumor biology, in addition to mere morphometric characteristics, has a critical impact in determining patient outcomes.4 Numerous studies have investigated various tissue and serum HCC biomarkers, including the presence of circulating cancerous cells and their correlation with patient survival, but further research is needed to incorporate these into clinical practice along with scoring systems capable of defining patient selection for LT
Abbreviations: AFP, alpha-fetoprotein; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; LRT, locoregional therapy; LT, liver transplantation; OS, overall survival; PIVKAII, protein induced by vitamin K absence; RFS, recurrence-free survival; TACE, transarterial chemoembolization; VEGF, vascular endothelial growth factor; VI, vascular invasion. Potential conflict of interest: Nothing to report. Address reprint requests to Federico Aucejo, M.D., Hepatobiliary & Liver Transplant Surgery, Cleveland Clinic Foundation, A100, 9500 Euclid Avenue, Cleveland, OH 44195. Telephone: 216-445-7159; FAX: 216-449-9375; E-mail: [email protected] DOI 10.1002/lt.24013 View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases
C 2014 American Association for the Study of Liver Diseases. V
102 ZHANG ET AL.
or other therapies.5 Vascular invasion (VI) is one of the strongest predictors of tumor relapse after LT.6 Unfortunately, the presence of VI can be determined only at the time of surgery via explant histopathological analysis; meanwhile, accurate pretransplant predictors of VI are lacking. Pretransplant HCC biopsy for assessing molecular biomarkers has been attempted,7 but HCC lesions in the same patient can be heterogeneous and are often not amenable to percutaneous biopsy; additionally, there is an associated risk of cancerous cell seeding, which makes this practice impractical.8 Reported serum biomarkers of a poor prognosis for patients with HCC include the following: alphafetoprotein (AFP), lens culinaris agglutinin–reactive AFP percentage of the total AFP concentration, and protein induced by vitamin K absence II (PIVKAII).9 However, many HCCs do not secrete these biomarkers or do so only in moderate amounts, and this undermines their clinical relevance. Additionally, these biomarkers can be elevated in patients with liver disease in the absence of HCC.10 Vascular endothelial growth factor (VEGF), a protein that promotes angiogenesis, has an important role in the development of HCC by fostering tumor cell survival, proliferation, and vessel formation.11 Previous studies have demonstrated that high serum VEGF levels can predict a poor prognosis after liver resection,12 radiofrequency ablation,13 transcatheter arterial chemoembolization,14,15 or hepatic arterial infusion chemotherapy.16 However, these experiences have been reported mostly from Asian centers, and the reproducibility of these results in the West and in the setting of LT is lacking. The aim of this study was to determine the association between plasma VEGF levels and the clinicopathological features and survival of patients with HCC undergoing LT at a single Western center.
LIVER TRANSPLANTATION, January 2015
Figure 1. Consolidated Standards of Reporting Trials diagram of the study population.
exhibit disease progression after locoregional therapy (LRT) were also eligible for LT. A histopathological examination of the explanted liver was performed by expert liver pathologists. The length of the longest axis of HCC was defined as the tumor diameter, and the number of lesions was determined via the counting of only viable ones. The degree of tumor necrosis and the differentiation and presence of lymphovascular invasion were described. The tumor burden was staged according to the pathological tumor-node-metastasis system and graded according to the Edmondson and Steiner system. For tumors with heterogeneous grades of differentiation, the most aggressive grade was considered.
PATIENTS AND METHODS From January 2007 to December 2013, 293 patients with HCC underwent LT at the Cleveland Clinic. With approval of the institutional review board, 164 of the 293 patients were analyzed in this study. Twenty-seven patients with incidental HCC that had not been diagnosed by pre-LT imaging, 5 patients undergoing transplantation for HCC recurrence after surgical resection, 5 patients undergoing transplantation for cholangiocarcinoma, and 92 patients without pre-LT plasma VEGF values were excluded, as outlined in Fig. 1. HCC was diagnosed preoperatively either by the characteristic appearance on cross-sectional imaging (computed tomography or magnetic resonance) and serum AFP levels in accordance with standard criteria17 or by biopsy in patients with atypical imaging characteristics. The diagnosis of HCC was confirmed by histopathological explant analysis. Patients with HCC within the Milan criteria were eligible for LT and granted exception Model for End-Stage Liver Disease points. Patients with HCC beyond the Milan criteria without evidence of vascular or extrahepatic spread who were successfully down-staged and did not
Assay of the Plasma VEGF Level Peripheral venous blood samples were taken from HCC patients every 3 months until the time of LT. Plasma was prepared from 3 to 5 mL of the peripheral venous blood collected in ethylenediaminetetraacetic acid–containing tubes. Samples were then centrifuged at 4 C for 15 minutes (3000 rpm), removed, aliquoted, and snapfrozen at 220 C. Plasma VEGF-A (VEGF165 isoform) was measured with an enzyme-linked immunoassay (Quantikine human VEGF enzyme-linked immunosorbent assay kit; R&D Systems, Minneapolis, MN). With this technique, a monoclonal antibody specific for VEGF was precoated onto a microplate. Samples and standards were pipetted into the wells of microtiter plates, and any VEGF present was bound by the immobilized antibody. After the unbound substances had been washed away, an enzyme-linked polyclonal antibody specific for VEGF was added to the wells. After a wash to remove any unbound antibodyenzyme reagent, a substrate solution was added to the wells, and color developed in proportion to the amount of VEGF bound in the initial step. The color
LIVER TRANSPLANTATION, Vol. 21, No. 1, 2015
ZHANG ET AL. 103
development was stopped, and the intensity of the color was measured. Most patients had 1 plasma VEGF level determination because of the short waiting time to LT from listing. For the small number of patients with more than 1 determination, the last level before LT was considered for analysis. For those patients who received pre-LT LRT, the plasma VEGF level obtained after LRT was considered for analysis.
level for the entire cohort was 15.5 ng/mL (3.02929.0 ng/mL). The median platelet count was 84.5 3 109/L (12.0-317.0 3 109/L). Among the 164 patients, 132 patients (80.5%) underwent LRT at least once before LT; this included transarterial chemoembolization (TACE), radiofrequency ablation, and radioembolization. The median waiting time for LT was 59.5 days (range 5 3-558 days).
Patient Follow-Up
Histopathological Analysis
The post-LT follow-up consisted of chest, abdominal, and pelvic contrast triphasic computed tomography and/or contrast magnetic resonance imaging at 3- to 6-month intervals for the first 3 years after transplantation and at 6- to 12-month intervals for years 4 and 5 after transplantation. Tumor markers including AFP, AFP-L3%, PIVKAII, and plasma VEGF were obtained at similar intervals. The post-LT tumor recurrence diagnosis was established on the basis of imaging and biopsy.
Necrosis of HCC was present in 121 patients (73.8%). Among them, 114 patients (94.2%) had received pre-LT LRT. Tumor necrosis was rated as 0% to 50%, 51% to 99%, and 100% in 88 (53.7%), 64 (39.0%), and 12 patients (7.3%), respectively. The median total tumor diameter was 3.8 cm (range 5 0.5-16.4 cm). The median number of lesions was 2 (range 5 1-9). Multifocal disease was present in 51.8% (n 5 85), and a bilobar distribution was present in 31.1% (n 5 51). Twenty-eight patients (17.1%) had well-differentiated HCC, 124 (75.6%) had moderately differentiated HCC, and 12 (7.3%) had poorly differentiated HCC. VI was present in 47 patients (28.7%). One hundred twenty-seven patients (77.4%) had HCC within the Milan criteria. Among the 37 patients with HCC beyond the Milan criteria, 22 patients (59.5%) had VI.
Statistical Analysis Continuous variables were expressed as medians and ranges and were compared with the Mann-Whitney U test. Categorical data were compared with the v2 test or Fisher’s exact test. The cutoff value for plasma VEGF was determined according to the best discrimination between patients with or without tumor VI with a receiver operating characteristic curve analysis. A logistic regression analysis was performed to identify independent variables for VI. Among 164 patients, the survival analysis was performed only for the 113 patients who reached a minimal post-LT follow-up of 24 months. Cumulative overall survival (OS) and recurrence-free survival (RFS) were calculated with Kaplan-Meier methods, and the difference between the curves was evaluated with log-rank testing. A multivariate analysis with a Cox model was performed for the factors identified as significant for OS and RFS in the univariate analysis. The differences were considered to be significant at P < 0.05. All statistical analyses were performed with SPSS 17.0 (SPSS, Inc., Chicago, IL).
RESULTS Patient Demographics The median age for the 164 patients (128 men and 36 women) was 62 years (range 5 41-78 years). Hepatitis C virus (HCV) infection was the most frequent etiology of cirrhosis [98 patients (59.8%)]. The Child-Pugh classification was A for 102 patients (62.2%), B for 49 patients (29.9%), and C for 13 patients (7.9%). The Barcelona Clinic Liver Cancer classification was A for 119 patients (72.6%) and B for 45 patients (27.4%). The median chemical Model for End-Stage Liver Disease score was 10.0 (range 5 6-19). The median plasma VEGF level for the entire cohort was 31.0 pg/ mL (30.0-272.0 pg/mL), and the median serum AFP
Association Between the Pre-LT Plasma VEGF Level and Clinicopathological Variables No significant association was observed between the plasma VEGF level and sex, age, HCV infection, Child-Pugh class, or ascites. A trend toward higher plasma VEGF levels was observed in patients with serum AFP levels > 100 ng/mL (median: 37.0 versus 30.5 pg/mL, P 5 0.07). The plasma VEGF level was significantly higher in patients who received LRT before transplantation versus those who did not (median: 34.0 versus 30.0 pg/mL, P 5 0.01). The median time between LRT and plasma VEGF measurement was 51 days. Significantly higher plasma VEGF levels were associated with multiple tumor lesions and a bilobar distribution (P 5 0.02 and P 5 0.03, respectively). The plasma VEGF level in patients with a total tumor diameter 5 cm was higher than the levels in patients with a total tumor diameter < 5 cm (median: 36.5 versus 30.0 pg/mL, P 5 0.01). Patients with HCC with VI had higher plasma VEGF levels than patients without VI (median: 48.0 versus 30.0 pg/mL, P < 0.001). A significantly higher plasma VEGF level was also associated with HCC beyond the Milan criteria (median: 48.0 versus 30.0 pg/mL, P < 0.001; Table 1). Predictors of tumor VI are shown in Table 2. The cutoff value of 44 pg/mL for VEGF obtained from the receiver operating characteristic curve analysis was used in the univariate analysis. The VI group included more patients under the age of 65 years old in comparison with the group without VI (P 5 0.02). The
104 ZHANG ET AL.
LIVER TRANSPLANTATION, January 2015
TABLE 1. Association Between Pre-LT Plasma VEGF Levels and Clinicopathological Variables Variable Sex Male Female Age 65 years >65years Etiology HCV-positive HCV-negative Child-Pugh class Child A Child B Child C Ascites Present Absent AFP 100 ng/mL >100 ng/mL Pre-LT LRT Present Absent Tumor number Solitary Multiple Total tumor diameter 44 pg/mL, a total tumor diameter 5 cm, and poor/moderate tumor differentiation (P < 0.001, P 5 0.001, and P 5 0.013, respectively; Table 3).
Post-LT OS and RFS For the 113 patients, the median post-LT follow-up was 45.8 months (range 5 3-86.3 months). The 1-, 3-, and 5-year OS rates were 93.8%, 81.2%, and 69.4%, respectively. The 1-, 3-, and 5-year RFS rates were 86.7%, 71.9%, and 64.5%, respectively. By the time of the data analysis, 27 patients (23.9%) had developed post-LT HCC recurrence. The median time to HCC recurrence was 22.5 months (range 5 3.1-54.6 months). Seventy-seven patients were alive without recurrence (68.1%), 7 patients were alive with recurrence (6.2%), and 29 patients had died (25.7%). Among these 29 patients, 20 died of tumor recurrence, 2 died of liver failure, 2 died of cardiac failure,
LIVER TRANSPLANTATION, Vol. 21, No. 1, 2015
ZHANG ET AL. 105
1 died of respiratory failure, 2 died of multiorgan failure from graft-versus-host disease, 1 died of throat cancer, and 1 died of an undetermined cause.
Prognostic Significance of Pre-LT Plasma VEGF Level Figure 2 shows the prognostic value of the plasma VEGF level for OS and RFS in 113 patients with the cutoff value of 44 pg/mL. Patients with a plasma VEGF level > 44 pg/mL had significantly worse
TABLE 2. Univariate Analysis of Clinicopathological Factors Related to VI
Variable
No VI
VI
P
(n 5 117)
(n 5 47)
Value
92 25
36 11
0.78
69 48
37 10
0.02
81 36
32 15
0.89
97 20
30 17
0.008
92 25
18 29
44 pg/mL Child-Pugh class A B C Waiting time 0-3 months >3 months Pre-LT LRT Present Absent Tumor number Solitary Multiple Total tumor diameter 44 pg/mL (P 5 0.04), a total tumor diameter 5 cm (P 5 0.004), tumor VI (P < 0.001), poor/moderate tumor differentiation (P 5 0.04), and a tumor beyond the Milan criteria (P 5 0.002) were significant risk factors that adversely affected OS. Significant prognostic factors for RFS included a plasma VEGF level > 44 pg/mL (P 5 0.02), a total tumor diameter 5 cm (P 5 0.002), tumor VI (P < 0.001), and a tumor beyond the Milan criteria (P 5 0.001; Table 4). In the multivariate analysis, VI was an independent predictor of OS (P < 0.001). A plasma VEGF level > 44 pg/mL and a total tumor diameter 5 cm were independent predictors of RFS (P 5 0.03 and P 5 0.005, respectively; Table 5). Eleven patients who had VI with a low plasma VEGF level (44 pg/mL, P 5 0.02; Fig. 3). With respect to Milan criteria classification, patients were stratified into 4 groups (Fig. 4A): (1) within the Milan criteria with a VEGF level 44 pg/mL (n 5 58), (2) within the Milan criteria with a VEGF level > 44 pg/mL (n 5 25), (3) beyond the Milan criteria with a VEGF level 44 pg/mL (n 5 11), and (4) beyond the Milan criteria with a VEGF level > 44 pg/mL (n 5 19). The 5-year RFS rates for patients who met the Milan criteria with a VEGF level 44 pg/mL and for patients who met the Milan criteria with a VEGF level > 44 pg/ mL were 93.9% and 68.0%, respectively (P 5 0.04). For patients beyond the Milan criteria with a VEGF level < 44 pg/mL, 5-year RFS was 72.7%; for patients beyond the Milan criteria with a VEGF level > 44 pg/ mL, it was 18.4% (P 5 0.003). Patients beyond the Milan criteria with a VEGF level 44 pg/mL had RFS similar to that of patients within the Milan criteria (P 5 0.31; Fig. 4B). Among the 113 patients for whom the survival analysis was performed, 88 patients received LRT. Patients with a VEGF level 44 pg/mL had better RFS than those with a VEGF level > 44 pg/mL (P 5 0.002). For the patients who did not receive LRT (n 5 25), better RFS was also observed when the VEGF level was 44 pg/mL (P 5 0.002; Fig. 5A,B).
Post-LT Plasma VEGF Level The patients who experienced post-LT HCC recurrence had higher median post-LT plasma VEGF levels than the patients without post-LT HCC recurrence (53.5 versus 33.0 pg/mL, P 5 0.001). Notably, in the group of patients with HCC recurrence, post-LT plasma VEGF levels dropped significantly in comparison with pre-LT levels (112.5 versus 45.0 pg/mL, P < 0.001) and increased significantly at the time of HCC recurrence (45.0 versus 71.0 pg/mL, P 5 0.001). With respect to the patients without post-LT HCC
106 ZHANG ET AL.
LIVER TRANSPLANTATION, January 2015
TABLE 3. Multivariate Analysis of Independent Predictors of VI Variable Plasma VEGF: >44 versus 44 pg/mL Total tumor diameter: 5 versus 100 ng/mL) Plasma VEGF (>44 pg/mL) Child-Pugh class (C) Ascites (absent) Waiting time (>3 months) Previous LRT (yes) Tumor number (multiple) Total tumor diameter (5 cm) Tumor location (bilobar) VI (present) Tumor differentiation (poor/moderate) Milan criteria (beyond)
RFS
Interval)
P Value
1.11 (0.47-2.62) 2.04 (0.89-4.80) 1.04 (0.46-2.34) 1.24 (0.55-2.82) 2.13 (1.00-3.98) 1.07 (0.61-1.85) 0.57 (0.28-1.19) 1.73 (0.70-4.25) 2.10 (0.72-6.10) 1.39 (0.66-2.94) 2.85 (1.34-6.06) 1.53 (0.73-3.21) 3.82 (1.80-8.10) 6.47 (0.88-47.65) 3.05 (1.47-6.31)
0.81 0.10 0.93 0.61 0.04 0.95 0.13 0.23 0.16 0.39 0.004 0.26
