ABSTRACT
Objective:To evaluate the accuracy and practicability of hepatocellular carcinoma prediction score (PAGE-B) and modified hepatocellular carcinoma prediction score (mPAGE-B) in predicting the development of hepatocellular carcinoma in patients with hepatitis B virus (HBV)-associated liver cirrhosis and received nucleos(t)ide analogue (NA) treatment.Methods:From June 2009 to December 2014, at Department of Hepatology, the First Affiliated Hospital of Fujian Medical University, the clinical data of 707 patients with HBV-associated liver cirrhosis and received NA treatment were retrospectively collected, and the patients were followed up. The risk factors of development of hepatocellular carcinoma were analyzed. PAGE-B (including platelet count, age, gender), mPAGE-B (including platelet count, age, gender and albumin), Child-Turcotte-Pugh (CTP) score and aspartate aminotransferase to platelet ratio index (APRI) were compared in area under receiver operator characteristic curve (AUROC) for predicting the occurrence of hepatocellular carcinoma within 5 years. Risk stratification analysis was carried out for mPAGE-B and PAGE-B. Multivariate Cox regression analysis, receiver operator characteristic curve, Mann-Whitney U test and Kaplan-Meier method were used for statistical analysis. Results:The age of 707 patients was (46.7±12.2) years old, including 567 males (80.2%) and 140 females (19.8%). The positive rate of hepatitis B e antigen was 56.4% (399/707). The scores of PAGE-B, mPAGE-B, CTP and APRI were 15.90±4.24, 12.39±3.58, 6.88±2.15 and 1.80 (0.85, 3.79), respectively. The overall follow up time was (38.14±20.97) months and the incidence of hepatocellular carcinoma was 8.1% (57/707). The results of multivariate Cox regression analysis showed that advanced age, low platelet count and quantitative reduction of HBV DNA were independent risk factors of development of hepatocellular carcinoma (Wald=20.44, 5.64 and 9.25; HR(95% confidence interval (95% CI) 1.056(1.031 to 1.081), 0.994(0.989 to 0.999) and 0.769(0.649 to 0.911); P<0.001, =0.018 and 0.002). The AUROCs (95% CI) of PAGE-B, mPAGE-B, CTP score and APRI for predicting the occurrence of hepatocellular carcinoma within 5 years were 0.708 (0.639 to 0.778), 0.724 (0.657 to 0.778), 0.576 (0.500 to 0.652) and 0.516 (0.443 to 0.589), respectively. There were no statistically significant differences in AUROCs for predicting the occurrence of hepatocellular carcinoma within 5 years between mPAGE-B and PAGE-B, between APRI and CTP score (both P>0.05). The AUROC for predicting the occurrence of hepatocellular carcinoma within 5 years of CTP score was less than those of PAGE-B and mPAGE-B, and the differences were statistically significant ( Z=3.00 and 3.79; P=0.003, <0.001). The AUROC for predicting the occurrence of hepatocellular carcinoma within 5 years of APRI was less than those of PAGE-B and mPAGE-B, and the differences were statistically significant ( Z=4.75 and 5.46, both P<0.001). There were 51 cases (7.2%), 394 cases (55.7%) and 262 cases (37.1%) in the low-risk (<10) group, medium-risk (10 to 17) group and high-risk (>17) group as assessed by PAGE-B. The incidence of hepatocellular carcinoma was 0(0/51), 4.8% (19/394) and 14.5% (38/262), respectively the annual average incidence of hepatocellular carcinoma was 0, 1.6% and 5.5%, respectively, the 5-year cumulative incidence of hepatocellular carcinoma was 0, 7.3% and 31.3%, respectively. The 5-year cumulative incidence of hepatocellular carcinoma of high-risk group was higher than those of medium-risk group and low-risk group (log-rank test=19.27, P<0.001). There were 97 cases (13.7%), 246 cases (34.8%) and 364 cases (51.5%) in the low-risk group (<9), medium-risk group (9 to 12) and high-risk group (>12) as assessed by mPAGE-B. The incidence of hepatocellular carcinoma was 2.1% (2/97), 3.7% (9/246) and 12.6%(46/364), the annual average incidence of hepatocellular carcinoma was 0.6%, 1.1% and 4.7%, respectively, the 5-year cumulative incidence of hepatocellular carcinoma was 2.4%, 5.1% and 26.7%, respectively. The 5-year cumulative incidence of hepatocellular carcinoma of high-risk group was higher than those of medium-risk group and low-risk group (log-rank test value=18.64, P<0.001). Conclusions:Both PAGE-B and mPAGE-B can predict the occurrence of hepatocellular carcinoma within 5 years in patients with HBV-associated liver cirrhosis treated with antiviral therapy, identify liver cirrhotic patients at high risk of development of hepatocellular carcinoma and guide clinicans to use more efficient screening strategies.