Abstract: Objective Radiation-induced liver disease (RILD) has been considered as the most severe complication in liver irradiation.However,RILD does not have effective predictors.This study was to identify dosimetric predictors for RILD in primary liver carcinoma (PLC) patients with Child-Pugh grade A cirrhosis after being treated with hypofractionated conformal radiotherapy (CRT).Methods Between August 2000 and June 2007,a total of 114 eligible patients were included.The median age was 45 years (23~79 years).Portal vein thrombosis (PVT) was detected in 22 patients.The mean gross tumor volume (GTV) was (378.3 ± 308.1) cm3.A median dose of 53 Gy (40~68 Gy) was delivered to the PLC by hypofractionated CRT (three fractions per week) with a median fraction size of 4.6 Gy (4~6 Gy).Thirty-eight patients had previously received transcatheter arterial chemoembolization.All patients had intact treatment planning data,including liver dose-volume histograms before treatment.Results Patients were followed up for 1-79 months (median 19 months) after completion of the irradiation.RILD was diagnosed in 9 (7.9%) patients.Univariate analyses revealed that GTV and the percent of normal liver volume exceeding 5-40 Gy (V5-40) were related to the risk of developing RILD.Multivariate analyses showed that only GTV (P=0.015) and V20 (P=0.003) were independent predictors.When using V20= 48.5% as tolerated dose to predict RILD,the accuracy,sensitivity and specificity was 76.3%,88.9% and 75.2%,respectively.Conclusions V20 is a unique significant dosimetric predictor to evaluate RILD risks for PLC patients with Child-Pugh grade A cirrhosis after hypofractionated CRT.

Key words: Liver neoplasm, Radiotherapy, Conformal radiotherapy, Hypofractionation, Radiation-induced liver disease