Non-classic radiation-induced liver disease after intensity-modulated radiotherapy for Child–Pugh grade B patients with locally advanced hepatocellular carcinoma

Abstract

Abstract Background The incidence of classic radiation-induced liver disease (cRILD) has been significantly reduced. However, non-classic radiation-induced liver disease (ncRILD) remains a major concern following radiotherapy in patients with hepatocellular carcinoma (HCC). This study evaluated the incidence of ncRILD following intensity-modulated radiotherapy (IMRT) for Child–Pugh grade B (CP-B) patients with locally advanced HCC and established a nomogram for predicting ncRILD probability. Methods Seventy-five CP-B patients with locally advanced HCC treated with IMRT between September 2014 and July 2021 were included. The max tumor size was 8.39 cm ± 5.06, and the median prescribed dose was 53.24 Gy ± 7.26. Treatment-related hepatotoxicity was evaluated within three months of completing IMRT. A nomogram model was formulated to predict the probability of ncRILD, using univariate and multivariate analysis. Results Among CP-B patients with locally advanced HCC, ncRILD occurred in 17 (22.7%) patients. Two patients (2.7%) exhibited a transaminase elevation of ≥ G3, fourteen (18.7%) exhibited a Child–Pugh score increase of ≥ 2, and one (1.3%) demonstrated both a transaminase elevation of ≥ G3 and a Child–Pugh score increase of ≥ 2. No cRILD cases were observed. A mean dose to the normal liver of ≥ 15.1 Gy was used as the cutoff for ncRILD. Multivariate analysis revealed that the prothrombin time before IMRT, tumour number, and mean dose to the normal liver were independent risk factors for ncRILD. The nomogram established on the basis of these risk factors displayed exceptional predictive performance (AUC = 0.800, 95% CI 0.674–0.926). Conclusions The incidence of ncRILD following IMRT for CP-B patients with locally advanced HCC was acceptable. A nomogram based on prothrombin time before IMRT, tumour number, and mean dose to the normal liver accurately predicted the probability of ncRILD in these patients.