Nomogram predicting survival of hepatocellular carcinoma with portal vein tumour thrombus after curative resection

Background The aim of this study was to combine clinicopathologic variables associated with overall survival and disease-free survival after curative resection for hepatocellular carcinoma (HCC) with portal vein tumour thrombus (PVTT) into a prediction nomogram. Methods We retrospectively analysed 358 patients who underwent curative resection for HCC with PVTT at the Eastern Hepatobiliary Surgery Hospital, the Second Military Medical University, Shanghai, China. Two-thirds of the patients were randomly assigned to the training set (n = 252) and one-third were assigned to the validation set (n = 106). Multivariate analysis by Cox proportional hazards regression was performed using the training set, and the nomogram was constructed. Discrimination and calibration were performed using the validation set. Results The multivariate Cox model identified alpha fetoprotein, hepatitis B surface antigen (HBsAg), tumour diameter, tumour capsule, PVTT type and TNM stage as covariates associated with 1-year survival, and alpha fetoprotein, HBsAg, tumour diameter, tumour capsule and PVTT type with half-year disease-free survival. In the validation set, the nomogram exhibited superior discrimination power (Harrell's C-index 0.78) compared with the American Joint Committe on Cancer TNM classification, the Cancer of the Liver Italian Program grade and the Japan Integrated Staging grade. Calibration of the nomogram-predicted survival corresponding closely with the actual survival, the predicted survival was within a 10% margin of ideal nomogram. Conclusion We developed a nomogram predicting 1-year overall survival and half-year disease-free survival after curative resection for HCC with PVTT. Validation data sets revealed good discrimination and calibration, suggesting good clinical utility. The nomogram improved individualized predictions of survival.