Artificial Neural Network Model to Predict Post-Hepatectomy Early Recurrence of Hepatocellular Carcinoma without Macroscopic Vascular Invasion
Abstract Background: The accurate prediction of post-hepatectomy early recurrence (PHER) for hepatocellular carcinoma (HCC) is of great significance in determining postoperative adjuvant treatment and monitoring. This research aimed to develop and validate an artificial neural network (ANN) model to predict PHER in HCC patients without macroscopic vascular invasion.Methods: 903 patients who underwent curative liver resection for HCC were collected. They were randomly divided into a derivation cohort (n = 679) and a validation cohort (n = 224). The ANN model was then developed in the derivation cohort and verified in the validation cohort.Results: The morbidity of PHER in the derivation and validation cohorts was 34.8% and 39.2%, respectively. Multivariate analysis revealed that hepatitis B virus DNA load, γ-glutamyl transpeptadase, α-fetoprotein, tumor diameter, tumor differentiation, microvascular invasion, satellite nodules and blood loss were significantly associated with PHER. Incorporating these factors, the ANN model had greater discriminatory abilities than conventional Cox model, existing recurrence models and commonly used staging systems for predicting PHER. Stratification into two risk groups indicated a statistically significant discrepancy in recurrence-free survival curves. Conclusion: The ANN model has a significant advantage in predicting PHER for HCC patients without macroscopic vascular invasion when compared to other models and staging systems.