OBJECTIVE: To establish a clinical prediction model of the mid-term fatality risk after radical resection in patients with primary hepatocellular carcinoma (HCC) based on the albumin-bilirubin (ALBI) grade and to assess its prediction value. METHODS: Clinical data of 533 patients who received HCC radical resection in Jinhua Hospital of Zhejiang University, Jinhua People's Hospital, Jinhua Hospital of Traditional Chinese Medicine and Jinhua Guangfu Hospital from January 2010 to August 2016 were retrospectively reviewed. In the training group ( =407), Cox model was used to screen the clinical risk factors of postoperative death, and a predictive model based on ALBI grade was established and then examined in the validation group ( =126). The value of the prediction model was assessed by ROC curve and calibration curve; the prediction results of the model were visualized by the nomogram for the convenience of clinical use. RESULTS: Cox model showed that ALT ≥ 80 U/L, tumor maximum diameter ≥ 5 cm, portal vein tumor thrombus and ALBI grade 2 were independent risk factors for the prognosis of patients with HCC radical resection. The prognosis index (PI) was 0.550×ALT+0.512×ALBI grade+0.872×maximum tumor diameter+1.377×portal vein tumor thrombus. The AUCs for predicting the risk of death in 12, 36 and 60 months were 0.872, 0.814 and 0.810, respectively (all < 0.01), and the goodness of fit ( ) of the established model were 0.953, 0.976 and 0.994. AUC of the established model for predicting risk of death in 36 months after resection was 0.814, which was higher than those of ALBI (AUC=0.683), BCLC (AUC=0.713), CLIP (AUC=0.689), Child-Pugh (AUC=0.645), TNM (AUC=0.612) ( < 0.05 or < 0.01). CONCLUSIONS ALT ≥ 80 U/L, maximum tumor diameter ≥ 5 cm, portal vein tumor thrombus and ALBI grade 2 are independent risk factors of patients after HCC resection, and ALBI grade-based prediction model is satisfactory in prediction of mid-term death risk of the patients.
Albumins ; Bilirubin ; Carcinoma, Hepatocellular ; Humans ; Liver Neoplasms ; Prognosis ; Retrospective Studies