A novel model in predicting very early recurrence after surgical resection of hepatocellular carcinoma / 中华肝胆外科杂志

ABSTRACT

Objective:To study the risk factors of very early recurrence (VER, within 3 months) after R 0 resection of hepatocellular carcinoma (HCC), and to establish a predictive model. Methods:Of 427 HCC patients [with 368 males, 59 females, aged (52.7±12.1) years] who developed early recurrence (within 2 years) after R 0 resection from January to December 2008 at Zhongshan Hospital, Fudan University were enrolled in the test cohort. Another 590 patients [with 525 males, 65 females, aged (54.7±11.0) years] who underwent R 0 resection from January to June 2009 were enrolled in the validation cohort. Risk factors were investigated and a predictive model was established. Results:In the test cohort, 126 patients (29.5%) developed VER and their survival outcomes were extremely poor. Serum α-fetoprotein (AFP) level >827 μg/L, multiple tumors, microvascular invasion (MVI) and tumor number were independent risk factors for VER. A new predictive model (0.809·AFP+ 1.262·tumor number+ 0.983·MVI) was established by logistic regression in predicting VER after surgery. The receiver operating characteristic curve showed that the area under the curve (AUC) in predicting VER was 0.722 (95% CI 0.669-0.774, P<0.001). In the validation cohort, the AUC of this model was 0.785 (95% CI 0.715-0.855, P<0.001). Conclusions:A high AFP level, multiple tumors, and MVI were independent risk factors for VER of HCC after R 0 resection. The prediction model consisting of these three factors demonstrated robustness and it has the potential in clinical application.