Abstract
INTRODUCTION: This study aimed to validate the prognostic value of alpha- fetoprotein (AFP) in AFP-negative hepatocellular carcinoma (HCC) and develop an AFP-integrated nomogram for post-resection recurrence-free survival (RFS) and overall survival (OS). MATERIAL AND METHODS: This retrospective study analyzed 453 HCC patients with preoperative AFP ≤ 20 ng/ml who underwent curative resection, divided into training (n = 317) and validation (n = 136) cohorts. The optimal AFP cutoff was determined using maximized χ(2) values. Nomograms were developed with repeated least absolute shrinkage and selection operator variable selection and stepwise Cox regression. Model performance was assessed using concordance (C-) indices, time-dependent area under the receiver operating characteristic curves (AUCs), calibration plots, and Kaplan-Meier (KM) curves. RESULTS: An alpha-fetoprotein cutoff of 7 ng/ml stratified patients for both RFS and OS (p < 0.001). The recurrence-free survival nomogram included AFP, age, sex, multiple tumors, and cirrhosis, while the OS nomogram incorporated AFP, albumin-bilirubin score, the up-to-7 criterion, des-γ-carboxy prothrombin, vascular invasion, and histological grade. In the training cohort, the nomograms demonstrated C-indices of 0.64 (95% CI: 0.60-0.68) for RFS and 0.72 (0.67-0.76) for OS. Validation cohort C-indices were 0.64 (0.62-0.65) for RFS and 0.67 (0.65-0.68) for OS. Time-dependent AUCs and calibration plots confirmed the predictive accuracy of the nomograms, and KM curves showed clear separation between high- and low-risk groups, further highlighting their clinical utility. CONCLUSIONS: Alpha-fetoprotein retains prognostic value even within the clinically normal range for HCC. The AFP-integrated post-resection nomograms demonstrated acceptable predictive performance for AFP-negative patients, potentially enhancing personalized management strategies.