There is currently no established model for predicting the recurrence of hepatocellular carcinoma (HCC) in patients with negative alpha‐fetoprotein (AFP) after curative resection. Therefore, the objective of this study was to establish a nomogram to identify the risk of recurrence in AFP‐negative (<or = 20 ng/mL) patients with HCC.

A retrospective study was conducted to establish the recurrence‐free survival (RFS) nomogram in a training cohort of 326 AFP‐negative HCC patients. The results were validated on a well‐matched validation cohort in the literature.

Macrovascular tumour invasion (P = 0.018, HR = 1.642), macronodular cirrhosis (P < 0.001, HR = 2.128), tumor size (P = 0.004, HR = 1.691), and γ‐glutamyl transferase (P = 0.039, HR = 1.496) were found to be independent risk factors for RFS in the training cohort, and all these factors were included in the nomogram. The C‐index for RFS in the nomogram was 0.661, which was higher than that of the BCLC system (0.551), the CLIP score (0.537), and the prediction model of Ju (0.618). The high consistency between the nomogram prediction and actual observation was further demonstrated by the calibration curve. In the subsequent study, the better net benefit and higher threshold probability of the nomogram were determined by decision curve analysis, and these advantages were confirmed in the validation cohort.

The present RFS nomogram for AFP‐negative HCC patients after curative resection provides an accurate and reliable prognostic model to facilitate recurrence surveillance. Once AFP‐negative patients are predicted to have a high recurrence score, additional high‐end imaging examinations, such as MRI or CT exams, should be considered, and the interval time of regular folow‐up should be reduced.