Log-transformed α-Fetoprotein and Des-γ-carboxy Prothrombin Predict Early Recurrence After Hepatectomy for Hepatocellular Carcinoma.

BACKGROUND/AIM: We evaluated the impact of log-transformed serum α-fetoprotein (AFP) and des-γ-carboxy prothrombin (DCP) levels on early hepatocellular carcinoma (HCC) recurrence after hepatectomy and developed a nomogram for risk stratification.

PATIENTS AND

METHODS

We examined 993 patients who underwent initial curative hepatectomy. A nomogram was constructed to predict early recurrence within 2 years using a Cox proportional hazards model (five-factor log-marker model) based on five preoperatively assessable oncological factors: log AFP, log DCP, multiple tumors, tumor size >3 cm, and macroscopic vascular invasion (MVI). Its time-dependent area under the curve (AUC) at 2 years was compared with a five-factor raw-marker model (replacing log-transformed AFP/DCP with their raw values) and a three-factor model (multiple tumors, tumor size, MVI). Performance was evaluated by time-dependent AUC, bootstrap calibration (1,000 resamples), and observed outcomes across tertiles (T1-T3) of total nomogram points.

RESULTS

The five-factor log-marker model (AUC 0.69) outperformed the five-factor raw-marker model (0.67, p=0.013) and the three-factor model (0.65, p<0.001). The nomogram showed good calibration (slope=0.98). Two-year recurrence rates by tertile were 25% (T1), 38% (T2), and 58% (T3); 2-year extrahepatic recurrence rates were 1.5%, 4.2%, and 11%, respectively. The trend across tertiles was significant (p<0.001).

CONCLUSION

The nomogram developed using preoperative oncological factors enabled precise prediction of early recurrence after hepatectomy for HCC by incorporating log-transformed AFP and DCP. Our nomogram may be useful in determining treatment strategies for patients with HCC.