Abstract
BACKGROUND: Alpha-fetoprotein (AFP) is an established biomarker for liver cancer, but its role in gastric cancer (GC) remains unclear. This study evaluated AFP's prognostic value in GC and developed a survival prediction model incorporating AFP and other clinical factors. METHODS: We analyzed 766 GC patients from Changzhou Traditional Chinese Medicine Hospital, categorizing them as AFP-positive (>20 ng/mL) or AFP-negative (≤20 ng/mL). Kaplan-Meier and Cox regression analyses assessed the association between AFP levels and overall survival (OS). A nomogram based on identified prognostic factors was created and evaluated using ROC curves, calibration curves, and decision curve analysis (DCA). RESULTS: Among 766 gastric cancer (GC) patients, 3.3% (n=25) exhibited elevated AFP levels (>20 ng/mL). The AFP-positive group demonstrated significantly more aggressive clinicopathological features, including larger tumor size (p < 0.05), deeper invasion (higher T-stage), increased lymph node metastasis (higher N-stage), and higher rates of distant metastasis (p = 0.035). Survival analysis revealed markedly worse outcomes for AFP-positive patients (Log-rank P < 0.001), with a 68% higher mortality risk (unadjusted HR =1.68, 95% CI: 1.27-2.23). Multivariate Cox regression confirmed AFP positivity as an independent prognostic factor (adjusted HR = 1.8, 95% CI: 1.03-3.14, p =0.04), alongside T4-stage, N3-stage, and distant metastasis. A prognostic nomogram integrating AFP levels and TNM staging achieved superior predictive accuracy (AUCs: 0.80-0.84) compared to TNM staging alone (AUCs: 0.70-0.74) across 1-, 3-, and 5-year survival. Calibration and decision curve analyses further validated the model's clinical utility, supporting its role in risk stratification and treatment planning. CONCLUSIONS: AFP is a significant independent prognostic factor in gastric cancer, and its inclusion in a multivariate model enhances survival prediction. The prognostic nomogram developed in this study offers clinicians a valuable tool for predicting patient outcomes and guiding treatment decisions. Further validation and prospective studies are necessary to confirm the model's clinical applicability.