Abstract
Tumor markers are commonly used in gastric cancer patients, but their effectiveness in monitoring recurrences is not optimal. This study aims to evaluate the recurrence predictive performance of carcinoembryonic antigen (CEA) across different baseline groups and establish a new surveillance protocol. We analyzed data from 1708 patients at Sun Yat-Sen University Cancer Center with stage I-III gastric adenocarcinoma. The research findings were subjected to longitudinal validation (expanded dataset comprising 6422 follow-up records of patients) and external validation (109 patients from the Sixth Affiliated Hospital, Sun Yat-Sen University). The 5-year disease-free survival (DFS) rate was 61.1% (95% CI: 57.4%-65.0%) for the normal baseline group and 42.1% (95% CI: 36.4%-48.6%) for the elevated baseline group. The normalization group had DFS similar to the persistently normal group (p = 0.117). For recurrence prediction efficacy, the elevated baseline group showed significantly higher sensitivity than the normal baseline group (0.73 vs. 0.32, p = 0.001), while the normal baseline group had superior specificity (0.87 vs. 0.59, p = 0.031). These results were confirmed in longitudinal validation and external validation cohorts. Different monitoring strategy should be used for different CEA baseline patients. Incorporating CEA monitoring into gastric cancer surveillance protocols with the above method may be included in the surveillance protocols of NCCN gastric cancer guideline.