Staging laparoscopy in locally advanced gastric adenocarcinoma: diagnostic yield, predictors of occult metastatic disease, and development of a clinical prediction score

Background: Computed tomography (CT) fails to detect occult peritoneal and distant metastases in a significant minority of patients with locally advanced gastric cancer (LAGC). Staging laparoscopy (SL) improves staging accuracy and redirects patients with occult M1 disease away from futile curative surgery. We determined the diagnostic yield of SL, identified independent predictors of positive findings, and developed a clinical prediction score. Methods: This was a single-center retrospective observational cohort study of 113 consecutive LAGC patients who underwent SL between January 2023 and December 2025 after non-metastatic staging on CT. Multivariable logistic regression identified independent predictors; an integer prediction score was derived and evaluated by receiver operating characteristic analysis in the same derivation cohort. Results: Positive SL findings were identified in 27 patients [23.9%; 95% confidence interval (CI): 16.5–32.8%], redirecting all 27 from curative to non-curative management. Peritoneal metastases were the most common finding (15.9%). Independent predictors on multivariable analysis were CT-detected ascites [odds ratio (OR) 6.84; 95% CI: 1.82–25.63), cT4 disease (OR 4.12; 95% CI: 1.65–10.31), elevated serum carbohydrate antigen 19-9 (OR 3.21; 95% CI: 1.24–8.29), and diffuse/signet-ring histology (OR 2.78; 95% CI: 1.08–7.17). The prediction score stratified patients into low-risk (8.2%; 4/49), intermediate-risk (22.2%; 10/45), and high-risk (68.4%; 13/19) groups ( P < 0.001; area under the curve (AUC) = 0.81; 95% CI: 0.73–0.89). This model was developed and evaluated in the same cohort without internal validation; performance estimates should be interpreted with caution, and external validation is required. Conclusion: SL identified radiologically occult M1 disease in nearly one-quarter of LAGC patients. The proposed four-variable prediction score provides a practical risk-stratification framework to inform selective SL utilization, with particular relevance to resource-constrained settings. External validation is required before clinical adoption.