S3183 Targeted vs Random Biopsy in Individuals Undergoing Surveillance for Hereditary Diffuse Gastric Cancer: A Systematic Review and Meta-Analysis of Endoscopic Detection Strategies

Introduction: Surveillance endoscopy is essential in managing individuals at increased risk for hereditary diffuse gastric cancer (HDGC). However, it remains unclear which biopsy technique—random or targeted—offers greater diagnostic yield for detecting signet ring cell carcinoma (SRCC). This meta-analysis aimed to compare the diagnostic performance of both approaches. Methods: PubMed, Embase, Scopus, and Web of Science were searched for studies that reported both random and targeted biopsies performed during the same surveillance endoscopy for HDGC. Odds ratio (OR) and 95% confidence interval (CI) were used to assess yield outcomes. Case reports, cross-sectional studies, and publications with inadequate methodology were excluded. Results: Nine cohort studies were included, comprising 736 patients undergoing surveillance, during which a total of 1,717 upper gastrointestinal endoscopies were performed. The pooled analysis showed no significant difference in the diagnostic yield of patients with SRCC between random and targeted biopsies (OR 0.45; 95% CI 0.10-2.01; P = 0.296; I2 = 68.8%). However, a sensitivity analysis using the leave-one-out method—excluding a study with very low adherence to systematic protocols—revealed a significant advantage favoring random biopsies (OR 0.29; 95% CI 0.15-0.55; P = 0.0002; I2 = 0.3%). In the per-sample analysis, targeted biopsies showed superior effectiveness in detecting SRCC foci (OR 6.03; 95% CI 1.81-20.17; P = 0.004; I2 = 93.7%), especially in pale areas (OR 18.07; 95% CI 13.50-24.18; P < 0.001; I2 = 7.7%). Conclusion: A greater performance of random biopsies in the sensitivity analysis highlights their clinical importance in identifying patients with HDGC. The apparent equivalence in the primary outcome may reflect poor adherence to systematic protocols rather than true parity—a hypothesis supported by the exclusion of the study with limited random sampling. However, expanding this approach requires careful evaluation of clinical, technical, and environmental impact, which remain underexplored in HDGC surveillance. Targeted biopsies were more efficient per sample, underscoring the value of meticulous inspection and adjunctive tools like narrow band imaging to visualize subtle lesions. These findings support a complementary role for both strategies and reveal gaps in current protocol standardization. This manuscript was prepared with the assistance of generative artificial intelligence.