Clinical decision tree for optimising endoscopic assessment of signet ring cell carcinoma in hereditary diffuse gastric cancer surveillance

Background and study aims Prophylactic total gastrectomy (PTG) is the definitive treatment for hereditary diffuse gastric cancer syndrome (HDGC). Endoscopic surveillance informs the requirement and optimal timing to surgery. However, endoscopic recognition of early signet ring cell carcinoma (SRCC) remains challenging. We aim to develop an endoscopic framework to optimise SRCC assessment during HDGC surveillance. Patient and methods We retrospectively analysed data from 147 HDGC individuals undergoing endoscopic surveillance to evaluate the diagnostic accuracy of endoscopic Cambridge criteria,We used machine learning to develop a clinical decision tree (cDT) to guide the application of Cambridge criteria. We then prospectively validated cDT in 66 CDH1 pathogenic-variant carriers. The inter-observer agreement and diagnostic accuracy of Cambridge criteria and cDT were assessed through multi-reader multi-case study. Results Retrospective analysis of 537 endoscopies showed Cambridge criteria achieved 82.8% (48/58) sensitivity and 78.2% (140/179) specificity for SRCC diagnosis. The presence and number of neoplastic pale areas are independent predictors of higher cancer burden in HDGC individuals. In prospective study, cDT had 77.8% (21/27) sensitivity and 90.7% (49/54) specificity and improved performance of both experts and non-experts. Conclusion We developed and validated a practical endoscopic framework to enhance SRCC assessment during HDGC endoscopic surveillance.