Background: Gallbladder cancer (GBC) is a highly lethal disease lacking reliable prognostic biomarkers. This study aimed to evaluate the clinical utility of artificial intelligence (AI)-powered spatial analysis of tumor-infiltrating lymphocytes (TILs) and the tumor microenvironment (TME) in resected GBC. Methods: This multicenter retrospective study analyzed a total of 225 patients with R0-resected GBC and an external validation cohort of 41 patients with GBC. Hematoxylin and eosin-stained slides were analyzed using Lunit SCOPE IO, an AI-powered whole-slide image analyzer, to map TME features. TME-related risk factors – low TIL density, low tertiary lymphoid structure count, and high fibroblast density – were identified, and their associations with overall survival (OS) and disease-free survival (DFS) were evaluated. Results: Compared with the immune-desert phenotype, both immune-inflamed and immune-excluded phenotypes were associated with longer OS ( P = 0.013). As the number of TME-related risk factors increased, OS and DFS progressively worsened. Using the group with three risk factors as the reference, adjusted hazard ratios (HRs) for OS were 0.40 (95% CI, 0.19–0.85; P = 0.017) for two risk factors, 0.34 (95% CI, 0.16–0.74; P = 0.007) for one risk factor and 0.20 (95% CI, 0.06–0.67; P = 0.009) for no risk factors. Similarly, in the analysis of DFS, with three risk factors as the reference group, the adjusted HRs were 0.37 (95% CI, 0.18–0.74; P = 0.005) for two risk factors, 0.30 (95% CI, 0.15–0.62; P = 0.001) for one risk factor, and 0.13 (95% CI, 0.04–0.41; P < 0.001) for no risk factors. External validation confirmed consistent prognostic patterns. Conclusion: We demonstrated that AI can be used to analyze TME components to predict the prognosis of resected GBC, and this tool may serve as a useful prognostic biomarker for resected GBC. Graphical abstract : http://links.lww.com/JS9/H298.
Choi et al. (Tue,) studied this question.