RESUMEN
BACKGROUND: Changes in the expression of genes related to glycosyltransferases may lead to alterations in N-glycan structure abundance, potentially acting as markers for diagnosis and prognosis in biliary tract cancer (BTC). METHODS: This study was divided into cross-sectional and longitudinal approaches. The cross-sectional study included 316 BTC and 301 non-BTC. Propensity score matching was applied to adjust for sex and age differences between BTC and non-BTC. Univariate and multivariate logistic regression identified independent risk factors for BTC and constructed the BTC-G model. The ROC curve was used to validate the diagnostic performance of BTC-G. Longitudinal follow-up studies included postoperative (N = 50) and immunotherapy (N = 43) follow-up cohorts. Cox regression analysis identified N-glycan structures impacting BTC prognosis postoperative and immunotherapy, with further confirmation through Kaplan-Meier curves. RESULTS: Univariate and multivariate analyses identified Peak3 (OR: 0.790, 95% CI: 0.658-0.949), Peak9 (OR: 1.646, 95% CI: 1.409-1.922), and Peak9p (OR: 2.467, 95% CI: 1.267-4.804) as independent BTC risk factors, leading to the creation of the BTC-G. The ROC curve confirmed that BTC-G performed well in training (AUC: 0.753, 95% CI: 0.703-0.799), validation (AUC: 0.811, 95% CI: 0.740-0.870), and CA19-9 negative cohorts (AUC: 0.717, 95% CI: 0.664-0.767). Cox regression analysis and Kaplan-Meier curves established that Peak12 (HR: 5.578, 95% CI: 1.145-27.170) and Peak11 (HR: 1.104, 95% CI: 0.611-1.994) are independent risk factors for BTC prognosis following surgery and immunotherapy, respectively. CONCLUSIONS: Our NGFP technology supplements BTC diagnostics, distinguishing survival and recurrence subtypes for postoperative and immunotherapy, thereby supporting the development of treatment strategies.