Dynamic changes in gut microbiota and discovery of prognostic biomarkers in locally advanced pancreatic cancer during chemoradiotherapy.

Pancreatic cancer has an exceptionally poor prognosis, with the majority of cases diagnosed at an advanced stage. Concurrent chemoradiotherapy (CCRT) remains the standard of care for locally advanced pancreatic cancer (LAPC); however, its therapeutic efficacy is limited. Emerging evidence suggests that the gut microbiota is an important modulator of cancer progression and treatment response. Nevertheless, the dynamic changes in the gut microbial ecosystem in LAPC patients undergoing CCRT remain poorly understood. This study aimed to characterize longitudinal alterations in the gut microbiota during CCRT, identify candidate microbiome-based prognostic markers, and explore their potential associations with host responses. This longitudinal study included 16 patients with LAPC. Fecal and peripheral blood samples were collected at three predefined time points: before CCRT initiation, during CCRT, and after CCRT completion. Gut microbiota composition and community structure were analyzed using 16 S rRNA sequencing targeting the V3–V4 region. Bioinformatic analyses were performed to assess taxonomic distribution, alpha and beta diversity, and microbial co-occurrence patterns. The prognostic relevance of microbial features was further evaluated using machine learning models integrating clinical parameters to predict overall survival. CCRT was associated with dynamic changes in specific microbial taxa across multiple taxonomic levels, with marked inter-individual heterogeneity in microbiota responses. In addition, microbial co-occurrence network complexity was reduced during treatment. Certain microbial taxa during CCRT showed associations with tumor-related serum biomarkers. Using integrated machine learning models, we identified candidate microbiota-based prognostic markers. In particular, Bifidobacterium and Bacteroides were associated with survival prediction, achieving areas under the receiver operating characteristic curve (AUC) of 0.833 and 0.722, respectively. CCRT is associated with longitudinal alterations in the gut microbiome of patients with LAPC, involving both compositional and structural changes. These findings suggest that microbiome dynamics may have potential value as exploratory prognostic indicators. However, given the limited sample size and observational design, further validation in larger cohorts and mechanistic studies are required before clinical application.