Prognostic Significance and Immune Landscape of Migrasome-Related Genes in Pancreatic Cancer.

Pancreatic cancer is characterized by early metastasis and high aggressiveness, highlighting the need for novel molecular biomarkers to improve prognostic assessment and therapeutic strategies. In this study, we constructed a migrasome-related prognostic model through multiple cohorts using 101 machine learning algorithms, and applied Shapley Additive Explanations (SHAP) to interpret the model's feature contributions. Six key genes - tetraspanin 5 (TSPAN5), bone morphogenetic protein 1 (BMP1), integrin subunit alpha 3 (ITGA3), integrin subunit alpha 5 (ITGA5), Wnt family member 11 (WNT11), and tetraspanin 2 (TSPAN2) -were identified, mainly enriched in extracellular matrix (ECM)-receptor interaction and cell adhesion pathways. Further analyses showed that these genes were preferentially expressed in tumor cells and fibroblasts, and were associated with an altered immune microenvironment. Drug sensitivity prediction and molecular docking indicated that the monocarboxylate transporter 1 (MCT1) inhibitor AZD-3965 may have therapeutic potential in this context. In general, our findings suggest that migrasome-related genes may contribute to prognostic stratification of pancreatic cancer and point to mechanisms of stroma-immune crosstalk, thereby offering exploratory avenues for personalized treatment.