Figure 9 Consensus clustering based on prognosis-related alternative splicing (AS) events and immune landscape analysis. (A, B) Consensus clustering matrix and silhouette plot identifying 3 clusters. (C) Kaplan-Meier survival plot showing significant differences among the 3 clusters (log-rank t test, P<0.05). (D) Classification of pancreatic ductal carcinoma samples based on 5 immune subtypes in 3 clusters: wound healing, IFN-γ dominant, inflammatory, lymphocyte depleted, and TGF-β dominant. (E, F) Comparison of immune and stromal scores of the 3 clusters with analysis of variance. (G) Heat map of clinicopathological characteristics and gene phenotypes in the 3 clusters. Driver gene mutations, including KRAS, CDKN2A, and TP53, showed distinct distribution in the 3 clusters with chi-squared test.