Delineation of molecular subtypes based on pathway activation signatures reveals intrinsic tumor heterogeneity and pan-cancer shared features

Abstract

Purposes Individual molecular features are insufficient to elucidate the complex biological properties of pan-cancer. Accordingly, this study performed pan-cancer molecular subtyping based on pathway activation signatures to dissect the heterogeneity and commonality across malignancies. Methods Transcriptomic data for 7,138 pan-cancer samples were retrieved from TCGA database for molecular subtyping. In parallel, transcriptomic data for 597 cell lines and drug sensitivity data for 269 lines were obtained from CCLE database for drug sensitivity analysis, and transcriptomic profiles of 16 in-house cell lines were incorporated for combined analysis. Results Using unsupervised clustering, we classified pan-cancer into four distinct molecular subtypes based on the activity of 43 key cancer-related pathways. C1 was enriched for upregulated pathways involved in oncogenic signaling, cellular processes-related and immune pathways. C2 was characterized by upregulated activation of metabolic and DNA damage response (DDR) pathways. C3 displayed upregulated activity of metabolic, DDR and immune pathways. C4 showed elevated activity of oncogenic signaling, cellular processes-related, and DDR pathways. Notably, compared with C2 and C4, C1 and C3 displayed enhanced immune score, increased CD8+T cell infiltration and higher expression of genes involving in antigen processing and presentation, immunogenic cell death and immune checkpoint. C3 demonstrated the most favorable overall survival and the lowest proportion of progressive/stable disease patients, followed by C1, whereas C2 and C4 were associated with inferior clinical outcomes. C3 was susceptible to multiple agents, while C1 showed multidrug resistance. Moreover, in vitro cellular experiments demonstrated that combined treatment with inhibitors targeting activated pathways yielded superior inhibitory effects relative to monotherapy for C1, C2 and C4. Conclusions Overall, our study defined four molecular subtypes based on the pathway activation profiles, with each subtype displaying unique features in pathway activation, clinical manifestations, treatment response, survival prognosis, immune microenvironment, and drug sensitivity. These findings may provide potential valuable insights for the precise diagnosis and treatment of cancer.