Circulating EVs long RNA-based subtyping and deconvolution enable prediction of immunogenic signatures and clinical outcome for PDAC

Abstract

Identification of clinically applicable molecular subtypes of pancreatic ductal adenocarcinoma (PDAC) is crucial to improving patient outcomes. However, the traditional tissue-dependent transcriptional subtyping strategies are invasive and not amenable to routine clinical evaluation. In this study, we developed a circulating extracellular vesicle (cEV) long RNA (exLR)-based PDAC subtyping method and provided exLR-derived signatures for predicting immunogenic features and clinical outcomes in PDAC. We enrolled 426 individuals, among which 227 PDACs served as an internal cohort, 118 PDACs from two other medical centers served as an independent validation cohort, and 81 healthy individuals served as the control. ExLR sequencing was performed on all plasma samples. We found that PDAC could be categorized into three subtypes based on plasma exLR profiles. Each subpopulation showed its own molecular features and was associated with patient clinical prognosis. The immunocyte-derived cEV fractions were altered among PDAC subtypes and interconnected with tumor-infiltrating lymphocytes in cancerous tissue. Additionally, we found a significant concordance of immunoregulators between tissue and blood EVs, and we harvested potential PDAC therapeutic targets. Most importantly, we constructed a nine exLR-derived, tissue-applicable signature for prognostic assessment of PDAC. The circulating exLR-based features may offer an attractive platform for personalized treatment and predicting patient outcomes in multiple types of cancer.