Background: Long RNAs have been recently identified in human blood exosomes, posing clinical implications. Whether exosomal long RNAs (exoLRs) could constitute key future biomarkers for noninvasive diagnosis, therapeutic evaluation, and prognosis in cancer remains unknown. The study aimed to explore the exoLR landscape of human blood exosomes and evaluate the feasibility of developing a diagnostic or prognostic signature for early detection and prognostic prediction of pancreatic ductal adenocarcinoma (PDAC) based on exoLR profiling. Methods: A case-control study of 267 cases including 137 patients with PDAC and 39 with chronic pancreatitis (CP) plus 91 blood donors as healthy participants was conducted. The exoLR profile of pretreated blood samples was analyzed by exoLR-sequencing (exoLR-seq). Results: An average of 15,000 exoLRs were reliably detected for each sample through exoLR-seq, and 1,053 exoLRs were differentially expressed in PDAC. Based on these data, we constructed a diagnostic signature (d-signature) that showed high accuracy with an area under the curve (AUC) of 0.977 (95% CI: 0.958–0.996), a sensitivity of 92.42% (95% CI: 83.2%–97.49%), and a specificity of 95% (95% CI: 87.69%–98.62%) in a training cohort (n=146), which was further confirmed in a validation cohort (n=93). Notably, the combination of d-signature and CA19-9 yielded an AUC of 0.963 (95% CI: 0.909–1.017), with a sensitivity of 98.13% (95% CI: 93.41%–99.77%) and specificity of 94.59% (95% CI: 81.81%–99.34%). Additionally, we constructed a prognostic prediction model (exoLR p-signature) that effectively predicted prognosis and survival in patients with PDAC (P=9.838e-08). Conclusions: This study clearly demonstrated the value of exoLR profiling in cancer marker discovery and the feasibility of developing a diagnostic or prognostic signature for early detection and prognostic prediction of PDAC.
Early Diagnosis of Pancreatic Ductal Adenocarcinoma by Combining Relative Expression Orderings With Machine-Learning Method
