Long-read sequencing of pancreatic adenocarcinoma transcriptome uncovered aberrant isoforms and tumor progression

Pancreatic cancer remains one of the most lethal malignancies, with a dismal 5-year survival rate of only 13%. Despite multi-omics studies uncovering critical variants and regulatory mechanisms in driver genes and oncogenes, effective therapeutic options remain limited. To address the urgent unmet clinical need for a deeper understanding of the disease, we employed long-read sequencing (LR-seq) to identify complex structural variants and alternative splicing events that are undetectable using short-read sequencing methods.In total, we identified 150,904 isoforms, including 62,111 novel variants. Among these novels, 51.5% showed tumor-specific expression. Novel isoforms were frequently detected in oncogenes such as CD74 , B2M , and DAXX , suggesting distinct driver events in pancreatic cancer. Notably, novel alternative transcription start sites were enriched in chromatin-accessible regions marked by H3K4me3 and H3K27ac in tumor cells, and these isoforms appeared to disrupt MHC-associated functions involving CALR , PTK6 , and TAPBP .Expression profiling of the novel isoforms clearly distinguished classical and quasi-mesenchymal (QM) subtypes ( P = 0.001). Furthermore, global analysis of alternative splicing and switch-like isoforms across molecular subtypes revealed that splicing events modulated metastatic characteristics between the classical and QM subtypes through Rho GTPase signaling, GPCR signaling, and extracellular matrix organization.In conclusion, our study underscores the critical role of long-read sequencing in uncovering novel isoforms and alternative splicing events that define the molecular heterogeneity of pancreatic cancer. These findings provide new insights into the regulation of key oncogenes and reveal potential therapeutic targets with implications for improving the diagnosis and treatment of this devastating malignancy.