Abstract
The accurate classification of pancreatic cystic lesions remains clinically challenging due to overlapping imaging features and variable malignant potential. Mucinous cystic neoplasms, in particular, require early identification given their premalignant nature. RNA profiling presents a promising alternative to current diagnostic limitations-a molecular lens sharpened by AI-driven pattern recognition. This study aimed to evaluate the diagnostic potential of RNA signatures for differentiating pancreatic cyst subtypes and to clarify their roles in their pathophysiology. The study included 31 patients with pancreatic lesions who underwent endoscopic ultrasound-guided fine-needle aspiration. RNA was extracted from cyst fluid, tissue, and peripheral blood. Expression of 17 target genes was analyzed using qPCR. Gene expression patterns were compared across mucinous cystic neoplasms, serous cystic neoplasms, pseudocysts, adenocarcinoma, and chronic pancreatitis cohorts. Diagnostic accuracy was evaluated via ROC analysis. Mucinous cysts exhibited significant overexpression of MUC1, ITGA2, ELOVL6, and MUC5AC genes compared to serous cysts and pseudocysts. PKM gene expression correlated with increasing malignant potential. In blood plasma, only MUC1, MUC4, and PYGL were elevated in adenocarcinoma compared to mucinous neoplasms. We identified a distinct RNA signature that can distinguish mucinous cystic neoplasms from benign cystic lesions (serous cysts and pseudocysts), which could be useful for guiding patient management and improving clinical outcomes. Validation in broader cohorts is essential for clinical implementation.