Thermal Liquid Biopsy: A Promising Tool for the Differential Diagnosis of Pancreatic Cystic Lesions and Malignancy Detection.

Simple Summary: Mucinous epithelial pancreatic cystic lesions (PCLs) are premalignant lesions detectable through imaging techniques; however, distinguishing them from other PCLs with lower malignancy potential is challenging. Current methods like biochemical markers and genomic studies are not always reliable. Thermal liquid biopsy (TLB) is an innovative tool that analyzes the thermal profile of biological samples to detect disease-related alterations. In a retrospective study of 35 intracystic fluid samples obtained via fine needle aspiration, predictive models were developed using machine learning algorithms. Two classification models were created: TLB1, which differentiates mucinous from non-mucinous PCLs, demonstrating 92% sensitivity and 86% negative predictive value, and TLB2, which identifies benign and malignant mucinous lesions, achieving an area under the curve of 1.00. TLB shows promise in improving the differential diagnosis of PCLs and in detecting malignant transformations. Background/Objectives: Mucinous epithelial pancreatic cystic lesions (PCLs) are premalignant lesions readily detectable through imaging techniques such as multidetector computed tomography, magnetic resonance imaging, and endoscopic ultrasound (EUS). However, distinguishing these from other PCLs with lower or no malignant potential, and the early identification of those undergoing malignant transformation, remains a diagnostic challenge. Current methods, including biochemical markers in intracystic fluid (ICF) and genomic studies, offer some assistance but are not always reliable or accessible. Thermal liquid biopsy (TLB) is a novel diagnostic tool that examines the thermal profile (thermogram) of biological samples, reflecting their response to heat and thereby revealing characteristics of their overall composition or disease-induced alterations. Methods: In this retrospective study, a total of 35 ICF samples (49% mucinous) obtained via EUS-FNA (fine needle aspiration) were analyzed using TLB. Thermogram data were utilized to develop predictive models for differential diagnosis between mucinous and non-mucinous PCLs or malignancy detection through machine learning algorithms. Results: Two classification models were developed: TLB1 ("mucinous" vs. "non-mucinous" PCLs) and TLB2 ("benign mucinous" vs. "malignant mucinous" PCLs). The TLB1 model demonstrated a sensitivity of 92% and a negative predictive value of 86%, with an area under the curve (AUC) of 0.79 (0.59–0.99), indicating good discriminative ability between the two groups. The TLB2 model exhibited excellent predictive capability, with an AUC of 1.00. Conclusions: TLB analysis of PCLs is a promising tool that could significantly enhance the differential diagnosis of PCLs, enabling the efficient identification of mucinous lesions and even those undergoing malignant transformation. [ABSTRACT FROM AUTHOR]