Beyond Imaging: Integrated Clinical, Endocrine, and Molecular Risk Stratification in Pancreatic Cystic Lesions: A Literature Review of Current Evidence

Pancreatic cystic lesions (PCLs) are increasingly detected as incidental findings on high-resolution imaging. While many are benign, entities like intraductal papillary mucinous neoplasms (IPMNs) carry a significant risk of progression to pancreatic ductal adenocarcinoma (PDAC). The central clinical dilemma is balancing the need for early surgical intervention in high-risk patients against the high morbidity of pancreatic surgery, which is often performed on patients with low-grade or benign lesions.

Traditional management relies on morphological features identified via CT, MRI, and endoscopic ultrasound (EUS), specifically focusing on 'high-risk stigmata' and 'worrisome features.' However, these structural assessments often lack the sensitivity to predict malignant transformation accurately. The current literature suggests a paradigm shift toward integrating biological and computational data. Notably, new-onset diabetes (NOD) is now recognized as a clinically accessible marker that may reflect tumor-induced metabolic changes or paraneoplastic phenomena, serving as a potential indicator of high-risk IPMNs.

Recent advancements in cyst fluid analysis have moved beyond traditional cytology and carcinoembryonic antigen (CEA) measurements. Next-generation sequencing (NGS) allows for the detection of recurrent somatic mutations (e.g., KRAS, GNAS, TP53, SMAD4) that provide prognostic insight into the likelihood of high-grade dysplasia. Simultaneously, artificial intelligence (AI) and radiomic models are being developed to integrate these high-dimensional datasets. These computational tools can identify subtle patterns in imaging and molecular profiles that are not apparent to the human eye, potentially offering a more personalized approach to patient triage and surveillance.