[Predictors of malignant intraductal papillary mucinous neoplasm of the pancreas].

Intraductal papillary mucinous neoplasm (IPMN) is the most common pancreatic cystic neoplasm with a tendency to malignancy,and the 5-year survival rate of patients with invasive IPMN is less than 40%. Therefore,early detection of malignant cases and surgical intervention will significantly improve patient outcomes. The clinical guidelines recommended risk factors for IPMN malignancy according to imaging characteristics,clinical manifestations and serum tumor markers,among which dilation of the main pancreatic duct≥10 mm,the presence of solid components or enhanced mural nodules,and obstructive jaundice were recognized as high-risk factors. However,clinical practice has shown that the risk factors recommended by the current guidelines are not highly accurate in predicting IPMN malignancy,and the main problem is that the guidelines are less specific and allow follow-up cases to undergo unnecessary surgical resection. In recent years,the development of new technologies and methods such as genomics,radiomics,and artificial intelligence has greatly promoted the research progress of IPMN. Endoscopic ultrasound-guided fine needle aspiration of cystic fluid combined with molecular detection such as DNA and RNA,as well as radiomics combined with machine learning modeling,have shown superiority in improving the prediction accuracy of malignant IPMN,and will become an important tool to assist the diagnosis and treatment of IPMN in the future.