Abstract 3387: A plasma metabolomic signature improves prediction of malignant intraductal papillary mucinous neoplasm

Abstract Pancreatic cancer is one of the deadliest cancers primarily because most patients are diagnosed at advanced, unresectable stages. To reduce the incidence and mortality of pancreatic cancer, it is crucial to timely identify and optimally treat its precursors. One such precursor is intraductal papillary mucinous neoplasm (IPMN) as some of these cyst lesions have potential to progress to invasive cancer. Current clinical guidelines for predicting and treating malignant IPMN have a satisfactory sensitivity (>90%) but a dismal specificity (25-30%), leading to a number of unnecessary pancreatic resections in benign IPMN. Therefore, there is an urgent unmet need to identify molecular biomarkers to improve malignant IPMN prediction. The present study was thus conducted to identify metabolomic biomarkers that can preoperatively differentiate malignant IPMN from benign IPMN. A global plasma metabolomic analysis of 607 metabolites was performed on Xevo GS-X2 quadrupole time-of-flight (TOF) mass spectrometers (MS) using a 2D column configuration in 125 cases of low/moderate-grade IPMN and 50 cases of high-grade/invasive IPMN, confirmed by surgical pathology. Of the 607 metabolites, 19 were selected based on their p values for the differences between the two groups compared (all p<0.005). All potential combinations (1 to 8 metabolites) of the 19 metabolites were examined by 3-fold cross validation. The training and test sets included 67% and 33% of total samples, respectively, while keeping the same ratio of low/moderate-grade IPMN to high-grade/invasive IPMN. We repeated such cross validation 100 times by random selection of the training and test sets. The metabolite combination that can distinguish malignant IPMN from benign IPMN with the highest average AUC (area under the receiver operating characteristics curve) (=0.70) consisted of four lipid-related metabolites. AUCs (95% CIs) were 0.63 (0.53 - 0.72), 0.64 (0.55 - 0.73), 0.65 (0.56 - 0.74), and 0.64 (0.54 - 0.73) for the metabolites 1-4, respectively. CA19-9 is the only established biomarker for pancreatic cancer. As shown above, AUCs for the four individual metabolites and for their combination were greater than the AUC of CA19-9 [0.58 (0.47 - 0.69)]. Adding the four metabolites and CA19-9 to the model that contained clinical and imaging variables (including age, sex, BMI, main pancreatic duct diameter, and radiographic main duct involvement) yielded an AUC (95% CI) of 0.82 (0.75 - 0.90). In summary, we identified a plasma signature of four metabolites that outperformed CA19-9 in the discrimination of malignant IPMN from benign IPMN. If replicated in other independent patient populations, the findings of the present study can aid in the risk stratification of IPMN patients prior to surgery. Citation Format: Johannes Fahrmann, Hao Fan, Michele Yip-Schneider, Huangbing Wu, Ziyue Liu, Jun Wan, Sheng Liu, Samir Hanash, C. Max Schmidt, Jianjun Zhang. A plasma metabolomic signature improves prediction of malignant intraductal papillary mucinous neoplasm [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 3387.