New onset diabetes as a predictive factor of focal lesions in the pancreas in a high-risk screening program.

Abstract

1512 Background: Genetic evolution studies have suggested the existence of a window of opportunity to improve clinical outcomes by intercepting pre-malignant lesions. This study reports the outcomes of Pancreatic Cancer (PC) surveillance in a high risk (HR) cohort followed at The University of Texas MD Anderson Cancer Center (MDA) between 2014 and 2018. Methods: The MDA PC High-Risk Clinic (MDA-PCHRC) performs surveillance based on risk stratification. This study reports 54 months of surveillance. The patients were stratified based on PC family history, personal history of other cancers, and germline mutations ( BRCA1/2, PALB2, STK11, CDKN2A, TP53) with high susceptibility to PC. Low risk patients were not offered surveillance while patients in the moderate or high-risk category were enrolled into a program which included annual screening with blood markers (CA 19-9, fasting glucose, HbA1C, amylase, and lipase) and magnetic resonance imaging/ cholangiopancreatography (MRI/MRCP). High risk patients also had a baseline endoscopic ultrasound (EUS). We evaluated the yield of premalignant lesions detection and our main goal was to detect predictive factors of focal pancreatic lesions and to validate our risk stratification strategy. Results: A total of 206 patients have been referred to our clinic during this time period. From this group, 126 (61%) patients completed at least one cycle of baseline surveillance, for the purposes of the analysis we only focus in the high risk (n=71) and moderate risk group (n=38). We have identified de novo pancreatic focal lesions in 22 patients, 20 from the high risk group (28%) and 2 from the moderate group (5%). Those lesions included 7 patients with simple cysts, 9 with side-branch IPMN, 3 with main duct IPMN, 1 with pseudocyst, 1 with mucinous cyst and 1 with a solid nodule (pancreatic neuroendocrine tumor). We compared demographic information (age, gender, and ethnicity) as well as family and personal medical history between patients with focal pancreatic lesions vs negative or diffuse findings. We found that new onset diabetes was significantly correlated with presence of focal pancreatic lesions 5 (22%) of patients with focal lesions versus patients without non focal lesion 2 (2%) ( P=0.003). Conclusions: Screening at the MDA-PCHRC detect pancreatic premalignant lesions in 20% of the patients in our cohort. We validated our risk stratification methodology and found that new-onset diabetes is predictive of pancreatic lesions, thus suggesting that this factor could be an important biomarker of focal lesions in a HR population.