Abstract
This study aimed to develop and validate a simple scoring system to determine the high-risk group for pancreatic cancer (PC) in the asymptomatic general population. The scoring system was developed using data from PC cases and randomly selected non-PC cases undergoing annual medical checkups between 2008 and 2013. The performance of this score was validated for participants with medical checkups between 2014 and 2016. In the development set, 45 PC cases were diagnosed and 450 non-PC cases were identified. Multivariate analysis showed three changes in clinical data from 1 year before diagnosis as independent risk factors: ΔHbA1c ≥ 0.3%, ΔBMI ≤ -0.5, and ΔLDL ≤ -20 mg/dL. A simple scoring system, incorporating variables and abdominal ultrasound findings, was developed. In the validation set, 36 PC cases were diagnosed over a 3-year period from 32,877 participants. The AUROC curve of the scoring system was 0.925 (95%CI 0.877-0.973). The positive score of early-stage PC cases, including Stage 0 and I cases, was significantly higher than that of non-PC cases (80% vs. 6%, p = 0.001). The simple scoring system effectively narrows down high-risk PC cases in the general population and provides a reasonable approach for early detection of PC.