A prediction rule for estimating pancreatic cancer risk in chronic pancreatitis patients with focal pancreatic mass lesions with prior negative EUS-FNA cytology

Abstract

Objective. Considerable false-negative endoscopic ultrasound guided fine needle aspiration (EUS-FNA) findings exist in chronic pancreatitis patients with focal pancreatic mass lesions. Our aim was to develop a prediction rule to stratify risk for pancreatic cancer in chronic pancreatitis patients with focal pancreatic mass lesions with prior negative EUS-FNA cytology. Material and methods. A total of 138 eligible consecutive patients were identified from three hospitals between January 2000 and May 2008. A final diagnosis of pancreatic mass lesions was confirmed histologically or verified by a follow-up of at least 12 months. A prediction rule was developed from a logistic regression model by using a regression coefficient-based scoring method, and then internally validated by using bootstrapping. Results. The rate of pancreatic cancer in the cohort was 18.1%. The prediction rule, which was scored from 0 to 10 points, comprised five variables: sex, mass location, mass number, direct bilirubin, and CA 19-9. Among the 87.7% of patients with low-risk scores (≤3), the risk of pancreatic cancer was 13.2%; by comparison, this risk was 52.9% (p < 0.001) among the 12.3% of patients with high-risk scores (>3). If further invasive tests were used for patients with high risk, 36% of patients with pancreatic cancer would not be missed. The prediction rule had good discrimination (area under the receiver operating characteristic curve, 0.72) and calibration (p = 0.96). Conclusions. The prediction rule can provide available risk stratification for pancreatic cancer in chronic pancreatitis patients with focal mass lesions with prior negative EUS-FNA cytology. Application of risk stratification may improve clinical decision making.