Risk of Pancreatic Cancer in Glycemically Defined New-Onset Diabetes: A Prospective Cohort Study

Increased 3-year incidence of pancreatic cancer following new-onset diabetes (NOD) observed in retrospective studies needs prospec4ive validation. It is unknown if incidence varies by race/ethnicity. In a prospective, observational study using active real-time surveillance of electronic health records we identified 18,838 adults >50 years of age with NOD defined by glycemic criteria (GNOD). In this interim analysis, we report 3-year Kaplan-Meier estimates of proportion diagnosed with pancreatic cancer following GNOD (absolute incidence (95% Confidence Intervals)) and associated Standardized Incidence Ratio (SIR) by race/ethnicity; overall 3-year incidence of pancreatic cancer adjusting for racial distribution of incident diabetes in the United States; and interval between GNOD and pancreatic cancer diagnosis. During median follow-up of 2.3 years, 82 pancreatic cancers have been diagnosed (60% male, mean age 71±8 years). The 3-year estimates for proportion diagnosed with pancreatic cancer (95% CI) and associated SIR (95% CI) by race/ethnicity were: non-Hispanic Whites (n=6,518) 0.84% (CI: 0.60, 1.07) (SIR 6.4 (CI: 4.8, 8.4)); Hispanics (n=5,984) 0.40% (0.20, 0.60) (SIR 4.2 (2.6, 6.3)); African Americans (n=2,192) 0.37% (0.07, 0.67) (SIR 2.4 (1.0, 5.0), and Asian/Pacific Islander (n=3,360) 0.22% (.06, 0.39) (SIR 3.0 (1.4, 6.0). Overall, race-adjusted 3-year pancreatic cancer incidence was 0.62%. On average GNOD occurred 8 months prior to clinical diagnosis; 0-4 months in 30.5%, 4-12 months in 31.3%, 12-24 months in 19.5%, and 24-36 months in 18.7%. Glycemically defined new-onset diabetes, identifiable in real-time using active surveillance of electronic health records, is associated with a high 3-year incidence of pancreatic cancer with marked racial/ethnic differences. Longer term risk needs further study.