Diabetes medications and pancreatic cancer risk: A population-based cohort study

Background: Studies of the relationship between diabetes medications and pancreatic cancer risk have produced inconclusive results. We aimed to examine associations between classes, subclasses, and individual diabetes medications with pancreatic cancer risk in a population-based retrospective cohort study. Methods: Among British Columbians aged > 35 (1996-2019), prescriptions for diabetes medications were categorised by ever/never use, cumulative duration, and dose. Time-varying Cox proportional hazards models adjusted for demographics were used to estimate hazard ratios (HRs) and 95 % confidence intervals (CIs) for associations between new diabetes medication use and pancreatic cancer. Confounding by indication was explored using active comparator analysis of ever/never associations relative to pioglitazone use. Results: The cohort consisted of 3,118,538 people (52,088,644 person-years), 7,540 of whom were diagnosed with pancreatic cancer. For every one-year increase in cumulative dose, diabetes medications in the insulin secretagogue class, and glyburide; an individual medication within the class, were associated with 2 % (HR=1.02, 95 % CI=1.02-1.03) and 3 % (HR=1.03, 95 % CI=1.02-1.05) increased risk of pancreatic cancer. For every one-year increase in cumulative dose, medications within the insulins and analogues class and insulin subclasses (basal and bolus insulins) were linked to a 4 % higher risk (HR=1.04, 95 % CI=1.03-1.05) of pancreatic cancer. In the active comparator analysis, elevated risk for basal insulins (HR=1.49, 95 % CI=0.33-6.63) was observed, consistent with the main analysis, although the risk was not statistically significant. Conclusion: Basal insulins may be associated with higher pancreatic cancer risk. Although confirmatory studies are needed, this finding may be informative for prescribing practices for high-risk populations with diabetes.