ABSTRACT
AIM: To investigate the association between the use of incretin agents and the risk of pancreatic cancer. METHODS: A retrospective population-based cohort study, using data from the Clinical Practice Research Datalink, 2007-2012, was conducted. Patients (n = 182 428) with at least one non-insulin antidiabetic drug (NIAD) prescription and aged ≥18 years during data collection, were matched one-to-one to control patients without diabetes. Multivariable Cox proportional hazards models and a new user design were used to estimate the hazard ratio (HR) of pancreatic cancer in incretin users (n = 28 370) compared with control subjects without diabetes and other NIAD-treated patients. Time-dependent adjustments were made for age, sex, lifestyle, comorbidities and drug use. RESULTS: The mean duration of follow-up was 4.1 years for incretin users. Current NIAD use was associated with a fourfold increased risk of pancreatic cancer [HR 4.28, 95% confidence interval (CI) 3.49-5.24]. This risk was almost doubled among current incretin users as compared with control subjects. Incretin use was not associated with pancreatic cancer when compared with control subjects with diabetes (HR 1.36, 95% CI 0.94-1.96); however, the 'new user' design did show an association between incretin use and pancreatic cancer when compared with control subjects with diabetes. In both cohorts with prevalent and incident users of antidiabetic drugs, the risk of pancreatic cancer almost doubled in those who had recently initiated incretin therapy (up to seven prescriptions), whereas this elevated risk dropped to baseline levels with prolonged use. CONCLUSIONS: We found that incretin use was not associated with pancreatic cancer after adjustment for the severity of the underlying Type 2 Diabetes Mellitus (T2DM). The elevated risk of pancreatic cancer in those recently initiating incretin agents is likely to be caused by protopathic bias or other types of unknown distortion. The presence of considerable confounding by disease severity and the lack of a duration-of-use relationship do not support a causal explanation for the association between incretin agents and pancreatic cancer.