Initiation and shift of antidiabetic therapy and pancreatic cancer.

Abstract

4126 Background: Concerns have been raised on the risk of pancreatic cancer associated with specific anti-diabetic therapies. We have examined the risk of pancreatic cancer among patients with diabetes prescribed with an oral anti-diabetic drug (OAD) or an incretin drug (DPP4i and GLP-1 RA) or insulin. Methods: The public health insurance databases of Belgium and of Lombardy Region, Italy include nearly 100% of the population living in these countries. We created within these databases two cohorts that included adult patients who were first prescribed an incretin drug or another noninsulin antidiabetic drug (NIAD) during 01/07/2008-31/12/2013 in Belgium and during 01/01/2008-31/12/2012 in Lombardy Region. The risk of pancreatic cancer after prescription of an anti-diabetic drug was evaluated using multivariate adjusted Cox models including time-dependent variables. Adjusted hazard ratios (aHRs) from Belgium and Italy were pooled using fixed effects meta-analyses. Results: Results in both cohorts were similar. Among those patients prescribed an OAD, 45% of pancreatic cancers occurred within the 6 months following first prescription, 20% in months 7 to 12 after first prescription and proportions decreased progressively during follow-up. The aHR of pancreatic cancer among subjects prescribed an incretin compared to an OAD was 2.14 [95% CI, 1.71 to 2.67]. The aHR decreased from 3.35 [CI, 2.32 to 4.84] in the first 3 months after first incretin prescription, 2.12 [CI, 1.22 to 3.66] in months 3 to 5.9, 1.95 [CI, 1.20 to 3.16] in months 6 to 11.9, to 1.69 [CI, 1.12 to 2.55] after 12 months. The risk of pancreatic cancer among subjects who were subsequently prescribed insulin was 6.89 [CI, 6.05 to 7.85]. The time from OAD prescription to a shift to incretins or to insulin was significantly lower in patients who were subsequently diagnosed with a pancreatic cancer. Conclusions: The increased risk of pancreatic cancer associated with anti-diabetic therapies could be the consequence of an occult pancreatic cancer that provokes diabetes (reverse causation bias). The search for an occult pancreatic cancer in subjects with newly diagnosed diabetes or patients shifting to more potent anti-diabetic therapy may lead to earlier detection of this cancer.