Twenty-five studies were included: 22 RCTs (sample size range 54 to 1,087) and three retrospective cohort studies, all of exenatide (sample size range 22,789 to 482,034). Study quality was considered high in 15 studies and moderate in 10 studies (scores ranged from 13 to 25).
Acute pancreatitis: Neither exenatide (13 studies) nor liraglutide (12 RCTs) were associated with increased incidence of acute pancreatitis. Sensitivity analyses yielded similar results. There was evidence for publication bias in the analysis of the 10 exenatide RCTs.
Any cancer: In the main analyses neither exenatide (10 RCTs) nor liraglutide (10 RCTs) were associated with increased incidence of cancer. Sensitivity analyses showed similar results for exenatide but liraglutide was associated with a statistically significant increased risk of cancer when only the five high quality studies were analysed (OR 2.60, 95% CI 1.08 to 6.27; Ι²=0%); neither the comparator treatments nor the number of cancers were explicitly stated for this analysis (although it appeared the comparators were other anti-diabetic drugs).
Thyroid cancer: There were no cases of thyroid cancer in patients taking exenatide. Liraglutide was not significantly associated with increased incidence of thyroid cancer.