The results add important pooled and patient-level data to our knowledge regarding the pancreatic safety of incretin-based therapies. In this study, the incidence of reported acute pancreatitis was numerically greater with liraglutide than with comparators (1.6 cases/1,000 PYE in the liraglutide group, 0.7 cases/1,000 PYE in the total active comparator group, and 0.9 cases/1,000 PYE in the active comparator group excluding sitagliptin and exenatide). Prior studies in patients with type 2 diabetes (not being treated with incretin-based therapy) reported an incidence of acute pancreatitis of 0.5–5.6 events/1,000 PYE.[22,32–34] A trend toward an increase in the relative risk of acute pancreatitis was seen in the liraglutide group versus the active comparator groups (2.1 [95% CI 0.3, 16.0], P = 0.4478 vs. all active comparators; 1.7 [0.2, 13.2], P = 0.6241 vs. active comparators excluding sitagliptin and exenatide), although this trend was based on very few cases. The CIs for the relative risk were wide, indicating an imprecise estimate of risk and low power to detect a difference in the incidence of acute pancreatitis between groups. Therefore, these statistical analyses should be considered as preliminary and inconclusive. Chronic pancreatitis also occurred more frequently with liraglutide than with comparators (four cases vs. zero cases; no statistical analyses conducted). To date, no drug has been implicated as causing chronic pancreatitis.
Upon review of all the reported pancreatitis cases, it was evident that not all of the cases fulfilled established diagnostic criteria. In one case of acute pancreatitis reported with liraglutide (case 1), acute pancreatitis was diagnosed in an individual with elevated lipase levels (more than three times the upper limit of the normal range) but without significant abdominal pain or imaging suggestive of the condition (Table 3). Therefore, this case did not fulfill the Atlanta criteria for the diagnosis of acute pancreatitis, as agreed on by international consensus (Table 2); note that as for all reported pancreatitis cases, it was nonetheless included in the statistical analyses. Serum lipase and amylase levels are elevated in ~13–20% and 6–8% of individuals with type 2 diabetes, respectively, who are not receiving GLP-1 receptor agonists or DPP-4 inhibitors and who do not have gastrointestinal symptoms (1–2% have threefold or greater lipase elevations).[36–39] Furthermore, in a 1-year longitudinal study, liraglutide raised serum lipase levels by a median of ~10 units/L (from 40 to 49 units/L; normal limit <60 units/L). This increase in lipase activity was unaccompanied by pancreatitis in 99.8% of cases and returned to baseline when liraglutide treatment was stopped. The significance of elevated lipase levels in patients with type 2 diabetes before starting a GLP-1 receptor agonist and the further rise after drug initiation is presently unclear. However, these data show that the presence of elevated serum levels of pancreatic enzymes in patients with type 2 diabetes not taking or taking GLP-1 receptor agonists needs to be interpreted with caution when diagnosing acute pancreatitis.
There is a substantial body of literature on drug-induced acute pancreatitis. The large majority of these reactions occur within the first 1–12 weeks of exposure to the drug and are believed to be due to idiosyncratic or allergic reactions. In contrast, in six of the cases reported in the present study, the patients had been receiving liraglutide for 6–18 months before the acute pancreatitis event occurred. A mechanism has been proposed to explain a long latency between exposure to a GLP-1 receptor agonist or a DPP-4 inhibitor and an event of acute pancreatitis. GLP-1 receptor agonists have been reported to induce pancreatic duct gland hyperplasia or pancreatic intraepithelial neoplasia lesions in certain animal models. Authors have hypothesized that these structures, which develop over time, obstruct the small pancreatic ductules, leading to late-presenting acute pancreatitis. From a pathophysiologic point of view, how slow occlusion of the peripheral pancreatic ductules (should this occur with GLP-1 receptor agonists in humans) over a prolonged period may lead to acute pancreatitis is unclear. The main model of occlusion of pancreatic ducts leading to acute pancreatitis is gallstone occlusion, during which abrupt blockage by a gallstone leads to a sudden increase of pressure in the pancreatic ductular system.
One or more risk factors or causes for acute pancreatitis (gallstones, ERCP with sphincterotomy, chronic pancreatitis) were present in one-half of the acute pancreatitis cases observed with liraglutide, and all patients were taking multiple drugs. It is possible that GLP-1 receptor agonists may potentiate other established causes of acute pancreatitis. For example, it is possible that liraglutide, in combination with other medications, may predispose to pancreatitis. It is also possible that liraglutide may induce gallstone formation or cause a stone to migrate down the common bile duct. GLP-1 receptor agonists may cause a patient with established chronic pancreatitis to have an acute exacerbation of pancreatitis (case 6) (Table 3). Of the 37 of 3,137 subjects with a baseline history of pancreatitis in the observational EVIDENCE study, only one patient developed renewed symptoms of pancreatitis during 12 months of liraglutide therapy.[43,44] However, the number of subjects was small, and any effects of liraglutide on pancreatitis recurrence over a period of several years remain unknown. The U.S. label for liraglutide states that other antidiabetic therapies should be considered in patients with a history of pancreatitis, whereas the European label for liraglutide does not make specific recommendations with regard to treating these patients.
The safety profile of liraglutide is under continuous surveillance through meticulous routine pharmacovigilance, with any cases of pancreatitis being analyzed and reported to the health authorities. In addition, two pharmacoepidemiological studies of liraglutide are under way (using OptumInsight in a U.S. study and Clinical Practice Research Datalink in a U.K. study) that will supply substantial scientific information to this debate once completed. In an interim analysis of the study using OptumInsight, no significant elevation in the incidence of unadjudicated acute pancreatitis diagnoses with liraglutide versus non-GLP-1–based comparators over a median follow-up of 15 months has been found (187.5/100,000 PYE vs. 154.4/100,000 PYE, adjusted rate ratio 1.10 [0.81, 1.49]).
The long-term cardiovascular outcome trial LEADER (Liraglutide Effect and Action in Diabetes: Evaluation of Cardiovascular Outcome Results—A Long Term Evaluation) in which 9,340 patients with type 2 diabetes are enrolled is expected to provide information regarding safety of up to 5 years' liraglutide exposure by 2016. Similar safety outcome trials are ongoing or have been completed for other incretin-based therapies. Of interest, the SAVOR-TIMI 53 (Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus–Thrombolysis in Myocardial Infarction 53) and EXAMINE (Examination of Cardiovascular Outcomes with Alogliptin versus Standard of Care) studies (long-term outcomes trials of DPP-4 inhibitors in patients with type 2 diabetes) indicated low incidences of pancreatitis with saxagliptin (confirmed acute pancreatitis: n = 17 [0.2%] vs. 9 [0.1%], P = 0.17) and alogliptin (acute pancreatitis: n = 12 [0.4%] vs. 8 [0.3%], P = 0.50) compared with placebo, with a numerical imbalance between arms.[47,48]
In the present study, the incidences of reported acute and chronic pancreatitis were numerically greater with liraglutide than with comparators, but the small number of cases and the confounding variables observed in several of the cases preclude firm conclusions. Combined data obtained from completed and multiple ongoing studies should provide more definitive evidence about whether incretin-based therapies cause pancreatitis. We share the current EMA and FDA view that any potential risk of pancreatitis is adequately reflected in the liraglutide label, which still carries a warning statement regarding pancreatitis.[2,9]
The authors thank Henrik F. Thomsen (Novo Nordisk A/S) for contributing to the data analysis and Laura Elson (Watermeadow Medical, U.K. [supported by Novo Nordisk A/S, Copenhagen, Denmark]) for assistance with the preparation of the manuscript.
Parts of this study were presented at the 2012 Joint Meeting of the American Pancreatic Association and International Association of Pancreatology, Miami, FL, 31 October–3 November 2012. These data were subsequently incorporated into studies published by other authors.[16,24] A short statement on six of the cases included in the case reviews was included in a letter to The Annals of Pharmacotherapy.
Diabetes Care. 2015;38(6):1058-1066. © 2015 American Diabetes Association, Inc.