Pancreatitis Risk Associated with GLP-1 Receptor Agonists, Considered as a Single Class, in a Comorbidity-Free Subgroup of Type 2 Diabetes Patients in the United States: A Propensity Score-Matched Analysis

This study was approved by the Institutional Board Review Committee at Charleston Area Medical Center. Written informed consent from patients was waived due to the de-identified nature of the TriNetX clinical database. The TriNetX (Cambridge, MA, USA) database is a global federal research network that combines real-time data with electronic medical records. Data used in our study were retrieved from the TriNetX research database using the US Collaborative Network, which included information from 61 healthcare organizations within the U.S. Using the International Classification of Diseases 10th revision codes (ICD-10), adult patients aged ≥ 18 years with type 2 diabetes mellitus (T2DM) were identified. Included patients were divided into two cohorts that underwent propensity score matching (PSM) and were subsequently compared. We matched patients of both cohorts based on baseline demographics, comorbidities, medications known to cause drug-related pancreatitis, and laboratory values.

After performing PSM, outcome analysis was performed. Our outcome for this study was the risk of pancreatitis between the two cohorts, which was analyzed using Kaplan–Meier curves and log-rank tests. Risk ratios (RR) with 95% confidence intervals (CI) were calculated for our outcome. A p-value of <0.05 was considered statistically significant. All statistical analyses were conducted on the TriNetX platform on https://live.trinetx.com/↗ (accessed date 31 March 2024, 23:35:21 UTC).

Patients with T2DM were divided into two cohorts: the first cohort comprised patients who received GLP-1 RAs, and the second cohort comprised patients who did not receive GLP-1 RAs. The selected agents of the GLP-1 RA class that were used in our study were dulaglutide, lixisenatide, exenatide, liraglutide, and semaglutide. We excluded patients with type 1 diabetes mellitus, heart failure, ischemic heart diseases, hypertension, and chronic kidney disease. We compared the risk of pancreatitis between the two cohorts over time using a 1:1 PSM model using patients' baseline characteristics as well as risk factors commonly associated with the development of pancreatitis. PSM components included demographics, comorbidities, medications used that are known to cause drug-related pancreatitis, and lab values. Demographics included age at index, race, gender, and ethnicity. Comorbidities included, but were not limited to, hypertensive disorders, ischemic heart disease, gallstones, annular pancreas, alcohol use disorders, hypertriglyceridemia, hypercalcemia, cystic fibrosis, and cannabis use. Medications included most medications and drug classes that were reported in the literature to be associated with pancreatitis. Lab values included hemoglobin A1C level, body mass index (BMI), triglyceride level, and calcium level. The full list of items used for PSM is shown in the. Supplementary Materials

A total of 969,240 patients with type 2 diabetes mellitus were identified. Of those, 93,608 (9.7%) patients were on a GLP-1 RA, and 875,632 (90.3%) patients were not on a GLP-1 RA. The mean age in the GLP-1 group was 47.3 with a standard deviation (SD) of 11.9. More than half the cohort comprised females at 62.6%. The mean body mass index (BMI) in the GLP-1 group was 35.6 kg/m² with a SD of 6.9. In the GLP-1 group, alcohol use disorder was found in 1.4% and cholelithiasis was found in 2.1%. The mean triglyceride level in the GLP-1 group was 187 mg/dL with a SD of 175, and a calcium level of 9.4 mg/dL with a SD of 0.5. Of the GLP-1 group, 53.4% received metformin, 38.9% received glucocorticoids or mineralocorticoid, 32.5% received penicillins or a beta-lactam antibiotic, and 22.9% received non-steroidal anti-inflammatory analgesics (NSAIDs). A full comparison of the cohorts' baseline demographics, comorbidities, medications, and lab values between the two cohorts before and after PSM is highlighted in the Supplementary Materials.

Analysis of the cohorts' baseline demographics, comorbidities, medications used, and lab values did not show any significant differences after PSM. We compared the rate of pancreatitis between the two cohorts after PSM over time. The rate of pancreatitis was similar in patients receiving GLP-1 RAs and those who did not receive GLP-1 RAs (0.1% vs. 0.1%, p = 0.035). Patients receiving GLP-1 RAs had a significantly lower rate of pancreatitis at one year (0.1% vs. 0.2%, p = 0.022), three years (0.2% vs. 0.3%, p < 0.001), and five years (0.3% vs. 0.4%, p < 0.001). The lifetime risk of pancreatitis was significantly lower in patients receiving GLP-1 RAs compared to those who did not receive GLP-1 RAs (0.3% vs. 0.4%, p < 0.001). The hazard ratio of the risk of pancreatitis over the study duration is shown in Table 1 and a summary of the results is highlighted in Table 2 and Figure 1.

Proglucagon is a protein in enteroendocrine cells, alpha cells of the pancreas, and in the brainstem. Proglucagon undergoes post-translational processing by a convertase enzyme to produce glucagon, Glicentin-related pancreatic polypeptide (GRPP), and the major proglucagon fragment (MPGF) [26]. This MPGF contains both glucagon-like peptide (GLP)-1 and GLP-2. Peripherally, the main source of GLP-1 is the enteroendocrine GLP-1-producing cells in the small intestine and colon [27]. Centrally, GLP-1 is also produced by the brain in a similar fashion to neuroendocrine cells [28].

GLP-1 secretion from enteroendocrine cells is triggered by carbohydrates, proteins, and fats [27,29,30,31]. Each of these act differently on the enteroendocrine cells and eventually lead to the extraction of GLP-1. This is crucial for its release as studies proved that IV administration of glucose, for instance, does not lead to a change in GLP-1 levels in the blood [32]. This means that these nutrients have to pass through the gut to trigger GLP-1 secretion.

GLP-1 is also secreted centrally from the brain by the release of preproglucagon neurons. They become activated after gastric distension or in a response to enzymes or hormones such as cholecystokinin and leptin [33,34,35]. Furthermore, the peripheral secretion of GLP-1 triggers its own central secretion [26]. Peripherally-released GLP-1 activates GLP-1 receptors on the vagus nerve, which in turn stimulates centrally-located preproglucagon neurons and the subsequent secretion of GLP-1 [36,37,38].

GLP-1 has a very short half-life with a maximum of 11 min [39,40,41]. Peripherally, GLP-1 is mainly metabolized by the liver and endothelial cells [42]. In order to be able to clinically contribute to diabetes management, a long-acting version of GLP-1 is needed to be synthesized. Exendin-4 was discovered in a venom of a lizard in southwestern USA and was found to have a 53% resemblance to the innate GLP-1 [43]. The synthetic version of exendin-4 is currently available commercially as Exenetide, which is slower to be metabolized with a half-life of 4 h [44]. Further biochemical adjustments led to the synthesis of Liraglutide, which has 97% similarity to the innate GLP-1 [44]. Further advancements and biochemical improvements led to the synthesis of the remainder of the commercially available GLP-1 RAs.

Exenatide and lixisenatide are exendin-4 derivatives and are typically administered daily 60 min before meals. Liraglutide and oral semaglutide are modified human GLP-1 and are taken any time daily. Dulaglutide and subcutaneous semaglutide are administered once weekly [45].

From a metabolic standpoint, multiple clinical trials were conducted to compare these GLP-1 RAs [46]. One meta-analysis showed that once-weekly GLP-1 RA administration had better hemoglobin A1c (HbA1c) reduction when compared to twice daily exenatide. However, it was not better than once daily liraglutide injection [47]. Another meta-analysis showed that liraglutide had better HbA1c reduction than exenatide but was not significantly different when compared to dulaglutide. The LEAD-6 trial and the DURATION-6 trial further confirmed the previous statement and showed that liraglutide, when compared to exenatide, had better reduction in HbA1c and weight with fewer adverse effects [48,49].

The AWARD-1 trial showed that dulaglutide was superior to exenatide, while the AWARD-6 trial showed that it was not inferior to liraglutide [50]. The SUSTAIN-7 trial compared semaglutide to dulaglutide which showed the superior effect of semaglutide in glycemic control and weight loss, it also showed they have a similar safety profile [50].

From a cardio–renal standpoint, there were seven clinical trials done on seven GLP-1 RAs that showed that liraglutide, subcutaneous semaglutide, and dulaglutide all showed significant cardiovascular risk reduction [51]. The LEADER trial showed that liraglutide had a cardiovascular benefit in high-risk patients, while the SUSTAIN-6 trial showed a cardiovascular-associated mortality reduction with semaglutide use as well as albuminuria reduction [46]. The SUSTAIN and PIONEER trials showed that semaglutide has a significant risk reduction in major adverse cardiovascular events [52]. A list of the previously mentioned trials with the studied GLP-1 RA used in each and their overall benefit is shown in Table 3 and Table 4 [53].

Due to their mechanism of action via attachment to alpha and beta cells in the pancreas and their subsequent stimulation, the concern for the occurrence of pancreatitis was raised [59]. The Food and Drug Administration (FDA) issued a warning due to concerns about GLP-1 RAs and the increased risk of pancreatitis and pancreatic cancer [60]. This was further reinforced in their latest release in 2024 [61,62]. This was recently debunked by very recent studies that did not show any association between GLP-1 RA use and pancreatic cancer [63,64,65].

Dulaglutide carries a boxed warning that it may increase the risk of pancreatitis [66,67]. Lixisenatide, according to its manufacturing label, was associated with cases of pancreatitis, and an alternative antidiabetic regimen should be pursued [68,69]. There were also post-marketing reports of acute pancreatitis in patients taking exenatide which were submitted to the Food and Drug Administration (FDA) and were published in the literature [70,71,72,73]. Some reports showed that liraglutide at different doses is associated with an increase in lipase levels up to 30% [74,75,76]. This prompted the manufacturer's labeling to recommend the prompt discontinuation of liraglutide if pancreatitis is suspected [77]. Similarly, some other case reports were published associating liraglutide with pancreatitis, which prompted their labeling to recommend against their use if pancreatitis is suspected [78,79,80,81].

Other conflicting studies debunked such associations [7,8,73,82,83]. A systematic review in 2014 of randomized and non-randomized trials by Ling Li et al. was unable to find convincing evidence that GLP-1 RAs increase the risk of pancreatitis [84]. The FDA and the European Medicines Agency (EMA) conducted a comprehensive evaluation of GLP-1 RAs due to the rising post-marketing reports of pancreatitis and pancreatic cancer [85]. The FDA evaluated more than 250 toxicology studies microscopically, which did not reveal any findings of pancreatitis [85]. Both the FDA and the EMA were not able to reach a final conclusion about such an association, however, they both agreed that the totality of the available data at that time was reassuring [85]. Another study by Meier and Nauck evaluated the pooled pancreatitis ratio in clinical trials ongoing at that time [86]. They saw a slightly elevated risk of pancreatitis with GLP-1 RA use; however, the number of incident cases were very small, and the statistical power was limited [86]. These findings aligned with our study outcome over time. The risk of pancreatitis was very similar between the cohort receiving GLP-1 RAs and those who did not receive GLP-1 RAs. Over the course of five years, the patients who received GLP-1 RAs had a statistically significant lower risk of pancreatitis than those who did not. The risk of pancreatitis in both cohorts was very low, which also confirmed the findings of the current literature [84,85]. This debunked the association of pancreatitis incidence with GLP-1 use. In fact, patients receiving GLP-1 RAs had a significantly lower rate of pancreatitis in their lifetime. Our findings should allow physicians to safely prescribe GLP-1, especially as their cardiovascular and renal benefits outweigh the very minimal risk of a pancreatitis incident. A list of some of the available GLP-1 RA studies with the reported pancreatic outcomes is shown in Table 5.

Some of our study strengths included the use of a nationwide database and the inclusion of a large cohort of patients. Additionally, using a 1:1 PSM model allowed us to reduce the effects of confounding the baseline covariates by creating matched cohorts, in which the distribution of the measured baseline covariates was similar in treated and control participants. This method created very similar cohorts after PSM and allowed us to clarify a more precise association while minimizing confounders. Furthermore, the exclusion of patients with CKD and using BMI in the PSM excluded and accounted for other indications for the use of GLP-1 RAs and ensured that their use in our patient population was for glycemic control.

Our study did not come without limitations. One of our study limitations was the use of the GLP-1 RA class as a whole without segmentation into different agents. This limited us from studying the variations between each single agent and their safety profile. Further studies and clinical trials are needed to explore such outcomes. Furthermore, despite the inclusion of a large number of patients and using a 1:1 PSM model, the use of a U.S.-based database and the exclusion of patients with multiple comorbidities limited the generalizability of our findings. We also focused on the risk of pancreatitis with GLP-1 RA use and did not explore other possible associations, such as the risk of solid organ malignancies, which warrant more clinical studies to further clarify the GLP-1 RA safety profile.