The research team, led by Issei Kazume and Kenji Onda of Tokyo University of Pharmacy and Life Sciences, mined the U.S. Food and Drug Administration’s Adverse Event Reporting System (FAERS), the world’s largest spontaneous reporting database, covering an extraordinary 26.5-year window from the fourth quarter of 1997 through the first quarter of 2024. After data cleaning through the Japan Pharmaceutical Information Center’s curated version of the database, the final dataset comprised nearly 16 million individual adverse event reports, of which 61,673 involved acute pancreatitis. The scale matters enormously here, because drug-drug interaction signals are inherently rare and subtle, and only massive datasets can hope to detect them above statistical noise.

The methodological backbone of the study is disproportionality analysis, a statistical technique that compares how often a drug and an adverse event co-occur in reports against how often they would be expected to co-occur by chance. For individual drugs, the team calculated crude reporting odds ratios (cRORs) from 2 x 2 contingency tables built across all 8,585 distinct agents in the database. For pairwise drug-drug interactions, they employed the Omega shrinkage measure, a conservative interaction detection method developed by Norén and colleagues in 2008, using the lower bound of the 95% confidence interval (Omega025) as the signal detection criterion. The study followed the READUS-PV reporting guidelines for disproportionality analyses, a transparency framework designed to improve the reproducibility of pharmacovigilance signal detection.

The individual-drug analysis produced findings that largely align with established clinical knowledge while offering some striking numbers. Liraglutide showed the highest reporting signal among the incretin agents, with a cROR of 19.86, meaning reports of pancreatitis involving liraglutide were nearly twenty times more frequent than expected based on overall reporting patterns. Sitagliptin, the most widely prescribed DPP-4 inhibitor, showed a cROR of 10.90, while exenatide, the first GLP-1 receptor agonist to reach the market in 2005, registered 8.99. Metformin, the global workhorse of type 2 diabetes therapy, showed a more modest but still elevated cROR of 4.50. Perhaps most eye-catching was eluxadoline, a treatment for diarrhea-predominant irritable bowel syndrome, which posted a cROR of 40.23, consistent with prior post-marketing reports that prompted a U.S. label change for that drug.

Yet the numbers also reveal the fundamental interpretive challenge of spontaneous reporting data. Patients prescribed incretin-based therapies typically carry underlying conditions such as diabetes, obesity, or metabolic syndrome, each of which independently elevates baseline pancreatitis risk. Elevated signals also appeared for fenofibrate, prescribed for hypertriglyceridemia, itself a recognized pancreatitis trigger, and for oncology drugs such as pegaspargase and asparaginase, antiretrovirals such as didanosine and stavudine, and antipsychotics such as quetiapine and olanzapine. The authors are careful to note that these cRORs cannot be interpreted as quantitative risk estimates or evidence of causation; they are reporting associations that require external validation through mechanistic studies and epidemiological analyses in healthcare databases.

The truly novel contribution, however, lies in the pairwise drug-drug interaction analysis. The team systematically evaluated one-to-one combinations of seven GLP-1 receptor agonists (albiglutide, dulaglutide, exenatide, liraglutide, lixisenatide, semaglutide, and tirzepatide) and five DPP-4 inhibitors (alogliptin, linagliptin, saxagliptin, sitagliptin, and vildagliptin) against all other reported medications, generating 103,008 evaluated drug pairs. In the pooled analysis spanning the entire observation period, positive interaction signals based on the Omega025 criterion emerged for 308 pairs, roughly 0.3% of all combinations. The most frequently reported pair was insulin combined with exenatide, with 532 co-occurring reports, though its Omega025 value of 0.04 sat at the very margin of the detection threshold, suggesting the association may reflect confounding by indication, since insulin is often administered to manage hyperglycemia secondary to pancreatitis rather than representing a true pharmacological interaction.

Other top-ranked combinations told a more intriguing story. Exenatide paired with acetylsalicylate produced 436 reports with an Omega025 of 1.10, meaning the observed co-occurrence substantially exceeded the expected count. Exenatide with lisinopril, simvastatin, hydrochlorothiazide, paracetamol, and fenofibrate all generated robust positive signals with Omega025 values ranging from 1.15 to 1.75. Notably, several of the concomitant drugs involved, including statins, ACE inhibitors such as lisinopril and enalapril, and fenofibrate, have themselves been previously implicated in pancreatitis case reports, raising the possibility that these combinations genuinely compound pancreatic stress rather than merely reflecting reporting artifacts.

The critical methodological innovation came next. Rather than relying solely on the pooled analysis, which offers maximal statistical power but cannot distinguish stable associations from transient reporting phenomena, the team stratified the 26.5-year observation window into six consecutive multi-year periods and recalculated Omega025 values separately for each interval. This period-stratified approach is designed to mitigate the impact of stimulated reporting, the well-documented phenomenon in which regulatory actions, media coverage, or label changes trigger waves of reports that inflate apparent associations. The results were sobering for the pooled findings: no drug pair demonstrated a consistently robust interaction signal across five or six periods, and the 308 positive signals from the pooled analysis narrowed dramatically to just 11 pairs showing signals in three or more periods.

Among those 11 recurrent pairs, the atorvastatin-dulaglutide combination showed the highest temporal consistency, with positive signals observed consecutively across four periods, from Period 2 through Period 5. The remaining pairs, including acetylsalicylate-exenatide, liraglutide-lisinopril, exenatide-paracetamol, esomeparazole-exenatide, enalapril-exenatide, exenatide-ondansetron, exenatide-lorazepam, bisoprolol-exenatide, exenatide-morphine, and enoxaparin-exenatide, all showed signals in three periods. Strikingly, all 11 recurrent pairs involved GLP-1 receptor agonists rather than DPP-4 inhibitors, and most involved exenatide, the oldest agent in the class. The authors suggest this historical pattern may reflect heightened awareness following exenatide’s early post-marketing pancreatitis reports, which initiated the broader debate about the entire GLP-1 drug class, and the possibility of transient stimulated reporting during those formative years.

The bottom line, however, is reassuring in its clarity: no concomitant medication combination involving GLP-1 receptor agonists or DPP-4 inhibitors was identified as being robustly associated with increased acute pancreatitis reporting. The period-stratified analysis effectively reduced false-positive signals that had emerged in the pooled data, demonstrating the complementary value of the two analytical strategies. Pooled analyses maximize sensitivity but risk capturing reporting artifacts, while period-specific analyses sacrifice statistical power for temporal validity. The authors emphasize that the 11 recurrent weak signals, including the dulaglutide-atorvastatin pair, warrant continued pharmacovigilance and longitudinal monitoring rather than immediate clinical action. For the millions of patients now taking GLP-1-based therapies alongside statins, antihypertensives, and analgesics, the study offers a measured but important message: the massive FDA reporting database, analyzed with rigorous temporal scrutiny, has not yet produced convincing evidence that common concomitant medications dramatically amplify the pancreatitis risk of incretin-based drugs, though the question remains open for ongoing post-marketing surveillance.

Subject of Research: Comprehensive pharmacovigilance assessment of pairwise drug–drug interactions and acute pancreatitis in the FDA adverse event reporting system: focus on incretin-based drugs

Article Title: Comprehensive pharmacovigilance assessment of pairwise drug–drug interactions and acute pancreatitis in the FDA adverse event reporting system: focus on incretin-based drugs

Article References: Kazume, I., Onda, K., Tanaka, S., Yokokawa, T., & Suzuki, K. (2026). Comprehensive pharmacovigilance assessment of pairwise drug–drug interactions and acute pancreatitis in the FDA adverse event reporting system: focus on incretin-based drugs. Frontiers in Drug Safety and Regulation, 6, Article 1827098. https://doi.org/10.3389/fdsfr.2026.1827098

Image Credits: AI Generated

DOI: 10.3389/fdsfr.2026.1827098

Keywords: GLP-1 receptor agonists, acute pancreatitis, drug-drug interactions, FAERS, pharmacovigilance, DPP-4 inhibitors, Comprehensive, assessment, pairwise, drug, interactions, acute