Glucagon-like peptide-1 receptor agonists (GLP-1 RAs) are widely prescribed for type 2 diabetes mellitus (T2DM) and for obesity. Prior observational studies have reported conflicting findings regarding whether GLP-1 RA exposure alters long-term risk of pancreatic cancer. The present study aimed to evaluate the association between initiation of GLP-1 RA therapy and incident pancreatic cancer risk compared with six classes of antidiabetic agents using a large federated electronic health record network.
The investigators performed a retrospective, new-user cohort study using the TriNetX global federated electronic health record network. The design contrasted adults with T2DM initiating GLP-1 RA therapy against incident users of six comparator antidiabetic drug classes in separate analyses. The study period for ascertaining incident pancreatic cancer included events occurring between 365 and 7,300 days after the index prescription.
Each comparison (GLP-1 RA versus one comparator class) used 1:1 propensity score matching with a greedy nearest-neighbor algorithm and a caliper of 0.1 pooled standard deviations. Matching accounted for 39 baseline covariates spanning demographics, comorbidities, procedures, and medication exposures. The matching approach was intended to balance measured confounders between GLP-1 RA initiators and comparator drug initiators.
The primary outcome was incident pancreatic cancer defined by ICD-10-CM code C25. To reduce reverse causation from occult malignancy and to focus on longer-term incidence, the analytic window for counting incident cases began 365 days after the index prescription and extended up to 7,300 days (approximately 20 years).
Hazard ratios (HRs) with 95% confidence intervals (CIs) were estimated using Cox proportional hazards regression in each propensity score–matched cohort. Absolute risks and absolute risk differences (ARDs) for pancreatic cancer in matched groups were reported in the abstract for several comparisons.
GLP-1 RA versus insulin: After propensity score matching, GLP-1 RA users had a lower hazard of pancreatic cancer compared with insulin users (HR 0.43, 95% CI 0.35–0.53). The abstract reports absolute risks as GLP-1 RA 0.15% versus insulin 0.11%, with an ARD of 0.04 percentage points.
GLP-1 RA versus metformin: Compared with GLP-1 RAs, metformin users had a higher hazard of pancreatic cancer (HR 1.39, 95% CI 1.16–1.66); the abstract reports an ARD of 0.04 percentage points favoring GLP-1 RAs.
GLP-1 RA versus sulfonylureas: Sulfonylurea users had a higher hazard relative to GLP-1 RA users (HR 1.37, 95% CI 1.13–1.65), with an ARD of 0.07 percentage points.
GLP-1 RA versus thiazolidinediones: Thiazolidinedione users showed a higher hazard compared with GLP-1 RA users (HR 1.30, 95% CI 1.001–1.678), with an ARD of 0.15 percentage points.
GLP-1 RA versus DPP-4 inhibitors: DPP-4 inhibitor users had a higher hazard than GLP-1 RA users (HR 1.31, 95% CI 1.06–1.61), with an ARD of 0.08 percentage points.
GLP-1 RA versus SGLT2 inhibitors: No statistically significant difference was observed between GLP-1 RA users and SGLT2 inhibitor users (HR 1.08, 95% CI 0.87–1.34).
In this large real-world, propensity score–matched analysis from the TriNetX network, the observed association between initiation of GLP-1 RAs and incident pancreatic cancer varied depending on the comparator antidiabetic agent. Specifically, GLP-1 RA use was associated with a lower hazard of pancreatic cancer relative to metformin, sulfonylureas, thiazolidinediones, and DPP-4 inhibitors; there was no significant difference versus SGLT2 inhibitors; and GLP-1 RA use was associated with a lower hazard relative to insulin users in the HR reported (note: the abstract phrased a lower hazard for GLP-1 RAs compared with insulin, with reported HR 0.43).
The absolute risks reported in the abstract were small (fractions of a percent) and absolute risk differences between groups were numerically small (reported in percentage points). These results indicate variation by comparator class rather than a simple uniformly increased or decreased pancreatic cancer risk associated with GLP-1 RA initiation.
The abstract provides key results and the matching strategy but does not report several details in the provided source text. The following items were not reported in the abstract text available here and therefore are not included in this summary: exact cohort sizes before and after matching for each comparator pair, event counts and person-time denominators used in HR estimation, details on censoring and competing risks handling, sensitivity or subgroup analyses, and potential sources of residual confounding beyond the 39 matched covariates. Also, the abstract does not provide granular information on individual GLP-1 RA agents, duration of therapy, dose, or adherence measures. Readers should consult the full article for the complete methods, full numeric results, and discussed limitations.