AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
GLP-1 Receptor Agonists and Arterial and Venous Thromboembolism in Patients with Type 2 Diabetes and Cancer: A Propensity-Matched Cohort Study
By Saraiya A, Pancholy M, Alswayed K et al.
Introduction: Cancer-associated thrombosis is a major cause of morbidity in oncology, and a growing number of patients live longer with active cancer and type 2 diabetes (T2DM), which compounds thrombotic risk. GLP-1 Receptor Agonists (GLP-1 RA) reduce major adverse cardiac events in patients with T2DM, but a prior meta-analysis raised concern for increased risk of deep vein thrombosis (DVT) compared to placebo. Their effect on thromboembolic events in patients with T2DM and cancer is not well known. Research Question: Among patients with T2DM and cancer, are GLP-1 RAs associated with lower risk of arterial and venous thromboembolic events than dipeptidyl peptidase-4 inhibitors (DPP-4i)? Methods: In a new user, active comparator cohort study using TriNetX, a health research network capturing data from 173 international healthcare organizations, we evaluated adults who were recently diagnosed with T2DM and cancer, and had initiated GLP-1 RA versus DPP-4i after 2016 with no prior thromboembolic event. 1:1 propensity-score matching balanced 28 demographic, comorbidity, cancer-site, treatment, and laboratory variables. Acute myocardial infarction (MI), ischemic stroke, DVT, and PE were assessed over 730 days using Kaplan-Meier estimates, log-rank tests, Cox hazard ratios (HR), and tests of the proportional-hazards (PH) assumption. Results: Of 14,325 GLP-1 RA and 13,949 DPP-4i new users entering matching, 7,275 per arm were propensity matched (median age 68; 51% male; all standardized differences <0.1 except residual higher adiposity in the GLP-1 RA arm). Over 2 years, GLP-1 RA was associated with lower risk of MI (142 vs 212 events; HR 0.72, 95% CI 0.58-0.89) and PE (58 vs 91; HR 0.68, 0.49-0.95), with no difference in ischemic stroke (HR 0.86, 0.68-1.07). DVT events were fewer with GLP-1 RA (74 vs 109; log-rank p=0.03), but this association violated the PH assumption, so its time-averaged HR (0.72, 0.54-0.97) is interpreted cautiously. PH held for all other outcomes. The residual adiposity imbalance would bias towards greater thrombosis in the GLP-1 RA arm; suggesting the observed prediction is conservative. Conclusion: In this large, real-world cohort of patients with T2DM and cancer, GLP-1 RAs were associated with lower risk of MI and PE when compared to DPP-4i medications with no difference in ischemic stroke. The DVT signal favored GLP-1 RA but was time varying. These findings counter prior concern for GLP-1 RA-associated venous thrombosis and support prospective evaluation of GLP-1 RA for thrombotic risk reduction in patients with cancer and diabetes.