AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
Preoperative Sodium Glucose Cotransporter Inhibitor Exposure and Outcomes After Isolated Coronary Artery Bypass Grafting in Type 2 Diabetes
By Mitar L, Gregory A, Bakaeen F et al.
Introduction/Background: Sodium-glucose cotransporter-2 inhibitors (SGLT2i) reduce cardiovascular mortality in type 2 diabetes mellitus (T2DM), but concerns about perioperative euglycemic diabetic ketoacidosis (DKA) prompted guideline recommendations to hold these agents before surgery. Their association with outcomes after isolated coronary artery bypass grafting (CABG) remains incompletely defined. Research Questions/Hypothesis: We evaluated whether SGLT2i exposure within 90 days before isolated CABG was associated with 5-year mortality, perioperative complications compared with matched T2DM patients without SGLT2i exposure. Methods/Approach: We performed a multicenter retrospective TriNetX cohort study of adults with T2DM undergoing isolated CABG. Patients with SGLT2i exposure within 90 days before surgery were compared with no SGLT2i controls after 1:1 propensity score matching. The primary outcome was all-cause mortality through 5 years. Secondary outcomes included 30-day mortality, acute kidney injury, atrial fibrillation, stroke or transient ischemic attack, pneumonia, respiratory failure, sepsis, perioperative DKA, a 30-day composite outcome (all listed 30-day outcomes except DKA and the late coronary ischemic event endpoint), and a late coronary ischemic event endpoint (repeat CABG, PCI, myocardial infarction, or angina). Results/Data: Matched cohorts included 2,170 SGLT2i-exposed and 2,170 no SGLT2i patients. Five-year mortality was lower in the SGLT2i cohort (139/2,160 [6.43%] vs 279/2,157 [12.94%]; RD –6.50%, 95% CI –8.25% to –4.75%; RR 0.498, 95% CI 0.410-0.604; HR 0.566, 95% CI 0.461-0.694; log-rank p<0.0001). Thirty-day mortality was numerically lower without conventional significance (1.20% vs 1.81%; OR 0.662; p=0.1030). The 30-day composite was similar by risk analysis (OR 0.892; p=0.0663) with a favorable Kaplan-Meier signal (HR 0.905; log-rank p=0.0396). Acute kidney injury, atrial fibrillation, stroke or TIA, pneumonia, respiratory failure, sepsis, and the late coronary ischemic event endpoint did not differ between groups. Perioperative DKA was rare, with fewer than 10 total events across the matched cohort lmiting comparative estimates according to database policy. Conclusion(s): Among matched T2DM patients undergoing isolated CABG, SGLT2i exposure within 90 days before surgery was associated with substantially lower 5-year mortality. Perioperative DKA was uncommon, with fewer than 10 total events across both matched groups, supporting that this safety signal was rare in the observed cohort while precluding reliable between-group inference. These findings support prospective evaluation of preoperative SGLT2i use in cardiac surgery.