Cardiovascular outcomes of SGLT2 inhibitors versus GLP-1 receptor agonists in patients with type 2 diabetes and a history of myocardial infarction: a systematic review and network meta-analysis of randomized controlled trials
In brief
SGLT2 inhibitors and GLP-1 agonists cut major cardiac events similarly in post-MI diabetics
In a network meta-analysis of randomized trials, both drug classes lowered major adverse cardiovascular events by roughly 13% compared with placebo, with no statistically significant difference between them. SGLT2 inhibitors hinted at a slightly larger reduction in heart-failure hospitalizations, but the evidence was not robust, especially for cardiovascular death. Individualized therapy remains advised pending direct head-to-head trials.
- Journal
- Frontiers in endocrinology (Q1)
- Published
- 29 July 2026
- Study design
- Systematic review / meta-analysis of RCTs
- Evidence level
- Level 1, High (CEBM 1a)
- Authors
- Yanxia Ren, Hui Yang, Shuang Zhang, Tianxiao Hu, Qingying Tan, Jing Wang, et al.
- PMID
- 42591121
- DOI
- 10.3389/fendo.2026.1902802
Why clinicians should know about it
- Picked for Cardiology and Cardiovascular Medicine (top studies of the week, 16 August 2026): Network meta-analysis of SGLT2i vs GLP-1RA post‑MI
Abstract
BACKGROUND: Patients with type 2 diabetes (T2DM) and a history of myocardial infarction (MI) face a high risk of recurrent cardiovascular events. Sodium-glucose cotransporter-2 inhibitors (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP-1 RAs) reduce cardiovascular risk, but no head-to-head trials exist. We performed a systematic review and network meta-analysis (NMA) to indirectly compare their cardiovascular efficacy in this high-risk population. METHODS: Randomized controlled trials (RCTs) published up to February 1, 2026, were systematically searched and evaluated. The primary outcome was major adverse cardiovascular events (MACE); secondary outcomes included hospitalization for heart failure (HFH), MI, and cardiovascular death (CV death). A frequentist NMA was used as the primary comparative analysis, complemented by Bucher interaction tests as sensitivity analyses. Evidence certainty was assessed with GRADE. RESULTS: Eleven RCTs were included; after exclusion of overlapping populations, the final analyses comprised 7 studies for MACE (2 SGLT2i, 5 GLP-1 RA), 6 for HFH (2 SGLT2i, 4 GLP-1 RA), 4 for MI (1 SGLT2i, 3 GLP-1 RA), and 6 for CV death (2 SGLT2i, 4 GLP-1 RA). Both drug classes significantly reduced MACE compared to placebo (SGLT2i HR 0.87, 95% CI 0.77-0.97; GLP-1 RA HR 0.83, 95% CI 0.77-0.89). The NMA indirect comparison for MACE yielded an HR of 1.05 (95% CI 0.92-1.20, p = 0.44), indicating no statistically significant difference between classes. For HFH, SGLT2i exhibited a numerically greater reduction (indirect HR 0.83, 95% CI 0.67-1.03, p = 0.09) that did not reach conventional significance. MI risk was similarly reduced by both classes (indirect HR 1.05, 95% CI 0.83-1.33, p = 0.68). For CV death, the indirect comparison showed no statistically significant difference between classes (HR 0.94, 95% CI 0.76-1.16, p = 0.57), but this estimate is based on limited SGLT2i data (only two trials). The GRADE certainty was low for CV death, primarily due to indirectness (predominantly remote MI populations) and imprecision (wide confidence intervals crossing the line of no effect). CONCLUSIONS: In patients with T2DM and a history of MI, indirect evidence suggests that both SGLT2i and GLP-1 RAs reduce MACE, HFH, and recurrent MI, with no statistically significant difference between classes. SGLT2i show a numerically larger HFH reduction, but the difference is not statistically robust. Evidence is largely derived from remote MI, limiting its applicability to the acute post-infarction setting. Treatment choice should be individualized based on patient risk profiles. Dedicated head-to-head trials are urgently needed. SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261299756, identifier CRD420261299756.
Abstract as published, via PubMed.
For healthcare professionals. The summary is generated by AI from the published abstract, and the evidence level is assigned automatically from the study design on the Oxford CEBM hierarchy. Neither is medical advice. Read the full paper before changing practice.