Class effect beyond glucose: Implications of the LEGEND‐T2DM findings for cardiovascular outcomes Authors: Yuexing Liu, Weiping Jia Published: 2026 The management of type 2 diabetes mellitus has entered an era in which the choice of pharmacologic approaches to antidiabetic therapy is inseparable from cardiovascular and kidney risks reduction. Some large randomized clinical trials have established sodium‐glucose cotransporter‐2 inhibitors (SGLT2is) and glucagon‐like peptide‐1 receptor agonists (GLP‐1 RAs) as therapies that extend far beyond glycemic control. These trials have found that SGLT2is and GLP‐1 RAs reduced major adverse cardiovascular events (MACEs) and renal outcomes 1 . However, comparable studies with older antihyperglycemic agents on cardiovascular efficacy or safety are lacking. So far, no direct head‐to‐head comparisons of SGLT2is and GLP‐1 RAs with sulfonylureas (SUs) or dipeptidyl peptidase‐4 inhibitors (DPP4is) have been reported before. In addition, the comparative cardiovascular efficacy of SGLT2is and GLP‐1 RAs has not been evaluated yet. To address this important evidence gap, Khera et al. 2 reported findings from a Large‐scale Evidence Generation and Evaluation across a Network of Databases for type 2 diabetes mellitus (LEGEND‐T2DM) in J Am Coll Cardiol in 2024, which represented a real‐world comparative study of these antidiabetic agents. Their study offers clinically relevant evidence, answering the question of how clinicians should choose among newer and older second‐line agents in those with type 2 diabetes mellitus and cardiovascular disease (CVD) on metformin monotherapy. In this large‐scale real‐world study 2 across six administrative claims and four electronic health record databases from four countries, including the United States, Germany, Spain and the United Kingdom, nearly 1.5 million patients with type 2 diabetes mellitus and established CVD who escalated from metformin monotherapy to a second‐line agent were included in an active‐comparator and target trial emulation for pairwise comparisons. The main findings summarized in the Table 1 were straightforward. Both SGLT2is and GLP‐1 RAs were associated with an 11% to 28% lower risk of 3‐point MACE (myocardial infarction, stroke, and sudden cardiac death), compared with DPP4is and SUs, while DPP4is was associated with a lower 3‐point MACE risk vs SUs (hazard ratio [HR] 0.87 [95% CI, 0.79–0.95]). Similar patterns were found in terms of 4‐point MACE (3‐point MACE + heart failure hospitalization). In addition, no significant differences were found between SGLT2is and GLP‐1 RAs in terms of 3‐point or 4‐point MACE. Thus, this study found comparable cardiovascular risk reduction in terms of SGLT2is and GLP‐1 RAs, with both agents more effective than DPP4is and SUs, adding further evidence regarding the use of SGLT2is and GLP‐1 RAs being prioritized as second‐line agents in those with type 2 diabetes mellitus and established CVD. The major strength of this study lies in its rigorous methods. Rather than relying on indirect estimates from clinical trials or single‐database analyses, the authors applied a federated analysis across multiple health systems using standardized data structures, data‐driven and large‐scale propensity score adjustment, and negative‐control calibration to address unmeasured confounding and systematic bias, with multiple ‘leave‐one‐out’ sensitivity analyses confirming the findings of the primary analysis. Moreover, the federated framework improves generalizability and reduces the possibilities of overinterpreting findings from one single population with potential biases due to a specific healthcare system. In an era full of real‐world evidence, the LEGEND‐T2DM study stands out for methodological discipline rather than large data scale alone 2 . Table 1Comparative hazard ratio estimates and 95% CI for 3‐point MACE and 4‐point MACE among new users of SGLT2is, GLP‐1 RAs, DPP4is, and SUs from the LEGEND‐T2DM studySGLT2isGLP‐1 RAsDPP4isvs SUsvs DPP4isvs GLP‐1 RAsvs SUsvs DPP4isvs SUs3‐point MACE (HR, 95% CI)0.76, 0.65–0.890.89, 0.79–1.001.06, 0.96–1.170.72, 0.58–0.880.83, 0.70–0.980.87, 0.79–0.954‐point MACE (HR, 95% CI)0.71, 0.61–0.830.84, 0.75–0.951.05, 0.97–1.130.67, 0.55–0.820.81, 0.73–0.910.87, 0.79–0.95Cited from Khera et al 2 .3‐point MACE, myocardial infarction, stroke, and sudden cardiac death; 4‐point MACE, 3‐point MACE plus heart failure hospitalization; CI, confidence interval; DPP4is, dipeptidyl peptidase‐4 inhibitors; GLP‐1 RAs, glucagon‐like peptide‐1 receptor agonists; HR, hazard ratio; MACE, major adverse cardiovascular event; SGLT2is, sodium‐glucose cotransporter‐2 inhibitors; SUs, sulfonylureas. Comparative hazard ratio estimates and 95% CI for 3‐point MACE and 4‐point MACE among new users of SGLT2is, GLP‐1 RAs, DPP4is, and SUs from the LEGEND‐T2DM study Cited from Khera et al 2 . 3‐point MACE, myocardial infarction, stroke, and sudden cardiac death; 4‐point MACE, 3‐point MACE plus heart failure hospitalization; CI, confidence interval; DPP4is, dipeptidyl peptidase‐4 inhibitors; GLP‐1 RAs, glucagon‐like peptide‐1 receptor agonists; HR, hazard ratio; MACE, major adverse cardiovascular event; SGLT2is, sodium‐glucose cotransporter‐2 inhibitors; SUs, sulfonylureas. Nevertheless, the null comparison between SGLT2is and GLP‐1 RAs in this study should be interpreted with caution. Absence of a statistically significant difference in composite MACEs does not imply therapeutic interchangeability. Prior trial‐level evidence has suggested that these agents express benefit through different clinical pathways. In a large network meta‐analysis 1 , both classes reduced all‐cause mortality, cardiovascular mortality, non‐fatal myocardial infarction, and kidney failure, but SGLT2is showed a greater reduction in heart failure hospitalization, whereas GLP‐1 RAs appeared more favorable for non‐fatal stroke. The LEGEND‐T2DM echoes a bit different pattern between these two agents on individual MACE components. Although overall composite outcomes were similar, SGLT2is was associated with a higher hazard of acute myocardial infarction than GLP‐1 RAs (HR 1.19 [95% CI, 1.05–1.35]) and no significant difference between SGLT2is and GLP‐1 RAs was found regarding the hazards of stroke, sudden cardiac death, and heart failure hospitalization 2 . These findings highlight that composite endpoints can mask biologically and clinically meaningful differences between drug classes. Moreover, the divergence between the LEGEND‐T2DM study and the prior meta‐analysis of the randomized controlled trials of SGLT2is and GLP‐1 RAs may partly reflect the inherent differences between real‐world observational and trial‐derived estimates. Emerging clinical evidence continues to broaden our understanding of the cardiovascular effects of GLP‐1 RAs. Together with the findings from the LEGEND‐T2DM study, these data suggest that GLP‐1 RAs may have a broader role in heart failure management, supporting the continued evaluation for reducing heart failure risk and improving outcomes in patients with heart failure 2 . Yet, the study also has limitations that matter not only as technical caveats but for clinical interpretation. First, it is a class‐level analysis, which risks overgeneralization. Agents within each class differ in molecular structure, route of administration, potency, and safety, etc. A class‐level estimate may obscure clinically meaningful heterogeneity among individual drugs. Thus, there is an unmet need for further evidence generation on drug‐level comparative effectiveness, which was suggested as the goals of the LEGEND‐T2DM study group's future work. Second, the study uses an on‐treatment design, censoring follow‐up at treatment discontinuation. That approach may better isolate direct treatment exposure, but it may not align with the treatment questions clinicians often face in practice and limit inferences on long‐term outcomes. Third, combination and sequencing strategies of different agents are increasingly needed. Future research should move beyond class rivalry toward combination therapy. Equally important, comparative studies should include tolerability, adherence, and affordability, because therapies with proven efficacy can still fail at the population level when they are difficult to access or sustain. Last but not least, further evidence from Asian populations is needed to clarify whether the comparative cardiovascular effectiveness observed in this predominantly Western, real‐world cohort is generalizable to populations with distinct genetic backgrounds, metabolic phenotypes, and healthcare systems. This study primarily focused on comparative cardiovascular efficacy among antidiabetic agents, whereas medication safety received relatively limited discussion. The safety of SGLT2is and GLP‐1 RAs should also be considered when applying the findings from the LEGEND‐T2DM study in clinical practice. GLP‐1 RAs use slows intestinal motility and increases the risk of biliary disease 3 . In addition, GLP‐1 RAs induces clinically significant loss of lean and skeletal muscle mass with fat mass during weight loss 3 . This concern is particularly relevant in older adults with type 2 diabetes mellitus and established CVD, who are most likely to receive second‐line antidiabetic strategies for cardiovascular risk reduction. In these patients, increased muscle loss may contribute to sarcopenia, frailty, and fragility fracture 3 . Finally, the psychiatric safety of GLP‐1 RAs remains an area of active surveillance 3 . Additionally, SGLT2is use was associated with increased risks of genitourinary infections and diabetic ketoacidosis 4 . Routine monitoring of ketone levels and urinalysis is advisable to enable timely identification of treatment‐related adverse events. Notably, emerging evidence suggests that the clinical significance of ketosis may vary under different status. Under conditions of carbohydrate restriction, reductions in fasting plasma glucose have been shown to correlate inversely with elevations in blood ketone concentrations 5 . Furthermore, continuous glucose monitoring and continuous ketone monitoring have revealed a pronounced inverse relationship between glucose and ketone trajectories 5 . These observations suggest that, in the absence of overt ketoacidosis, mild‐to‐moderate increase of ketosis may play a physiological role in glucose homeostasis regulation 5 . Accordingly, precise characterization of ketosis severity should inform clinical decision‐making regarding SGLT2is use. Collectively, these findings highlight that the cardiovascular and metabolic benefits of SGLT2is and GLP‐1 RAs demonstrated in the LEGEND‐T2DM study should be interpreted with their safety considerations, particularly in the elderly and comorbid population most likely to receive these agents in clinical practice. Taken together, the LEGEND‐T2DM study provides strong real‐world evidence supporting the preferential use of SGLT2is and GLP‐1 RAs as second‐line therapies in patients with type 2 diabetes mellitus and established CVD. These findings further support the current treatment paradigm that second‐line therapy in patients with type 2 diabetes mellitus and established CVD should extend beyond glucose lowering alone. Importantly, the findings across multiple healthcare systems enhance their relevance to clinical practice. More importantly, this study highlights the value of high‐quality real‐world evidence with rigorous methodology in addressing clinically relevant questions that are unlikely to be answered by randomized clinical trials. As therapeutic options expand, future comparative effectiveness research should move beyond comparisons between drug classes to better determine which patients benefit most from specific treatment strategies. DISCLOSURE The authors declare no conflict of interest. Approval of the research protocol: N/A. Informed consent: N/A. Registry and the registration no. of the study/trial: N/A. Animal studies: N/A.