SGLT2 inhibitors reduced composite CVD risk by 18% versus DPP4 inhibitors in diabetics without CVDs, with significant efficacy variation across patient clusters.
Do SGLT2 inhibitors reduce the composite of incident myocardial infarction, stroke, and heart failure in individuals with diabetes without established CVDs compared to DPP4 inhibitors?
SGLT2 inhibitors are associated with a lower risk of incident cardiovascular events compared to DPP4 inhibitors in primary prevention for diabetes, with machine learning identifying specific subgroups (e.g., females, insulin users) who derive greater benefit.
Abstract Background The cardioprotective effects of sodium-glucose cotransporter-2 (SGLT2) inhibitors are well-documented. However, understanding of their efficacy in primary prevention of cardiovascular diseases (CVDs) remains limited, necessitating further investigation. Purpose This study aims to evaluate the effectiveness of SGLT2 inhibitors over dipeptidyl peptidase-4 (DPP4) inhibitors for individuals with diabetes who have no established CVDs. The focus is on the unbiased identification of heterogeneity in the individual drug efficacy using a machine learning approach. Methods The present study included individuals with SGLT2 inhibitors or DPP4 inhibitors newly prescribed for diabetes without established CVDs using a nationwide epidemiological cohort. A new-user, active comparator design was employed, and 1:2 propensity score matching was performed. The primary outcome was the composite of incident myocardial infarction, stroke and heart failure. To investigate the heterogeneity in the drug efficacy, dimension reduction of the individual background information was performed using principal component analysis and uniform manifold approximation and projection (PCA-UMAP). Unsupervised clustering was further performed using hierarchical clustering. The efficacy of SGLT2 inhibitors over DPP4 inhibitors was evaluated and compared across the identified clusters. Results We analyzed 5,196 individuals with SGLT2 inhibitor prescription and 10,392 matched individuals with DPP4 inhibitor prescription. Overall, individuals with SGLT2 inhibitors demonstrated significantly lower risk of developing the composite endpoint events compared to individuals with DPP4 inhibitors hazard ratio, 0.82; 95% confidence interval (CI) 0.72 to 0.93; p=0.002, primarily driven by heart failure events hazard ratio, 0.77; 95%CI 0.67 to 0.88; p0.001. Hierarchical clustering analysis identified 8 individual clusters, demonstrating significant interaction effects across the clusters on the efficacy of SGLT2 inhibitors over DPP4 inhibitors for the composite endpoints (p for interaction=0.033) and heart failure (p for interaction=0.035). The clusters with higher efficacy of SGLT2 inhibitors were characterized with female sex, use of insulin or sulfonylureas, and higher prevalence of diabetes related complications. Conclusion This study underscores the potential of SGLT2 inhibitors in primary prevention of CVDs for individuals with diabetes. The application of machine learning to real-world data revealed the heterogeneity in individual drug efficacy, highlighting the importance of patient selections for effective usage of SGLT2 inhibitors in the prevention of CVDs.Graphical abstract
Jimba et al. (2025) studied this question. SGLT2 inhibitors reduced composite CVD risk by 18% versus DPP4 inhibitors in diabetics without CVDs, with significant efficacy variation across patient clusters.