Dapagliflozin and ventricular repolarization across heart failure phenotypes

Type 2 diabetes mellitus (T2DM) is associated with abnormal ventricular repolarization and increased arrhythmic risk. Sodium-glucose cotransporter-2 (SGLT-2) inhibitors may exert electrophysiological effects beyond glycemic control. We investigated changes in electrocardiographic (ECG) markers of ventricular repolarization after dapagliflozin initiation in SGLT-2 inhibitor-naïve patients with T2DM according to heart failure (HF) phenotype.

In this prospective, non-randomized, controlled observational study, 700 patients with T2DM were enrolled between June 2023 and February 2026. Patients initiating dapagliflozin (n = 350) were classified as HFrEF (LVEF <50%, n = 87), HFpEF (LVEF ≥50% with HF, n = 93), or no HF (LVEF ≥50% without HF, n = 170); 350 controls continued pre-existing antidiabetic therapy. Twelve-lead ECG and echocardiography were performed at baseline and 3 months. Prespecified endpoints included QT, QTc, QT dispersion, Tp-e, Tp-e/QT, and Tp-e/QTc. Inverse probability of treatment weighting and Holm-Bonferroni correction were applied.

Dapagliflozin was associated with greater reductions in all prespecified ECG parameters than controls (all between-group p <0.001; Cohen's d = 0.48-1.08), with consistent findings after IPTW. Reductions were greatest in HFrEF (QTc = -14.3 ms; Tp-e = -9.1 ms), followed by HFpEF (-11.2 and -7.4 ms) and no HF (-7.6 and -3.8 ms); all between-subgroup comparisons p <0.001. HFrEF independently predicted greater QTc reduction (β = -4.8 ms vs no HF, p < 0.001). Change in Tp-e correlated weakly with HbA1c change (r = -0.145, p = 0.037). LVEF increased in HFrEF (+3.2%) and HFpEF (+0.9%).

Dapagliflozin initiation was associated with reductions in multiple ECG markers of ventricular repolarization, with the greatest changes observed in HFrEF. These findings concern surrogate ECG measures and do not establish a reduction in clinical arrhythmic events. Confirmation in prospective randomized trials is warranted.