Hemodynamic changes in response to SGLT2 inhibitor treatment in ICD and CRT patients: insights from HeartLogic sensor data

Background Sodium-glucose cotransporter-2 inhibitors (SGLT2i) reduce hospitalizations in patients with heart failure (HF) hospitalizations. Cardiovascular Implantable Electronic Devices (CIED) can provide daily collected sensor data (e.g., thoracic impedance, heart sounds, and heart rate), allowing for high-resolution physiological changes and mechanistic understanding following SGLT2i initiation. Methods In a nationwide registry-based cohort (2018 - 2024), patients initiated on SGLT2i with HeartLogic compatible CIED were matched 1:3 to non-initiators, matched by sex, age, device type, baseline HF status, recent HF admission and diabetes status. Daily sensor data were analysed from 30 days pre-initiation to 90 days post-initiation. Between-group differences in weekly trajectories of HeartLogic index, night-time heart rate, S3 heart sound, and thoracic impedance, were assessed using generalized linear mixed-effects models. HeartLogic data were not available to patients or treating clinicians. HF-related hospitalizations were analyzed using an age-adjusted self-controlled case series (SCCS) design within an observation window spanning from 180 days before to 180 days after SGLT2i initiation. Subgroup analyses included baseline HF status and no concurrent changes in HF medications around the time of SGLT2i initiation. Results We included 363 SGLT2i initiators and 1039 matched controls (mean age 60 12 years, 79% male, 82% ICD, ejection fraction 39 14 vs 31 9%, table 1). In the pre-initiation period, initiators had higher S3 amplitude, higher night-time heart rate, and lower thoracic impedance than controls (e.g., week -3: S3 0.05 mG [95% CI 0.02 to 0.08], p<0.01; thoracic impedance -1.62 Ω [95% CI -2.91 to -0.33], p<0.01; night HR 1.5 bpm [0.15 to 2.79], p=0.02). After SGLT2i initiation, thoracic impedance increased significantly compared with controls (week 12 mean difference 1.73 Ω [0.24 to 3.04], p<0.01; figure 1). No other sensor-derived changes differed significantly. In SCCS analyses, HF hospitalization rates were elevated in the months prior to SGLT2i initiation (-180 to -60 days: IRR 5.17 [2.37 to 11.28], p<0.01). Rates were not increased after initiation (0 to 30 days: IRR 1.19 [0.29 to 4.93], p=0.82) and tended to be lower from 30 to 180 days (IRR 0.49 [0.15 to 1.61], p=0.24). Findings were consistent across prespecified subgroups. Conclusion In ICD/CRT-D patients initiating SGLT2i, thoracic impedance increased progressively relative to matched controls, despite a higher-risk HF profile at baseline. Pre-initiation elevations in S3 amplitude and night-time heart rate attenuated relative to matched controls.