Heart-rate variability is one of the cleaner non-invasive readouts of cardiac autonomic balance. High variability between beats means the parasympathetic side has a strong vote. Low variability means the sympathetic side is winning. GLP-1 receptor agonists raise resting heart rate by a few beats per minute and trim HRV indices, a pattern reproduced across the large cardiovascular outcomes trials and a long series of smaller mechanistic studies. The mechanism is not fully worked out. Vagal modulation, direct cardiac actions, and central sympathetic input all share blame.
A prospective observational study from a Chinese type-2-diabetes clinic, published online May 11 in Frontiers in Endocrinology ↗, asked whether sodium-glucose co-transporter 2 inhibitors (a class many of the same patients take for separate cardiovascular indications) can buffer that decline. Forty-five adults with type 2 diabetes were split by whether they had been taking dapagliflozin (an SGLT2 inhibitor) before the study started. Twenty-two had not. Twenty-three had. Both groups then initiated GLP-1 receptor agonist ↗ therapy and wore 24-hour ambulatory ECGs before and after twelve weeks of treatment.
The dapagliflozin-naive group lost autonomic function across the board. SDNN, the standard deviation of all normal RR intervals over 24 hours and the most-used global HRV index, fell. SDANN, the same statistic but averaged over five-minute bins to pick up slower modulation, fell. RMSSD and pNN50, which reflect short-term parasympathetic activity, fell. Log high-frequency power, the spectral readout of vagal tone, fell. Log low-frequency power and the LF/HF ratio, both biased toward sympathetic activity, rose. Every direction matched what the prior literature predicts when a GLP-1 receptor agonist is added to standard care.
The dapagliflozin group moved on none of them. After twelve weeks on the same drug class, the patients who had been on dapagliflozin before enrollment showed no statistically significant within-group shift in any of the HRV indices the study tracked.
The between-group analysis tightened the contrast. After covariate adjustment, four indices still separated significantly between groups. SDNN, SDANN, log high-frequency power, and the LF/HF ratio. Inverse probability of treatment weighting sensitivity analyses gave the same answer. In a multivariable regression model, baseline dapagliflozin use was the variable most clearly associated with a more favorable change in SDNN.
Forty-five patients is not large. The design was observational, not randomized. The dapagliflozin group was, by construction, the group whose physician had already decided they needed an SGLT2 inhibitor (usually for heart-failure protection, sometimes for kidney indications), and that selection effect cannot be fully scrubbed out by propensity weighting alone. The investigators acknowledge as much, calling the result hypothesis-generating and asking for larger prospective studies.
What the study does is make a specific testable claim. GLP-1 receptor agonists modestly bend the cardiac autonomic dial toward sympathetic dominance over the first three months of therapy. Dapagliflozin, on board first, blunts that bend across at least four indices, in a small cohort, with three different statistical approaches converging on the same answer. The mechanism candidates are not exotic. SGLT2 inhibitors lower sympathetic outflow in animal work, reduce noradrenaline spillover in human heart-failure studies, and shift cardiac substrate metabolism in ways that can improve mitochondrial efficiency in the autonomic ganglia. Any of those is biologically plausible.
The sequencing implication is what makes the result clinically interesting at this size. SGLT2 inhibitors and GLP-1 receptor agonists are both first-line additions to metformin in patients with diabetes and high cardiovascular risk, and the choice of which to start first is more often driven by insurance coverage and physician preference than by direct head-to-head data. If prior SGLT2 inhibitor exposure protects against GLP-1-induced autonomic decline, the order may matter more than the field has assumed.
The study did not record cardiovascular events, only HRV indices. HRV is a strong correlate of cardiovascular mortality in older adults with diabetes but is not itself a regulatory endpoint. The next step is whether a larger trial can connect the autonomic finding to specific arrhythmia signals or event rates, in a population where the two drug classes increasingly arrive together.