The journey of sodium-glucose cotransporter-2 (SGLT2) inhibitors from glucose-lowering agents to a cornerstone of heart failure (HF) therapy was one of the more striking paradigm shifts in contemporary cardiovascular medicine. Developed initially to treat type 2 diabetes, these agents have consistently reduced HF hospitalization, cardiovascular death, and renal disease progression — benefits that extend well beyond patients with diabetes.1–5

In this issue of the Portuguese Journal of Cardiology, Abd El Hady et al.6 add data to this story by examining an often-overlooked corner of it: the effects of SGLT2 inhibitors on blood glucose in patients who do not have diabetes. In a randomized comparison of dapagliflozin and empagliflozin in 130 patients with HF but no diabetes and reduced ejection fraction (HFrEF), the authors report small but statistically significant reductions in several glycemic parameters — more consistently with empagliflozin than with dapagliflozin — with no hypoglycemia, no ketoacidosis, and with parallel improvements in left ventricular dimensions and ejection fraction over three months.

The headline finding is reassuring rather than surprising. Mean HbA1c fell from approximately 5.3% to 5.0% with empagliflozin and shifted little with dapagliflozin — changes that are statistically real but clinically trivial, occurring in patients who began the study essentially normoglycemic. This fits a concept established over the last decade: SGLT2 inhibitors do not force glucose down through pharmacological pressure but through a self-limiting mechanism tied to the filtered glucose load, so their glucose-lowering effect fades as blood glucose approaches the normal range.7 Hypoglycemia in patients without concomitant insulin or sulfonylurea therapy therefore remains vanishingly rare — the safety profile that this study confirms once again.

The more interesting question the study raises is not whether glucose falls, but whether it matters. The totality of evidence from DAPA-HF, EMPEROR-Reduced, DELIVER, and EMPEROR-Preserved tells us, with increasing confidence, that it largely does not matter.1–4 Patients with and without diabetes derive remarkably similar reductions in cardiovascular death and HF hospitalization, and two large, pooled analyses — one restricted to HFrEF8 and one spanning the full ejection-fraction spectrum9 — found no meaningful treatment-by-diabetes-status interaction, a result corroborated by dedicated subgroup analyses from the individual trials.10 Glycemic status, in other words, is not the variable that predicts who benefits.

Attention has accordingly shifted to mechanisms that have nothing to do with glycemic control. Contemporary reviews describe a multifactorial model of cardiorenal protection built on osmotic diuresis and natriuresis, favorable loading conditions, improved myocardial substrate utilization and mitochondrial efficiency, reduced inflammation and oxidative stress, and direct modulation of cardiomyocyte sodium and calcium handling through inhibition of the sodium-hydrogen exchanger.7,11 None of these pathways requires hyperglycemia to operate, which is precisely why the benefit is preserved — and, in several outcome measures, of similar magnitude — in people who did not have diabetes.

Read against this background, the findings of Abd El Hady et al.6 are consistent with a much larger and more consistent evidence base: benefit across the ejection-fraction spectrum, largely irrespective of baseline glucose metabolism.8,9 HF itself, rather than diabetes, is the target these drugs are treating.

One observation in the present study deserves more attention than it typically receives: the rise in serum magnesium seen with both agents. This finding is not confined to a single small trial. A pre-specified analysis of EMPEROR-Reduced, published in the past few months, found that low baseline serum magnesium was itself associated with worse HF and renal outcomes, that empagliflozin reliably raised serum magnesium within four weeks and lowered the odds of frank hypomagnesaemia, and that the drug's benefit on HF events was numerically more pronounced in patients who started with low magnesium levels.12 Magnesium is central to myocardial electrophysiology, vascular tone, and arrhythmia prevention; whether its correction is a bystander effect of SGLT2 inhibition or an active contributor to the mortality and arrhythmia benefit seen in the major trials remains unresolved, but it is no longer an implausible hypothesis. The parallel, more modest rise in phosphorus with empagliflozin in the present study likewise merits mechanistic follow-up, given its relevance to vascular calcification and cellular energetics.

Several limitations temper how far these results can be extrapolated. The trial was open-label, lacked a placebo arm, and followed patients for only three months in a relatively small, single-country cohort — a design that can establish safety and describe metabolic trends but cannot, on its own, establish that those trends translate into fewer hospitalizations or deaths. Because the cardiovascular and renal benefits of SGLT2 inhibitors in HFrEF are already established beyond reasonable doubt, the more useful next step for this line of research is not another safety study but a mechanistic study: biomarkers tracking the magnitude of magnesium and phosphate change, imaging correlates of the reverse remodeling already visible at three months, and predictors of which patients derive the largest cardiorenal benefit independent of glycemic status.

None of this changes the clinical message that already exists: SGLT2 inhibitors are one of the four foundational pillars of guideline-directed therapy for HFrEF, to be started early and, wherever possible, simultaneously with the other three.13 The STRONG-HF trial showed that rapid, protocolized up-titration of these therapies after an HF admission reduces death and readmission at 180 days.14 The practical challenge today is not identifying which patients benefit from SGLT2 inhibitors; rather, it is closing the persistent gap between patients who are eligible for the four pillars and those who are prescribed them.

Ultimately, this study adds a useful, if modest, brick to the safety case for SGLT2 inhibitors in non-diabetic HFrEF. Its more lasting contribution may be the reminder that the field has already moved past the question it set out to answer. The central issue is no longer whether these drugs lower glucose in people without diabetes — they do, trivially and safely — but how they protect the heart and kidney so consistently across metabolic phenotypes.5 The clearest current answer is that they do so mostly without help from glucose.