Does the eGFR dip explain heart benefit?

A retrospective cohort published in Expert Review of Cardiovascular Therapeutics on 31 July 2026 asks a narrow mechanistic question. When patients take the SGLT2 inhibitor dapagliflozin alongside semaglutide, does the early change in kidney filtration explain what happens to their hearts? Across 376 adults treated at a Saudi tertiary hospital between 2020 and 2024, the answer was no. Early change in estimated glomerular filtration rate (eGFR, the standard estimate of how much blood the kidneys filter per minute) showed what the authors call minimal mediation, with an indirect risk ratio of 1.013 and a 95% confidence interval of 0.901 to 1.165^[1].

What a mediation analysis is trying to do

Mediation analysis splits a treatment effect into two paths. The indirect path runs through a proposed intermediate variable: drug changes the kidney, changed kidney changes the heart. The direct path is everything else the drug does. If the indirect path carries most of the effect, the intermediate variable is doing the work and becomes a candidate marker to watch in clinic. If the indirect path is close to nothing, as here, the proposed mechanism is not where the action is. The authors ran theirs on top of inverse probability of treatment weighting, a method that reweights patients so the treated and untreated groups look more alike on measured characteristics^[1].

One caveat governs everything that follows. This is an observational cohort, not a randomised trial. Causal mediation analysis borrows the vocabulary of causation, but the arithmetic only holds if the model has accounted for every confounder that matters, and nobody can verify that in a records-based dataset. The title of the paper says hypothesis-generating, and that is the correct reading.

The numbers

The headline comparison was combination therapy against dapagliflozin alone, measured on recorded heart failure and myocardial infarction clinical status over 90 to 365 days of follow-up. Combination therapy was not associated with a statistically significant difference: adjusted risk ratio 0.787, with a 95% bootstrap confidence interval of 0.320 to 1.813 and p equal to 0.516^[1]. Propensity score analyses produced comparable neutral results. The mediation figure quoted above sits on top of that neutral primary result.

Why the kidney was the obvious suspect

Both drugs have a kidney story, which is why anyone would think to test this. SGLT2 inhibitors produce a well-described acute dip in eGFR in the first weeks of treatment. In a prespecified analysis of the DAPA-CKD trial published in the Journal of the American Society of Nephrology, 49.4% of patients on dapagliflozin had an acute eGFR reduction greater than 10% at two weeks, against 23.7% on placebo. The patients who dipped then declined more slowly afterwards, roughly 1.58 mL/min per 1.73 m2 per year against 2.44 to 2.48 in those with little or no initial reduction, and the dip was not associated with higher rates of kidney disease progression^[2]. The dip looks alarming on a lab report and turns out to be a haemodynamic adjustment rather than injury.

Semaglutide has its own kidney evidence. The FLOW trial, published in the New England Journal of Medicine in 2024, tested semaglutide in type 2 diabetes with chronic kidney disease and reported a 24% reduction in the primary composite kidney outcome (hazard ratio 0.76, 95% CI 0.66 to 0.88, p equal to 0.0003), alongside a hazard ratio of 0.71 for cardiovascular death and 0.82 for major cardiovascular events^[3]. Two drugs, two kidney signals, both with cardiovascular benefit attached. A shared renal pathway is a reasonable thing to go looking for. This cohort went looking and did not find it. The wider evidence base for the drug, including the trials that carry the cardiovascular claims, is summarised on the semaglutide peptide page.

The problem with the question as asked

Mediation analysis decomposes an effect. If there is no effect to decompose, the exercise has limited room to say anything. The primary comparison here was null and the confidence interval ran from 0.320 to 1.813^[1], which spans a large benefit and a large harm. A null indirect effect on top of a null total effect is consistent with the eGFR change being irrelevant, and it is equally consistent with the study being too small to see anything at all. Readers should not take this as a demonstration that renal mechanisms do not matter.

Follow-up is the second constraint. Heart failure and myocardial infarction accumulate over years. A window of 90 to 365 days in 376 patients yields very few events, which is exactly what a confidence interval that wide is telling you. Prescriber selection is the third: nobody was randomised, and the clinicians who added semaglutide were choosing patients for reasons the records do not fully capture.

What to take from it

Two misreadings are worth heading off. The first is treating this as evidence that adding semaglutide to dapagliflozin does nothing for the heart. It is not designed to answer that, and the randomised evidence for semaglutide points the other way in the populations where it has been tested. The second is treating the early eGFR dip as a warning sign in its own right. The DAPA-CKD analysis suggests the opposite^[2].

What the paper does add is a small piece of real-world evidence from a region under-represented in cardiometabolic trials, and a reminder that plausible mechanisms need testing rather than assuming. The same lead author reported a related Saudi analysis of BMI response across obesity classes on the same drug pairing^[4], and the randomised test of the combination in a different population is running as the TTT1 trial in type 1 diabetes. The open questions on what we still do not know about semaglutide sit alongside this one.

This article is educational and is not medical advice. Semaglutide and dapagliflozin are prescription medicines in every jurisdiction we cover, and the regulatory status of semaglutide is on its peptide page. Decisions about starting, stopping or combining either drug belong with a qualified healthcare professional who knows your history.