FLOW Trial: Semaglutide Kidney Subgroup Detail (2026)

The headline from FLOW (Perkovic et al., New England Journal of Medicine 2024) is well known: weekly semaglutide 1.0 mg cut the risk of major kidney events by 24% in people with type 2 diabetes and chronic kidney disease, and the trial was stopped early for benefit (HR 0.76, 95% CI 0.66-0.88)^[1]. That single number earned semaglutide the first-ever FDA kidney indication for a GLP-1 drug. But the primary paper was only the beginning — over 2024 and 2025 the FLOW investigators published a series of prespecified secondary and subgroup analyses that answer the questions clinicians actually ask: Does it slow the rate of kidney decline (eGFR slope)? Does it lower cardiovascular death and all-cause mortality? Does it prevent heart failure? And critically — does the benefit hold across the whole spectrum of kidney severity, and on top of SGLT2 inhibitors and mineralocorticoid receptor antagonists? This article is a deep dive into those follow-on analyses. For the trial design, FDA-approval story, and the SGLT2-trial comparison, start with our FLOW trial overview; this piece picks up where that one leaves off.

The honest summary

• Primary result (the anchor). The major-kidney-disease composite fell 24% (HR 0.76, 95% CI 0.66-0.88; P=0.0003) over a median 3.4 years in 3,533 participants — trial stopped early for efficacy (Perkovic 2024^[1]).
• eGFR slope slowed meaningfully. The mean annual decline in eGFR was 1.16 mL/min/1.73m² per year slower with semaglutide (95% CI 0.86-1.47; P<0.001) — a measure of how much the drug flattens the trajectory of kidney decline (Perkovic 2024^[1]).
• Cardiovascular death and all-cause death both dropped. Death from cardiovascular causes fell 29% and death from any cause fell 20% in the primary report (Perkovic 2024^[1]). A dedicated cardiovascular secondary analysis confirmed 3-point MACE HR 0.82 (95% CI 0.68-0.98) and all-cause mortality HR 0.80 (95% CI 0.67-0.95) (Mann 2025^[2]).
• Heart failure was reduced. The composite of heart-failure events or CV death fell 27% (HR 0.73, 95% CI 0.62-0.87; P=0.0005), with consistent benefit whether or not participants had heart failure at baseline (Mc Causland 2024^[3]).
• Kidney benefit was uniform across CKD severity. The kidney composite was reduced regardless of baseline eGFR or albuminuria stratum (p-interaction 0.83 for eGFR, 0.42 for UACR) — supporting use across the CKD spectrum including advanced disease (Mann 2026 CJASN^[4]; Mann 2025 EHJ^[2]).
• The SGLT2-inhibitor subgroup is the big caveat. Only ~15.6% were on an SGLT2 inhibitor at baseline; in that small group the kidney HR was 1.07 (not significant) vs 0.73 in non-users, p-interaction 0.109 — underpowered, not proof of no benefit (Apperloo 2024^[5]).
• Benefit persisted on top of an MRA. Among the ~7% on a mineralocorticoid receptor antagonist, the kidney HR was 0.51 vs 0.79 in non-users (p-interaction 0.12) — consistent direction (Rossing 2025^[6]).

First, the anchor: what the primary FLOW paper actually showed

FLOW (NCT03819153) randomized 3,533 participants with type 2 diabetes and chronic kidney disease (eGFR 50-75 with UACR >300, or eGFR 25 to <75 with UACR >100) to weekly semaglutide 1.0 mg or placebo. The primary composite was the first occurrence of major kidney disease events: kidney failure (dialysis, transplant, or sustained eGFR <15), a sustained ≥50% reduction in eGFR, or death from kidney or cardiovascular causes. After a median 3.4 years the independent data-monitoring committee recommended stopping early for efficacy. The primary outcome occurred in 331 semaglutide participants (5.8 per 100 patient-years) versus 410 on placebo (7.5 per 100 patient-years) — HR 0.76, 95% CI 0.66-0.88, P=0.0003 (Perkovic 2024^[1]). That is the number every secondary analysis below is built on. The remainder of this article is about the questions the primary paper raised but could not fully answer in its main results table.

eGFR slope: how much does it flatten the decline?

A binary composite tells you whether a hard event happened, but not how fast a kidney is failing. For that, nephrologists look at the eGFR slope — the rate of decline in filtration over time. In FLOW, the total eGFR slope was 1.16 mL/min/1.73m² per year slower in the semaglutide arm (95% CI 0.86-1.47; P<0.001) (Perkovic 2024^[1]). To put that in context, the SGLT2-inhibitor trials (CREDENCE, DAPA-CKD, EMPA-KIDNEY) produced chronic-slope benefits in a broadly similar range, which is part of why semaglutide is now discussed as a complementary — not redundant — kidney-protective drug. The slope finding also matters mechanistically: a difference that accrues year after year is what ultimately separates the event curves, and it is the kind of continuous endpoint regulators and guideline bodies weigh heavily.

Cardiovascular death and all-cause mortality

FLOW enrolled a high-cardiovascular-risk population, so the cardiovascular and mortality signals were closely watched. In the primary report, death from cardiovascular causes was reduced 29% and death from any cause was reduced 20% (Perkovic 2024^[1]). A dedicated cardiovascular secondary analysis (Mann et al., European Heart Journal 2025) reported the three-point major adverse cardiovascular event composite (CV death, nonfatal MI, nonfatal stroke) at HR 0.82 (95% CI 0.68-0.98; P=0.03) and all-cause mortality at HR 0.80 (95% CI 0.67-0.95; P=0.01)^[2]. These are clinically meaningful effect sizes for hard outcomes in a CKD population that is otherwise difficult to protect.

One nuance from that cardiovascular analysis is worth flagging honestly. When all-cause mortality was broken out by baseline albuminuria, the benefit was concentrated in higher-albuminuria participants (UACR ≥300 mg/g: HR ~0.70), while the lowest-albuminuria stratum showed a non-significant trend in the opposite direction (HR ~1.17), producing a p-interaction of 0.01 for mortality across UACR — even though the composite cardiovascular outcome itself was consistent across strata (Mann 2025^[2]). Subgroup interactions like this are hypothesis-generating, not definitive — low-albuminuria participants were a smaller, lower-event group — but it is the kind of detail that the headline 24% number hides and that this deep dive exists to surface.

Heart failure: a distinct secondary endpoint

A separate prespecified analysis (Mc Causland et al., Journal of the American College of Cardiology 2024) examined heart-failure outcomes — relevant because diabetic CKD patients carry a heavy heart-failure burden, and about 19% of FLOW participants had heart failure at baseline (Mc Causland 2024^[3]). The composite of heart-failure events or cardiovascular death was reduced 27% (HR 0.73, 95% CI 0.62-0.87; P=0.0005). Heart-failure events alone fell to HR 0.73 (95% CI 0.58-0.92; P=0.0068) and cardiovascular death alone to HR 0.71 (95% CI 0.56-0.89; P=0.0036). Crucially, the benefit was consistent whether participants had heart failure at baseline (HR 0.73) or not (HR 0.72) — the drug was not simply treating prevalent heart failure, it appeared to reduce new events too^[3].

Does the benefit hold across the whole CKD spectrum?

One of the most clinically useful secondary questions is whether semaglutide works as well in advanced kidney disease as in milder disease. The CKD-severity analyses examined outcomes across four baseline eGFR bands (down to <30) and five albuminuria bands (up to ≥2000 mg/g). The answer was reassuringly uniform: the kidney composite was reduced regardless of baseline eGFR (p-interaction 0.83) or UACR (p-interaction 0.42), and all-cause mortality was likewise consistent across eGFR strata (Mann et al., CJASN 2026^[4]). The companion cardiovascular-by-severity paper found the MACE and mortality benefits similarly consistent across eGFR and UACR strata (Mann 2025^[2]). Practically, this means a clinician does not need to withhold semaglutide's kidney protection from someone with a low eGFR or heavy proteinuria — the relative benefit was preserved at the severe end of the spectrum, where absolute risk (and therefore absolute benefit) is highest.

The two concomitant-therapy subgroups everyone asks about

On top of an SGLT2 inhibitor

Because SGLT2 inhibitors are now first-line for diabetic CKD, the obvious question is whether semaglutide adds anything on top. FLOW cannot answer this cleanly: only ~15.6% of participants (550 of 3,533) were on an SGLT2 inhibitor at baseline (Apperloo et al., Nature Medicine 2024). In that small subgroup the primary kidney HR was 1.07 (95% CI 0.69-1.67; P=0.755), versus 0.73 (95% CI 0.63-0.85; P<0.001) in those not on an SGLT2 inhibitor, with a non-significant p-interaction of 0.109^[5]. It is tempting to read the 1.07 as "no benefit when added to an SGLT2 inhibitor," but that interpretation is almost certainly wrong: the subgroup is small, the event count low, the confidence interval enormous, and the interaction test non-significant. The honest read is that FLOW was underpowered to detect added benefit in SGLT2-inhibitor users — a question the ongoing combination-therapy literature is still working out — not that benefit is absent.

On top of a mineralocorticoid receptor antagonist (MRA)

A second prespecified analysis (Rossing et al., Diabetes Care 2025) looked at the ~7.3% (257 participants) taking a mineralocorticoid receptor antagonist — spironolactone, eplerenone, or finerenone — at baseline. The primary kidney HR was 0.51 (95% CI 0.30-0.86) in MRA users and 0.79 (95% CI 0.68-0.92) in non-users, with a non-significant p-interaction of 0.12^[6]. The direction of effect was consistent — semaglutide's kidney protection was preserved (numerically even larger) when stacked on an MRA — though, as with the SGLT2 subgroup, the MRA group was too small to draw firm conclusions about a true difference in effect size. The takeaway is that there is no signal that concomitant MRA use blunts semaglutide's benefit.

How to read all of this together

• The kidney benefit is robust and broad. The 24% primary reduction is backed by a favorable eGFR slope, and it holds across baseline eGFR and albuminuria strata — this is not a fragile finding confined to one patient type (Perkovic 2024^[1]; Mann 2026^[4]).
• It is more than a kidney drug. Cardiovascular death, all-cause mortality, and heart-failure events all moved in the right direction with meaningful effect sizes (Mann 2025^[2]; Mc Causland 2024^[3]).
• Subgroup interactions are signals, not verdicts. The low-albuminuria mortality trend and the SGLT2-user kidney HR of 1.07 are products of small, underpowered subgroups — interesting to note, dangerous to over-interpret (Mann 2025^[2]; Apperloo 2024^[5]).
• It layers onto existing therapy. No analysis showed that a baseline SGLT2 inhibitor or MRA abolished semaglutide's benefit; the data are consistent with additive, not redundant, protection (Apperloo 2024^[5]; Rossing 2025^[6]).

Bottom line

FLOW's 24% kidney-event reduction is the headline, but the secondary analyses are where the clinical picture fills in. Semaglutide slowed the eGFR slope by 1.16 mL/min/1.73m² per year, cut cardiovascular and all-cause death, reduced heart-failure events, and delivered that kidney protection consistently across the full range of baseline kidney function and albuminuria (Perkovic 2024^[1]; Mann 2025^[2]; Mc Causland 2024^[3]; Mann 2026^[4]). The two honest caveats are the underpowered SGLT2-inhibitor subgroup (HR 1.07, but tiny and non-significant) and a low-albuminuria mortality interaction — both hypothesis-generating, neither a reason to withhold a drug that, across the rest of the spectrum, performed remarkably consistently (Apperloo 2024^[5]; Rossing 2025^[6]). For the trial design and FDA-approval context, see our FLOW trial overview, and for the practical CKD-prescribing picture our GLP-1 and CKD patient guide.

This article is educational and is not medical advice. Every PMID, hazard ratio, and confidence interval above was verified against the live PubMed database and the published FLOW primary and secondary-analysis papers before publication. Treatment decisions about semaglutide in chronic kidney disease should be made with a nephrologist or endocrinologist who knows your kidney function, albuminuria, and full medication list.