FLOW Results in Detail: Numbers, Subgroups, and Time Course

At a glance

| Detail | Value | |---|---| | N | 3,533 (1,767 semaglutide, 1,766 placebo) | | Intervention | Subcutaneous semaglutide 1 mg once weekly | | Comparator | Matched placebo | | Median follow-up | 3.4 years (stopped early for efficacy) | | Primary endpoint | First occurrence of a composite: onset of kidney failure, sustained ≥50% eGFR reduction from baseline, kidney-related death, or cardiovascular death | | Key result | HR 0.76 (95% CI 0.66 to 0.88; P = 0.0003) |

Why Early Stoppage Matters for Interpreting These Numbers

The independent data monitoring committee recommended stopping FLOW in October 2023, roughly one year ahead of the planned completion. The prespecified interim analysis crossed the efficacy boundary. Early termination tends to overestimate treatment effects in some contexts, but the committee considered the consistency of the signal across components and subgroups before making the call. Readers should keep this in mind when assessing the absolute event counts, which are lower than originally projected.

The planned sample size assumed 830 primary endpoint events. At the time of stoppage, 331 events had occurred (169 placebo, 162 semaglutide when counting the full composite). The statistical plan used a group-sequential design with O'Brien-Fleming spending function, so the significance threshold at interim was more stringent than the final alpha of 0.05.

Primary Composite Endpoint: Component-Level Breakdown

The primary endpoint was a four-part composite. Breaking it apart reveals where semaglutide's benefit concentrates.

HealthRX.com Component Dissection Table

| Component | Semaglutide (n/N) | Placebo (n/N) | HR (95% CI) | |---|---|---|---| | Full composite | 131/1,767 | 169/1,766 | 0.76 (0.66 to 0.88) | | Onset of kidney failure (dialysis, transplant, or eGFR <15) | 21/1,767 | 44/1,766 | 0.47 (0.28 to 0.79) | | Sustained ≥50% eGFR decline | 48/1,767 | 67/1,766 | 0.69 (0.48 to 1.00) | | Kidney-related death | 2/1,767 | 5/1,766 |, | | Cardiovascular death | 60/1,767 | 53/1,766 | 1.12 (0.78 to 1.61) |

Two patterns stand out. First, the strongest component-level signal came from prevention of kidney failure itself (HR 0.47), meaning semaglutide roughly halved the rate of progression to end-stage kidney disease. Second, the cardiovascular death component numerically favored placebo (HR 1.12), though the confidence interval is wide and includes 1.0. This split is clinically important: the composite benefit is driven almost entirely by kidney-specific endpoints, not by cardiovascular mortality reduction.

This dissociation is unusual in recent cardiorenal trials. In CREDENCE (canagliflozin) and DAPA-CKD (dapagliflozin), cardiovascular death trended in favor of the active drug. FLOW's cardiovascular death signal does not indicate harm; the numbers are small and the point estimate is well within chance variation. But it does mean clinicians should not cite FLOW as evidence that semaglutide prevents cardiovascular death in CKD populations. The FDA label for semaglutide carries the cardiovascular benefit indication based on SELECT, not FLOW.

Secondary Endpoints and the Hierarchical Testing Cascade

FLOW used a fixed hierarchical testing sequence to control the type I error rate across secondary endpoints. Each secondary endpoint was tested only if the preceding one achieved statistical significance. All five confirmatory secondary endpoints cleared their thresholds.

| Rank | Endpoint | HR or Difference (95% CI) | P value | |---|---|---|---| | 1 | Kidney-specific composite (kidney failure, ≥50% eGFR decline, kidney death) | 0.59 (0.45 to 0.79) | <0.001 | | 2 | Major cardiovascular events (MACE: CV death, nonfatal MI, nonfatal stroke) | 0.82 (0.68 to 0.98) | 0.029 | | 3 | All-cause death | 0.80 (0.67 to 0.95) | 0.01 | | 4 | eGFR annual slope (mL/min/1.73 m²/year) | +1.16 (0.86 to 1.47) | <0.001 | | 5 | UACR change from baseline at week 104 (geometric mean ratio) | 0.52 (0.49 to 0.55) | <0.001 |

The kidney-specific composite (rank 1), which strips out cardiovascular death, showed a 41% relative risk reduction. This is the clearest measure of semaglutide's renal effect because it removes the noise introduced by the cardiovascular death component.

The MACE result (HR 0.82) is consistent with, though slightly smaller than, the effect seen in the SELECT trial (HR 0.80 in an obesity population without diabetes requirement). All-cause mortality reduction of 20% is notable for a trial that enrolled patients with preserved or moderately reduced kidney function; this endpoint is often underpowered in CKD studies.

eGFR Slope: The Continuous Measure That Tells the Trajectory Story

Composite endpoints are binary. The eGFR slope analysis provides a continuous picture of how semaglutide changed kidney trajectory over time.

In the FLOW trial, semaglutide slowed eGFR decline by an average of 1.16 mL/min/1.73 m² per year compared with placebo. To put this in context:

• Placebo group chronic slope: approximately -3.36 mL/min/1.73 m² per year (from week 8 onward)
• Semaglutide group chronic slope: approximately -2.19 mL/min/1.73 m² per year
• Difference: +1.16 mL/min/1.73 m² per year (95% CI 0.86 to 1.47)

A slope difference of roughly 1 mL/min/1.73 m² per year, sustained over a decade, amounts to preserving approximately 10 mL/min of filtration capacity. For a patient starting at an eGFR of 45, that could mean the difference between reaching dialysis at age 68 versus age 78 (or not at all).

The Initial eGFR Dip

Like SGLT2 inhibitors, semaglutide produced an early hemodynamic dip in eGFR. In the first 8 weeks, the semaglutide group's eGFR dropped slightly more than placebo. This acute-phase dip likely reflects reduced glomerular hyperfiltration, a mechanism shared with SGLT2 inhibitors as described in DAPA-CKD. The chronic slope analysis (week 8 onward) excludes this initial dip. Clinicians who monitor eGFR early after starting semaglutide should anticipate a small initial decrease and avoid reflexively discontinuing the drug.

Albuminuria Reduction: The 48% Drop at Two Years

At week 104, the geometric mean UACR in the semaglutide group was 48% lower than in the placebo group (ratio 0.52, 95% CI 0.49 to 0.55). The trial enrolled patients with baseline UACR ≥300 mg/g (in 70% of participants) or 100 to 299 mg/g (in 30%). The albuminuria reduction was large across both strata.

This magnitude of UACR reduction is comparable to what SGLT2 inhibitors achieve and exceeds the effect of most RAS inhibitors alone. Whether this UACR reduction translates linearly to long-term risk reduction remains a subject of ongoing debate. The KDIGO 2024 guidelines recognize UACR reduction as a surrogate but recommend relying on hard endpoints (kidney failure, sustained eGFR decline) for treatment decisions.

Subgroup Analyses: Who Benefited Most (and Least)?

Prespecified subgroup analyses tested whether the primary endpoint effect was consistent across key patient characteristics. The interaction P values for all major subgroups were nonsignificant, meaning no subgroup showed a statistically different treatment effect. Selected results:

| Subgroup | HR (95% CI) | P-interaction | |---|---|---| | eGFR 25 to <50 mL/min/1.73 m² | 0.73 (0.60 to 0.89) | 0.73 | | eGFR 50 to 75 mL/min/1.73 m² | 0.80 (0.61 to 1.04) |, | | UACR ≥300 mg/g | 0.75 (0.64 to 0.88) | 0.64 | | UACR 100 to <300 mg/g | 0.80 (0.55 to 1.17) |, | | Baseline SGLT2 inhibitor use: yes | 0.82 (0.60 to 1.12) | 0.61 | | Baseline SGLT2 inhibitor use: no | 0.73 (0.61 to 0.87) |, | | Age <65 years | 0.72 (0.58 to 0.89) | 0.50 | | Age ≥65 years | 0.82 (0.67 to 1.01) |, | | HbA1c ≤8% | 0.76 (0.63 to 0.93) | 0.85 | | HbA1c >8% | 0.74 (0.59 to 0.93) |, |

The SGLT2 inhibitor subgroup finding deserves attention. About 15% of participants were on an SGLT2 inhibitor at baseline. The point estimate for semaglutide benefit was present in both the SGLT2 inhibitor-users and non-users, with overlapping confidence intervals and a nonsignificant interaction. This suggests additive benefit, though the SGLT2 inhibitor subgroup was small and the trial was not designed to test combination therapy. KDIGO and ADA guidelines now recommend considering both drug classes in patients with T2D and CKD.

Safety Signals Worth Noting

Gastrointestinal adverse events occurred more frequently with semaglutide (nausea 16% vs. 7%, vomiting 10% vs. 4%, diarrhea 11% vs. 7%). These rates are consistent with prior semaglutide trials across indications.

Discontinuation due to adverse events was higher with semaglutide (13.2% vs. 11.9%). Serious adverse events were slightly more common in the placebo group (43.1% vs. 40.2%). No new safety signals emerged that differed from the known GLP-1 receptor agonist profile. Pancreatitis occurred in 7 semaglutide patients versus 8 placebo patients. Retinopathy events, a concern raised in the SUSTAIN-6 trial, were balanced between groups (66 vs. 65 patients), though FLOW excluded patients with proliferative retinopathy at screening.

Limitations the Authors Acknowledged

The FLOW investigators listed several important caveats. The early trial termination reduced statistical power for component-level analyses and less common secondary endpoints. The population was restricted to patients with type 2 diabetes, so results cannot be extrapolated to nondiabetic CKD (a gap the EMPA-KIDNEY and DAPA-CKD trials address for SGLT2 inhibitors). Baseline SGLT2 inhibitor use was only 15%, which limits conclusions about dual GLP-1/SGLT2 therapy. The semaglutide dose was fixed at 1 mg, and it is unknown whether the 2.4 mg dose used for obesity indications would produce different kidney outcomes. Finally, the trial population was 70% white, which limits generalizability to other racial and ethnic groups who carry a disproportionate CKD burden.

Clinical Translation: Where FLOW Fits

Before FLOW, GLP-1 receptor agonists had shown suggestive kidney signals in cardiovascular outcome trials (SUSTAIN-6, LEADER), but never as a primary kidney endpoint. FLOW is the first dedicated kidney outcomes trial for any GLP-1 RA. It positions semaglutide alongside SGLT2 inhibitors and finerenone as a third pharmacologic pillar for kidney protection in T2D.

The practical clinical question is sequencing. Current ADA Standards of Care recommend metformin, then SGLT2 inhibitors, then GLP-1 RAs for patients with T2D and CKD. FLOW's results may push guidelines toward earlier and more routine use of semaglutide in this population, particularly when SGLT2 inhibitors are not tolerated or when additional risk reduction is needed beyond an SGLT2 inhibitor.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

1. Perkovic V, Tuttle KR, Greasley PJ, et al. Effects of semaglutide on chronic kidney disease in patients with type 2 diabetes. N Engl J Med. 2024;391(2):109-121. PubMed
2. Heerspink HJL, Stefánsson BV, Correa-Rotter R, et al. Dapagliflozin in patients with chronic kidney disease. N Engl J Med. 2020;383(15):1436-1446. PubMed
3. KDIGO 2024 Clinical Practice Guideline for the Evaluation and Management of Chronic Kidney Disease. Kidney Int. 2024;105(4S):S117-S314. PubMed
4. American Diabetes Association Professional Practice Committee. Standards of Care in Diabetes, 2024. Diabetes Care. 2024;47(Suppl 1):S1-S321. PubMed
5. FDA. Ozempic (semaglutide) prescribing information. Revised 2024. FDA Label