FLOW Extension Data and What Happened After the Trial Ended

By
Published Updated Last reviewed
GLP-1 medication and metabolic health image for FLOW Extension Data and What Happened After the Trial Ended

At a glance

| Parameter | Detail | |-----------|--------| | N | 3,533 (1,767 semaglutide, 1,766 placebo) | | Intervention | Subcutaneous semaglutide 1.0 mg weekly | | Comparator | Matched placebo | | Duration | Median 3.4 years (planned 5 years, stopped early) | | Primary endpoint | Composite: onset of kidney failure, sustained ≥50% eGFR decline, kidney death, or cardiovascular death | | Key result | HR 0.76 (95% CI 0.66, 0.88), 24% relative risk reduction |

Why Early Termination Matters for Follow-Up Interpretation

FLOW was stopped early by the Data Monitoring Committee in October 2023 after a pre-specified interim analysis showed overwhelming efficacy. The trial had originally planned for five years of follow-up. Participants had accumulated a median of 3.4 years on treatment when unblinding occurred.

Early stopping for benefit creates a specific interpretive problem. The effect estimate at the point of termination may overestimate the true long-term treatment effect because interim analyses that trigger early stopping tend to catch the trial at a moment of peak separation between curves. This phenomenon, sometimes called the "random high" problem, means the 24% risk reduction could represent the upper boundary of the real effect rather than a stable point estimate.

The FLOW investigators addressed this concern through several pre-specified sensitivity analyses. The primary result held across all of them, including analyses that censored at interim analysis date and analyses using different handling of missing data. The consistency across these approaches provides some reassurance, but none of them can substitute for additional years of observation.

eGFR Slope as a Durability Signal

The most informative data about sustained benefit comes from the eGFR slope analysis reported in the primary publication. Semaglutide reduced the annual rate of eGFR decline by approximately 1.16 mL/min/1.73m² per year compared with placebo over the total treatment period.

Critically, the slope difference was consistent across time intervals:

| Time interval | Semaglutide eGFR slope | Placebo eGFR slope | Difference | |---------------|------------------------|--------------------|--------------------| | Weeks 0, 104 | −2.19 mL/min/1.73m²/yr | −3.36 mL/min/1.73m²/yr | 1.16 mL/min/1.73m²/yr | | Weeks 104, end | Continued separation | Continued decline | Maintained |

The absence of slope convergence during the observed period suggests the mechanism of kidney protection was not a transient hemodynamic effect that would fade over time. In contrast, the acute eGFR dip seen with SGLT2 inhibitors (which reverses on discontinuation) raises questions about durability in that drug class. For semaglutide, the slope data imply a disease-modifying effect on kidney function decline itself.

The Post-Trial Transition Problem

When FLOW was stopped early, participants in the placebo group became eligible for open-label semaglutide. This standard ethical practice creates a well-known problem for long-term follow-up: once both groups receive active treatment, between-group differences in outcomes become uninterpretable.

No formal FLOW extension study with continued randomized follow-up has been published as of mid-2026. The trial's ClinicalTrials.gov record does not list a registered extension phase with blinded comparator groups. Any future analyses of FLOW participants will therefore be observational or registry-based, not randomized.

This contrasts with trials like CREDENCE (canagliflozin), which also stopped early but had its full dataset available for post-hoc time-course analyses. FLOW's slightly longer observation period before termination (3.4 vs. 2.6 years) provides modestly more data, but neither trial reached its originally planned duration.

What "Durability" Means in Context

For clinicians interpreting FLOW, the durability question splits into two distinct clinical scenarios:

Scenario 1: Continued treatment. If a patient stays on semaglutide indefinitely, does kidney protection persist? The slope data from FLOW suggest yes through 3.4 years. Supporting evidence comes from the SELECT cardiovascular outcomes trial, where semaglutide 2.4 mg showed sustained cardiorenal benefit through a median 39.8 months of follow-up, and from the SUSTAIN-6 post-trial analysis showing eGFR preservation at 2 years.

Scenario 2: Treatment discontinuation. If a patient stops semaglutide, does the kidney benefit persist, partially reverse, or fully reverse? FLOW did not include a structured washout period. No randomized data exist to answer this question for semaglutide specifically.

The biological plausibility of sustained benefit after discontinuation depends on mechanism. If semaglutide protects kidneys primarily through weight loss, glycemic improvement, and blood pressure reduction, some benefit likely persists after stopping (weight regain is partial, not complete). If protection requires ongoing GLP-1 receptor activation in kidney tissue, benefit would likely diminish after drug clearance.

Regression to the Mean Considerations

FLOW enrolled patients with eGFR 25 to 75 mL/min/1.73m² and UACR 300, 5 to 000 mg/g. These entry criteria selected patients who were, by definition, already in decline. Some portion of the observed placebo-group deterioration reflects the natural history of the selected population rather than a worsening that would have occurred in an unselected cohort.

The trial's design mitigated this concern by requiring a run-in period and confirmatory laboratory values. Participants needed two qualifying eGFR and albuminuria measurements, reducing the likelihood that transient fluctuations drove enrollment. The protocol paper published in Nephrology Dialysis Transplantation describes these procedures in detail.

Regression to the mean would be most problematic if the baseline values represented temporary extremes. For albuminuria (which fluctuates substantially day-to-day), this is a real concern. For eGFR in established CKD, it is less so, because eGFR decline in this population is largely monotonic rather than oscillatory.

Safety Signals Through Extended Observation

The FLOW safety dataset through 3.4 years of exposure provides one of the longest randomized safety datasets for semaglutide in a CKD population. Key findings:

| Safety parameter | Semaglutide | Placebo | |-----------------|-------------|---------| | Serious adverse events | 49.6% | 53.7% | | Discontinuation for AE | 13.2% | 11.9% | | GI events (any) | 51.2% | 36.2% | | Pancreatitis (adjudicated) | 8 events | 8 events | | Neoplasms (all) | Similar rates | Similar rates |

The GI burden was substantial (a 15 percentage-point excess), consistent with all GLP-1 RA trials. The discontinuation rate difference (1.3 percentage points higher with semaglutide) was modest, suggesting most GI events were manageable.

No new safety signals emerged during extended observation compared with the semaglutide FDA prescribing information. Specifically, the trial did not confirm the thyroid C-cell tumor signal seen in rodent studies. Pancreatitis rates were balanced. Retinopathy events were not notably increased in this CKD-specific population, unlike the SUSTAIN-6 signal.

Implications for Guideline Updates

Following FLOW's publication, KDIGO (Kidney Disease: Improving Global Outcomes) issued a practice point recommending GLP-1 RAs as second-line kidney-protective therapy for patients with T2D and CKD who are already on maximally tolerated RAS inhibition and SGLT2 inhibition. This recommendation hinged on FLOW being the first dedicated kidney outcomes trial for any GLP-1 RA.

The early termination did not prevent guideline endorsement because the effect size was large and the p-value (p = 0.0003) was far below the stopping boundary. Guidelines committees accepted that additional follow-up time would have been informative but was not necessary to establish clinical benefit.

What Remains Unknown

Several questions about long-term FLOW outcomes remain open:

  1. CKD stage-specific durability. Did the benefit persist equally in participants with eGFR 25, 44 vs. 45, 75? Pre-specified subgroup analyses showed consistent point estimates, but power within subgroups was limited by early stopping.

  2. Albuminuria trajectory beyond 3.4 years. The 24% reduction in UACR observed with semaglutide could theoretically attenuate over time if tubular compensation mechanisms develop.

  3. Cardiovascular death contribution. CV death was included in the composite. Whether the kidney-specific components alone would have triggered early stopping is not publicly known.

  4. Interaction with SGLT2 inhibitors over time. Approximately 15.6% of FLOW participants used SGLT2 inhibitors at baseline. The combination's durability profile has never been studied beyond 3 years.

  5. Off-treatment follow-up. No data exist on what happens to kidney function after semaglutide discontinuation in this population.

Comparison With Other Kidney Trials Stopped Early

FLOW's early termination is not unique. CREDENCE (canagliflozin), DAPA-CKD (dapagliflozin), and EMPA-KIDNEY (empagliflozin) were all stopped early for efficacy. The field now faces a recurring problem: the most important kidney protection trials in the modern era have all been truncated, creating a systematic gap in long-term evidence.

| Trial | Drug | Planned duration | Actual median | % shortfall | |-------|------|-----------------|---------------|-------------| | CREDENCE | Canagliflozin | 5.5 years | 2.6 years | 53% | | DAPA-CKD | Dapagliflozin | Event-driven | 2.4 years | N/A | | EMPA-KIDNEY | Empagliflozin | Event-driven | 2.0 years | N/A | | FLOW | Semaglutide | 5.0 years | 3.4 years | 32% |

FLOW achieved the longest median follow-up of these four trials. Its 32% shortfall is the smallest. This relative advantage gives it slightly more durability evidence than its comparators, even though it too was terminated before plan completion.

Clinical Translation

For prescribers weighing whether to start semaglutide for kidney protection in a patient with T2D and CKD stages 3, 4, the available durability data support confidence through at least 3.4 years of continuous treatment. The eGFR slope data do not show waning benefit during observation.

The honest clinical answer about what happens beyond year four is: we do not know. The trial was stopped. No extension is running. Registry data may eventually fill this gap, but with inherent confounding.

Given the biological rationale (weight-mediated and direct anti-inflammatory mechanisms), the slope consistency, and the absence of late safety signals, most nephrologists and endocrinologists have treated the FLOW result as sufficient for long-term prescribing. The absence of extension data has not prevented clinical adoption.

Frequently asked questions

›
›
›
›
›
›
›
›
›
›

References

  1. Perkovic V, Tuttle KR, Engerson S, et al. Effects of semaglutide on chronic kidney disease in patients with type 2 diabetes. N Engl J Med. 2024;391(2):109-121. https://pubmed.ncbi.nlm.nih.gov/38785350/
  2. Perkovic V, Tuttle KR, Engerson S, et al. Design and baseline characteristics of the FLOW trial. Nephrol Dial Transplant. 2023;38(3):575-583. https://pubmed.ncbi.nlm.nih.gov/36459489/
  3. KDIGO 2024 Clinical Practice Guideline for Diabetes Management in Chronic Kidney Disease. Kidney Int. 2024;105(4S):S1-S127. https://pubmed.ncbi.nlm.nih.gov/38785209/
  4. FDA Prescribing Information: Ozempic (semaglutide) injection. Revised 2024. https://www.accessdata.fda.gov/drugsatfda_docs/label/2024/209637s020lbl.pdf
  5. The EMPA-KIDNEY Collaborative Group. Empagliflozin in patients with chronic kidney disease. N Engl J Med. 2023;388(2):117-127. https://pubmed.ncbi.nlm.nih.gov/36482093/
  6. Marso SP, Bain SC, Consoli A, et al. Semaglutide and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. 2016;375(19):1834-1844. https://pubmed.ncbi.nlm.nih.gov/27633186/