FLOW Trial Update: GLP-1 for CKD Stage 3-4 — The Evidence

FLOW (Perkovic 2024 NEJM^[1]) stopped early for efficacy after a pre-specified interim analysis: semaglutide 1 mg weekly reduced major kidney events — a composite of kidney failure, sustained ≥50% eGFR decline, and kidney or cardiovascular death — by 24% in 3,533 adults with type 2 diabetes and chronic kidney disease. The eGFR slope was attenuated by about 1.2 mL/min/1.73 m² per year on semaglutide vs placebo. That result moved GLP-1 receptor agonists from “reasonable for glycemic control in T2D with CKD” to a guideline-level kidney-protective therapy alongside SGLT2 inhibitors. This article walks through what FLOW actually showed, how it sits next to DAPA-CKD and EMPA-KIDNEY, the practical dose-by-stage prescribing protocol, and the dialysis-imminent decision tree.

The honest summary

• FLOW reduced major kidney events by 24%. In 3,533 patients with T2D and CKD (eGFR 50–75 with UACR 300–5,000, or eGFR 25–<50 with UACR 100–5,000), semaglutide 1 mg weekly vs placebo produced a hazard ratio of 0.76 (95% CI 0.66–0.88, p=0.0003) for the primary composite (Perkovic 2024^[1]).
• The eGFR slope improved. The annual eGFR decline was about 1.2 mL/min/1.73 m² per year slower on semaglutide vs placebo — clinically meaningful over 3–5 years of follow-up.
• SGLT2 evidence is larger but compatible. DAPA-CKD (Heerspink 2020^[2]) reported a 39% reduction in a similar composite, and EMPA-KIDNEY (Herrington 2023^[3]) reported a 28% reduction. The mechanisms differ — SGLT2 acts on tubuloglomerular feedback, GLP-1 acts on systemic inflammation and weight — so combination therapy is now the AACE/ADA/KDIGO default for T2D + CKD (KDIGO 2022^[7]).
• Most GLP-1s need no renal dose adjustment. Semaglutide, tirzepatide, liraglutide, and dulaglutide all have label-supported use down to CKD stage 4 with no dose adjustment. Exenatide is the exception — avoid at eGFR <30. Data in CKD 5 / dialysis remain thin across the class.

What FLOW actually showed

FLOW enrolled 3,533 adults with type 2 diabetes and CKD defined by eGFR 50–75 mL/min/1.73 m² with urinary albumin-to-creatinine ratio (UACR) 300–5,000 mg/g, or eGFR 25–<50 with UACR 100–5,000 mg/g (Perkovic 2024^[1]). Participants were randomized 1:1 to semaglutide 1 mg subcutaneously once weekly or placebo on top of guideline-directed RAS blockade. The primary composite endpoint was the first occurrence of kidney failure (dialysis, transplantation, or sustained eGFR <15), a sustained ≥50% reduction in eGFR from baseline, or kidney or cardiovascular death.

At a median follow-up of 3.4 years the trial was stopped early at the pre-specified interim analysis for efficacy. The primary composite occurred in 331 of 1,767 semaglutide participants vs 410 of 1,766 placebo participants — hazard ratio 0.76 (95% CI 0.66–0.88, p=0.0003), a 24% relative risk reduction and roughly a 4.4 percentage-point absolute reduction over the follow-up window. The kidney-specific component (kidney failure plus ≥50% eGFR decline plus kidney death) showed a hazard ratio of 0.79. Cardiovascular death alone was reduced by 29%, and all-cause death by 20% — effect sizes consistent with the SUSTAIN-6 (Marso 2016^[4]) and LEADER (Marso 2016^[5]) cardiovascular outcomes trials but now in a higher-risk CKD population.

The eGFR slope analysis is what changed nephrology practice. Total slope was attenuated by 1.16 mL/min/1.73 m² per year in the semaglutide arm vs placebo; the chronic slope (after the initial acute “dip” that some kidney- protective agents produce) was attenuated by 1.34 per year. Over a 5-year horizon, that translates to roughly 6 extra mL/min preserved — the difference between borderline CKD 3b and definite CKD 4 for many patients.

FLOW vs DAPA-CKD vs EMPA-KIDNEY: how the kidney trials line up

DAPA-CKD (Heerspink 2020^[2]) randomized 4,304 patients with eGFR 25–75 and UACR 200–5,000 to dapagliflozin 10 mg or placebo; the primary kidney composite was reduced by 39% (hazard ratio 0.61). EMPA-KIDNEY (Herrington 2023^[3]) randomized 6,609 patients across a broader eGFR range (20–<45 regardless of albuminuria, or 45–<90 with UACR ≥200) and showed a 28% reduction in the primary composite of kidney progression or cardiovascular death (hazard ratio 0.72).

The mechanistic stories differ. SGLT2 inhibitors act proximally on tubuloglomerular feedback — the initial acute eGFR dip is the signature of restored feedback and predicts long-term preservation. GLP-1 receptor agonists appear to work through weight loss, blood-pressure reduction, reduction in albuminuria, and a direct anti-inflammatory signal in the kidney; there is no acute eGFR dip with semaglutide. Because the mechanisms are non-overlapping, the consensus is that the renal effects of GLP-1 plus SGLT2 in the same patient are likely additive — not yet proven head-to-head in a stacking RCT, but the rationale is strong enough that KDIGO 2022^[7] and the ADA 2024 Standards of Care^[9] both recommend the combination as first-line for T2D with CKD and persistent albuminuria.

Magnitude: relative kidney event reduction by therapy

CKD stage breakdown and dose adjustment

Chronic kidney disease is staged by eGFR plus albuminuria (UACR). Stage 1 is eGFR ≥90 with kidney damage markers (proteinuria or imaging); stage 2 is eGFR 60–89 with damage; stage 3a is 45–59; stage 3b is 30–44; stage 4 is 15–29; stage 5 is <15 or on dialysis. The renal effects of GLP-1 receptor agonists, and their prescribing implications, vary by stage.

• Semaglutide (Ozempic, Wegovy, Rybelsus). No dose adjustment for any eGFR. Label studied to CKD 4 (FLOW enrolled to eGFR 25). Data in CKD 5 and dialysis are limited; case-level use has not flagged safety issues but formal trial data are not available.
• Tirzepatide (Mounjaro, Zepbound). No dose adjustment for any eGFR. SURPASS-4 included patients with eGFR down to ~30; CKD 4–5 experience is thinner than for semaglutide. The dedicated dual GIP/GLP-1 CKD outcomes trial has not yet read out.
• Liraglutide (Victoza, Saxenda). No dose adjustment. The LEADER trial (Marso 2016^[5]) reported a secondary renal composite reduction of 22% in a T2D population with prevalent CKD subgroups.
• Dulaglutide (Trulicity). No dose adjustment. AWARD-7 (Tuttle 2018^[6]) specifically enrolled 577 patients with moderate-to-severe CKD (stage 3–4) and showed comparable A1c reduction with attenuated eGFR decline vs insulin glargine.
• Exenatide (Byetta, Bydureon). Avoid at eGFR <30 (CKD 4–5). Renal elimination is significant and accumulation risk plus nausea-driven dehydration risk make this the one class member to switch off below stage 4.

The practical prescribing protocol

1. Confirm staging. eGFR (CKD-EPI 2021 equation, race-free) plus spot urine ACR within 90 days of starting. Document the stage (3a, 3b, 4) and the albuminuria category (A1 <30, A2 30–300, A3 >300 mg/g).
2. Optimize the base. Maximally tolerated ACE inhibitor or ARB; SGLT2 inhibitor (dapagliflozin or empagliflozin) if eGFR ≥20 and not already on one; finerenone if T2D + albuminuria (KDIGO 2022^[7], ADA 2024^[9]).
3. Add the GLP-1. Start at the standard induction dose (semaglutide 0.25 mg weekly × 4 weeks, tirzepatide 2.5 mg weekly × 4 weeks). No renal dose adjustment is required, but the dose- escalation pace should be unhurried in CKD 3b–4 because nausea-driven dehydration is the dominant short-term risk.
4. Recheck eGFR at week 4, week 12, and week 24. A small <15% eGFR drop during titration is not unusual and does not warrant discontinuation; a sustained drop >30% should prompt nephrology review and a hold while the cause is sorted out.
5. Adjust diuretics down. Patients on loop diuretics or thiazides for blood pressure or edema often need a step-down when a GLP-1 reduces appetite, food-derived sodium load, and body weight. Falling orthostatics or acute pre-renal AKI on titration is usually a diuretic problem, not a GLP-1 problem.
6. Re-stage at month 6 and month 12. Repeat eGFR and UACR. Falling UACR is one of the early signals that the GLP-1 is doing renal work.
7. Nephrology referral. Refer at CKD 3b with rapid progression, all CKD 4, all CKD 5, or any patient considering pre-emptive transplant evaluation.

The dialysis-imminent decision tree

For patients within 6–12 months of likely dialysis initiation, the GLP-1 conversation gets harder. Three considerations dominate.

First, AV fistula maturation typically takes 8–12 weeks. GLP-1 therapy does not interfere with vascular access creation or healing, and there is no published reason to hold the medication peri-operatively beyond standard fasting requirements. Continue the GLP-1 unless surgical or anesthesia consultation has documented a specific concern (e.g., severe gastroparesis with anticipated airway risk).

Second, dehydration sensitivity is amplified in patients with low residual renal reserve. Counsel explicitly on volume status; use lower-end titration steps; and consider holding a dose if the patient develops a gastroenteritis-equivalent illness with vomiting or diarrhea.

Third, data on GLP-1 use during hemodialysis are limited but not negative. Pharmacokinetic studies show minimal removal by dialysis for semaglutide and tirzepatide; clinical experience in observational series has not flagged excess hypoglycemia or other safety signals. The current practical default at most U.S. dialysis centers is to continue the GLP-1 if it has been tolerated, and to re-evaluate the indication every 6 months. For pre-emptive-transplant candidates, weight loss to BMI <35 often determines listing eligibility — a strong argument for continuation through the dialysis window.

Hyperkalemia, anemia, and the indirect benefits

GLP-1 receptor agonists do not directly alter serum potassium or hemoglobin. The indirect effects are more interesting. Weight loss and improved insulin sensitivity reduce compensatory hyperinsulinemia, which in turn modestly reduces aldosterone-driven potassium retention — some patients see their pre-RAS-blocker hyperkalemia margin widen. Anemia of CKD responds in a similar indirect way: reduced inflammation and reduced adipose-tissue-derived hepcidin signaling can lower erythropoiesis-stimulating- agent requirements at the margin. Neither effect is large enough to be a primary indication, but both are reasons the labs trend better than the medication list alone would predict.

Post-transplant and pre-emptive transplant patients

Pre-emptive transplant candidates with obesity face listing thresholds typically at BMI 35–40 depending on center. GLP-1 therapy to facilitate listing is now common practice and is consistent with the 2024 ADA guidance on comorbidity management^[10]. Post-transplant GLP-1 use is also expanding. The theoretical concern is tacrolimus absorption variability from delayed gastric emptying; published case series have not documented clinically meaningful tacrolimus trough disruption, but more-frequent trough monitoring during titration is prudent. The SELECT trial (Lincoff 2023^[8]) established cardiovascular benefit in non-diabetic obesity — relevant for post-transplant patients whose all-cause mortality is dominated by cardiovascular events.

Coverage, cost, and the value case

Medicare Part D covers Ozempic for type 2 diabetes including T2D + CKD; some Part D plans have begun expanding coverage criteria explicitly for CKD prevention following FLOW. Wegovy is not currently covered by Medicare for obesity alone, but the SELECT label expansion has opened coverage for established cardiovascular disease. Commercial coverage for renal indication is patchier and often requires prior-authorization documentation of CKD stage and progression.

The cost-effectiveness math favors GLP-1 therapy in this population. Annual U.S. dialysis cost runs roughly $90,000 per patient; any therapy that defers dialysis by even 12 months is high-value. Whether your payer agrees is a separate question.

Related research and tools

• GLP-1 + SGLT2 stacking — the additive renal and CV evidence behind the combination protocol
• SELECT trial cardiovascular benefits — semaglutide CV outcomes in obesity without diabetes (Lincoff 2023)
• FLOW trial primer — the original headline numbers and patient-facing summary
• GLP-1 for CKD patient guide — the patient-facing version of the prescribing protocol
• Retatrutide and kidney protection — the next-generation triple agonist's renal signal
• SGLT2 vs GLP-1 — head-to-head class comparison for the T2D + CKD patient
• GLP-1 insurance coverage — Medicare Part D and commercial PA pathways for renal indications

Important disclaimer. This article is educational and does not constitute medical advice. CKD management requires individualized assessment by a nephrologist or qualified clinician familiar with the patient's comorbidities, medications, and dialysis access plan. GLP-1 receptor agonists are not a substitute for RAS blockade or SGLT2 inhibition in patients who can tolerate them. Dose adjustments for tacrolimus and other narrow-therapeutic-index medications during GLP-1 titration should be coordinated with the transplant or renal team. PMIDs were verified live against the PubMed E-utilities API on 2026-05-29.

Last verified: 2026-05-29. Next review: every 12 months, or sooner if new prospective CKD outcomes trial data on tirzepatide (SURPASS-CVOT or a dedicated renal substudy), retatrutide, or oral semaglutide become available.