Metabolic HealthResearch PaperPaywall

Semaglutide Protects Kidneys in Type 2 Diabetes by Repairing Vessels and Halting Fibrosis

A 52-week randomized trial reveals semaglutide reduces kidney vascular resistance and prevents fibrosis progression, uncovering the molecular mechanisms behind its kidney-protective effects.

Saturday, October 3, 2026 0 views
Published in Nat Med
A nephrologist reviewing a kidney MRI scan on a lightbox in a clinical office, with a syringe and vial on the desk beside them

Summary

Chronic kidney disease affects about 40% of people with type 2 diabetes. The FLOW trial showed weekly semaglutide protects the kidneys, but not how. The REMODEL trial randomized 106 adults to semaglutide or placebo for 52 weeks and used specialized MRI, kidney biopsies and single-cell gene analysis to find out. The main MRI measures of oxygenation, blood flow and inflammation did not change. But semaglutide lowered resistance in the kidney's arteries, stabilized an MRI marker of scarring, shrank fat around the kidney, cut albumin in urine by 40%, and shifted gene activity in the cells lining the kidney's filters. Fewer immune cells sat near those cells. The findings point to direct kidney effects beyond weight and glucose control, though the study was small and exploratory.

Detailed Summary

Chronic kidney disease (CKD) affects nearly 800 million people worldwide and develops in about 40% of those with type 2 diabetes (T2D), making diabetes the leading cause of kidney failure. In the FLOW trial, weekly semaglutide cut the risk of major kidney events by 24%. Yet mediation analyses show that lower blood sugar and weight loss explain only part of that benefit. The REMODEL trial was built to uncover what semaglutide actually does inside the kidney.

REMODEL randomized 106 adults with T2D and CKD (71 to semaglutide 1 mg weekly, 35 to placebo) at 28 sites in eight countries. Participants averaged 65 years old, and 25 (23.6%) were women. Mean eGFR was 51 ml/min/1.73 m² and median urinary albumin-to-creatinine ratio (UACR) was 187 mg/g. Over 52 weeks, researchers combined multiparametric kidney MRI with paired biopsies in 33 participants, using histology, single-nucleus RNA sequencing (22 paired samples, about 617,000 nuclei) and spatial transcriptomics (13 participants).

The three coprimary MRI outcomes missed significance. Cortical oxygenation (R2*, treatment ratio 0.98), global perfusion (ratio 1.10, 95% CI 0.98–1.24) and T1 mapping (ratio 1.00) did not differ from placebo. Secondary outcomes were more revealing. Renal artery resistive index fell (ratio 0.96, P=0.008), and cortical apparent diffusion coefficient stayed higher (ratio 1.05, P=0.005), consistent with halted fibrosis progression. Perirenal and sinus fat volumes dropped 25% and 13% versus placebo, out of proportion to the roughly 4.5 kg of weight loss.

Clinically, semaglutide lowered UACR by 40% (P=0.008) and raised 24-hour creatinine clearance by 12 ml/min versus placebo (P=0.038). The eGFR differences (about 2–3 ml/min/1.73 m²) did not reach significance. HbA1c fell 0.88 percentage points and weight 4.45 kg. In biopsies, the intimal area of the most diseased arterioles shrank by 10% (P=0.032), and periglomerular immune infiltrates appeared reduced.

Single-nucleus sequencing showed that 7 of 31 cell clusters responded, with glomerular endothelial cells the most affected. Genes tied to lipid metabolism, ER stress, TGFβ, NF-κB and MAPK signaling were mostly downregulated. Pathway enrichment did not survive multiple-testing correction. Spatial analysis found fewer natural killer/NKT cells within 25 μm of these endothelial cells (13.6% to 7.7%), plus stronger PECAM1 and VE-cadherin autocrine signaling, which suggests better endothelial barrier integrity.

In a post hoc analysis, participants also taking SGLT2 inhibitors saw an eGFR gain of 8.1 ml/min/1.73 m² versus placebo, while those not taking them saw 0.2. This is a hypothesis-generating signal for combination therapy. Treatment discontinuation due to adverse events was higher with semaglutide (12.7% vs 2.9%), and no biopsy-related serious events occurred. The authors stress the associations are not causal, the trial was underpowered for its imaging endpoints, no multiplicity adjustment was applied, and women were under-represented.

Key Findings

  • Coprimary MRI outcomes were not significantly changed: R2* oxygenation ratio 0.98 (95% CI 0.96–1.01), perfusion 1.10 (0.98–1.24), T1 mapping 1.00
  • Renal artery resistive index fell versus placebo (ratio 0.96, 95% CI 0.93–0.99; P=0.008), indicating lower renal vascular resistance
  • Cortical ADC stayed higher with semaglutide (ratio 1.05; P=0.005), suggesting fibrosis stabilization
  • Perirenal fat fell 25% and renal sinus fat 13% versus placebo (P=0.027 and P=0.012), more than the ~5% body weight loss
  • UACR dropped 40% versus placebo (ratio 0.60, 95% CI 0.41–0.88; P=0.008); 24-hour creatinine clearance rose 12 ml/min (P=0.038)
  • Fractional intimal area of the most diseased arterioles was 10% lower with semaglutide on biopsy (P=0.032)
  • Glomerular endothelial cells were the most semaglutide-responsive cell type, with fewer NKT cells within 25 μm of them (13.6% to 7.7%); discontinuation for adverse events was 12.7% vs 2.9%

Methodology

REMODEL was a 52-week, double-blind, placebo-controlled trial (NCT04865770) in which 106 adults with T2D and CKD were randomized 2:1 to semaglutide 1 mg weekly (n=71) or placebo (n=35) at 28 sites in 8 countries. Kidney MRI (BOLD, phase-contrast, T1 mapping, diffusion) was done in all participants, and 33 had paired biopsies for histology, snRNAseq (22 paired) and spatial transcriptomics (13 analyzed). MRI outcomes were analyzed with ANCOVA on the log scale, with baseline as covariate and no multiplicity adjustment. Differential gene expression used a generalized linear model with Benjamini-Hochberg FDR below 0.1, and ligand-receptor and spatial proximity analyses were post hoc.

Study Limitations

The trial was exploratory and underpowered for its imaging endpoints, and sample size assumptions for R2* were likely too optimistic. No multiplicity adjustment was used, so nominal P values need caution. The links among MRI, histology, transcriptomics and clinical outcomes are associative, with no formal mediation analysis. The SGLT2 subgroup, biopsy and spatial cohorts were small, and GFR was not measured directly by iohexol clearance. Women made up only 23.6% of participants. Funding and conflict-of-interest disclosures were not included in the supplied text.

Enjoyed this summary?

Get the latest longevity research delivered to your inbox every week.

Enter your email to subscribe: