Chronic kidney disease (CKD) is a major complication of type 2 diabetes (T2D), affecting hundreds of millions worldwide and developing in a substantial proportion of people with T2D. Clinical outcome data from the FLOW trial demonstrated that once-weekly subcutaneous semaglutide 1.0 mg reduced risk of a composite kidney outcome by 24% in people with T2D and CKD. However, the mechanisms underlying kidney protection by GLP-1 receptor agonists are incompletely understood. Mediation analyses suggest that improvements in glycemia and body weight only partly account for the observed kidney benefits, and preclinical work has implicated effects on inflammation, fibrosis and the renal vasculature.
The REMODEL (REnal MODE of action of semagLutide) trial was designed as a mechanistic, tissue-centered study to identify kidney-specific mechanisms of semaglutide in a population analogous to FLOW. The trial combined multiparametric kidney magnetic resonance imaging (MRI), paired kidney biopsies for histology and transcriptomics, and spatial transcriptomics to map cell-type–specific and spatially resolved responses to treatment.
REMODEL was a randomized, placebo-controlled trial of subcutaneous semaglutide 1 mg once weekly versus placebo over 52 weeks. Between April 2021 and October 2023, 106 participants were randomized at 28 sites across eight countries (71 assigned to semaglutide, 35 to placebo). The cohort included 25 women and 81 men. Key baseline characteristics were mean age 65.3 ± 9.9 years, mean eGFRcreatinine 51.1 ± 10.4 ml min−1 per 1.73 m2, median urinary albumin-to-creatinine ratio (UACR) 187.3 mg g−1 (IQR 60.5–546.4), mean body weight 89.6 ± 18.9 kg, mean body mass index 30.8 ± 5.7 kg m−2 and mean glycated hemoglobin (HbA1c) 7.1 ± 0.9%. Use of renin–angiotensin-system inhibitors was nearly universal (99.1%). Sodium–glucose cotransporter-2 (SGLT2) inhibitor use was 39.6% overall and 54.5% in the biopsy subgroup.
A subset of participants underwent paired kidney biopsies and molecular profiling: 33 participants (22 semaglutide, 11 placebo) had biopsies, 22 contributed single-nucleus RNA sequencing (snRNAseq) data, and 13 had spatial transcriptomics on paired samples with pre- and post-treatment measurements.
ClinicalTrials.gov identifier: NCT04865770.
REMODEL used an integrative pathophysiological approach. Multiparametric MRI assessed renal tissue oxygenation (R2*), global renal perfusion and tissue characteristics including T1 mapping as a marker of inflammation or altered tissue composition. Diffusion-weighted metrics such as the apparent diffusion coefficient (ADC) were included to probe microstructural changes consistent with fibrosis progression. Paired kidney biopsies enabled histologic assessment and high-resolution molecular profiling: snRNAseq characterized cell-type–specific transcriptional responses and spatial transcriptomics mapped cell populations and their local interactions within tissue architecture.
The trial prespecified coprimary MRI outcomes of renal oxygenation (R2*), global perfusion and tissue inflammation by T1 mapping. After 52 weeks, semaglutide did not significantly alter these coprimary MRI measures compared with placebo. In other words, the primary imaging endpoints designed to capture bulk oxygenation, perfusion changes and T1-derived tissue characteristics were not significantly different between treatment groups in this 52-week period.
Secondary outcomes revealed physiologically and structurally relevant changes with semaglutide. Semaglutide treatment was associated with a significantly reduced renal artery resistive index, suggesting decreased intrarenal vascular resistance. In addition, the apparent diffusion coefficient (ADC) was stabilized in the semaglutide group relative to placebo, an imaging signal interpreted as prevention or slowing of fibrosis progression. These secondary imaging and Doppler-based vascular findings point to hemodynamic and microstructural mechanisms that may not be captured by the coprimary MRI measures used.
Paired molecular analyses provided cell-type–resolved insights. SnRNAseq and spatial transcriptomics in paired biopsy samples identified pronounced effects of semaglutide on glomerular endothelial cells. Transcriptional programs in these endothelial cells were modulated by semaglutide treatment, and spatial analyses revealed a reduction in the number of immune cells in proximity to glomerular endothelial cells after semaglutide. Together, the molecular data indicate improved endothelial cell–related programs and reduced parenchymal immune cell proximity, consistent with decreased local inflammation or altered cell–cell interactions that may support renal protection.
Combining imaging, hemodynamic and molecular readouts, REMODEL suggests several putative mechanisms by which semaglutide may protect the kidney in people with T2D and CKD. The main signals reported were reduced renal vascular resistance (lower renal artery resistive index), stabilization of diffusion measures consistent with attenuated fibrotic progression, and cell-specific transcriptional changes in glomerular endothelial cells with reduced nearby immune cell presence. These observations align with preclinical data that GLP-1 receptor agonists can modulate inflammation, fibrosis and vascular function in the kidney.
REMODEL provides mechanistic context for the clinical kidney benefits observed in the FLOW outcome trial, which reported a 24% reduction in the prespecified composite kidney endpoint with semaglutide. REMODEL’s findings support a model in which semaglutide-mediated kidney protection involves vascular and cellular programs beyond glycemic and weight effects alone.
The coprimary MRI endpoints (R2*, perfusion, T1 mapping) did not change significantly with semaglutide, while secondary hemodynamic, diffusion and molecular measures did. The source article reports the trial design, participant numbers, principal imaging and molecular findings and contextualizes them against prior trials and preclinical work. Detailed statistical values, subgroup analyses and longer-term tissue outcomes beyond 52 weeks are reported in the full article but are not reproduced here. Further work will be needed to confirm causality, to determine the relative contribution of each mechanism to clinical outcomes and to explore interactions with background therapies such as SGLT2 inhibitors.