Skip to main content

Endocrinology Research Analysis

3 papers

In May 2025, endocrinology research emphasized translational advances spanning metabolic liver disease and diabetic complications. A pivotal phase 3 RCT showed that once-weekly semaglutide 2.4 mg achieved histologic benefits in MASH with F2–F3 fibrosis. Complementary mechanistic work mapped a mycobiome-driven CerS6–ceramide axis as a druggable pathway in steatohepatitis. In diagnostics, a multi-ethnic, prospectively tested deep-learning system using retinal images enabled scalable detection and

Summary

In May 2025, endocrinology research emphasized translational advances spanning metabolic liver disease and diabetic complications. A pivotal phase 3 RCT showed that once-weekly semaglutide 2.4 mg achieved histologic benefits in MASH with F2–F3 fibrosis. Complementary mechanistic work mapped a mycobiome-driven CerS6–ceramide axis as a druggable pathway in steatohepatitis. In diagnostics, a multi-ethnic, prospectively tested deep-learning system using retinal images enabled scalable detection and triage of diabetic kidney disease. Collectively, the month underscored weight-centric and microbiome-lipid interventions plus noninvasive AI tools for risk stratification.

Selected Articles

1. Phase 3 Trial of Semaglutide in Metabolic Dysfunction-Associated Steatohepatitis.

88.5The New England journal of medicine · 2025PMID: 40305708

A 72-week interim analysis of a multicenter, double-blind phase 3 RCT in biopsy-proven MASH (F2–F3) showed once-weekly semaglutide 2.4 mg significantly increased rates of steatohepatitis resolution and fibrosis improvement versus placebo, with substantial weight loss; GI adverse events were more frequent with semaglutide.

Impact: First large phase 3 RCT to demonstrate histologic benefit of a GLP-1RA in MASH, addressing a major unmet need and potentially redefining standard of care pending long-term outcomes.

Clinical Implications: Semaglutide 2.4 mg weekly may be considered for biopsy-proven MASH with F2–F3 fibrosis to achieve NASH resolution and fibrosis benefit; monitor GI tolerability and await longer-term data and regulatory guidance for routine adoption.

Key Findings

  • Steatohepatitis resolution without fibrosis worsening: 62.9% vs 34.3% (P<0.001).
  • Fibrosis improvement without steatohepatitis worsening: 36.8% vs 22.4% (P<0.001).
  • Mean body weight change: −10.5% with semaglutide vs −2.0% with placebo; GI AEs more common with semaglutide.

2. A symbiotic filamentous gut fungus ameliorates MASH via a secondary metabolite-CerS6-ceramide axis.

87Science (New York, N.Y.) · 2025PMID: 40310917

Preclinical work identified a symbiotic gut filamentous fungus whose secondary metabolite modulates host CerS6–ceramide signaling to ameliorate MASH, establishing a causal mycobiome–lipid axis and nominating metabolite- and fungus-targeted therapies.

Impact: Expands the therapeutic landscape from bacteria-focused microbiome concepts to a defined mycobiome–sphingolipid pathway with translational potential.

Clinical Implications: Motivates development of mycobiome- or metabolite-based diagnostics and CerS6-modulating therapeutics; requires human validation and biomarker assay development.

Key Findings

  • Isolation and characterization of a gut filamentous fungus linked to improved MASH phenotypes.
  • A fungal secondary metabolite attenuates steatohepatitis via CerS6–ceramide signaling.
  • Causal linkage between a mycobiome member and host sphingolipid metabolism.

3. Non-invasive biopsy diagnosis of diabetic kidney disease via deep learning applied to retinal images: a population-based study.

84.5The Lancet. Digital health · 2025PMID: 40312169

DeepDKD, a retinal image–based deep learning system pretrained on >700,000 images and externally validated across multi-ethnic cohorts, detected DKD with AUCs ~0.79–0.84 and distinguished diabetic from non-diabetic nephropathy; prospective and 4.6-year longitudinal analyses showed improved sensitivity and divergent renal outcomes by AI classification.

Impact: Demonstrates scalable, multi-ethnic, prospectively validated AI for noninvasive DKD screening and biopsy-decision support.

Clinical Implications: Integrate retinal-image AI with albuminuria and eGFR screening to prioritize referrals, intensify renoprotective therapy, and flag suspected non-diabetic pathology; implementation trials should assess outcomes and fairness.

Key Findings

  • Internal DKD AUC 0.842; external AUCs 0.791–0.826 across multi-ethnic datasets.
  • Differentiation of isolated diabetic nephropathy vs NDKD: internal AUC 0.906; external 0.733–0.844.
  • Prospective study sensitivity 89.8% vs 66.3% vs metadata model; 4.6-year analysis showed divergent eGFR decline by AI-defined groups.