HIV trial points to semaglutide as a new anti-aging contender

By Vijay Kumar MalesuReviewed by Susha Cheriyedath, M.Sc.Aug 4 2025

A diabetes drug may hold the key to slowing the aging process. New research finds semaglutide decelerates biological aging in people with HIV, sparking hope for broader longevity breakthroughs.

Study: Semaglutide Slows Epigenetic Aging in People with HIV-associated lipohypertrophy: Evidence from a Randomized Controlled Trial. Image Credit: myskin / Shutterstock

*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.

In a study posted to the medRxiv preprint* server, researchers tested whether once-weekly semaglutide, a glucagon-like peptide-1 (GLP-1) receptor agonist, slows biological aging measured by deoxyribonucleic acid (DNA) methylation–based epigenetic clocks in adults with Human Immunodeficiency Virus (HIV)-associated lipohypertrophy.

This study provides the first randomized clinical trial evidence that a GLP-1 receptor agonist can modulate epigenetic biomarkers of aging in this population. These results represent a post hoc, exploratory analysis of a previously completed randomized controlled trial.

Background

Aging underlies most chronic diseases, so interventions that slow biological aging have outsized public-health value. People living with HIV often show accelerated aging and persistent low-grade inflammation despite effective antiretroviral therapy (ART), elevating risks for cardiovascular, hepatic, and neurocognitive complications. GLP-1 receptor agonists such as semaglutide reduce body weight, visceral adiposity, and cardiometabolic risk, outcomes that correlate with epigenetic age acceleration.

Whether GLP-1 receptor agonists also modulate DNA methylation-based “epigenetic clocks” remains uncertain. If confirmed, a familiar therapy might serve as a health-span agent. Further research is needed to test durability, clinical outcomes, and generalizability.

These questions are particularly relevant for populations, such as people living with HIV and lipohypertrophy, who are at increased risk of accelerated biological aging. The HIV population is considered an “ideal model” for geroscience research, as they experience features of accelerated aging and persistent inflammation despite effective viral suppression.

About the Study

This single-center, double-blind, randomized, placebo-controlled, phase 2b trial enrolled adults with documented HIV-1 infection who were on stable ART, had controlled HIV-1 ribonucleic acid (RNA) (<400 copies/mL), a body mass index (BMI) ≥25 kg/m², and clinically defined lipohypertrophy.

Individuals with diabetes mellitus or established cardiovascular disease were excluded. Participants were randomized 1:1 to once-weekly subcutaneous semaglutide (dose escalation from 0.25 mg to 0.5 mg to 1.0 mg, then maintained at 1.0 mg through week 32) or matching placebo; all injections were administered in clinic to ensure adherence and blinding.

Of 108 randomized participants, 84 (semaglutide n=45; placebo n=39) contributed paired samples for epigenetic analyses. The parent trial’s primary outcomes assessed adipose distribution using computed tomography (CT) and dual-energy X-ray absorptiometry (DEXA). The present exploratory, post hoc analysis focused on epigenetic aging.

Peripheral blood mononuclear cells (PBMCs) were collected at baseline and week 32. DNA was extracted and profiled on the Infinium MethylationEPICV2 BeadChip, quantifying over 850,000 cytosine-phosphate-guanine (CpG) sites. Quality control employed the minfi pipeline and Enmix; all samples passed prespecified checks.

Multiple epigenetic clocks were evaluated: first-generation (Horvath, Hannum), second-generation (PhenoAge, GrimAge, including principal-component GrimAge (PCGrimAge), a mortality-linked estimator), the third-generation DunedinPACE, which indexes the “pace of aging,” and OMICmAge (multi-omic methylation age). Principal-component (“PC”) versions were used to enhance longitudinal reliability. In parallel, Biolearn implementations of GrimAge V1 and V2 were applied to provide methodological triangulation.

Annualized changes were compared between groups using analysis of covariance (ANCOVA) with adjustment for baseline clock value, chronological age, sex, BMI, high-sensitivity C-reactive protein (hsCRP), and soluble cluster of differentiation 163 (sCD163). Sensitivity analyses with alternative covariates confirmed the robustness of the findings.

Study Results

Semaglutide significantly slowed epigenetic aging across multiple independent measures relative to placebo. In adjusted models, PCGrimAge increased by 3.08 fewer “years per year” in the semaglutide arm (95% confidence interval (CI)= –5.29 to –0.86; p=0.007), indicating a notable deceleration in mortality-linked epigenetic aging. DunedinPACE declined by 0.09 units (approximately 9% slower; p=0.01), consistent with a biologically meaningful reduction in the rate at which physiological systems accrue age-related changes. PhenoAge decreased by 4.90 years per year (p=0.004), and OMICmAge fell by 2.20 years per year (p=0.009), reinforcing a broad signal of age deceleration.

Additional clocks pointed in the same direction. RetroClock, which emphasizes retrotransposon-associated methylation patterns implicated in aging, dropped by 2.18 years per year (p=0.030). Using an orthogonal computational pipeline (Biolearn), semaglutide also reduced GrimAge V1 by 1.39 years and GrimAge V2 by 2.26 years versus placebo after covariate adjustment (both statistically significant), corroborating the mortality-risk signal captured by second-generation clocks and reducing concerns that results reflect a single analytic approach.

System-specific “organ clocks” shifted favorably as well. These epigenetic clocks are designed to predict both all-cause mortality and organ-specific functional decline. Adjusted differences favored semaglutide in Blood (–4.37 years; p=0.011), Brain (–4.99; p=0.0049), and Inflammation (–5.01; p=0.0056), with meaningful reductions in Heart (–4.34; p=0.0088), Kidney (–4.20; p=0.014), Liver (–4.19; p=0.042), and Metabolic (–4.72; p=0.0090) domains.

Lung, Hormone, Immune, and Musculoskeletal clocks trended lower but did not achieve statistical significance. This multi-organ pattern aligns with clinical priorities: slowing inflammatory and cardiovascular aging could translate into fewer hospitalizations, improved cognition, and better functional status.

Not all biomarkers changed, as AdaptAge, CausAge, and DamAge exhibited small, heterogeneous, non-significant shifts, and the Intrinsic Capacity (IC) clock, linked to physical and cognitive resilience, did not differ between groups (estimated group difference: 0.003 IC score; p=0.31). These null findings set realistic expectations and suggest that certain aging axes may require more prolonged exposure, different doses, or larger samples to detect change.

Baseline characteristics of the epigenetic subset were balanced and representative of the target population: mean age approximately 49 years; 42% women; racially diverse; durable virologic suppression; and obesity (median BMI approximately 32.9 kg/m²). Low-grade inflammation (median hsCRP approximately 4.1 μg/mL) and immune activation markers were common, underscoring the clinical relevance of slowing biological aging in this setting.

Analytically, the primary models quantified annualized change from baseline to week 32 and adjusted for age, sex, BMI, hsCRP, and sCD163.

Sensitivity analyses substituting visceral fat mass for BMI, or interleukin-6 or soluble CD14 for hsCRP, yielded consistent results, strengthening confidence that effects were not driven by baseline imbalances or a single inflammatory marker.

Finally, correlations linked epigenetic aging measures with adiposity, supporting a mechanistic pathway whereby central fat and inflammation accelerate biological age, an axis that GLP-1 receptor agonists are designed to modify. In this analysis, correlations between central adiposity and epigenetic aging were generally stronger than those with systemic inflammation.

The authors also discuss the possibility that semaglutide may disrupt “obesogenic epigenetic memory” in adipose tissue, a proposed mechanism by which adipose tissue retains aging-related DNA methylation signatures even after weight loss.

Conclusions

Among adults with HIV on ART and lipohypertrophy, semaglutide (GLP-1) slowed DNA methylation aging across PCGrimAge, PhenoAge, DunedinPACE, and organ clocks, while IC was unchanged. Effects persisted after adjusting for BMI, hsCRP, and sCD163 and replicated with Biolearn GrimAge.

These shifts may improve outcomes and justify larger trials testing durability beyond 32 weeks and generalizability, including effects on mortality benefits. It remains unknown whether these results will generalize to people without HIV or without lipohypertrophy, and the findings should be considered preliminary pending further peer review and clinical outcome studies.

*Important notice: medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be regarded as conclusive, guide clinical practice/health-related behavior, or treated as established information.