Semaglutide slowed epigenetic aging by up to 9% in a randomized trial in people with HIV

GLP-1 receptor agonists such as semaglutide are widely used for weight loss and cardiometabolic protection, and interest is growing in whether they act as gerotherapeutics that slow biological aging. Until now, randomized trial evidence on their effects on aging biomarkers has been absent. People with HIV are a useful model because they show accelerated biological aging, persistent low-grade inflammation, and metabolic dysfunction even when antiretroviral therapy suppresses the virus. HIV-associated lipohypertrophy, an excess of visceral and ectopic fat, may compound these processes.

This was a post hoc, exploratory epigenetic analysis of a 32-week, randomized, double-blind, placebo-controlled phase 2b trial (NCT04019197). The parent trial randomized 108 adults 1:1 to once-weekly subcutaneous semaglutide (8-week titration, then 24 weeks at 1.0 mg) or placebo. Eight participants (15%) in each arm withdrew early. Paired peripheral blood mononuclear cell DNA methylation profiles at baseline and week 32 were available for 84 participants (45 semaglutide, 39 placebo). The team computed 17 first-, second-, and third-generation clocks, including principal-component versions designed for better reliability in longitudinal trials.

Participants averaged 49 years old, 42% were women, 58% were Black, and median BMI was 32.9 kg/m². Median HOMA-IR was 2.9, median hsCRP was 4.1, and about 35% were current smokers. Antiretroviral therapy had lasted roughly 14 years on average. The arms were well matched, although men made up 67% of the semaglutide group versus 49% of the placebo group. Baseline epigenetic and chronological ages correlated strongly (r = 0.83–0.99).

In adjusted analyses, semaglutide reduced epigenetic aging across multiple second- and third-generation clocks relative to placebo. PhenoAge fell by 4.9 years/year (p = 0.004), PCGrimAge by 3.1 (p = 0.007), GrimAge V2 by 2.3 (p = 0.009), OMICmAge by 2.2 (p = 0.009), and RetroAge by 2.2 (p = 0.030). DunedinPACE, which measures pace of aging, dropped by 0.09 units, equivalent to a 9% slower pace (p = 0.01). Systems-based clocks showed parallel reductions in inflammation, brain, and heart aging measures.

These are the first randomized-trial data suggesting that a licensed GLP-1 receptor agonist can shift epigenetic aging biomarkers. The clocks showing effects are those trained on mortality, morbidity, and physiological phenotypes, so the signal may be clinically relevant. The paper does not establish whether the change reflects direct effects on aging biology or secondary effects of reduced visceral fat and inflammation. Epigenetic clocks are surrogate markers, and slower clock readings have not been shown to translate into longer healthspan.

The authors stress important limitations. Epigenetic aging was not pre-specified, and the analysis was post hoc with a modest sample. The cohort was HIV-specific, mostly obese, and followed for only 32 weeks, which limits generalizability. Prospective, adequately powered, mechanism-focused trials are needed before GLP-1 drugs can be considered gerotherapeutics.