A STING-knockout lifespan preprint, an immune-aging map, and two reasons to be careful with GLP-1 clock claims
The most provocative mechanistic item this week is not yet peer reviewed: a bioRxiv preprint from the Gorbunova/Seluanov/Sedivy groups reports that germline STING deletion extends median lifespan in mice. The best human data are observational: a 12.4-million-cell Immunity atlas links the granzyme B to granzyme K effector-memory CD8 T-cell ratio to mortality in UK Biobank. Meanwhile, Lilly and Novo presented unpublished GLP-1 aging-clock data at ARDD, and a separate analysis argues that single-clock intervention results should be trusted only when specific criteria are met. Everything on the preclinical side remains mouse or cell work.
age-unit epigenetic clocks reached significance in Lilly's single-arm tirzepatide sub-study (n=71 paired; unpublished, ARDD 2026) although all 15 trended slower than elapsed time
Where the human evidence is
Count by statusEvidence-stage map
Evidence stage × investigator promise (1–10)GLP-1s and the validity of aging clocks
Lilly and Novo present unpublished GLP-1 aging-clock data at ARDD 2026
Lilly reported an epigenetic sub-study within SURMOUNT-5 analysing the tirzepatide arm only (71 participants with paired baseline and week-72 samples, no placebo or control arm). All 15 age-unit clocks moved less than the roughly 1.38 years of elapsed time, but only two reached statistical significance (median gains of about 0.7 and 1.1 years), and DunedinPACE fell from a median of 1.13 to 1.08; Lilly itself called the work preliminary and hypothesis-generating. Novo presented proteomic-clock data for semaglutide that press accounts summarize as roughly 2 to 3 years of slower biological age (up to about 4 years for a heart-protein clock); the trial identity, sample size and effect sizes differ between accounts and could not be verified. Clock readouts are not clinical outcomes, and the size of effect varied widely by clock and organ.
Read the paperExtending single-clock intervention studies to more epigenetic clocks adds no significant results; three criteria proposed for trusting a clock result
Per a secondary summary, the authors re-analysed previously published single-clock intervention studies with additional clock models and found that no additional clocks reached statistical significance. Using positive-control datasets of age-accelerating events they propose that a result earns more confidence when reliable principal-component-based clocks respond, when mortality-trained or pace-of-aging clocks respond, and when multiple clocks move in the same direction. Some single-clock findings may therefore deserve re-examination. This is directly relevant to reading the GLP-1 clock claims above; the primary paper's title, authors and sample sizes were not verified.
Read the paperNeuro-longevity: lipids, myelin and resilience
APOE4 suppresses Wnt signaling in oligodendrocytes; GSK3b inhibition reduces lipid droplets and improves myelination in models
A genome-wide CRISPR screen in APOE4 human oligodendrocytes implicated Wnt signaling; APOE4 cells had less beta-catenin and more GSK3b, and disrupting Wnt in either direction caused lipid droplet accumulation. GSK3b inhibition with the research compound CHIR99021 reduced lipid droplets, and in organoids and in APOE4 plus tau-pathology mice it increased myelination. Caveats: preclinical only, no cognitive outcomes reported from the treated mice, the tau model does not replicate human Alzheimer disease, and CHIR99021 is a tool compound rather than a therapeutic candidate. Link points to the MIT News summary; the primary DOI was not verified.
Read the paperPla2g2f, an aging-sensitive secreted phospholipase, confers neuroprotection and cognitive resilience in mice
Pla2g2f rises in the aging hippocampus and with inflammation; reducing it worsened synapse loss, inflammation, lipid-droplet accumulation and memory, while increasing it protected against many of these changes, including when boosted in middle-aged mice. In human microglia, higher activity reduced lipid droplet and cholesterol-ester accumulation, and expression was associated with cognitive resilience in older adults with Alzheimer-type pathology (association only). No safe way to raise it in patients exists yet, and sample sizes and effect sizes were not available from the sources checked.
Read the paper