"Irreversible" damage turns out to be reversible — and biological age finally gets an absolute risk number
Two things worth your attention this cycle. First, a Nature Communications paper from Revel Pharmaceuticals and Calico reports an engineered enzyme (CMLase) that strips carboxymethyl-lysine off long-lived proteins in ex vivo human lens, skin and artery tissue — the first credible proof-of-concept that a glycation adduct long considered permanent is enzymatically repairable. Second, a JAMA pooled analysis across six cohorts converts plasma p-tau217 from a relative-risk marker into time-specific absolute risk, which is what a clinician actually needs at the bedside. Everything this week is peer-reviewed — no preprints, no company press releases dressed as data — but the balance is heavily preclinical or observational, and the one evidence-synthesis item is a corrective: a systematic review of 27 human nutrition trials finds essentially no reliable effect of dietary supplements on senescent-cell burden.
reduction in carboxymethyl-lysine glycation achieved by an engineered enzyme in ex vivo human lens, skin and artery tissue from elderly donors (Nature Communications 2026)
Where the human evidence is
Count by statusEvidence-stage map
Evidence stage × investigator promise (1–10)Human risk prediction gets quantitative
Plasma p-tau217 converted into time-specific absolute risk of cognitive impairment (6 cohorts, n=2,684)
Buckley et al. harmonized six longitudinal cohorts across North America, Japan and Australia and followed 2,684 cognitively unimpaired older adults (median age 69.6; 63% female) for a median 5.4 years, with 478 progression events. Each 1-SD increase in baseline plasma p-tau217 carried HR 1.38 (95% CI 1.30–1.46), and 1.32 (1.24–1.41) after adjustment including amyloid-PET Centiloids. Critically, they report absolute rather than relative risk: 24% (20–28%) 5-year risk of progression at high p-tau217 (1.1–2.4 SD) and 38% (33–43%) at very high (>2.5 SD). These are selected research cohorts, so the numbers will shift in unselected primary-care populations — but this is the first version of this biomarker that supports an actual prognostic conversation.
Read the paperProteomic age clocks predict 24 incident diseases and mortality across 28 years (EPIC, n=17,473)
Robinson et al. tested plasma SomaScan-based proteomic clocks, including organ-specific clocks, against 24 incident chronic diseases and all-cause mortality in 17,473 participants of the European Prospective Investigation into Cancer and Nutrition. A composite "global proteomic age gap" showed the strongest mortality association of any clock tested and tracked smoking, alcohol, physical inactivity, and risk of cardiovascular disease, dementia, and liver, upper aero-digestive, lung and kidney cancers. Organ-specific gaps were preferentially associated with cancers of the matching organ. Notable caveat from the authors themselves: mortality prediction was only comparable to classical lifestyle risk factors — the clocks add mechanistic resolution, not yet incremental predictive power.
Read the paperEpigenetic aging — an atlas, and a microbiome caveat
Cross-tissue methylation atlas (>15,000 profiles, 17 tissues) nominates NAD+ metabolism as the modifiable cluster
Jacques et al. (Eynon lab, with Horvath, Gladyshev, Teschendorff) meta-analyzed more than 15,000 human methylation profiles across 17 tissues, finding both conserved and tissue-specific aging signatures plus systemic shifts in methylation level, rising methylation variability and increasing molecular disorder. Network analysis separated tightly connected gene clusters that beneficial interventions do NOT move from a more tractable cluster linked to NAD+ metabolism, and identified the cell-adhesion gene PCDHGA1 as a conserved cross-tissue hub. Read the NAD+ finding as a network-level hypothesis about which parts of the epigenome are addressable at all — it is not evidence that NAD+ precursor supplementation changes human aging outcomes.
Read the paperGut microbiome composition tracks DunedinPACE — but not Horvath, Levine or GrimAge2
Paired 16S rRNA sequencing and DNA methylation from 123 monocyte-enriched samples in a cohort including Native Hawaiian and Pacific Islander participants were used to build "EpiBiome" models. Models predicting residuals of first- and second-generation clocks (Horvath, Levine, GrimAge2) showed no predictive signal at either taxonomic rank; only DunedinPACE was predictable (species-level R²=0.152, genus-level R²=0.099), and adding chronological age did not improve performance, indicating age-independence. Bifidobacterium adolescentis was the dominant SHAP contributor to decelerated aging, Succinivibrio dextrinosolvens to accelerated aging. Small, single-cohort, hypothesis-generating — and a useful reminder that "epigenetic age" is not one measurement.
Read the paperRepairing damage previously deemed irreversible
Engineered enzyme (CMLase) reverses carboxymethyl-lysine glycation in aged human tissue ex vivo
Trabosh et al. used directed evolution across more than 500 million variants of a bacterial glycine oxidase scaffold to engineer CMLase, an enzyme that specifically oxidizes Nε-carboxymethyl-lysine and restores the native lysine. CML is the most abundant stable AGE adduct in aged tissue and a RAGE ligand sustaining chronic inflammation; it has been treated as chemically irreversible. CMLase removed up to 97% of CML from model proteins and reduced CML burden by roughly 45–70% in ex vivo human lens, skin and artery tissue from elderly donors. This is bench-stage proof-of-concept in explanted tissue — there is no delivery route, no pharmacokinetics and no in vivo functional endpoint yet — but it is the strongest evidence to date that the damage-repair arm of geroscience is chemically tractable.
Read the paperSarcopenia reframed: neuromuscular junction transmission failure from NaV1.4 loss, reversed by ClC-1 inhibition
Weak older adults show measurable NMJ transmission failure that scales with weakness severity; aged rodents reproduce it, and the defect localizes to loss of muscle-fiber excitability at the junction rather than to cholinergic transmission. Immunohistochemistry across species pinned a localized reduction in the skeletal-muscle sodium channel NaV1.4 at the post-synaptic membrane, and acute NaV1.4 inhibition in adult rats phenocopied the aged deficit. Small-molecule inhibition of the muscle-specific ClC-1 chloride channel — an approach already shown to improve function in myasthenia gravis patients — enhanced contractile and motor function in aged, weak rats. A genuinely novel and druggable sarcopenia mechanism; the human data so far are observational, the intervention data rodent.
Read the paperImmune aging & inflammaging
GPR40 agonism (GW9508) restores thymic structure and function in 17-month-old mice
Li et al. report marked loss of GPR40 (FFAR1) expression in thymic epithelial cells from aged mice and in doxorubicin-induced senescent TECs, and show that the selective agonist GW9508 triggers Ca²⁺ influx and AMPK signaling while repressing the ERK1/2-MAPK cascade, restoring TEC viability and thymic architecture in 17-month-old mice via intraperitoneal dosing over a few weeks. A practical small-molecule route to thymic rejuvenation would matter a great deal, since FOXN1 gene therapy and KGF have both foundered on delivery. Flagging clearly: group sizes were as low as 3–5 animals for some readouts and the dose-response was non-monotonic. This needs replication at 12+ per group before it should change anyone's model.
Read the paperPeripheral inflammaging reaches the brain via DNA-carrying extracellular vesicles and STING
Analyzing Parkinson's patients and LRRK2 gain-of-function (G2019S) mice, the authors argue PD behaves as an accelerated-aging disorder driven by STING-dependent inflammation that starts peripherally, disrupts the blood-brain barrier and produces dopaminergic neurodegeneration. Mechanistically, aging or LRRK2-GoF causes endolysosomal decline, cytosolic self-DNA accumulation, and release of DNA-containing extracellular vesicles that activate cGAS-STING within and between cells. This extends the cGAS-STING inflammaging thread this digest has tracked into a long-range, EV-mediated body-to-brain route — and it sharpens the same therapeutic-window question: damp STING signaling without disabling antiviral defense.
Read the paper