Valine, cell-resolved clocks, and a ceiling on how far this goes
The most useful item this cycle is a rodent diet study with an unusually clean number: restricting valine alone — not all three branched-chain amino acids — raised median male mouse lifespan 23% and cut frailty, cancer prevalence and senescent-cell burden in both sexes. The human-side news is measurement: plasma proteomics now resolves biological age at the level of individual cell types across 60,542 people, with extreme-decile effect sizes large enough to be clinically legible. On mechanism, two papers move the SASP from a cytokine story to a metabolic one and hand us a senomorphic route through an already-approved oncology drug. The corrective comes from npj Aging: even if every reversible hallmark were abolished, somatic mutation alone would cap median lifespan around 146–194 years — which both bounds escape-velocity rhetoric and implies mutation is only about half the problem.
median male mouse lifespan from lifelong dietary valine restriction, with reduced frailty, cancer prevalence and senescent-cell burden in both sexes (Nature Aging 2026)
Where the human evidence is
Count by statusEvidence-stage map
Evidence stage × investigator promise (1–10)Human biomarkers at cellular resolution
Plasma proteomics resolves the biological age of 40+ cell types in 60,542 people and predicts incident disease over 15 years
Ding et al. (Wyss-Coray lab) trained models on more than 7,000 plasma proteins in 60,542 individuals to estimate biological age for over 40 cell types spanning neuronal, glial, immune, endocrine, epithelial and musculoskeletal lineages. Between 20–25% of individuals showed accelerated aging confined to a single cell type and 1–3% across ten or more; signatures tracked disease status and predicted incident disease and mortality over 15 years of follow-up. Extreme-decile effects are large and clinically legible: extreme astrocyte aging tripled incident Alzheimer's risk in APOE4 homozygotes while youthful astrocytes lowered it, extremely aged skeletal myocytes carried a 12.7-fold higher ALS risk, and in smokers extreme respiratory epithelial aging added 58% to lung-cancer risk beyond smoking alone. This is observational cohort work — a risk-stratification framework, not a validated intervention endpoint, and tail effects will attenuate in unselected populations.
Read the paperFive-protein plasma signature predicts incident MASLD up to 16 years before clinical onset (>50,000 participants, four cohorts)
Yu, Chen, Feng, Li et al. identified and validated a five-protein panel — FUOM, ACY1, KRT18, CDHR2 and GGT1 — across more than 50,000 participants from Southern UK, Northern UK, EPIC-Norfolk and a Southern China inception cohort. The model achieved 5-year AUC 0.838 and 16.6-year AUC 0.756, with long-horizon performance sustained in EPIC-Norfolk (AUC 0.710) and 0.912 in the Chinese cohort; integrating routine clinical data raised these to 0.904 and 0.822. For a metabolic-health practice this is the most credible ultra-early MASLD stratification tool published to date. It remains an observational prediction model — no trial has shown that acting on the signature changes hepatic outcomes.
Read the papercGAS-STING, DNA sensing & the origins of inflammaging
FAD is an endogenous brake on cGAS and RIG-I; losing it drives sterile inflammation and cellular senescence
Wang et al. screened endogenous metabolites for inhibition of cytosolic nucleic-acid sensing and found flavin adenine dinucleotide binds directly into the catalytic pockets of both cGAS and RIG-I, blocking cGAMP production and ATPase activity respectively. Physiologically FAD prevents self-nucleic-acid-driven sterile inflammation; ablation of FAD synthase (FLAD1) exacerbated autoinflammation and cellular senescence, while reduced FLAD1 activity during viral infection releases the brake and amplifies type-I interferon signaling. This is the first metabolite-level rheostat identified on the cGAS-STING inflammaging axis this digest has been tracking, and it runs through riboflavin (vitamin B2) metabolism. Two cautions before anyone extrapolates to B2 supplementation: the causal manipulation here is intracellular FLAD1-dependent FAD availability in mice and cells, not oral riboflavin in humans; and raising FAD would be expected to blunt antiviral interferon as well as inflammaging — the same therapeutic-window problem as cGAS partial loss-of-function.
Read the paperHSC-intrinsic SIRT3 loss installs maladaptive 'trained immunity' that drives systemic inflammaging
Mu, Barthez et al. show the mitochondrial deacetylase SIRT3 — highly expressed in hematopoietic stem cells and reduced with age — suppresses an HSC response to aging that programs maladaptive trained immunity in myeloid progeny. Overexpressing SIRT3 in HSCs improved HSC function and, via myeloid cells with modulated inflammatory programs, attenuated age-related declines in cognition and motility in distant tissues. This relocates the origin of inflammaging upstream to an epigenetically trained stem-cell compartment rather than to senescent cells accumulating in the affected tissue — a plausible explanation for why tissue-directed senolytics have been inconsistent. Entirely murine and genetic: no small molecule, no human data, and SIRT3 is not addressable by NAD+ precursor supplementation in any demonstrated way.
Read the paperSenescence & the SASP — druggable, and double-edged
Abemaciclib suppresses the NF-κB arm of the SASP in pre-existing senescent cells — a senomorphic that does not require killing
Wang, Piccolantonio et al. report that short-term CDK4/6 inhibition with abemaciclib suppresses the established NF-κB-associated SASP (NASP) in cells that are already senescent, in vitro and in vivo, reducing the pro-tumorigenic activity of chemotherapy-induced senescent cells and improving physical function in mice after chemotherapy; genetic CDK4/6 knockdown phenocopied the effect. Mechanistically, CDK4/6 inhibition suppresses retinoic acid signaling — CDK4 and CDK6 interact with NF-κB and CDK4 additionally binds RARα, and abemaciclib disrupts both interactions. The RARα antagonist agn194310 reproduced the NASP reduction, and in aged mice both compounds lowered systemic NASP and improved physical performance. Notable because abemaciclib is an already-approved oncology agent, making this the most directly repurposable senomorphic lead in some time — but the aging data are murine, and CDK4/6 inhibitors carry non-trivial neutropenia, diarrhoea, ILD and VTE risk in humans. Not a rationale for off-label use.
Read the paperFructose is a metabolic component of the chemotherapy-induced SASP that promotes metastatic dissemination
Cole, Buj et al. show the cisplatin-induced SASP enhances detachment of high-grade serous ovarian cancer cells in vitro and dissemination in vivo, and identify fructose as a metabolic SASP component driving that detachment — a high-fructose diet increased dissemination in vivo. Mechanistically the effect runs through complex I and SASP-mediated inhibition of an NAD–SIRT–SREBP axis, lowering plasma-membrane cholesterol and loosening adhesion. Two things worth holding onto: the SASP is a metabolic phenotype and not only a cytokine one, and the NAD axis sits inside this pathway, which complicates any simple 'more NAD is better' framing in an oncology context. Preclinical ovarian-cancer models only — this is not a general claim about dietary fructose and cancer risk in humans.
Read the paper'Senotypes': a proposed taxonomy of senescent cells by origin, molecular features and adaptive vs maladaptive effect
Schafer, Basisty and Baker propose replacing the monolithic 'senescent cell' with senotypes — classes defined by inducing stimulus, molecular signature and downstream effect, explicitly including states that are adaptive rather than pathogenic. It publishes alongside the SenNet Collection mapping senescent cells across tissues, species and lifespan. Clinically this is the cleanest published articulation of why broad senolysis has produced inconsistent human results, including the null D+Q Alzheimer's biomarker data covered here previously: current agents do not distinguish maladaptive from beneficial senescent populations. Flagging plainly — a Perspective, not new data.
Read the paper