Nothing this week changes the basics of eating well, moving and sleeping, but one item is worth acting on today. Chemists who tested products sold as the herbal supplement cat's claw found undeclared kratom alkaloids, which act on opioid receptors, and tianeptine. If a bottle like that is in your cupboard, stop taking it. A new antibody also protected muscle during GLP-1 weight loss, but the result was lean mass, not strength. And one aging clock rose with tumour stage in cancer patients -- the number may be reading the disease, not the aging. The results worth waiting for measure strength, walking speed and fractures.

TAKE

Products sold as cat's claw contained undeclared kratom alkaloids and tianeptine

Who: Products marketed as the herbal supplement cat's claw (Uncaria tomentosa). How many: Number of products tested not stated. How long: A single survey, no follow-up. Design: Cross-sectional analytical chemistry survey of marketed products: the tablets were tested, not people. Measured: What the tablets contained, not any effect in users, STAND-IN.

A supplement label is the only information most buyers have about what is in the bottle. Here it was wrong.

Chemists analysing products marketed as cat's claw found undeclared kratom alkaloids, including the opioid-active 7-hydroxymitragynine, and tianeptine. The work followed US reports of overdoses and deaths linked to mislabelled tablets. This is a chemical survey of products rather than a study in people, so no evidence stage is assignable. The analysis is a preprint and has not been peer reviewed.

Nobody counted how many users were exposed or what happened to them. The problem sits in the product category and its supply chain, not in one brand a single recall would fix. (Measurement)¹

Harms: Reported. The analysis followed US overdose and death reports linked to mislabelled tablets, and one compound found, 7-hydroxymitragynine, acts on opioid receptors.

What would move this: Testing that named which brands and batches were contaminated, and how common contamination is across products on sale.

DOES THIS CHANGE WHAT I SHOULD DO?

Yes. If a cat's claw product is in your cupboard, stop taking it. How many products on sale are affected is unreported, so the label cannot tell you which bottles are clean. If you have been using one and feel unwell, tell your doctor exactly what you took.

Regeneron says trevogrumab preserved muscle mass during GLP-1 weight loss, and nobody reported strength

Who: Adults losing weight on a GLP-1 drug (Lab)². How many: Not disclosed in the announcement. How long: Not disclosed. Design: Company-reported Phase 2 topline, through press coverage, not peer-reviewed. Stage not assignable: no trial design or results have been published. Measured: Lean and muscle mass, STAND-IN.

Funding flag: The result comes from the party with a commercial stake in the drug, not from an independent analysis.

If you are losing weight on a GLP-1 drug, some of what you lose is muscle. What matters is whether an add-on keeps you climbing stairs and carrying shopping.

Regeneron says trevogrumab, an antibody aimed at the myostatin pathway, preserved lean and muscle mass during GLP-1 weight loss, without meaningfully increasing total weight loss. The announcement is the company's own and has not been peer-reviewed. No strength, function or fracture result is reported, and a review the same week adds synthesis rather than data.

India's food regulator issued a notice over lean-muscle-preservation claims on Celevida GLP+, a Dr Reddy's and Nestle Health Science product sold alongside GLP-1 therapy. A prescription-drug result does not license a supplement claim. In the US, BodySpec has opened an Oakland storefront selling 12-minute DEXA scans for under $60, which report lean mass, not strength. (People, tested on a stand-in, Review, Not a study)^3–6

Harms: None reported. The company's announcement carries no adverse-event data, which is an absence of reporting rather than a safety record. (none reported)

What would move this: A trial showing people on the drug walked faster, carried more or broke fewer bones than those on a GLP-1 drug alone.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. Ask your doctor to track your strength and walking speed, not only the scale. No study tested whether training or protein changes what GLP-1 users lose, and trevogrumab remains an experimental asset with no strength or fracture result reported.

MEASURE

A PhenoAge clock reading rose with tumour stage, and may be tracking cancer rather than aging

Who: Patients undergoing primary tumour resection. How many: 3,067. How long: Not stated. Design: Observational cohort, biomarker clock construct-validity analysis, People, watched: watched, not tested. A preprint, not yet peer reviewed.. Measured: PhenoAge acceleration against tumour stage, STAND-IN.

• patients undergoing tumour resection: 3,067
• people in the kidney-disease clock analysis: 1,259

Biomarkers flag: The clock reading stands in for the rate of aging, and here it moved with tumour stage instead.

A biological age test promises to tell you how fast you are aging, which is why people pay for one. This week shows what such a number can quietly be measuring instead.

In 3,067 patients having a primary tumour removed, PhenoAge acceleration rose with tumour stage. The authors conclude it may track cancer burden rather than biological aging. These patients were observed during routine surgical care, not assigned to anything.

Two other papers this week used the same kind of clock in people already ill: one in diabetic kidney disease, the other in head and neck cancer. If the clock moves with disease, these studies may be counting the disease twice. (Measurement, People, watched)^7–9

Harms: Not measured. None of these studies followed what patients did after being given a biological age number.

What would move this: A trial that moved people's clock readings with treatment and then showed fewer deaths, fractures or hospital admissions.

DOES THIS CHANGE WHAT I SHOULD DO?

Probably. If you were about to pay for a biological age test, keep the money. With an active illness, the number may be reading the illness. None of this week's clocks were tested against a clinical outcome, and the tumour-stage result is one unreplicated preprint.

WHAT DIDN'T MOVE

Cold exposure.

Cold extended lifespan in worms through a biopterin signal passed between tissues (Animals). No human exposure, dose or outcome was tested, and this one study reached us under three headlines. (Animals)¹⁰

Sun exposure.

A reanalysis of existing human skin data argues non-burning sun may not measurably accelerate skin aging. Two parameters, no new data, nothing about skin cancer (People, watched). Nothing here supports changing sun-protection behaviour. (People, watched)¹¹

Longevity clinics.

Large health systems are opening longevity programmes (Not a study)¹², and at the fourth Roundtable of Longevity Clinics the sector's own leaders argued for moving from luxury diagnostics to standardised care (Not a study)¹³. A hospital logo on a longevity panel does not change what the panel predicts. (Not a study)^12,13

AYUSH patent filings.

The Ministry of AYUSH reported 4,077 traditional-medicine patent applications filed between ^15,16. Filing activity is not evidence of efficacy, safety, approval or a product on a shelf. (Not a study)¹⁴

THE THREAD THIS WEEK

Three items turn on a cheap measurement standing in for the thing you actually want to know: lean mass for strength, a clock for aging, a label for what is in the bottle.

Who: Regeneron, with BodySpec selling the measurement (Not a study)¹. What: A Phase 2 topline (COURAGE) and a retail scan launch (People, tested on a stand-in, Not a study)^1,2. How much: No trial numbers released; scans under $60, about 12 minutes (Not a study)¹. Evidence: Stage not assignable: a company topline carried by press coverage (People, tested on a stand-in)², and a retail scan launch (Not a study)¹. Product, price: BodySpec DEXA body-composition scan, under $60 (Not a study)¹; trevogrumab has no approved product or price.

• scan price: under $60 (Not a study)¹
• functional endpoints reported: none (People, tested on a stand-in)²

Biomarkers flag: Lean mass stands in for strength, walking speed and fracture risk, none reported (People, tested on a stand-in)².

Regeneron reported Phase 2 COURAGE topline results for trevogrumab, an antibody against the myostatin pathway. The company says it preserved lean and muscle mass during GLP-1-induced weight loss, without meaningfully increasing total weight loss. No effect size was released, and the result reached us through press coverage.

The adjunct is now a named Phase 2 asset rather than a thesis, and BodySpec is selling the matching measurement: DEXA body-composition scans from an Oakland storefront, aimed at GLP-1 users worried about muscle.

The indication's yardstick is unstable: repeating a grip-strength measurement changes which patients count as low-strength under the 2025 Asian Working Group for Sarcopenia criteria. No study design is in the available reporting (Measurement)³, so no evidence stage is assignable. (People, tested on a stand-in, Not a study, Measurement)^1–3

Harms: No adverse-event data accompany the topline, and no source reports whether preserved mass prevented a fall (People, tested on a stand-in)². No harms, outcome or interpretation data accompany the retail scan either: a lean-mass number does not tell a buyer whether strength or function changed (Not a study)¹.

What would move this: A peer-reviewed COURAGE report carrying strength, gait-speed or fracture outcomes alongside the mass data.

DOES THIS CHANGE WHAT GETS BUILT, FUNDED, APPROVED OR PRICED?

Maybe. Funders now have a named Phase 2 asset in the muscle-preservation class. With no effect size, no peer-reviewed report and no strength or fracture outcome released, there is nothing yet to price against, and prescribers get nothing (People, tested on a stand-in)².

Longevity clinic leaders called for standardised care as health systems open their own programmes

Who: New York Times reporting, a Roundtable of Longevity Clinics discussion, CNBC (Not a study)^4–6. What: Market structure: health-system programmes, a standardisation push, concierge pricing (Not a study)^4–6. How much: Up to $250,000 a year per patient, per CNBC (Not a study)⁶. Evidence: Stage not assignable: press reports, a conference discussion and a commentary carry no study design (Not a study)^4–7. Product, price: Concierge longevity care, up to $250,000 a year (Not a study)⁶; no other price list appears.

concierge spend: up to $250,000 a year (Not a study)⁶

The New York Times reported that longevity medicine is gaining credibility as major health systems open their own programmes. At the fourth Roundtable of Longevity Clinics, chaired by Buck Institute CEO Eric Verdin, clinic leaders argued for moving from luxury diagnostics toward standardised care.

A sector asking for standards is telling you its current offer is not standard. CNBC reported some wealthy Americans spending up to $250,000 a year on concierge care, while a commentary argues hospital longevity clinics monetise time and attention rather than proven medicine.

Distribution widened, not prediction: none of the four sources carries outcome data, and a teaching-hospital logo does not make a panel predict more. (Not a study)^4–7

Harms: None of the four sources reports harms, testing cascades or the cost of chasing incidental findings (Not a study)^4,5.

What would move this: A published outcome dataset from one of these programmes: who was tested, what changed, at what cost per patient (Not a study)⁵.

DOES THIS CHANGE WHAT GETS BUILT, FUNDED, APPROVED OR PRICED?

Maybe. Health systems and clinic operators are building on this; patients paying up to $250,000 a year still buy a panel with no outcome data behind it (Not a study)⁶.

CONTRADICTIONS

Is epigenetic age acceleration measuring aging, or the disease the patient already has?

Reading A: It is tracking disease burden. PhenoAge acceleration rose with tumour stage in 3,067 patients undergoing primary tumour resection (People, watched, observational cohort biomarker clock construct-validity analysis, preprint). The authors conclude it may track cancer burden rather than biological aging, which would make that readout hard to interpret in a patient with active disease, on one unreplicated preprint.

Reading B: It is tracking aging, and disease is how aging shows up. Epigenetic age acceleration tracked diabetic kidney disease progression in an observational cohort comparing 10 epigenetic clocks across two cohorts, n=1259 (People, watched). Pretreatment blood DNA methylation profiling was associated with survival on immune checkpoint inhibitors in a retrospective observational cohort of 100 head and neck cancer patients (People, watched). If aging drives disease, these correlations are expected.

What would settle it: A clock measured before disease onset that predicts incident disease, or a clock moved in a trial with a clinical endpoint. Neither appears in this week's record. (Measurement, People, watched)^8–10

BUSINESS SIGNALS

Regulation:

The Trump administration ordered a new NIH division and task force on women's health research; no budget, remit or timeline reported. (Not a study)¹¹

Money:

Mark Cuban's Cost Plus Drugs says it can lower manufacturing costs for custom rare-disease drugs and n-of-1 gene therapies; no figures or contracts reported. (Not a study)¹²

Pipeline:

XellSmart presented Phase I data for XS411, an off-the-shelf cell therapy, in nine people with moderate-to-advanced Parkinson's: single-arm, open-label, 3+3 dose escalation (People, tested on a stand-in). (People, tested on a stand-in)¹³

Pipeline:

SBC Medical says it has begun developing a DNA methylation biological age test, disclosing no validation data, cohort or timeline. Stage not assignable: a stated development intention with no study design in the available reporting. (Not a study)¹⁴

Pipeline:

A randomised, blinded, controlled Phase I/II trial was registered for SYS6052, a recombinant CHO-cell herpes zoster vaccine, in adults 40 and above; stage not assignable. (Not a study)¹⁵

Pipeline:

Genflow reported that its gene therapy shifted an aging-clock biomarker in dogs; company-reported and unpublished (Animals). (Animals)¹⁶

Pipeline:

Elysium launched Balance, adding S-equol and ergothioneine to Basis for menopause symptoms, citing its own open-label research; stage not assignable, no harms data. (Not a study)¹⁷

INDIA WATCH

• FSSAI issued a notice to the Dr Reddy's and Nestle Health Science product Celevida GLP+ over lean-muscle-preservation claims made alongside GLP-1 therapy. (Not a study)¹⁸
• Mapmygenome acquired Canadian genetic testing firm dnaPower to enter North America. No deal value disclosed; consumer genetic testing carries no clinical outcome evidence. (Not a study)¹⁹
• India's Clinical Establishment Rules tier diagnostic labs as basic composite, medium or advanced. Stricter standards for CGHS-empanelled labs have drawn a double-standards complaint. (Not a study)²⁰
• The Ministry of AYUSH reported 4,077 traditional-medicine patent applications filed between ^30,31. A filing count is not efficacy, safety or approval. (Not a study)²¹

WATCHING

Nothing this week changes what you should do. The headline you may have seen: a US service will now set your GLP-1 plan against a blood test that scores how old your body is, and retest you every three months. Those scores do carry information about groups of people. But a study using inherited genetic differences as a natural test found no sign that DNA-mark or blood-protein age scores cause scarring lung disease. Genetically short telomeres did raise its risk. If you are offered one, ask what the clinic would do differently if your number fell. Two supplement reports are worth knowing before your next purchase: the collagen case remains unmade, and a belly-fat extract missed the target it set out to hit. And if anabolic steroids are in your gym conversation, the heart findings in people who had already stopped are the part to raise with a doctor.

TAKE

Steroid users' hearts were still underperforming after they stopped

Who: People using anabolic-androgenic steroids outside medical care, compared with non-users. How many: Not stated. How long: Not reported. Design: Systematic review and meta-analysis of comparative cohort studies: users watched, not randomised. Measured: All-cause mortality, OUTCOME; heart pumping measures, STAND-IN.

Anabolic steroids get discussed in gyms in terms of what they do while you are on them. What happens after you stop is the part users hear least about.

A pooled analysis of cohort studies comparing non-medical users with non-users found excess deaths from all causes, and weaker heart pumping in people who had already quit. Nobody was randomised, so training habits, other drugs taken alongside, and who starts using at all are tangled into the numbers.

That tangle could inflate a mortality figure. It could equally produce the weaker heart pumping seen in former users: nobody measured their hearts before they started. (Review)¹

Harms: Reported: the affected readings are global longitudinal strain and ejection fraction, two measures of how well the heart squeezes, and the review formally assessed its own risk of bias.

What would move this: Prospective follow-up after people stop, with imaging and events, showing heart function returning to normal.

DOES THIS CHANGE WHAT I SHOULD DO?

Maybe. Nothing here shows that checking a former user's heart function changes anything, so tell a doctor you have used these drugs rather than waiting for symptoms.

MOVE

Training frequency mattered less than weekly volume for muscle growth, across 16 pooled trials

Who: 380 participants across 16 trials. How many: 380. How long: Varies by trial, pooled across 16 studies. Design: Systematic review and meta-analysis of volume-equated resistance-training trials. Measured: Muscle size (hypertrophy), STAND-IN.

• trials pooled: 16
• participants: 380
• sessions per muscle per week: 1 to 3

Biomarkers flag: Muscle size stands in for the strength and function outcomes a reader likely actually cares about.

A systematic review pooled 16 trials, 380 participants total, that held weekly training volume constant while varying how many times a week each muscle was trained, from once to three times. Once volume was equated, splitting it across more sessions had little independent effect on how much muscle grew, with GRADE-rated certainty behind the comparison.

Participants were not specifically older adults, so this should not be read as guidance for preserving muscle against age-related loss specifically -- it says something about hypertrophy training in general, not about sarcopenia prevention. The outcome measured was muscle size, not strength or physical function. (Review)²

Harms: Resistance-training harms are not reported in this meta-analysis; it compares outcomes between training schedules, not training against no training.

What would move this: The same comparison repeated in older adults specifically, measuring strength or function rather than muscle size.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. If you already train, hitting your weekly volume looks more important than how you split it across the week -- but this was not tested in older adults specifically.

EAT

A belly-fat extract missed its target, and the collagen case still isn't made

Who: People taking tamarind seed husk extract; separately, trials of oral collagen. How many: 112 in the tamarind trial; not stated for the collagen review. How long: 12 weeks. Design: Randomised, double-blind, placebo-controlled trial, plus a narrative review of RCTs, systematic reviews and meta-analyses. Measured: Visceral fat area, STAND-IN; skin, body-composition and joint claims, STAND-IN.

primary endpoint at week 12: p=0.077

Funding flag: The collagen review names industry funding as one reason the positive results in that literature are hard to trust.

If you have been eyeing collagen powder for skin and joints, or a herbal capsule for belly fat, two reports this week are worth a minute.

A review of the randomised trials and meta-analyses on oral collagen found the evidence insufficient for its skin, body-composition and joint claims. It named industry funding among the reasons positive results there are hard to read. A trial of tamarind seed husk extract in 112 people missed its main target, visceral fat at twelve weeks, after earlier timepoints had looked significant.

When a product leads with a secondary result, the thing it set out to test usually did not come in. Visceral fat is a stand-in anyway: even a hit would not have told you about illness or lifespan. (Review, People, tested on a stand-in)^3,4

Harms: Not measured: neither the collagen review nor the tamarind trial carries adverse-event data.

What would move this: A pre-registered trial hitting the primary endpoint it declared, published whether or not it worked.

DOES THIS CHANGE WHAT I SHOULD DO?

No. Save your money on both, and keep the test you now have: with the next claim, check whether the number quoted is the one the trial set out to measure.

MEASURE

A US service now sets GLP-1 plans by a blood aging score, retested every three months

Who: Adults in genetic datasets; separately, buyers of a US telehealth aging score. How many: Not stated. How long: Not applicable to a genetic analysis. Design: Two-sample Mendelian randomisation with pleiotropy dissection and replication: inherited genetic differences used as a natural experiment. Separately, retrospective validation of an imaging-derived biological age estimator against observational mortality outcomes. Measured: Fibrotic lung disease risk against six ageing measures, OUTCOME.

ageing measures tested: 6

Biomarkers flag: The aging score stands in for health itself, and no study here shows that moving it changes how long or how well anyone lives.

If a clinic wants to run your treatment off a biological age score, settle one thing first: what would the number change? One US telehealth service now sets GLP-1 plans against a blood-based aging score, and retests every three months.

These scores are not empty. A biological age estimate read off CT scans was tested against mortality records and did carry information about who died. But this week's causal test found that scores built from chemical marks on DNA and from blood proteins did not raise the risk of fibrotic lung disease, while short telomeres did.

Predicting death across a group is a different claim from being a dial worth turning in one person. (Not a study, People, watched, Measurement)^5–7

Harms: Not measured: no study here reports outcomes, good or bad, for anyone whose care was steered by an aging score.

What would move this: A trial randomising people to care guided by an aging score against usual care, reporting illness, independence or death.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. If a clinic ties your GLP-1 plan to a quarterly aging score, ask what it would do differently at a lower number, and what shows that helps. Nothing in the evidence says the prescribing decision needs the score.

A new blood-cell ratio was linked to immune aging and health trajectories, in 2,609 people

Who: 2,609 people aged 18 to 98, across 8 international cohorts. How many: 2,609. How long: Cross-sectional: one point in time per person. Design: Observational cohort: single-cell sequencing of over 12 million white blood cells; watched, not tested. Measured: Ratio of two T-cell subtypes (GZMK+ to GZMB+), STAND-IN.

• people profiled: 2,609
• white blood cells analysed: over 12 million

Biomarkers flag: The ratio being associated with health trajectories does not mean shifting it would change them.

A single-cell analysis of blood from 2,609 people, ages 18 to 98, across eight international cohorts, measured the balance between two subtypes of a common immune cell. One subtype (marked GZMK+) rose with age in people who stayed healthier; the other (GZMB+) rose more in people with worse health trajectories. The ratio between them tracked better with health outcomes than age alone.

This record is Eric Topol's newsletter describing the peer-reviewed paper, not the paper itself. Nobody was assigned to anything here. The cohorts were watched, and the cell ratio was measured alongside whatever health outcomes people already had or went on to have. (People, watched)⁸

Harms: A blood draw carries its own ordinary risk. No harms specific to this analysis are reported, because nothing was administered.

What would move this: A clinical test built on this ratio, validated against a real outcome like illness or survival in a separate cohort, not just correlated with health status in the same one.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. This is a pattern worth knowing about, not yet a test you can ask for or a number you can act on.

Wildfire and traffic pollution components were linked to millions of new dementia diagnoses in Medicare data

Who: Medicare fee-for-service beneficiaries, 2005 to 2019. How many: nearly 40 million. How long: 14 years. Design: Observational cohort: national claims data linked to modelled air-pollution exposure; watched, not tested. Measured: New Alzheimer's and dementia diagnoses, OUTCOME.

• beneficiaries tracked: nearly 40 million
• new Alzheimer's cases: 3.5 million
• new dementia cases: 8.4 million

Researchers linked claims records from nearly 40 million Medicare beneficiaries to detailed estimates of the PM2.5 pollution components they breathed, from wildfire smoke and other combustion sources, over 14 years. The analysis found 3.5 million new Alzheimer's diagnoses and 8.4 million new dementia diagnoses in that population. It then asked which specific pollution components carried the strongest association.

This is observational: nobody's exposure was assigned or controlled, so the finding is an association across a huge population, not proof that any one component causes dementia. A companion analysis found a similar pattern for mortality. The transferable part is which chemical components matter, not wildfire specifically. India's own exposure comes mostly from different combustion sources, including crop burning, so the component-level finding may travel even where the wildfire source does not. (People, watched)⁹

Harms: The study measures an existing, involuntary exposure's association with disease; it is not evaluating a product or therapy with its own side effects.

What would move this: A natural experiment, such as a region-level pollution-reduction policy, tested directly against dementia incidence.

DOES THIS CHANGE WHAT I SHOULD DO?

No. This is about air you breathe involuntarily. The response belongs to regulators, not to anything you can individually buy or do.

WHAT DIDN'T MOVE

NAD+ supplements.

A ten-person crossover compared liposomal with conventional oral dosing; in worms, an NMN-producing probiotic matched purified NMN on most measures. Absorption and worms, no human function, disease or survival. (People, tested on a stand-in, Animals)^10,11

Longeveron's Alzheimer's cell therapy.

A post-hoc reanalysis of brain-inflammation markers from the completed 49-participant trial, posted as a preprint during an investor tour. A hypothesis for the next trial, not evidence of benefit. (People, tested on a stand-in)¹²

Statins and muscle.

In liver-cell cultures, lovastatin and atorvastatin cut IGF-1 release, proposed as the route to statin-linked muscle loss. Cells, not people, and no reason to stop a prescribed statin. (Lab)¹³

Exercise proteomics.

Two papers mapped thousands of coordinated protein changes in blood, saliva and urine after a single bout of exercise. Detailed biology, and still nothing that reproduces exercise. (People, tested on a stand-in)^14,15

THE THREAD THIS WEEK

All three items turn on the same gap: a number that predicts something, and a number worth acting on.

Who: Deep Longevity (launch); separately Rubedo and an unnamed cream brand, no relationship reported (Not a study, People, tested on a stand-in)^2–4. What: Telehealth launch, plus two company biomarker claims (Not a study, People, tested on a stand-in)^2,3. How much: Quarterly retesting; 11 paired skin samples at Rubedo. Evidence: Stage not assignable: a launch has no study design; supporting readings are company-reported biomarkers (Not a study, People, tested on a stand-in)^2,3. Product, price: Accrua GLP-1 treatment plans; no price announced (Not a study)².

skin epigenetic age: 2.1 years lower across 11 paired samples, p=0.032

Biomarkers flag: A surrogate with no intervention-response validation is sold as a treatment-management tool.

Deep Longevity launched Accrua, a US telehealth service setting GLP-1 treatment plans against a blood-based aging score, retested quarterly (Not a study)². The score now sits inside a prescribing decision.

Rubedo reported skin epigenetic age 2.1 years lower across 11 paired photoaged samples, p=0.032, company-reported and unpublished. An unnamed cream claims two years off a marker, no sample size (People, tested on a stand-in)^3,4.

Genetic evidence pointed the other way on one outcome. A two-sample Mendelian randomisation across six ageing measures (People, watched) found short telomeres raised fibrotic lung disease risk, while epigenetic and proteomic age acceleration did not (People, watched)⁵. That is one disease, not a verdict on clocks as a class. Two mouse reports tie clock readings to tissue state, but record no design, so both are stage not assignable (Animals)^6,7. (Not a study, People, tested on a stand-in, People, watched, Animals)^2–7

Harms: Neither Deep Longevity nor Rubedo reports adverse events; the nearer risk is a drug plan set by an unvalidated reading.

What would move this: A trial where clock-guided GLP-1 dosing beats standard dosing on a clinical outcome.

DOES THIS CHANGE WHAT GETS BUILT, FUNDED, APPROVED OR PRICED?

Probably. Builders of clock-based services have a working template. No regulatory action on Accrua has been reported, and whether such a score counts as a clinical claim is untested (Not a study)².

A non-dopaminergic Parkinson's adjunct met its Phase 3 primary by 37 minutes a day

Who: Cerevance (People, tested on what matters)⁸. What: Phase 3 topline result (People, tested on what matters)⁸. How much: 341 patients, 12 weeks. Evidence: People, tested on what matters. Randomised, double-blind, placebo-controlled Phase 3, company topline, not peer-reviewed (People, tested on what matters)⁸. Product, price: Solengepras 150 mg once daily, adjunctive; no price announced (People, tested on what matters)⁸.

OFF time versus placebo: 0.61 hours less per day, p=0.0350

randomised: 341 patients, 12 weeks

Funding flag: The result is a company topline; the full dataset is unpublished.

Cerevance reported that once-daily solengepras 150 mg, added to existing therapy, cut daily OFF time by 0.61 hours more than placebo at week 12 (People, tested on what matters)⁸. The trial was randomised, double-blind and placebo-controlled, 341 patients, p=0.0350. OFF time is when symptoms return between doses, so the endpoint is what a patient feels.

Non-dopaminergic action is the commercial point: a win widens Parkinson's treatment beyond dopamine replacement. Thirty-seven minutes a day is modest at a clean p-value.

The data establish a 12-week symptom difference in one company-reported trial. Durability, disease modification and any advantage over optimising existing dopaminergic dosing are untested here. (People, tested on what matters)⁸

Harms: Insomnia occurred in 6.2% of treated patients versus 0.9% on placebo, and dyskinesia in 4.4% versus 1.8%. Durability past 12 weeks is untested.

What would move this: Peer-reviewed publication with longer follow-up showing the OFF-time gain holds against optimised dopaminergic dosing.

DOES THIS CHANGE WHAT GETS BUILT, FUNDED, APPROVED OR PRICED?

Yes. Cerevance has a filing-grade adjunct outside dopamine replacement and payers a pricing question; if the difference survives peer review, the average patient loses 37 minutes of daily OFF time (People, tested on what matters)⁸.

CONTRADICTIONS

Has senolytics actually weakened, or did it just get a randomised win we never carried?

Reading A: Still weakening. This week's senolytic news is a company-reported skin biomarker and a preclinical completion promising clinical work in 2026 (People, tested on a stand-in, Lab)^3,9. Neither reaches a patient-relevant endpoint.

Reading B: The call is out of date. A randomised dasatinib-plus-quercetin trial in MASH sat in our week-41 miss list, never carried. It reported fibrosis improvement in 47% versus 7% on placebo, MASH resolution 53% versus 7%, adverse ¹% versus 43%, called self-limiting, in 31 patients. Unpublished, so no stage is assignable, and 31 patients across those endpoints leave few events per cell.

What would settle it: Peer-reviewed publication of that MASH trial, plus replication large enough to weigh an 82% adverse-event rate. (People, tested on a stand-in, Lab)^3,9

BUSINESS SIGNALS

Regulation:

Three trials of a CHO-cell recombinant herpes zoster vaccine were registered; endpoints are zoster protection and safety, not cognition. (Not a study)¹⁰

Money:

Genentech will pay Alector $100 million upfront and up to $1.17 billion in milestones for preclinical AL050 in Parkinson's. (Not a study)¹¹

Money:

Tiny Health raised an oversubscribed $33 million Series B led by B Capital; adult testing volume now exceeds infant. (Not a study)¹²

Pipeline:

Digbi Health reached all 50 US states via OpenLoop Health and MD Integrations: 300-plus employers, 700-plus prescribers. (Not a study)¹³

Pipeline:

A disproportionality analysis flagged an audiovestibular adverse-event reporting pattern for GLP-1s, which cannot establish causation. (People, watched)¹⁴

Pipeline:

Retatrutide and zenagamtide, neither approved, are sold online for unsupervised use with no clinical data. (Not a study)¹⁵

Pipeline:

Arrowhead reported deep, durable tau reduction in healthy volunteers from ARO-MAPT, a Phase 1/2a interim analysis (People, tested on a stand-in). (People, tested on a stand-in)¹⁶

Pipeline:

Monte Rosa's NEK7-targeting NLRP3 inhibitor moved several heart-risk inflammatory markers toward healthy levels in 108 people with obesity; company-reported. (People, tested on a stand-in)¹⁷

INDIA WATCH

• India's prime minister launched IIT Madras's mission to map over 100 human brains at cellular resolution, plus a 1,000-section stroke atlas. (Not a study)¹⁸
• FSSAI issued a notice to Dr Reddy's and Nestle Health Science over Celevida GLP+ muscle-preservation claims. (Not a study)¹⁹
• A 1,200-person randomised mHealth lifestyle trial in Colombo (People, tested on a stand-in) reported 10-year risk scores, not events: the closest South Asian test available. (People, tested on a stand-in)²⁰
• AWGS 2025's BMI-adjusted muscle index flagged vulnerable patients the height-adjusted index missed, an argument for cohort-calibrated cut-offs. (Measurement, People, watched)^21,22

WATCHING

Nothing this week changes what you should do. You may have seen headlines that cell reprogramming reached its first human test. Three glaucoma patients each had one injection, the company reported early visual-field signals in two of them and no serious adverse events so far, and nothing has been published. Treat that as a start, not a result. Among more than half a million adults who had taken an exercise test, how fit they were sharpened the prediction of who would die within ten years. Keep doing what makes you fitter and stronger. If you take a GLP-1 drug and worry about muscle, the add-on drugs being built for that have produced company-reported body-composition numbers and nothing yet about strength or falls. And if you have been offered a longevity panel costing thousands a year, none of the services selling one has reported what happened to the people who bought it.

TAKE

Muscle-sparing add-ons to GLP-1 drugs reported muscle mass and nothing about function

Who: Adults taking GLP-1 drugs; the ATR-258 study enrolled overweight or obese subjects. How many: 13 in the ATR-258 study; not stated for trevogrumab. How long: 8 weeks for ATR-258; not stated for trevogrumab. Design: Clinical trial of trevogrumab added to GLP-1 therapy, randomisation not disclosed; ATR-258 open-label, single-arm, dose-escalation. Both people, tested on a stand-in.. Measured: Muscle loss and body composition, STAND-IN.

ATR-258 participants: 13

Funding flag: Both results were released by the companies developing these investigational drugs, and neither has been published. The ATR-258 study is reported as investigator-initiated.

If you are taking a GLP-1 drug, you have probably wondered whether you are losing muscle along with the fat. Companies are building add-on drugs for that worry, and one reported results this week.

The trevogrumab report says muscle loss fell during GLP-1 treatment and regain improved afterwards. How patients were assigned to treatment was not disclosed, so the finding cannot be weighed. A second drug, oral ATR-258, was given to 13 people for eight weeks with no comparison group, which can show tolerability and not benefit.

In India, FSSAI has issued notices to Dr Reddy's Laboratories and Nestlé Health Science over claims that Celevida GLP+ supports lean muscle during GLP-1 and GIP therapy. No clinical outcome data accompanies that claim in the available reporting. (People, tested on a stand-in, Not a study)^1–3

Harms: None reported. The trevogrumab report carries no adverse-event data, which is an absence of reporting rather than a clean safety record. (none reported)

What would move this: A randomised trial in people on GLP-1 drugs showing measured strength holding up, or fewer falls, rather than a better body-composition scan.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. If you take a GLP-1 drug, ask your doctor whether anyone is tracking your strength and physical function, which these products haven't shown they protect.

MOVE

Fitness and grip strength predicted who died, in two very large groups of adults

Who: Adults in a clinical exercise-testing registry, and UK Biobank adults tested before any cancer diagnosis. How many: 562,234 and 10,462. How long: Ten-year mortality prediction; median 12.5 years of follow-up. Design: Observational cohorts, both people-watched studies: watched, not tested. Measured: Death from any cause, OUTCOME.

exercise-test cohort: 562,234 adults

UK Biobank cohort: 10,462 adults

A fitness number carries real prognostic information, which is the argument for measuring it routinely. Nothing in this week's record shows that clinicians have started to. Two large studies this week say that number carries real information.

Among 562,234 adults in a clinical exercise-testing registry, peak fitness, meaning the hardest work the body can sustain, improved ten-year death-risk prediction beyond the standard cardiovascular risk factors. These were people already referred for testing, not a general population sample. Among 10,462 UK Biobank adults tested before any cancer diagnosis, fitness and grip strength tracked with death from any cause and with survival fifteen years after diagnosis. (People, watched)^4,5

Harms: Not measured. Neither study tested an exercise programme, so neither reports injuries or anything else training can cause.

What would move this: A randomised trial in which people assigned to raise their fitness then lived longer, or stayed independent longer, than those who were not.

DOES THIS CHANGE WHAT I SHOULD DO?

No. Keep doing what makes you fitter and stronger: these studies show fitness predicts your risk, not what training itself changes.

MEASURE

Longevity panels went on sale at $5,500 to $45,000 a year, with no report of what buyers gained

Who: People buying preventive testing out of pocket. How many: No study population; these are commercial services. How long: No follow-up reported. Design: Stage not assignable: news reports of commercial services, no study design. Measured: Test results sold to buyers, with no outcomes reported, STAND-IN.

• hospital longevity packages: $5,500 to $45,000 a year
• direct-to-consumer calcium scan: $249

Biomarkers flag: The hospital packages and the $249 calcium scan sell measurements, and neither has been shown to change what happens to the person measured.

A longevity assessment is now something a nonprofit hospital system will sell you for $5,500 to $45,000 a year, and a coronary artery calcium scan can be bought directly for $249 with no referral. The question is whether a number you pay for changes anything.

None of these services reports what happened to the people who bought them. The reporting on the $249 scan does explain the limits of a zero calcium score, which is the right caveat and a rare one here. A third company is building software to turn test data into continuing care, which is a plan, not a result. (Not a study)^6–8

Harms: Not measured. Testing people without symptoms turns up incidental findings, and those lead to further scans and procedures. Neither the hospital packages nor the calcium scan reports how often that happens to buyers.

What would move this: A study that followed buyers of one of these panels. It would need to report what was found, what was done about it, and whether buyers ended up healthier than people who did not buy one.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. If you've wondered about a longevity panel or a calcium score, ask a doctor what either result would change before paying for one.

Brain waste clearance was measured directly in people for the first time, and overlaps with amyloid buildup

Who: 96 adults undergoing standard neuroimaging. How many: 96. How long: Cross-sectional: one point in time. Design: Human neuroimaging with an injected contrast agent, matched against gene activity; people, watched. Measured: Where clearance was weakest versus where amyloid protein was building up.

participants: 96

Biomarkers flag: A clearance-amyloid overlap is two measurements moving together, not a measured effect on a person.

Researchers injected a contrast agent into the fluid around the brain and tracked, region by region, how well 96 people cleared it, then compared that map to where amyloid protein was building up. The two tracked together: regions clearing waste more slowly also carried more amyloid.

This is the first time this clearance process has been mapped directly in people rather than inferred from animals. Separate mouse work this week showed the channels waste normally drains through can collapse in Alzheimer's-like brains, and a rat study combined imaging with modelling of the same system. All of it points at one mechanism; none of it tests whether helping the brain clear waste changes anything a person experiences. (People, watched, Animals, Review)^9–12

Harms: No harms data is reported for the imaging or contrast procedure in this record.

What would move this: A trial testing whether a treatment that improves glymphatic clearance changes a real outcome, not just the clearance measurement itself.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. Researchers can now see this process in a living brain, but nothing here says what to do about it.

WHAT DIDN'T MOVE

Reprogramming.

First human data: three glaucoma patients, one injection each, open-label and company-reported, with no serious adverse events so far. Informal versions given outside trial structures are unpublished, so no safety data on them exist. Nothing to act on. (People, tested on a stand-in, Not a study)^13–15

Senolytics.

A systematic review of human senolytic trials found the trials predominantly report biomarkers rather than clinical outcomes. Whether a senolytic has improved a clinical outcome in people is not settled by that review. (People, tested on a stand-in)¹⁶

Creatine.

A review of randomised trials in older adults adds no new trial data. The evidence you were working from in May is the evidence you have now. (Review)¹⁷

Hyperbaric oxygen.

A systematic review pooled existing studies in adults over 50, with no new trial. It is heavily marketed on a mixed study base, and no harm estimate for this population appears in the available reporting. (Review)¹⁸

Sodium.

A US cohort found no significant association between reported sodium intake and death after adjustment. An observational null is weak evidence of no effect, and salt's blood-pressure trials stand. (People, watched)¹⁹

Who: Life Biosciences. What: First-in-human dose-escalation trial of ER-100. How much: 3 patients, one injection each. Evidence: People, no comparator: single-arm Phase I dose escalation, company-reported. Product, price: None announced (People, tested on a stand-in)¹.

• patients dosed: 3 (People, tested on a stand-in)¹
• visual-field signal: 2 of 3 (People, tested on a stand-in)¹

Funding flag: The only account of this trial comes from the company that ran it.

Life Biosciences dosed three glaucoma patients with ER-100 in a first-in-human dose escalation, the first human data from a formal reprogramming trial (People, tested on a stand-in)¹. Earlier informal human use was never published (Not a study)². The company reported early visual-field improvement signals in two of the three patients (People, tested on a stand-in)¹. Two reports describe one trial, not two (People, tested on a stand-in)³. In mice, a 16-day course of Yamanaka-factor induction reversed features of haematopoietic stem cell aging (Animals)⁴.

Three open-label patients with no serious adverse events so far do not establish tolerability, and the follow-up duration is unreported. A visual-field change in two people with no comparator is an observation, not an effect, and the company is the only source. (People, tested on a stand-in, Animals, Not a study)^1–4

Harms: No serious adverse events to date in three patients (People, tested on a stand-in)^1,3; three people cannot make a safety profile. A separate unpublished report describes AAV-delivered partial reprogramming given informally outside trial structures (Not a study)².

What would move this: A controlled trial with prespecified visual-field endpoints, or a dataset published outside the company's own release.

DOES THIS CHANGE WHAT GETS BUILT, FUNDED, APPROVED OR PRICED?

Not yet. Funders get their first human signal; no clinician, regulator or buyer can act on three open-label patients (People, tested on a stand-in)¹.

Scholar Rock refiled apitegromab in Europe and a GLP-1 add-on reported a trial in one week

Who: Scholar Rock, Atrogi. What: An EMA resubmission plus two company-reported readouts (Not a study, People, tested on a stand-in)^5–7. How much: 13 subjects in the ATR-258 study; trevogrumab trial size not disclosed. Evidence: People, no comparator: an 8-week single-arm study for ATR-258, and a company-reported trevogrumab readout whose randomisation and design are undisclosed (People, tested on a stand-in)^6,7. Product, price: None announced (Not a study, People, tested on a stand-in)^5,6.

ATR-258 subjects: 13, open-label single-arm (People, tested on a stand-in)⁷

• trevogrumab design disclosed: none (People, tested on a stand-in)⁶
• apitegromab resubmission: indication not stated (Not a study)⁵

Biomarkers flag: Lean mass is standing in for the strength and function outcomes the sarcopenia worry is about.

Scholar Rock resubmitted its apitegromab marketing authorisation application to the European Medicines Agency (Not a study)⁵. The indication covered by that resubmission is not stated in the available reporting. It cannot be read as a filing for muscle preservation during GLP-1 therapy.

A company-reported trial says trevogrumab reduced muscle loss in patients on a GLP-1 agonist and aided muscle regain after treatment stopped (People, tested on a stand-in)⁶. The report discloses no randomisation and no design, so the claim cannot be graded. Atrogi's 8-week, open-label, single-arm study of oral ATR-258 in 13 overweight or obese subjects is company-reported and unpublished (People, tested on a stand-in)⁷.

Every result here measures muscle mass, not strength or function. (People, tested on a stand-in, Not a study)^5–8

Harms: No adverse-event data accompanied the trevogrumab report (People, tested on a stand-in)⁶. The ATR-258 study is company-reported and unpublished, so its safety data cannot be read (People, tested on a stand-in)⁷.

What would move this: A randomised trial in older patients reporting strength, falls or physical function (People, tested on a stand-in)⁶.

DOES THIS CHANGE WHAT GETS BUILT, FUNDED, APPROVED OR PRICED?

Maybe. Funders see the category move; prescribers get no falls, strength or function data to act on (People, tested on a stand-in, Not a study)^5,6.

Nonprofit hospital systems began selling longevity packages at $5,500 to $45,000 a year

Who: Nonprofit hospital systems, Mito, Longevitix. What: Clinic launches and a direct-to-consumer scan. How much: $5,500 to $45,000 a year; $249 per scan. Evidence: Not a study: a clinic launch and a price carry no design (Not a study)^9,10. Product, price: Packages at $5,500 to $45,000 a year; $249 calcium scan (Not a study)^9,10.

• annual packages: $5,500 to $45,000 (Not a study)⁹
• calcium scan: $249 (Not a study)¹⁰
• outcome data on buyers: none

Nonprofit hospital systems are opening longevity clinics selling personalised assessments and advanced testing at $5,500 to $45,000 a year (Not a study)⁹. The operator Mito sells a direct-to-consumer coronary artery calcium scan at $249 with no physician referral (Not a study)¹⁰. Longevitix is building an AI layer meant to turn preventive-health data into continuous care (Not a study)¹¹.

The channel is what changed. Selling these packages inside nonprofit systems puts cash-pay testing in institutions that carry charity-care obligations. Neither the hospital clinics (Not a study)⁹, nor Mito (Not a study)¹⁰, nor Longevitix (Not a study)¹¹ reports outcome data on buyers. (Not a study)^9–11

Harms: Broad testing in people without indications produces incidental findings, and findings produce procedures. Neither the hospital clinics (Not a study)⁹ nor Mito (Not a study)¹⁰ reports what happened to buyers.

What would move this: Follow-up on a cohort of package buyers: downstream procedures, findings acted on, outcomes.

DOES THIS CHANGE WHAT GETS BUILT, FUNDED, APPROVED OR PRICED?

Probably. Builders and investors in preventive testing gain an institutional channel; buyers get a priced panel with no outcome evidence (Not a study)⁹.

A skin-longevity category is forming on reviews and lab assays, with one real trial behind it

Who: A skin-care trial sponsor, plus seven review and lab-assay publishers. What: A randomised, vehicle-controlled facial-photoaging trial, surrounded by reviews, an in vitro assay and opinion pieces, all using 'hallmarks of aging' language. How much: 1 controlled trial, sample size not disclosed; 7 non-trial publications in the same window. Evidence: People, tested on a stand-in: one randomised vehicle-controlled trial on skin biomechanics; the rest are reviews, a lab assay, or not a study at all. Product, price: None announced.

• real trials this week: 1 of 8
• trial length: 12 weeks

Biomarkers flag: Skin biomechanics and extracellular-matrix markers stand in for any health outcome; a cosmetic result is not a longevity result.

A 12-week randomised, double-blind, vehicle-controlled trial tested a topical TORC1 inhibitor against facial photoaging, measuring skin biomechanics and structural protein markers. It is a real controlled design, run against a placebo cream rather than nothing. Seven other pieces published the same week sell the same 'aging hallmark' language. They include reviews of marine collagen peptides, the skin microbiome, and skin-aging cell stress; an in vitro extract screen; a clinical framework proposal; and an opinion piece on lab-grown collagen.

None of the seven is a study of a person. The category is building commercial and scientific-sounding language around aging biology faster than it is building trial evidence for it, and the one real trial measured a cosmetic surrogate, not a health outcome. (People, tested on a stand-in, Review, Not a study, Lab)^12–19

Harms: No adverse-event data is reported for the trial, and the review and assay pieces carry no human safety data at all.

What would move this: A trial that ties a skin-aging marker to a result a buyer would actually notice, or to a claim a regulator would recognise.

DOES THIS CHANGE WHAT GETS BUILT, FUNDED, APPROVED OR PRICED?

Not yet. Builders get one real trial to point to; buyers get a category mostly selling on language borrowed from aging biology.

BUSINESS SIGNALS

Regulation:

WHO issued its first global childhood obesity guidelines, putting diet, exercise and counselling ahead of GLP-1 agonists for younger children. (Not a study)²⁰

Money:

Caribou Biosciences is shutting down after failing to fund a late-stage trial of its allogeneic CAR-T lymphoma therapy. (Not a study)²¹

Money:

UCLA Health won a five-year, $25 million National Institute on Aging grant to lead national testing of AI dementia tools. (Not a study)²²

Pipeline:

Longevica reports a four-year, $13 million lifespan screen in roughly 16,000 mice (animals), claiming more than 60 candidate compounds. Company-reported and unpublished: survival curves, effect sizes and maximum lifespan are unreported. (Animals)²³

Pipeline:

Genflow presented interim methylation and biological-age signals from SIRT6c gene therapy in aged beagles. Sarcopenia endpoints pending. (Animals)²⁴

Pipeline:

A Phase II trial of GAL-101 2% eyedrops in non-foveal geographic atrophy was registered, no results. (Not a study)²⁵

Pipeline:

EsoGuard, a methylation assay on balloon-collected cells, was evaluated in 11,991 patients screened for Barrett's oesophagus. Retrospective preprint. (People, watched)²⁶

Pipeline:

The ELITE pilot gave 31 adults aged 50-80 repurposed metabolic drugs plus two AgeBiotics supplements for eight weeks. Feasibility only. (People, tested on a stand-in)²⁷

INDIA WATCH

• FSSAI issued notices to Dr Reddy's Laboratories and Nestle Health Science over Celevida GLP+ lean-muscle claims during GLP-1 therapy. No clinical data accompanies the claim. (Not a study)²⁸
• A scoping review maps parenteral placental extracts sold for healthy aging in India and three other markets. Heterogeneity prevents pooling, so neither benefit nor harm is estimable. (Not a study)²⁹

WATCHING

Nothing this week changes what you should do. Six new ways to estimate biological age arrived at once, from speech, from an ECG, from blood and from medical records. One was tested against how long people actually lived; the other five were not tested against any outcome at all. Either way, the number is not a plan. If a clinic offers you one, nothing here says the result would change a decision, so declining costs you nothing. You may also have seen weight-loss drugs credited with preventing cancer and dementia; those claims came from reviews of existing studies, not new trials, and are not a reason to start a drug. Nine years of hard cycling did not stop fitness or muscle from falling, and the riders' weight stayed flat while their body fat rose. Keep training, and watch what you can do rather than what the scale says.

MOVE

Keep training, and change what you count

If you exercise to stay independent, track what your body can still do: stairs, shopping bags, getting off the floor unaided. That holds even for people who have trained hard for decades.

Researchers followed 82 of 125 highly active older cyclists for nine years; at follow-up they were 64 to 86 (People, watched)¹. This was observational, so it shows patterns rather than causes. Peak fitness fell 19.7% and lean tissue mass 8.1%, and body fat rose from 23.8% to 29.2% while body weight stayed flat.

Two other items this week bear on exercise, and neither tests whether training preserves independence. In 4,477 prefrail and frail older adults, walking-test fitness and grip strength were each independently associated with cognitive impairment at a single timepoint (People, watched)². The MoTrPAC consortium also reported what 12 weeks of training does inside the body. The parallel-arm trial randomised 206 sedentary adults aged 18 to 74 to endurance training, resistance training or no exercise (People, tested on a stand-in)³. A companion paper profiled acute responses in 175 participants from the same cohort (People, tested on a stand-in)⁴.

What none of it shows is that training arrests the decline. Harms: no adverse events from the training programme are reported in the MoTrPAC papers (People, tested on a stand-in)^3,4.

DOES THIS CHANGE WHAT I SHOULD DO?

No. Keep exercising: standard advice to stay active is unchanged, and nothing this week weakens it. The cyclist cohort was observational and the fitness-cognition link was cross-sectional, so neither tests whether tracking function rather than weight changes any outcome. Weight alone can mislead: in the cyclists it stayed flat while body fat rose.

MEASURE

Six new biological age tests

Six research groups published new ways to estimate biological age this week. If and when a clinic offers one of them, the question is whether that number would change anything a doctor does.

The six estimators:

• speech, in 2,928 people across five Latin American countries, separating healthy controls from mild cognitive impairment and dementia subtypes (People, watched)⁵
• organ age from blood and tissue gene activity (People, watched)⁶
• diagnosis codes from 9.5 million Hungarian records (People, watched)⁷
• ECG patterns, from a database of 399,786 recordings (People, watched)⁸
• blood methylation modules in 2,050 people (People, watched)⁹
• death over 16 years in 861 people from the Lothian Birth Cohort 1936, the only one here tested against a hard endpoint (People, watched)¹⁰

Which of these measures performed best against that mortality endpoint is unreported. None of the six is reported as validated against a clinical outcome it would be used to predict.

A separate methods paper looked at six established epigenetic clocks (People, watched)¹¹. It quantified how much of their signal comes from blood cell composition, specifically naive and memory T cell proportions, rather than from aging biology. In plain terms, a shift in your immune cell mix can move your age result without necessarily changing anything you would feel.

No longevity-clinic protocol in the available reporting has been linked to a health outcome, at any price point, and none reports adverse-event data (Not a study)¹². No clinic is yet reported to offer any of these six measures.

DOES THIS CHANGE WHAT I SHOULD DO?

No. None of these numbers changes a screening interval, a drug or a dose. Declining costs you nothing; if you already have a result, do not act on the gap it reports. This moves when one clock predicts something a clinician acts on, and beats the cheap measures already in use.

Alzheimer's testing is moving to blood and phone-based tools, before the reliability question is settled

If a cheap blood test or a phone app could tell you your Alzheimer's risk, the real question is whether one reading means anything. This week's evidence says: not quite yet.

A Mayo Clinic analysis of 744 patients found a leading blood test for Alzheimer's-related proteins (plasma p-tau217) gave consistent results across different test batches over 18 months (Lab)¹³. A separate study validated a self-administered cognitive test, taken on a phone or computer, against brain scans and clinical diagnosis in 2,249 people (People, watched)¹⁴. Two companies, Amprion and Prema Cognition, are already packaging a digital cognitive test with a blood-based biomarker test into one screening product (Not a study)¹⁵.

What complicates this: a smaller study measuring the same blood test and the same cognitive test three times over six months found both varied meaningfully between visits in the same people (Lab)¹⁶. Separately, the brain scans used as the reference standard for validating these tests were found to bind proteins they were not designed for (Lab)¹⁷. That could affect how the reference standard itself is read.

Harms: none of this evidence reports harm from taking either test. The risk is treating one reading as a diagnosis when the tests themselves are shown to move between visits.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. These tests are getting cheaper and easier to access, which is real progress. But the same record that validates them also shows results vary from visit to visit, so a single result is not yet something to act on alone. If you are offered one, ask whether repeat or confirmatory testing is part of the process.

Pollution is becoming a measurable aging input, and one trial is testing whether removing it helps

Most of what is known about pollution and aging says people exposed to more of it do worse, which cannot tell you whether removing the exposure helps. One new trial is testing removal directly, and it includes India. A four-country randomised trial is distributing LPG stoves and fuel to half of 3,200 pregnant-women households in India, Guatemala, Peru and Rwanda (Not a study)¹⁸. The other half keep cooking with wood or dung smoke, and children in both groups are followed from birth. No results are reported yet. In India specifically, the policy lever already exists: LPG subsidy and distribution programmes already run, so a positive result would argue for using what exists more, not for building something new.

Why this reaches beyond lung disease: this week also brought evidence that pollution-level exposures can touch aging biology directly. Microplastic exposure triggered cellular senescence in human brain tissue grown in the lab, alongside after-death evidence linking plastic buildup to dementia-affected brains (Lab)¹⁹. In 403 children followed from mid-childhood, higher blood levels of PFAS ("forever chemicals") were associated with faster biological aging that persisted into early adolescence (People, watched)²⁰. Against that, more green space near home was associated with less dementia over 16 years in nearly 3,000 older Swedish adults (People, watched)²¹.

Harms: none of these studies report harm from an intervention, because only the stove trial tests one; the rest are laboratory or observational findings.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. The lab and cohort findings make pollution exposure a plausible target, not a proven one; none of it shows that reducing an exposure changes how you age. The stove trial (Not a study)¹⁸ is the one result worth watching: it tests the exposure directly, in households like many in India.

TAKE

What the newer shingles vaccine does to T cells

Zoster vaccination is one of the few interventions in this field with an established clinical outcome. This week's study offers a candidate mechanism for the recombinant vaccine's reported advantage. It measured T cell responses, not shingles cases, so it does not test protection.

Researchers put a weakened chickenpox-family virus into the skin of older adults vaccinated five or more years earlier (People, tested on a stand-in)²². Those who had received the recombinant adjuvanted vaccine mounted faster and larger T cell responses at the site than those given the live attenuated vaccine.

That was measured in people rather than inferred from animals. Harms: the study reports immune responses only, with no safety comparison between the two vaccines.

DOES THIS CHANGE WHAT I SHOULD DO?

No. Nothing here changes vaccination advice: the study measured T cell responses, not shingles cases, and it is a preprint. If you are due for zoster vaccination, that decision rests on existing guidance and on which vaccine your programme supplies, neither of which this result alters.

WHAT DIDN'T MOVE

• GLP-1 drugs for cancer or dementia prevention. Reviews widened the claim list to seven obesity-associated cancers (People, watched)²³ and pooled cognition data (People, tested on a stand-in)²⁴. Neither added trial data, so neither supports starting one for prevention. Harms: a meta-analysis pooling trials, cohorts, database analyses and pharmacovigilance reports found the evidence on semaglutide and depression risk inconsistent across study types (People, watched)²⁵.
• Stem cell therapy. A randomised, placebo-controlled crossover trial of a single injection into the spinal fluid in progressive multiple sclerosis found no neurological benefit, only immune changes in blood and spinal fluid (People, tested on a stand-in)²⁶. Adverse events are not reported.
• Rapamycin. A commentary restates that it extends mouse lifespan by 10 to 20% while human anti-aging benefit remains unproven (Not a study)²⁷. Two drugs nominated as mTOR inhibitors failed a standardised worm lifespan protocol (Animals)²⁸.
• Beetroot supplements. A systematic review found beetroot extract reliably raised nitrate and nitrite in older adults while pooled blood pressure results stayed inconsistent (People, tested on a stand-in)²⁹. Harms for long-term use are not reported.
• PB125. Its maker announced a randomised trial in 44 people, 24 of them older, where two weeks improved artery dilation (People, tested on a stand-in)³⁰. Artery dilation is a stand-in measure and the window was two weeks; no adverse-event data appear.

• FDA officials said the agency plans to put aging on its regulatory science agenda (Not a study)¹.
• Six new biological age measures were published; a methods paper quantified how much of a clock reading reflects blood cell mix (People, watched)². No product launch or price accompanies any of the six.
• A T-cell-engager senolytic reached non-human primates, with a dose-limiting liver toxicity produced by the same mechanism that kills the senescent cells (Animals)³.

↑ Six new age clocks, and what a clock reading contains

Six new biological age measures appeared this week.

• Speech, from 2,928 people across five Latin American countries (People, watched)⁴.
• Organ-level transcriptomic clocks from blood and GTEx tissue (People, watched)⁵.
• ICD-10 codes from a national health registry of 9.5 million Hungarians (People, watched)⁶.
• An ECG clock trained on 399,786 recordings (People, watched)⁷.
• Blood methylation modules, developed in 656 people and replicated in 1,394 (People, watched)⁸.
• A Lothian Birth Cohort analysis benchmarked against 16-year all-cause mortality in 861 people; which measure predicted best is unreported (People, watched)⁹.

None of the six, including the Lothian mortality analysis above, has been validated against a clinical outcome it would be used to guide. Predicting death in a cohort is not the same as a reading that changes a decision. At the same time, a methods analysis asked what a clock reading actually contains. It quantified how much of six epigenetic clocks' age and age-acceleration signal comes from blood cell composition, specifically naive and memory T cell proportions, rather than intrinsic aging biology (People, watched)². The size of that share is not stated in the available reporting.

Transplanted hearts shifted epigenetic age toward the recipient's (People, watched)¹⁰, and fat transplanted between young and old mice took on the host's aged or youthful state (Animals)¹¹. Both are preprints, one is mouse, and neither reports a functional gain.

Voice, ECG and registry codes need no new sample collection (People, watched)^4,6,7. Entry costs fall for consumer and insurer products while validation lags. The confound is quantified in epigenetic clocks, so it bears on the methylation entrant (People, watched)⁸ and not on the speech, ECG, registry or transcriptomic models. For all six, no clock here is tied to a harm or a benefit, so a buyer gets a number with no decision attached to it.

→ Senolytics changed modality, and the primate data came with a dose-limiting toxicity

After years of small-molecule disappointment, the pipeline moved to immune targeting. A bispecific T-cell engager against uPAR-positive senescent cells produced dose-dependent senolytic effects in mice and non-human primates (Animals)³. The harms sit in the same study: high doses caused T-cell-mediated hepatotoxicity, detectable as raised serum aminotransferase activity. Efficacy and toxicity scale together, because a T cell kills wherever its antigen appears.

A review separately proposes autologous cytokine-induced NK cells as senolytics, listing the obstacles outstanding (Not a study)¹². Immune senolysis is further along than the small molecules got, and still short of any human trial. A monitorable liver enzyme gives a first-in-human dose escalation something to titrate against.

CONTRADICTIONS

Is GLP-1 lean-mass loss a problem? A perspective argues the lean mass lost on incretin therapy, roughly a quarter to a third of total weight lost, matches other effective weight-loss methods. It should not be called treatment-induced sarcopenia (Not a study)¹³. A narrative review treats the same loss as a risk to weigh (People, tested on a stand-in)¹⁴. Nobody has shown which reading holds. A registered single-arm trial will measure body composition, bone and physical performance under nine months of weekly tirzepatide in older adults (Not a study)¹⁵. With no comparator arm it cannot show whether the lean loss is drug-specific.

Exercise biology and exercise outcomes diverged. The parallel-arm MoTrPAC trial randomised 206 sedentary adults aged 18 to 74 to endurance training, resistance training or no exercise (People, tested on a stand-in)¹⁶. A companion paper profiled acute responses in 175 participants from the same cohort (People, tested on a stand-in)¹⁷. Against that, 82 of an original 125 highly active older cyclists were followed nine years, with 43 lost to follow-up and no untrained comparison group. They lost 19.7% of VO2peak and 8.1% of lean tissue (People, watched)¹⁸. The cohort shows decline continuing under sustained training, not that training leaves its rate unchanged.

BUSINESS SIGNALS

• Senior FDA officials said at a Boston meeting that the agency plans to include aging in its regulatory science agenda. They discussed possible approval pathways, naming no drug, trial or guidance (Not a study)¹. This changes what investors fund before it changes what regulators approve.
• SMART-MS, a randomised placebo-controlled crossover trial of one intrathecal mesenchymal stem cell injection in progressive multiple sclerosis, found no neurological benefit alongside immune-compartment changes (People, tested on a stand-in)¹⁹. No adverse-event data for the intrathecal injection appear in the available reporting. Clinics selling MSC infusions are selling a modality that just failed a randomised test in progressive multiple sclerosis, by a different route of administration.
• Novartis and Abogen Biosciences signed an mRNA collaboration and licensing deal worth up to $7.8 billion (Not a study)²⁰. Deal values that size are mostly milestones; large pharma is still paying platform prices for the mRNA delivery that in vivo reprogramming needs.

INDIA WATCH

• India's Union Health Minister said 420 million people have been screened for hypertension and diabetes under a central scheme (Not a study)²¹. What share of those screened entered treatment is unreported.
• Vytalyou raised Rs 9 crore, about $1 million, to expand its longevity clinic network (Not a study)²².
• A Supreme Court bench asked the central government to examine hospitals requiring patients to buy medicines from affiliated pharmacies, citing markups up to ten times on cancer drugs (Not a study)²³.
• No clinic protocol outcome or adverse-event data accompany these reports, for Vytalyou or for operators elsewhere (Not a study)^22,24.

BRIEFLY

• Arvinas will present Phase 1 biomarker data on ARV-102, a LRRK2 degrader for Parkinson's disease, in Seoul (People, tested on a stand-in)²⁵. A Phase 1 biomarker shows target engagement, not an effect on disease.
• Basecamp Research raised $140 million to train AI models on biological data beyond conventionally studied human genomes (Not a study)²⁶.
• A benchmark of ten general-purpose and three aging-specific single-cell foundation models found a simple highly-variable-gene method beat every model on average (Lab)²⁷.
• Variant Bio expanded its AI target-discovery licence with Boehringer Ingelheim to cardiovascular disease (Not a study)²⁸.
• Junevity will apply its gene-network target-discovery platform to skin aging in Amorepacific skincare products (Not a study)²⁹.
• Prenuvo launched a research network built on its whole-body MRI screening data (Not a study)³⁰. Whole-body screening generates incidental findings, and no overdiagnosis or downstream-procedure data accompany the launch.
• Pathways Bioscience announced a two-week randomised, double-blind, placebo-controlled trial of PB125 in 44 people, 20 younger and 24 older, reporting improved flow-mediated dilation; no adverse-event data accompany it (People, tested on a stand-in)³¹. Flow-mediated dilation is a surrogate, and two weeks in 24 older adults supports no cardiovascular claim.
• A Phase I open-label trial will give oral mesenchymal stem cell secretome capsules to 20 people aged 60 to 75 with frailty (Not a study)³².
• A trial was registered testing the ExPEC9V vaccine against placebo in adults 60 and over with a recent urinary tract infection (Not a study)³³.
• The ELITE trial was registered to test deep biological profiling plus algorithmic polytherapy in healthy ageing (Not a study)³⁴. Giving healthy people several interventions at once carries interaction risk, and the registration record reports no safety data.
• A controlled human infection study in previously vaccinated older adults found faster, greater VZV-specific T cell infiltration after the recombinant adjuvanted zoster vaccine than the live attenuated one (People, tested on a stand-in)³⁵. The endpoint is immune infiltration, and no safety comparison between the two vaccines is reported.
• A Johns Hopkins immune-aging team led by Sean Leng won a $1 million Milestone 2 award in the $101 million XPRIZE Healthspan competition (Not a study)³⁶. It was one of ten teams recognised at this stage. A second awardee the same week was a Tokyo startup working with parsley-derived compounds (Not a study)³⁷.

WATCHING

Nothing this week is a reason to start a new treatment, and one item confirms what already works. Resistance training remains the most consistent story in this newsletter's record. Judge it by what you can lift, walk and climb, not by a muscle-mass number, since two different tools can disagree on mass anyway. Two very large studies measured how well a blood test detects cancer signals, not whether finding one earlier helps you live longer. Keep the screening your age and risk already qualify you for, and treat this one as unproven rather than an upgrade. One joint statement is worth raising with your doctor if hormone therapy was ruled out on heart-risk grounds.

MOVE

Resistance training keeps working. The number most people track for it doesn't.

Three studies this week agree on how to judge it: by strength and function, not by a muscle-mass reading.

• Resistance exercise consistently improved strength and functional performance in older adults, while mass gains were smaller and more variable (People, tested on a stand-in)¹.
• A self-selected lacto-ovo vegetarian diet was non-inferior to an omnivorous one for protein intake and appendicular muscle mass (Not a study, People, tested on a stand-in)^2,3.
• An anthropometric skeletal muscle index equation failed to identify DXA-defined low muscle mass at existing thresholds in 501 Chinese adults (People, watched)⁴. No Indian validation of these thresholds appears in the available reporting.

Not shown: that raising a muscle-mass reading improves strength, falls or independence.

Harms: neither the review nor the diet trial reports adverse events (People, tested on a stand-in)^1,3.

IndiaA large share of the population eats lacto-ovo vegetarian by default, and sarcopenia advice built on omnivorous cohorts assumed a diet many Indian patients will not adopt (People, tested on a stand-in)³.

DOES THIS CHANGE WHAT I SHOULD DO?

Maybe. Nothing here argues against training. It is the most consistent finding in this record, and nothing about it is in question. What changes is which number tells you it's working: strength, gait and stair-climbing move more reliably than muscle mass, and two different tools can disagree on mass anyway. Both arms in the diet trial trained, so it compares diets against each other, not diet against training. After a recent fall, or with a cardiac condition, have a training plan cleared by a physician.

TAKE

A joint obstetrics and cardiology statement links hormone therapy started early to long-term heart benefit

Two proteomic studies arrived alongside it, tying the menopause transition to brain aging markers.

• The statement covers women with cardiovascular risk factors or established ischaemic heart disease. In them, systemic therapy begun early after symptom onset is associated with long-term cardiovascular benefit (Not a study)⁵.
• Hormone-linked molecular changes correlated with poorer later-life cognitive outcomes (People, watched)⁶.
• Inflammatory, synaptic, metabolic and Alzheimer's-related pathway changes tracked hormone status more closely than chronological age (People, watched)⁷.

Not shown: that hormone therapy prevents dementia or cardiovascular events, or that either molecular signature is a treatment target.

Harms: no harms figures accompany the benefit claim, and it does not remove the need for individual cardiovascular and cancer risk assessment (Not a study)⁵.

DOES THIS CHANGE WHAT I SHOULD DO?

Maybe. If you have menopausal symptoms and were told cardiovascular risk rules hormone therapy out, raise this statement with your doctor. Bring your own cardiovascular and cancer history to that conversation. An association in early starters is not a randomised cardiovascular outcome, and nothing here supports taking hormone therapy to protect the brain.

MEASURE

Two large studies of one multi-cancer blood test reported how well it detects, not whether it helps

Both were published with a commentary saying the question that matters is still open.

• PATHFINDER 2 tested the blood test in an intended-use population of adults aged 50 and older (People, tested on a stand-in)⁸.
• NHS-Galleri reported secondary performance analyses nested inside a randomised trial of participants without suspicion of cancer, added to usual care (People, tested on a stand-in)⁹.
• The accompanying commentary concludes that demonstrating population-level benefit remains the unresolved challenge (Not a study)¹⁰.

Not shown: that finding a cancer signal earlier makes anyone live longer. Not priced in the published performance results: overdiagnosis, false positives and the imaging they trigger (People, tested on a stand-in)^8,9.

Harms: no adverse-event or false-positive figures appear in the available reporting (People, tested on a stand-in)^8,9.

DOES THIS CHANGE WHAT I SHOULD DO?

No. This test has not been shown to add anything to the screening you already qualify for. If a clinic offers it, ask what a positive result triggers, who reads the follow-up imaging, and who pays for it. The randomised trial's clinical-outcome result would change this answer.

WHAT DIDN'T MOVE

Aging clocks and alcohol.

Organ-specific proteomic clocks read alcohol as slowing aging in some organs and accelerating it in others. The author attributes the split to clock construction, alcohol phenotyping and confounding control, not biology (Not a study)¹¹. A reading that depends on which clock a vendor licensed cannot direct a decision.

Shatavari for hot flushes.

A widely sold botanical got the right design, in 70 peri- and postmenopausal women over 56 days (People, tested on a stand-in)¹². No effect size is reported and safety was assessed without figures.

NAD+ precursors.

Nicotinamide riboside restored ACC1-deficient adipose immune cells in mice (Animals)¹³, beside two reviews (Lab, Not a study)^14,15. Still no human outcome data.

Senolytics.

A review argued senescence is too dynamic to target confidently (Lab)¹⁶, alongside a fisetin pilot registration and a company hair-loss statement (Not a study)^17,18. No senolytic has a human functional outcome.

• Organ-specific proteomic clocks gave opposite answers on alcohol (Not a study)¹.
• Pooled randomised trials found blood pressure and HbA1c changes do not fully explain the GLP-1 stroke reduction (People, tested on what matters)².
• Two large blood-based cancer-screening studies published together report performance, not benefit (People, tested on a stand-in, Not a study)^3–5.
• Menopause biology gained a 15,000-woman proteomic correlation and a cardiovascular position paper (People, watched, Not a study)^6,7.

Aging clocks ↑ two clocks, one exposure, opposite answers

What changedAn Aging Cell Perspective examined organ-specific proteomic clocks where alcohol appeared to slow aging in some organs and accelerate it in others (Not a study)¹.

• The author blames clock construction, alcohol phenotyping and confounding control, not biology (Not a study)¹.
• More exposure correlates landed the same week, from pesticide metabolites to immigration age (People, watched)^8–10, plus clocks built across 49 tissue and cell-line categories (Lab)¹¹.

Not tested: that any clock tracks a common process, or that moving a reading changes an outcome.

Harms: no clock in this week's evidence was validated against a clinical outcome, so no reading should move a treatment decision.

Commercial: a consumer result depends on which clock the vendor licensed, and that dependency is not disclosed at point of sale (Not a study)¹.

GLP-1 agonists ↑ a randomised stroke benefit blood pressure and HbA1c do not fully explain

What changedA meta-regression of randomised trials with adjudicated stroke endpoints in type 2 diabetes tested two candidate mediators (People, tested on what matters)².

Trial-level changes in systolic blood pressure and HbA1c do not fully explain the stroke reduction. Trial-level data cannot exclude mediation in individuals.

Two observational extensions: adults 65 and older with MASLD (People, watched)¹², and people with HIV, obesity and type 2 diabetes (People, watched)¹³.

Both are new-user electronic-health-record designs, so confounding by indication stays live. Neither tests aging.

Harms: none of the three reports adverse-event data.

Commercial: a randomised benefit these two mediators do not fully explain widens the indications a sponsor can pursue, and supports no aging claim.

Ovarian aging ↑ a 15,000-woman correlation and new hormone-therapy guidance

What changedTwo proteomic studies tied the menopause transition to brain aging markers, the same week FIGO and the World Heart Federation issued cardiovascular hormone-therapy guidance.

• The larger correlated hormone-linked molecular changes with poorer later-life cognitive outcomes (People, watched)⁶; the smaller staged 80 women by STRAW+10 (People, watched)¹⁴.
• In women with cardiovascular risk factors or established ischaemic heart disease, the position paper reports systemic therapy begun early after symptom onset as associated with long-term cardiovascular benefit (Not a study)⁷. An association is not a randomised cardiovascular outcome. Harms: no figures this week, and individual cardiovascular and cancer risk assessment still applies.
• Nothing shows hormone therapy prevents dementia, or that any signature is a target. Commercial: menopause companies gain a large-cohort correlation to cite, not a validated target.

CONTRADICTIONS

Rapamycin may not test its own hypothesis.

Two Nature fly studies identify Lsp2 linking mTORC1 activity, TOP-motif mRNA translation and lifespan, proposed to explain ribosomal protein synthesis that resists rapamycin (Animals)^15,16. Rapamycin itself cut plaque inflammation in aged atherosclerotic mice (Animals)¹⁷. A mammalian test of that fly output would show whether rapamycin trials can settle the mTOR question.

Autophagy's lifespan effect depends on conditions.

RNAi knockdown of autophagy genes in C. elegans shortened, lengthened or left lifespan unchanged by condition, including in long-lived daf-2 mutants (Animals)¹⁸. The authors offer that as why prior findings conflict; which conditions generalise, nobody has shown.

BUSINESS SIGNALS

Multi-cancer early detection has performance data and no benefit data. PATHFINDER 2 was a single-arm interventional study in more than 35,000 adults aged 50 and older, reporting safety and performance (People, tested on a stand-in)³. A single arm cannot show benefit. NHS-Galleri reported secondary performance analyses nested in a randomised trial of more than 142,000 participants (People, tested on a stand-in)⁴. A commentary alongside concludes population-level benefit remains unresolved (Not a study)⁵. Harms: no adverse-event or false-positive figures appear in the available reporting.

A mortality win, in one rare disease. A Phase III randomised trial of anselamimab improved all-cause mortality as a primary outcome in kappa light chain amyloidosis (People, tested on what matters)¹⁹. No effect size and no adverse-event data appear in the available reporting. It supports pricing and priority-review positions, not an aging claim.

INDIA WATCH

India is considering an amendment to Rule 30(Animals)²⁰ of the Medical Devices Rules, 2017, letting regulators halt manufacture and sale of unsafe or spurious devices and order their destruction (Not a study)²¹. Wearables and diagnostics sold into India would fall under it if adopted.

A multicentre randomised, double-blind, placebo-controlled trial gave Shatavari root extract at 400 mg/day or placebo to 70 peri- and postmenopausal women over 56 days, for climacteric symptoms (People, tested on a stand-in)²². No effect size is reported, and safety findings are reported as assessed, without figures. Shatavari is sold widely as a women's-health botanical, and a placebo-controlled trial is a genuine step above the testimonial evidence most of that category rests on. Fifty-six days and 70 participants support no long-term safety or healthspan claim.

BRIEFLY

• Mayo Clinic and Thermo Fisher Scientific launched Precure LLC to build a multi-omics dataset from up to one million biospecimens, Mayo holding majority ownership (Not a study)²³.
• Oracle and the Buck Institute announced a collaboration to build what they call an FDA-grade Intrinsic Capacity Score (Not a study)²⁴. FDA-grade is the companies' own phrase, and the announcement carries no study design. No aging marker has been qualified as a regulatory endpoint (Lab)²⁵.
• UnitedHealth and CVS are opposing a Medicare proposal that would change how physicians bill for remote patient monitoring (Not a study)²⁶.
• Mabwell reported Phase Ib/IIa data for an experimental COPD antibody, stating severe flare-ups fell by up to 100% in its best-performing dose group (People, tested on a stand-in)²⁷. That is a dose-group maximum, not a trial-level effect, and no safety data are reported.
• Telangana's Industries and IT Minister outlined state plans for a precision metabolic care pilot across state-run hospitals (Not a study)²⁸.

WATCHING

Nothing this week changes what you should buy or take. Two items are worth a question to a clinician. The AI-designed drug that made blood look "biologically younger" is a company re-analysing its own markers, and the lifespan headline was about mice. Hold off buying a biological-age test: no clock in the record has been validated against a clinical outcome as an endpoint. An independent pooling of 51 studies also found about half the anti-ageing products tested moved no clock at all. If you or a parent has non-urgent surgery coming, ask the surgical team whether they run a prehabilitation programme. Which exercise type and dose ranked best for bone is in the paper, not in the available reporting. Any of this would mean more if someone showed that moving the number moves how long or how well people live.

MOVE

Prehabilitation before surgery is being tested on the outcome that matters

A multicentre randomised trial tested three weeks of individualised prehabilitation before elective surgery (People, tested on what matters)¹. Patients were frail or pre-frail and aged 70 and over, with standard care as the comparator.

The endpoint is maintained independence and avoided long-term care dependency, not a biomarker (People, tested on what matters)¹.

Effect size: in the paper, not in the available summary. It has to be read before anyone changes a pre-surgical pathway.

Exercise in frail patients carries fall and cardiac risk, and the adverse-event profile is not reported (People, tested on what matters)¹.

DOES THIS CHANGE WHAT I SHOULD DO?

Maybe. Asking whether a surgical unit offers prehabilitation costs nothing, and the trial aims at independence rather than a surrogate. Until the effect size is read, that is a question to raise, not a service to buy privately.

Exercise for bone now has a dose-response ranking

A Bayesian network meta-analysis of randomised trials ranked exercise types and doses against bone mineral density and fracture risk in adults over 40 (People, tested on a stand-in)².

• Dose and modality ranking is what a clinician needs, and almost nothing sold for longevity has it (People, tested on a stand-in)².
• Bone mineral density is a surrogate. Fracture is the outcome.
• High-impact loading carries fracture risk in people who already have low bone density (People, tested on a stand-in)².

DOES THIS CHANGE WHAT I SHOULD DO?

Maybe. Which modality and dose ranked highest is not stated in the available reporting, so this is a reason to read the paper, not to change protocol. If you have diagnosed osteoporosis, have the loading prescribed rather than self-selected.

MEASURE

Half the anti-ageing products tested moved no clock at all

A Yale-led reanalysis pooled 51 studies of anti-ageing interventions through one harmonised pipeline (People, tested on a stand-in)³.

About half the tested products moved no epigenetic clock (People, tested on a stand-in)³. A Nature analysis separately ranked clocks by responsiveness to human interventions (People, tested on a stand-in)⁴.

Supply grew anyway: sex-specific multi-organ clocks (People, watched)^5,6, a retinal age gap with 2.66-year mean absolute error (People, watched)⁷, a brain-age product called Brain Years (Not a study)⁸.

• None of the clocks in this record has been validated against a clinical outcome as an endpoint. Methylation clocks tracked women's survival to 90 but not cognitive healthspan (People, watched)⁹.
• The blood draw is not the harm. What gets bought on the strength of the number is, and none of those purchases has outcome data.

INDIA. Clinic formats scaled this week. A physician-led longevity centre is opening in Whitefield, Bengaluru (Not a study)¹⁰, an operator has rebranded as The Longevity Centre with a celebrity ambassador (Not a study)¹¹, and Temple has acquired the UK clinic Longevous (Not a study)¹². None of these items reports clinical outcome data, and none states a price. Harms are the standard ones for unindicated testing: incidental findings and the investigations they trigger. No harms data exist for these specific services.

DOES THIS CHANGE WHAT I SHOULD DO?

No. A clock reading changes no management decision, and this week made the case against paying for one stronger, not weaker. Predicting death is a different claim from predicting good years, and these products are sold on the second.

TAKE

Senolytics gained mouse targets and lost ground in humans

Three preclinical studies refilled the pipeline in the week a pooled reanalysis cast doubt on some senolytic claims (People, tested on a stand-in)³.

• Blocking PD-L2 cut senescent-cell burden and improved metabolic and physical function in aged mice (Animals)^13,14. Soluble PD-L2 rose with human age and fell after senolytic treatment, a biomarker observation, not an outcome.
• UA-30 reduced frailty and extended lifespan in aged mice (Animals)¹⁵. Genetic clearance of p16-positive cardiac cells improved remodelling in mice at n=6 per group (Animals)¹⁶.
• No new human senolytic trial result appeared.
• None of the mouse studies reports adverse-event data.

DOES THIS CHANGE WHAT I SHOULD DO?

No. Nothing you can buy as a senolytic changed. If you already take one, nothing changed: the pooled reanalysis found many tested products moved no clock, which is a biomarker result, and no human senolytic outcome result has been reported.

WHAT STILL HOLDS

Standing positions, restated here because this week did not move them. The highest-stage evidence in this issue is a randomised prehabilitation trial, not a drug (People, tested on what matters)¹. Its effect size is not in the available reporting.

• No epigenetic clock has been validated against a clinical outcome as an endpoint (People, tested on a stand-in, People, watched)^3,9. More clocks arrived this week (People, watched, Not a study)^5–8, two of those reports possibly describing one study, and none changed that.
• No new human senolytic trial result appeared this week (People, tested on a stand-in)³. Every new target this week was in mice (Animals)^13,15,16.
• No randomised trial has tested metformin for aging endpoints in people without diabetes, and the observational signal is confounded by indication (Not a study)^17,18.

WHAT DIDN'T MOVE

• The AI-designed drug claim widened from one clock to six proteomic clocks, in a company-reported, unpublished post-hoc biomarker analysis with no peer review reported (People, tested on a stand-in)¹⁹, then recirculated through several outlets (People, tested on a stand-in, Not a study)^20–22. Repetition is not replication, and no adverse-event data accompany the claim.
• Semaglutide extended lifespan in female mice and slowed aging markers (Animals)^23,24. One sex, and maximum lifespan is unreported. Nobody should start or stop a GLP-1 because of it.
• Metformin drew two journalism pieces and no new data (Not a study)^17,18.
• A T-cell "longevity vaccine" programme was announced with no efficacy data (Not a study)²⁵. Stage not assignable.
• Trials were registered for calcium alpha-ketoglutarate (Not a study)²⁶, young-donor faecal transplant with a fermented diet (Not a study)²⁷, and diet, exercise and metformin in frail obese veterans (Not a study)²⁸. The supplements are on sale already; the trials that might justify them have not reported.

ONE FOOD, INTERPRETED. — Turmeric

What the evidence shows. As a culinary spice, turmeric is safe and its use goes back centuries; nothing here changes that. As a disease-treating supplement, the case is thinner. Cell and animal work is extensive, but no single human study is in view here, so stage not assignable, and the human trials that exist are small, short and confounded by poor oral bioavailability.

What is oversold. The jump from mechanism to disease claims. A culinary dose is not a supplement dose, and a supplement dose is not a tested drug.

INDIA. Culinary use is ancient and uncontroversial. No turmeric product, seller or trial appears in the record.

DOES THIS CHANGE WHAT I SHOULD DO?

No. Keep cooking with it. That part was never in question. Concentrated extracts are where harms would matter, and the human trials are too small and short to characterise them.

• A pooled reanalysis of 51 studies found about half the tested anti-aging products moved no epigenetic clock (People, tested on a stand-in)¹.
• Three or four more clocks arrived in the same week (People, watched, Not a study)^2–5. Supply grew faster than validation.
• The FDA accepted a Reasonable Expectation of Effectiveness determination for Loyal's LOY-003 in large- and giant-breed dogs (Not a study)⁶. Acceptance, not approval.

AGE CLOCKS ↑ The first pooled test a product claim can fail

What changed. An independent reanalysis put clock-moving claims on one measuring stick, while more clocks appeared, one of them a company-reported product (Not a study)³.

A Yale-led harmonised reanalysis pooled 51 studies of epigenetic-clock effects from tested anti-aging interventions; about half the products moved no clock (People, tested on a stand-in)¹. A separate meta-analysis ranked how responsive clocks are to human interventions (People, tested on a stand-in)⁷.

New entrants: sex-stratified multi-organ clocks (People, watched)^4,5 and a deep-learning retinal age gap with 2.66-year mean absolute error (People, watched)². A company-reported functional brain-age product called Brain Years also appeared (Not a study)³.

Not tested: no clock here has been validated against a clinical outcome as an endpoint. Methylation clocks tracked women's survival to 90 but not cognitive healthspan (People, watched)⁸.

Harms: the test itself is not the risk. What gets bought on the strength of the number has no outcome data.

Commercially, sellers of clock panels now face a pooled reanalysis in which about half of tested products moved no clock (People, tested on a stand-in)¹.

SENOLYTICS ↓ Targets keep arriving, humans do not

What changed. The preclinical pipeline refilled and the human record did not move.

• PD-L2 acts as a senescence-associated immune checkpoint; blockade cut senescent-cell burden and improved metabolic and physical function in aged mice (Animals)^9,10. Soluble PD-L2 rose with human age and fell after senolytic treatment, a biomarker, not an outcome.
• UA-30 inhibited RalA, reduced frailty and extended lifespan in aged mice (Animals)¹¹. Genetic clearance of p16-positive cardiac cells improved remodeling in aged mice at n=6 per group (Animals)¹².

Not tested: no new human senolytic trial result appeared this week.

Harms: none of the mouse studies reports adverse-event data.

Commercially, a buyer of senolytic supplements now has a pooled reanalysis showing about half of tested products moved no clock (People, tested on a stand-in)¹. That is a null on the biomarker these products are sold on, not an outcome result.

DOG LONGEVITY ↑ A second regulatory precedent

What changed. The FDA accepted a Reasonable Expectation of Effectiveness determination for Loyal's LOY-003, an oral drug lowering GH and IGF-1 in large- and giant-breed dogs (Not a study)⁶.

It follows LOY-001, reported as the first drug to receive an RXE acceptance for lifespan extension in any species (Not a study)⁶.

Not established: an RXE acceptance speaks to expected effectiveness only, and this one reported no efficacy data (Not a study)⁶. No product is on sale on this basis.

Commercially, lifespan extension now has two RXE acceptances as a workable indication in animals (Not a study)⁶.

CONTRADICTIONS

Do marketed interventions move clocks?

The 51-study reanalysis says about half do not (People, tested on a stand-in)¹. In the same week a company reported its drug reversing biological age across six proteomic clocks (People, tested on a stand-in)¹³. One is pooled and independent, the other unreplicated and self-reported. A pre-specified clock endpoint in a controlled trial would resolve it.

Semaglutide as a geroprotector.

It extended lifespan in female mice (Animals)^14,15. The human case rests on a narrative review citing the SELECT trial second-hand (Lab)¹⁶ and observational associations such as lower fracture risk in type 2 diabetes (People, watched)¹⁷. This is the same mouse result logged in W36, recirculating through new outlets. One sex, and maximum lifespan is unreported. The mouse studies report no adverse-event data, and nobody should start or stop a GLP-1 on this evidence.

BUSINESS SIGNALS

Novartis bought Sironax's blood-brain-barrier delivery platform for $125M (Not a study)¹⁸.

Paying for the pipe rather than a molecule says where Novartis thinks the CNS constraint sits, not that any drug crossing it will work.

A four-country prospective study of 1,785 people reported the PANXEON exosome biomarker plus CA19-9 detecting early-stage pancreatic cancer with high sensitivity (People, watched)¹⁹.

Early-detection assays with prospective data are the diagnostics most likely to reach reimbursement. No specificity figure is reported, so the false-positive burden and the biopsies it would trigger cannot be judged.

HexemBio raised $15.5M to move a stem-cell rejuvenation platform from transplant medicine into injectables and skin treatments (Not a study)²⁰.

The record does not say which product, or which market, comes first. A cosmetic or injectable launch would not require the efficacy evidence a transplant indication does. No efficacy or safety data accompany the raise.

INDIA WATCH

Temple, Deepinder Goyal's startup, acquired UK clinic Longevous and brought its cofounders in-house (Not a study)²¹.

• Prajeeva Healthcare is opening a physician-led longevity centre in Whitefield, Bengaluru (Not a study)²². A third-party estimate put Decode Age's 2025 annual recurring revenue at $2.9M (Not a study)²³, an estimate rather than an audited disclosure.
• No clinical outcome data or pricing accompany these services. Harms are the standard ones for unindicated testing: incidental findings and downstream investigation, with none reported for these specific services.

BRIEFLY

• Insilico published an open AI benchmark and language-model toolkit for aging biology as a Cell cover study (Lab)²⁴. A benchmark measures models, not aging.
• PRAEP-GO, a multicentre randomised trial of three weeks of individualised prehabilitation before elective surgery in frail and pre-frail patients aged 70 and over, reported in Lancet Healthy Longevity (People, tested on what matters)²⁵. The effect estimate is not in the available reporting. Exercise programmes in frail patients carry fall and cardiac risk, and the adverse-event profile is not reported.
• A BMJ Bayesian network meta-analysis ranked exercise types and doses against bone mineral density in adults over 40 (People, tested on a stand-in)²⁶. Which modality won is not in the available reporting. High-impact loading carries fracture risk in people who already have low bone density.

WATCHING

Nothing this week changes what you should do. You may have seen that an AI-designed drug made patients biologically younger: that analysis covered 42 patients over 12 weeks with no comparison group, so it cannot tell you the drug did anything. Vitamin D after a heart attack now has 4.2 years of outcome data and did not clearly cut heart events (People, tested on what matters)¹. If you take it for bone health or a measured deficiency, this does not apply. Nothing here justifies buying a biological age test.

TAKE

Vitamin D after a heart attack did not clearly reduce heart events

TARGET-D reported its 4.2-year result, and it did not show a clear reduction in cardiovascular events.

• Targeted vitamin D management did not clearly reduce major cardiovascular events after myocardial infarction, over 4.2 years of follow-up (People, tested on what matters)¹.
• It says nothing about vitamin D for bone health, or for a measured deficiency without a heart attack.
• Two company cardiovascular programmes also stopped short of an outcome. One missed its endpoint in a randomised outcomes trial, company-reported and unpublished (People, tested on what matters)². The other, an anti-inflammatory drug's two randomized trials, was discontinued for reasons not stated (Not a study)³.

Harms: no adverse-event data appear in the available reporting for any of the three.

DOES THIS CHANGE WHAT I SHOULD DO?

No. TARGET-D did not find a clear reduction in major cardiovascular events over 4.2 years (People, tested on what matters)¹. A null trial does not prove no effect, but nothing here supports taking vitamin D for the heart.

Six clocks fell in a single-arm trial of an AI-designed drug

Insilico published blood-sample analysis from its completed Phase IIa of rentosertib in idiopathic pulmonary fibrosis.

• Predicted biological age fell across six proteomic clocks over 12 weeks in 42 patients (People, tested on a stand-in)⁴.
• The trial was single-arm. With no control group, the fall cannot be attributed to the drug.
• No new efficacy data on the drug itself appeared. Several outlets dropped this from their coverage (Not a study, People, tested on a stand-in)^5,6.

Harms: safety and tolerability were not what this analysis measured (People, tested on a stand-in)⁴.

DOES THIS CHANGE WHAT I SHOULD DO?

No. This is a fibrosis drug reported on an aging endpoint it was never tested against. It tells no one, fibrosis patient or otherwise, that they will age more slowly.

Folate and homocysteine tracked biological aging in Indian older adults

An Indian cohort measured nutrient status against aging clocks over 4.5 years (People, watched)⁷.

• In adults aged 60 and over, higher folate was associated with slower biological aging and higher homocysteine with faster aging, on Phenotypic Age and methylation-based clocks (People, watched)⁷.
• This is association within a cohort. No supplement was given, so no benefit or harm from taking folate or B12 is established.

IndiaThe data come from LASI-DAD, a study built for Indian dementia research, so the population is the right one for Indian screening debates about folate and homocysteine (People, watched)⁷. The study measured folate and homocysteine, not B12, and an association cannot tell you whether screening or treating helps.

DOES THIS CHANGE WHAT I SHOULD DO?

No. Low B12 or folate on a blood test is a clinical matter with its own treatment, and that has not changed. A supplementation trial against an aging endpoint would be the thing to watch; none is in the record.

MEASURE

Clocks moved this week for cash, saunas and exercise

Three results show how easily an epigenetic age reading shifts. Two more show how little the reading carries.

• A randomized trial gave low-income US mothers $333 or $20 a month for four years. Children in the high-cash group showed a small reduction in DunedinPACE aging (People, tested on a stand-in)⁸.
• Eighteen firefighters using a portable sauna showed higher DNAmVO2max and AdaptAge, and higher DamAge as well (People, tested on a stand-in)⁹.
• Six months of exercise lowered epigenetic age in proportion to volume, plateauing near WHO guideline levels (People, tested on a stand-in)¹⁰.
• Proteomic clock age acceleration tracked disease and mortality risk, but added limited predictive value beyond standard lifestyle risk indicators (People, watched)¹¹.
• Genflow's beagle trial (Not a study, People, watched)^12,13 used a methylation clock as primary endpoint, the same week clock-to-outcome value was reported as limited (Animals, People, watched)^11,14.

Harms: none of these clock studies reports adverse-event data. The risk is what someone does with the number.

One study compared T-cell p16INK4a against five methylation clocks in 251 women treated for early breast cancer (People, watched)¹⁵. Whether the methods ranked the same women differently is not reported in the available coverage. The demonstrated discordance is elsewhere: 1,752 Alzheimer's polygenic score configurations placed the same individuals in both top and bottom risk strata (Lab)¹⁶.

DOES THIS CHANGE WHAT I SHOULD DO?

No. Treat any clock reading you pay for as one configuration among many, taken at one moment. Clock age acceleration tracks disease and mortality risk, but adds little beyond standard risk factors a doctor already measures (People, watched)¹¹, and no study shows moving the number changes the outcome. Even the most useful result, exercise, only moved a clock: the curve flattened near activity guideline levels (People, tested on a stand-in)¹⁰.

WHAT STILL HOLDS

Standing positions this week. Nothing moved them.

No clock reading has been shown to predict how long or how well you live better than ordinary risk factors already do.

Proteomic clock age acceleration tracked disease and mortality risk, but added limited predictive value beyond standard lifestyle risk indicators (People, watched)¹¹. What would change this is a clock that improves on standard indicators, or an intervention that moves a clock and an outcome together.

Animal lifespan results are still not human results.

One arrived this week: midlife suppression of growth hormone receptor production extended lifespan in mice (Animals)¹⁷. A second mouse study reduced colitis severity with systemic senolytics (Animals)¹⁸, which is a disease model, not a lifespan result.

A trial can meet its primary endpoint without anyone living longer or better.

Genflow's beagle trial used a methylation clock as that endpoint (Animals)¹⁴.

WHAT DIDN'T MOVE

GLP-1 drugs and lifespan.

Mouse coverage continued, and a 75-adult pilot randomizing GLP-1 patients to standard care, protein, or protein plus spices was registered (Not a study)¹⁹. No human outcome data on GLP-1 for aging appeared.

Muscle and bone during weight loss.

A review sets out how weight loss raises bone turnover and lowers bone density, a drug-driven counterweight still without outcome data (Not a study)²⁰.

Senolytics.

A Phase II of dasatinib plus quercetin in 80 frail or prefrail adults with HIV, aged 50+, was registered (Not a study)²¹; systemic senolytics also reduced colitis severity in mice (Animals)¹⁸. Registrations are plans.

NAD+ precursors.

NAD+ dysregulation was implicated in mouse colitis via T-cell senescence (Animals)¹⁸. Human trials still measure metabolites, not outcomes.

Metformin and rapamycin.

No new trial data for either.

Population-level interpretation of published evidence. Not medical advice, and not a recommendation to start or stop any prescription.

• Genflow reports that its randomized, blinded beagle trial met its primary endpoint, a methylation clock reading (Animals)¹.
• Insilico published a blood-sample analysis from its completed single-arm Phase IIa of rentosertib: predicted biological age fell across six proteomic clocks in 42 patients over 12 weeks (People, tested on a stand-in)².
• A separate analysis found proteomic clock age acceleration tracked disease and mortality risk, but added limited predictive value beyond standard lifestyle risk indicators (People, watched)³.
• Three cardiovascular programmes reported failure or discontinuation in one week (People, tested on what matters, Not a study)^4–6.
• ARPA-H plans to put $62.7 million into AI for directing heart failure treatment, aimed at FDA authorization (Not a study)⁷.

Clocks are being adopted as trial endpoints ↑

What changedA methylation clock served as the primary endpoint of a randomized animal trial, and a proteomic clock panel carried a human drug's aging story.

• Genflow reports its blinded SLAB trial in 24 beagles aged over 10 met its primary endpoint of reduced biological age, measured by the GRIM clock (Animals)¹. Two naked-DNA SIRT6 doses and one AAV8 arm were compared against saline.
• Insilico published a blood-sample analysis from its completed Phase IIa of rentosertib in idiopathic pulmonary fibrosis (People, tested on a stand-in)². Predicted biological age fell across six proteomic clocks over 12 weeks in 42 patients.
• Neither study tested survival or physical function. The Insilico trial was single-arm, so no control group exists to attribute the change to the drug. Coverage repeated the finding, some of it without the design (Not a study, People, tested on a stand-in)^8,9.

Harms: adverse events were not the endpoint in the dog trial, and safety and tolerability were not what the Insilico analysis measured (Animals, People, tested on a stand-in)^1,2.

Commercially: clock readings now sit at three layers, consumer tests, trial endpoints and marketing copy. Veterinary approval is a shorter path than a human one, which is why dog data carry commercial weight. This endpoint was a methylation clock, not survival or function, which is a weaker foundation for that path than it looks (Animals)¹.

The disagreement between biological age scores is now measured ↑

What changedThree papers ran the same people through competing scoring methods. Only one has so far reported the size of the disagreement.

• Six proteomic clocks were run on the same rentosertib serum samples and tested for agreement; the agreement result is not stated in the available reporting (Lab)¹⁰.
• T-cell p16INK4a was compared against five methylation clocks in 251 women treated for early breast cancer; the comparison result is not stated in the available reporting (People, watched)¹¹.
• 1,752 Alzheimer's polygenic score configurations placed the same individuals in both top and bottom risk strata, depending on method and ancestry (Lab)¹².
• None of the three reports a treatment result, so no harms question arises.

Commercially: a product selling a single biological age number is selling one configuration among many that disagree. Standardization would make those products comparable, and would make some of them look worse.

Three cardiovascular programmes failed or were halted in one week ↓

What changedThree cardiovascular strategies reported failure or discontinuation in one week.

• Novartis's lipoprotein(a)-lowering drug missed its randomised, controlled cardiovascular outcomes trial (People, tested on what matters)⁴, prompting questions about the target itself.
• Novo Nordisk halted two randomized controlled trials of an anti-inflammatory cardiovascular drug (Not a study)⁵.
• TARGET-D found targeted vitamin D management did not clearly reduce major cardiovascular events after myocardial infarction over 4.2 years (People, tested on what matters)⁶.

Harms: safety data are not reported for any of the three (People, tested on what matters, Not a study)^4–6.

Commercially: two of the three targeted a plausible mechanism or biomarker and missed the outcome. The Novo trials were halted, and the reason for the discontinuation is not stated in the available reporting (Not a study)⁵. Vitamin D management sold for cardiac prevention now has a randomized test that did not find the benefit (People, tested on what matters)⁶.

CONTRADICTIONS

Clocks as a drug-response measure.

Predicted biological age fell across six proteomic clocks in 42 single-arm patients (People, tested on a stand-in)^2,13. A separate analysis found proteomic clock acceleration tracked disease and mortality risk but added little beyond standard lifestyle risk indicators (People, watched)³. The first supports clocks as a response measure, the second says they add little to what is already measured.

Sauna.

In a pilot of 18 firefighters using a portable sauna, DNAmVO2max and AdaptAge rose, and DamAge rose too (People, tested on a stand-in)¹⁴. Whether a control group was used is not stated in the available reporting. One set of readings says benefit, the other says damage.

BUSINESS SIGNALS

ARPA-H, $62.7 million, heart failure AI

(Not a study)⁷. The agency plans to fund AI systems for directing heart failure treatment, aimed at FDA authorization. Public money buying regulatory-grade evidence for clinical AI is rarer than private money buying AI products. No clinical result exists yet. Track whether an authorization follows, and on what evidence.

C15:0 gets a Phase II after launch

(Not a study)¹⁵. A registered trial will test 200 mg or 300 mg daily against placebo in 93 adults aged 55 and over, with red blood cell distribution width as the primary outcome. The supplement is already on sale. A registration is a plan, carrying no efficacy and no harms data, and the result can go either way.

INDIA WATCH

Front-of-pack warnings.

India's food industry asked the Supreme Court to base warnings on a per-serving benchmark instead of FSSAI's per-100-gram plan (Not a study)¹⁶. Manufacturers define serving size, so the denominator decides how many products carry a warning. The All India Food Processors' Association argues the per-100-gram standard misclassifies traditional foods and burdens smaller manufacturers. A per-serving rule would let repackaging substitute for reformulation. Neither benchmark has been tested against health outcomes.

LASI-DAD.

In Indian adults aged 60 and over, higher folate was associated with slower biological aging and higher homocysteine with faster aging over 4.5 years (People, watched)¹⁷. The measures were Phenotypic Age and methylation-based clocks. This is association within a cohort. No supplement was tested, so neither benefit nor harm from taking folate or B12 follows from it.

BRIEFLY

• Oura filed for a US IPO targeting up to a $3B raise and a $16B+ valuation (Not a study)¹⁸.
• Qualcomm led a $70M round in Ultrahuman's smart-ring computing platform (Not a study)¹⁹.
• Temple acquired longevity clinic Longevous ahead of a wearable device launch (Not a study)²⁰.
• Neko Health reported $21.9M in 2025 ARR at a $1.8B valuation (Not a study)²¹.
• A market-research report projects the biological age and longevity diagnostics market at $5.58B by 2030 (Not a study)²².
• Project Serotonin launched a preventive-health platform for Indian hospitals and clinics (Not a study)²³.

WATCHING

Nothing this week changes what you should do. You have probably seen that Ozempic-type drugs "extended lifespan". That happened in mice, and dosing began when the mice were already old. Nobody has run the equivalent human trial, and it would take years. If you already take one of these drugs, ask your doctor how your strength and physical function are being checked. Two trials of muscle-protection adjuncts are registered, and neither has reported results. Biological age tests predict mortality in cohorts. None of this week's clocks was tested as a treatment target, so a reading buys a prediction and no plan.

TAKE

Semaglutide extended lifespan in mice that were already old

Two mouse studies gave GLP-1 drugs a lifespan result. The human evidence stayed observational.

• Semaglutide started at 20 months of age extended lifespan in female mice and slowed aging hallmarks (Animals)¹.
• A second study extended median lifespan in mice through GLP-1 receptor agonism (Animals)². Maximum lifespan is unreported.
• Three other mouse interventions begun in middle or old age also extended lifespan (Animals)^3–5. Lifelong dosing may not be required, in rodents.
• The human side stayed observational. A records-based cohort linked GLP-1 initiation around a transient ischaemic attack to lower later stroke and death (People, watched)⁶. No absolute risks are given.

Not shown: no human trial has tested these drugs against an aging endpoint, and none of the mouse studies reported harms.

Harms in people: bone density tracked the degree of weight loss over 12 months in 70 people with type 1 diabetes (People, watched)⁷, a population that is not the typical weight-loss user. A review treats sarcopenia during incretin therapy as a real risk (Not a study)⁸; a perspective disputes it (Not a study)⁹.

A review states bariatric surgery remains superior to high-dose semaglutide and tirzepatide (Not a study)¹⁰. Those drugs give roughly 15-22% weight reduction at one year.

DOES THIS CHANGE WHAT I SHOULD DO?

No. A mouse lifespan result is not a reason to seek a prescription (Animals)^1,2. If you already take one of these drugs for an approved indication, nothing here changes that decision. Dose, duration and safety for an aging purpose in older adults are untested.

Muscle-protection products entered trials before the muscle harm was measured

Three clinical bets were placed in one week against muscle loss on GLP-1 drugs.

Registered, no results: enobosarm alongside GLP-1 therapy, with physical function among the endpoints (People, tested on a stand-in)¹¹.

Registered, no results: a placebo-controlled ubiquinol trial for fatigue and strength, aimed at Wegovy and Mounjaro users (People, tested on a stand-in)¹².

Reported, weakly: a 12-week coaching programme improved body composition, without randomisation (People, tested on a stand-in)¹³.

The disagreement: a perspective argues lean mass loss need not mean worse muscle quality, strength or function (Not a study)⁹. A review treats sarcopenia during incretin therapy as a real risk (Not a study)⁸.

• Measured strength and function outcomes are unreported in all four of those studies (People, tested on a stand-in, Not a study)^9,11–13.
• In obese mice matched for weight loss, tirzepatide preserved blood stem cell function while calorie restriction caused cytopenias (Animals)¹⁴.

Harms: neither registered adjunct has reported safety results in GLP-1 users.

DOES THIS CHANGE WHAT I SHOULD DO?

No. Nothing here supports adding a second product to a prescription, and sellers reach you faster than trials report. The question worth asking your doctor is how your strength and physical function are being measured, because none of this week's work measured them.

An mTOR inhibitor enters a human aging trial that measures markers

A randomised trial of low-dose everolimus was registered against the hallmarks of aging.

The design: 24 weeks of dosing in insulin-resistant or prediabetic adults aged 55 to 80, with a younger untreated comparison group (People, tested on a stand-in)¹⁵.

Endpoints are physiological and molecular hallmarks, not disease or death (People, tested on a stand-in)¹⁵.

Harms: immunosuppression and metabolic effects in this drug class are real and dose-dependent (People, tested on a stand-in)¹⁵.

DOES THIS CHANGE WHAT I SHOULD DO?

Not yet. Even a positive result would show markers moving over 24 weeks, not longer or healthier life (People, tested on a stand-in)¹⁵. If you already take a drug in this class, infection risk and glucose control are worth reviewing with a physician (People, tested on a stand-in)¹⁵.

MEASURE

Three new biological age tests, and a paper attacking how all of them are built

New age measures arrived from urine, chest X-rays and immune cells, alongside an objection to the method behind them.

The measures: urine peptides (People, watched)¹⁶, chest radiographs (People, watched)¹⁷ and immune cell profiling (People, watched)¹⁸. Each predicts mortality or surgical risk in retrospective data.

The objection: the machine-learning regressions used to build such clocks carry systematic prediction bias (Lab)¹⁹. That bias can distort or reverse the associations the clocks are sold on.

• Neither side cites the other. This is an open argument about method, not a failed replication.
• One number also averages over real differences. Tissue aging ran organ-specific across 40 tissues (People, watched)²⁰, and immune aging differed between men and women (People, watched)²¹.

Harms: none of these studies reports harms from testing (People, watched)^16–18. The risk is acting on a number that was never tested as a treatment target.

IndiaA clock built on 4,240 adults in the Phenome India-CSIR cohort targets cardiometabolic risk, because existing models were built mostly on Western populations (People, watched)²².

DOES THIS CHANGE WHAT I SHOULD DO?

No. A ranking across a cohort is not an instruction for one person. If a reading would not change a decision you and your physician could otherwise make, it buys worry. Indian buyers carry a second problem: the models were built mostly on Western populations (People, watched)²².

WHAT STILL HOLDS

Standing positions, restated. The only section here not reporting this week's evidence.

No human trial tests any geroprotector against an aging endpoint.

Five mouse lifespan results arrived this week (Animals)^1–5, and the human counterpart exists for none of them. The gap has not closed.

A clock predicts, it does not direct.

Ranking mortality risk in a cohort is not evidence that moving the number moves the outcome. None of this week's clocks was tested as a treatment target (People, watched)^16–18.

Observational human data stayed observational.

The strongest human evidence tying GLP-1 drugs to lower stroke and death is a records-based cohort (People, watched)⁶. Cohorts rank risk; they do not establish cause.

WHAT DIDN'T MOVE

NAD+ precursors and senolytics.

A review concluded the evidence is insufficient to support NAD+ supplementation for delaying aging (Not a study)²³. A senescence roadmap listed what the field still lacks (Not a study)²⁴, and a fisetin sepsis trial was registered (People, tested on a stand-in)²⁵. No human results appeared this week for senolytics, metformin, rapamycin, NAD+, taurine or spermidine.

Biological age diagnostics forecast.

A syndicated projection put the market at $5.58 billion by 2030 (Not a study)^26,27, with no clinical data on any panel.

Tea and cardiovascular risk.

A review concluded causal inference remains uncertain (Not a study)²⁸. Trials used surrogate outcomes, and Mendelian randomisation has not consistently supported an effect.

Vitamin D3 versus D2.

One news report described D3 outperforming D2 on muscle and cardiovascular markers (People, tested on a stand-in)²⁹. No methods are given, so stage is not assignable, and markers are not outcomes.

Vegetarian and vegan diets.

A Dutch study reported inadequate protein and micronutrient intake in long-term vegetarians and vegans (People, watched)³⁰. It measured 20 vegetarians and 6 vegans with 3-day food records at one time point. That cannot guide anyone, including the Indian vegetarians it will be quoted at.

Stem cell and exosome products.

US manufacturing of MuseCells, exosomes and secretome products was agreed (Not a study)³¹, with no clinical data attached. If a clinic offers you these, supply exists and evidence of benefit does not.

INDIA, dialysis.

Hepatitis B, pneumococcal and influenza vaccination gaps, plus a substantial tuberculosis burden, were found at dialysis initiation in a North Indian tertiary centre cohort (Not a study)³². Cheap to fix, and worth raising at initiation.

_Population-level interpretation, not individual medical advice. Nothing here recommends starting or stopping a prescription._

• GLP-1 drugs acquired a rodent lifespan result. Label, price and reimbursement are unchanged (Animals)^1,2.
• Three mouse studies began dosing in middle or old age and still extended lifespan: semaglutide at 20 months, bacterial peptidoglycan at 18, growth hormone receptor ablation at 12 (Animals)^1,3,4.
• Four new biological age measures shipped, and one methods paper attacked the class all four belong to (People, watched, Lab)^5–9.
• A gene-editing cholesterol therapy entered humans in Australia (Animals)¹⁰.
• A Canadian reimbursement advisory committee recommended public funding for Leqembi (Not a study)¹¹.

GLP-1 CROSSES FROM METABOLIC OUTCOMES TO RODENT LIFESPAN ↑

What changed: GLP-1 drugs now have a mouse lifespan result.

• Semaglutide started at 20 months of age extended lifespan in female mice and slowed aging hallmarks (Animals)¹. A second study reports GLP-1 receptor agonism extending median mouse lifespan (Animals)².
• Maximum lifespan is unreported in both. Neither tested a human aging outcome.
• The human addition was one retrospective cohort. It linked GLP-1 initiation after transient ischaemic attack to lower subsequent stroke and death (People, watched)¹². The design is open to confounding by indication.

Harms: muscle and bone change during weight loss remain the live human harms (Not a study, People, watched)^13,14. Neither mouse study reported adverse events.

Commercially: manufacturers gain a citation, not a claim. No regulator grades mouse lifespan, and no pricing decision should rest on one.

AGE CLOCKS MULTIPLY WHILE THE METHOD IS ATTACKED ↑

What changed: four new biological age measures appeared in one week, alongside a paper challenging how the class is built.

• A urinary peptidomic clock predicted all-cause mortality in 1,811 people (People, watched)⁵. An immune-aging score predicted cardiovascular risk in UK Biobank (People, watched)⁶. A deep-learning model read biological age off chest X-rays, adding prediction beyond EuroSCORE II (People, watched)⁷. A fourth swapped epigenetic age for chronological age inside the PREVENT equation (People, watched)⁸; whether discrimination improved is unreported.
• All four are prediction studies. None tested whether acting on the number improves the outcome it predicts.
• A methods paper argues the machine-learning approach these clocks share carries systematic prediction bias that can distort or reverse downstream associations (Lab)⁹. A perspective asks what should count as proof of age reversal (Not a study)¹⁵.

Harms: no study this week reported harms from age-clock testing.

Commercially: four measures became licensable, one objection became answerable. Separately, four studies describe aging as organ-specific and sex-specific trajectories (People, watched, Animals)^16–19. If that holds, one biological age number is the wrong shape for the product.

MUSCLE-PROTECTION ADJUNCTS OUTRUN THE EVIDENCE UNDER THEM ↑

What changed: three products and programmes now sell muscle protection to GLP-1 users, and a perspective questioned the premise they share.

• A proof-of-concept trial of enobosarm alongside GLP-1 therapy was registered (People, tested on a stand-in)²⁰. A placebo-controlled ubiquinol trial for fatigue and strength in GLP-1 users was registered (People, tested on a stand-in)²¹. Both are registrations, so stage is not assignable.
• A 12-week quasi-experimental coaching programme reported better body composition than GLP-1 therapy alone (People, tested on a stand-in)²². It was not a randomised comparison.
• A perspective argues DXA lean mass loss during incretin therapy need not mean worse muscle quality, strength or function (Not a study)²³. A narrative review sets out the sarcopenia concern (Not a study)¹³. An observational study tracked 12-month bone density by degree of weight loss in 70 people with type 1 diabetes (People, watched)¹⁴.

Harms: neither registered adjunct has reported safety or efficacy results in GLP-1 users.

Commercially: every endpoint in this work is lean mass, a surrogate. If the perspective (Not a study)²³ is right, the case for these adjuncts rests on a measure that may not track strength or function.

CONTRADICTIONS

Do the clocks predict, or does the method manufacture the prediction?

Four new biological age measures reported associations with mortality or disease (People, watched)^5–8, and a fifth association study (People, watched)²⁴ is in the same week's record without an assignable stage. A methods paper argues their shared approach carries bias that can distort or reverse exactly those associations (Lab)⁹. This is a fight about method, not a failed replication. Re-analysis of the same cohorts with bias-corrected estimation would settle it.

Is a GLP-1 drug a calorie-restriction mimetic?

The semaglutide lifespan study frames its effects as paralleling calorie restriction (Animals)¹. A matched-weight-loss preclinical comparison found tirzepatide preserved blood stem cell cycling where calorie restriction caused multilineage cytopenias (Animals)²⁵. Different drugs, and nobody has shown which reading holds.

How much actually happened?

Three findings arrived twice from separate sources: one market projection (Not a study)^26,27, one infection-burden cohort (People, watched)^28,29, and one transcriptomic entropy result carrying conflicting stage tags of Lab and People, watched (Lab, People, watched)^30,31. Counting syndicated copies separately overstates the week.

BUSINESS SIGNALS

Scribe Therapeutics began a Phase 1 of STX-1150 in Australia

(Animals)¹⁰. The company also presented non-human primate data showing up to 90% PCSK9 silencing and LDL-C reductions up to 68% (Animals). Those numbers describe the edit, not a cardiovascular outcome. Harms: no human safety data exist, and a permanent edit cannot be withdrawn. If a single dose produced durable LDL lowering, it would compete with lifelong statin prescribing, which this week's evidence covers only as a clinical question (People, watched)³². No spending data appear in this week's evidence.

A Canadian drug reimbursement advisory committee issued a final positive recommendation to publicly fund Leqembi

in mild cognitive impairment or mild dementia due to early Alzheimer's disease (Not a study)¹¹. Provincial and territorial decisions are the next gate. No efficacy or safety data for lecanemab appear in this week's evidence, so no harms position follows from it. Amyloid antibodies are won or lost at reimbursement, not approval.

Celularity and MuseCell Innovations agreed to establish US manufacturing of Dezawa MuseCells

at Florham Park, New Jersey, covering the cells plus exosome and secretome derivatives from perinatal, bone marrow and adipose sources (Not a study)³³. The agreement carries no claim about clinical effect. Capacity is being committed ahead of outcome data, and that gap is where clinic marketing lives.

INDIA WATCH

GLP-1 generics are already in India.

Whether uptake is rising or slowing is not established by anything in this week's evidence. The only India item on these drugs is a BMJ commentary on what generic competition could mean for access, and it carries no company figures (Not a study)³⁴. Anyone sizing an India opportunity off the mouse result has nothing here to size it with.

An age clock trained in India was published.

It was built on 4,240 adults in the Phenome India-CSIR cohort to predict cardiometabolic risk (People, watched)³⁵. It was trained in the population it would be applied to. Whether it predicts better than a Western-trained clock in Indian adults is unreported, so the transferability gap is addressed in design rather than demonstrated in results. The prediction bias described in 2074 applies to it too.

Burden lands on a younger population with thinner infrastructure.

Aging-related disease was the largest share of global burden, including in low-income countries (People, watched)³⁶. A national survey documented complex multimorbidity and palliative care gaps among older Indians (People, watched)³⁷.

WATCHING