Nir Barzilai

Nir Barzilai is an Israeli-American physician-scientist at the Albert Einstein College of Medicine in New York, where he directs the Institute for Aging Research and leads a long-running study of exceptionally long-lived Ashkenazi Jewish families. He is best known as the principal designer of TAME, a proposed trial of Metformin and the TAME trial whose endpoint was constructed to force a regulator to treat aging as something a drug can be approved to address. As of 2026 TAME had not been funded at the scale it requires, had not enrolled, and had reported no results.

§Overview

Barzilai's two bodies of work rest on different kinds of evidence. One is observational human genetics: a cohort of centenarians and their children, assembled to ask which inherited variants separate people who reach extreme age from people who do not. The other is regulatory strategy: an argument that the Geroscience hypothesis cannot be tested until an endpoint exists that a health authority will accept for a drug aimed at aging rather than at a named disease. The genetics is descriptive and largely uncontested. The trial built to settle the second question has never been funded at the scale it requires.

§Career

Barzilai took his MD at the Technion in Haifa and trained in internal medicine and endocrinology before joining the faculty of the Albert Einstein College of Medicine in the Bronx. His early laboratory work concerned insulin action, visceral fat, and the metabolic changes that accompany aging in rodents, which is the route by which he arrived at the question of why some people escape them. He is the founding director of Einstein's Institute for Aging Research and directs its Nathan Shock Center of Excellence in the Basic Biology of Aging, one of a small number designated by the US National Institute on Aging. He co-founded CohBar, a company formed to develop mitochondria-derived peptides as drugs, and holds advisory roles across the sector. His 2020 trade book Age Later sets out the case for treating aging as the shared upstream cause of the diseases of old age.¹

1. Late 1990s
   Longevity Genes Project begins
   Barzilai's group at Albert Einstein College of Medicine starts recruiting Ashkenazi Jewish people aged 95 and over, their adult children, and age-matched controls.
2. 2003
   Lipoprotein genotype reported
   A CETP variant is found over-represented among the long-lived, whose carriers have larger lipoprotein particles. The association later fails to yield a drug.
3. 2011
   Lifestyle comparison published
   People in the cohort who passed 95 report drinking, exercise, diet and weight histories no better than those of a national survey sample from the same birth cohort, a result widely over-read in popular coverage.
4. 2015
   Composite endpoint discussed with US regulators
   Barzilai and colleagues present the TAME design; regulators indicate a composite of age-related events could be acceptable in principle.
5. 2017
   Growth hormone receptor result
   A deletion in exon 3 of the growth hormone receptor is associated with longer life in men across several cohorts, and with taller stature.
6. 2018
   TAME biomarkers framework
   A convened workgroup publishes which blood-based measures a geroscience trial should collect, building infrastructure a future application would need.
7. 2020
   Age Later published
   A trade book arguing that aging is the shared upstream cause of the diseases of old age.
8. 2026
   TAME still unreported
   More than a decade after it was designed, the trial has not been funded at scale and has reported no results.

§The Longevity Genes Project

From the late 1990s Barzilai's group recruited several hundred Ashkenazi Jewish people aged 95 and over, along with their adult children and a comparison group of similar age whose parents had ordinary lifespans. The children are the analytically useful part of the design: young enough to be measured directly, sharing half the genome of a demonstrated survivor, and comparable on disease rates with their contemporaries, which is how the project connects to compressed morbidity rather than to lifespan alone.

The most-quoted result is negative. On self-reported alcohol use, physical activity, diet and body weight, the people in the cohort who had passed 95 looked no more virtuous than a general-population sample from their own birth cohort.² The finding says that in this population extreme longevity is not explained by behaviour; it does not say that behaviour is irrelevant to ordinary lifespan, and Barzilai has been careful in print to keep the two apart. The distinction is routinely lost in popular coverage, where the result is read as licence to ignore the evidence behind exercise, and it sits awkwardly beside the lifestyle-first framing of long-lived population studies.

The genetic findings are associations, and are described here as such. A variant of the cholesteryl ester transfer protein gene, CETP, was over-represented among the long-lived, and carriers had larger high- and low-density lipoprotein particles.³ Functionally significant mutations in the insulin-like growth factor 1 receptor were later reported as enriched among centenarian women, connecting the human cohort to the insulin and IGF-1 signalling axis that governs lifespan in nematodes and mice, the pathway opened by Kenyon's daf-2 work.⁴ A deletion in the third exon of the growth hormone receptor was associated with longer life in men across several cohorts, and with taller stature in carriers, a male-specific result that complicates the simple story that less growth signalling means more years.⁵

What a founder population buys and costs

A genetically homogeneous population reduces the background variation that swamps association studies, so a modest cohort can detect variants a general-population study would miss. The price is generalizability: a variant enriched in Ashkenazi centenarians may be rare, differently penetrant, or absent elsewhere, and several of these associations have replicated unevenly.

§From variant to drug

The CETP story shows how far an association sits from a therapy. Pharmaceutical companies pursued CETP inhibition hard on the strength of its lipid biology. The first such drug to reach a large outcome trial, torcetrapib, raised HDL cholesterol and also raised mortality, and the trial was stopped.⁶ Later compounds in the class were safer and did little; none became a longevity drug. The inference that a variant carried by long-lived people names a druggable target has not held.

Association is not mechanism

None of the variants identified in centenarian cohorts has been shown to cause exceptional longevity, and no drug derived from one has extended human healthy life. The cohorts also carry a survivorship problem: people who reach 95 are, by construction, unrepresentative in ways no covariate captures.

§TAME and the regulatory argument

TAME, Targeting Aging with Metformin, was designed by Barzilai and colleagues and organized through the American Federation for Aging Research, where he has held a scientific leadership role. As proposed it would randomize roughly 3,000 adults aged 65 to 79 to metformin or placebo across multiple US sites, follow them for about six years, and count time to the first of a composite of unrelated age-related events.⁷ The design and the evidence behind the drug are set out under Metformin and the TAME trial.

The composite is the substance of the proposal. Regulators approve drugs for diseases, and there is no approval pathway for aging, so a drug that modestly delayed several conditions at once would fail every existing trial design while doing exactly what geroscience predicts. US regulators met the investigators in the mid-2010s and indicated that such an endpoint could be acceptable in principle. A related workgroup Barzilai convened set out which blood-based biomarkers a geroscience trial should collect, an attempt to build the measurement infrastructure such an application would need.⁸ Neither an Epigenetic clocks reading nor any other composite age measure is accepted as a surrogate endpoint, which is why TAME was built around hard clinical events.

The trial has never been fully funded. Metformin is generic, so no sponsor can recover a budget in the tens of millions of dollars from sales, and public money has arrived in amounts far short of what a multi-centre outcome trial costs. That the field's most discussed trial has gone a decade without enrolling, through years of fundraising, is itself a substantive fact: the interventions cheap enough to give to everyone are the ones nobody owns.

§Reception

Barzilai is taken seriously as a clinician-scientist and is among the field's most effective advocates with funders and regulators. The criticism is aimed at the vehicle rather than the design. Metformin's geroprotective evidence in people without diabetes is weak, resting largely on observational comparisons vulnerable to confounding by indication; the US National Institute on Aging's Interventions Testing Program has not reported a lifespan extension from metformin in genetically heterogeneous mice, in contrast to its result for Rapamycin. A trial published in 2019 found that metformin blunted the mitochondrial adaptation of older adults to aerobic exercise training,⁹ which would be an unfortunate property in a drug given to healthy people for decades.

Right argument, wrong drug

Critics generally accept that a multi-disease endpoint is what geroscience needs and doubt that metformin should carry it. Their argument is that a null result would be read as evidence about aging as an indication rather than about one weak biguanide, so a cheap drug with thin preclinical support puts the precedent at risk. Barzilai's answer has been that metformin was chosen for a safety record accumulated in millions of older people and for a price that would make a positive result usable everywhere.

His public framing is more restrained than that of several contemporaries. He argues for delaying the onset of disease and extending Healthspan rather than for indefinite lifespan, a position well short of escape-velocity claims of the kind associated with Aubrey de Grey, and unaccompanied by the supplement advocacy that has drawn criticism onto David Sinclair. Colleagues who reject the drug rarely reject the framing, which is part of why the proposal keeps a hearing.

§Legacy

The design has outlasted the effort to pay for it. Composite age-related endpoints, offspring-of-centenarian cohorts, and the language of the The longevity dividend — the economic argument that delaying aging returns more than curing any single disease — now appear routinely in geroscience proposals. Aging-biology centres such as the Buck Institute for Research on Aging frame their work as one upstream target with many diseases downstream, and the XPRIZE Healthspan competition defines winning as function restored across several domains rather than one disease improved. His insistence that regulatory acceptance, not molecular novelty, is the binding constraint is now close to consensus among people who want geroprotectors deployed rather than discussed, and it shapes how Senolytics and other candidate classes plan their trials.

The counterfactual is instructive. While TAME waited, the GLP-1 receptor agonists accumulated large randomized outcome data across weight, cardiovascular, and kidney endpoints in adults with obesity or diabetes, because a patented class had sponsors who could pay. Whether those results say anything about aging as such is unsettled, but they show how multi-organ evidence actually gets produced, and it is not the route TAME was designed to take. The open question Barzilai's career poses is not whether the Hallmarks of aging describe a shared biology. It is whether a public good with no owner can be tested at all, and who is left out if the answer arrives only through drugs someone can sell.

§See also

• Metformin and the TAME trial
• Geroscience hypothesis
• The longevity dividend
• Aging biomarkers
• Compression of morbidity
• Cynthia Kenyon
• Healthspan
• Access and inequality

§References