Interventions Testing Program

The Interventions Testing Program (ITP) is a programme of the US National Institute on Aging that tests whether candidate compounds extend the lifespan of mice, running every test in parallel at three independent sites under an identical protocol. Since its first cohorts in 2004 it has become the closest thing biogerontology has to a benchmark: a lifespan claim that survives the ITP is taken seriously, and one that fails there is substantially weakened however striking the original report. Its most consequential outputs include the 2009 demonstration that Rapamycin extends mouse lifespan when begun late in life, and a long list of famous candidates that did nothing.

§Overview

The ITP exists because the mouse-longevity literature it inherited was unreliable. A lifespan result from one laboratory, in one inbred strain, on one diet, frequently failed to reproduce anywhere else; a compound could look life-extending simply because it corrected a quirk of a single genotype. The programme's design answers each failure mode directly. Every compound is tested simultaneously at the Jackson Laboratory, the University of Michigan, and the University of Texas Health Science Center at San Antonio. The mice are genetically heterogeneous UM-HET3 animals, bred as a four-way cross so that every mouse is a genetically distinct sibling of every other, which prevents any result from being an artifact of one inbred genome. Both sexes are tested, doses are verified in blood, and survival is pooled across the three sites before analysis.¹

Anyone may propose a compound. The programme runs an annual open call, and nominations have come from academic laboratories, companies, and independent researchers; the sponsor of a compound is typically involved in designing the dose and interpreting the result. This openness is deliberate. The ITP is a public filter for the field's hypotheses, including the ones its own investigators doubt.

§History

The National Institute on Aging launched the programme in 2004, after several years of planning among investigators who had watched single-laboratory lifespan claims collapse on replication. The founding design — three sites, heterogeneous mice, pooled analysis, pre-specified statistics — was described by Richard Miller, David Harrison, Randy Strong and colleagues in a 2007 interim report that remains the reference for how the programme works.¹ The Methuselah Foundation's Mprize, created around the same period, approached the same problem from the opposite direction, rewarding record lifespans rather than standardising their measurement.

The programme's defining moment came in 2009, when it reported in Nature that rapamycin fed to mice from 600 days of age — roughly equivalent to a human in their early sixties — extended lifespan in both sexes at all three sites.² It was the first drug shown to extend mouse lifespan robustly when started that late, and the result, an accident of a formulation delay that pushed the start date back, did more than any other single finding to make aging look pharmacologically tractable. It is a core exhibit for the Geroscience hypothesis, and it launched the wave of interest in mTOR inhibition and Autophagy that continues today.

§How a compound is tested

UM-HET3 mice are the offspring of two F1 hybrid parents, drawing on four inbred grandparental strains. Compounds are almost always delivered in food, at a dose negotiated with the sponsor and confirmed by measuring blood levels; treatment usually begins in early adulthood, though some agents are deliberately started late to model treatment of already-aged animals. The primary endpoint is survival, pooled across sites and analysed separately by sex, with secondary studies of Healthspan measures and pathology in parallel cohorts.¹ The design trades breadth for rigour: a compound is typically tested at one or a few doses, and a null result is a statement about those doses and that schedule, not about the molecule in every possible regimen.

§Results

1. 2004
   First cohorts enrolled
   Testing begins at three sites in genetically heterogeneous UM-HET3 mice under a shared protocol.
2. 2009
   Rapamycin extends lifespan late in life
   Begun at 600 days of age, rapamycin raises age at 90% mortality by 14% in females and 9% in males, at all three sites.
3. 2011
   Resveratrol fails
   Rapamycin extends lifespan again at a different dose; resveratrol and simvastatin do not, at the doses tested.
4. 2014
   Acarbose and 17-α-estradiol
   Acarbose raises median male lifespan 22% but female lifespan only 5%; 17-α-estradiol and NDGA extend lifespan in males only.
5. 2016
   Metformin alone fails
   Metformin at 0.1% of diet does not significantly extend lifespan in either sex, though combined with rapamycin it does.
6. 2020
   Canagliflozin, males only
   The SGLT2 inhibitor raises median male lifespan 14% with no significant effect in females, despite similar glucose lowering in both sexes.
7. 2021
   Nicotinamide riboside fails
   The NAD+ precursor does not affect lifespan in either sex; 17-α-estradiol works in males even when started at 20 months.
8. 2024
   Astaxanthin and meclizine in males; fisetin fails
   Two new male-only extensions are reported, while the senolytic fisetin and four other agents show no effect at the doses used.
9. 2026
   Replications at new doses fail
   Astaxanthin, meclizine, and mitoglitazone show no benefit when retested at different doses or starting ages, alongside eight other null results.

Two decades of testing have produced a short list of validated life-extending drugs and a much longer list of failures. After rapamycin, the diabetes drug acarbose extended lifespan in both sexes, though far more strongly in males.³ A striking pattern emerged and persisted: 17-α-estradiol, a weakly feminising estrogen isomer, extends lifespan only in males,⁴ and the same male-only pattern holds for canagliflozin,⁵ nordihydroguaiaretic acid, and the 2024 additions astaxanthin and meclizine.⁶ Why most ITP-positive drugs work better in males is an open question in the field, with proposed explanations running from sex differences in drug metabolism to interactions with growth signalling. In 2026 the programme reported that astaxanthin, meclizine, and mitoglitazone failed to repeat their benefits when given at different doses or starting ages, a reminder that even within mice, dose and timing can make or unmake a longevity drug.⁷

§The null results

The negatives are arguably the programme's most valuable product. Resveratrol, the sirtuin activator whose promise David Sinclair's work had made famous, did not extend lifespan at the doses the ITP tested.⁸ Metformin alone did not significantly extend lifespan in either sex — a result that sits awkwardly beneath the TAME trial proposed by Nir Barzilai, which rests substantially on human observational data rather than mouse survival.⁹ Nicotinamide riboside, the most-studied of the NAD+ precursors, had no effect on lifespan in either sex.⁴ Fisetin, widely sold as a senolytic supplement, joined the null list in 2024 at the dose and schedule used⁶ — a result that matters for how Senolytics claims are graded, even though other senolytic agents and regimens remain untested there. Dozens of other agents, from green tea extract to curcumin to methylene blue, have likewise shown nothing.

The base rate

The large majority of compounds tested since 2004 have produced no significant lifespan extension in either sex. Several of the failures — resveratrol, metformin, nicotinamide riboside, fisetin — are among the best-selling ingredients of the longevity supplement market.

The nulls carry a caveat the programme itself states: a compound tested at one dose, one formulation, and one starting age has not been tested exhaustively. Metformin's advocates note that the ITP dose produced blood levels below those of treated diabetic humans. But the asymmetry is the point. Under identical conditions rapamycin worked and these agents did not, and a marketplace of Dietary supplements sold against aging has produced few molecules that clear the same bar.

§Reception and limits

Within biogerontology the ITP is routinely described as the field's most rigorous test of longevity interventions, and its three-site replication in heterogeneous animals is the standard against which single-laboratory claims are discounted. Regulators have no approval pathway for aging, so the programme functions as the de facto arbiter of which compounds deserve human study, informing trial design from TAME to the veterinary trials run by Loyal. Its data are public, its statistics are pre-specified, and its failures are published as prominently as its successes — practices that remain rare in the surrounding literature.

Its limits are equally plain. Mice in specific-pathogen-free housing, fed one diet and dying largely of cancer, are an imperfect model of human aging; effect sizes are percentages of a two-to-three-year lifespan, not steps toward record human lifespans; and lifespan itself is a blunt endpoint that can miss what a compound does to function. The programme measures survival superbly and mechanism not at all, which is why its results are read alongside the Hallmarks of aging framework and validated Aging biomarkers rather than in place of them.

The best mouse evidence is still mouse evidence

The ITP is what makes the rule "animal results are not human results" enforceable with data: it defines what a real mouse result looks like. No ITP finding, positive or negative, establishes anything about human aging, and no compound the programme has validated has been shown to slow aging in people.

§Outlook

The programme's current direction is toward combinations and refinements: rapamycin plus acarbose outperformed either drug alone in males, and the ACE inhibitor captopril produced a modest extension, in results reported in 2022.¹⁰ A companion NIA effort applies the same multi-site logic to Caenorhabditis worms, and proposals keep arriving faster than the three sites can test them. The unresolved question is the one the ITP was never designed to answer. Caloric restriction, rapamycin, and acarbose all extend mouse lifespan under its protocol, yet no human trial has demonstrated that any drug slows human aging, and none of the programme's male-only results has an accepted explanation. Whether the ITP's short list of validated compounds translates into human healthspan is the wager on which much of the longevity field — companies, trials, and off-label use alike — currently rides.

§See also

• Geroscience hypothesis
• Rapamycin
• Caloric restriction
• Hallmarks of aging
• Methuselah Foundation
• XPRIZE Healthspan
• Loyal

§References