Longevity Compounds, Scored by Human Evidence: A Brutal Honest Scorecard
A deflationary scorecard of popular longevity compounds ranked by real human RCT evidence: rapamycin, metformin, NAD precursors, spermidine, urolithin A and more.
Ranked by human evidence, almost every popular longevity compound sits at the biomarker-only tier or worse. Only a few (urolithin A, NMN) have any functional-endpoint RCT, and none has a replicated, healthspan-changing human trial. Raising a biomarker is not the same as extending healthspan.
Every longevity compound arrives with the same shape of promise: it extends lifespan in a model organism, it moves a fashionable biomarker, and therefore it should slow human aging. The gap between those first two facts and the third is where most of the field lives — and it is enormous. This is a scorecard graded not on mechanism or mouse data but on what has actually been measured in people, in randomized trials, on endpoints that matter. Everything below describes laboratory and clinical-literature findings, not use in any person.
How do you honestly score a longevity compound?
Grade on the endpoint, not the enthusiasm. A defensible human-evidence hierarchy for geroprotective claims runs in four tiers: (1) a randomized controlled trial with a functional or clinical endpoint — strength, walking distance, insulin sensitivity, a validated cognitive score; (2) a biomarker-only RCT, where a molecule moves NAD or an autophagy marker but nothing you can feel; (3) observational or epidemiological association; and (4) animal, in-vitro, or mechanistic data with no human trial at all. The geroscience literature itself frames this hierarchy explicitly and concedes that human functional and lifespan endpoints for repurposed longevity drugs remain largely unproven.14
Apply that ruler and the popular list collapses toward the middle and bottom. Almost every compound that trends in longevity circles sits at tier 2 or worse in humans. The scorecard below is deliberately deflationary, because the honest headline is uncomfortable: raising a biomarker is not the same as extending healthspan.
1 Of the popular longevity compounds, essentially one class — urolithin A and, marginally, NMN — has any positive functional-endpoint human RCT at all.
Rapamycin: best case on paper, thin on human endpoints
Rapamycin (sirolimus) has the strongest theoretical case in the entire field. It inhibits mTOR, the nutrient-sensing pathway most consistently tied to lifespan across yeast, worms, flies and mice, and it extends lifespan in mice even when started late. The human flagship is the PEARL trial, a 48-week decentralized RCT of intermittent low-dose rapamycin in healthy adults.1 PEARL is a real, prospectively registered trial — but read what it measured. Its endpoints were safety, tolerability, and self-reported or wearable-derived healthspan metrics, not a hard clinical outcome and certainly not lifespan. It demonstrates that intermittent low-dose rapamycin was tolerated over roughly a year in this population. It does not demonstrate that rapamycin extends human healthspan. On the scorecard, rapamycin earns the best mechanistic and animal grade of any compound here, but its human tier remains biomarker-and-self-report.
Metformin: a rationale trial that hasn’t reported, and a caution flag
Metformin’s longevity reputation rests almost entirely on rationale plus observational data. The foundational case is the TAME framework — Targeting Aging with Metformin — which argues from repurposing logic and epidemiology that a cheap, well-characterized drug could serve as a proof-of-concept geroprotector.9 The critical detail is that TAME’s definitive RCT has still not reported. So the geroprotective claim for metformin is, in human-evidence terms, a well-argued hypothesis rather than a finding.
There is also a countervailing human signal. In the MASTERS substudy, a randomized trial in older adults, metformin blunted the skeletal-muscle transcriptome adaptation to resistance training.10 In plainer terms: in that trial, adding metformin appeared to antagonize the muscle benefit of exercise. That is a genuine complication for the “metformin for healthspan” story, because exercise is the one intervention with unambiguous functional benefit in aging humans.
The chemistry moves reproducibly; the function stays flat. That single sentence describes the entire human record of NAD precursors.
NAD precursors: the textbook biomarker-up, function-flat case
Nicotinamide riboside (NR) and nicotinamide mononucleotide (NMN) are the clearest illustration of why a biomarker shift is not evidence of benefit. Both raise blood NAD in humans — that part is reproducible. What repeatedly fails to appear is any functional payoff. In obese men, NR was safe and well tolerated but did not improve insulin sensitivity, mitochondrial function, or other metabolic endpoints.4 A follow-up muscle-biopsy analysis from the same program found NR raised NAD-metabolome markers yet left skeletal-muscle mitochondrial respiration, content, and morphology unchanged.5 A separate physiologic RCT in overweight and older adults robustly increased whole-blood NAD but produced no significant improvement in insulin sensitivity, mitochondrial function, or muscle.7 An exploratory NR trial in older adults with subjective cognitive decline and mild cognitive impairment established no clear cognitive benefit either.8
The best-case NAD data comes from NMN. In a dose-ranging RCT in healthy middle-aged adults, NMN raised blood NAD dose-dependently and improved six-minute-walk distance along with a subjective aging questionnaire.6 Six-minute walk is a genuine functional endpoint, which is why NMN edges above NR on the scorecard. But the effect sizes were small and one of the two positive endpoints was a self-report instrument. It is a real signal, not a settled result.
| Compound | Best human trial type | Functional result | Evidence tier |
|---|---|---|---|
| Rapamycin | Safety/tolerability RCT (PEARL) | Self-report/wearable only | 2 (best animal case) |
| Urolithin A | Functional-endpoint RCT | Modest strength gain | 1 (small, industry-funded) |
| NMN | Functional-endpoint RCT | Modest walk gain (partly soft) | 1–2 |
| NR | Biomarker RCT(s) | Null on function | 2 |
| Metformin | Rationale + observational | Definitive RCT pending; caution signal | 3 (mixed) |
| Spermidine | Functional-endpoint RCT | Null (SmartAge) | Null |
| Resveratrol | Pilot RCT + systematic review | Largely null/inconsistent | Null/weak |
| Fisetin / D+Q | Early pilot/mechanistic | Not yet tested on healthspan | 4 (early) |
| Epitalon / epithalamin | Single-lineage, no Western RCT | Not independently established | 4 (unreplicated) |
Tiers reflect rigorous human evidence only. All results are from clinical or preclinical literature; none constitutes a use recommendation, and several positive trials were industry-sponsored with modest effect sizes.
Spermidine and resveratrol: where the null results live
Spermidine had a clean chance to prove itself and did not. The SmartAge RCT ran 12 months in 100 older adults with subjective cognitive decline and found no significant benefit versus placebo on its primary memory endpoint or on biomarkers.2 That is a well-designed, adequately-long trial reporting a null — the most honest anchor available for spermidine’s human case, and one that autophagy-marker enthusiasm tends to skip past.
Resveratrol has arguably the weakest track record among the famous compounds. A pilot RCT combining resveratrol with exercise in late life found no enhancement of exercise-induced functional gains, and on some measures a blunting effect.11 A comprehensive systematic review of resveratrol RCTs in older adults concluded that effects on clinically meaningful aging and functional endpoints were inconsistent and largely null.12 The molecule that launched a thousand longevity headlines has, in humans, mostly produced disappointment.
Urolithin A: the rare positive functional endpoint
Urolithin A is the exception that proves how empty the rest of the shelf is. A four-month RCT in middle-aged adults reported improved muscle strength (hamstring) alongside some exercise-performance and mitophagy biomarker changes versus placebo.3 That is a genuine functional-endpoint win, which lands urolithin A at tier 1 on the scorecard. The honest caveats: the effect sizes were modest, the trial was industry-sponsored, and the strongest results sat on secondary endpoints. It is the best of the popular longevity compounds by human evidence — and even it is a modest, single-trial story.
Epitalon, fisetin, and the senolytics: early or unreplicated
Epitalon (the AEDG tetrapeptide, Ala-Glu-Asp-Gly) and its parent preparation epithalamin carry striking human claims — reduced mortality, restored circadian rhythm — that need to be handled carefully. Those data derive almost entirely from a single Russian research lineage, reviewed by the originating group itself, with no independent Western RCT replication.13 Single-lineage, decades-old, methodologically opaque, and never reproduced by an outside team is not the profile of established human evidence. For a rigorous scorecard, epitalon/epithalamin scores at the animal-or-none tier, and the mortality figures should not be cited as if settled.
Fisetin and the dasatinib-plus-quercetin (D+Q) senolytic combination are earlier still. Human work so far is dominated by small pilot and mechanistic trials probing whether senescent cells can be cleared, not RCTs with healthspan or lifespan endpoints. That is legitimate early science, but on this ruler it is tier 4 — promising mechanism, no functional human result yet.
An honest read of the evidence
Set the trials side by side and a consistent pattern emerges, and it is not a flattering one. NAD precursors are the textbook case of a biomarker moving while function does not: NAD rises reproducibly across NR and NMN trials, yet insulin sensitivity, mitochondrial respiration, and muscle outcomes stay flat.457 The lone positive functional signal — NMN’s six-minute-walk gain — is small and partly subjective.6 The two clearest positive functional-endpoint trials, urolithin A and NMN, are both industry-sponsored with modest effect sizes on secondary endpoints; those deserve cautious interest, not a healthspan verdict.36
The best-branded compounds do not escape. PEARL is a decentralized trial with self-reported and wearable outcomes and no hard clinical or lifespan endpoint; it shows rapamycin was tolerated, not that it extends healthspan.1 Metformin has no completed longevity RCT — TAME remains pending — and carries a human signal that it can blunt exercise adaptation, so its geroprotective claim is unsettled and may even conflict with the one intervention that unambiguously works.910 Epitalon’s human data are single-lineage and unreplicated in the West, which is a replication gap, not a body of evidence.13 Spermidine has a definitive null in an adequately-powered RCT, resveratrol has broadly null and inconsistent reviews, and fisetin and the senolytics have only early pilot work.21112 None of these compounds has a robust, replicated, functional-endpoint human RCT. That is the field, told straight.
All materials supplied by Condor Research are Research Use Only (RUO). Everything above describes findings from clinical and preclinical literature and is intended for laboratory and literature purposes only. Nothing here is a dosing protocol, clinical guidance, benefit claim, or safety assessment for any person, animal, or other organism. The compounds discussed here are described strictly as research-use-only reference materials; several are not Condor Research catalogue products and are covered for scientific reference only.
Condor Research · Sportello scientifico
Atrio Sciences s.r.o., IČO 57 669 651, Nitra (SK) · info@condorresearch.com
- The honest evidence hierarchy runs: functional/clinical-endpoint RCT > biomarker-only RCT > observational data > animal/mechanistic/none. Most named longevity compounds sit at tier 2 or worse in humans.
- Rapamycin has the best animal and mechanistic case, and the PEARL trial (2025) is the flagship human RCT — but it measured safety and self-reported healthspan, not a hard clinical or lifespan endpoint.
- NAD precursors (NR, NMN) reliably raise blood NAD yet repeatedly fail to move insulin sensitivity or mitochondrial function; the one positive functional signal (NMN 6-minute walk) is small and partly subjective.
- Spermidine's SmartAge RCT (12 months, n=100) was null on its primary memory endpoint — the clearest deflation of spermidine hype in humans.
- Urolithin A is one of the very few longevity compounds with a positive functional-endpoint human RCT (muscle strength over 4 months), though the effect was modest and the trial industry-sponsored.
- Metformin has no completed longevity RCT (TAME still pending), and one human trial showed it can blunt muscle adaptation to resistance training in older adults.
- Epitalon/epithalamin human data trace to a single Russian research lineage with no independent Western RCT; the mortality claims are not established evidence.
Which longevity compound has the strongest human evidence?
None qualifies as strong by clinical-trial standards. Urolithin A and NMN have small positive functional-endpoint RCTs, and rapamycin has the best mechanistic and animal case backed by a safety-focused human trial. No popular compound has a replicated RCT with a hard healthspan or lifespan endpoint.
Do NAD boosters like NMN and NR actually work?
They reliably raise blood NAD, but multiple RCTs show no improvement in insulin sensitivity or muscle mitochondrial function. One NMN trial reported a modest walking-distance gain alongside a subjective questionnaire. In humans, the chemistry moves while the function largely does not.
Is rapamycin proven to extend human lifespan?
No. Rapamycin extends lifespan in mice and has a coherent mTOR-inhibition mechanism, but the human PEARL trial measured safety, tolerability, and self-reported metrics over roughly a year — not lifespan or a hard clinical endpoint. It shows tolerability, not healthspan extension.
Why is spermidine on the low end of the scorecard?
Its best human trial, the 12-month SmartAge RCT in 100 older adults with subjective cognitive decline, found no significant benefit versus placebo on its primary memory endpoint or biomarkers. A well-designed, adequately-long null result carries real weight against the mechanistic hype.
What about epitalon and epithalamin?
The human mortality-reduction claims come almost entirely from one Russian research group and have never been replicated in an independent Western RCT. For rigorous human evidence, epitalon and epithalamin score at the animal-or-none tier, and those figures should not be presented as established.
Does a compound raising a biomarker mean it works?
No, and that is the central lesson here. NAD levels and autophagy markers can be moved in humans without any measured change in strength, metabolism, or cognition. A biomarker shift is a hypothesis to be tested on a functional endpoint, not proof of healthspan benefit.