The Peptides Aimed at Ageing Itself

LONGEVITY LATEST ISSUE 26 · 2 SEPTEMBER 2026

LONGEVITY LATEST · DEEP DIVE

The Peptides Aimed at Ageing Itself

Five compounds, five different ways the evidence gives way — and three rules you can take to any peptide claim you meet next.

By Christian Thomsen · Companion to Issue 26 · 2 September 2026 · ~7-minute read

I expected the ageing peptides to be the exotic ones — epitalon, humanin, MOTS-c, the compounds whose entire pitch is the ageing process itself. That is not where the evidence turned out to be.

It comes instead from a diabetes drug, in a trial not designed to look for it, in a population nobody would have chosen. The five entries below run downhill from there, and the descent is the point: each gives way for a different reason, and the reasons are worth more than the grades.

What “targeting ageing” actually means

A blood pressure drug is judged against strokes. A drug aimed at ageing has no equivalent endpoint you can reach inside a trial — you would need to follow healthy people for forty years and outlive your own funding. So the field uses a proxy, and for a decade that proxy has been the epigenetic clock.

The clocks read DNA methylation: tags that appear and disappear across the genome in patterns tracking with age. First-generation clocks (Horvath, Hannum) predict chronological age, which makes them accurate and close to useless — a perfect one would only tell you your birthday. Second-generation clocks (PhenoAge, GrimAge) were trained against mortality instead, so they predict how long you have rather than how long you have had. DunedinPACE measures the rate.

The simplification worth flagging: a clock is a statistical model, not a measurement of damage. Moving one is a bet that the correlation runs causally in the direction you want, and that bet is not settled.

Why peptides in particular? Because they are the body's own signalling vocabulary — hormones, growth factors, the messengers mitochondria send to the nucleus. If ageing is partly a failure of signalling, a class of molecules that carries signals is the obvious place to intervene. That is the theory. Here is the file.

1. Semaglutide — a proxy moved, in a real trial

Nature Communications, May 2026: a post-hoc analysis of a 32-week randomised, double-blind, placebo-controlled phase 2b trial in HIV-associated lipohypertrophy — 45 on semaglutide, 39 on placebo, mean age 49. Across seventeen methylation clocks, PhenoAge fell 4.9 years against placebo, PCGrimAge 3.1, DunedinPACE 0.09 units — a pace of ageing roughly 9% slower. The direction held across nearly every measure, which is the hard part to explain away.

The caveats stack up fast. Post hoc, not pre-specified. One clinical population. Thirty-two weeks. And a good share of the effect may simply be weight loss and reduced inflammation — good for you, certainly, but not a distinct mechanism of ageing.

It is still a randomised, placebo-controlled trial, and nothing below has that. This is the high-water mark, and the high-water mark is a surrogate endpoint found by accident.

Grade: B for the clock effect. A for the drug — SELECT randomised 17,604 people without diabetes and cut major cardiovascular events by 20%, still the only hard outcome any peptide on either list has produced.

2. Growth hormone and the thymus — the design problem

Fahy's 2019 paper in Aging Cell is the most cited result in this field, resting on the weakest design of the five. Nine men, 51 to 65, on recombinant human growth hormone plus DHEA plus metformin for a year. Mean epigenetic age came out about 1.5 years below baseline at twelve months, 2.5 below where untreated ageing would have put them, with thymic fat replaced by functioning tissue on MRI.

There are several things wrong with that, and only one of them decides the question. Not the sample size, and not the three drugs given together. The missing control arm. Clocks drift, assays vary batch to batch, and without a comparison group there is no way to know how much of 2.5 years belonged to the treatment and how much to the measurement.

The extension trial, TRIIM-X, was registered in 2020. Six years on, still no controlled published result. And last week's tension sits squarely in the way: the active agent raises IGF-1, while the Ecuadorian Laron cohort is twenty-two years of human evidence that lower IGF-1 signalling travels with less cancer and less diabetes.

A striking result from a design that cannot support it is not evidence. It is a reason to run the trial.

Grade: D. Thymic involution is a real feature of immune ageing and the hypothesis deserves testing. Six years is a long time to leave a nine-person uncontrolled pilot as the best answer available.

3. Elamipretide — the indication problem

Now a compound with none of those weaknesses, which fails a different test entirely.

SS-31 is a four-amino-acid peptide that binds cardiolipin, the lipid holding the respiratory complexes in place inside the mitochondrion. Cardiolipin degrades with age; the theory is that you stabilise it. In September 2025 the FDA approved it for Barth syndrome, a rare genetic disorder of cardiolipin remodelling, on a 28-week randomised trial of twelve patients plus an extension in which eight of ten reached 168 weeks.

So: randomised, controlled, licensed, and acting directly on a core mechanism of ageing. Then the same compound missed its primary endpoint in primary mitochondrial myopathy, and again in dry age-related macular degeneration. Two well-funded shots at broader indications, two misses.

A peptide correcting a specific inherited defect in cardiolipin is not therefore a treatment for the diffuse mitochondrial decline of ordinary ageing. Different problems sharing a vocabulary — and the vocabulary is what sells.

Grade: A for Barth syndrome. D for ageing. The same molecule, and the gap between those two grades is the indication, not the evidence quality.

4. Klotho — the dose problem

Named after the Greek Fate who spins the thread. Mice lacking klotho age rapidly, mice overexpressing it live longer, and humans carrying the KL-VS variant have higher circulating klotho and better cognitive ageing.

Castner, Dubal and colleagues, in Nature Aging in 2023, gave a single injection to aged rhesus macaques averaging 22 years — roughly a human 65 — and found improved memory on a maze task. The detail that makes this entry different from the ones above it: the low dose worked and the high dose did not.

That inversion breaks the assumption every protocol quietly relies on: that more of a beneficial molecule is more benefit. Nobody knows where the window sits in a human, and the first klotho gene therapy studies in people are now registered — some outside conventional trial oversight.

Grade: D in humans, on an unresolved dose-response curve rather than an absence of biology. The one here I would most like to be wrong about.

5. Humanin and MOTS-c — the best story, the emptiest file

Peptides encoded inside the mitochondrial genome itself, which is about as elegant as ageing biology gets. Humanin overexpression extends life in C. elegans through daf-16/FOXO — the pathway caloric restriction runs on. Levels fall with age in mice, monkeys and humans, and conspicuously do not fall in the naked mole rat. Offspring of centenarians, eighteen of them, carry higher humanin than nineteen age-matched controls.

That last sentence is the entire consumer case. It is an association in thirty-seven people.

And from last week's file: FDA reviewers, going through the July compounding nominations, identified no human clinical study for MOTS-c at all. Not a weak one. None. That is the bottom of the descent — randomised trial, missing control arm, wrong indication, unknown dose, and now no human file to criticise.

Grade: D for both — not on the biology, which is the most elegant here, but because nobody has given either to a person in a published controlled trial.

Three rules the file gives you

The grades above will date. These won't, and they are the reason the article is worth the time rather than the table.

1. A proxy is not an outcome. Every reversal figure here is a methylation model, not a life extended. DO-HEALTH found vitamin D, omega-3 and exercise all move the clocks a little — useful, and not an outcome. That distance is where most of this market lives.

2. A mechanism is not an efficacy. Elamipretide is the cleanest case: real drug, real target, real approval, two failures the moment it left the indication it was built for. “Acts on a pathway involved in ageing” is the commonest sentence in peptide marketing and the least informative.

3. Endogenous is not safe. “Your body already makes it” is the commonest safety argument offered for the mitochondrial peptides, and it is not a safety argument at all. Your body already makes insulin, cortisol and aldosterone. Give any of the three from outside in the wrong quantity and you will find out quickly why the dose was regulated.

The frontier

Four things worth watching. Whether the semaglutide clock finding replicates outside an HIV cohort, pre-specified. Whether TRIIM-X ever produces a controlled readout. Whether the first-in-human klotho work resolves the dose inversion before somebody sells it. And the next FDA compounding committee, due before the end of February 2027, taking up GHK-Cu, LL-37, dihexa acetate, melanotan II and pegylated mechano growth factor — a list that still contains dihexa, whose evidence base was retracted in April 2025 after a misconduct finding.

What this means for you

Run the three rules across the file and the practical answer writes itself. Rule one takes the semaglutide result from a reason down to a bonus. Rule two removes elamipretide unless you have Barth syndrome. Rule three is all you have left for the mitochondrial peptides, because there is nothing else in their file to use.

What survives is short and dull:

1. If you have a metabolic condition a GLP-1 is licensed to treat, that conversation belongs with a doctor, and the Issue 23 cardiovascular numbers belong in the room. The clock data is a bonus, not a reason.

2. If you don't, the interventions with randomised human evidence for the same mechanisms are unglamorous and already in your week: resistance and isometric work, sleep, enough protein.

3. Nothing at the D end of this list belongs in your body on what is currently published. That may change. It has not yet.

The peptide with the strongest claim to slowing human ageing is the one never marketed as an ageing drug. That tells you something about where to look — and rather more about where not to.

Sources and further reading

Where a claim below rests on the FDA compounding briefing documents discussed in Issue 25, the citation is carried there rather than repeated here.

1. Corley MJ, Dwaraka VB, Pang APS, Labbato D, Smith R, Eckard AR, McComsey GA. Semaglutide slows epigenetic aging in a randomized trial of HIV-associated lipohypertrophy. Nat Commun. 2026 — post-hoc exploratory analysis of NCT04019197; n=84 (45 semaglutide, 39 placebo); PhenoAge −4.9 years, PCGrimAge −3.1, GrimAge V2 −2.3, DunedinPACE −0.09 units (≈9% slower pace). Epigenetic ageing was not a pre-specified endpoint.

2. Lincoff AM, Brown-Frandsen K, Colhoun HM, et al. Semaglutide and cardiovascular outcomes in obesity without diabetes (SELECT). N Engl J Med. 2023;389:2221–2232 — n=17,604; MACE HR 0.80.

3. Fahy GM, Brooke RT, Watson JP, Good Z, Vasanawala SS, Maecker H, Leipold MD, Lin DTS, Kobor MS, Horvath S. Reversal of epigenetic aging and immunosenescent trends in humans. Aging Cell. 2019;18(6):e13028. PMID: 31496122 — n=9, uncontrolled; mean epigenetic age ≈1.5 years below baseline at 12 months, −2.5 years relative to no treatment.

4. Thymus Regeneration, Immunorestoration, and Insulin Mitigation Extension Trial (TRIIM-X), NCT04375657, Intervene Immune, registered 2020 — no controlled results published as at September 2026.

5. Guevara-Aguirre J, Balasubramanian P, Guevara-Aguirre M, et al. Growth hormone receptor deficiency is associated with a major reduction in pro-aging signaling, cancer, and diabetes in humans. Sci Transl Med. 2011;3(70):70ra13. PMID: 21325617 — 22-year follow-up of the Ecuadorian Laron cohort.

6. Szeto HH. First-in-class cardiolipin-protective compound as a therapeutic agent to restore mitochondrial bioenergetics. Br J Pharmacol. 2014;171(8):2029–2050. PMID: 24117165 — mechanism of SS-31/elamipretide.

7. US Food and Drug Administration. Approval of elamipretide for Barth syndrome, September 2025 — based on the 28-week randomised TAZPOWER trial (n=12) and its long-term open-label extension, reported in Genetics in Medicine (2023); 8 of 10 initial participants reached the 168-week visit.

8. Elamipretide missed its primary endpoint in primary mitochondrial myopathy (MMPOWER-3) and in dry age-related macular degeneration (ReCLAIM-2); development in both indications was discontinued.

9. Castner SA, Gupta S, Wang D, Moreno AJ, Park C, Chen C, Poon Y, Groen A, Greenberg K, David N, Boone T, Baxter MG, Williams GV, Dubal DB. Longevity factor klotho enhances cognition in aged nonhuman primates. Nat Aging. 2023;3(8):931–937. PMID: 37400721 — single administration in aged rhesus macaques; memory enhanced at low dose but not high dose.

10. Arking DE, Krebsova A, Macek M Sr, et al. Association of human aging with a functional variant of klotho. Proc Natl Acad Sci USA. 2002;99(2):856–861. PMID: 11792841 — the KL-VS variant.

11. Safety and Efficacy of Injectable Klotho Plasmid Gene Therapy in Humans, NCT07216781 — first-in-human klotho gene therapy in healthy volunteers.

12. Yen K, Mehta HH, Kim S-J, et al. The mitochondrial derived peptide humanin is a regulator of lifespan and healthspan. Aging (Albany NY). 2020;12(11):11185–11199. PMID: 32575074 — lifespan extension in C. elegans via daf-16/FOXO; humanin stable across two decades in the naked mole rat; centenarian offspring (n=18) versus age-matched controls (n=19).

13. Kim S-J, Guerrero N, Wassef G, et al. Mitochondria-derived peptides in aging and healthspan. J Clin Invest. 2022 — review of the humanin and MOTS-c literature.

14. Bischoff-Ferrari HA, Gängler S, Wieczorek M, Belsky DW, et al. Individual and additive effects of vitamin D, omega-3 and exercise on DNA methylation clocks of biological aging in older adults from the DO-HEALTH trial. Nat Aging. 2025;5:376–385.

15. Retraction notices, J Pharmacol Exp Ther, April 2025 — two dihexa mechanism papers retracted following a Washington State University investigation; a third carries an expression of concern. See Issue 25 for the full note.

16. FDA Pharmacy Compounding Advisory Committee, meeting of 23–24 July 2026; briefing documents — no human clinical study identified for MOTS-c. Next meeting scheduled before the end of February 2027 (LL-37, GHK-Cu, dihexa acetate, melanotan II, pegylated mechano growth factor).