A trial's primary endpoint is the one question it agreed to answer before it could see the answers. PEARL asked whether weekly rapamycin reduces visceral fat. The answer was no, at p = 0.942.
Why This Matters
Rapamycin has the strongest animal evidence of any longevity candidate. It extends median lifespan in genetically heterogeneous mice even when started late in life, and that finding has replicated across sites and doses (1, 2). People are taking it off label now, on that basis, while no human lifespan trial exists.
PEARL was the first randomized placebo-controlled trial of rapamycin run specifically for longevity endpoints in healthy adults (3). So how it turned out matters, and how it gets described matters nearly as much, because the description is what reaches the person deciding whether to ask a doctor for a prescription.
The description has drifted. In most coverage, including our own earlier explainer, PEARL is summarized as having found improvements in lean tissue mass and pain. That is a real result in the paper. It is not the primary endpoint, and the primary endpoint did not move.
What PEARL Actually Registered
Before enrollment, PEARL registered as NCT04488601 with one primary outcome measure: change in visceral fat as measured by dual-energy x-ray absorptiometry. Bone density, lean body mass, adverse events, blood counts and a long list of serum chemistry measures were all registered as secondary.
That ordering is not bookkeeping. Pre-registration exists so that a trial has to name its target before it sees where the data landed. Everything registered as secondary is a question the trial is permitted to ask, and every answer it gets is worth less, because there are many such questions and some of them come back positive by chance alone.
PEARL randomized adults to placebo, 5 mg compounded rapamycin weekly, or 10 mg compounded rapamycin weekly, double blind, for 48 weeks, run as a decentralized trial. 129 were enrolled, 114 completed: 39 placebo, 40 on 5 mg, 35 on 10 mg (3).
The Primary Result
Visceral adiposity did not change. Partial eta squared was 0.001 at p = 0.942 (3).
Two pieces of jargon there, and both are worth a sentence, because the rest of this article turns on them.
A p-value is the chance of seeing a gap at least this large IF there were genuinely no effect. At p = 0.942, a gap this size or bigger would turn up about 94 times in 100 in a world where rapamycin does nothing at all to visceral fat. That is not a trial that came close. It is a trial whose result is almost exactly what you would expect from noise.
Partial eta squared is the share of the variation in the outcome that tracks which arm a participant was in. At 0.001, essentially none of the differences in visceral fat across this study went with the treatment assignment.
The distinction that matters is between a null and a near miss. Underpowered trials usually produce a suggestive direction and a wide confidence interval, the kind of result that says "possibly something, we could not tell." This produced neither. It produced a flat line.
The honest reading is not "rapamycin does not reduce visceral fat." It is that this trial, at these doses, in these people, over 48 weeks, found no detectable difference on the question it set out to answer. What it could have detected is the open part: with 114 completers split three ways, only a fairly large effect would have shown up reliably, so a small real benefit could sit inside this null unseen.
Where The Positive Findings Came From
The results people quote are real, and they are all downstream of that null.
One more piece of notation before the numbers. A 95 percent confidence interval is the range of values compatible with this data. For a difference between two groups, the no-difference point is 0: if the interval includes 0, the data are compatible with there being no difference at all. Watch how close some of these intervals come to it.
Lean tissue mass improved in women on 10 mg at 48 weeks, against placebo (mean difference 6.194, 95 percent CI 0.877 to 11.511, p = 0.018) and against the 5 mg arm (mean difference 5.565, 95 percent CI 0.531 to 10.598, p = 0.026). Self-reported pain improved in women on 10 mg at both 24 and 48 weeks (48 week mean difference 8.071, 95 percent CI 3.044 to 13.098, p < 0.001). Self-reported general health improved in the 5 mg arm across all participants (mean difference 5.882, 95 percent CI 1.350 to 10.415, p = 0.007) (3).
Read that list again and notice the shape of it. The lean mass and pain findings are in women only. The general health finding is in the 5 mg arm only, which is the arm that did not produce the lean mass finding. Emotional well-being improved in the 5 mg arm and also in the placebo arm.
Women were 35.1 percent of the cohort (3). Split across three arms, the female 10 mg group is somewhere around a dozen people. This is what a subgroup analysis looks like: a result that is statistically significant, meaning unlikely to be noise on its own terms, sitting inside a slice of a trial that found no detectable difference overall. The standard reference on this is blunt about how often such findings fail to replicate and about the conditions under which they deserve to be believed (4).
The One Reason To Take The Sex Split Seriously
Here is where PEARL gets more interesting than a routine subgroup story.
If you had asked, before PEARL reported, which sex would respond more to rapamycin, the animal literature had already answered. The NIA Interventions Testing Program found that rapamycin's lifespan effect in mice is both dose dependent and sex dependent, with females showing the larger response, and published that in 2014 (2).
A subgroup finding that matches a prediction made eleven years earlier in a different species is not the same object as a subgroup finding fished out of a results table. It is still one small trial. But it is a coherent one, and it tells you what the next trial should be powered to detect.
That is the correct use of PEARL: not as evidence that rapamycin builds lean mass in women, but as a reason to design a trial that could establish whether it does.
The Dose Problem
Buried in the limitations is a fact that complicates every number above. The authors report that their compounded rapamycin achieved roughly one third of the blood concentration of commercial formulations, which makes the effective doses about 66 percent lower than the 5 mg and 10 mg labels suggest (3).
So the arms may not be the arms. A trial nominally testing 5 and 10 mg weekly may have delivered something closer to 1.7 and 3.3 mg weekly in commercial terms. If rapamycin's effect on visceral fat is dose dependent in humans the way its effect on mouse lifespan is dose dependent (2), a null at an unintentionally low dose is a much weaker piece of evidence than a null at the intended one.
This does not rescue the primary endpoint. It does mean the null should not be generalized past the exposure that was actually delivered, and that exposure is not the one in the trial's own arm labels.
What Human Rapamycin Data Existed Before This
PEARL was first for longevity endpoints, not first in humans. Two earlier trials from a different research line tested mTOR inhibition on immune function in older adults, and both found something. The 2014 trial found improved vaccine response in the elderly (5). The 2018 follow-up found reduced respiratory tract infections (6).
Those are real, positive, adequately reported human results, and they are about immune function rather than body composition or lifespan. They are the reason the mTOR hypothesis in humans is taken seriously at all. They are also a useful contrast: those trials hit endpoints they had set out to hit.
What VITAL-H Will And Will Not Settle
The trial the field has been waiting for is now funded. VITAL-H, led by the Sam and Ann Barshop Institute at UT Health San Antonio with up to 38 million dollars from ARPA-H, will enroll 726 generally healthy adults aged 60 to 65 into four arms: rapamycin, dapagliflozin, semaglutide and placebo (7, 8).
Three things about its design are worth noting now, because they determine what the result will mean.
It is a head-to-head. Running three repurposed drugs against one placebo in a single protocol removes the usual excuse that two trials disagreed because they enrolled different people. Dapagliflozin is an SGLT2 inhibitor, which puts that drug class and the GLP-1 class in the same room as rapamycin under identical conditions for the first time.
Its endpoints are function, not biomarkers. Physical performance, cognition, vitality and sensory function (7). Not epigenetic clocks, not visceral fat, not a composite score. That choice makes the trial harder to pass and the result harder to spin.
It will not measure lifespan. None of these trials do. VITAL-H asks whether these drugs keep people working better in their sixties and seventies, which is a healthspan question, and the distinction between that and lifespan is the whole reason the endpoints were chosen this way.
Results are expected between 2028 and 2031 (7, 8).
A Correction To Our Earlier Coverage
Our May 2026 article on rapamycin and mTOR inhibition stated that PEARL's primary finding was improvement in lean tissue mass and pain scores, and framed the epigenetic clock null as the notable negative result. Both of those are wrong.
PEARL's registered primary outcome was visceral fat by DXA (NCT04488601), and it returned p = 0.942. Lean body mass was registered as a secondary outcome, and the result reported for it is confined to women in the 10 mg arm. That article is being amended to say so.
We are noting this in a separate piece rather than quietly editing the old one because the error is the exact error this article is about. Describing a secondary subgroup result as a trial's primary finding is how a null trial becomes a positive one in the retelling, and we did it in our own pages.
What This Means For You
If you are taking rapamycin off label, PEARL is not the reassurance it is often presented as, and it is not a reason to stop either. It is a small trial that missed its target, found something in a subgroup that the mouse data had predicted, and may have delivered a third of the dose it intended.
If you are deciding whether to start, the honest position is that the human evidence for rapamycin as a longevity intervention currently consists of one trial with a null primary endpoint plus two positive immune-function trials at a different question. VITAL-H is the first study designed well enough to change that, and it will not report for at least two years.
The transferable skill here is smaller and more useful than any of that. When a trial result reaches you, find the primary endpoint before you read anything else. It is in the registration, it is public, and it takes about ninety seconds to look up. If the headline finding and the primary endpoint are different things, you are reading about a subgroup, and you should price it accordingly.
Frequently Asked Questions
Did the PEARL trial fail?
It missed its primary endpoint, which is not the same as failing. PEARL was designed to find out whether weekly rapamycin reduces visceral fat over 48 weeks, and it answered that question with a clear no (p = 0.942). A trial that answers its question in the negative has done its job. What it did not do is establish any of the benefits it is usually cited for.
Does this mean rapamycin does not work?
No. It means one small trial, at doses that may have delivered about a third of the intended blood concentration, found no detectable difference on one body-composition measure over one year. Rapamycin's mouse lifespan data remain the strongest in the field (1, 2), and two human trials of mTOR inhibition found real improvements in immune function (5, 6). None of that is overturned by a null on visceral fat.
Why does it matter which endpoint was primary?
Because the primary endpoint is chosen before anyone can see the results, and every other question is asked afterwards. A trial that tests one pre-committed hypothesis and a few dozen secondary measures will usually turn up something statistically significant somewhere by chance. Treating a secondary subgroup result as the headline converts a null trial into a positive one without any new evidence (4).
Should the sex difference in PEARL be believed?
More than a typical subgroup finding, less than a primary result. It is based on roughly a dozen women per arm, which is very thin. But the direction was predicted in 2014 by the mouse data, which found rapamycin's effect to be sex dependent with females responding more (2). A subgroup result that matches a prior prediction is worth designing the next trial around. It is not worth acting on yet.
When will there be a better answer?
VITAL-H reports between 2028 and 2031 (7, 8). It enrolls 726 adults aged 60 to 65, runs rapamycin against dapagliflozin, semaglutide and placebo in one protocol, and measures physical performance, cognition, vitality and sensory function rather than biomarkers. Until it reports, the human evidence base for rapamycin as a longevity drug does not change.
Sources
- Harrison DE, Strong R, Sharp ZD, et al. Rapamycin fed late in life extends lifespan in genetically heterogeneous mice. Nature. 2009;460(7253):392-395. PMID: 19587680. DOI: 10.1038/nature08221
- Miller RA, Harrison DE, Astle CM, et al. Rapamycin-mediated lifespan increase in mice is dose and sex dependent and metabolically distinct from dietary restriction. Aging Cell. 2014;13(3):468-477. PMID: 24341993. DOI: 10.1111/acel.12194
- Moel M, Harinath G, Lee V, et al. Influence of rapamycin on safety and healthspan metrics after one year: PEARL trial results. Aging (Albany NY). 2025;17(4):908-936. PMID: 40188830. DOI: 10.18632/aging.206235
- Wang R, Lagakos SW, Ware JH, Hunter DJ, Drazen JM. Statistics in medicine: reporting of subgroup analyses in clinical trials. N Engl J Med. 2007;357(21):2189-2194. PMID: 18032770. DOI: 10.1056/NEJMsr077003
- Mannick JB, Del Giudice G, Lattanzi M, et al. mTOR inhibition improves immune function in the elderly. Sci Transl Med. 2014;6(268):268ra179. PMID: 25540326. DOI: 10.1126/scitranslmed.3009892
- Mannick JB, Morris M, Hockey HP, et al. TORC1 inhibition enhances immune function and reduces infections in the elderly. Sci Transl Med. 2018;10(449):eaaq1564. PMID: 29997249. DOI: 10.1126/scitranslmed.aaq1564
- UT Health San Antonio. UT San Antonio to lead $38 million national trial testing drugs to extend healthspan. News release, 2026. https://news.uthscsa.edu/ut-san-antonio-to-lead-38-million-national-trial-testing-drugs-to-extend-healthspan/
- Kulkarni AS, Aleksic S, Berger DM, et al. Geroscience-guided repurposing of FDA-approved drugs to target aging: a proposed process and prioritization. Aging Cell. 2022;21(4):e13596. PMID: 35343051. DOI: 10.1111/acel.13596
- PEARL trial registration. Participatory Evaluation of Aging with Rapamycin for Longevity. ClinicalTrials.gov identifier NCT04488601. https://clinicaltrials.gov/study/NCT04488601
Funding Transparency
PEARL (source 3) was funded by crowdfunding through Lifespan.io, with major donations from Vitalik Buterin, Micah Zoltu, Brad Armstrong and anonymous contributors, plus administrative support from AgelessRx. The paper discloses that seven of its authors are employees and shareholders of AgelessRx, a telehealth company that sells rapamycin prescriptions.
Disclosure is the right behaviour and the authors did it. The reason to weigh it anyway is structural rather than personal: when the people reporting a trial have a commercial interest in the intervention, the risk is not usually fabricated data. It is which result ends up in the abstract and which ends up in the limitations. In this case the null primary endpoint and the one-third bioavailability problem are both in the paper, and both are absent from most secondhand coverage, including our own.
The NIA Interventions Testing Program (sources 1, 2) is federally funded through the National Institute on Aging, with no commercial sponsor for the rapamycin arms.
Sources 5 and 6 were funded by Novartis, which was developing the mTOR inhibitors tested. Both papers disclose this. The same weighting applies.
VITAL-H (source 7) is funded by ARPA-H, a federal agency, at up to 38 million dollars. It has no commercial sponsor, and none of the three drugs it tests are on patent in a way that gives a manufacturer a stake in the outcome.
This publication has no financial relationship with AgelessRx, Novartis, any rapamycin supplier, or any party to VITAL-H, and sells no supplements or prescriptions.
Related Reading
- Rapamycin and mTOR Inhibition, the general explainer on the drug and the pathway, now corrected on the PEARL endpoint
- SGLT2 Inhibitors and Longevity, on dapagliflozin's drug class, one of VITAL-H's other three arms
- GLP-1 Drugs and Longevity Science, on semaglutide, the third VITAL-H arm
- Healthspan vs Lifespan, on why VITAL-H measures function rather than survival
- Autophagy and Mitophagy, on the cellular process mTOR inhibition is thought to act through
The primary endpoint is public, it is timestamped before the data existed, and it takes ninety seconds to check. Do that first.
Written with the help of AI tools, shaped and verified by humans. Every number in this article was checked against the primary publication and the trial registration rather than a summary of either.
This article is for information only and is not medical advice.