Urolithin A, Spermidine, and NAD+: Sorting the Real Evidence From the Marketing
Three supplements share a marketing shelf and almost nothing else. Here's what the human trial evidence for urolithin A, spermidine, and NAD+ precursors actually supports.
Stand in front of the longevity shelf at any half-decent supplement store and you'll find the same three names sitting inches apart: spermidine, urolithin A, and some flavor of NAD+ booster. The labels use almost identical language — cellular renewal, mitochondrial support, healthy aging — and nothing on any of them tells you that these three compounds do genuinely different jobs inside your cells, backed by genuinely different amounts of human evidence.
I went looking for a straight answer to a simple question: if I only trusted my money to one of these, which one would that be. The honest answer took some digging, because the marketing is built to make three very different evidence tiers look like one product category.
Three different jobs wearing one marketing label
The word doing the most hiding here is "cellular." All three compounds get filed under it, but they intervene in three separate cleanup-and-repair systems that happen to share a neighborhood.
Autophagy is the cell's general housekeeping process — breaking down and recycling damaged proteins and worn-out components. Spermidine's proposed mechanism runs through here, largely by inhibiting an enzyme (a class of acetyltransferases, EP300 chief among them) that otherwise keeps housekeeping switched down.
Mitophagy is a specific, narrower version of that same cleanup system, aimed only at damaged mitochondria — the components that generate a cell's energy and, when they degrade, are thought to drive a meaningful share of age-related muscle and metabolic decline. This is urolithin A's target.
NAD+ salvage isn't a cleanup system at all. NAD+ is a molecule every cell needs for basic energy metabolism and for DNA repair enzymes to function, and its levels measurably decline with age. NMN and NR, the two most common NAD+ precursor supplements, feed into the salvage pathway your cells already use to rebuild NAD+ from its breakdown products.
Three different systems, three different plausibility stories, and — this is the part the shelf doesn't tell you — three very different amounts of actual human trial data behind them.
Urolithin A: the one with the most coherent human data
Urolithin A isn't something you eat directly. It's a metabolite your gut bacteria produce from ellagitannins, compounds found in pomegranates, walnuts, and some berries — and a large share of people simply don't have the right gut flora to produce much of it, which is the actual justification for taking it as a supplement rather than just eating more pomegranate.
The preclinical case for urolithin A driving mitophagy is solid, built up over years of cell and animal work. What sets it apart from the other two is that it's also the one with published human trials designed around a real, measurable outcome rather than a blood biomarker. An early trial established that oral urolithin A was safe and detectable in muscle tissue at achievable doses. A later randomized trial in middle-aged and older adults, run over several months, measured actual muscle endurance — how many repetitions someone could do, how far they could go — and reported improvement on some of those measures relative to placebo.
That's a meaningfully different kind of evidence than "a biomarker moved in the right direction." It's not overwhelming — the trials remain few, industry-funded (largely by Amazentis, the company behind the Mitopure ingredient), and modest in size relative to what you'd want before calling anything settled. But among the three compounds here, urolithin A is the one where a human trial actually measured something you'd feel in daily life and found a signal.
Spermidine: biologically compelling, clinically still unproven
Spermidine has the most dramatic origin story of the three. It's a naturally occurring polyamine — found in wheat germ, soy, mushrooms, aged cheese — and in yeast, worms, flies, and mice, supplementing it or boosting it genetically has reliably extended lifespan across a remarkable range of species. That kind of cross-species consistency is exactly what makes aging researchers pay attention to a compound in the first place.
The gap shows up the moment you move to humans. There's real epidemiological data here: population studies, mostly out of Europe, have found that people who eat more spermidine-rich foods tend to have lower cardiovascular and all-cause mortality over long follow-up periods. That's suggestive, but dietary-pattern studies always carry the same asterisk — people who eat more polyamine-rich whole foods also tend to differ from the rest of the population in a dozen other ways that have nothing to do with spermidine specifically.
The interventional evidence — actually giving people spermidine and measuring an outcome — is thin. The most-cited human trial is a small randomized study out of Germany that gave older adults with mild memory complaints a spermidine-rich supplement and tracked subjective and objective memory measures; it reported some encouraging signals, but it was small, short, and has not been the kind of finding that gets reliably reproduced at scale since. Nothing in the human literature resembles the muscle-endurance trial urolithin A has behind it. Spermidine remains the compound with the best animal story and the least-settled human story of the three.
NAD+ precursors: the middle tier, and a moving target
NMN and NR sit between the other two in a specific way: the mechanistic case is strong, the human trials are more numerous than spermidine's, and the results are more consistent than spermidine's — but the endpoints those trials measure are mostly still one step removed from something you'd notice.
NAD+ decline with age is well documented, and both NMN and NR reliably do the one thing they're supposed to do: raise measured NAD+ levels in blood, consistently, across a fair number of small randomized trials now. Where the evidence gets murkier is what that increase actually buys you. Trials on insulin sensitivity, blood pressure, exercise capacity, and general metabolic markers have produced a genuinely mixed record — modest improvement in specific subgroups (postmenopausal women and people with mild insulin resistance show up more than once), no meaningful difference from placebo in plenty of others. Short-term safety across these trials has generally looked clean, which is worth something on its own. There's also a regulatory wrinkle specific to NMN worth knowing: its status as a legally marketable dietary supplement in the United States has been contested, tied to an earlier drug-development filing on the same molecule, and the situation has shifted more than once. If you're shopping for NMN specifically, check its current standing rather than assuming the shelf product looked the same a year ago.
Side by side
| Compound | Primary mechanism | Strength of human evidence | Typical studied dose | Biggest open question |
|---|---|---|---|---|
| Urolithin A | Mitophagy (mitochondrial cleanup) | Strongest of the three — small but real trials on functional outcomes like muscle endurance | ~500–1,000 mg/day in trials | Trial count and sample sizes are still small; mostly industry-funded so far |
| NAD+ precursors (NMN/NR) | NAD+ salvage pathway | Middle tier — reliably raises NAD+ levels; downstream functional benefit is mixed across trials | ~250–500 mg/day (NMN); ~250–300 mg/day (NR) in trials | Whether raising NAD+ translates into an outcome you'd actually feel |
| Spermidine | Autophagy (general cellular cleanup) | Weakest human data — strong animal/cross-species evidence, mostly observational and small-trial evidence in humans | ~1–6 mg/day spermidine-equivalent in the main human trial | No human trial yet resembles the animal lifespan-extension findings |
What we actually know about dosing and safety
Across all three, the honest summary is: short-term safety looks reasonably clean in the trials that exist, and long-term safety data — the kind you'd want before treating any of these as a permanent daily habit — simply doesn't exist yet for any of them. These compounds haven't been on the market, at these doses, for long enough to have it.
A few specifics worth knowing. Urolithin A trials have mostly reported mild, infrequent gastrointestinal complaints and nothing more serious at studied doses. Spermidine, because it ramps up autophagy broadly rather than in one targeted system, carries a theoretical caution worth taking seriously if you're on immunosuppressive therapy or being treated for certain cancers — autophagy modulation cuts both ways in active cancer research, and "more autophagy is always better" is not a settled claim. NAD+ precursors have looked clean in short trials, though a few flag mild GI or flushing-type effects at higher doses.
The bigger practical risk with all three isn't the molecule — it's the unregulated market they're sold in. Dose accuracy, purity, and whether the bottle contains what the label claims vary a lot between brands, and none of it is verified the way an actual pharmaceutical would be. If you take any of these, third-party testing (a certificate of analysis, not just a badge on the label) matters more than which brand's marketing was more convincing.
Why the marketing keeps blurring them together
None of this stops companies from selling all three in one "longevity stack," bundled and priced together, marketed with the same handful of words rotating across every panel. It's a reasonable business move and a genuinely misleading one at once, because it invites you to treat three different evidence tiers as interchangeable ingredients in the same recipe.
I've spent a fair amount of time in software, and the analogy I keep landing on is shipping three modules in one release when only one of them actually has integration tests. They all compile. They all sit in the same package. But "in the same release" and "equally trustworthy" are not the same claim, and a good engineer reads the test coverage before trusting the build — the same discipline is worth applying before trusting a supplement claim, not because the industry is uniquely dishonest, but because nobody else is going to sort the tiers out for you.
An honest longevity supplement label would separate "backed by a trial that measured something you'd notice" from "raises a blood marker reliably" from "extends lifespan in five other species and hasn't really been tested in yours yet." That label doesn't exist, so the sorting is on you.
Where that leaves me, personally, is closer to urolithin A than to the other two — not because it's proven, but because it's the one where somebody actually measured whether people could do more with their bodies afterward, and found something. That's a low bar for "proven." It's still higher than the other two have cleared.
FAQ
Can I just eat pomegranates and walnuts instead of taking urolithin A? You can raise the raw material your gut bacteria use to make it, but many people don't carry the gut flora needed to convert those precursors efficiently — the actual argument for taking the metabolite directly rather than the food.
Is it safe to combine all three? Nobody has tested spermidine, urolithin A, and an NAD+ precursor together, so any claim about combined safety or benefit is speculation dressed up as science. Taking more than one is a reason for more caution, not less.
Which one has the best evidence for muscle and physical function specifically? Urolithin A, currently. It's the only one of the three with a published randomized trial that measured a functional muscle outcome directly rather than a blood marker.
Should I stop eating spermidine-rich foods since the supplement evidence is weak? No — different question. Wheat germ, aged cheese, mushrooms, and soy are ordinary foods with their own nutritional case, independent of the longevity claim. The weak evidence concerns concentrated supplementation, not the foods themselves.
Does any of this replace exercise, sleep, or diet as the foundation? No, and none of the researchers behind this work would claim otherwise. Every compound here is being studied as a possible addition on top of the basics, not a substitute for them — the trial designs assume a baseline of ordinary healthy behavior, not a replacement for it.