Spermidine and longevity: a striking survival link, and a trial that fell flat

Spermidine, a compound found in wheat germ, aged cheese, mushrooms and soy, is among the more scientifically interesting geroprotector candidates — and a clean example of why an eye-catching population association is not proof a supplement works. In one prospective cohort, people eating the most spermidine had less than half the death rate of those eating the least [s1]. But the first year-long randomised trial of a spermidine supplement, in older adults at risk of dementia, found no effect on memory or on the biomarkers it measured [s2]. The gap between those two results is the point.

Spermidine's appeal is mechanistic: it induces autophagy, the cellular housekeeping process that clears damaged components and that declines with age, and supplementation extends lifespan in yeast, flies, worms and mice. Whether any of that carries into humans is what the two studies below actually tested.

The observational signal is large

The survival link comes from the Bruneck study, a prospective community cohort of 829 adults aged 45 to 84, whose diets were assessed by repeated validated questionnaires in 1995, 2000, 2005 and 2010 [s1]. Over follow-up to 2015, 341 deaths occurred, and all-cause mortality fell steeply across thirds of increasing spermidine intake: from 40.5 deaths per 1,000 person-years (95% CI 36.1 to 44.7) in the lowest third, to 23.7 (20.0 to 27.0) in the middle, to 15.1 (12.6 to 17.8) in the highest [s1]. Adjusted for age, sex and caloric intake, each one-standard-deviation increase in spermidine intake was associated with a lower risk of death (hazard ratio 0.74, 95% CI 0.66 to 0.83) [s1].

That is a striking gradient — the death rate in the lowest-intake group was more than double that in the highest [s1]. But it is diet, not a supplement, and it is observational. People who eat spermidine-rich whole foods differ in countless ways from those who do not, and no adjustment fully removes that confounding. The study can show a correlation; it cannot show that swallowing spermidine causes the benefit.

The randomised trial did not confirm it

That is exactly the test a randomised trial exists to run, and the first long one came back negative. The SmartAge trial randomised 100 older adults (mean age 69) with subjective cognitive decline — a group at elevated dementia risk — to 12 months of a spermidine-rich supplement extracted from wheat germ (0.9 mg spermidine a day) or a cellulose placebo [s2]. Eighty-nine completed the intervention. On the primary outcome, a memory task, there was no significant change (between-group difference −0.03, 95% CI −0.11 to 0.05, P = .47), and the secondary neuropsychological, behavioural and physiological outcomes were null as well [s2]. Adverse events were balanced between groups; exploratory analyses hinted at possible effects on inflammation and verbal memory, but those are the kind of after-the-fact signals that need their own trial before they mean anything [s2].

One detail deserves emphasis because supplement labels tend to blur it: the survival association was with spermidine from food, measured across 2,540 dietary assessments over 15 years, whereas the trial tested a purified wheat-germ extract delivering 0.9 mg a day [s1][s2]. A whole-food diet rich in spermidine is also rich in fibre, polyphenols and everything else that travels with legumes, whole grains, mushrooms and aged cheese — so the mortality gradient may reflect a dietary pattern rather than the single molecule a capsule isolates. Untangling those is exactly what a randomised trial is for, and the one that has reported did not find the predicted benefit [s2].

Reading the two together

The honest summary is that spermidine has a promising mechanism and a large observational survival association, and that the one long randomised trial of a supplement to date did not find the cognitive benefit its rationale predicted [s1][s2]. That does not falsify the whole idea — the trial tested a low dose, one outcome and one at-risk group over one year, and dietary spermidine may act differently from a purified supplement — but it should discipline any marketing that cites the mortality figure while omitting the null trial. A geroprotector's story is only as strong as its randomised evidence, and here that evidence is, so far, a null [s2].

The same observational-strong, trial-weak pattern runs through much of this market — it recurs in the evidence for urolithin A and muscle, for NAD+ IV therapy, and for the NMN and NR supplements sold on the same anti-ageing shelf.