Phosphatidylserine supplement improved short-term memory in Chinese MCI trial

Phosphatidylserine is one of those supplements with a mechanism so elegant it feels like it should work. An endogenous phospholipid concentrated in neuronal membranes, it keeps cell walls fluid, supports neurotransmitter release, and declines with age. The hypothesis that topping it up might preserve cognition in older adults has been kicking around since the 1990s, backed by a handful of small trials and a 2015 narrative review that called it “required for healthy nerve cell membranes and myelin.”

A new randomized trial from China puts that hypothesis to its longest test yet. The results are encouraging but come with a catch that matters for anyone reaching for a bottle.

The trial, published in the Journal of Affective Disorders in January 2025, randomized 190 older adults with mild cognitive impairment in Tianjin to 12 months of either a daily supplement or a placebo. The supplement contained phosphatidylserine (PS), alpha-linolenic acid (ALA), Ginkgo flavonoids, and three B vitamins. The intervention group showed statistically significant improvements in short-term memory, arithmetic performance, and a measure of executive function called the similarity test, alongside increases in serum omega-3 fatty acids and three neurotransmitters: acetylcholine, GABA, and serotonin.

The numbers are solid for a trial of this size. Short-term memory improved with a beta coefficient of 0.600 (95 percent CI 0.399 to 0.800), the similarity test registered a beta of 1.070 (95 percent CI 0.472 to 1.667), and arithmetic testing improved by a beta of 0.688 (95 percent CI 0.103 to 1.274). All three crossed the significance threshold cleanly. Serum ALA, DHA, and EPA all rose in the treatment group, confirming the n-3 PUFA component was bioavailable. Neurotransmitter changes were modest in absolute terms but all moved in the expected direction: acetylcholine rose by a beta of 0.441, GABA by 0.009, and serotonin by 0.160.

But none of those results answer the question the paper’s title invites, which is whether the phosphatidylserine specifically did anything. The supplement was a five-ingredient matrix. Each daily dose delivered 144 milligrams of ALA, 31.5 milligrams of PS, 3.6 milligrams of Ginkgo total flavonoids, and small amounts of thiamine, pyridoxine, and folic acid. The cognitive signal could be driven by any one of those components, or any combination of them. And the mediation analysis the authors ran points partly away from PS.

The mediation model found that the increase in serum ALA explained roughly 20 percent of the short-term memory improvement (average causal mediation effect of 0.132, 95 percent CI 0.053 to 0.225). That leaves 80 percent of the memory benefit unaccounted for by any single measured pathway. It could be the Ginkgo flavonoids. Ginkgo has its own clinical evidence on cerebral blood flow and cognition, independent of PS. It could be the B vitamins quietly correcting subclinical deficiencies in an older population. Or it could be a synergistic effect of the whole matrix that would vanish if you isolated any one ingredient.

This is not an academic quibble. It goes to what a consumer would actually buy based on the headline. Someone who reads “phosphatidylserine improved memory” and purchases a standalone 100-milligram PS capsule is not replicating the intervention. They are taking roughly three times the PS dose used in the trial without the ALA, Ginkgo, or B vitamins that accompanied it.

The dose discrepancy runs in both directions. The Duan trial used 31.5 milligrams of PS per day, roughly one-tenth of the effective dose range established by prior research. A 2015 review by Glade and Smith in the journal Nutrition, covering 127 articles on PS and the human brain, concluded that exogenous PS is absorbed efficiently, crosses the blood-brain barrier, and supports cognitive function at doses of 300 to 800 milligrams per day. If that dose-response relationship is correct, the PS in the Duan trial was essentially a garnish. If 31.5 milligrams works when paired with ALA and Ginkgo, then the prior consensus dose needs revision.

The broader PS clinical literature does not settle the question either. Most prior trials tested doses at least three times higher than the Duan regimen, used soybean-derived PS rather than the unspecified source in this study, and ran for weeks rather than months. A 2013 pilot by Richter and colleagues gave 300 milligrams of soybean-derived PS to elderly participants with subjective memory complaints for six weeks and found improvements in immediate recall, but the sample was small and the duration short. The Duan trial’s contribution is its length and its biochemical depth, not its dosing precision.

The neurotransmitter data add biological plausibility but do not resolve the attribution problem. Acetylcholine matters for attention and memory encoding, GABA is the brain’s main inhibitory signal and its dysregulation tracks with age-related cognitive decline, and serotonin modulates mood and sleep, which influence test performance independently of any direct nootropic effect. That all three moved in the expected direction over 12 months says something in the supplement was crossing into the central nervous system and shifting neurochemistry. Which ingredient crossed, and how, remains unclear.

The trial also carried an industry footprint. Two of the 18 authors, Zhongbao Yue and Fei Ma, are affiliated with the BYHEALTH Institute of Nutrition and Health in Guangzhou. BYHEALTH is a major Chinese supplement manufacturer. The paper declares no conflict of interest, and the academic leads are from Tianjin Medical University’s School of Public Health, a credible institution. But the test product was supplied by the institute that employs two co-authors. A 12-month RCT with 190 participants and serum biomarker assays is not cheap to run, and the direction of the funding is not disclosed.

The single-site design brings its own limits. The population was Chinese older adults in Tianjin, mean age 68, roughly two women for every man. Results may not generalize to younger populations or people without existing cognitive impairment. The supplement matrix also makes the trial essentially unreplicable without BYHEALTH’s specific formulation. No independent lab can confirm the finding with the same product.

So what does the trial actually contribute? A genuine 12-month signal that a food supplement containing PS, ALA, and Ginkgo improved cognitive test scores and altered neurotransmitter levels in people with MCI. The dropout rate was zero among the 190 who completed both assessments. The mediation analysis, while not solving the attribution problem, at least acknowledges it. But the gap between what the trial name-checked in its title and what it actually tested is too wide to ignore.

If you are already taking PS at 300 to 800 milligrams per day, this trial does not change the evidence base you are relying on. If you are not taking PS and the Duan trial headlines are making you curious, know that the product in the study is not the one on the shelf. It is a five-ingredient matrix made by a Chinese supplement manufacturer that employed two of the paper’s authors. The science of phospholipids and the aging brain is real and advancing. This trial adds a data point, but not the one the title suggests.

References

1. Duan H, Xu N, Tong Y, et al. Effects of a food supplement containing phosphatidylserine on cognitive function in Chinese older adults with mild cognitive impairment: A randomized double-blind, placebo-controlled trial. Journal of Affective Disorders. 2025;369:35-42. https://pubmed.ncbi.nlm.nih.gov/39317299/
2. Glade MJ, Smith K. Phosphatidylserine and the human brain. Nutrition. 2015. https://pubmed.ncbi.nlm.nih.gov/25933483/