GOS prebiotic shifted gut microbiome and brain GABA in young women, but anxiety scores held steady

A 28-day course of the prebiotic galacto-oligosaccharides shifted the gut microbiome and altered brain chemistry in young women, a new randomized trial has found. Anxiety scores did not budge.

The trial, published in Brain, Behavior, and Immunity by Nicola Johnstone and Kathrin Cohen Kadosh of the University of Surrey, is one of the most comprehensive human probes of the gut-brain axis to date. It tracked 83 women aged 17 to 25 through four weeks of daily GOS supplementation, measuring not just what participants reported feeling but what was happening inside their brains and their guts at the chemical level.

Galacto-oligosaccharides, or GOS, are prebiotic fibers that the human gut cannot digest. They pass through the small intestine intact and ferment in the colon, where they feed Bifidobacterium species. The same fibers occur naturally in breast milk and, at lower concentrations, in legumes and some dairy products. The supplement form is made from lactose by enzymatic conversion. Animal work has shown that prebiotics can shift concentrations of GABA and glutamate in brain regions that govern emotional regulation. Whether that axis works the same way in humans, and at a dose someone could actually take, has been the question hanging over the field.

The Surrey team randomized 83 women aged 17 to 25 to receive either 7.5 grams of GOS per day, containing 5.5 grams of the active ingredient Biotis GOS, or a maltodextrin placebo. The trial was double-blind. Assessments were taken at baseline, after 28 days of supplementation, and again 28 days after the supplement was withdrawn. The design is clean: three time points, an active prebiotic against a matched control, and enough participants to detect a moderate effect if one existed.

The primary endpoint was trait anxiety, scored on the State-Trait Anxiety Inventory. GOS did not beat placebo (p = 0.443). At the follow-up visit, four weeks after the supplement stopped, anxiety scores in the GOS group trended lower (p = 0.069), but the result stayed shy of statistical significance.

The null on anxiety masked a more interesting set of secondary results. The researchers used proton magnetic resonance spectroscopy (1H-MRS) to measure GABA and glutamate in three brain regions. GABA, the brain’s main inhibitory neurotransmitter, dropped at trend significance in the GOS group. In the inferior occipital gyrus, the reduction reached p = 0.053 across all participants. When the analysis was restricted to women who started the trial with above-average anxiety, GABA also fell in the dorsolateral prefrontal cortex, a region tied to executive function and emotional regulation (p = 0.088). Both effects were still present at the 28-day follow-up.

The direction of change matters. Cortical GABA reductions have been linked to better cognitive flexibility and less rumination in prior work. Less GABA in these regions is not necessarily a bad outcome. The authors do not call the shifts beneficial, but the fact that they persisted four weeks after the supplement stopped suggests the signal was real, not noise.

How a fiber in the gut changes a neurotransmitter in the brain is the central question of gut-brain axis research. The leading hypothesis points to short-chain fatty acids produced when bacteria ferment prebiotics. Those SCFAs enter circulation, cross the blood-brain barrier, and may influence the synthesis or release of GABA. Johnstone and Cohen Kadosh did not measure SCFAs directly, so the mechanistic link in this particular trial remains inferred rather than demonstrated.

The gut results were less ambiguous. Bifidobacterium abundance spiked during supplementation (p = 0.001) and dropped back to baseline after the supplement stopped. This matches the known profile of GOS as a bifidobacteria substrate. The supplement did what it was supposed to do in the colon. Overall microbial diversity did not change. Neither did what participants ate during the trial. That matters because it rules out the simpler explanation: that GOS just pushed out worse food choices rather than doing anything itself.

Social anxiety and depression scores did not move, but reaction times did. High-anxiety participants on GOS got faster. Simple reaction times improved (p = 0.036), and choice-based tasks saw a sharper gain (p < 0.001). The effect was on speed, not accuracy. The authors suggest it may point to a mild attentional benefit, not a direct anti-anxiety effect. Most nutritional interventions do not shift cognitive performance in healthy people, so the reaction-time finding, if replicated, would matter beyond this trial.

The gut-brain axis has produced hundreds of animal studies showing that changing the gut microbiome alters behavior, stress responses, and neurotransmitter levels. Far fewer human trials have tested the same question. The Johnstone and Cohen Kadosh paper stands out because it did not stop at asking people how they felt. It scanned their brains for the chemical intermediates.

A 2024 mouse study by Tang and colleagues, published in the Journal of Agricultural and Food Chemistry, found that GOS reduced skin inflammation and anxiety-like behaviors in mice with atopic dermatitis. The mechanism ran through restructured gut microbiota and increased short-chain fatty acid production. The Surrey trial asked whether any of that preclinical signal reaches a healthy human. The answer: some of it does, but not the part someone would notice.

The results come with real limits. Every participant was a young woman between 17 and 25. That is not a criticism; it is a study population, and a relevant one, since anxiety disorders emerge disproportionately in young women. But it means the findings do not automatically extend to men, to older adults, or to people with diagnosed anxiety disorders rather than subclinical trait variation. The dose of 5.5 grams of active GOS per day sits at the lower end of what preclinical protocols have used, though it matches what is sold in commercial prebiotic products. The trial ran for four weeks. The anxiety trend at the eight-week mark, four weeks after stopping, hints that a longer intervention might have separated the groups more cleanly. The trial was preregistered on ClinicalTrials.gov (NCT03554694), which cuts against the worry that the outcomes were selected after seeing the data.

None of this means GOS is useless. The Bifidobacterium increase and the GABA shifts were real. The fibers did what they are supposed to do in the gut, and the brain registered the change. What did not follow was a drop in how anxious participants felt. That gap, between a biological signal and a subjective experience, is where the whole gut-brain axis field has been stuck. Closing it will take trials that run longer than four weeks, enroll men as well as women, test higher doses, and include people whose anxiety is severe enough that a measurable drop is possible. The Johnstone and Cohen Kadosh trial is a good example of how to report a null result: publish the secondary outcomes, share the data, and let the field build on what actually happened rather than what the researchers hoped would happen. As always, consult your doctor before starting any supplement.

References

1. Johnstone N, Cohen Kadosh K. A randomised controlled trial of the effects of galacto-oligosaccharides on the gut brain-axis of young females. Brain, Behavior, and Immunity 129:573-584. 2025. https://doi.org/10.1016/j.bbi.2025.06.020
2. Tang L, Cao X, Chen S, et al. Dietary galacto-oligosaccharides ameliorate atopic dermatitis-like skin inflammation and behavioral deficits by modulating gut microbiota-brain-skin axis. Journal of Agricultural and Food Chemistry 72(15):8375-8390. 2024. https://doi.org/10.1021/acs.jafc.4c00205