A probiotic given to very premature babies took hold in the gut within a week. The title says it is linked to better brain white matter. The brain scans of the two groups did not differ

A baby born at full term and breastfed usually ends up with a gut full of Bifidobacterium within weeks. A baby born three months early does not. The first settlers tend to be Staphylococcus, then Enterococcus and the Enterobacteriaceae family, and the bifidobacteria arrive late. Those same early weeks are when the brain's wiring is being laid down and insulated, work that would otherwise have happened in the womb. So neonatal researchers have a reasonable question: if you get the friendlier bacteria in sooner, does the brain do better? A Dutch team from Wageningen University, the Wilhelmina Children's Hospital in Utrecht and Danone Research & Innovation published their attempt at an answer in the journal Gut Microbes on 7 October 2026.

The trial behind it was double-blind and ran in the neonatal intensive care unit in Utrecht, with babies followed into seven regional hospitals when they were transferred. Parents of 338 eligible babies were approached, and between October 2018 and May 2022, 99 babies born at 24 to 30 weeks were randomised. From two to three days after birth until 36 weeks, or going home, half were given a daily dose of Bifidobacterium breve M-16V by feeding tube or syringe, along with a prebiotic mix of two oligosaccharides and the amino acid L-glutamine added to their milk. The other half got lookalike powders made of maltodextrin and milk proteins. Stool was collected from the first days onward, and each baby had a brain MRI at around the date they should have been born.

On the gut side, the result is clear. One week in, the given strain could be found in 97% of the babies who received it and 9% of those who did not. Bifidobacterium as a whole was more abundant in the supplemented babies through week 6, Staphylococcus was lower at week 1 and Enterococcus lower at week 3. The bacteria in the supplemented babies also carried more of the genetic equipment for making amino acids and for fermenting pyruvate into acetate and lactate. After week 6 the two groups could no longer be told apart on any of this. The authors call the colonisation 'robust but transient'. Weight gain was the same in both groups.

Now the brain, which is what the title is about. The paper's own sentence is: 'No significant differences in the assessed brain injury markers were observed between the intervention and control groups.' The babies who got the supplement did not have better scans than the babies who got placebo. What the team did next was a post hoc analysis, meaning one decided on after the data were in. They set the randomised groups aside and compared babies by what their scan showed. The 20 babies with delayed myelination of white matter had a different mix of gut bacteria at week 3 from the 45 without it, with less Bifidobacterium breve and less capacity for pyruvate fermentation. The 11 babies with a thinned corpus callosum differed at week 1 from the 52 without. The fermentation pathways that were lower in the babies with delayed myelination were the same ones the supplement had raised.

That is two separate findings placed next to each other: the supplement changed the gut bacteria, and the gut bacteria looked different in babies with two particular scan findings. It is not a finding that the supplement changed the scans, because when that was tested directly it showed nothing. The second finding is also an observation, not an experiment. A baby who is sicker in the first weeks may have both a disturbed gut and slower brain development for reasons that have nothing to do with one causing the other. And the team looked at a long list of scan measures. Two were tied to the bacteria. The rest, including the direct measurements of white-matter structure taken from diffusion imaging, were not.

There are practical complications as well. By week 3 the probiotic strain was showing up in 44% of the placebo babies. Bacteria travel around a neonatal unit on hands and equipment, and this is common in probiotic trials. It blurs the comparison, and the authors say it may have hidden some of the effect. That is possible, though it cannot be shown from this trial. Of the 99 babies randomised, 84 had usable stool data and 76 of those had a usable scan. The trial's size was set for its main question, not for this analysis. And because three things were given together, nothing here says which of the bacterium, the fibres or the glutamine did what.

Who paid matters. The work was funded by a Dutch public grant with co-financing from Danone Research & Innovation, and five of the authors are Danone employees. The paper states that they took no part in the microbiome or MRI analysis. The abstract ends by saying the study 'demonstrated that early microbial modulation could influence brain development'. On the evidence in the paper, 'could' is doing a lot of work in that sentence.

Our reading: this is good evidence that a daily Bifidobacterium breve with prebiotics settles into the gut of a very premature baby within a week and shifts the early bacterial mix for about six weeks. It is a lead, not evidence, that this does anything for the brain. The same trial compared the scans and found no difference between the groups. The honest next step is a trial sized to answer the brain question, with the children's development followed for years. For parents of a premature baby, the practical point is short. Whether a baby in a neonatal unit gets a probiotic, and which one, is a decision for the neonatal team. Products sold to the public are not made or tested for babies this small and should never be given without the team's say.