In 2023, 164 volunteers in St. Louis, aged 68 to 94, all of them performing normally on cognitive tests, did something unremarkable: they collected a stool sample and handed it in. Forty-nine of them, brain scans and spinal fluid would show, already carried the amyloid and tau signatures of preclinical Alzheimer's disease. They did not know it. Their families did not know it. But the bacteria in their intestines, the study by Aura Ferreiro, Beau Ances, Gautam Dantas and colleagues at Washington University reported in Science Translational Medicine, already looked different from those of their neighbours in the same study eating broadly the same food. The microbial differences tracked with amyloid and tau, and not with markers of neurodegeneration, which is to say they showed up early, before the brain had started visibly losing tissue. That was a cross-sectional study, one snapshot, no follow-up, and the authors said so.
That snapshot is now one panel in a larger picture. On 29 July 2026, a team led by D. M. Sithara Dissanayaka at Edith Cowan University in Western Australia, with co-authors including Stephanie Rainey-Smith, Hamid Sohrabi, Colin Masters and Ralph Martins, published a narrative review in Nutrients titled "Gut Microbiome Changes in Preclinical Alzheimer's Disease". It is worth being precise about what that is: a narrative review generates no new data and follows no pre-registered search protocol. It is a reading of the field by people who work in it. What they read is a consistent thinning of short-chain fatty acid producers, the genera Faecalibacterium, Roseburia and Eubacterium, in people with Alzheimer's and with mild cognitive impairment, alongside disrupted fermentation pathways, altered lipid and amino acid metabolism, and reduced short-chain fatty acid biosynthesis. And then the harder claim: that a small number of studies in cognitively unimpaired people with biomarker evidence of amyloid have found the same kinds of early shifts, before any symptom.
The reason this belongs in a column about the grammar of living systems is that short-chain fatty acids are not a gut story that happens to touch the brain. They are, as the review puts it, a major interface between microbial activity and host physiology. Acetate, propionate and butyrate dominate; branched forms such as isobutyrate and valerate appear at lower concentrations. They act locally in the intestine and systemically once absorbed, touching gut barrier integrity, immune regulation and inflammatory signalling. Read that way, the colon and the cortex are not two organ systems exchanging messages. They are one continuous chemical sentence, and dysbiosis is a change in its syntax. The review is careful to add the qualifier that undercuts any neat reading: direct human evidence in preclinical Alzheimer's remains limited, and the effects of these metabolites appear to depend on concentration and biological context, which is a scientist's way of saying more is not automatically better.
Which leaves the question the review is honest enough to put at the centre. Did the microbiome change and then the amyloid accumulate? Did an already-changing brain, or the behaviour and metabolism of a body in the early years of a slow disease, change the microbiome? Or does some third thing, diet, medication, sleep, inflammation, move both? Almost all of the available evidence is cross-sectional, which cannot answer that. Most of it comes from people who already have Alzheimer's or mild cognitive impairment, not from biomarker-confirmed preclinical cases. Most of the sequencing stops at genus level, too coarse to name a species, let alone a mechanism. The studies also disagree with each other: a 2022 study of 78 cognitively normal Korean participants, 18 of them amyloid-positive, published in PLOS ONE by Joon Hyung Jung and colleagues, found Megamonas, Serratia, Leptotrichia and a Clostridium genus enriched in the amyloid-positive group, a different cast of characters from the SCFA producers. The review's prescription is specific rather than rhetorical: longitudinal, species-level metagenomic studies in cognitively unimpaired people grouped by amyloid status, measuring diet, microbial functional pathways, short-chain fatty acids and amyloid pathology in the same cohort, with faecal microbiota transplantation named among the interventions worth testing.
For a family reading this over breakfast, the practical content is smaller than the headline suggests, and it is important not to pretend otherwise. Mediterranean, DASH, MIND and prudent dietary patterns are associated with the microbial profiles that look better on paper, and with better cognitive outcomes in observational work, but the review states that direct evidence connecting diet to the microbiome to amyloid pathology is still thin. Nobody has shown that changing what is in your gut changes what is in your brain. There is no stool test for preclinical Alzheimer's, no probiotic with an evidence base for preventing it, and no dose of anything that this paper supports. What there is, is a reasonable hypothesis with a real research programme attached, and a set of investigators asking for the studies that could break it.