14 blood metabolites in midlife predict brain health decades before Alzheimer’s symptoms

Brain changes that eventually surface as Alzheimer's disease begin accumulating years, possibly decades, before any symptom appears. A new study published in Nature Aging suggests that blood metabolites measured in midlife could serve as early sentinels of that silent deterioration, and that a meaningful share of those metabolites are shaped by factors you can actually change: your gut microbiome, your medications, and your lifestyle habits.

14 blood metabolites tied to midlife cognition and brain structure

Researchers from the Rotterdam Study analyzed 1,387 plasma metabolites in 1,082 dementia-free participants with a mean age of around 63, then cross-referenced those measurements against cognitive test scores and brain MRI data. After adjustment for age, sex, BMI, and lipid-lowering medication, 14 metabolites showed significant associations with general cognition. Three of them stood out on the protective side: ergothioneine, uridine, and 2-deoxyuridine, all of which were higher in people with better cognitive scores. Seven sulfated xenobiotic metabolites went the other direction; lower levels of these compounds tracked with sharper thinking.

The MRI picture added a parallel story. Twenty-one metabolites were linked to total brain volume, and S-adenosylhomocysteine (SAH), a compound involved in methylation reactions in the body, was associated with greater white matter lesion volume, a marker of small-vessel disease. Importantly, the metabolite signatures for cognition and for brain structure did not overlap much, yet their effect patterns correlated significantly: the same metabolites that tracked with better cognition tended to track with larger total brain volume (r = 0.33) and hippocampal volume (r = 0.27). Research on blood metabolites as cognitive aging biomarkers has long suggested that metabolic shifts precede clinical impairment, and this study adds a finely resolved midlife snapshot to that picture.

Why these metabolite signals could matter for Alzheimer's prevention

The study's strongest translational signal came from connecting midlife metabolite levels to actual Alzheimer's diagnoses years later. The overall metabolite signature for cognition correlated with incident AD diagnosis at r = -0.73 (P = 1.03 x 10-13), meaning the metabolites that predicted better cognition in midlife were strongly inversely associated with who went on to develop dementia. Two metabolites showed the largest individual effects on incident AD: beta-cryptoxanthin, a carotenoid found in foods like papaya and tangerines, and imidazole propionate, a compound produced when gut bacteria break down the amino acid histidine.

The fact that both of these compounds connect to gut microbiome activity is notable. Gut microbiota explained more than 5% of the variance in imidazole propionate levels (11.3%) and in beta-cryptoxanthin (5.1%), placing them in a category of metabolites where the microbiome is a meaningful driver. Work on metabolomics for Alzheimer's disease early detection has consistently highlighted the need to understand which metabolite signals are stable, causal markers versus downstream noise; this data suggests gut-linked metabolites deserve particular attention in future intervention work.

How gut microbiome and lifestyle shape the brain-linked metabolome

The study went beyond identifying the metabolites and asked what drives them. Lifestyle factors, clinical variables, and medication use together explained considerable variance in several key compounds. Smoking was the dominant driver for sulfated xenobiotics: it raised their levels, matching the worse-cognition direction of those metabolites. Ergothioneine, by contrast, was higher in people who drank moderate amounts of alcohol, had more years of education, lower BMI, and lower blood pressure.

Medication use produced one of the paper's most concrete findings. Antacid use, including proton pump inhibitors (PPIs), was strongly associated with lower blood ergothioneine levels. A mediation analysis estimated that ergothioneine accounted for 31.5% of the negative effect of antacid medication on cognition, a finding that aligns with prior reports linking PPI use to dementia risk. Six microbial genera, including Clostridium innocuum group, Sellimonas, and Hungatella, were linked to lower ergothioneine levels; all six have separately been associated with risk factors for Alzheimer's disease.

Lifestyle factors alone explained 28.6% of the variance in 2-naphthol sulfate (a sulfated xenobiotic) and 4.7% of ergothioneine variance, numbers that are modest but meaningful given how complex metabolite regulation is. Findings like these illustrate why blood metabolites predicting dementia risk may ultimately be most useful not just as biomarkers but as readouts of modifiable risk.

Replication and the study's real limitations

Nine of the 14 cognition-associated metabolites replicated in an older Dutch cohort (mean age 76), and 7 of 10 tested replicated in a US-based Alzheimer's Disease Research Centers cohort, lending reasonable confidence to the core findings across populations and age groups. Elastic net models using all measured metabolites explained 17.5% of hippocampal volume variance, even though no single metabolite reached significance for hippocampal volume on its own.

The limitations are real and the authors acknowledge them clearly. The study is cross-sectional at its core: blood and imaging were not collected at the same moment, and causal direction cannot be inferred. Dietary data and physical activity records were not available, which is a meaningful gap given how directly diet shapes both the metabolome and the microbiome. Sex-stratified analyses suggested that lower bile acid levels were associated with better cognition and larger hippocampal volume only in women, but statistical power was insufficient to confirm that finding after multiple testing correction. As challenges in metabolomics for neurodegeneration research persist, translating these associations into clinical tools will require longitudinal studies that track both metabolites and cognitive change in the same individuals over time.

What the ergothioneine finding opens up for research

Ergothioneine is a sulfur-containing amino acid that humans cannot synthesize themselves; dietary intake, primarily from mushrooms and some fermented foods, and gut microbial activity are the main sources. Lower brain levels of ergothioneine have been reported in prior studies of people with worse cognition and Alzheimer's disease, and this paper extends that association into a midlife, presymptomatic population while also pointing to antacid use as a modifiable upstream factor.

The finding that a common medication class could suppress a cognition-linked metabolite, and that the suppression explains nearly a third of the medication's apparent cognitive effect, is the kind of mechanistic thread that warrants a dedicated clinical follow-up. Whether supplementing ergothioneine, adjusting antacid prescribing in at-risk patients, or targeting the gut genera that deplete it could shift cognitive trajectories remains an open question. For now, the study offers a metabolic map of midlife brain health that is more detailed, more actionable-looking, and more connected to everyday exposures than most previous work in this space.

Evidence rating

EMERGING — 76/100

Caveats:

• Cross-sectional design: this study shows associations, not cause-and-effect. Ergothioneine levels and cognition were measured at the same time, so it is not possible to confirm which came first.
• No funding source was disclosed by the authors, so independent verification of financial interests could not be performed.
• The gut microbiome explained only 5–11% of variance in the most microbiome-linked metabolites; genetics and lifestyle were the dominant drivers of most metabolite levels.
• Dietary data and physical activity were not available, which limits conclusions about how modifiable lifestyle choices shape these metabolites.
• Mediation analysis assumed a temporal sequence that the cross-sectional data cannot confirm — causal inference is not warranted.

Large cross-sectional study (discovery n=1,082; replicated in two independent cohorts of 874 and 512 older adults) examining associations between 991 blood metabolites and cognitive and MRI markers of brain health in dementia-free midlife adults. The design establishes association, not causation, and mediation estimates rely on assumed temporal ordering that the cross-sectional data cannot confirm. Key findings — particularly ergothioneine and antacid-related associations — replicated across cohorts, though longitudinal metabolomics data are limited and dietary confounders were not measured.

References

1. Ahmad S, Wu T, Arnold M, Hankemeier T, Ghanbari M, Roshchupkin G, Uitterlinden AG, Borkowski K, Neitzel J, Kraaij R, Alzheimer’s Disease Metabolomics Consortium, van Duijn CM, Ikram MA, Kaddurah-Daouk R, Kastenmüller G. (2026). The blood metabolome of brain health in midlife and influences of genes, microbiome and exposome. https://doi.org/10.1038/s43587-026-01149-4
2. Zhiguang Huo, Brinda K. Rana, Jeremy A. Elman, et al. (2020). Metabolic Profiling of Cognitive Aging in Midlife. Frontiers in Aging Neuroscience. https://doi.org/10.3389/fnagi.2020.555850
3. Prabhakar Tiwari, Anu Gupta, Meenakshi Kaushik, et al. (2025). Comprehensive metabolomics profiling reveals novel biomarkers and pathways for early detection of Alzheimer’s disease. Brain Communications. https://doi.org/10.1093/braincomms/fcaf410
4. Michael Sasner, Christoph Preuss, Ravi S. Pandey, et al. (2024). In vivo validation of late‐onset Alzheimer's disease genetic risk factors. Alzheimer's & Dementia. https://doi.org/10.1002/alz.13840
5. Md Abdul Hakim, Li Li (2026). Decoding the Metabolic Signatures of Neurodegeneration Diseases: Advances in Mass Spectrometry-Based Metabolomics. Metabolites. https://doi.org/10.3390/metabo16030206