Plasma Metabolites Predict Accelerated Brain Aging Independent of Alzheimer's Gene
A large UK Biobank study links 77 plasma metabolites to brain age gap, with HDL lipid fractions showing the strongest signals.
Summary
Using data from 17,770 UK Biobank participants, researchers measured 249 plasma metabolites and estimated brain age from over 1,000 MRI phenotypes across six imaging modalities. They identified 77 metabolites linked to brain age gap (BAG)—the difference between predicted brain age and chronological age. Lipids in small and medium HDL particles were tied to older-appearing brains, while cholesterol fractions in VLDL, LDL, and HDL correlated with younger-appearing brains. Critically, these metabolite-brain age associations held regardless of whether participants carried the APOE ε4 Alzheimer's risk allele, suggesting metabolic profiling could serve as a broad, genetics-independent early warning system for accelerated brain aging.
Detailed Summary
Brain aging—how quickly the brain structurally deteriorates relative to chronological age—is a powerful predictor of neurological decline. Yet the biochemical signals in blood that track this process remain incompletely mapped. This large-scale study set out to fill that gap by linking comprehensive plasma metabolomics to machine learning–derived brain age in a middle-aged to older adult cohort.
Researchers drew on the UK Biobank, enrolling 17,770 participants aged 40–69 who were free of chronic brain disorders. Plasma metabolites (249 in total) were measured at baseline using nuclear magnetic resonance (NMR) spectroscopy. Approximately nine years later, participants underwent multi-modal brain MRI covering six modalities: T1-weighted structural MRI, T2*, resting-state fMRI, diffusion MRI, task fMRI, and T2-FLAIR. Brain age was estimated via LASSO regression trained on 1,079 imaging-derived phenotypes (IDPs) from a healthy subset of 3,484 participants. Brain age gap (BAG = predicted brain age minus chronological age) was the primary outcome.
Linear regression analyses, adjusted for age, sex, ethnicity, deprivation, BMI, smoking, alcohol, physical activity, and relevant medications, identified 64 metabolites significantly associated with brain age and 77 with BAG, with 55 overlapping. The dominant pattern involved lipid fractions: total lipids, cholesterol, cholesteryl esters, free cholesterol, and phospholipids in small and medium HDL particles were each linked to larger BAG (i.e., an older-appearing brain). Similarly, phospholipids and triglycerides as a percentage of total lipids across multiple lipoprotein classes correlated with larger BAG. Conversely, the proportions of cholesterol, cholesteryl esters, and free cholesterol relative to total lipids in VLDL, LDL, and variously sized HDL particles were associated with smaller BAG—suggesting these compositional ratios may reflect a more favorable brain aging trajectory.
A key secondary finding concerned the APOE ε4 allele, the strongest genetic risk factor for Alzheimer's disease. Stratified analyses and interaction testing showed that the associations of linoleic acid (LA)/fatty acid ratio, omega-6/fatty acid ratio, saturated fatty acid (SFA)/fatty acid ratio, and phospholipids-to-total lipids in large HDL with brain age were statistically consistent between APOE ε4 carriers and non-carriers (all p for interaction >0.05). This indicates that metabolic influences on brain aging are largely independent of this genetic risk factor, broadening the potential applicability of metabolic biomarkers across genetic backgrounds.
The study's scale, multi-modal imaging approach, and comprehensive metabolite panel represent important advances over prior work. However, temporal limitations and measurement timing mean causality cannot be established. Nonetheless, these findings position plasma metabolomics—particularly lipoprotein lipid composition—as promising early, actionable indicators of brain aging trajectories, potentially enabling intervention before overt neurological symptoms emerge.
Key Findings
- 77 plasma metabolites were significantly associated with brain age gap (BAG) in 17,770 UK Biobank adults.
- Lipids in small/medium HDL particles were linked to larger BAG, indicating an older-appearing brain.
- Cholesterol fractions relative to total lipids in VLDL, LDL, and HDL correlated with smaller (younger) BAG.
- Metabolite–brain age associations were consistent across APOE ε4 carriers and non-carriers (all interaction p>0.05).
- Brain age was estimated from 1,079 MRI phenotypes across six imaging modalities using LASSO regression.
Methodology
Cross-sectional metabolite-to-brain-age analysis in 17,770 UK Biobank participants; 249 plasma metabolites measured by NMR at baseline; brain age estimated ~9 years later via LASSO regression on 1,079 MRI phenotypes across six modalities. Linear regression with FDR correction assessed metabolite–BAG associations; APOE ε4 interaction tested via stratified analysis.
Study Limitations
Metabolites were measured at baseline while MRI occurred ~9 years later, precluding firm causal inference; reverse causation cannot be excluded. The cohort is predominantly white British, limiting generalizability. Residual confounding from unmeasured lifestyle and dietary factors remains possible despite extensive covariate adjustment.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: