CSF Proteomics in Healthy Adults Reveals Early Neurodegeneration Signatures
A landmark study of 6,175 CSF proteins in 994 cognitively normal adults uncovers age, sex, and amyloid-linked clusters foreshadowing Alzheimer's.
Summary
Researchers profiled 6,175 proteins in cerebrospinal fluid from 994 cognitively normal adults aged 43–91, identifying thousands of proteins significantly associated with age, sex, APOE ε4 status, and amyloid burden. Many proteins were co-affected by multiple factors, revealing deep interplay among these AD risk drivers. Network analysis grouped proteins into modules tied to neuropsychiatric and aging-related diseases. CSF signatures were distinct from plasma, underscoring the brain-specific nature of these changes. The findings illuminate the earliest molecular shifts preceding clinical Alzheimer's disease and highlight novel therapeutic targets.
Detailed Summary
Alzheimer's disease (AD) is the leading cause of dementia, with amyloid plaques accumulating silently for decades before symptoms appear. Age, female sex, and carrying the APOE ε4 allele are major risk factors, yet the molecular landscape of the brain during this preclinical window remains poorly understood. Cerebrospinal fluid (CSF), which bathes the brain, offers a more direct window into central nervous system biology than blood, but its invasive collection has historically limited large-scale studies.
This study profiled 6,175 proteins using the SomaScan 7k aptamer platform in CSF from 994 cognitively normal individuals (ages 43–91) drawn from the Knight ADRC cohort (discovery, n=660), with replication in ADNI and FACE cohorts (n=334), and orthogonal validation via mass spectrometry in 101 participants from the Emory Diversity cohort. Associations with age, sex, APOE ε4 carrier status, and CSF amyloid status were tested using linear regression with FDR correction.
In discovery, 4,464 proteins were age-associated (64% increasing, 36% decreasing with age), 1,347 were sex-associated (roughly equal male/female enrichment), 344 were APOE ε4-associated, and 3,212 were amyloid-associated. Replication confirmed 2,172 age-, 711 sex-, 193 APOE ε4-, and 1,807 amyloid-associated proteins, with Pearson correlations between discovery and replication exceeding 0.80 for all four factors. Strikingly, age and amyloid associations were largely in opposite directions (ρ=−0.77), suggesting that proteins rising with age tend to fall with amyloid accumulation and vice versa. Forty-two proteins were co-affected by all four factors, including FOXO1—a transcription factor central to longevity and metabolic regulation—whose levels were strongly reduced in APOE ε4 carriers and individuals with abnormal amyloid.
Weighted gene co-expression network analysis (WGCNA) identified protein modules with distinct biological signatures. Module M2 was associated with age, sex, and amyloid status and was enriched for neuropsychiatric disease pathways. Module M6 was associated with age and sex and linked to general aging-related diseases. Critically, these CSF proteomic signatures were largely distinct from those found in plasma proteomics from 1,382 Knight ADRC participants, confirming that blood cannot substitute for CSF in capturing brain-specific aging biology. The study also constructed CSF- and plasma-based proteomic aging clocks, demonstrating their complementary utility in predicting biological age.
The identification of 2,013 previously unreported age-associated proteins and 1,763 previously unreported amyloid-associated proteins—including TGFB1, NPTX2, NLGN1, and CARTPT—substantially expands the landscape of candidate AD biomarkers and drug targets. While the cross-sectional design and aptamer-based measurement technology introduce some caveats, the multi-cohort replication and orthogonal mass spectrometry validation substantially strengthen confidence in these findings.
Key Findings
- 2,172 age-, 711 sex-, 193 APOE ε4-, and 1,807 amyloid-associated CSF proteins replicated across cohorts.
- Age and amyloid associations were strongly inversely correlated (ρ=−0.77), revealing opposing biological trajectories.
- FOXO1, a longevity-linked transcription factor, was the most significantly APOE ε4-associated protein, also reduced with amyloid accumulation.
- Network modules M2 and M6 linked to neuropsychiatric and aging diseases identified from co-expression clustering.
- CSF proteomic signatures were largely distinct from plasma, highlighting brain-specific molecular aging patterns.
Methodology
CSF from 994 cognitively normal adults was profiled using SomaScan 7k (6,175 proteins) with discovery in Knight ADRC (n=660) and replication in ADNI/FACE (n=334). Orthogonal validation used mass spectrometry in 101 Emory Diversity cohort participants. Linear regression with FDR correction tested associations with age, sex, APOE ε4, and amyloid status.
Study Limitations
The study is cross-sectional, limiting causal inference about temporal protein dynamics during aging and AD progression. Aptamer-based SomaScan measures may have platform-specific biases, though mass spectrometry orthogonal validation partially addressed this. Lumbar puncture sampling may introduce selection bias toward healthier, more research-engaged individuals.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: