Associations of plasma lipids with CSF sTREM2 in Alzheimer’s disease
Danni Li, Matthew Glittenberg- Psychiatry and Mental health
- Cellular and Molecular Neuroscience
- Geriatrics and Gerontology
- Neurology (clinical)
- Developmental Neuroscience
- Health Policy
- Epidemiology
Abstract
Background
We hypothesize that plasma lipids interact with brain endothelial cells, which in turn activate microglia. To test our hypothesis, we examined cross‐sectional associations between plasma lipids and cerebrospinal fluid (CSF) soluble triggering receptor expressed on myeloid cells 2 (sTREM2), a biomarker of microglial activation.
Method
We included 357 non‐demented participants (197 males and 160 females) from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) with available plasma lipidomics data (Baker Heart and Diabetes Institute, Metabolomics laboratory), CSF Ab42, CSF sTREM2, amyloid PET (18F‐florbetapir/18F‐florbetaben), demographics and cognitive status. Participants with baseline CSF and PET data obtained beyond a window of 6 months were excluded. Participants were stratified into early‐ (Ab CSF+/PET‐; n = 53) or late‐amyloid beta accumulators (CSF+/PET+; n = 196) based on CSF Ab42 (positivity when CSF Ab42 < 976.6 pg/ ml) and amyloid‐PET status (positivity when 18 F‐florbetapir > 1.11 SUVR or 18Fflorbetaben PET > 1.08 SUVR). Participants classified as Ab CSF‐/PET‐ were included as healthy controls (n = 108); participants as Ab CSF‐/PET+ were excluded. Plasma lipids and CSF sTREM2 were log10 transformed. Multiple linear regression models were used to examine the associations between plasma lipids and CSF sTREM2, adjusting for age, sex, BMI and # of APOE4 alleles. ChemRICH software was used for lipid cluster enrichment analysis and enriched clusters were defined based on FDR p value of < 0.01 and altered ratio (# of lipids altered/# of lipids in a cluster) of 0.1 or greater.
Result
781 plasma lipids were grouped into 37 clusters based on chemical ontologies and structural similarity. We identified 12 plasma lipid clusters enriched in early accumulators (Figure 1) for their associations with CSF sTREM2, only 1 (lysoPC) for late accumulators, none for health controls. All the associations were positive (e.g., higher levels of lipids were associated with higher CSF sTREM2). We further examined the associations between plasma lipids and CSF Ab42 in early accumulators and identify a different set of plasma lipid clusters (i.e., phosphatidylethanolamine and phosphatidylinositol).
Conclusion
Our results suggest strong connections between plasma lipids and brain microglial activation before abnormal brain deposition begins, which were not mediated by decreased CSF Ab42 clearance.