Identification of concordant plasma lipid signatures in Alzheimer’s disease: Validation between two independent studies of Alzheimer’s disease

Document Type

Journal Article

Publication Title

Alzheimer's & Dementia, Biomarkers - Part 2




School of Medical and Health Sciences


Huynh, K., Lim, W. L. F., Giles, C., Jayawardana, K. S., Salim, A., Mellett, N. A., … Meikle, P. J. (2020). Identification of concordant plasma lipid signatures in Alzheimer’s disease: Validation between two independent studies of Alzheimer’s disease. Alzheimer's & Dementia, Biomarkers - Part 2, 16(S5 Supplement), article e042275. https://doi.org/10.1002/alz.042275


Background: Changes to lipid metabolism are tightly coupled to the onset and pathology of Alzheimer’s disease (AD). Lipidomics has allowed for unprecedented characterisation of the lipidome within biological systems. To explore the lipid dysregulation associated with AD, we utilised our expanded lipidomics platform, examining 569 lipid species across 32 lipid classes, and applied it to the Australian Imaging, Biomarker & Lifestyle Study of Ageing (AIBL) study and the Alzheimer's Disease Neuroimaging Initiative (ADNI) study. Method: We examined over 4000 samples between the AIBL and ADNI studies. Baseline samples comprises of 1112 and 804 from the AIBL and ADNI respectively. Univariate associations between lipids and AD were examined using logistic or linear regressions accounting for variables including anthropometric measures and APOE genotype. Cox regression analysis was used to identify lipids associated with future onset of AD from baseline samples. We utilised a fixed effect model to meta analyse the results, identifying lipids which are concordantly associated between the two studies. Result: Exploratory analysis of the data identified strong associations between plasma lipids and AD risk factors such as age (increasing acylcarnitine species), sex (sphingoid base specific) and APOE genotype (ether lipids). Combined meta‐analysis of the two cohorts identified lipid class and subclass‐specific associations with incident and established AD. In the fully adjusted analysis (anthropometric measures, APOE genotype, clinical lipids and medication), a total of 218 species were associated with established AD and 71 with incident AD after correction for FDR. Sphingomyelin, ceramide, ganglioside, plasmalogen and other ether lipids were concordantly associated with AD and its risk factors in both studies. In particular, ether lipids were consistently associated with both established and future AD independent of their subclass (plasmalogen vs non‐plasmalogen) highlighting impairment of peroxisomal function at the earliest stages of AD. Whole lipid class associations were seen with phosphatidylethanolamine and triglyceride, despite adjustments for clinical lipids, implicating dysregulation in liver lipid metabolism. Conclusion: By leveraging off two independent studies, we were able to highlight concordant associations between lipids and both established and incident AD. The significantly broader coverage of the lipidome from our platform provides new hypotheses and directions for future research.



Access Rights