Identifying Alzheimer's Disease Biomarkers by Analyzing Fatty Acids in Cerebrospinal Fluid and Urine

Abstract

Although fatty acid oxidation is associated with Alzheimer's disease (AD) pathology, no links have been established between brain lipids and excreted oxidized products. We hypothesize that oxidation of brain polyunsaturated fatty acids (PUFA) can generate dicarboxylic acids (DCA) excreted in urine by AD subjects at different rates compared with cognitively healthy (CH) subjects. Furthermore, this trend can be extended to differentiate CH subjects with normal cerebrospinal fluid (CSF) Aβ42/tau ratios (CH‐NAT) and CH subjects with pathological CSF Aβ42/tau ratios (CH‐PAT).Subjects > 70 years were classified as CH (n=64) or AD (n=26) after neuropsychological assessment. CSF Aβ42 and Tau levels were determined using an ELISA‐based method and a logistic regression of Aβ42/tau ratios used to classify CH‐NAT and CH‐PAT. CSF was fractionated and PUFA levels quantified in supernatant fluid (SF), nanoparticles (NP), and unesterified fatty acid (UFA). DCA was derivatized to pentafluorobenzyl ester, and quantified using gas chromatography and isotope dilution negative ion chemical ionization mass spectrometry. PUFA data is expressed as a percentage of fatty acids detected in CSF fractions while DCA is calculated as a percentage of 7 DCAs (C4–C10) in urine. Links between urinary DCAs and CSF PUFAs were determined using Spearman's ranked correlation.C10 DCA negatively correlated with 6 of 8 PUFAs in SF from CH and also negatively correlated with C20:3n‐3 and C22:6n‐3 in SF from CH‐NAT. In contrast, C9 DCA negatively correlated with 7 of 8 PUFAs in SF from AD. C6 and C8 DCAs positively correlated with SF C18:3n‐3 and C20:3n‐3 in AD, respectively, while C6 DCA negatively correlated with SF C18:2n‐6 from CH‐NAT. C4 DCA negatively correlated with C18:2n‐6, C20:4n‐6, and C22:6n‐3 in SF from CH‐PAT, while C7 DCA positively correlated with C20:4n‐6 in SF from CH‐PAT. In the NP fraction of CH, C4 DCA positively correlated with C20:2n‐6, C20:3n‐3, C20:5n‐3, and C22:5n‐3 PUFAs, while C8 DCA negatively correlated with C20:5n‐3. Only C9 DCA positively correlated with C20:4n‐6 in AD NP fractions. Interestingly, no correlations were found in the NP fraction from CH‐NAT. However, C8 DCA negatively correlated with NP C20:3n‐3, C20:5n‐3, and C22:5n‐3 and C4 DCA positively correlated with NP C20:5n‐3 from CH‐PAT. For UFA of CH, C6 DCA negatively correlated with C20:2n‐6, C20:3n‐3, and C22:6n‐3, while C5 and C7 DCAs positively correlated with homo‐γ‐C20:3n‐6 and C18:2n‐6, respectively. For UFA of CH‐NAT, C4 DCA positively correlated with homo‐γ‐C20:3n‐6, C5 and C8 DCAs negatively correlated with homo‐γ‐C20:3n‐6, and C7 DCA negatively correlated with C18:2n‐6. C5 DCA positively correlated with C18:2n‐6 in UFA from CH‐PAT. C8 DCA positively correlated with unesterified C22:6n‐3, while C9 DCA negatively correlated with unesterified C20:5n‐3 in AD subjects.Differential correlation of urinary DCAs with CSF PUFAs in clinical groups suggests a link with AD pathology. PUFAs and DCAs that segregate CH‐NAT from CH‐PAT are potential biomarkers of early AD.Support or Funding InformationL.K. Whittier and the Helen Posthuma FoundationsThis abstract is from the Experimental Biology 2018 Meeting. There is no full text article associated with this abstract published in The FASEB Journal.