CSF markers as predictors for Alzheimer's disease in subjects with MCI: Effect of APOE genotype-adjusted cut offs

Article Outline

• Copyright

Background: Decreased Aβ_1-42 and increased levels of tau and phosphorylated tau (ptau) in cerebrospinal fluid (CSF) are key markers for Alzheimer's disease (AD). Previous studies showed that the concentration of these markers correlate with the apolipoprotein E (APOE) genotype. The aim of the present study was to investigate whether correction for APOE genotype could improve the predictive accuracy of CSF markers for AD in subjects with mild cognitive impairment (MCI). Methods: Subjects with MCI for which data were available on CSF markers, APOE genotype, and follow-up were selected from the ADNI study (n = 198), DESCRIPA study (n = 98), and VU University medical center (n = 96). Subjects aged <55 years and with obvious causes for MCI were excluded. Follow-up was performed every 6-12 months up to 3 years. CSF markers were analyzed using xMAP technology (INNO-BIA AlzBio3, ADNI) or ELISA (Innotest Aβ_1-42, hTAU-Ag, Phospho-tau [181P], other cohorts). For each assay the cut-off was determined that maximized the area under the curve (AUC) for the distinction between subjects with and without AD at follow-up. Results: At baseline, subjects (n = 391) had a mean age of 72 years, MMSE score of 26.6, and 13.2 years of education. 204 Subjects (52%) were carriers of at least 1 APOE-e4 allele. Average follow-up was 2.4 years and 146 subjects (37%) converted to AD. At baseline, the APOE-e4 allele was associated lower Aβ_1-42 and higher tau and ptau levels (p < 0.0001). Without correction for APOE genotype, the AUC ranged from 0.63 (ptau) to 0.66 (tau). Compared to the optimal cut-off for subjects without APOE-e4 allele, the cut-off for APOE-e4 allele carriers was lower for Aβ_1-42 (15% to 26%; the lower estimate applies to xMAP, the higher estimate to ELISA) and higher for tau (24% to 48%) and ptau (48% to 54%). When APOE-e4 allele-corrected cut-offs were used, the AUC for the prediction of AD remained the same or decreased compared to cut-offs without correction for APOE genotype. Conclusions: While APOE-e4 allele has a strong effect on Aβ_1-42, tau, and ptau levels in CSF, correction for APOE genotype may not improve the predictive accuracy for AD.