Serial MRI and CSF biomarkers in normal aging, MCI, and AD

Abstract

Objective: To compare the annual change in MRI and CSF biomarkers in cognitively normal (CN), amnestic mild cognitive impairment (aMCI), and Alzheimer disease (AD). Comparisons were based on intergroup discrimination, correlation with concurrent cognitive/functional changes, relationships to APOE genotype, and sample sizes for clinical trials.

Methods: We used data from the Alzheimer's Disease Neuroimaging Initiative study consisting of CN, aMCI, and AD cohorts with both baseline and 12-month follow-up CSF and MRI. The annual change in CSF (total-tau [t-tau], Abeta(1-42)) and MRI (change in ventricular volume) was obtained in 312 subjects (92 CN, 149 aMCI, 71 AD).

Results: There was no significant average annual change in either CSF biomarker in any clinical group except t-tau in CN; moreover, the annual change did not differ by clinical group in pairwise comparisons. In contrast, annual increase in ventricular volume increased in the following order, AD > aMCI > CN, and differences were significant between all clinical groups in pairwise comparisons. Ventricular volume increase correlated with concurrent worsening on cognitive/functional indices in aMCI and AD whereas evidence of a similar correlation with change in CSF measures was unclear. The annual changes in MRI differed by APOE epsilon4 status overall and among aMCI while annual changes in CSF biomarkers did not. Estimated sample sizes for clinical trials are notably less for MRI than the CSF or clinical measures.

Conclusions: Unlike the CSF biomarkers evaluated, changes in serial structural MRI are correlated with concurrent change on general cognitive and functional indices in impaired subjects, track with clinical disease stage, and are influenced by APOE genotype.

Figures

Figure 1 Box plots of annual change in ventricular volume (cm3), Aβ1–42 (pg/mL), and t-tau (pg/mL) by clinical diagnosis Individual points have been randomly shifted along the x-axis to reduce overlap. Boxes represent the 25th, 50th, and 75th percentiles of the data. Whiskers represent the range of the non-outlier data estimated using Tukey method. The horizontal line indicates the reference of annual change of zero. p Value indicates the results from the test of whether the annual change was different from zero. AD = Alzheimer disease; aMCI = amnestic mild cognitive impairment; CN = cognitively normal; t-tau = total-tau.

Figure 2 Area under the receiver operating characteristic curve (AUROC) with 95% confidence intervals for (A) between-group discrimination and (B) between-APOE genotype discrimination within clinical groups based on annual change in ventricular volume, Aβ1-42, and t-tau A reference line has been drawn at 0.50, which indicates no discriminative power for the measure. p Values along the right side of each panel are from Wilcoxon signed-rank tests. AD = Alzheimer disease; aMCI = amnestic mild cognitive impairment; CN = cognitively normal; t-tau = total-tau.