Structural MRI biomarkers for preclinical and mild Alzheimer's disease

Abstract

Noninvasive MRI biomarkers for Alzheimer's disease (AD) may enable earlier clinical diagnosis and the monitoring of therapeutic effectiveness. To assess potential neuroimaging biomarkers, the Alzheimer's Disease Neuroimaging Initiative is following normal controls (NC) and individuals with mild cognitive impairment (MCI) or AD. We applied high-throughput image analyses procedures to these data to demonstrate the feasibility of detecting subtle structural changes in prodromal AD. Raw DICOM scans (139 NC, 175 MCI, and 84 AD) were downloaded for analysis. Volumetric segmentation and cortical surface reconstruction produced continuous cortical surface maps and region-of-interest (ROI) measures. The MCI cohort was subdivided into single- (SMCI) and multiple-domain MCI (MMCI) based on neuropsychological performance. Repeated measures analyses of covariance were used to examine group and hemispheric effects while controlling for age, sex, and, for volumetric measures, intracranial vault. ROI analyses showed group differences for ventricular, temporal, posterior and rostral anterior cingulate, posterior parietal, and frontal regions. SMCI and NC differed within temporal, rostral posterior cingulate, inferior parietal, precuneus, and caudal midfrontal regions. With MMCI and AD, greater differences were evident in these regions and additional frontal and retrosplenial cortices; evidence for non-AD pathology in MMCI also was suggested. Mesial temporal right-dominant asymmetries were evident and did not interact with diagnosis. Our findings demonstrate that high-throughput methods provide numerous measures to detect subtle effects of prodromal AD, suggesting early and later stages of the preclinical state in this cross-sectional sample. These methods will enable a more complete longitudinal characterization and allow us to identify changes that are predictive of conversion to AD.

Figures

Figure 1

Example coronal sections from FreeSurfer volumetric segmentation in an individual's native space to demonstrate regions‐of‐interest including hippocampus (gold), amygdala (sky blue), caudate (blue), putamen (pink), cerebellum (brown), nucleus accumbens (orange), body of the lateral ventricle and temporal horn of the lateral ventricle (purple).

Figure 2

Group differences in average thickness (mm) for left hemisphere. Top row: NC vs. SMCI; middle row: NC vs. MMCI; bottom row: NC vs. AD. Left mesial views, right lateral views. The scale ranges from +0.3 (cyan) mm thickness. Areas on the red‐yellow spectrum indicate regions of thinning with disease: approximate color scale in mm is −0.05 to −0.15 dark red, −0.20 bright red, −0.25 orange, and

Figure 3

Group differences in average thickness (mm) for left hemisphere. Top row: AD vs. SMCI; bottom row: AD vs. MMCI. Left mesial, right lateral views. The scale ranges from +0.3 (cyan) mm thickness. Areas on the red‐yellow spectrum indicate regions of thinning with disease: approximate color scale in mm is −0.05 to −0.15 dark red, −0.20 bright red, −0.25 orange, and

Figure 4

Medial (left) and lateral (right) views of the cortical parcellation scheme. Regions‐of‐interest on the mesial surface include entorhinal cortex (red), four cingulate areas (shades of purple from retrosplenial, rostral posterior, caudal anterior, and rostral anterior), inferior parietal (violet). ROIs on the lateral surface include inferior (purple), middle (brown), and superior (light blue) temporal gyri, caudal (brown) and rostral (blue) midfrontal.

Figure 5

Estimated marginal mean volume (mm3) for hippocampus and thickness (mm) for remaining regions by group for left hemisphere regions‐of‐interest accounting for sex and age effects. Significant group effects were present for all ROIs displayed. For a–e, all groups differed from each other. For medial orbitofrontal, rostral midfrontal, and retrosplenial regions (f–h), NC and SMCI were significantly thicker relative to MMCI and AD; NC and SMCI were not different nor were MMCI and AD. Error bars = std. error of the mean. Solid bar = left hemisphere; outlined open bar = right hemisphere.