Subcortical Shape Changes, Hippocampal Atrophy and Cortical Thinning in Future Alzheimer's Disease Patients

Abstract

Efficacy of future treatments depends on biomarkers identifying patients with mild cognitive impairment at highest risk for transitioning to Alzheimer's disease. Here, we applied recently developed analysis techniques to investigate cross-sectional differences in subcortical shape and volume alterations in patients with stable mild cognitive impairment (MCI) (n = 23, age range 59-82, 47.8% female), future converters at baseline (n = 10, age range 66-84, 90% female) and at time of conversion (age range 68-87) compared to group-wise age and gender matched healthy control subjects (n = 23, age range 61-81, 47.8% female; n = 10, age range 66-82, 80% female; n = 10, age range 68-82, 70% female). Additionally, we studied cortical thinning and global and local measures of hippocampal atrophy as known key imaging markers for Alzheimer's disease. Apart from bilateral striatal volume reductions, no morphometric alterations were found in cognitively stable patients. In contrast, we identified shape alterations in striatal and thalamic regions in future converters at baseline and at time of conversion. These shape alterations were paralleled by Alzheimer's disease like patterns of left hemispheric morphometric changes (cortical thinning in medial temporal regions, hippocampal total and subfield atrophy) in future converters at baseline with progression to similar right hemispheric alterations at time of conversion. Additionally, receiver operating characteristic curve analysis indicated that subcortical shape alterations may outperform hippocampal volume in identifying future converters at baseline. These results further confirm the key role of early cortical thinning and hippocampal atrophy in the early detection of Alzheimer's disease. But first and foremost, and by distinguishing future converters but not patients with stable cognitive abilities from cognitively normal subjects, our results support the value of early subcortical shape alterations and reduced hippocampal subfield volumes as potential markers for the early detection of Alzheimer's disease.

Keywords: Alzheimer's disease; cortical thickness; hippocampal subfields; mild cognitive impairment; subcortical shape analysis.

Figure 1

Surface labels for automated segmentation of thalamic subnuclei (L = left hemisphere; R = right hemisphere), based on expert neuroanatomical labeling of serial histology (Chakravarty et al., 2006). Reprinted from Journal of Alzheimer's Disease 49(1) Leh SE, Kälin AM, Schroeder C, Park MT, Chakravarty MM, Freund P, Gietl AF, Riese F, Kollias S, Hock C, Michels L “Volumetric and shape analysis of the thalamus and striatum in amnestic mild cognitive impairment” 237–249, Copyright 2015, with permission from IOS Press.

Figure 2

Coronal views of the hippocampus and hippocampal subfields in magnetic resonance images from a healthy control subject (HC) and a mild cognitive impairment converter subject at time of conversion (MCI-CC) of the present study.

Figure 3

Cortical thickness differences in patients with stable mild cognitive impairment (MCI-S), future converters at baseline (MCI-CB) and converters at time of conversion (MCI-CC) when compared with healthy control subjects (HC). Images were generated after including age and gender in the model, and after correction using FDR at q = 0.15 to better illustrate the anatomical localization. Bar shows FDR-values, with blue /light blue indicating reduced cortical thickness.

Figure 4

Differences in thalamic (A) and striatal (B) shape alterations in future converters at baseline (MCI-CB) and at time of conversion (MCI-CC) when compared with healthy control subjects (HC). Images were generated after including age and gender in the model, and after correction using FDR at q = 0.15 to better illustrate the anatomical localization. Bars show FDR-values, with blue/light blue indicating inward displacements (contractions) and pink/light pink indicating outward displacements (expansions).

Figure 5

Receiver operating characteristic curve analyses of contractions in the left striatum (area under the curve [AUC] = 0.96, p < 0.001) and the right thalamus (AUC = 0.98, p < 0.001) for discriminating future converters at baseline from healthy control subjects. ————- ROC curvereference lines.