O5‐04‐06: Multicenter stability, diagnostic accuracy and regional specificity of cholinergic forebrain atrophy in Alzheimer's disease: An EDSD study

Severe neurofibrillary degeneration and loss of basal forebrain (BF) cholinergic neurons is a major pathological hallmark of Alzheimer's disease (AD). Post mortem autopsy studies suggest a regional specificity of neuronal degeneration within the subnuclei of the cholinergic BF in AD. Previous studies using in-vivo MRI have confirmed significant atrophy of the BF nuclei in AD. Here, we studied the multicenter stability and diagnostic accuracy of MRI derived BF atrophy in a large multicenter data set. Additionally, we investigated regional specificity of the pattern of atrophy within different cholinergic nuclei using a newly created cytoarchitectonic map based on combined post mortem in-situ MRI and histology. The European DTI Study in Dementia (EDSD) includes high-resolution MRI and DTI data of 143 healthy controls (HC) and 137 patients with the diagnosis of Alzheimer's disease (AD) dementia. Nine scanners in eight European centres have been involved in this study. In-vivo volumetry of the BF was performed using a novel cytoarchitectonic map and both ROI-based and voxel-wise analyses of high-dimensionally normalized MRI scans. For comparison we also investigated hippocampus volume using ROI-based volume extraction. BF cholinergic nuclei (Ch2 to Ch4p according to Mesulam) were mapped into MRI standard space by projecting the locations of these nuclei from histological sections of a post-mortem brain to corresponding MRI sections in the space followed by a non-linear transformation to the MNI standard template. Patients with AD had statistically significant volumetric reductions in areas mapping to the BF subregions Ch2, Ch3, Ch4al, Ch4am and Ch4p. These effects remained significant after controlling for age, gender and scanner. ROC analysis showed areas under the curve (AUC), as a measure of diagnostic accuracy, ranging from 0.55 (anterior commissure - juxtacommissural cells) to 0.90 (Ch4p region). AUC for hippocampus volume was 0.87. MRI volumetry of the BF shows multicenter stability of BF cholinergic system atrophy in AD. Consistent with post mortem data, our in vivo analysis suggests a selective involvement of cholinergic nuclei in AD. Atrophy of the most affected cholinergic compartment (Ch4p) showed high diagnostic accuracy, which was comparable to the diagnostic potential of hippocampus atrophy.