P2–158: Combining morphometric measurements of the basal forebrain cholinerigc system and the cortical thickness for better diagnostic accuracy

Cortical thickness and volumetric measurements of the basal forebrain cholinergic system (BFCS) show highly significant changes in patients with Alzheimer's disease (AD) compared to healthy controls (HC). The combination of volumetry of BFCS subnuclei and regional measurement of cortical thickness will allow (i) determining the corticotopy of BFCS nuclei using in vivo imaging, and (ii) combining diagnostic accuracy from both key markers of structural changes in AD. We used 280 high-resolution 3-D structural MRI datasets from the European DTI Study on Dementia (EDSD) including 137 patients with AD following the NINDS-ADRDA criteria and 143 HC. Nine scanners in eight European countries have been involved. BFCS volumes were derived by atlas-based image segmentation using high-dimensional image normalisation and a newly created subregion specific atlas of the BFCS based on post-mortem MRI in-cranio. Cortical thickness was analysed using Freesurfer software. We focused our analysis on cholinergic projections known from post-mortem data such as Ch4p to transverse temporal gyrus (TTG) and Ch2 to hippocampus. Ch4p volume showed significant correlation with thickness of the TTG only in HC; volume of the Ch2 region correlated with thickness of the parahippocampal gyrus (PG) in HC and AD groups. Cortical thickness of precuneus and PG showed an accuracy of 81.7 % for separating AD and HC. The combination with volumetry of the BFCS contributed significantly to diagnostic accuracy in the logistic regression model (85.3 %). In contrast, hippocampus volume, the best established structural imaging marker, did not add to the diagnostic accuracy of cortical thickness measures (80.7 %). Our data support the notion of a corticotopic organization of BFCS subnuclei as proposed by post mortem findings in non human primates. BFCS volume and cortical thickness measurements carry complementary diagnostic accuracy that can help to increase diagnostic accuracy of each single marker alone. Diagnostic utility remained largely unaffected by variability of scan parameters across multiple sites. Presently, an analysis is ongoing on the use of these markers to predict AD in mild cognitive impairment.