Print  Close

Brain Connectivity Measured Through Spatial Covariance as a Biomarker for Neurodegeneration in AD

Poster No:

252 

Submission Type:

Abstract Submission 

Authors:

Débora Peretti1, Cecilia Boccalini1, Max Scheffler2, Cristelle Rodriguez2,1, Marie Montandon2,1, Sven Haller1,3,4,5, Panteleimon Giannakopoulos2, Giovanni B. Frisoni2,1, Valentina Garibotto2,1

Institutions:

1University of Geneva, Geneva, Switzerland, 2Geneva University Hospitals, Geneva, Switzerland, 3Centre d'Imagerie Médicale de Cornavin, Geneva, Switzerland, 4Uppsala University, Uppsala, Sweden, 5Capital Medical University, Beijing, China

First Author:

Débora Peretti  
University of Geneva
Geneva, Switzerland

Co-Author(s):

Cecilia Boccalini  
University of Geneva
Geneva, Switzerland
Max Scheffler  
Geneva University Hospitals
Geneva, Switzerland
Cristelle Rodriguez  
Geneva University Hospitals|University of Geneva
Geneva, Switzerland|Geneva, Switzerland
Marie Montandon  
Geneva University Hospitals|University of Geneva
Geneva, Switzerland|Geneva, Switzerland
Sven Haller  
University of Geneva|Centre d'Imagerie Médicale de Cornavin|Uppsala University|Capital Medical University
Geneva, Switzerland|Geneva, Switzerland|Uppsala, Sweden|Beijing, China
Panteleimon Giannakopoulos  
Geneva University Hospitals
Geneva, Switzerland
Giovanni B. Frisoni  
Geneva University Hospitals|University of Geneva
Geneva, Switzerland|Geneva, Switzerland
Valentina Garibotto  
Geneva University Hospitals|University of Geneva
Geneva, Switzerland|Geneva, Switzerland

Introduction:

Scaled Subprofile Modelling using Principal Component Analysis (SSM/PCA) is a voxel-based technique that uses spatial covariance maps to identify disease patterns (DP) that best differentiate between two groups of subjects. This technique has been applied to 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) scans to classify Alzheimer's disease (AD) patients. AD is a neurodegenerative disorder characterised by a hypometabolic pattern that affects temporal and parietal lobes, and, in advanced cases, the frontal lobe. A specific region of interest (ROI) combining angular gyrus, posterior cingulate, and temporal lobe has been shown to measure specific AD hypometabolism. Furthermore, a distinctive pattern of atrophy, observed in the temporal lobe on magnetic resonance imaging (MRI) scans, can also be used for AD identification. Atrophy and FDG-PET ROI uptake are established biomarkers of neurodegeneration in AD. The aim of this study was to compare these methods with a connectivity metric measured through SSM/PCA to assess the AD neurodegeneration pattern in a memory clinic patient cohort.

Methods:

A cohort of 333 subjects from the memory clinic of the Geneva University Hospitals underwent FDG-PET, T1-MRI, and neuropsychological assessment within one year. Cognitive stage varied between cognitively unimpaired (CU), mild cognitive impairment (MCI), dementia, and psychiatric disorders (other). AD was diagnosed based on clinical assessment and available biomarkers. PET images were registered to a stereotactic space using subjects' respective MRI. Three approaches to measure neurodegeneration were used: SSM/PCA, SUVR uptake in the Landau meta-ROI, and AD cortical thickness signature. SSM/PCA was applied to a subset of 15 CU and 15 AD-dementia subjects to generate an AD-specific DP (ADDP). ADDP was validated using bootstrapping and leave-one-out cross validation. Remaining subjects were tested against the generated ADDP to retrieve pattern expression. FDG-PET images were then converted to standardised uptake value ratios (SUVR, with vermis and pons as reference) and regional uptake in the Landau meta-ROI was extracted. Finally, an AD cortical thickness signature was extracted from MRI scans using Freesurfer. For each approach, a receiver operating characteristic (ROC) curve was generated to estimate an AD-dementia threshold and calculate its performance for identifying AD-dementia patients using an independent set of CU and AD-dementia subjects. Spearman correlations between neurodegeneration methods and baseline mini-mental state examination (MMSE) score and MMSE annual rate of change were estimated in the whole group and in a subset of 165 subjects who underwent a follow-up neuropsychological assessment, respectively.

Results:

Mean (SD) age of participants was 72 (7) years, and 57% of subjects were female. The threshold for AD classification for SSM/PCA scores was of -763, with an area under the curve (AUC) of 0.96, a sensitivity of 0.9, and specificity of 0.95. The meta-ROI had an SUVR threshold of 1.4 (AUC=0.9, sensitivity=0.85, specificity=0.86), and the cortical thickness threshold was 2.6 (AUC=0.89, sensitivity=0.78, specificity=0.86). All methods were significantly correlated to baseline MMSE scores (SSM/PCA=-0.43, meta-ROI=0.38, cortical thickness=0.37, p<0.01). Patients diagnosed with dementia (regardless of aetiology) showed a significant MMSE annual rate of change (average=-3.8±3.8 points/year) compared to other groups. SSM/PCA scores showed the highest correlation with MMSE annual rate of change (-0.52, p<0.01), followed by the meta-ROI (0.46, p<0.01), and cortical thickness (0.36, p<0.01).
Supporting Image: Figure1.png
Supporting Image: Figure2.png
 

Conclusions:

SSM/PCA ADDP expression provides a strong and specific marker for AD neurodegeneration, outperforming more conventional metrics. As SSM/PCA is a connectivity method that can be applied on a single-subject basis, it has a potential to be applied in clinical practice as a characterisation and prognostic biomarker.

Disorders of the Nervous System:

Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 1

Modeling and Analysis Methods:

Connectivity (eg. functional, effective, structural) 2
Multivariate Approaches
PET Modeling and Analysis

Keywords:

Cognition
MRI
Neurological
Positron Emission Tomography (PET)
Statistical Methods
STRUCTURAL MRI

1|2Indicates the priority used for review

Provide references using author date format

Landau, S. M., Harvey, D., Madison, C. M., Koeppe, R. A., Reiman, E. M., Foster, N. L., Weiner, M. W., & Jagust, W. J. (2011). 'Associations between cognitive, functional, and FDG-PET measures of decline in AD and MCI'. Neurobiology of Aging, 32(7), 1207–1218.
Peretti, D. E., Renken, R. J., Reesink, F. E., de Jong, B. M., de Deyn, P. P., Dierckx, R. A. J. O., Doorduin, J., Boellaard, R., & Vállez García, D. (2021). 'Feasibility of pharmacokinetic parametric PET images in scaled subprofile modelling using principal component analysis'. NeuroImage: Clinical, 30, 102625.
Mattsson-Carlgren, N., Leuzy, A., Janelidze, S., Palmqvist, S., Stomrud, E., Strandberg, O., Smith, R., & Hansson, O. (2020). The implications of different approaches to define AT(N) in Alzheimer disease. Neurology, 94(21), e2233–e2244.
Meles, S. K., Pagani, M., Arnaldi, D., de Carli, F., Dessi, B., Morbelli, S., Sambuceti, G., Jonsson, C., Leenders, K. L., & Nobili, F. (2017). The Alzheimer’s disease metabolic brain pattern in mild cognitive impairment. Journal of Cerebral Blood Flow and Metabolism, 37(12), 3643–3648.