P1‐252: Characteristic Metabolic Brain Pattern in Slovenian Alzheimer's Disease Patients with Low CSF Beta‐Amyloid

Neuroimaging techniques offer various biomarkers to support clinical diagnosis of Alzheimer’s disease (AD) and to differentiate AD patients from normal controls (NC). Specific metabolic brain network associated with Alzheimer Disease – Alzheimer’s Disease Related Pattern (ADRP) has previously been identified (Habeck 2008, Teune 2014) as one of the imaging biomarkers. Our objective was to identify and validate this pattern in a Slovenian cohort of AD patients (ADRP-Slovenia) recruited on the basis of decreased ß amyloid concentration in their cerebrospinal fluid (CSF) – an analysis not done previously. 18F-flourodeoxyglucose (FDG) PET brain images were acquired using Siemens Biograph mCT in 13 AD patients (70 ± 10 yrs old, 9 female, ß amyloid concentration 454 ± 71 pg/ml and MMSE score 21 ± 5) and 13 NC (65 ± 7 yrs old, 10 female, ß amyloid concentration 1190 ± 151 pg/ml and MMSE score 28 ± 1). To identify ADRP-Slovenia network, Scaled Subprofile Model/Principle Component Analysis (SSM/PCA) was performed using customized neuroimaging software scanvp (Moeller 1991, Spetsieris 2011). Correlation between ADRP-Slovenia expression for each individual subject (subject score) and corresponding Mini-Mental State Examination (MMSE) score was assessed by computing Pearson correlation coefficient. Receiver operating characteristic (ROC) curve was plotted, to evaluate the diagnostic power of ADRP-Slovenia expression for discrimination of AD patients from NC. ADRP-Slovenia metabolic brain network was characterized with decreased metabolic activity in the posterior parietal and lateral temporal cortical regions associated with hypermetabolism in primary cortices and cerebellum (Fig 1). ADRP-Slovenia expression was significantly higher in AD patients compared to NC (p<0.001, Fig. 2) and correlated significantly with corresponding MMSE score (p<0.0001, Fig 3). ROC curve (AUC 0.994, 95% confidence interval 0.975 - 1.013) confirmed significant discrimination between AD and NC subjects and revealed an optimum subject score cut-off value 1.4 (sensitivity 92%, specificity 92%) (Fig. 4). ADRP-Slovenia was identified using FDG PET brain imaging and SSM/PCA analysis in AD patients with decreased CSF ß amyloid concentration and NC. Its expression accurately discriminates between AD patients and NC, and correlates with MMSE score. We have demonstrated ADRP as a promising functional imaging biomarker of Alzheimer’s disease. ADRP-Slovenia: Alzheimer's Disease Related Pattern identified by network analysis of FDG PET scans of AD patients and NC. Relative metabolic decreases are colour coded blue to green (the posterior parietal and lateral temporal cortical regions), whereas associated metabolic increases are colour coded red to yellow (primary cortices and cerebellum). ADRP-Slovenia expression (Subject score) in AD and NC subjects. Correlation between subject scores of ADRP-Slovenia expression and corresponding MMSE scores for AD patients and NC. ROC curve. The point corresponding to optimal cut-off value is marked by asterix on the curve.