Principal Components Analysis of Brain Metabolism Predicts Development of Alzheimer Dementia.

Abstract

The value of 18F-FDG PET for predicting conversion from mild cognitive impairment (MCI) to Alzheimer dementia (AD) is currently under debate. We used a principal components analysis (PCA) to identify a metabolic AD conversion-related pattern (ADCRP) and investigated the prognostic value of the resulting pattern expression score (PES). Methods:18F-FDG PET scans of 544 MCI patients were obtained from the Alzheimer Disease Neuroimaging Initiative database and analyzed. We implemented voxel-based PCA and standard Statistical Parametric Mapping analysis (as a reference) to disclose cerebral metabolic patterns associated with conversion from MCI to AD. By Cox proportional hazards regression, we examined the prognostic value of candidate predictors. Also, we constructed prognostic models with clinical, imaging, and clinical and imaging variables in combination. Results: PCA revealed an ADCRP that involved regions with relative decreases in metabolism (temporoparietal, frontal, posterior cingulate, and precuneus cortices) and relative increases in metabolism (sensorimotor and occipital cortices, cerebellum, and left putamen). Among the predictor variables age, sex, Functional Activities Questionnaire, Mini-Mental State Examination, apolipoprotein E, PES, and normalized 18F-FDG uptake (regions with significant hypo- and hypermetabolism in patients with conversion vs. those without conversion), PES was the best independent predictor of conversion (hazard ratio, 1.77, per z score increase; 95% CI, 1.24-2.52; P < 0.001). Moreover, adding PES to the model including the clinical variables significantly increased its prognostic value. Conclusion: The ADCRP expression score was a valid predictor of conversion. A combination of clinical variables and PES yielded a higher accuracy than each single tool in predicting conversion from MCI to AD, underlining the incremental utility of 18F-FDG PET.