P1‐416: Use of an Alzheimer's Disease‐Related Hypometabolic Convergence Index to Predict Progression from Mild Cognitive Impairment to Alzheimer's Dementia

Using fluorodeoxyglucose positron emission positron emission tomography (FDG PET) data from the Alzheimer's Disease Neuroimaging Initiative (ADNI), we introduce the concept of an AD-related “hypometabolic convergence index (HCI),” a single voxel-based index which, in comparison with data from normal control (NCs), reflects the extent to which the pattern and magnitude of hypometabolism in an individual subject converges with the pattern and magnitude of hypometabolism in probable AD patients. After using a cross-validation procedure to identify the threshold for each biomarker, cognitive, or clinical measurement to distinguish between MCI converters and non-converters, we characterize and compare the ability of abnormal HCIs, other magnetic resonance imaging (MRI) and cerebrospinal fluid (CSF) measurements, memory scores and clinical ratings in predicting 18-month progression from mild cognitive impairment (MCI) to Alzheimer's dementia. The Cox proportional hazards model was used to characterize and compare the prediction ability of “abnormal” HCIs, hippocampal volumes, CSF Aß1-42, t-tau, p-tau181p levels and ratios, auditory verbal learning test total and long-term memory (LTM) scores, and three different clinical ratings to predict time-to-progress to Alzheimer's dementia. Abnormally high HCI's and small hippocampal volumes were associated with the highest odds of 18-month progression from MCI to Alzheimer's dementia (OR = 7.38 and 6.35, respectively), more likely to progress to AD than those who were normal. ORs for the other biomarker, cognitive and clinical measurements were between 1.33 and 4.94. MCI patients with both an abnormally high HCI and abnormally small hippocampal volume had an even higher odds ratio of clinical progression (OR = 36.72), and each of these measurements were correlated with cognitive and clinical measurements of disease severity in the overall group of probable AD, MCI and NC subjects. While additional studies are needed, the HCI offers promise in automatically characterizing the AD-related pattern of hypometabolism in FDG PET images in a single measurement, predicting progression from MCI to Alzheimer's dementia alone or in combination with hippocampal volume measurements, and providing an indicator of disease severity in different clinical and research settings. Among other things, it raises the possibility of generating Alzheimer's disease-related convergence indices using imaging modalities and voxel-based data analysis algorithms.