Abstract
283
Objectives: Neuroimaging biomarkers of neurodegeneration, including regional brain volumes measured with magnetic resonance imaging (MRI) as well as pathological amyloid (A) and tau (T) measured with positron emission tomography (PET), are part of the research framework for Alzheimer’s disease. However, neuroimaging exams are expensive, and the costs and burden to patients and caregivers from multiple neuroimaging studies in large patient populations are significant. We hypothesized that early frame intervals from amyloid-PET studies with [18F]florbetapir could provide an estimate of cerebral blood flow (eCBF) to predict the presence or absence of pathological tau deposition measured with [18F]flortaucipir-PET.
Methods: 39 participants (8 A-/T-, 8 A+/T-, 23 A+/T+) with dynamic florbetapir-PET and static flortaucipir-PET scans within 6 months of each other were obtained from Washington University in St. Louis (WUSTL) for our training dataset in order to determine the appropriate cutoffs and flow-phase parameters for eCBF. Regional brain segmentation based on volumetric MRI and regional PET quantitation was performed using the BLAzER workflow1 developed by our group to extract regional standardized uptake value ratios (SUVRs), followed by computational analysis in Matlab vR2019b (MathWorks, Natick, MA) to calculate eCBF values. Receiver Operating Characteristic (ROC) analysis was used to determine the effectiveness of eCBF to classify tau-positive subjects relative to a reference standard SUVR cutoff of 1.22.
Results: The strongest within-subject correlations between decreased eCBF and increased cortical flortaucipir binding were observed with the 30-90 second interval after florbetapir injection across 12 individual FreeSurfer-based regions (Figures 1 and 2). This optimum eCBF interval also revealed strong, significant region of interest (ROI) inter-subject Pearson correlations. One-to-one inverse correspondence and ability to stratify tau positivity by ROC was strongest in the bilateral temporal cortex (r = -0.51, p = 0.001, AUC = 0.87) across 10 ROIs (Braak 1 - 6 and frontal, temporal, parietal, cingulate, and frontal cortices). Decreased eCBF in the bilateral temporal cortex also correlated with increased bilateral flortaucipir binding in brain regions relevant to tau staging (Braak 3, r = -0.35, p = 0.03, AUC = 0.80; Braak 4: r = -0.49, p = 0.002, AUC = 0.87; and Braak 5 with basal ganglia: r = -0.36, p = 0.02, AUC = 0.83, Figure 3). Across a weighted average of major cortical areas (temporal, parietal, frontal, and cingulate) for flortaucipir, decreased eCBF in both Braak 2 and the temporal cortex correlated independently with increased global cortical tau (SUVR: r = -0.50, p = 0.001, AUC = 0.83 and SUVR: r = -0.41, p = 0.01, AUC = 0.86, respectively).
Conclusions: Training data suggests that eCBF measures from dynamic florbetapir-PET studies strongly correlate with tau positivity measured with flortaucipir-PET across all subjects irrespective of amyloid status. Successful development of this approach could simplify and decrease the cost of acquiring neuroimaging biomarkers of amyloid and tau. References: Raman F, et al. (2019) Biomarker Localization, Analysis, Visualization, Extraction, and Registration (BLAzER) Methodology for Research and Clinical Brain PET Applications. J Alzheimers Dis 70, 1241-1257. **NOTES: Data used in preparation of this article were obtained from the Charles F. and Joanne Knight Alzheimer’s Disease Research Center (Knight ADRC) at Washington University in St. Louis School of Medicine. A complete listing of Knight ADRC investigators and goals can be found at https://knightadrc.wustl.edu