One-Stop Shop: F-18-Flortaucipir Pet Differentiates Amyloid-Positive And -Negative Forms Of Neurodegenerative Diseases

Tau protein aggregations are a hallmark of amyloid-associated Alzheimer disease and some forms of non-amyloid-associated frontotemporal lobar degeneration. In recent years, several tracers for in vivo tau imaging have been under evaluation. This study investigated the ability of F-18-flortaucipir PET not only to assess tau positivity but also to differentiate between amyloid-positive and-negative forms of neurodegeneration on the basis of different F-18-flortaucipir PET signatures. Methods: The F-18-flortaucipir PET data of 35 patients with amyloid-positive neurodegeneration, 19 patients with amyloidnegative neurodegeneration, and 17 healthy controls were included in a data-driven scaled subprofile model (SSM)/principal-component analysis (PCA) identifying spatial covariance patterns. SSM/PCA pattern expression strengths were tested for their ability to predict amyloid status in a receiver-operating-characteristic analysis and validated with a leave-one-out approach. Results: Pattern expression strengths predicted amyloid status with a sensitivity of 0.94 and a specificity of 0.83. A support vector machine classification based on pattern expression strengths in 2 different SSM/PCA components yielded a prediction accuracy of 98%. Anatomically, prediction performance was driven by parietooccipital gray matter in amyloid-positive patients versus predominant white matter binding in amyloid-negative patients. Conclusion: SSM/PCA-derived binding patterns of F-18-flortaucipir differentiate between amyloid-positive and-negative neurodegenerative diseases with high accuracy. F-18-flortaucipir PET alone may convey additional information equivalent to that from amyloid PET. Together with a perfusion-weighted early-phase acquisition (F-18-FDG PET-equivalent), a single scan potentially contains comprehensive information on amyloid (A), tau (T), and neurodegeneration (N) status as required by recent biomarker classification algorithms (A/T/N).