NEUROIMAGING BIOMARKERS OF THE A/T/N CLASSIFICATION SYSTEM PREDICT CROSS-SECTIONAL AND LONGITUDINAL CHANGE IN PRECLINICAL AND EARLY ALZHEIMER DISEASE

Abstract

Assessment of β-amyloid (A), tau (T) and neurodegeneration (N) biomarkers has recently been proposed by Jack and colleagues (2016, Neurology) to stage Alzheimer disease (AD). In isolation, biomarkers for A/T/N have shown associations with cognitive changes in the preclinical and early symptomatic stages of AD. However, the relative contribution of each A/T/N imaging biomarker to cross-sectional cognitive impairment and longitudinal decline has not been well-characterized. The current study assesses unique variance in cross-sectional and longitudinal cognition explained by each neuroimaging biomarker. Cognitive data were available on 104 individuals from an average of 5 assessments over an average 6-year period. All participants had a [18F]AV-1451 (T807, flortaucipir) tau PET scan during follow-up and both a [18F]-AV-45 b-amyloid scan and hippocampal volume from a structural MRI within one year from the tau scan. Unique variance explained (measured with R2) by cross-sectional imaging biomarkers in both cross-sectional and longitudinal performance on the Free and Cued Selective Reminding Test (fcSRT) was calculated from linear regression models that included age, education, gender and Clinical Dementia Rating as covariates. Continuous variables were standardized prior to analysis. Cross-sectionally, tau PET significantly predicted a relatively large portion of the variance in fcSRT performance (ΔR2 = .13, beta = -1.38, p < .001). As expected, this relationship was much stronger in individuals at high risk for AD (Amyloid +) relative to low risk participants (Amyloid -). Neither amyloid (A) nor hippocampal volume (N) predicted cognition (ΔR2 < .01). Longitudinally, tau was again a significant predictor of decline (beta = -.08, ΔR2 = .03, p = .001) whereas both amyloid and hippocampal volume were not (ΔR2 < .01). Similar relationships between tau PET and cognitive decline were found in semantic memory (category fluency) and executive function (Trail Making B).