CSF TOTAL TAU AS A BIOMARKER FOR NEURONAL INJURY IN ALZHEIMER’S DISEASE: ALIGNING RATES OF CSF CHANGE WITH RATES OF HIPPOCAMPAL AND CORTICAL GRAY MATTER ATROPHY

Abstract

Assessment of disease progression via cerebrospinal fluid (CSF) and magnetic resonance imaging (MRI) biomarkers allows for an approximate estimation of disease stage along the Alzheimer's disease (AD) spectrum. While a cross-sectional assessment of these biomarkers provides information regarding the presence/absence of disease, investigation of these biomarkers at the earliest stages of the disease might provide prognostic information about disease progression. Such biomarker investigations are key to understanding their importance earlier in the pathological process. We assessed correlations (Spearman's Rho (r)) between CSF (total-tau (T-tau), phosphorylated-tau (P-tau181P), Aβ1–42 and the T-tau/Aβ1–42 ratio) and MRI biomarkers (hippocampal and gray matter volumes) from 111 Australian Imaging, Biomarkers and Lifestyle (AIBL) participants at first follow up, 36 at second follow up and 11 at third follow up. For those non-demented (12 HC, 1 MCI) participants with both longitudinal CSF and MRI measures, we assessed the rates of change (Linear Mixed Models, (LMM)) for each of these biomarkers over 36 months. Hippocampal volume was most strongly correlated with CSF T-tau at first (r=-0.37, p=0.0002), second (r=-0.70, p<0.0001) and third (r=-0.86, p=0.002) time-points. Cortical gray matter was inversely correlated with T-tau at first (r=-0.26, p=0.007), but not second (r=-0.31, p=0.07) or third time-points (r=-0.25, p=0.47). For those participants who had at least two follow up data-points for both CSF and MRI biomarkers, individual rates of change from the LMM's were strongest for the CSF T-tau/Aβ1–42 ratio (standardized b=-0.49, p=0.001), and for cortical gray matter (standardized b=-0.39, p=0.00002). In an extended model both time and T-tau alone were able to predict cortical grey matter atrophy (T-tau: standardized b=-0.38, p=0.011, time: standardized b=-0.41, p=0.000004). Using data from the well characterized AIBL study, we confirm the relationship between CSF and MRI biomarkers is consistent over time. Furthermore, the rates of change for the strongest CSF biomarker -T-tau/Aβ1–42 ratio- are similar to, albeit a little stronger, than those for the rates of cortical gray matter atrophy, suggesting that CSF measures could be used to predict grey matter atrophy at the pre-symptomatic stage.