Biomarkers offer the potential for an unbiased, objective picture of underlying disease pathology. Classical AD biomarkers such as cerebrospinal fluid (CSF) beta-amyloid (Aβ1–42) and total-tau (t-tau) are considered suitable endpoints in clinical trials of preventative treatments, providing a clear picture of the development of AD pathology in the clinically silent stage (pre-clinical) of the disease. We performed a correlation analysis on a selection of biochemical (CSF Aβ1–42, and CSF t-tau), neurophysiological (FDG-PET), volumetric (hippocampal volume), and cognitive markers, using the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset, at baseline and longitudinally. We describe 4 biological (biochemical) phenotypes (CSF Aβ1–42 >= 192 pg/mL and CSF t-tau <= 93 pg/mL as normal) at baseline, and compare these with clinically derived diagnostic groupings. Per capita rate of changes in CSF Aβ and t-tau are quantified, and population estimates, in addition to individual trajectories, are assessed in relation to their biological phenotype and change in cognitive diagnosis. A strong, inverse correlation was found between CSF Aβ1–42 and CSF t-tau (r = -0.43), as well as CSF Aβ1–42 and ADAScog-11 (r = -0.38). A ‘phase’ plot of the concentrations of Aβ1–42 and t-tau, show a highly non-linear “right angle” relationship. We observe strong but imperfect concordance between biological phenotypes and cognitive diagnosis, 63% of AD-diagnoses individuals present with an ‘AD typical’ biological phenotype while 49.8% of cognitively normal (CN) individuals present with a ‘CN typical’ phenotype at baseline. A small per capita annual rate of change is observed in CSF Aβ1–42 and CSF t-tau concentrations relative to variance. The greatest per capita rate of change in Aβ occurs in CN individuals who do not progress to MCI/AD across follow-up (CN-NP), mean rate of -2.4 pg/mL/year, and those with progressive MCI (MCI-P) accumulate the largest amount of t-tau, mean rate of 2.2 pg/mL/year. Our findings support the ordering of biomarker changes suggested by the amyloid cascade hypothesis. The high variance in CSF Aβ and t-tau concentrations, within subjects longitudinally, impairs quantification and visualisation of individual biomarker trajectories. The phase plots of Aβ1–42 versus t-tau suggest a highly non-linear dynamic between the two variables.
Read full abstract