Ante mortem cerebrospinal fluid tau levels correlate with postmortem tau pathology in frontotemporal lobar degeneration

Abstract

Objective: To test the hypotheses that (1) antemortem cerebrospinal fluid (CSF) tau levels correlate with postmortem tau pathology in frontotemporal lobar degeneration (FTLD) and (2) tauopathy patients have higher phosphorylated-tau levels compared to transactivation response element DNA-binding protein 43 (TDP-43) proteinopathy patients while accounting for Alzheimer's disease (AD) copathology.

Methods: Patients had autopsy-confirmed FTLD with tauopathy (n = 31), TDP-43 proteinopathy (n = 49), or AD (n = 26) with antemortem CSF. CSF tau levels were compared between groups and correlated with digital histology measurement of postmortem tau pathology averaged from three cerebral regions (angular gyrus, mid-frontal cortex, and anterior cingulate gyrus). Multivariate linear regression tested the association of ante mortem CSF tau levels with postmortem tau pathology adjusting for demographics.

Results: Multivariate regression found an independent association of ante mortem CSF phosphorylated tau levels with postmortem cerebral tau pathology in FTLD (Beta = 1.3; 95% confidence interval = 0.2-2.4; p < 0.02). After excluding patients with coincident AD-associated tau pathology accompanying sporadic FTLD, we found lower CSF phosphorylated tau levels in the TDP-43 group (median = 7.4pg/ml; interquartile range [IQR] = 6.0, 12.3; n = 26) compared to the tauopathy group (median = 12.5pg/ml; IQR = 10.7, 15.0; n = 23; Z = 2.6; p < 0.01).

Interpretation: CSF phosphorylated-tau levels are positively associated with cerebral tau burden in FTLD. In vivo detection of AD copathology in sporadic FTLD patients may help stratify clinical cohorts with pure neuropathology in which low CSF phosphorylated-tau levels may have diagnostic utility to distinguish TDP-43 proteinopathy from tauopathy. Autopsy-confirmed samples are critical for FTLD biomarker development and validation. Ann Neurol 2017;82:247-258.

Figure 1. Digital image analysis of cerebral tau pathology

Representative photomicrographs depict raw images (top) and digital image analysis thresholding quantification (%AO; bottom- red/orange/yellow overlay) of tau pathology in the mid-frontal cortex for A–B) AD, C–D) FTLD-Tau and E–F) FTLD-TDP with coincident AD pathology. There is higher overall and GM burden of cerebral tau pathology in AD compared to FTLD-Tau and FTLD-TDP, while FTLD-Tau has higher white-matter tau pathology compared to both groups.

Figure 2. Grey and white matter cerebral tau pathology in FTLD-Tau, FTLD-TDP and AD

Boxplots depict values of %AO of tau immunohistochemistry from grey and white matter in each region sampled for each neuropathological group. *=p<0.05, #=p<0.01, ‡≤0.001.

Figure 3. Correlation of ante mortem CSF phosphorylated tau measurements with postmortem cerebral tau severity

Scatterplot depicts individual patient data points coded for by primary pathology (Blue= FTLD-TDP, Orange= FTLD-Tau, Red= AD), the presence of hereditary mutations (open circles) and AD tau co-pathology (large circles= AD tau Braak B2/B3) for natural log transformed CSF p-tau levels (y-axis) compared to natural-log transformed total cerebral tau %AO pathology measurement (x-axis) in A) the total FTLD and AD cohort (r=0.5, p<0.01) and B) the total FTLD cohort (r=0.3, p=0.02). and C) pure sporadic FTLD cohort (r=0.4, p=0.02) excluding patients with AD co-pathology or a hereditary mutation.

Figure 4. CSF p-tau levels in FTLD neuropathological groups with and without coincident AD neuropathology or hereditary disease

Scatter plots depict individual data points in each group coded for the primary pathology (Blue= FTLD-TDP, Orange= FTLD-Tau, Red= AD), presence AD tau co-pathology (large circles= AD tau Braak B2/B3) and hereditary mutations (open circles) for CSF p-tau values (pg/ml) in A) the total FTLD and AD cohort B) pure FTLD patients (excluding those cases with coincident AD-associated tau pathology (n=14) and C) pure sporadic FTLD additionally excluding those pure FTLD patients with a hereditary mutation (n=15) and data points coded for age younger than 65 years (triangles). Bars represent median and interquartile range values from box-plot. *=p<0.02, **p<0.01.

Figure 5. Diagnostic accuracy to distinguish FTLD-TDP in sporadic patients without Alzheimer’s disease co-pathology

Receiver operating characteristic curve depicts the diagnostic accuracy of CSF p-tau levels for sporadic FTLD-TDP patients with pure pathology (i.e. AD tau Braak B0/B1) (n=26) compared to the combined AD and pure sporadic FTLD-Tau group (n=49; red solid line) (AUC=0.85 95% CI=0.76–0.94 p<0.001) and pure sporadic FTLD-Tau group alone (n=23; blue solid line) (AUC=0.72 95% CI=0.57–0.87 p<0.01). Sub-analysis of pure sporadic FTLD cases younger than 65 years old finds increased diagnostic accuracy using CSF p-tau to differentiate FTLD-TDP from FTLD-Tau (AUC=0.86, 95%CI=0.71–1.0. p=0.01) (blue dashed line).