Cortical tau load is associated with white matter hyperintensities

Abstract

Introduction: Cerebral white matter lesions (WML), visualized as white matter hyperintensities (WMH) on T2-weighted MRI, encompass structural damage and loss of integrity of the cerebral white matter (WM) and are commonly assumed to be associated with small vessel disease (SVD). However, it has been suggested that WM damage may also be the result of degenerative axonal loss that is secondary to cortical Alzheimer's disease (AD) pathologies i.e., hyperphosphorylated tau (HPτ) and amyloid-beta (Aβ). Here we investigate the influence of HPτ, Aβ and SVD on WMH severity.

Results: 36 human post-mortem right fixed cerebral hemispheres (mean age 84.4 ± 7.7 years; male: 16, female: 20) containing varying amounts of AD-pathology (AD: 23, controls: 13) underwent T2- weighted MRI with WMH assessed according to the age related white matter change scale (ARWMC). After dissection, using tissue samples from the frontal, temporal, parietal and occipital regions from the right hemisphere, we quantitatively assessed cortical HPτ and Aβ pathology burden by measuring the percentage area covered by AT8 immunoreactivity (HPτ-IR) and 4G8 immunoreactivity (Aβ-IR), and assessed the severity of WM SVD by calculating the sclerotic index (SI) of WM arteries/arterioles. HPτ-IR, Aβ-IR, and SI were compared with ARWMC scores. HPτ-IR, Aβ-IR and WM ARWMC scores were all significantly higher in AD cases compared to controls, while SI values were similar between groups. ARWMC scores correlated with HPτ-IR, Aβ-IR and SI in various regions, however, linear regression revealed that only HPτ-IR was a significant independent predictor of ARWMC scores.

Conclusions: Here we have shown that increasing cortical HPτ burden independently predicted the severity of WMH indicating its potentially important role in the pathogenesis of WM damage. Moreover, our findings suggest that in AD patients the presence of WMH may indicate cortical AD-associated pathology rather than SVD. Further studies are warranted to elucidate the pathological processes that lead to WM damage and to clarify if WMH may serve as a general biomarker for cortical AD-associated pathology.

Fig. 1

Cartoon illustrating the 24 locations used for image analysis encompassing sulci, midsection, and gyri from prefrontal (a), mid frontal (b), entorhinal (c), temporal (d), parietal (e) and occipital (f) cortices. White dots indicate the sampling location; numbers refer to Brodman areas, which are marked with coloured patterns (modified from [46])

Fig. 2

Total cortical percentage area covered by HPτ-IR (AT8 antibody for hyperphosphorylated tau burden) and 4G8 immunoreactivity (Aβ-IR; for amyloid-β burden), total white matter sclerotic index (WM SI; measure of small vessel disease severity), and total age related white matter change scale (ARWMC; measure for white matter hyperintensities) scores in Alzheimer’s disease (AD) i.e., cognitively impaired, and controls. a, total HPτ-IR is significantly higher in AD as compared to controls. (ai), hyperphosphorylated tau in the form of neurofibrillary tangles and neuropil threads is immunopositive for AT8 antibody (neurofibrillary tangle - arrow, neuropil thread - arrow head) and (aii) the area covered by AT8 immunopositivity (shaded in red) was measured to calculate HPτ-IR. b, total Aβ-IR was significantly higher in AD as compared to controls. (bi), Aβ plaques/depositions are immunopositive for 4G8 antibody and (bii) the area covered by 4G8 immunopositivity (shaded in red; arrow indicates physiological APP that is not included in the analysis) was measured to calculate Aβ-IR. c, No differences were seen in the severity of small vessel disease as measured by sclerotic index (SI) between controls and AD. (ci), small white matter artery, histologically stained with H&E. (cii) example for calculating the SI: using VasCalc software [61] the internal (Dint) and external (Dext) diameters were measured three times to yield a SI value (for details see main text). d, Total age related white matter change (ARWMC) scores [60] were significantly higher in AD compared to controls. Examples of post mortem T2 MRI: (di), no WMH (ARWMC score 0) and (dii) confluent white matter hyper intensity in the parieto-occipital region (arrow head; ARWMC score 3). MR images were captured in sagittal plane. *,p < 0.05; ***,p < 0.001; scale bar, 20 μm, valid for ai and bi, 50 μm for ci

Fig. 3

Partial Spearman’s (ρ') (HPτ-IR and Aβ-IR) and Pearson’s (r') (WM SI) correlation coefficients controlling for the effect of age at death. Scatter graphs show correlations between cortical percentage area covered by HPτ-IR (AT8 antibody for hyperphosphorylated tau burden), Aβ-IR (4G8; for amyloid-β depositions/plaques) and total white matter sclerotic index (WM SI; measure of small vessel disease severity) with age related white matter change score (ARWMC; measure for white matter hyperintensities) in the frontal (a-c), temporal (d-f), parietal (g-i) and occipital (j-l) regions, as well as total values representing the entire hemisphere (m-o). Analysis revealed that in all regions, and the entire hemisphere, significant correlations were observed between ARWMC scores and cortical HPτ-IR and Aβ-IR values, respectively. A significant relationship between ARWMC scores and WM SI were observed only in the occipital lobe. Of note, a correlation was observed between total SI and total ARWMC scores but this is likely due to the correlation in the occipital region. Only significant p values (and associated correlation coefficients (ρ’, r’)) are shown; for all correlation please see main text

Fig. 4

Correlations between total white matter sclerotic index (a) (WM SI; measure of small vessel disease severity) as well as total cortical percentage area covered by HPτ-IR (b) (AT8 antibody for hyperphosphorylated tau burden) and 4G8 (c) immunoreactivity (Aβ-IR; for amyloid-β depositions/plaques) with age related white matter change score (ARWMC; measure for white matter hyperintensities) in cases with minimal hyperphosphorylated tau pathology (classified as Braak NFT stage 0-II [6]). Only total WM SI (a) correlated with total ARWMC score; no correlation was seen between total cortical HPτ-IR and Aβ-IR with total ARWMC scores. Only significant p values (and associated correlation coefficients (ρ)) are shown