[18f]av-1451 pet imaging of tau pathology predicts white matter hyperintensities in healthy aging, mild cognitive impairment and alzheimer's disease: the nimrod study

White matter hyperintensities (WMH) are commonly associated with normal aging and Alzheimer's disease (AD). However, the aetiology of these lesions is still unclear as both vascular factors and AD pathology have been implicated. Whereas previous evidence has come from histopathological studies, we aim to elucidate the in vivo associations between tau pathology and WMH. 42 participants (12 controls, 19 MCI, 11 AD) underwent structural T1-MPRAGE, T2-weighted volumetric FLAIR and dynamic [18F]AV-1451 PET imaging as part of the NIMROD study. T1 data were processed in Freesurfer v6-beta and parcellated into cortical regions of interest (ROIs) (Desikan-Killiany atlas) (Fischl et al., 2002). [18F]AV-1451 binding was quantified at voxel level using a simplified referenced tissue model with cerebellum as reference region. Parametric [18F]AV-1451 images were co-registered to T1 and cortical ROIs reverse transformed into PET native space, from which mean ROI binding potential values were extracted and summed to estimate lobar tau deposition. We used the Lesion Prediction Algorithm implemented in the Lesion Segmentation Toolbox (Schmidt et al., 2012) (SPM12) to segment WMH and an in-house script to quantify total and lobar WMH (Firbank et al., 2003). We investigated the associations between lobar tau deposition and WMH with partial correlations, correcting for age, gender and estimated total intracranial volume, followed by stepwise linear regression models, including lobar [18F]AV-1451 binding and age, across the total sample. AD patients had significant increased parietal, occipital and temporal [18F]AV-1451 binding compared to controls and MCI (p<0.001) and frontal [18F]AV-1451 binding compared to controls (p<0.05). Parietal and occipital WMH were significantly higher in AD compared to controls (p<0.05). Across the total sample, global [18F]AV-1451 binding correlated with total WMH (r=0.36,p=0.027). Regional partial correlations between [18F]AV-1451 binding and WMH were significant for the parietal (r=0.37,p=0.023), temporal (r=0.45,p=0.005) and occipital (r=0.43,p=0.008) lobes. Stepwise linear regression indicated that increased parietal [18F]AV-1451 binding predicted parietal WMH independent of age (F(2,41)=13.05,p<0.001,R2=0.40). We demonstrate the association between tau pathology and WMH in vivo, particularly in the parietal lobe. These findings provide support to post-mortem studies suggesting a role of tau pathology in the pathogenesis of WMH (McAleese et al., 2015). Illustration of segmentation of white matter hyperintensities (WMH) from the volumetric T2-weighted FLAIR using the Lesion Segmentation Toolbox (Schmidt et al., 2012) (SPM12) and an in-house developed lobar mask (Firbank et al., 2003). Group comparisons of [18F]AV-1451 binding potential (BP) within the lobes and total tau (average of the lobes). Blue = healthy controls (HC), green = mild cognitive impairment (MCI) patients, red = Alzheimer's disease (AD) patients. Bars indicate mean [18F]AV-1451 binding ± standard error. * p < 0.05 with ANCOVA correcting for age and gender. Group comparisons of white matter hyperintensities (WMH) within the lobes (left) and total WMH (right). Blue = healthy controls, green = mild cognitive impairment (MCI) patients, red = Alzheimer's disease (AD) patients. Bars indicate mean WMH volumes ± standard error. * p < 0.05 with ANCOVA correcting for age, gender and estimated total intracranial volume. Partial correlations between white matter hyperintensities (WMH) log volumes and [18F]- AV1451 binding correcting for age, gender and estimated total intracranial volume. Blue = healthy controls, green = mild cognitive impairment patients, red = Alzheimer's disease patients.