Amyloid β influences the relationship between cortical thickness and vascular load

Abstract Introduction Cortical thickness has been proposed as a biomarker of Alzheimer's disease (AD)– related neurodegeneration, but the nature of its relationship with amyloid beta (Aβ) deposition and white matter hyperintensity volume (WMHV) in cognitively normal adults is unclear. Methods We investigated the influences of Aβ status (negative/positive) and WMHV on cortical thickness in 408 cognitively normal adults aged 69.2 to 71.9 years who underwent 18F‐Florbetapir positron emission tomography (PET) and structural magnetic resonance imaging (MRI). Two previously defined Alzheimer's disease (AD) cortical signature regions and the major cortical lobes were selected as regions of interest (ROIs) for cortical thickness. Results Higher WMHV, but not Aβ status, predicted lower cortical thickness across all participants, in all ROIs. Conversely, when Aβ‐positive participants were considered alone, higher WMHV predicted higher cortical thickness in a temporal AD‐signature region. Discussion WMHV may differentially influence cortical thickness depending on the presence or absence of Aβ, potentially reflecting different pathological mechanisms.


INTRODUCTION
Cortical thickness has been proposed as a biomarker of neurodegeneration across the pathophysiological continuum of Alzheimer's disease (AD). [1][2][3] It is vital to understand the pathological factors that may influence such biomarkers, particularly in asymptomatic individuals, who represent an increasingly important target population for potential disease-modifying therapies 4 and where objective biomarkers are likely to play key roles in the interpretation of such therapeutic trials in the future. 5 Although the relationship between amyloid beta (A ) deposition 6-18 -one of the earliest hypothesized changes in the AD-continuum 19 -and cortical thickness, as well as the relationship between white matter hyperintensity volume (WMHV) 20-24largely considered to be a surrogate marker of cerebral small vessel vasculopathy 25 -and cortical thickness, have been investigated in cognitively normal adults in isolation, there has been limited research looking at the relative influences of A deposition and WMHV on cortical thickness concurrently.
We report a cross-sectional analysis of 408 cognitively normal individuals (ages 69.2 to 71.9 years) all born in mainland Britain in the same week of 1946 who underwent clinical assessment, 18 F-Florbetapir positron emission tomography (PET) and volumetric structural magnetic resonance imaging (MRI). The aim of this analysis was to investigate the hypothesis that A deposition and WMHV predict lower cortical thickness and to investigate potential interactions in their effects.

Participants
Data were analyzed from individuals who participated in a neuro- Participants were assessed for a history of cognitive impairment, and major neurological or psychiatric illness. If a participant reported memory or cognitive difficulties they perceived as more than others the same age, or if they felt they would seek medical attention regarding cognitive difficulties, they were defined as having subjective memory complaints. A collateral cognitive history was acquired using the an eight-item informant interview validated to differentiate aging and dementia (AD8) questionnaire 29 and informant cognitive concern was defined as an AD8 score ≥2. Exclusions from the analysis were the following: failure to complete the PET/MRI scan (n = 31); PET acquisition failure (n = 8); WMHV segmentation quality control failure (n = 4); cortical thickness imaging quality control failure (n = 3); and participants with evidence of dementia, MCI, or major neurological (including clinical history or radiological evidence of a cortical stroke) or psychiatric disorder (n = 48).   Table 2). 28 In unadjusted analyses, there was no evidence of an association at the 5% level between WMHV and sex, childhood cognitive ability, educational attainment, MMSE, logical memory delayed recall, matrix reasoning, 12-item Face-Name test score, APOE genotype, diabetes, hypercholesterolemia, or smoking ( Table 2). There was some evidence of a positive association between greater age at time of scan and WMHV (P = 0.013), a negative association between digit-symbol substitution score and WMHV (P = 0.03), and that individuals with a history of hypertension had higher WMHV (P = 0.004).

Cortical thickness and A deposition
There was no evidence that A status or SUVR predicted cortical thickness in any cortical ROI investigated following adjustment for age at time of scan, sex, and WMHV (Table 3). To explore whether adjustment for WMHV had removed a statistically significant association between A cortical thickness, the same analysis without adjustment for WMHV was performed. However, this made no material difference to the results obtained (data not shown).  with increased cortical thickness in the ADsig Mayo ROI (P = 0.027) ( Table 5).

ROIs
In a post hoc analysis we also explored whether the patterns observed were related to APOE genotype by adding APOE 4 carrier status as a covariate, as well an interaction term between APOE

DISCUSSION
In this study of 408 age-matched cognitively individuals, we found no evidence, even at a trend level, for cross-sectional associations between A deposition and cortical thickness in any of the ROIs investigated. Although some studies have shown evidence of associations between A deposition and cortical thickness in cognitively normal individuals, 7,[11][12][13]35 several studies have also reported null findings. [14][15][16][17] Because AD is associated with cortical thinning, the lack of association in this cohort may reflect the age of the participants Although we found no relationship between A deposition and cortical thickness, higher WMHV strongly predicted lower cortical thickness across all participants. This is consistent with previous studies, which have provided evidence for cross-sectional associations between increasing WMHV and gray matter atrophy [20][21][22][23][24] ; this study confirms and extends these findings, allowing for associations with A deposition to be explored.
We found evidence for an interaction between A status and WMHV in terms of their effects on cortical thickness. When Apositive participants were considered alone, there was some evidence that higher WMHV predicted higher cortical thickness in the ADsig Mayo ROI. Although the statistical effect in A -positive participants was relatively weak (P = 0.027), the interaction effect between A status and WMHV across all participants was strong (P < 0.001 in temporal ROIs). Although paradoxical increases in cortical gray matter structural metrics with A deposition have been reported in cognitively normal individuals, [8][9][10]18 to our knowledge, a positive association between WMHV and cortical thickness has not been reported previously in individuals with asymptomatic A deposition. However, a similar interaction in the setting of clinically established AD consistent with our findings, also localized to the temporal lobe, has been reported. 42 The mechanism(s) leading to positive associations between WMHV and cortical thickness are unclear. Elevated WMHV may reflect pathological processes other than conventional cerebrovascular disease in the context of AD. 43,44 In a large cohort of non-demented older adults, Graff-Radford et al. found white matter hyperintensities to be correlated with A load, with evidence implicating cerebral amyloid angiopathy rather than small vessel vasculopathy. 45 It is notable that in our study hypertension was strongly related to WMHV in the A -negative population, consistent with this being due to small vessel vasculopathy, whereas there was no evidence of an association in the A -positive sample, noting differences in sample size between A -negative and Apositive groups. Another possibility for this unexpected relationship is neuroinflammation. Recent work has demonstrated positive associations between PET radiotracer uptake of markers of microglial activation and gray matter volumes in the setting of MCI. 46 It is also possible that this is reflective of a selection effect, whereby those with a thicker cortex are more likely to be recruited to an intensive neuroimaging study and be defined as cognitively normal in the context of a higher WMHV load and A positivity. Longitudinal imaging and neuropsychology data will be important to further explore the consequence of this observed effect, whereas ultimately large-scale postmortem data will be vital to investigate further the pathological heterogeneity underlying white matter hyperintensities in the context of A deposition.
This study has a number of strengths and weaknesses. Because all subjects were of near identical age-separated only by the ≈2 to 3 years required to complete scanning-the effects of age on any relationships are limited. We do not have a measure of neurofibrillary tangle deposition such as tau PET, which is more closely linked to neurodegeneration and cognitive decline than A 47,48 ; recent work has shown that tau PET tracer uptake is strongly correlated with lower cortical thickness and may mediate relationships between A and cor-tical atrophy. 49 However, the study by Graff-Radford and colleagues demonstrated no relationship between increased tau and white matter hyperintensity burden, 45 suggesting that it is unlikely to play a role in the interactions between A and WMHV in terms of their effects on cortical thickness in our data set. We considered A and WMHV on a global scale, and future work looking at regional differences in pathological biomarkers and cortical thickness will be of interest, as will work looking at relevant white matter pathways using diffusion MRI. 50 Although the broader NSHD is highly representative of individuals born in mainland Britain in 1946, 26 intensive neuroimaging studies such as Insight 46 are at risk of introducing an element of recruitment bias; we have previously shown that Insight 46 participants are on average slightly more educated, as well as more likely to being defined as having a non-manual occupation. 51 However, compared to many other intensive data-rich neuroimaging studies, which rely on convenience sampling methodology, 52 this is likely to remain a fairly representative samples, noting that all NSHD participants are white, which is reflective of the British population in 1946, but not the richer ethnic diversity present in the modern United Kingdom.
In conclusion, we found that WMHV is a stronger predictor of cortical thickness than A deposition in normal 70-year-olds. This, and the demonstration that WMHV may be differentially associated with cortical thickness according to A status, has implications for the use of cortical thickness as a biomarker of AD-related neurodegeneration in the preclinical phase and suggests that WMHV may reflect different pathological mechanisms in the presence of A accumulation.

ACKNOWLEDGMENTS
We are very grateful to those study members who helped in the design of the study through focus groups, and to the participants both for their contributions to Insight 46 and for their commitments to research over the last 7 decades. We are grateful to the radiographers and nuclear medicine physicians at the UCL Institute of Nuclear Medicine, and to the staff at the Leonard Wolfson Experimental Neurology Centre at UCL. We are particularly indebted to the support of the late Dr Chris Clark of Avid Radiopharmaceuticals who championed this study from its outset.

FUNDING AND DISCLOSURES
This study is principally funded by grants from Alzheimer's Research