Participants
This study is part of the Korean Brain Aging Study for Early Diagnosis and Prediction of Alzheimer’s Disease (KBASE), an ongoing prospective, community-based cohort study [16]. As of February 2017, a total of 480 older adults consisting of 281 CN and 199 CI (MCI and AD dementia) subjects were initially recruited. The inclusion criteria for the CN group were (a) aged 55–90 years, (b) no diagnosis of MCI or dementia, and (c) Clinical Dementia Rating (CDR) score of 0. For the MCI group, individuals 55–90 years old who fulfilled the core clinical criteria for diagnosis of MCI according to the recommendations of the National Institute on Aging-Alzheimer’s Association (NIA-AA) guidelines [17] were included as follows: (a) memory complaints corroborated by the patient, an informant, or clinician; (b) objective memory impairment for age, education, and gender (i.e., at least 1.0 SD below the respective age, education, and gender-specific mean for at least one of the four episodic memory tests included in the Korean version of Consortium to Establish a Registry for Alzheimer’s Disease (CERAD-K) neuropsychological battery [Word List Memory, Word List Recall, Word List Recognition, and Constructional Recall test]); (c) largely intact functional activities; and (d) no dementia. The global CDR score of all MCI individuals was 0.5. For the AD dementia group, participants 55–90 years old who fulfilled the following inclusion criteria were recruited: (a) criteria for dementia in accordance with the Diagnostic and Statistical Manual of Mental Disorders 4th Edition (DSM-IV-TR), (b) the criteria for probable AD dementia in accordance with the NIA-AA guidelines [18], and (c) a global CDR score of 0·5 or 1. For all groups, individuals with the following conditions were excluded from the study: (1) presence of major psychiatric illness; (2) significant neurological or medical condition or comorbidities that could affect mental function; (3) contraindications to MRI (e.g., pacemaker, claustrophobia); (4) illiteracy; (5) presence of significant visual/hearing difficulty, and severe communication or behavioral problems that would make clinical examination or brain scan difficult; (6) taking an investigational drug; and (7) pregnant or breastfeeding. More detailed information on recruitment of the KBASE cohort was described in our previous report [16]. This study protocol was approved by the Institutional Review Boards of Seoul National University Hospital and SNU-SMG Boramae Medical Center, Seoul, South Korea. The participants and/or their legal representatives provided written informed consent.
Clinical assessment
All participants were administered comprehensive clinical and neuropsychological assessments by trained psychiatrists and neuropsychologists based on the KBASE assessment protocol which incorporates the CERAD-K [16]. Blood samples were collected to determine apolipoprotein E ε4 allele (APOE4) carrier status. Vascular risk factors, including hypertension, diabetes mellitus, hyperlipidemia, coronary artery disease, transient ischemic attack, and stroke, were evaluated via systematic interview by trained nurses, and vascular risk score was calculated for the number of vascular risk factors present and reported as a percentage [16].
Image acquisition, preprocessing, and measurement of vessel stenosis and AD biomarkers
All subjects underwent simultaneous three-dimensional (3D) [11C] Pittsburgh compound B (PiB)-positron emission tomography (PET), 3D T1-weighted MRI, fluid-attenuated inversion (FLAIR) images, and 3D time-of-flight (TOF)-MR angiography using the 3.0-T Biograph mMR (PET-MR) scanner (Siemens, Washington DC, USA). Acquisition parameters for MRI and MR angiography are described in Methods S1 (Additional file 1).
Systematic evaluation of stenosis on MR angiography
Diagnosis of extracranial carotid and intracranial arterial stenosis was reached by the consensus between two qualified neuroradiologists (KMK and CHS) blinded to the participants’ clinical information. We recorded the overall presence, the number, and the degree of detectable stenotic lesions in the following 13 arterial segments: right and left proximal cervical internal carotid artery (ICA); right and left intracranial ICA; right and left anterior, middle, and posterior cerebral arteries; right and left intracranial vertebral artery; and basilar artery. For the extracranial carotid artery, the degree of stenosis was measured according to the North American Symptomatic Carotid Endarterectomy Trial (NASCET) criteria [19] using maximum-intensity projections and source images of the bifurcation of the carotid artery. In cases of intracranial arterial stenosis, the degree of stenosis was calculated based on maximum-intensity projections and source images using the method published for the Warfarin-Aspirin Symptomatic Intracranial Disease Study [20]: percent stenosis = [(1 − (Dstenosis/Dnormal)] × 100. In the case of an artery with multiple stenotic lesions, the most severe degree was selected. Based on the above quantitative data, participants were categorized into stenosis-positive (stenosis+) vs. stenosis-negative (stenosis−) groups according to the stenosis measurements for extracranial carotid and intracranial arteries as follows: (i) overall presence of any detectable stenosis and (ii) severity (i.e., the degree of stenosis ≥ 50%, and number of stenotic arteries ≥ 2). In terms of the location of intracranial arterial stenosis, the presence of detectable stenosis in the anterior circulation and posterior circulations was also evaluated. Anterior circulation stenosis was defined as any detectable stenosis in ICA and anterior or middle cerebral arteries. Posterior circulation stenosis included any detectable stenosis in intracranial vertebral or basilar arteries. As there were only very limited numbers of cases with ≥ 50% stenotic lesions in the extracranial carotid arteries (1 of 281 subjects in the CN group and 1 of 196 subjects in the CI group), and those with bilateral extracranial carotid stenosis (6 of 281 subjects in the CN group and 6 of 196 subjects in the CI group) in our sample, these measurements could not be applied to the extracranial carotid arteries and only available for evaluation of intracranial arterial stenosis.
Beta-amyloid (Aβ) biomarker
For measurement of Aβ biomarker of AD, a 30-min emission scan was obtained 40 min after injection of intravenous administration of 555 MBq of [11C] PiB (range, 450–610 MBq). The [11C] PiB-PET data collected in list mode were processed for routine corrections such as uniformity, UTE-based attenuation, and decay corrections, and were reconstructed into a 344 × 344 image matrix using iterative methods (5 iterations with 21 subsets). The image preprocessing steps were performed using Statistical Parametric Mapping 8 (SPM8; http://www.fil.ion.ucl.ac.uk/spm) implemented in MATLAB 2014a (MathWorks, Natick, MA, USA). Static [11C] PiB-PET images were coregistered to individual T1 structural images, and transformation parameters for spatial normalization of individual T1 images to a standard Montreal Neurological Institute (MNI) template were calculated. The inverse transformation of parameters to transform coordinates from the automatic anatomic labeling (AAL) 116 atlas [21] into an individual space for each subject (resampling voxel size = 1 × 0.98 × 0.98 mm) was performed using IBASPM (Individual Brain Atlases using Statistical Parametric Mapping) software in MATLAB. To extract gray matter (GM) and exclude the non-GM portions of the atlas (i.e., white matter [WM] and cerebrospinal fluid space), a GM mask, which is a binary probabilistic GM map generated by preprocessing step using SPM8, was applied for each individual. The mean regional [11C] PiB uptake values from cerebral regions were extracted using the individual AAL 116 atlas from T1-coregistered [11C] PiB-PET images. Cerebellar GM was used as the reference region for quantitative normalization of cerebral [11C] PiB uptake values, due to its relatively low Aβ deposition [22], with a probabilistic cerebellar atlas (Institute of Cognitive Neuroscience, UCL; Cognitive Neuroscience Laboratory, Royal Holloway, University of London, UK) which was transformed into individual space as described above. The AAL algorithm and a region combining method [23] were applied to determine regions of interest (ROIs) to characterize the [11C] PiB retention levels in the frontal, lateral parietal, posterior cingulate-precuneus, and lateral temporal regions. A global Aβ retention value (standardized uptake value ratio, SUVR) was generated by dividing the voxel-weighted mean value of the four ROIs by the mean cerebellar uptake value [23,24,25]. Aβ positivity was defined if [11C] PiB SUVR value was > 1.4 in at least one of the abovementioned four ROIs [23,24,25].
Neurodegeneration biomarker
All T1-weighted MR images were automatically segmented using FreeSurfer version 5.3 (http://surfer.nmr.mgh.harvard.edu/) with manual correction of minor segmentation errors. As AD-related neurodegeneration biomarkers, both AD-signature cortical thickness (AD-CT; i.e., mean cortical thickness obtained from AD-signature regions) and hippocampal volume adjusted for intracranial volume (HVa) were measured as described previously [24]. First, AD-CT was defined as the mean cortical thickness values of AD-signature regions including the entorhinal, inferior temporal, middle temporal, and fusiform gyrus, based on the Desikan–Killiany atlas [26]. Second, to obtain HVa, left and right hippocampi volume from the FreeSurfer extracted output were first added together to yield the total hippocampal volume (HV). Then, the volume deviating from the expected total HV according to intracranial volume (ICV) based on the reference group (i.e., young CN group of the study cohort [KBASE]), which was not included in the present study, was calculated to obtain HVa as described in the previous study [24, 27]. Briefly, a linear regression line was derived based on the young control group (i.e., the reference group) using their ICV and total HV. HVa, then, is the residual from the linear regression of HV (y-axis) vs. ICV (x-axis); therefore, the HVa is interpreted as the amount of deviation in a subject’s hippocampal volume from what is expected given their ICV.
White matter hyperintensities
The volume of white matter hyperintensities (WMH) on FLAIR images was calculated using a validated automatic procedure [28] with two modifications, as follows: First, an optimal threshold of 70 instead of 65 in the original reference was applied, as the original study [28] recommended adjustment of the threshold for each study dataset to capture the voxels with WMH better than 65 without including non-WMH voxels. More detailed information on the adjustment of the threshold is described in Methods S2 (Additional file 1). Second, diffusion-weighted imaging was not used in the present automated procedure as there were no participants with acute cerebral infarcts in our study population. WMH candidate images were used to extract WMH volumes based on lobar ROIs in the native space for each subject [29].
Statistical analysis
Clinical characteristics were compared between CN and CI, and between MCI and AD dementia using the chi-square and Fisher exact tests to compare data distributions, and independent t test to compare means of continuous variables. Interobserver agreement for stenosis was determined by calculating Cohen’s kappa correlation coefficient from 125 randomly selected individuals.
The association between a measure of extracranial carotid or intracranial stenosis and AD biomarkers (i.e., global Aβ deposition, AD-CT, and HVa) was investigated focusing on CN and CI separately, with two steps of analysis including exploratory and confirmative steps. Exploratory univariable analyses were performed with independent t test to compare the quantitative values of AD biomarkers between the stenosis+ and stenosis− groups. Cohen’s d was determined to imply the effect size of the discrepancy between sequences. The AD biomarkers with p < 0.05 in exploratory univariable analyses were selected for the next confirmatory multivariable analyses. Confirmative multivariable analyses using general linear model (GLM) were conducted for the selected biomarker adjusting for age, sex, and APOE4 carrier status for the CN group, and for age, sex APOE4 carrier status, and clinical diagnosis (MCI or AD dementia) for the CI group. The Bonferroni correction method was applied to multiple comparisons using p < 0.05/no. of confirmatory analyses within each cognitive group. The same analyses with WMH volume as an additional covariate were also conducted to evaluate the mediating effect of WM lesions.
All statistical analyses except calculating Cohen’s d were performed using IBM SPSS Statistics 23 (SPSS Inc., Chicago, IL, USA), and p < 0.05 (two-sided) was taken to indicate statistical significance unless otherwise specified. Cohen’s d was calculated using effsize package in the R 4.0.0.