Effect of carotid artery stenosis on cortical microinfarcts, white matter integrity, and brain volume: An interhemispheric comparison within the population-based Rotterdam Study

Highlights

• •Carotid artery stenosis (CAS) contributes to the occurrence of cortical microinfarcts.
• •CAS is associated with lower downstream parenchymal tissue volume.
• •Findings do not support effects of CAS on white matter microstructural integrity.
• •Future studies should assess relevance to cognitive and functional impairment.

Abstract

Background

Carotid artery stenosis could contribute to gradual loss of brain function through chronic hypoxia and ischemia.

Methods

We included consecutive participants of the population-based Rotterdam Study with unilateral ≥50 % stenosis at the carotid artery bifurcation on time-of-flight carotid MR angiography, and compared between hemispheres the presence of ischemic lesions, tissue volumes, and white matter integrity on structural brain MRI.

Results

Among 50 participants (mean age 76 years, 50 % women), flow was lower in the affected carotid artery than on the unaffected side (160mL/min versus 202mL/min; flow reduction [95 %CI] per 1 % increase in stenosis: 1.7 mL/min [1.0–2.5]). Twelve individuals had radiographic evidence of cortical infarction, of whom 8 had cortical microinfarcts, all on the side of the stenosis (P = 0.001). Downstream of the stenotic artery, parenchymal volume was lower than in the contralateral hemisphere (mean difference: -2.7 mL [-4.9;-0.4]), similar for grey and white matter. Differences were most profound in the frontoparietal lobes, and increased with severity of stenosis to roughly 5 mL in individuals with ≥70 % stenosis. White matter hyperintensity volume and microstructural integrity did not differ between hemispheres.

Conclusions

Carotid artery stenosis is associated with downstream presence of cortical microinfarcts as well as lower parenchymal tissue volume.

Introduction

Extracranial stenosis of the internal carotid artery (ICA) is an important risk factor for stroke, and surgical or endovascular intervention in the early phase after TIA and stroke is a key part of secondary prevention to reduce stroke recurrence risk [1,2]. The majority of ICA stenosis, however, is asymptomatic, with prevalence of asymptomatic stenosis ranging from 2.4 % after age 50 in the general population [3], to 7.2 % in patients with TIA or stroke [4]. Similar to patients with symptomatic stenosis, people with asymptomatic ICA stenosis are at increased risk of ischemic stroke [5], but optimal treatment is undetermined as immediate procedural risks can outweigh long-term reductions in stroke risk [6,7]. The balance might be tilted towards early intervention if consequences of stenosis on brain health extend beyond stroke risk, including for example cognitive and functional consequences of covert brain infarction, cerebral small vessel disease, and neurodegeneration. Indeed, reduced cerebral blood flow is associated with risk of neurovascular and neurodegenerative brain changes [8,9], rendering ICA stenosis a potential new target for preventive interventions against stroke and dementia.

Few studies have investigated asymptomatic stenosis of the ICA in relation to subclinical cerebral small vessel disease. These generally cross-sectional studies suggest a higher burden of lacunes [10], perivascular spaces [11], and cortical microinfarcts [12], but not white matter hyperintensities [13,14], in people with stenosis compared to those without. Moreover, severe CAS was associated with accelerated brain volume loss [15], which might translate into worse cognitive function and higher prevalence of dementia [16,17]. Yet, studies comparing patients with stenosis to reference subjects remain susceptible to (residual) confounding, as ICA stenosis shares many risk factors with vascular and neurodegenerative brain injury that invariably contribute via pathways other than stenotic disease. This confounding may be overcome by the comparison of brain regions downstream of a unilateral haemodynamically significant stenosis, to the equivalent regions in the other, unexposed hemisphere in the same individual. In a prior meta-analysis of 5 studies that used this approach, mixed analytical approaches and results precluded a conclusion on the link between CAS and white matter disease [18,19]. Another single study suggesting higher prevalence of cortical microinfarcts on the side of stenosis awaits further replication [20], while no published studies have yet applied this method to compare brain tissue volumes between hemispheres.

We therefore aimed to determine interhemispheric differences in markers of vascular brain injury and neurodegeneration in individuals with unilateral carotid artery bifurcation stenosis, nested within a prospective population-based cohort.

Methods

Study population

This study is embedded within the Rotterdam Study, a large population-based cohort study in the Netherlands [21]. The study population consists of 14,926 individuals aged ≥45 years from the Ommoord area, a suburb of Rotterdam, who undergo extensive routine examinations at four-yearly visits to a dedicated research facility. At each centre visit, participants undergo structural brain magnetic resonance imaging (MRI), and are screened for subclinical atherosclerosis by ultrasound measurement of the carotid artery intima-media thickness [22]. From October 2007 onwards, all individuals with an intima-media thickness >2.0 mm in either carotid artery were invited for time-of-flight magnetic resonance angiography (MRA) to assess carotid artery bifurcation stenosis [23]. Of invitees, 1982 (74 %) underwent MRA. Internal carotid artery (ICA) stenosis at the bifurcation was present in 238 (12 %) individuals, of whom 110 had no stenosis in the contralateral ICA. Of these 110 individuals, 53 with structural brain MRI within 6 months of arterial imaging were considered eligible for the current study. For a post-hoc analysis of brain asymmetry in the healthy general population, we selected a random sample of 1000 participants without carotid artery stenosis from the overall Rotterdam Study population undergoing brain MRI between 2005–2016. The Rotterdam Study has been approved by the medical ethics committee according to the Population Study Act Rotterdam Study, executed by the Ministry of Health, Welfare and Sports of the Netherlands. Written informed consent was obtained from all participants.

Carotid artery MRI

Magnetic resonance imaging of the carotid arteries was performed on a 1.5T MR scanner (General Electric Healthcare, Milwaukee, WI, USA) with a bilateral phased-array surface coil (Machnet, Eelde, The Netherlands) [23]. The position of the participants was stabilised in a custom-designed head coil to reduce motion artefacts. High-resolution images were obtained using a standardized protocol. First, both carotid bifurcations were identified by means of two-dimensional time-of-flight MRA. Next, high-resolution MRI sequences captured the carotid bifurcations on both sides, as described in detail previously [23]. All images were assessed for the presence of stenosis by a trained physician under the supervision of an experienced neuroradiologist, blinded to the clinical status, and the severity of stenosis was graded according to NASCET criteria (Fig. 1)[24].

Fig. 1.

Visualisation of carotid imaging and cortical microinfarcts.

Legend: Exemplary images of one participant, showing carotid artery stenosis on time-of-flight MRI (A) and corresponding brain MRI sequences (T1 and PD) with the presence of a cortical microinfarct (B).

Brain MRI scan protocol

MRI of the brain was performed on a 1.5T scanner (General Electric Healthcare, Milwaukee, WI, USA), using an 8-channel head coil [25]. We acquired high-resolution axial T1-weighted sequence, proton-density-weighted (PD) sequence, and fluid attenuated inversion recovery (FLAIR) sequence. For flow measurement, 2D phase-contrast imaging was performed as described previously [26]. In brief, a sagittal 2D phase-contrast angiographic scout image was performed. On this scout image, a transverse imaging plane perpendicular to both the pre-cavernous portion of the internal carotid arteries and the middle part of the basilar artery was chosen for a 2D gradient-echo phase-contrast sequence (repetition time = 20 ms, echo time = 4 ms, field of view = 19 cm2, matrix=256 × 160, flip angle = 8°, number of excitations = 8, bandwidth = 22.73 kHz, velocity encoding = 120 cm/sec, slice thickness = 5 mm). Acquisition time was 51 seconds and no cardiac gating was performed. For diffusion-MRI, we performed a single shot, diffusion-weighted spin echo-planar imaging sequence. Maximum b-value was 1000 s/mm2 in 25 non-collinear directions; three volumes were acquired without diffusion weighting (b-value = 0 s/mm2) [27].

Brain asymmetry on MRI was calculated for each brain lobe separately in a random sample of 1000 participants without significant carotid artery stenosis using the brain asymmetry index [28]. The asymmetry index is a measure of grey matter asymmetry, calculated by voxel-wise comparison of grey matter content on the original and flipped image: $=\frac{originalimage-flippedimage}{0.5*\left(originalimage-flippedimage\right)}$ . We depicted index values for the left hemisphere, and a positive value consequently indicates more grey matter in the left hemisphere (leftward asymmetry), whereas a negative value indicates rightward asymmetry.

Assessment of cerebral blood flow, brain volumes, white matter injury, and cortical infarcts

Flow was calculated from the phase-contrast images using interactive data language-based custom software (Cinetool version 4; General Electric Healthcare). Two independent, experienced technicians manually drew the regions of interest (i.e. surface area of the left and right carotid artery, and basilar artery) in a plane perpendicular to the arteries at the level of the carotid clinoid segment, and performed subsequent flow measurements per vessel (mL/min; interrater correlations >0.94 for all vessels) [26]. For the assessment of brain volume, the structural MR sequences (T1-weighted, PD-weighted, and FLAIR) were transferred to a Linux workstation. Pre-processing steps and the classification algorithm have been described previously [29]. Quantification of cerebrospinal fluid, total parenchymal volume, and white matter hyperintensity (WHM) volume was done using an automated tissue segmentation method, based on a k-nearest-neighbour brain tissue classifier algorithm [29]. All segmentation results were visually inspected and if needed manually corrected by investigators who were blinded to the carotid artery stenosis measurements. Parenchymal brain volume was calculated by adding up grey and white matter volumes, converted to millilitres. For secondary analyses, total parenchymal tissue volumes were also calculated using segmentations acquired with FreeSurfer version 6.0 (http://surfer.nmr.mgh.harvard.edu/). All diffusion data were pre-processed using a standardized pipeline [27], fitting diffusion tensors per voxel to compute mean fractional anisotropy (FA) and mean diffusivity (MD) in the normal-appearing white matter through combination with the tissue segmentation. In general, lower FA and higher MD values are considered indicative of lower microstructural integrity. All scans were furthermore inspected by trained research physicians, blinded to stenosis site, for the presence of cortical infarction. In accordance with rating standards, microinfarcts were defined as strictly intracortical lesions of less than 4 m in dimension that were hyperintense on T2-weighted MRI, hypointense on T1-weighted MRI, and isointense on blood-sensitive scans [30,31].

Other measurements

Medical history and information on cardiovascular risk factors was obtained during study centre visits on the basis of measurement assessed by interview, with verification in medical records if appropriate [21].

Statistical analysis

Among all eligible participants, we first determined to which extent flow was reduced downstream of the ICA stenosis, compared to the contralateral side using a paired t-test, and subsequently assessed whether degree of stenosis in the stenotic (≥50 %) artery was related to blood flow using linear regression.

We compared prevalence of any cortical infarction and microinfarcts specifically across hemispheres, using McNemar’s χ² test for paired data. We then performed a paired t-test to determine interhemispheric differences in total parenchymal volume in relation to the presence of ICA stenosis, before and after exclusion of participants with radiographic evidence of cortical infarction, or a clinical history of TIA or stroke. For reliability of tissue segmentations, and generalisability to individuals without (covert) brain infarcts, we excluded individuals with cortical macro-infarcts from the subsequent analyses. We tested for interhemispheric differences, segregated by grey and white matter, and stratified results by brain region. Given the arterial territory supplied by the ICA, we expected the largest interhemispheric differences in the frontoparietal lobes. We repeated these analyses for increasing severity of stenosis (i.e. 50–59 %, 60–69 %, and ≥70 %), and for increasing reduction in downstream blood flow (i.e. flow through the affected ICA as a percentage of flow through the unaffected ICA). We furthermore assessed whether associations differed by systemic mean arterial pressure, age at time of brain imaging, and side of the stenosis. For robustness of results and comparison across studies, parenchymal tissue comparisons were repeated using FreeSurfer segmentations. We further performed an interhemispheric comparison of WMH volume, mean diffusivity and fractional anisotropy as measures of white matter integrity. Finally, in a post-hoc analysis, we computed the asymmetry index for comparison of the left and right hemispheric lobes of Rotterdam Study participants without carotid artery stenosis.

Analyses were done using IBM SPSS Statistics version 28.0 (IBM Corp, Armonk, NY, USA). Alpha-level (type 1 error) was set at 0.05.

Results

Of 53 eligible participants, two were excluded because of the presence of a large meningioma, and one because of substantial movement artefacts on structural brain imaging, leaving 50 participants for analysis. Characteristics of the study population are shown in Table 1. Ten participants reported prior (transient) neurological symptoms consistent with ischaemia. A stenosis of ≥50 % was present in the left ICA in 24 (48.0 %) participants, and was right-sided in 26 (52.0 %) participants. Flow through the ICA was lower downstream of the stenosis than on the contralateral side (160 mL/min versus 202 mL/min, P = 0.0002), and the degree of stenosis was related to the degree of flow reduction (reduction in flow [95 % CI] per 1 % increase in stenosis: 1.7 mL/min [1.0–2.5], P < 0.0001).

Table 1.

Participant characteristics.

	Overall sample (n = 50)
Age, years	75.8 (±7.6)
Female sex	25 (50.0 %)
Smoking
Former	33 (66.0 %)
Current	10 (20.0 %)
Systolic blood pressure, mmHg	150 (±21)
Diastolic blood pressure, mmHg	82 (±10)
Blood pressure lowering medication	30 (60.0 %)
Total cholesterol, mmol/L	5.4 (±1.3)
High density lipoprotein cholesterol, mmol/L	1.4 (±0.4)
Lipid lowering medication	23 (46.0 %)
Diabetes	13 (28.9 %)
Body-mass index, kg/m2	27.1 (±3.6)
History of TIA or stroke	10 (20.0 %)
Antithrombotic medication	27 (54.0 %)
Carotid artery stenosis, % (median, IQR)	57 (51–65)
50–59 %	28 (56.0 %)
60–69 %	14 (28.0 %)
≥70 %	8 (16.0 %)
Contralateral stenosis, % (median, IQR)	11 (0–23)

TIA=transient ischaemic attack. Values are depicted as mean±standard deviation for continuous variables, and absolute numbers ( %) for categorical variables, unless indicated otherwise.

Cortical infarction

A total of 12 individuals (24.0 %) had radiographic evidence of cortical infarction, of whom eight (16.0 %) had cortical microinfarcts and four (8.0 %) had larger cortical infarction. Six of eight people had a single microinfarct, whereas the others had two and seven microinfarcts, respectively. All cases of cortical infarction occurred on the side of the stenosis, except for one person with bilateral cortical microinfarcts. As such, cortical infarcts were significantly more common downstream of ICA stenosis than on the contralateral side (P = 0.001), similar for cortical microinfarcts only (P = 0.016). Of all 12 individuals with cortical infarction, 7 (58.3 %) reported having had prior (transient) neurological symptoms consistent with ischemia.

Parenchymal volume

Overall, parenchymal volume was lower on the side of the stenosis than on the contralateral side (mean difference [95 % CI]: −3.2 mL [−5.4;−1.0]). This difference slightly attenuated after excluding the four individuals with evidence of cortical macro-infarction on MRI (−2.7 mL [−4.9;−0.4]), but was unaffected by further exclusion of ten participants with clinically overt cerebrovascular disease (−2.7 mL [−5.2;−0.2]). Proceeding with region-specific analyses among individuals without cortical macro-infarction on MRI, interhemispheric differences in tissue volumes were most profound in the frontal and parietal lobes, and were similar for grey and white matter tissue (Table 2). Interhemispheric differences increased both with degree of stenosis, and with percentage of flow reduction, to a roughly 5 mL difference in individuals with the haemodynamically most significant stenosis (Table 3).

Table 2.

Interhemispheric difference in parenchymal tissue volume.

	Parenchymal volume Δ mL (95 % CI)	Volume share* ( %)	Grey matter Δ mL (95 % CI)	Volume share* ( %)	White Matter‡ Δ mL (95 % CI)	Volume share* ( %)
Hemisphere	−2.7 (−4.9;−0.4)	−0.62	−1.6 (−3.3;0.2)	−0.64	−0.9 (−1.8;0.1)	−0.50
Frontal lobe	−1.5 (−2.4;−0.6)	−1.02	−0.7 (−1.2;−0.1)	−0.85	−0.8 (−1.3;−0.2)	−1.24
Temporal lobe	−0.6 (−2.6;1.4)	−0.66	−0.3 (−1.6;1.1)	−0.51	−0.3 (−1.0;0.3)	−0.92
Parietal lobe	−1.0 (−2.4;0.4)	−1.10	−0.7 (−1.4;0.1)	−1.42	−0.2 (−1.0;0.6)	−0.49
Occipital lobe	0.4 (−0.5;1.3)	+0.74	−0.001 (−0.6;0.6)	−0.003	0.4 (−0.04;0.8)	+1.89

Numbers reflect the difference between affected and unaffected hemisphere, i.e. a negative sign means volumes were lower on the affected side.

^ΔmL=interhemispheric difference in millilitres.

^⁎Percentage of mean hemispheric volume;.

^‡Normal-appearing white matter; CI=confidence interval.

Table 3.

Interhemispheric difference by severity of stenosis.

		Parenchymal volume
	N*	Difference in mL (95 % CI)
Stenosis degree
50–59 %	25	−1.4 (−4.6;1.8)
60–69 %	13	−3.5 (−8.3;1.3)
≥70 %	8	−5.1 (−10.9;0.7)
Flow reduction
No flow reduction	10	0.3 (−3.8;4.4)
1–14 %	12	−1.7 (−6.3;3.0)
15–29 %	12	−3.1 (−8.7;2.6)
≥30 %	12	−5.7 (−10.6;−0.8)

^⁎Four individuals with evidence of cortical macro-infarction on MRI were excluded.

Differences in parenchymal volume were most prominent in older than in younger participants (≥75 years: −5.5 [−8.4;−2.6], versus 0.1 [−3.0;3.3] below age 75). Volumetric differences were also more pronounced in those with left-sided stenosis (−7.3 mL [−9.9;−4.8] for left-sided, versus 0.6 mL [−2.3;3.6] for right-sided stenosis), despite a similar degree of luminal narrowing on both sides (median [IQR] for left– 57 % [51–69], versus right– 58 % [54–63], P = 0.84). Associations did not differ with levels of mean arterial pressure (data not shown). The use of FreeSurfer total parenchymal tissue volumes yielded similar results (overall mean difference −3.1 mL [−5.2;−1.1]).

White matter injury and integrity

We observed no interhemispheric differences in the volume of WMH (2.8 mL [1.7 % of white matter volume] on the side of stenosis, versus 3.0 mL [1.8 %] on the contralateral side; P = 0.23). This was similar on regions-specific analyses by lobe (data not shown). Microstructural white matter integrity was similar between hemispheres (standardised difference [95 % CI] in mean diffusivity: −0.002 [−0.097;0.093], and fractional anisotropy: 0.004 [−0.100;0.108]).

Brain asymmetry in the general population

In the post-hoc analysis of 1000 individuals without carotid artery stenosis, we observed rightward occipital asymmetry (i.e. more grey matter in the right compared to the left occipital lobe), and less profound leftward asymmetry in the frontal lobe (Supplementary figure 1).

Discussion

In this study among community-dwelling individuals with mostly asymptomatic unilateral carotid artery stenosis, we found an increased burden of cortical ischemia, including cortical microinfarcts, and smaller parenchymal volumes downstream of the stenosis, in comparison to the contralateral hemisphere. Volumetric differences were larger with increasing severity of stenosis, and most profound in the internal carotid artery flow territory. White matter integrity was not affected by stenosis in this population.

Brain infarcts on imaging were present in one-fourth of participants in our study, all downstream of the stenosis, further establishing the increased risk of silent brain ischemia with carotid artery stenosis. Prior studies had shown that silent macro-infarcts are more common with stenosis [19,32]. Yet, ischemic complications of stenosis may be more subtle than what is routinely visualised on MRI, emphasised by the higher prevalence of cerebral microinfarcts in the current study. Microinfarcts have recently emerged as a risk factor for future stroke and cognitive impairment [31], and they may well occur commonly with stenosis owing to micro-emboli, hypoperfusion, and cerebral small-vessel disease [31]. In view of potential preventive interventions, it is important to determine whether these observations reflect haemodynamic or thromboembolic consequences of stenosis. Small cortical infarcts on brain imaging and microinfarcts on neuropathology appear predominantly in the watershed areas, supporting a haemodynamic origin [33,34]. This is further supported by animal studies, in which microinfarcts are seen along with microhaemorrhages, white matter injury, (vascular) amyloid deposition, neuronal loss and memory impairment [[35], [36], [37]], following bilateral partial carotid artery coiling. Future studies assessing plaque characteristics in humans may help to differentiate between haemodynamic and thromboembolic origin.

Presence of infarcts alone did not explain the observed interhemispheric differences in parenchymal volume in our study. These volumetric differences could be of neurodegenerative nature, thereby explaining worse cognitive function with carotid artery stenosis in prior studies [17]. However, we cannot rule out a contribution of reduced interstitial volume with lower tissue perfusion, for which reasons longitudinal studies with repeated imaging, for example before and after endarterectomy, are warranted. Numerous reports of varying methodological rigor have suggested that cognitive performance may improve after carotid artery desobstruction. In the absence of randomised controlled trials, these should be no basis for routine intervention, but together with the current findings, they do advocate the incorporation of cognitive endpoints in trials of carotid desobstruction for stroke prevention in the statin era. Whilst for stroke prevention the efficacy of such interventions is often measured in a relatively short timeframe, the prolonged exposure contributing to neurodegeneration and cognitive decline may well pass unnoticed for a long time and fit a different paradigm. Although our study provides certain evidence for a causal contribution of carotid artery stenosis to brain tissue loss, the anticipated impact of any intervention should be estimated cautiously in view of the various alternative mechanisms that also contribute to the imaging phenotype.

The lack of association between stenosis and white matter hyperintensities in the current study is consistent with 4 out of 6 prior studies assessing interhemispheric differences [18,19]. Differences in study design may explain part of the discrepant results, as studies differed in terms of degree of stenosis (ranging from >30 % to >70 %) and outcome measurement (i.e., visual inspection versus volumetric segmentation, and location of WMH) [18]. Relatively mild stenosis may be insufficient to detect differences in small studies, especially in the absence of a sensitive quantitative outcome measure. In two studies comparing individuals with and without stenosis, carotid artery stenosis was associated with volume of WMH in one (using automated segmentation) [32], but not another study (using visual rating) [38]. Alternatively, at least part of the effect of ICA stenosis on cerebral small vessel disease could be due to arterial stiffening and increased downstream pulsatility, rather than luminal narrowing [[39], [40], [41]]. Given the often systemic presence of atherosclerosis, such effects would remain undetected when assessing interhemispheric differences. Regarding white matter microstructure, a meta-analysis of six small studies suggested that carotid artery stenosis relates to changes in FA and the apparent diffusion coefficient, but not MD [42]. We did not find evidence of such an association in the present study, and all in all it remains to be determined whether hypoperfusion itself is in fact a contributor to small-vessel disease [43].

Remarkably, we found associations mainly for left-sided stenosis. Given the similar associations using FreeSurfer segmentation, these differences are unlikely to be due to systematic error in the tissue segmentations. To rule out a physiological asymmetry that accounts for the observed differences, we post-hoc assessed interhemispheric differences in a random sample of 1000 participants of the Rotterdam Study without carotid artery stenosis. In this group we found rightward occipital asymmetry (i.e. more grey matter in the right lobe), with less profound leftward asymmetry in the frontal lobe (Supplementary figure 1), similar to what was recently observed in the large ENIGMA consortium [44]. This implies that the smaller frontal and temporal lobe volumes with left-side stenosis in our study are not merely a natural physiological phenomenon, but might even be a mild underestimation of a true interhemispheric difference. Of note, the Cardiovascular Health Study reported associations of high-grade stenosis of the left, but not the right carotid artery with cognitive impairment and cognitive decline [45]. Taken together, this could point to preferential flow of microemboli from concomitant aortic stenosis into the left common carotid artery, higher plaque vulnerability on the left side [46], or more profound effects of flow impairment on the dominant, metabolically perhaps more active hemisphere. However, numbers of individuals with stenosis in both the Cardiovascular Health Study and the study presented here are small, and lateralisation due to chance may well overrule these rather speculative explanations.

A strongpoint of the current study is the elimination of confounding of concomitant vascular disease by design, in an unselected well-characterised sample of the general population. Certain limitations of this study need to be taken into account when interpreting the findings. First, structural MRI and MRA were not performed concurrently, and although time differences were limited to six months by design, the progressive nature of arterial stenotic disease may have biased our findings to the null if structural MRI was acquired before MRA. Second, we did not measure stenosis of the cerebropetal vessels beyond the carotid artery bifurcation. Concurrent intracranial stenosis may have caused bias in either direction, whereas vertebrobasilar stenosis may notably affect occipital lobe volumes. Third, by design we were unable to assess the effect of stenosis on cognitive performance, but the magnitude of effects, equivalent to about 2 years of atrophy, suggests such effects could be clinically meaningful. Fourth, the limited sample size hampers precision, in particular for subgroup analyses, and larger studies are needed to determine the importance, for example, of plaque characteristics among patients with symptomatic and asymptomatic stenotic disease.

In conclusion, carotid artery stenosis is associated with cortical microinfarcts and parenchymal tissue volume, but in this relatively small sample not with markers of white matter integrity. Future studies are needed to disentangle haemodynamic from thromboembolic effects, and assess the effect of medical treatment and revascularisation procedures on neurodegeneration and cognitive performance.

Sources of funding

The Rotterdam Study is sponsored by the Erasmus Medical Centre and Erasmus University Rotterdam, The Netherlands Organization for Scientific Research (NWO), The Netherlands Organization for Health Research and Development (ZonMw), the Ministry of Education, Culture and Science, the Ministry of Health, Welfare and Sports, the European Commission (DG XII), and the Municipality of Rotterdam. Further support was obtained from the Netherlands Consortium for Healthy Ageing and the Dutch Heart Foundation (CVON 2012-06 and 2018-28). Dr. Wolters was supported by funding from The Netherlands Organization for Scientific Research (NWO Veni-09150162010108). None of the funding organizations or sponsors were involved in study design, in collection, analysis, and interpretation of data, in writing of the report, or in the decision to submit for publication.

Figure S1. Brain asymmetry index (AI) in the Rotterdam Study

Legend: in a random sample of 1000 Rotterdam Study participants without significant carotid artery stenosis, we determined the asymmetry index (AI; vertical axis) for each brain lobe separately. The AI is a measure of grey matter asymmetry, calculated by voxel-wise comparison of grey matter content on the original and flipped image:: $=\frac{originalimage-flippedimage}{0.5*\left(originalimage-flippedimage\right)}$ . AI values are depicted here for the left hemisphere, and a positive AI value consequently indicates more grey matter in the left hemisphere (leftward asymmetry), whereas a negative value indicates rightward asymmetry.

CRediT authorship contribution statement

Frank J. Wolters: Writing – original draft, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Meike W. Vernooij: Writing – review & editing, Funding acquisition, Data curation. Gennady V. Roshchupkin: Writing – review & editing, Methodology, Data curation. M․Arfan Ikram: Writing – review & editing, Methodology, Funding acquisition. Maryam Kavousi: Writing – review & editing, Methodology. Peter J. Koudstaal: Writing – review & editing, Methodology, Data curation. Aad van der Lugt: Writing – review & editing, Funding acquisition, Data curation. Daniel Bos: Writing – review & editing, Visualization, Methodology, Data curation.

Declaration of competing interest

None of the authors report any disclosures.