Association of Brain Arterial Elongation with Risk of Stroke and Death in Stroke-free Individuals: Results from the Northern Manhattan Study

Abstract

Background:

Brain arterial dilation and elongation characterize dolichoectasia, an arteriopathy associated with risk of stroke and death. We aim to determine whether brain arterial elongation increases the risk of stroke and death independent of brain arterial diameters.

Methods:

We analyzed 1,210 stroke-free participants (mean age 71±9 years, 41% men, 65% Hispanic) with available time-of-flight MR-angiogram from the Northern Manhattan Study, a population-based cohort study across a multiethnic urban community. We obtained baseline middle cerebral artery M1-segment (MCA-M1) and basilar artery (BA) mean lengths and diameters using a semi-automated software. Cox proportional hazards models adjusted for brain arterial diameters and potential confounders yielded adjusted hazards ratios (aHR) with 95% CIs for the primary outcomes of incident stroke and all-cause mortality, as well as secondary outcomes including non-cardioembolic stroke, vascular death, and any vascular event.

Results:

Neither MCA-M1 or BA lengths correlated with incident stroke or all-cause mortality. Both MCA-M1 and BA larger diameters correlated with all-cause mortality (MCA-M1 aHR=1.52; 95%CI 1.03–2.23, BA aHR=1.28; 95%CI 1.02–1.61), as well as larger MCA-M1 diameters with vascular death (aHR=1.84, 95%CI 1.02–3.31). Larger MCA-M1 and BA diameters did not correlate with incident stroke. However, larger BA diameters were associated with posterior circulation non-cardioembolic stroke (aHR 2.93; 95%CI 1.07–8.04). There were no statistical interactions between brain arterial lengths and diameters in relation to study outcomes.

Conclusions:

In a multiethnic cohort of stroke-free adults, brain arterial elongation did not correlate with risk of stroke or death, nor influenced the significant association between brain arterial dilation and vascular risk.

Graphical Abstract

Introduction

Brain arterial dilation and elongation characterize intracranial dolichoectasia, a cerebral arteriopathy linked to increased risk of vascular morbidity and mortality.¹ Moreover, larger brain arterial diameters are increasingly recognized as a marker of vascular risk, even in the spectrum of what is considered normal. One study across ambulatory patients in Japan showed that 1 mm increase in basilar artery (BA) diameter was associated near 50% higher risk of cardiovascular events during a mean follow-up of 6 years.² Another study among French patients with ischemic stroke found that BA diameters greater than 3.1 mm were associated with a two-fold risk of stroke-related mortality.³ Our group previously reported that, among stroke-free individuals living in the community, averaged brain arterial diameters greater than two standard deviations from the population mean were at higher risk of vascular death.⁴ The aforementioned studies are complemented by investigations that have found that larger brain arterial diameters correlate with cerebral white matter disease, worse cognitive performance, and risk of dementia.^5–7

Although brain arterial elongation often coexists with larger brain arterial diameters, both in the general population and in individuals with defined intracranial dolichoectasia,^8,9 much less is known about the independent pathogenic implications of this arterial phenotype. The specific aim of this study is to determine whether brain arterial elongation, as measured by arterial length, is associated with an increased risk of incident stroke and death in a multiethnic cohort of stroke-free community-dwelling individuals independent of brain arterial diameters, and weather brain arterial lengths influence the relationship between brain arterial diameters and vascular risk.

Methods

The data that support the findings of this study are available from the NOMAS Steering Committee upon reasonable request.

Study Design, Setting and Participants.

The Northern Manhattan Study (NOMAS) is an ongoing epidemiological prospective cohort study designed to investigate stroke incidence and risks in an ethnically diverse population living in Northern Manhattan. Details of the NOMAS cohort and methods have been previously described.^6,8 In brief, the original NOMAS cohort included a total of 3,298 participants identified by random-digit dialing between 1993 and 2001. Key inclusion criteria were no history of stroke, age ≥40 years-old, and resident for >3 months in northern Manhattan, New York. Between 2003 and 2008, surviving NOMAS participants plus 199 participants’ household members were invited to undergo brain MRI during annual follow-up as part of the NOMAS imaging sub-study, which included a total of 1,290 participants. To be eligible, participants had to have remained clinically stroke-free, be ≥50 years-old without MRI contraindications, and provide written informed consent. For the current analysis, we considered 1,210 participants who had available 3-dimensional time-of-flight (TOF) MR-angiogram (MRA). Data of this analysis is available upon reasonable request to the NOMAS Steering Committee. The study was approved by the Institutional Review Boards of Columbia University and the University of Miami, and followed the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) reporting guidelines for cohort studies.

Baseline clinical data

Demographics and vascular risk factors were obtained at the time of MRA using structured questionnaires by trained study personnel during in-person visits. Fasting blood samples were obtained and evaluated for glucose and a lipid profile. Demographics collected included age at the time of MRA, sex, self-reported race/ethnicity, and insurance status. Prevalent hypertension, diabetes mellitus, and hypercholesterolemia were established by a combination of self-report, treatment of any of these conditions, or physical examination and laboratory data. For hypertension, we used a cutoff of ≥140/90 mm Hg averaged from at least two separate brachial blood pressure measurements. For diabetes, we used a cutoff of fasting glucose ≥126 mg/dl. For hypercholesterolemia, we used a cutoff of total cholesterol ≥ 240 mg/dl. In addition, the duration in years of any of these diagnoses was recorded. Additional information obtained were smoking status quantified by years of smoking and use of medications (antiplatelet agents, antihypertensives, statins and other anti-lipidemic agents, and diabetic medications).

Arterial Imaging Analysis

Brain MRIs were performed in a 1.5T MRI system (Phillips Medical Systems) following a standardized protocol at the Columbia University Medical Center. We obtained TOF MRAs under the following parameters: field of view of 15 cm, 1 mm effective slice thickness, acquisition matrix interpolated to 256×228 matrix, flip angle of 25 degrees, and TR/TE of 20 and 2.7 ms, respectively. Brain arterial diameters and lengths were obtained in millimeters using commercially available software (LAVA, Leiden University Medical Center, The Netherlands, build date October 19^th, 2018) previously validated by our group and others.^8,10 For automatic arterial segmentation, the software uses deformable tubular 3D Non-Uniform Rational B-Splines (NURBS) derived from TOF MRAs to identify the arterial lumen margins based on voxel intensity. A trained operator determined each arterial segment of interest by marking a beginning and ending point. For the middle cerebral artery M1-segment (MCA-M1), the main bifurcation was used as ending point. The software generated a center line between the marks to calculate arterial segment length. Brain arterial diameters were estimated by averaged measurements at the first millimeter of length of each arterial segment. Arterial segments of interest were the ascending portion of the cavernous internal carotid arteries; the anterior, middle and posterior cerebral arteries; the posterior communicating arteries; the intracranial portion of the vertebral arteries (VA); and the BA; (maximum 13 arterial segments per individual). Because of heterogeneity in the composition of the circle of Willis and the arteries we measured, we first focused on the length and diameter of the MCA-M1 and the BA, which were consistently present in all participants. In addition, we calculated the length and diameter Z-scores for each arterial segment and averaged them by the number of identified arterial segments per individual to estimate an overall length and diameter Brain Arterial Remodeling (BAR) score.

To control for confounders, we ascertained additional morphological parameters including total cranial volume (TCV), presence of intracranial stenosis, and completeness of the circle of Willis.^11–13 TCV was obtained using a custom-designed image analysis package (QUANTA 6.2 using a Sun Microsystems Ultra 5 workstation).¹⁴ The degree of intracranial stenosis was quantified based on the narrowest lumen area compared to the immediately preceding segment with normal lumen (or the next normal appearing lumen if the stenosis was at the arterial origin), and was considered clinically relevant when greater than 50%.¹⁵ Completeness of the COW was determined based on the presence or absence (including hypoplasia) of proximal intracranial arteries on the 3D format MRA reconstruction. Presence of VA dominance was determined when the contralateral VA was either hypoplastic or non-visualized. In addition, the presence or absence of uni- or bilateral fetal-type posterior cerebral artery (fPCA), as well as the anterior temporal artery branching from the MCA, were included as potential confounders based on demonstrated relatedness with intracranial arterial diameters and lengths.⁸

Study Outcomes

The primary outcomes studied were any ischemic stroke and all-cause mortality. Other outcome studied included non-cardioembolic stroke in the territory of the analyzed circulation (i.e., anterior circulation and posterior circulation for MCA-M1 and BA analyses, respectively), vascular death (cause of death was attributed to either stroke, myocardial infarction (MI), heart failure, pulmonary embolus or cardiac arrhythmia), and any vascular event defined as the composite of any stroke, MI or vascular death. For the purpose of this study, we used outcomes data until July 2020. Study outcomes were defined at the initiation of the NOMAS cohort and adjudicated by investigators blinded to vascular imaging analyses during annual in-person visits or standardized telephone interviews. In addition, we used continuous surveillance of medical records from the Columbia University Irving Medical Center (where most participants obtained medical care), as well as other institutions when applicable, to provide outcome data. Data obtained included vital status (dead or alive), interval hospitalizations, and presence of symptoms or signs suggestive of stroke or MI. Stroke outcome was defined according to the World Health Organization criteria as the acute onset of focal (or global) disturbance of cerebral function lasting >24 hours, with no apparent non-vascular cause.¹⁶ Two independent study vascular neurologists blinded to the MRA subanalysis adjudicated stroke subtypes. MI was established by a study cardiologist according to the adapted criteria from the Lipid Research Clinics Coronary Primary Prevention Trial.¹⁷

Statistical analysis

Data management and analysis were performed using SAS software, version 9.4 (SAS Institute Inc., Cary, NC). To test whether baseline characteristics differed between NOMAS participants with and without TOF MRA, we used linear models for continuous variables and the chi-squared test for categorical variables. We obtained the overall crude incidence with 95% CI for first-occurring study outcome per 1,000 person-years. We built serial Cox proportional hazards regression models to obtain the hazard ratio (HR) and 95% CI to evaluate whether MCA-M1 length, BA length, or the overall length BAR score correlated with study outcomes, with estimates calculated using Fine & Gray regression to account for competing risks. The first regression models were adjusted for anatomical covariates known to be associated with brain arterial lengths including diameters of the arterial segments of interest, TCV, and presence of fPCA.⁸ Secondly, the fully-adjusted models included model 1 variables plus demographics and risk factors associated to vascular events including age, sex, race/ethnicity, insurance status, hypertension, years of hypertension, diabetes, years of diabetes, hypercholesterolemia, smoking status, years of smoking, antiplatelet use, and presence of significant intracranial stenosis. In addition, for MCA-M1 analyses we included the averaged diameter of the anterior temporal arteries, whereas for BA analyses we included VA dominance. For each outcome, we tested for interaction by adding a length × diameter term as an independent variable in the models. A two-tailed p-value <0.05 was considered statistically significant.

Results

Of 1290 participants enrolled in the NOMAS imaging sub-study cohort, 1,210 (mean age 70.5 ±8.9 years, 41.0% men, 64.5% Hispanic) were included in the analysis. Only three (0.4%) participants were lost to follow-up and 11 (1.4%) withdrew from the study. Baseline characteristics of the included participants are shown in table 1. Compared to participants excluded due to absence of available TOF MRA, included participants were younger and less likely to be women, Hispanic, have diabetes, or be current smokers (Supplemental Table S1). Post-MRA median follow-up time was 12 years (range 0.1–17). Overall, there were 112 (9.3%) ischemic strokes, 61 (5.0%) MIs, 200 (16.5%) vascular deaths, and 454 (37.5%) deaths, resulting in a crude incidence rate per 1000 person-years of 8.8 (95%CI 7.4–10.5), 14.6 (95%CI 12.8–16.7), 24.9 (95%CI 22.4–27.7) and 32.9 (95%CI 30.1–36.0) for any ischemic stroke, vascular death, any vascular event, and all-cause mortality, respectively.

Table 1.

Baseline characteristics of NOMAS participants included in the analysis.

	Total Sample (N=1210)
Demographics
Age in years, mean ± SD	70.5 ±8.9
Male, n (%)	496 (41.0)
Ethnicity, n (%)
Non-Hispanic White	185 (15.3)
Non-Hispanic Black	213 (17.6)
Hispanic	785 (64.9)
Other	27 (2.2)
Insured, n (%)	1029 (85.0)
Clinical characteristics
Hypertension, n (%)	943 (77.9)
Diabetes Mellitus, n (%)	296 (24.5)
Hypercholesterolemia, n (%)	979 (80.9)
Current smoker, n (%)	136 (11.2)
Measurements
SBP in mmHg, mean ± SD	136.2 ± 17.5
DBP in mmHg, mean ± SD	78.1 ± 9.7
Glycemia in mg/dL, mean ± SD	100.7 ± 98.8
LDL-C in mg/dL, mean ± SD	115.5 ± 35.5
HDL-C in mg/dL, mean ± SD	53.4 ± 17.0
Triglycerides in mg/dL, mean ± SD	126.5 ± 112.0
Smoking packs per day, mean ± SD	0.4 ± 0.6
Years of smoking, mean ± SD	16.1 ± 19.5
Imaging measurements
MCA-M1 length in mm, mean ± SD	16.7 ± 5.2
BA length in mm, mean ± SD	24.1 ± 6.4
MCA-M1 diameter in mm, mean ± SD	2.7 ±0.3
BA diameter in mm, mean ± SD	2.8 ± 0.5
TCV in cm3, mean ± SD	1156.5 ± 121.8

Abbreviations: DBP diastolic blood pressure, HDL-C high-density lipoprotein cholesterol, LDL-C low-density lipoprotein cholesterol, PP pulse pressure, SD standard deviation, SBP systolic blood pressure, TCV total cranial volume.

In the first models adjusted for TCV and fPCA presence, neither MCA-M1 or BA length correlated with any of the study outcomes. Conversely, larger diameters in both MCA-M1 and BA showed a significant association with all-cause mortality, whereas the association with incident ischemic stroke was not significant. Larger MCA-M1 diameters correlated with risk of vascular death and any vascular event, while larger BA diameters correlated risk of non-cardioembolic stroke in the posterior circulation, risk of vascular death and any vascular event (Table 2).

Table 2.

Risk of stroke and death according to brain arterial lengths and diameters

	All-cause Mortality	Ischemic stroke	Non-cardioembolic stroke*	Vascular Death	Any Vascular Event^
	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)
Middle Cerebral Artery
Length
Model 1	1.01 (0.99–1.03)	1.01 (0.97–1.05)	1.02 (0.97–1.08)	1.02 (0.99–1.04)	1.02 (0.99–1.04)
Model 2	1.00 (0.98–1.02)	1.00 (0.96–1.04)	1.01 (0.96–1.07)	1.00 (0.98–1.03)	1.00 (0.98–1.03)
Diameter
Model 1	2.05† (1.40–3.01)	1.39 (0.62–3.12)	1.80 (0.54–5.91)	2.52† (1.42–4.49)	1.86† (1.20–2.91)
Model 2	1.52† (1.03–2.23)	0.99 (0.43–2.20)	1.48 (0.44–4.94)	1.84† (1.02–3.31)	1.34 (0.86–2.11)
Basilar Artery
Length
Model 1	1.00 (0.98–1.02)	1.03 (0.99–1.06)	0.94 (0.87–1.03)	1.00 (0.98–1.02)	1.00 (0.99–1.02)
Model 2	1.00 (0.98–1.01)	1.02 (0.99–1.06)	0.93 (0.85–1.02)	1.00 (0.97–1.02)	1.00 (0.98–1.02)
Diameter
Model 1	1.48† (1.19–1.83)	1.19 (0.76–1.89)	2.98† (1.14–7.78)	1.54† (1.11–2.13)	1.32† (1.02–1.71)
Model 2	1.28† (1.02–1.61)	1.02 (0.64–1.64)	2.93† (1.07–8.04)	1.30 (0.97–1.85)	1.19 (0.86–1.46)

^*Non-cardioembolic stroke in the anterior circulation for middle cerebral artery analyses and in the posterior circulation for basilar artery analyses

^{^}Any vascular defined by the composite outcome of any stroke, myocardial infarction or vascular death.

Model 1 was adjusted for total cranial volume and presence of fetal posterior cerebral artery; additional covariates included anterior temporal artery diameter for middle cerebral artery analyses and vertebral artery dominance for basilar artery analyses.

Model 2 was adjusted for variables included in model 1 plus age at the time for MRI, sex, ethnicity, insurance status, hypertension, years of hypertension, diabetes, years of diabetes, hypercholesterolemia, smoking status, years of smoking, antiplatelet use, and presence of significant intracranial stenosis.

^†p-value <0.05

Statistical note: Hazard ratios represents per 1 mm increase in arterial length or diameter. Estimates calculated with Fine & Gray regressions

In the fully-adjusted models, larger MCA-M1 and BA diameters showed an independent association with all-cause mortality (MCA-M1 aHR=1.52; 95%CI 1.03–2.23, BA aHR=1.28; 95%CI 1.02–1.61) but not with incident ischemic stroke. For secondary outcomes, larger MCA-M1 diameters remained associated with risk of vascular death (HR=1.84; 95%CI 1.02–3.31), and larger BA diameters with risk of non-cardioembolic stroke in the posterior circulation (HR=2.93; 95%CI 1.07–8.04).

Consistent with the primary analyses, grouping all arterial segments using the BAR score showed that averaged brain arterial lengths had no effect on study outcomes. Larger averaged brain arterial diameters were independently correlated with all-cause mortality (HR=1.24; 95%CI 1.06–1.46) but not with incident ischemic stroke (Table 3; Figure 1).

Table 3.

Risk of vascular events and death according to the Brain Arterial Remodeling (BAR) score for arterial lengths and diameters

	All-cause Mortality	Ischemic stroke	Non-cardioembolic stroke	Vascular Death^	Any Vascular Event^
	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)
BAR score
Length
Model 1	1.13 (0.96–1.34)	1.07 (0.76–1.51)	1.01 (0.66–1.59)	1.07 (0.83–1.38)	1.03 (0.84–1.25)
Model 2	1.01 (0.85–1.20)	0.96 (0.67–1.36)	0.93 (0.60–1.44)	0.94 (0.71–1.23)	0.90 (0.73–1.11)
Diameter
Model 1	1.36† (1.17–1.59)	1.13 (0.82–1.56)	1.28 (0.66–1.56)	1.46† (1.15–1.85)	1.25† (1.04–1.50)
Model 2	1.24† (1.06–1.46)	0.99 (0.71–1.37)	1.18 (0.78–.1.77)	1.36† (1.06–1.74)	1.12 (0.92–1.35)

^{^}Any vascular defined by the composite outcome of any stroke, myocardial infarction or vascular death.

Model 1 was adjusted for total cranial volume, presence of fetal posterior cerebral artery and anterior temporal artery diameter.

^†p-value <0.05

Model 2 was adjusted for variables included in model 1 plus age at the time for MRI, sex, ethnicity, insurance status, hypertension, years of hypertension, diabetes, years of diabetes, hypercholesterolemia, smoking status, years of smoking, antiplatelet use, and presence of significant intracranial stenosis.

Statistical note: Hazard ratios represents per 1 mm increase in arterial length or diameter. Estimates calculated with Fine & Gray regressions

Figure 1.

Cumulative incidence according to averaged brain arterial diameters and lengths stratified in deciles (top decile, middle 8 deciles, and bottom decile) for (a) all-cause mortality and (b) ischemic stroke.

We found no significant interaction between brain arterial lengths and diameters for any of the outcomes studied (supplemental Table S2).

Discussion

In a community-based racially and ethnically diverse cohort of initially clinically stroke-free adults, we found that increased brain arterial diameters of the anterior and posterior circulation increased the risk of all-cause mortality during follow-up. Conversely, despite a previously-reported direct relationship between brain arterial lengths and diameters in our cohort [8], brain arterial lengths did not correlate with vascular risk or influence the association between increased brain arterial dimeters and adverse vascular outcomes.

Increased brain arterial length is considered a marker of brain arterial remodeling and can be used as a surrogate of brain arterial tortuosity, a phenotype capable of influence cerebral blood flow and cause mechanical kinking of penetrating vessels.^18,19 In our study, these mechanisms played no significant role on vascular risk in community dwelling individuals. Our results differ from studies conducted among patients with established vertebrobasilar dolichoectasia which have reported that both BA height-of-bifurcation and lateral displacement are associated with increased risk of cerebrovascular events.^3,20,21 Contrary to our study, the aforementioned reports used anatomical landmarks (rather than continuous measurements) to determine BA elongation, thus focused on the extreme phenotypes of brain arterial elongation.

Our results concur with previous studies conducted in hospital-based populations that identified larger brain arterial diameters as a marker of vascular risk in both individuals with and without defined dolichoectasia. One study across 493 outpatients with atherosclerotic risk factors conducted in Japan reported that increased BA diameters correlated with incident cardiovascular events including vascular death, stroke, and coronary events during a mean follow-up of 6 years, despite BA diameters considered to be within normal limits (BA diameter range 1.1–5.2 mm, with less than 0.8% considered to have dolichoectasia).² The GENIC study in a French cohort of 466 subjects with ischemic stroke found that the 5-year risk of stroke mortality increased 23% per 1 mm increase in BA diameters.³ Similarly, one study in China found that a larger BA diameter (≥ 5.3 mm) was associated with a four-fold increased risk of stroke recurrence among 115 stroke patients with vertebrobasilar dolichoectasia.²¹

The correlation of larger brain arterial diameters with vascular risk and mortality in the general population is likely complex and multifactorial. Brain arterial dilation has been linked to abnormal vascular remodeling associated with dysregulation of matrix metalloproteinases (MMPs).^22–25 These proteolytic enzymes regulate extracellular proteins in the tunica media and can lead to increased elastin and collagen degradation, resulting in increased arterial diameters.²⁶ Some MMPs (i.e., MMP-9) associated with brain arterial dilation have been proposed to play a role in brain disorders closely related to cerebrovascular morbidity, such as small vessel disease and Alzheimer’s disease.^27–29 This shared underlying pathophysiology is supported by robust evidence on the relationship between larger brain arterial diameters and imaging signatures of small vessel disease,^5,30–33 as well as worse cognitive performance.^6,7 MMP dysregulation has also been found in patients with abdominal aortic aneurysms and coronary ectasia,^34,35 both conditions reportedly to correlate with larger BA diameters.^36–38 Therefore, it is plausible that dilatative arteriopathies in other vascular beds contribute to vascular risk among those with larger brain arterial diameters. Another potential contributor to vascular risk is the bidirectional pathogenic association between larger arterial diameters and atherosclerosis. Larger diameters can disturb blood flow causing low wall sheer stress and formation of atherosclerotic plaques,³⁹ whereas early atherosclerotic plaque formation can result in compensatory arterial enlargement to maintain luminal diameter.⁴⁰ Finally, brain arterial dilation by itself can be a direct cause of both ischemic and hemorrhagic stroke, as well as less frequent neurological complications including brainstem compression and hydrocephalus.¹ In our study, we did not find an independent correlation between brain arterial dilation and incident ischemic stroke. However, larger BA diameters were independently associated with a 3-fold higher risk of non-cardioembolic stroke in the posterior circulation.

A fundamental step to advance dolichoectasia research is to establish unified diagnostic criteria practical for clinicians and useful for researchers. Our findings suggest that the vascular risk associated with the arterial phenotypes that feature dolichoectasia is mainly driven by brain arterial dilation rather than arterial elongation. This is relevant since quantification of brain arterial length and tortuosity usually require advanced software, whereas arterial diameters are easily determined on routine brain vascular imaging. Although diameter cutoffs to establish dolichoectasia have been proposed,^9,11,20 larger studies including heterogeneous populations and detailed clinical follow-up are needed to establish diameter cutoffs relevant for the effect of major intracranial arteries of the anterior and posterior circulation on clinical outcomes.

Strengths of our study include its population-based design, inclusion of an ethnically diverse population, systematic ascertainment of study outcomes, and low rate of loss to follow-up. However, our study has some limitations. Our population underrepresents individuals of Asian or Amerindian ancestry, as well as those from younger age groups. The low incident rate of individual vascular outcomes such as ischemic stroke may limit our power to demonstrate associations. Although it is reasonable to assume that arterial length is tightly correlated with arterial tortuosity, arterial bending was not directly analyzed. For MCA-M1 arterial segment, well-established anatomical variations in MCA branching pattern may have influenced MCA-M1 length assessment.³¹ We included intracranial stenosis as a potential confounder for brain arterial lengths an diameters association with study outcomes. However, we lacked reliable assessment of stenosis degree in the extracranial vasculature which may influence brain arterial diameters. Finally, the lack of imaging follow-up prevents us to assess dynamic changes in brain arterial dimensions as an additional marker of risk for vascular events and death.

In conclusion, we confirmed the association between increased brain arterial diameters and vascular risk in a multiethnic community-based sample, while brain arterial elongation did not demonstrate such association. A definition of dolichoectasia based on brain arterial dilation rather than on elongation criteria may represent a practical approach for researchers and clinicians. The mechanisms by which increased brain arterial dilation is a marker of vascular risk and mortality need to be elucidated in order to develop disease-specific therapies for this unexplored cerebral arteriopathy.

Highlights:

• Previous research recognizes brain arterial dilation as a marker of arterial remodeling linked to vascular risk. However, much less is known about brain arterial elongation, an arterial phenotype that often coexist with larger brain arterial diameters.

• We found that brain arterial elongation did not correlate with risk of stroke or death, nor influenced the significant association between increased brain arterial diameters and vascular risk in a population-based multiethnic cohort of stroke-free adults.

• In the general population, vascular risk related to dolichoectasia seems to be determined primarily by bran arterial dilation, which have relevant implications for dolichoectasia definition and research.

Non-standard Abbreviations and Acronyms

BA

Basilar artery

BAR

Brain Arterial Remodeling

fPCA

Fetal-type posterior cerebral artery

HR

Hazard ratio

MCA-M1

middle cerebral artery M1-segments

MRA TOF

MR-angiogram time-of-flight

NOMAS

Northern Manhattan Study

TCV

Total cranial volume