Anatomical Effects on the Relationship Between Brain Arterial Diameter and Length: The Northern Manhattan Study

Abstract

Background and Purpose:

In patients with dolichoectasia, it is uncertain how dilatation and/or elongation relate to each other. We aimed to examine the correlation between arterial diameter and length within arteries and across the Circle of Willis (COW).

Methods:

We included stroke-free participants in the Northern Manhattan Study who underwent Magnetic Resonance Angiography. Intracranial artery diameters and lengths were obtained with semi-automated commercial software and were adjusted for head size. We first investigated the correlation between diameters and length using Pearson’s correlation coefficient. We then built generalized linear models adjusted for demographics and risk factors.

Results:

Among 1210 participants included in the analysis (mean age 71 ± 9 years, 59% women, 65% Hispanic), a larger basilar artery (BA) diameter correlated with greater BA length (r=0.3), and left and right middle cerebral artery (MCA) diameters correlated with one another (r=0.4). Across the COW, BA diameter correlated with MCA diameters (r=0.3 for both). In adjusted analyses, MCA diameters were associated with larger posterior circulation diameters (β= 0.07), MCA and BA lengths (β=0.003 and β=0.002, respectively), presence of fetal posterior cerebral artery (PCA), (β=0.11), and a complete COW (β= −0.02). Similarly, BA length was associated with a fetal PCA (β=1.1), and BA diameter was associated with anterior circulation diameters (β=0.15) and presence of fetal PCA (β= −0.4).

Conclusions:

COW configuration should be considered when using arterial diameter cutoffs to define dolichoectasia. Further studies are needed to discern whether arterial diameter or length best identify individuals at risk of vascular events attributable to dolichoectasia.

Introduction

Intracranial arteriopathies are a known risk factor for vascular events. Most commonly, we associate intracranial large artery stenosis, from atherosclerosis or other causes, with a future risk of stroke.¹ However, recent data shows that arterial dilatation, as measured by larger diameters, is also associated with risk of vascular events and death.² Arterial dilatation frequently accompanies arterial elongation and tortuosity, which is measured by length and lateral deviation. Arterial dilatation and elongation have frequently been studied in the context of brain large artery dolichoectasia (DE), a disease known to cause strokes.³ While DE is an extreme phenotype that may affect any of the brain arteries, much of the literature on DE focuses on basilar artery (BA) dilatation because of the ease of diagnosis based on visual assessment.⁴ As a result, studies examining DE are less likely to include elongation and/or tortuosity of the affected vessels in their measurements. Apart from DE, the relationship between more subtle degrees of arterial elongation and tortuosity with cerebrovascular disease is less clear.^5,6 A confounder in the relationship between arterial diameter and lengths is Circle of Willis (COW) completeness. Our prior study found that BA dilatation was more common in participants that lack antero-posterior communicating arteries in their COW, which suggests an expected compensatory increase in diameters as a function of greater flow.⁷ To date, there is little data on how arterial dilatation, elongation, and/or tortuosity all correlate with one another and with COW configuration. If, as we hypothesize, arterial dilatation is correlated with arterial elongation and/or tortuosity, it would be important for future studies to account for both to determine which of these geometric features may be helpful in stratifying the risk of vascular events in arteriopathies such as DE.

The objective of our analyses was to examine the relationship between intracranial arterial diameters and length of each artery and across the circle of Willis to provide normative data that would inform future studies about the correlation of these measures and the anatomical expectations that need to be taken into account to better define abnormal phenotypes.

Methods

Study Design

The Northern Manhattan Study (NOMAS) includes stroke-free participants identified using random digit dialing with dual-frame sampling to identify published and unpublished telephone numbers.⁸ Brain MRIs were obtained in 1290 NOMAS stroke-free participants >50 years between 2003 and 2008. The institutional review boards at Columbia University Irving Medical Center and the University of Miami approved the study. Imaging was performed on a 1.5-Tesla MRI system (Philips Medical Systems) at the Columbia University Irving Medical Center following a standardized protocol. Eligible participants also underwent 3-dimensional time-of-flight (TOF) Magnetic Resonance Angiography (MRA) with 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 repetition time/echo time of 20/2.7ms. For the purposes of this study, we only included participants that had an available TOF MRA. An investigator (J.G.) rated MRA quality on a 4-point scale (excellent, good, poor, very poor) in all the included participants. The MRA grading quality was based on the presence and degree of motion artifact: “excellent” quality meant a well flow-enhanced MRA without any artifact, “good” indicated flow-based artifact or streak artifacts in only curved portions of the arteries without motion artifact, “poor” was used if there was some motion artifact that didn’t impair the ability to distinguish anatomical structures, and “very poor” indicated extreme motion artifact.

To control for confounders, we ascertained the following additional demographic and vascular risk factors: age, sex, race/ethnicity, hypertension, diabetes mellitus, hyperlipidemia, smoking, coronary heart disease, peripheral vascular disease, and atrial fibrillation.⁸ Age, sex, and race/ethnicity were self-reported. Vascular risk factors were defined at the time of MRI and during follow-up by either self-reported diagnosis, treatment, blood pressure, or laboratory measurements. Additional information including total cranial volume (TCV), use of certain medications (hypertension or statin medications), number of years with vascular risk factors (diabetes mellitus, hypertension, smoking years), and height were also obtained.

Measurements

Brain arterial diameters and lengths were obtained from all available MRA images. We used commercially available software (LAVA, Leiden University Medical Center, The Netherlands, build date Oct 19, 2018). Briefly, this software uses a flexible 3D tubular Non-Uniform Rational B-Splines model to automatically identify the margins of the arterial lumen based on voxel intensity, in our case, derived from TOF MRA.⁹ The software operator marks the beginning and the end of the desired arterial segment, and the software calculates the length between the seed points as well as the volume, cross-section area and diameters of the individual segment in between the seed points (Figure). The software has been previously used by others and validated with excellent reliability.¹⁰ Lengths were estimated from the total number of voxels included between the origin of a given arterial segment origin and the first bifurcation. We calculated the average diameter of the first mm in length to keep it consistent with previous analyses in this cohort.¹¹ In this analysis, we focused on both middle cerebral arteries (MCAs), anterior cerebral arteries (ACAs), and the BA. Similar to prior studies, arterial diameters were adjusted for head size.¹² In addition to obtaining individual artery diameters, we used a similar method as in our prior study to compute average Z-scores for each participant by adding all measured diameters for the anterior and posterior circulation arteries and dividing by the total number of identified arteries.²

Figure:

Demonstration of the software used for arterial morphometric evaluation. The software operator marks the beginning (red circle) and end (blue circle) of the desired arterial segment, and the software calculates the length between the seed points as well as the volume, cross-section area, and diameters of the individual segment in between the seed points. A and C) Beginning and end of the measured portion of the basilar artery and left middle cerebral artery, B and D) calculated diameter and length of basilar and left middle cerebral artery. Abbreviations: BA, basilar artery; L MCA, left middle cerebral artery.

We carried out additional measurements of the BA. The laterality of the BA and height of the BA bifurcation were based on visual inspection of the artery and its surrounding structures,^{12, 13} and were individually scored from 0–3 based on Smoker’s criteria. Additionally, we measured BA laterality manually by measuring the lateral displacement of the midpoint of the BA at three locations along its course: 1) vertebrobasilar junction, using the distance to the clivus, 2) mid-pons at the level of the trigeminal nerve exit, using the distance to the edge of each sella or internal carotid artery (if sella was not visible), and 3) basilar tip, using deviation from the center of the axial image. The total BA curvature was calculated using the sum of measurements from these three locations, and was used to assess the degree of BA tortuosity. All measurements were obtained perpendicular to a central sagittal axis and recorded in millimeters. We used a 10% random sample for intra- and inter-reader agreement, and reproducibility was assessed with kappa values and intra-class correlation coefficients for categorical and continuous variables, respectively. Visual inspection of the BA was done by two neurologists (S.S.O, C. Z.), with very good reliability for lateral deviation (α = 0.82) and moderate reliability for height of BA bifurcation (α = 0.63) based on Smoker’s criteria, and good to excellent reliability (α = 0.86) based on manual measurement.

We included previously obtained morphological parameters of various intracranial arteries in the COW.⁷ Similar to prior studies, COW completeness was defined based on its visibility on MRA.^7,14 These arteries were described as either present on the 3D format reconstruction, not visible on the 3D format due to small size but visible on the MRA axial source images (i.e., hypoplastic arteries), or not visible in either the 3D format or MRA axial source images (i.e., absent arteries). The total number of arteries per patient that were hypoplastic and absent were recorded. Arteries included in the anterior part of the COW included the MCAs, the ACAs, the anterior temporal arteries branching off the MCA, and anterior communicating arteries (AComm). Arteries included in the posterior part of the COW included the posterior cerebral artery (PCA), anterior-posterior arteries in the form of fetal PCA, and the BA. The AComm, anterior temporal arteries, and fetal PCAs were described as either “present” or “absent.”

Statistical Analysis

Baseline characteristics were reported with standard descriptive statistics. We performed three distinct analyses using the MRA data from the included participants.

The first analysis explored the interrelatedness of arterial diameter and length within each artery using Pearson’s correlation coefficients. Subsequently, we explored the inter-relationship between arterial diameters and length across arteries in the COW and whether arterial diameter and length related to the presence or absence of the AComm artery, anterior temporal arteries, and fetal PCAs using student’s t-test.

The second analysis built multivariable models to examine the association between the configuration of the COW and either the diameter or length of a given intracranial artery. For these analyses, we collected information on five anatomical variants of the COW: Presence or absence of an anterior temporal artery (branching off from the M1 segment), presence or absence of an AComm, and presence or absence of a fetal PCA. Additionally, we rated the number of arteries that were absent (not visualized using MRA) or hypotrophic, as described above. Linear regression was used for continuous variables and a generalized linear model with Poisson distribution was used for categorical variables; the beta estimate and its p-value were reported for both analyses. To ensure quality control, we adjusted for MRA quality to see whether imaging quality affects the association.

Lastly, we built adjusted models to evaluate whether the associations noted between length and diameter were independent. Therefore, we progressively adjusted for confounders including demographics and vascular risk factors in all models.

A two-sided p value of <0.05 was considered statistically significant. All analyses were done using the statistical software SAS (Cary, NC, USA, Version 9.4).

Results

Among the total NOMAS participants, 1290 had MRA data available for semi-automated arterial diameter measurements. Of these patients, 80 were excluded from the analysis due to severe motion artifact or missing data, leaving 1210 subjects within the analysis (Table 1). MRA imaging quality was excellent (37%) or good (54.6%) in the majority of included participants; 7.7% had poor and 0.7% had very poor imaging quality. Among participants included in the analysis, the mean age was 71 ± 9 years (59% women; 65% Hispanic). In the sample, 60% had a right anterior temporal artery, 66% had a left anterior temporal artery, 92% had an Acomm, 13% had a right fetal PCA and 11% had a left fetal PCA.

Table 1:

Characteristics of participants within the NOMAS MRI sub-study that were included in this analysis versus excluded participants

Characteristic†	Total Sample (N=1290)*	Included participants (N=1210)	Excluded participants (N=80)	P value
Demographics
Age, years	70±9	71±9	73±10	0.03
Male Sex,%	40	41	18	<0.001
Ethnicity
Non-Hispanic White, %	15	15	8	0.06
Non-Hispanic Black, %	17	18	13	0.24
Hispanic, %	66	65	78	0.02
Insured, %	85	85	83	0.54
Vascular Risk Factor
HTN, %	78	78	84	0.22
DM, %	26	23	41	<0.001
Hypercholesterolemia, %	90	90	90	0.98
Current smoker, %	12	11	19	0.04
CAD, %	24	24	38	0.005
Afib, %	5	4	6	0.43
PAD, %	0	0	0	0.56
Measurements‡
Height, inches	64±4	64±4	62±4	<0.001
TCV, cm3	1152±123	1157±122	1089±119	<0.001
Pulse pressure, mmHg	58±15	58±15	63±15	0.005
SBP, mmHg	136±18	136±18	139±18	0.13
MAP, mmHg	98±11	98±11	97±11	0.89
DBP, mmHg	78±10	78±10	76±10	0.11
LDL, mg/dL	115±35	116±36	112±34	0.44
HDL, mg/dL	53±17	53±17	52±18	0.53
Triglycerides, mg/dL	127±78	127±79	128±65	0.86
DM duration, years	2±6	2±6	5±9	<0.001
HTN duration, years	9±12	9±12	13±13	<0.001
Smoking duration, years	16±20	16±20	17±23	0.72

CAD, Coronary Artery Disease; Afib, Atrial Fibrillation; PAD, Peripheral Artery Disease; TCV, Total Cranial Volume; SBP, Systolic Blood Pressure; MAP, Mean Arterial Pressure; DBP, Diastolic Blood Pressure; LDL, Low Density Lipoprotein; HDL, High Density Lipoprotein; SD, Standard Deviation; DM, Diabetes Mellitus; HTN, Hypertension

^*N represents the number of stroke-free participants in the sample

^†All the data represent mean±standard deviation unless otherwise indicated

^‡Measurements obtained at the time of MRI

Relationship between Arterial Diameters and Length

At the arterial level, BA diameter correlated with BA length (r=0.3, P<0.05), and BA length correlated with BA lateral deviation (r=0.4, P<0.05) (Table 2). Across the COW, BA diameter correlated with right MCA diameter (r=0.3, P<0.05) and left MCA diameter (r=0.3, P<0.05). Left MCA diameter correlated with right MCA diameter (r=0.4, P<0.05).

Table 2:

Correlation between dilatation, elongation, and tortuosity of various intracranial arteries^§

	BA diameter	BA laterality†	BA length	Left MCA diameter‡	Right MCA diameter‡	Left ACA diameter‡	Right ACA diameter‡	Left MCA length	Right MCA length
BA diameter‡		0.19*	0.27*	0.27*	0.33*	0.06*	0.14*	0.04	0.05
BA laterality†	0.19*		0.42*	0.06	0.02	−0.06	0.06	0.13*	0.10*
BA length	0.27*	0.42*		0.04	0.04	−0.03	0.00	0.05	0.02
Left MCA diameter‡	0.27*	0.06	0.04		0.43*	0.14*	0.17*	0.02	0.07*
Right MCA diameter‡	0.33*	0.02	0.04	0.43*		0.15*	0.09*	0.07*	0.00
Left ACA diameter‡	0.06	−0.06	−0.03	0.14*	0.15*		0.05	0.02	0.04
Right ACA diameter‡	0.14*	0.06	0.00	0.17*	0.09*	0.05		0.03	0.10*
Left MCA length	0.04	0.13*	0.05	0.02	0.07*	0.02	0.03		0.23*
Right MCA length	0.05	0.10*	0.02	0.07*	0.00	0.04	0.10*	0.23*

MCA, middle cerebral artery; BA, basilar artery; ACA, anterior cerebral artery

^*P <0.05

^†BA laterality was calculated manually by measuring the lateral displacement of the midpoint of the BA at the vertebral junction, mid-pons, and tip.

^‡Measurement obtained from the first segments of the arteries.

^§Using Pearson correlation coefficient. Numbers represent r.

Arterial Diameters and Length as a Function of the Circle of Willis Configuration

In the anterior circulation, the MCA M1 segments were longer in participants with an anterior temporal artery (P<0.0001) or in the absence of ipsilateral fetal PCA (P=0.02) and more dilated in the absence of an ipsilateral fetal PCA (P=0.02). The ACA A1 segments were more dilated in participants with left anterior temporal arteries (P=0.012) and with absent fetal PCAs (P<0.0001). In the posterior circulation, participants with longer BA were less likely to have anterior temporal arteries (P=0.005), and participants with larger BA diameters were less likely to have right or left fetal PCAs (P<0.0001 for both associations).

In multivariable analyses, the MCA M1 segment length was associated with anterior circulation diameters (B= 1.3, p<0.0001), presence of anterior temporal arteries (B= 0.6, P<0.05), absent fetal PCA (B= −1.0, P=0.02) and greater number of hypotrophic arteries in other regions of the COW (B= −0.4 per hypotrophic artery, P<0.05) (Table 3). MCA M1 segment diameter was associated with larger posterior circulation diameters (B= 0.07, P<0.0001), larger intracranial volume (ICV; B= 0.0006, P<0.0001), MCA and BA lengths (B= 0.003, P=0.02; B=0.002, P=0.03), presence of fetal PCA (B= 0.11, P<0.001), and complete COW (B= −0.02, P=0.01).

Table 3:

Multivariable Analysis of Anthropomorphic Features Pertaining to the Anterior Circulation

	ATA	Fetal PCA	AComm	Posterior Circulation Diameter†	Anterior Circulation Diameter†	MCA Length	BA Length	Absent Score‡	Hypoplastic Score‡
MCA Diameter, β, SE	−0.009, 0.01	0.11*, 0.02	0.004, 0.03	0.07*, 0.009		0.003*, 0.001	0.002*, 0.001	−0.02*, 0.01	0.01, 0.01
MCA Lengths, β, SE	0.64*, 0.21	−0.95*, 0.44	−0.89, 0.57	−0.35, 0.20	1.27*, 0.33		0.05, 0.03	−0.08, 0.23	−0.38*, 0.16

MCA, middle cerebral artery; ATA, anterior temporal artery; PCA, posterior cerebral artery; AComm, anterior communicating artery; BA, basilar artery; β, Beta Coefficient; SE, Standard Error

^*P <0.05

^†Represents the average Z-score of all the measured arteries in the anterior or posterior circulation.

^‡The absent and hypoplastic score represent the sum of all absent or hypoplastic large arteries in the brain.

In the posterior circulation, BA length was associated with presence of fetal PCA (B= 1.1, P=0.005), as well as diameter, height of bifurcation, and lateral deviation of the BA (B= 2.3, P<0.0001; B= 1.3, P<0.001; B=2.8; P<0.001) (Table 4). BA diameter was associated with the diameter of arteries in the anterior circulation (B= 0.15, P<0.001), larger ICV (B= 0.0007; P<0.0001), height of bifurcation of BA (B= 0.2, P<0.0001), and absence of fetal PCA (B= −0.4, P<0.0001). BA tortuosity, assessed here by the degree of lateral deviation, was associated with presence of anterior temporal artery (B= 1.03, P=0.008), MCA and BA lengths (B= 0.12, P=0.04; B= 0.67, P=<0.0001), and BA diameter (B= 2.3, P<0.002), and higher height of BA bifurcation (B= 0.08, P<0.0001). Our findings remained unchanged when adjusting for MRA image quality.

Table 4:

Multivariable Analysis of Anthropomorphic Features Pertaining to the Anterior and Posterior Circulations

	ATA	Fetal PCA	AComm	Anterior Circulation Diameter†	MCA length	BA length	BA diameter	BA height of bifurcation§	BA laterality§
BA diameter, β, SE	−0.009,	−0.41*,	−0.02,	0.15*,	−0.002,	0.01*,		0.21*,	0.02,
	0.02	0.02	0.05	0.02	0.003	0.002		0.02	0.02
BA length, β, SE	−0.37,	1.08*,	−0.19,	0.02,	−0.02,		2.34*,	1.29*,	2.78*,
	0.23	0.38	0.65	0.33	0.03		0.45	0.26	0.22
BA lateral deviation‡, β, SE	1.03*,	0.97,	−0.1,	0.34,	0.12*,	0.67*,	2.30*,	−0.60,
	0.39	0.69	1.05	0.54	0.06	0.05	0.75	0.43
BA laterality§, β, SE	−0.02,	0.03,	−0.14,	0.07,	0.01*,	0.05*,	0.06,	0.08*,
	0.04	0.06	0.09	0.05	0.005	0.005	0.07	0.04
Height of BA bifurcation§, β, SE	0.006,	−0.05,	0.03,	−0.004,	0.006,	0.01*,	0.46*,		0.02,
	0.02	0.04	0.06	0.03	0.003	0.003	0.04		0.02

BA, basilar artery; ATA, anterior temporal artery; PCA, posterior cerebral artery; AComm, anterior communicating artery; MCA, middle cerebral artery; β, Beta Coefficient; SE, Standard Error

^*P <0.05

^†Represents the average Z-score of all the measured arteries in the anterior circulation.

^‡BA laterality was calculated manually by measuring the lateral displacement of the midpoint of the BA at the vertebral junction, mid-pons, and tip.

^§Defined using Smoker’s criteria

Relationship between Arterial Lengths and Diameters and Demographics and Vascular Risks

Larger MCA diameter was associated with increased age and use of hypertensive medications, whereas smaller MCA diameters were associated with smoking, use of statin medications, and elevated low density lipoprotein (LDL) and high density lipoprotein levels at the time of MRI (Table 5). Larger BA diameters were associated with age, hypertension, and greater diastolic blood pressure at the time of MRI, while smaller BA diameters were associated with non-Hispanic Black race/ethnicity. BA length was associated with age, sex, and higher diastolic blood pressure values and greater blood triglyceride levels at the time of MRI. BA lateral deviation was associated with lower LDL levels at the time of MRI.

Table 5:

Multivariable analysis examining association of MCA and BA lengths and diameters with demographics and vascular risk factors

Characteristic	MCA Diameter	MCA Length	BA Diameter	BA Length	BA Lateral Deviation‡	BA Laterality§	BA Height of Bifurcation§
Demographics
Age	0.01*	0.02	0.01*	0.05*	0.05	0.01	0.01*
Male Sex	0.02	1.31*	0.05	1.21*	0.55	0.09	−0.09*
Ethnicity
Hispanic	0.03	0.80	−0.03	−0.07	0.75	0.07	0.11*
Non-Hispanic Black	0.02	1.86*	−0.11*	0.49	1.09	0.13	0.22*
Vascular Risk Factor†
HTN	0.02	0.32	0.10*	0.26	0.60	−0.04	0.07
Duration of HTN	<−0.01	0.02	<0.01	0.01	0.05	<−0.01	<0.01
Antihypertensive use	0.04*	0.15	0.02	0.25	−0.92	<0.01	0.03
SBP	<−0.01	−0.01	<−0.01	−0.01	−0.03	<−0.01	<0.01
DBP	<0.01	<−0.01	<0.01*	0.06*	0.05	0.01	<0.01
DM	<0.01	0.65	−0.03	−0.41	0.41	−0.09	−0.01
Duration of DM	<0.01	0.01	<−0.01	−0.01	0.06	<−0.01	<0.01
Diabetes medication use	−0.02	0.74	<−0.01	0.20	0.06	−0.07	−0.05
Serum glucose	<0.01	−0.01	<0.01	−0.01	<0.01	<−0.01	<−0.01
Hypercholesterolemia	−0.03	−0.51	−0.01	−0.08	−0.09	<0.01	0.03
Statin use	−0.04*	0.11	−0.03	−0.34	−1.42	−0.07	−0.01
LDL level	<−0.01*	<0.01	<−0.01	<−0.01	−0.02*	<0.01	<−0.01
HDL level	<−0.01*	<0.01	<−0.01	0.02	−0.03	<−0.01	<−0.01
Triglyceride level	<−0.01	<−0.01	<−0.01	0.01*	<0.01	<−0.01	<0.01
Current smoker	<0.01	−0.22	0.06	−0.75	−0.16	−0.03	0.04
Duration of smoking	< −0.01*	−0.01	<0.01	−0.02	<0.01	<−0.01	<−0.01

HTN, Hypertension; DM, Diabetes Mellitus; SBP, Systolic Blood Pressure; DBP, Diastolic Blood Pressure; LDL, Low Density Lipoprotein; HDL, High Density Lipoprotein; BA, basilar artery; MCA, middle cerebral artery

^*P <0.05

^†Measurements obtained at the time of MRI

^‡BA laterality was calculated manually by measuring the lateral displacement of the midpoint of the BA at the vertebral junction, mid-pons, and tip.

^§Defined using Smoker’s criteria

In adjusted models now including vascular risk factors, MCA length was associated with presence of anterior temporal arteries (B= 0.78± 0.22, p <0.001) (Table 6). BA diameter was associated with BA length (B= 0.02 ± 0.00, p <0.001), presence of fetal PCA (B=−0.43 ± 0.02, p <0.001), BA lateral deviation (B= 3.02 ± 0.64, p <0.001) and height of bifurcation (B= 0.48 ± 0.04, p=0.007).

Table 6:

Association between length and diameter in the MCAs and BA after adjusting for demographics and vascular risk factors

Predictor	Artery Characteristic	Parameter Estimate ± SE
MCA Diameter	MCA Length
		Model 1	0.00 ± 0.00
		Model 2	0.00 ± 0.00
MCA Length	MCA Diameter
		Model 1	0.96 ± 0.60
		Model 2	0.98 ± 0.12
MCA Length	Presence of ATA
		Model 1	0.77 ± 0.21*
		Model 2	0.78 ± 0.22*
BA Diameter	BA Length
		Model 1	0.02 ± 0.00*
		Model 2	0.02 ± 0.00*
BA Diameter	Presence of Fetal PCA
		Model 1	−0.43 ± 0.02*
		Model 2	−0.43 ± 0.02*
BA Length	BA Diameter
		Model 1	3.16 ± 0.34*
		Model 2	3.09 ± 0.35*
BA Lateral Deviation†	BA Diameter
		Model 1	3.13 ± 0.63*
		Model 2	3.02 ± 0.64*
BA Laterality‡	BA Diameter
		Model 1	0.29 ± 0.06*
		Model 2	0.28 ± 0.06*
BA Height of Bifurcation‡	BA Diameter
		Model 1	0.49 ± 0.04*
		Model 2	0.48 ± 0.04*

SE, standard error; BA, basilar artery; MCA, middle cerebral artery; ATA, anterior temporal artery; PCA, posterior cerebral artery

Model 1: adjusted for demographics, vascular risk factors

Model 2: adjusted for demographics, vascular risk factors, medication use, duration of vascular risk factor, and intensity of risk factor

^*P <0.05

^†BA lateral deviation was calculated manually by measuring the lateral displacement of the midpoint of the BA at the vertebral junction, mid-pons, and tip.

^‡Defined using Smoker’s criteria

Discussion

In this study, we report several cross-sectional associations of brain arterial diameters and length with anatomical, physical, demographic, and clinical factors. We found that the diameters and lengths of various arteries were directly related to one another at the arterial level and across the COW. This interrelatedness could be developmental, with common underlying mechanisms causing vessel dilatation or elongation. Changes in one artery may lead to flow-related adaptations in another artery. Notably, BA diameter correlates with BA length, right and left MCA diameters, while BA length correlates with BA lateral deviation. Similarly, there was a direct relationship between the diameters of both MCAs. In multivariable analysis, MCA length was associated with large anterior circulation diameters, while MCA diameter was associated with large posterior circulation diameters, and MCA and BA lengths. In the posterior circulation, BA diameter is associated with the diameter of arteries in the anterior circulation, and BA length is closely linked to the diameter, height of bifurcation, and lateral deviation of the BA.

Our findings are supported by existing knowledge of arterial dilatation and elongation, which has most extensively been studied in the context of DE. While DE preferentially affects the BA,¹⁵ nearly two thirds of cases have concomitant anterior circulation involvement.^6,16,17 A single center study showed that participants with diffuse intracranial DE, defined by arterial diameter >2 times the mean diameter of a given vascular segment, were older and more likely to have vascular risk factors and signs of systemic arteriopathy compared to participants with vertebrobasilar DE alone.¹⁶ The presence of systemic arteriopathy in participants with diffuse intracranial DE suggests that diffuse intracranial DE may represent a more advanced systemic arteriopathy and/or have a shared etiology with these other vascular beds.¹⁸ While several studies analyzed the presence of anterior and posterior circulation DE using a diameter-based definition, our study looked at elongation, tortuosity, and TCV-adjusted diameter of the involved vessels, thereby providing a more robust assessment of DE. More importantly, we aim to shed light on the common use of arterial cutoffs to define DE, e.g. a BA diameter of >4.5 mm to define BA DE.¹⁹ Without taking into account the expectation of head size, age, and COW configuration, fixed universal arterial cutoffs may increase error and attribute possible pathological significance to larger diameters that appear to be the result of expected compensatory changes. Our study builds on prior studies that were focused mainly on the incidence of concomitant anterior and posterior circulation DE by showing how elongation, dilatation, and tortuosity of each large intracranial artery are correlated.^3,20,21 Our findings should be applied to future clinical studies that define thresholds of dilatation, elongation, and tortuosity that are associated with cerebrovascular risk.

We found an association between COW anatomical variations and arterial dilatation and elongation such as the MCA length with presence of anterior temporal arteries, absent fetal PCA and greater number of hypotrophic arteries or MCA diameter with fetal PCA. These findings are supported by prior studies,⁵ including one by our group showing COW configuration as a determinant of intracranial DE.⁷ We propose that in those with a fetal PCA, the anterior circulation tends to be larger and shorter whether the BA tends to be smaller and shorter. These changes are likely to be the result of the shifting blood volume and associated flow-induced remodeling of those respective vascular systems. The coupling of (brain) demand with (blood) supply is present during the embryological stage of the COW and the brain,²² and it is likely to determine the COW configuration from birth and onward. Based on these findings, we believe that future studies in DE should explore the use of head size, age, and COW configuration-adjusted definitions of “abnormally” dilated brain arteries.

The primary cause of DE is aberrant vascular remodeling that results from increased blood flow causing increased wall shear stress in blood vessels. Over time, this leads to dilatation and remodeling of the affected blood vessels through signal transduction pathways. In DE, this process might become pathological, with breakdown and remodeling of the internal elastic lamina and tunica media leading to weakening instead of strengthening of the arterial walls that are prone to dilatation.²³ Despite the uncertainty of the mechanism, our findings suggest that development or adaptive changes in the COW may lead to (allow for) arterial dilatation and/or elongation. Brain arterial remodeling also occurs with aging and it is further impacted by vascular risk factors.^12,17,24 In our sample, arterial diameter was associated with increased age, whereas smoking was associated with smaller arterial diameters. Both diameters and lengths of the anterior and posterior circulation arteries were associated with markers of hypertension and hypercholesterolemia, including greater diastolic blood pressure, greater cholesterol levels, and use of hypertensives and statin medications.

Our study has several strengths. Our participants were drawn from a diverse population-based community stroke-free sample in New York City. In our study, in addition to using TCV-adjusted diameter, we also included several other factors that give a more comprehensive look at the degree of disease pathology. These data points were systematically and reliably obtained by clinicians who were blinded to the clinical outcomes. The results should be interpreted in the context of the urban, multiethnic, older adult US population represented in NOMAS. Arterial remodeling causing differences in arterial caliber is a chronic process and therefore we lack data points to evaluate repeated diameter or length measurements or the influence of risk factors earlier in life.

Using a stroke-free diverse cohort, we found that the diameter and length of large intracranial arteries were correlated at the arterial level and across the COW, suggesting that changes in one artery may lead to dilatation or elongation in others. Dilatation in the anterior circulation was associated with concomitant changes in the posterior circulation, and vice versa. Posterior circulation DE was associated with absent intracranial arteries in the posterior circulation, while an inverse relationship was found in the anterior circulation. Our findings add to our understanding of anterior and posterior circulation arterial dilatation and elongation and/or tortuosity in diseases such as DE, including the interrelatedness of arterial changes, the need for considering COW configuration in DE, and the role of various demographic and clinical factors. Furthermore, future studies should examine elongation, tortuosity, and dilatation thresholds that are associated with cerebrovascular risk.