Carotid Artery Remodeling is Segment Specific: An in Vivo Study by Vessel Wall Magnetic Resonance Imaging

Abstract

Objective

Early atherosclerosis is often undetected due in part to compensatory enlargement of the outer wall, termed positive remodeling. Variations in hemodynamic conditions and clinical factors influence the patterns of remodeling. The carotid artery provides an opportunity to examine these variations due to the unique geometry of the carotid bulb. This study aimed to determine differences in remodeling of the common, internal and bifurcation segments of the carotid using magnetic resonance imaging (MRI).

Approach and Results

Carotid arteries of 525 subjects without history of cardiovascular disease were imaged by MRI. The carotid artery was divided into three segments: common carotid artery (CCA); bifurcation; and internal carotid artery (ICA). Remodeling patterns were characterized using linear regression analysis of lumen and total vessel areas (dependent variables) compared to maximum wall thickness (independent variable) for each segment, adjusted for age, sex, and height.

The CCA demonstrated a pattern consistent with positive remodeling, while the bifurcation demonstrated negative remodeling. The ICA demonstrated a mixed pattern of outer wall expansion and lumen constriction.

Females and subjects with diabetes showed more positive remodeling, hypertension was associated with attenuated positive remodeling, and those with hypercholesterolemia showed more negative remodeling.

Conclusions

In this cohort of 55–80 year-old individuals without history of cardiovascular disease, the pattern of early carotid artery remodeling was segment specific and appeared to be associated with sex and clinical characteristics. These findings provide the groundwork for longitudinal studies to define local and systemic factors such as hemodynamic and clinical conditions on carotid artery remodeling.

Introduction

In response to early atherosclerotic plaque formation, the arterial wall has the potential for outer wall expansion to preserve luminal area, termed positive remodeling, first described in the coronary arteries by Glagov.¹ Failure to remodel outwardly can lead to luminal narrowing and potential reduction in blood flow.

Factors influencing arterial remodeling are still not well understood. Location appears as an important factor in the pattern of arterial remodeling. Negative remodeling, where there is constriction of the outer wall and lumen area, is observed more frequently in peripheral arteries, compared to renal and coronary arteries.² A possible explanation is the variation in hemodynamic conditions between vascular beds, with triphasic flow in normal peripheral arteries, lower resistance flow patterns in the renal, and flow predominantly occurring during diastole in the coronary arteries.³

Further, studies by ultrasound have shown positive remodeling in the common carotid artery (CCA) but not in the internal carotid artery (ICA).^4–6 However these studies did not examine the bifurcation segment, which is unique from a hemodynamic perspective due to the unusual geometry of the carotid bulb, and where plaque is often first formed. Comparing remodeling patterns between the CCA, the bifurcation, and the ICA segments may provide new insights into the role of hemodynamic factors in atherosclerosis.

Black-blood carotid magnetic resonance imaging (MRI) can precisely measure arterial lumen area, total vessel area, and vessel wall thickness, thereby providing a more comprehensive method to characterize patterns of remodeling in early carotid atherosclerosis.^7–9 Unfortunately, data from population-based studies have rarely examined the relationship between subject demographic and clinical characteristics and remodeling patterns in different segments of the extracranial carotid artery.⁸ In this study, we aimed to characterize patterns of carotid arterial remodeling in the CCA, bifurcation, and ICA segments, and examine how demographic and clinical characteristics influence carotid remodeling patterns in a large cohort of individuals with subclinical atherosclerosis.

Materials AND Methods

Study population

This is a cross-sectional, MRI sub study of the BioImage study (NCT00738725). The overall design and objectives of the BioImage study have been published in detail.¹⁰ In brief, the BioImage Study is a prospective, observational study designed to evaluate associations between imaging and circulating biomarkers, and their ability to predict atherothrombotic events in asymptomatic at-risk subjects. Subjects were recruited using the Humana Health Plan database. Inclusion criteria of the BioImage study are: resident of Chicago, Illinois, or Fort Lauderdale, Florida; 55–80 years of age for males and 60–80 years for females; no history of clinical cardiovascular disease, cerebrovascular disease or peripheral arterial disease, or associated interventions. A subset of 525 subjects underwent MRI of the carotid artery if abnormal IMT was found on the initial ultrasound examination. Abnormal IMT was defined as: 1) presence of carotid plaque, defined as a focal structure that encroaches into the arterial lumen by at least 0.5 mm or 50% of the surrounding IMT value, or 2) a mean thickness of >0.8 mm (55–65 years old), >0.9 mm (65–75 years old), or >1.0 mm (75–80 years old) over the 1.0 cm segment being evaluated. The ultrasound protocol has been previously described.¹¹ The protocol of the BioImage study was approved by the Western Institutional Review Board, Olympia, Washington. All study participants provided written informed consent.

MR imaging

Cross-sectional imaging of bilateral carotid arteries was performed using a 3.0T whole-body MR scanner (Achieva, Philips Healthcare, Best, Netherlands) with a bilateral 8-channel phased array carotid surface coil. Detailed protocol information is provided in Table I (Online-only Data Supplement).

A standardized protocol was used to obtain cross-sectional images of the carotid arteries in multiple contrast weightings: T1-weighted (T1W), T2-weighted (T2W), proton density weighted (PDW), 3D time-of-flight angiography (TOF), and magnetization-prepared rapid acquisition with gradient echo (MP-RAGE). All images were obtained with field-of-view of 160 mm and matrix size of 160 × 160 mm², for an in-plane acquisition resolution of 0.63 × 0.63 mm². Axial images of bilateral carotid arteries were acquired with a 2 mm slice thickness and no inter-slice gap over a longitudinal coverage of 32 mm for T1W and T2W, and 2 mm slice thickness and 1 mm overlap between adjacent slices over a longitudinal coverage of 48 mm for TOF and MP-RAGE. Total scan time was approximately 25 minutes.

Image analysis

Two experienced reviewers (F.L., X.Z.) who were blinded to the subject demographic and the clinical information, analyzed the images with consensus. Image quality was rated per artery on a four-point scale (1 = poor, 4 = excellent). Images with an image quality < 2 were excluded from the study. Different contrast weighted images were registered during image review using the carotid bifurcation as a reference. Scans without a carotid bifurcation within the field-of-view were excluded.

Lumen and total vessel areas of the carotid artery were measured on all axial image slices using CASCADE, a custom-designed image analysis software package (University of Washington, Seattle, Washington, USA).¹² The total vessel area included the lumen, intima, media and adventitia. The maximum wall thickness was derived using the luminal and outer wall boundary contours.

Statistical analysis

The summary statistics of the data are presented as mean ± standard deviation for continuous variables and count with percentage for categorical variables. We classified three segments as previously described:¹³ 1) bifurcation segment, which is 0–6 mm (4 slices) proximal to the carotid bifurcation; 2) CCA segment, which is the 1-cm (5 slices) segment proximal to the bifurcation segment; and 3) ICA segment, which is the 1-cm (5 slices) segment distal to the bifurcation segment (Online-only Data Supplement Figure I). Lumen area, total vessel area, and maximum wall thickness were averaged for each segment. Segments were excluded if 2 or more imaging slices were unavailable.

Linear regression analyses were used to investigate the relationship between a) lumen and b) total vessel areas (dependent variables) and maximum wall thickness (independent variable), a method that has been used to characterize coronary artery remodeling patterns in previous studies.^14,15 Because previous studies have described associations of age, sex, and height with arterial diameters,¹⁶ those factors were adjusted in multivariable models. Additional models with interaction terms between maximum wall thickness and each clinical characteristic were tested to explore whether any clinical characteristics might have influenced the arterial remodeling pattern of each segment. Generalized estimating equations (GEE) were used to account for dependence between different arterial segments of the same subject.¹⁷

The regression model slope parameter (β₁) relating lumen or total vessel wall area and maximum wall thickness was interpreted to indicate whether lumen or total vessel area increased (slope significantly greater than zero) or decreased (slope significantly less than zero) with increasing maximum wall thickness. Figure 1 shows examples of the two remodeling patterns. Pattern A is an example of positive remodeling, where total vessel area increases significantly (β₁> 0) without change in lumen area (β₁≈ 0), as maximum wall thickness increases. Pattern B is an example of negative remodeling, where lumen area decreases significantly (β₁< 0), without change in total vessel area (β₁≈ 0), as maximum wall thickness increases.

Figure 1.

Hypothetical example schematics and graphs of remodeling patterns that may occur with increased wall thickness, as assessed by the linear regression model utilized in this study. Remodeling Pattern A (positive remodeling): Expansion in only the outer boundary without luminal narrowing, characterized by a positive sloping regression line between total vessel area and maximum wall thickness and a relatively flat regression line between lumen area and maximum wall thickness; and Remodeling Pattern B (negative remodeling): Luminal narrowing without compensatory expansion in outer boundary, characterized by a relatively flat regression line between total vessel area and maximum wall thickness and a negatively sloping regression line between lumen area and maximum wall thickness.

Two-sided p< 0.05 was considered statistically significant. All data analyses were performed using STATA ver 13.0 software (StataCorp; College Station, Texas, USA).

Results

Clinical characteristics

Of the 525 scanned subjects, one was a duplicate, 12 were missing clinical information, six were missing MR images; and 16 had insufficient image quality or coverage (Figure 2). Of the remaining 490 subjects, 419 subjects (557 arteries) had sufficient coverage of the CCA segment, 490 subjects (950 arteries) had sufficient coverage of the bifurcation segment, and 481 subjects (899 arteries) had sufficient coverage of the ICA segment. Clinical characteristics of this study are shown in Table 1. Approximately half of the subjects were female and 74% of subjects were Caucasian.

Figure 2.

Flow Chart of Carotid Arteries by Segment

Table 1.

Clinical Characteristics and Dimensions of Each Segment on MRI (N=490 subjects)

Characteristic*
Age (years)	67.7 ± 6.1
Female, n (%)	242 (49.4)
Caucasian, n (%)	363 (74.1)
Current Smoking, n (%)	6 (1.2)
Height (cm)	167.6 ± 9.0
Hypertension, n (%)	235 (48.0)
Taking Hypertension medication, n (%)	119 (24.3)
Hypercholesterolemia, n (%)	286 (58.4)
Cholesterol medication, n (%)	164 (33.5)
Diabetes, n (%)	45 (9.2)
Chronic Kidney Disease, n (%)	72 (14.7)
MRI measurement
CCA
LA (mm2)	39.5 ± 11.5
TVA (mm2)	62.2 ± 15.1
Maximum WT (mm)	1.1 ± 0.2
Bifurcation
LA (mm2)	67.5 ± 26.1
TVA (mm2)	99.8 ± 31.9
Maximum WT (mm)	1.3 ± 0.5
ICA
LA (mm2)	36.4 ± 12.3
TVA (mm2)	58.2 ± 16.0
Maximum WT (mm)	1.1 ± 0.4

CCA, Common Carotid Artery; ICA, Internal Carotid Artery; LA, Lumen Area; TVA, Total Vessel Area; Maximum WT, Maximum Wall Thickness.

^*Data were expressed as means ± standard deviation or number (percentage).

Carotid dimensions and arterial remodeling patterns

Mean lumen area, total vessel area and maximum wall thickness in each segment are shown in Table 1. Carotid arterial remodeling patterns in each segment are shown in Table 2 and Figure 3. In the CCA segment, the total vessel area increased significantly by 17.6 mm² for every millimeter increase of maximum wall thickness (P=0.003). Lumen area did not change significantly (P=0.21). This demonstrates expansion in the outer boundary without luminal narrowing to compensate for atherosclerotic change, consistent with positive remodeling (Remodeling Pattern A in Figure 1). In the bifurcation segment, lumen area decreased significantly by 9.4 mm² for every millimeter increase of maximum wall thickness (P< 0.001) while total vessel area did not change significantly (P= 0.98), consistent with negative remodeling (Remodeling Pattern B in Figure 1). In the ICA segment, lumen area decreased significantly by 3.3 mm² (P=0.007), while total vessel area increased significantly by 5.9 mm² for every millimeter increase of maximum wall thickness (P<0.001). Thus, the remodeling pattern in the ICA segment was mixed with expansion of the outer boundary and constriction of the lumen.¹⁸

Table 2.

Remodeling Patterns for Each Segment

Segment	Lumen Area			Total Vessel Area			Pattern
	β1*	95%CI	P-value	β1*	95%CI	P-value
CCA	4.8	−2.8, 12.4	0.21	17.6†	5.9, 29.2	0.003	Expansion of the outer boundary
Bifurcation	−9.4†	−13.6, −5.2	<0.001	0.1	−5.4, 5.5	0.98	Constriction of the lumen
ICA	−3.3†	−5.6, −0.9	0.007	5.9†	2.6, 9.1	<0.001	Expansion of the outer boundary and constriction of the lumen

^*Slope of regression line between lumen or total vessel area (dependent variable) and maximum wall thickness (independent variable); units are mm² for every millimeter increase in thickness;

^†Lumen area or total vessel area increased (β₁>0) or decreased (β₁<0) significantly with increasing maximum wall thickness.

Figure 3.

Carotid maximum wall thickness compared to both total vessel area (black) and lumen area (red)

CCA: Total vessel area increased significantly with increasing maximum wall thickness while lumen area did not change significantly (positive remodeling pattern; Remodeling Pattern A in Figure 1). Bifurcation: Lumen area decreased significantly with increasing maximum wall thickness while total vessel are did not change significantly (negative remodeling pattern; Remodeling Pattern B in Figure 1). ICA: Lumen area decreased significantly and total vessel area increased significantly with increasing maximum wall thickness (mixed remodeling pattern).

Arterial remodeling with clinical characteristics

Clinical characteristics in various subgroups are shown in Table 3 and remodeling patterns are illustrated by slope parameters (β₁). No significant difference in remodeling patterns occurred in Caucasian versus non-Caucasian, or in subjects with or without chronic kidney disease. However, sex was significantly associated with the degree of remodeling. Total vessel area increased significantly by 12.3 mm² in females compared to 4.3 mm² in males for every millimeter increase of maximum wall thickness in the ICA segment (difference between slopes, P=0.02). The lumen area did not change significantly in women, but decreased significantly by 4.5 mm² in men for every millimeter increase in maximum wall thickness in the ICA segment (difference between slopes, P=0.01).

Table 3.

Remodeling Patterns with Clinical Characteristics by Each Segment

		Lumen Area			Total Vessel Area
Clinical Characteristic	Segment	Female	Male	P-Value*	Female	Male	P-Value*
Sex	CCA	8.7†	3.8	0.86	24.9†	15.6†	0.26
	Bifurcation	−8.6†	−9.8†	0.76	0.1	0.1	0.99
	ICA	2.1	−4.5†	0.01	12.3†	4.3†	0.02
		Caucasian	Non Caucasian	P-Value*	Caucasian	Non Caucasian	P-Value*
Race	CCA	8.6	−0.5	0.15	23.1†	9.7	0.90
	Bifurcation	−8.8†	−10.8†	0.62	0.8	−2.2	0.53
	ICA	−2.6	−5.8†	0.24	6.9†	1.8	0.18
		Present	Absent	P-Value*	Present	Absent	P-Value*
Hypertension	CCA	0.6	12.8†	0.02	11.2†	29.4†	0.02
	Bifurcation	−8.9†	−10.0†	0.79	0.8	−1.1	0.71
	ICA	−3.8†	−2.0	0.42	5.1†	7.6†	0.40
Hyper-cholesterolemia	CCA	5.9	3.8	0.78	19.8†	15.2	0.68
	Bifurcation	−14.6†	−4.6	0.002	−6.5†	6.1	0.003
	ICA	−1.5	−4.9†	0.13	7.8†	4.0†	0.20
Diabetes	CCA	9.9†	4.4	0.39	25.1†	16.9†	0.37
	Bifurcation	−2.5	−10.4†	0.03	9.3†	−1.3	0.01
	ICA	9.0†	−3.8†	0.03	22.8†	5.2†	0.03
Chronic Kidney Disease	CCA	5.7	4.7	0.90	22.6†	17.3†	0.61
	Bifurcation	−5.6	−10.7†	0.35	4.9	−1.5	0.37
	ICA	−3.5	−3.2†	0.93	7.3†	5.6†	0.62

Data were expressed as β₁ (slope of regression line) between mean lumen or total vessel area (dependent variable) and maximum wall thickness (independent variable), adjusted for age, sex, and height; units are mm² for every millimeter increase in thickness;

^*Test for difference in β₁ (slope of regression line) between the two groups;

^†Lumen area or total vessel area increased (β₁>0) or decreased (β₁<0) significantly with increasing maximum wall thickness.

Presence or absence of hypertension, hypercholesterolemia and diabetes were also associated with differences in patterns of remodeling. Subjects without hypertension demonstrated significant increases in total vessel area (29.4 mm²) for every millimeter increase in maximum wall thickness in the CCA segment. However, this pattern of positive remodeling was attenuated in subjects with hypertension (11.2 mm², difference between slope, P=0.02). Lumen area increased significantly by 12.8 mm² for those without hypertension, but did not change significantly for those with hypertension for every millimeter increase in maximum wall thickness (difference between slope, P=0.02).

A pattern consistent with negative remodeling was observed in subjects with hypercholesterolemia in the bifurcation segment. Total vessel area decreased significantly by 6.5 mm² for each millimeter increase in maximum wall thickness in subjects with hypercholesterolemia, but no significant change was noted for those without hypercholesterolemia (difference between slope, P=0.003). Further, lumen area decreased significantly by 14.6 mm² in subjects with hypercholesterolemia, but no significant lumen area change was noted for those without hypercholesterolemia (difference between slope, P=0.002).

A history of diabetes was associated with positive remodeling patterns, compared to those without diabetes. In the bifurcation segment, the total vessel area increased significantly by 9.3 mm² in subjects with diabetes, but no significant change was noted in subjects without diabetes (difference between slope, P=0.01). Lumen area did not change significantly in those with diabetes, but decreased significantly by 10.4 mm² in subjects without diabetes with increasing in maximum wall thickness (difference between slope, P=0.03).

Discussion

In this study, we sought to determine whether the pattern of carotid artery remodeling was segment specific and associated subject demographic and clinical characteristics in a large cohort of individuals with subclinical atherosclerosis. Our results show that patterns of remodeling differ by carotid arterial segment: the CCA demonstrated a pattern consistent with positive remodeling, while the bifurcation demonstrated negative remodeling. The ICA demonstrated a mixed pattern of outer wall expansion and lumen constriction. The CCA and ICA findings are consistent with those reported by Aster et al in a 2010 study that showed that the CCA had greater compensatory remodeling capacity than the ICA.⁸ To our knowledge, ours is the first study that characterizes the segment-specific carotid arterial remodeling pattern in three distinct arterial segments (CCA, bifurcation and ICA) in a large cohort of subjects with subclinical atherosclerosis. Additionally, we found several potential associations between clinical risk factors and remodeling patterns. Specifically, female sex and history of diabetes were associated with a pattern consistent with more positive remodeling, a history of hypertension was associated with a pattern of attenuated positive remodeling, and hypercholesterolemia was associated with a negative remodeling pattern.

Several studies have shown that intima-media thickness (IMT) in the carotid artery was positively correlated with interadventitial and lumen diameters in the CCA segment,^5,6,19 consistent with our findings of increasing lumen and total vessel areas with increasing wall thickness in the CCA. Terry et al.⁴ also reported that the common carotid interadventitial and lumen diameters were larger with increasing IMT, and internal carotid lumen diameter was smaller with greater IMT. However, internal carotid interadventitial diameter was not associated with IMT,⁴ contrary to our findings where the outer boundary in the ICA segment demonstrated expansion with increasing wall thickness. Of note, the majority of earlier studies examining carotid remodeling patterns involved ultrasound imaging which may be influenced by a variation in incident angle of the ultrasound beam.²⁰ MRI has been shown to be highly reproducible for measuring the size and composition of atheroma in the extracranial carotid arteries.^21,22

An explanation for the specific remodeling pattern in different segments may be related to differences in hemodynamics^23,24 and vessel wall composition.²⁵ A pattern of significant expansion in the outer vessel wall boundary with increasing wall thickness was present in the common carotid. The CCA has a dense network of elastin fibers, which are distributed through the media and organized in layers.²⁵ Pulse pressure related cyclic stretch can lead to increased mechanical stress, resulting in dilation of large elastic arteries such as the common carotid.²⁶ Disruption of the large elastic fibers due to aging or plaque formation may also cause outer wall expansion in the CCA. In the ICA segment, we observed expansion in the outer boundary and constriction in lumen area with increasing wall thickness. The vessel wall in the ICA segment is a more muscular structure compared to the CCA segment,²⁵ which may affect the capacity of the vessel segment for compensatory remodeling due to increase in vascular smooth muscle mass with aging and effects of blood pressure.²⁷

The unique geometry of the carotid bulb and branching of the common carotid artery can cause disturbed, non-uniform flow patterns in the bifurcation segment. This unique geometry alters flow in all directions, creating vortices, lower and oscillatory hemodynamic wall shear stress, and endothelial cell barrier dysfunction.^{23–25,28,29} Decreased hemodynamic wall shear stress can induce increasing low-density lipoprotein uptake, inflammation, and produce a large necrotic core.^24,30–32 Lower hemodynamic shear stress has been shown to induce a decrease in internal vessel radius,³⁰ similar to our findings where the lumen area in the bifurcation segment shows constriction.

Our results suggest that the female gender is associated with more expansion in lumen and outer boundary with increasing maximum wall thickness in the ICA segment, compared to males. Previous studies have been reported that sex differences influence plaque size, arterial size, and the degree of stenosis.^33,34 To our knowledge, a difference in remodeling patterns between men and women has not been reported. Subjects with hypertension and hypercholesterolemia demonstrated more constriction in the lumen and outer vessel wall boundary compared to those without, as has been previously reported.^5,35 On the other hand, subjects with diabetes exhibited more expansion in the lumen and outer vessel wall boundary in both the bifurcation and ICA segments compared to those without, as has been reported in coronary^15,36,37 and carotid arteries.³⁵ One might speculate that this occurs because diabetes may be associated with more vascular wall inflammation, and positive remodeling has been histologically associated with inflammation.³⁸ While it has been reported that smoking influences remodeling patterns in carotid arteries,³⁵ we did not assess smoking in our analysis due to the small number of current smokers (six subjects).

Limitations

Our study has potential limitations. First, due to lack of sufficient imaging coverage, all three segments of the carotid artery were not completely imaged bilaterally in all the subjects. Second, the tortuous nature of the internal carotid artery in some cases may result in partial volume effects in the axial direction. More recently developed large-coverage, 3D isotropic carotid MRI techniques will permit coverage of the extracranial carotid artery bilaterally, with reconstruction perpendicular to the center line in future studies.³⁹ Third, our analysis correlating clinical characteristics with remodeling parameters involved a large number of statistical comparisons and we did not perform any P-value adjustments (e.g., Bonferroni correction). Therefore, this analysis should be considered more exploratory and hypothesis- generating and the corresponding results should be confirmed in subsequent studies. And last, findings from this cross sectional study will require corroboration in prospective longitudinal assessment with serial imaging.

Conclusions

In this large cohort study of subjects with subclinical atherosclerosis, we found that early carotid artery remodeling differs between the common carotid, carotid bifurcation, and internal carotid segments. This finding corroborates the observation that patterns of remodeling vary not only by the arterial bed (cerebrovascular, coronary, peripheral), but also within arterial segments, indicating a role of hemodynamics and vessel wall composition in arterial remodeling. Our data also showed that the early carotid artery remodeling patterns appear to be associated with sex and clinical characteristics. These findings provide the basis for future studies that incorporate serial imaging and flow modeling to study the potential role local (hemodynamic, histopathologic) and systemic determinates of carotid artery remodeling.

Highlights.

• Early carotid arterial remodeling differs by segment, with a pattern of positive remodeling observed in the common carotid artery, negative remodeling in the carotid bifurcation, and a mixed pattern of outer wall expansion and lumen constriction in the internal carotid artery.

• Early carotid artery remodeling patterns appear to be influenced by sex and clinical characteristics.

• These findings suggest that local (e.g. hemodynamic, histopathologic) and systemic factors influence patterns of remodeling.

Abbreviations

ICA

internal carotid artery

CCA

common carotid artery

MRI

magnetic resonance imaging

IMT

intima-media thickness