Assessment of Complex Flow Patterns in Patients With Carotid Webs, Patients With Carotid Atherosclerosis, and Healthy Subjects Using 4D Flow MRI

Abstract

Background:

Carotid webs (CaWs) are fibromuscular projections in the internal carotid artery (ICA) that cause mild luminal narrowing (<50%), but may be causative in up to one-third of seemingly cryptogenic strokes. Understanding hemodynamic alterations caused by CaWs is imperative to assessing stroke risk. Time-Average Wall Shear Stress (TAWSS) and Oscillatory Shear Index (OSI) are hemodynamic parameters linked to vascular dysfunction and thrombosis.

Purpose:

To test the hypothesis: “CaWs are associated with lower TAWSS and higher OSI than mild atherosclerosis or healthy carotid bifurcation.”

Study Type:

Prospective study.

Population:

A total of 35 subjects (N = 14 bifurcations with CaW, 11F, age: 49 ± 10, 10 mild atherosclerosis 6F, age: 72 ± 9, 11 healthy 9F, age: 42 ± 13).

Field Strength/Sequence:

4D flow/STAR-MATCH/3D TOF/3T MRI, CTA.

Assessment:

4D Flow velocity data were analyzed in two ways: 1) 3D ROI in the ICA bulbar segment (complex flow patterns are expected) was used to quantify the regions with low TAWSS and high OSI. 2) 2D planes were placed perpendicular to the centerline of the carotid bifurcation for detailed analysis of TAWSS and OSI.

Statistical Tests:

Independent-samples Kruskal–Wallis-H test with 0.05 used for statistical significance.

Results:

The percent surface area where low TAWSS was present in the ICA bulb was 12.3 ± 8.0% (95% CI: 7.6–16.9) in CaW subjects, 1.6 ± 1.9% (95% CI: 0.2–2.9) in atherosclerosis, and 8.5 ± 7.7% (95% CI: 3.6–13.4) in healthy subjects, all differences were statistically significant (ƞ² = 0.3 [95% CI: 0.05–0.5], P-value CaW vs. healthy = 0.2). OSI had similar values in the CCA between groups (ƞ² = 0.07 [95% CI: 0.0–0.2], P-value = 0.5), but OSI was significantly higher downstream of the bifurcation in CaW subjects compared to atherosclerosis and normal subjects. OSI returned to similar values between groups 1.5 diameters distal to the bifurcation (ƞ² = 0.03 [95% CI: 0.0–0.2], P-value = 0.7).

Conclusion:

Lower TAWSS and higher OSI are present in the ICA bulb in patients with CaW when compared to patients with atherosclerotic or healthy subjects.

Evidence Level:

2

Technical Efficacy:

Stage 2

Atherosclerosis is considered to be the main etiology of stroke and is characterized by plaque accumulation resulting in arterial wall thickening and stenosis formation.¹ An arterial wall injury can cause plaque cap rupture, resulting in thrombus formation, and eventually distal vessel occlusion. Vessel stenosis is not limited to atherosclerosis but may also be due to other etiologies including fibromuscular dysplasia (FMD). Certain atypical FMD lesions appear exclusively in the carotid as triangular-shaped or shelf-like projections into the internal carotid artery (ICA) bulbar segment lumen and are commonly known as carotid artery webs (CaWs) (Fig. 1). CaWs are not related to atherosclerotic plaques and generally cause <50% stenosis.² Atherosclerosis lesions with <50% stenosis are usually asymptomatic and are managed medically, without surgical intervention.³ However, studies have shown that despite having <50% stenosis, CaWs have a strong correlation with transient ischemic attacks (TIAs) and strokes, particularly affecting younger patients.^2,4,5 One study found that young subjects (<60 years) with cryptogenic stroke demonstrated 9.4%–37% prevalence of CaW ipsilateral to the stroke.⁶ Additionally, CaWs have been shown to have a prevalence of 1.2%–2.5% of all subjects presenting with acute stroke.^4,6,7 According to the European Carotid Surgery Trial (ECST) subjects with atherosclerosis and CaW with <70% stenosis are not candidates for surgical intervention.³ It is important to identify those at high risk for stroke to improve patient clinical management. This is particularly important in the setting of an incidentally discovered CaW.

FIGURE 1:

Black-blood image of a patient with Carotid Web. Red arrow indicates Carotid web location.

Recent studies suggest that the composition of emboli forming due to CaW differs from other arterial clots.^8–10 In particular, CaW-associated thrombi obtained after mechanical thrombectomy show a higher content of red blood cells on histological analysis.^9–11 Other arterial clots, such as those related to atherosclerosis, contain a higher percentage of platelets. The shape of the CaWs induces alteration of blood flow leading to stagnation in the bulbar segment of the ICA.^12–14 Together, the presence of flow stagnation and the presence of large numbers of red blood cells (RBCs) in the clots suggest the formation of low shear thrombus (red clots) distal to the web.^8–10

Three-dimensional, three-directional, ECG-gated, time-resolved, phase-contrast magnetic resonance imaging (4D Flow MRI) has shown the ability to measure complex blood flow.^15–19 Low time-averaged wall shear stress (WSS) and high oscillatory shear index (OSI) have been linked to disturbed blood flow and vascular dysfunction, two requirements for thrombus formation.^20,21 WSS is the tangential force per unit area that is exerted on the endothelial cells by blood flow.²² Since WSS varies across the cardiac cycle, the time average WSS (TAWSS) is determined as the temporal mean over the cardiac cycle in a single location at the wall.^15,23,24 Additionally, to understand the variation of WSS due to pulsatile flow within a cardiac cycle, OSI can be calculated. These hemodynamic parameters have been quantified in the literature in the aorta,^15,24 normal carotid artery,^18,23,25 and carotid artery with atherosclerosis.²⁶

On the proximal side of the CaW, its shape will cause flow acceleration, while distal to the CaW, there is a sudden expansion in the lumen that creates an unstable, separated, vortical flow pattern that gives rise to a large recirculation region, including areas of near stasis. Therefore, larger complex flow regions are expected in patients with CaWs compared to healthy carotid bifurcation or atherosclerosis lesions (Fig. 2). Depending on the severity of the luminal narrowing, varying degrees of flow separation can cause stagnation regions that facilitate thrombosis formation.^12,14 Individuals with disturbed flow patterns may be at higher risk of stroke due to this phenomenon.

FIGURE 2:

Artist depiction of streamlines and velocity profile in (a) normal carotid bifurcation compared to (b) a carotid bifurcation with an atherosclerotic lesion (c) a carotid bifurcation with a Carotid web (CaW) lesion. The larger recirculation region in patients with CaWs due to the sudden expansion.

The aim of this study is to utilize 4D flow MR to characterize the altered hemodynamics produced by CaWs and compare it to patients with mild atherosclerosis and healthy subjects. The work hypothesizes that CaWs are associated with lower TAWSS and higher OSI than mild atherosclerosis or healthy carotid bifurcations.

Material and Methods

Patient Population

The prospective study was approved by IRB and all subjects enrolled in the study were provided informed consent. The inclusion criteria of subjects with CaW included a consensus read of a prior computed tomography angiography (CTA) scan confirming CaW diagnosis, while the exclusion criteria included a history of prior carotid surgical intervention or the presence of devices incompatible with MRI. The same exclusion criteria were used for subjects with atherosclerosis, while the inclusion criteria included luminal narrowing of mild to moderate measured using the ECST criteria.³ Additionally, healthy subjects (aged matched to CaW subjects) were recruited as a control group.

MRI Scans

All subjects were imaged in a 3 Tesla MRI scanner (Siemens PrismaFit, Erlangen Germany) using a 4-channel (Machnet BV, the Netherlands) neck coil in combination with a 20-channel head/neck coil. Multi-slab, transverse, 3D time of flight (TOF) images were acquired to cover the bifurcation (field of view (FOV) = 151 × 200 mm², isotropic spatial resolution = 0.5 mm³, repetition time (TR) = 23 msec, echo time (TE) = 3.1 msec, flip angle (FA) = 20°, GRAPPA acceleration factor = 2, bandwidth = 250 Hz/Px, acquisition matrix = 384 × 203, number of slabs = 6, number of slice/slab = 24, scan time = 5.20 minutes). 4D flow MRI scans were acquired in a parasagittal orientation planned in the plane of the bifurcation with the following parameters: FOV = 162 × 200 mm², isotropic spatial resolution = 1 mm³, FA = 7°; TR = 47 msec; TE: 5 msec, GRAPPA acceleration factor = 2, bandwidth = 490 Hz/Px, acquisition matrix = 192 × 108, number of slice/slab = 16, velocity encoding (VENC) = 60–80 cm/s in all directions, number of k-space segments = 2, scan time = 4.83 minutes and all images were retrospectively reconstructed with 24 timeframes over the cardiac cycle. Black-blood images for each patient were acquired using a prototype STAR-MATCH (Multi-contrast Atherosclerosis CHaracterization) imaging sequence without intravenous contrast in the coronal plane with anisotropic spatial resolution of 0.8 mm³ isotropic resolution (TR/TE = 1884.16/2.44 msec, FOV = 160 × 160, number of slices = 64, base resolution = 192 with 384 radial views, bandwidth = 441 Hz/Px, FA = 8°, scan time = 12.1 minutes).²⁷ CaW were confirmed with a previously acquired, clinically indicated CTA scan acquired prior to enrollment in the study. JWA (a neuroradiologist with 16 years post-fellowship experience) and DCH (neurointerventionalist with 9 years post-fellowship experience) conducted the CTA consensus read.

4D Flow MRI Processing

The 4D flow MRI data was analyzed using a custom-written MATLAB code and EnSight (2021R1, ANSYS, Inc, PA, USA) software.²⁸ A MATLAB (R2022b MathWorks, Inc, MA, USA) program was used to correct for eddy currents and aliasing, perform noise filtering,^16,29,30 and calculate a 3D magnetic resonance angiogram (MRA) from the 4D flow MRI data as the mean of the sum of squares of the three-directional velocity encoding images (Fig. 3a). The carotid bifurcation was segmented using the combined velocities MRA image loaded into Materialise MIMICS (version 22.0.0.524, Belgium) (Fig. 3b). A clinical CTA scan, black-blood STAR-MATCH MRI scan and MRA images were used to help with visualizing low velocities during segmentation (Fig. 3c). Based on the MRA segmentation, an EnSight file and mask were generated by RES (5 years’ experience) (Fig. 3d). EnSight was used for the visualization of velocity streamlines/pathlines and the extraction of 2D cross-sections at different locations along the carotid bifurcation (Figs. 3d and 4). The TAWSS heat maps in the carotid bifurcation and the bulbar segment were also generated using MATLAB program previously published (Fig. 3e).³¹

FIGURE 3:

(a) 4D flow MRI sequence: 3D magnitude, with 3D velocity encoding directions, acquired across multiple time points in the cardiac cycle. (b) The combined velocities directions into a magnetic resonance angiogram (MRA) image. (c) The segmentation of the MRA image, in Carotid web cases, where patients undergone a computed tomography angiography scan. (d) Particle pathlines velocity magnitude color-coded in a carotid bifurcation. (e) Time-average wall shear stress heat map of the carotid bifurcation with the bulbar segment region analyzed.

FIGURE 4:

(a) Shows a computed tomography angiography image of a patient with Carotid web. The CCA diameter and maximum stenosis location were measured. (b) Shows the 2D cross-sectional planes inserted perpendicular to the centerline of the vessel one in CCA and three in the ICA located 0.5D, 1D, and 1.5D above maximum stenosis. (c) Shows an example of extracted planes with manual ROI segmentation for the calculation of wall shear stress.

Low TAWSS and High OSI Area Quantification in the ICA Bulb

For each case, particle pathlines were generated, which qualitatively showed larger regions of recirculating helical velocity patterns in the ICA of patients with CaW compared to atherosclerotic lesion patients and normal subjects (red arrow in the left of Fig. 5a). TAWSS, OSI, low TAWSS, and high OSI heatmaps were generated over the entire computational domain for all patients. WSS was calculated based on previously published methods in the literature.^22,32 A spline was fitted to the three velocity values along the inward normal vector from the wall. The derivative of the spline curve at the vessel wall is proportional to WSS.³² The WSS was calculated over all the cardiac time frames, then averaged to calculate the TAWSS (Fig. 3e).³¹ OSI is an index of the temporal variation in arterial WSS due to pulsatile flow within a cardiac cycle, which was calculated using the definition based on the model developed by He and Ku.³¹ OSI index ranges from 0 to 0.5, where 0 indicates no oscillation in WSS throughout the cardiac cycle and 0.5 indicates purely oscillatory WSS with a zero mean value.³¹

FIGURE 5:

Selected cases from the patient population. (a) Shows a carotid bifurcation from a patient with Carotid web, (b) a patient with mild atherosclerosis, and (c) a carotid bifurcation from a normal subject. Velocity pathlines, time-average wall shear stress, Oscillatory Shear Index (OSI), low wall shear stress, and high OSI Heat maps.

A segmentation was done to create a specific region of interest in the ICA bulbar region where complex flow is expected (Fig. 3e). The region of interest (herein ICA bulb) was specified to be the volume between the maximum stenosis created by the web or atherosclerotic lesions, to a location 1.5 CCA diameters downstream of the maximum stenosis or bifurcation in the case of normal subjects.³³ This region was chosen since it is where complex blood flow is expected based on previous studies.³⁴ In this region, the area exposed to low TAWSS and high OSI was determined. Low TAWSS threshold was defined as the lowest 10% of WSS magnitude values based on all subjects’ (CaW, atherosclerosis, healthy) combined data; high OSI was defined as the highest 10% of all OSI values over all subjects.¹⁸ The area of low TAWSS and high OSI was then quantified in each subject’s ICA bulb by dividing the number of pixels below/above the threshold by the total number of pixels in the ICA bulb.

TAWSS and OSI Quantification in 2D Cross-Sections in the CCA and ICA

Two dimensional cross-sectional planes were sampled from the 4D flow domain. For each subject’s segmentation, the centerline of the vessel, the diameter of the CCA, the bifurcation location, and the maximum stenosis location were determined. Four 2D cross-sectional planes were analyzed, one plane was located in the CCA, and three planes were located in the ICA. The 2D cross-sectional planes were placed perpendicular to the centerline of the vessel segmentation, based on the bifurcation location or the maximum stenosis location and the CCA diameter (Fig. 4a). In patients with CaWs and patients with atherosclerotic lesions, the cross-sections were located at one CCA diameter below the bifurcation (CCA), 0.5 CCA diameter (ICA1), 1.0 CCA diameter (ICA2), and 1.5 diameters (ICA3) above the maximum stenosis (Fig. 4b). In normal subjects, a similar method was used to extract the cross-sections but based on the carotid bifurcation location. The plane in the CCA was chosen as a representative of a laminar, nearly parabolic flow profile in a cylindrical tube.³⁴ The multiple planes in the ICA were chosen to evaluate flow in areas where complex flow patterns were expected.³⁴

The analysis of the 2D planes was conducted based on a previously developed method.²⁹ This analysis was conducted using MATLAB code which includes: ROI segmentation of the 2D planes, and the quantification of different hemodynamics parameters including TAWSS, and OSI.²⁹ The WSS was calculated using the derivative of the local velocity vector field based on the deformation at the vessel lumen, or the segmented contour. Due to the limited spatial resolution, the data was interpolated using cubic B-spline, where the B-spline interpolation model provided continuous velocity and derivatives on the vessel contour.²⁹ TAWSS and OSI were quantified in a similar method discussed above. TAWSS was averaged across all the subjects and four cross-sections (group average) to understand the difference across the subjects’ population. OSI is expected to have a higher variation in certain 2D cross-sections distal to the region of maximum stenosis, so this analysis was done to fully understand the spatial the temporal variation of hemodynamics associated with the pulsatility of the cardiac cycle.

Interobserver Agreement

To assess the variation in segmentation between observers, the 2D analysis was done by two observers for 10 subjects for a total of 40 segmentations over the subject groups. Both observers have had extensive segmentation training RES (>5 years’ experience) ZS (>3 years’ experience).

Statistical Analysis

All numbers were reported as mean ± standard deviation. Statistical analysis was performed using IBM SPSS Statistics software (v.29 Chicago, SPSS Inc). Independent-samples Kruskal–Wallis test was used as a nonparametric test to determine the statistical significance by pairwise comparisons of group type and 0.05 was used as a significance level. The lower and upper bond of 95% confidence interval for mean (CI) was computed as well. One way ANOVA was used for group wise comparison and the quantification of effect size (Eta-squared [ƞ²]). Additionally, intraclass correlation coefficient (ICC) was used to understand the interobserver agreement.

Results

The study cohort included patients with CaWs (N = 14, 11F, age 49 ± 10 years, luminal narrowing of 29 ± 9%) and history of stroke/TIA, patients with mild atherosclerosis with a similar degree of stenosis (N = 10, 6F, age 72 ± 9, luminal narrowing of 39 ± 9%), and healthy volunteers (N = 11, 9F, age 42 ± 13 years).

Low TAWSS and High OSI Area Quantification in the ICA Bulb

The low TAWSS and high OSI regions based on the 10% thresholds were found to be 0.08 N/m² and 0.43 (scale of 0–0.5), respectively. Figure 5a–c shows selected cases of a patient with CaW, a patient with atherosclerosis, and a healthy bifurcation, respectively. The heat maps of TAWSS, OSI, Low TAWSS, and high OSI were also shown in each of the subjects.

Figure 6a shows the results of the percent area exposed to low TAWSS, while Fig. 6b shows the results of the percent area exposed to high OSI comparing patients with CaW to those with atherosclerotic lesions and healthy subjects. Patients with CaW have a larger region exposed to low TAWSS compared to patients with atherosclerosis (12.3 ± 8.0% [95% CI: 7.6%–16.9%] vs. 1.6 ± 1.9% [95% CI: 0.2%–2.9%]) or compared to healthy subjects (8.5 ± 7.7% [95% CI: 3.6%–13.3%], ƞ² = 0.3 [95% CI: 0.05–0.5], P-value = 0.2), all other comparisons were statistically significant (Fig. 6a).

FIGURE 6:

The area of Disturbed Complex Flow Patterns is represented by a box and whisker plot. (a) Shows the area percent of the carotid bifurcation that is exposed to low time-average wall shear stress values in three subject groups (all comparisons were statistically significant). (b) Shows the area percent that is exposed to high Oscillatory Shear Index in all subjects’ groups. Asterisk (*) shows statistically significant comparisons.

Patients with CaW have significantly larger regions of high OSI compared to patients with atherosclerosis (9.9 ± 3.5% [95% CI: 7.9%–11.9%] vs. 6.2 ± 4.9% [95% CI: 2.7%–9.7%]), but the difference between CaW subjects and healthy subjects was not statistically significant (8.4 ± 5.0% [95% CI: 6.7%–10%], ƞ² = 0.1 [95% CI: 0–0.3], P-value: 0.12).

TAWSS and OSI Quantification Based on 2D Cross-Sections

The interobserver agreement showed an overall good reliability in the ICC measurement. The interobserver agreement based on 2D segmentations in the TAWSS measurement resulted in ICC = 0.86 (95% CI: 0.7–0.9) showing moderate to excellent reliability. In OSI measurements ICC = 0.85 (95% CI: 0.4–0.9) resulting in good reliability. TAWSS were averaged across all patients in each of the subject groups in each of the cross-sections (group average) (Table 1). The results were statistically significant when comparing CaW subjects to patients with atherosclerosis at all four cross-sections (ƞ² = 0.3 [95% CI: 0.2–0.4], 0.3 ± 0.1 N/m² [95% CI: 0.27–0.34] vs. 0.6 ± 0.3 N/m² [95% CI: 0.5–0.7]) and to normal subjects (0. 4 ± 0.1 N/m² [95% CI: 0.3–0.4]). When comparing normal subjects with patients with atherosclerosis, TAWSS values were also significant (0.4 ± 0.1 N/m² vs. 0.6 ± 0.3 N/m²) (Table 1).

TABLE 1.

TAWSS Results in Different Subject Groups in Cross-Sections CCA, ICA1, ICA2 & ICA3 and Across the Group Population and the Cross-Sectional Planes (Group Average)

TAWSS (N/m2)
Cases/Plane Locations	CCA	ICA1	ICA2	ICA3	Group Average
CaW	0.39 ± 0.12	0.21 ± 0.07	0.27 ± 0.09	0.37 ± 0.13	0.31 ± 0.12
Atherosclerosis	0.59 ± 0.20	0.67 ± 0.40	0.64 ± 0.34	0.62 ± 0.33	0.63 ± 0.32
Normal	0.52 ± 0.17	0.28 ± 0.07	0.34 ± 0.10	0.40 ± 0.11	0.39 ± 0.14
CaW vs. Atherosclerosis (P-value)	0.01 *	0.00 *	0.00 *	0.01 *	0.00 *
CaW vs. normal (P-value)	0.05 *	0.09	0.09	0.70	0.01 *
Normal vs. Atherosclerosis (P-value)	0.44	0.00 *	0.05 *	0.02 *	0.00 *

All results were averaged across the subject group with standard deviation indicating the variation in each subject group.

BOLDED* indicates statistically significant results. ICA1, ICA2, ICA3: 0.5, 1.0, 1.5 CCA diameter downstream max stenosis or bifurcation, respectively. CaW = Carotid web; TAWSS = time-average wall shear stress.

OSI results are shown across different cross-sections in the three subject groups in Fig. 7. OSI between all three groups was not statistically different at the cross-sections in CCA (proximal) (ƞ² = 0.68 [95% CI: 0–0.2], CaW: 0.02 ± 0.02 [95% CI: 0.01–0.03], atherosclerosis: 0.01 ± 0.01 [95% CI: 0.005–0.02], normal: 0.01 ± 0.01 [95% CI: 0.003–0.02] | P-value: 0.5) and ICA3 (distal) (ƞ² = 0.03 [95% CI: 0–0.2], CaW: 0.03 ± 0.02 [95% CI: 0.01–0.04], atherosclerosis: 0.02 ± 0.02 [95% CI: 0.005–0.04], normal: 0.01 ± 0.01 [95% CI: 0.02–0.01] | P-value: 0.7). However, OSI in patients with CaW was significantly higher compared to patients with atherosclerosis and normal subjects in ICA1 cross-sections (ƞ² = 0.4 [95% CI: 0.1–0.6], CaW: 0.09 ± 0.04 [95% CI: 0.07–0.12], atherosclerosis: 0.03 ±0.04 [95% CI: 0.003–0.06], normal: 0.06 ± 0.02 [95% CI: 0.05–0.08]) and in ICA2 (ƞ² = 0.3 [95% CI: 0.1–0.5], CaW: 0.06 ± 0.031 [95% CI: 0.04–0.08], atherosclerosis: 0.02 ± 0.02 [95% CI: 0.007–0.04], normal: 0.03 ± 0.01 [95% CI: 0.03–0.05] | CaW vs. normal P-value = 0.07).

FIGURE 7:

Oscillatory Shear Index results in patients with Carotid web, atherosclerosis, and normal carotid bifurcation in CCA, ICA1, ICA2 & ICA3 represented by a bar plot. Statistical significance observed in ICA1 and ICA2 is represented by an asterisk (*).

Discussion

CaWs have been linked to TIA and ischemic stroke in younger patients.^2,4,5,13 This study utilized 4D flow MRI to quantify the parameters in patients with CaW to help understand the hemodynamic differences between subjects with CaW, subjects with mild atherosclerosis, and healthy subjects. The study’s major findings were: 1) ICA bulb region in patients with CaW has a larger region exposed to low TAWSS and high OSI compared to patients with mild atherosclerosis or normal subjects, 2) CaW subjects had higher OSI in cross-sections from the web to 1.0 CCA diameters distal to the lesion compared to patients with mild atherosclerosis and normal bifurcations; but in the CCA and at cross-sections 1.5 diameters downstream of the lesion, all subjects had similar OSI values; 3) Based on the group average of the 2D cross-sectional analysis, subjects with CaW had lower TAWSS compared to subjects with mild atherosclerosis and normal subjects (Table 1). The results indicate that OSI is similar before the web and 1.5 diameters after the web among all groups, but OSI is significantly higher in the region just distal to the web. The high OSI and low TAWSS in the ICA of subjects with CaW may predispose the region to clot formation. These factors indicate that the unique CaW geometry introduces a higher degree of blood flow disturbance represented by lower TAWSS, higher OSI, and larger regions exposed to low TAWSS and high OSI.

The findings of this study may help to lay the ground-work to explain why patients with CaW may be predisposed to recurrent ischemic strokes.^2,5 Individuals with CaW with a higher degree of disturbed flow patterns may be at a higher risk of stroke.^2,4 The larger regions of disturbed flow represented by low TAWSS and high OSI indicate regions of blood stasis. In combination with endothelial activation related to alterations in blood flow during stasis, platelets and other clotting factors in the blood contact the endothelium for prolonged periods leading to adhesion and ultimately coagulation.³⁵ Typically, venous clots have large amounts of RBCs and are termed “red clots.”⁸ Even though CaWs affect arteries, recent studies have suggested CaW-associated clots have a high proportion of RBC,^8–10 which suggests that these clots form in low WSS regions similar to that seen in veins, as opposed to high WSS regions in the ICA bulb typically found in patients with atherosclerosis. The difference between high WSS affecting atherosclerosis lesions compared to low WSS areas induced by CaW geometry may help our understanding of the differences between the thrombus formation mechanism resulting from atherosclerosis lesions compared to CaW.

Multiple studies have utilized 4D Flow MRI to study the blood flow in healthy carotid bifurcations and carotid arteries with atherosclerosis lesions.^18,23,25 Markl et al¹⁸ investigated in vivo distribution of WSS and OSI in healthy compared to atherosclerotic lesions in the carotid bifurcation to evaluate the dependency of hemodynamics parameters on bifurcation geometry metrics such as diameter, tortuosity, and bifurcation angle. The study found that increased bifurcation angle and diameter resulted in much larger areas with reduced WSS and higher OSI.¹⁸ This study did not investigate geometric parameters of the carotid bifurcation due to the limited number of patients, but similar patterns were observed in the results.

Other studies have investigated the hemodynamics parameters in the carotid bifurcation using computational fluid dynamics (CFD) compared to 4D Flow MRI.^36–38 Ngo et al showed that 4D flow MRI has good velocity magnitude agreement to CFD and reliable blood flow visualization in 4D flow MRI.³⁸ Cibis et al³⁶ have compared the WSS based on 3D PCMR to CFD based on healthy volunteers scans of the carotid bifurcations. WSS was found to be lower in MRI than CFD 0.62 ± 0.18 Pa vs. 0.88 ± 0.30 Pa but closer to downsampled CFD 0.56 ± 0.18 Pa. However, flow patterns based on MRI matched well with CFD.³⁶ The findings of this study agree with their results, where the TAWSS in the CCA was 0.48 ± 0.18 Pa. This study did not compare 4D Flow MRI results to CFD, but previous studies have validated 4D flow MRI results to CFD in a patient-specific carotid model.^17,19 Although CFD is a very powerful tool, it is limited by long processing time for segmentation and solution convergence, high computational requirement, and the need for data from multiple image modalities.^39,40 4D flow MRI has a clinical advantage over CFD, in which it has the potential to provide the same 3D, time-resolved velocity field without the heavy preprocessing and postprocessing required by CFD.

Few studies have evaluated CaW hemodynamics parameters using CFD with a limited number of patients.^12,14 Compagne et al,¹² have investigated flow patterns of patient-specific CFD models derived from CTA scans. They quantified several hemodynamic parameters such as the recirculation area, TAWSS, transverse WSS, and OSI. Their results showed that the recirculation zones distal to the web, WSS, and OSI were significantly larger compared to controls. Averaged WSS distal to the CaW and in controls were comparable.¹² The results of this work have shown similar patterns as higher regions areas of low TAWSS and high OSI were observed in patients with CaW. Ozaki et al,¹⁴ performed CFD with patient-specific data acquired through a CTA of a patient who had undergone a carotid endarterectomy, and the thrombus was preserved and analyzed. Their findings indicate that thrombus formation in CaWs is due to blood stagnation.¹⁴ Similar results were found using 4D flow MRI including low TAWSS in the bulbar segment of the ICA, which may induce blood stagnation leading to thrombus formation.

Limitations

Some limitations of this study include the fact that it was done in single center on a single type of MRI scanner. Additional limitations are related to 4D flow MRI limitations such as low spatial resolution which make visualization of CaW geometry difficult. To overcome this challenge, CTA images acquired prior to the enrollment of the study and black blood MRI scans were used to assist visualization. Additionally, errors in WSS measurements might be introduced due to partial volume effect and based on the fitting functions used and the accuracy of the segmentation.³² Therefore, comparing to the normal population makes the accuracy of the results relative. Another limitation WSS quantification method used constant viscosity, to overcome this limitation in WSS quantification in MRI, shear thinning non-Newtonian modeling of fluids is suggested. Another limitation is the search for subjects with mild atherosclerosis in a similar age group. Atherosclerosis usually affects older population and atherosclerosis lesions with <50% stenosis is generally asymptomatic and may not come to clinical attention. To overcome the implication of age as confounding, the correlation of TAWSS with age within groups was quantified. The Pearson correlation coefficient within each subject group resulting in a 0.01 coefficient in patients with CaW, 0.01 in patients with atherosclerosis, and 0.07 in normal subjects showing an overall poor correlation between age and WSS measurements. Moreover, in this study CaW population is predominantly female, and previous studies have shown that CaW is more common in female subjects.^4,5,7 Finally, a multi-center with larger number of patients study is needed to verify these results and to eventually build a stroke risk assessment tool.

Conclusion

This study compared the hemodynamics parameters (TAWSS and OSI) in patients with CaW, patients with mild atherosclerosis, and healthy subjects using 4D flow MRI. The results of this work showed that despite a low degree of stenosis, the CaW shape induces complex flow patterns represented by larger regions exposed to low TAWSS, and higher OSI when compared to mild atherosclerosis lesions and healthy subjects. Additionally, the findings of this study may suggest that the thrombus formation mechanism between CaW and atherosclerosis is different due to the different hemodynamic environments.