Multi-Year Interval Changes in Aortic Wall Shear Stress in Patients with Bicuspid Aortic Valve Assessed by 4D Flow MRI

Abstract

Background:

In patients with bicuspid aortic valve (BAV), 4D flow MRI can quantify regions exposed to abnormal aortic hemodynamics, including high wall shear stress (WSS), a known stimulus for arterial wall dysfunction. However, the long-term multi-scan reproducibility of 4D flow MRI-derived hemodynamic parameters is unknown.

Purpose:

To investigate the long-term stability of 4D flow MRI-derived peak velocity, WSS, and WSS-derived heatmaps in patients with BAV undergoing multi-year surveillance imaging.

Study type:

Retrospective.

Population:

20 BAV patients (mean age 48.4±13.9 years; 14 males) with five 4D flow MRI scans, with intervals of at least six months between scans, and 125 controls (mean age: 50.7±15.8 years; 67 males).

Field strength/sequence:

1.5 and 3.0T, prospectively ECG and respiratory navigator-gated aortic 4D flow MRI.

Assessment:

Automated AI-based 4D flow analysis pipelines were used for data preprocessing, aorta 3D segmentation, and quantification of ascending aorta (AAo) peak velocity, peak systolic WSS, and heatmap-derived relative area of elevated WSS compared to WSS ranges in age and sex-matched normative control populations. Growth rate was derived from the maximum AAo diameters measured on the first and fifth MRI scans.

Statistical tests:

One-way repeated measures analysis of variance. P < 0.05 indicated significance.

Results:

One hundred 4D flow MRI exams (5 per patient) were analyzed. The mean total follow-up duration was 5.5±1.1 years, and the average growth rate was 0.3±0.2 mm/year. Peak velocity, peak systolic WSS, and relative area of elevated WSS did not change significantly over the follow-up period (p=0.64, p=0.69, and p=0.35, respectively). The patterns and areas of elevated WSS demonstrated good reproducibility on semi-quantitative assessment.

Conclusion:

4D flow MRI-derived peak velocity, WSS and WSS-derived heatmaps showed good multi-year and multi-scan stability in BAV patients with low aortic growth rates. The robustness of these metrics demonstrates their stability when monitoring BAV patients for potential risk of dilation.

INTRODUCTION

Bicuspid aortic valve (BAV) is the most common congenital heart defect, affecting 1–2% of the population (1). While many patients with BAV have a stable disease course, 35% to 68% develop ascending aorta (AAo) dilation, which carries an 8-fold higher risk of aortic dissection compared to the general population (2,3). Consequently, current guidelines recommend close lifelong monitoring of patients with BAV and prophylactic repair if the maximal aortic diameter is ≥5.5 cm, or ≥5.0 cm with associated high-risk features, such as aortic coarctation, systemic hypertension, family history of dissection, and rapid aortic growth (i.e., dilation, defined as an increase of >0.3 cm/year) (4–6). However, these recommendations are largely based on expert consensus because of limited observational data. Furthermore, complications such as dissection and rupture have been shown to occur even in patients with a normal-sized aorta (7,8). Thus, there is a need for improved biomarkers to better risk-stratify BAV patients and identify those who could benefit from prophylactic surgery.

In recent years, 4D flow MRI studies have demonstrated that BAV morphology can lead to abnormal transvalvular flow patterns characterized by eccentric outflow jets (9,10). These altered flow patterns can significantly increase the wall shear stress (WSS) in the AAo, which represents the shear force produced by tangential blood flow on the arterial wall. WSS is a well-known trigger of outward vascular remodeling through endothelial mechanotransduction. In fact, multiple studies have demonstrated that high WSS can cause regional elastic fiber thinning, media degeneration, and decreased aortic wall strength, all of which can lead to aortic dilation (11–13). These findings suggest that WSS is a key factor in the development of aortopathy and may serve as a promising risk measure for this condition.

However, it has been suggested that WSS changes with age which poses a challenge for interpreting WSS data and deriving actionable information to improve patient management (14). To overcome these limitations, the concept of WSS heatmaps was introduced (15). Heatmaps delineate regions with high WSS in a patient relative to WSS maps obtained from population averages of age and sex-matched healthy control cohorts. They account for changes with age while providing an easy-to-interpret 3D visualization of aortic regions exposed to elevated WSS.

Emerging evidence has suggested that WSS and WSS heatmaps may serve as promising predictors of aortic dilation, potentially improving risk stratification and identifying patients who would benefit from prophylactic surgery (16–19). However, for these measures to become reliable clinical tools, it is important to assess their robustness and reproducibility. Previous studies have shown that WSS is reliable in terms of test-retest and interobserver reproducibility (20–22). However, these test-retest studies were limited to two scans separated by a short time interval of up to 2 weeks and did not fully assess the reproducibility of WSS heatmaps.

Thus, the aim of this study was to establish the long-term multiscan stability of absolute WSS and WSS heatmaps in the AAo using a cohort of clinically followed stable BAV patients who underwent five consecutive surveillance imaging studies that included a 4D flow MRI, with a time interval of at least 6 months between two subsequent scans. We hypothesized that 4D flow MRI-derived peak velocity, WSS and heatmaps would demonstrate stability over time in BAV patients with low aortic growth rates.

MATERIALS AND METHODS

Study Cohort - Patients

Patients were retrospectively selected from an institutional database approved by the Institutional Review Board (IRB) of a large academic medical center that routinely employs aortic 4D flow MRI as a standard-of-care for patients with aortic valve disease and/or aortic dilatation. The database is maintained longitudinally and includes information from patients who received clinical cardiovascular MRI scans, including 4D flow MRI of the thoracic aorta, starting from April 1, 2011. Individuals with BAV who underwent five cardiothoracic MRI scans for surveillance of aortic dimensions, with intervals of at least six months between scans, were included in the study. Patients were excluded if they had a time interval between two scans of less than six months, had connective tissue disorders, or underwent aortic or valve surgery prior to their fifth MRI examination. Informed consent of BAV patients was waived because the study was retrospective, and all scans were acquired as part of clinical routine.

The study identified 20 BAV patients (14 males, mean age: 48.4 ± 13.9 years) with a total of 100 4D flow MRI exams (5 per patient) acquired between 2011 and 2019. All demographic and clinical data were extracted from electronic medical records, including information on body mass index, left ventricular volumes, valve fusion morphology, and valvular disease. Patients’ demographic and clinical data are detailed in Table 1. Severity of aortic regurgitation and stenosis were extracted from radiologists’ clinical reports and assessed using 2D phase-contrast MRI, following quantitative grading guidelines (23). Levels of severity included four categories: none, mild, moderate, and severe.

Table 1.

Characteristics of the 20 BAV subjects in this study.

Characteristics (n = 20)
Age (years)	48.4 ± 13.9
Male sex	14 (70%)
Growth Rate (mm/year) Maximum AAo diameter on the 5th scan (cm)	0.3 ± 0.2 4.4 ± 0.4
Time interval between 2 consecutive scans (years)	1.3 ± 0.7
Total follow-up duration (years)	5.5 ± 1.1
BMI (kg/m2)	26.4 ± 3.6
Heart rate (beats/min)	65.1 ± 9.9
Left ventricle
End diastolic volume (ml)	152.7 ± 45.0
End systolic volume (ml)	69.7 ± 29.3
Stroke volume (ml)	83.1 ± 48.3
Ejection fraction (%)	59.9 ± 4.5
Cardiac output (l/min)	6.3 [5.2 – 7.6]
Aortic regurgitation / insufficiency grade
None to trace	9
Mild	5
Moderate	6
Aortic stenosis grade
None to trace	17
Mild	3
Valve morphology
Type 0	2
Type 1
R-L	14
R-N	3
L-N	1

Note.— R-L = fusion of right and left-coronary leaflets; R-N = fusion of right and non-coronary leaflets; L-N = fusion of left and noncoronary leaflets.

Study Cohort – Controls

To determine regionally resolved 95% confidence interval (CI) values for physiologically normal aortic WSS, healthy individuals with a normally functioning trileaflet aortic valve and no history of cardiovascular disease were recruited to undergo a research aortic 4D flow MRI as part of an ongoing study. IRB-approval was granted for this study and informed consent was obtained from all individuals. From this control cohort, we included all individuals whose age fell within a ± 5-year range of a patient’s age at the time of one of their scans. This selection criterion resulted in the inclusion of a total of 125 subjects (67 males, mean age: 50.7 ± 15.8 years).

Imaging

For both patients and controls, scans were acquired on 1.5 or/and 3.0T systems (MAGNETOM Aera, Avanto, Skyra, Siemens Healthcare, Germany) and included a prospectively ECG and respiratory navigator-gated aortic 4D flow MRI acquisition in sagittal oblique orientation covering the entire thoracic aorta. The scan parameters were as follows: spatial resolution = 1.8 to 2.7 × 1.8 to 2.7 × 2.4 to 3.5 mm³, 3D slab thickness = 62 to 96 mm, temporal resolution = 33 – 42ms, repetition time = 4.1 – 5.3 ms, echo time = 2.2 to 2.8 ms, flip angle = 7° to 15°, velocity sensitivity = 150 to 300 cm/s. For BAV patients, the 4D flow acquisition was taken 15 to 20 minutes after the administration of gadolinium contrast agent (Gadavist or Magnevist, Bayer Healthcare, Berlin, Germany). Controls did not receive any contrast agent. Moreover, in all BAV scans, an aortic angiogram was obtained along with the 4D flow sequence, using either 3D contrast-enhanced MR angiography (spatial resolution = 0.6 –1.2 × 0.6 –1.2 × 1.0 – 2.0 mm³, flip angle = 25° – 40°) or 3D cardiac gated steady state free precession (SSFP) in free breathing with a diaphragm navigator (spatial resolution = 0.7–1.5 × 0.7–1.5 × 1.2–2.0 mm³, flip angle = 18° – 70°).

4D Flow MRI Data Analysis

4D flow MRI data of patients and controls were processed by using an AI-based pipeline capable of automatically performing corrections for eddy currents, velocity noise, and velocity aliasing, as previously described (24) (Figure 1.A). After data processing, a 3D aortic segmentation was automatically generated using a previously trained and tested convolutional neural network (25). One primary observer (AM, three years of experience) inspected all 100 segmentations to determine if any manual corrections were needed at the level of the AAo. Corrections required at the arch or descending aorta were not considered, as these segments were not part of the region of interest for this study. To evaluate interobserver variability, an external observer (two years of experience) inspected a subset of 25 randomly selected segmentations.

Figure 1.

4D Flow MRI analysis workflow in a 45-year-old male patient. (A) Data was processed using an AI pipeline to correct for eddy currents, velocity noise and aliasing, and to generate a 3D aortic segmentation. (B) A centerline was automatically calculated and an orthogonal 2D plane was placed proximal to the brachiocephalic trunk to separate the ascending aorta (AAo). (C, D) Peak velocity and peak systolic wall shear stress (WSS) were quantified in the AAo. (F) A patient-specific WSS heatmap was computed relative to a map constructed from >10 age and sex-matched controls (E). Identified regions with normal and elevated WSS were mapped onto the 3D aorta visualization, and the relative area of elevated WSS was calculated (G).

From these AI-generated segmentations, a regional hemodynamic analysis in the AAo was conducted and focused on three 4D flow MRI-derived parameters: systolic peak velocity (m/s), peak systolic absolute WSS (Pa), and relative area of elevated WSS on heatmaps in the AAo (%). All analyses were performed using in-house tools programmed in Matlab (v.2019b, MathWorks, Natick, MA, United States).

Region of Interest

To perform a regional analysis focusing on the AAo, a centerline was automatically calculated in the AI-generated segmentation, followed by the manual placement of a 2D plane orthogonal to the lumen (AM, three years of experience). This plane was placed immediately proximal to the brachiocephalic trunk to separate the AAo from the entire aorta (see Figure 1.B). This planar method employed for isolating the AAo has been used in prior studies and demonstrated excellent interobserver agreement (16, 26).

Peak Velocity and Wall Shear Stress

Peak systole was defined as the time point of the cardiac cycle that results in the highest mean velocity in the AAo. Systolic peak velocity in the AAo was then calculated as the maximum velocity (98th percentile) within the AAo segmentation at peak systole (27). For visual interpretation, maximum intensity projection (MIP) images of velocities in the entire aorta at peak systole were generated (Figure 1.C). Next, the 3D systolic WSS magnitude on the surface of the entire aorta was calculated at peak systole for each scan after basic assumptions (i.e., no flow occurring through the wall). Peak systolic WSS was defined as the maximum WSS (98th percentile) in the AAo segmentation (Figure 1.D).

Wall Shear Stress Heatmap

WSS heatmaps were calculated based on previously described strategies by van Ooji et al (15,28). Briefly, to establish the range of normal aortic WSS for each patient’s scan, we selected a minimum of 10 sex-matched control scans whose age at the time of imaging fell within 5 years of the patient’s age. For example, as shown in Figure 1.E, for a 45-year-old male patient, at least 10 male controls were selected with their age ranging from 40 to 50 years. Subsequently, 3D peak systolic WSS was calculated for all selected controls and a healthy ensemble-averaged WSS map was generated to determine “normal” 3D WSS based on the patient’s age and sex. Next, the patient’s aorta was registered to the corresponding healthy ensemble-averaged WSS map, and regions with WSS outside the age- and sex-matched healthy control 95% CI were identified and deemed abnormal, as previously described (15). These regions of the patient’s aorta with increased WSS relative to the tailored healthy ensemble-averaged WSS map are referred to in this study as areas with elevated WSS. To provide a readily interpretable report, regions of normal and elevated WSS were mapped onto a 3D visualization of the patient’s aorta (Figure 1.F). Finally, the relative area (in %) of the AAo exposed to increased WSS was calculated by dividing the area of elevated WSS in the AAo by the total AAo area (Figure1.G).

Heatmaps 3D Reproducibility

To evaluate the 3D reproducibility of areas with elevated WSS in the AAo, we applied a semi-quantitative approach. This approach consisted of segmenting the AAo into 12 distinct parts: six in the left view and six in the right view, identified as proximal-anterior, proximal-posterior, middle-anterior, middle-posterior, distal-anterior, and distal-posterior (Figure 2). All 100 aortas were segmented following this layout. Next, each segment was visually assessed to determine whether elevated WSS regions occupied more or less than 50% of the segment’s area (AM, three years of experience). An interval change in the areas of elevated WSS was noted for a segment when a shift was observed from occupying more than 50% to less than 50% of the segment’s area (or vice versa) on two consecutive scans.

Figure 2.

Scan 3D reproducibility of areas with elevated wall shear stress across ascending aorta segments. The analysis included examinations of the aorta from both left and right views. Within each view, the ascending aorta was divided into proximal, middle, and distal sections using four planes. Additionally, a central line subdivided each of these sections into anterior and posterior segments, resulting in a total of 12 distinct segments. In this case, changes were observed at one interval and in two segments: the middle-anterior and distal-anterior segments in the left view. The change occurred at interval 4, where the areas with elevated wall shear stress transitioned from covering more than 50% on scan 4 to less than 50% on scan 5 for the two segments.

Aortic Growth Rates

To quantify aortic growth rates, a radiologist (RC, eight years of experience) measured the aortic diameters on the first and fifth scans of each BAV patient using 3D contrast-enhanced MR angiography or 3D cardiac gated SSFP images, depending on availability. Measurements were performed using multiplanar reformatting on a dedicated analysis software (Visage 7, Visage Imaging, Inc., San Diego, CA, United States). Two orthogonal measurements were made at three levels on the first and fifth scan: the sinotubular junction, mid-AAo, and proximal arch. To determine the maximum diameter of the AAo, we calculated the average of these two orthogonal measurements at each of the three levels. Subsequently, among these averages, the one representing the greatest length between all levels was identified as the maximum AAo diameter. Growth rates were then derived by dividing the maximum AAo diameter change by the time interval of the total follow-up duration for each patient. Patients were considered stable if the maximal aortic diameter was <5.5 cm and the growth rate was less than the 0.3 cm/year threshold for surgery (5). To investigate interobserver variability, a second observer (AM, three years of experience), measured aortic diameters and growth rates for 10 randomly selected patients.

Statistical analysis

Continuous variables are expressed as mean ± standard deviation if normally distributed or otherwise as median [1st-3rd] quartiles. Categorical variables are presented as frequency (percentage). The Kolmogorov–Smirnov test was used to assess normality. One-way repeated measures ANOVA was conducted to determine if there was a difference in peak velocity, peak systolic WSS, and relative areas of elevated WSS in the AAo between scans. Interobserver agreement was assessed using Bland Altman analysis. A p < 0.05 was considered significant.

RESULTS

Study Cohort

The AI-based pipeline successfully processed all 100 4D flow MRI scans. The primary observer evaluated the AI-generated segmentations for the entire cohort and determined that no manual corrections were needed at the level of the AAo. Similarly, the external observer, reviewing a subset of 25 scans, concluded that no corrections were required.

For every patient scan, a sufficient number of control data was available, allowing for the construction of an ensemble-averaged WSS using data from more than 10 controls. Aortic regurgitation was present in 11 patients, while 3 had aortic valve stenosis (Table 1). The mean total follow-up duration was 5.5 ± 1.1 years, and the average time interval between two consecutive scans was 1.3 ± 0.7 years.

The maximum AAo diameter was 4.2 ± 0.4 cm on the first scan, and 4.4 ± 0.4 cm on the fifth scan. These measurements yielded an average growth rate of 0.3 ± 0.2 mm/year. The interobserver analysis for growth rate calculation in a subset of 10 patients showed a non-significant bias and good agreement between the two observers (Bias: −0.1 mm/year; limits of agreement: −0.6 to 0.4 mm/year; p = 0.63).

Multi-Year Stability of Aortic Peak Velocity and WSS

The distributions of peak velocity and peak systolic WSS in the AAo were normal on all five scans. As summarized in Figure 3, peak velocity and peak systolic WSS in the AAo showed no significant differences over the multi-year follow-up time (p = 0.64 and p = 0.69, respectively). By visually analyzing the patterns of connecting lines on a patient basis, peak velocity showed less variability between scans compared to peak systolic WSS.

Figure 3.

Boxplots showing the temporal evolution of peak velocity (A), peak systolic wall shear stress (WSS) (B), and relative area of elevated WSS (C) across the five consecutive 4D flow MRI scans, all quantified from the ascending aorta. The solid red line in each boxplot denotes the median value, while the dashed green line represents the mean. Each box encompasses the 25th to 75th percentiles of the dataset, with black whiskers extending to the 5th and 95th percentiles. Markers outside these bounds indicate outliers. Data points from the same patient are connected by lines, facilitating the tracking of measurement variations at the individual level and showcasing how parameters change over time in relation to all other scans. In the bottom right (D), the table shows the mean ± SD of the three quantified measures for each scan, and the average time interval between consecutive scans.

The BAV cases with the highest and lowest reproducibility are presented in Figures 4 and 5, respectively. In each figure, regionally resolved 3D WSS at peak systole and the corresponding WSS heatmap for each of the five consecutive scans are shown. Values of peak systolic WSS and the relative area of elevated WSS quantifications are also included.

Figure 4.

Heatmaps of the patient with the highest heatmap reproducibility across five consecutive 4D flow MRI scans over a time span of five years. The top row shows the color-coded peak systolic wall shear stress (WSS) maps of the entire aorta. The second row displays the corresponding heatmaps delineating areas with normal WSS level (grey) and areas with a higher WSS than the mean+1.96*SD of WSS in the sex and age-matched control population (red).

Figure 5.

Heatmaps of the patient with the lowest heatmap reproducibility across five consecutive 4D flow MRI scans over a time span of five years. The top row shows the color-coded peak systolic wall shear stress (WSS) maps of the entire aorta. The second row displays the corresponding heatmaps delineating areas with normal WSS level (grey) and areas with a higher WSS than the mean+1.96*SD of WSS in the sex and age-matched control population (red).

For the best case (Figure 4), visual interpretation of 3D peak systolic WSS indicated that WSS followed the same pattern across the AAo between scans. Furthermore, the boundaries of the areas with high WSS were sustained, as indicated by the black arrows. For the worst case (Figure 5), regionally resolved 3D WSS showed consistent patterns across follow-ups, however, WSS values in the distal AAo were attenuated on the first and fifth scans (see green arrows). These differences resulted in lower overall peak systolic WSS in the AAo for the first and fifth examinations compared to the intermediate scans (1.3 vs. 1.4 or 1.5 Pa). Similar observations were noted in four additional patients across six scans.

Multi-Year Stability of Heatmaps

The distribution of the relative area of elevated WSS was normal across all five scans. In line with findings for peak velocity and peak systolic WSS, a one-way repeated measures ANOVA revealed no significant difference in the relative area of elevated WSS between scans (p = 0.35).

The findings from the heatmaps’ 3D reproducibility analysis are detailed in Table 2. The data is presented across the four scan intervals and includes both the number of segments showing changes per interval and the number of patients exhibiting these changes. The analysis revealed that during intervals 1, 2, 3, and 4, respectively, 50%, 40%, 50%, and 60% of patients experienced changes in elevated WSS areas. However, of the total 240 segments inspected across all patients, only 14, 12, 11, and 19 segments, respectively, showed changes on these intervals. In fact, on average, there was a change in elevated WSS areas in 0.7 segment/scan per patient. The segment that showed the worst reproducibility was the distal-anterior segment in the left view, followed by the middle-anterior segment in the same view.

Table 2.

Patients and aortic segments showing interval changes of areas with elevated wall shear stress.

	Interval 1	Interval 2	Interval 3	Interval 4
Number of patients	10/20	8/20	10/20	12/20
Number of segments	14/240	12/240	11/240	19/240

Note.— The total number of segments at each interval refers to the total count of segments in all patients (12 segments/patient x 20 patients = 240 segments).

For the best performing case, areas with elevated WSS on heatmaps were reproducible across all segments, with two exceptions (Figure 4). These exceptions were: (1) the right view middle-anterior segment on the third scan (indicated by a green arrow); and (2) the left view proximal-posterior segment, where areas with high peak systolic WSS adjacent to the aortic valve were not identified as having elevated WSS in four consecutive scans (see blue arrows). In fact, among six other patients displaying high WSS in the left view proximal-posterior segment, five similarly did not demonstrate elevated WSS on their heatmaps.

In the worst performing case, heatmap findings agreed with the previously described 3D WSS results, as smaller areas with elevated WSS were identified around the distal part of the AAo for scans with reduced WSS, as indicated by the green arrows (Figure 5). This underestimation caused the distal-anterior and distal-posterior segments in the left view to have the worst reproducibility in this case. Another notable interval change was in the left view middle-anterior segment, which showed relatively more areas with elevated WSS on the third scan (see black arrow).

DISCUSSION

In patients with BAV undergoing long-term surveillance imaging and exhibiting slow AAo growth rates, this study presents several findings: (1) peak velocity and systolic WSS demonstrate consistent multi-year stability in subjects with minimal aortic growth; (2) 3D patterns are also largely stable with some variation observed in the distal AAo; (3) WSS-derived heatmaps remain robust over five consecutive assessments. Given this relative stability, our findings suggest that changes in 4D flow MRI metrics could potentially serve as imaging biomarkers to identify BAV patients at increased risk of aortic growth.

In recent years, accumulating evidence has highlighted the importance of 4D flow MRI-derived parameters, including circumferential WSS, WSS angle, and WSS heatmaps, in the surveillance of BAV aortopathy (16–18). However, for these metrics to serve as reliable tools it is important to determine their robustness. To this end, this study evaluated the long-term multi-scan stability of 4D flow MRI-derived hemodynamic parameters in BAV patients.

All patients included in this study exhibited a growth rate below 0.3 cm/year and a maximum aortic diameter below 5.5 cm on the fifth scan, indicating a slow growth and stability throughout the follow-up period. For peak velocity and peak systolic WSS the results revealed no significant differences between scans. Prior 4D flow MRI studies assessed the test-retest reproducibility of these metrics using different quantification methods, varying from a time-resolved 2D planar quantification to a volumetric 3D assessment, with time intervals between the two scans ranging from 10 minutes to a year (20–22,29). Consistent with our findings, these studies demonstrated good reproducibility for quantification. Notably, when tracking changes across the five scans on a case-by-case basis, peak systolic WSS appeared to show greater variability compared to peak velocity. This observation aligns with findings from a previous study conducted by van Ooij et al., which reported a lower intra-class correlation for peak systolic WSS on test-retest analysis in comparison to systolic velocity (21). This increased variability is expected, as WSS is a derivative of the velocity field, and thus, any velocity noise effects will be propagated to WSS measurements. In addition, WSS quantification is very sensitive to contour delineation, which may explain the decreased reproducibility in the results (30).

Upon visual interpretation of 3D peak systolic WSS, we observed that WSS values remained consistent between scans at the proximal to mid AAo. For cases with lower performance, we observed higher WSS variability at the distal AAo, with some scans showing a reduction in WSS values. This finding is consistent with a previous test-retest study by van der Palen et al. where AAo segments demonstrated lower consistency in WSS than aortic arch or descending aortic segments (22). While the factors contributing to this variability remain unclear, the spatial resolution differed between scans and the VENC adopted ranged from 150 to 300 cm/s, which may have affected the accuracy of WSS calculations in the low velocity range near the aortic wall.

Despite the attenuation of peak systolic WSS at the distal AAo, our results showed consistent patterns of 3D WSS between scans across the AAo. This preserved distribution has also been highlighted in previous studies. For instance, in a study employing 2D planar quantification, plane-wise aortic WSS was divided into four planar segments and on test-retest was shown to be spatially consistent (20). Another study found low directional changes on test-retest analysis by measuring the angle between WSS vectors of two consecutive scans (21). Demonstrating the consistency of 3D WSS distribution across the aorta between scans is important because the impact of WSS on extracellular matrix architecture and aortic growth likely take many years to manifest (16). Therefore, a persistent pattern of increased WSS in one area could have important prognostic implications, as it could indicate a higher risk of growth for that region.

In this study, we also present the multi-year reproducibility of WSS heatmaps. We found no significant difference between scans for the relative areas of elevated WSS with robust 3D heatmap reproducibility across the follow-up period. The two segments that showed the highest variability were the distal-anterior and middle-anterior segments in the left view. Lower reproducibility in these two segments was, however, anticipated given that several cases showed attenuation in WSS at the distal AAo. On average, only 0.7 segment/scan showed a change, which strongly indicates that areas of elevated WSS display similar distribution across scans. However, it is noteworthy that areas with high peak systolic WSS adjacent to the aortic valve in the left view proximal-posterior segment were often not identified as elevated on heatmaps.

Lastly, it is important to note that in order to facilitate clinical translation of WSS techniques for BAV surveillance, elimination of time-consuming cumbersome manual analysis is needed. In this study, we employed automated AI-based pipelines for 4D flow data preprocessing, aorta segmentation, and parameter quantification, which enabled prompt analyses and minimized interobserver variability. In fact, while the literature demonstrates satisfactory interobserver reproducibility in vessel segmentation for 4D flow MRI-derived parameters, conflicting data arises when considering interobserver variability for the entire workflow, which includes the selection of appropriate background phase offset correction and noise reduction (20,21,31,32). These findings, coupled with the considerable time and effort required for manual analysis, highlight the need for complete workflow automation.

Limitations

First, the study cohort was relatively small, consisting of 20 BAV patients. Similarly, a larger number of controls may be needed to define normal WSS ranges more consistently across age and sex groups. Second, our study was retrospective and included only clinically followed stable BAV patients with slow aortic growths rates, which may limit the generalizability of our findings to more diverse and less medically compliant BAV patient populations. Third, our cohort included a mix of patients both with and without valvular disease, spanning various severity levels. The limited sample size in each category precluded any subgroup analysis. Fourth, a time-averaged segmentation was performed. Although this method has demonstrated the ability to generate heatmaps capable of predicting BAV aortopathy, performing a time-resolved segmentation could be a future direction to account for the aortic movement over the cardiac cycle. Finally, a mix of scans acquired on 1.5T and 3.0T were included which could have affected the results. However, a prior study has shown identical performance in 4D flow MRI quantitative analysis for different MRI field strengths (33).

Conclusion

4D flow MRI-derived peak velocity, WSS, and WSS-derived heatmaps demonstrated long-term multi-scan reproducibility in stable BAV patients. Additionally, WSS heatmaps were consistent in the detection of regions of elevated WSS in the aorta, exhibiting similar patterns and boundaries across scans. This suggests that these techniques may be used to monitor WSS changes in BAV patients over time, potentially assisting in the prediction of aortic complications.