Bicuspid aortic valve morphology and hemodynamics by same-day echocardiography and cardiac MRI

This study investigated the impact of bicuspid aortic valve (BAV) on valve morphology and motion as well as proximal and aortic hemodynamics using a same-day echocardiography and cardiac MRI. Transthoracic echocardiography, two-dimensional cine MRI of the aortic valve, and aortic 4D flow MRI were performed on the same day in 9 normofunctional BAV patients (age = 41 ± 12, 3 female), 4 BAV with moderate to severe aortic stenosis (AS) (age = 63 ± 5, 1 female), and 36 healthy tricuspid aortic valve controls (age = 52 ± 10, 21 female). Valve opening and closing timings and transvalvular peak velocity were measured using B-mode and Doppler echocardiogram, respectively. Valve orifice morphology at a fully-opened state was characterized using cine MRI. Ascending aortic (AAo) wall shear stress (WSS) was measured using 4D flow MRI data. Valve motion timings were similar between BAV and controls. BAV was associated with an increased orifice aspect ratio (1.44 ± 0.11 vs. 1.10 ± 0.13, P < 0.001), transvalvular peak velocity (1.5 ± 0.3 vs. 1.2 ± 0.2 m/s, P < 0.001) and maximum AAo WSS (1.62 ± 0.31 vs. 0.91 ± 0.24 Pa, P < 0.001). The increased orifice aspect ratio was associated with the increase in transvalvular peak velocity (r = 0.80, P < 0.0001) and maximum AAo WSS (r = 0.83, P < 0.0001). Transvalvular peak velocity was also positively correlated with maximum AAo WSS (r = 0.83, P < 0.0001). A same-day echo and MRI imaging allows for a comprehensive assessment of the impact of aortic valve disease on valve function and hemodynamics. In this pilot application to BAV, we found increased orifice aspect ratio may be responsible for increased transvalvular peak velocity and maximum AAo WSS.


Introduction
Bicuspid aortic valve (BAV) is the most common congenital cardiovascular defect that occurs in 1-2% of the general population in the United States [1]. BAV is associated with a wide range of complications such as valvular stenosis and regurgitation and aortic aneurysm and dissection [2,3]. Progression to secondary complications is thought to result from genetic defects in vascular connective tissue [4]. However, additional evidence suggests that BAV can result in abnormal ascending aortic (AAo) flow associated with the development of aortopathy, even in patients without aortic valve stenosis (AS) [5][6][7][8].
Wall shear stress (WSS), which represents the viscous friction force exerted on the aortic wall by blood flow, has been regarded as an important flow metric associated with vascular remodeling [10]. Analysis of aortic WSS by 4D flow MRI has shown asymmetric and elevated WSS in the AAo of BAV patients [11][12][13]. In addition, recent studies provide evidence that elevated WSS can be associated with degeneration of the aortic wall (elastic fiber thinning on histopathology) [11,14]. However, 4D flow MRI has insufficient spatial and temporal resolution to accurately assess BAV morphology and function, potentially important physiologic mechanisms implicated in AAo flow and WSS changes. The impact of valve metrics such as opening and closing dynamics, valve orifice shape, and near-valve flow on changes in AAo WSS in BAV is thus not fully understood.
This study presents a multimodality approach that combines echocardiography (echo), 2D cine MRI, and 4D flow MRI to investigate in detail aortic valve morphology, valve motion, as well as near-valve flow and downstream AAo WSS affected in BAV patients compared to healthy subjects with normal tricuspid aortic valve (TAV). The specific objectives were (1) to assess dynamic characteristics of valve leaflet opening and closing motion and aortic valvular flow using echo, (2) to assess morphological characteristics of valve orifice shape using cine MRI, and (3) to assess downstream 3D WSS using aortic 4D flow MRI. We hypothesized that this combined echo and MRI approach could detect the relationships between BAV morphology, valve motion dynamics, near-valve and downstream aortic hemodynamics.

Study cohort
A total of 13 BAV patients and 36 healthy volunteers with TAV (age = 52 ± 10, female = 20) were prospectively enrolled for same-day transthoracic echo and cardiac MRI, including 4D flow MRI. Based on American Society of Echocardiography guidelines [15], BAV patients were divided into two subgroups with and without echocardiography proven aortic valve stenosis (AS): 9 BAV patients without AS and 4 BAV patients with AS (2 moderate-tosevere and 2 severe). Subject demographics are summarized in Table 1. The study was approved by our institutional review board and informed consent was obtained from all participants.

Data acquisition-echo
All echo exams following a standard-of-care protocol were conducted using an echocardiography scanner (Vivid E95, GE Healthcare, Horten, Norway) and a transducer for adult cardiac imaging (M5Sc-D 1.4-4.6 MHz). In each set of echocardiograms of a subject, parasternal long-axis and short-axis B-mode images that captured aortic valve motion was used to assess aortic valve motion and continuous-wave Doppler velocity spectrum of flow through the aortic valve was used to assess near-valve flow dynamics.

Data analysis-echocardiography
Echo data post-processing and measurements were performed using EchoPAC (v2.02, GE Healthcare, Horten, Norway). Valve leaflet motion was analyzed by measuring the timings of three leaflet motion phases: valve rapid opening time, valve slow closing time and valve rapid closing time suggested previously by De Paulis et al. [16] and Leyh et al. [17] They quantified the timings based on an aortic valve motion diagram (Fig. 1A) seen on aortic valve M-mode echocardiogram. However, we noted that M-mode often fails to visualize all of the three phases due to the use of a single fixed beamline. In this study, we used parasternal long or short-axis aortic valve two B-mode echocardiogram to reconstruct the valve motion diagram B-mode echocardiogram ( Fig. 1B, C). The measurement procedure was as follows: (1) Place an interrogation line across the aortic valve leaflets, (2) extract B-mode signals along the line and plot them over time (Fig. 1C), (3) measure the duration of each phase observable in the plot. If there is a missed phase in the diagram, the position of the interrogation line was adjusted and the step (2) and (3) were repeated. For near-valve flow dynamics characterization, peak velocity and time-to-peak velocity (TTP) of transvalvular flow were measured as a peak point and a duration between the inception of velocity increase to the peak point, respectively, as described in Fig. 2. All timings were divided by √ RR interval according to Bazett's formula to minimize the effects of heart rate variability on the timing measurements [18].

Data analysis-2D cine MRI
Valve morphology was characterized based on 2D cine MRI images at the level of the aortic valve. The orifice and aortic sinus region at mid-systole (largest orifice area) were manually contoured (Fig. 3A) and used to quantify the following morphological parameters (Fig. 3B): • Relative orifice area: orifice area divided by the area of aortic sinus • Orifice aspect ratio: ratio of the long-axis ( d l ) vs. shortaxis diameter ( d s ). The short axis was measured based on a line from a commissural point of a conjoined leaflet on

4D flow MRI-wall shear stress
4D flow MRI datasets were preprocessed for phase offset correction, noise masking, and velocity anti-aliasing as described previously [19] using in-house MATLAB (R2018b, Mathworks Inc., Natick, USA) code. The thoracic aorta region within the 4D flow MRI dataset was automatically identified using a deep learning approach described previously [20] and was used to create a mask to remove velocity vectors outside of the aorta. Peak systole was defined as the time point in systole when the average velocity magnitude in the aorta was the highest. Peak-systolic aortic WSS on the aorta surface was calculated by using a previously developed method [21]. Briefly, velocity gradient vectors were calculated at the surface of the aorta 3D segmentation mask and multiplied the dynamic viscosity of blood (3.2 cP) to calculate WSS vectors at the aorta wall. Figure 4 shows maximum intensity projection images of 3D aortic WSS magnitude at peak systole for a subject in each cohort. Median WSS values in the ascending aorta were quantified by a manually defined region-of-interest from the sinus of Valsalva to the onset of the brachiocephalic artery. In addition, maximum WSS was determined as the mean of the 5% highest WSS values.

Statistical methods
Numerical data are reported in mean ± standard deviation or median ± interquartile depending on data normality determined by the Lilliefors test. Differences between the three groups were evaluated using one-way analysis of variance or Kruskal-Wallis test. When the difference between all three groups was statistically significant (P < 0.05), the pair-wise difference was tested using a two-tailed unpaired t-test (normal distribution) or Wilcoxon rank-sum test with an adjusted significance level (P < 0.017) using Bonferroni correction. Pearson correlation coefficients were calculated to investigate associations between valve morphology, valve leaflet motion timings and hemodynamic parameters. Correlation analysis was performed with data points from (1) all groups and (2) BAV and healthy TAV controls (BAV AS excluded).

Study cohort
Patient demographics are summarized in Table 1. Age, body surface area, systolic blood pressure, diastolic blood pressure between all three groups were not statistically significantly different. BAV AS patients had lower ejection fraction than TAV controls (56.3 ± 5.0 vs. 62.0 ± 4.5, P < 0.007). Echocardiography-based valve leaflet motion timings could not be analyzed for cases when image quality was insufficient to identify the aortic valve and/or when out-of-plane valve movement made the aortic valve disappear during the valve closing phase. Valve rapid opening time could not be obtained from 2 BAV AS patients and 8 healthy controls, and valve slow closing time and valve rapid closing time could not be obtained from 1 BAV patient, 2 BAV AS patients and 13 TAV controls.

Echocardiographic measurements-valve leaflet motion and near-valve flow dynamics
As summarized in Table 2

MRI measurements-aortic hemodynamics
An example of WSS distribution along the aorta representing each cohort is illustrated in Fig. 4 using maximum intensity projection along the anterior to posterior direction. Local high WSS regions were noted in the ascending aorta of BAV and BAV AS patients, whereas TAV controls have overall lower WSS. As shown in Table 2

Relationships between valve dynamics, morphology and aortic hemodynamics
Relationships between morphological and hemodynamic parameters are summarized in Table 3. Significant strong correlations were found between increased relative valve orifice area and reduced transvalvular peak velocity (r = − 0.77, P < 0.0001), between larger orifice aspect ratio and higher transvalvular peak velocity (r = 0.80, P < 0.0001) and between larger orifice aspect ratio and increased maximum AAo WSS (r = 0.79, P < 0.0001). As expected by moderate to severe AS status in the BAV AS group, orifice aspect ratio, peak velocity and WSS were far off from BAV and control groups as shown in Fig. 5, possibly driving the correlations. When BAV AS group was excluded to remove the impact of AS on the analysis, moderate correlations remained between orifice aspect ratio and transvalvular peak velocity (r = 0.43,  Fig. 5A and B, respectively. No significant associations were found between valve leaflet motion timings and valve morphology. Relationships between hemodynamic parameters are summarized in Table 4. A strong positive association was found between maximum AAo WSS and transvalvular peak velocity (r = 0.83, P < 0.0001) which remained significant when BAV AS patients were excluded (r = 0.51, P < 0.001) (Fig. 5C).

Discussion
This pilot study demonstrated the feasibility and the utility of same-day echo and cardiac MRI for comprehensive characterization of BAV and associated hemodynamics. Superior spatiotemporal resolution of echo was advantageous for evaluating rapid opening and closing valve leaflet motion and near-valve flow, while superior image contrast in 2D cine MRI and full volumetric thoracic aorta coverage of 4D flow MRI were advantageous for characterizing BAV morphology and 3D downstream hemodynamics. The sameday acquisition minimized potential variation in physiologic conditions between the two modalities. Despite the small number of BAV patients, we found significant differences in valve morphology (orifice aspect ratio), and hemodynamics near the valve (transvalvular peak velocity) and in the downstream (ascending aortic WSS) in BAV patients compared to healthy controls with TAV. Furthermore, moderate to strong correlations were observed between orifice aspect ratio by 2D cine MRI, transvalvular peak velocity by echo and ascending aortic WSS by 4D flow MRI. Near-valve flow characterized using echo demonstrated significantly elevated transvalvular peak velocity in normally functioning BAV patients. This finding has been documented previously [22,23], but its cause has not been well understood. We found that increased peak velocity was associated with the orifice aspect ratio. This relationship may be explained by an elliptical orifice providing a smaller jet flow area at the location of its narrowest section (i.e., vena contracta, where peak velocity occurs) compared to a circular orifice for a given orifice area, as shown by prior in-vitro studies [24,25]. Flachskampf et al. have further demonstrated that the vena contracta area reduces as the orifice aspect ratio increases [25]. In addition, acceleration of transvalvular flow assessed by transvalvular TTP was maintained despite the increase in peak velocity in normal BAV compared to TAV controls. Previous studies have reported that a prolonged TTP is found in severe AS patients and is associated with a poor prognosis [26,27]. In line with the studies, we observed a prolonged TTP in BAV AS patients suggesting that BAV might not alter temporal near valve flow characteristics unless it has progressed into AS.
The observation of elevated AAo WSS in normally functioning BAV patients with further increase in BAV AS patients is a common finding in previous literature [6,11,23,28]. Increased flow derangement manifested as helical flow pattern, and eccentric and skewed flow toward the wall has been shown to contribute to the increase in AAo WSS in BAV patients [5][6][7]. In addition to previous findings, we observed that transvalvular peak velocity is associated with increased local AAo WSS (i.e., maximum WSS). The association with peak velocity may be a direct consequence since a faster valvular flow velocity is likely to create a greater velocity gradient on the wall elevating WSS. Furthermore, BAV has been shown to demonstrate higher flow displacement in the ascending aorta [6]. The skewed velocity profile will induce higher velocity gradients on the aortic surface closer to fast velocities, increasing WSS consequently, as illustrated in Fig. 4. We also noted maximum AAo WSS increased with increased orifice aspect ratio. The higher aspect ratio of the valve orifice may account for the faster transvalvular peak velocity causing the elevation in maximum AAo WSS, as discussed in the previous paragraph.
This study was the first attempt to characterize temporal BAV leaflet motion. We found similar valve motion timings between the groups implying that abnormal leaflet structure due to leaflet fusion might not significantly impact the opening and closing characteristics of the valve. The change in aortic valve opening and closing timings have been observed previously in patients who underwent a valve-sparing procedure with different techniques (David or Yacoub) [17,29] or implanted with different aortic root prostheses [16] which produced different aortic root shapes and distensibility. Previous studies based on the observations in dogs suggested that expansion of the aortic root due to an increased left ventricular pressure at the beginning of systole triggers the valve to open before the onset of aortic forward flow [30,31]. A study based on in vitro observation suggested a vortex trapped in the sinus activates the valve closing [32]. In light of these previous studies, valve leaflet motion would be primarily affected by structural abnormalities in the valve surrounding, not the valve itself. It should be noted that the temporal resolution of approximately 10 ms of the B-mode echocardiograms may limit the precision of rapid opening and closing  [16,17] would resolve valve motion with higher temporal resolution (1-2 ms) [16,17]. However, the use of retrospective B-mode allowed more subjects to be analyzed (39 vs. 16 out of total 49 subjects) using B-mode that enabled retrospective adjustment of valve leaflet motion interrogation line. Patients with BAV are known to have a higher risk of developing aortic valve dysfunction and progressive proximal aortic dilation that could lead to aortic rupture or dissection [2,3]. The combined echo and MRI assessment presented in this study is useful for a comprehensive and quantitative diagnosis of valvular function and the development of aortopathy in BAV patients. As demonstrated by this study, this dual-modality approach could provide measures of abnormal valve morphology (e.g., orifice aspect ratio) and valvular hemodynamics (e.g., transvalvular peak velocity) to assess valvular dysfunction, and measures of abnormal AAo hemodynamics (e.g., WSS distribution) to predict the risk of progressive AAo dilation [33]. Furthermore, novel insights into the link between valvular dysfunction and the development of aortopathy in BAV patients may be discovered.
A major limitation of this study is the small number of BAV patients with a predominant left and right leaflet fusion pattern. Another limitation is that the ascending aorta size was not controlled in this study which would impact the distribution and magnitude of WSS [5,6]. Lastly, the BAV AS group consisted of mostly severe AS patients whose valve morphology and flow characteristics may be dominated by AS rather than BAV. However, within this limited cohort, we were still able to observe some significant relationships between valve morphology and aortic hemodynamics. Future work is warranted to confirm the study findings in a larger BAV cohort with controlled aortic size and BAV phenotype.
In conclusion, we have demonstrated that a same-day echo and cardiac MRI could serve as a useful tool for a comprehensive evaluation of valve morphology and motion, and near-valve and distal 3D hemodynamics. In this pilot application to BAV, we found increased orifice aspect ratio of BAV may be responsible for increased transvalvular peak velocity and maximum AAo WSS. This multimodality approach could be expanded to various valvular heart diseases to identify temporal and spatial characteristics of valve morphology and associated hemodynamics.