Supravalvar aortic stenosis (SVAS) severity guides management, including decisions for surgery. Physiologic and technical factors limit the determination of SVAS severity by Doppler echocardiography and cardiac catheterization in Williams syndrome (WS). We hypothesized SVAS severity could be determined by the sinotubular junction-to-aortic annulus ratio (STJ:An). We reviewed all preintervention echocardiograms in patients with WS with SVAS cared for at our center. We measured STJ, An, peak and mean Doppler gradients, and calculated STJ:An. We created 2 mean gradient prediction models. Model 1 used the simplified Bernoulli’s equation, and model 2 used computational fluid dynamics (CFD). We compared STJ:An to Doppler-derived and CFD gradients. We reviewed catheterization gradients and the waveforms and analyzed gradient variability. We analyzed 168 echocardiograms in 54 children (58% male, median age at scan 1.2 years, interquartile range [IQR] 0.5 to 3.6, median echocardiograms 2, IQR 1 to 4). Median SVAS peak Doppler gradient was 24 mm Hg (IQR 14 to 46.5). Median SVAS mean Doppler gradient was 11 mm Hg (IQR 6 to 21). Median STJ:An was 0.76 (IQR 0.63 to 0.84). Model 1 underpredicted clinical gradients. Model 2 correlated well with STJ:An through all severity ranges and demonstrated increased pressure recovery distance with decreased STJ:An. The median potential variability in catheterization-derived gradients in a given patient was 14.5 mm Hg (IQR 7.5 to 19.3). SVAS severity in WS can be accurately assessed using STJ:An. CFD predicts clinical data well through all SVAS severity levels. STJ:An is independent of physiologic state and has fewer technical limitations than Doppler echocardiography and catheterization. STJ:An could augment traditional methods in guiding surgical management decisions.
Supravalvar aortic stenosis (SVAS) is the most common form of congenital heart disease (CHD) in Williams syndrome (WS). Defining SVAS severity is crucial as mild or moderate SVAS is likely to have a stable course, whereas severe SVAS typically requires surgery. , Severity of valvar aortic stenosis (AS), and by extension SVAS, was initially defined based on systolic pressure gradients measured at cardiac catheterization and correlation with clinical outcomes. Doppler echocardiography has allowed noninvasive estimation of gradients. However, both Doppler and catheter-based assessments of gradients are affected by transient physiologic and technical factors, subject to error, and are frequently discrepant. The stage of disease also can impact gradients if left ventricular (LV) function is compromised. Additionally, the risk of sudden death in children with WS is up to 100 times that of the general population, particularly with sedation which is needed for catheterization and sometimes for echocardiography. In the face of these challenges, we sought to develop a noninvasive method of SVAS severity assessment, independent of hemodynamic factors. We hypothesized that the ratio of the sinotubular junction (STJ) to the aortic annulus (An) would be a simple, reliable, and novel method for accurately assessing SVAS severity.
Methods
With the approval of the Institutional Review Board of Stanford University (protocol number 50981), we reviewed data on all patients with genetically confirmed WS and SVAS evaluated through our institution. Patients with other sources of potential gradients along the LV outflow tract or aorta, such as bicuspid aortic valve, aortic valve stenosis, aortic hypoplasia, or coarctation, were excluded. We reviewed all preintervention echocardiograms available and measured the An and STJ on long-axis, 2-dimensional echocardiographic images using recommended methods (inner edge-to-inner edge during systole at maximum diameter). , ( Figure 1 ) Peak and mean gradients were determined from continuous wave Doppler tracings obtained from the modified apical 4-chamber or suprasternal notch view. If the gradients were inconsistent between the views, the higher of the 2 was used. , STJ:An was calculated and compared with Doppler-derived mean gradients. LV function was assessed by fractional shortening or 4-chamber ejection fraction, depending on available images. When height, weight, and M-mode imaging were available, LV mass index was calculated. We reviewed all cardiac catheterizations which were completed within 2 months of acquisition of an echocardiogram. We compared reported catheterization and Doppler-derived gradients and reviewed available waveform data. The highest and lowest peak pressures captured by waveform in the LV and ascending aorta were recorded. ( Figure 1 ) Using these, the maximum possible gradient (highest LV minus lowest ascending aorta pressure) and minimum possible gradient (lowest LV minus highest ascending aorta pressure) were calculated. Outcomes of surgical intervention and in-hospital mortality were recorded. Data were maintained in a REDCap database ( https://redcap.stanford.edu , version 9.3.1). Descriptive and analytic statistics were performed using Microsoft Excel. Gradients from catheterization were compared with those measured by Doppler echocardiography using two-tailed, paired t tests.
We created a gradient prediction model (model 1) based on the Continuity and simplified Bernoulli equations along 2 points, “1” and “2” meant to simulate the An and STJ, on a streamline in the flow field:
Continuity: Area 1 × Velocity 1 = Area 2 × Velocity 2
Velocity = peak flowrate/cross-sectional area of channel
Simplified Bernoulli: ΔPressure [mm Hg]=4 × Velocity [m/s 2 ]
An initial tube diameter of 10 mm was chosen based on the median age of patients in the study (1.2 years). Using WS-specific growth charts, this predicts a height of 80 cm and weight of 8 kg. These growth parameters correspond to a median An diameter of 10.1 mm based on Boston Z score data. A peak flowrate of 10 L/min was chosen for the model based on a peak systolic flowrate for an ejection time of 180 milliseconds and cardiac output of 9.5 ml. Data points for theoretical mean gradients were generated by sequentially reducing the tube diameter by 1% (0.1 mm).
To better understand the flow dynamics along the stenosis, we performed computational fluid dynamics (CFD) simulations using the COMSOL Multiphysics software (model 2). The Navier-Stokes equations were solved numerically using the finite-volume method.
Navier-Stokes equations: <SPAN role=presentation tabIndex=0 id=MathJax-Element-1-Frame class=MathJax style="POSITION: relative" data-mathml='∇·u=0,ρ∂u∂t+ρ(u·∇)u=−∇p+μ∇2u+ρg’>𝛻·𝑢=0,𝜌𝜕𝑢𝜕𝑡+𝜌(𝑢·𝛻)𝑢=−𝛻𝑝+𝜇𝛻2𝑢+𝜌𝑔∇·u=0,ρ∂u∂t+ρ(u·∇)u=−∇p+μ∇2u+ρg
∇ · u = 0 , ρ ∂ u ∂ t + ρ ( u · ∇ ) u = − ∇ p + μ ∇ 2 u + ρ g
Here, u is velocity, p is pressure, ρ is the fluid density, μ is the fluid viscosity, and g is an external force. The domain consists of a circular, cross-section channel with stenosis as depicted in Figure 2 . The CFD models assumed an idealized SVAS geometry with fixed cavity length and position and rigid, smooth walls. The fluid was treated as incompressible. A time-dependent flowrate was applied at the inlet using a simple half-sine waveform with a period of 0.225 seconds and a peak flowrate of 6.833 L/min, whereas a constant outlet pressure of 100 mm Hg served as boundary conditions. The time-averaged flowrate was 4.35 L/min, which is roughly half the estimated peak systolic flowrate used in model 1. A total of 998.2 kg/m 3 was used for density, and 3.5e-3Pa*s was used for the viscosity of whole blood. The pressures and velocities along the cavity centerline were interrogated once a steady-state flow regime was established. We assessed gradients and pressure recovery (PR) at various STJ:An by varying the diameter of the three-dimensional flow cavity’s constriction while holding all other geometry constant. Simulated gradients from models 1 and 2 were plotted alongside clinical data for varying STJ:An ratios.
