The bicuspid aortic valve (BAV) is a common congenital malformation of the aortic valve (AV) affecting 1% to 2% of the population. The BAV is predisposed to early degenerative calcification of valve leaflets, and BAV patients constitute 50% of AV stenosis patients. Although evidence shows that genetic defects can play a role in calcification of the BAV leaflets, we hypothesize that drastic changes in the mechanical environment of the BAV elicit pathological responses from the valve and might be concurrently responsible for early calcification. An in vitro model of the BAV was constructed by surgically manipulating a native trileaflet porcine AV. The BAV valve model and a trileaflet AV (TAV) model were tested in an in vitro pulsatile flow loop mimicking physiological hemodynamics. Laser Doppler velocimetry was used to make measurements of fluid shear stresses on the leaflet of the valve models using previously established methodologies. Furthermore, particle image velocimetry was used to visualize the flow fields downstream of the valves and in the sinuses. In the BAV model, flow near the leaflets and fluid shear stresses on the leaflets were much more unsteady than for the TAV model, most likely due to the moderate stenosis in the BAV and the skewed forward flow jet that collided with the aorta wall. This additional unsteadiness occurred during mid- to late-systole and was composed of cycle-to-cycle magnitude variability as well as high-frequency fluctuations about the mean shear stress. It has been demonstrated that the BAV geometry can lead to unsteady shear stresses under physiological flow and pressure conditions. Such altered shear stresses could play a role in accelerated calcification in BAVs.
- native aortic valve
- shear stress
- laser doppler velocimetry
aortic valve (av) calcification is a degenerative disease with a high prevalence amongst the elderly (40). It can lead to aortic stenosis and heart failure and is a major cause of mortality and morbidity. The disease has been shown to be an active process akin to atherosclerosis, involving lipoprotein deposition, chronic inflammation, ectopic osteoblast-like cell differentiation, and ectopic osteogenesis (13, 25). One of the most significant risk factors for AV calcification is the congenital bicuspid AV (BAV) (40). The BAV is the most common congenital malformation of the heart and great vessels, affecting 1% to 2% of the population, or about four million US citizens (34, 43).
Patients with a BAV experience leaflet calcification about two decades earlier than people with normal AVs (5, 13), and BAV patients constitute 50% of AV stenosis patients (34, 43), which is out of proportion to the prevalence of BAV, indicating an increased predisposition of BAVs to calcification. The occurrence of BAVs follows significant patterns of familial clustering (12, 14). Recent studies have identified NOTCH1 haploinsufficiency defects in families with BAV and AV calcification (14, 15, 27), and that NOTCH1 signaling disruption can increase AV calcification in the mouse model as well as in vitro cell cultures (29). NOTCH1 haploinsufficiency alone, however, did not present the BAV phenotype (30), but other studies have shown endothelial nitric oxide synthase (eNOS)−/− or NKX2.5+/− or GATA−/− defects can result in the BAV phenotypes (21). eNOS, in particular, was shown to play a role in preventing calcification in animal AV (33) and in AV cell cultures (20). Taken together, these studies suggest that genetic anomalies could cause a congenitally fused valve, as well as accelerated calcification disease in BAV patients.
On the other hand, the role of mechanics in calcification of BAVs cannot be neglected. Nongenetic factors such as adverse mechanical force environments have been implicated to play a role in the disease (7). Mechanics studies of human BAVs have shown that the leaflet kinematics of the BAV were different from normal AVs (35) and that the forward flow can be stenotic and skewed (11, 35), indicating that the mechanical environment in the BAV can be quite different from that of a normal AV. It has also been reasoned that the increased forward flow jet velocity in stenotic BAV will cause alterations in the leaflet shear stress environment (18). Furthermore, previous ex vivo mechanobiology studies have shown that AV leaflets are sensitive to changes in mechanical force stimuli (including pressure, membrane tension, and fluid shear stress) (3, 8, 18, 24, 41, 46) and that biological responses such as inflammation and apoptosis are observed with adverse mechanical conditions, consistent with responses found in diseased human valves (31). It has been shown that the cellular and molecular mechanisms in BAV calcification are similar to normal trileaflet AV (TAV) calcification (42), suggesting that BAV leaflets would respond similarly to mechanical forces as normal TAV leaflets. We further noted that fluid shear has been shown to influence bone morphogenetic protein expressions (39, 41), which is in the NOTCH signaling pathway, and eNOS expressions (17), and these two pathways were shown to be influential on the formation of BAV and the calcification of BAV, as discussed above. Considered as a whole, these studies suggest that the altered geometry of BAVs causes an altered mechanical environment for the valve and that the valve leaflet biology is likely sensitive to this change, possibly contributing to accelerated calcification. Thus, mechanical factors could be acting in addition to genetic factors toward causing early calcification.
We hypothesize that the mechanism of BAV calcification is caused by a combination of genetic and hemodynamic factors, instead of genetic factors alone. The drastic alteration in the valve geometry in the BAV may lead to drastic alterations in the fluid mechanical environment, and elicit inflammation and calcification response on a chronic basis. The focus of the current study is to investigate the effects of BAV geometry on the fluid mechanical environment. An in vitro model of the BAV was constructed through surgical manipulation of a normal fresh porcine TAV and was tested in a pulsatile flow loop. Fluid shear stresses experienced by the valve leaflet were measured using laser Doppler velocimetry (LDV), using a previously established methodology (49, 50). The resultant measurements were compared with those obtained from an in vitro TAV model.
MATERIALS AND METHODS
The procedure for constructing the BAV model is shown in Fig. 1A. A fresh porcine aortic root was harvested from the local slaughterhouse and surgically modified as follows: 1) the aorta was opened longitudinally between the right and left coronary leaflets; 2) 2- to 3-mm strips of the aortic wall were excised from the longitudinally cut edge; 3) triangular portions of the right and left coronary leaflets were excised on the side of the longitudinal edge, and the same two leaflets were then sutured together to form the fused leaflet of the BAV; and 4) the sinus walls of the root were trimmed away, and the valve was sutured to a dual-stented plastic ring. Care was taken that the resultant fused leaflet was mobile and did not create excessive stenosis. Details about the model construction and the characterization can be found in Ref. 37.
In the similar manner, the TAV model was constructed for comparison with the BAV model (Fig. 1B). Preparation of this valve involved trimming of the sinus walls of the root only, without affecting the leaflets or their commissural attachments, and suturing to a plastic ring with three stents.
The valve models were fixed in 0.1% glutaraldehyde solution for 24 h under ∼25 mmHg pressure in preparation for lengthy experiments. The valves were then inserted into the respective aortic chamber with idealized sinus geometry. The BAV was inserted into a chamber with bilobed sinuses (Fig. 2A), whereas the TAV was inserted into a chamber with trilobed sinuses (Fig. 2B). The chamber was made of acrylic to provide optical access during LDV measurements (Fig. 2C). The valves were then tested in a pulsatile in vitro flow loop.
Pulsatile flow loop.
The Georgia Tech Left Simulator was used as the pulsatile flow loop for testing the two valve models. The loop components and setup were explained in detail in previous publications (22, 47). The valve was exposed to mean aortic pressure of 100 mmHg and cardiac output of 5 L/min at a heart rate of 70 beats/min. A water-glycerin solution containing 36% glycerin by volume was used as the working fluid to match the density and viscosity of blood (dynamic viscosity of 3.8 cP and density of 1.08 g/cm3).
Shear stress measurement technique.
The technique for measuring shear stresses on these valve models has been described in a previous publication (49, 50). LDV was used for velocity measurements, which were performed along a radial line from the sinus wall to the center (“belly” region) of the valve at a spatial resolution of 89 μm. Two components of velocity, the axial or streamwise velocity and the lateral or nonstreamwise velocity, were measured using this technique in this study. The leaflet location was tracked by tracing the location of the highest back-scattered LDV laser light intensity, utilizing the principle that the leaflet would reflect the most LDV laser light back to the LDV probe when it is situated in the probe volume, where the LDV laser pair crossed. The orientation of the valve leaflet was determined by tracking a slightly lateral and slightly downstream location of the valve leaflet, and shear stresses were corrected for the misalignment between the valve leaflet surface and the LDV probe alignment by the squared cosine of the misalignment angle, all the while making sure that the misalignment angle remained smaller than 0.15 radians. Finally, the same method of computing shear stress based on viscosity and velocity profile gradient was used as in our previous study (48).
Analysis of velocity and shear stress variability.
At each measurement location along the radial line of measurement, at least 40,000 data points were taken over at least 60 cardiac cycles. Measured velocities were binned into 86 time phases according to the time point within the cardiac cycle that they were measured. Since at least 40,000 data points were measured at each location, there were ∼400 velocity measurements in each of 86 time phases, at each measurement location. Velocities measured at the same time phase and at the same location were ensemble averaged to give the mean and the standard deviation. The standard deviation of velocities served to indicate the variability of flow during a particular instant in the cycle at a particular measurement location. Mathematically, the standard deviation is equivalent to root-mean-square of the fluctuating component of velocity (which is also the deviation of instantaneous velocity from the ensemble average velocity). where where ui is the ith instantaneous velocity data within a time bin (denoted by t), Ū is the ensemble average velocity, σ is the standard deviation of velocities, and u′ is the fluctuating component of velocity. Summations are sums over all phase-locked data points within the same time bin.
With the shear stresses measuring technique described in the previous subsection, only phase-locked shear stresses ensemble-averaged over many cycles can be obtained. Instantaneous shear stresses cannot be measured since the velocities at each spatial location are measured during different cardiac cycles, and consequently, it is not possible to examine temporal evolution of instantaneous shear stresses. However, we can examine the temporal evolution of instantaneous velocities very close to the valve leaflet surface to infer characteristics of shear stresses experienced by the leaflets.
We examined time-sequential velocity-measurements ∼1 mm away from the valve leaflet surface to assess temporal unsteadiness in the fluid velocities. Velocity measurements analyzed had ∼400 data points per cardiac cycle and spanned over at least 60 cardiac cycles. The power spectral density of the fluctuating component of the velocity was also calculated to see if the BAV has higher magnitude velocity fluctuations and if there were specific dominant frequencies in the velocity fluctuations. To account for irregular temporal spacing of the input data, which is typical of LDV measurements, the sample and hold method was adopted, as described by Adrian et al. (1) and Adrian and Yao (2). Under the sample-and-hold method, velocities were resampled at regular intervals, and at each resampling time point, velocity was assumed to have the same value as the immediate previous measurement. The autocorrelation function of the resampled velocity data was then calculated, and the power spectrum was calculated by the Fourier transformation of the autocorrelation function. To account for the bias due to the nonuniform spacing of time, Moreau et al. (28) proposed a refinement of the autocorrelation function of the sample and held velocities before the calculation of power spectrum. Power spectrum was then calculated based on unrefined sample-and-hold method, as well as the refined sample-and-hold method.
Finally, with use of the ensemble average velocities at the points closest to the leaflet surface, the ensemble averaged velocity profile is reconstructed. This ensemble averaged velocity profile is utilized to calculate the ensemble averaged shear stresses on the valve leaflets. Additionally, the velocity profile constructed using the ensemble average plus one standard deviation of velocity provides an upper bound estimate for the shear stress, whereas the ensemble average minus one standard deviation of velocity provides a lower bound estimate for the shear stress. This unsteadiness represents both cycle-to-cycle variability and temporal unsteadiness within individual cycles.
Particle image velocimetry.
Phase-locked two-dimensional particle image velocimetry (PIV) was performed in the downstream and sinus regions of the BAV and TAV models to obtain the two-dimensional instantaneous flow fields during midsystole. This was performed to compare the differences in the flow fields immediately downstream of the BAV versus the TAV, and the flow fields in the sinuses. Detailed PIV data are presented in earlier work from our group (37).
The PIV cameras were positioned to view the measurement region normal to the laser illuminated plane. The laser sheet thickness was estimated to be about 0.5 mm. The same flow loops used for LDV were used for PIV analysis. The resolution of the images taken was ∼30 μm/pixel, with the particles images size being between 3 and 5 pixels.
Fifty image pairs were acquired during midsystole, and Davis 7.11 software (Lavision, Germany) was used to analyze the PIV images. PIV cross-correlation vectors were computed with an initial pass of 64 × 64 pixels interrogation regions with 50% overlap reduced down to a final pass of interrogation region corresponding to 32 × 32 pixels with 50% overlap. A median filter was applied upon acquisition, where spurious vectors with velocities exceeding five times the average velocities of their immediately neighbor vectors were removed, and filled in with the mean of the neighboring velocities.
Characteristics of the BAV and TAV models.
The hemodynamic conditions simulated for both the BAV and TAV models are shown in Fig. 3. The BAV valve had an effective orifice area of 1.0 cm2, with mean and peak systolic pressure gradients of 24 and 40 mmHg, respectively, and thus could be classified as a moderately stenotic valve (6). The TAV had an effective orifice area of 1.4 cm2, with mean and peak systolic pressure gradients of 12 and 20 mmHg, respectively, and could be classified as a mildly stenotic valve. During the experiments, the flow loop was controlled such that the volumetric flow waveform and the aortic pressure waveforms for the two valves matched reasonably. Because the BAV was more stenotic than the TAV, it had higher transvalvular pressure. This resulted in differences in the ventricular pressures for the two valves.
The radial motions of the valve leaflets were obtained with the back-scattered LDV laser light intensity technique as described above. Measurements indicated that all valve leaflets had stable positions for the majority of systole and diastole, and were mostly free from fluttering or oscillatory motions, except for the transient after valve closure, where slight oscillations were observed. Between systole and diastole, the leaflet moved quickly between the open and closed positions. Shear stress computations were omitted for these durations of rapid leaflet motion and for the duration of the oscillatory motion after valve closure, because it was not possible to reliably locate the leaflet for accurate shear stress computation during these times.
Figure 4, A and B, shows the ventricular view of the BAV in the flow loop, recorded with a high-speed camera through an en face viewing window in the ventricular chamber. These images of the BAV model compared well with those of human BAVs tested in an in vitro flow loop by Robicsek et al. (35), as shown in Fig. 4, B–E. At valve closure, the ratio of the fused to nonfused leaflet area was ∼60:40, corresponding to the most common clinically observed BAV morphology (36). When the valve was open, the valve had a skewed oval flow orifice rather than a round central one and the fused leaflet demonstrated impaired mobility and tended to obstruct flow.
Flow fields downstream of the valves and in the sinuses.
Figure 5 shows the ensemble average flow fields during midsystole for both the BAV and TAV obtained using PIV. In the BAV, forward flow is skewed away from the fused leaflet, which results in the forward flow impinging on the sinotubular junction. Consequently, part of the forward flow moves into the nonfused leaflet sinus, directly driving the sinus vortex flow in that sinus. On the other side of the valve, the forward flow jet leads to the formation of another vortex adjacent to it, downstream of the fused-leaflet sinus. This vortex in turn induced the vortex in the fused leaflet sinus, which is thus a secondary vortex. This mechanism of vortex induction resulted in the vortex in the fused leaflet sinus having the same sense of rotation as that in the nonfused leaflet sinus. This would result in the streamwise shear stress on fused leaflet having an opposite direction as that on the nonfused leaflet. In the TAV, flow was noneccentric, and was wider and lower in magnitude than that in the BAV model. The flow in the TAV sinus is similar to the fused leaflet sinus flow from the BAV model and has the same sense of rotation. The skewed forward flow observed in our BAV model is similar to the flow patterns commonly observed in clinical BAVs using PC-MRI (35, 36, 38).
Shear stress computation.
The fluid shear stresses on the leaflets of the BAV model are plotted in Fig. 6A. Streamwise shear stresses are shown for both the fused leaflet and the nonfused leaflet, and nonstreamwise shear stress on the fused leaflet is also shown. In the streamwise direction, shear stresses were elevated for both leaflets during systole, when flow in the sinus forms the sinus vortex. Systolic streamwise shear stress resembled a sinusoid, peaking at about the 205-ms time point in the cardiac cycle (refer to Fig. 6A, which is about 105 ms from the beginning of systole). For the fused leaflet of the BAV, the systolic streamwise shear stress waveform appeared skewed toward late systole: shear stresses were low during early systole and elevated rapidly during midsystole to the peak of 20 dyn/cm2, and remained high until end diastole. For the nonfused leaflet of the BAV, systolic streamwise shear stress waveform was not skewed toward late systole. The peak shear stress was much higher on the nonfused leaflet (36 dyn/cm2) and opposite in direction compared with that on the fused leaflet (19.9 dyn/cm2). This can be explained by the opposite direction of vortices in the two sinuses, as discussed earlier. Diastolic streamwise shear stress resembled a decaying exponential curve, starting at a higher initial magnitude during early diastole and reducing to zero during late diastole. Diastolic streamwise shear stress had a higher peak on the nonfused leaflet (19 dyn/cm2) than on the fused leaflet (3 dyn/cm2), but there were no negative shear stresses during this time. In the nonstreamwise direction, shear stress was initially elevated to ∼13 dyn/cm2 during early systole, and decreased to an absolute magnitude of less than 4 dyn/cm2 for the rest of systole. During diastole, shear stress was less than 2 dyn/cm2.
Figure 6A also shows the streamwise shear stress waveform for the TAV experimented under similar hemodynamic conditions as the BAV. These data were presented in our previous study, where the variation of these shear stresses with cardiac outputs and heart rates were documented (49). The streamwise shear stress waveform of the BAV-fused leaflet was similar to that of the TAV. Systolic shear stress elevated only after a delay from the beginning of systole, and thus both waveforms were skewed toward late systole, although the delay was slightly longer for the TAV (shear stress started elevating about 95 ms after onset of systole, as opposed to 70 ms for the BAV-fused leaflet). Furthermore, peak systolic shear stress for the TAV was 19.6 dyn/cm2, which is close to that for the BAV-fused leaflet streamwise shear stress. The TAV streamwise shear stress waveform had slightly different shape from that of the BAV-fused leaflet, where shear stress peaked later (about 155 ms after the onset of systole, as opposed to 105 ms for the BAV-fused leaflet). In the BAV waveform, shear stress magnitudes remained high for a longer period of time during the mid- to late-systole period, giving the waveform a blunt-top shape, whereas the shear stress waveform for the TAV had a shape with a sharper peak.
Shear stresses calculated from the ensemble average velocities ±1 SD are shown in Fig. 6B for the BAV-fused leaflet. Shear stress variability was low throughout diastole and early systole, but saw a large increase during midsystole, when average shear stress was peaking. During this period, the 1 SD shear stress bound was found to be substantially wide, which corresponds to the increased levels of velocity variability as shown above. The peak of the upper-bound shear stress was about twice the peak of the average shear stress, whereas the lower-bound shear stress reversed in direction for the first half of systole.
Two BAVs and two TAVs were evaluated altogether, and the following observations were representative of both sets of valves.
To assess the variability of shear stresses due to flow unsteadiness as well as cycle-to-cycle variability, it is instructive to first analyze the variability in the velocities used to compute the shear stresses. Figure 7 plots the raw fluctuating velocity (u′) data measured near the two BAV leaflets and the TAV leaflet. These plots qualitatively showed that during systole, velocities near the BAV leaflets had a tendency to experience high-frequency fluctuations from the ensemble mean than velocities near the TAV leaflet. Figure 7, D and E, shows quantitative comparison of the magnitudes of the fluctuating components for the three leaflets from the two valves, for systole and diastole, respectively. These statistics were obtained by averaging the time-specific fluctuating velocity over all the cardiac cycles. Figure 7D demonstrates that the fluctuating component of velocities near the BAV leaflets were significantly higher than that near the TAV leaflet at various time points during systole. At all other time points in the cardiac cycle, there were no statistically significant differences between the fluctuating components of velocities between the different leaflets.
To further quantify the level of velocity fluctuations, the SD of velocities (u′RMS) is quantified at a representative location ∼1 mm away from the systolic position of the valve leaflet. Figure 8A plots u′RMS within all time bins for three different leaflets: the BAV-fused leaflet, the BAV nonfused leaflet, and a leaflet of TAV. During systole, the nonfused leaflet of BAV showed largest velocity SD, with a peak value of about 0.18 m/s. The fused leaflet of BAV had a peak velocity SD of 0.12 m/s, whereas the TAV leaflet showed very small velocity SD at this time point (<0.02 m/s). Velocities near the fused and nonfused leaflets of BAV were especially variable, particularly during midsystole, and this was distinctly different from the leaflet of the TAV, which showed only minor variability throughout systole. During valve closure, velocity variability increased for both the BAV and the TAV, most likely due to fluid mixing from reversing flow. Variability of flow at this location during diastole showed little differences between the three cases (TAV leaflet, BAV-fused leaflet, and BAV nonfused leaflet). However, during diastole, this location was far away from the leaflet and was well into the sinus.
Figure 8B shows the SD of velocities or u′RMS measured ∼1 mm away from the diastolic position of valve leaflets. Note that there were no measurements during systole at this spatial location because optical access to this location was blocked by the opened leaflet during systole. At this location, variability of flow gradually decreased from the early diastolic value to the low value at the end of diastole, coinciding with the dissipation of fluid motion in the sinuses. Variability of flow was higher near the fused and nonfused leaflets of BAV than that on the TAV leaflet during early diastole. However, from mid-diastole onward, the variability near all the leaflets of both models became similar.
The power spectral density analysis of the velocity deviation from the ensemble mean is shown in Fig. 9. Figure 9A shows the power spectrum for the sample and hold method without refinement, whereas Fig. 9B shows the power spectrum for the same, but with refinement proposed by Moreau et al. (28). The results showed that velocity deviation from ensemble mean had broad-band characteristics without any distinctive features at specific frequencies. Nonetheless, power spectrum of fluctuating velocities near the BAV leaflets had higher magnitudes than those near the TAV leaflet, further indicating higher levels of velocity unsteadiness near the BAV leaflets, and thus higher levels of unsteadiness of fluid shear stresses experienced by the BAV leaflets.
Relevance of the BAV model.
The current BAV model was modeled to be representative of the most common BAV morphology seen clinically: the fused leaflet covered between two thirds and half of the valve area (36) and had restricted motion, and blocked flow even in the fully open position (35), and forward flow was skewed away from the fused leaflet sinus (11).
The BAV model described here has a peak systolic gradient of 40 mmHg and a mean systolic gradient of about 24 mmHg. In a patient, a BAV with these characteristics will not be replaced or treated until the mean systolic gradient reaches about 40 mmHg, per current American Heart Association guidelines (32). Hence, a lower gradient BAV model as described here can cause a chronic change in the mechanical environment of the valve over a long period of time, potentially resulting in calcification of the valve. A higher gradient BAV model is not relevant for a hemodynamic study as such a valve is typically stenotic due to geometric or congenital reasons, and will mostly be replaced before calcification can develop on the leaflets.
Fluid dynamic environment of the BAV and TAV.
To date, the determination of the fluid mechanical environment experienced by the native AV is incomplete and is an area of ongoing research. Estimates of this environment have been presented using different methods: experimental measurements in transparent plastic valves under steady flow (45), fully coupled fluid structure interaction computer simulations at low Reynolds numbers (10, 44), quasi-steady computational fluid dynamic simulations without coupling of leaflet dynamics and mechanics (16). More recently, experimental measurements of shear stresses using in vitro valve models were performed (49, 50). The current study is an addition to this list, which adds to the knowledge of shear stresses experienced by the congenitally fused bicuspid AV.
In the current study, shear stresses on the aortic surface of the valve leaflets were the result of flow in the sinus, which is circular or vortical in nature. Vortical flow in the sinuses is a clinically observed phenomenon (23), and these vortices are induced during systole by the forward flow jet downstream of the valve orifice, as explained by previous authors (4, 26, 32), and dissipates during diastole after valve closure, due to viscous dissipation. This explains the higher magnitude of fluid shear stresses on the valve leaflets during systole than diastole, and why shear stresses peaked at about midsystole, approximately the same time as peak forward flow.
The PIV measurements showed that for the BAV, the vortex in the nonfused leaflet sinus was directly driven by the skewed forward flow, and thus had higher velocities than the vortex in the fused leaflet sinus. Consequently, the nonfused leaflet experienced higher magnitude shear stresses than the fused leaflet sinus. Furthermore, due to the indirect induction of the sinus vortex in the fused leaflet, it took longer for the vortex to form, thus explaining the delay in the elevation of shear stress on the fused leaflet compared with that on the nonfused leaflet during systole. In the TAV, PIV measurements showed that the vortices in the sinus were in the same direction as that in the BAV-fused leaflet sinus, potentially due to the mild stenosis of the TAV model.
The BAV experienced greater unsteadiness in shear stresses than the TAV.
The peak Reynolds number of flow through the valve models can be calculated using the radius of the aorta and mean velocity within the jet during peak flow, and this parameter is often used to determine the flow regime of a fluid phenomena: whether there has been onset of turbulence, and how intense the turbulence is. The BAV model had a peak flow Reynolds number of 12,400, which could be classified as being fully turbulent, whereas the TAV had a peak flow Reynolds number of 7800, which is near the transition to turbulence. The turbulent BAV flow could cause greater variability in the velocity and pressure fields in the downstream flow field. The direct driving of flow within the nonfused sinus of the BAV by the forward flow potentially caused larger magnitudes of fluctuation in the nonfused sinus of this model. The fused sinus had lower levels of fluctuations due to the indirect forcing of the vortex in this sinus. The fluctuations in the TAV model were the lowest amongst all valve models, and this might be expected since this model had a lower peak Reynolds number.
Plots of u′ (Fig. 7) demonstrated that there were greater high-frequency fluctuations in the flow field near the BAV leaflets than near the TAV leaflet, which would most likely result in high-frequency fluctuations in shear stresses on the BAV leaflets. The u′RMS plots of velocities close to the leaflets (Fig. 8) and the power spectral analysis (Fig. 9) both indicated that flow near the BAV had a greater tendency to deviate from the ensemble average value. From these observations, it could be concluded that the additional shear stress unsteadiness observed in the BAV was a combination of magnitude variability between different cardiac cycles as well as high-frequency fluctuations within the same cardiac cycle. It is interesting to note that the power spectral analysis did not show any specific dominating frequencies, which is a feature similar to the power spectra of turbulent flows. This spectral analysis hence indicates the presence of turbulence-like flow in the sinus regions of the valve.
Possible role of mechanics in BAV calcification.
Due to data that demonstrate that NOTCH1 and eNOS signaling are associated with BAV calcification (12, 14), some may hold the view that genetic factors are solely responsible for the development of BAV morphology and associated secondary diseases such as early leaflet calcification. We believe, however, that it is in fact the combination of both mechanics and genetic factors that causes the early calcification of BAV leaflets. In the current study, we present supporting data for the mechanics view to encourage consideration of this view. We demonstrate that it is possible for the BAV geometry to experience altered fluid shear stress characteristics, such as the excessive unsteadiness due to turbulent and skewed forward flow. In arteries, low and oscillatory shear stresses are known to cause atherosclerosis due to active endothelial response to the fluid shear stress environment (9). This could be similar for the AV endothelium, where the increased unsteadiness in fluid shear stresses could lead to accelerated sclerosis in the BAV leaflets, leading to early calcification. This notion is corroborated by a recent study on the genetic expressions of AV endothelial cells, showing that there are specific genetic expressions when endothelial cells are exposed to low-frequency oscillatory shear stress versus steady shear stress (19).
We note, however, that the oscillatory shear stress used in this study and associated with atherosclerosis has much lower frequency than the unsteady shear stress we observed in our BAV model. To the best of our knowledge, there has been no study on whether such high-frequency unsteady shear stresses can affect endothelium biology to promote calcification. This notion, however, requires further investigation for confirmation. Ex vivo mechanobiology studies, for example, can be performed to investigate whether unsteady shear stresses are pathogenic to the AV endothelium. Furthermore, it may be possible to correlate the level of stenosis of BAVs to their tendency to calcify using clinical data. If such a correlation is observable, it could be demonstrated that increased turbulence due to greater stenosis could lead to increased calcification by imposing shear stresses of greater unsteadiness on the valve leaflets.
The current study has the following limitations. First, the in vitro flow loop could not completely mimic in vivo conditions. Nonphysiologic factors included the rigidity of channel walls, the lack of natural motions of the aortic root leading to the lack of the physiologic inertial frame, and minor deviations from the in vivo flow and pressure waveforms. Nonetheless, the conditions simulated were sufficiently close to native conditions to allow an estimation of the in vivo shear stress characteristics.
Second, being a surgical modification from a normal AV, even though the geometry was controlled as much as possible, our BAV model might have differences with actual clinical BAV geometry. The model might also have different material properties from the clinical BAV. Furthermore, the sutures along the centerline of the fused leaflet may result in differences in leaflet mobility from clinical BAV.
Third, it should be noted that it is difficult to create valve models from the native AVs that have characteristics exactly as desired, such as valve orifice area and orifice shape. However, through the creation of several valve models, it is possible to be able to repeatably manufacture models with desired characteristics. Generally, the manufacture of two to three valves was necessary to obtain the valve of desired characteristics for an experienced worker. We note, however, that modification of the valve to have a smaller orifice area was possible, if upon testing, the valve was found to have a valve area that was too big.
Fourth, in the current study, only one type of BAV was studied: one which is moderately stenotic and has an eccentric systolic orifice. It should be noted that the observed macroscopic fluid mechanics characteristics may apply only to this category of BAV, and can vary with the BAV geometry. It should also be noted that the BAV valve model was more stenotic than the TAV valve model and that the resulting data are applicable specifically to valves of these characteristics. Nonstenotic BAV found clinically may not experience similar unsteady leaflet surface shear stresses found in our BAV valve model; on the other hand, stenotic TAV may experience such unsteady shear stresses. A nonstenotic TAV, however, is unlikely to experience such unsteady shear stresses. Because only one model of BAV was studied, the results may apply only to the BAV with characteristics similar to our BAV model.
LDV was used for measuring shear stress due to its ability to make measurements very close to the valve leaflet. LDV, however, was a point-by-point measurement technique, and the computation of shear stresses could only be performed using the ensemble-averaged velocities at all measurement locations, as well as the ensemble average leaflet location. The results could be interpreted as estimations of the mean shear stresses. This technique could not be used to elucidate instantaneous shear stresses experienced by the valve leaflets. Given the variability encountered for some of the cases, the actual instantaneous shear stress could deviate considerably from the shear stress calculated using the ensemble average velocity profile most of the times.
In the current study, shear stresses experienced by the leaflets of the BAV were measured and compared with that experienced by the leaflets of the TAV. The BAV geometry could alter hemodynamics that can cause significant changes in the shear stresses experienced by the BAV leaflets, introducing substantial unsteadiness in shear stresses.
The current project is partially funded by the National Heart, Lung and Blood Institute Grant HL-070262 and by donations by Tom and Shirley Gurley.
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: C.H.Y., N.S., N.V.V., and A.P.Y. conception and design of research; C.H.Y., G.T., and N.V.V. performed experiments; C.H.Y. analyzed data; C.H.Y. and N.S. interpreted results of experiments; C.H.Y. and N.S. prepared figures; C.H.Y. drafted manuscript; C.H.Y., N.S., and A.P.Y. edited and revised manuscript; A.P.Y. approved final version of manuscript.
We thank Mr. Holifield of Holifield Farms for the fresh porcine aortic valves.
- Copyright © 2012 the American Physiological Society