Professional Documents
Culture Documents
Bio 141 11 111002
Bio 141 11 111002
net/publication/332902529
CITATIONS READS
3 121
5 authors, including:
Some of the authors of this publication are also working on these related projects:
Numerical Prediction of Thrombosis Risk in Left Atrium under Atrial Fibrillation View project
All content following this page was uploaded by Yonghui Qiao on 23 March 2020.
intentionally during TEVAR, and LSA revascularization was not of 1080 kg/m3 [10]. The non-Newtonian viscosity of the blood
carried out prior TEVAR due to a dominant right vertebral artery. flow was defined by the Carreau–Yasuda viscosity model [11].
After 1 yr follow-up, the coverage of LSA was well tolerated. The The flow was considered to be laminar.
preoperative and postoperative fluid domains were generated from A pulsatile flow waveform extracted from a previous study [1]
the original computed tomography angiography images using was applied at the inlet of the ascending aorta. Three-element
MIMICS 19.0 (Materialise, Leuven, Belgium), with supra-aortic Windkessel model was adopted at the outlets of the supra-aortic
branch arteries retained, and other small vascular branches branches and the descending aorta, and the values of the model
excluded (Fig. 1). All the branches on the aortic arch were parameters were calculated in accordance with Pirola et al. [12],
extended upward by 50 mm for the rationality of the geometric as shown in Table 1. However, the necessary patient-specific flow
models. Notably, there was no LSA in the postoperative geometry rate data were not available in the present study. Thus, simulations
because the connection between LSA and the aortic arch gradu- with zero pressure outlets were preconducted to obtain approxi-
ally disappeared during follow-up. The descending aorta was mate flow rate distribution [13]. Detailed workflow to determine
remodeled by the stent graft implanted in the true lumen. The the model parameters is illustrated in Fig. 1(f). The postoperative
false lumen gradually became thrombotic as the proximal tear was parameters of Windkessel model were consistent with preopera-
blocked. However, thrombosis has not occurred completely due to tive simulation.
the presence of the distal tear. So, the residual false lumen was In the FSI analysis, the arbitrary Lagrange–Eulerian method
observed in the follow-up. was applied to describe vessel wall displacement [14,15]. The ves-
The thickness of vessel wall was simplified to be a uniform sel wall was assumed to behave as a linear elastic material with a
thickness and the outer wall of the solid domain was generated by uniform density of 2000 kg/m3, a Young’s modulus of 2.7 MPa,
extruding the wall of the fluid geometry outward 2 mm in GEOMA- and a Poisson ratio of 0.45 [16]. The FSI simulation in our study
GIC STUDIO (GeoMagic, SC). Then the outer wall and the inner wall was two-way and the diastolic pressure in the dissected region
replicated from fluid domain could be combined into a solid was preloaded on the external vessel wall [17]. The center points
domain by closing all the openings. It should be noted that there of solid boundaries were fixed to achieve numerical stability
was a gap between the true lumen and false lumen in preoperative and save computational resources [17,18]. In the postoperative
geometry. Therefore, the thickness of the reconstructed solid flap
was equal to the width of the gap, which was thinner than the
thickness of the vessel wall. This method of constructing a solid Table 1 The parameter values of the three-element Windkessel
domain has been reported in our previous work [9]. Another post- model used in the pre-TEVAR aorta, post-TEVAR aorta and
operative geometry was constructed by virtually adding LSA for post-TEVAR aorta with the LSA added
comparison. All the computational domains were meshed using
ANSYS-ICEM (ANSYS, Canonsburg, PA) and the solid and fluid
Outlet R1 (107 Pa s m 3) C (10 10
m3 Pa 1) R2 (108 Pa s m 3)
domains had more than 100,000 and 3,000,000 elements, respec-
BT 6.23 28.99 5.55
tively. Meanwhile, five boundary layers were defined at the near- LCA 21.05 9.05 17.68
wall region. Grid-independence tests were performed for all three LSA 12.88 14.12 11.38
cases and the grid number chosen here was reasonable. DA 1.82 100.37 1.60
simulations, the effect of stent graft had been considered. Due to preoperative aorta, it can be seen that the majority of the flow
the absence of the geometry and mechanical properties of the (57.37%) actually goes through the false lumen due to the consid-
stent-graft, we assumed that the vessel wall covered by the stent erable cross-sectional area. However, only 8.43% of the blood
graft was totally stiff and cannot deform during a cardiac cycle. flows through the true lumen, which could result in ischemia to
Considering the high stiffness of the stent graft, this strategy was those vascular branches connecting with the true lumen. The
a reasonable assumption. Specifically, a fixed support setting was amount of flow crossing the supra-aortic branches, which supplies
applied at the composite stent-wall region, which could make the the cerebrovascular and upper limb, is 34.20%. After TEVAR, the
solid domain approximately rigid. The FSI simulation was devel- descending aorta is remodeled and the LSA disappears gradually.
oped on ANSYS WORKBENCH 16.1 (ANSYS, Inc.) by coupling Tran- The blood supply to the true lumen is strengthened and increases
sient Structural with CFX and the three-element Windkessel to 42.33%. It should be noted that the blood flow going through
model was defined with CFX expression language. the supra-aortic branches has risen to 43.90%. Compared with the
The constant time-step was set as 0.001 s and the maximum presence of the LSA, the amount of aortic branches blood supply
residual for convergence is 10 5. All the simulations were run for further increases (44.97%). However, the blood supply to the true
three cardiac cycles to reach a periodic solution and the third car- lumen, which is adjacent to the root of the LSA, drops to 36.49%.
diac cycle data were used for analysis. Energy loss (EL) is related to cardiac workload and the defini-
tion could be found in previous study [21]. The comparison of EL
throughout a cardiac cycle is shown in Fig. 4. It could be seen that
Results the EL in all cases reaches the maximum before peak systole. The
Wall Pressure. Wall pressure is a key factor in the develop- preoperative average EL is 0.0327 W and the minimum is the low-
ment of aortic dissection. Figure 2 shows the comparison of wall est relative to the other two postoperative cases. After TEVAR,
pressure distribution between the preoperative and postoperative the value of EL gets higher and average value is 0.0578 W. A
aortas at peak systole. It can be observed that the preoperative small decrease (2.7%) on the magnitude of EL is observed and
pressure reduces along the descending aorta and the highest pres- average value drops to 0.0563 W when the LSA is not covered.
sure region locates at the ascending aorta and the aortic arch,
which is consistent with previous studies [19,20]. In addition, the Wall Shear Stress. Wall shear stress is critical as it affects the
proximal pressure in the true lumen is higher than that in the false endothelial cells lining the vessel wall [22] and modifies the
lumen. After TEVAR, the maximum wall pressure is increased structure of the inner vessel wall [23]. Two WSS-based indices,
and the highest pressure region extents to the descending aorta. It time-averaged WSS (TAWSS) and oscillatory shear index (OSI),
should be emphasized that the pressure is relatively lower in the are analyzed to describe the WSS characteristics [24]. The effect
distal residual false lumen. When the LSA is added virtually, the of TEVAR on TAWSS distribution is illustrated in Fig. 5. It can
overall wall pressure distribution is similar, while a decrease in be seen that the region with high TAWSS value is located at the
the magnitude of pressure on the aortic branches and residual false distal part of the true lumen at the preoperative aorta. Notably,
lumen can be observed. TAWSS in the proximal false lumen is higher than that of the true
lumen. After TEVAR, the high TAWSS region gets wider and the
Blood Flow and Energy Loss. Figure 3 investigates the effect TAWSS in the stenotic region located at the distal part of
of TEVAR with LSA coverage on the proportion of blood flow descending aorta is elevated. When the LSA is added, there is no
crossing each outlet throughout a cardiac cycle. In the considerable difference on the overall distribution of TAWSS,
and only the TAWSS maximum reduces slightly. The comparison OSI (more than 0.25) is also shown in Fig. 6. When the LSA is
of low TAWSS (less than 0.4 Pa [25]) is also shown in Fig. 5. At added, the region exposed to high OSI at the top of residual false
the preoperative aorta, the surface area exposed to low TAWSS lumen gets wider, indicating that the coverage of LSA could
appears at the proximal true lumen adjacent to the entry tear. After weaken the oscillatory nature of the flow.
TEVAR, the low TAWSS region mainly locates at the stented
segment and the top of residual false lumen, which is more prone Wall Displacement. In general, the greatest vessel wall dis-
to thrombosis [1]. It should be noted that the low TAWSS area placement takes place around peak systole when the inlet velocity
expands slightly when the LSA is present. of blood reaches the highest value. The comparison of the aortic
Figure 6 shows the comparison of OSI distribution, which is wall displacement among pre- and post-TEVAR aortas at peak
related to flow oscillation. High value of OSI can be observed at systole is shown in Fig. 7. The preoperative aortic arch and
the ascending aorta, supra-aortic branches, and the true lumen at branches deform the most while the distal descending aorta shows
the preoperative aorta. Most of the false lumen exhibits relatively small displacement. Notably, the maximum wall deformation
low OSI. After TEVAR, the high OSI region concentrates on the (1.034 mm) is observed at the proximal part of descending aorta
aortic arch intrados and the top of residual false lumen. The latter and it is approximate to the range of Karmonik’s report [26]. After
position coincides with the low TAWSS region. The OSI in the the stent graft is implanted, the average deformation of the aorta
distal descending aorta is close to zero indicating that blood flows decreases sharply and high displacement is observed at the
pass this zone with nearly no disturbance. The comparison of high ascending aorta, aortic branches, and distal descending aorta.
When the LSA is virtually added, the distribution of wall displace-
ment is analogous but the deformation of distal descending aorta
further diminishes. Moreover, the surface area exposed to high
displacement locates at the left carotid artery while the other two
branches show small deformation.
Discussion
The clinical goals of TEVAR for acute type B aortic dissection
include coverage of the entry tear, expansion of the true lumen
with restoration of blood supply, and obliteration of false lumen
flow with thrombosis. However, few literatures have investigated
the pre- and post-treatment hemodynamics in patient-specific aor-
tic dissection [27]. Karmonik et al. [28] reported a hemodynamics
comparison between pre- and poststent graft placement in an
acute type B aortic dissection basing on the dynamic magnetic
resonance image data, and the elimination of large WSS magni-
tude and flow rate was observed in the postoperative residual false
lumen. Menichini et al. [29] predicted TEVAR-induced thrombus
Fig. 4 EL profiles during a cardiac cycle. Horizontal dashed evolution in type B aortic dissection, which is consistent with clin-
lines represent respective average value. ical follow-up. The present study focuses on the efficacy of
Fig. 6 The comparison of OSI characteristics (top) and the comparison of high OSI regions (>0.25) (bottom): (a) OSI distribu-
tion in the preoperative aorta, (b) OSI distribution in the postoperative aorta, and (c) OSI distribution in the postoperative aorta
with the LSA added
TEVAR and the influence of LSA coverage on the important branches’ blood supply increases further (44.97%), indicating the
hemodynamic parameters. Moreover, a new workflow, which is coverage of LSA has a major impact on the blood flow distribu-
suitable for cases with the lack of in vivo data, is proposed to cal- tion. In addition, the region exposed to low TAWSS expands
culate the model parameters of three-element Windkessel. slightly, which means an increased probability of thrombosis
From the preoperative simulation results, we can see that the when the LSA is not covered. The high OSI region also gets
proximal wall pressure in the true lumen is higher than in the false wider, indicating that the presence of LSA could strength the
lumen and the pressure difference may promote true lumen oscillatory nature of the flow and promote the remodeling of the
enlargement [19]. The flow rate crossing the compressed true residual false lumen. The magnitude of EL is not remarkably
lumen is only 8.43%, which could lead to ischemia in related altered, while the wall deformation of distal residual false lumen
organs. Obvious EL fluctuation through a cardiac cycle is further decreases when the LSA is present, which may be benefi-
observed, which may result from complex dissection structure. cial for the patient [30].
Moreover, proximal TAWSS in the false lumen is higher than that The chief limitation of this work is that only a single patient is
of true lumen, indicating a potentially expansion of the false analyzed. However, the coverage of LSA during TEVAR is still a
lumen while the OSI in the proximal true lumen is relatively high, controversial issue in the clinic and it is only suitable for a small
which means that the blood flow is unstable. For vessel wall, the number of people [4]. The value of this study is to demonstrate
maximum deformation (1.034 mm) is located at the proximal the effect of LSA through the combination of virtual surgery and
descending aorta. In summary, the preoperative patient suffering computational fluid dynamics. More patient-specific studies
from acute aorta dissection has a high risk of further deterioration. would be conductive to arrive at some clinical utility. Second, the
After TEVAR, the pressure distribution is altered due to the arterial wall and the intimal flap are assumed to behave as a linear
remodeling of the descending aorta and the region of high pres- elastic material and the center points of the boundaries are fixed in
sure expands to the distal stenosis where the stent-graft minimizes our model. As a result, the deformation of the ascending aorta is
the subsequent possibility of wall expansion. The blood supply limited due to the relatively short length. Finally, considering of
going through the supra-aortic branches is strengthened and the high stiffness of the stent-graft, the vessel wall covered by the
increases to 43.90%. However, average EL is elevated, which stent-graft is assumed to be rigid during a cardiac cycle in the
may be associated with the presence of distal stenosis and residual postoperative cases. Real mechanical properties of the stent graft
false lumen. The high TAWSS region extends obviously and an would be considered in future study.
elevated TAWSS in descending aortic stenotic region is observed
owing to the diameter reduction. It should be noted that the top of
residual false lumen shows relatively low TAWSS, which is more Conclusion
prone to thrombosis [1,25]. Meantime, the average deformation of The present study investigates the blood flow in an acute type B
the aorta is significantly reduced when the impact of the stent aortic dissection undergoing TEVAR with LSA coverage inten-
graft is considered. The comparison between preoperative and tionally. From the hemodynamic comparison between the pre-
postoperative results shows that TEVAR is able to improve hemo- and post-TEVAR aortas, most of the hemodynamic parameters
dynamic parameters, especially the blood supply and wall are obviously improved, indicating that TEVAR may have a good
deformation. efficacy on acute type B aortic dissection, which is consistent with
When the LSA is virtually added in the geometry model, the the clinical phenomena that most clinical patients are well toler-
wall pressure distribution is similar and the magnitude reduces ated after intentional LSA coverage. After virtually adding LSA
slightly. It is worth emphasizing that the amount of aortic in post-TEVAR model, notable changes in the blood flow to