Carotid artery stenting (CAS) is undergoing rapid evolution. New devices specially developed for treating carotid artery stenosis have facilitated CAS over the past few years. Design improvement in second-generation devices, such as new stent materials, small application systems, monorail devices or protections systems, as well as sufficient cumulative experience, have contributed to a safer and easier procedure [1, 2, 3, 4]. This is especially important when dealing with a tortuous vessel anatomy. Over the past few years CAS has gathered clinical acceptance, especially when dealing with patients who are poor candidates for surgery.

In the vascular system the carotid bifurcation is an area with a complex hemodynamic configuration. As the flow is divided by the internal (ICA) and external carotid artery (ECA) a separation from the unidirectional laminar flow in the common carotid artery (CCA) occurs. In such locations, secondary flow patterns with complex flow profiles, such as vortices and recirculation zones, tend to form and have significant impact on the localization of atherosclerotic lesions [5, 6, 7]. To avoid additional flow disturbances and thus increase the risk of embolization or restenosis, carefully determined stent deployment adapted to the anatomic configuration is necessary, especially when the stent is deployed across the bifurcation and the ECA origin [8].

When performing CAS the use of self-expanding stents in combination with protection devices is common. A new concept for preventing embolization into the brain is a self-expanding nitinol stent with a porous membrane serving as a filter. As described previously by Müller-Hülsbeck et al. [9], this prototype has shown promising results with low embolization rates during stent deployment in an ex vivo. study.

The purpose of this experimental study was to investigate the influence of this newly developed membrane stent design on the flow behavior in a physiologic model of the carotid artery.

Materials and Methods

Models and Model Fluids

Three stents of differing design were investigated (Fig. 1A–D): a stainless steel Wallstent (Boston Scientific International, Paris, France), a nitinol SelfX stent (Abbott, Behringen, Switzerland), and a new prototype nitinol MembraX stent (Abbott, Behringen, Switzerland) covered inside with a semipermeable silicone-polyurethane copolymer (Elast-Eon). Elast-Eon is specially developed for long-term blood contact. Filaments of this material are woven into a mesh. According to the number of layers different porosities can be achieved. In this study the pore size was around 100 μm. In a second MembraX the membrane was modified at the outflow segment. Here the membrane ended within the last row of Z-shaped stent struts, thus increasing the contact area of the membrane with the vessel wall (Fig. 1E). All stents were the same size (6 mm in diameter and 32 mm in length, 6/32) and were positioned at identical locations in the CCA and ICA, spanning the ECA (Fig. 1F).

Figure 1
figure 1

The three different stent designs included in this study: (A) a stainless steel Wallstent, (B) a nitinol SelfX stent, (C) a prototype MembraX stent based on the SelfX but covered inside with a silicone-polyurethane copolymer, and (D) a modified version of the MembraX in which the membrane ended within the last Z-shaped row of the stent struts. E The modification to the MembraX increased the contact area of the membrane with the vessel wall. F Stent positioning in the CCA and ICA, spanning the ECA.

Full size image

Experimental models consisted of the common carotid artery (CCA), the internal carotid artery (ICA) and the external carotid artery (ECA) with an angle of 37° between ICA and ECA (Fig. 1F). These were 1:1 true-to-scale models made of transparent silicone rubber that were molded, postmortem, from human carotid arteries. The preparation technique is described in Liepsch et al. [10]. The carotid arteries are pressurized up to 60–100 mmHg prior to casting the silicone rubber core from an autopsy subject. Models prepared with this method have a similar roughness to the intima, and identical geometry and similar compliance as the original vessel [10].

As laser light is absorbed by the hemoglobin of the red cells, a blood-like model fluid is necessary for the simulation of blood. The fluid is transparent and has the same refractive index as the model wall. We used a dimethylsulfoxide (Merck-Schuchardt, Höhen-brunn, Germany)/water solution and added polyacrylamides (Seperan, DOW Chemical Rheinwerke, Rheinmünster, Germany). The density of the solution (ρ) was 1050 kg/m3. Titanoxide particles, with a diameter of 1 μm, were used as tracer particles for LDA measurements. The representative viscosity of the fluid was η = 4.9 mPa·sec. The elastic and viscous components were also similar to those of blood. The fluid is stable for approximately 3 weeks.

Experimental Setup

A physiologic, pulsatile flow was simulated with an experimental circulatory system. Fluid from an upstream tank was pumped to a straight tube almost 1.5 m long to achieve fully developed laminar flow at the entrance to the bifurcation. Pulsatile flow was created by a computer-controlled piston pump. The pulsatile waveform of a healthy human common carotid artery was obtained by noninvasive ultrasound Doppler velocimetry in a 25-year-old man and was used to drive the piston pump. A mean flow of 0.39 l/min (Reynolds number = 220) with a frequency of 60 min−1 was measured with an inductive electromagnetic flowmeter 50 mm in front of the CCA. The pressure was measured 30 mm upstream of the model with inductive pressure transducers. The experimental setup with flow, pressure and velocity settings is shown in Fig. 2. The flow rate ratio of the ICA to ECA was 70:30. The flow rate ratio was regulated by means of two moving flow regulation containers. The model was installed in a Plexiglas container and embedded in a glycerin/water solution with the same refraction index as the model wall. The measuring volume of the one-dimensional laser system was exactly positioned when the whole model with the Plexiglas box was mounted on an x-y-z moving table.

Figure 2
figure 2

Experimental setup with pressure (A) and velocity (B) setting.

Full size image

Velocity was measured with a 5-mW laser Doppler anemometer (LDA) system (BBC Goerz, Spectrophysics, Munich, Germany). LDA does not disturb the flow, has a high spatial resolution and a fast linear response. The system is not affected by temperature, pressure or density of the fluid. For the laser measurements, the whole system—including model container, container fluid, model and model fluid—must be transparent and have the same refraction index.

For the calculation of identical measurement planes, the apex of the bifurcation was defined as reference point X0 (Fig. 3A). LDA measurements were performed in 14 cross-sections in the CCA, ICA and ECA as shown in Fig. 3A. For each cross-section 69 measuring points were exactly defined (Fig. 3B). The axial velocity component was measured at each position over eight pulse cycles. For every pulse cycle, 100 measurements were performed and a mean velocity profile from the 800 velocity values for each point was calculated (e.g., Fig. 2B). For the analysis of the flow profile, the systolic peak velocity and the minimum diastolic velocity were compared with the velocities in the model without a stent. For the whole cross-section, a velocity profile of the axial velocity components can be calculated for any time point of the pulse cycle (Fig. 3C).

Figure 3
figure 3

A Cross-sections measured with LDA in the CCA, ICA and ECA. B LDA measurement points per cross-section. The upper wall is continued by the ICA. C For the whole cross-section, a velocity profile of the axial velocity components can be calculated for any time point of the pulse cycle (Fig. 2B). Here the flow velocity profile in the CCA 20 mm proximal to the flow divider (X0) at peak systolic velocity is shown. Proximal to the stents in the CCA typical laminar flow is found. All velocity profiles in the CCA show a laminar flow profile with similar flow velocities (C).

Full size image

Definitions

Flow separation is a rheologic term describing the tendency for fluid slipstreams to continue in a straight line as flow enters a bend, curve or bifurcation. If the velocity is high enough or the radius of the curve is small enough, the slipstreams break away or separate from the wall with the lesser curvature [6, 7]. The region along the lesser curve is called a separation zone, which typically contains fluid that is moving sluggishly, swirling, or even sometimes moving retrograde. In bifurcations, the slipstreams pass towards the wall adjacent to the carina and, because fluid is incompressible, velocity increases, whereas in the separation zone at the outer wall, low shear stresses and, sometimes, reversed flow may be found. These flow characteristics of a separation zone are portrayed by a typical “armchair profile” (Fig. 4A), with high velocities at the wall adjacent to the carina and low velocities at the opposite side. High-velocity shear gradients are found at the border between low and high velocities.

Figure 4
figure 4

Flow velocity profiles in the ICA 2.5 mm distal to the flow divider (D) at peak systolic velocity. Without the stent (A), the typical “armchair profile” of the separation zone in the ICA was clearly appreciated: low profiles at the outer wall (planes 1–4) with decreased velocities in the central slipstreams. At the inner wall, high velocities with laminar flow profiles were seen (plane −1 to −4). The SelfX (B) showed a similar flow profile without any hemodynamically relevant changes in the separation zone, whereas in the Wallstent (C) a more laminar flow profile with high velocities in the central slipstreams was found.

Full size image

Results

Flow Rate Ratio

With the Wallstent and the SelfX the flow volume in the ICA increased slightly. The stent grid augmented the resistance at the entrance of the ECA but the flow rate alterations were not significant. With the MembraX a significant (p < 0.05) shift of the flow rate occurred, but seemed to stay within acceptable limits. These data correlate with the findings of the flow profiles in the ECA (see below). The figures are shown in Table 1.

Table 1 Flow rate ratio ICA:ECA
Full size table

Inflow

Proximal to the stents a laminar flow profile was found. The velocity profiles in the CCA of all models were similar (Fig. 3C).

In the Stent

In the MembraX no measurements were possible as the laser light was completely absorbed by the membrane. In the Wallstent and the SelfX, flow profiles within the stents could be analyzed. However, the mesh grid of the stents sometimes interfered with the laser beam, resulting in signal reduction. Such points were eliminated from flow analysis.

In the proximal ICA the typical “armchair profile” of the separation zone was displayed in the reference model without a stent. At the outer wall low profiles with decreased velocities in the central slipstreams were found, whereas at the inner wall, high velocities with laminar flow profiles were observed (Fig. 4A). The SelfX (Fig. 4B) showed a nearly unchanged flow profile without any hemodynamically relevant changes in the separation zone. At the inner wall (Fig. 5A) the velocities decreased up to 0.14 m/sec and at the outer wall (Fig. 5B) a maximum increase up to 0.13 m/sec was observed. In the Wallstent a more laminar flow profile was found with a “diminishing” of the separation zone (Fig. 4C). At the inner wall (Fig. 5A) a similar flow profile was found compared with the reference model and the SelfX, but at the outer wall a laminar flow profile was developed with a maximum increase up to 0.38 m/sec (Fig. 5B). At the end of the stent the flow in the reference model returned to a more laminar profile at the outer wall. In the SelfX an increase in the velocities in the separation zone up to 0.22 m/sec was found, whereas in the Wallstent a decrease in the flow velocities, especially in the central slipstreams, up to 0.16 m/sec could be seen (Fig. 6). This means the center of the separation zone in the reference model and in the SelfX was located at the beginning of the ICA, whereas in the Wallstent the separation zone shifted to the distal end of the stent. Compared with the reference model the configuration of the separation zone with strongly decreased velocities in the central slipstreams seemed not to be that extensive in the Wallstent model (Figs. 5B, 6).

Figure 5
figure 5

Flow velocity profiles 2.5 mm distal to the flow divider in the ICA (shown in Fig. 4D) at peak systolic velocity. Near the inner wall (A) similar flow profiles were seen in the Wallstent and the SelfX with slightly decreased velocities compared with the model without a stent. Near the outer wall in the separation zone (B) a typical low profile with decreased velocities in the central slipstreams is shown in the model without a stent. With the SelfX a slight increase in velocities was found, but the profile shape stayed nearly unchanged, whereas in the Wallstent significant changes occurred with a high-velocity increase in the central slipstream.

Full size image
Figure 6
figure 6

Flow velocity profiles 11.5 mm distal to the flow divider in the ICA at the distal end of the stent. Near the outer wall in the separation zone the flow in the reference model returns to a more laminar profile. Compared with Fig. 5B there was an increase in the velocities in the SelfX, whereas in the Wallstent a decrease in the flow velocities was found, especially in the central slipstreams. That means the separation zone had shifted from the entrance of the ICA to the distal end of the stent.

Full size image

Outflow

Distal to the stents the flow profile in the SelfX and the Wallstent returned to a normal shape within 4 mm from the end of the stent (Fig. 7A,B). At the outer wall the velocities were increased to an almost normal level; however, the influence of the separation zone with decreased velocities in the central slipstreams was still found. With the MembraX a velocity increase of up to 0.59 m/sec, especially of the central slipstreams, was seen (Fig. 7C). Near the inner wall (planes −4 to −2) a flow separation was found in this model (Fig. 7C). The maximum intensity occurred 10 mm distal to the end of the stent. By modifying the membrane at the end of the stent struts (Fig. 1D) the flow separation at the inner wall distal to the stent was clearly reduced and a more laminar flow profile was observed (Fig. 7D).

Figure 7
figure 7

Systolic flow profile 15 mm distal to the flow divider in the ICA (E). The flow profile in the SelfX and the Wallstent returned to a normal shape within 4 mm from the end of the stent (A, B). The influence of the separation zone at the outer wall decreased compared with Fig. 6A and B. With the MembraX a velocity increase was found, especially of the central slipstreams (C). Near the inner wall (planes −4 to −2) a flow separation occurred in the MembraX with a maximum intensity 10 mm distal to the stent. With the modified MembraX the flow separation at the inner wall was clearly reduced (D).

Full size image

Side Branch (ECA)

With the SelfX (Fig. 8B) the flow in the separation zone at the outer wall was almost unaffected. At the inner wall (planes 3 to 4) velocity fluctuations up to 0.31 m/sec occurred caused by the stent struts. With the Wallstent (Fig. 8C) and the MembraX (Fig. 8D) the amplitude between the maximum and minimum velocity decreased, i.e., a calming of the flow with a more laminar profile was found. With the Wallstent the velocity fluctuations at the inner wall were not as extensive as in the SelfX. The MembraX showed a calming of the flow over the whole cross-section with a more laminar flow profile in the ECA and only small velocity fluctuations, especially at the inner wall. In the central slipstreams even a velocity increase of about 10% was found (Fig. 8D). In the diastolic phase a velocity decrease over the whole cross-section compared with the other models was found in the MembraX. Especially at the outer wall even negative velocities with values around 0.05 m/sec were seen in the MembraX but also in the reference model. With the SelfX and the Wallstent no negative velocities were found. Ten millimeters distal to the bifurcation the velocity fluctuations vanished in all stent models and a more parabolic profile developed, whereas in the reference model the effect of the separation zone was still evident.

Figure 8
figure 8

Systolic flow velocity profiles 2.5 mm distal to the flow divider (E) in the ECA. A Without the stent, the separation zone at the outer wall (planes −4 to −1) was clearly displayed. With the SelfX (B) the flow in the separation zone was almost unaffected. At the inner wall (planes 3 to 4) velocity fluctuations with decreased velocities occurred, caused by the stent struts. With the Wallstent (C) a velocity increase at the outer wall with a more laminar flow profile was found. At the inner wall also, velocity fluctuations as in the SelfX were seen but not as large. With the MembraX (D) a calming of the flow over the whole cross-section was observed despite small velocity fluctuations.

Full size image

Discussion

In clinical practice the use of self-expanding stents in combination with protection devices is common as there is considerable evidence, both clinical and experimental, that embolization takes place universally during all CAS procedures [11]. A review of the literature showed significantly lower stroke rates with protection than without [11, 12]. However, there are clinical limitations such as embolization around the filter or during filter retrieval. Also spasm, dissection or even “embolectomy” in the ICA may occur when the filter is accidentally pulled. Placing the device often leads to prolonged procedural times. Also late embolization may be a risk for delayed stroke. Based on these findings a self-expanding nitinol stent with a porous membrane was developed to serve as an integrated filter for preventing embolization into the brain during endovascular reconstruction. As described previously by Müller-Hülsbeck et al. [9] this prototype has shown promising results, with low embolization rates in an ex vivo study.

In this study the influence of this new stent design on the flow behavior in the carotid artery was investigated. As a reference model the standard Wallstent was tested. The SelfX is not common in CAS; but the MembraX stent concept is based on the SelfX design. It has a segmental architecture with rows of Z-shaped stent struts. This leads to a high flexibility and good adaptation to the vessel anatomy. To increase the contact area of the stent membrane with the vessel wall a small modification of the MembraX was made at the outflow tract as described above. The influence of this modification was also analyzed.

Velocity distribution in bifurcations seems to have significant impact on the localization of atherosclerotic lesions and thrombus formation. In separation regions hemodynamic alterations such as low wall shear stress, oscillations in shear directions, and long particle residence times are observed [5, 6, 7]. Thrombus formation is triggered in such regions as particles may stick more easily to the wall and to existing stenosis as a result of pathologic alterations in substrate exchange, thrombocyte activity and endothelial surface structure [13, 14, 15, 16, 17, 18, 19]. Flow separation is dependent on flow volume, flow rate ratio and vessel geometry [20]. Architecture of the stent struts, surface structure, longitudinal stiffness, flexibility and adaptation to the original anatomy all interfere with these parameters and induce hemodynamic changes after stent implantation.

In most clinical applications the ECA has to be bridged, as most atherosclerotic lesions are located at the entrance of the ICA. When the stent is placed across the bifurcation, it must adapt to arteries of different diameters and fit to the bend of the ICA. That implies changes in the original vessel anatomy, the flow rate ratio and the flow volume. In addition, the flow is disturbed by the mesh grid itself. Incorrectly positioning a stent or misjudging stent size and vessel geometry also lead to relevant disturbances of the flow behavior [8]. As a consequence such flow alterations may induce thrombus formation with the risk of early embolization as well as atherosclerotic plaque formation or intimal hyperplasia—the main cause of in-stent restenosis.

There were no changes at the inflow proximal to the stents in all models. With the SelfX and the Wallstent a shift in the flow in the ECA of 2.0% and 1.7%, respectively, was found. In these models the flow ratio stayed nearly unchanged. No relevant changes were seen in the ICA within the SelfX. With the Wallstent a shift of the separation zone from the entrance of the ICA to the distal end of the stent was found. Four millimeters distal to the SelfX and the Wallstent the flow profile returned to normal. In the ECA flow disturbances were seen at the inner wall distal to the stent struts in the Wallstent and more extensively in the SelfX. In contrast to the Wallstent with its twisted mesh grid the SelfX consists of a segmental stent architecture. Due to its higher flexibility and lower longitudinal stiffness the SelfX adapts better to the original vessel anatomy. That can be seen in the shift of the separation zone within the ICA in the Wallstent caused by a smooth straightening of the angle between the CCA and ICA. That means the bend of the proximal ICA in the bifurcation has shifted to the distal end of the stent as the longitudinal stiffness of the stent is rather high. This may lead to elongation or even kinking of the ICA distal to the stent and as a consequence increase the risk of restenosis in the elongated or kinked vessel segment. On the other hand the segmental architecture of the SelfX leads to a protrusion of the stent struts into the vessel lumen when the stent is deployed at a bend. That causes small disturbances in the border slipstreams as is shown in the ECA. To reduce such borderline disturbances the membrane of the MembraX could be useful. With the membrane a significant decrease in the flow in the ECA of 12.7% was found. That means a slight canalization of the flow in the ICA occurred in that model. A velocity increase of the central slipstreams was found, with creation of a flow separation distal to the end of the membrane. Therefore, a modified version of the MembraX was tested. With this modification the conformability of the membrane to the stent struts was improved and the flow separation at the distal end of the stent was significantly reduced (Fig. 7D). In the ECA the membrane induced a calming of flow with just a slight loss of flow volume. In contrast to the velocity gradients in the other models the MembraX showed a more laminar flow profile over the whole cross-section in the ECA. In the central slipstreams even a velocity increase about 10% was found. Thus an occlusion of the ECA orifice should not be expected from the hemodynamic point of view.

In clinical practice the membrane may fulfill two functions. On the one hand the disturbances caused by the stent struts in the border slipstreams are reduced, as with the membrane a smooth surface is created inside the stent. On the other hand the membrane works as an integrated protection device during stent deployment. Additionally it may prevent late embolization through the stent struts as the atherosclerotic material is protected permanently from dislodging by the membrane.

Limitations of This Experimental Study

With the MembraX flow measurements were done at the inflow, the outflow and in the side branch segment. As the membrane absorbed the laser light no signal could be received within the stent itself. In this blind window no data on the flow velocities were available, and flow analysis had to be done indirectly on the basis of the flow rate and the flow behavior distal to the stent.

There are still several additional limitations associated with the experimental setup: flow parameters were studied in models of a healthy carotid artery including pulsatile flow with physiologic blood pressure. In clinical settings, however, there are further important factors such as cardiac disease, hypertension, the degree, nature and location of the plaque, the presence of calcification, ulceration or even the presence of intimal flaps beyond the stent. Finally, many questions remain as to the influence of the stent and the porous membrane on the interaction with blood clotting reactions and the endothelial cell layer.

In spite of the above-mentioned limitations this well-established flow model offers a method for reproducibly testing the flow alterations of different stent designs under pulsatile flow conditions. The results provide information about the performance of the tested stents that might be expected in vivo. This information may be used for further developments in stent design. Additional aspects that can not be studied in a single in vitro model should be considered for evaluation in further in vitro trials and animal models prior to human use. Therefore these results should be extrapolated with care to the clinical situation.

Conclusion

Hemodynamically, stent placement across the carotid bifurcation induces alterations of the physiologic flow behavior. Depending on the stent design the changes are located in different regions. There is no “ideal” stent for the carotid artery. All the stents tested were suitable for the carotid bifurcation. The MembraX prototype—a new concept of CAS protected by a membrane stent—has shown promising hemodynamic properties ex vivo. Further evaluations of its biocompatibility are warranted prior to human use.