Turbulent stress measurements downstream of three bileaflet heart valve designs in pigs☆

Implantation of mechanical heart valves is a routine treatment for valvular heart disease, which dramatically prolongs life expectancy of patients and increases their quality of life. Unfortunately, mechanical heart valves still have several shortcomings. Complications such as thromboembolisms, valve thrombosis, hemolysis, tissue overgrowth, calcification and thrombocytopenia remain problems to be solved. Thromboembolic complications account for 75% of the complications seen after implantation of a mechanical aortic valve, and even though treatment has improved, there is still an overall 10-year mortality of 30–55% [1,2]. Initiation of thromboembolic events is caused both by the artificial material of the valve and by the valves’ interference with normal blood flow [3]. Therefore, continuous evaluation and testing of mechanical heart valves in terms of material and design is needed to provide input to future valve designs which entail minimal biologic damage. After implantation of a mechanical heart valve the blood flow is partially obstructed by the leaflets and the sewing ring of the new valve. This disturbs the flow and causes abnormal flow patterns such as turbulence. Several studies have associated the shear stresses produced by turbulence with pathophysiological changes such as platelet activation and damage [3–7] and hemolysis [4,8]. Whether the elements of the blood become activated and damaged by turbulent stresses depends both on the magnitude and the exposure time of the stresses. In healthy individuals with normal valves Reynolds normal stresses (RNS) have been measured to be below 4 N/m2[9]. Studies have shown that the absolute minimal critical turbulent shear tress (TSS) level for altering platelet function is around 50 N/m2 for exposures times within normal physiological ranges [4,10]. For red blood cells (RBC) the critical level is around 600–800 N/m2 for any direct and immediate alteration [7,11,12]. However, it has been suggested that in the presence of a foreign material, RBCs and platelets can be damaged as well as activated at lower values [13]. The formation of TSS is thought to depend on cardiac output. When blood flow is increased, more blood passes through the obstructed area per time unit and therefore velocity increases and more turbulence is formed. Thus, TSS levels will be higher during exercise with high cardiac output, than during rest when cardiac output is low. Several studies have tested different mechanical valves in order to identify optimal design and orientation to minimize interference with blood flow [9,13–16]. In order to minimize formation of TSS it is necessary to identify a valve design that interferes minimally with the blood flow. The Medtronic Advantage® valve is a bileaflet valve that differentiates itself from formerly known bileaflet valves by having a larger central opening with a wider space between the two leaflets. The larger central opening is hypothesized to minimize obstruction of the blood flow and thus formation of TSS. In order to investigate whether this design optimization is reflected by less turbulence we aimed to compare turbulence levels with two widely used bileaflet valve prostheses in a porcine model. The study comprised of 19 female Danish Landrace/Yorkshire pigs all weighing approximately 90 kg. This size was chosen to mimic the aortic size of human beings and to ease the access to the aortic orifice [17]. According to our own unpublished experience and other previous studies, native annulus diameters were assumed to be approximately the same size and were therefore not measured in order to reduce preoperative stress on the animals [18]. Turbulence measurements were conducted right after weaning from the extracorporeal circulation and under open chest conditions. The animals were euthanized immediately after the measurements were performed. Experiments were conducted at the Institute of Clinical Medicine, Aarhus University Hospital and were performed according to the guidelines and with approval from the Inspectorate of Animal Experimentation under the Danish Ministry of Justice. All valves were factory labelled 21 mm bileaflet aortic mechanical heart valves (Medtronic Advantage®, CarboMedics© Orbis™ Universal and St. Jude Medical® Standard). Since we have previously performed turbulence measurements on the SJM Standard and CM valves these were chosen as a reference for comparing with the MA valve. A fixed valve size was chosen because the pigs were of the same size and the results would therefore be easier to compare. General anesthesia was maintained by continuous infusion of ketamine/midazolam (Ketalar® 250 mg/Dormicum® 50 mg/h) and fentanyl (Haldid® 900 μg/h). The heart was exposed through a midline sternotomy. ECC was established via cannulation of the aortic arch and the caval veins and a vent (Sarns®) was placed through an apical incision. After cross clamping the ascending aorta one liter of a 4 °C crystalline cardioplegic solution (Kardioplex®) was given in the aortic root to induce cardiac arrest. The native aortic valve was removed and the mechanical valve was placed with one of the two large openings towards the non-coronary sinus. The placement of the valve is opposite recommendations on bileaflet valve placement and previous studies have shown a tendency towards higher turbulent shear stress values in this position [18,19]. Although the impact of valverotation is minor we chose this position as a marginal worst case scenario. During the operation cardioplegic solution (Kardioplex®) was given every 20 min directly into the coronary ostia. After aortic closure the cross clamp was removed and the heart was defibrillated and reperfused for approximately 60 min to allow stabilizing after the CPB. The aortic cannula and the two stage venous catheter were removed as was the left ventricular vent catheter. Micro-tip catheters (Millar Instruments Inc., Houston, TX) for ventricular pressure and aortic pressure measurements were put through the left ventricular apex and through the carotid artery. A 18 mm transit-time flow probe (Cardiomed, Norway) was placed around the pulmonary trunk to measure cardiac output. Turbulence measurements were performed as soon as the pigs had spontaneously stabilized hemodynamically in one of the required CO ranges. If CO was outside the selected range we interrupted the measurements and started the measurement series all over again for that CO range. Further details in anesthesia and surgery has been described elsewhere [17]. Turbulence is known as fluid motions that are random both in time and space. This means that to prove the existence of turbulence, simultaneous three-dimensional measurements of velocity fluctuations are theoretically needed. However, feasible methods of three-dimensional measurement of turbulence in vivo are not yet available. Thus, the existence of turbulence has to be disclosed otherwise. Results in this study will therefore be shown as RNS values rather than TSS values. The radial velocity components were measured by two 10 MHz ultrasonic probes. The probes were placed in a C-shaped PVC shell that was placed on top of the ascending aorta with the probes in −45° and +45° position (Fig. 1a). Schematic drawing of a section of the aorta with the two positions of the C-shaped PVC shell placing the probes in the four different positions. Each probe was used to record blood velocity in five fixed sample volumes through the diameter of the vessel. The size of the aorta was measured to find a compatible shell and to set the range gate delay (depths) of the pulsed ultrasound system. The correct position of the shell was ensured by marking 0° on the aorta. By rotating the shell from 0° to 45° the probes were positioned at 0° and 90° (Fig. 1b). Seventeen different points of measurement were obtained by switching between the two probes, rotating the shell and changing the depth of the measurement (Fig. 1c). Mean sample volumes were approximately 2 mm in diameter and 1 mm in axial length. The ultrasonic probes were connected to a modified pulsed Doppler ultrasound (PDU) velocimeter (ALFRED, Vingmed, Norway). The velocity measurements from the PDU as well as ECG, cardiac output, left ventricular pressure and aortic pressure were recorded on a datarecorder and digitized. At least 20 heart cycles were recorded from each of the 17 measuring points to calculate turbulence components. The animals were held hemodynamically stable and measurements were performed at three different cardiac outputs (3.5, 5 and 6.5 L/min, all ±0.5 L/min). Data were analyzed by a specially developed LabView program (National Instruments, Austin, TX; Version 6.0i). The velocity signal was divided into eight time windows each lasting 50 ms. Each time window overlapped with the previous window by 25 ms. Thus measurements of the formation of RNS through 225 ms of peak systole were obtained. Acquisition of data was triggered by the rise of pressure in the left ventricle or by the ECG (Fig. 2). The basic principles for calculating the development of turbulent stresses (RNS) are schematically shown in this figure. (A) The electrocardiogram was used as a time reference for signal analysis. (B) The velocity component from each of the 17 points recorded was computed in eight time windows (C). This was used to calculate RNS values in each of the eight time windows (D), so that RNS levels for the 17 points throughout the systole are obtained. A: anterior; P: posterior; L: left; R: right; t: time; u′(t) turbulent velocity component; w(t) time window. Groups were compared by Bonferrani correlated t-test after ANOVA analysis at more than two groups. Of the 19 animals in the study, the first two failed due to technical flaws in our measurement equipment. No subsequent technical difficulties were seen. Animal number three died from metabolic acidosis and four animals died from trying to obtain a high cardiac output under extreme hemodynamic conditions (experiments number 6, 7, 9 and 15). The pigs died from ventricular fibrillation or from bleeding. None of these conditions were related to the implanted valve which was normally functioning in all cases (verified by post-mortem examination). In total data from 12 animals were obtained, distributed evenly between the three valve types. Mean cross clamp time was 68.5 min (57–91) and reperfusion time was 61 min (37–150). Measurements at each cardiac output level were performed within approximately 15 min. The highest instantaneous RNS value observed in these experiments was 47 N/m2, which was observed at high cardiac output (Table 1). Mean spatial RNS values measured in the cross sectional area of the ascending aorta was between 3 and 15 N/m2 (Table 2). Thus, all valves produced levels of turbulence clearly below the stated critical threshold for blood cell damage. In the middle cardiac output range the Carbomedics valve exhibited statistically significant higher maximum RNS values than the other two valves (p = 0.02) and the St. Jude Medical valve statistically significant lower mean turbulence values than the Medtronic and Carbomedics valves (p = 0.007). Maximum Reynolds normal stress values for the investigated mechanical heart valves Mean Reynolds normal stress values for the investigated mechanical heart valve prostheses Also in the high cardiac output range the St. Jude Medical valve exhibited statistically significant lower turbulence values than the other two valves (maximum turbulence, p = 0.03 and mean turbulence, p = 0.016). At low cardiac output there was no statistically significant difference between the three valves at neither mean nor maximum values (p > 0.05). There was a general tendency towards higher RNS values with increasing cardiac output. This tendency was statistically significant and was seen at both maximum (p = 0.0033) and mean (p = 0.00037) values of RNS. Since the formation of turbulence depends on stroke volume rather than cardiac output, this was calculated for all groups. There was no statistically significant difference in mean pulse or mean stroke volume between the three valve types at any of the cardiac output groups (p > 0.05). Measurement of turbulent stresses has been established as an important part of evaluation of new valve type designs. While the St. Jude Medical and the Carbomedics valves have been tested widely through the past years [9,13,14,22] turbulent stresses have so far not been measured in the newer Medtronic Advantage valve. In this study we measured turbulent stresses over a wide physiologic range and found that none of the three valve designs exceeded the lowest critical threshold for thrombocyte and erythrocyte damage or activation. We did find a statistical significant difference in turbulence formation between the three valves especially in the high CO ranges. It was not possible to increase CO further than 7 L/min and therefore information on whether the turbulence curves of the three valves would converge at higher COs is not available. In the tested ranges of CO it is obvious that though statistically significant different there is no clinically relevant difference between the three valves in terms of turbulence formation. Two main onsets of thromboembolic complications associated with mechanical heart valves are known. Firstly, turbulent shear stress above a critical level that causes damage to and direct activation of the formed elements of the blood. Secondly, the presence of foreign non-biological material in the blood stream induces activation of the platelets and causes them to aggregate. Some studies suggest that the presence of non-biological materials play the primary role in activation while other investigators believe that turbulence is the most important factor in the activation of platelets [3]. However, both turbulence and non-biological material is present at all times after implantation of a mechanical heart valve whereby both factors influence the elements of the blood at all times. Nygaard et al. [13] has suggested that the two factors actually intensify each other. Alteration of the influence of both non-biological material and turbulent shear stresses in addition to increased exposure times due to the vortex phenomenon and continuous passage through the valve suggests that the critical limit for damage or activation of platelets and RBCs may be lower than the value needed for damage from turbulence alone. The relation between exposure time and formation of turbulence has been investigated by Brown et al. who found a linear correlation between exposure time and damage to the platelets. Acid phosphatase release doubled after one passage through the shear stress area. This indicates that the damage and activation of the thrombocytes accumulates by every passage of the valve [23]. The fact, that there is a cumulative damage and activation caused by shear stresses could mean that the thrombocytes are altered even though the critical lower threshold for damage is not reached. Furthermore, a vortex is created downstream of the valve due to its obstruction of the blood flow. Some of the blood re-enters into this vortex and into the area of high TSS values. In this way blood that has been exposed to TSS once and is containing already activated and damaged platelets and RBCs, re-enters the area where they are affected by high levels of TSS several times [10,24]. The concept of a vortex distal to the valve implies that the exposure times and cumulative stress affections might be increased [10]. This means that even though turbulent shear stress values are lower than the critical threshold level alteration may still occur. RNS values measured preoperatively in human beings by Nygaard et al. [9] were below 4 N/m2. The mean RNS values for all three valve types in this study are similar to the mean RNS values measured in native valves in 17 pigs by Hasenkam et al. [14]. Hasenkam et al. also measured RNS values in six pigs after cardioplegia but without implantation of a valve. Here the means of mean RNS values were 18 N/m2 (SD 12.9) with maximum values reaching as high as 57 N/m2. The mean cardiac output was 5 L/min (SD 1.8). Comparing the RNS values of the present study to those for native valves before and after cardioplegia in the study by Hasenkam et al. show that the RNS values in the three valve types in our study is within normal physiological ranges in pigs. Nygaard et al. tested the St. Jude Medical valve in nine human beings and measured spatial mean RNS values between 4 and 13 N/m2. They furthermore tested the Carbomedics valve in six human beings and found spatial mean RNS values between 4 and 18 N/m2[14]. In comparison we found RNS values for the St. Jude Medical valve to be between 6 and 8 N/m2 and Carbomedics values to be between 6 and 13 N/m2 depending on cardiac output. Other pig studies also show values that are similar to those found in our study [19]. In vitro studies have shown higher turbulent stresses downstream of mechanical heart valves than in vivo [13]. This can be due to difference in distance between the measuring points and the valve. When measurements of velocity are performed in vivo it is not possible to keep the transducers as close to the valve as it is in vitro. RNS values decrease dramatically with distance from the valve which suggests that values measured in vivo are underestimated [13]. Also, turbulence measurements in vitro are performed by measuring TSS directly instead of estimating them from RNS values which could result in TSS values that are more inaccurate than those measured in vitro. Furthermore, keeping the animals hemodynamically stable for the time required to perform the measurements was difficult. Reperforming measurements every time CO was outside the set ranges, even though only for a very short period of time, led to long hours post CPB. The stress on the hearts was in some cases too much and not all measurements were performed. Another limit of the animal studies was that the highest possible cardiac output to reach immediately after heart surgery was approximately 7 L/min. However, a porcine model was chosen in this project to mirror the anatomic and physiologic conditions of the human heart as close as possible; especially during increasing work loads on the heart mimicking exercise. In this study we found that at low cardiac output all three valve types had low mean RNS values comparable to those seen in pigs with native valves. When we increased cardiac output we observed that the mean RNS values of all three valve types increased. Although statistically significant different, none of the three valves exhibited RNS values exceeding the critical limit for immediate blood cell damage at any of the cardiac outputs measured. In conclusion, all three valve types have low turbulent shear stress values that do not theoretically cause any direct or immediate alteration of the formed elements of the blood. Therefore, the difference in the turbulent stresses between the three valves has no clinically relevance. The authors wish to thank Tanja Thomsen for her time and expertise in performing these experiments and Niels Trolle Andersen for statistical consult. The authors also wish to thank The Danish Research Agency, The OTICON Foundation, Helga and Peter Kornings Foundation and Snedkermester Sophus Jacobsen og hustru Astrid Jacobsen Foundation for their generous financial support. Medtronic Inc. kindly supported the project by providing the Medtronic Advantage valve and by partial payment of the experimental costs.