| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 1994;107:438-0446
© 1994 Mosby, Inc.
Surgery for Acquired Heart Disease |
Aarhus, Denmark
Supported by the Danish Heart Foundation and Thomas B. Thriges Foundation.
Received for publication Feb. 25, 1993. Accepted for publication June 22, 1993. Address for reprints: Hans Nygaard, ME, Professor of Biomedical Engineering, Department of Cardiothoracic Surgery, Skejby Sygehus, Aarhus University Hospital, 8200 Aarhus N, Denmark.
Abstract
High levels of turbulent stresses resulting from disturbed blood flow may cause damage to red blood cells and platelets. The purpose of this study was to evaluate the spatial distribution and temporal development of turbulent stresses downstream of three mechanical aortic valve prostheses in human subjects: the St. Jude Medical, the CarboMedics, and the Starr-Edwards silicone rubber ball. Blood velocity measurements were taken at 17 measuring points in the cross-sectional area of the ascending aorta 5 to 6 cm downstream of the aortic anulus with the use of a perivascular pulsed Doppler ultrasound system. Turbulence analysis was done for each of the 17 measuring points by calculating the radial Reynolds normal stresses within 50 msec overlapping time windows during systole. By coordinating the calculated Reynolds normal stress values for each time window and for all measuring points, computerized two-dimensional color-coded mapping of the turbulent stress distribution during systole was done. For the St. Jude Medical valves the highest Reynolds normal stress (27 to 63 N/m2) were found along the central slit near the vessel walls. The temporal development and spatial distribution of Reynolds normal stresses for the CarboMedics valves were quite similar to those of the St. Jude Medical valves with maximum Reynolds normal stress values ranging from 19 to 72 N/m2. The typical Reynolds normal stress distribution for the Starr-Edwards silicone rubber ball valves was asymmetric, revealing the highest Reynolds normal stresses (11 to 56 N/m2) at various locations in the annular region between the ball and the vessel wall. The spatial distribution and temporal development of turbulent stresses downstream of the three investigated mechanical aortic valve prostheses correlated well with the superstructure of the valves. The maximum Reynolds normal stresses for the three valve types were in the same order of magnitude with exposure times sufficient to cause sublethal damage to red blood cells and platelets. (J THORAC CARDIOVASC SURG 1994;107:438-46)
Velocity fields and turbulent stresses downstream of various artificial heart valve designs have been studied in detail during the past decade, and results have suggested the possibility that elevated levels of turbulent stresses may contribute to thromboembolic complications, hemolysis, and endothelial damage.
1-7 The relationship between hemodynamic parameters that indicate high levels of turbulent stresses and the degree of intravascular hemolysis after implantation of artificial heart valves is today well documented. 8 Thromboembolic complications include thrombosis, thromboembolism, and anticoagulation-related hemorrhage and are major causes of morbidity and mortality after heart valve replacement. 9
High levels of shear stresses inherent in turbulent flow have a direct effect on platelets and on erythrocytes, resulting in the liberation of thrombospecific enzymes that are of special importance for thrombus formation. 10, 11 Therefore the role of fluid mechanics in thromboembolic complications and thrombus formation has been given increasing attention, because thromboembolic complications and thrombus formation represent about 75% of all valve-related complications in patients with mechanical aortic valve prostheses and about 50% in patients with aortic bioprostheses. 12
Damage to blood corpuscles seems to be associated with the magnitude, the exposure time, and the spatial distribution of the turbulent shear stresses (TSS). According to in vitro studies the critical TSS levels for lethal erythrocyte and thrombocyte damage is 150 to 400 N/m2 for exposure times within physiologic ranges. 13 However, in the case of mechanical prostheses, because of the presence of foreign surfaces, erythrocytes can be damaged by shear stresses on the order of 1 to 10 N/m2, 14, 15 and platelet function can be altered by shear stresses on the order of 20 to 60 N/m2. 16-18 Platelet damage seems to increase linearly with time of exposure to a constant level of shear stress, which indicates that shear-induced platelet damage is cumulative. 19 Yoganathan, Wick, and Reul 20 suggested that during an individual excursion through the valve prosthesis, the combination of shear stress magnitude and exposure time may not induce platelet aggregation. However, as a result of multiple journeys through the artificial valve, shear-induced damage may accumulate to a degree sufficient to promote thrombosis and embolization. Thrombi can form near the prosthesis or directly on parts of the foreign, nonbiologic surface of the valve. The cause of thrombus formation near the prosthesis is considered to be associated with local hemodynamic disturbances such as elevated levels of turbulent shear stresses and regions of stasis. 9, 20, 21 Rams and associates 22 found with a newly developed transcranial Doppler ultrasonography method that approximately 50% of patients with mechanical aortic valves had detectable microemboli in the basal intracranial vessels. Butchart 23 suggested that microthrombus deposited on atheromatous plaques in the ascending aorta as the result of disturbed blood flow downstream of an aortic prosthesis could be the source of microemboli detected in the middle cerebral artery in patients with mechanical aortic valves. Wurzinger and Schmid-Schönbein 11 found from in vitro experiments that shear stresses exceeding 50 N/m2 lead to irreversible damage of platelets within exposure times of 0.1 second and that the presence of red cells enhances the effects of shear forces on platelets by biophysical and biochemical mechanisms. Previously, Schmid-Schönbein and associates 24 found that thrombotic events involving the activation and aggregation of platelets take place under conditions of abnormally high shear stresses and that rupture of as few as 0.05% of red cells, subjected to shear stresses greater than 50 N/m2, liberated sufficient adenosine diphosphate to activate the platelets chemically in flowing blood. Stein and Sabbah 25 established arteriovenous shunts in dogs and showed that more thrombi occurred within a vascular shunt that produced turbulent flow than within a shunt in which the flow was laminar. The weight of thrombi that accumulated within the turbulent system appeared to be related to the intensity of turbulence.
These and other studies that show a relationship between shear stress levels and platelet damage strongly indicate that hemodynamic disturbances in the vicinity of prosthetic heart valves may be at least a contributing factor to some of the thromboembolic problems observed after heart valve replacement. Therefore, additional in vivo studies are required to characterize the hemodynamics of different valve types to propose new designs with smaller areas of high shear stresses and regions of flow stasis.
Numerous in vitro studies have clearly demonstrated TSS values downstream of different types of artificial heart valves sufficiently high to cause lethal or sublethal damage to blood cells. 2-6 The question is whether threshold limits for blood cell damage are reached in the human circulation after heart valve replacement. This is a difficult question to answer because quantitative human studies on turbulent stresses are scarce, probably because suitable perioperative measuring techniques have not been available. For practical and ethical reasons, measuring techniques applied perioperatively on humans have to be safe, simple, fast, accurate, and minimally invasive. A technique with these qualities was recently developed by Nygaard and associates 26 and was used in the present study with the purpose of evaluating the temporal development of the turbulent stress distribution downstream of St. Jude Medical (St. Jude Medical, Inc., St. Paul, Minn.), Starr-Edwards silicone rubber ball (Baxter Healthcare Corp., Santa Ana, Calif.), and CarboMedics (CarboMedics, Inc., Austin, Tex.) aortic valve prostheses implanted in human patients.
Theoretic considerations.
Turbulent flow is characterized by an unsteady and irregular eddying motion that is superimposed on the mean flow. These fluctuations appear random in space and time and possess a continuous spectrum of length scales and frequencies. They can only be described in statistical terms.
Hinze 27 defined turbulence as "motion in an irregular condition of flow in which the various quantities show a random variation with time and space coordinates so that statistically significant average values can be discerned." According to this strict definition, the presence of turbulence could be proved only by simultaneous registration of random velocity fluctuations in all three dimensions. In practice, however, the process has been so well characterized in many fluid dynamic studies that random fluctuations in one plane can be accepted as an evidence of turbulence. 28
The fluid stresses associated with turbulence, the so-called Reynolds stresses, can be interpreted as stresses that acts on an element of fluid. The turbulent stress acting normally on an element of the fluid is called Reynolds normal stress (RNS) and is defined as
where u'n is the turbulent velocity component in the axial direction and p the fluid density. The TSS or Reynolds shear stress is defined as
and can be interpreted as the stresses acting tangentially on a surface of an element of the fluid. u'i and u'j are orthogonal turbulent velocity components recorded simultaneously. TSS can be measured directly in vitro with a two-component laser Doppler anemometer system, 2, 7, 29 but it is so far impossible to measure in vivo, because of lack of feasible methods. RNS, on the other hand, can be calculated from a single velocity component, which can be measured in vivo by hot-film anemometry 30 or pulsed Doppler ultrasonography (PDU). 26
If the radial and axial turbulence intensities in the flow field downstream of a heart valve are considered to be of the same order of magnitude, the radial RNS can then be used to estimate the maximum TSS by multiplying the calculated RNS values by 0.5. 31 This correlation factor between TSS and RNS was confirmed by correlating 25 maximum TSS and RNS values measured at different locations downstream of four different mechanical aortic valve prostheses from the work of Woo. 32
PATIENTS AND METHODS
Patients.
The study comprised 20 patients (9 women and 11 men) aged 62 ± 10 years (mean plus or minus the standard deviation; range 41 to 75 years) referred to us with the primary diagnosis of aortic valve stenosis. They all underwent heart operation with aortic valve replacement. Intraoperative measurements were made in nine patients after insertion of a St. Jude Medical aortic valve, in five patients after insertion of a Starr-Edwards silicone rubber aortic ball valve, and in six patients after insertion of a CarboMedics aortic valve.
Blood pressures were measured with liquid-filled catheters in all patients and cardiac output was measured in 10 of the patients by the thermodilution method with Swan-Ganz catheters (Baxter) at the beginning and at the end of each measurement. After a midline sternotomy and pericardiotomy, stay sutures in the pericardium were used to lift the heart anteriorly in an anatomically more correct position. The St. Jude Medical and CarboMedics valves were implanted with the two major orifices oriented anteriorly and posteriorly.
Turbulence measurements.
Blood velocity measurements were taken in the ascending aorta 5 to 6 cm downstream of the aortic valve (approximately 1.5 vessel diameters) with a PDU technique. The perivascular ultrasound transducer consisted of a half circular polyvinylchloride shell with two holes to fit two 10 MHz ultrasonic probes, which were directed at measuring points along a diameter perpendicular to the vessel axis to obtain the radial velocity component (Fig. 1). The shells were designed for vessel diameters from 20 to 40 mm in 2 mm increments. The ultrasonic probes were self-designed, equipped with 3 mm unfocused piezoelectric crystals, and could be sterilized in a formalin oven at 85° C. The mean sample volume was approximately 2 mm in diameter and 1 mm in axial length. The two probes were alternately operated by a PDU velocimeter (Alfred, Vingmed, Norway), which was modified to measure turbulent velocity fluctuations with frequencies up to at least 200 Hz. By changing the range gate delay (depth), switching between the two probes, and rotating the shell from 0 to 45 degrees, the radial blood velocity component was measured in five points along four diameters, covering most of the cross-sectional area in 17 measuring points (Fig. 2, upper left). The electrocardiogram and the arterial blood pressure signals were recorded on an instrumentation recorder (TEAC XR 510) together with the velocity signal from the PDU velocimeter. At least 20 heart cycles were recorded from each of the 17 measuring points. The blood velocity measurements were conducted during stable hemodynamic conditions after cardiopulmonary bypass and lasted from 6 to 10 minutes. The study was approved by the Danish Local Ethical Committee, and the patients were informed and gave written and oral consent according to the "Helsinki Declaration II."
|
|
33
Compensation for Doppler ambiguity was done in accordance with previous findings 26 as follows:
RNS = 0.77RNSPDU - 2.5
where RNSPDU is the RNS measured by PDU.
To characterize the flow through the ascending aorta the mean systolic Reynolds number (Remean) was estimated as
Remean = (
Usd) / µ
where p is blood density, µ is dynamic blood viscosity, d is the inner diameter of the ascending aorta calculated as 85% of measured outer diameter, and Us is the mean systolic cross-sectional average velocity calculated from cardiac output, heart frequency, and the systolic duration.
RESULTS
The key results from the study are listed in Table I. RNSmax is the maximum local RNS measured in the cross-sectional area downstream of the valves in top systole and RNSmean is the mean systolic RNS averaged over the 17 measuring points.
|
Fig. 3. shows typical examples of the RNS distribution during top systole for the three valve types. They are presented as two-dimensional color-coded maps with indications of the anterior, posterior, left, and right vessel wall of the ascending aorta.
|
CarboMedics valve.
The RNS distributions for the CarboMedics valves were quite similar to those in the St. Jude Medical valves. Like the St. Jude Medical valves the CarboMedics valves were implanted with the two major orifices oriented anteriorly and posteriorly with the central slit from left to right. The highest RNS values were seen along the central slit and the lowest RNS values occurred over the two major orifices. A typical example of the RNS distribution during top systole is seen in Fig. 3. The maximum local RNS values were from 19 to 72 N/m2 and the systolic spatial mean RNS values were from 4 to18 N/m2.
Starr-Edwards silicone rubber ball valve.
The RNS distributions for the Starr-Edwards silicone rubber ball valves were asymmetric, revealing the highest RNS values between 11 and 56 N/m2 at various locations in the annular region between the ball and the vessel wall. A typical example is shown in Fig. 3. The maximum local RNSs were maintained during most of the systole from top systole to the last part of the deceleration phase (about 200 msec). The lowest RNS values occurred in the opposite half plane of the highest RNSs. The spatial mean RNS values were from 1 to 8 N/m2.
DISCUSSION
Detailed comparative studies of velocity fields and turbulent stresses downstream of different designs of artificial heart valves have been done almost exclusively in vitro. 1-7 The only comparative in vivo study reported was done by Hasenkam and associates, 30 who used a hot-film anemometry technique to evaluate different types of biologic and mechanical aortic valve prostheses implanted in large pigs. For the implanted St. Jude Medical valves the maximum RNS was 24 N/m2 (8 to 59) averaged over the systolic ejection phase. Quantitative measurements of turbulent stresses downstream of normal, diseased, and artificial aortic valves in human patients have been done by Stein, Walburn, and Sabbah 34 with a catheter-mounted hot-film anemometry probe to register point velocities in the ascending aorta. In a patient with a Björk-Shiley valve (Shiley, Inc., Irvine, Calif.) the mean systolic RNS was 19 N/m2. Nygaard and associates 35 used a hot-film anemometry needle probe to register blood velocities at 41 evenly distributed measuring points in the cross-sectional area of the human ascending aorta downstream of normal, stenotic, and artificial valves. Maximum and mean values of RNS for St. Jude Medical valves were 38 to 113 N/m2 and 10 to 37 N/m2, respectively, whereas the corresponding values for the Starr-Edwards silicone rubber ball valves were 37 to 120 N/m2 and 11 to 38 N/m2.
The main results of the aforementioned in vivo studies are summarized in Table II and reveal maximum turbulent stresses downstream of the investigated mechanical aortic valve prostheses on the same order of magnitude, not exceeding 100 N/m2.
|
On the basis of the results of the present study the two investigated bileaflet aortic valve prostheses seem to perform equally well regarding turbulent stress generation. The small differences in the measured mean and maximum RNS could equally well be a result of different flow and pressure conditions and various sizes of valve and vessel diameters. The magnitudes, distribution, and temporal development of RNSs downstream of the two valves were, not surprisingly, very similar and for the St. Jude Medical valve almost the same as those obtained in pigs. The hinge area in the middle of the flow field creates high-velocity gradients and is the obvious reason for the high levels of RNS at locations along the central slit near the vessel wall.
The RNS distribution for the Starr-Edwards silicone rubber ball valve was different from that of the bileaflet valves because of a totally different design. According to in vitro studies the maximum RNS values should be found in the annular region between the ball and the vessel wall. 3, 4, 13 However, probably because of tilting of the valve ring in the aortic anulus, the curvature of the ascending aorta, and poststenotic dilation, the maximum RNSs were located only in one half plane in the annular region between the ball and the vessel wall. In the opposite half plane the RNSs were very low, indicating that the main flow was concentrated on one side of the ball. The elevated levels of RNS were present for a larger part of systole than those observed for the bileaflet valves, which is in agreement with results from in vitro studies. 3, 4, 7, 13 This could be because of the greater inertia of the ball slowing the closure of the valve in contrast to the lighter bileaflet valves that close more rapidly.
It is difficult to compare the results from in vitro studies with the results of the present study, because of different test conditions such as various flow and pressure conditions, different valve and vessel diameters, compliant vessels in contrast to rigid tubes, and blood in contrast to the glycerol-water mixture typically used during in vitro tests. These differences, in addition to the different measuring techniques used, may account for some of the discrepancies observed between in vitro and in vivo studies of artificial heart valves. However, three main causes for the diverging results are (1) the turbulence intensity decreases with the distance from the valve and in vitro measurements are typically taken closer to the valve than is possible in vivo; (2) skewed mounting of the prostheses in vivo combined with poststenotic dilations; (3) it is not possible to measure turbulent shear stresses (TSS) directly in vivo, rather, they have to be estimated from turbulent (Reynolds) normal (RNS) stress measurements. For these reasons RNSs measured in vitro are typically higher than RNSs measured in vivo.
Maximum RNSs of 450 N/m2 measured 0.5 vessel diameter downstream of St. Jude Medical valves and 1060 N/m2 downstream of Starr-Edwards silicone rubber ball valves have been found in vitro. 3, 4 Several in vitro studies have shown decreasing levels of TSS with axial distance downstream of mechanical heart valve prostheses. 2-4, 7, 13, 36 According to these studies the turbulence intensities measured 1.5 to 2 vessel diameters downstream of the valves in the present study could have been at least twice as high if they were measured closer to the valve, which means that the calculated RNS would have been more than four times higher. Consequently, the TSS closer to the investigated valves could have been higher than the critical shear stress level for lethal damage to erythrocytes and thrombocytes. Horstkotte and associates 8, 37 found hemolysis somewhat higher after implantation of Starr-Edwards silicone rubber ball valves than with St. Jude Medical valves and suggested that intravascular hemolysis is only of minor clinical importance if the function of the valve is not impaired.
Lund 38 found no significant difference in hemolysis between the Starr-Edwards silicone rubber ball and the St. Jude Medical valve and suggested that innate hemodynamic characteristics of different prostheses are not a major hemolytic factor and that a hypertrophic, malfunctioning left ventricle might be responsible for higher degrees of turbulence in the vicinity of prosthetic valves in the aortic position and could explain the increased tendency to hemolysis in these patients. Lund and associates 39 also reported comparable and acceptable long-term performance for the Starr-Edwards silicone rubber ball and St. Jude Medical valves used in our institution from 1980 to 1986 and found the incidence rate of thromboembolism to be 1.4% per patient-year for both valves types.
The present study is the first detailed human comparative study in which the clinically experimental data seem to support the findings of clinical follow-up studies. The measuring and analysis procedures are now so well established that the investigation has been extended to a Danish multicenter project involving several other mechanical and biologic aortic valve prostheses.
CONCLUSION
The spatial distribution and temporal development of turbulent stresses downstream of the three mechanical aortic valve prostheses correlated well with the superstructure of the valves, when the mounting of the valves is taken into consideration. The two bileaflet valves (St. Jude Medical and CarboMedics) revealed the highest RNSs along the central slit near the right or left vessel wall, with elevated levels of RNS maintained from top systole and halfway through the deceleration phase. The RNS distribution for the Starr-Edwards silicone rubber ball valve was asymmetric, revealing the highest RNSs at various locations in the annular region between the ball and the vessel wall. The maximum local RNSs were maintained during almost the entire systole. The magnitudes of RNS downstream of the two bileaflet valves were on the same order of magnitude and lowest for the Starr-Edwards silicone rubber ball valve. The estimated TSS values revealed magnitudes and corresponding exposure times sufficient to cause sublethal and maybe lethal damage to red blood cells and platelets. The surprisingly low RNS values downstream of the Starr-Edwards silicone rubber ball valve might be a result of the skewed mounting of the valve ring in the aortic anulus combined with a poststenotic dilation. The spatial distribution of the highest turbulent stresses disclose the drawback in the design of mechanical heart valves involving a central occluder mechanism.
Footnotes
From the Department of Thoracic and Cardiovascular Surgery, Skejby Sygehus, Aarhus University Hospital;a Engineering College, Aarhus;b and the Institute of Experimental Clinical Research, Aarhus University,c Aarhus, Denmark. All authorsare affiliated with the Cardiovascular Research Center, Aarhus University, Aarhus, Denmark. 
References
This article has been cited by other articles:
|
|
 |
|
  C. Nyboe, J. A. Funder, M. H. Smerup, H. Nygaard, and J. M. Hasenkam Turbulent stress measurements downstream of three bileaflet heart valve designs in pigs. Eur. J. Cardiothorac. Surg., June 1, 2006; 29(6): 1008 - 1013. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. N Andersen, S. Ringgaard, J M. Hasenkam, and H. Nygaard Quantitative haemodynamic evaluation of aortic cannulas Perfusion, September 1, 2004; 19(5): 323 - 330. [Abstract] [PDF]
|
|
|
|
 |
|
 Y. Naito, M. Nakajima, H. Inoue, N. Hibino, E. Mizutani, and K. Tsuchiya Unexpected durability of Smeloff-Cutter aortic ball valve prosthesis Ann. Thorac. Surg., May 1, 2003; 75(5): 1633 - 1635. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Kozerke, J. M. Hasenkam, H. Nygaard, P. K. Paulsen, E. M. Pedersen, and P. Boesiger Heart Motion-adapted MR Velocity Mapping of Blood Velocity Distribution Downstream of Aortic Valve Prostheses: Initial Experience Radiology, February 1, 2001; 218(2): 548 - 555. [Abstract] [Full Text]
|
|
|
|
 |
|
 F. I. Pareti, A. Lattuada, C. Bressi, M. Zanobini, A. Sala, A. Steffan, and Z. M. Ruggeri Proteolysis of von Willebrand Factor and Shear Stress-Induced Platelet Aggregation in Patients With Aortic Valve Stenosis Circulation, September 12, 2000; 102(11): 1290 - 1295. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. M. Bernal, J. M. Rabasa, F. Gutierrez-Garcia, C. Morales, J. F. Nistal, and J. M. Revuelta The CarboMedics Valve: Experience With 1,049 Implants Ann. Thorac. Surg., January 1, 1998; 65(1): 137 - 143. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. Lund, K. Emmertsen, T. T. Nielsen, F. T. Jensen, C. Flo, H. K. Pilegaard, B. S. Rasmussen, O. K. Hansen, and L. H. Kristensen Impact of Size Mismatch and Left Ventricular Function on Performance of the St. Jude Disc Valve After Aortic Valve Replacement Ann. Thorac. Surg., May 1, 1997; 63(5): 1227 - 1234. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
