Design Considerations for the In Vitro Testing of Cardiovascular Prosthesis
Elaine R. Strope, Ph.D.
Dynatek Laboratories, Inc.
Prosthetic cardiovascular devices are unique in that, in addition to the requirement of functioning for nearly a billion cycles or more, failure can result in death of the user. As a result, in vitro testing becomes critically important. The special considerations in design of instrumentation to test these devices will be reviewed and examples of data generated will be given.
For nearly a decade, personnel now associated with Dynatek Laboratories, Inc., have been involved in the multifaceted areas of cardiovascular flow dynamics and cardiovascular prosthetic device design. During this time, a certain logic to sequential in vitro testing has been developed. This logic results in generation of data which reliably reflects information that can be utilized for the appropriate design or modification of prosthetic cardiovascular devices. Also during these years, it has become apparent that the textbooks and reference sources that one might access in order to give guidance to the investigator with an interest in designing a prosthetic device are usually too comprehensive. A clear understanding of appropriate testing becomes lost in the mire of dozens of recommendations. We will attempt in this publication to set up step-by-step recommendations to potential investigators on the types of in vitro testing that will guide them in the design of their devices. Included in these recommendations are considerations of the publications circulated by both the National Heart, Lung, and Blood Institute (NHLBI) and, in the case of heart valve testing, the FDA guidelines on data to be submitted to the agency in support of application of a pre-market approval for a new heart valve.
One of the greatest problems facing the biomaterial community is the establishment of appropriate testing protocols that can be utilized by all investigators such that meaningful data is generated in a reproducible manner from laboratory to laboratory. This inter-laboratory irreproducibility problem has been underscored by Reul (1983) when he reviewed the design criteria of several heart valve testers and noted that the greatest disadvantage of all of these simulators is the different designs and thereby the very limited comparability of measured results. The most recent report of the NHLBI working group which works through the administrative offices of the National Institutes of Health, the Public Health Service, and the United States Department of Health and Human Services is entitled “Guidelines for Blood Material Interactions” (1985). This handbook now serves as the primary reference document that all well-known academic and industrial biomaterial laboratories utilize when establishing various protocols necessary for the characterization of the biomaterials. Included in this handbook is a description of a variety of tests that are recommended on the bulk materials before they are fabricated into the final product such as a vascular graft, heart valve, or artificial heart. This characterization of bulk materials recommended on films or strips and includes a variety of the standard tests that materials scientists or engineers would probably use for this characterization.
Dynamic mechanical testing consists of both static transient tests (such as creep) and periodic oscillatory tests. Ferry’s text (1971) is considered the best single reference for information on this type of testing.
Also considered to be a standard in the testing of soft materials for creep are the procedures developed by Placek (1968). The working group also felt that the experimental methods and procedures for stress relaxation reviewed in detail by Bergen (1985) are appropriate for most biomaterial applications. The aforementioned tests, however, do not address the problems that many polymers as biomaterials will confront with respect to large cyclic deformations for long lifetimes. This application is particularly true in those cardiovascular areas which include the heart valve, artificial heart, left ventricular assist device, and vascular graft products. Before a material can be qualified for clinical usage where a cyclic deformation will occur, two levels of fatigue testing are recommended: a uniaxial screening test to rank candidate materials and provide preliminary information for design, and a performance test which mimics the actual use conditions as closely as possible. Regardless of the type of test used, the statistical spread of the cycles to failure data at any one stress condition and frequency must be taken into account to provide a measure of the reliability of the material.
The ASTM has three (1979, 1980, 1983) publications dealing with the tests necessary for uniaxial testing of the biomaterials. In addition, the factors that affect polymer fatigue life which include molecular weight distribution, crystallinity, etc., are discussed in some detail in the publication by Hertzberg (1980).
Cyclic biaxial testing is an area in which little has been published even though it is, in general, more closely matched to performance conditions. Descriptions of bubble-type experiments by Green (1970) allows one to measure anisoptropic stiffness and biaxial stress-strain curves. There are several ways to set up a biaxial test and an excellent review of this type of testing has been given by Bert (1975).
Prosthetic Vascular Grafts
Bulk Material Screening:
When a prosthetic vascular device or synthetic artery is implanted into a patient, it experiences a very unique biaxial stress situation, cycling approximately 40 million times per year. Under normal circumstances, the pressure differential during each cycle will be approximately 120/70mm Hg. As a result of this type of motion, it is most appropriate that candidate materials for the device be tested utilizing what is referred to as a “blister” test. Basically, this involves a film of the test material in a diaphragm morphology. Cyclic stress loads are applied from each side, forming a blister during the maximum pressure differential across the membrane (thereby the reference to blister testing). Although in the actual application, the material will be stressed primarily from one side only, symmetrical testing can still generate appropriate information to eliminate materials that are completely unsatisfactory.
The speed of testing will be not only a function of the instrument that has been built to do the testing, but also the response time of the material that will be tested. In any system that contains little or no compliance except that of the materials to be evaluated, an accurate pressure wave vs. cycle rate plot can usually detect any response time problems associated with testing at higher speeds. In general, as the speed of the testing goes beyond the ability of the material to respond, the pressure trace reflects what appears to be a more and more rigid material being tested. It is a simple matter, therefore, to compare pressure traces vs. speed and choose as the highest speed that point at which the pressure traces just begin to change qualitatively as well as quantitatively.
The nature of the fluid to be used during testing can be quite important, particularly when evaluating materials such as polyester polyurethanes or polyether polyurethanes whose viscoelastic properties are affected by the presence and uptake of materials such as lipids. Since the final product will be utilized in an environment with blood on one side, it seems most appropriate to test the materials with a plasma or plasma substitute. This, of course, involves the utilization of a device that can be sterilized.
Although the area of accelerated testing is always fraught with controversy and disagreement, it seems clear that a comparison can readily be made between any two materials regardless of the effect of increased testing rate. This assumes, of course, that the testing rate is within the response time of the materials to be tested. This simply means that, although the actual cycles to failure might be different in the in vitro vs. in vivo situation, two materials tested under the same circumstances at an accelerated rate will usually fail in the same order that they will fail in the in vivo situation. In this way, considerations such as molecular weight distribution and general type of polymer to be chosen can be decided upon fairly quickly. Rates of 1200 per minute are easily attained with most materials. This gives a rate multiplication factor of about 20 to the in vivo rate.
In order to improve the reliability of the data and to also reduce the amount of time required for all bulk material screening, it would be advantageous to have devices that can test more than one sample at a time. In addition, a controlled temperature environment would be critical.
Although bulk material testing will give a user an indication of the relative ranking of materials with respect to durability, fatigue, creep, etc., a synthetic artery must be subjected to fatigue/durability testing after it has been formed into its final shape. At this point, considerations such as post-fabrication processing, sterilization, and packaging effects will come into play.
Many of the general considerations mentioned in the bulk material section still come into play in this particular area of testing. These include temperature control, nature of liquid environment, sterilizability, and maximum speed as determined by response time of the device. It has now reached the point in the product development process where tubes have been fabricated in a variety of diameters and lengths and are under consideration for certain preclinical trials. It needs to be determined whether or not certain steps in the post-fabrication processing have had any impact on the overall fatigue/ durability of the prostheses as well as the general understanding of the long-term durability of the graft.
Maximum speed should again be determined for each particular device and in this particular situation, an additional piece of data can be generated on the material itself. This measurement is referred to as compliance and is a measure of the change in length or diameter of a product vs. the pressure differential experienced by that device. It is recommended that an internal compliance measurement be made on each device before they are placed on the accelerated tester. In this way, changes in dynamic mechanical properties of the device can be measured after the durability testing has been completed.
Testing to failure is usually the most conservative approach to generating durability data on this kind of an implant. In some instances, however, properly fabricated devices may never fail and will simply creep and general changes in the mechanical properties will occur. As a result, the number of cycles that the products will be exposed to has to be decided. At 1200 cycles per minute, ten year data can be generated in 6 months. This probably represents the shortest test period that should be considered if a one-to-one relationship between in vitro and in vivo failure can be established. Of course, if a comparison between two devices is being made, then one simply needs to carry the test out until one of the devices fails.
Again, multiple device testing apparatuses are very helpful in this case. Not only can the statistical significance of the data be increased by increasing then, but also direct comparisons can be made between various prototypes tested at the exact same time.
As a quick summary, the vascular grafts will be fabricated and processed, including sterilization if necessary, and then mounted on an appropriate device that can deliver an accurate pressure curve to the device at speeds up to which the device can no longer respond quickly enough. Before being put on this testing device, a variety of mechanical properties of the polymers should be determined, including the internal compliance in a radial and longitudinal manner. The grafts should then be tested for a period which will include that time at which one of the devices fails or, as suggested, a period that represents 10 years implantation period. The devices should then be removed from the accelerated tester and again all of the mechanical properties redetermined.
In this way a reliable reflection of the effect of long-term cyclic stress can be determined before an implantation is required.
Prosthetic Heart Valves
Bulk Material Screening:
Prosthetic heart valves experience a very much different trajectory in the motion during each cycle than do vascular grafts. As a result, general screening of the bulk material must include a durability test that exposes the materials to a different kind of motion than the aforementioned blister test. In this particular case, synthetic leaflets will usually go through a 90 degree bend per cycle. As a result, a strip of material can be mounted across a port whose diameter represents the flow diameter of a particular heart valve. The strips should be long enough such that, during deflection in either direction, a 90 degree bend at the connecting point will be attained. Devices can be fabricated that can direct flows at these strips such that, during each half cycle, the strips bend the appropriate number of angles. If a 90 degree angle is not wanted, then the strip holders can be modified to restrict bending to any angle desired. As before, the testing solution should contain materials that are known to affect mechanical properties of the bulk materials. Temperature control and multi-position testing is important in this area also. Mechanical testing before and after durability exposure as well as microscopic examination of the flex area will allow one to determine the effect of this kind of motion on the integrity of the tested materials.
This is somewhat of a diversion from the kinds of testing that has been described for vascular grafts. In the design of prosthetic heart valves, another very important consideration is the efficiency of the heart valve from the perspective of cardiovascular flow dynamics. One needs to measure various parameters such as steady flow pressure drop across the valve, mean pressure drop during systole vs. mean flow rate, and the root mean square of the flow rate during systole for the aortic valves. For mitral valves, the mean pressure drop during diastole vs. the mean flow rate and the root mean square of the flow rate during diastole should be measured. These tests should be run at a normal pulse rate (70 beats per minute) with systole occupying about 35% of the cycle time. In addition, dynamic regurgitation at various beat rates and various flow rates for experimental valves should be generated. These tests are those that are recommended in the guidelines of data to be submitted to the Food and Drug Administration in support of applications for pre-market approval. This testing must be accomplished on a cardiovascular system simulator that provides accurate and sensitive information to allow the assessment of the above parameters.
Of course the purpose of the above tests is simply to determine the amount of stress that the components of the blood will feel going through the valve, as well as the additional amount of work that the heart must do during each beat to overcome the presence of this valve in the human flow loop.
Once the bulk material has been chosen and the design of the valve has been proven to be efficacious, then the last phase of testing includes the fatigue/durability evaluation of the valve design.
In this particular area of testing, the FDA as well as investigators in the area agree that a cycle-for-cycle comparison between in vitro and in vivo testing of mechanical valves can be made. However, it is felt that accelerated testing causes tissue valves or composite synthetic tissue valves to fail approximately twice as fast as in the in vivo situation. It has been found by several investigators, however, that the mechanism of failure of the leaflet valves is the same even though they fail at an earlier time period when tested in an accelerated manner.
Considerations that one must make when designing a tester that will cycle heart valves in an accelerated manner include the temperature and fluid composition considerations. In addition, when testing an aortic valve, the geometry of the exit side of the valve is critical since the bulbous shape of the aortic root very much contributes to the backflow characteristics of the liquid which assist the closing motion of the leaflets.
Probably the most important consideration when testing valves is the pressure trace experienced by the valve throughout the cycle. Included in this consideration is not only the magnitude of the pressure across the valve that is felt, but also the change in pressure with time (dP/dt). Maximum cycle rate must also be determined for each individual valve design and can only be accurately assessed if the valves are stroboscopically analyzed at the various potential cycle rates.
When testing tissue valves, full opening and full closing must be reached during each cycle. At maximum flow through the valve, it is very important that the leaflets experience no flutter. In general, one will find that careful tuning of an appropriately designed machine will result in full opening and full closing without leaflet flutter. It must be stressed that leaflet flutter will lead to early failure of the leaflet materials and must be avoided at all costs.
While flutter is not a problem when testing mechanical valves, cavitation can be. Cavitation along various portions of the valve can lead to very quick erosion of materials regardless of the hardness of those experimental substances. As a result, the valves again must fully open and fully close during each cycle and the valve must experience no cavitation over its surfaces during maximum flow.
In summary, valves should be mounted in chambers of appropriate geometry and should be tested as quickly as possible while still experiencing full opening and full closing without flutter or cavitation. In addition, the maximum pressure across the valve at closing and the dP/dt must be carefully adjusted.
We have not mentioned any clinical testing nor the generation of safety, biocompatability, or hemocompatability information for any of these materials. The purpose of this paper was to address specifically the step-by-step evaluation of materials from a mechanical/durability perspective. We have shown that the testing of synthetic vascular grafts should include preliminary bulk materials screening with respect to the change in various mechanical properties of the materials that occur after cycling through biologically relevant parameters for hundreds of millions of cycles. This testing is followed by the testing of the actual device to determine how cyclic stresses affect its mechanical properties. Our description switched to prosthetic heart valves which, in addition to bulk materials screening and long-term durability, included the very important determination of the performance of the valve from a hydrodynamic perspective.
We have generated this stepwise process after years of experience with prosthetic device design and FDA submissions. We hope that this presentation will clarify the steps necessary to quickly generate appropriate data in evaluating or designing new cardiovascular products.
American Society for Testing and Materials (1979): ASTM D3479, Annual Book of ASTM Standards, Part 361–Plastics- Materials, Film, Reinforced and Cellular Plastics; High Modulus Fibers and Composites, Philadelphia, American Society for Testing and Materials.
American Society for Testing and Materials (1980): ASTM D671, Annual Book of ASTM Standards, Part 35–Plastics – General Test Methods, Nomenclature, Philadelphia, American Society for Testing and Materials.
American Society for Testing and Materials (1983): ASTM D430 Annual Book of ASTM Standards, Section 9, Volume 09 01–Rubber, Natural and Synthetic – General Test Methods; Carbon Black, Philadelphia, American Society for Testing and Materials.
Bergen, R.L. In: Testing of Polymers, Vol 2, Schmitz, J.V., ed. New York, Wiley, p1.
Bert, C.W. (1975): Composite Materials, vol. 8. Chamis, C.C., Broutman, L.J., and Krock, eds. p.73.
Ferry, J.D. (1971): Viscoelastic Properties of Polymers, New York, Wiley.
Green, A.E., and Adkins, J.E. (1970): Large Elastic Deformations, 2nd ed., Oxford, Clarendon Press, p. 161.
Hertzberg, R.W., and Manson, J.A. (1980): Fatigue of Engineering Plastics, New York, Academic Press.
Placek, D.J. (1968): Magnetic Bearing Torsional Creep Apparatus. J. Polymer Sce., 6:621.
Reul, H. (1983): In-Vitro Evaluation of Artificial Heart Valves. Advances in Cardiovascular Physics, Cardiovascular Engineering, Part IV: Prostheses, Assist and Artificial Organs, D.N. Chista, ed. S. Karger, New York , p.16.