Bioengineering December 1, 1997 Volume 27 Issue 4 Fall Issue of The Bridge on Bioengineering. The LVAD: A Case Study Monday, December 1, 1997 Author: Victor L. Poirier By solving a host of design challenges, engineers have brought the heart-assist device to successful application in humans. From an engineering point of view, one would consider the development of a simple fluid pump to propel blood through the body to be a minor task. Our experience tells us that this minor task has taken close to 100 years to accomplish. Perhaps this is a short time relative to evolution, but it is an extraordinary length of time relative to engineering practices. Why has it taken so long? To answer that question, we need to look at what the artificial cardiac system is required to do. The design goals for a left ventricular assist device (LVAD) are a flow rate of 10 liters per minute (lpm) at a mean arterial pressure of 120 mmHg at a beat rate of no more than 120 per minute with a filling pressure of 20 mmHg. The initial design goals were set for 2 years of operation and are now being extended to 5 years. The system must be nontoxic to the human body. It must operate continuously without stopping for preferably 10 years at the minimum, cycling at a rate of 40 million times a year. It must accelerate and decelerate blood on every beat from velocities of 0 to 25 lpm, and it must provide sufficient energy on a beat-by-beat basis to increase the pressure from 0 to 150 mmHg. This must be accomplished without damaging the delicate cell membranes of red blood cells or altering their biologic function. It must also be done in a way that prevents damage to other cellular components or blood elements such as platelets, and it must not produce anomalies in cellular function. The system is to pump, on an average, 6 lpm with excursions of up to 10 to 11 lpm; it should be responsive to biologic demand and automatically alter its output consistent with the patient's needs. The system must not endanger the patient in any way and should not produce blood clots (thrombi) that can lead to stroke. Structural Compatibility, Stability The structure of the system must be compatible with the varying anatomy of the human race. It must operate in small children as well as large adults. It has to be structurally stable operating in a warm saline environment without corrosion, and its rigid structures must be capable of being joined to the flexible anatomy, which also supports this large and heavy mass. The system must operate maintenance free for the duration of use. Mechanical components need to be small and carry very high loads without failing. Due to lack of space, redundancy is impractical. Systems must be efficient to reduce the volume of batteries that are needed as well as to minimize the amount of waste heat that is transferred to the delicate surrounding tissue, which cannot maintain viability above 41 degrees Centigrade. In addition, the system must not degrade with time as a result of enzymatic attack. Investigation into the development of blood pumps was under way by 1900. The late 1940s and early 1950s were a particularly important period in the development of circulatory support systems. A significant event was the establishment, in 1948, of the National Heart Institute (Poirier, 1993). Progress was very sporadic in the early days, until the advent of cardiopulmonary bypass in 1953. But it was not until 1963 that Congress approved money specifically for research on an artificial heart, to be funded through the National Heart Institute (Poirier, 1993). Six contracts totaling $581,000 were awarded to support comprehensive analyses of technical development issues and needs for the artificial heart. Thermo Electron, the predecessor of Thermo Cardiosystems, received one of these six contracts, which was my first exposure to this technology. Researchers were actively looking at the possibility of replacing the entire heart with a mechanical device as well as supporting only the left ventricle, thereby assisting the natural heart. We investigated both concepts and early on decided it would be most prudent to take the simplest approach and assist the natural heart. The concept of the LVAD is valid from an engineering point of view, for the device simply assists the natural heart by taking over the pumping function of the left ventricle. The left ventricle must produce considerably more pressure for the same flow as the right ventricle and therefore does the majority of the cardiac work, approximately 80 percent. By assuming the work of the left ventricle, the LVAD improves cardiac output, increases coronary perfusion, and reverses the complex hemodynamic and metabolic disturbances responsible for heart failure (Frazier, 1994). Initial work with LVADs was discouraging. Dennis and colleagues were among the earliest investigators to undertake direct unloading of the left ventricle by cannulation. By 1964, their series included 12 patients, most of whom died within a few hours after undergoing the procedure (Dennis et al., 1962). Spencer also reported his experience using cannulation techniques for left-heart bypass in 4 patients, but only 1 patient survived (Spencer and Eiseman, 1964). One of the most successful short-term devices developed by several teams was the intra-aortic balloon pump. Research conducted by Moulopoulos and associates, Kantrowitz and associates, as well as Clauss and coworkers led to an inflatable balloon pump that is positioned in the lumen of the thoracic aorta. This balloon pump could augment the cardiac output by as much as 25 percent. These pioneering researchers laid the groundwork for us to build upon. As our work progressed, it became clear that we were facing major barriers. These barriers were overcome one by one. As we advanced on a technical front, the economic concerns came forth. Saddled with these two barriers in the late 1970s, another obstacle was put before us: regulatory requirements by the Food and Drug Administration (FDA). With the passage of the 1976 Medical Device Amendments, we had to face and satisfy new requirements that we never before considered. This was another hurdle to contend with, but we continued and pushed forward. Now, after more than 30 years, we approach what I hope is the final barrier: acceptance by the medical community, patients, and society in general. The first barriers we had to overcome were technological in nature. These entailed the development of special materials, blood-contacting surfaces, blood-pump designs, power sources, and methods to store energy. The first problem that confronted us was to find materials that could be safely used in the body and that possessed the strength and flexion capabilities required for our application. Bladders and diaphragms had to be able to flex at least 40 million times per year, had to withstand the hostile environment of the body, and had to be biologically acceptable and nontoxic. To achieve these goals, we evaluated several materials and coating processes. Initially, the only materials available were industrial grade. The first material compounded specifically for medical purposes was silicone rubber, which was introduced in 1953. Gradually, however, better materials with a variety of different properties became available. We initiated our research with two types of silicone polymers, both of which were predominantly polydimethyl-siloxane. This material, which seemed ideal, proved to be inadequate: In tests, bladders made with the material failed within 25 million cycles, far short of the 80?100 million cycle goal. New formulations using an improved platinum curative yielded improved properties; however, these materials were not as promising as the family of polyurethane materials that were under development. We evaluated several polyurethane formulations (Poirier, 1980). Initially, industrial grade materials were evaluated because medical grade urethane was not available. These materials proved to be toxic and exhibited hydrolytic instability. We undertook a study to evaluate aliphatic and aromatic urethane as well as linear segmented formulations, cross-linked versions, and polyether-urea types. Extensive testing of this class of materials demonstrated that the polyether-urea types were most promising. Testing Pump Configurations We also undertook an extensive evaluation of several configurations. We examined eight using five different urethanes, one silicone, and one polyolephine rubber. The bladders and diaphragms were mounted in pumps and tested on mock circulatory loops. The results of this long-term testing revealed no significant differences between a geometric configuration for bladders and diaphragms made from the same material; however, there were differences between materials. Hexsyn (Goodyear Rubber), Pellethane (Upjohn Chemical), and Biomer (Ethicon, Dupont) proved to have superior flex life to Tecothane (Thermo Electron), Tecoflex HR (Thermo Electron), Silastic (Dow Corning), and SRI (Stanford Research Institute) materials. This evaluation of 214 bladders and diaphragms continuously tested over an 8-year period led us to choose Biomer as our material of choice for diaphragm fabrication (Poirier, 1979). The solution was short-lived, however. After approximately 15 years of testing, evaluation, and clinical trials, Biomer was removed from the market by Dupont, the manufacturer, due to fears of litigation. It took us 3 more years to develop a replacement, Cardioflex, which is now used exclusively in all of our implantable devices. We also needed to find suitable material for the device's rigid and structural components. We looked at both polymeric materials, such as polysulfone, and metallic materials. Because of our choice of blood-contacting surfaces, as will be described later, we focused our selection on metallic components. We evaluated nickel, stainless steel, aluminum, and titanium. We quickly learned that stainless steel was more appropriate than nickel and aluminum; however, some time was spent until we discovered that high carbon content and improper heat treating led to intergranular corrosion when the materials were exposed to the warm saline solution in the body. To avoid these problems, we coated these materials with a protective polymeric coating. This procedure improved our long-term implant results; however, it did not eliminate corrosion problems. In 1973, we switched to a titanium alloy, an outstanding material that we continue to use today. Tackling the Thrombus Problem Perhaps the most significant technical problem in the development of the LVAD related to the blood-contacting surface of the pump. Initial studies undertaken in 1966 used an axisymmetric blood pump with smooth silastic bladders. These were soon terminated because of rapid accumulation of thrombus in the pump and massive thromboembolic damage to vital organs, despite anticoagulant therapy. Studies involving pumps and other devices also resulted in thrombus formation at joints or interfaces between the rigid and flexible members and between these materials and the vasculature. Attempts to make hydrodynamically smooth joint transitions, and the use of anticoagulants, did not eliminate the formation of tenuously held thrombi and the threat of thromboembolism. Defects or imperfections as small as 10 microns were sufficient to form platelet clumps leading to thromboembolic complications. A new approach was obviously necessary. Initial work by other researchers (DeBakey et al., 1966; Hall et al., 1967; Liotta et al., 1966) laid the groundwork for our next steps. This research, which encouraged a cellular layer to form on the interface of a vascular graft, was a possible solution to the thrombus problem. Thus, the joints would be entirely covered by this living, autologous lining. To use this concept, we had to develop a means of reliably anchoring the lining to all pump surfaces, and the lining had to be adequately nourished from the blood stream. Our initial effort involved the development of a new interface using short fibers imbedded in an adhesive. The resulting tangled network of fibers was bonded to a substrate. This interface was applied uniformly, by a process called flocking, to all surfaces of the pump, including such complex surface shapes as valve guide pins. Initial in vivo results were encouraging; however, we experienced great difficulty providing reliable adhesion of the flocked Dacron fibrils to the silastic bladders. To overcome this problem, polyurethane was evaluated for both the bladder and adhesive. Three problems persisted: flock release, excessively thick biologic linings, and adhesive failures. The problem of flock release was overcome by the development of a flock overcoating process that provided secondary bonding at all fibril contact points (Poirier and Keiser, 1977). The adhesive problem was solved by changing the adhesive from a polyester- to a polyether-based urethane. The concerns related to the thickness of the biologic linings proved more difficult to resolve. After extensive studies of all the parameters affecting lining thickness, it became clear that the technology as it stood could only allow implant durations of 2 to 3 months in animals, barring a few exceptions. During this time, evaluation of our textured biomaterials revealed spontaneously occurring islands in the pseudo-neotima. Subsequent microscopic examination indicated at these locations the presence of fibroblast cells imbedded in a loose matrix of collagen fibrils. These islands were thin, and firmly attached, and appeared to be an ideal lining for animal use. In an effort to accelerate collagen formation on larger surface areas, we tried seeding with cultured bovine, fetal fibroblasts derived from female Holstein donors (Bernhard, 1987). Immediately before we implanted a device, these fetal allogenic cells were distributed on all blood-contacting surfaces. These preseeded surfaces greatly improved the results of our long-term experiments with animals, extending useful pump life to 1 year. The mature linings were thin, continuous, and firmly attached. We tested a variety of textured surfaces to determine if we could improve on the Dacron fibril surface, which was adhesively bonded to the substrate. We evaluated plasma sprayed surfaces, textured surfaces created by ion-beam drilling, surfaces formed by molding techniques, as well as integrally textured surfaces. This latter configuration proved to be the most promising. Fibrils are extruded from the base membrane and subsequently over-coated to provide a secondary bond at fibril contact points. This structure eliminates the adhesive layer and the use of two different materials with different moduli of elasticity. The fibrils and base membrane are continuous and of the same material. To improve the textured surfaces on the rigid components, we followed the same philosophy: one material and no adhesives. We accomplished this by diffusion-bonding titanium spheres to the titanium substrate. The resultant structure was continuous and provided the necessary matrix for cellular and protein penetration and attachment. After more than 30 years of development, I am pleased to say our textured surfaces have proved quite satisfactory and require no preseeding. The surface geometry has overcome the major problem of thromboembolic complications that have plagued this technology for so many years. Implant patients have no need for prescription anticoagulants. Aspirin is the only drug recommended with the use of the device. With more than several hundred patient years of experience on 1,300 patients who have been supported for periods averaging 3 months and extending to more than 2 years, we can now say that we have solved the thromboembolic problem. It is important to point out that surfaces alone cannot eliminate thromboembolic complications. Careful pump design is necessary, and we certainly have evaluated many different configurations. The most challenging design considerations were associated with the inlet conduit assembly. The low-flow conditions coupled with the need for flexibility to accommodate anatomic variations resulted in designs that were susceptible to pannus formation and conduit occlusion. This problem, coupled with valve positioning and valve selection, retarded the development of these systems for many years. Mechanical valve fracture and failure were common. Another obstacle in our quest was the development of reliable power sources, as well as means to transfer this energy into the body. We evaluated every conceivable power source, including pneumatics, low- and high-pressure hydraulics, thermal conversion systems, including the Rankine cycle and electric systems. Every system was evaluated for size, efficiency, reliability, noise production, implantability. Evaluating Power Sources The most challenging power source we developed was an implantable system consisting of a miniature, closed-cycle steam engine powered by a Plutonium-238 fuel capsule. This power source produced a high-pressure hydraulic fluid that could be used to power a blood pump. The system was tested extensively both in vitro and in vivo until the concept was abandoned due to public fears over the use of radioactive fuel. After many years of testing and evaluations, it became quite clear that, for now at least, the most appropriate power sources are pneumatic and battery-operated electrics. LVAD technology has now been developed to the point of successful application in humans (Frazier et al., 1995). More than 1,300 implants have been done, with minimal (3 to 5 percent) thromboembolic complications. Patients using the device look good, feel good, and are able to go back into society. In my opinion, this is wonderful technology, but what did it cost and can we afford to use it? If we look back at history, it was indeed feasible at one point, just before the introduction of FDA regulatory requirements, to develop a system and to carry out clinical trials without spending excessive amounts of capital and time. Commercial companies could bring technologically advanced products to the marketplace. Today, it is another story. Companies now face huge barriers to entry, barriers that make it impractical to develop a system from concept to commercialization and expect a financial return for the investor. Development costs associated with time to market, safety concerns, and technological barriers are truly staggering. In the case of the HeartMate blood-pump system, we initiated the design in 1975. It took 10 years to complete the development and testing needed to obtain an investigational device exemption from the FDA. Our clinical trial took 9 years to complete, from 1985 to 1994. The cost of this trial was $41.6 million. Coupled with development costs, our total investment was $62 million to develop a simple fluid pump that can take over the pumping function of the left ventricle. Clinical Trial Costs Prohibitive The cost associated with running a clinical trial is beyond what industry can bear for these complex technical systems. To push forward, companies like ours have had to rely on their clinical colleagues to absorb the expense associated with data acquisition and hardware support. These are costs that should be borne by our government. To some extent, they are. In the early 1960s, within a few years of the establishment of the artificial heart program, annual federal expenditures in this area totaled approximately $10 million, about 5 percent of the budget of the National Heart, Lung and Blood Institute (NHLBI). This was a considerable amount compared with what we spend today. Over the past decade, expenditures on the targeted programs were stabilized in the range of $10 million to $12 million annually, or slightly more than 1 percent of the total NHLBI budget. The collective total of all expenditures on the artificial heart program is now approximately $250 million. Government-sponsored research is essential if we are to arrive at a practical artificial heart. We must continue this research. We should not shut it down to satisfy a few vocal individuals who question the cost effectiveness or lack of economic benefit. We should not assume that funds used to support artificial heart or ventricle assist device research would be more effectively spent on another technology or social need. Industry is also strapped with satisfying regulatory requirements. At Thermo Cardiosystems, data collection and analysis requires a team of 9 individuals, and meeting requirements related to good manufacturing practices requires the effort of another 10. This represents an annual cost to us of more than $1.5 million. Long development cycles produce another dilemma. With so much invested in testing of existing systems, industry is reluctant to make changes, even if they are for the better. Are we justified in changing some aspect of our FDA-approved system if by doing so we invalidate all the testing we have done? Can we justify the added cost of qualification and lost time, even though I know that I can make a better or safer system? I am afraid that we have created a system in this country that discourages advancement and improvements. Our technology is on the forefront and yet it is technology that in many ways is outdated. The system that we are now using clinically was first designed in 1975, 22 years ago. I, as well as others, have learned much over the years and need to express that knowledge in advanced concepts. The long regulatory path is not only an economic burden, it forces obsolescence. Components and materials that were available in 1975 are not necessarily available today. Given that advances in some fields, such as electronics, are occurring very rapidly, how can a design from 1975 continue to be useful in 1997? Change is necessary as components become obsolete or withdrawn from the marketplace. Yet, developers face a Catch-22: If we use new components and start testing all over again, we will reach a point where what has been tested is out of date. What is needed, which does not exist at this time, is a rapid approval process for devices that have the unique characteristic of making the difference between life and death. No one can argue that an effective AIDS drug should reach patients as quickly as possible. I see no good reason why the same standard should not apply for lifesaving technologies such as artificial hearts or heart assist devices. We now have systems that appear to perform well. They are not perfect, but they have been granted marketing approval. Do we now stand still and reap the benefits or our hard work, or do we continue? As engineers, we can do nothing else but continue to improve and expand the technology. References Bernhard, W. F. 1987. A fibrillar blood-prosthetic interface for both temporary and permanent ventricular assist devices: Experimental and clinical observations. The 11th Hasting Lecture. Devices and Technology Branch Contractors Meeting, Bethesda, Md. December 7. DeBakey, M. E., D. Liotta, and C. W. Hall. 1966. Mechanical Devices to Assist the Failing Heart. National Academy of Sciences, National Research Council. Washington, D.C.: National Academy Press. Dennis, C., D. P. Hall, and J. R. Moreno. 1962. Reduction of the oxygen utilization of the heart by left heart bypass. Circ. Res. 10:298-305. Frazier, O. H. 1994. Ventricular assistance: A perspective on the future. Heart Failure 6:259-264. Frazier, O. H., E. Rose, P. McCarthy, N. Burton, A. Tector, H. Levin, H. L. Kayne, V. L. Poirier, and K. A. Dasse. 1995. Improved mortality and rehabilitation of transplant candidates treated with a long-term implantable left ventricular assist system. Annals of Thoracic Surgery 222(3): 327-338. Hall, C. W., D. Liotta, J. Ghidoni, and M. E. DeBakey. 1967. Velour fabrics applied to medicine. J. Biomedical Material Research 1:179. Liotta, D., C. W. Hall, and M. E. DeBakey. 1966. A permanent autologous lining for implantable blood pumps. Cardiovascular Res. Bulletin 4:69-75. Poirier, V. L., and J. Keiser. 1977. Method of Flocking Blood-Contacting Artificial Implant Devices. U.S. Patent No. 4,016,033. Poirier, V. L. 1979. Comprehensive, real time, endurance testing of cardiac assist pump bladders. ASAIO 25:319-324. Poirier, V. L. 1980. Fabrication and testing of flocked blood pump bladders. Pp. 73-115 in Synthetic Biomedical Polymers: Concepts and Applications. M. Szycher and W. G. Robinson, eds. Westport, Conn.: Technomic Publishing. Poirier, V. L. 1993. The quest for a solution. We must continue. We must push forward. The 16th Hasting Lecture. Artificial Organs 4:856-863. Spencer, F. C., and B. Eisemen. 1964. Quantization of effectiveness of cardiac assistance. Pp. 27-41 in Mechanical Devices to Assist the Failing Heart. Publication No. 1283. National Research Council. Washington, D.C.: National Academy of Sciences. About the Author:Victor L. Poirier is president, director, and CEO of Thermo Cardiosystems, Inc. He presented this paper at the 1997 NAE Annual Meeting Technical Symposium, held 8 October.