ARTICLE IN PRESS Biomaterials 29 (2008) 385–406 www.elsevier.com/locate/biomaterials Leading Opinion Prosthetic heart valves: Catering for the few$ Peter ZillaÃ, Johan Brink, Paul Human, Deon Bezuidenhout Christian Barnard Department of Cardiothoracic Surgery, University of Cape Town Medical School and Groote Schuur Hospital, Anzio Road, 7925 Observatory, Cape Town, South Africa Received 2 August 2007; accepted 23 September 2007 Available online 24 October 2007Abstract Prosthetic heart valves epitomize both the triumphant advance of cardiac surgery in its early days and its stagnation into aretrospective, exclusive ﬁrst world discipline of late. Fifty-two years after the ﬁrst diseased heart valve was replaced in a patient,prostheses largely represent the concepts of the 1960s with many of their design-inherent complications. While the sophisticated medicalsystems of the developed world may be able to cope with sub-optimal replacements, these valves are poorly suited to the developingworld (where the overwhelming majority of potential valve recipients reside), due to differences in age proﬁles and socio-economiccircumstances. Therefore, it is the latter group which suffered most from the sluggish pace of developments. While it previously took lessthan 7 years for mechanical heart valves to develop from the ﬁrst commercially available ball-in-cage valve to the tilting pyrolytic-carbondisc valve, and another 10 years to arrive at the all-carbon bi-leaﬂet design, only small incremental improvements have been achievedsince 1977. Similarly, bioprosthetic valves saw their last major break-through development in the late 1960s when formalin ﬁxation wasreplaced by glutaraldehyde cross linking. Since then, poorly understood so-called ‘anti-calciﬁcation’ treatments were added and thehomograft concept rediscovered under the catch-phrase ‘stentless’. Still, tissue valves continue to degenerate fast in younger patients,making them unsuitable for developing countries. Yet, catheter-delivered prostheses almost exclusively use bioprosthetic tissue, therebyreducing one of the most promising developments for patients of the developing world into a fringe product for the few ﬁrst worldrecipients. With tissue-engineered valves aiming at the narrow niche of congenital malformations and synthetic ﬂexible leaﬂet valvesbeing in their ﬁfth decade of low-key development, heart valve prostheses seem to be destined to remain an unsatisfying and exclusiveﬁrst world solution for a long time to come.r 2007 Elsevier Ltd. All rights reserved.Keywords: Heart valve; Bioprosthesis; Calciﬁcation; Immune response; Degradation; Thrombogenicity1. Introduction prosthetic heart valves were implanted into patients [1,2]. By the second half of the 1970s, these crude initial The emergence of heart surgery was closely linked to the prototypes had already evolved into the valve prosthesestechnological advances of the post-second world war we know today: bioprosthetic tissue (BPT) valves had longperiod. Pioneering spirit and innovative academic minds undergone the quantum leap from unstable formalinbroke new ground on an almost daily basis. Once the treatment  to the far superior glutaraldehyde ﬁxationheart–lung machine was invented, it did not take long until  and mechanical prostheses had moved from the original ball-and-cage valve  to tilting disc  and even bi-leaﬂet $ Editor’s Note: Leading Opinion: This paper is one of a newly designs . Against this background it is disappointinginstituted series of scientiﬁc articles that provide evidence-based scientiﬁc that, more than three decades later, commercial productsopinions on topical and important issues in biomaterials science. They are still based on the largely unchanged concepts of thehave some features of an invited editorial but are based on scientiﬁc 1970s. Glutaraldehyde-ﬁxed tissue valves have been onlyfacts, and some features of a review paper, without attempting to be marginally improved by eliminating design ﬂaws andcomprehensive. These papers have been commissioned by Editor-in-Chiefand reviewed for factual, scientiﬁc content by referees. adding so-called ‘anti-calciﬁcation’ treatments. Mechanical ÃCorresponding author. Tel.: +27 21 406 64 76; fax: +27 21 448 59 35. valves are still mostly of the bi-leaﬂet design and E-mail address: email@example.com (P. Zilla). improvements were restricted to better-polished pyrolytic0142-9612/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.biomaterials.2007.09.033
ARTICLE IN PRESS386 P. Zilla et al. / Biomaterials 29 (2008) 385–406carbon coatings or some minor reductions in turbulences in countries these complications are multiplied by a lack ofthe crucial hinge area. In contrast to the pioneering years, compliance with anti-coagulation due to infrastructuralcourage and incentive towards a new developmental and educational deﬁcits [13,14]. Together with the youngerquantum leap have been mostly absent from commercial age of patients (Fig. 1) the individual likelihood ofproduct strategies. experiencing a thrombotic or thromboembolic disaster in the course of a life-time is therefore distinctly higher than in ﬁrst-world patients [13,14]. Unfortunately, the alter-2. Population needs and obstacles native use of contemporary tissue valves is equally over- shadowed in these patients due to the fast degeneration of Unfortunately, socioeconomic circumstances still limit bioprostheses in young recipients.access to valve surgery to a fraction of the patients who Taking all these factors into account, it is obvious that:actually need it. While the bulk of the estimated 275,000 to 370,000  annual valve replacements beneﬁt predomi- The overwhelming majority of potential recipients ofnantly the older patients of the First World, developing prosthetic heart valves reside in developing countries.countries with their much higher incidence of rheumatic The fact that 85% of all open-heart procedures arefever  often have no access to heart surgery. Related to performed in countries representing 11% of theEuropean service levels, for instance, cardiac surgery is world population emphasizes how unrepresentative theonly available to 8.1% of the Chinese and 6.9% of the current heart valve recipients are with regard to globalIndian population . However, given the fast growing needs .economies of some of these countries, it is predictable that The fact that the economies of many of these thresholdthey will soon have a high demand for prosthetic heart countries are growing fast makes it predictable thatvalves that are affordable and that address the speciﬁc cardiac surgery will be increasingly accessible to thoseneeds of their mainly young patients. Since performance millions who are in need of a valve replacement. Sincedemands, complications and degeneration distinctly differ improved socio-economic circumstance will only have abetween age groups, the largely geriatric experience proﬁle delayed impact on reducing the incidence of rheumaticfrom developed regions  does not allow generalized fever, a huge demand for valve replacements will exist inconclusions as to ‘‘what is the best replacement heart these ‘newly developed’ countries for a relatively longvalve?’’ for recipients in threshold countries. The often- time.deadly complications associated with mechanical valve The unattractive proﬁt margins for ‘ﬁrst-world-pros-prostheses accumulate over patient years. In developing theses’ in developing countries as well as the relativelyFig. 1. Typical age-distribution of patients undergoing heart valve replacement in the First World and in a Developing Country. While prosthetic valverecipients in a First World population are predominantly in the age group of 60–69 years (red line) they are broadly disseminated over an age-spectrumfrom 20 to 70 years in a Developing Country such as South Africa (blue line). As the age distribution of 2000 consecutive heart valve recipients at theGroote Schuur Hospital (University of Cape Town) shows, a signiﬁcant proportion of patients is even younger than 20 years.
ARTICLE IN PRESS P. Zilla et al. / Biomaterials 29 (2008) 385–406 387 small market in developed countries, together with the prostheses. A ‘mechanical’ problem such as a leaking valve exceptionally long ‘bench-to-proﬁt’ delays for heart of a ‘pump’ naturally seemed amenable to an ‘engineering’ valve prostheses have hampered major research invest- solution. Without this ability to exploit concomitant ments in this ﬁeld during the past three decades. developments in engineering and material sciences, pre- Inasmuch as locally manufactured prosthetic heart vious pioneers were doomed to fail. Ten years after the ﬁrst valves will help to partially overcome the affordability known operation on a heart valve by Harvey Cushing at factor for threshold countries, they will be replicas of the Peter Bent Brigham hospital in Boston in 1913, Elliott valves which were designed to address the speciﬁc Carr Cutler and the cardiologist Samuel A. Levine circumstances of ﬁrst world patients. attempted a surgical valvotomy in a young female patient Therefore, the discrepancy between the world’s needs with mitral valve stenosis. This operation, hailed as a and industry’s commitment will continue to widen for milestone by the British Medical Journal, had a subsequent many more years: an ever increasing number of young mortality of 90% and was soon abandoned. patients from threshold countries in need of a new In stark contrast to this rather gloomy era, the heady era concept of a truly long-lasting valve prosthesis on the of replacement valves was the decade from the early 1950s one side as opposed to the regardless pursuit of to the early 1960s, when break-through developments in continuously stepped-up sale-efforts of non-innovative, science and engineering were vigorously applied to clinical long-established products of yesterday on the other. needs. Insight into the physiology of hypothermia allowed A good example of the latter is the ‘sales-speak’ of Floyd Lewis of Minnesota to gain open access to an atrium ‘actual’ freedom from structural valve failure as opposed in 1952 followed by Charles Hufnagel’s implantation of a to ‘actuarial’. While the actuarial median failure time of caged-ball heart valve into the descending aorta of ten bioprosthetic valves in elderly ﬁrst-world patients is patients (six survived the operation) in the same year. By approximately 14 years , the ‘actual’ failure-rate 1953, electrical engineering had provided the pacemaker, after the same period of time as reported by industry- and the miniaturization of industrial technologies had supported research is 8% in AVRs and 11% in MVRs allowed John Gibbon Jr. of Philadelphia to introduce the ! As a damage-control measure, the Journal of Heart heart lung machine, thereby for the ﬁrst time allowing Valve Disease announced in an Editorial its intention to unlimited open access for the surgical replacement of a reject any manuscripts in the future that report ‘actual’ heart valve on July 22, 1955 in Shefﬁeld, UK. The epitomes rather than ‘actuarial’ data . of this pioneering spirit were the use of synthetic materials for the ﬁrst mitral valve replacement by Nina Braunwald in For the time being, we need to accept that the limited 1960 and the joint effort of the surgeon Albert Starr andmarket for valve prostheses in today’s world of high proﬁts the electrical engineer Miles ‘‘Lowell’’ Edwards—founderwith mass-selling low-tech products such as coronary stents of the medical device company ‘American Edwardswill continue to perpetuate a situation whereby heart valve Laboratories’ in 1950—to improve the ball-and-cage valveresearch is too cumbersome and unattractive for the main to a level where it could reliably be used as a replacementplayers to sufﬁciently invest in truly disruptive technologies. heart valve (Fig. 2). The ﬁrst human implant of this valveThis is opposed by a ﬂurry of innovative ideas by on September 21, 1960 concluded the most vibrant decadeUniversity-based entrepreneurs and start-up companies. of development in this ﬁeld. During the subsequent 15However, the particularly long and costly lead-time between years major insights into ﬂuid dynamics, material sciencesproduct development and marketing as well as the moderate and engineering were still carried over into the improve-proﬁt margins will unfortunately prevent many of these ment of replacement heart valves, but they were no matchinnovations from becoming a mainstream product. Until for the quantum leap of the preceeding era. The two crucialthen, the hype around the endovascular delivery of yester- elements of these post-pioneer years were the adaptation ofday’s concepts may ironically and unintentionally turn into technologies that allowed the production of tilting discs ora success story for the huge number of potential recipients of semi-discs (Fig. 2) as occluders and the utilization ofheart valve prostheses in threshold countries: In contrast to materials developed in the course of atomic energythe heavily calciﬁed heart valve pathology in the ﬁrst world, research. The scientist J.C. Bokros from the ‘Generalyoung patients with rheumatic valves almost naturally offer Atomic Company’ had previously investigated pyrolyticthemselves as ideal recipients of endovascularly or transa- carbon materials for the coating of nuclear fuel particles.pically placed prostheses as their diseased native valves are Since this material’s atomic microstructure was found tomostly delicate and non-calciﬁed. With the incentive of a increase resistance to cracking and distinctly loweredpotential mass market, durability issues in young patients thrombogenicity, it was introduced into heart valvemay eventually be confronted. prostheses. For more than three decades all commercial mechanical heart valves have either been using it for3. Mechanical heart valves surface coating (by depositing gaseous hydrocarbons— usually methane—onto a heated graphite substrate at The groundbreaking initial years in the development of temperatures of between 1800 and 2300 1C) or for entireprosthetic heart valves are tightly linked to mechanical leaﬂets and housings of the valves (made from 100%
ARTICLE IN PRESS388 P. Zilla et al. / Biomaterials 29 (2008) 385–406 3.1. Mechanical durability Out of 86,000 convexo-concave Bjork-Shiley tilting-disc valves, 619 experienced a fracture of the outﬂow strut, which led to the patient’s death in two-thirds of the cases. On a smaller scale, 46 leaﬂet escapes were reported out of 200,000 implants of the Edwards–Duromedics bileaﬂet valve . As unsettling as these failures were, and still are, for the many patients living with one of these prostheses (e.g. 23,000 Bjork-Shiley valves in North America alone), they provided crucial insight into key elements of the three major challenges of engineering, namely closing load, material fatigue and cavitation. As far as the ﬁrst is concerned, the major load on mechanical heart valves generally arises from transvalvularFig. 2. The development of mechanical heart valves experienced an pressure generated at and after valve closure leading toexponential slow-down over it’s more than 50-year history: while it took both ‘impact wear’ and ‘friction wear’. Impact wear usuallyonly 7 years from the ﬁrst commercially available ball-in-cage valve to the occurs between occluder and ring in tilting discs and theera of pyrolytic carbon-coated tilting disc valves and 10 years from the hinge regions in bileaﬂets. At the same time, friction wearlatter to the ﬁrst all-carbon bileaﬂet design, progress of the past 30 yearswas incremental at most with leaﬂets being either more parallel aligned or occurs between occluder and struts in tilting discs andmildly curved and hinge mechanisms slightly less thrombogenic. between leaﬂet pivots and hinge cavities in bileaﬂets. In the case of the Bjork-Shiley valve, the damage occurred on the outlet strut, causing bewilderment as to how the closing load could damage the outlet strut. In 1984, Shiley itselfpyrolytic carbon using ﬂuidized bed processes). Overall, provided the answer by calling the discovery of a transientmechanical heart valve developments were largely con- (o0.5 ms) occlusion impact on the outlet strut tip due tocluded in 1977 with the introduction of an all-pyrolytic over-rotation of the disc ‘bimodal closure phenomenon’.carbon bileaﬂet valve (St. Jude Medical/SJM). What This force was ten times higher than the actual openingfollowed was attention to engineering detail rather than impact, causing bending stresses that exceeded the strutgrand new concepts. Not surprisingly, the one aspect of wire’s fatigue endurance limit. In Duromedic’s case, whereengineering which represented the monopoly component the leaﬂet’s seat was created through a lip or shelf on theof production, pyrolytic carbon, became the centre of housing (rather than a cavity in the housing like in all othercorporate interest and ‘merry-go-round’ acquisitions: ﬁrst, models), erosion contributed to the breakages. All theseDr. Bokros of the ‘General Atomic Company’ helped to set insights were addressed by modern strut- and hinge-up their medical division under the name ‘‘Carbomedics’’. designs, some of them (Bjork-Shiley) by slightly increasingAt one stage, ‘‘Carbomedics’’ manufactured 17 heart the outlet strut clearance, some of them (On-X) by evenvalves for 16 different companies! Subsequently, Dr. claiming a ‘two-point landing system’ that reduces the loadBokros split off ‘‘Carbon Implants’’. When Medtronic impact.bought it in 1992 to obtain their ‘‘Parallel Valve’’, Bokros The second most important cause of failure, materialsplit off the ‘‘Medical Carbon Research Institute’’ (MCRI) fatigue as a consequence of the polycrystalline character-in 1994, which designed the On-X valve and later changed istics of metals, was successfully addressed by a gradualthe name MCRI to On-X. At the same time, ‘‘Carbome- shift from metal alloys to pyrolytic carbon, a material thatdics’’ (Austin, TX) itself was bought by ‘‘Sulzer Medica’’ of belongs to the ‘turbostatic carbons’ and thus has a similarSwitzerland in the late 1980s, and then in turn by ‘‘Sorin’’, structure to graphite. Initially, strength and wear resistancethe medical division of Fiat (the Italian automobile of pyrolytic carbon were increased by adding up to 8% ofmanufacturer), which eventually transferred the produc- its weight as silicon, co-deposited from silicon carbides, buttion from Austin to an FDA approved site in Milan, Italy. recent improvements by the same group, which has beenThe incestuous nature of corporate acquisitions in this driving pyrolytic carbon development for the past 40 years,ﬁeld is further highlighted by another pioneer ending in led to what they call ‘pure carbon’. Allegedly, theItaly: After Pﬁzer took over Shiley, the very durable elimination of the silicon component resulted in a 50%Bjork-Shiley tilting disc valve was re-designed, leading to stronger material. The switch to carbon was ﬁrst pioneeredone of the biggest disasters in the history of medical in the occluder discs when pyrolytic carbon-coated graphiteimplants when many of the convexo-concave Bjork-Shiley replaced the original Delrin disc in Bjork-Shiley valves invalves developed mechanical failure [19–21]. Eventually 1968 (ﬁrst implanted in 1971). Since then, all manufac-Pﬁzer sold Shiley to Sorin, but Sorin chose to exclude the turers have been using this carbon-combination for therights to the failing ‘convexo-concave’ valve from this production of their mono/bileaﬂet discs except for MCRIpurchase. (On-X) which introduced leaﬂets made of solid, silicon-free
ARTICLE IN PRESS P. Zilla et al. / Biomaterials 29 (2008) 385–406 389pyrolytic carbon in 1996. In contrast to leaﬂets, the clinical trials before obtaining the European CE Mark andhousings and struts were made out of metals until well American FDA approval. Unfortunately, this rigorous pre-into the 1980s. Initial attempts at overcoming the fatigue of market testing added signiﬁcantly to the research andmetals involved either using alloys or particularly strong development costs of the prostheses. The resulting costpure metals such as titanium. ‘Starr-Edwards’ ball-and- increases of mechanical heart valves added anothercage valves were the ﬁrst to replace the original stainless obstacle to the widespread use of these prostheses insteel by a cobalt–chromium–molybdenium–nickel-alloy, a developing countries.material that was later also used for the Bjork-Shileyvalves. The introduction of titanium to mechanical heart 3.2. Weighing life and deathvalves also took place during the days of ball-in-cage valveswhen pure titanium was used for the Smeloff-Cutter valve The Bjork-Shilry ‘‘convexo-concave debacle’’ did notas early as in 1965 . Later, tilting-disc designs such as only help to identify the weak points of mechanical heartthe Medtronic Hall (1977), Sorin Monoleaﬂet (1977), valve engineering but also taught us to deal calmly withLilleihei-Kastner (1978) and Omniscience (1978) valve potential failure by balancing the risk of valve fracturemachined their housings out of a solid block of titanium. against the risk of surgery to replace the valve. The decisionEventually, the introduction of the bileaﬂet design in 1977 of both the University of Alabama and the Mayo Clinic tocoincided with the advent of the ‘all carbon’ valves where remove every concavo–convex valve may not hold retro-both leaﬂets and housing were made of pyrolytic carbon. spective scrutiny. An average re-operation mortality of 8%This bileaﬂet concept has continuously dominated the may well outweigh the actually occurred 0.9% incidence ofmarket for the past 30 years: St. Jude Medical (SJM; since strut breakage, even if risk-stratiﬁcation takes the three1977; 800,000 sold); Duromedics (1982–1988; 200,000 sold particularly prone groups into account which were retro-; reintroduced in 1990); Carbomedics (introduced in spectively identiﬁed : (a) 33 mm mitral valves, for1986 in France, 1993 in USA; 500,000 sold by 2004); instance, were 23 times more likely to fracture thanAdvanced Tissue Science (ATS; since 1992; 135,000 sold); 21–25 mm aortic valves as were (b) valves with moreMedtronic (since 1999, after ‘Parallel’ valve was bought ﬂexible outlet struts, as determined by hook deﬂection andfrom ‘Carbon Implants in 1992) and On-X (since 1996 load deﬂection tests during manufacture and (c) valvesoutside USA, 2001 USA; 50,000 sold). Yet, Titanium is still produced by one particular welder.part of the housing in some designs (e.g. Lockring andLockwires in ATS and the housing body in Sorin Bicarbon 3.3. Biological limits of mechanical solutionswhich later became the Edwards Mira in 1997). Overallit appears as if material science at least has provided us Given the fact that both the ‘ﬂuid’ ﬂowing through thewith hardly improvable, optimal materials for mechanical valves and the ‘pump’ into which these valves areheart valves. incorporated consist of living cells, problems are likely to The third most important mode of failure is ‘cavitation’. be augmented by cascades of biological events. Fluid shearThis well-known phenomenon is deﬁned by the rapid forces and turbulences, for instance, do not only lead toformation of vaporous microbubbles in a ﬂuid due to a energy losses but to an activation of the platelet andlocal reduction of pressure below vapour pressure—similar coagulation system which has the potential to immobilizeto boiling . When the pressure conditions for ‘cavita- the entire valve or embolize into vital organs. Similarly,tion’ are present, bubbles will start to form and grow. small valvular oriﬁces do not only lead to higher pressuresWhen the pressure recovers, the formed bubbles will in the outlet chamber but to a remodelling process of theimplode, producing pressure and thermal shock waves contractile tissue that may eventually lead to irreversiblethat can impinge on solid surfaces. Pressure oscillations, pump failure.ﬂow decelerations, tip vortices and ‘squeeze jets’ are allcapable of inducing cavitation, the last one being the most 3.3.1. Effective oriﬁce area (EOA)contributive factor. Squeeze jets are formed when the valve Overall, it appears as if minimizing the energy requiredis closing and the blood between the occluder and the valve to eject the blood through the valve has been morehousing is ‘squeezed’ out to create a high-speed jet. This in successful than eliminating the activation of the coagula-turn creates vortices that can lead to additional cavitation. tion cascade. The ‘effective oriﬁce area’ (EOA) is a meansDamages found on failed Duromedics valves, for instance, of expressing the degree to which a prosthesis impedeswere associated with conﬁned areas of erosion and pitting blood ﬂow through the valve and thus increases the energyon leaﬂets and housing  caused by cavitation. loss. In contrast to the energy efﬁcient central ﬂow of Overall, the disasters of the 1970s and 1980s seem to natural heart valves, caged-ball valves completely blockedhave helped eliminate mechanical failure. The fact that not the central ﬂow. While the ﬁrst tilting disc valve (Melrose)a single failure was reported on later designs (e.g. after half in 1964 introduced some degree of central ﬂow, the majora million Carbomedics implants) indicates that combined break-through occurred in 1977 with the ﬁrst bi-leaﬂetefforts were successful. Part of these efforts is the valve. The magnitude of this development is best expressedrequirements for much more extended animal tests and by the increase of EOA from 1.5 to 2.1 cm2 in tilting disc
ARTICLE IN PRESS390 P. Zilla et al. / Biomaterials 29 (2008) 385–406valves to 2.4–3.2 cm2 in bileaﬂet designs of the same outer activates platelets and factor VIII. In parallel, plateletdiameter. Again, improvements after 1977 were incremen- activation is triggered when shear stresses reach a leveltal rather than disruptive. By moving the sewing ring supra higher than 60–80 dyn/cm2, followed by von Willebrandannularly, for instance, additional increases of the inner factor binding to the platelet receptor GPIb and calciumdiameter were possible (e.g. 17% On-X; Carbomedics; etc). release. GP IIb–IIIa facilitates platelet–platelet adhesion.Furthermore, by successively aligning the leaﬂets parallel Last, but not least, contact activation begins when factorwith the blood stream, unobstructed central ﬂow became XII binds to the non-endothelialized artiﬁcial surfaces ofalmost a reality. Again, most of the progress was already prostheses. This in turn activates prekallikrein and high-made before 1978 when the tilting disc opening-angle of 601 molecular weight kininogen. Although ‘back-wash’ designsof the ﬁrst Bjork-Shiley valve was increased to 751 (in have aimed at reducing the thrombus formation in theaortic; 701 in mitral valves) in the Medtronic Hall in 1977 hinge area [31–35], mechanical prostheses still require anti-and to 801 in the Lilleihei-Kaster valve (later Omniscience) coagulation and the promise of freedom from thromboem-in 1978. The introduction of the bileaﬂet design only bolism is still elusive and remains the Achilles heel of allincreased this angle to 851 (SJM 1977), with some makes mechanical prostheses, even in the ﬁrst world. Althougheven having a lower opening angle than tilting discs the incidence of thromboembolism has changed from the(Duromedics 771/731 in aortic/mitral valves, respectively, 4.5% per patient year [36–41] in the days of the ball-in-cageand Carbomedics 801/781). Even with a 901 opening angle valves, it is still 0.6–2.5% in Omnicarbon tilting disc(Medtronic Parallel; On-X), however, the actual travel arc valves [42,43]; 1.3–4.2% in Medtronic Hall valvesthrough which the leaﬂets swing is/was mostly less than 601 [13,32,33,37,44–47]; 1.1–3.7% in SJM valves [14,48–54];(e.g. 551/601 SJM; 531/551 Carbomedics; Duromedics 0.6–4.3% in Carbomedics valves [55,56]; 0.0–3.3% in ATS581/621 and 501 Medtronic Parallel), resulting in a steeper valves [57,58] and 1.8% in On-X valves . The mostclosing angle and thus carrying the potential of a higher lethal and debilitating forms of thromboembolism, how-regurgitation volume. ever, is thrombotic prosthetic obstruction. While the previously reported estimates for the tilting disc valves3.3.2. Thromboembolic risks were between 0.1% and 1.2% per patient year in the ﬁrst Mechanical heart valves remain vulnerable to thrombus world, there is general recognition that this became anformation due to high shear stress (one of the strongest extremely rare complication in the era of bileaﬂet valvesplatelet activators), ﬂow separation/stagnation (since unless anti-coagulation had been stopped. Therefore, givenVirchow’s days the dreaded cause of coagulation) and the unreliable degree of anti-coagulation in developingblood damage (through the release of pro-coagulant countries, it is not surprising that thrombotic valvesubstances). As a consequence, anti-coagulation therapy obstruction is still a feared occurrence outside thehas been a hitherto sine qua non. In general, thromboem- developed world. Unfortunately, the true number ofbolism (e.g. stroke) is more pronounced in mitral valve patients dying as a consequence of valve clotting will oftenreplacements, and within this group, more so in patients remain unrecognized in these geographic regions due to thewith large left atria and in those in atrial ﬁbrillation and/or lack of centralized death registries and cultural resistancescardiac failure [28,29]. to autopsies. In tilting-discs, ﬂow separation occurs behind the valvestruts and distal to the leaﬂet edge as a result of a 3.3.3. Coumadin-related haemorrhagecombination of high velocity and stagnant ﬂows. Of all The necessity for anti-coagulation with Coumadindesigns, the largest turbulent stresses of about 150 Pa were (Warfarin) that minimizes, but never completely eliminatesfound behind tilting disc valves in the minor ﬂow region the risk of thromboembolism also exposes patients to theparallel to the tilt axis . The bileaﬂet models have high additional risk of major haemorrhagic complications. Thestress during forward ﬂow and leakage regurgitation as most lethal and debilitating of these is intracranialwell as adjacent stagnant ﬂow in the hinge area. As it turns haemorrhage, which occurs at a rate of between 0.8%out, the hinge area is the most critical part of bileaﬂets and and 3.7% per year . Although INR requirements haveis where the thrombus formation usually commences. come down from 3.5–4.5 in ball-in-cage-valves to 2.5–3.5, Prosthetic valve-induced clot formation-triggered by and anti-coagulation-caused haemorrhages are signiﬁ-tissue factor exposure, platelet activation and contact cantly fewer being reported by various authors [61,62],activation by foreign materials—occurs in three steps: the numbers are still worrying. While they were reported toinitiation, ampliﬁcation and propagation. Tissue factor be between 0.6% and 3.7% per patient year in Starr(TF) release usually occurs through cell rupture with blood Edwards valves [38,40], they were 0.0–2.7% in Omnicarbontrauma playing a central role. High stresses during leakage valves [63–65], 1.4–3.2% in Medtronic Hall valves [47,66];ﬂow in aortic valves and forward ﬂow in mitral valves lead 0.3–2.8% in SJM valves [53,54], 0.0–2.8% in Carbomedicsto blood damage releasing both TF and the platelet- valves [53,67], 0.0–4.9% in ATS valves  and 0.0% inactivating ADP. Plasma factor VII binds to TF, setting off On-X valves . Naturally, these complications occur at aa chain reaction which activates factor Xa and Va which distinctly higher rate in the developing world where thebind to each other to produce thrombin which in turn ability to ensure good anti-coagulation control is very
ARTICLE IN PRESS P. Zilla et al. / Biomaterials 29 (2008) 385–406 391limited and often impossible due to poor socio-economic anti-coagulation of mechanical heart valves for the ﬁrstcircumstances [13,14,68–70]. Typically, half of these time ventured into the ‘grey zone’ of INRsp2.2 .patients presenting with obstructive valve dysfunction havean INR of 1.4 or less at the time of failure . 4. Tissue valves3.3.4. Pannus overgrowth Bioprosthetic heart valves were the answer to the Excess tissue growth across the sewing ring can lead thromboembolic complications of mechanical valves. Sinceeither to a narrowing of the oriﬁce or to leaﬂet commercial tissue quantities could be only obtained fromimmobilization. Although widely underestimated, it is animal sources, antigen masking through cross-linking waslikely the primary cause of obstructive prosthetic valve an integral part of the concept from the beginning.failure in patients on correct anti-coagulation therapy. In general, few improvements have been added since thePure pannus overgrowth is estimated to be behind introduction of glutaraldehyde ﬁxation in the late 1960s obstructive valve failure in between 31%  and 53% and its translation into commercial valves (Hancock 1972; of cases. Even in geographic regions where the INR Carpentier Edwards Porcine 1970/1976). Other than thecontrol is sub-optimal, the reason for obstructive valve elimination of design ﬂaws such as ill-placed suspensionfailure is still overwhelmingly pannus rather than thrombus stitches (Ionescu Shiley 1976)  or over-zealous efforts torelated . In the aortic position, pannus formation is reduce the valve proﬁle (BioImplant) , poorly under-mainly on the inﬂow side  while in the mitral position it stood ‘anti-calciﬁcation’ treatments were added [81–83]. Atoccurs both on the atrial and the ventricular side . the same time, market-driven preferences for either bovineIn order to prevent tissue ingrowth into the clearance of pericardial or porcine aortic valves cultivated a pseudo-leaﬂets, one contemporary valve design introduced a longer sense of progress amongst surgeons while in fact theyhousing cylinder with the goal of creating an ingrowth merely reﬂected the circumstances of the time rather thanbarrier. scientiﬁc insight. While in previous eras clinical failures due Although pannus formation is a complex event, inﬂam- to design ﬂaws affected the surgeon’s preferences formation certainly contributes to a sustained growth signal bovine or porcine products, prion and virus concerns asfor the hyperplastic tissue. One of the sources of growth well as tissue availabilities and production advantagesstimulation is the chronic foreign body reaction against the determine today’s marketing thrusts towards ‘bovine’ orsewing ring material. The dominant cells in this reaction ‘porcine’ valves. Last, but not least, the much-hailed ‘new’are macrophages and giant cells which are known to secrete concept of ‘stentless-valves’ (SJM Toronto SPV 1991;an extended range of powerful cytokines and growth Edwards Prima 1991; Medtronic Freestyle 1991; BioCorfactors, including IL-1, PDGF, bFGF, etc. . 1994), which emulated the homograft concept of the 1960s grossly overstated the biomechanic advantages. Slowly3.4. Way forward for mechanical valves emerging mid-term results conﬁrm the largely unfulﬁlled expectations . The fact that stentless roots do not Modern investigative technologies (microvascular ﬂow calcify more than homograft roots  is less a reﬂection ofvisualization, computational ﬂuid dynamics modelling, the success of stentless valves than the degree of degenera-laser Doppler velocimetry and anemometry measurements) tion occurring in homografts.have contributed to the latest bileaﬂet hinge pocket designwith the goal of further reducing turbulences and pro- 4.1. Insight into failure modescoagulant ‘pockets’. The two most recent bileaﬂet valves onthe market, the On-X and the Medtronic Advantage valve, The absence of living tissue, as well as the necessity tohave been designed and evaluated using these modalities mask antigenicity, represents the system-inherent problems. This has already resulted in one study reporting of contemporary tissue valves. Instead of recognizing theselowered rates of valve thrombosis and thromboembolism core issues and addressing them from the beginning,in a third world population . Whether attention to however, there was a widely prevailing perception thatdetail such as lower turbulence levels and better polished immunogenicity non-vital tissue was acceptable as long asmaterials may eventually allow the safe implantation of the right ‘anti-calciﬁcation’ treatment was found. On themechanical valves into the young patients of developing basis of these two assumptions, bioprosthetic heart valvecountries, regardless of their poor anti-coagulation control, research focused almost exclusively on ‘down-stream’will need to be elucidated in large, dedicated studies. One problems such as the intrinsic potential of glutaraldehydesuch study pursued by SJM has just been approved. to trigger calciﬁcation although it was well knownAnother reduced anti-coagulation trial by On-X, using that calciﬁcation affects less than half of failed tissuelowered anti-coagulation targets in the mitral position and valves while tears, as a consequence of inﬂammation,only Aspirin (without Coumadin) in the aortic position, collagen degradation and a lack of repair mechanisms werehas also been approved. In the meantime, an incremental the predominant modes of failure [86,87]. Eventually, thelowering of the INR-bar may be on the cards for many role of inﬂammation was revisited and a link betweenof the current prostheses. Most recent guidelines for the remnant-immunogenicity and tissue degeneration was
ARTICLE IN PRESS392 P. Zilla et al. / Biomaterials 29 (2008) 385–406established , but recognition and implementation areslow.4.1.1. Remnant tissue immunogenicity It has long been known that the low-dose glutaraldehydetreatment used for the ﬁxation of bioprosthetic heart valvesreduces immunogenicity but does not abolish it (Fig. 3). The low-grade ﬁxation used in commercial valvepreparations fails to signiﬁcantly alter membrane-boundreceptors or structural glycoproteins . Therefore, thetissue continues to elicit both cellular and humoral immuneresponses . By increasing crosslink density, tissueantigenicity could be reduced [88,89,91] and the humoralantibody-response mitigated. Most importantly, by de-monstrating a link between immune response and calciﬁcdegeneration, a key aspect of bioprosthetic degenerationcould be clearly traced to insufﬁciently masked immuno-genicity [88,92–95] (Fig. 4). After identifying porcine Fig. 4. Support for a possible role of graft-speciﬁc IgG in the calciﬁcation of bioprosthetic tissue provided by atomic absorption spectroscopyﬁbronectin as a major persisting antigen (Fig. 3) , the analysis of calcium levels in subdermally implanted glutaraldehyde-ﬁxedmain challenge ahead will be to identify more tissue porcine aortic wall tissue in the rabbit. An almost three-times higher levelantigens and elucidate the exact mechanisms through of calciﬁcation was found in bioprosthetic tissue that was exposed towhich antibody binding facilitates degeneration. serum containing graft-speciﬁc antibodies Ref.  with permission.4.1.2. Inﬂammatory degradation of macrophages are serious culprits of destruction. In In the vast majority of explanted tissue valves, inﬂam- bovine pericardial heart valve prostheses, for instance,matory cells are either found to cover the surfaces  or macrophages are regularly found invading and focallyfocally inﬁltrate into the tissue . Since the initial signs of degrading the prosthetic collagen [90,101] (Fig. 5). Simi-xenograft rejection in the form of polymorphnuclear larly, clinically implanted porcine valves also showinﬁltrates [99,100] give way to a more macrophage- and inﬂammatory cells inside the BPT . Conspicuously,foreign body giant cell-dominated phenomenon, a percep- dense inﬁltrates of inﬂammatory cells are found in valvestion of harmlessness ensued. Given the destructive poten- which had failed due to tissue tearing [100,102]. Macro-tial of giant cells to erode even synthetic materials, phage-mediated tissue degradation is further evidenced byhowever, it becomes obvious that these ‘war-formations’ the direct proof of collagen phagocytosis [99,103]. In clinical series, as many as 82% of failed valves showed signs of collagen phagocytosis by macrophages in trans- mission electron microscopy  (Fig. 6). 4.1.3. Mechanical damage/lack of repair The consequences of BPT being crosslinked and non- vital—and thus deprived of repair mechanisms—are most obvious when it comes to mechanical wear and tear: in the mechanically more stressed mitral position, as many as 75% of failed porcine prostheses show a rupture of a free cusp edge and 43% of the cusp belly . In vital, native valves both a macromechanical and a micromechanical principle reduce the impact of mechanical load. Macro- mechanically, the distinct dilatation of the annulus during systole results in the triangular stretching of cusp tissue,Fig. 3. Demonstration of the need for increased crosslink density to thereby avoiding acute bending at the commissures beyondmitigate residual immunogenicity of glutaraldehyde ﬁxed porcine aortic 601 . Micromechanically, a sophisticated tri-layeredwall tissue. Commercially applied concentrations of the dialdehyde (up to ultrastructural architecture ascertains optimal stress reduc-1.0%) failed to avoid an IgG response to porcine bioprosthetic tissue—in tion whereby the lamina spongiosa provides a ‘sliding gap’particular the glycoprotein ﬁbronectin—in a rabbit subdermal implant between the relatively smooth lamina ventricularis on themodel. Only glutaraldehyde concentrations in excess of 1.0% (here 3.0%ideally with L-lysine extension) were able to quench the response outﬂow side and a folded lamina ﬁbrosa on the inﬂow side.(Densitometry plot of a porcine Fibronectin Western blot using 1:100 Bioprosthetic heart valves—stented or stentless—defydilutions of rabbit sera) Ref.  with permission. both these principles of valve mechanics: by either
ARTICLE IN PRESS P. Zilla et al. / Biomaterials 29 (2008) 385–406 393 Fig. 6. Typical transmission electron micrograph of a macrophage inﬁltrating into the depth of a porcine bioprosthetic heart valve. The phagocytotic vesicles containing remnants of collagen are evidence that these macrophages are no ‘innocent bystanders’ but rather active players in the degeneration of bioprosthetic tissue. overgrowth onto leaﬂet tissue and cross-link density has been shown . 4.2. Scope for improvements 4.2.1. Anti-calciﬁcation treatments Although the unifocal attention to intrinsic tissue calciﬁcation was disproportionate, it remains an integral component of bioprosthetic degeneration and as such needs to be part of improvement strategies. Tissue calciﬁcation is a multifactorial process, of which insufﬁ- cient immune masking is only one component. Glutar-Fig. 5. Torn leaﬂet of a pericardial prosthesis leading to acute mitral aldehyde is another factor, as this dialdehyde is known toregurgitation 56 months after implantation. All retrieved tissue samples intrinsically elicit calciﬁcation. Initially, the glutaralde-showed signiﬁcant inﬂammatory inﬁltration into the depth of the leaﬂet. hyde-based crosslinks themselves were suspected of beingAlthough the inﬂammation was macrophage-dominated [(a) CD 68/Ham pro-calciﬁc . However, when extraction methods for56; 40 Â ] there was also signiﬁcant involvement of granulocytesdominated [(b) neutrophil elastase; 20 Â ]. the unbound derivates of glutaraldehyde were developed, it became clear that the intrinsic potential of glutaraldehyde to contribute to the calciﬁcation process rests in the free dialdehydes and their polymerization products rather thanstent-mounting, or through glutaraldehyde ﬁxation, annu- the crosslinks. By extracting unbound glutaraldehyde, twolar dilatation during systole is either eliminated or major improvements were achieved: tissue became non-dramatically diminished. As a consequence, the bending toxic allowing even endothelial cells to grow on it  andangle becomes acute and reverse curvature and buckling calciﬁcation could be distinctly mitigated, even if muchensues. By eliminating the ‘sliding gap’ between the higher concentrations of glutaraldehyde were used forventricularis and the ﬁbrosa as a result of crosslinking, ﬁxation [109–111]. The ﬂurry of approaches marketablya major means of stress distribution is lost. termed ‘‘anti calciﬁcation treatments’’ often represented variations on glutaraldehyde detoxiﬁcation. Substances4.1.4. Pannus overgrowth such as diphosphonates, for instance, are believed to act Inasmuch as similar triggers initiate the tissue over- twofold: aldehyde stabilization occurs through binding viagrowth of the sewing ring as in mechanical valves, the Schiff’s bases and subsequent reduction by NaBH4, whileprocess is augmented by the inﬂammatory potential of the restriction of crystal growth is thought to occur throughBPT. Amongst other factors, the extent of this inﬂamma- direct diphosphonate binding to developing hydroxyapa-tion is a reﬂection of the degree of insufﬁcient immune tite nucleation sites . The precise mechanism of themasking of the BPT. A clear correlation between pannus anticalciﬁcation effect of a-amino-oleic-acid (AOAs)—the
ARTICLE IN PRESS394 P. Zilla et al. / Biomaterials 29 (2008) 385–406anticalciﬁcation substance incorporated into Medtronic’s However, regardless of whether the inﬂammatory responseMosaic and Freestyle valves—is not fully understood is co-triggered by matrix immunogenicity and byproductseither, but it is thought that it also binds with its two of the extraction process or only by matrix immunogeni-amino groups to free aldehyde groups. Residual aldehyde city: the latter will need to be addressed. The most obviousis pivotal for the binding of AOA . way of dealing with it is an additional cross-linking step. Interpreting contemporary research programs, however, it4.2.2. Inﬂuence of crosslinking on degeneration seems unlikely that the cross-linking of decellularized BPT Cross-linking appears to have a direct and an indirect will be eagerly embraced. The reason for this reluctance lieseffect on tissue calciﬁcation: while the density of cross-links in the fact that ﬁxation would conﬂict with one of theindirectly acts through different degrees of immune unfulﬁlled hopes behind decellularization. This unfulﬁlledmasking [88,89,91], there is an additional direct effect hope is the misconception that—once implanted—anthrough the chemistry of the cross-linking process itself. ‘ideal’, ‘natural’ scaffold such as a decellularized matrixOne empirical approach was therefore to experiment with would be eagerly populated by host cells. Whereas somealternative crosslinking procedures such as Photoﬁx evidence appears to support the ingrowth potential of[113,114] or alternative crosslinking agents such as Epoxy acellular xenogenic matrices, at least in the sheep [127–129]compounds , Carbodiimide , Diglycidyl , and rat , the debate is skewed by the multitude ofReuterin , Genipin  and many others. An decellularization methodologies in use, the wide range ofalternative approach to this empirical permutation of extraction times employed and the questionable animalcrosslinking agents emerged in the wake of tissue- and models used to validate the process. Our own experiencesbio-engineered valves. The realization that engineering (unpublished) with exhaustively washed decellularizedprinciples can be applied to hitherto empirical processes porcine aortic valves have conﬁrmed a failure to repopulate led to a more fundamental and mechanistic analysis in rat, rabbit and primate models, with any ingrowthof the ‘engineerable’ elements of cross-linking. After potential confounded by a massive granulocytic response.anchoring the cross-links to the carboxyl rather than the This is supported by the ﬁndings of others conﬁrming theamino groups, for instance, it could be demonstrated that ability of decellularized tissue to attract mononuclear cellsunblocked amino groups in the tissue have a strong pro- and granulocytes [131,132]. The last blow to the concept ofcalciﬁc effect  and vice versa, the length and tissue ‘vitalization’ came with the clinical implantationhydrophobicity of the blocking agent determine the degree of SynerGraftTM Valves, where no evidence of matrixof suppression. Once such rational ‘engineering’ repopulation was seen even after a year of implantation.approaches allowed the almost complete elimination of Stock et al.  even suggested that, in an inert acellulartissue calciﬁcation , other consequences of tissue matrix devoid of MMPs and their inhibitors (TIMPs),ﬁxation, such as stiffness, became amenable to a similar repopulation by cells from adjacent tissue is biologicallyapproach . unlikely .4.2.3. Cellular antigen removal (decellularization processes) 4.3. Way forward for tissue valves Cell extraction has been tried since the 1950s . Fromthe beginning, the primary goal was to extract cellular The mantra of bioprosthetic research must be theimmunogenicity and thereby eventually manage to avoid recognition of its system-inherent limitations such ascrosslinking. The misconception of this approach was that matrix immunogenicity and perpetual non-vitality. Inas-it exclusively targeted cell surface molecules (MHC-II in much as the removal of the xenogenic cells reduces thethe case of allografts and the Gal-a1,4-Gal xenoantigen in immune burden, the continual immunogenicity of theheterografts), implying that the extracellular matrix is non- extracellular matrix will always require masking. Similarly,immunogenic. Apart from the fact that extracellular matrix believing that the extraction of cells will leave voids behindhas been shown to be immunogenic , the decellulariza- which are inviting spaces for ingrowing host cells means totion process itself holds a pro-inﬂammatory potential. ignore the decades-old knowledge that even the best ofApart from the uncertainty with which solubilized antigens decellularized heart valves do not get repopulated:and remnant detergents are eluded from the tissue, the Although homografts are non-crosslinked they remaindecellularization process itself may even liberate pro- largely acellular even after years of implantation .inﬂammatory substances. Arachidonic acid, for instance, However, if our goal is less ambitious and we just want toa major component of the membranes of cellular orga- signiﬁcantly extend the longevity of a non-vital bio-nelles, may potentially not be removed by aqueous buffers prosthesis, engineered crosslinking already offers a solu-and thereby play a role in neutrophil chemotaxis through tion, potentially augmented by combining the process withleukotriene B4, a product of the arachidonic acid pathway. a preceeding immune burden reduction such as decellular-The latter may partly explain the massive polymorph- ization. A more recent discovery may even allow one tonuclear response to the clinically implanted ‘‘Syner- achieve immune masking without cross-linking: by ﬁllingGraftTM’’ valves [125,126] that were used as right non-crosslinked tissue with a hydrogel, antigens andventricular outﬂow tract replacement in Ross procedures. proteins seem to have been spatially rendered inaccessible
ARTICLE IN PRESS P. Zilla et al. / Biomaterials 29 (2008) 385–406 395Fig. 7. A novel approach to abrogate bioprosthetic tissue calciﬁcation and degradation involves the ﬁlling of the tissue with hydrogel rather than cross-linking the amino or carboxyl groups of the tissue as conventionally done in tissue ﬁxation. The ramiﬁed blue lattice of the hydrogel ﬁlling the spacebetween two yellow collagen strands schematically shows how a ‘space ﬁller’ has replaced cross-links. The tissue structure is well preserved after ﬁlling with20% acrylamide (+ bis-acrylamide crosslinker; UV initiation) (top left; Goldner; original magniﬁcation: 40 Â ) and calciﬁcation (60 days ratsubcutaneous) is well below that seen after standard glutaraldehyde ﬁxation (bottom right) .by cells, enzymes or immune molecules. It was hypothe- numerical modelling subsequently allowed for optimiza-sized that it was this inaccessibility that prevented the tion of ﬂow dynamics and leaﬂet stress. Variations rangedin vivo degradation, inﬂammatory inﬁltration and calciﬁca- from hemi-cylindrical cusps  to leaﬂets in half-opention of hydrogel-inﬁltrated tissue (Fig. 7) . position [143,144]; from variable-curvatures  to elliptical and hyperbolic shapes  and from a conical5. Polymeric ﬂexible-leaﬂet valves base with spherical upper part  to high proﬁle cusps (Fig. 8b) [137,148]. Notable exceptions to the tricuspid The promise of synthetic heart valves was to have design include the very ﬁrst bileaﬂet mitral valve implantedprostheses available that have the durability of mechanical by Braunwald , the single-leaﬂet silicone-covered Dacronvalves and the haemocompatibility of tissue valves. In cusps implanted by Hufnagel in the 1970s  and anreality, most of the polymeric valves were the opposite, asymmetric bileaﬂet mitral valve [137,150] comprising acombining the durability of tissue valves with the kidney-shaped stent-ring containing a large anterior andthrombogenicity of mechanical valves. Therefore, years small posterior cusp, thereby mimicking mitral ﬂowafter the ﬁrst ﬂexible leaﬂet polymeric heart valves were conditions  (Fig. 8c). Overall, design and materialimplanted into patients [2,136], they have yet to reach a improvements led to a gradually increased durabilityperformance level which makes them clinically acceptable (up to 900 million cycles) and fatigue resistance nowbeyond the short-term use in artiﬁcial hearts (e.g. equals that of bioprostheses . There still seems to beAbiomed, Berlin-Heart, Medos ). high variability in cycle life not only between different valve designs and different studies [137,146,148,150]5.1. Valve design but also between similar valves within one and the same study . Similar to synthetic vascular grafts, material and designaspects stood in the foreground during the 1960s and 1970s 5.2. Leaﬂet materialswhile biocompatibility was added as a third pillar from the1980s onwards. Initially, most researchers used the basic Since 1959 [137,138], the overwhelming majority oftrileaﬂet (Fig. 8a) aortic design for their prototype valves polymer valves were made of polyurethanes (PUs). These[138–141]. Advances in ﬂuid-dynamic measurements and user-friendly materials have continuously been favoured
ARTICLE IN PRESS396 P. Zilla et al. / Biomaterials 29 (2008) 385–406Fig. 8. Polymeric heart valves: (a) a frame machined from polyetheretherketone (PEEK) and coated with a thin layer of leaﬂet polyurethane. Leaﬂets of acommercially available polyetherurethane suitable for animal implantation (Estane 58315, BF Goodrich, Westerlo-Oevel, Belgium) were dip-coated ontothe frame. This valve design has achieved durabilities in excess of 400 million cycles (10.5 years) during in vitro fatigue testing ; (b) and (c):polycarbonate urethane (PCU) tri-leaﬂet and bi-leaﬂet valves intended for the aortic and mitral positions. These particular designs achieved in vitrodurabilities of up to 600 million (15.8 years) and 1 billion (26 years) cycles, respectively .for blood contacting applications in spite of their initially 5.3. In vivo performancepoor long-term chemical stability . However, new-generation materials such as Elasteon  seem to have Initial results of the 1960’s with synthetic ﬂexible leaﬂetlargely overcome the problem of bio-degradation. Simi- valves were catastrophic [136,163]. During the subsequentlarly, improved fabrication technologies [145,154] resulted decade improvements were only moderate  or short-in even and constant leaﬂet thickness thereby improving lived. An example of the latter was the re-discovery offatigue resistance and durability [151,154,155]. Segmented ePTFE for a tricuspid heart valve prosthesis . AfterPUs are generally ﬂexible and durable at approx 100 mm promising initial results, its leaﬂets were soon found to beleaﬂet thickness, although there was some concern that a prone to stiffening, free-edge inversion and calciﬁcationleaﬂet thickness of less than 150 mm may not be sufﬁcient  conﬁrming a previous clinical trial that had led to a. Overall, the choice of thickness and material high mortality rate, associated with leaﬂet rupture andmodulus is a trade-off between tensile strength and thickening almost 15 years earlier .imparted bending stresses. The difﬁculty of this choice is Until the mid to late 1980s, calciﬁcation [142,164,165]reﬂected in the broad range of leaﬂet thicknesses had almost inevitably led to leaﬂet immobilization, rupture(60–400 mm) reported [137,147,148,150,151,155]. and perforation . This seemingly insurmountable The second most widely used polymer family was limitation brieﬂy resulted in an almost complete disconti-Silicone rubbers . Both the ball of the ﬁrst successful nuation of polymer valve programs. Subsequent materialmechanical valve and the leaﬂets of one of the ﬁrst improvements during the 1990s [146,151,167] allowed theclinically implanted polymeric heart valves  were re-commencement of some of these programs culminatingmade from this material. Disappointing results [136,157], in a mitral implant study  where no valve relatedhowever, led to silicones eventually being abandoned. deaths occurred. Furthermore, bioprosthetic anticalciﬁca- In general, most of the biomaterials of the 1960s and tion treatments of the 1980s and early 1990s were1970s were either tried for vascular grafts or synthetic heart incorporated in an attempt to modify the leaﬂet surfacesvalves. Expanded Teﬂon was one of the materials which towards more ‘biocompatibility’. Covalently bound bipho-was used for both applications [139,158,159]. Most sphonate [168,169], the extraction of low molecular weightrecently, a novel polyoleﬁn, poly(styrene-b-isobutylene- components  from methanol-extracted polyetherb-styrene) (SIBS), with excellent chemical stability, has urethane (PEU) and the modiﬁcation of a polyurethanebeen proposed as alternative to the traditional polymers valve with polyethylene oxide and sulfonate groups[160,161]. The material was found to match PU in platelet (PU-PEO-SO3) all showed reduced polymer calciﬁcation.deposition tests, and was comparable with PU in terms of The latter also showed a decrease in thrombogenicity andtensile and fatigue properties, albeit only after reinforce- crack formation . Similarly, heparin, taurine andment with polypropylene ﬁbres [160,161]. Altogether, the aminosilane modiﬁcations equally resulted in extendedpromising result with both the latest generation of durability . Yet, even the latest generations of polymerbiostable urethanes  and innovative new materials valves continue to experience some degree of extrinsicsuch as ﬁbre-reinforced SIBS  for the ﬁrst time calciﬁcation [137,148,150] indicating that the goal of aindicate that after 50 years of failure a new door may long-lasting synthetic heart valve may remain elusive aseventually have opened a small gap. long as no cellular components such as endothelial cells are
ARTICLE IN PRESS P. Zilla et al. / Biomaterials 29 (2008) 385–406 397an integral part of the concept. The ﬁrst successful attempts up requirements of future clinical trials with tissue-to grow endothelial cells on polyurethanes and PU-silicone engineered valves (Fig. 9).rubber co-polymers were made in the 1980s [172–174]. While the majority of researchers pursue a conceptA promising current approach to endothelialization in- whereby a scaffold is seeded in vitro with autologous cells,volved the pre-seeding of ECs onto cholesterol-modiﬁed an alternative approach aims at the in vivo recruitment ofpolyurethane cusps. This modiﬁcation resulted in increased cells  through the incorporation of biological signalscollagen synthesis and cell retention in vitro and into a degradable scaffold. The latter is supported by thein vivo [175,176]. observation that circulating endothelial progenitor cells (ECP) led to the spontaneous endothelialization of the5.4. Way forward for polymer valves blood surface of clinically implanted ventricular assist devices [185,186]. Similarly, ECPs have been shown to Over the course of half a century, polymeric valve home to stents coated with CD 34 antibodies (CD34 beingdesigns have been improved regarding ﬂow characteristics cell surface molecules of ECPs) .and stress reduction. During the same period of time, Although scientiﬁcally representing a quantum-leapdegradation-resistant materials were developed, processing improvement to today’s heart valve prostheses, tissue-methods evolved that allow reproducible high-quality engineered heart valves still have a long way to go towardsmanufacturing, and thrombogenicity and calciﬁcation have competing with contemporary heart valve prosthesesbeen partly addressed through surface modiﬁcations. Yet, regarding safety, functionality, logistic feasibility andwithout the integration of living cells into these con- ﬁnancial viability. A clearly unmet medical need andstructs—either through blood borne fallout healing  or indication, however, is in paediatric applications due to theactive incorporation through seeding—it seems unlikely inherent growth-potential of tissue-engineered valves.that the full potential of polymeric valves will ever come to Furthermore, recent studies have demonstrated the feasi-fruition. bility of using prenatal foetal cells—which can be obtained during pregnancy and used for the tissue-engineering6. Tissue-engineered valves procedure prior to birth—to provide living, autologous heart valves for early correction of congenital heart defects Tissue-engineering technologies providing living auto- [178,188] (Fig. 10). Therefore, should tissue-engineeringlogous heart valves with the capacity of regeneration and technologies become an accepted clinical tool, it willgrowth have shown promising experimental results and most likely happen through the ‘‘niche’’ of paediatricinitial human applications have been reported [178–181]. applications.Resorption of the porous leaﬂet scaffold initially led toovershooting ﬁbrosis [182,183] but ﬁne-tuning of the 7. Catheter-based valvesresorption process through the use of poly-glycolic-acid(PGA)/poly-4-hydroxybutyrate (P4HB) instead of poly- As attractive as the concept of a catheter-delivered heartlactic acid (PLA) co-polymers led to living heart valves valve is, the emperor’s new clothes syndrome remains:with a tri-layered structure and many features of a native while it potentially reduces the implantation trauma, itleaﬂet . The ability to visualize the vitality and cellular perpetuates the unsatisfying situation regarding availableactivity of heart valve cells in vivo is a backbone of follow- prostheses. By using mainly conventional bioprostheticFig. 9. Novel multimodality molecular imaging techniques enable the monitoring of cell activation and remodelling enzymes during valve developmentand disease. The long-axis view (a) shows the aortic root and arch, followed by a short axis view (b) which shows negative signal enhancement (darkening)caused by the uptake of nanoparticles by activated cells. Colour-coded signal intensities (c) show focused uptake of the nanoparticles in the commissures(modiﬁed from  with permission/mouse model/VCAM-1).
ARTICLE IN PRESS398 P. Zilla et al. / Biomaterials 29 (2008) 385–406 tube at a fraction (approximately a third) of their expanded diameter, and then be able to unfold into the precise geometries required for function and durability. Although ball-and-cage  and tilting disk concepts  were proposed and evaluated, ﬂexible-leaﬂet designs are prob- ably the most suitable candidates to fulﬁll the required criteria, and it follows that transcatheter valve leaﬂets are preferentially fabricated from BPT or ﬂexible polymers used in more conventional designs. Although the ﬁrst type (BPT) suffers from degenerative ﬂexural failure, 50 years of research have yet to produce a valve of the second type (polymeric) that approximates its success (as seen in sections above). Hyperelastic metal leaﬂets comprise a third option, but relatively little is currently known about the suitability of this emerging technology for valve applications. 7.2. For few of the few? If there was ever a development that was tailor-made for the millions of patients in threshold countries who need aFig. 10. Ideally, an autologous cardiovascular substitute would be used valve replacement, it was the catheter-based delivery offor the repair of congenital cardiovascular malformations which has the heart valve prostheses. By offering patients with earlypotential to grow and regenerate, thereby avoiding secondary damage to regurgitation who have no access to open heart surgery butthe immature heart and re-operations with their associated mortality and access to a modern C-arm visualization system a tool formorbidity over a life-time. If used at or shortly after birth, the process of the placement of a functioning valve and thereby prevent-engineering such an autologous construct would need to be initiatedprenatally. Conceptually, foetal stem cells could be harvested from extra- ing ventricular dilatation, morbidity patterns of entireembryonic foetal tissues such as chorionic villi, the umbilical cord or populations could change. Yet, the available bio-materialsamniotic ﬂuid. After differentiation and proliferation in vitro, cells could are so unsuitable for these patients that catheter-basedbe seeded onto a biodegradable scaffold and conditioned in a bioreactor, heart valves turned from a promise for millions into themimicking physiological conditions. The feasibility to fabricate foetal most exclusive ﬁrst world prostheses of all. Even worse,tissues according to this concept has been recently demonstrated by thein vitro fabrication of autologous living heart valve tissues based on contemporary catheter-based concepts cater for two ﬁrst-prenatal progenitor cells derived from umbilical cord , chorionic villi world fringe-indications: severely diseased patients who are and amniotic ﬂuid . inoperable due to co-morbidities  and pulmonary placements in patients with congenital heart disease in order to obviate or delay the comorbidity associated withmaterial, delivery-associated hype is not matched by re-operation [192,193]. Even if indications may eventuallyprognostic expectations. This is particularly true for the be extended to the majority of ﬁrst world patients, thelargely young patients in developing countries, whose non- market is small. Given this small potential market, it iscalciﬁed rheumatic valves would theoretically make them astonishing how all the major players rushed to the scene.ideal recipients for endovascularly placed prostheses. Since the ﬁrst transcatheter implants in animals occurred inGiven the rapid degeneration of all contemporary ﬂexible 1992, with Andersen et al. and Pavcnik et al. delivering aleaﬂet valves in young recipients, however, the very patient stented porcine valve and a ball-and-cage valve (withgroup that would dramatically beneﬁt from transcatheter inﬂatable ball), respectively [189,194], there are reported todelivery is the one that hardly qualiﬁes for it. be more than 20 companies currently working on this technology . A Pubmed search for ‘‘percutaneous7.1. Principle heart valve’’ receives over 1500 hits, which include more than 150 reviews! The concept behind catheter-based valve delivery is Since the ﬁrst human catheter-based replacement of asimple: a valved stent is collapsed into a catheter, the pulmonary valve in pediatric patients by Bonhoeffer et al.catheter tip is positioned at the site of the valve to be  and an aortic valve by Cribier et al. , eventsreplaced, and the stent is balloon-expanded to unfurl the follow hot on the heels of one another. Bonhoeffer hasnew valve. The execution of the concept, however, is very since treated more than 120 patients with pulmonarycomplex. In addition to the strenuous demands already insufﬁciencies using a porcine jugular vein valve in aplaced on surgically implanted valves, and to the ingenious platinum stent  (now the Medtronic Melodys valve)stent and catheter designs required to deliver them, (Fig. 11), while the Edwards Sapiens (equine pericardialcatheter-based valves must withstand being forced into a tricuspid valve in a stainless steel stent)  (Fig. 11), and
ARTICLE IN PRESS P. Zilla et al. / Biomaterials 29 (2008) 385–406 399 from Vancouver performed a hybrid procedure utilized a transapical access to deploy the valve antegradely under radiographic control . There were no intraprocedural mortalities or morbidities but one patient died on day 12 of pneumonia and 2 died later on days 51 and 85, respectively of non-cardiac causes; the remaining 4 have done well and completed 6-month follow-up—only one has more than trivial or mild aortic incompetence and none have had valve-related complications. A larger series of 30 very high risk patients (average age 8275.1 years) from Leipzig using a similar method of transapically introduced valves wasFig. 11. Two examples of percutaneously delivered heart valves, theEdwards Lifesciences ‘‘Sapien’’ aortic valve prosthesis (left) and the more favourable with successful deployment of the deviceMedtronic ‘‘Melody’’ transcatheter pulmonary valve prosthesis (right). in 29 of the 30 patients, experiencing 3 deaths .Although both models use unconventional animal sources (equine Extracorporeal circulatory support was only used in thepericardial for the Edwards valve and bovine jugular for the Medtronic initial 43% patients and thereafter the procedures werevalve), the debut of two major commercial players highlights how successfully performed without circulatory support. Over-‘conventional’ the actual valves are—even if the delivery is catheterbased. Reprinted from  with permission of Universal Medical Press, all, this initial experience conﬁrms transcatheter placementInc. of heart valves as a promising approach for fringe groups.Corvalves (bovine pericardial valve in a Nitinol (NiTi) 7.3. Design/procedural challengesself-expanding stent) [76,199,200] are arguably the mostcommonly used aortic prostheses. Other valves include the In addition to securing seating—in ﬁrst world patients3F (Enables and Entratas) (bioprosthetic leaﬂets in self- mostly on top of heavy calcium deposits—and possibleexpanding NiTi stents) and Sadra Medical Lotuss interference with other cardiac structures, access and valve(pericardium/NiTi) aortic valves, the Shelhighs (porcine placement are important considerations in valve develop-pulmonic/NiTi) pulmonary valve, the Aortexs valve, and ment . In the pulmonary position, access is typicallythe Palmaz-Baileys valve composed entirely of Nitinol gained via the femoral vein , although transventricular. A ﬁrst assessment of the clinical experience with more pulmonary placement  has been advised in senescentthan 120 transvenous placements of pulmonary valves patients when the RVOT is large and the valve and catheteryears after correction of congenital malformations, was size precludes traditional percutaneous placement.favourable in comparison to surgical pulmonary valve In the aortic position, the initial antegrade transseptalreplacement in a historical cohort from the same institution procedure via the femoral vein  was technically. The median hospital stay was 2 days (compared to 7 demanding encountering difﬁculties in accurately seatingin the surgical group). There was no mortality and the early and deploying the device via this tortuous route (via anmorbidity was 5.8% (vs. 8.5% in the surgical group). atrial septostomy, through the mitral valve and having toPercutaneous aortic valve replacement in humans was ﬁrst negotiate an acute curvature in the left ventricle). There-performed as a femoral transvenous, transseptal procedure fore, the transseptal approach was subsequently aban-with antegrade access to the aortic annulus . In the doned for retrograde arterial placement [202,207]. Thisinitial attempt in 26 high-risk patients (turned down for procedure was also later extended to mitral valvesconventional aortic valve surgery due to high risk), 22 [208,209]. Although better tolerated by patients, it had itsdeployments were successful and 4 failed (2 patients could own challenges—especially in older patients—in requiringnot tolerate the guidewire across the mitral valve and in 2 a large caliber femoral artery capable of receiving a 24Fthe valve migrated). Complications were high with 27% (8 mm) catheter (commonly used for currently preferredearly (o30days) and 41% late deaths. The survivors have 26 mm valves). Moreover, the long catheter route alsoreturned to a normal life with mean transvalvular gradients made accurate deployment of the device difﬁcult. This ledbeing 11 mmHg and paravalvular leak grade 0–1 in seven Ye et al.  from Vancouver to utilize a similar catheterand grade 2 in four patients. Subsequently, Webb et al. mounted valve to implant an aortic valve via the apex of from Vancouver reported a femoral transarterial the left ventricle antegradely via a small left anteriorprocedure with retrograde access to the aortic annulus. thoracotomy and without the use of cardiopulmonaryTheir initial results in 18 patients showed successful bypass. Although not strictly complying with the conven-deployment in 14 (the 2 failures were due to iliac arterial tional interpretation of ‘endovascular’ placement, transa-injury in two and prosthetic valve embolization in another pical placement via endoscopy or small thoracotomytwo patients). There were 2 post-procedural deaths and one [203,210,211] allows the catheter-mounted valve conduitsurgical aortic valve replacement for failure—in the 14 to be introduced via the left ventricular apex more directly.patients with successful deployment paravalvular leakage The shorter, more rigid and larger deployment systemvaried from none to moderate. Because of the technical allows more accurate and secure prosthetic valve deploy-difﬁculty in deploying the device, Ye, Webb and colleagues ment in the aortic annulus. This is done via radiographic