This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Vesely, I. Search for Related Content PubMed PubMed Citation Articles by Vesely, I. Pubmed/NCBI databases Substance via MeSH Medline Plus Health Information Heart Valve Diseases Related Collections CV surgery: valvular disease Pediatric and congenital heart disease, including cardiovascular surgery

(Circulation Research. 2005;97:743.)
© 2005 American Heart Association, Inc.

Review

Heart Valve Tissue Engineering

Ivan Vesely

From The Saban Research Institute of Children’s Hospital Los Angeles, Keck School of Medicine, University of Southern California.

Correspondence to Ivan Vesely, PhD, The Saban Research Institute of Children’s Hospital Los Angeles, Keck School of Medicine, University of Southern California, 4650 Sunset Blvd–MS137, Los Angeles, CA 90027. E-mail ivesely{at}chla.usc.edu

This Review is part of a thematic series on Cardiovascular Tissue Engineering, which includes the following articles:

Custom Design of the Cardiac Microenvironment With Biomaterials

Heart Valve Tissue Engineering

Engineering a Small-Diameter Artificial Artery

Engineering Myocardial Tissue

Regenerative Cardiomyocytes for Cardiovascular Tissue Engineering
Richard T. Lee Guest Editor

	Abstract

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 
Tissue-engineered heart valves have been proposed by physicians and scientists alike to be the ultimate solution for treating valvular heart disease. Rather than replacing a diseased or defective native valve with a mechanical or animal tissue–derived artificial valve, a tissue-engineered valve would be a living organ, able to respond to growth and physiological forces in the same way that the native aortic valve does. Two main approaches have been attempted over the past 10 to 15 years: regeneration and repopulation. Regeneration involves the implantation of a resorbable matrix that is expected to remodel in vivo and yield a functional valve composed of the cells and connective tissue proteins of the patient. Repopulation involves implanting a whole porcine aortic valve that has been previously cleaned of all pig cells, leaving an intact, mechanically sound connective tissue matrix. The cells of the patients are expected to repopulate and revitalize the acellular matrix, creating living tissue that already has the complex microstructure necessary for proper function and durability. Regrettably, neither of the 2 approaches has fared well in animal experiments, and the only clinical experience with tissue-engineered valves resulted in a number of early failures and patient death. This article reviews the technological details of the 2 main approaches, their rationale, their strengths and weaknesses, and the likely mechanisms for their failure. Alternative approaches to valvular tissue engineering, as well as the role of industry in shaping this field in the future, are also reviewed.

Key Words: cardiac valves • tissue engineering • review • acellular matrix • scaffold

	Introduction

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 
Every year, more than 100 000 US patients need to have their dysfunctional or diseased valves replaced with a prosthetic valve. Where there is a need, there is a technological solution. The heart valve industry is the US is vibrant and healthy, enjoying a growth in the market of 5% per year, selling roughly 300 000 valves worldwide.¹ Worldwide sales were $910 million in 2002 and are most likely past the $1 billion mark in 2005. Faced with such tremendous market opportunities, many companies, clinicians, and scientists alike have taken serious interests in developing a new type of heart valve that can potentially revolutionize the industry and the practice of medicine. A tissue-engineered valve promises to be a living implant with a potential to grow and last a lifetime, like most native valves do. Rather than being a device that palliates a disease, it promises to be curative: a living replacement for a diseased component of our physiology.

As reviewed here, however, tissue engineering has promised much more than it has delivered. Indeed, heart valve tissue engineering has been hyped to such an extent that there is roughly 1 review article for every 4 true research reports published in this field. A simple PubMed search on "heart valve tissue engineering" reveals 164 citations since 1995, 32 of which are actually review articles.^{2– 33} This article, part of a series of reviews on tissue engineering in Circulation Research, intends to examine what has been accomplished and what is realistically possible in the coming years. Patients and physicians alike are beginning to ask "When can I get a tissue-engineered heart valve?" Indeed, even the American Heart Association Web site has an article entitled "Tissue-engineered valves give diseased hearts new life,"³⁴ suggesting that clinical use of tissue-engineered valves is around the corner. The American Heart Association document fails to report the clinical outcomes of the cited experiment: that the technology has largely failed and that the claims of the surgeon have been discredited. This review thus aims to present a much more critical report on the real progress in this field, pointing out the specific mechanism by which tissue-engineered heart valves have failed in clinical and animal experiments.

As a discipline of its own, tissue engineering is surprisingly old. The term "tissue engineering" was coined by Fung in October 1987 at a National Science Foundation workshop in Washington, DC.³⁵ The different approaches to tissue engineering appeared to originate independently as early as the 1960s, when advanced tissue culture technologies were used to propagate skins cells.³⁶ In the mid-1970s, the work of Rheinwald and Green at the Massachusetts Institute of Technology set the stage for skin grafting with sheets of cultured, autologous keratinocytes.³⁷ Tissue engineering has had a number of definitions, both simple and complex. One of the simplest is that found in The Biomedical Engineering Handbook³⁸ (D. Williams, personal communication, 2005) and is stated as follows:

The application of scientific principles to the design, construction, modification, growth and maintenance of living tissue.

A more complex definition is given in a World Technology Panel Report³⁹ funded jointly by the NIH, the Food and Drug Administration (FDA), and other government agencies, in which tissue engineering was defined as:

The application of principles and methods of engineering and life sciences to obtain a fundamental understanding of structure-function relationships in novel and pathological mammalian tissues and the development of biological substitutes to restore, maintain or improve tissue function.

For the purposes of this review, tissue engineering relevant to heart valves is defined as the manipulation of biological molecules and cells for the purpose of creating new structures capable of metabolic activity.

A new type of heart valve thus fabricated must, therefore, contain material of biological origin in a configuration that did not emerge naturally. Heart valves containing trace amounts of biological material, such as pannus overgrown, are not considered tissue-engineered valves. Valves consisting of inert materials covered with a cellular coating, developed for the purpose of improving the performance of the valve, could be considered tissue-engineered devices. Before exploring the universe of tissue-engineered valves, it is best to first review the spectrum of existing prosthetic valves based on biological materials.

	Bioprosthetic Valves

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 
There are generally three types of bioprosthetic valves available commercially: (1) porcine xenograft valves, (2) bovine pericardial valves, and (3) allograft or homograft valves (Figure 1). The porcine xenograft valve consists of an intact pig aortic valve that is preserved in low-concentration glutaraldehyde solution.⁴⁰ These valves are prepared by valve manufacturers in various configurations, such as with or without integrated sewing cuffs, to maximize either ease of implantation or effective orifice area. Occasionally, these valves are assembled from up to 3 separate segments of aorta and the associated cusp material in an effort to improve valve symmetry and hence perceived performance.⁴¹ The bovine pericardial valve is fabricated from up to 3 separate pieces of glutaraldehyde-treated calf pericardium, affixed to a supporting stent and sewing cuff, in a configuration very similar to that of the porcine xenograft. Both the porcine and bovine valve tissues are crosslinked in low concentrations of glutaraldehyde to reduce their antigenicity and to stabilize the tissue against the proteolytic degradation that would otherwise occur following implantation into the recipient. Both types of valve tissues are also treated with various other chemical agents to minimize their propensity to calcify over the duration of implantation and hence improve their longevity.⁴² The homograft valves are intact human valves obtained from organ and tissue donors, usually stored cryopreserved as entire aortic or pulmonary roots, and trimmed to size and shape before implantation in the recipient.^43,44 These 3 types of valves are used primarily in the aortic and pulmonary positions and are occasionally inverted and used to replace the mitral valves. Mitral valve repair is used more frequently than mitral replacement during the first surgery to address a dysfunctional mitral valve. In addition to these 3 "device-related" approaches, there are also completely surgical approaches to reconstruct valves, making use of autologous or commercially available bovine pericardium to augment defective cusps or fabricate monocusp valves.⁴⁵ These highly varied approaches are used primarily in children who do not tolerate prosthetic devices nearly as well as adults do. The Ross procedure, the surgical removal of the autologous pulmonary valve and its reimplantation in the aortic position, is a hybrid approach that makes use of surgical reconstruction of the aortic valve and a replacement of the missing pulmonary valve with a cadaveric homograft.⁴⁶ Although these surgical approaches often place biological tissues in new configurations, they are not considered tissue engineering, as there is little manipulation to the internal molecular structure for the purpose of enhancing their biological performance.

View larger version (64K): [in this window] [in a new window]   Figure 1. Images of a porcine bioprosthetic valve xenograft (A), bovine pericardial valve (B), and a human aortic valve allograft (C), also called a homograft. Both the porcine and bovine valves are treated with glutaraldehyde before implantation. The homograft is stored frozen and implanted without any other chemical preparation and often without any tissue type matching.

	Why the Need for a Tissue-Engineered Valve?

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 
Some may argue that there is not much of a need for tissue-engineered valves, because conventional valve technology is very mature with well-described performance criteria. Indeed, the original motivation for the continued development of prosthetic valve technologies appears to have lessened considerably with the improvements in surgical approaches and hence surgical outcomes. In the early days of bioprosthetic valve development, durability and hence longevity, was the main motivating factor. The reason why durability was the main object of research and development is that prosthetic valves are meant to be implanted once and should last the life of the patient. Historically, the main complication associated with the use of prosthetic valves has been operative mortality during prosthetic valve replacement surgery. Whereas the mortality during the first surgery—the surgery to replace the disease native valve—is often less than 1%, the second surgery is considerably more risky, with repeat operations having mortality as high as 20%, varying highly across institutions.^47,48 With improvements in surgical technique and technologies, reoperative mortality has been reduced considerably during the past decade. Many surgeons now view surgical mortality during reoperation to be lower than the cumulative risk of thromboembolism associated with the use of mechanical valves (approximately 4% per patient-year^49,50) and now opt for repeated use of bioprosthetic valves in their patients. Indeed, the durability of the Edwards pericardial valve, perceived by many to be the most durable bioprosthesis, is close to 20 years^51,52 and almost equivalent to the aortic valve homograft which, because of its excellent longevity, is considered by many to be the gold standard.⁴³

Because of the relatively good performance of current generation prosthetic valves, and the excellent quality of life they provide, both surgeons and valve manufacturers have become quite conservative. Surgeons are often reluctant to switch from a proven valve to a new device, and manufacturers are wary of introducing a new product that will have clinical performance that is worse than its current product. A number of clinical failures of new products have all but stifled innovation in conventional heart valve technologies. The failure of the Carbomedics Photofix- valve (cusp abrasion and perforation),⁵³ the Medtronic Parallel mechanical valve (thromboembolism),⁵⁴ and the St Jude Silzone coating (tissue necrosis and perivalvular leak)^55,56 have made the valve industry highly aware of the possibility that innovation can breed disaster. Most recently, Edwards Lifesciences suspended the clinical trial of its catheter-deployable valve⁵⁷ because of problems with delivery and anchoring of the valve in the aorta of the patient.⁵⁸

In the world of conventional valve technologies, the bar for the adaptation of a new valve is very high. Current bioprosthetic valves have a lifespan of 15 to 20 years,⁵⁹ highly predictable failure patterns⁶⁰ that can be managed, and negligible early complications. A tissue-engineered valve will, therefore, need to compete in this arena, where change is slow and methodical, often taking a generation to establish. Indeed, the Edwards valve has increased its market share steadily over the past 20 years, primarily because of clinical reports on its very good long-term performance.⁵¹

There may not be any immediate need for a tissue-engineered valve in the adult patient. Given the good service life that conventional valves offer, experimental use of new tissue-engineered valve concepts is questionable. The more realistic option for the use of tissue-engineered valves in the near future is in the pediatric population. The performance of the many surgical corrections for valvular defects is highly variable and depends on the age of the child.⁶¹ Allograft valves are difficult to obtain for children because they require the death of other children of similar size. Surgical reconstruction and monocusp valves tend to fibrose and contract early.⁶² In the child, there is a need for new materials and new approaches and thus an opportunity for tissue-engineered valves. Unfortunately, the market size for pediatric products is very small, less than 10% of the adult valve market, and thus not commercially viable. Accordingly, established heart valve manufacturers expend relatively few resources developing tissue-engineered valve technologies (neither Medtronic nor Edwards Lifesciences has a significant research program in heart valve tissue engineering; Medtronic and Edwards Lifesciences, personal communications, 2005). Clinical use of tissue-engineered valves will thus most likely happen first in pediatric hospitals on an ad hoc basis and will augment the portfolio of surgical options currently available to treat these very challenging patients.

	Three Approaches to Valvular Tissue Engineering

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 
Perhaps the first examples of heart valve tissue engineering were the experiments with seeding cells on otherwise inert substrates. This approach has been relatively successful in vascular grafts⁶⁴ but not at all for cardiac valves. Compliant synthetic materials have failed as valve leaflet substitutes, not because of thromboembolism but because of material fatigue, often related to microcracks, plasma protein insudation, and subsequent mineralization.⁶⁵ Because of these fundamental material issues, glutaraldehyde-preserved biological matrixes have dominated the valve field. Because of the residual toxicity of glutaraldehyde, glutaraldehyde-treated valve tissues have remain cell free. For reasons not completely clear, they do not induce thromboembolism and remain an essentially passivated structured in the blood stream, which does not elicit any adverse reaction.

Perhaps the first example of heart valve tissue engineering came out of the University of Vienna in 1991, when Grimm et al presented their success in inducing endothelium to grow on glutaraldehyde-fixed bovine pericardium.⁶⁶ Sustaining cells on the otherwise toxic pericardial tissue was made possible by inactivation of the aldehyde by crosslinking with L-glutamic acid before cell seeding. For unknown reasons, this technology was never implemented clinically and the last article on this topic was published in 1993.⁶⁷

The approaches that have been sustained by investigators around the world can be grouped into the following two areas: (1) decellularization of xenogenic tissues followed by cell seeding or direct implantation and (2) use of bioresorbable synthetic scaffolds. A third, less popular approach, involves fabrication of cell matrix constructs by way of polymerization and cell entrapment.

Acellular Matrix Xenograft
This is perhaps the oldest approach to mainstream valvular tissue engineering, the first reports being patents by Brendel and Duhamel of the University of Arizona, Tucson, filed in 1984,⁶⁸ and by Klement et al⁶⁹ from Toronto, filed in 1987. Both of these approaches claimed the much larger field of acellular matrix use for applications ranging from valves to vascular grafts to bone, teeth, ligament, and skin. Decellularization of heart valves has since been attempted all over the world. In the US and Canada, this approach was attempted by Vesely and Noseworthy⁷⁰ and Wilson and colleagues⁷¹ very early on in this field and more recently by the thorough investigation of Hilbert et al.⁷² In Europe, this approach has been used more widely by Dohmen et al of Berlin, Germany,^73,74 Steinhoff and colleagues of Rostok, Germany,⁷⁵ Stock and colleagues of Jena, Germany,⁷⁶ Haverich and colleagues from Hanover, Germany,⁷⁷ Weigel et al from Vienna, Austria,⁷⁸ Gittenberger-de Groot and colleagues from Leiden, The Netherlands,⁷⁹ Fisher and colleagues of Leeds, UK,⁸⁰ and Spina, Gerosa, and colleagues from Padua, Italy.⁸¹ In Asia, the approach has been adopted by Hong and colleagues⁸² and Ye et al⁸³ from Shanghai, China, and by Wu et al from Beijing, China.⁸⁴

The rationale for this approach is the assumption that the antigenicity of xenogenic tissues originates in the cellular debris. Recall that porcine aortic valves need to be crosslinked with glutaraldehyde before implantation and that human allograft valves do not. Clearly, the immunohistocompatability mismatch between humans and pigs is far more severe than it is between unmatched human subjects. Pig cells express the gal-^1,3 epitope, whereas humans do not.⁸⁵ Indeed, human allograft valves are almost never matched to recipients, even though matching has been shown to have better long-term graft survival.⁸⁶ Lack of matching is apparently not too detrimental because homograft valves can last 15 to 20 years. It has been theorized that perhaps the valve is in a privileged location, in a high flow environment where monocytes and other immune system cells cannot readily attach. With all of these issues in mind, suggesting that cell-extracted porcine valves can exist in human patients, without any appreciable antigenic response, may sound quite reasonable. This approach also assumes that these acellular matrixes will become repopulated with recipient cells, either before or immediately after implantation in the patient. A fully repopulated matrix would thus become "invisible" to the host immune system, because it would be enveloped in an endothelial cell layer recognized as "self," and possibly remodeled by the invading cells, ultimately becoming self as the host cells lay down a new matrix in place of the degraded porcine matrix. This approach has been developed to the greatest extent by Goldstein et al⁸⁷ of CryoLife Inc, a company well known for its homograft valve cryopreservation business. CryoLife eventually brought the technology to clinical use in Europe,⁸⁸ unfortunately with disastrous results that are reviewed later in this article.

The typical approach to generating an acellular matrix tissue is to first break apart the cell membranes through lysis in hyper- and hypotonic solutions, followed by extraction with various detergents. The detergents used by most investigators include the anionic Sodium dodecyl sulfate, the zwitterionic CHAPS and CHAPSO, and the nonionic BigCHAP, Triton X-100, and Tween family of agents. The enzymes that have accompanied these detergent treatments have focused mainly on cleaving and removing the DNA that is part of the cellular debris. Because these enzymes can potentially degrade the useful matrix, enzyme inhibitors, such as trypsin inhibitor, can be used, although some have used trypsin-EDTA alone to decellularize the matrix.⁸⁹ It is important to note that the agents used for cell extraction can be quite detrimental to the matrix: they can degrade or denature the matrix proteins or leave toxic residues or residual charge, any of which can detrimentally affect mechanical function or cellular response. The parameters that the many investigators in this field have varied involve mainly the sequence of steps, the specific detergents to use, and the time duration of the various soaking periods. For example, the laboratory of Fisher in Leeds, UK,^80,90 uses a series of baths in PBS and hypotonic buffer, along with trypsin and nucleases, to decellularize the tissue. The entire extraction procedure can last up to a week. Mechanical testing is used most of the time to determine the mechanical integrity of the processed tissues. Parameters such as burst pressure or failure strength of test trips is compared with unprocessed controls.^80,91 In most cases, mechanical properties remain well preserved after cell extraction. Histological morphology, however, varies greatly from process to process, often showing a highly porous, locally collapsed microstructure, suggesting that there are features of the matrix that are not readily measurable using conventional techniques. The patents that describe the creation of these acellular matrixes are perhaps the largest collection in this field and number in the dozens, if not hundreds.

To seed or not to seed cells before implantation has become 1 of the important variables to consider in this approach. Perhaps the most disconcerting issues in this approach are reports that homograft valves not only fail to repopulate with recipient cells but actually become completely acellular within months of implantation.⁹² This is of particular concern as the human aortic valve is expected to be the ideal matrix for repopulation: it has the right mechanical properties and is far less antigenic than porcine matrixes. Possibly in view of these pathological findings, or possibly for other reasons, a number of investigators have attempted to repopulate these scaffolds in vitro, in advance of implantation into animal models.⁹³ The CryoLife technology,⁸⁷ however, was implanted into patients without any prior cell seeding in vitro. Indeed, technologies attractive to the valve industry are those that do not involve handling of patient cells. It is far easier to get a new valve approved by the FDA as a device, if there is no reliance on cells contributing to its long-term durability.

Bioresorbable Scaffold
This approach to valvular tissue engineering is perhaps the most conventional. The bioresorbable scaffold has been used in applications as varied as skin, bone, vessels, spinal chord, tendon, bladder, vagina, muscle, and solid organs like liver and pancreas. The concept is simple: cells of a particular phenotype are seeded on a porous material, implanted in the body, and are expected to generate the organ of interest as the scaffold degenerates. The scientific rationale for this approach is unclear and based primarily on empirical evidence that it appears to work in certain applications. Perhaps the oldest most successful application of this approach is tissue-engineered skin, dating back to the late 1970s.⁹⁴ This approach launched the tissue-engineering industry and remains the most, if not the only, successful product.³⁶

The original scaffolds for heart valves were borrowed from skin: polylactic and polyglycolic acid and copolymers thereof.⁹⁵ These materials have been largely abandoned for heart valve applications because they were too stiff, and newer, more compliant materials like polyhydroxyalkanoate, have been used.⁹⁶ Natural scaffolds, like small intestine submucosa have also been used for valvular matrixes.⁹⁷ In most cases, the valve candidates have been implanted in the pulmonary, rather than the aortic position, because the degrading scaffold cannot bear left ventricular pressures before new tissues being regenerated. Besides the option of scaffold material and shape of the leaflets, the other variable is the decision of whether to preseed in vitro or not and, specifically, how to do it.⁹³ As discussed above for the acellular matrix valve, preseeding in vitro appears to yield better in vivo results.

Collagen-Based Constructs Containing Entrapped Cells
This is perhaps the least explored area and the one that involves the greatest amount of engineering. This approach is based on the relatively old observations that cells entrapped in collagen gels contract and compact the gels, increasing the density of the collagen many fold.^98–100

The principle involves first mixing soluble, fibrillar collagen with the appropriate cells. After the collagen–cell mixture is neutralized, soluble collagen reassembles into fibrils and a gel is created. Cells become entrapped within the collagen gel and begin to interact with the collagen fibrils and contract the matrix, excluding water.^98,99,101 In many ways, this in vitro contraction mimics wound healing in vivo.^102,103 When the gel is mechanically constrained, the collagen fibrils align in the direction of constraint,¹⁰⁴ and a highly aligned, compacted collagenous construct can thus be fabricated.

One of the early applications of this technology has been as blood vessels,¹⁰⁵ where the fabrication of collagen tubes is relatively straightforward. All that is required is a rod affixed centrally within the lumen of a tube, creating an annular space that is filled with the collagen–cell mixture. Once the mixture gels, it begins to contract around the central rod and peels off the inner surface of the tube. At that point, the rod with its adherent gel coating can be removed from the tube. Because the gel coating on the rod cannot shrink in circumference, it compacts only through its thickness and along the length of the rod. Collagen fibrils thus align circumferentially, much like in the adventitia of a blood vessel.

A very logical extension of this approach to heart valves is to fabricate a mold for leaflets, contiguous with the aortic annulus. This is what has been done by the laboratory of Tranquillo et al.¹⁰⁶ Tranquillo developed an interesting mold with inner and outer parts that produced the shape of a bileaflet valve within a tube (Figure 2). Like the vessel wall, the valve leaflets also shrank along the direction of the tube, becoming shorter and developing an aligned fiber structure.¹⁰⁷

View larger version (54K): [in this window] [in a new window]   Figure 2. A, Image of the original mold by Tranquillo (patent 6,666,886). B, Photograph of real mold fabricated from Teflon. C, Image of bileaflet valve fabricated by casing a collagen or fibrin gel within this mold.

The work in the laboratory of this author is a simplification of the original work of Tranquillo. Rather than forming complex 3D structures, we have focused on generating simple, 1D strings that could be used for surgical reconstruction of mitral valve chordae or for future use in more complex constructs. We began by selecting neonatal rat aortic smooth muscle cells as the experimental cell line, because they are well known for producing considerable amounts of matrix, particularly elastin.¹⁰⁸ We found empirically, however, that cells with highly varied phenotype can be extracted from the minced aortic tissues. Digesting these tissues with trypsin apparently liberates a mixture of cells, many of which are not highly contractile and thus unsuitable for use with directed collagen gel shrinkage. The best outcomes were obtained with cells isolated by the outgrowth method, in which tissue is minced into small pieces and plated onto to dishes. The most highly motile cells apparently grow out from these tissue fragments and can be amplified over several weeks of culture. Optimal shrinkage of the constructs occurred when these cells were added to the collagen suspension at a cell-seeding concentration of 10⁶ cells/mL.¹⁰⁹ The collagen suspension consists of sterile acid-soluble type I collagen at an initial concentration of 2.0 mg/mL. For our application, the cell/collagen suspension is pipetted into rectangular wells of variable geometries with microporous holders at their ends. Like the inner mandrel of the tubular constructs, the anchors at the ends of the wells prevent longitudinal contraction and allow shrinkage to occur only transversely to the long axis of the wells. This gives rise to well-aligned collagen constructs with relatively high collagen fibril density (Figure 3).

View larger version (45K): [in this window] [in a new window]   Figure 3. Images of collagen constructs during the shrinkage process, showing rapid compaction within a few days and more gradual, yet continuing, compaction over the next 8 weeks of culture. Reproduced from Shi and Vesely109 with permission.

After 8 weeks of culture, the collagen constructs have the typical nonlinear stress/strain curve of tendinous materials, an extensibility of 14%, a stiffness of 5 MPa, and failure strength of 1.1 MPa. Although the stiffness and strength are still about an order of magnitude lower than what is required, our constructs are already 10 to 100 times stronger than similar collagen-based materials fabricated previously.^110,111 Ultrastructural analyses have shown that the main reason for the good strength of our constructs is the very high collagen fibril density. Because the constructs are relatively simple 1D collagen bundles, they compact from 2 directions, producing an area shrinkage ratio that is greater than 99%.

In an effort to improve the strength of these constructs further, we have explored different sizes and aspect ratios, different materials for the anchors for these constructs, and different forms of application of external tension. For example, triangular-shaped holders that appear to channel the tension from the holder material to the construct lead to stronger constructs, as does the application of external forces. Dynamic loading, in particular, increases construct strength by a factor of 3. Although these constructs are clearly not ready for human use, they are nearing use in animal models. One point of concern is the use of rat collagen and rat cells, and efforts are underway to translate this technology from the rat to the sheep model, making use of sheep collagen and sheep cells. The core technology, however, is also being used in a more ambitious approach to develop a composite aortic valve cusp. This is discussed below.

Other Substrates in Early Development
The main problems with using reconstituted collagen as a substrate for tissue engineering is the observation that cells entrapped in collagen gels rapidly enter apoptosis¹¹² and synthesize matrix metalloproteinases. Whereas strategies for overcoming these phenomena have been attempted (ie, mechanical loading), use of collagen alone has thus not been widely embraced. A number of investigators are exploring cell adhesion and phenotype on thin flat films of candidate materials, as a prelude to using the bulk material as a scaffold. For example, Giachelli and colleagues have explored the use of chitosan, an interesting material derived from crustacean shells and used for wound dressings, and found that mixtures of chitosan and collagen work better than either alone.¹¹³ Anseth and colleagues have used crosslinked polyvinylalcohol¹¹⁴ and hyaluronan as a substrate for valve tissue engineering.^115,116 Rothenburger and colleagues from Muenster, Germany, have used collagen films as scaffolds for tissue engineering^117,118 and found that myofibroblasts and endothelium cocultured produced significant amounts of collagen and structural proteoglycans. Fibrin is being considered an alternative to collagen in valvular tissue engineering by a number of groups,¹¹⁹ including that of Tranquillo.

	History of Each Approach: Successes and Failures

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 
Acellular Matrix Valve
Judging from the volume of literature, the acellular matrix valve has received the most interest. This is not surprising, because it seems to make the most sense from a biomechanical point of view. The aortic valve is only that: a mechanical structure designed to open and close with minimal pressure drop and reverse leakage.¹²⁰ What is unique to the aortic valve, compared with most other connective tissues, is that it does not require any metabolic activity or self-repair to provide a good service life. Recall that human aortic valve allografts (homografts) are transplanted without any tissue-type matching and consequently become acellular within a few months.⁹² The homograft can last 20 years, essentially as a dead piece of tissue, free from any mechanical reinforcement by crosslinking agents. Indeed, it is remarkable that the homograft valves last longer than the glutaraldehyde-fixed porcine xenografts; there must be something about the microstructure and composition of the native aortic valve that makes it very resistant to mechanical fatigue. Indeed, many have studied the structural basis for this remarkable durability of the aortic valve.^120–143 Although it still remains unknown exactly what features of the native valve tissue give it such remarkable durability, the importance of its internal complexity is being appreciated more and more. Most likely, the presence of interconnected sheets of collagen, layers and tubes of elastin, highly nonlinear mechanics, anisotropy, and viscoelasticity endow the valve tissue with is unique longevity. Not knowing exactly which features are important, the logical approach is to duplicate all of them in a prosthetic device. Indeed, the use of intact porcine aortic valves has spawned the whole bioprosthetic valve industry. Early in its history, engineering analyses clearly demonstrated that the porcine xenograft, with its inherent structural complexity, is theoretically better than the pericardial valve¹⁴⁴ and should thus last much longer. Interestingly, this conventional wisdom ultimately proved to be quite wrong: a particularly good design of the pericardial valve (the Carpentier-Edwards valve) is remarkably long lived^51,52 and appears to be more durable than glutaraldehyde-fixed porcine valves.

Now it could be argued, of course, that a glutaraldehyde-fixed aortic valve, mounted on a stent, is hobbled somehow and its theoretical advantage over the pericardial valve disappears. That could very well be the case, because some studies have shown certain disadvantages of glutaraldehyde-fixed aortic valve cusps relative to similarly prepared bovine pericardium.¹⁴⁵ The intact aortic valve may thus still have the upper hand when finally revitalized with freshly seeded cells. Regrettably, there have been few, if any successes, with this approach. Rather than happily repopulating the waiting matrix, the cells that come into contact with acellular valves have reacted badly and destroyed the intricate microstructure and its associated perfect mechanics. The reality of this, unfortunately, has come to light only recently. For almost 20 years, research groups around the world have independently tried various chemical processing approaches and published the successes in short-term animal models, but none led to any real clinical success.

The first reports that these protocols ultimately fail in long-term animal studies were only anecdotal at first, shared among scientists during personal discussion at the various heart valve conferences. No negative results were actually published until Hilbert, a well-recognized valve scientist working at the laboratories of the NIH, did the "ultimate" comparative study of various extraction protocols in a long-term sheep implantation study.¹⁴⁶ Hilbert copied the protocols of a number of investigators and implanted 2 sets of differently processed valves into the sheep model, as pulmonary artery interposition grafts. After 20 weeks of implantation, the valves were explanted and examined grossly and histopathologically. With some variation between treatment protocols, all valves underwent considerable tissue overgrowth and infiltration with inflammatory cells. There was also some evidence of aneurismal dilatation. The second such report came from the laboratory of Stock,¹⁴⁷ who reported similar complications with his valves, even though these valves were seeded with cells before implantation and were thus more "ready" for implantation than the valves prepared by Hilbert.

It could be argued, however, that the sheep model was destined for failure and that the acellular xenograft approach would work in the human condition. The reason for this argument is that it has become clear over the past decade that the sheep model generates an exuberant fibrotic response to valve implants. Valves that are implanted in sheep overgrow rapidly with fibrotic tissue, certainly much more than they do in humans.⁵³ Indeed, a painful lesson learned in this area was experienced by Sulzer Carbomedics, whose Photofix- pericardial valve developed severe abrasions to the leaflets against the sewing cuff during its clinical trials in the mid-1990s. The reason for the leaflet abrasion was attributable to a very specific design flaw, which was not anticipated.⁵³ Indeed, this flaw did not manifest in the preclinical sheep studies because the sheep rapidly covered the sewing cuff with pannus and thus protected the delicate valve leaflets from the more abrasive sewing cuff material. It could be thus argued that testing acellular matrix valves in the sheep model is a waste of time and destined to fail, because of this severe fibrotic response.

Perhaps this was the motivation behind the use of the CryoLife Synergraft in patients directly, without any reported long-term animal studies, something quite unusual for the heart valve field. The CryoLife approach was similar to the others: the removal of cellular antigens through dissolution and extraction.⁸⁷ Unlike the somewhat harsh chemistry of the original Brendel and Duhamer approach,⁶⁸ and most of those who tried to improve on it, CryoLife favored a more gentle approach that has not been revealed publicly but most likely involves multiple freeze-thaw cycles, cell lysis through variation in osmotic pressure, use of enzymes such as DNase, and prolonged extraction in aqueous solutions. The CryoLife product was thus expected to be the least affected and the most mechanically sound acellular matrix. In October of 2000, CryoLife Inc obtained CE Mark approval to sell the Synergraft product in Europe⁸⁸ to a limited group of patients who had few alternatives: the neonate with congenital valvular malformations.

Although the numbers are not widely reported, a number of patients in Vienna, Austria, received these valves. Within a few weeks to months, many of these children began to experience serious valvular complications.¹⁴⁸ Although some remained apparently unaffected, many valves became highly regurgitant, and some children died. CryoLife rapidly withdrew the product from the market and disclosed the negative clinical outcomes at the 2004 Florence Heart Valve meeting. In their 2003 article, Simon et al¹⁴⁸ reported that 1 patient (aged 7 years) died 7 days after implantation of the valve, 1 patient (aged 9 years) died 6 weeks after surgery, and 1 patient (age 2.5 years) died 1 year after surgery. Death resulted from various cardiac complications related to inflammation, valve rupture, and stenosis. One patient was reoperated on 2 days after the primary surgery, in view of what happened to the others. All valves showed severe inflammation, both inside and out, fibrosis, encapsulation, perforation, and deterioration of the leaflet tissues. Grossly, this reaction was not markedly different from that shown by Stock and colleagues¹⁴⁷ in the sheep model. The sheep model was thus not to blame; it was the matrix itself after all. Apparently, there is some abnormal signaling generated by the acellular matrix that induces the invading cells to remodel, fibrose, and contract what otherwise appears to be a perfect, mechanically sound valve matrix. The exact reasons for this remain unclear.

Although its first clinical use was disastrous, proponents of this approach remain undeterred. Apparently, another variant of this approach is doing well in Berlin, Germany, and is being implanted by Dohmen, Konertz, and colleagues^73,74 with 50 implants in patients for more than 2 years. In this series, 1 patient died postoperatively, 2 required reoperation for valve-related complications, and the rest of the patients appear to be doing fine. The published short-term results in adults (17 to 70 years; mean, 46 years) therefore appear to be good, and the surgeons are satisfied with this technology. Personal reports obtained at the Society for Heart Valve Disease Meeting in Vancouver in June of this year, however, indicate that a number of their patients have indeed died from valve-related complications. In their article,⁷³ the authors also disclose that Konertz and Dohmen are shareholders of AutoTissue GmbH, the company that produces these valves.

Resorbable Scaffold
The failure of this approach is much less dramatic, as apparently none of these scaffolds have been tried in patients. Interestingly, the main and still dominant reason for favoring the bioresorbable polylactic/polyglycolic acid (PLA/PGA) scaffold approach is the FDA approval of the materials for implantation in the US, as bioresorbable sutures. Although this may be true, it most likely does not lessen the regulatory burden of a device constructed from these materials. Valves are life-sustaining class III devices and must be taken through the full set of preclinical and clinical trials before market approval. Perhaps it is thought that as limited-use surgical materials, valves fabricated from these matrixes could be used in limited volumes by surgeons directly, as part of an investigative study in their hospital. Perhaps the attractiveness of this approach is the legacy of Robert Langer who pioneered the resorbable matrix approach in the artificial skin product and launched the tissue engineering industry. Perhaps the entire approach is flawed. Jeff Hubbel, a noted chemical engineer with expertise in developing materials with controlled release chemistry noted recently at a conference that biological matrixes do not degrade through hydrolysis, but rather through proteolysis. Accordingly, he has been engineering his matrixes to be cleaved by enzymes.¹⁴⁹ Hydrolysable matrixes therefore are clearly not biomimetic and their use as biodegradable substrates should thus give rise for concern. Matrixes cleavable by proteolytic enzymes, however, are also far from clinical reality and may indeed experience the same fate as the others.

The fate and the failure mechanism of the bioresorbable matrix approach to heart valves, unfortunately, is less clear than it is for the acellular xenograft. Like for the acellular matrix valves, there have been no clear reports of "failed experiments." Personal communications with the principal investigators, however, confirm that fibrosis, retraction, and incompetence have hampered the progress of valves based on resorbable matrixes. No good histological pictures of failed valves appear in the literature, and thus little can be learned from the 10 years of experience with this approach. Each group working in this field appears to publish promising short-term results with a particular matrix material and yet abandons it in favor of a more promising one with no report of how the old material fared. What remains unclear to most of the scientific community is specifically how the previous approach failed. Although PLA/PGA⁹⁵ was abandoned by the pioneers in this field—the group of Vacanti and Mayer of Harvard/Children’s Hospital—in favor of a polyhydroxyalkanoate, a better, more compliant material,⁹⁶ other groups around the world continue to work with PLA,¹⁵⁰ most likely because it is the only FDA-approved, clinically popular material. As the various research groups working in this field continue to use resorbable materials, the questions that remain can be unpleasant: Are PLA/PGA and their variants inappropriate materials, or have they simply been used improperly by the first groups? Are the other groups going to see the same results and waste a lot of time? What is the responsibility of the research community in helping others avoid approaches that failed and perhaps choose a better path? In an environment where the prize is a piece of a $1 billion annual market, competition for attention and credibility is clearly felt at scientific meetings. Reporting about failures, rather than successes, also does not make for exciting reading and negative results seldom get published in mainstream journals. The ability for scientists to learn from the mistakes of the past is therefore quite limited.

	Hybrid Approaches

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 
A possible alternative to the acellular valve and the bioresorbable matrix approaches is the fabrication of complex structures by manipulating biological molecules. With sufficient fidelity, one could potentially fabricate structures as complex as the aortic valve cusps. This approach is a derivative of that pioneered by Tranquillo,¹⁰⁰ except that the directed collagen shrinkage method is simplified to a process that forms essentially 1D strings of collagen. These collagen fiber bundles can then be used as building blocks for the development of a composite aortic valve cusp.¹⁵¹ Whereas collagen structures in heart valves can be found in the form of sheets,¹⁵² most of the load-bearing components of the valve cusp are relatively thick, dense collagen fiber bundles. Having developed this technology already^109,153 the next step was to develop the other components required to make this approach work.

The aortic valve consists of the three basic building blocks of all connective tissues: collagen fibers, elastin sheets, and glycosaminoglycan matrix schematically arranged as shown in Figure 4. The collagen exists to bear tensile load and provide some ultimate stiffness and strength to the valve, so that it can withstand diastolic loads.¹²⁵ The elastin matrix exists to return the collagen structures back to their resting states between loading cycles,^124,126 and the glycosaminoglycans likely maintain hydration and the intrinsic viscoelasticity of the tissue.^123,142,154 For the latter component, hyaluronan was chosen as the working material.

View larger version (78K): [in this window] [in a new window]   Figure 4. Schematic diagram of the multilayered configuration of an aortic valve cusp, showing the location of the Collagen fibers in the fibrosa, the elastin sheets in the ventricularis, and the GAG-rich matrix of the watery spongiosa.

Hyaluronan is a glycosaminoglycan polymer with a repeating disaccharide structure (glucuronic acid–b1,3-N-acetylgalactosamine-b1,4-)n, where n can reach 25 000 or more. In solution, hyaluronan forms large, random coil structures that occupy large solvent volumes. When constrained within a matrix, like a collagen network, hyaluronan exerts a swelling pressure that traps water, gives tissues compressive resistance, and imparts viscoelastic properties. Hyaluronan also plays a role during embryonic cardiac development as invading cells transform "cardiac jelly" into specialized cardiac structures, like the myocardium and the cardiac valves.¹⁵⁵ Hyaluronan is exceptionally biocompatible.^156–158 Unlike collagen that has both tissue- and species-specific markers, hyaluronan exhibits structural homology across species. For example, hyaluronan made by bacteria is the same as hyaluronan made by humans. Because of its interesting viscoelastic properties and its broad biocompatibility, hyaluronan has been used in a number of clinical applications and also as a scaffold for heart valve tissue engineering¹¹⁵ by others.

The formulation of hyaluronan gels used in our laboratory is based on a modification of a previously patented protocol.¹⁵⁹ Details of the preparation are published elsewhere,^160,161 but in brief, the process involves mixing commercially available sodium salt of long chain hyaluronan with 1 mol/L NaOH at low temperatures and crosslinking with divinyl-sulfone. These gels can have a stiffness as high as 30 kPa, similar to that of the elastin structures of the aortic valve cusps, and thus can form highly hydrated, elastic sheets. It is expected that appropriately cast sheets of crosslinked hyaluronan can find use in the central spongiosa layer of the aortic valve cusp.

Two spurious findings occurred when working with the collagen constructs and the crosslinked hyaluronan; both stimulated the formation of elastin sheets. Elastin sheets formed spontaneously around the periphery of the collagen constructs¹⁵³ and on hyaluronan substrates texturized through dehydration and texturization with ultraviolet light¹⁶² (Figure 5). Whereas the growth of elastin sheets on the collagen constructs is straightforward, growth of elastin sheets atop the hyaluronan gels has been difficult to reproduce. The success of this technique appears to be highly dependant on the specific cell type, gel formulation, and degree of surface texturization by UV light irradiation. Standardization of these protocols is an ongoing process in our laboratory, and no animal implants have yet been done.

View larger version (153K): [in this window] [in a new window]   Figure 5. Histological images of longitudinal sections of a collagen construct showing the new elastin sheath under low (A) and high (B) magnification. Transmission electron micrographs of hyaluronan gels with an adherent elastin sheet just underneath the cell layer, shown under low (C) and high (D) magnification. Panel B reproduced from Shi and Vesely153 with permission; panel C modified from Ramamurthi and Vesely162 with permission.

	Stem Cells and Other Future Prospects

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 
The cell is clearly an important component of the tissue-engineered heart valve. A very good review of the types of cells used in tissue-engineered valves is provided by Flanagan and Pandit.⁹ Much of the work identifying the phenotype of interstitial cells in the aortic valve has been done outside the US, in the laboratories of Yacoub,^163,164 Gerosa,¹⁶⁵ and Boughner.¹⁶⁶ Which cell to use for seeding scaffolds remains unclear. Stem cells and various other progenitor cells are being increasingly used in tissue-engineered valve applications. The 2 main cell types are mesenchymal stem cells and circulating endothelial progenitor cells. These cells are harvested from either experimental animals or patients, expanded in culture, and then seeded on the various valve leaflets substrates, be they the acellular matrix valves or the resorbable scaffolds. From presentations at recent conferences, all of these studies are very preliminary, showing no real differences in histological morphology over conventional interstitial cells used in previous studies.

Many investigators believe that cells are "smart," somehow recognizing the substrate and behaving in the appropriate way. The fibrotic overgrowth and failure of valvular matrixes described above clearly points to the contrary. With the emergence of stem cell science, many investigators believe that stem cells are going to be the "silver bullet" and are going to be smarter and act appropriately on the substrates. Embryonic stem cells, when injected into infracted hearts, have not behaved "smartly," did not rebuild the damaged myocardium, and, instead, created calcific deposits or teratomas.¹⁶⁷ There is no evidence that undifferentiated stem cells will behave in a more intelligent way on valvular substrates. Most likely, considerable effort will need to be expended to differentiate embryonic stem cells along a valvular lineage in advance of seeding on valvular substrates, before the promise of stem cells can be realized in this field.

Stem cell, however, are clearly the wave of the future. Considerable evidence exists in the literature that matrixes implanted without cells resorb, fibrose, and fail to produce any clinical benefit, whereas matrixes seeded with relevant cells incorporate and offer therapeutic benefit. The work of Atala in the field of urology is perhaps the best example of cell-seeded matrixes providing real clinical benefit to patients in augmenting large soft tissue defects.¹⁶⁸ Stem cells, once differentiated to the proper end point, are expected to provide a broader source of autologous cell lines, along with the appropriate matrixes, for therapeutic use in the cardiovascular field.

The use of autologous cells, cell expansion, preimplantation culture, and other cell "management," however, is not the preferred approach in the medical device and therapeutics industry. Because of regulatory and cost issues, a tissue-engineered valve that does not require any cell management is the preferred choice for medical device companies. This is the reason for which CryoLife and originally St Jude and Advanced Tissue Sciences and currently Medtronic have pursued the acellular matrix approach; it can be qualified, manufactured, and sold as a device, a far simpler approach than the biological or combination product that most other tissue engineering therapies are likely to follow. The fact that the CryoLife approach failed, however, has not deterred others from trying it. The literature clearly shows that the majority of research projects prefer the acellular matrix approach, particularly now with the greater promise of stem cells.

	Conclusions

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 
Perhaps the most often asked question by outsiders is, "When will tissue-engineered valves be ready for clinical use?" Given that heart valve tissue engineering is a field already almost 20 years old, and the only real clinical experience (the use of the Synergraft technology in children) has been disastrous, it may take another 20 years before the many complex challenges are finally solved. Although occasional experimental use of tissue-engineered valves in children will likely occur sooner, because children with congenital outflow tract abnormalities have few options, it will take 20 years to demonstrate that the long-term performance of tissue-engineered valves is comparable or better than conventional glutaraldehyde-treated porcine xenografts or pericardial valves. Children and adult patients with no option for conventional treatment will thus continue to serve as the proving ground for tissue-engineered solutions for cardiovascular defects for the foreseeable future, hopefully with less tragic consequences.

	Acknowledgments

 
The author is The H. Russell Smith Foundation Endowed Chair of Cardiothoracic Research, Professor of Cardiothoracic Surgery, at The Saban Research Institute of the Children’s Hospital Los Angeles and is grateful to the Saban Family, the Smith Family, and Dr Vaughn Starnes, the Chair of Cardiothoracic Surgery at the Keck School of Medicine, University of Southern California, for creating an environment in which outstanding research can be performed. The author thanks colleagues in the heart valve industry for candid thoughts on the history and the future of heart valve tissue engineering. Without their help, this article would be far less interesting and controversial.

	Footnotes

 
The author has in the past, and is now, providing expert opinion on heart valve litigation, including tissue-engineered valves.

Original received June 29, 2005; resubmission received July 8, 2005; revised resubmission received August 22, 2005; accepted August 29, 2005.

	References

Top Abstract Introduction Bioprosthetic Valves Why the Need for... Three Approaches to Valvular... History of Each Approach:... Hybrid Approaches Stem Cells and Other... Conclusions References

 

1. Medtronic Inc. News Release: Medtronic Receives CE Mark, Releases New Mechanical Heart Valve to European Surgeons. Minneapolis, Minn: Medtronic Inc; 2000. Available at: http://wwwp.medtronic.com/Newsroom/NewsReleaseDetails.do?itemId=1096566769805=en_US.
2. Barron V, Lyons E, Stenson-Cox C, McHugh PE, Pandit A. Bioreactors for cardiovascular cell and tissue growth: a review. Ann Biomed Eng. 2003; 31: 1017–1030.[CrossRef][Medline] [Order article via Infotrieve]
3. Breuer CK, Mettler BA, Anthony T, Sales VL, Schoen FJ, Mayer JE. Application of tissue-engineering principles toward the development of a semilunar heart valve substitute. Tissue Eng. 2004; 10: 1725–1736.[CrossRef][Medline] [Order article via Infotrieve]
4. Breuer C, Anthony T, Fong P. Potential tissue-engineering applications for neonatal surgery. Semin Perinatol. 2004; 28: 164–173.[CrossRef][Medline] [Order article via Infotrieve]
5. Elkins RC. Tissue-engineered valves. Ann Thorac Surg. 2002; 74: 1434.[Free Full Text]
6. Elkins RC. Is tissue-engineered heart valve replacement clinically applicable? Curr Cardiol Rep. 2003; 5: 125–128.[Medline] [Order article via Infotrieve]
7. Fath R. ["Tissue engineering": cultivated heart valves are still a future vision]. Dtsch Med Wochenschr. 2003; 128: 912.[Medline] [Order article via Infotrieve]
8. Flameng W. A new approach to heart valve tissue engineering. Int J Cardiol. 2004; 95 (suppl 1): S55–S56.[Medline] [Order article via Infotrieve]
9. Flanagan TC, Pandit A. Living artificial heart valve alternatives: a review. Eur Cell Mater. 2003; 6: 28–45;discussion 45.
10. Haverich A. Tissue engineering. Eur J Cardiothorac Surg. 2004; 26 (suppl 1): S59–S60;discussion S60–S61.[CrossRef]
11. Hopkins RA. Tissue engineering of heart valves: decellularized valve scaffolds. Circulation. 2005; 111: 2712–2714.[Free Full Text]
12. Jankowski RJ, Wagner WR. Directions in cardiovascular tissue engineering. Clin Plast Surg. 1999; 26: 605–616.[Medline] [Order article via Infotrieve]
13. Korossis SA, Fisher J, Ingham E. Cardiac valve replacement: a bioengineering approach. Biomed Mater Eng. 2000; 10: 83–124.[Medline] [Order article via Infotrieve]
14. Mayer JE Jr, Shin’oka T, Shum-Tim D. Tissue engineering of cardiovascular structures. Curr Opin Cardiol. 1997; 12: 528–532.[Medline] [Order article via Infotrieve]
15. Mitka M. Tissue engineering approaches utility. JAMA. 2000; 284: 2582–2583.[Free Full Text]
16. Mol A, Bouten CV, Baaijens FP, Zund G, Turina MI, Hoerstrup SP. Review article: tissue engineering of semilunar heart valves: current status and future developments. J Heart Valve Dis. 2004; 13: 272–280.[Medline] [Order article via Infotrieve]
17. Mol A, Hoerstrup SP. Heart valve tissue engineering–where do we stand? Int J Cardiol. 2004; 95 (suppl 1): S57–S58.[Medline] [Order article via Infotrieve]
18. Morsi YS, Birchall IE, Rosenfeldt FL. Artificial aortic valves: an overview. Int J Artif Org