Chapter 69. Tissue Engineering for Cardiac Valve Surgery

Tissue engineering is a developing science, comprising elements of engineering and biology, whose aim is to build replacement tissues de novo from individual cellular and structural components. The impetus for our work on tissue engineered cardiovascular structures arises from the need to replace cardiovascular tissues that failed to develop normally during embryogenesis or have become dysfunctional as a consequence of disease. In pediatric patients, the cardiovascular structures most often afflicted by congenital anomalies are the cardiac valves and great vessels.

Diseases of the heart valves and large “conduit” arteries account for approximately 60,000 cardiac surgical procedures each year in the United States, and all of the currently available replacement devices have significant limitations.1,2 Ideally, any valve or artery substitute would function like a normal valve or artery, that is, to allow the passage of pulsatile blood flow without stenosis or regurgitation. The theoretical advantage of engineered tissue replacements is that they could also display other desirable characteristics, such as: (1) durability, (2) growth (for infants and children), (3) compatibility with blood components and the absence of thrombosis or destructive inflammation, and (4) resistance to infection. None of the currently available devices, constructed from either synthetic or biologic materials, meet these criteria. Mechanical heart valves are very durable, but they require anticoagulation to reduce the risk of thrombosis and thromboembolism.1,2 Therapeutic anticoagulation carries associated morbidity, and even among patients who receive therapeutic anticoagulation, the incidence of thromboembolic complications of mechanical heart valve replacement is not zero.1,2 Biologic valves, whether of allograft or heterograft origin, remain subject to structural deterioration after implantation.2–4 Neither mechanical nor biologic valves have any growth potential, and this limitation represents a major source of morbidity for pediatric patients who must undergo multiple reoperations to replace valves and/or valved conduits during the period of maximum somatic growth.

Tissue engineering is an approach based on the hypothesis that when properly designed, living devices will simulate the biology of normal cardiovascular structures, thereby overcoming the shortcomings of currently available heart valve replacements. Of particular relevance to pediatric heart surgery is the long-term function of the engineered valve over time: its capacity to grow, self-repair, and remodel. This chapter summarizes some of the progress that has been made in tissue engineering research as it relates to cardiac valves and conduit arteries, and then outlines the areas where additional efforts must be focused in order to direct cardiovascular tissue engineering toward clinical utility.

Adult Valve Structure and Function

Heart valves open and close approximately 40 million times per year; this coordinated function occurs under the demands of hemodynamics. The normal valve presents minimal resistance to opening and no pressure gradient during systolic forward flow. In diastole, the same structure is responsible for rapid and complete closure in order to prevent valve regurgitation. Semilunar valves must additionally resist pressure differences between the diastolic arterial pressure and the diastolic ventricular pressure.

Like many other tissues, valve cusps are composed of cells residing within, and acting reciprocally on, the extracellular matrix (ECM). Semilunar valve cusps are thin, flexible structures with impressive microscopic and molecular complexity. Much of the strength and flexibility of normal heart valve cusps is owing to the specialized proteins and polysaccharide-protein complexes of the extracellular matrix, produced by resident valve interstitial cells.5,6 The microscopic and molecular structure of adult valves reflects the regional mechanical forces experienced by the valve cusps; valve cusp ECM is not homogeneous, and the arrangement of the extracellular matrix provides a high degree of flexibility during systole, and a high degree of strength to resist pressure loads during diastole.5 On the surfaces, a specialized endothelium prevents thrombosis, and acts as a transducer of mechanical force.7,8 Under the endothelium, valve interstitial cells receive signals sent by the endothelium, and respond by secreting suitable matrix.9,10

Valve ECM is stratified and related to valve mechanics.11 Facing the sinus of Valsalva, where eddy currents occur with diastolic pressure loads, dense collagen is found in the fibrosa layer. A middle layer of connective tissue, the spongiosa, is particularly rich in glycosaminoglycans, large complex molecules that associate with water and act as “shock absorbers,” bearing largely compressive mechanical loads. The ventricularis, an elastin-rich layer facing the great artery, is specialized to stretch as the cusps elongate during ejection in systole.

Valve structure and function are therefore closely related, and strategies to engineer valves are aimed to mimic normal valve structure and function, potentially at the subcellular, cellular, tissue, and whole heart valve levels.

Embryologic and Postnatal Development

Valves do not begin as stratified tissues; as circulatory patterns evolve in fetal and postnatal life, valve morphology also changes.12 Embryologic valve development, therefore, has relevance to tissue engineering, as a model for normal in vivo valve remodeling as programmed by gene regulatory pathways and by applied biomechanical force.

By day 15 of human embryo development, specification of myocardial and endothelial cardiac cells occurs. At day 21, formation of a linear heart tube occurs. These events are followed by rightward looping of the heart, resulting in orientation of the cardiac chambers into their final adult positions, and opposition of two specialized segments of ECM, known as cardiac jelly.13 The first evidence of valvulogenesis occurs in this cardiac jelly when a subset of endothelial cells delaminate, invade the ECM, and undergo endothelial-to-mesenchymal transformation (EMT).14 In early fetal valve development, valve interstitial cells are highly proliferative, and reside in a glycosaminoglycan-rich, homogeneous microenvironment. Over the course of fetal development, cell proliferation slows, and production of extracellular matrix results in valve cusp elongation. During the late stages of fetal development (20 to 36 weeks of human gestation) and in early postnatal life, valve stratification into a trilaminar structure occurs, and organization and maturation of collagen fibers begins. At the time of birth, through a series of largely undefined molecular steps, changes in oxygenation and blood pressure distinguish the aortic from the pulmonary sides of the circulation.12 Transitional neonatal circulation has been associated with a change in phenotype from activated to quiescent interstitial cells in pulmonary, but not aortic, valve cusps.15 Valve maturation and remodeling continue during childhood; cellularity of valve interstitial cells continues to decrease into adulthood.12 Once thought to be passive structures, increasing evidence shows cardiac valves to be dynamic organs. Mechanistically, the finding that cellular and extracellular components mature is thought to reflect the dramatic changes in flow and biomechanical loading conditions from fetal to adult life.12,16 Further elucidation of the genetic regulatory events defining valve growth and maturation is critical to designing biomimetic replacement devices.

Approaches to tissue engineered heart valves (TEHV) can be divided into two paradigms, based on the type of structural scaffold employed: bioresorbable scaffolds (nonwoven felts, electrospun scaffolds, knitted meshes, hydrogel-based, and combinations) and decellularized tissue-based scaffolds. The bioresorbable scaffold approach begins with seeding cells and culturing under appropriate biomechanical conditions (static flow, pulsatile flow) and nutrient medium. With regard to tissue mechanical properties, the goal of this approach is fabrication of an adequately strong tissue with approximately constant mechanical properties, requiring that the process of scaffold degradation occurs with a reciprocal increase in extracellular matrix production. Porosity is an important characteristic of bioresorbable scaffolds as it provides a permeable framework for cell migration, nutrient supply, and waste removal. Insufficient porosity leads to nutrient deprivation and cell death. Once a sufficiently stable tissue is formed, in vivo implantation would ideally follow, placing a newly synthesized tissue, devoid of foreign elements, in the required site.

This de novo approach is the one that has been predominantly employed in our laboratory at Children's Hospital, Boston.17–26 This approach treats components of tissue as “building blocks,” constructing valves from a variety of cell types and scaffold materials. Most in vivo studies have been carried out in the lower-pressure pulmonary circulation, which is a more tolerant system than the systemic circulation. This type of approach considers several basic questions as its foundation: (1) What cell type or combination of cell types is necessary to allow the production and maintenance of an appropriate extracellular matrix? (2) To what extent can cellular phenotype be altered, guided, or “engineered” to replicate cells found in the normal valve? (3) How can these cells be spatially organized during the development of this tissue engineered structures until the cells in the construct produce sufficient and appropriate extracellular matrix? (4) What biochemical signals are necessary during the development of these structures to ensure proper extracellular matrix production? (5) What mechanical signals are necessary for optimal tissue development and growth? (6) Should a tissue engineered valve construct be completely developed and appear as an adult valve before implantation, or can further maturation of an engineered construct occur in vivo after implantation?

Proponents of the decellularized tissue method base their approach on the premise that extracellular matrix geometry directs cell behavior, and that the closest structure to the normal valve scaffold is the normal valve scaffold itself. Human aortic valve homografts are currently implanted without tissue-type matching, become acellular after several months in vivo, yet retain their mechanical properties for decades. As the exact features endowing the aortic valve with such durability are unknown, it is hypothesized that these features are retained after ex vivo decellularization. Scaffolds are either seeded with cells before implantation, or implanted without seeded cells, in the expectation that appropriate circulating cell populations will populate the scaffolds in vivo. Without seeding of cells before implantation, however, some decellularized matrices are insufficiently endothelialized by circulating cells to resist surface thrombus formation in vivo.27

The literature of tissue engineered heart valves demonstrates the selection of a variety of cell types, scaffold conditions, and preconditioning regimens, and overall, methods are difficult to compare and results have been inconsistent. Systematic screening of engineered tissue elements and combinatorial approaches to building tissues have recently emerged.28,29 A complete review of past investigations is beyond the scope of this chapter; however, a summary of the fundamental elements of synthetic, biomaterial-based TEHV will be considered in detail in the following sections. In vivo outcomes of these methods will also be considered.

Cell Origin and Phenotype

As in embryologic valvulogenesis, the ideal cell type for a tissue engineered heart valve would fulfill the function of both the valve interstitial and endothelial cells. Conceptually, these cells could be derived from fully differentiated cells capable of extracellular matrix synthesis, or from less committed, multipotent or pluripotent stem cells with the additional potential for differentiation into multiple cell types.

The first tissue engineered heart valve experiments were performed with differentiated cells from artery or vein, including vascular smooth muscle cells, fibroblasts, and endothelial cells derived from the vasculature of immature animals.17–21 These cells were chosen because they were readily accessible, are derived from a cardiovascular source, and can synthesize extracellular matrix proteins. A comparison of myofibroblasts from the wall of the ascending aorta with those from segments of saphenous vein revealed that the latter cells exhibit superior collagen formation and mechanical strength when cultured on biodegradable polyurethane scaffolds.30 In our laboratory at Children's Hospital, Boston, tissue engineered heart valves (TEHV) based on these differentiated cells from systemic blood vessels functioned for periods of up to 4 months in vivo.17,19 However, enhanced collagen formation may be a double-edged sword. The rapid formation of new tissue in the early culture period could give rise to an overabundance of matrix elements, leading to tissue stiffness and potential tissue contraction. Early studies with dermal fibroblasts demonstrated that valve cusps constructed from these cells developed tissue contraction, which limited the ability of valve leaflets to coapt with each other, resulting in valve regurgitation.31 In addition to production of excessive or unfavorable types of extracellular matrix, mature cells may present a problem of senescence in long-term cell cultures in vitro, which limits the ability to quickly produce sufficient numbers of cells to seed a TEHV construct. Finally, the prospect of harvesting segments of artery from an otherwise normal peripheral circulation in order to obtain cells for a TEHV represented an undesirable clinical situation, and therefore led to a search for alternative cell sources for engineered valves.

The emergence of the field of stem cell biology has changed the paradigm for candidate cell types for heart valve tissue engineering. As stem cells differentiate from their embryonic state, they lose pluripotency with each subsequent step. Multiple steps of differentiation form lineages of cells. In embryonic development, differentiation down a specific lineage occurs with biochemical signaling, occurring in a specific mechanical microenvironment.32 How lineage specification occurs in the developing heart is currently an area of active research, with the goal of understanding the necessary cues to replicate differentiation down valve cell lineages. In normal development, differentiating cells are subject to a rapidly changing three-dimensional extracellular environment, and development is thought to occur by signals originating outside of the cell, from molecules originating in neighboring cells, and in the extracellular matrix. These reciprocal interactions between cells and their environment in developing valves, the regulatory mechanisms that induce cells to secrete and respond to components of appropriate extracellular matrix, are thought to be the underlying mechanisms stratifying valve cusps into their known compartments. These mechanisms are thought to be largely driven by hemodynamics in developing valves.

There is also recent evidence that stem cells with proliferative and regenerative capacities reside in many adult tissues. These stem cells are capable of not only acting locally on the tissues in which they reside, but they also may be recruited out of the circulation and enlisted in the regeneration of diverse tissues at distant sites. A recent review details emerging evidence that bone marrow–derived endothelial, hematopoietic stem and progenitor cells can also contribute to tissue vascularization during both embryonic and postnatal life.33 Visconti and coworkers have made the intriguing observation that cardiac valve interstitial cells in mice appear to originate in the bone marrow as hematopoietic cells.34 The idea that bone marrow contains cells capable of repairing damaged tissue has been applied to regeneration of cardiac muscle after myocardial infarction, whereby progenitor cells have been isolated from sites outside the heart, then injected back into the heart in an attempt to regain contractile function of ischemic or infarcted myocytes. Results of these trials have been equivocal, marginal, or negative, suggesting that the process of regeneration does not occur by the simple addition of multipotent cells.35 The early experience with the SynerGraft decellularized heterograft valved conduits that were implanted in children as right ventricle-to-pulmonary artery conduits occurred with the expectation that circulating cells from the bloodstream, or ingrowth from adjacent normal tissue, would repopulate the grafts and grow. In this instance, repopulation of the graft by circulating cells occurred, but did not result in adequate function or tissue growth.36,37 Nonetheless, the progenitor cell populations represent an attractive source of cells for tissue engineering and regeneration, because they have the plasticity necessary to fulfill critical cell functions and are potentially programmable for lineage specification. In addition, these cells can be obtained less invasively than differentiated cells.33 Our initial experience with progenitor cells in the Mayer laboratory was gained with autologous endothelial progenitor cells (EPCs) isolated from circulating blood in lambs and seeded onto decellularized arterial segments.17 These seeded arterial grafts were then implanted as an interposition graft in the carotid artery of the donor lamb. These grafts remained patent and functional for up to 130 days. Subsequent animal studies by Sutherland and associates used bone marrow mesenchymal stem cells (MSC) to seed a bioresorbable scaffold formed into a three-leaflet valve within a conduit (Fig. 69-1). These valved conduits were implanted as valved conduits into the main pulmonary artery of neonatal sheep, and remained in place for up to 8 postoperative months.24 This study was followed by that of Gottlieb et al., who implanted valves of similar elements into larger numbers of sheep and followed changes in function with growth of the animals using MRI and echocardiography. Although the valves had trace to mild regurgitation at the time of implantation (Fig. 69-2), a loss of valve leaflets surface area was observed, which correlated with increasing valve regurgitation over time.26 Importantly, the valve leaflets underwent a remodeling process in vivo after implantation, seen also in earlier experiments using myofibroblasts and endothelial cells from systemic arteries, and in both sets of experiments a layered histologic appearance developed after implantation.19,24,26

Several types of progenitor cells have been used for tissue engineering applications in congenital heart disease. Cebotari and colleagues have recently reported an initial experience seeding EPCs onto homograft valves followed by implantation into two children.38 Matsumura and associates have shown that when seeded onto a copolymer of lactic acid and ε-caprolactone, green fluorescent protein (GFP)–labeled cells contributed to the histogenesis of their explanted tissue engineered vascular graft. These grafts remained patent, and explanted constructs contained GFP-labeled cells expressing both endothelial and mesenchymal markers.31 The group at Tokyo Women's Medical College has also carried out implants of tissue engineered vascular grafts in children with congenital heart disease using whole mononuclear cell fractions, seeded onto bioresorbable scaffolds.39 This work is now ongoing at Yale University, because of encouraging early results.

Increasing evidence supports the idea that for cells, “geography is destiny.” Many cell types including bone marrow–derived MSC exhibit surprising plasticity, and cell phenotype seems to be related to the microenvironment in which cells reside.40–42 However, many of the factors controlling differentiation in these environments remain incompletely defined. There is evidence that endothelial progenitor cells are able to transdifferentiate in response to biochemical signals. Dvorin and colleagues showed that in the presence of transforming growth factor (TGF)-β<sb>1</sb>, EPCs express α-smooth muscle actin (α-SMA), after seeding on a bioresorbable copolymer scaffold.43 This behavior is not characteristic of endothelium; SMA expression suggests a mesenchymal phenotype, and this observation is a potentially related phenomenon to EMT seen in embryologic valvulogenesis.43 Human aortic valve endothelial cells, but not vascular endothelial cells, respond to TGF-β<sb>1</sb> in a similar fashion, suggesting that endothelial progenitor cells may be suitable as a replacement for valve endothelium. Though stem cells of many types remain promising candidates for tissue engineering, their full potential will be harnessed with an understanding of their normal generative and regenerative roles in vivo.

Structural Scaffold

Although it has been possible to grow individual cell types in culture for decades, it is more difficult to induce cells to assemble or organize into complex three-dimensional structural arrangements that are found in normal tissues, and to induce cells to produce specific extracellular matrix components on demand. Implantation of engineered tissues requires structural integrity, and for this reason, most tissue engineers employ a structural scaffold, in addition to cellular material.

Given this fundamental requirement, two main strategies for development of a three-dimensional tissue have been used: (1) de novo scaffold synthesis and arrangement into a three-dimensional structure; and (2) decellularization of a whole (usually xenograft) tissue.

Any scaffold for tissue engineering applications must be biocompatible and allow cells to adhere and proliferate. For congenital cardiac applications, tissue growth is our target, and therefore the scaffold must either degrade or be remodeled in vivo. The advantage of the biopolymer approach is that the chemistry of these scaffolds allows for in vivo degradation, usually by hydrolysis.44 The disadvantage is that heart valves are structurally complex and anisotropic; designing the structural features of the normal heart valve de novo has presented a substantial engineering challenge. The obvious advantage of the decellularization approach is that the three-dimensional complexity is largely preserved; however, there is a shortage of homograft material relative to the clinical demand, and immunogenicity remains a concern with xenografts. Perhaps most importantly, the extracellular matrix of decellularized xenografts is dense, and may prevent the penetration of seeded cells into the interstices of the matrix. In our laboratory, we have constructed trileaflet valved conduits from small intestinal submucosa and seeded them with EPCs.23 After implantation in an ovine model, there was satisfactory short-term function, but there was no penetration of any cells into the depths of the small intestinal submucosa scaffold. Extracellular matrix proteins, when exposed to blood components, can induce inflammation, thrombosis, and calcification, so this material has not been pursued further in our laboratory.

Our primary laboratory efforts to develop heart valves and large arteries have utilized the approach of seeding cells onto alternate bioresorbable polymer scaffolds. Ideally, there exists an inverse relationship between scaffold degradation and extracellular matrix formation by seeded cells. Polymer degradation is related to polymer chemistry, fabrication method, and mechanism of degradation, and therefore polymers used in tissue engineering differ in their degradation times. The first generation of scaffolds used in heart valve tissue engineering were highly porous, nonwoven felts produced from fibers of polyglycolic acid (PGA). PGA and related polymers continue to be the most widely used for multiple tissue engineering applications.45–49 The advantage of PGA and related aliphatic polyesters, including poly-L-lactic acid (PLLA), is their safety, biocompatibility, lack of toxicity, and commercial availability. PGA has been used as the commercially available Dexon suture material since the 1970s.49 PGA can be extruded as a fiber, which allows fabrication of nonwoven sheets with large, open pores. Open pore structures facilitate cell delivery and proliferation by allowing a large surface area for cell attachment, free diffusion of nutrients and dissolved gases, and removal of waste products of metabolism. Material properties of these nonwoven felts are well established, with reproducible hydrolytic degradation times; PGA alone degrades in 2 to 4 weeks, whereas the majority of fibers of the more hydrophobic PLLA degrade within 4 to 6 weeks. These scaffolds lose strength before losing mass, which challenges tissue engineers to seed sufficient numbers of matrix-producing cells so as to replace the scaffold strength as it is lost. Our laboratory has experimented with PGA coated with the thermoplastic polymer poly-4-hydroxybutyrate (P4HB), assembled into a trileaflet structure by attaching leaflets to a flat scaffold sheet, then wrapping the scaffold around a mandrel and heat-welding a seam of attachment. Despite promising early results using the PGA/P4HB composite, subsequent studies showed loss of structural integrity with longer periods of in vitro culture, followed by difficulties with suture retention and hemostasis in vivo. For these reasons, Sutherland and colleagues in our laboratory developed a scaffold composed of equal parts of PGA and PLLA fibers. Because PGA is a stronger but more rapidly degrading polymer, and PLLA is a less strong polymer with a longer degradation time, we expected more uniform strength from the tissue when constructed from this material. The composite nonwoven felt was fabricated into a valved conduit with a trileaflet valve, and had substantially improved surgical handling characteristics.24

Although satisfactorily strong, polymer fiber-based scaffolds are significantly stiffer than normal valve leaflets, and with the addition of cell-secreted extracellular matrix, these constructs are notably stiff.44 Scaffold stiffness has been shown to affect cell behavior, and tissue engineers have sought less more flexible materials.42 Wang and colleagues at the Massachusetts Institute of Technology designed an elastic polymer with a rapid degradation time, based on sebacic acid, a derivative of castor oil. Polyglycerol sebaceate is a strong but elastomeric material, and is currently under investigation for use in heart valve tissue engineering applications.50

In addition to their stiffness, polymer-based scaffolds are thick, relative to normal valve leaflets. A thick scaffold leads to a nutrient gradient in culture, and many tissue engineering studies based on nonwoven materials have been limited by the lack of nutrient delivery to the deepest areas of engineered tissues. Two solutions are proposed to overcome this limitation: addition of a blood supply and design of thinner scaffolds.

Hydrogel-based scaffolds, including collagen, alginate, agarose, gelatin, fibrin, chitosan, polyethylene glycol, hyaluronic acid, and dehydrated sheets of extracellular matrix have formed the basis of many experiments in tissue engineering. When gels become solid, cells are trapped in the cell, providing a homogeneous distribution of cells embedded in a temporary matrix. In addition, hydrogels can be laid into thin layers. A bileaflet heart valve was produced in the Tranquillo laboratory using a collagen-based scaffold seeded with dermal fibroblasts.51 Recent advances in drug delivery technologies and microfluidics have resulted in further control of scaffold characteristics, including orchestration of scaffold polymerization with changes in temperature, pH, or exposure to light; engineering of nano- and micro-scale cell environments in order to direct cell distribution throughout a scaffold, and microencapsulation of growth factors and adhesion peptides on scaffolds for improved cell attachment and proliferation.52–56

Finally, cells and scaffold have been incorporated and directed in nanoscale fabrication techniques such as electrospinning. Still, no perfect material has been identified for heart valve tissue engineering, and in this regard, the search continues.

The complex anisotropy and three-dimensional structure of heart valves has provided another challenge for a biomimetic device. Even if the optimal scaffold material is identified, its fabrication into a three-dimensionally accurate valve geometry is not trivial. Using normal heart valve anatomy derived from computed tomographic images, Sodian and associates employed stereolithography to print a three-dimensional model for use as a mold for the thermoplastic polymer P4HB.57 The mold was used for curing the polymer into three-dimensional valve anatomy. As improvements in imaging technology yield higher spatial and temporal resolution, anatomic definition of thin, moving, anisotropic structures in the heart such as valve leaflets will be more feasible. In addition, evolving microfabrication methods will further our ability to fabricate microscale features of valve anatomy. With defined anatomic dimensions and an understanding of outflow tract, great artery, and leaflet motion and growth, tissue engineers will have clear targets for three-dimensional fabrication of heart valve scaffolds.

Biochemical Signals

The migration and transdifferentiation of endothelial cells in the early stages of valvulogenesis (EMT) has been experimentally modeled in an ex vivo chick cardiac cushion explant system.14 Much progress has been made to understand the signals required for the proper sequencing and execution of these early events. Vascular endothelial growth factor is one such molecule, thought to regulate EMT in an environment of adequate tissue glucose and oxygen saturation. Although VEGF is secreted by endothelial cells, other important signals, such as bone morphogenetic protein 2 and 4, are expressed by myocardium. Hyaluronic acid, a component of the extracellular matrix, is thought to regulate downstream signaling through its large, hydrated structure, which regulates ligand availability for receptor binding. Therefore, valvulogenesis is dependent in vivo on signals from myocardium, local extracellular matrix, and endothelium.11,32 Though very early growth of endocardial cushions and valve primordia are less understood, and late regulatory events in valve growth are poorly understood, regulators of endothelial and mesenchymal cell proliferation have been identified. These known pathways have been largely studied in isolation, and as gene regulation in organogenesis is a highly complex process, synthesis of critical gene pathway data is necessary for developing a big-picture view of required events. Current microarray technologies will yield large volumes of data, and will likely provide additional insight into the critical regulatory steps involved in valve growth. This type of information will be high yield in future generations of engineered valves.

Mechanical Signals

Identification of a suitable cell phenotype is necessary but not sufficient for engineering replacement tissues. Proper cell orientation and three-dimensional microstructure are also required for tissue function; tissues demonstrate organization of cells and matrix across multiple levels of scale. In this regard, engineered tissues, like native tissues, require coordination. Hemodynamics are fundamental to the development and ongoing function of cardiovascular structures, and biomechanical signals are epigenetic regulators of tissue growth and development. Increasing evidence supports a role for endothelial cells as mechanotransducers, sending signals through the underlying extracellular matrix to the more deeply embedded valve interstitial cells.16 Endothelial cells respond to shear stress and cyclic strains.58 In an environment in which concentrations of soluble growth factors are held constant, endothelial cell programs can be switched between growth, differentiation, and apoptosis by varying the extent to which the cell is spread or stretched.59 As cells are embedded in, and coupled to, their extracellular matrix, ECM serves as a vehicle through which signals must pass. Biochemical signals can be sequestered or amplified by ECM, and as cells bind to surrounding ECM via integrin receptors, the binding itself can induce phenotypic changes.32 When human MSCs are plated onto large tissue islands that promote cell spreading, they efficiently differentiate into bone cells; when plated onto small islands, the same cells in the same culture medium differentiate into adipocytes.60

Clinical observations made from patients undergoing the Ross procedure have provided further evidence for the responsiveness of vascular and valve tissues to hemodynamic forces. In the Ross procedure, a pulmonary valve from the lower-pressure pulmonary circulation is transplanted into the aortic position and subjected to systemic pressure, leading to significant changes in the phenotype of valve interstitial cells, including an increase in matrix metalloproteinase activity, indicating ECM remodeling.15

Biophysical signaling is therefore considered fundamental to engineered tissue organization and conditioning, or training, for the in vivo environment. Mechanical forces can be applied to growing engineered tissues in a flow-loop containing tissue culture medium, known as a bioreactor. Preconditioning of tissues is thought to be important for the biology and monitoring of engineered cardiovascular tissues. Bioreactors are thought to serve two predominant purposes for engineered heart valves: first, mechanical forces influence cell phenotype and gene expression, and therefore tissue development, and potentially growth, are controlled by biomechanical signals. In experiments from our laboratory reported by Hoerstrup and colleagues, flow and pressure were demonstrated to increase the production of collagen in tissue engineered semilunar valves.19 In separate experiments, Lee and associates demonstrated the variation of extracellular matrix gene transcripts with changes in tightly controlled mechanical strains. Vascular smooth muscle cells seeded onto biodegradable scaffolds, then subjected to cyclic flexure produced more collagen and were stiffer than controls.61 Similar findings were reproduced using mesenchymal stem cells, and porcine heart valves.62,63

In addition, because valves must not be regurgitant at the time of implantation, observation of engineered valve mechanics in a bioreactor before implantation has allowed investigators to monitor and predict in vivo valve function.

Progress toward a clinically translatable engineered heart valve is not possible without in vitro study of valve development and remodeling. Developmental biology and in vitro experiments are the fundamental elements of this emerging field. However, in vivo experiments are the necessary complement to in vitro investigation, and represent the true test of engineered device function over time. The in vivo environment is more complex than can be approximated in vitro, and remodeling is dependent on the contributions of circulating progenitor and immune cells only present in whole living organisms.

Four major implantation studies of tissue engineered heart valves have been undertaken using the de novo engineered approach.19,24,26,64 Several groups have implanted decellularized and recellularized valves into the juvenile lamb circulation,65,66 and two major studies evaluate the growth of vascular conduits in the pulmonary circulation.67,68 Several representative studies will be reviewed in terms of valve function, evaluation of valve/conduit growth, and observation of engineered tissue maturation.

In Vivo Valve Function, Growth, and Maturation

Heart valve tissue engineering was conceptually established through proof-of-concept experiments in large animals. The first in vivo tissue engineered heart valve experiment involved implantation of a single leaflet into the pulmonary circulation.17 Cells were isolated from ovine arteries and sorted into endothelial and fibroblast populations. Cells were labeled, then seeded onto PGA scaffold and subsequently implanted. Labeled cells were visualized in explanted specimens after 6 hours and after 1, 6, 7, 9, or 11 weeks. Cells produced extracellular matrix, and leaflets persisted in the circulation at all time points.

A second set of experiments involved implantation of a pulmonary valve composed of autologous ovine endothelial cells and myofibroblasts seeded on a PGA-P4HB composite scaffold.19 After 14 days of exposure to gradually increasing flow and pressure conditions in a pulse duplicator, valves were implanted into sheep for 1 day, 4, 6, 8, 16, and 20 weeks. Valves demonstrated acceptable hemodynamics and had no evidence of thrombosis, stenosis or aneurysm formation for up to 20 weeks. Central valvar regurgitation was noted after 16 weeks. Tissue analysis showed a layered structure with central glycosaminoglycans, collagen on the outflow surface, and elastin on the inflow surface. Elastin was detectable in the leaflets by 6 weeks in vivo. Over the 20-week study period, scaffold degradation corresponded to decreased leaflet stiffness. Histologic, biochemical, and biomechanical parameters were similar to those of native pulmonary artery and leaflet tissue.

Using bone marrow–derived mesenchymal stem cells on a PGA/PLLA scaffold, investigators fabricated and implanted valved conduits in the pulmonary position of juvenile sheep.24 Valves were evaluated by in vivo echocardiography; explanted tissues were analyzed by histology and immunostaining. At the time of implantation, echocardiograms demonstrated a maximum instantaneous gradient of 17.2 ± 1.33 mm Hg, with estimates of regurgitation ranging from trivial to mild. Four animals survived the immediate postoperative period. After 4 months in vivo, there were no statistically significant difference between maximum instantaneous gradient, mean gradient, or effective orifice area from the time of implantation. Histologic analysis at 8 postoperative months showed explanted tissue to have a layered organization, with elastin fibers identified on the inflow surface and collagen dominating the outflow surface. Glycosaminoglycans were distributed throughout the remainder of the valve. At the time of implantation, cells uniformly expressed the mesenchymal cell marker a-SMA, but at the time of valve explant, these cells were confined to the subendothelial layer, whereas surface cells expressed von Willebrand factor, suggesting in vivo endothelialization of the graft. This study demonstrated the feasibility of implantation of a pulmonary valve using stem cells.

Further work in our laboratory has reproduced some of these initial findings, and studied valve function over time in larger numbers of animals.26 Using the same scaffold material (PGA-PLLA composite nonwoven felt) and cells (MSCs), 19 animals underwent implantation of valved conduits into the main pulmonary artery of neonatal sheep after resection of the native pulmonary leaflets. Valve function, cusp, and conduit dimensions were evaluated at implantation (echocardiography), at the experimental midpoint (MRI), and at explant at 1 day, 6, 12, or 20 weeks postoperatively. At implantation, valved conduit function was excellent; maximum transvalvar pressure gradient by Doppler echocardiography was 17 mm Hg; most valved conduits showed trivial pulmonary regurgitation. At ≥12 weeks, valved conduit cusps were increasingly attenuated and regurgitant. Valved conduit diameter remained unchanged over 20 weeks. Dimensional measurements by MRI correlated with direct measurement at explant. These studies demonstrated autologous engineered valved conduits that functioned well at implantation, with subsequent monitoring of dimensions and function in real time by MRI. The valves underwent structural and functional remodeling without stenosis, but worsening pulmonary regurgitation was noted after 6 weeks.

When fibrin-based valves were implanted in sheep for 3 months, valved conduits exhibited the gross appearance of intact tissue; however, leaflets demonstrated pulmonary regurgitation because of tissue contraction.64

To date, no group has implanted valve leaflets that demonstrated in vivo growth. In many experiments, however, the microscopic structure of the valve leaflets evolved from a relatively homogeneous appearance to a layered structure. The mechanisms by which this in vivo evolution of structure and cellular activity occurs remain completely unexplored, but these observations suggest that a tissue engineered valve may not have to be in a mature, adult form at the time of implantation into the circulation.

Present and Future

After a decade of effort toward a tissue engineered heart valve replacement, the clinical need for a more durable valve replacement still exists. Modes of failure of currently available heart valve replacements are well described and provide insight into strengths and weaknesses of valve designs. Though the appropriate cell source and scaffold material for heart valve tissue engineering have been areas of substantial experimental effort, the optimal cell type, scaffold material, and in vitro culture conditions for a tissue engineered heart vale still remain unknown. Stem cells and adult progenitor cells offer promise toward this end, because they possess the capacity for self-renewal and multilineage differentiation, and therefore, growth potential. Evolving imaging technologies will allow a more complete understanding of postimplantation remodeling and growth. Though many details of in vitro preconditioning regimens are yet to be understood, bioreactor preconditioning has the potential to reduce in vitro culture times by stimulating tissue formation, and may offer a controllable environment for evaluation of valve function before implantation.

Complex biological events, such as heart valve development, occur through both biochemical and biophysical cues, occurring in parallel at the cellular, tissue, and organ levels. A detailed understanding of the molecular sequence of critical developmental events may allow engineering of appropriate microenvironments and three-dimensional structures leading to the ultimate goal: a clinically translatable heart valve replacement with the capacity to grow and remain durable for a lifetime of cardiac cycles.