Acellular vascular tissues: natural biomaterials for tissue repair and tissue engineering

Transcription

1 Biomaterials 21 (2000) 2215}2231 Acellular vascular tissues: natural biomaterials for tissue repair and tissue engineering Christine E. Schmidt*, Jennie M. Baier Department of Chemical Engineering, University of Texas at Austin, 26th and Speedway Austin, TX, USA Abstract Various research groups around the world are actively investigating cardiovascular prostheses of biological origin. This review article discusses the need for such bioprosthetics and the potential role for natural tissues in cardiovascular applications such as cardiac valves and vascular grafts. Upon implantation, unmodi"ed natural materials are subject to chemical and enzymatic degradation, seriously decreasing the life of the prosthesis. Therefore, methods such as glutaraldehyde and polyepoxide crosslinking treatments and dye-mediated photooxidation have been developed to stabilize the tissue while attempting to maintain its natural mechanical properties. Also, residual cellular components in a bioprosthetic material have been associated with undesired e!ects, such as calci"cation and immunological recognition, and thus have been the motivation for various decellularization processes. The e!ects of these stabilization and decellularization treatments on mechanical, biological and chemical properties of treated tissues have been investigated, speci"cally with regard to calci"cation, immunogenicity, and cytotoxicity concerns. Despite signi"cant advances in the area of cardiovascular prostheses, there has yet to be developed a completely biocompatible, long-lasting implant. However, with the recent advent of tissue engineering, the possibility of applying selective cell seeding to naturally derived bioprosthetics moves us closer to a living tissue replacement Elsevier Science Ltd. All rights reserved. Keywords: Tissue engineering; Cardiac bioprosthesis; Vascular prosthesis; Decellularization; Glutaraldehyde; Photooxidation 1. The need for cardiovascular bioprosthetics For more than 40 years, materials to replace malfunctioning or diseased cardiovascular tissues have been under investigation. Arti"cial prostheses were introduced into cardiovascular surgery in 1952 when Hufnagel implanted the "rst arti"cial heart valve, and in the same year, Voorhees introduced the "rst arti"cial vascular graft. Subsequent reconstructive procedures have been developed with the intent of increasing implant biocompatibility, including the transfer of healthy tissue from one site or individual to another and the use of living tissue prosthetics fabricated through tissue engineering. However, despite signi"cant advances in implant technology, clinical experience has revealed the developmental challenges of prosthetics in the cardiovascular system. As reviewed below, the current therapies for cardiovascular diseases are not perfect, and the development of improved prosthetics and living cell-based devices is actively underway. * Corresponding author. Tel.: # ; fax: # address: schmidt@che.utexas.edu (C.E. Schmidt). Over 175, 000 prosthetic cardiac valves are implanted annually [1]. Presently, there are two primary choices for heart valve substitutes: mechanical and xenogeneic bioprosthetic (tissue) valves. Cryopreserved allograft valves are also used clinically, but to a lesser extent because of limited supply [2,3]. With regard to mechanical valves, several styles exhibit long-term durability ('25 years), but in general, the key drawbacks to these mechanical devices include the need for life-long anticoagulation therapy, bleeding disorders, and #ow dynamics that are distinctly di!erent from those of normal heart valves. Xenogeneic bioprosthetic valves consist of chemically crosslinked intact porcine aortic valves or valves created from crosslinked bovine pericardial tissue [1,4]. These bioprosthetic valves display better hemodynamics than do mechanical valves, and their use does not require life-long anticoagulation therapy. Unfortunately, these tissue-based valves tend to have shorter lifetimes (&7}10 years) due to complications with calci"cation and lea#et wear [5,6]. In addition to the large number of cardiac valves required every year, over 600,000 vascular grafts are implanted annually to replace damaged blood vessels /00/$ - see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S ( 0 0 )

2 2216 C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215}2231 [7]. Arterial damage typically results from atherosclerotic plaques that restrict the lumen of the blood vessel and subsequently damage the endothelial lining of the artery. This damage ultimately results in the formation of a thrombus and obstruction of blood #ow. For reconstruction of large arteries, such as the aorta or iliac artery, synthetic grafts made from eptfe or Dacron are the vascular grafts of choice. However, synthetic materials are not suitable for reconstruction of smaller diameter arteries ((6 mm diameter), as required for lowerextremity bypass and coronary artery bypass grafting procedures, because they carry a substantial risk for thrombosis. For these small-diameter arterial bypass procedures, physicians will typically use an autologous vein graft (e.g., saphenous vein), and in the case of the coronary artery, may alternatively use an autologous arterial graft (i.e., internal thoracic artery, gastroepiploic artery, inferior epigastric artery, radial artery) [8}10]. Cryopreserved allografts have also been used in the clinic as coronary artery bypass conduits, but poor patency rates (i.e., high rates of occlusion) and problems with aneurysm have limited the use of these grafts to situations in which no other autologous conduits are available [11,12]. Although venous and arterial autografts currently yield the best results, disadvantages include the need for multiple surgical procedures, with increased risk and cost to the patient. In addition, vein grafts have thin walls that may be damaged when transplanted into the arterial system, and suitable vessels are not available in all patients due to disease, amputation, or previous vessel harvest. Thus, there remains a clear need for a vascular prosthesis that would be suitable for small-diameter vessel reconstruction. The creation of such a graft has been the focus of considerable research for many years [13], yet there is still no adequate alternative to the autograft for small-caliber vessel reconstruction procedures. 2. The potential role of natural tissues as cardiovascular biomaterials The use of xenograft and allograft tissue as part of bioprosthetic vascular devices such as heart valves and vascular grafts has long been the focus of research [14]. The use of these natural biomaterials has typically required chemical or physical pretreatment aimed at (1) preserving the tissue by enhancing the resistance of the material to enzymatic or chemical degradation, (2) reducing the immunogenicity of the material, and (3) sterilizing the tissue. Multiple crosslinking techniques have been explored in an attempt to "nd the ideal procedure to stabilize the collagen-based structure of the tissue while maintaining its mechanical integrity and natural compliance. In addition to crosslinking techniques, decellularization approaches may reduce host immune response to bioprosthetics and generate natural biomaterials for use in cell seeding and tissue engineering applications. Naturally derived materials o!er many mechanical, chemical and biological advantages over synthetic materials, and thus hold tremendous potential for use in tissue engineering therapies. 3. Glutaraldehyde treatment: early e4orts in preservation of natural tissues The most commonly accepted crosslinking reagent is glutaraldehyde, a "ve-carbon bifunctional aldehyde, whose use has dominated since its introduction into biomedicine in the late 1960s [15]. Glutaraldehyde reacts with the ε-amino group of lysyl residues in proteins (e.g., collagen), which induces formation of interchain crosslinks [16] and stabilizes tissues against chemical and enzymatic degradation depending on the extent of crosslinking [17,18]. Unfortunately, the exact mechanism by which this occurs is complex and only partially understood [19,20]. Glutaraldehyde's success as a "xative and sterilant is in part due to its mixed hydrophobic and hydrophilic character, which allows the molecule to rapidly penetrate both aqueous media and cell membranes [21]. Crosslinking with glutaraldehyde has also been shown to suppress immunological recognition of the tissue [18,22], presumably preventing the display of antigenic determinants by killing viable tissue cells and by controlling the stability of the collagen triple helix [16]. However, glutaraldehyde treatment does not completely eliminate the immune response to allografts and xenografts [23,24]. For instance, glutaraldehyde-tanned bovine pericardial tissue elicits a cytotoxic T cell and a humoral response when implanted into rats [24]. Both residual cellular debris and extracellular matrix (ECM) proteins in glutaraldehyde-treated tissue are thought to contribute to such an immune response [25]. Thus, immunogenicity can be only partially attenuated by glutaraldehyde crosslinking. In fact, the use of glutaraldehyde to minimize antigenicity comes at the expense of inducing other undesirable e!ects (e.g., compromised mechanical properties, cytotoxicity, calci"cation) Mechanical properties of glutaraldehyde-treated tissues Glutaraldehyde-treated tissue exhibits altered mechanical properties compared to untreated tissue [26,27]. Porcine aortic valves crosslinked with glutaraldehyde tend to be sti!er than fresh tissue and have stress relaxation rates about 60% of those for fresh valves [28]. Treated tissues also show increased apparent tensile extensibility associated with shrinkage during "xation [29,30]. These mechanical changes may facilitate early

3 C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215} failure of the device, either through direct tearing of the tissue (e.g., lea#et) or by accelerating calci"cation. For example, failure of the bioprosthetic valves may be related to compressive buckling of the aldehyde-treated tissue and breaking of collagen "bers at the site of buckling [31]. To modulate the mechanical properties of glutaraldehyde-treated porcine aortic valves and bovine pericardium, several studies have investigated the utility of stress and pressure during the "xation process [32}37]. For example, the application of uniaxial and biaxial stress during glutaraldehyde treatment of bovine pericardium results in permanent changes in the mechanical behavior of the crosslinked tissue, depending on the magnitude of the initial load [34}37]. This process enforces an alternative conformation of the collagen and elastin structure present during the reaction [30,33}35,37,38]. Additionally, changes in the pressure at which the "xation process is carried out also appear to impact the organization of the matrix (i.e., the collagen crimp geometry) and the mechanical behavior of the resulting tissue [32]. These data suggest that applied stress and pressure during crosslinking may permit the engineering of bioprosthetic materials with desired mechanical properties for speci"c medical applications. Other approaches have been used in an attempt to improve the mechanical properties of glutaraldehyde- "xed tissues. For example, altering the solvent dielectric constant during "xation of pericardium and aortic tissue alters collagen conformation and subsequently a!ects the extent of crosslinking and the resulting mechanical properties of the tissue [39,40] Cytotoxicity of glutaraldehyde-treated tissues Host endothelial cells do not typically grow onto bioprosthetic tissues when implanted into patients, and the cytotoxic e!ects of glutaraldehyde treatment presumably contribute to this lack of endothelialization. Solutions of glutaraldehyde contain not only its linear monomer but also dihydrated forms of monomeric glutaraldehyde, monomeric and polymeric hemiacetals, and aldol condensation products such as polymeric aldehydes [19]. These crosslinking chemicals have been shown to leach slowly from tissue-derived bioprostheses "xed in glutaraldehyde, producing cytotoxic e!ects [41,42]. Many approaches are under investigation as a means to detoxify or neutralize the toxic e!ects of glutaraldehyde on processed tissues. Alternate storage solutions such as hydroxy-benzoate, rather than the typical glutaraldehyde or formaldehyde storage solutions, in addition to extensive bioprosthetic rinsing, have partially reduced the cytotoxic e!ects [41]. However, regardless of storage solutions and prosthetic rinsing, the degradation of the actual glutaraldehyde-derived crosslinks [43] and the continual release of cytotoxic aldehydes appear to contribute to prolonged toxic e!ects of glutaraldehydetreated tissues [44,45]. Thus, more comprehensive approaches to reduce the e!ects of glutaraldehyde are being pursued. Several groups have demonstrated a detoxi"cation process in which a low ph L-glutamic acid treatment is used to extract and neutralize aldehyde groups from treated tissues, thus improving the ability of these materials to support endothelial cell growth [19,46,47]. L-glutamic acid reacts with aldehydes to form acetals and esters, and the acidic ph favors depolymerization of polymeric glutaraldehyde and subsequent extraction of excess aldehyde [48]. Other detoxi"cation approaches include the use of chondroitin sulfate [18], protamine [49], diphosphonates [18,50], and homocysteic acid [51]. In addition to their e!ects on cytotoxicity, extraction and neutralization of these glutaraldehyde products are also linked to reduced calci"cation [48,52,53] Calcixcation of glutaraldehyde-treated tissues Calci"cation is the formation of calcium-containing mineral deposits, which results in cusp or vessel sti!ness, loss of pliability, and blockage of the valve or vessel opening (Fig. 1). This detrimental side e!ect of glutaraldehyde processing severely limits the lifetime of treated tissues and is the primary cause of failure of bioprosthetic heart valves [20,54]. Furthermore, calci"cation is much more di$cult to reduce in vascular grafts than in lea#ets and pericardium [55]. The exact mechanism for calci"cation is not known, and the process has been attributed to several factors: (1) the degree of crosslinking [17,18], although this theory has been refuted [56}58]; (2) the presence of unreacted aldehyde groups [52,59]; (3) the attraction of calcium ions to glutaraldehyde [60]; (4) the disruption of cellular calcium regulatory mechanisms by glutaraldehyde, leading to nucleation sites for calci"cation [61]; (5) localized areas of stress [62,63]; and (6) the presence of lipids and cellular debris [64}68]. Refer to Rao and Shanthi for a detailed review of the proposed mechanisms of valve calci"cation [20]. Many approaches are being investigated in an e!ort to minimize calci"cation of glutaraldehyde-treated tissues. It has been hypothesized that positive charge modi"cation of tissues will prevent the in"ltration of Ca and minimize tissue calci"cation. Thus, several such approaches to minimize Ca di!usion have been explored, including the use of trivalent metallic cations (e.g., Fe and Al ) [69,70], protamine sulfate [49,70], L-glutamic acid [71], 2-amino oleic acid [72], and diphosphonate [50,73]. Residual alkaline phosphatase activity in glutaraldehyde-treated tissues presumably is a factor in the pathogenesis of mineralization [74], and tissues treated with levamisole, an alkaline phosphatase inhibitor, have shown some reduction in calci"cation [70]. Furthermore, the addition of proteoglycans, such as

4 2218 C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215}2231 calci"cation [76}78]. Ethanol pretreatment resulted in: (1) an irreversible conformational change in type I collagen, (2) a decrease in the amount of water in the tissue samples, (3) an increase in tissue stability as assessed by increased resistance to collagenase digestion, (4) no change in glutaraldehyde content in the tissue (i.e., glutaraldehyde is not extracted), and (5) a nearly complete extraction of lea#et cholesterol and phospholipid. Ethanol pretreatment, in synergistic combination with other anticalci"cation agents, may be of bene"t for preventing bioprosthetic tissue calci"cation. Although these various treatments seem to retard calci"cation of glutaraldehyde-treated tissues, some evidence suggests that in most cases the mineralization process is not completely inhibited [78,79]. Thus, studies are actively underway to "nd the appropriate combination of agents and improved methods to attenuate calci- "cation in glutaraldehyde-treated tissues, and to develop alternative "xation and preservation procedures that prevent this undesired side e!ect Glutaraldehyde-treated vascular grafts Fig. 1. Calci"ed glutaraldehyde-treated cardiac valve. This clinical porcine bioprosthetic valve was removed because of extensive calci"cation, a condition which can result in cuspal tears and stenosis or narrowing of the valve opening. (a) Gross photograph of explanted valve with calci"cation and cuspal tear. (b) Radiograph of valve specimen in (a), showing calci"c deposits. (Reproduced by permission from Schoen FJ, Hobson CE: Hum Pathol 1985;16:549}559). chondroitin sulfate, to tissues after glutaraldehyde "xation [18] and the selective removal of lipids prior to glutaraldehyde treatment [75] also help to attenuate calci"cation. However, the glutaraldehyde crosslinking reagent on its own, regardless of cells and soluble proteins in the tissue, likely contributes to the formation of calcium salts, as suggested by studies in which puri"ed (i.e., cell-free) collagen sponges calci"ed when treated with glutaraldehyde and implanted into rats [56]. More recently, investigators have demonstrated that ethanol pretreatment of aortic-wall tissue and bioprosthetic aortic-valve cusps prior to glutaraldehyde-crosslinking signi"cantly, but not completely, inhibits Glutaraldehyde-treated human umbilical vein grafts have been used with some success for lower limb revascularization [80}83]. These grafts typically display lower graft patency compared to autologous saphenous vein, but higher patency rates than those for synthetic materials. For example, in one comparative study of saphenous veins, glutaraldehyde-treated human umbilical vein grafts, and eptfe grafts, patency rates after 5 years for popliteal revascularization were 76, 74, and 41%, respectively [84]. In a separate study comparing umbilical vein grafts and eptfe for below-knee femoropopliteal bypass, patency rates of 75% for umbilical vein graft and 40% for eptfe were reported at 1 year [85]. Biodegradation of the umbilical vein grafts represents the most serious potential barrier to long-term patency. With increasing time after implantation, the umbilical vein grafts undergo progressive dilation and exhibit a higher frequency of aneurysms, with reported rates of occurrence reaching 18% after 5 years [81]. However, more recent studies suggest that these concerns have been greatly exaggerated [82]. Although the autologous saphenous vein is still considered the material of choice for smalldiameter vessel reconstruction, the glutaraldehydetreated human umbilical vein graft may be an acceptable alternative in cases when an autologous vein is absent or de"cient, where life expectancy is limited, and where expediency may be a critical factor [81,82] Summary and future directions for glutaraldehyde treatment of tissues The use of natural tissue, derived either from a human donor or from an animal source, requires that the tissue be

5 treated to minimize immunogenicity of the graft and to stabilize the tissue against rapid enzymatic and chemical degradation in the body. Glutaraldehyde crosslinking accomplishes these goals for the most part, but its use in preserving tissues has also been associated with several problems, including altered mechanical properties and early mechanical failure, calci"cation, cytotoxicity, and incomplete suppression of immunological recognition [86]. Many new approaches are being explored to reduce or eliminate the undesired side-e!ects of glutaraldehyde treatment of tissues, including methods to neutralize or extract glutaraldehyde products from the treated tissue, as already described, and the use of alternative, nonaldehyde crosslinking and preservation procedures. In addition, methods are being explored to produce completely acellular-tissue matrices by speci"cally removing cells and cell fragments, which are believed to promote calci"cation and to give rise to a residual immunological response. The subsequent stabilization of these acellular tissues with glutaraldehyde or other crosslinking techniques may prove essential for certain applications. However, because the fundamental mechanisms underlying cytotoxicity, calci"cation, and immunological recognition are not completely understood, the development of new approaches to create the ideal natural biomaterial is an even larger challenge. C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215} Alternative procedures for preserving tissues There is considerable interest in identifying and investigating alternative tissue treatments that preserve natural tissue but do not result in the deleterious side e!ects typically associated with glutaraldehyde treatment. Several of these alternative crosslinking and "xation approaches include the use of carbodiimides such as cyanimide [44,87,88] and 1-ethyl-3(-3 dimethyl aminopropyl) carbodiimide hydrochloride (EDC) [89,90], adipyl dichloride [91}93], hexamethylene diisocyanate (HMDC) [94], glycerol [59,95,96], alginate azide [20], and dehydration approaches (heat drying and freeze drying) to induce the formation of crosslinks [44,88,97]. Two particularly promising methods are (1) chemical crosslinking using polyepoxy compounds, and (2) crosslinking catalyzed by dye-mediated photooxidation Polyepoxy compounds Poly(glycidyl ether) (i.e., polyepoxy) compounds have been investigated extensively as crosslinking agents for porcine and pericardium cardiac valves [98}103] and vascular grafts [104}109]. Polyepoxy reagents are a family of chemicals that have greatly varying backbone lengths and functionalities. Several di!erent polyepoxy compounds have been used, including glycerol polyglycidyl ether and polyethylene glycol diglycidyl ether Fig. 2. Photo-Fix vascular graft. (a) Image of a sheep carotid artery that has been treated using a dye-mediated photooxidation process. (Photo courtesy of Diane Hern-Anderson, Sulzer Innotec, Austin, TX). (b) Histological cross-sectional image of a section of PhotoFix sheep carotid artery tissue that has been seeded with endothelial cells (EC). The endothelial cells form a complete, well-adhered monolayer on the PhotoFix material. (PGDE). In particular, the polyepoxy compounds marketed under the trade name Denacol (Nagase Chemical Company, New York) have been used frequently in tissue crosslinking applications. Some of these reagents include Denacol EX-313, EX-512, EX-521, EX-810, and EX-861 (Fig. 2) [88,101,104,110}115]. Polyepoxy compounds di!er from glutaraldehyde in their mechanism of crosslink formation: glutaraldehyde reacts only with the ε-amino group of lysyl residues in proteins, whereas polyepoxy compounds react with amide, carboxyl, phenol, and alcohol groups [116,117]. Polyepoxy reagents with intermediate length backbones and 4 or 5 epoxide groups are most e!ective in producing intrahelical crosslinking [115]. More speci"cally, reagents with short molecular backbones (EX-313 and EX-810) or a very long backbone (EX-861) displayed a slow rate of crosslinking, whereas molecules with backbones of 17 or 25 atoms (corresponding to a 4 or 5-mer of glutaraldehyde, EX-512 and EX-521) were more

6 2220 C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215}2231 e!ective. This suggests an optimal molecular length for crosslinking. As with glutaraldehyde, treatment of tissue with polyepoxy compounds (e.g., glycerol polyglycidyl ether) reduces the antigenicity and immunogenicity of the tissue proportionally to the reaction time [111,118]. Polyepoxy treatment of cardiovascular tissue represents an e!ort to overcome some of the drawbacks that are typically encountered with glutaraldehyde, but it is not clear whether, or to what extent, these problems are addressed. For example, in terms of mechanical properties, polyepoxy-treated vessels and valves have typically been softer, more #exible, and have shown better retention of natural residual strains compared to glutaraldehyde-treated tissues [103,113,119]. However, others claim that the mechanical properties of polyepoxytreated tissues are similar to those of glutaraldehydetreated tissues, including increased extensibility and decreased stress relaxation in pericardium and decreased compliance in arterial grafts [68,88,99,115,118]. Earlier observations of increased compliance in polyepoxytreated tissues compared to glutaraldehyde-treated tissues may be due to incomplete crosslinking with the slower reacting polyepoxy reagents [115,118]. In terms of cytotoxicity of polyepoxy-treated tissues, several studies demonstrate improved biocompatibility compared to glutaraldehyde-treated tissue, as suggested by observations of good recellularization of polyepoxy-treated tissues implanted in vivo [109,110,120,121]. On the other hand, a separate investigation has shown low endothelial cell coverage of polyepoxy-"xed vascular graft tissue [122]. The reason for this contradiction is not clear. With respect to calci"cation, almost all studies have shown a de"nite reduction in calci"cation for tissues treated with polyepoxide compounds [98,100,112,115]. Polyepoxide-treated vascular grafts have shown increased antithrombogenicity and improved patency compared to other vascular graft materials [109,110,113]. Perhaps polyepoxy (i.e., Denacol EX- 313) treatment results in the formation of a polyethylene glycol hydrogel layer, which reduces the adherence of blood components and promotes the in"ltration of plasma proteins and endothelial cells, thus aiding antithrombogenicity [110]. One study in dogs showed three-month patency rates of 100% for Denacol -treated small-diameter vascular grafts (i.e., Baxter Dena#ex, Baxter Edwards CVS Division) implanted into the carotid arteries, compared to 40% for eptfe [113]. In a separate experiment, also in dogs, patency was 71% after 145 days for polyepoxy-treated arterial grafts compared to 10% for glutaraldehyde-tanned vascular grafts [110]. Other studies have demonstrated about the same patency rates for polyepoxy-treated grafts as for eptfe grafts (&40}50%) [104,122]. Aneurysm formation also appears to be reduced in polyepoxy-treated vascular grafts compared to glutaraldehyde-tanned grafts in the carotid artery position in rabbits [109] and dialdehyde starch crosslinked grafts in the coronary artery bypass position in dogs [104,123]. Introduced as an e!ective "xative alternative to glutaraldehyde, polyepoxy treatment of cardiovascular tissues seems to o!er the advantages of reduced calci"cation, and improved resistance to thrombosis and aneurysm. However, the mechanical and cytotoxicity e!ects of this treatment have not been fully elucidated Dye-mediated photooxidation Dye-mediated photooxidation is an alternative tissue preservation method that requires no harsh chemical "xatives such as glutaraldehyde [124,125]. Certain amino acids such as tryptophan, histidine, tyrosine, and methionine in a protein can be speci"cally oxidized by irradiation with visible light in the presence of a suitable photosensitizer (e.g., methylene blue and rose bengal dyes) [126}129]; this reaction does not cause cleavage of the peptide bond [130]. With respect to photooxidation of histidine, it has been reported that aspartic acid and urea were the "nal products of this reaction, and a detailed mechanism for the oxidation reaction has been described [131]. Furthermore, other studies show that photooxidation of collagen solution stabilizes the collagen to denaturation and to enzymatic degradation, presumably by the formation of protein crosslinks [132]. More recent studies have shown that dye-mediated photooxidation can be used to stabilize intact collagenbased tissues such as bovine or sheep pericardium [133] and small-diameter arteries (Fig. 2a). For these applications, photooxidation serves as a catalytic process that induces modi"cation and crosslink formation within the existing matrix components, resulting in a more natural material with little added matrix complexity [134]. Photooxidized bovine pericardial tissue (PhotoFix, Sulzer Carbomedics, Austin, TX) is nonimmunogenic, biocompatible [135], and unlike glutaraldehyde-treated tissues, PhotoFix is also noncalcifying [125,136,137]. In one study in juvenile sheep, four of six valves with photooxidized tissue remained free of any signs of calci- "cation for up to 1.5 years, while none of the glutaraldehyde-treated valves did, suggesting that photooxidation is a promising method of preserving and "xing tissue for use in bioprostheses [137]. Furthermore, PhotoFix tissue is similar to untreated tissue in texture, pliability, and shrinkage temperature, but unlike untreated tissue, it possesses chemical, enzymatic, and in vivo stability [133,134]. Photooxidized pericardium and vascular grafts have supported endothelial cell growth in vitro (Fig. 2b), demonstrating the non-cytotoxic nature of dye-mediated photooxidation [54,138,139]. In vivo studies have further demonstrated the potential utility of PhotoFix tissue for medical applications. For example, PhotoFix bioprosthetic heart valves implanted into sheep are partially endothelialized after 2 years, demonstrating the ability of

7 C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215} these tissues to adequately support cell adhesion and migration [125,135}137]. Furthermore, studies of Photo- Fix small-bore sheep carotid artery implants show 1-year patency rates of 100% in the pig carotid artery position, and 150-day patencies of 60% in the dog coronary artery bypass model [140]. Fixation of cardiovascular tissues with dye-mediated photooxidation results in a stabilized material that is noncalcifying, nonimmunogenic, and noncytotoxic, yet retains natural physical properties and biomechanical integrity. Therefore, dye-mediated photooxidation holds promise as a longterm implantable biomaterial. 5. Extraction of cellular components from natural tissues In addition to alternative treatments that preserve (i.e., crosslink) natural tissue, methods are being explored to produce completely acellular tissue matrices by speci"- cally removing cellular components that are believed to promote calci"cation and to give rise to a residual immunological response. These decellularization techniques include chemical, enzymatic and mechanical means of removing cellular components, leaving a material composed essentially of extracellular matrix components. For the most part, these acellular tissues retain natural mechanical properties and promote remodeling of the prosthesis by neovascularization and recellularization by the host. Ultimately, the creation of actual living tissue replacements for cardiovascular applications would solve many of the existing problems associated with cardiac heart valve replacements and vascular prostheses The role of cellular remnants in calcixcation During typical processing and crosslinking treatment of tissue, cells are ruptured, but the cellular debris is largely retained. Various cell extraction methods (e.g., detergent treatments, enzymatic digestion, sonication) have been pursued as a means to create completely acellular tissues for use as biomaterial implants, since residual cellular components and lipids within processed tissue may promote undesired e!ects such as calci"cation [64,65,68,141]. Cell remnants have been microscopically associated with early calcium crystal formation [64,142]. Furthermore, glutaraldehyde-treated tissues that were pre-treated with acidi"ed sulfuric ether to remove phospholipids, as demonstrated by electron microscopy, showed signi"cantly attenuated calci"cation compared to normal glutaraldehyde-treated tissues [75]. Presumably, the removal of the phospholipids from the treated tissue reduces the number of available sites for deposition of calcium-based hydroxyapatite crystals. In a separate study, the selective extraction of lipids from bovine pericardium by either chloroform/methanol or sodium dodecyl sulfate (SDS) treatment decreased calci"cation of the tissue in the rat model [67]. The authors hypothesized that the initiation of calci"cation involves an electrostatic interaction of calcium ions with phospholipids. Regardless of the exact mechanism of calci"cation, which still remains unclear, these studies suggest the need to account for the role of lipids and, in particular, the membranous phospholipids in strategies to minimize calci"cation of bioprostheses. Cell extraction procedures, such as detergent treatments, often remove proteoglycans, which also appear to play a signi"cant role in calci"cation [143]. In one set of studies, the removal of free proteoglycans from glutaraldehyde-treated pericardium using the reagent guanidine hydrochloride resulted in reduced calci"cation, but increasing the extent of proteoglycan extraction produced a greater accumulation of calcium [67]. The authors hypothesized that extensive extraction procedures, which presumably removed some of the more tightly bound proteoglycans and resulted in loss of collagen-proteoglycan interactions, generated a porous matrix into which calcium salts were deposited [66,67]. Moreover, the addition of exogenous proteoglycans (i.e., chondroitin sulfate) to tissues after glutaraldehyde "xation has been shown to attenuate calci"cation [18]. Thus, the role of proteoglycans in calci"cation is not clear, although it appears that the removal of the more tightly bound, collagenassociated proteoglycans in tissue is detrimental The role of cellular remnants in immunological response In addition to its role in calci"cation, the presence of cellular debris (e.g., lipids) and soluble proteins (e.g., proteoglycans) in treated biological tissue contributes to an immunological response by the host toward the implanted bioprosthesis. For instance, glutaraldehydetreated bovine pericardial tissue elicits both a cytotoxic T cell and a humoral response when implanted in rats [24], presumably due in part to residual cellular components. Thus, the extraction of cellular components from tissues should minimize immunologically induced in- #ammatory processes. In fact, acellular matrix vascular allografts that were produced by detergent and enzymatic extraction of natural canine arteries showed no in#ammation when implanted into dogs of a di!erent breed [144,145]. It has also been hypothesized that the removal of antigens, such as found on cell surface proteins, will reduce in vivo cellular attack and possibly eliminate the need for extensive crosslinking [68]; studies have demonstrated that aortic allografts implanted into rats displayed signi"cant in#ammatory cell in"ltration, medial thinning due to degradation, dilation, and aneurysm formation [146]. Thus, reduction and possible elimination of in#ammation, by the removal of cell components, may diminish biodegradation of implanted

8 2222 C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215}2231 bioprostheses and minimize the need for extensive stabilization of allografts using crosslinking procedures. It is also important to keep in mind that even after the removal of cells and cell debris, the intact ECM of the acellular tissue itself may elicit an immune response [25]. In the laboratory, the immunogenicity of collagen is measurable, but the actual clinical signi"cance of such immunogenicity is very low due to the small species di!erence among di!erent type I collagens [16]. On the other hand, research has shown that other ECM molecules (i.e., "bronectin) in allografts may play a fundamental role in the host immune cascade triggered by organ transplantation [25]. The immune response for xenografts appears to be even more pronounced. For example, a series of studies have shown that decellularized arterial xenografts dilate, whereas decellularized arterial isografts and allografts do not [147,148]. The authors hypothesize that an interspecies, rather than intraspecies, immunogenicity of the arterial extracellular matrix leads to in#ammation, biodegradation, dilation, and ultimately chronic rejection. Similarly, recent studies have suggested that, although a decellularization process may attenuate a severe xenogeneic immune response [149], the removal of cellular components may not be su$cient to eliminate in#ammation, and "xation techniques may still be necessary to prevent degradation [150]. For instance, crosslinking may prove e!ective at lowering immunogenicity by altering the display of antigenic determinants [16], but at the possible expense of inducing other undesirable e!ects (e.g., calci"cation). On the other hand, glutaraldehyde-treatment of acellular matrix vascular xenografts speci"cally does not reduce antigenicity [150]. Regardless of these drawbacks, considerable e!ort is underway to process xenograft tissue, which is readily available, so that it can be used for clinical applications. In fact, CryoLife, Inc. plans to have a decellularized, uncrosslinked porcine valve (SynerGraft heart valve) available for human implant trials soon [2,149] Mechanical properties of acellular tissues The mechanical properties of uncrosslinked, cell-extracted tissues are remarkably similar to those of fresh tissue. Acellular pericardial matrix prepared using a four-step detergent extraction and enzymatic digestion process displayed similar mechanical properties to fresh tissue, except for slightly increased stress relaxation [68]. Subsequent crosslinking of the acellular matrix with glutaraldehyde produced mechanical changes consistent with the same treatment of fresh tissue. The authors found that although their extraction procedure removed most soluble proteoglycans, there was little e!ect on the viscoelastic properties of the tissue. A similar extraction process applied to porcine aortic valve lea#ets produced a tissue with a modest 20% reduction in fracture tension and increased tissue extensibility [151]. However, extraction preserved tissue structure and mechanics over the physiological loading range, whereas glutaraldehyde "xation produced markedly larger changes in mechanical properties including increased extensibility, increased elastic behavior, and a signi"cant drop in fracture tension. In a separate study, decellularized aortic and pulmonic valves were implanted as allografts in dogs; after one month, echocardiography showed lea#et motion and 3}5 mm Hg transvalvular gradients [145]. These results suggest that the acellular-valve lea#ets have the mechanical integrity to resist physiological-#ow conditions for at least short-term implantation. More recently, studies have demonstrated key di!erences in mechanical properties of acellular tissues depending on the method of decellularization. For example, a detailed characterization of decellularized aortic tissue using thermogravimetric analysis and di!erential scanning calorimetry showed that SDS treatment destabilizes the collagen triple helix and swells the elastin network, whereas Triton X-100 and cholate treatments did not a!ect the structural integrity of either collagen or elastin [152] Acellular vascular grafts To address the particular concerns with small-diameter vascular prostheses, investigators have developed sequential detergent extraction techniques [153,154] and multistep-detergent-enzymatic procedures [68,144,145, 150] to create acellular vascular grafts. Both techniques utilize Triton X-100 and SDS to remove the cellular components of natural vasculature while preserving the ECM foundation. The sequential detergent extraction technique involves a nondenaturing step with Triton X-100, followed by a denaturing SDS treatment, and results in a material composed of collagen, elastin, "bronectin and laminin [154]. The detergent-enzymatic procedure includes three steps: (1) suspension in a hypotonic solution containing protease inhibitors, (2) highsalt extraction involving Triton X-100, DNAse, and RNAse, and (3) treatment with SDS. This process yields a material primarily made of elastin, insoluble collagen and tightly bound glycosaminoglycans [144,145]. Both treatments had high patency rates in canine allograft applications: the sequential detergent extraction technique yielded a 90-day patency of 80% [153], while the detergent-enzymatic procedure resulted in a six-year patency of 94% [144]. In addition, no in#ammation, aneurysm or dystropic calci"cation was noted for the acellular matrix vascular prosthesis [144], and complete re-endothelialization of the graft in canines has been reported [153]. Along similar lines, LifeCell Corporation (The Woodlands, TX) is also working to develop an acellular vascular graft using detergent treatment combined with freeze drying (Fig. 3). LifeCell's "rst commercial product is

9 C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215} Fig. 3. Acellular vascular graft. (a) Normal caprine (i.e., goat) carotid artery before processing and (b) acellular caprine carotid artery following a decellularization and preservation process that involves detergent treatment combined with freeze drying. Both images are of tissue stained with hematoxylin and eosin. Original magni"cation, 100X (Photos courtesy of Lawrence E. Boerboom, LifeCell Corporation, The Woodlands, TX). AlloDerm, a decellularized tissue matrix derived from donated human tissue that retains the essential biochemical and structural composition of human dermis and provides a template for revascularization, cell repopulation and normal-tissue regeneration [155,156]. Allo- Derm is processed from allograft skin using detergent extraction to remove the cell components, which are the targets of the rejection response, and freeze drying to maintain the ultrastructural integrity of the extracellular matrix. When transplanted into a patient, AlloDerm is incorporated into the surrounding tissue, revascularized, and repopulated with the patient's own cells. Moreover, the AlloDerm matrix guides the new cells to remodel and maintain the matrix, establishing a new functional living tissue. AlloDerm is distinguishable from allograft skin only by its lack of cell components, as de"ned by electron microscopic analysis and staining for major histocompatibility complexes (MHC) class I and class II. Using a similar technology to that used for dermis, Life- Cell has recently developed a vascular graft that is an acellular arterial matrix in which the cells are removed to avoid an immune response and the extracellular matrix is preserved to serve as a sca!old for host cell reconstitution. The graft becomes integrated into host tissues and is repopulated by host endothelial cells, smooth muscle cells, and "broblasts [157] Small intestinal submucosa as a biomaterial Small intestinal submucosa (SIS) is a unique acellular matrix derived from porcine small intestine (Fig. 4). SIS is prepared by mechanically removing layers of mucosa and muscle from the small intestine, lysing the native cells with hypotonic solution, treating the material with 0.15% peracetic acid, and "nally rinsing with bu!ered saline or sterile water. This process results in an extracellular matrix material composed of about 90% collagen (primarily Type I), "bronectin, growth factors, glycosaminoglycans, proteoglycans and glycoproteins [158,159]. Various studies in animal models have shown suitability in several applications: vascular graft [160}166], pulmonary valve lea#et [167], lower urinary tract [168,169], tendon [170], dural mater [171], dermal tissue [172], and body wall tissue [173]. The primary advantage to SIS grafts is their ability to promote site-speci"c tissue remodeling and regeneration by the host. Upon implantation, rapid neovascularization and the in"ltration and spatial organization of host

10 2224 C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215}2231 by inconsistent patency results between microvessels and other larger diameter vessels, the thrombogenic response of SIS is not completely understood. 6. Tissue engineering prospects for natural biomaterials Fig. 4. Small intestinal submucosa (SIS). (a) SIS still in its native tubular shape. The appropriate super"cial mucosal and external muscular layers have been removed leaving the submucosal and basilar mucosal extracellular matrix. The SIS can then be `engineereda for particular applications by con"guring into single or multiple layer sheets, dehydrated, powdered, cut into strips for braiding, or formed into a gel for injection. (b) Histologic appearance of SIS. The SIS}ECM is approximately 80 μm thick, acellular after processing, and consists of organized bundles of structural collagen admixed with the naturally occurring components of the extracellular matrix. Note the remnants of blood vessels that once occupied this portion of the native small intestine. (Photos courtesy of Stephen F. Badylak, Purdue University, West Lafayette, IN). cells are presumably promoted by the natural complexity of the ECM structure and composition [158,159,174]. Furthermore, the authors suggest that the immediate formation of a direct blood supply enables the graft's resistance to infection. In a canine study comparing the e!ects of intentional graft infection with Staphylococcus aureus, 56% of eptfe grafts and 0% of SIS autografts had positive culture results after 30 days of implantation [175]. SIS autografts, allografts and xenografts have high patency rates when implanted in canine aorta, carotid and femoral arteries, and superior vena cava locations [176]. Patency rates at 6 months implantation were comparable for autogenous saphenous vein grafts (88%) and SIS xenografts (83%) in the canine model [163]. In a similar study, SIS xenografts had a higher patency rate than eptfe grafts (88% and 25%, respectively) [165]. However, when SIS microvessel grafts were implanted in rats, all grafts failed to remain patent beyond the "rst hour [177]. As a biomaterial composed primarily of acellular collagen, SIS has been shown to promote host tissue remodeling in a variety of tissue applications, including vascular grafts, and therefore is a promising candidate for cardiovascular applications. However, as demonstrated The "eld of tissue engineering has evolved rapidly over the last decade, and many parallel research e!orts are underway to create a vast array of living tissue replacements for therapeutic applications [178,179]. Most of these approaches involve the use of synthetic polymer sca!olds that serve to guide cell growth and tissue morphogenesis [180}182]. Although various biodegradable synthetic polymers show great promise, there is reason to believe that naturally derived materials provide additional bene"ts that also warrant their investigation as biomaterials for tissue engineering applications. Natural biomaterials are composed of ECM proteins that are conserved among di!erent species and that can serve as intrinsic templates for cell attachment and growth. On the other hand, synthetic materials do not contain epitopes that are directly bound by cellular adhesion receptors, and as a result, many investigations focus speci"cally on modifying polymers with various biological and adhesion moieties to enhance their cell and tissue compatibility [183,184]. In addition to inherent cell compatibility, natural materials possess the desired shape and, for the most part, the strength of the tissues from which the materials are derived. This can be a large advantage over synthetic materials in terms of materials processing. Although natural materials on their own possess properties that are desirable for biomedical applications, cell seeding of these materials and their use in tissue engineering applications will likely enhance long-term function. In other words, a material or bioprosthesis that is repopulated with the patient's cells, either naturally in vivo or by cell seeding in vitro, has the potential to develop into a living tissue that can adapt and respond to changes in the body. In particular, a vascular graft will almost certainly produce better results in the clinic if it is lined with an endothelium to prevent thrombosis and therefore increase graft patency [185,186]. Thus, e!orts are now underway to repopulate natural biomaterials with cells (ideally, derived from the patient) to create living tissue replacements. Several recent studies have focused on cellularizing naturally derived biomaterials with endothelial cells and "broblasts in an e!ort to create living vascular tissues [138,154,187}191]. An acellular vascular matrix produced by a series of detergent extractions and enzymatic digestions was able to support endothelial cell adherence and spreading without pretreatment of the matrix with "bronectin or serum [154]. Freeze-dried porcine valves have also demonstrated promising results with regard to

11 C.E. Schmidt, J.M. Baier / Biomaterials 21 (2000) 2215} in vitro cellularization with endothelial cells and "broblasts [189]. In this particular study, endothelial cells formed a con#uent layer on the surface of the porcine lea#ets, and viable "broblasts were present within the substance of the lea#et. Furthermore, photooxidized bovine and porcine pericardium have supported endothelial cell growth in vitro [138]. Studies are also underway to endothelialize Sulzer Carbomedics' PhotoFix sheep carotid arteries in an attempt to improve patency (Fig. 2b) [140,192]. Preliminary investigations have demonstrated uniform endothelial cell attachment and good retention ('85% after 24 h) under #ow conditions that mimic physiological, resting #ow in the coronary arteries (68 ml/min) [193]. Based on these preliminary results, it is clear that natural biomaterials, when processed using techniques that do not induce cell cytotoxicity, serve as an ideal template for cell growth and tissue engineering applications. In addition to these matrices' ability to support cell growth, it may be important that natural biomaterials be remodeled by the body to incorporate the prosthesis as living tissue. Highly crosslinked matrices are more stable to degradation, and therefore are less likely to be remodeled as part of tissue morphogenesis. However, modi"cation of the crosslinking process to produce a lower crosslinking rate can produce `growablea materials that are easily in"ltrated with cells, degraded, and remodeled as part of the wound healing response [105]. Alternatively, acellular tissues that are not crosslinked in any form, and thus are susceptible to remodeling in the body, may prove to be ideal biomaterials for cell-repopulation and new tissue growth [68,158]. The extent to which natural biomaterials that are processed with various preservation and decellularization techniques are remodeled in the body needs to be more fully explored. 7. Summary Many attempts have been made to produce longlasting, biocompatible cardiovascular implants. To overcome the mechanical and biological limitations of synthetic implants, various researchers have begun to focus on the development of a naturally derived biomaterial for the fabrication of heart valve replacements and vascular grafts. In order for materials to be transplanted to a patient from a donor, especially an animal donor, the tissue must be modi"ed to increase resistance to degradation and to decrease immunogenicity, while maintaining natural mechanical properties. Some of these approaches for tissue modi"cation are summarized in Table 1. Glutaraldehyde is the "rst and most prevalent "xative of naturally derived biomaterials. This crosslinking reagent is e!ective for stabilizing tissue and has proven a comparable alternative to autologous saphenous vein in vascular grafting applications. However, this treatment causes altered mechanical properties, increased calci"cation, and cytotoxicity. As alternatives to glutaraldehyde treatment, other methods to chemically modify natural tissue include crosslinking with polyepoxy compounds and dye-mediated photooxidation. These methods show improved resistance to calci"cation while providing stabilization against in vivo degradation. While dye-mediated photooxidation has been shown to be noncytotoxic, nonimmunogenic and to retain natural mechanical properties, the cytotoxicity and mechanical properties have not been fully elucidated for polyepoxide compounds. Cells are ruptured during typical chemical and thermal processing of natural materials, but residues such as phospholipids and proteoglycans remain, presumably causing increased calci"cation and antigenicity. Therefore, various cell extraction techniques are used to create decellularized tissues, including chemical, enzymatic and mechanical methods of removing cellular components, and leave a material comprised primarily of extracellular matrix components. Surprisingly, these acellular tissues retain natural mechanical properties and, in some cases, promote remodeling of the prosthesis by neovascularization and recellularization by the host. Although the antigenic response to acellular xenograft tissue is pronounced in comparison to allogeneic material, considerable e!ort is currently focused on treating readily available animal-derived tissues. Ultimately, the creation of living tissue replacements for cardiovascular applications would solve many of the existing problems associated with cardiac heart valve replacements and vascular prostheses. Although the use of synthetic materials in tissue engineering applications is being actively pursued and shows great promise, naturally derived biomaterials may o!er advantages in such applications. Naturally derived biomaterials are composed primarily of ECM components, which o!er a biological foundation suitable for cell attachment and growth. Furthermore, these natural materials may provide improved mechanical and shape compatibility compared to synthetic sca!olds. Thus, natural biomaterials, when repopulated with autologous or genetically engineered cells, can serve as the ideal template for the design of living implants for speci"c applications such as longlasting prostheses, growable grafts, or cell-based drugdelivery devices. 8. Notes For an excellent and comprehensive review written on tissue heart valves, refer to Schoen and Levy [4], and for an earlier general review on the use of tissue-derived biomaterials for cardiovascular prosthetics, refer to Hilbert et al. [14].