Transpecies Heart Valve Transplant: AdvancedStudies of a Bioengineered Xeno-AutograftSteven Goldstein, PhD, David R. Clarke, MD, Steven P. Walsh, PhD,Kirby S. Black, PhD, and Mark F. O’Brien, MDCryoLife, Inc, Kennesaw, Georgia, University of Colorado Health Science Center, Denver, Colorado, and Prince Charles Hospital,Brisbane, Australia

Background. Tissue engineering approaches utilizingbiomechanically suitable cell-conductive matrixesshould extend xenograft heart valve performance, dura-bility, and growth potential to an extent presently at-tained only by the pulmonary autograft. To test thishypothesis, we developed an acellular, unfixed porcineaortic valve-based construct. The performance of thisvalve has been evaluated in vitro under simulated aorticconditions, as a pulmonary valve replacement in sheep,and in aortic and pulmonary valve replacement inhumans.

Methods. SynerGraft porcine heart valves (CryoLifeInc, Kennesaw, GA) were constructed from porcine non-coronary aortic valve cusp units consisting of aorta,noncoronary aortic leaflet, and attached anterior mitralleaflet (AML). After treatment to remove all histologi-cally demonstrable leaflet cells and substantially reduceporcine cell-related immunoreactivity, three valve cuspswere matched and sewn to form a symmetrical rootutilizing the AML remnants as the inflow conduit. Syn-erGraft valves were evaluated by in vitro hydrodynam-ics, and by in vivo implants in the right ventricularoutflow tract of weanling sheep for up to 336 days.

Cryopreserved allograft valves served as control valvesin both in vitro and in vivo evaluations. Valves were alsoimplanted as aortic valve replacements in humans.

Results. In vitro pulsatile flow testing of the Syner-Graft porcine valves demonstrated excellent valve func-tion with large effective orifice areas and low gradientsequivalent to a normal human aortic valve. Implants insheep right ventricular outflow tracts showed stableleaflets with up to 80% of matrix recellularization withhost fibroblasts and/or myofibroblasts, and with no leaf-let calcification over 150 days, and minimal deposition at336 days. Echocardiography studies showed normal he-modynamic performance during the implantation pe-riod. The human implants have proven functional forover 9 months.

Conclusions. A unique heart valve construct has beenengineered to achieve the equivalent of an autograft.Short-term durability of these novel implants demon-strates for the first time the possibility of an engineeredautograft.

(Ann Thorac Surg 2000;70:1962–9)© 2000 by The Society of Thoracic Surgeons

Preparation of porcine aortic valves for human valvereplacement has always included either protein de-

rivitization or crosslinking. The earliest porcine valveimplants were enabled by modification with either mer-curials [1] or formalin [2]. When the poor durability ofthese valves became apparent, covalent bonding(crosslinking) within the cellular and extracellular struc-tures of the valve with glutaraldehyde was instituted [3],which reduced immunologic recognition of the xenoge-neic tissue by human recipients, and also stabilized it todegradative enzymes [4]. Over time, these bioprostheticvalves have demonstrated adequate durability in olderage groups but still have a high risk of early structuraldeterioration in younger patients [5].

Frequently underappreciated is the importance of thelack of cellular viability in such valves. Glutaraldehyde-crosslinked valves are, and remain, nonviable tissues

without the opportunity for either tissue renewal orgrowth, explained by the cytotoxicity of residual alde-hydes [6] and the inability of cells to migrate through thefixed, nondegradable matrix. It is clear that the cellularcomponents of the native heart valve leaflet repair andmodify the matrix resulting in lifetime durability. Over 20years of experience with aortic homografts has shownthat careful preservation of both matrix and cellularcomponents correlates well with clinical durability. An-tibiotic-stored grafts may not have had both matrix andcell integrity preserved, resulting in poor wear character-istics in comparison with cryopreserved homografts.Clearly nonviable homografts sterilized by chemicalagents (b-propiolactone [7]) or physical techniques(freeze drying and irradiation [8]) also demonstratedpoor durability. These considerations suggest that a re-

Presented at the Thirty-sixth Annual Meeting of The Society of ThoracicSurgeons, Fort Lauderdale, FL, Jan 31–Feb 2, 2000.

Address reprint requests to Dr Goldstein, CryoLife, Inc, 1655 RobertsBlvd, NW, Kennesaw, GA 30144; e-mail: goldstein.steven@cryolife.com.

Doctors Steven Goldstein, Steven P. Walsh, and KirbyS. Black are employees of Cryolife, Inc. Doctors DavidR. Clarke and Mark F. O’Brien are paid consultants ofCryoLife, Inc.

© 2000 by The Society of Thoracic Surgeons 0003-4975/00/$20.00Published by Elsevier Science Inc PII S0003-4975(00)01812-9

placement valve design combining a stable leaflet con-nective tissue matrix with a viable cellular componentwould be optimal for producing long-term durability.

Additionally, a biologically engineered heart valveshould have performance characteristics similar to thenatural valve. Glutaraldehyde-crosslinked heart valveleaflets and conduit are markedly stiffened. Associatedwith the altered biomechanical characteristics are de-monstrable changes in leaflet motion, which produceabnormal stress patterns and cause buckling, acceleratedcalcification, and eventual tissue failure. Fixation alsoaffects the interaction of the leaflet/conduit unit resultingin limited valve orifice opening, which impairs valveperformance. The normal aortic valve mechanics mini-mize leaflet stress during the cardiac cycle, especially asimposed at the commissural posts and leaflet free mar-gins. A substantial advantage in terms of valve perfor-mance and durability should be anticipated with the useof non-glutaraldehyde-crosslinked tissue (normalmatrix).

To optimize replacement heart valve characteristics,w e h a v e d e v e l o p e d a d e c e l l u l a r i z e d ( n o n -glutaraldehyde-fixed) composite porcine aortic valve.This valve was designed to provide the same leaflet-conduit interaction that allows optimal valvular mechan-ics and hemodynamics found in the human aortic valve,while at the same time sufficiently reducing the immu-nologic potential of the porcine valve by removing cellu-lar constituents and soluble proteins. The low immuno-genic potential of collagen in tissues is clear [9–11], andan acellular connective tissue matrix was hypothesized tobe the basis of a stable replacement valve.

Valve design and performance was first assessed in anin vitro pulsatile flow loop system simulating physiologicaortic flow conditions. The premise of an immunologi-cally neutral matrix was evaluated by using a porcinetissue-based graft in a weanling sheep pulmonary im-plant model of 5 or more months’ duration. Finally, thevalves were implanted as intraaortic or pulmonary valvesin human recipients with excellent near-term (9 months)results.

Material and Methods

Valve PreparationPorcine hearts were transported from United States De-partment of Agriculture-approved slaughterhouses inphysiologic buffer. The noncoronary leaflets of the aorticvalves were visualized; those free of pathologic andanatomic anomalies were removed from the aortic valvealong with the attached anterior mitral leaflet and aorticconduit. Noncoronary cusp units were treated by a pro-prietary process designed to substantially reduce aorticleaflet cellularity. The steps included cell lysis, enzymaticdigestion of nucleic acids, and washout in neutral buffer.Trileaflet valved conduits (SynerGraft porcine heartvalve [SHV] (CryoLife Inc, Kennesaw, GA) were sewnfrom three noncoronary leaflet units, matched to opti-

mize leaflet symmetry and coaptation [12]. Completedvalves were cryopreserved, terminally sterilized, andmaintained in liquid nitrogen until implantation.

Hydrodynamic TestingHuman aortic valves and SHV were sewn into compliantsilicone tubes and mounted as intact roots into a modi-fied left heart model pulse duplicator (ViVitro Systems,Victoria, British Columbia, Canada). Flow evaluationswere performed using a blood analog at ambient tem-perature, with system parameters adjusted to obtainstandard adult aortic flow conditions (70 cycles perminute, 30% to 40% systolic fraction, nominal aorticpressures of 120/80 mm Hg, with mean aortic pressure of100 mm Hg). Stroke volume was adjusted to achieve flowrates from 2 to 7 L per minute. A minimum of three testvalves of each size and one equivalently sized referencevalve were studied under all simulated cardiac outputs.

Aortic and left ventricular and aortic flow rates weredigitally collected and averaged over 10 cycles. Addition-ally, average closure and leakage volumes (per stroke)were collected. High-speed video images of the outflowaspect were obtained (Phantom v.2.0; Vision Research,Wayne, NJ) using capture rates of 200 to 500 frames persecond. Mean pressure gradient versus root-mean-square flow rate (Qrms) or simulated cardiac output werecalculated and plotted. Effective orifice areas (EOA) weregenerated for each valve at each flow rate using themodified Gorlin equation, and were averaged to providea nominal EOA over the range of flow rates investigated.Closure volume and net retrograde flow per cycle werereported for each valve at each flow rate.

Sheep ImplantationPreclinical SynerGraft valve performance was evaluatedby implantation into the right ventricular outflow tract(RVOT) of nine 4- to 6-month-old female or neuteredmale weanling Suffolk sheep. The choice of model wasbased on accessibility of the implant site, its suitability tostentless valve implantation, and studies of bioprostheticvalve performance in similar models recently summa-rized [13]. All implanted valves measured 19 mm at outerannulus diameter (OD); the internal diameter (ID) was 2to 3 mm smaller depending on conduit wall thickness.For comparison with standard tissue valves used inpulmonary valve repair and replacement, 2 animals wereimplanted with a cryopreserved sheep aortic valve ofsimilar annular dimensions. Exposure of the RVOT wasobtained via a left thoracotomy, and the grafts wereimplanted as pulmonary valve conduit replacements.

All animals successfully weaned from bypass under-went open chest echocardiographic assessment of valvu-lar function immediately after implant. At 3 months, andagain just before sacrifice at 150 days, a closed chestechocardiogram was obtained with the animals standing.Complete examinations were obtained including four-chamber (two-dimensional [2D] and M-mode) and short-axis (2D and M-mode) views from the right side, short-axis views from the left side, and continuous-waveDoppler of the conduit from the left side. These views

1963Ann Thorac Surg GOLDSTEIN ET AL2000;70:1962–9 BIOENGINEERED HEART VALVE XENO-AUTOGRAFT

provided measurements of all wall thicknesses, chambervolumes, shortening fractions, and peak flow velocities,and allowed detection of any regurgitation. A singleanimal was evaluated at 150 days, and then allowed tosurvive until it died at 336 days of non-valve-relatedcauses.

Hemodynamic studies were performed upon conclu-sion of the implant procedure and on all successfulimplants with the animal in the left decubitus positionjust before sacrifice and necropsy. A Swan catheter waspassed through the jugular vein into the right heart andpulmonary artery to measure right atrium, right ventri-cle, pulmonary artery, and pulmonary wedge pressures.The same catheter was used to obtain cardiac output bythermodilution, and after repositioning, to determinetransvalvular pulmonary valve gradient. The valve orificearea was determined utilizing a modified Gorlinequation.

All sheep involved in this study received humane carein compliance with the “Guide for the Care and Use ofLaboratory Animals” prepared by the Institute of Labo-ratory Animal Resources and published by the NationalInstitutes of Health (NIH Publication No. 86-23, revised1985).

Histology/ImmunohistochemistryAt sacrifice, a portion of each leaflet was submitted forhistologic evaluation, while a separate region was equil-ibrated in 20% sucrose in phosphate-buffered saline for 1hour at 15°C, then trimmed and frozen in embeddingcompound (Tissue-Tek O.C.T. Compound, Sakura Fi-netek U.S.A., Inc, Torrance, CA) for cryosectioning andimmunostaining. Antibodies for immunohistochemistryincluded monoclonal antibody to class II major histocom-patibility antigen (H42A; VMRD, Inc, Pullman, WA),monoclonal antibody to fibroblastoid cells (RCV508B;VMRD, Inc), and monoclonal antibody to smooth muscleactin (VMRD, Inc). Specimens were scored on a five-point system (0 5 minimal, 5 5 high incidence).

StatisticsThe significance of differences between treatment groupswas evaluated by analysis of variance using the statisticalprogram for the IBM-PC (SPSS for Windows, v. 8.0).Means with differences of p less than 0.05 were judgedsignificant.

Clinical PerformanceFive patients have been implanted with the SHV at ThePrince Charles Hospital, Brisbane, Australia. The studyprotocol was reviewed and approved by the Hospital EthicsCommittee, and Clinical Trail Notification was filed withthe Therapeutic Goods Administration of Australia.

Results

In Vitro Hydrodynamic PerformanceHigh-speed video images of both SHVs and humanaortic valves showed smooth and symmetric leaflet mo-

tion with annular dilation and outward movement ofthe commissures during systole to yield a large floworifice. Valve closure was likewise smooth and sym-metric, with no observable central insufficiencythrough diastole. Closure volumes for SHV were lessthan 5 mL per stroke, with no observable leakage (lessthan 1 mL per stroke). As seen in Figure 1, meanpressure gradients for the tested SHVs were low, andequivalent to size-matched human aortic valves. For allvalves evaluated, mean pressure gradients were,10 mm Hg at peak simulated cardiac output. Calcu-lated EOA (modified Gorlin equation) yielded valuesof 2.52 and 2.40 cm2 for the 19-mm OD/16-mm IDSynerGraft and 16-mm ID human aortic valves, respec-tively. Under similar flow conditions, the calculatedEOA for a 19-mm Hancock MO (model 250) stentedporcine valve was found to be 1.41 cm2.

Sheep RVOT ImplantsEleven sheep (7 female, 4 male) were successfully im-planted, two with control valves (cryopreserved, aorticallograft) and 9 with the bioengineered SHVs. The recip-ients of the control valves were slightly, but not signifi-cantly, older and heavier (178 days, 61 kg) than thebioengineered valve recipients (160 days, 49 kg) at im-plant. Both control valve recipients were terminated at161 days postimplantation, while 6 of the composite valverecipients were explanted at an average of 150 days(range 147 to 156 days). There were three other compositevalve implants: 1 sheep died 2 days after surgery from afatal arrhythmia; the second died at 77 days from block-age of the RVOT as a consequence of thrombus forma-tion secondary to a mycotic infection; and the finalcomposite valve recipient died at 336 days of non-valve-related causes.

Fig 1. Comparison of in vitro valve performance as measured by themean peak systolic pressure gradient across the valve measured un-der simulated aortic conditions (19-mm OD/16-mm ID SynerGraftand 16-mm ID cryopreserved human aortic valve in a full rootconfiguration).

1964 GOLDSTEIN ET AL Ann Thorac SurgBIOENGINEERED HEART VALVE XENO-AUTOGRAFT 2000;70:1962–9

In Vivo Valve Performance: Sheep RVOT ModelECHOCARDIOGRAPHY. After implantation of the pulmonaryvalve graft and after weaning from cardiopulmonarybypass, 2D and M-mode ultrasound studies were per-formed on each animal through the left thoracotomyincision with the transducer placed directly upon theheart and/or great vessels. The long-axis and cross-sectional images of each sheep demonstrated excellentgraft leaflet mobility. A 2D cross section of the SynerGraftconduit at the level of the valve demonstrated a valvearea range of 1.82 to 3.18 cm2 in 8 sheep. The valve areaof the 2 sheep with the control bioprosthesis was 2.09 and3.57 cm2. These orifice areas were consistent with 15- to21-mm pulmonary valve sizes and were reasonable forthe conduit size implanted in each sheep. Valve orificesize determined by intraoperative ultrasound correlatedwell with Gorlin’s mathematical calculation of the valvesize.

A complete ultrasound examination was performedapproximately 3 months after device implant and at thetime of valve explant. These ultrasonographic studieswere performed with the animal awake and standing. Inevery instance but two, the sonographically determinedorifice size remained correlated to the conduit diameterat the time of implantation. During the ultrasound exam-ination, 22-satisfactory sonographic measurements of theconduit above the graft leaflets were obtained. Peakvelocity of pulmonary artery blood flow in these sheepranged from 1.8 to 2.77 m/s. There was no significantdifference between SynerGraft and allograftmeasurements.

The estimated peak pulmonary artery valve gradientrange of the sheep of this study as determined from theinterim and preexplant echocardiographic studies was 5to 25 mm Hg. The intraoperative gradient was lower andranged from 0.8 to 3.3 mm Hg, as would be expectedgiven the stress and the compromising affects of surgeryon cardiac output and the fact that the animals hadgrown significantly in the interim.

HEMODYNAMIC EVALUATION. At implantation and at explant,each animal had catheters placed through the left jugularvein into the right heart. Pressures were recorded fromthe right atrium, right ventricle, and pulmonary artery.Simultaneously, electrocardiography was recorded andcardiac output determined by thermodilution. The heartrate, valve orifice, and peak and mean gradient across theprosthetic heart valve were determined from recordedpressures and flows. The peak gradients measured di-rectly across the bioprosthetic pulmonary valve at thetime of implant and explant correlated well with thegradients determined by ultrasound. When measureddirectly, 2 sheep in the SynerGraft group had a measuredgradient of less than 5 mm Hg, and the remaining 5sheep had gradients between 10 and 19 mm Hg. Asshown in Table 1, there was no statistical differencebetween average peak gradients or valve areas of sheepallograft implants and SynerGraft composite valves at150 days postimplant. The relatively high gradients maybe due to the fact that the animals were indeed growing

during this period. Valve areas at 150 days postimplantwere also the same.

MACROSCOPIC VALVE APPEARANCE. Explanted valves were vi-sually inspected and showed no evidence of dehiscence,hematoma, thrombi, vegetations, suture interactins,tears, abrasions, or structural lesions. There was nostatistical difference between the scores when comparingthe SHV and sheep allografts for these gross features aswell as for the minimal tissue overgrowth of the proximalanastomosis. Stiffness was noted in the distal conduit ofboth allografts and SynerGraft valves, with a mean scoreof 5 in the allografts and 3.7 in the SynerGraft valves. Thestiffness was confined to the distal conduit. The valveleaflets were pliable and coaptive with minimal visiblechanges from the time of implant.

MICROSCOPIC VALVE APPEARANCE. Figure 2 compares the rep-resentative histologic appearance of preimplant SHV andfresh porcine leaflets showing the effective removal ofvirtually all histologically evident cells (fibroblast, myo-fibroblast, and endothelial). After implantation, therewas a reappearance of cells within the SHV leaflets. Atthe earlier explant time (150 days), cells were foundmainly toward the base of the leaflets near the aortic wall.After 336 days, there was a more widespread distributionof cells, with cellular elements found 60% to 80% of the

Fig 2. Histology of porcine aortic leaflets before and after processingby SynerGraft technology (top panels, left and right) and after im-plantation of the processed tissue in sheep for 150 days (bottom, left)or 336 days (bottom, right). All specimens were stained with hema-toxylin and eosin (objective magnification, 103).

Table 1. Average Pressure Gradients and Valve Areas ofRight Ventricular Outflow Tract Implants at 150 Days inSheep

Peak Gradient(mm Hg)

Valve Area(cm2)

Allograft aortic valvedconduit (n 5 2)

11.8 2.2

SynerGraft porcineheart valve (n 5 6)

13.3 2.2

1965Ann Thorac Surg GOLDSTEIN ET AL2000;70:1962–9 BIOENGINEERED HEART VALVE XENO-AUTOGRAFT

distance to the free margin of the leaflets. The density ofcells after 336 days resembled that of the fresh leaflettissue.

In contrast to the SynerGraft processed leaflet tissue,cryopreserved allografts did not show extensive cellular-ity at explant. Although Figure 3 demonstrates that thecellularity of the preimplant sheep leaflets was similar tothat of the unprocessed porcine tissue, samples from150-day explants were cell free or had a cell populationlimited to leaflet margins. Unlike the SynerGraft pro-cessed tissue, it was impossible to define the donor orrecipient as the source of the cells in the allograft.

The cells within the SHV leaflets were predominantlyfibroblastic with a small (, 5%) population of lympho-cytes and monocytes (Fig 4). Similarly, at 150 days, asignificant percentage of the new cell population in themiddle of the leaflet matrix was expressing smoothmuscle actin (Fig 5). After 336 days, almost the entirety ofthe cells were fibroblasts or smooth muscle cells. Theleaflet structure was normal and leaflet thickness wasnormal throughout the length of the leaflet.

Clinical ImplantsSHVs have been successfully implanted in 5 patients aseither aortic or pulmonary valve replacements (Table 2).

Comment

Over 30 years ago, Mohri and colleagues [14] speculatedthat host repopulation of leaflets had the potential tocreate a “new viable leaflet of autogenous tissue utilizingthe transplanted nonviable valve skeleton.” Even thoughviable valves provided better long-term durability, theynoted that host repopulation was observed only when anonviable homograft was implanted. Since then, thecryopreserved homograft has proved to be the mostdurable of the unfixed tissue valves, but evidence forlong-term recellularization is scant [15]. The goal of totalrestoration of heart valve function with a living, self-repairing replacement has been elusive, and only partlyachieved by the use of the pulmonary autograft in theRoss procedure [16]. The regeneration of a viable valvewith autologous cells evidenced in the present study withthe SynerGraft heart valve suggests an alternative path-way to a living valve replacement.

The present studies expand the previous successes increating a cell-conductive matrix that generates a viablevalve via the in-growth of host cells. In our sheep model,we have shown a progressive recellularization of the

Fig 3. Immunohistochemistry and histology of allograft aortic leaf-lets before (left, sheep antibody stain) and after (right, hematoxylinand eosin stain) implantation for 150 days in sheep recipient (objec-tive magnification, 103).

Fig 4. Antifibroblast immunohistochemistry of representative SynerGraft processed porcine aortic leaflets after implantation in sheep for either150 (left, objective magnification, 103) or 336 days (right, objective magnification, 403).

Fig 5. Anti-smooth muscle actin immunohistochemistry of represen-tative SynerGraft processed porcine aortic leaflets after implantationin sheep for 150 days (objective magnification, 103).

1966 GOLDSTEIN ET AL Ann Thorac SurgBIOENGINEERED HEART VALVE XENO-AUTOGRAFT 2000;70:1962–9

porcine leaflet matrix. With time, this recellularizationtakes on the character of interstitial cells present in thenative leaflet, with fibroblasts and myofibroblasts as thepredominant cell phenotypes present. A small comple-ment of inflammatory cells present in the 150 explantleaflets had disappeared by the 336-day time point,suggesting that an early phase of remodeling had beencompleted by the later time.

The histological findings with the decellularized SHVcontrast with those of cryopreserved allograft tissuesused as controls. Though cellular at implantation, theexplanted allograft leaflets contained only narrow cellu-lar areas near the leaflet base and along the leafletmargins (Fig 3). Because the grafts were allogeneic, it wasimpossible within this study to assess whether the cellsoriginated in the implant or from the recipient. The cellnumbers and distribution appeared to be following thepattern observed with human allograft valve implants asreported by Mitchell and associates [17] and Armiger[18], who report a steady decline in the cellularity ofcryopreserved allografts eventually leaving a cell-free,but structurally intact, matrix. These findings imply thatthe cryopreserved allograft matrix is not cell conductive.

A viable cell complement is vital to the turnover ofextracellular matrix components of the leaflet [19]. Therelevant cells have been identified as having character-istics of fibroblasts [20] or myofibroblasts [21] with re-spect to their synthesis of fibrillar collagens and/orsmooth muscle actin. The synthesis of types I, III, and Vcollagen is considered most important to the long-termdurability of the leaflets, whereas the participation ofthese interstitial cells in the moment-to-moment contrac-tile function of the leaflets remains speculative [20].Myofibroblasts also participate in wound healing, pro-viding contraction or retraction phases of tissue remod-eling [22]. Thus, it is expected that the cells of therepopulated leaflets would contain cells having charac-teristics of smooth muscle cells (Fig 5) as well as fibro-blasts. Through immunohistochemical staining, actin-positive cells were typically found in the leafletspongiosa, representative of normal leaflet hierarchy. At150 days, a small complement of inflammatory cells werestill present in the leaflets, making it difficult to specifyeither a functional (contractile) role for these cells, or amatrix remodeling role from a porcine to sheep collage-

nous base. However, it is important to note that smoothmuscle cells are a part of the cell complement of normalheart valve leaflets, and the implanted SynerGraft matrixwas able to serve as the scaffold for in vivo recellulariza-tion with appropriate cell phenotypes.

The SynerGraft porcine heart valve design incorpo-rates several attributes of human aortic heart valves thatconfer optimal functionality and performance. Amongthese are leaflet symmetry, native leaflet-conduit-sinusgeometry, and natural tissue biomechanics. The principalfocus of the in vitro hydrodynamic studies was to dem-onstrate the integrity of the valve design. In these inves-tigations, SynerGraft porcine heart valves viewed underaortic pulsatile parameters using high-speed digitalvideo recordings showed excellent leaflet symmetry andvalve function under all flow rates evaluated. Leafletmotion upon opening was even and smooth, with fullleaflet retraction to yield a large flow orifice. Dilation ofthe conduit was noted, resulting in leaflet-free margintensioning. Valves closed symmetrically showing no sig-nificant central insufficiency or retrograde flow and asmooth regular coaptive margin. In each instance, leafletbehavior was shown to be analogous to that found withhuman aortic valves. In contrast, stented fixed porcinevalves showed no systolic annular dilation and thus arestricted flow orifice, often with asymmetric leaflet mo-tion and significant folding of the free margin duringmaximal ejection.

Pressure gradients for SynerGraft porcine heart valveswere substantially lower than the same clinically sizedstented bioprosthetic heart valves (eg, 19-mm SHV and a19-mm stented porcine heart valve) over all flow ratestested. The measured gradients for the SHVs, and thustheir calculated effective orifice areas, were statisticallyindistinguishable from similar clinically sized humanaortic heart valves. Calculated EOA (modified Gorlinequation) yielded values of 2.52 and 2.40 cm2 for the19-mm OD/16-mm ID SynerGraft and 16-mm ID humanaortic valves, respectively. Under similar flow conditions,the calculated EOA for a 19-mm Hancock MO (model250) stented porcine valve was found to be 1.41 cm2. Theability of the compliant SHV to respond to changingpressures of the cardiac cycle results in normal valvularfunction and flow efficiency equal to the native aorticvalve. For larger sized SHVs and human aortic valves

Table 2. Clinical Implant Summary

Patient 1 Patient 2 Patient 3 Patient 4 Patient 5

Diagnosis Calcific AS Rheumatic AI AS/bicuspid PS RVOTreconstruction

Implant Aortic subcoronary Aortic subcoronary Aortic subcoronary Pulmonary root Pulmonary rootSize (mm OD) 25 25 23 23 2312-week follow-up Gradient peak:

13 mm Hgmean: 7 mm HgMild AI

Gradient mean:5 mm HgnonprogressiveAI ejectionfraction improved

Intraoperativegradients:peak: 17 mm Hgmean: 8 mm Hg

AI 5 aortic insufficiency.

1967Ann Thorac Surg GOLDSTEIN ET AL2000;70:1962–9 BIOENGINEERED HEART VALVE XENO-AUTOGRAFT

(annular diameters 23 mm and greater), mean pressuregradients were very low, often below 5 mm Hg, atsimulated cardiac outputs of 6 L/min and above. Thesimilarities in tissue mechanics and valve geometry be-tween the SHVs and native human aortic valves resultedin this close matching of the measured hydrodynamicperformance.

Typical of all trileaflet tissue valves, retrograde flowfractions of the SHVs evaluated were small, generallyless than 5% of the forward flow volume, and wereindependent of cardiac output. All regurgitation wasattributable to closure of the valve mechanism, with nomeasurable leakage through the closed valve or throughthe SHV conduit construction suture lines. As expected,closure volumes increased modestly as the valve sizeincreased. Variations in cardiac output or cycle rate didnot significantly affect closure volumes or leakage for agiven valve size. Regurgitant flow parameters for theSHVs were similar to size-matched human aortic valves.Based on these hydrodynamic findings, we concludedthat the design of the SHV produces performance char-acteristics indistinguishable from normal human aorticvalves.

The weanling sheep have proven to be an excellentmodel for evaluation of candidate heart valves, and itsrelevance to valve implantation in humans is well under-stood [23]. Cuspal calcification is a major cause of glutar-aldehyde-crosslinked valve failure in humans, and thisleaflet pathology can be demonstrated even after short-term (3-month) implants of typical fixed bioprostheses.We previously reported that the SHV leaflet undergoesno calcification over 5 months of implantation [12].Therefore, we anticipate that valve performance similarto that shown for the SHV in this study will be replicatedin human surgery. The short-term clinical experience ofthis valve in human patients (Table 2) lends credibility tothis assertion.

In summary, animal connective tissue can be madenonimmunogenic, and thus acceptable for xenografting,by means other than chemical crosslinking. As shown inthis report, the residual connective tissue matrix is stablein two xenogeneic species. This stability is not gained atthe cost of creating a crosslinked matrix into which cellscannot migrate. We found that the acellular matrix sup-ports recellularization with interstitial cells that appar-ently are capable of remodeling the implant. Whenchemical fixation is avoided, the natural biomechanicalproperties of the tissue are retained, thus enabling con-struction of a composite valve with human aortic valve-like properties. In total, these features demonstrate sig-nificant progress toward the engineering of an idealtissue-based replacement heart valve. The SynerGraftvalve is nonthrombogenic, nonhemolytic, has excellentperformance characteristics, and repopulates in vivo tocreate a viable matrix. As a result, this valve designprovides the greatest opportunity to clinically demon-strate extended durability and growth, two characteris-tics long sought after in replacement heart valves.

The authors wish to acknowledge the excellent technical assis-tance of Carriebeth Bair, Stacey Bode, Joseph Hamby, SandraKalmbach, and Karen Sylvester.

This work was supported, in part, by grant numberIR44HL53088-02 from the National Institutes of Health to Cryo-Life, Inc.

References

1. Binet JP, Duran CG, Carpentier A, Langlois J. Heterologousaortic valve transplantation. Lancet 1965;ii:1275.

2. O’Brien MF, Clarebrough JK. Heterograft aortic valve trans-plantation for human valve disease. Aust Med J 1966;2:228–230.

3. Carpentier A, Lemaigre G, Robert L, Carpentier S, Dubost C.Biological factors affecting long-term results of valvularheterografts. J Thorac Cardiovasc Surg 1969;58:467–83.

4. Nimni ME, Cheung D, Strates B, Kodama M, Sheikh K.Chemically modified collagen: a natural biomaterial fortissue replacement. J Biomed Mat Res 1987;21:741–71.

5. Spray TL, Roberts WC. Structural changes in porcine xeno-grafts used as substitute cardiac valves. Fross and histologicobservations in 51 glutaraldehyde-preserved hancock valvesin 41 patients. Am J Cardiol 1977;40:319–30.

6. Grimm M, Eybl E, Ing D, et al. Biocompatibility of aldehyde-fixed bovine pericardium. J Thorac Cardiovasc Surg 1991;102:195–201.

7. Barnes RW, Rittenhouse EA, Mohri H, Merendino KA. Aclinical experience with the betapropiolactone-sterilized ho-mologous aortic valve followed up to four years. J ThoracCardiovasc Surg 1970;59:785–93.

8. Beach PM, Bowman FO, Jr., Kaiser GA, Parodi E, Malm JR.Aortic valve replacement with frozen irradiated homografts.Long term evaluation. Circulation 1972;40/41(Suppl I):29–35.

9. Villa ML, De Biasi S, Pilotto F. Residual heteroantigenicity ofglutaraldehyde-treated porcine cardiac valves. Tissue Anti-gens 1980;16:62–9.

10. Timpl R. Immunological studies on collagen. In: Ramachan-dran GN, Reddi AH, eds. Biochemistry of collagen. NewYork: Plenum Press, 1976:319–75.

11. Schwick HG, Heide K. Immunochemistry and immunologyof collagen and gelatin: Modified gelatins as plasma substi-tutes. Bibltlaematol 1969;35:111–25.

12. O’Brien MF, Goldstein S, Walsh S, Black KS, Elkins R, ClarkeD. The SynerGraft valve: a new acellular (nonglutaralde-hyde-fixed, tissue heart valve for autologous recellulariza-tion. First experimental studies before clinical implantation.Sem Thorac Cardiovasc Surg 1999;4(Suppl 1):194–200.

13. Herijgers P, Ozaki S, Verbeken E, et al. The No-Reactanticalcification treatment: a comparison of Biocor No-ReactII and Toronto SPV stentless bioprostheses implanted insheep. Sem Thorac Cardiovasc Surg 1999;11(Suppl 1):171–5.

14. Mohri H, Reichenbach DD, Barnes RW, Merendino KA. Ho-mologous aortic valve transplantation. Alterations in viableand nonviable valves. J Thorac Cardiovasc Res 1968;56:767–74.

15. O’Brien MF, Stafford EG, Gardner MAH, Pohlner PG, McGrif-fin DC. A comparison of aortic valve replacement with viablecryopreserved and fresh allograft valves, with a note on chro-mosomal studies. J Thorac Cardiovasc Surg 1987;94:812–23.

16. Elkins RC, Knott-Craig CJ, Ward KE, McCue C, Lane ML.Pulmonary autograft in children: realized growth potential.Ann Thorac Surg 1994;57:1387–94.

17. Mitchell RN, Jonas RA, Schoen FJ. Structure-function corre-lations in cryopreserved allograft cardiac valves. Ann ThoracSurg 1995;60:108–12.

18. Armiger LC. Postimplantation leaflet cellularity of valveallografts: are donor cells beneficial or detrimental? AnnThorac Surg 1998;66:S233–5.

19. Schneider PJ, Deck JD. Tissue and cell renewal in the naturalaortic valve of rats: an autoradiographic study. CardiovascRes 1981;15:181–9.

20. Messier RH, Jr., Bass BL, Aly HM, et al. Dual structural and

1968 GOLDSTEIN ET AL Ann Thorac SurgBIOENGINEERED HEART VALVE XENO-AUTOGRAFT 2000;70:1962–9

functional phenotypes of the porcine aortic valve interstitialpopulation: characteristics of the leaflet myofibroblast.J Surg Res 1994;57:1–21.

21. Bairati A, DeBiasi S. Presence of a smooth muscle system inaortic valve leaflets. Anat Embryol (Berl) 1981;161:329–40.

22. Desmouliere A. Factors influencing myofibroblast differen-

tiation during wound healing and fibrosis. Cell Biol Int 1995;19:471–1.

23. Ouyang DW, Salerno CT, Pederson TS, Bolman RM, III,Bianco RW. Long-term evaluation of orthotopically im-planted stentless bioprosthetic aortic valves in juvenilesheep. J Invest Surg 1998;11:175–83.

DISCUSSION

DR DAVID A. FULLERTON (Chicago, IL): Doctor Clarke, doyou envision that the collagen matrix is a permanent structure oris that replaced as well?

DR CLARKE: Well, I think in order for the durability to bemaintained, that has to be replenished over time. I also thinkthat we still have some problems to deal with, particularly in theaortic conduit. The leaflets seem to be doing extremely well,however, there was some slight calcification in the conduits. Itwas not nearly as much as in the wall of the homograft conduit,however.

DR AXEL HAVERICH (Hannover, Germany): You may knowthat we follow a similar concept in our laboratory, with theexception that we try to recellularize those porcine matrices exvivo prior to implantation. We feel this might be the betterapproach because we have full endothelialization of the graft atthe time of implantation. My questions for you would be: Didyou see full reendothelialization of the grafts in all anatomicalaspects at the time of explantation long term, and did you seecalcification of the matrix or the valve leaflets in your long-termimplants? Thank you.

DR CLARKE: In answer to your first question, no, we did not seeextensive endothelialization. The cells were identified as almostexclusively fibroblasts and smooth muscle cells. In terms ofcalcification, there was absolutely no calcification observed in

the leaflet portion of the graft at all. The only calcification thatwas observed was in the aortic wall, and this calcification, as Imentioned, was significantly less than the calcification in thecorresponding controls.

DR RICHARD D. WEISEL (Toronto, Canada): This is truly aremarkable way of approaching things. Have you tried the valvein the immature sheep to see if it does in fact grow?

DR CLARKE: These were weanling sheep, and, as a matter offact, there is some evidence that in this model they did not reallygrow since the gradients increase, and that may be because thevalve did not increase in size as the animal grew.

DR WEISEL: Did you see any clots in the valves at all?

DR CLARKE: The only one that clotted was the 77-day death,which probably clotted as a result of a fungal infection at thetime of surgery, and it had a thrombosis in the conduit, but therewas no evidence of thrombosis in any of the others.

DR WEISEL: Do you think there will be endothelializationeventually?

DR CLARKE: I do not know the answer to that. It is certainlypossible.

1969Ann Thorac Surg GOLDSTEIN ET AL2000;70:1962–9 BIOENGINEERED HEART VALVE XENO-AUTOGRAFT