|Home | About | Journals | Submit | Contact Us | Français|
This study assesses in a baboon model the hemodynamics and HLA immunogenicity of chronically implanted bioengineered (decellularized with collagen conditioning treatments) human and baboon heart valve scaffolds.
Fourteen baboons underwent pulmonary valve replacement, eight with decellularized and conditioned (bioengineered) pulmonary valves derived from either allogeneic (N=3) or xenogeneic (human) (N=5) hearts; for comparison, six baboons received clinically relevant reference cryopreserved or porcine valved conduits. Panel reactive serum antibodies (HLA Class I&II), complement fixing antibodies (C1q binding), and C-reactive protein titers were measured serially until elective sacrifice at 10 or 26 weeks. Serial transesophageal echocardiograms (TEE) measured valve function and geometry. Differences were analyzed with Kruskal-Wallis and Wilcoxon Rank Sum. P≤ 0.05 significant.
All animals survived and thrived, exhibiting excellent immediate implanted valve function by TEE. Over time, reference valves developed smaller indexed effective orifice areas, EOAI=0.84(1.22) cm2/m2 median (range) while all bioengineered valves remained normal, EOAI=2.45 (1.35) cm2/m2; P=0.005. None of the bioengineered valves developed elevated peak transvalvular gradients, 5.5(6.0) versus 12.5(23.0) mmHg, P=0.003. Cryopreserved valves provoked the most intense antibody responses. Two of five human bioengineered and two of three baboon bioengineered valves did not provoke any Class I antibodies. Bioengineered human (but not baboon) scaffolds provoked Class II antibodies. C1q+ antibodies developed in four recipients.
Valve dysfunction correlated with markers for more intense inflammatory provocation. The tested bioengineering methods reduced antigenicity of both human and baboon valves. Bioengineered replacement valves from both species were hemodynamically equivalent to native valves.
There is increasing interest in subhuman primates for biomedical research.1 However, they are rarely used for evaluating bioengineered cardiovascular constructs. The Genus Papio (baboons) is potentially valuable for comparative tissue engineering research given significant similarities with humans in cardiac anatomy, physiology, valve dimensions, resident antigens, innate and acquired immune systems. Yet there are few reported baboon chronic survival cardiac surgical replacement valve performance studies.2, 3 A baboon orthotopic valve replacement model could be a significant preclinical tool for bridging cardiovascular tissue engineering to “real world” clinical applications. Papio hamadryas Anubis, an Old World monkey species, is genetically more similar to homo sapiens as compared to other non-hominoid primates and far more so than hoofed food stock ungulate mammals.(eFigure1) The latter are often used both as valve test models (e.g. sheep) and as source material (e.g. porcine) for bioprosthetic valves. International standards and regulatory guidelines for implantable cardiovascular devices typically require large animal preclinical in vivo validation studies.4 For evaluating bioprosthetic heart valves, the classical juvenile sheep model is typically chosen as it is very robust and sensitive for predicting valve structural failure due to dystrophic calcification.5 However, all ungulates express powerful non-HLA xeno-epitopes such as alpha-galactosyl to which all catarrhines (humans, apes and Old World Monkeys) possess natural antibodies. Conversely ungulates cannot be used to directly test human derived uncrosslinked tissue due to profound xenotransplant rejection.6 After removal of cells, it is not known whether heart valve ECM scaffolds remain functionally antigenic and proinflammatory across catarrhine species. Thus for both mechanistic research and possible regulatory utility, these studies were undertaken to evaluate the suitability of the baboon as a subhuman primate model for testing clinical prototype bioengineered heart valves.
The current clinical standard for pediatric valved conduit reconstructions is the cryopreserved, cadaver derived allograft “biologic” pulmonary valve, historically termed valve “homografts”. As traditionally prepared, donor cells are retained through surgical implantation, yet these valves do not grow, and the donor cell population disappears; durability is limited especially in infants and children, typically failing due to dystrophic calcification driven by chronic inflammation.7 Cryopreserved human homografts have been conclusively shown to contain HLA antigens capable of provoking antibodies in recipients.8 To reduce inflammatory responses, various tissue decellularization treatments of allogeneic tissues have been devised and some show promise.9-11 Unlike allografts, the clinical results with decellularized porcine valve xenografts have been very poor with accelerated inflammatory destruction.12, 13 This research utilizes valve scaffolds bioengineered (decellularized and conditioned) to be design optimal, minimally pro-inflammatory and attractive biologically to cells both in vitro and in vivo.14, 15 As tested here, these are not “tissue engineered valves” (TEHVs). TEHVs are a subset of bioengineered in which the scaffolds are “seeded” with cells in a bioreactor before implantation.
The bioengineered valve scaffolds tested in these studies are acellular but otherwise inherently similar to native valves in design, hemodynamic performance, matrix chemistry and material properties.10, 16, 17 As such, they could function as the platform or scaffold for a tissue engineered valve, and since a scaffold without cells at the time of surgical implantation is the “worst case” scenario for a tissue engineered construct in which seeded cells have failed to repopulate, it is the appropriate starting point for the development of new evaluative tools for preclinical assessment of tissue engineered heart valves.
Fourteen healthy post pubescent male baboons weighing 28-34 kg were selected as recipients. Donors were unrelated animals 2-4 kg larger than the recipients. Anesthesia and cardiopulmonary bypass methods were refined as previously reported by us.18 All recipients had excision of native valves and main pulmonary artery segments, then test valve conduits were inserted with standard homograft surgical techniques using end-to-end anastomoses. Five were replaced with human bioengineered pulmonary valves for 10 weeks (N=3) or 26 weeks (N=2); three received bioengineered baboon pulmonary valves for a duration of 10 weeks. Three types of clinically analogous alternatives (N=6) were used as reference valves, cryopreserved pulmonary valves from humans (N=2), baboons (N=2) and stentless porcine (N=2) bioprostheses (size 21 Medtronic Freestyle®, Medtronic Corporation, Minneapolis, MN). Studies were performed with IACUC approval (SNPRC), in accordance with the Guide for the Care and Use of Laboratory Animals, (National Research Council, 2011).
Baboon pulmonary valves were harvested and cryopreserved with clinical methods.7 Seven cryopreserved human valves at end of storage time limits, were obtained from a clinical tissue bank (LifeNet Health, Virginia Beach, VA). Human and baboon bioengineered valves were prepared by subsequent decellularization using a previously reported laboratory developed, double-solvent, multi-detergent, enzyme-assisted, reciprocating osmotic cell fracture method optimized for pulmonary valves.10 Prior to implantation, all decellularized valves were conditioned in solutions designed to acidify tissue pH, rehydrate the collagen helix moisture envelope, align and compact collagen fibrils, and restore soluble proteins to the matrix.16 These decellularized-conditioned valves are termed “bioengineered”.
Valve performance was evaluated by transesophageal (TEE) and dobutamine stress echocardiography with geometric dimensions and functional parameters measured and calculated as recommended for evaluation of prosthetic heart valves by the American Society of Echocardiography.19 A 5.0/ 3.7 MHz omni-plane TEE probe (Philips Medical Systems; Andover, MA) was positioned after induction of general anesthesia. Images were captured in real time on an HP SONOS 2500 platform (Philips Medical Systems) and recorded for quantitative off-line analyses. Baseline measurements were performed before cardiopulmonary bypass and after sternotomy closure during stable hemodynamics. Dobutamine was administered intravenously as a continuous infusion (Sigma Spectrum Volumetric Infusion pump, Sigma International Inc.; Medina, NY), with measurements repeated at dosages of 2 and 4 mcg/kg/min. Terminal measurements were similarly performed. Leaflet thicknesses were measured in triplicate by 2D and M-mode echocardiography.
Immediate pre and post operative as well as serial weekly post-operative serum samples were assayed qualitatively for the development of panel reactive antibodies using LABScreen® mixed Class I&II HLA bead assay (One Lambda, Inc., Canoga Park, CA, USA) run on Luminex 200 System (Luminex, Inc., Austin, TX) after serum purification (HiTrap, Amersham Biosciences, Buckinghamshire, England). All sera were also assayed with a second assay, LABScreen® PRA (One Lambda, Inc), to confirm positivity and calculate the clinically familiar percentage PRA developing to Class I&II antigens. This %PRA assay contains bead sites to 55 Class I and 32 Class II phenotypes and allows for some discrimination of responsible antigen specificities. Serum C–Reactive Protein titers were measured by ELISA at day 0 and days 7, 35, 70 postoperatively (Monkey CRP, Life Diagnostics, West Chester, PA). Class I complement fixing IgG antibodies were measured with a solid state assay (C1q Screen, One Lambda, Inc).
At necropsy, the heart lung block was removed and severity of pericardial adhesions scored (absent, mild, moderate, severe). Test valves were excised with cuffs of native tissue beyond both anastomoses for explant pathology evaluations. Explanted valves were fixed with HistoChoice® MB (Amresco, Solon, OH) and photographed without magnification (Sony® Cybershot, San Diego, CA) and at macroscopic 6.3x (SteREO Discovery V12, Carl Zeiss, Thornwood, NY) to document visual observations and imaged with radiographic methods for calcium mineral formation (Faxitron-LR, Lincolnshire, IL). Longitudinal sections were cut through each cusp, from the free edge to the base, including the corresponding sinus and arterial wall. Paraffin-embedded sections were used for routine hematoxylin and eosin (H&E) stains (American Histo Laboratory, Gaithersburg, MD). For immunohistochemistry (IHC), following antigen retrieval (10 minutes, 90°C; Antigen Unmasking Solu tion, Vector Laboratories, Burlingame, CA) deparaffinized sections of each valve were incubated overnight with a 1:25 dilution of CD79 (B-cell; mouse monoclonal; Dako) followed by application of a secondary antibody (alkaline phosphatase-conjugated; Vectastain ABC-AP Kits, Vector Laboratories). Slides were examined via light and fluorescence microscopy with digital camera and imaging software (Axio Imager.Z1; Axio Vision; Carl Zeiss). Multiple low magnification (2.5x) H&E images from each entire valved conduit were blindly reviewed (SLH) and the inflammatory response scored based on maximum percentage of tissue area per low power field infiltrated with inflammatory cells: minimal (0-10%), mild (10-25%), moderate (25-50%) or marked (50-100%).
Kolmogrov–Smirnoff test for normality indicated these data sets were not normally distributed so nonparametric tests were employed and values reported as median(range). Differences between groups were analyzed for continuous variables with Kruskal-Wallis and ordinal variables by Wilcoxon Rank Sum Tests. P≤0.05 was considered significant (SAS 9.2 and SPSS v.17); Bonferroni correction was applied for multiple comparisons.
All baboons were successfully weaned from CPB, survived and thrived. One recipient easily tolerated emergency open surgical removal of an intercurrent gastric bezoar (hair).
Compared to bioengineered, reference valves developed higher peak (20.0(29.0) versus 5.5(6.0) mmHg, P=0.0003) and mean (12.5(23.0) versus 3.5(4.0) mmHg, P=0.003) pressure gradients by 10 weeks post-implant (Table 1). None of the bioengineered valves developed elevated resting valve gradients (Figure 1, data and notes). The majority (5 of 6) reference valves developed cusp thickening (≥ 1.5 mm) and restriction in cusp excursion (Table 2). In contrast, mild cusp thickening was seen in only one bioengineered valve (animal #10), and all had normal cusp mobility. Bioengineered valves maintained stable normal effective orifice areas (EOA) to 10 and 26 weeks. Reference valves developed markedly decreased EOA as compared to the bioengineered valve group (0.72(1.08) cm2 versus 1.77(1.19) cm2; P=0.005). Bioengineered cusp dysfunctional regurgitation was typically trace to mild, but was moderate for one C1q+ recipient receiving a human derived bioengineered valve (animal #10).
Dobutamine stress testing demonstrated marked differences in EOA and EOAI between reference and bioengineered valves (Figure 1). Peak and mean transvalvular pressure gradients were consistently lower for all bioengineered valves. At maximum dobutamine stimulation, the reference valves had peak and mean pressure gradients of 44.5(76.0) mmHg and 29.5(49.0) mmHg, respectively. In contrast, the human bioengineered valves had functionally low valve gradients (12.0(12.0) mmHg peak and 7.0(8.0) mmHg mean) despite cardiac indexes ≥ 5 L/min/m2 at the maximum dobutamine stimulation (p≤0.005 for all comparisons). For all bioengineered valves, transvalvular gradients during stress echocardiography were comparable to those across native valves in each animal, suggesting the bioengineered valves were functionally equivalent to native valves. The geometric valve annulus diameters (Table 1) by direct TEE measurement after 10 weeks for bioengineered (1.76(0.54) cm and reference valves (1.75(0.52) were the same; P>0.05). Thus the reductions in EOA measured in reference valves were likely not due to anatomical shrinkage, but rather to increasing cuspal stiffness, restricted cusp opening and loss of conduit wall compliance.
The mixed screening assay accurately predicted whether Class I or II PRA titers would exceed 10% when measured by quantitative LABScreen® PRA % assay (Table 2). Both allogeneic and xenogeneic cryo reference valves provoked Class I&II antibodies, typically as “strongly positive”, beginning relatively early (weeks 1-6) after implantation. In contrast, 2/5 human and 2/3 baboon bioengineered valves did not provoke Class I antibodies. None of the bioengineered baboon valves provoked Class II antibodies while all human tissues did.
All cryopreserved donor valves provoked Class I&II titers exceeding 60% by week 10 (Table 2) regardless of species of origin. Baboon bioengineered valves were negative for both Class I&II elevations except for one animal (#9) that developed 75% Class I (and C1q+) by week 6 but which declined by week 10 to 47%. Two human bioengineered valves remained Class I negative (<10%) to six months, but all eventually provoked Class II antibodies. PRA time courses were variable (Figure 2): bioengineered valves that elicited positive PRA titers typically did so by 4 weeks and then returned to zero by 10 weeks, while cryopreserved valves (even allogeneic) tended to evoke more persistent elevations. Valves maintaining superior hemodynamics typically had absent or shorter elevations of PRA.
Both human and baboon bioengineered valves appeared to have virtually normal cusps qualitatively except for one human valve in a strongly Class II+ and C1q+ recipient (#10) that was described as “slightly thickened”. By echo, this valve's leaflets were measured as “mildly thickened” (Table 2). Only one valve (#3 – cryopreserved) had discordant leaflet scoring between echo and explant observations (compare Tables and e1), although this particular valve did exhibit gross findings of inflammatory degradation with sinus wall thickening (Table 2). In contrast, all but one reference valve exhibited cusp thickening quantitatively by echo (eTable 1). Frank calcifications were not observed by radiographic examination in any of the explanted valves. Inflammatory histologic scores demonstrated some variability yet tended to correlate with the PRA titers (Table 2 and eTable1). With the exception of a single baboon valve, all cryopreserved valves elicited a marked inflammatory infiltrative response (mimicking clinical experience) while only two human bioengineered valves did. Immunohistochemical staining confirmed the presence of B-cells in these inflammatory infiltrates (eFigure3).
The key findings of this study include: 1) bioengineering using valve decellularization and conditioning reduces antigenicity and prolongs normal functional valve performance especially well for baboon (allogeneic) but also for human (xenogeneic) sourced valve scaffolds; 2) the intensity of immune and inflammatory responses is correlated with valve functional durability; 3) prolonged and higher HLA titers appear to predict accelerated functional degradation; 4) C1q+ complement fixing antibodies were provoked by four different test valves, all developed 2+ to 3+ valve regurgitation; 5) technical implant surgery in baboons is analogous to human clinical surgery; 6) development of HLA antibodies in baboons can be quantified with commercially available solid state assays used in transplantation surgery. Our rationale for selecting the baboon was similarity to humans in ontogeny, phylogeny, immunology (eg, 4 IgG subclasses, complement fixation, macrophage, T and B cell functions), absence of α-Gal epitope, similar semilunar valve microstructure, semi-upright posture, lack of susceptibility to simian herpes B virus, and robust tolerance of cardiopulmonary bypass.18 Our data suggest that bioengineered cross species human to baboon valve scaffold transplants are different immunologically (eg, provoking Class II+ antibodies), than allo-transplants, but functionally much less provocative than cryopreserved valves of either species.9
The baboon valve diameters as measured at implant scaled allometrically to normalized (BSA) pediatric valve annulus diameters with values similar to human (Figure 3). The indexed effective valve orifice areae increased by the time of explant for bioengineered valves (papio ↑ 29%, human ↑ 19%), but decreased for cryopreserved (papio ↓ 68%, human ↓ 57%) and porcine (↓ 76%). The differences in resting pressure gradients across the clinical reference valves versus the bioengineered valves are likely under reported since the cardiac outputs at 10 weeks were generally higher in the animals with bioengineered valves. There was no echocardiographic evidence of impending bioengineered valve stenotic failure as transvalvular gradients did not increase over time, nor did calculated valve EOA's decrease significantly at either 10 weeks or 26 weeks post-implant. The dobutamine driven elevated cardiac outputs only provoked physiologic flow gradients comparable to expected values for human native valves. In contrast, echocardiographic evidence of critical dysfunction with decreasing EOA was manifest in all 3 types of reference valves even by 10 weeks suggesting that the baboon is a severe test species and robust model for such assessments. Dysfunction by TEE correlated with measurements of more intense PRA responses. Only one bioengineered valve (#10) demonstrated cuspal thickening. This recipient had a high Class II+ response and was Class I C1q+. The two valves that developed leaflet dysfunction with greater than 2+ regurgitation (#5, #10) were both associated with higher PRA titers and C1q+ antibodies. To our knowledge, correlation of the appearance of C1q complement fixing antibodies and transplanted biological valve dysfunction has not previously been reported. Intuitively, C1q+ antibodies may be implicated in more consequential antibody mediated clinical tissue transplant rejection and may thus be an observation perhaps worthy of evaluation clinically in homograft valve recipients exhibiting accelerated valved conduit dysfunction. Thus various indicators of the severity of provoked immunogenic inflammatory responses appeared to correlate with the severity of valve dysfunction, which decellularization and conditioning appeared to mitigate. These observations corroborate current mechanistic theories emphasizing the immune and inflammatory pathogenesis of valve degradation..20
Baboon and human HLA molecules have been compared and are almost 90% identical with cross-species differences concentrating in positions typically demonstrating polymorphisms in human alleles that serve to activate T cells.21 Cryopreserved valves provoked strong HLA I and II responses regardless of donor species. Interestingly, in the baboon model, as in human transplantation, suspect culprit antigen specificities tended to group with known shared epitopes (eg, B8 with B64 and B65; DQ7 with DQ8; B7 with B42).22 The duration of the PRA elevations was more sustained for cryo-implants, but when occurring in recipients of bioengineered valves, the transient elevations, were as brief as five weeks duration, highlighting the need for well timed blood draws to identify sensitization. (Figure 2) This has significant implications for the interpretations of PRA titers in older reports and assay timing in future clinical studies.
The 10 week duration of these studies was too short to adequately assess in situ autologous in vivo recellularization but was intentionally chosen to best capture early immune-inflammatory responses to correlate with echo function and PRA titers. Only two recipients with human bioengineered scaffolds were kept to 6 months; both had persistent excellent valve functionality without MHCI+ or C1q+ antibodies, suggesting that when antigenicity is minimized, longer term hemodynamic performance studies are feasible. Why antigenicity can be variably retained at all in decellularized valves (albeit typically mild) is not clear from these studies. All mammalian semilunar valves have microscopic interdigitations of ventricular muscle deep within the fibrous annulus where some fragmentary cell remnants could remain despite 97-99% removal of DNA and other indicators of decellularization effectiveness. Epitopes could be associated with extracellular matrix (as seen with α-gal). Alternatively, mass spectrometry of detergent decellularized equine carotid arteries has revealed small residuals of over 300 cell associated proteins.23 These could include epitopes accounting for some retention of xenotransplant immunogenicity despite absence of cell fragments. Our complete cardiac experience with baboons has provided some unique observations (eTable 2) that may aid other investigators considering this species for cardiac preclinical studies.18 Cardiac surgery studies in baboons cost 500% more than sheep. As in this study, this limits feasible animal numbers per test group.
The need for chronic animal models that are especially relevant to humans has been identified by multiple authorities as critical to “bench to bedside” translation of tissue engineering strategies.1,2,24 A principal tenet in the field of tissue engineering is that the more a re-established cellular population resembles normal tissue, in cell density, location and phenotypes, the more likely it is that such tissues will effectively mimic native, particularly with the capacity for constructive and adaptive remodeling. One pathway to a “personal” heart valve would be to “tissue engineer” by repopulating a decellularized allogeneic scaffold with the putative recipient's own cells. This could be achieved by one or more strategies: 1) preimplant direct bioreactor cell seeding (classical tissue engineering); 2) employing physical, chemical, and biological conditioning treatments of scaffold matrix to promote cell adhesion and in-migration in vivo as an in situ postimplantation autologous process; 3) introducing into the scaffold “homing” molecules (“breadcrumbs” strategy) that accelerate in vivo cell seeding.6, 25 In this study, Method 2 was utilized without ex vivo cell seeding. However, regardless of technique, the base scaffold must be minimally pro-inflammatory or the result may be scar tissue not physiologic, healthy, reconstituted tissue structures. Previous research has demonstrated that autologous cell in-migration, even without preseeding, effectively repopulates decellularized conduit vascular walls.10 Cuspal re-endothelialization is improved with conditioning, but the matrix of all semilunar cusps of such valve scaffolds are not consistently (ie, all three cusps every time) fully repopulated with valve interstitial cells by just autologous postsurgical in vivo cell migration and proliferation.10, 16 This is an important distinction as repopulated cells are needed to provide matrix remodeling protein synthesis capacity for growth and repair, which suggests enhanced recellularization strategies could be useful.14, 25 Development of viable personal heart valves using any combination of these strategies will be strengthened by assessment methods designed to emulate putative clinical paradigms as closely as possible. Both clinical hemodynamic failure modes for biological replacement valved conduits (insufficiency due to leaflet dysfunction and stenosis with progressive gradients) were replicated in this model. Thus, the baboon could be useful for both mechanistic studies and as a bridge from traditional large animals to clinical trials of tissue engineered cardiovascular constructs.
The bioengineering processes employed appeared to reduce antigenicity of semilunar valve scaffold tissues and prolong functional performance, especially within species, but also across genera from two Primate Families of high genomic congruence. Absent or evolving postoperative MHC antibody titers could be assessed and appeared to correlate with functional measurements. This experience suggests that the baboon model may be useful in the future for chronic functional testing of clinical prototype human bioengineered or even “tissue engineered” replacement valves for which human decellularized scaffolds could be recellularized with recipient (ie, baboon) cells to closely simulate clinical paradigms.
This research was supported in part by a subgrant award from National Institutes of Health, Southwest National Primate Research Center base grant P51RR013986-11.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.