A biomimetic multilayered polymeric material designed for heart valve repair and replacement

Abstract

Materials currently used to repair or replace a heart valve are not durable. Their limited durability related to structural degeneration or thrombus formation is attributed to their inadequate mechanical properties and biocompatibility profiles. Our hypothesis is that a biostable material that mimics the structure, mechanical and biological properties of native tissue will improve the durability of these leaflets substitutes and in fine improve the patient outcome. Here, we report the development, optimization, and testing of a biomimetic, multilayered material (BMM), designed to replicate the native valve leaflets. Polycarbonate urethane and polycaprolactone have been processed as film, foam, and aligned fibers to replicate the leaflet’s architecture and anisotropy, through solution casting, lyophilization, and electrospinning. Compared to the commercialized materials, our BMMs exhibited an anisotropic behavior and a closer mechanical performance to the aortic leaflets. The material exhibited superior biostability in an accelerated oxidization environment. It also displayed better resistance to protein adsorption and calcification in vitro and in vivo. These results will pave the way for a new class of advanced synthetic material with long-term durability for surgical valve repair or replacement.

1. Introduction

Valvular heart disease (VHD) affects more than 100 million people worldwide and is associated with substantial morbidity and mortality [1–3]. Although the incidence of VHD is high, current therapeutic approaches are limited to valve replacement or repair, either through surgical or percutaneous approaches [4,5]. Two types of artificial valves are commercially available for valve replacement: mechanical and bioprosthetic valves. Mechanical valves have higher durability, but their rigid metal leaflets and limited geometric design create non-physiological flow and predispose to thrombosis and related embolic events [6]. Consequently, patients treated with mechanical valve replacement require lifelong anticoagulation therapy and thus are at increased risk of adverse events related to bleeding. Moreover, the complication specific to mechanical valves are devastating, and the permanent neurological sequelae of embolic or hemorrhagic stroke can result in catastrophic changes in the patients’ life [7–9]. Bioprosthetic valves, on the other hand, are made with fixed exogenous cardiac tissue and have better biocompatibility without lifelong anticoagulant therapy. However, their tissue processing conditions and inadequate material properties make them susceptible to calcification and structural valve degeneration (SVD) in 15 years, depending on patient ages, co-morbidities and demographics, etc. [10,11]. Durability issues are attributed to calcification and fatigue-induced structural deterioration of the tissue leaflets, caused in part by localized leaflet damage at stressed regions [12,13]. Valve repair becomes an alternative treatment option for patients with incompetent aortic valves, which preserves the native valve tissue, and decreases the risks of mortality and infection [14]. Such repairs frequently require the use of polymeric or tissue-derived patches, both of which have intrinsic limitations that affect their long-term durability and mechanical performance, leading to structural degeneration (SD) of the repaired valve leaflet and subsequent reoperation [15–23]. Pavy et al. summarized that the discrepancy between the elastic modulus of the patch and native tissue is linked to severe aortic stenosis and prosthetic material failure by alteration of pressure and flow dynamics throughout the valve [24,25]. This mismatch results in perturbed blood flow and turbulence, which may lead to thrombosis, and which also produces pulsatile mechanical stresses at anastomoses leading to suture line disruption [26]. Martin and Sun utilized a computational modeling to identify adverse impacts of these mismatched mechanical properties: leaflet substitutes with isotropic behavior and unmatched elastic property resulted in higher leaflet stresses and accelerated fatigue damage along the commissures and suture attachments [27,28]. This phenomenon is also in line with clinical findings that mechanical failures of bioprosthetic valves are often associated with leaflet tears near the commissures and suture attachments. Moreover, the adjacent native tissue has to biologically remodel itself to compensate for this mechanical discrepancy, and the resulting excessive remodeling (thickening) narrows the valve opening and occludes the blood flow [29].

Different groups have attempted to fabricate synthetic valve substitutes that can more closely resemble native tissue, but with limited success [30–33]. Early studies explored the use of polymeric materials such as silicon and polyurethanes due to their favorable mechanical properties and chemically defined composition, but clinical translation was hindered by premature degradation, thrombosis, and calcification [34]. Advances in polymer synthesis have recently led to new generations of polyurethanes with more favorable thrombogenic and calcification profiles, such as polycarbonate urethane (PCU) and polyether urethane urea (PEUU), that show clinical promise [35]. More recently, hydrogel-based valve substitutes fabricated from materials such as a polyvinyl alcohol (PVA) and poly(ethylene glycol) (PEG) have also been examined due to their flexible mechanical and biological properties [30, 31]. However, the degradable nature of hydrogels and lack of durability data makes this family of materials questionable as a long-term valve substitute [32]. In addition, many of these proposed polymers or polymer composites do not account for the unique three-layer architecture of the native leaflets. Native leaflets have a highly organized architecture with three specific layers. Leaflet mechanical stiffness and nonlinear stress-strain behavior are attributed to two surface fibrous layers: the fibrosa and ventricularis [36]. The middle layer, the spongiosa, accommodates the shear forces between the two surface layers and absorbs the load during valve opening and closing [37]. This unique architecture is critical to offering mechanical properties that withstand high trans-valvular pressures with low flexural stiffness. Masoumi et al. developed a tri-layered scaffold to mimic the structural and anisotropic mechanical characteristics of the native leaflet [33]. However, this tri-layered scaffold was fabricated from a poly (glycerol sebacate) (PGS)-polycaprolactone (PCL) composite that degraded at a fast rate with a significant loss of mechanical strength in 4 weeks. Despite aiming at replicating the valve architecture, this polymer composite was not a mechanically stable option for clinical use.

In summary, the reasons for those unideal performances of the academic trials may be attributed to these factors: 1) focusing on tissue engineering and regenerative medicine to create living autologous heart valve leaflets are still far from ideal despite many decades of research, mainly related to the lack of control of the balance between polymer degradation and tissue formation in-vivo: the mechanical properties of the applied degradable materials change over time, leading to a loss of mechanical stability and a progressing mismatch between artificial leaflet substitutes and the native leaflets, if the degradation and regeneration are unexpected or out of control; 2) some attempts did not fully replicate the complex structure-function relationship of the native valves, and neglect the essential role of the architecture in the functionality; 3) mechanical data such as ultimate tensile strength and break strain, were recorded at a higher strain level than the physiological level, which is not appropriate to assess the mechanical performance under physiological conditions; 4) the impact of the cyclic deformation that the material is exposed in a valve leaflet position was not fully studied; Hence, there continues to be a significant need for the development of durable valve materials that can mechanically and structurally mimic native tissue, and ultimately improve the outcomes of the surgical and transcatheter treatment of valvular heart disease patients. In this present work, we aim to fabricate a stable, functional and biomimetic material based on advanced polymer composites. Our biomimetic, multilayered material (BMM) was fabricated with a biostable polymer as the main composition, using a combination of solution casting, electrospinning and lyophilization techniques to form three distinct layers. We expect to tailor the proper mechanical properties and anisotropic performance of BMM, through fabricating the composite with the designed structure replicating the architecture of the native valve, in order to maintain its matched properties and durable functionality in the long term. This material may possess potential for translation towards new cardiovascular patches for surgical valve repair and new polymeric heart valve prostheses for surgical and transcatheter valve replacement.

2. Materials and method

2.1. Materials

Carbothane^™ AC-4075A, Polycarbonated-based polyurethane (PCU) (Lubrizol, Wilmington, Massachusetts, Mw = 480 kDa) was dissolved in dimethylacetamide (DMAC) (Acros Organics, Fair Lawn, New Jersey). Polycaprolactone (PCL, Mw = 80,000; Sigma-Aldrich, St. Louis, Missouri) was used to create fibers and dissolved with a mix of chloroform (Sigma-Aldrich, St. Louis, Missouri) and methanol (Fisher Scientific, Hampton, New Hampshire) with a 3:1 M ratio. Three commercially available patches were selected for comparison: Gore-Tex^® (W. L. Gore and Associates, Flagstaff, Arizona, USA), CorMatrix^® (Cardiovascular, Inc, Atlanta, Georgia, USA) and CardioCel^® (Admedus, Toowong, Queensland, Australia). The CryoValve^® aortic human valve (CryoLife Inc., Kennesaw, Georgia, USA) was used as the control sample, after being dissected and kept intact in PBS.

2.2. Fabrication of the biomimetic multilayered material (BMM)

2.2.1. Fibrosa-mimic layer and ventricularis-mimic layer fabrication

Using a 15% PCL solution prepared in a mixed solvent, PCL fibers were produced by electrospinning with the following parameters: a flow rate of 1 mL/h, a voltage of 20 kV voltage and a distance of 15 cm between the nozzle and drum collector. The solution was spun towards a rotating collector at a rate of 1600 rpm to collect the aligned fibers. The fibers were dried overnight in a chemical hood for solvent evaporation. The collected, aligned PCL fibers were embedded in a solution-casted PCU film. The 15% PCU solution was casted by a doctor-blade coater through a 500 μm gap to control the film thickness. The fiber-solution composites, fibrosa-mimic (F-mimic) layer and ventricularis-mimic (V-mimic) layer, were cured overnight in a chemical hood to evaporate the solvent and form the fiber-enhanced layers.

2.2.2. Spongiosa-mimic layer fabrication

15% PCU solution was casted by a doctor-blade coater to create a film with a fixed thickness of 1500 μm. Subsequently, the film was immersed in deionized water for 24 h in order to replace the solvent with water. The film was frozen at − 80 °C and lyophilized at 0.1 mBar and − 40 °C for 72 h, leading to the formation of a porous structure (or foam) [38].

2.2.3. BMM fabrication

The F-mimic layer was used as the bottom layer of the BMM. It was fabricated first as described above. After 1 h drying in the hood, the spongiosa-mimic (S-mimic) layer was placed over the half-cured composite and fully cured with this F-mimic layer overnight. Then this two-layer composite was stacked on top of the V-mimic layer to fabricate the three-layered BMMs using the same strategy.

2.3. Morphology characterization

The specimens (PCL aligned fibers, each mimic layer and the BMMs) were sputter-coated with gold/platinum and imaged with a Zeiss Sigma VP scanning electron microscope (SEM) at an accelerating voltage of 3 kV. SEM images were used to assess the fibers’ orientation, mimic layers and the BMMs’ structures and surface morphology.

2.4. Tensile mechanical tests

Heart valve leaflets are subjected to a cycle of mechanical loadings like flexure (leaflet opening), shear (flow of blood through the valve), flexure (leaflet closing) and tension (leaflets suppressing back-flow of blood) [39]. The tissue-level mechanical properties of native valve leaflets are highly anisotropic and reflect the directional arrangement of collagen and elastin fibers. Tensile tests, thus, are widely used to assess tissue or tissue substitutes along either circumferential or radial direction. Tensile mechanical tests were performed using an Instron 5848 mechanical tester with a 50 N load cell at a strain rate of 10% s^{− 1}. The specimens were cut as 5 mm × 20 mm stripes (for non-tissue samples) or 3 mm × 10 mm ones (for the native tissue samples) in two different directions, circumferentially (C-direction) and radially (R-direction). The thickness was measured at three different points with a digital caliper (Mitutoyo 547–526s, Mitutoyo America Corp, Aurora, IL, USA) and the values were averaged. Four to six specimens for each sample were repeatedly stretched for 20 cycles, either to a maximal strain of 15% in the C-direction or to a maximal strain of 40% in the R-direction. Missirlis and Chong, Brewer et al., Thubrikar et al. and Li et al. have all reported in vivo AV leaflet strains of physiological level to be approximately 10–15% and 30–40% in the circumferential and radial directions respectively [39–42]. After the first 5 preconditioning cycles, the subsequent 15 cycles of stress-strain curves were recorded and averaged and the tensile modulus E were calculated as Equation (1) below:

$$E=\frac{\Delta \sigma }{\Delta \epsilon }$$ (1)

where $\Delta \sigma =\frac{\Delta F}{w_0{T}_0}$ and $\Delta \epsilon =\frac{\Delta l}{l_0}$ are engineering stress and engineering strain. $l_0$, $w_0$, $T_0$ are the dimensions (length, width and thickness) of the specimen, Δl is the change in length (elongation), and ΔF is the change in force. The average stress-strain curves and the tensile modulus at the strain of 15% or 40% were then used to compare mechanical performance in different directions and to assess anisotropy.

2.5. Flexural mechanical tests

Flexural mechanical tests are more oriented towards the physiological behavior of the valve leaflet. The flexural properties are a measure for the open-close movement of the cusp [43]. On the other hand, bulge test method is capable of repeatable measurements of the anisotropic nonlinear properties of human tissue [44]. Thus, flexural bulge tests, were first introduced to assess the flexural properties of the HAV, commercial patches and our BMM. Samples for the flexural mechanical tests were cut as planar specimens (n = 3 for each type of samples) with enough area to fully cover the test hole (diameter = 6 mm). Thickness was evaluated by averaging three measurements taken at specimen’s center with a digital caliper (Mitutoyo 547–526s). The specimens were speckled with black India ink to allow for digital image correction (DIC) tracking deformation and glued between two plates with holes of 6 mm diameter (Fig. S1A). For tissue specimens, the ventricular side was selected to face the camera. The embedded specimen was secured onto a custom inflation chamber through the holder (Fig. S1B).

Specimens were inflated by a custom-made displacement-driven syringe injection of PBS into the custom-made pressurization chamber. The pressure was monitored by a pressure transducer with a range of 0–8 kPa. The loading regimen was programmed using LabView (V2020, National Instruments, Austin, TX). The specimen was brought to a baseline pressure of 0.2 kPa and held for 30 s prior to cyclic testing to ensure the specimen was at equilibrium [44]. The specimens were subjected to 30 load-unload cycles at a rate of 3.5 kPa/s from the baseline pressure to a maximum pressure of 7.2 kPa (Fig. S1C) to mimic the leaflet deformation during the cardiac cycle. The deforming specimen surface was imaged by two stereoscopically arranged cameras with 20 mm focus lengths at an aperture of f/4. The optical axes of the cameras were positioned 35 cm above the chamber and fixed with a total angle of 12°. This configuration had a depth of field in front over 1.5 cm, sufficient to capture the deformation. Images were collected during testing at a rate of 10 Hz by VicSnap 2009 and correlated by Vic3D (V8, Correlated Solution, Inc. Columbia, SC, USA).

The flexural bulge test measured the components of the displacements in a 3D coordinate plane, providing the U, V and W components of the displacement field in X, Y and Z directions (Fig. S1D). The elastic modulus measured with this flexural bulge test, E_flex., was calculated through the change of the applied pressure (ΔP) and the change of the out-of-plane displacement component (ΔW). The sample in this test was modeled as a circular thin plate with edges fully fixed. The pressure was evenly distributed on the bottom surface of the sample. The governing equation and boundary conditions of this case could be expressed in cylindrical coordinates (r, φ, z) as [45]

$$\begin{gathered} {\nabla }^2{\nabla }^{2}w=\frac{-\Delta P}{D} \\ \text{s.t}.\begin{array}{c}w(R)=0\\ \phi (R)=0\end{array} \end{gathered}$$ (2)

where w is the displacement of z direction (defined as the out-of-plane direction) at a point of the thin plate, φ is the rotation, R is the radius of the plate, and ΔP is the change of the pressure exerted. D is the flexural rigidity defined as $\frac{E_{flex}{T}_0^3}{12\left(1-{\nu }^2\right)}$. is the thickness of the specimen. The solution to this equation is derived as

$$w(r)=\frac{-\Delta P}{64D}\left(R^2-r^2\right)$$ (3)

At the center point (r = 0), E_flex. could be expressed as

$$E_{flex}=\Delta P\Delta \frac{3R^4\left(1-{\nu }^2\right)}{16T_0^3\Delta W}$$ (4)

where ΔW is the change of the displacement in z direction. Here, all the materials were assumed to be incompressible, so the Poisson’s ratios ν were all set as 0.5.

2.6. Suture retention tests

The resistance to tearing is essential to evaluate the feasibility of patches or alternative materials. Suture retention strength was determined in accordance with the straight across procedure described in the American National Standard Institute – Association for the Advancement of Medical Instruments VP20–1994 [46]. Briefly, prolene 5–0 suture was inserted 2 mm from the end of the 10 × 15 mm specimen and through the specimen to form a half loop. The suture was pulled at a rate of 50 mm/min crosshead speed. Five specimens were tested in each group. The force (N) required to pull the suture through and/or cause the specimen to fail was recorded as the suture retention strength (SRS). A thickness normalized suture retention strength (TN-SRS, N/mm²) was also applied to eliminate the effect of sample thickness and needle size [47,48]. TN-SRS is calculated by dividing the suture retention strength by the area of the sample over which the load was applied [49]:

$$TN-SRS=\frac{SRS}{SutureThreadDiameter*SampleThickness}$$ (5)

and compared among all the samples.

2.7. Biostability tests

Specimens were pre-cut as 5 mm × 30 mm and submerged into 2 mL vials filled with an in vitro solution of 20% hydrogen peroxide (H₂O₂)/0.1 M cobalt chloride (CoCl₂) [50,51]. The in vitro solution was refreshed twice a week, and all tests were done at 37 °C. After a period of 5, 10, 14, 15, 20, 24, and 30 days, the specimens were removed, rinsed thoroughly in deionized water, dried in the hood, then cut into two parts (5 mm × 25 mm and 5 mm × 5 mm). The former was tested via the tensile tests and the modulus at strain = 15% was calculated. The latter was analyzed by SEM to inspect the surface quality. The specimen number of each customized samples (PCU film, PCU foam and BMM) at different time point is 6. The specimen number of each commercial samples (Gore-Tex^®, CardioCel^® and CorMatrix^®) at different time point is 4.

2.8. Biocompatibility tests

2.8.1. Bovine serum albumin (BSA) static protein-adsorption experiments

For static protein-adsorption tests, 1 mg mL^{− 1} BSA solution was prepared in PBS (pH 7.4). BMMs and three commercial patches were cut into specimens (50 mm × 10 mm, n = 5 for each specimen) and immersed in 10 mL 1 mg mL^{− 1} BSA solution in a test tube. BSA adsorption was conducted under vibration at 37 °C for 3 h to allow for adsorption equilibrium. Then the specimens were rinsed with PBS, and the remaining proteins adsorbed on the surfaces were removed with a 1 wt% aqueous solution of sodium dodecylsulfate (SDS), as described by Song et al [52]. The experiments were performed with five measurements for each specimen. BSA content was measured using a NanoDrop^™ spectrophotometer at a wavelength of 280 nm, and then the amount of adsorbed BSA on specimens was calculated.

2.8.2. Calcium deposition experiments

The calcium deposition experiments were performed in a metastable calcium phosphate (MCP) solution. The MCP solution has been previously described in detail [53]. In brief, 3.87 mmol (mM) CaCl₂, 2.32 mM K₂HPO₄ and 0.05 M Tris buffer were solved in 1000 mL of de-ionized water, to yield a Ca/PO₄ ratio of 1.67. This solution is more physiologically representative of hydroxyapatite, which is the most common form of calcium minerals in the vascular calcification process. BMM, PCU film, Gore-Tex^®, CardioCel^®, and CorMatrix^® were cut into disc specimens (diameter = 8 mm, n = 5 for each specimen) and immersed in 2 mL MCP solution individually. This experiment was conducted under vibration at 37 °C, and the solution was changed every 48 h to ensure an adequate ion concentration. The specimens were removed after 30 days and rinsed with water to remove excess solution and loosely attached deposits. The specimens were lyophilized overnight, weighed, and hydrolyzed in 1 mL of 6 N HCl overnight at 90 °C for decalcification. Once evaporated, the remaining solutes were dissolved in 0.6 N HCl for analysis. The calcium concentration was determined from HCl hydrolysate, using a calcium colorimetric assay.

2.9. Rat subcutaneous implant model

In accordance with NIH guidelines for the care and use of laboratory animals (NIH Publication #85–23 Rev. 1985), all animal protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of Columbia University (Protocol #AC-AABD5614).

Eighteen specimens (diameter = 8 mm) of PCU film (n = 6), Gore- Tex^® patch (n = 6) and CardioCel^® patch (n = 6) were implanted in the subcutaneous position of three rats. Following induction of anesthesia, fur clipping, standard sterile prepping and draping, six subcutaneous pockets were created on the dorsal surface of each rat. One specimen was implanted into each pocket, after which all wounds were re-approximated with surgical clips. The rats were sacrificed at 8 weeks with an overdose of isoflurane (Euthenase).

2.9.1. Histology

The implanted specimen was retrieved while still contained in host tissue, fixed in 10% neutral buffered formalin and processed using paraffin-embedding techniques. Slides were stained with Hematoxylin and Eosin and Alizarin Red stains. In each specimen, both the patch and the surrounding host tissue were evaluated.

2.9.2. Calcium content & mechanical test

Samples were analyzed for calcium content using calcium colorimetric assay as described in section 2.8. Briefly, the specimen disks were removed from host tissue, fixed in formalin and solvent-exchanged in DI-water. Following the lyophilization, the net weight of the specimen disks was acquired. After hydrolyzing in nitric acid, the calcium content was quantified (microgram calcium per milligram dry specimen weight). The PCU disks were separated from the host tissue after lyophilization. This specimen’s mechanical performance was also evaluated as described in section 2.4 and its tensile modulus at the strain = 15% was recorded, to compare with the control, unimplanted sample.

2.10. Statistical analysis

For studies including various groups of samples, like the tensile property studies, the suture retention tests, the biocompatibility studies (protein adsorption and calcium deposition), etc., two-sided t-tests for parametric data with Welch’s correction were conducted and used for analysis. The biostability studies, which included the same group of BMM samples, were analyzed using the one-way ANOVA followed by Tukey’s post-hoc tests (GraphPad Prism 7, San Diego, CA, USA). Results are showed as means ± standard deviation. P values less than 0.05 were considered statistically significant (*P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001).

3. Results

3.1. Structure

The BMM was designed as a tri-layer polymeric structure that was specifically developed to mimic the tri-layer anatomy of the native valve (Fig. 1A): an F-mimic layer, an S-mimic layer and a V-mimic layer. Fig. 1B shows that the aligned PCL fibers predominantly exist with a highly-orientated distribution, while random PCL fibers are electrospun with a random direction (Fig. 1C). The cross-section image (Fig. 1D) shows the aligned fibers were embedded in the PCU film to form two fiber-enhanced layers (F-mimic and V-mimic layers). PCU foam is used as the S-mimic layer showing a porous structure created after lyophilization in the cross-sectional view (Fig. 1E). The three mimic layers were tightly bound together via “glue”–PCU solution used for F-mimic layer and V-mimic layer–to form the BMM, with two fiber-enhanced layers outside and the foam layer inside (Fig. 1F).

Fig. 1.

Schematic drawing for the design of biomimetic, multilayered material (BMM) (A). SEM images of fiber morphologies of aligned fibers (B) and random fibers (C). The SEM image of the cross-section of the F/V-mimic layer, the white arrows and the enlarged image display the PCL fibers embedded in the PCU film (D). The cross-section of the S-mimic layer is displayed and shown its porous structure (E). BMM illustrates the tri-layer structure ‘Film-Foam-Film’ via the cross-section view (F), which parallels the design of BMM.

3.2. Tensile properties

Cyclic, uniaxial tensile tests were performed to assess the tensile properties of BMM and its component layers (Table S2). The aligned PCL fibers exhibit a highly anisotropic performance (35.74 ± 9.81 MPa vs 1.63 ± 0.38 MPa), compared to the random PCL fibers electrospun from the same solution (7.37 ± 0.30 MPa). Due to the incorporation of the aligned PCL fibers, the fiber-enhanced layers also demonstrate an anisotropic behavior, stronger along the fiber-aligned direction (green solid curve) and similar performance to pure PCU film along the fiber- perpendicular direction (green dash curve), shown in Fig. 2A. Conversely, the PCU foam, which is used to mimic the spongiosa layer of the native leaflet, exhibits a more compliant mechanical behavior and a significant lower tensile modulus (0.55 ± 0.22 MPa), compared to the PCU film (6.48 ± 0.17 MPa) made from the same solution (Fig. 2B). Combining three layers together, the BMM also demonstrates an anisotropic mechanical behavior and has a tensile modulus of 6.20 ± 1.83 MPa at 15% strain in the C-direction and 1.80 ± 0.21 MPa at 40% strain in the R-direction (Fig. 2E, Table S2).

Fig. 2. Representative stress-strain curves and tensile modulus of all samples.

Representative stress-strain curves of Random PCL fibers (black), Aligned PCL fibers (red) and Fiber-enhanced Film (green) (A). Representative stress-strain curves of PCU Film (red) and Foam (orange), which are fabricated from the same concentration of PCU solution (B). Stress-strain average curves of BMM, HAV and three commercial patches: Gore-Tex^®, CorMatrix^® and CardioCel^® along circumferential direction (C-Direction) (C) and radial direction (R-Direction) (D). Tensile modulus of all samples at strain level = 15% and 40%. *P < 0.05, **P < 0.01 and ****P < 0.0001 indicate significant difference between commercial patches and HAV (E). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Native tissues and commercial patches were also tested under the same conditions to compare with BMM. For native leaflets, the average stress-strain loading curves exhibit a residue deformation and then an increase in the slope of the stress-strain curves which is attributed to the deformation and stretch of fiber networks. This increase is accentuated in the C-direction compared to the R-direction because of the existence of oriented collagen fibers (Fig. 2C&D, black curves). The human aortic valve leaflets (HAVs) have a tensile modulus value of 16.34 ± 0.42 MPa in the C-direction, while a quite low modulus of 0.03 ± 0.01 MPa in the R-direction (Table S2 and Fig. 2E). On the other hand, three selected commercial patches are generally much stiffer and display a non- anisotropic behavior, with similar tensile curves at the same strain level in these two directions (Table S2 and Fig. S2). Gore-Tex^® is the most isotropic and the stiffest sample among the three commercial patches. CorMatrix^® and CardioCel^® are less stiff, and CardioCel^® even has a similar tensile modulus to those of HAV in the C-direction. The latter two samples are also relatively anisotropic given their biological nature and the presence of residual extracellular matrix fibers. Nevertheless, the commercial patches still possess much stiffer properties than HAVs, especially in the R-direction.

3.3. Flexural properties

Bulge tests were performed to assess the flexural properties of the HAV, commercial patches and our BMM. Table 1 summarized the data of thickness and displacement of specimens in the out-of-plane direction. The commercial patches generally possessed higher E_flex: Gore-Tex^® was the stiffest among the three commercial patches (16.73 ± 4.28 MPa) and CardioCel^® was the most compliant (4.25 ± 2.26 MPa). HAV was more compliant than commercial patches during the bulge tests. BMMs had a similar E_flex range (2.99 ± 2.43 MPa) as HAV (2.54 ± 1.22 MPa), and displayed better compliance than commercial patches. Commercial products have a generally stiffer performance than native tissues and BMMs, whether from tensile tests in two directions or from flexural bulge test (Fig. 3).

Table 1.

Flexural properties of BMM, native tissue and commercial patches.

Sample	Thickness(mm)	ΔW (mm)	Eflex (MPa)
BMM	0.688 ± 0.186	0.231 ± 0.166	2.99 ± 2.43
HAV	0.347± 0.038	1.196 ± 0.472	2.54 ± 1.22
Gore-Tex®	0.382 ± 0.011	0.120 ± 0.031	16.73 ± 4.28
CorMatrix®	0.309 ± 0.109	0.366 ± 0.030	9.89 ± 0.98
CardioCel®	0.364 ± 0.101	0.722 ± 0.137	4.25 ± 2.26

All data is acquired from the 16th loading-unloading cycles during the bulge test; n = 3 for each type of samples.

Fig. 3.

The summary of tensile modulus (red and blue bars) and E_flex (orange line) of all tested samples. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.4. Suture retention of samples

The resistance to tearing of the BMM and its main component, PCU film, compared to the three commercial patches was determined by suture retention strength (SRS) measurements. The mean SRS of the BMM and the PCU film were 6.58 ± 0.97 N and 6.25 ± 0.88 N respectively. There was no significant difference on SRS of the BMMs in two directions, reflecting a uniform resistance to tearing. The mean SRS of Gore-Tex^®, CardioCel^® and CorMatrix^® were 5.35 ± 1.25 N, 8.99 ± 1.77 N and 4.07 ± 1.38 N respectively (Fig. 4A–B). On the other hand, the TN-SRSs of the BMMs in C- and R-directions were 89.91 ± 13.25 N/mm², and 79.1 ± 11.1 N/mm² respectively and were not significantly different from the three commercial patches (Fig. 4C–D). By calculating the area under the stress-strain curve, it was also noted that the BMM had a significantly higher toughness than the commercial patches (Fig. S3, Table S3), which indicated that the BMM was able to withstand more energy from tear to fracture.

Fig. 4. Suture Retention Strength and Thickness-normalized Suture Retention Strength.

Representative SRS curves of the commercial patches, the PCU films and the BMMs (A). SRS of BMM, PCU film and each commercial patch. The difference was found among the commercial patches (*p < 0.05 for Gore-Tex^® vs CardioCel^® and **p < 0.01 for CardioCel^® vs CorMatrix^®) (B). Representative TN-SRS curves of the commercial patches, the PCU films and the BMMs (C). The TN- SRS of the BMM had no significant difference compared to the commercial options, while PCU film has a significant higher TN-SRS compared to the rest of samples (**p < 0.01 for PCU film vs the remaining groups) (D).

3.5. Biostability

The biostability of the BMMs, PCU film, PCU foam, and three commercial patches were assessed via measuring the degradability of samples in the accelerated oxidative solution. The polymer-based samples (BMM, PCU film, PCU foam and Gore-Tex^®) remained stable throughout the 30 days in the accelerated oxidization solution (Fig. 5A–B) with no significant difference in the mechanical properties. On the other hand, the two tissue-based materials fully degraded in the oxidization solution within Day 1. After 20–30 days, the formation of oxidization spots and dents on the PCU films’ surface was apparent, as shown in Fig. 5C, suggesting the beginning of a slow degradation process starting at the outside surface layer.

Fig. 5. Biostability and biocompatibility performance of three commercial patches, the PCU films, foams and BMMs.

The tensile modulus change of all samples in 30 days in the accelerated oxidization solution (A). The tensile modulus of BMM in 30 days showed a stable mechanical performance (ns = no significance, via One-way ANOVA) (B). SEM images unveiled details on the surface morphology of BMM from 0 to 30 days (C). The BSA protein adsorption of the BMM and commercial patches. CorMatrix^® and CardioCel^® patches have a higher BSA adsorption level compared to the BMM (****p < 0.0001 and **p < 0.01) (D). The Ca deposition of the PCU film, BMM and commercial patches. The PCU film and BMM had lower Ca contents in a unit of dry samples, compared to the Gore-Tex^® patch (vs BMM, **p < 0.01), the CardioCel^® patch (vs BMM, ****p < 0.0001) and the CorMatrix^® patch (vs BMM, ****p < 0.0001). (E).

3.6. Biocompatibility

3.6.1. Bovine serum albumin (BSA) adsorption

A BSA protein adsorption test was applied to assess the blood compatibility of the artificial material surface. Fig. 5D illustrates the amounts of adsorbed protein on the surface of BMM and three commercial patches. The polymer-based materials (BMM and Gore-Tex^®) showed similarly low adsorbed BSA amounts. Conversely, the two tissue-derived patches exhibited a much higher concentration of adsorbed albumin compared to the BMMs. The BMM therefore demonstrates a low level of protein adsorption that compares favorably to the three commercial patches.

3.6.2. Ca²⁺ deposition

To study the material’s susceptibility to calcification, the in vitro deposition of Ca²⁺ ions on BMM, PCU film, Gore-Tex^®, CardioCel^® and CorMatrix^® samples was evaluated in a MCP solution, as previously described [53,54]. Fig. 5E shows that both BMM and its main component material PCU have a lower level of calcification compared to commercial patches in 30-day in vitro tests (Table S4), which demonstrate a better resistance to calcification. It also provides a solid foundation for further evaluation in in vivo tests, shown in Section 3.7.

3.7. In vivo studies: subcutaneous implantation

The rat subcutaneous implant model was used for screening cellular infiltration, inflammation and calcification resistance in vivo. Fig. 6A illustrates a set of H&E staining images from PCU film and two commercial patches. Transparent PCU film had delaminated with the adjacent neo-tissue during cutting, due to its elastic properties in comparison with the surrounding tissue. Both the PCU and Gore-Tex samples had a dense layer of tissue capsuled around the samples, and both samples kept a relatively intact morphology after 8-weeks of implantation (Fig. 6A1-A3, Fig. 6A4-A6). There was no cell or tissue growth into the PCU film, whereas cellular nuclei were found infiltrated into the Gore-Tex^® and CardioCel^® patches (Fig. 6A5 and 6A6, Fig. 6A8 and 6A9). CardioCel^® displayed a different tissue response: the patch displayed signs of early degradation and loss of structural integrity at 8 weeks after implantation.

Fig. 6. Histological characterization, mechanical property and calcium quantification of the PCU film, Gore-Tex^® patch and CardioCel^® Patch after in vivo implantation.

Histological sections of the PCU film (A1-A3, B1& B2), Gore-Tex^® (A4-A6, B3& B4) and CardioCel^® (A7-A9, B5& B6) samples following a 8-weeks subcutaneous implantation in rats. Hematoxylin-eosin staining (A1-A9), alizarin red staining (B1–B6) The PCU specimen (dash line) had delaminated with the adjacent neo-tissue during microtome cutting due to the elastic properties in comparison with the surrounding tissue. Calcium quantification data of three samples (C). Mechanical property of PCU film before and after implantation (D). P-values of <0.01 (**) and <0.001(***) were considered statistically significant. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Sections of the three samples also displayed signs of calcification (red and auburn color) in materials and their surrounding tissues (Fig. 6B). PCU film had no evidence of calcification as indicated in Fig. 6B1-B2. Additionally, alizarin red staining showed a far higher degree of calcification in the two commercial patches. The calcification also appeared to extend into surrounding tissues, especially at the interface between the tissue and the two commercial samples (Fig. 6B3-B6). Little to no calcification was present in the rest of the encapsulated tissue.

A calcium content assay was subsequently conducted to confirm the histological findings. A significant increase in Ca²⁺ level was found in Gore-Tex^® and CardioCel^® samples compared to the PCU film (Fig. 6C, Table S4). There was no significant difference in tensile modulus of the PCU film before and after implantation, indicating an intact and non- degradable structure of the raw material of the BMM (Fig. 6D).

4. Discussion

Heart valve leaflets have a highly organized architecture with three specific layers. The fibrosa and ventricularis consist of circumferentially oriented collagen fibers and radially oriented elastin sheets, which constitute their primary load-bearing properties [36]. The spongiosa is inherently soft and compliant with a much lower stiffness. It acts as a cushion, absorbing the load resulting in minimal stress [37]. In this present work, we designed and fabricated a biomimetic, multilayered material to replicate the architecture of those specific layers: The fiber-enhanced PCU films are used as F-mimic layer and V-mimic layers to provide the appropriate mechanical strength and anisotropic properties. The F-mimic and V-mimic layers are the same structure on both sides of the S-mimic layer. These two identical layers do not mimic the mechanical and architectural differences between the two corresponding native F and V layers. These two layers have embedded aligned PCL fibers in order to enhance the anisotropic properties in circumferential direction. A PCU-foam was fabricated via a lyophilization process to create the porous structure from the same polymer solution [55,56]. It was applied to replicate the load-bearing mechanical role, confer flexibility and tune the overall mechanical properties of BMM. Those three layers are based on the same PCU solution, and utilize DMAC as the solvent. Neither delamination nor detachment was founded during the fabrication and tests (Fig. S4 & Video S1). PCU is known as a biostable and biocompatible polymer for heart valve and vascular graft applications [57,58]. It was found to have superior resistance to degradation under biological conditions when compared with common poly(ether urethane) (PEU) and poly(ether urethane urea) (PEUU) [59]. The selection of Carbothane^™ AC-4075A as our PCU resin was not only because of its biostable nature, but also due to its mechanical properties in the range of the native tissue (Fig. S5). In order to offer anisotropic behavior and increase the mechanical stiffness in specific directions, aligned PCL fibers are used since electrospun PCL fibers are widely applied in the fabrication of biomedical devices [60–62]. Although PCL has a biodegradable nature, its fibers were embedded within the PCU films in our design. It probably explains why the biostability test has not shown signs of PCL degradation and why BMM maintained its mechanical properties (Fig. 5A and B). In our future work, fibers made of a biostable polymer will be investigated to ensure the biostability of the mechanical properties of the next version of BMM long-term in vivo. Nevertheless, utilizing PCU as the main component and PCL as supporting fibers, we fabricated the first generation of BMM and assessed its mechanical and biological performance in vitro and in vivo.

Supplementary video related to this article can be found at https://doi.org/10.1016/j.biomaterials.2022.121756

The mechanical assessment utilized cyclic uni-axial tensile tests, flexural bulge tests, and suture retention tests for characterization. For the tensile test, our averaged stress-strain curves and modulus data displayed that anisotropic behavior and mechanical properties of native HAVs were not achieved by the commercial patches. Compared to the native tissue, three selected commercial patches are either too stiff or isotropic and are therefore far from a satisfactory material to match the native tissue: Gore-Tex^® is the expanded polytetrafluoroethylene (e-PTFE) made through a thermal extrusion and it has the most homogeneous performance (e.g. isotropic) among the three commercial patches. It is also the stiffest sample since the carbon atoms in the ePTFE chain are enclosed within a sheath of fluorine atoms [63]. CorMatrix^® and CardioCel^® are two tissue-derived products: The former is composed of porcine small intestinal submucosa extracellular matrix and the latter is a tissue-engineered ADAPT^™ bovine pericardial patch [18,24]. Both of them are less stiff due to the tissue nature, and CardioCel^® even has a similar tensile modulus to those of HAV in the C-direction. BMM, in comparison, demonstrated a superior, stable performance with valve-mimicking architecture, anisotropic behavior, and stable tensile modulus. The capability of BMM to match the mechanical performance of the native tissue is important to optimize leaflet stresses and decrease tears and perforations [30]. Mismatched properties, especially high stiffness from a rigid material, will lead to fibrosis, inflammation, and loss of elasticity and functionality [64].

For the flexural properties, bulge tests were first introduced to study the native leaflet tissue and its artificial alternatives in the literature, to the best of our knowledge. The three commercial patches generally displayed either isotropic or uncontrolled and variable anisotropic performance, and they possessed much higher modulus than HAVs. Both BMM and HAV have a lower E_flex and this performance is also in line with the trend of tensile modulus data, especially in the C-direction (Fig. 3E, orange line vs red bar). Due to the limitation of the pressure transducer and the customized syringe pump, the reproducible maximum pressure of our testing system is 7.2 kPa (54 mmHg). Although this value is lower than the adult physiological transvalvular peak pressure range (90–120 mmHg), our results still demonstrate the performance of flexural deformation under this condition and revealed a correlation between the flexural and tensile properties. BMM, therefore offers a flexural property closer to the native leaflets, compared to the three commercial materials.

Punctures and defects are generated during suturing the artificial materials, leading to mechanical failures through crack propagation [47]. The resistance to tearing is therefore essential to evaluate the feasibility of patches or alternative materials. The SRS and TN-SRS measurements exhibited that the BMMs have a comparable tear resistance to the commercial products. The BMM also displayed a higher toughness (Table S3) than most commercial patches, which emphasizes its durability and capacity to withstand more tear energy than other samples during suturing. A customized heart valve prototype is also fabricated via suturing the BMMs to the 3D-printed valve struts (Fig. S6) and this prosthesis has been tested via the pulse duplicator and more data about its hydrodynamic performance will be reported in our future work.

The biological assessment of the BMMs and commercial patches included their biostability and biocompatibility, in vitro and in vivo. PCU was selected due to its expected stable and compatible in vivo profile and its stable mechanical properties over time. The degradation of polyurethane-based materials in vitro and in vivo was attributed to several mechanisms, including metal ion-induced accelerated oxidative degradation [65], hydrolytic degradation [66] and enzymatic degradation [67]. It has been demonstrated that oxidative degradation is the more dominant mechanism over others [68,69]. Consequently, a 0.1 M CoCl₂/20% H₂O₂ solution was applied to accelerate oxidative degradation of the PCUs. Degradation results after 24 days is shown to correlate to 12 months of in vivo implantation [51,69]. The modulus of the BMM and PCU film/foam displayed no significant change in mechanical properties for 30 days in this accelerated oxidization system. It demonstrated that the BMM has a stable performance which is equivalent to 15 months of in vivo implantation. Using polycarbonate macrodiols as the soft segments, the PCU is designed with better hydrolytic stability and anti-oxidization capability than PEU and PEUU [70]. A stable mechanical performance is essential to maintain the mechanical functionality of the valve or patch over time, and to avoid potential failure and repeated reimplantation procedures. These results confirm the biostability of the BMM, which is comparable to the biostable FDA-approved patch (Gore-tex^®). However, it is also noted that minor signs of oxidative degradation (Fig. 5C) were found on the surface layer, suggesting potential susceptibility of the BMM to long-term oxidization starting from the surface. The biostability of the BMM needs to be assessed in long-term in vivo experiment and may need to be further improved through a surface modification process targeting resistance to oxidation. Coating poly(ethylene glycol) or heparin over the PCU have been verified to defer the oxidation procedure and also improve the hemocompatibility [71,72]. Synthesis of new PCU-based polymer resins to introduce chemical groups like siloxane group is another way to help improving the biostability. The introduction of a siloxane segment to form a large part of the soft segment in PU improves resistance to oxidative degradation, a major pathway leading to degradation of polyurethanes in vivo [73].

Serum protein adsorption and calcium deposition were examined to evaluate the samples’ biocompatibility. Protein adsorption is a significant factor to determine the thrombogenicity of an implanted material. When blood gets in contact with the material’s surface, protein adsorption occurs first, which can then provoke the adhesion of platelets and immune cells on the protein layer. Platelets may aggregate continuously and eventually lead to the generation of a non-soluble fibrin network and thrombus formation [74]. An ideal valve leaflet material should have a low protein adsorption profile to limit or cut the path of thrombin formation and potential subsequent thrombogenic reactions. We performed a BSA adsorption test and found that the BMM exhibited a lower level of protein adsorption compared to three commercial patches. Although the difference is not significant, BMM (main composition PCU) displayed a lower surface tension with improved hydrophilicity (Fig. S7) compared to Gore-Tex^® [75,76], which reduces protein adsorption. Its smooth, non-permeable surface is another aspect that limits albumin adsorption. It is also significant to evaluate the resistance to calcification when developing any biomaterial for heart valve application since calcification is the leading reason of failure of bioprosthetic heart valves and grafts [77]. It is a complex phenomenon influenced by a series of mechanical and biochemical factors [78]. Blood or serum were not used in biostability testing in this round, and this may represent a limitation. We will further assess the protein adsorption on BMM in a simulated blood sample under dynamic conditions in the future study. Calcification limits the durability of synthetic polymer materials used in heart valve devices and blood contact application in general [79]. A 30-day test showed that the BMM and its main component PCU film had a lower level of Ca²⁺ ion accumulation compared to commercial patches. The BMM displayed a higher mean value than the pristine film, which may be attributed to its porous S-mimic layer embedded between films offering more areas for Ca²⁺ ion accumulation. The BMM should therefore be expected to have a slightly higher calcification level than pristine PCU film, but much lower than commercial patches.

The implantation of any artificial material inevitably provokes a host response. The formation of encapsulated tissue (stained as the pink color in Fig. 6A) indicates the end stage of a foreign body reaction [80]. The fibrous tissue capsules around PCU film and Gore-Tex^® were more organized and denser than that around CardioCel^®. The presence of infiltrated cells in Gore-Tex^® and CardioCel^® samples may be associated with the formation of calcium deposits, and the findings in Fig. 6B supported this hypothesis, as calcium concentrations in these two commercial patches were significantly higher than the value for PCU film. Polymers have relatively superior resistance to calcification compared to tissue-derived materials due to their lack of mineralization, which interacts with phosphorus-rich cellular debris and destroyed collagen [81]. In our comparison of two polymer samples, PCU film had a superior resistance to calcification than Gore-Tex from the in vitro and in vivo quantitative analysis. This indicates that PCU may be option for the development of polymeric heart valve devices including patches and heart valve prostheses. In future studies, the BMM material will be implanted in a large animal model in the valvar position in order to assess its biological performance when exposed to the biological and mechanical microenvironment of a heart valve leaflets.

5. Conclusion

Materials currently used for heart valve repair or replacement display very limited durability related to their suboptimal mechanical and biocompatibility performance. There is a clinical need for a new type of biostable material that can achieve better durability after implantation in patients. In this work, our team developed a polymeric, biomimetic multilayered material that replicates the structure-function driven architecture of native valve leaflets via a series of processing methods: solution-casting, lyophilization and electrospinning. Compared to three commercial patches, this BMM demonstrated an anisotropic mechanical behavior and mechanical stiffness which was much closer to the native aortic valve leaflets than the commercial patches. This BMM also showed an excellent durability in an in vitro accelerated oxidization solution and displayed excellent biocompatibility with a lower in vitro protein adsorption level and a lower calcium deposition level. In vivo rat subcutaneous modeling confirmed the mechanical biostability and superior resistance to inflammation and calcification of the main component material, PCU, compared to the commercial patches. This BMM will pave the way for a new clinical- grade biomaterial to be used for surgical valve repair and the development of a new polymeric surgical or transcatheter valve device.

Data availability

Data will be made available on request.