To provide a cost- and time-saving method for the assessment of the calcification potential of bioprosthetic heart valves, several attempts towards an accelerated in-vitro model were undertaken. Due to the problem of superficial calcification or spontaneous precipitation, which occurred in the fluids applied,3,10,14 we focused our investigation on the development of a near-physiological non-spontaneously precipitating calcification fluid and the assessment of its calcification behavior in contact with prosthetic materials under dynamic conditions. Calcification behavior includes the initiation, progression, structural appearance and histological localization of the deposits.
In (avital) bioprosthetic valve replacement, which is often preferred to mechanical valve replacement because of the lack of need for lifelong anticoagulation, failure of the valve due to calcification-related stiffening of the leaflets after about 10 to 12 years is a serious problem.26 The cause of this phenomenon is attributed to mechanical, chemical and biological processes.5 Decades of research in this field has led to various explanations regarding the process and different approaches of valve conditioning. The explanatory approaches and influencing factors are still controversially discussed, e.g. the question whether the initial crystallization process is a purely physical-chemical precipitation of a metastable solution (blood) with possibly subsequent cell and protein involvement, or whether the nucleation itself is already cell-controlled.5,23,24,27,30 In the healthy organism, a balanced ratio of calcification inducers, (high concentration of mineral ions that lead to metastability of the blood in terms of hydroxyapatite crystallization, additional enzymatic release of inorganic phosphates from organophosphates) and inhibitors (Fetuin-A, pyrophosphate, matrix Gla protein19) is assumed, so that ectopic mineralization is prevented.5 For the pathological occurrence of calcification, a disturbance of the mechanism of inhibition is assumed, which may be both, systemically caused or induced by potentially calcifying materials. Thus, in bioprosthetic heart valve replacement, it is believed that after glutaraldehyde (GA) fixation of the valve tissue, the dead cells contained therein may act as nucleators of the calcification due to the phosphate groups of their lipid membranes, thus contributing additional inducers into the system.4,5
Fluid-Reference-Test of Four Different Calcification Fluids by Dynamic Contact with a Potentially Calcifying Material
As expected from the preliminary fluid study,13 the spontaneously precipitating fluids C and E showed a distinct macroscopic calcification of the pericardium patches after 3 to 4 weeks (Fig. 1) whereas the non-spontaneously precipitating fluids revealed no calcification within 4 weeks. Only after raising the calcium concentration in Fluid F, while the ionic product of the fluid was deliberately kept below the solubility product of DCPD13(Table 1), calcification of the patches began in the new Fluid L after 6 weeks of the entire test duration. Fluid L itself revealed no spontaneous precipitation. Here, the material influence of the pericardium on the calcification was clearly visible. The fluid alone did not show homogeneous nucleation, but in the presence of the potentially calcifying material heterogeneous nucleation and secondary seed adsorption took place.3,20,21 The calcifying potential of the pericardium may be ascribed to the tissue texture (collagen fibers, elastic fibers, glutaraldehyde treatment, debris of dead cells, roughness). Moreover, the calcification of all calcified patches commenced at the transition from the firmly clamped to the movable area of the patches where the mechanical bending stress applied to the patch material was highest.25 Thus an additional material stress is suspected here, which obviously had an influence on the location of calcification initiation. Compared to Fluid L, Fluid G had an even lower calcification potential due to its saturation level (Table 1), which was noticeable in a conspicuous delay in the start and extent of calcification (Fig. 2, P10–P12). Even after 9 weeks of testing only scattered spots of calcification were recognizable, whereby the region of highest mechanical bending stress25 was also affected here. In terms of accelerated tests, Fluid L would therefore be preferable, as it fulfills all the necessary conditions (no spontaneous precipitation, almost physiological composition) despite its higher calcification potential. Nevertheless, the use of Fluid G should be further considered in the dynamic testing of bioprosthetic heart valves, as they may have a higher tendency to calcify due to their different load distributions compared to patches. In this case, a fluid with a lower calcification potential could possibly be used to reveal even more detailed differences between different bioprostheses. This will be verified in further tests with bioprosthetic heart valves.
Besides the calcification potential and the macroscopic localization of the occurring calcification, the structural characterization and microscopic or histological localization of the deposits played a decisive role in the assessment of the fluids.
Structural Analysis of the Deposits by XRD
According to the comparison with the reference maps for DCPD, OCP and HAP13 and the literature diagrams of synthetic apatite and the deposits on a human aortic valve (Fig. 8),15 HAP represented the main phase of deposition of all patches (compare Figs. 3a–3e).
Even Fluid C, which in the previous fluid study13 produced OCP as the predominant phase (see Fig. 3b), showed HAP as the predominant phase in material contact. Since this fluid is a spontaneously precipitating fluid, it is assumed that in the presence of the pericardium, both homogeneous and heterogeneous nucleation took place, whereby the homogeneously formed nuclei also found attachment sites on the materials’ surface. The fact that HAP was the predominant phase here compared to pure fluid precipitation, could be due to the active involvement of the pericardial collagen textures and residues of charged sidechains in the initiation, orientation and phase transformation of the crystallization process,17,29 so that HAP formation was favored both thermodynamically and kinetically. According to the literature,17 ACP is considered a precursor of collagen mineralization, which can transform into HAP (the thermodynamically most stable phase at pH 7.4 and 37 °C)8 both, directly and via OCP. The conversion from ACP or OCP to HAP depends on the Ca2+-availability and takes place within 24–72 h as stated in literature.2,8,28 In our study, all diffractograms of patch-deposits (Figs. 3b–3e) showed HAP as the predominant phase, suggesting that at the time of sampling the conversion had already taken place, so that the precursor phases could not be identified in the diffractograms. Fluid C itself had the lowest Ca2+ concentration of the considered fluids, so that it can be assumed that in the pure fluid study (without heterogeneous catalysis) the conversion of OCP to HAP was delayed or prevented as a function of the lower Ca2+-availability, which was exceeded by the material effect in the patch study. Fluid E, which had a distinctly higher Ca2+-content, already showed HAP as the predominant phase of the spontaneous precipitates in the fluid study13 (Fig. 3c blue line). This certainly supports the theory of Ca2+-availability as a controlling factor in OCP/HAP transformation.
All von Kossa stains (Figs. 4d.1–4f.6) of the patch cross sections showed both, calcification on the patch surface and collagen matrix internal intrinsic calcification, which underlines the active participation of the patch material (collagenous fibers) as heterogeneous nucleator.17,29 The heterogeneous nucleation can take place within the collagen structure as well as on the patch surface. It was remarkable that in the spontaneously precipitating fluids C and E (patches P1, P2 and P5, P6) the calcification along the patch surfaces appeared more pronounced than in the non-spontaneously precipitating fluid L (patches P8 and P9) (see Figs. 4d.1–4e.6). This could probably be ascribed to an additional deposition of the homogeneously nucleated crystals on the patch surfaces, which led to a predominantly superficial calcification.
Fluid-Material-Differentiation-Test of the Non-spontaneously Precipitating Fluid L with Two Different Materials
As expected for fluid L from the Fluid-Reference-Test, the pericardial patches showed the macroscopic beginning of calcification after week 6 of testing. Comparable to the Fluid-Reference-Test, the calcification of the pericardial patches commenced at the area where the mechanical bending stress due to the clamping was highest25 (Fig. 5). The polyurethane patches, in contrast, with the exception of patch PU2 (as described in the results), exhibited no calcification even after 9 weeks of testing. This can be attributed to the fact that polyurethane calcification is a surface phenomenon,31,32 so that no deposition occurred after testing in the non-spontaneously precipitating fluid. Furthermore, this suggests that the patches had no surface defects, roughness or porosity to the extent necessary to trigger calcification. Even the mechanical stress applied apparently did not cause any surface defects that served as precipitation nuclei. Although patch PU2 was made from the same PU in the same way as the other two patches and was tested under the same conditions, a different picture emerged here. Since the compartment and the fixation rings of patch PU2 revealed a recurrent white deposit infestation, it was assumed that the test compartment or the fixation rings themselves had an unknown contamination that led to the white precipitate, especially since this phenomenon has not been observed in any of our further calcification tests. Therefore, the results of patch PU2 were not considered representative.
Based on the results for the patches PU1 and PU3, it was found that the tested polyurethane in the shape of foils did not represent a heterogeneous calcification nucleator.
Structural Analysis of the Deposits by XRD
As the comparison with the deposits of the Fluid-Reference-Test and the reference maps for DCPD, OCP, HAP13 as well as the literature diagram (Fig. 8) proved, HAP also formed the main phase of calcifications in the Fluid-Material-Differentiation-Test.
The result of von Kossa staining of the patch cross-section of Peri3 underlines the pronounced occurrence of intrinsic calcification (Figs. 7d.2–7d.4) with heterogeneous nucleation on collagen structures.
Since we have deliberately kept the fluid composition simple and have limited it to the essential physiological calcification factors, i.e. we have refrained from adding fluid-side inducers or inhibitors, a possible effect of non-collagenous proteins, as discussed in the literature,17,28,29 could not be investigated. As a further consequence, even systemically acting anti-calcification coatings, as they are partly considered for bioprosthetic heart valves, cannot be sufficiently investigated with the simplified fluid. These would require systemically active fluids based on cell culture medium with the addition of proteins and enzymes, possibly via fetal calf serum and addition of suitable cells. The in vitro use of these fluids is a challenge, especially because of the need to avoid long-term contamination, and is part of our ongoing studies. For further application in standard in vitro tests, the use of such fluids would also represent a cost issue.
Another limitation arose from the fact that we wanted to develop an accelerated test to get results within a reasonable time. The increased mechanical load applied for this purpose led to expansions of the firmly clamped patches, especially the pericardial patches, which are certainly not comparable with deformations under physiological conditions. In order to achieve a more realistic performance under accelerated conditions, further systematic calcification studies with heart valve prostheses that react differently to stress due to their opening and closing behavior need to be conducted. The influence of patch deformation on calcification behavior was not investigated in this study, as this study only focused on evaluating the fluid used, based on its ability to differentiate between different calcification propensities of the tested materials, regardless of what caused these different tendencies.