- Email: [email protected]
Chemical treatment and tissue selection: factors that influence the mechanical behaviour of porcine pericardium
Biomaterials 22 (2001) 2759}2767
Chemical treatment and tissue selection: factors that in#uence the mechanical behaviour of porcine pericardium JoseH M. GarcmH a PaH ez *, Eduardo Jorge-Herrero, Antonio Carrera, Isabel MillaH n, Aurora Rocha, P. Calero, Angeles CordoH n, JoseH Salvador, Natividad Sainz, JesuH s MeH ndez, JoseH L. Castillo-Olivares Services of Preventive Medicine, Puerta de Hierro Clinic (Cln& nica Puerta de Hierro), San Martin de Porres, 4, 28035 Madrid, Spain Experimental Surgery, Puerta de Hierro Clinic (Cln& nica Puerta de Hierro), San Martin de Porres, 4, 28035 Madrid, Spain Superior Technical School of Industrial Engineering (Escuela Te& cnica Superior de Ingenieros Industriales), Madrid, Spain Biostatistics, Puerta de Hierro Clinic (Cln& nica Puerta de Hierro), San Martin de Porres, 4, 28035 Madrid, Spain Received 19 July 2000; accepted 8 January 2001
Abstract Calci"cation and mechanical failure are the major causes of the loss of cardiac bioprostheses. The chemical treatments used to stabilize the tissue employed are considered to play a fundamental role in the development of these two phenomena, although the problem is multifactorial and the underlying causes are yet to be fully identi"ed. Currently, there is an ongoing search for chemical treatments capable of reducing or eliminating the process of calci"cation while preserving the mechanoelastic characteristics of the tissue. One of the approaches to this e!ort is the elimination of the phospholipid component from the biological tissue employed in prosthesis construction. There is evidence that this component may be responsible for the precipitation of calcium salts. The present study compares two delipidating chemical treatments involving chloroform/methanol and sodium dodecyl sulfate (SDS) with the use of glutaraldehyde (GA) alone. For this purpose, porcine pericardial tissue was subjected to tensile strength testing employing a hydraulic simulator. A total of 234 samples were studied 90 treated with GA, 72 treated with chloroform/methanol and 72 treated with SDS. The mean breaking strength was signi"cantly higher in the samples treated with GA (between 43.29 and 63.01 MPa) when compared with those of tissue treated with chloroform/methanol (29.92}42.30 MPa) or with SDS (13.49}19.06 MPa). In a second phase of the study, selection criteria based on morphological and mechanical factors were applied to the pericardial membranes employing a system of paired samples. The mathematical analysis of the "ndings in one fragment will aid in determining the mechanical behavior of its adjacent twin sample. In conclusion, the anticalci"cation chemical treatments tested in the experimental model conferred a lesser mechanical resistance than that obtained with GA. On the other hand, the utilization of paired samples was found to be useful in the prediction of the mechanical behavior of porcine pericardial tissue. Nevertheless, in order for our method of selection to be considered the most adequate approach, it will be necessary to validate these "ndings in dynamic studies involving a real, functional model. 2001 Elsevier Science Ltd. All rights reserved. Keywords: Biomaterials; Pericardium; Mechanical properties; Chemical treatments
1. Introduction The development of biomaterials for the medical "eld is increasingly necessary. To improve their resistance, compatibility and durability will be a great challenge for the coming years. One example of the need for these e!orts is illustrated by the limited use of cardiac bioprostheses due to problems concerning their durability. Valve * Corresponding author. Tel.: #34-91-316-4040; fax: #34-91-3737667. E-mail address: [email protected] (E. Jorge-Herrero).
failure makes their replacement necessary within 10}15 years of implantation in 30% of cases [1]. The calci"cation of biological tissues, due to unknown reasons [2}4], mechanical fatigue [5}7] or both processes simultaneously [8}10] are responsible for this situation. Structural failure of bioprostheses can be produced by early tears in the tissue [11]. On other occasions, the problem develops over the medium or long term as a consequence of the deposition of calcium salts on the tissue. To date, there is no chemical treatment capable of preventing this from occurring, although a number of approaches are currently being studied [12}14].
0142-9612/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 0 1 ) 0 0 0 1 9 - 9
2760
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
In 1986, Ross [15] attributed the injury to the vascular endothelium to the shearing force generated in patients with arterial hypertension as a consequence of the excessive tension and the perturbation of the blood #ow. He also reported that the lesion produced by shear stress was followed by the hardening of the vascular wall and, "nally, calci"cation. Shearing stress was most probably responsible for the limited durability of samples subjected to tensile testing, as we observed in trials carried out with a tissue similar to that used in the construction of the lea#ets for prostheses made of calf pericardium [6]. To reduce or impede calcium deposition is the objective of a number of studies [16,17], but this aim must be achieved with no loss of the mechanical properties of the tissue, and if possible, should result in their improvement. Glutaraldehyde (GA) is currently the "rst choice for the crosslinking of biological tissues [18], although a number of new chemical treatments are under investigation to test their capacity to eliminate the harmful e!ects of this dialdehyde. Thus, new forms of chemical stabilization are being introduced, such as treatment with carbodiimide [19,20], diphenylphosphorylazide [21,22] polyepoxy resins [23], aminooleic acid [24] and diisocyanate hexamethylene [25,26]. All these agents and procedures must be tested to demonstrate their e$cacy for crosslinking and their lack of any role in favoring the process of calci"cation, although GA , despite its disadvantages, presents an entire series of properties that make it the reagent of choice for the time being: it reduces biodegradation, it preserves the anatomic integrity and physical features of the tissue, its cost is low, it is easy to obtain, it is water soluble and it does not promote thrombogenicity. Moreover, the use of GA for tissue "xation is associated with a good hemodynamic behavior [27]. On the other hand, there are a number of clear disadvantages such as the calci"cation of the implants and the cytotoxicity of this reagent [3,18,27}29]. Changes in the mechanical properties of the treated tissue have also been reported, including increased rigidity, the appearance of zones of greater concentration of internal shear and reduced strength [30]. It is therefore necessary to search for alternatives to this agent, either the use of one or more of the new reagents mentioned above, or some additional chemical treatment that impedes the calci"cation of the tissue. In this respect, our group has studied several anticalci"cation procedures in bovine pericardial tissue [2,4,10,16]. In order for a chemical treatment to be used to reduce or impede the deposition of calcium on bioprostheses, it is increasingly important to study the solvents in which the biological tissues are prepared and their e!ects on both the mechanical and elastic properties of these tissues [31]. In this study, we have analyzed these mechanoelastic properties in porcine pericardium exposed to a standard treatment with GA, comparing them to those observed in tissue samples treated previously with chloroform/meth-
anol or with sodium dodecyl sulfate (SDS). The use of chloroform/methanol for lipid extraction was e!ective against mineralization of the tissue, reducing calci"cation [2,4,16,32]. SDS acts directly on the tissue phospholipids [33,34] and is capable of reducing calci"cation in tissue implanted subcutaneously, but it is not as e!ective when exposed to the circulation [34]. The trials involved the use of a hydraulic simulator capable of delivering the pressure caused by the compression of a saline solution by a piston to a chemically stabilized pericardial membrane. The "nal objective of this work is to develop a new in vitro system using a hydraulic simulator to establish criteria for the selection of tissues according to the type of chemical treatment to which they have been exposed, and to apply these criteria to predict the mechanical behavior.
2. Materials and methods Porcine pericardium was obtained directly from a local abattoir and transported in cold isotonic saline (0.9% sodium chloride) to the laboratory where it was cleaned. The parietal pericardial sac was employed and the tissue samples were obtained in such a way that they all presented similar morphology (Fig. 1). Each membrane thus obtained measured approximately 10}20 cm from base to apex and 10 cm wide. Six circles with a radius of 1 cm were cut from each pericardial sac (Fig. 1). In all, 234 circular membranes were employed for each assay, 117 from the region referred to as B (pericardium covering the right ventricle) and 117 from the region referred to as C (pericardium covering the left ventricle). Each region was divided into three
Fig. 1. Illustration of a pericardial sac employed in the trial. Three symmetrical pairs of circular membranes are obtained for a total of 39 specimens, one each from the following zones of the regions corresponding to right ventricle (region B) and to left ventricle (region C): upper outer zone (zone 1), central zone (zone 2) and lower outer zone (zone 3).
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
symmetrical zones*zone 1 (upper outer zone), zone 2 (central zone) and zone 3 (lower outer zone)*each of which yielded one of the six circular tissue samples. Three groups were established, including a control group, treated with glutaraldehyde alone (GA). It was composed of 90 samples (45 obtained from region B and 45 from region C) which were treated for 24 h with 0.625% GA, prepared from a commercial solution of 25% GA (Merck), at a ratio of 1/50 (w/v), in 0.1 M sodium phosphate bu!er, pH 7.4. Another group consisting of 72 samples (36 from region B and 36 from region C) was pretreated with chloroform/methanol (CM) 1 : 4 (v/v) for 2 h at a proportion of 1 : 30 (p/v) with respect to the amount of tissue, followed by a treatment similar to that employed in the GA group. The third group was made up of 72 membranes (36 from region B and 36 from region C) that were pretreated for 2 h with 1% SDS in 0.15 M NaCl at a ratio of 1/30 (w/v), followed by the same GA treatment as above. The thickness of each membrane was determined by serial readings at 10 di!erent points using a digital Mitutoyo micrometer (Elecount digital series E/A33/8) with a precision at 203C of $3 . Each membrane was subjected to increasing stress until rupture, which was accompanied by a loss of stress and con"rmed macroscopically by the presence of a tear in the tissue.
2761
Fig. 2. Clamping system to ensure the immobility of the samples during the trial.
2.1. Assay method The assay was carried out on a hydraulic simulator capable of delivering increasing stresses to the pericardial membranes secured with pressure clips (Fig. 2). The membranes were exposed to increasing hydrostatic pressure caused by the compression transmitted to saline solution by a piston. As the piston moved, the #uid deformed the membrane and a pressure gauge determined the pressure, ranging between 0 and 16 atm. The simulator consisted basically of a unit for measuring pressure equipped with a servomotor to drive the pump propelling the piston. The entire system is illustrated in Fig. 3. 2.2. General description of the function of system A piston was activated by means of a digital monitor based on a high-speed processor that controlled the direct current electric servomotor. The piston compressed the #uid and the pericardial membrane resisted the pressure. The tissue was subjected to continuous, increasing pressure until rupture. The controlling computer indicated the angular velocity of the activating system, which was maintained throughout the trial. The data acquisition system evaluated the #uid pressure and the movement of the piston at all times. The numerical data corresponding to these variables were transferred to a computer via a series interface, where they were stored for subsequent analysis.
Fig. 3. View of the equipment. Pressure pistons.
2.3. Technical features of the hydraulic simulator The most relevant technical speci"cations are as follows: Ampli"er. D-MOS technology; H-bridge con"guration; maximum working voltage: 53 V; maximum intensity in steady state: 3 A. Motor. Rated voltage: 24 DCV; rated output: 15 W; starting torque: 120 mN m; current in a vacuum: 21.7 mA; starting current: 3040 mA; maximum permanent torque: 30.46 mN m. Incremental position sensor: optical; two quadrature outputs and index impulse. Quadrature processor: programmable logic technology; two quadrature inputs; incremental/decremental monopolar impulse output; maximum working frequency of 4 MHz. Digital compensator: processor, RISC microcontroller, 24 MHz, 8 bits, 200 ns/instruction; maximum sampling frequency of 1 kHz; velocity range from 00 to 8388607 counts/sampling period *256; proportional action coe$cient (KP) of !32768 to 32767; di!erential action coe$cient (KD) of !32768 to 32767.
2762
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
Piston: 160-mm stroke; 32 mm in diameter; maximum pressure 16 atm. Pressure sensor: maximum pressure 16 atm; output signal: 4}20 mA. Computer: standard Pentium-75 con"guration. This system was developed by Sat Polar, S.L., a Spanish electronic engineering "rm.
tion, that is deformation, was not considered to be of interest).
2.4. Tensile strength
2.6.4. Selection criteria Selection criteria were established to ensure greater homogeneity of the samples. The purpose of these statistical selection criteria was to determine the probability that each membrane tested actually belonged to the region or zone to which it was assigned in the initial selection. Thus, those membranes with a minimum thickness greater than the mean value for the corresponding treatment group, region and zone plus one standard deviation or less than the mean value minus one standard deviation were excluded, as were those membranes in which the di!erence between the mean thickness for of a given zone in the two regions and the minimum thickness for the corresponding zone was greater than the mean value for this di!erence determined for the corresponding series, plus one standard deviation. The pairs of membranes (one each from regions B and C) in which the stress (MPa) for x"1 for each zone of region C was above or below the mean plus or minus one standard deviation for said zone were also excluded. It is assumed that the values for samples from B would be unknown since they would be obtained from the projection of the results in region C in each treatment group. Thus, the expected values for B (dependent variables), predicted on the basis of the "ndings in region C (independent variables) would be compared with the actual values for region B determined subsequently for each respective projection.
Once the pressure withstood by the pericardial membrane at each instant of the trial was known, its tensile strength was calculated using a formula similar to the Laplace equation, described by Timoshenko [35] for a thin-walled membrane subjected to pressure: T " pr/2e, where p is the pressure in kg/cm, r the radius of the spherical membrane expressed in cm, e the thickness of the membrane in cm and T the tensile strength in kg/cm. To express this value in MPa, we divided the result by 10.19. 2.5. Elongation The movement of the piston indicated the variation in the #uid volume at every moment and for each di!erent pressure applied and, thus, the changes in membrane geometry up to the moment of rupture. At that point, the shape was that of a round bonnet the base of which was a known circle (the frame on which the membrane to be tested was mounted). By measuring the changes in length of the longest arc of the bonnet, it was possible to determine the percentage of elongation at each moment of the trial. 2.6. Statistical study and mathematical analysis 2.6.1. Comparison of means at rupture The mean values at rupture for groups GA, CM and SDS were compared, taking into account the di!erent regions (B and C), by means of Student's t test and the Newman}Keuls test. 2.6.2. Mathematical xt of the tensile strength/elongation ratio The tensile strength (MPa)/elongation (per unit) ratio was studied using the least squares method. The best "t corresponded to a third-order polynomial, the shape of which is expressed as y"b x#b x#b x, where y is the tensile strength in MPa and x is the per unit elongation of the membrane (the value of the constant b was made to equal zero since due to biological considerations, the equation must pass through the origin; at zero stress, there would be no deformation). For these same considerations, the analysis was done for x(1 (the behavior of the function when the membrane had surpassed its elastic limit, entering the realm of irreversible elonga-
2.6.3. Mean overall xt for each treatment group in each of the two regions The tensile strength/elongation ratio was also studied by group, by region and by zone.
2.6.5. Mean overall xt in the selected zones On the basis of the aforementioned criteria, the following 41 of the 234 sample pairs tested (35.04%) were selected: GA group: region B, zone 1 (B1)/region C, zone 1 (C1): pairs 9, 11, 12 and 13 B2/C2: pairs 1, 2, 7, 8, 9 and 10 B3/C3: pairs 8, 9 and 10 CM group: B1/C1: pairs 4, 6 and 10 B2/C2: pairs 1, 2, 4 and 12 B3/C3: pairs 2, 4, 8, 10, 11 and 12 SDS group: B1/C1: pairs 1, 4, 9 and 11 B2/C2: pairs 1, 5, 10, 11 and 12 B3/C3: pairs 1, 3, 4, 5, 9 and 11
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
2.6.6. Predictive study A predictive study of the values for region B was performed after the selection process had been carried out and the values for the equivalent region C samples were known. This determination involved the mathematical calculation of the values for the selected pairs according to the aforementioned criteria, using linear regression analysis, where the values for region C were known independent variables and those of region B were predictive, dependent variables. The tensile strength (MPa) in region B (y ) was estimated on the basis of that of region C (y ), and the 95% con"dence intervals were calculated. The estimated values for y were then compared with the respective known values for y . Correlation and regression studies to compare the three treatment groups were also carried out according to region (B or C) and zone (1, 2 or 3). A regression curve gave the best mathematical "t.
3. Results 3.1. Breaking strength The results at breaking point are shown in Table 1. The pericardial membranes treated with GA, whether Table 1 Mean values at breaking point for each zone according to chemical treatment and region Trial
GA B C B C B C CM B C B C B C SDS B C B C B C
No. of samples
Mean breaking strength (MPa)
Standard deviation
Range
15 15 15 15 15 15
59.51 61.84 43.29 62.07 63.01 61.61
11.60 22.65 17.78 13.51 17.38 16.99
43.43}83.27 13.00}71.70 23.18}90.64 26.28}72.27 21.91}81.46 25.25}92.79
12 12 12 12 12 12
35.88 28.93 29.92 42.30 38.84 35.17
8.76 12.58 11.31 15.21 12.60 13.01
17.08}49.34 10.32}52.37 16.40}45.75 21.06}62.90 21.75}70.55 16.03}60.11
12 12 12 12 12 12
13.98 13.49 19.06 18.69 19.06 16.26
5.17 6.81 6.62 10.49 6.46 5.06
7.98}23.14 3.07}25.73 10.76}27.82 10.90}49.61 5.68}29.90 8.99}24.86
from region B or region C, were more resistant to breakage, in terms of mean values, than those treated with CM or with SDS. The mean breaking strength of the samples treated with GA and of those treated with CM was signi"cantly greater ( p(0.001) than that of the SDS series. However, when GA and CM treatment groups were compared, statistically signi"cant di!erences were observed only in zones 1 ( p(0.001) and 2 ( p(0.05) of region B. The values for the GA group ranged from 13 to 92.79 MPa, those of CM samples between 10.32 and 62.90 MPa and those of samples treated with SDS from 3.07 to 49.61 MPa. 3.2. Mathematical xtting of the tensile strength/elongation ratio Table 2 displays the coe$cient for the equations corresponding to the mean values for the GA, CM and SDS groups in each region and zone, as well as the coe$cients of determination (R). The model that best "ts the tensile strength/elongation ratio is a third-order polynomial. The application of the selection criteria described in the material and methods section revealed a clear improvement in the coe$cients of determination (Table 3), all of which were greater than or equal to 0.9.
Table 2 Mathematical "t of the tensile strength/deformation ratio for each sample series Trials
The values for the GA group and the CM group were signi"cantly higher than those of the SDS group in every case (p(0.001). The values in the GA group were signi"cantly higher than those of the CM group in zone 1 (p(0.001) and zone 2 (p(0.05) of region B. GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate.
2763
GA B C B C B C CM B C B C B C SDS B C B C B C
a
a
a
R
15.38 12.79 12.87 14.96 10.96 15.05
!28.47 !25.39 !24.99 !27.19 !19.38 !32.31
31.58 28.67 28.73 29.32 21.48 34.14
0.810 0.781 0.853 0.896 0.802 0.865
11.30 10.62 12.19 14.27 9.50 15.06
!17.79 !18.71 !21.87 !27.77 !18.69 !30.58
22.09 21.23 25.54 32.67 21.40 35.21
0.890 0.950 0.847 0.869 0.706 0.939
12.79 10.57 10.07 11.34 10.51 10.97
!24.46 !18.07 !16.88 !21.10 !21.26 !16.98
31.69 23.90 22.42 27.50 24.90 25.23
0.544 0.898 0.914 0.966 0.951 0.952
This table presents the coe$cients obtained after the mathematical "tting of the tensile strength/deformation curves, as well as the determination coe$cient of the "t. GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate; R, coe$cient of determination.
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
2764
Table 3 Mathematical "t of the tensile strength/deformation ratio for each sample series after application of the selection criteria
Table 4 Prediction of the tensile strength (MPa) after application of the selection criteria
Trials
Trials/region
GA B C B C B C CM B C B C B C SDS B C B C B C
a
a
a
R
15.55 10.33 13.95 16.43 12.49 18.11
!26.68 !15.45 !27.40 !30.06 !22.99 !38.31
29.98 17.17 31.13 31.56 26.41 43.34
0.942 0.961 0.939 0.844 0.920 0.967
12.49 9.74 11.82 11.25 8.77 16.16
!22.99 !13.17 !19.01 !20.79 !16.52 !33.24
26.41 15.49 22.11 23.15 18.56 38.21
0.920 0.918 0.930 0.843 0.931 0.930
6.91 19.10 11.08 8.84 10.51 10.02
!6.25 !53.63 !22.15 !7.58 !21.26 !13.13
11.66 60.22 27.14 15.27 24.90 21.90
0.950 0.974 0.980 0.962 0.951 0.946
The coe$cients expressed in Table 2 after application of the selection criteria (see Section 2). GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate; R, coe$cient of determination.
3.3. Prediction of the tensile strength The use of linear regression analysis to "t the data from each region and the corresponding zones for each treatment group revealed excellent correlation for the tensile strength values (Table 4). These results were obtained after the application of the selection criteria described above. The values for region B (dependent variables) for each trial were obtained from region C (independent variable). The coe$cients of determination (R) were greater than 0.99 in every case. The analysis of the correlation between the tensile strengths found for the di!erent treatments, regions and zones appear in Table 5. The coe$cients of determination were 0.94 or higher in every case (over 0.99 in all but two), indicating that the behavior of a given zone can be predicted on the basis of the "ndings in the corresponding zone and region of samples treated with di!erent chemical agents. By way of example, "gure illustrates the comparison of the mechanical behavior (tensile strength in MPa) between zones 1 and 3 of region B.
4. Discussion The introduction of di!erent crosslinking and anticalcifying chemical treatments for biological tissues has ren-
GA B B B CM B B B SDS B B B
a (95% CI)
a (95% CI)
R
!0.22 (!0.15, 0.11) !0.47(!0.27,!0.67) 0.35 (0.24, 0.46)
1.56 (1.53, 1.59) 0.98 (0.78, 1.02) 0.42 (0.41, 0.44)
0.999 0.996 0.996
!0.33(!0.08,!0.57) 0.06(!0.03, 0.15) 0.15 (0.08, 0.21)
1.25 (1.19, 1.30) 1.13 (1.11, 1.15) 0.52 (0.45, 0.62)
0.994 0.999 0.999
0.69 (0.18, 1.19) !0.16 (!0.45, 0.14) 0.10 (!0.06, 0.26)
0.56 (0.48, 0.63) 0.87 (0.82, 0.92) 0.69 (0.66, 0.72)
0.945 0.989 0.996
Regression curves where y is the dependent variable and y the independent variable (see Section 2.6.6). GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate; R, coe$cient of determination.
Table 5 Coe$cients of the regression curves that correlate the applied tensile strength (MPa) in terms of the di!erent treatment groups, regions and zones Independent variable (x)
Dependent variable (y)
Region Zone a
a
R
CM
GA
B
1.22 1.05 0.89 0.95 1.29 1.07 1.35 1.06 0.69 0.59 0.90 1.05 1.11 1.00 0.81 0.63 0.70 0.94
0.999 0.997 0.994 1.000 0.998 1.000 0.995 0.999 0.992 0.940 0.991 0.993 0.994 0.993 0.999 0.935 0.993 0.992
C
SDS
GA
B
C
SDS
CM
B
C
1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
0.14 !0.12 0.20 0.05 0.23 0.04 0.43 0.22 0.34 0.65 0.65 0.40 0.24 0.33 0.08 0.65 0.32 0.46
GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate; R, coe$cient of determination.
dered them for use in clinical applications, one of which is the construction of cardiac bioprostheses. The phenomena of calci"cation, stress and tissue perforation are intimately correlated, resulting in the degeneration of the native structure. Di!erent chemical treatments a!ect the tissue ultrastructure, often modifying the native conformations of the tissue's protein, mainly collagen, that is most prevalent in bovine or porcine pericardium. The degree of change produced by chemical solvents would
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
be especially important in the mechanical performance of biological tissues [36]. To improve the mechanoelastic properties of the tissue and reduce or prevent calcium deposition on the tissue would increase the functionality and durability of and structure made from that biomaterial. To achieve this, there are two complementary strategies that consist in designing e!ective chemical treatments and biomaterials that are more resistant or have less propensity toward calcium deposition. It has been demonstrated that cardiac bioprostheses implanted in tricuspid position and subjected to mild physiological stress do not become calci"ed [12}14,37]. On the other hand, in 1998, Myken et al. [38] observed that 16 of 30 dysfunctional porcine bioprostheses showed signs of calci"cation, all with a peak gradient of over 60 mmHg. As it is not always possible to reduce stress, it is obvious that we must prevent calci"cation by chemical means. This study, unfortunately, demonstrates a statistically signi"cant loss of breaking strength ( p(0.001) in tissue samples treated with SDS (Table 1). This loss is evident in both the pericardium covering right ventricle (region B) and that of left ventricle (region C). It is less marked, however, in samples treated with CM and is statistically signi"cant ( p(0.001) in only two zones (1 and 2) of region B. The width of the ranges, on the other hand, re#ects the lack of homogeneity of the tissue, making it necessary to select the most suitable specimens. The results obtained with recently described selection processes have not been de"nitive [39,40], not even those of Hiester and Sacks who, in 1998, utilized small-angle light scattering (SALS), a nondestructive optical technique, to quantify the collagen "ber architecture of soft tissues [39]. That same year, Braile et al. reported their conclusions using a method that combines mechanical and histopathological features [40], but were not able to provide a de"nitive selection strategy either. The technique proposed in the present report consists of the use of paired samples [41]. The process consists of the selection of two adjacent tissue samples, one of which will be subjected to mechanical assays, while the other (Fig. 1) is left intact and its theoretical parameters calculated on the basis of the mathematical "tting of the stress/elongation curve. This method substantially improves the "t of the mean stress/deformation, or tensile strength/deformation, curves, providing mathematical functions to de"ne the mechanical behavior with excellent determination coe$cients (Table 3). These functions can be employed to compare the viscoelastic properties achieved with di!erent chemical treatments (GA, CM, SDS) in tissue samples obtained from each region and zone of the pericardial sac. In addition, the use of paired samples, according to a statistical methodology described elsewhere [41] made it possible to perform a predictive analysis of the outcome in intact tissue prior to its treatment, based on the knowledge of the mechanical
2765
behavior of the other half of the pair. This method helps to establish the relationships between the di!erent chemical treatments and zones employed, as shown in Table 5, which presents the mathematical functions that correlate the tensile strength (MPa) applied in each assay and permit the comparison of the resulting mechanical behaviors. Figs. 4 and 5 display the regression curves resulting from this comparison. For a tensile strength of 0.2 MPa, which is similar to the working stress that a cardiac valve lea#et made of a tissue treated with SDS will withstand, the theoretic stress that a tissue treated with GA or with CM could withstand would range between 1.2 and 3.5-fold greater (0.24}0.70 MPa) for tissue from zones 1 and 3 of region B (Fig. 4) and 2.3}4.2-fold greater (0.46}0.83 MPa) for samples from zone 2 of region C (Fig. 5), with the higher values corresponding to samples treated with GA. This lack of strength at a low level of stress shown by tissue treated with SDS, as well as its low breaking strength, can severely in#uence the durability of devices made from biomaterials subjected to this treatment. In general, samples exposed to CM also show a loss of strength, although less marked than those treated with SDS, and only its anticalci"cation properties would argue in favor of recommending its use, but with precaution.
Fig. 4. Regression curves for zones 1 and 3 of region B, comparing the tensile strength (MPa) according to the chemical treatment received. Glut: glutaraldehyde. Cl/M: chloroform/methanol. SDS: sodium dodecyl sulfate.
Fig. 5. Regression curves for zone 2 of region C, comparing the tensile strength (MPa) according to the chemical treatment received. Glut: glutaraldehyde. Cl/M: chloroform/methanol. SDS: sodium dodecyl sulfate.
2766
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
In conclusion, any new method of "xation of a biomaterial requires a study of its impact on the mechanical behavior of the material. A loss of strength like that demonstrated in the present report in tissue exposed to SDS eliminates any advantages to the use of this substance. The method of selection employed is suitable for static testing, as described here, but it will have to be validated in dynamic studies of a real, functional model if it is to be considered an adequate approach to tissue selection.
Acknowledgements This study was "nanced by Grants Nos. 96/0250 and 00/0192 from the Fondo de Investigaciones Sanitarias (FIS), Spain. The authors are grateful to M. Messman for her translation of the text.
References [1] Vongpatanasin W, Hillis LD, Lange RA. Medical progress: prosthetic heart valves. N Engl J Med 1996;335:407}16. [2] Jorge-Herrero E, Fernandez P, GutieH rrez M, Castillo-Olivares JL. Study of calci"cation of bovine pericardium: analysis of the implication of lipids and proteoglycans. Biomaterials 1991;12:683}9. [3] Levy RJ. Glutaraldehyde and the calci"cation mechanism of bioprosthetic heart valves. J Heart Valve Dis 1994;3:101}4. [4] Jorge-Herrero E, FernaH ndez P, Escudero C, Castillo-Olivares JL. Inhibition of the calci"cation of porcine valve tissue by selective lipid removal. Biomaterials 1994;15:815}20. [5] Bortolotti U, Milano A, Thiene G, Guerra F, Mazzuco A, Valente M, Talenti E, Galluci V. Early mechanical failures of the Hancock pericardial xenograft. J Thorac Cardiovasc Surg 1987;94:200}7. [6] GarcmH a PaH ez JM, Carrera A, GarcmH a Sestafe JV, MillaH n I, Jorge E, Candela I, Castillo-Olivares JL. Is cutting stress responsible for the limited durability of heart valve bioprostheses? J Thorac Cardiovasc Surg 1990;100:580}6. [7] Shoen FJ, Horson CE. Anatomic analysis of removed prosthetic heart valves: causes of failure of 33 mechanical valves and 58 bioprostheses, 1980}1983. Hum Pathol 1985;16:549}59. [8] Thubrikar MJ, Deck DJ, Aouad J. Role of mechanical stress in calci"cation of aortic bioprosthetic valves. J Thorac Cardiovasc Surg 1983;86:115}25. [9] Gabbay S, Kadan P, Factor S, Cheung JK. Do heart valve bioprostheses degenerate for metabolic or mechanical reasons? J Thorac Cardiovasc Surg 1988;95:208}15. [10] Jorge-Herrero E, FernaH ndez P, Escudero C, de la Torre N, GarcmH a PaH ez JM, Castillo-Olivares JL. In#uence of stress on the calci"cation of delipidated bovine pericardial tissue employed in the construction of cardiac valves. J Biomed Mater Res 1996; 30:411}5. [11] Hariza F, Papouin G, Barratt-Boyes B, Christie G, Whitlock R. Tear in bioprosthetic heart valve lea#ets without calci"c degeneration. J Heart Valve Dis 1996;5:35}9. [12] Grunkemeier GL, Jamieson WRE, Miller DC, Starr A. Actuarial versus actual risk of porcine structural valve deterioration. J Thorac Cardiovasc Surg 1994;108:709}18. [13] Tyers GFO, Jamieson WR, Munro IA, Germann E, Burr LH, Miyagishima RT, Ling L. Reoperation in biological and mechanical valve populations: fate of the reoperative patient. Ann Thorac Surg 1995;60:S464}9.
[14] Kent PD, Tazelaar HD, Edwards WD, Orszulak TA. Temporal changes in the surgical pathology of prosthetic aortic valves: a study of 157 cases spanning 26 years (1970}1995). Cardiovasc Pathol 1998;7:9}23. [15] Ross R. The pathogenesis of atherosclerosis: an update. N Engl J Med 1986;314:488}500. [16] Jorge-Herrero E, GutieH rrez M, Castillo-Olivares JL. Inhibition of bovine pericardium calci"cation: a comparative study of Al> lipid removing treatments. J Mater Sci: Mater Med 1991;2:86}8. [17] Khor E, Wee A, Tan BL, Chew TY. The e$cacy of a dual anticalci"cation agent treatment in the prevention of calci"cation of glutaraldehyde-"xed porcine pericardium. J Mater Sci: Mater Med 1995;6:518}22. [18] Jayakrishan A, Jameela SR. Glutaraldehyde as a "xative in bioprostheses and drug delivery matrices. Biomaterials 1996;17:471}84. [19] Nimni ME, Cheung D, Strates B, Kodama M, Sheilkh K. Chemically modi"ed collagen: a natural biomaterial for tissue replacement. J Biomed Mater Res 1987;21:741}71. [20] van Wachem PB, van Luyn MJA, Olde Damink LHH, Dijkstra PJ, Feijen J, Nieuwenhius P. Biocompatibility and tissue regenerating capacity of crosslinked dermal sheep collagen. J Biomed Mater Res 1994;28:353}63. [21] Petite H, Frei V, Huc A, Herbage D. Use of diphenylphosphorylazide for cross-linking collagen based biomaterials. J Biomed Mater Res 1994;28:159}65. [22] Petite H, Duval JL, Abdul Malak N, Sigot MF, Frei V, Herbage D. Cytocompatibility of calf pericardium treated by glutaraldehyde and by acyl-azide methods in an organotypic culture model. Biomaterials 1995;16:1003}8. [23] Xi T, Liu F. E!ect of treatment with epoxy compounds on the mechanical properties of bovine pericardial bioprosthetic materials. J Biomater Appl 1994;7:61}75. [24] Chen W, Kim JD, Schoen FJ, Levy RJ. 2-Aminooleic acid derivation of glutaraldehyde pretreated bioprosthetic tissues for counteracting calci"cation: a review and comparison of its e$cacy and mechanism in various animal models. Cells Mater 1994;4:419}28. [25] Naimark WA, Pereira CA, Tsang K, Lee JM. HDMC crosslinking of bovine pericardial tissue: a potential role of the solvent environment in the design of bioprosthetic materials. J Mater Sci: Mater Med 1995;6:235}41. [26] Van Luyn MJ, Van Wachem PB, Dijkstra P, Olde Damnik LH, Feijen J. Calci"cation of subcutaneously implanted collagens in relation to cytotoxicity, cellular interaction and crosslinking. J Mater Sci: Mater Med 1995;6:288}96. [27] Duncan AC, Boughner D, Vesely I. Dynamic glutaraldehyde "xation of a porcine aortic valve xenograft. Biomaterials 1996; 17:1849}56. [28] Hughes H, Lilburn S, Tipton S, Aboul-Enein HY, Duran CG. Chemical assay of glutaraldehyde incorporation into pericardial tissue. J Heart Valve Dis 1994;3:105}9. [29] Olde Damnik LH, Dijkstra P, Van Luyn MJ, Van Wachem PB, Feijen J. Glutaraldehyde as a crosslinking agent for collagen based materials. J Mater Sci: Mater Med 1995;6:460}72. [30] Talman EA, Boughner DR. Glutaraldehyde "xation alters the internal shear properties of porcine aortic heart valve tissue. Ann Thorac Surg 1995;60:s369}73. [31] Gratzer PF, Pereira CA, Lee JM. Solvent environment modulates e!ects of glutaraldehyde crosslinking on tissue derived materials. J Biomed Mater Res 1996;31:533}43. [32] Jorge-Herrero E, GutieH rrez M, Castillo-Olivares JL. Calci"cation of soft tissues employed in the construction of heart valve prostheses: study of di!erent chemical treatments. Biomaterials 1991; 12:249}52. [33] Carpentier A, Carpentier S, Nashef A, Ahmed A, Goussef N. Techniques for prevention of calci"cation of valvular bioprostheses. Circulation 1984;79(Suppl 1):165}8.
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767 [34] Arbustini E, Jones M, Moses R. Modi"cation by the Hancock T6 process of calci"cation of bioprosthetic cardiac valves implanted in sheep. Am J Cardiol 1984;57:167}74. [35] Timoshenko S. In: Resistencia de Materiales, vol. 1 (Spanish translation of Strength of materials). Madrid: Espasa Calpe, 1970. p. 163, 272. [36] Schoen FJ, Levy RJ, Hilbert Sl, Bianco RW. Antimineralization treatments for bioprosthetic heart valves. Assessment of e$cacy and safety. J Thorac Cardiovasc Surg 1992;104:1285}8. [37] Munro AI, Jamieson WRE, Tyers GFO, Germann E. Tricuspid valve replacement: porcine bioprostheses and mechanical prostheses. Ann Thorac Surg 1995;59:S470}4.
2767
[38] Myken PS, Berggren HL, Larsson S, Roberts B, Wallentin I, Caidahl K. Long-term Doppler echocardiographic results of aortic or mitral valve replacement with Biocor porcine bioprosthesis. J Thorac Cardiovasc Surg 1998;116:599}608. [39] Hiester ED, Sacks MS. Optimal bovine pericardial tissue selection sites. II. Cartographic analysis. J Biomed Mater Res 1998;39:215}21. [40] Braile DM, Soares MJ, Souza DR, Ramirez DA, Suzigan S, Godoy MF. Mapping of bovine pericardium: physical and histopathologic test. J Heart Valve Dis 1998;7:202}6. [41] Mausner J, Bahn. In: Mausner J, Kramer S, editors. Epidemiology: an introductory text. Philadelphia, W.B. Saunders Company, 1985.
Chemical treatment and tissue selection: factors that in#uence the mechanical behaviour of porcine pericardium JoseH M. GarcmH a PaH ez *, Eduardo Jorge-Herrero, Antonio Carrera, Isabel MillaH n, Aurora Rocha, P. Calero, Angeles CordoH n, JoseH Salvador, Natividad Sainz, JesuH s MeH ndez, JoseH L. Castillo-Olivares Services of Preventive Medicine, Puerta de Hierro Clinic (Cln& nica Puerta de Hierro), San Martin de Porres, 4, 28035 Madrid, Spain Experimental Surgery, Puerta de Hierro Clinic (Cln& nica Puerta de Hierro), San Martin de Porres, 4, 28035 Madrid, Spain Superior Technical School of Industrial Engineering (Escuela Te& cnica Superior de Ingenieros Industriales), Madrid, Spain Biostatistics, Puerta de Hierro Clinic (Cln& nica Puerta de Hierro), San Martin de Porres, 4, 28035 Madrid, Spain Received 19 July 2000; accepted 8 January 2001
Abstract Calci"cation and mechanical failure are the major causes of the loss of cardiac bioprostheses. The chemical treatments used to stabilize the tissue employed are considered to play a fundamental role in the development of these two phenomena, although the problem is multifactorial and the underlying causes are yet to be fully identi"ed. Currently, there is an ongoing search for chemical treatments capable of reducing or eliminating the process of calci"cation while preserving the mechanoelastic characteristics of the tissue. One of the approaches to this e!ort is the elimination of the phospholipid component from the biological tissue employed in prosthesis construction. There is evidence that this component may be responsible for the precipitation of calcium salts. The present study compares two delipidating chemical treatments involving chloroform/methanol and sodium dodecyl sulfate (SDS) with the use of glutaraldehyde (GA) alone. For this purpose, porcine pericardial tissue was subjected to tensile strength testing employing a hydraulic simulator. A total of 234 samples were studied 90 treated with GA, 72 treated with chloroform/methanol and 72 treated with SDS. The mean breaking strength was signi"cantly higher in the samples treated with GA (between 43.29 and 63.01 MPa) when compared with those of tissue treated with chloroform/methanol (29.92}42.30 MPa) or with SDS (13.49}19.06 MPa). In a second phase of the study, selection criteria based on morphological and mechanical factors were applied to the pericardial membranes employing a system of paired samples. The mathematical analysis of the "ndings in one fragment will aid in determining the mechanical behavior of its adjacent twin sample. In conclusion, the anticalci"cation chemical treatments tested in the experimental model conferred a lesser mechanical resistance than that obtained with GA. On the other hand, the utilization of paired samples was found to be useful in the prediction of the mechanical behavior of porcine pericardial tissue. Nevertheless, in order for our method of selection to be considered the most adequate approach, it will be necessary to validate these "ndings in dynamic studies involving a real, functional model. 2001 Elsevier Science Ltd. All rights reserved. Keywords: Biomaterials; Pericardium; Mechanical properties; Chemical treatments
1. Introduction The development of biomaterials for the medical "eld is increasingly necessary. To improve their resistance, compatibility and durability will be a great challenge for the coming years. One example of the need for these e!orts is illustrated by the limited use of cardiac bioprostheses due to problems concerning their durability. Valve * Corresponding author. Tel.: #34-91-316-4040; fax: #34-91-3737667. E-mail address: [email protected] (E. Jorge-Herrero).
failure makes their replacement necessary within 10}15 years of implantation in 30% of cases [1]. The calci"cation of biological tissues, due to unknown reasons [2}4], mechanical fatigue [5}7] or both processes simultaneously [8}10] are responsible for this situation. Structural failure of bioprostheses can be produced by early tears in the tissue [11]. On other occasions, the problem develops over the medium or long term as a consequence of the deposition of calcium salts on the tissue. To date, there is no chemical treatment capable of preventing this from occurring, although a number of approaches are currently being studied [12}14].
0142-9612/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 2 - 9 6 1 2 ( 0 1 ) 0 0 0 1 9 - 9
2760
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
In 1986, Ross [15] attributed the injury to the vascular endothelium to the shearing force generated in patients with arterial hypertension as a consequence of the excessive tension and the perturbation of the blood #ow. He also reported that the lesion produced by shear stress was followed by the hardening of the vascular wall and, "nally, calci"cation. Shearing stress was most probably responsible for the limited durability of samples subjected to tensile testing, as we observed in trials carried out with a tissue similar to that used in the construction of the lea#ets for prostheses made of calf pericardium [6]. To reduce or impede calcium deposition is the objective of a number of studies [16,17], but this aim must be achieved with no loss of the mechanical properties of the tissue, and if possible, should result in their improvement. Glutaraldehyde (GA) is currently the "rst choice for the crosslinking of biological tissues [18], although a number of new chemical treatments are under investigation to test their capacity to eliminate the harmful e!ects of this dialdehyde. Thus, new forms of chemical stabilization are being introduced, such as treatment with carbodiimide [19,20], diphenylphosphorylazide [21,22] polyepoxy resins [23], aminooleic acid [24] and diisocyanate hexamethylene [25,26]. All these agents and procedures must be tested to demonstrate their e$cacy for crosslinking and their lack of any role in favoring the process of calci"cation, although GA , despite its disadvantages, presents an entire series of properties that make it the reagent of choice for the time being: it reduces biodegradation, it preserves the anatomic integrity and physical features of the tissue, its cost is low, it is easy to obtain, it is water soluble and it does not promote thrombogenicity. Moreover, the use of GA for tissue "xation is associated with a good hemodynamic behavior [27]. On the other hand, there are a number of clear disadvantages such as the calci"cation of the implants and the cytotoxicity of this reagent [3,18,27}29]. Changes in the mechanical properties of the treated tissue have also been reported, including increased rigidity, the appearance of zones of greater concentration of internal shear and reduced strength [30]. It is therefore necessary to search for alternatives to this agent, either the use of one or more of the new reagents mentioned above, or some additional chemical treatment that impedes the calci"cation of the tissue. In this respect, our group has studied several anticalci"cation procedures in bovine pericardial tissue [2,4,10,16]. In order for a chemical treatment to be used to reduce or impede the deposition of calcium on bioprostheses, it is increasingly important to study the solvents in which the biological tissues are prepared and their e!ects on both the mechanical and elastic properties of these tissues [31]. In this study, we have analyzed these mechanoelastic properties in porcine pericardium exposed to a standard treatment with GA, comparing them to those observed in tissue samples treated previously with chloroform/meth-
anol or with sodium dodecyl sulfate (SDS). The use of chloroform/methanol for lipid extraction was e!ective against mineralization of the tissue, reducing calci"cation [2,4,16,32]. SDS acts directly on the tissue phospholipids [33,34] and is capable of reducing calci"cation in tissue implanted subcutaneously, but it is not as e!ective when exposed to the circulation [34]. The trials involved the use of a hydraulic simulator capable of delivering the pressure caused by the compression of a saline solution by a piston to a chemically stabilized pericardial membrane. The "nal objective of this work is to develop a new in vitro system using a hydraulic simulator to establish criteria for the selection of tissues according to the type of chemical treatment to which they have been exposed, and to apply these criteria to predict the mechanical behavior.
2. Materials and methods Porcine pericardium was obtained directly from a local abattoir and transported in cold isotonic saline (0.9% sodium chloride) to the laboratory where it was cleaned. The parietal pericardial sac was employed and the tissue samples were obtained in such a way that they all presented similar morphology (Fig. 1). Each membrane thus obtained measured approximately 10}20 cm from base to apex and 10 cm wide. Six circles with a radius of 1 cm were cut from each pericardial sac (Fig. 1). In all, 234 circular membranes were employed for each assay, 117 from the region referred to as B (pericardium covering the right ventricle) and 117 from the region referred to as C (pericardium covering the left ventricle). Each region was divided into three
Fig. 1. Illustration of a pericardial sac employed in the trial. Three symmetrical pairs of circular membranes are obtained for a total of 39 specimens, one each from the following zones of the regions corresponding to right ventricle (region B) and to left ventricle (region C): upper outer zone (zone 1), central zone (zone 2) and lower outer zone (zone 3).
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
symmetrical zones*zone 1 (upper outer zone), zone 2 (central zone) and zone 3 (lower outer zone)*each of which yielded one of the six circular tissue samples. Three groups were established, including a control group, treated with glutaraldehyde alone (GA). It was composed of 90 samples (45 obtained from region B and 45 from region C) which were treated for 24 h with 0.625% GA, prepared from a commercial solution of 25% GA (Merck), at a ratio of 1/50 (w/v), in 0.1 M sodium phosphate bu!er, pH 7.4. Another group consisting of 72 samples (36 from region B and 36 from region C) was pretreated with chloroform/methanol (CM) 1 : 4 (v/v) for 2 h at a proportion of 1 : 30 (p/v) with respect to the amount of tissue, followed by a treatment similar to that employed in the GA group. The third group was made up of 72 membranes (36 from region B and 36 from region C) that were pretreated for 2 h with 1% SDS in 0.15 M NaCl at a ratio of 1/30 (w/v), followed by the same GA treatment as above. The thickness of each membrane was determined by serial readings at 10 di!erent points using a digital Mitutoyo micrometer (Elecount digital series E/A33/8) with a precision at 203C of $3 . Each membrane was subjected to increasing stress until rupture, which was accompanied by a loss of stress and con"rmed macroscopically by the presence of a tear in the tissue.
2761
Fig. 2. Clamping system to ensure the immobility of the samples during the trial.
2.1. Assay method The assay was carried out on a hydraulic simulator capable of delivering increasing stresses to the pericardial membranes secured with pressure clips (Fig. 2). The membranes were exposed to increasing hydrostatic pressure caused by the compression transmitted to saline solution by a piston. As the piston moved, the #uid deformed the membrane and a pressure gauge determined the pressure, ranging between 0 and 16 atm. The simulator consisted basically of a unit for measuring pressure equipped with a servomotor to drive the pump propelling the piston. The entire system is illustrated in Fig. 3. 2.2. General description of the function of system A piston was activated by means of a digital monitor based on a high-speed processor that controlled the direct current electric servomotor. The piston compressed the #uid and the pericardial membrane resisted the pressure. The tissue was subjected to continuous, increasing pressure until rupture. The controlling computer indicated the angular velocity of the activating system, which was maintained throughout the trial. The data acquisition system evaluated the #uid pressure and the movement of the piston at all times. The numerical data corresponding to these variables were transferred to a computer via a series interface, where they were stored for subsequent analysis.
Fig. 3. View of the equipment. Pressure pistons.
2.3. Technical features of the hydraulic simulator The most relevant technical speci"cations are as follows: Ampli"er. D-MOS technology; H-bridge con"guration; maximum working voltage: 53 V; maximum intensity in steady state: 3 A. Motor. Rated voltage: 24 DCV; rated output: 15 W; starting torque: 120 mN m; current in a vacuum: 21.7 mA; starting current: 3040 mA; maximum permanent torque: 30.46 mN m. Incremental position sensor: optical; two quadrature outputs and index impulse. Quadrature processor: programmable logic technology; two quadrature inputs; incremental/decremental monopolar impulse output; maximum working frequency of 4 MHz. Digital compensator: processor, RISC microcontroller, 24 MHz, 8 bits, 200 ns/instruction; maximum sampling frequency of 1 kHz; velocity range from 00 to 8388607 counts/sampling period *256; proportional action coe$cient (KP) of !32768 to 32767; di!erential action coe$cient (KD) of !32768 to 32767.
2762
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
Piston: 160-mm stroke; 32 mm in diameter; maximum pressure 16 atm. Pressure sensor: maximum pressure 16 atm; output signal: 4}20 mA. Computer: standard Pentium-75 con"guration. This system was developed by Sat Polar, S.L., a Spanish electronic engineering "rm.
tion, that is deformation, was not considered to be of interest).
2.4. Tensile strength
2.6.4. Selection criteria Selection criteria were established to ensure greater homogeneity of the samples. The purpose of these statistical selection criteria was to determine the probability that each membrane tested actually belonged to the region or zone to which it was assigned in the initial selection. Thus, those membranes with a minimum thickness greater than the mean value for the corresponding treatment group, region and zone plus one standard deviation or less than the mean value minus one standard deviation were excluded, as were those membranes in which the di!erence between the mean thickness for of a given zone in the two regions and the minimum thickness for the corresponding zone was greater than the mean value for this di!erence determined for the corresponding series, plus one standard deviation. The pairs of membranes (one each from regions B and C) in which the stress (MPa) for x"1 for each zone of region C was above or below the mean plus or minus one standard deviation for said zone were also excluded. It is assumed that the values for samples from B would be unknown since they would be obtained from the projection of the results in region C in each treatment group. Thus, the expected values for B (dependent variables), predicted on the basis of the "ndings in region C (independent variables) would be compared with the actual values for region B determined subsequently for each respective projection.
Once the pressure withstood by the pericardial membrane at each instant of the trial was known, its tensile strength was calculated using a formula similar to the Laplace equation, described by Timoshenko [35] for a thin-walled membrane subjected to pressure: T " pr/2e, where p is the pressure in kg/cm, r the radius of the spherical membrane expressed in cm, e the thickness of the membrane in cm and T the tensile strength in kg/cm. To express this value in MPa, we divided the result by 10.19. 2.5. Elongation The movement of the piston indicated the variation in the #uid volume at every moment and for each di!erent pressure applied and, thus, the changes in membrane geometry up to the moment of rupture. At that point, the shape was that of a round bonnet the base of which was a known circle (the frame on which the membrane to be tested was mounted). By measuring the changes in length of the longest arc of the bonnet, it was possible to determine the percentage of elongation at each moment of the trial. 2.6. Statistical study and mathematical analysis 2.6.1. Comparison of means at rupture The mean values at rupture for groups GA, CM and SDS were compared, taking into account the di!erent regions (B and C), by means of Student's t test and the Newman}Keuls test. 2.6.2. Mathematical xt of the tensile strength/elongation ratio The tensile strength (MPa)/elongation (per unit) ratio was studied using the least squares method. The best "t corresponded to a third-order polynomial, the shape of which is expressed as y"b x#b x#b x, where y is the tensile strength in MPa and x is the per unit elongation of the membrane (the value of the constant b was made to equal zero since due to biological considerations, the equation must pass through the origin; at zero stress, there would be no deformation). For these same considerations, the analysis was done for x(1 (the behavior of the function when the membrane had surpassed its elastic limit, entering the realm of irreversible elonga-
2.6.3. Mean overall xt for each treatment group in each of the two regions The tensile strength/elongation ratio was also studied by group, by region and by zone.
2.6.5. Mean overall xt in the selected zones On the basis of the aforementioned criteria, the following 41 of the 234 sample pairs tested (35.04%) were selected: GA group: region B, zone 1 (B1)/region C, zone 1 (C1): pairs 9, 11, 12 and 13 B2/C2: pairs 1, 2, 7, 8, 9 and 10 B3/C3: pairs 8, 9 and 10 CM group: B1/C1: pairs 4, 6 and 10 B2/C2: pairs 1, 2, 4 and 12 B3/C3: pairs 2, 4, 8, 10, 11 and 12 SDS group: B1/C1: pairs 1, 4, 9 and 11 B2/C2: pairs 1, 5, 10, 11 and 12 B3/C3: pairs 1, 3, 4, 5, 9 and 11
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
2.6.6. Predictive study A predictive study of the values for region B was performed after the selection process had been carried out and the values for the equivalent region C samples were known. This determination involved the mathematical calculation of the values for the selected pairs according to the aforementioned criteria, using linear regression analysis, where the values for region C were known independent variables and those of region B were predictive, dependent variables. The tensile strength (MPa) in region B (y ) was estimated on the basis of that of region C (y ), and the 95% con"dence intervals were calculated. The estimated values for y were then compared with the respective known values for y . Correlation and regression studies to compare the three treatment groups were also carried out according to region (B or C) and zone (1, 2 or 3). A regression curve gave the best mathematical "t.
3. Results 3.1. Breaking strength The results at breaking point are shown in Table 1. The pericardial membranes treated with GA, whether Table 1 Mean values at breaking point for each zone according to chemical treatment and region Trial
GA B C B C B C CM B C B C B C SDS B C B C B C
No. of samples
Mean breaking strength (MPa)
Standard deviation
Range
15 15 15 15 15 15
59.51 61.84 43.29 62.07 63.01 61.61
11.60 22.65 17.78 13.51 17.38 16.99
43.43}83.27 13.00}71.70 23.18}90.64 26.28}72.27 21.91}81.46 25.25}92.79
12 12 12 12 12 12
35.88 28.93 29.92 42.30 38.84 35.17
8.76 12.58 11.31 15.21 12.60 13.01
17.08}49.34 10.32}52.37 16.40}45.75 21.06}62.90 21.75}70.55 16.03}60.11
12 12 12 12 12 12
13.98 13.49 19.06 18.69 19.06 16.26
5.17 6.81 6.62 10.49 6.46 5.06
7.98}23.14 3.07}25.73 10.76}27.82 10.90}49.61 5.68}29.90 8.99}24.86
from region B or region C, were more resistant to breakage, in terms of mean values, than those treated with CM or with SDS. The mean breaking strength of the samples treated with GA and of those treated with CM was signi"cantly greater ( p(0.001) than that of the SDS series. However, when GA and CM treatment groups were compared, statistically signi"cant di!erences were observed only in zones 1 ( p(0.001) and 2 ( p(0.05) of region B. The values for the GA group ranged from 13 to 92.79 MPa, those of CM samples between 10.32 and 62.90 MPa and those of samples treated with SDS from 3.07 to 49.61 MPa. 3.2. Mathematical xtting of the tensile strength/elongation ratio Table 2 displays the coe$cient for the equations corresponding to the mean values for the GA, CM and SDS groups in each region and zone, as well as the coe$cients of determination (R). The model that best "ts the tensile strength/elongation ratio is a third-order polynomial. The application of the selection criteria described in the material and methods section revealed a clear improvement in the coe$cients of determination (Table 3), all of which were greater than or equal to 0.9.
Table 2 Mathematical "t of the tensile strength/deformation ratio for each sample series Trials
The values for the GA group and the CM group were signi"cantly higher than those of the SDS group in every case (p(0.001). The values in the GA group were signi"cantly higher than those of the CM group in zone 1 (p(0.001) and zone 2 (p(0.05) of region B. GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate.
2763
GA B C B C B C CM B C B C B C SDS B C B C B C
a
a
a
R
15.38 12.79 12.87 14.96 10.96 15.05
!28.47 !25.39 !24.99 !27.19 !19.38 !32.31
31.58 28.67 28.73 29.32 21.48 34.14
0.810 0.781 0.853 0.896 0.802 0.865
11.30 10.62 12.19 14.27 9.50 15.06
!17.79 !18.71 !21.87 !27.77 !18.69 !30.58
22.09 21.23 25.54 32.67 21.40 35.21
0.890 0.950 0.847 0.869 0.706 0.939
12.79 10.57 10.07 11.34 10.51 10.97
!24.46 !18.07 !16.88 !21.10 !21.26 !16.98
31.69 23.90 22.42 27.50 24.90 25.23
0.544 0.898 0.914 0.966 0.951 0.952
This table presents the coe$cients obtained after the mathematical "tting of the tensile strength/deformation curves, as well as the determination coe$cient of the "t. GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate; R, coe$cient of determination.
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
2764
Table 3 Mathematical "t of the tensile strength/deformation ratio for each sample series after application of the selection criteria
Table 4 Prediction of the tensile strength (MPa) after application of the selection criteria
Trials
Trials/region
GA B C B C B C CM B C B C B C SDS B C B C B C
a
a
a
R
15.55 10.33 13.95 16.43 12.49 18.11
!26.68 !15.45 !27.40 !30.06 !22.99 !38.31
29.98 17.17 31.13 31.56 26.41 43.34
0.942 0.961 0.939 0.844 0.920 0.967
12.49 9.74 11.82 11.25 8.77 16.16
!22.99 !13.17 !19.01 !20.79 !16.52 !33.24
26.41 15.49 22.11 23.15 18.56 38.21
0.920 0.918 0.930 0.843 0.931 0.930
6.91 19.10 11.08 8.84 10.51 10.02
!6.25 !53.63 !22.15 !7.58 !21.26 !13.13
11.66 60.22 27.14 15.27 24.90 21.90
0.950 0.974 0.980 0.962 0.951 0.946
The coe$cients expressed in Table 2 after application of the selection criteria (see Section 2). GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate; R, coe$cient of determination.
3.3. Prediction of the tensile strength The use of linear regression analysis to "t the data from each region and the corresponding zones for each treatment group revealed excellent correlation for the tensile strength values (Table 4). These results were obtained after the application of the selection criteria described above. The values for region B (dependent variables) for each trial were obtained from region C (independent variable). The coe$cients of determination (R) were greater than 0.99 in every case. The analysis of the correlation between the tensile strengths found for the di!erent treatments, regions and zones appear in Table 5. The coe$cients of determination were 0.94 or higher in every case (over 0.99 in all but two), indicating that the behavior of a given zone can be predicted on the basis of the "ndings in the corresponding zone and region of samples treated with di!erent chemical agents. By way of example, "gure illustrates the comparison of the mechanical behavior (tensile strength in MPa) between zones 1 and 3 of region B.
4. Discussion The introduction of di!erent crosslinking and anticalcifying chemical treatments for biological tissues has ren-
GA B B B CM B B B SDS B B B
a (95% CI)
a (95% CI)
R
!0.22 (!0.15, 0.11) !0.47(!0.27,!0.67) 0.35 (0.24, 0.46)
1.56 (1.53, 1.59) 0.98 (0.78, 1.02) 0.42 (0.41, 0.44)
0.999 0.996 0.996
!0.33(!0.08,!0.57) 0.06(!0.03, 0.15) 0.15 (0.08, 0.21)
1.25 (1.19, 1.30) 1.13 (1.11, 1.15) 0.52 (0.45, 0.62)
0.994 0.999 0.999
0.69 (0.18, 1.19) !0.16 (!0.45, 0.14) 0.10 (!0.06, 0.26)
0.56 (0.48, 0.63) 0.87 (0.82, 0.92) 0.69 (0.66, 0.72)
0.945 0.989 0.996
Regression curves where y is the dependent variable and y the independent variable (see Section 2.6.6). GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate; R, coe$cient of determination.
Table 5 Coe$cients of the regression curves that correlate the applied tensile strength (MPa) in terms of the di!erent treatment groups, regions and zones Independent variable (x)
Dependent variable (y)
Region Zone a
a
R
CM
GA
B
1.22 1.05 0.89 0.95 1.29 1.07 1.35 1.06 0.69 0.59 0.90 1.05 1.11 1.00 0.81 0.63 0.70 0.94
0.999 0.997 0.994 1.000 0.998 1.000 0.995 0.999 0.992 0.940 0.991 0.993 0.994 0.993 0.999 0.935 0.993 0.992
C
SDS
GA
B
C
SDS
CM
B
C
1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
0.14 !0.12 0.20 0.05 0.23 0.04 0.43 0.22 0.34 0.65 0.65 0.40 0.24 0.33 0.08 0.65 0.32 0.46
GA, glutaraldehyde; CM, chloroform/methanol; SDS sodium dodecyl sulfate; R, coe$cient of determination.
dered them for use in clinical applications, one of which is the construction of cardiac bioprostheses. The phenomena of calci"cation, stress and tissue perforation are intimately correlated, resulting in the degeneration of the native structure. Di!erent chemical treatments a!ect the tissue ultrastructure, often modifying the native conformations of the tissue's protein, mainly collagen, that is most prevalent in bovine or porcine pericardium. The degree of change produced by chemical solvents would
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
be especially important in the mechanical performance of biological tissues [36]. To improve the mechanoelastic properties of the tissue and reduce or prevent calcium deposition on the tissue would increase the functionality and durability of and structure made from that biomaterial. To achieve this, there are two complementary strategies that consist in designing e!ective chemical treatments and biomaterials that are more resistant or have less propensity toward calcium deposition. It has been demonstrated that cardiac bioprostheses implanted in tricuspid position and subjected to mild physiological stress do not become calci"ed [12}14,37]. On the other hand, in 1998, Myken et al. [38] observed that 16 of 30 dysfunctional porcine bioprostheses showed signs of calci"cation, all with a peak gradient of over 60 mmHg. As it is not always possible to reduce stress, it is obvious that we must prevent calci"cation by chemical means. This study, unfortunately, demonstrates a statistically signi"cant loss of breaking strength ( p(0.001) in tissue samples treated with SDS (Table 1). This loss is evident in both the pericardium covering right ventricle (region B) and that of left ventricle (region C). It is less marked, however, in samples treated with CM and is statistically signi"cant ( p(0.001) in only two zones (1 and 2) of region B. The width of the ranges, on the other hand, re#ects the lack of homogeneity of the tissue, making it necessary to select the most suitable specimens. The results obtained with recently described selection processes have not been de"nitive [39,40], not even those of Hiester and Sacks who, in 1998, utilized small-angle light scattering (SALS), a nondestructive optical technique, to quantify the collagen "ber architecture of soft tissues [39]. That same year, Braile et al. reported their conclusions using a method that combines mechanical and histopathological features [40], but were not able to provide a de"nitive selection strategy either. The technique proposed in the present report consists of the use of paired samples [41]. The process consists of the selection of two adjacent tissue samples, one of which will be subjected to mechanical assays, while the other (Fig. 1) is left intact and its theoretical parameters calculated on the basis of the mathematical "tting of the stress/elongation curve. This method substantially improves the "t of the mean stress/deformation, or tensile strength/deformation, curves, providing mathematical functions to de"ne the mechanical behavior with excellent determination coe$cients (Table 3). These functions can be employed to compare the viscoelastic properties achieved with di!erent chemical treatments (GA, CM, SDS) in tissue samples obtained from each region and zone of the pericardial sac. In addition, the use of paired samples, according to a statistical methodology described elsewhere [41] made it possible to perform a predictive analysis of the outcome in intact tissue prior to its treatment, based on the knowledge of the mechanical
2765
behavior of the other half of the pair. This method helps to establish the relationships between the di!erent chemical treatments and zones employed, as shown in Table 5, which presents the mathematical functions that correlate the tensile strength (MPa) applied in each assay and permit the comparison of the resulting mechanical behaviors. Figs. 4 and 5 display the regression curves resulting from this comparison. For a tensile strength of 0.2 MPa, which is similar to the working stress that a cardiac valve lea#et made of a tissue treated with SDS will withstand, the theoretic stress that a tissue treated with GA or with CM could withstand would range between 1.2 and 3.5-fold greater (0.24}0.70 MPa) for tissue from zones 1 and 3 of region B (Fig. 4) and 2.3}4.2-fold greater (0.46}0.83 MPa) for samples from zone 2 of region C (Fig. 5), with the higher values corresponding to samples treated with GA. This lack of strength at a low level of stress shown by tissue treated with SDS, as well as its low breaking strength, can severely in#uence the durability of devices made from biomaterials subjected to this treatment. In general, samples exposed to CM also show a loss of strength, although less marked than those treated with SDS, and only its anticalci"cation properties would argue in favor of recommending its use, but with precaution.
Fig. 4. Regression curves for zones 1 and 3 of region B, comparing the tensile strength (MPa) according to the chemical treatment received. Glut: glutaraldehyde. Cl/M: chloroform/methanol. SDS: sodium dodecyl sulfate.
Fig. 5. Regression curves for zone 2 of region C, comparing the tensile strength (MPa) according to the chemical treatment received. Glut: glutaraldehyde. Cl/M: chloroform/methanol. SDS: sodium dodecyl sulfate.
2766
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767
In conclusion, any new method of "xation of a biomaterial requires a study of its impact on the mechanical behavior of the material. A loss of strength like that demonstrated in the present report in tissue exposed to SDS eliminates any advantages to the use of this substance. The method of selection employed is suitable for static testing, as described here, but it will have to be validated in dynamic studies of a real, functional model if it is to be considered an adequate approach to tissue selection.
Acknowledgements This study was "nanced by Grants Nos. 96/0250 and 00/0192 from the Fondo de Investigaciones Sanitarias (FIS), Spain. The authors are grateful to M. Messman for her translation of the text.
References [1] Vongpatanasin W, Hillis LD, Lange RA. Medical progress: prosthetic heart valves. N Engl J Med 1996;335:407}16. [2] Jorge-Herrero E, Fernandez P, GutieH rrez M, Castillo-Olivares JL. Study of calci"cation of bovine pericardium: analysis of the implication of lipids and proteoglycans. Biomaterials 1991;12:683}9. [3] Levy RJ. Glutaraldehyde and the calci"cation mechanism of bioprosthetic heart valves. J Heart Valve Dis 1994;3:101}4. [4] Jorge-Herrero E, FernaH ndez P, Escudero C, Castillo-Olivares JL. Inhibition of the calci"cation of porcine valve tissue by selective lipid removal. Biomaterials 1994;15:815}20. [5] Bortolotti U, Milano A, Thiene G, Guerra F, Mazzuco A, Valente M, Talenti E, Galluci V. Early mechanical failures of the Hancock pericardial xenograft. J Thorac Cardiovasc Surg 1987;94:200}7. [6] GarcmH a PaH ez JM, Carrera A, GarcmH a Sestafe JV, MillaH n I, Jorge E, Candela I, Castillo-Olivares JL. Is cutting stress responsible for the limited durability of heart valve bioprostheses? J Thorac Cardiovasc Surg 1990;100:580}6. [7] Shoen FJ, Horson CE. Anatomic analysis of removed prosthetic heart valves: causes of failure of 33 mechanical valves and 58 bioprostheses, 1980}1983. Hum Pathol 1985;16:549}59. [8] Thubrikar MJ, Deck DJ, Aouad J. Role of mechanical stress in calci"cation of aortic bioprosthetic valves. J Thorac Cardiovasc Surg 1983;86:115}25. [9] Gabbay S, Kadan P, Factor S, Cheung JK. Do heart valve bioprostheses degenerate for metabolic or mechanical reasons? J Thorac Cardiovasc Surg 1988;95:208}15. [10] Jorge-Herrero E, FernaH ndez P, Escudero C, de la Torre N, GarcmH a PaH ez JM, Castillo-Olivares JL. In#uence of stress on the calci"cation of delipidated bovine pericardial tissue employed in the construction of cardiac valves. J Biomed Mater Res 1996; 30:411}5. [11] Hariza F, Papouin G, Barratt-Boyes B, Christie G, Whitlock R. Tear in bioprosthetic heart valve lea#ets without calci"c degeneration. J Heart Valve Dis 1996;5:35}9. [12] Grunkemeier GL, Jamieson WRE, Miller DC, Starr A. Actuarial versus actual risk of porcine structural valve deterioration. J Thorac Cardiovasc Surg 1994;108:709}18. [13] Tyers GFO, Jamieson WR, Munro IA, Germann E, Burr LH, Miyagishima RT, Ling L. Reoperation in biological and mechanical valve populations: fate of the reoperative patient. Ann Thorac Surg 1995;60:S464}9.
[14] Kent PD, Tazelaar HD, Edwards WD, Orszulak TA. Temporal changes in the surgical pathology of prosthetic aortic valves: a study of 157 cases spanning 26 years (1970}1995). Cardiovasc Pathol 1998;7:9}23. [15] Ross R. The pathogenesis of atherosclerosis: an update. N Engl J Med 1986;314:488}500. [16] Jorge-Herrero E, GutieH rrez M, Castillo-Olivares JL. Inhibition of bovine pericardium calci"cation: a comparative study of Al> lipid removing treatments. J Mater Sci: Mater Med 1991;2:86}8. [17] Khor E, Wee A, Tan BL, Chew TY. The e$cacy of a dual anticalci"cation agent treatment in the prevention of calci"cation of glutaraldehyde-"xed porcine pericardium. J Mater Sci: Mater Med 1995;6:518}22. [18] Jayakrishan A, Jameela SR. Glutaraldehyde as a "xative in bioprostheses and drug delivery matrices. Biomaterials 1996;17:471}84. [19] Nimni ME, Cheung D, Strates B, Kodama M, Sheilkh K. Chemically modi"ed collagen: a natural biomaterial for tissue replacement. J Biomed Mater Res 1987;21:741}71. [20] van Wachem PB, van Luyn MJA, Olde Damink LHH, Dijkstra PJ, Feijen J, Nieuwenhius P. Biocompatibility and tissue regenerating capacity of crosslinked dermal sheep collagen. J Biomed Mater Res 1994;28:353}63. [21] Petite H, Frei V, Huc A, Herbage D. Use of diphenylphosphorylazide for cross-linking collagen based biomaterials. J Biomed Mater Res 1994;28:159}65. [22] Petite H, Duval JL, Abdul Malak N, Sigot MF, Frei V, Herbage D. Cytocompatibility of calf pericardium treated by glutaraldehyde and by acyl-azide methods in an organotypic culture model. Biomaterials 1995;16:1003}8. [23] Xi T, Liu F. E!ect of treatment with epoxy compounds on the mechanical properties of bovine pericardial bioprosthetic materials. J Biomater Appl 1994;7:61}75. [24] Chen W, Kim JD, Schoen FJ, Levy RJ. 2-Aminooleic acid derivation of glutaraldehyde pretreated bioprosthetic tissues for counteracting calci"cation: a review and comparison of its e$cacy and mechanism in various animal models. Cells Mater 1994;4:419}28. [25] Naimark WA, Pereira CA, Tsang K, Lee JM. HDMC crosslinking of bovine pericardial tissue: a potential role of the solvent environment in the design of bioprosthetic materials. J Mater Sci: Mater Med 1995;6:235}41. [26] Van Luyn MJ, Van Wachem PB, Dijkstra P, Olde Damnik LH, Feijen J. Calci"cation of subcutaneously implanted collagens in relation to cytotoxicity, cellular interaction and crosslinking. J Mater Sci: Mater Med 1995;6:288}96. [27] Duncan AC, Boughner D, Vesely I. Dynamic glutaraldehyde "xation of a porcine aortic valve xenograft. Biomaterials 1996; 17:1849}56. [28] Hughes H, Lilburn S, Tipton S, Aboul-Enein HY, Duran CG. Chemical assay of glutaraldehyde incorporation into pericardial tissue. J Heart Valve Dis 1994;3:105}9. [29] Olde Damnik LH, Dijkstra P, Van Luyn MJ, Van Wachem PB, Feijen J. Glutaraldehyde as a crosslinking agent for collagen based materials. J Mater Sci: Mater Med 1995;6:460}72. [30] Talman EA, Boughner DR. Glutaraldehyde "xation alters the internal shear properties of porcine aortic heart valve tissue. Ann Thorac Surg 1995;60:s369}73. [31] Gratzer PF, Pereira CA, Lee JM. Solvent environment modulates e!ects of glutaraldehyde crosslinking on tissue derived materials. J Biomed Mater Res 1996;31:533}43. [32] Jorge-Herrero E, GutieH rrez M, Castillo-Olivares JL. Calci"cation of soft tissues employed in the construction of heart valve prostheses: study of di!erent chemical treatments. Biomaterials 1991; 12:249}52. [33] Carpentier A, Carpentier S, Nashef A, Ahmed A, Goussef N. Techniques for prevention of calci"cation of valvular bioprostheses. Circulation 1984;79(Suppl 1):165}8.
J.M. Garcn&a Pa&ez et al. / Biomaterials 22 (2001) 2759}2767 [34] Arbustini E, Jones M, Moses R. Modi"cation by the Hancock T6 process of calci"cation of bioprosthetic cardiac valves implanted in sheep. Am J Cardiol 1984;57:167}74. [35] Timoshenko S. In: Resistencia de Materiales, vol. 1 (Spanish translation of Strength of materials). Madrid: Espasa Calpe, 1970. p. 163, 272. [36] Schoen FJ, Levy RJ, Hilbert Sl, Bianco RW. Antimineralization treatments for bioprosthetic heart valves. Assessment of e$cacy and safety. J Thorac Cardiovasc Surg 1992;104:1285}8. [37] Munro AI, Jamieson WRE, Tyers GFO, Germann E. Tricuspid valve replacement: porcine bioprostheses and mechanical prostheses. Ann Thorac Surg 1995;59:S470}4.
2767
[38] Myken PS, Berggren HL, Larsson S, Roberts B, Wallentin I, Caidahl K. Long-term Doppler echocardiographic results of aortic or mitral valve replacement with Biocor porcine bioprosthesis. J Thorac Cardiovasc Surg 1998;116:599}608. [39] Hiester ED, Sacks MS. Optimal bovine pericardial tissue selection sites. II. Cartographic analysis. J Biomed Mater Res 1998;39:215}21. [40] Braile DM, Soares MJ, Souza DR, Ramirez DA, Suzigan S, Godoy MF. Mapping of bovine pericardium: physical and histopathologic test. J Heart Valve Dis 1998;7:202}6. [41] Mausner J, Bahn. In: Mausner J, Kramer S, editors. Epidemiology: an introductory text. Philadelphia, W.B. Saunders Company, 1985.