Tensile strength of the interface between hydroxyapatite and bone Lin Hong and Xu Hengchang Department of Dental Materials, School of Stornafology, Beijing Medical University,
Beijing, China K. de Groot* Department of Biomaterials, Biomaterials Research Group, Medical School, Leiden University, Leiden, The Netherlands Tensile strength of the interface between hydroxyapatite (HA) and bone was tested. Scanning electron microscopy was used to observe the tensile failure mode and the morphological change of hydroxyapatite ceramic surface in bone. The porosity of hydroxyapatite is 14% and pore size less than 2 pm. After 2 weeks of implantation, the tensile strength of the interface is 0.72 MPa. After 4, 8, and 16 weeks, the average tensile strength stayed at 1.5 MPa.
SEM showed that tensile failure occurred at the HA-bone interface at the second week, but after 4 weeks, the failure occ u r r e d between HA particles w i t h i n the bulk, and not at the HA-bone interface. Calcified tissue was directly deposited on the HA ceramic surface and exits also in the micropores. Near the interface, s i n t e r e d necks among HA ceramic p a r t i c l e s w e r e subjected t o biodegradation.
IN TRODUCTION
There are many biomaterials used for repair or reconstruction of the skeleton. Hydroxyapatite (HA) is one of those and subject of intensive investigation. HA is nontoxic and biocompatible with bony tissues. It is implanted either as bulk or as a coating on metal implants. The interface between bone and HA is an important topic of basic research. Many methods have been used to study the interface, such as histological, electron microscopic observation, and biomechanical test of its strength. With respect to the latter method, most researchers use the so-called push-out test to obtain an ultimate “shear-stress’’ of the bone-HA interface. However, data in the literature, obtained on plugs coated with HA, show a wide variability: from 5 MPa to more than 50 MPa. This variation does not depend on the nature of the interface but on the test itself.’ In addition, the shear strength may not reflect the actual stresses along a functioning implant-bone interface; not only shear but also tension occurs?,3 *To whom correspondence should be addressed. Journal of Biomedical Materials Research, Vol. 26, 7-18 (1992) CCC 0021-9304/92/010007-12$4.00 0 1992 John Wiley & Sons, Inc.
LIN, XU, AND DE GROOT
8
Therefore, the aim of the present study is to determine the tensile strength of the HA-bone interface. In addition, by means of SEM observation, we want to investigate the failure mode, and the changes of HA ceramic in bone. We have selected HA samples with a porosity similar to HA coatings, thus being able to draw conclusions with respect to those coatings, without the additional complication of having a second interface (HA-substrate).
MATERIALS
The implant material used was sintered microporous, crystallographically pure hydroxyapatite, with a porosity of 1470, pore size smaller than 2 pm. The tensile strength is 14.2 MPa and stoichiometry is Calo(P04)6(OH)2, the Ca/P ratio is 1.67.
METHODS
Evaluation of tensile strength of HA-bone interface The hydroxyapatite ceramic was shaped into a semicylindrical shape (Ushape) of 3 mm length and has an outer radius of 5.5 mm and inner radius of 3.5 mm [see Fig. l(a)]. Thereafter, the implant was polished with 400-grit abrasive paper and cleaned in an ultrasonic cleaner (ME 2.1, Mettler Electron Corp., USA), two times 5 min in distilled water, then 10 min in acetone. After cleaning again in distilled water, they were incubated in 75% ethanol for 2 h, and then autoclaved for half an hour. The semicylindrical implants were implanted into the diaphysis of the tibiae of mature male rabbits (2.0 to 3.0 kg). The rabbits were anesthetized by intravenous injection of 3% sodium pentobarbitol. The operations were performed under standard aseptic conditions. For implantation of HA ceramic sample into tibia, a longitudinal 3-cm skin incision was made on the anteromedial aspect of the left tibia of the rabbit. Extending from 1 cm to 4 cm below the knee, the skin, muscle, and periosteum were retracted. Using a parallel dental disc cutter (made of two disc cutters, separated from 3 mm) and a dental burr driven by a dental hand-drill machine and cooled with saline coolant, a 3-mm length semicylindrical bone defect was made [see Fig. l(b)]. In order to get a nonfunctional condition, the calcaneal tendon and the tendon of the quadriceps were tenotomized. The bone defect was irrigated with saline, then the semicylindrical HA ceramic was inserted in the cortical bone defect [see Fig. l(c) and Fig. 21. Forty-four experimental rabbits were randomly divided into four groups of 11 rabbits each. The groups were killed 2, 4, 8, and 16 weeks after the operation, respectively. A 4-cm-long segment of the tibia containing a HA ceramic in the middle was excised. The callus which covered the surface of HA ceramic and the
TENSILE STRENGTH OF HA-BONE INTERFACE
9
Bone
-
HA
Bone
Figure 1. Schematic drawing of the preparation of rabbit tibia bone for tensile strength test. (a) 3-mm-length semicylindrical (U-shape) HA ceramic, (b) 3-mm-length semicylindrical tibia1 cortical bone defect, (c) HA ceramic in situ, (d) tensile strength test sample. Tibia1 cortical bone segments contacted only with proximal and distal ends of HA ceramic. Bone segments were pulled perpendicularly to the interface in opposite direction. At the second tensile test, one of the two bone segments was replaced by a small plexiglass bar. (e) Interfacial area after HA ceramic detached from bone.
bony tissues located on the lateral sides of the HA ceramic were removed by a dental burr as shown in Figure l(d). Extreme care was taken to remove the bony tissues completely from the lateral sides and the surface of the HA ceramics. A hole of 1 mm diameter was made on each part of the bone segments, hemp ropes were through these holes and held by a hook of a Universal Test Machine (Autograph, DCS-5000, Shimadzu, Japan). The bone segments were pulled perpendicularly to the HA-bone interface in opposite direction at a cross-head speed of 2.5 m/min [see Fig. l(d) and Fig. 31. The load at which an implant was detached from the bone was recorded. After that, only one of the
10
LIN, XU, AND DE GROOT
Figure 2. HA implant in tibia before tensile test.
two HA-bone interfaces was detached (see Fig. 3). This is called "the first interfacial tensile test"; the HA ceramic which the unseparated bone-HA sample was glued to a small bar of plexiglass by cyanoacrylate adhesive. This second interface of HA-bone sample was tested again 1.5 h later. It was "the second interfacial tensile test." Most samples were done in this way. In this way we increased the number of measurements and, hence, the statistical relevance. The entire tensile test was performed within 3 h after the sacrifice of animals. Samples were kept moist during the test. The Image Analysis method (Interactive Image Analysis System, Semi Automatic Evaluation Unit, IBAS 2000, Conton Co., Germany) was used to mea-
Figure 3. Each sample has two HA-bone interfaces. This is after the first test situation, only upper bone was separated from HA, the lower bone segment still bonded to HA, allowing for a second test on this sample as described in the text.
TENSILE STRENGTH OF HA-BONE INTERFACE
11
sure the interfacial area after HA ceramic detached from bone [see Fig. l(e)]. Then the tensile strength of the interface HA-bone was calculated. Scanning electron microscopic (SEM) observation
After the tensile test, samples were prepared according to normal SEM samples preparation method. That is, samples were cleaned with normal saline, immersed in 3% glutaraldehyde-0.1M phosphorous buffer (PH 7.2) solution for 2-3 h at 4"C, then postfixed in 1% osmium tetroxide-0.1M phosphorous buffer solution at 4°C for 2 h, subsequently dehydrated in 50%, 70% 90%, and 100% ethanol for 1.5 h, iso-Amy1 acetate for 0.5 h. After that they were dried in COz critical point drying instrument (Autosamdri-810, Tousimis, USA), gold-plated by ion sputtering (JEC-1000, JEOL, Japan). The following samples were made: (1) Surface of HA before implantation. (2) Surface of fractured HA before implantation. (3) Surface of fractured HA-bone interface (bony side of the interface after the tensile failure). (4) Natural cross section of HA-bone interface. The natural cross section of HA-bone interface is made by the following procedures: One method is after the first tensile test, the undetached HAbone interface sample was split from the bone yielding a natural cross section of the interface (the split force is perpendicular to the interface). Then the samples were prepared according to the preparation procedure of SEM samples mentioned above. Another method is: after the first tensile test, the undetached HA-bone interface samples were fixed and postfixed according to the procedure mentioned above, then dehydrated in 50%, 70%,90%, 100% acetone for 1.5 h, and were embedded in epoxy resin Epon 812. After that they were split from bone side yielding a natural cross fracture of HA-bone interface. We believe, that natural cross section of implant-bone interface can reflect a more accurate relationship between implant and bone than do cut slides. The SEM used was Digital Scanning Electron Microscope, DSM 950, Opton Co., Germany. X-ray radiographic examination
After the animals were sacrificed, the implant sites were examined by xray. RESULTS
Gross observation and x-ray examination
Healing of implant site was uneventful in all animals. Four HA ceramic samples did not adhere to bone completely at one of the two interfacial areas:
LIN, XU, AND
12
DE
GROOT
small crevices could be seen at these areas. Besides these crevices, HA ceramics adhered to bone well. Periosteal and endosteal callus increased with the implantation time, some of them covered the whole HA ceramic surface. X-ray examination showed that the shape and density of HA implants did not change significantly with time, they were dense and even. There were no significant x-ray translucent areas at the interface and the density of interfacial tissues increased with implantation time.
Evaluation of tensile strength of HA-bone interface After the HA ceramic and bone were pulled apart, sometimes a small amount of fractured HA ceramic was left on the bony surface of some samples as could be seen with the naked eyes. Except that, for samples of 2, 4, and 8 weeks, fracture took place at the interface and not in either bone or HA. For the 16-week samples, there was a thin layer of HA ceramic left on the bone side after fracture, showing that failure occurred within the subsurface of HA. It seems that at 16 weeks the HA-bone interface is stronger in tension than the subsurface section of HA. In this study, each HA-bone sample has two HA-bone interfaces. The tensile strength of both interfaces were tested for most of the samples, as presented in Materials and Methods. Student’s t test showed no statistically significant differences between the mean of the first group data and the mean of the second group data in each experimental period (see Table I). So each periods strength value was obtained by putting the first group data and the second group data together (see Table 11). The results of Table I1 showed that 2,4,8, and 16 weeks after implantation, the tensile strength of HA-bone samples were 0.717 5 0.286 MPa, 1.375 5 0.627 MPa, 1.534 5 0.544 MPa, and 1.594 5 0.715 MPa, respectively. The data in Table I1 were analyzed using ANOVA analysis of variance and q-test. There were statistically significant differences between the data of 2 weeks and the data of other three periods ( p c 0.01). No statistically significant differences were observed among the data of 4, 8, and 16 weeks ( p > 0.05).
TABLE I Tensile Strength (MPa) of HA-Bone Interface Implant at ion Time (weeks)
1st Results (Means & SD)
2nd Results (Means + SD)
2 4 8 16
0.592 f 0.295 (8) 1.194 f 0.471 (7) 1.212 f 0.312 (6) 1.386 f 0.507 (9)
0.842 f 0.228 (8) 1.515 k 0.722 (9) 1.728 f 0.572 (10) 1.862 f 0.886 (7)
Numbers in parentheses indicate number of samples.
P >0.05 >0.05 >0.05
>0.05
TENSILE STRENGTH OF HA-BONE INTERFACE
13
TABLE I1 Tensile Strength (MPa) of HA-Bone Interface
Implantation Time (weeks)
Tensile Strength (Mean f SD)
2 4 8 16
0.717 f 0.286 (16) 1.375 2 0.627 (16) 1.534 k 0.544 (16) 1.594 2 0.715 (16)
Numbers in parentheses indicate number of samples.
SEM observation
Hydroxyapatite ceramic not implanted HA polyhedron crystal particles were dense and linked to each other by sintering. There were many pores among them; pore size was less than 2 pm. When fractured, the failure occurred through HA crystal particles themselves, not through necks connecting the particles (see Fig. 4).
HA-bone interface after various periods of implantation After tensile fracture of the 2, 4, and 8 weeks samples, some HA ceramic fragments were left on the bony surface as could be seen by naked eyes. These HA ceramic fragments had the same failure morphology as not implanted HA ceramic mentioned above, that is tensile failure occurred in the HA particles themselves, not through the sintered necks.
Figure 4. SEM photo of the fractured HA ceramic before implantation . failure occurred within the HA ce(original magnification ~ 5 0 0 0 )Tensile ramic particles themselves, not at the necks. Micropore size smaller than 2 pm.
14
LIN, XU, AND
DE
GROOT
The following SEM results of surfaces of fractured HA-bone interface regard the bony surface side of the failure. Two weeks after implantation: Bony side of the interface after the tensile failure: Surface of bone showed loose, irregular arranged fibrous tissues; amorphous substances; cells; and irregular and compact tissues. No HA ceramic crystal particles were left on bone surfaces after the tensile test. Cross sections: HA ceramic crystal particles had still a polyhedron shape, and were directly connected with the newly formed unmineralized fibrous tissues of the bony surface. Four, eight and sixteen weeks after implantation: Bony side of the interface after the tensile failure (see Fig. 5): At the 4th week most of the bone was compact with irregular arranged lamella, compact amorphous structures, and a few fibrous tissues. At 8th week and 16th week, the bone surfaces were mainly irregularly arranged, amorphous compact tissue. In these three periods, there were many HA ceramic implant particles left on the bony surface after the tensile test. Tensile failure occurred at the necks where HA ceramic particles sintered together (between the HA ceramic particles), rather than through the sintered particles themselves (inside of HA particles), showing that sintered necks of HA ceramic weakened after the HA ceramic implanted into the rabbit bone. At the 4th week, these HA ceramic particles were scattered on the bony surface, and increased at the 8th week. At the 16th week, almost the whole bony surface were covered by such HA ceramic particles. Cross sections of samples of 4, 8, and 16 week group were similar to each other (see Fig. 6): The tissues at the interface were compact. There was no distinct border between HA particles and bony tissues. The shape of HA particles did not
Figure 5. SEM photo of a failed HA-bone interface (bone side) of an , the center square original &week sample (original magnification ~ 2 0 0 0in magnification x 5000). Tensile faiiure occurred between the HA particles, i.e., the sintered necks fractured (see the center area).
TENSILE STRENGTH OF HA-BONE INTERFACE
15
Figure 6. SEM photo of a natural cross section of HA-bone interface (4 weeks after implantation: original magnification ~10,000).C: Fissurae occurred among HA particles near the interface. Tissue which identified with interfacial tissue exist in these fissurae. Bone: bony tissues at the interface. HA: HA ceramic particle.
change, but some fine fissura occurred among them near the interface, the necks among HA particles had disappeared. The micropores of the HA implant were not enlarged. Some irregular compact substances existed in these fissura and micropores and the morphology of the substances were the same as the bony tissues near the interface. DISCUSSION
The result of SEM observation With increasing implantation time, bony tissues at the interface continually formed, calcified and densified. This is in accordance with the observations of other researchers?
Bonding of the HA-bone interface Bony side of the interface showed that 4 weeks after the implantation there were some HA ceramic particles left on the bony end of the interface. At the cross sections, the osseous tissue directly attached to the particles of HA ceramic. Some tissuelike structure, identical with that on the interface, existed in the micropores and fissura among the HA ceramic particles. They were connected with those tissues at the interface. Hence, we conclude that HA is biocompatible with bony tissues. Calcified tissue can be directly deposited on the HA ceramic surface and may exist in the micropores within the HA. This was in accordance with the results of most researchers,>* who also found that HA is biocompatible with bony tissues.
16
LIN, XU, AND
DE
GROOT
Biodegradation of hydroxyapatite Cross sections showed that some changes had taken place within the HA ceramic near the interface. There were some small fissurae among the crystal particles of HA ceramic. The bony side of the interface of the 4 to 16-week samples showed that the sintered necks among H A ceramic crystal particles, left on the bone surface, had disappeared. However, the polyhedron shape of the HA crystal particles did not change. Tensile failure always occurred between, and not through, the HA ceramic particles, i.e., the sintered necks fractured. On the other hand, the tensile failure of not implanted H A ceramic occurred within the HA ceramic crystal particles themselves and not at the necks. It thus seemed that the biological environment changed the structure of the implanted HA near the interface, by weakening or dissolving the sintered necks among the HA particles. Thus the microporous H A used in this experiment may undergo some biodegradation after implantation in the rabbit tibia1 cortical bone for 2 to 16 weeks. Our findings are in agreement with those of van Blitterswijk, who found HA to resorb at a rate of 20-30 pm surface removal per year.' It can also be seen from this study that with increasing time, there were more HA particles left on the bony side of the interface after the tensile test. This can be explained by assuming that as time passed, the calcification degree of the tissues near the interface increased and the contact area between the calcified osseous tissue and HA ceramic enlarged. The tensile strength of the HA-bone interface The results of this study showed that the tensile strength of the HA-bone interface was 0.72 MPa two weeks after operation and more than 1.5 MPa 4 weeks later. The strength did not change during the period of 4 to 16 weeks after the operation. The average strength during this period was 1.5 MPa. Four weeks after the operation, the tensile strength of the interface was twice as much as that at the 2-week inverval. The result was influenced by the nature of the tissue at the interface and its degree of calcification, as well as the contact area of HA to the calcified osseous tissues. This was in accordance with the opinion of most researchers and also in agreement with our SEM observations. There are only a few reports about the bonding strength of the HA-bone interface. Most of them reIate to shear strength of the interface of which the results vary greatly. The results of our study are quite different from the shear strengths reported. We believe that interfacial tensile strength can reflect the bonding strength of the interface better than the push-out test. The SEM observation in this study showed that tensile failure of the HA-bone samples after 4 weeks implantation occurred by failure of the weakened necks among the HA ceramic crystal particles near the interface and not at the HA-bone interface itself. Thus the tensile strength of the HA-bone interface was higher than that of the HA ceramic near the interface.
TENSILE STRENGTH OF HA-BONE INTERFACE
17
The actual tensile strength of the interface measured in this study was much lower than that of HA which had not been implanted (14.2 MPa). The reasons for this are as follows: As we have already discussed, SEM showed that the structure of this microporous HA ceramic near the interface had changed during the implantation. Biodegradation occurred by weakening the necks between the HA ceramic crystal particles. The strength of the HA ceramic near the interface is lower than that of either bulk H A ceramic, or the osseous tissues, or the HA-bone interface. Therefore, tensile failure occurred at the weakened area near the interface. Hence, the results of the tests after 4 weeks of implantation in this study was the tensile strength of the locally degraded HA ceramic, while the in vitro tensile test of HA ceramic indicated the tensile strength of the (sintered) HA particles themselves, rather than that of the in vivo weakened necks. If we may translate these results to HA-coatings, reported to have the same porosity (5-15%) as our HA samples tested, we conclude that because the same weakness of exposed necks will take place, they will gradually degrade and be replaced by osseous tissue. In addition, their true in vivo tensile strength will be in the order of 1MPa. Push-out shear tests will therefore not reflect the interfacial strength but a stress need to pull an irregular surface out of bone. CONCLUSIONS
(1)The results of this implant-bone interfacial tensile strength test method reflects the actual tensile strength of the tensile failure, independent of the failure mode. (2) The microporous HA used in this study has a good biocompatibility with bone. Bony tissues can deposit directly on the HA ceramic surface, and may exist within the micropores of the ceramic. (3) Our HA ceramic was subjected to biodegradation in rabbit tibial cortical bone. The degradation is limited to the sintered necks between HA ceramic crystal particles, and happened mainly at the HA ceramic near the interface. (4) During the 16 weeks implantation of HA ceramic in the rabbit tibial cortical bone, its tensile strength near the interface dropped from 14.2 MPa to 1.5 MPa. Four weeks after the implantation, the failure mode shifts from the HA-bone interface to the subsurface of HA. References 1. T. Fujiu and M. Ogino, ”Difference of bond bonding behaviour among surface active glasses and sintered apatite,“ 1. Biomed. Mater. Res., 18, 845-859 (1984). 2. T. Kitsugi, T. Yamamuro, T. Nakamura, S. Higashi, Y. Kakutani, K. Hyakuna, S. Ito, T. Kokubo, M. Takagi, and T. Shibuya, “Bone bonding behaviour of three kinds of apatite containing glass ceramics,” 1. Biomed. Mater. Res., 20, 1295-1307 (1986). 3. J. B. Park, B. J. Kelly, G. H. Kenner, A. F. Von Recum, M. F. Grether, and W.W. Coffeen, ”Piezoelectric ceramic implants: in vivo results,” 1. Biomed. Mafer. Xes., 15, 103-110 (1981).
LIN, XU, AND DE GROOT
18
4. M. Jarcho, V. Jasty, K.I. Gumaer, J. F. Kay, and R. H. Doremus, ”Electron microscopic study of a bone-hydroxylapatite implant interface,” in: Trans. 4th Ann. Meet. SOC.Biomater. 10th Ann. Infer. Biomater. Symp. 1978. Vol. 11, San Antonio, Texas, pp. 112-113. 5. M. Jarcho, J.F. Kay, K.I. Gumaer, R.H. Doremus, and H.P. Drobeck, “Tissue, cellular and subcellular events at a bone-ceramic hydroxyapatite interface,” J. Bioeng. 1, 79-92 (1977). 6. C. A. Van Blitterswijk, J. J. Grote, W. Kuijpers, W.Th. Daems, and K. de Groot, ”Macropore tissue ingrowth: A quantitative and qualitative study on hydroxyapatite ceramic,” Biomaterials, 7, 137-143 (1986). 7. K. Kato, H. Aoki, T. Tabata, and M. Ogiso, ”Biocompatibility of apatite ceramics in mandibles,” Biomat. Med. Dev. A r f . Org., 7(2), 291-297 (1979). 8. J. F. Osborn and H. Newesely, “Bonding osteogeneses induced by calciumphosphate ceramic implants,” in Biomaterials 1980, Winter, G. D., Gibbons, D. F., and Plenk, H., Jr. (eds), 1982, John Wiley and Sons Ltd., New York, 1982, pp. 51-58. Received November 2,1990 Accepted June 3, 1991
Beijing, China K. de Groot* Department of Biomaterials, Biomaterials Research Group, Medical School, Leiden University, Leiden, The Netherlands Tensile strength of the interface between hydroxyapatite (HA) and bone was tested. Scanning electron microscopy was used to observe the tensile failure mode and the morphological change of hydroxyapatite ceramic surface in bone. The porosity of hydroxyapatite is 14% and pore size less than 2 pm. After 2 weeks of implantation, the tensile strength of the interface is 0.72 MPa. After 4, 8, and 16 weeks, the average tensile strength stayed at 1.5 MPa.
SEM showed that tensile failure occurred at the HA-bone interface at the second week, but after 4 weeks, the failure occ u r r e d between HA particles w i t h i n the bulk, and not at the HA-bone interface. Calcified tissue was directly deposited on the HA ceramic surface and exits also in the micropores. Near the interface, s i n t e r e d necks among HA ceramic p a r t i c l e s w e r e subjected t o biodegradation.
IN TRODUCTION
There are many biomaterials used for repair or reconstruction of the skeleton. Hydroxyapatite (HA) is one of those and subject of intensive investigation. HA is nontoxic and biocompatible with bony tissues. It is implanted either as bulk or as a coating on metal implants. The interface between bone and HA is an important topic of basic research. Many methods have been used to study the interface, such as histological, electron microscopic observation, and biomechanical test of its strength. With respect to the latter method, most researchers use the so-called push-out test to obtain an ultimate “shear-stress’’ of the bone-HA interface. However, data in the literature, obtained on plugs coated with HA, show a wide variability: from 5 MPa to more than 50 MPa. This variation does not depend on the nature of the interface but on the test itself.’ In addition, the shear strength may not reflect the actual stresses along a functioning implant-bone interface; not only shear but also tension occurs?,3 *To whom correspondence should be addressed. Journal of Biomedical Materials Research, Vol. 26, 7-18 (1992) CCC 0021-9304/92/010007-12$4.00 0 1992 John Wiley & Sons, Inc.
LIN, XU, AND DE GROOT
8
Therefore, the aim of the present study is to determine the tensile strength of the HA-bone interface. In addition, by means of SEM observation, we want to investigate the failure mode, and the changes of HA ceramic in bone. We have selected HA samples with a porosity similar to HA coatings, thus being able to draw conclusions with respect to those coatings, without the additional complication of having a second interface (HA-substrate).
MATERIALS
The implant material used was sintered microporous, crystallographically pure hydroxyapatite, with a porosity of 1470, pore size smaller than 2 pm. The tensile strength is 14.2 MPa and stoichiometry is Calo(P04)6(OH)2, the Ca/P ratio is 1.67.
METHODS
Evaluation of tensile strength of HA-bone interface The hydroxyapatite ceramic was shaped into a semicylindrical shape (Ushape) of 3 mm length and has an outer radius of 5.5 mm and inner radius of 3.5 mm [see Fig. l(a)]. Thereafter, the implant was polished with 400-grit abrasive paper and cleaned in an ultrasonic cleaner (ME 2.1, Mettler Electron Corp., USA), two times 5 min in distilled water, then 10 min in acetone. After cleaning again in distilled water, they were incubated in 75% ethanol for 2 h, and then autoclaved for half an hour. The semicylindrical implants were implanted into the diaphysis of the tibiae of mature male rabbits (2.0 to 3.0 kg). The rabbits were anesthetized by intravenous injection of 3% sodium pentobarbitol. The operations were performed under standard aseptic conditions. For implantation of HA ceramic sample into tibia, a longitudinal 3-cm skin incision was made on the anteromedial aspect of the left tibia of the rabbit. Extending from 1 cm to 4 cm below the knee, the skin, muscle, and periosteum were retracted. Using a parallel dental disc cutter (made of two disc cutters, separated from 3 mm) and a dental burr driven by a dental hand-drill machine and cooled with saline coolant, a 3-mm length semicylindrical bone defect was made [see Fig. l(b)]. In order to get a nonfunctional condition, the calcaneal tendon and the tendon of the quadriceps were tenotomized. The bone defect was irrigated with saline, then the semicylindrical HA ceramic was inserted in the cortical bone defect [see Fig. l(c) and Fig. 21. Forty-four experimental rabbits were randomly divided into four groups of 11 rabbits each. The groups were killed 2, 4, 8, and 16 weeks after the operation, respectively. A 4-cm-long segment of the tibia containing a HA ceramic in the middle was excised. The callus which covered the surface of HA ceramic and the
TENSILE STRENGTH OF HA-BONE INTERFACE
9
Bone
-
HA
Bone
Figure 1. Schematic drawing of the preparation of rabbit tibia bone for tensile strength test. (a) 3-mm-length semicylindrical (U-shape) HA ceramic, (b) 3-mm-length semicylindrical tibia1 cortical bone defect, (c) HA ceramic in situ, (d) tensile strength test sample. Tibia1 cortical bone segments contacted only with proximal and distal ends of HA ceramic. Bone segments were pulled perpendicularly to the interface in opposite direction. At the second tensile test, one of the two bone segments was replaced by a small plexiglass bar. (e) Interfacial area after HA ceramic detached from bone.
bony tissues located on the lateral sides of the HA ceramic were removed by a dental burr as shown in Figure l(d). Extreme care was taken to remove the bony tissues completely from the lateral sides and the surface of the HA ceramics. A hole of 1 mm diameter was made on each part of the bone segments, hemp ropes were through these holes and held by a hook of a Universal Test Machine (Autograph, DCS-5000, Shimadzu, Japan). The bone segments were pulled perpendicularly to the HA-bone interface in opposite direction at a cross-head speed of 2.5 m/min [see Fig. l(d) and Fig. 31. The load at which an implant was detached from the bone was recorded. After that, only one of the
10
LIN, XU, AND DE GROOT
Figure 2. HA implant in tibia before tensile test.
two HA-bone interfaces was detached (see Fig. 3). This is called "the first interfacial tensile test"; the HA ceramic which the unseparated bone-HA sample was glued to a small bar of plexiglass by cyanoacrylate adhesive. This second interface of HA-bone sample was tested again 1.5 h later. It was "the second interfacial tensile test." Most samples were done in this way. In this way we increased the number of measurements and, hence, the statistical relevance. The entire tensile test was performed within 3 h after the sacrifice of animals. Samples were kept moist during the test. The Image Analysis method (Interactive Image Analysis System, Semi Automatic Evaluation Unit, IBAS 2000, Conton Co., Germany) was used to mea-
Figure 3. Each sample has two HA-bone interfaces. This is after the first test situation, only upper bone was separated from HA, the lower bone segment still bonded to HA, allowing for a second test on this sample as described in the text.
TENSILE STRENGTH OF HA-BONE INTERFACE
11
sure the interfacial area after HA ceramic detached from bone [see Fig. l(e)]. Then the tensile strength of the interface HA-bone was calculated. Scanning electron microscopic (SEM) observation
After the tensile test, samples were prepared according to normal SEM samples preparation method. That is, samples were cleaned with normal saline, immersed in 3% glutaraldehyde-0.1M phosphorous buffer (PH 7.2) solution for 2-3 h at 4"C, then postfixed in 1% osmium tetroxide-0.1M phosphorous buffer solution at 4°C for 2 h, subsequently dehydrated in 50%, 70% 90%, and 100% ethanol for 1.5 h, iso-Amy1 acetate for 0.5 h. After that they were dried in COz critical point drying instrument (Autosamdri-810, Tousimis, USA), gold-plated by ion sputtering (JEC-1000, JEOL, Japan). The following samples were made: (1) Surface of HA before implantation. (2) Surface of fractured HA before implantation. (3) Surface of fractured HA-bone interface (bony side of the interface after the tensile failure). (4) Natural cross section of HA-bone interface. The natural cross section of HA-bone interface is made by the following procedures: One method is after the first tensile test, the undetached HAbone interface sample was split from the bone yielding a natural cross section of the interface (the split force is perpendicular to the interface). Then the samples were prepared according to the preparation procedure of SEM samples mentioned above. Another method is: after the first tensile test, the undetached HA-bone interface samples were fixed and postfixed according to the procedure mentioned above, then dehydrated in 50%, 70%,90%, 100% acetone for 1.5 h, and were embedded in epoxy resin Epon 812. After that they were split from bone side yielding a natural cross fracture of HA-bone interface. We believe, that natural cross section of implant-bone interface can reflect a more accurate relationship between implant and bone than do cut slides. The SEM used was Digital Scanning Electron Microscope, DSM 950, Opton Co., Germany. X-ray radiographic examination
After the animals were sacrificed, the implant sites were examined by xray. RESULTS
Gross observation and x-ray examination
Healing of implant site was uneventful in all animals. Four HA ceramic samples did not adhere to bone completely at one of the two interfacial areas:
LIN, XU, AND
12
DE
GROOT
small crevices could be seen at these areas. Besides these crevices, HA ceramics adhered to bone well. Periosteal and endosteal callus increased with the implantation time, some of them covered the whole HA ceramic surface. X-ray examination showed that the shape and density of HA implants did not change significantly with time, they were dense and even. There were no significant x-ray translucent areas at the interface and the density of interfacial tissues increased with implantation time.
Evaluation of tensile strength of HA-bone interface After the HA ceramic and bone were pulled apart, sometimes a small amount of fractured HA ceramic was left on the bony surface of some samples as could be seen with the naked eyes. Except that, for samples of 2, 4, and 8 weeks, fracture took place at the interface and not in either bone or HA. For the 16-week samples, there was a thin layer of HA ceramic left on the bone side after fracture, showing that failure occurred within the subsurface of HA. It seems that at 16 weeks the HA-bone interface is stronger in tension than the subsurface section of HA. In this study, each HA-bone sample has two HA-bone interfaces. The tensile strength of both interfaces were tested for most of the samples, as presented in Materials and Methods. Student’s t test showed no statistically significant differences between the mean of the first group data and the mean of the second group data in each experimental period (see Table I). So each periods strength value was obtained by putting the first group data and the second group data together (see Table 11). The results of Table I1 showed that 2,4,8, and 16 weeks after implantation, the tensile strength of HA-bone samples were 0.717 5 0.286 MPa, 1.375 5 0.627 MPa, 1.534 5 0.544 MPa, and 1.594 5 0.715 MPa, respectively. The data in Table I1 were analyzed using ANOVA analysis of variance and q-test. There were statistically significant differences between the data of 2 weeks and the data of other three periods ( p c 0.01). No statistically significant differences were observed among the data of 4, 8, and 16 weeks ( p > 0.05).
TABLE I Tensile Strength (MPa) of HA-Bone Interface Implant at ion Time (weeks)
1st Results (Means & SD)
2nd Results (Means + SD)
2 4 8 16
0.592 f 0.295 (8) 1.194 f 0.471 (7) 1.212 f 0.312 (6) 1.386 f 0.507 (9)
0.842 f 0.228 (8) 1.515 k 0.722 (9) 1.728 f 0.572 (10) 1.862 f 0.886 (7)
Numbers in parentheses indicate number of samples.
P >0.05 >0.05 >0.05
>0.05
TENSILE STRENGTH OF HA-BONE INTERFACE
13
TABLE I1 Tensile Strength (MPa) of HA-Bone Interface
Implantation Time (weeks)
Tensile Strength (Mean f SD)
2 4 8 16
0.717 f 0.286 (16) 1.375 2 0.627 (16) 1.534 k 0.544 (16) 1.594 2 0.715 (16)
Numbers in parentheses indicate number of samples.
SEM observation
Hydroxyapatite ceramic not implanted HA polyhedron crystal particles were dense and linked to each other by sintering. There were many pores among them; pore size was less than 2 pm. When fractured, the failure occurred through HA crystal particles themselves, not through necks connecting the particles (see Fig. 4).
HA-bone interface after various periods of implantation After tensile fracture of the 2, 4, and 8 weeks samples, some HA ceramic fragments were left on the bony surface as could be seen by naked eyes. These HA ceramic fragments had the same failure morphology as not implanted HA ceramic mentioned above, that is tensile failure occurred in the HA particles themselves, not through the sintered necks.
Figure 4. SEM photo of the fractured HA ceramic before implantation . failure occurred within the HA ce(original magnification ~ 5 0 0 0 )Tensile ramic particles themselves, not at the necks. Micropore size smaller than 2 pm.
14
LIN, XU, AND
DE
GROOT
The following SEM results of surfaces of fractured HA-bone interface regard the bony surface side of the failure. Two weeks after implantation: Bony side of the interface after the tensile failure: Surface of bone showed loose, irregular arranged fibrous tissues; amorphous substances; cells; and irregular and compact tissues. No HA ceramic crystal particles were left on bone surfaces after the tensile test. Cross sections: HA ceramic crystal particles had still a polyhedron shape, and were directly connected with the newly formed unmineralized fibrous tissues of the bony surface. Four, eight and sixteen weeks after implantation: Bony side of the interface after the tensile failure (see Fig. 5): At the 4th week most of the bone was compact with irregular arranged lamella, compact amorphous structures, and a few fibrous tissues. At 8th week and 16th week, the bone surfaces were mainly irregularly arranged, amorphous compact tissue. In these three periods, there were many HA ceramic implant particles left on the bony surface after the tensile test. Tensile failure occurred at the necks where HA ceramic particles sintered together (between the HA ceramic particles), rather than through the sintered particles themselves (inside of HA particles), showing that sintered necks of HA ceramic weakened after the HA ceramic implanted into the rabbit bone. At the 4th week, these HA ceramic particles were scattered on the bony surface, and increased at the 8th week. At the 16th week, almost the whole bony surface were covered by such HA ceramic particles. Cross sections of samples of 4, 8, and 16 week group were similar to each other (see Fig. 6): The tissues at the interface were compact. There was no distinct border between HA particles and bony tissues. The shape of HA particles did not
Figure 5. SEM photo of a failed HA-bone interface (bone side) of an , the center square original &week sample (original magnification ~ 2 0 0 0in magnification x 5000). Tensile faiiure occurred between the HA particles, i.e., the sintered necks fractured (see the center area).
TENSILE STRENGTH OF HA-BONE INTERFACE
15
Figure 6. SEM photo of a natural cross section of HA-bone interface (4 weeks after implantation: original magnification ~10,000).C: Fissurae occurred among HA particles near the interface. Tissue which identified with interfacial tissue exist in these fissurae. Bone: bony tissues at the interface. HA: HA ceramic particle.
change, but some fine fissura occurred among them near the interface, the necks among HA particles had disappeared. The micropores of the HA implant were not enlarged. Some irregular compact substances existed in these fissura and micropores and the morphology of the substances were the same as the bony tissues near the interface. DISCUSSION
The result of SEM observation With increasing implantation time, bony tissues at the interface continually formed, calcified and densified. This is in accordance with the observations of other researchers?
Bonding of the HA-bone interface Bony side of the interface showed that 4 weeks after the implantation there were some HA ceramic particles left on the bony end of the interface. At the cross sections, the osseous tissue directly attached to the particles of HA ceramic. Some tissuelike structure, identical with that on the interface, existed in the micropores and fissura among the HA ceramic particles. They were connected with those tissues at the interface. Hence, we conclude that HA is biocompatible with bony tissues. Calcified tissue can be directly deposited on the HA ceramic surface and may exist in the micropores within the HA. This was in accordance with the results of most researchers,>* who also found that HA is biocompatible with bony tissues.
16
LIN, XU, AND
DE
GROOT
Biodegradation of hydroxyapatite Cross sections showed that some changes had taken place within the HA ceramic near the interface. There were some small fissurae among the crystal particles of HA ceramic. The bony side of the interface of the 4 to 16-week samples showed that the sintered necks among H A ceramic crystal particles, left on the bone surface, had disappeared. However, the polyhedron shape of the HA crystal particles did not change. Tensile failure always occurred between, and not through, the HA ceramic particles, i.e., the sintered necks fractured. On the other hand, the tensile failure of not implanted H A ceramic occurred within the HA ceramic crystal particles themselves and not at the necks. It thus seemed that the biological environment changed the structure of the implanted HA near the interface, by weakening or dissolving the sintered necks among the HA particles. Thus the microporous H A used in this experiment may undergo some biodegradation after implantation in the rabbit tibia1 cortical bone for 2 to 16 weeks. Our findings are in agreement with those of van Blitterswijk, who found HA to resorb at a rate of 20-30 pm surface removal per year.' It can also be seen from this study that with increasing time, there were more HA particles left on the bony side of the interface after the tensile test. This can be explained by assuming that as time passed, the calcification degree of the tissues near the interface increased and the contact area between the calcified osseous tissue and HA ceramic enlarged. The tensile strength of the HA-bone interface The results of this study showed that the tensile strength of the HA-bone interface was 0.72 MPa two weeks after operation and more than 1.5 MPa 4 weeks later. The strength did not change during the period of 4 to 16 weeks after the operation. The average strength during this period was 1.5 MPa. Four weeks after the operation, the tensile strength of the interface was twice as much as that at the 2-week inverval. The result was influenced by the nature of the tissue at the interface and its degree of calcification, as well as the contact area of HA to the calcified osseous tissues. This was in accordance with the opinion of most researchers and also in agreement with our SEM observations. There are only a few reports about the bonding strength of the HA-bone interface. Most of them reIate to shear strength of the interface of which the results vary greatly. The results of our study are quite different from the shear strengths reported. We believe that interfacial tensile strength can reflect the bonding strength of the interface better than the push-out test. The SEM observation in this study showed that tensile failure of the HA-bone samples after 4 weeks implantation occurred by failure of the weakened necks among the HA ceramic crystal particles near the interface and not at the HA-bone interface itself. Thus the tensile strength of the HA-bone interface was higher than that of the HA ceramic near the interface.
TENSILE STRENGTH OF HA-BONE INTERFACE
17
The actual tensile strength of the interface measured in this study was much lower than that of HA which had not been implanted (14.2 MPa). The reasons for this are as follows: As we have already discussed, SEM showed that the structure of this microporous HA ceramic near the interface had changed during the implantation. Biodegradation occurred by weakening the necks between the HA ceramic crystal particles. The strength of the HA ceramic near the interface is lower than that of either bulk H A ceramic, or the osseous tissues, or the HA-bone interface. Therefore, tensile failure occurred at the weakened area near the interface. Hence, the results of the tests after 4 weeks of implantation in this study was the tensile strength of the locally degraded HA ceramic, while the in vitro tensile test of HA ceramic indicated the tensile strength of the (sintered) HA particles themselves, rather than that of the in vivo weakened necks. If we may translate these results to HA-coatings, reported to have the same porosity (5-15%) as our HA samples tested, we conclude that because the same weakness of exposed necks will take place, they will gradually degrade and be replaced by osseous tissue. In addition, their true in vivo tensile strength will be in the order of 1MPa. Push-out shear tests will therefore not reflect the interfacial strength but a stress need to pull an irregular surface out of bone. CONCLUSIONS
(1)The results of this implant-bone interfacial tensile strength test method reflects the actual tensile strength of the tensile failure, independent of the failure mode. (2) The microporous HA used in this study has a good biocompatibility with bone. Bony tissues can deposit directly on the HA ceramic surface, and may exist within the micropores of the ceramic. (3) Our HA ceramic was subjected to biodegradation in rabbit tibial cortical bone. The degradation is limited to the sintered necks between HA ceramic crystal particles, and happened mainly at the HA ceramic near the interface. (4) During the 16 weeks implantation of HA ceramic in the rabbit tibial cortical bone, its tensile strength near the interface dropped from 14.2 MPa to 1.5 MPa. Four weeks after the implantation, the failure mode shifts from the HA-bone interface to the subsurface of HA. References 1. T. Fujiu and M. Ogino, ”Difference of bond bonding behaviour among surface active glasses and sintered apatite,“ 1. Biomed. Mater. Res., 18, 845-859 (1984). 2. T. Kitsugi, T. Yamamuro, T. Nakamura, S. Higashi, Y. Kakutani, K. Hyakuna, S. Ito, T. Kokubo, M. Takagi, and T. Shibuya, “Bone bonding behaviour of three kinds of apatite containing glass ceramics,” 1. Biomed. Mater. Res., 20, 1295-1307 (1986). 3. J. B. Park, B. J. Kelly, G. H. Kenner, A. F. Von Recum, M. F. Grether, and W.W. Coffeen, ”Piezoelectric ceramic implants: in vivo results,” 1. Biomed. Mafer. Xes., 15, 103-110 (1981).
LIN, XU, AND DE GROOT
18
4. M. Jarcho, V. Jasty, K.I. Gumaer, J. F. Kay, and R. H. Doremus, ”Electron microscopic study of a bone-hydroxylapatite implant interface,” in: Trans. 4th Ann. Meet. SOC.Biomater. 10th Ann. Infer. Biomater. Symp. 1978. Vol. 11, San Antonio, Texas, pp. 112-113. 5. M. Jarcho, J.F. Kay, K.I. Gumaer, R.H. Doremus, and H.P. Drobeck, “Tissue, cellular and subcellular events at a bone-ceramic hydroxyapatite interface,” J. Bioeng. 1, 79-92 (1977). 6. C. A. Van Blitterswijk, J. J. Grote, W. Kuijpers, W.Th. Daems, and K. de Groot, ”Macropore tissue ingrowth: A quantitative and qualitative study on hydroxyapatite ceramic,” Biomaterials, 7, 137-143 (1986). 7. K. Kato, H. Aoki, T. Tabata, and M. Ogiso, ”Biocompatibility of apatite ceramics in mandibles,” Biomat. Med. Dev. A r f . Org., 7(2), 291-297 (1979). 8. J. F. Osborn and H. Newesely, “Bonding osteogeneses induced by calciumphosphate ceramic implants,” in Biomaterials 1980, Winter, G. D., Gibbons, D. F., and Plenk, H., Jr. (eds), 1982, John Wiley and Sons Ltd., New York, 1982, pp. 51-58. Received November 2,1990 Accepted June 3, 1991
