mechanical strength of poly(methyl methacrylate) cement-human bone interfaces

Mechanical strength of poly(methy1 methacrylate) cement-human bone interfaces

R. Kusleika Drpnrtiircnt of Mntcrials Scicncc and E?rglifccrfng, Northumtern University, Eunnston, Illinois 6020 1

Bir~crr,yiiircrir~g Program and Dcprtrncnt of Ccramic Enginccring and Polymer Group, University of llliirors at Url~aira-Cltampaigrr, Urhana, Illinois 61801

s. I. Stupp'

A device was constructed to test the interfacial strength of PMMA-based bone cement and human 'cancellous bone under pure tension. Two types of tissue were used in the investigation: (1) formalin-fixed vertebral bone as an I J I urtro model for weak cancellous bone, and (2) freshly removed metatarsal bone. Tissue-cement joints were allowed to solidify under two different pressures (0.11 and 0.47 MPa), and cement placement time on tissue surfaces was also controlled as a variable. The higher curing pressure only seemed to enhance the strength of interfaces .formed with mechanically weak fixed bone but had no significant effect for joints formed with the stronger, freshly extracted tissue. Cement placement time did not have a discernible

effect on interfacial strength regardless of the tissue used or the pressure applied during setting. An analysis of fracture morphology by optical microscopy revealed largely cement cohesive failure in some cases and bone or mixed fractures in others. Joints exhibiting mainly cement fracture had the highest interfacial tensile strengths (in the order of 7.5 MPa). Once measured values of tissue porosity were taken into account, the observed joint strength corre- lated well with cement tensile strength. Based on experimental findings, better stress-dissipating qualities and higher tensile strength are suggested as two important necessary improvements of bone cements based on poly(methy1 methacrylate).

INTRODUCTION

The formulation and use of poly(methy1 methacrylate)-based bone cements for fixation of metallic endoprostheses in joint surgery has been a significant development over the past two decades. Despite a reasonable Success rate in terms of prostheses fixation and generally improved stress distribution at the implant site, several problems are encountered in the use of acrylic cements. Thermal necrosis due to the setting exotherm1F2 and monomer-induced hy- potension3 are two examples. The mechanical strength of bone cements4 and their creep behavior5 have been characterized. It is not presently understood, however, which should be the ideal physical properties of this biomaterial. In spite of the difficulties described above, the use of acrylic bone cements ' To whom correspondence should be addressed

Journal of Biomedical Materials Research, Vol. 17,441-458 (1983) Q 1983 John Wiley & Sons, Inc. CCC 0021-9304/83/030441-18SO2.80

442 KUSLEIKA AND STUPP

continues to be the best alternative for fixation of endoprostheses in joint surgery.

An important problem encountered in the use of bone cements is their nonadhesive nature toward bone surfaces. It is widely thought that the source of adhesive strength of the interface is mechanical interlocking, which can result from flow of cement through pores on the surface of cancellous bone. The low bonding strength of cement-bone interfaces is believed to be a con- tributing factor in clinical loosening of prostheses. Clinical reports of postoperative loosening of prostheses are often found in the l i t e r a t ~ r e . ~ - ~ One- study6 collected information on over 6,000 reported cases of implanted prostheses and observed femoral stem loosening in only 1.4% of the cases. Al- though the report included cases with postoperative periods of up to 10 years, the median prostheses implantation period was 2 years. In a different investigation,’ over 300 cases were followed for periods of 4-7 years, and it was found that 24% of the femoral stems had loosened. A strong correlation between loosening rate and the nature of the cement-tissue bond was also found. Specifically, poor cement packing and insufficient removal of cancellous bone were cited as two major reasons causing postoperative femoral stem loosening. This latter work suggests that the nature of the cement-tissue bond is a critical factor in long-term prevention of femoral prostheses loosening.

At the present time, it is not known to what extent handling variables control the long-term mechanical strength of the bone-acrylic cement interface. Furthermore, with only a few exceptions,lOJ1 in vifro measurements of mechanical strength at the interface which could serve as reference values for improved formulations, improved handling techniques, or stabilization by osseous ingrowth are not readily available in the literature. Therefore, one objective of this work has been to develop a test method and a quantitative assessment of mechanical strength of the cement-bone interface. The effects of several variables on interfacial strength have been investigated, namely, time elapsed between cement mixing and cement placement onto the bone surface, the intensity of low curing pressures, and bone substrate orientations. Also, results have been compared for tests on both formalin-fixed and freshly removed human bone. A second major objective of the present work has been to seek hints from interfacial failure modes regarding beneficial changes in cement mechanical properties for improvement of clinical performance.

EXPERIMENTAL

Bone samples used in this work were obtained from freshly removed human metatarsals following amputation procedures. Bones were dissected of soft tissues and stored in refrigerated saline for a maximum time of 18 davs prior to testing. A cold cure acrylic resin was placed around the medial section of the bone to permit mounting in a vise (see Fig. 1 j. Metatarsals were then sectioned using an alumina cutoff wheel and copious water lubrication. Those sections containing portions of the medullary canal were discarded. Forma- lin-fixed bone samples were obtained from cadaveric vertebral bodies. The

PMMA-HUMAN BONE INTERFACES 443

CLEANED

READY TO SECTION

Figure 1. metatarsals.

Initial preparation of specimens from freshly extracted human

vertebral bodies were embedded in a cold-cure acrylic resin and sectioned using an alumina cutoff wheel and copious water lubrication (see Fig. 2). Samples of formalin-fixed bone were prepared in two different orientations. Longitudinal sections were cut from the bodies along the sagittal plane, and transverse ones were obtained by cutting normal to the spinal axis. All bone sections were trimmed to '/4 in. square samples using a jeweler's saw and water lubrication (Fig. 3). The thickness of the experimental bone specimens ranged between 4 and 6 mm. Samples were beveled with a scalpel and selected for approximately uniform pore size, then embedded into the face of an acrylic cylinder for subsequent adhesion testing. Cortical bone was not present within the surface area used for testing. In order to insure minimal heating of bone samples due to the setting exotherm of cold-cure mounting acrylic, the cylinders were prefabricated with a cavity at one end and bone samples were later mounted into the cavity using only a small amount of cold-cure acrylic. Extreme care was taken to assure alignment of the bone surface per- pendicular to the axis of the acrylic cylinder. At this stage, the sections prepared were ready for adhesion tests.

A commercial PMMA-based bone cement' was used for all adhesion tests. The cement powder was divided into 100 mg quantities and mixed according

Simplex-P, Howmedica, Inc.


0 I

YOUKTED BODY

L 0 N G IT UMN A L TRANSVERSE SECTION SECTION

Figure 2. Initial preparation of specimens from fixed vertebral bodies.

A BONE SECTION 0

1

EMBEDDED 5*ypLE

1

Figure 3. samples obtained as indicated in Figs. 1 and 2.

Final preparation of bone substrates for adhesion testing, cut from

PMMA-HUMAN BOh’E INTERFACES 445

to manufacturers’ instructions. A glass dappen dish and a stainless steel spatula were used for mixing. The powderlliquid ratio was carefully main- tained a t 2:1, and time-related handling variables were carefully monitored using an electronic timer, accurate to f0.1 s.

Bone surface porosity was estimated through intercept analysis of photomicrographs. A polished and dyed surface of bone was photographed, and nine random lines drawn on the photo. Percentage porosity was calculated as the average percent of a line’s dimension that crossed porous regions on the photomicrographs. Average pore size was calculated from the average length of line that resided within a pore. Assuming spherical pores, these data were used to estimate the average number of pores per square centimeter.

ADHESION TESTING APPARATUS

Tensile adhesive butt joints of human bone and acrylic bone cement were prepared using an apparatus specially constructed in the laboratory (see Fig. 4). The apparatus was made of stainless steel throughout, except for a Teflon ring at the end of the compressor. The bone sample, embedded within the acrylic cylinder, is positioned in a sample holder and held in place with set

TO LOAD CELL

STEEL ROD-

S-HOOK-

U P _____., LOCATING SHOULDERS

TO CROSSHEAD BASE _____*

Figure 4. measure cement-bone interfacial strength under pure tension.

Schematic representation of adhesion testing apparatus used to


WEIGHT - A PLUNGER - T=

I

. SAMPLE HOLDER

-SETSCREW

ACRYLIC -YOUNTING

CYLINOER

Figure 5. the preparation of a cement-bone joint for adhesion tests.

Schematic representation indicating the various steps involved in

screws. The sample holder is then inserted into the shell and held in place with set screws as well. The compressor is next inserted into the shell and tightened by clockwise twisting. This tightening causes the Teflon ring on the compressor end to contact the bone sample. Through the application of suitable pressure, the Teflon masks the bone surface such that only a circular region of lo2 mm is exposed to the acrylic cement to be tested. Typical mask effectiveness on cancellous bone substrates is shown in figure 13. The cement is now mixed and applied to an undercut hole in the postend. This undercut is necessary because bone-bone cement interfacial strength frequently exceeds metal-bone ceme.nt interfacial strength. The post is inserted into the compressor and, with the aid of the plunger, the unset cement contacts the bone sample. Next, a weight is placed on the plunger to insure contact between cement and bone substrate. Two different chromeplated steel weights were used, one was 100 g {ca. 1.7 cm diam), the other weighed 454 g (ca. 3 cm diam). The cement is then aliowed to set for 30 min. Prior to the adhesion test, the shell set screws are loosened and the fabricated test joint is removed from the preparation assembip (a light pressure is sometimes applied to the plunger to facilitate extraction). A prepared adhesive test joint is shown in Figure 5. Cement thicknesses in the various samples ranged 0.25-1 .O mm.


The prepared adhesive joints were tested in tension to failure using an In- stron Universal testing machine. . Special locating fixtures (see Fig. 5) were built to insure uniaxial’ application of tensile forces to the test joint. The prepared joint is placed into the base in contact with a circular locating shoulder. A cap is then screwed down over the sample holder, the body of which protrudes through a hole in the cap. The base is attached directly to the Instron crosshead and preloaded with coil springs to prevent error which can arise because of base lifting at the light loads these tests require. The post is attached to a long steel rod with a piano wire S-hook. (This hook is very light and produces negligible loading to the test joint prior to testing and helps to maintain uniaxial transmission of tensile forces to the joint.) The steel rod is attached to a load cell via a universal joint to further insure uniaxial transmission of tensile forces to the joint being tested. Loads as small as 0.1 lb are easily measured with the setup, and linear force versus elongation behavior occurs up to loads of 20 lb (beyond this point the S-hook tends to yield).

RESULTS AND DISCUSSION

Several limitations of values obtained for joint failure stresses must be rec- ognized. First of all, joint failure stresses were calculated dividing the ultimate sustained load by the macroscopic cross-sectional area of the test joint, that is, 10 mm2. Therefore, tabulated values represent a macroscopic failure stress and do not allow for differences between the true interfacial fracture area and the original cross-sectional area. Secondly, the extent to which tensile butt joint strength is sensitive to adhesive layer thickness depends on the specific values of modulus for the adhesive and substrate i n v o l ~ e d . ’ ~ J ~ In order to assess the importance of this effect, joint strengthsas a function of adhesive layer thickness were measured for bone cement between two aluminum substrates and compared to results obtained for similar measurements between human dentin substrates, in this case the adhesive used was a PMMA-based dental restorative. Cancellous bone cannot be used as a substrate of lower modulus to investigate the thickness effect, since variations in porosity do not allow accurate control of adhesive layer thickness. Results of thickness effect studies are shown in Figure 6. For bone cement on aluminum substrates, these measurements reveal a strong dependence of joint strength on adhesive thickness. This is expected on the basis of adhesive and substrate moduli differing by roughly a factor of 300. This dependence, however, is nearly absent for a PMMA adhesive on dentin, an adhesive-substrate combination which closely resembles (in terms of difference between moduli) the joint of interest in this work. The moduli of cancellous bone and bone cement differ by roughly a factor of 2. This fact precludes development of lateral contraction shear stresses, the effect responsible for thickness dependence of joint strength.14

Finally, a coefficient of variation was estimated for joint strengths obtained. This coefficient is approximately 20% for our testing apparatus, when deter- -mined from joint strengths of bone cement on aluminum substrates using a

44s KUSLEIKA AND STUPP

‘*I T

P ALUMINUM

SUBSTRATE

DENTIN SUBSTRATE

0.2 0.4 0.6 0.8 LO ADHESIVE LAYER THICKNESII, (1111

Figure 6. taining adhesives placed on aluminum and dentin surfaces.

Dependence of tensile joint strength on thickness of PMMA-con-

constant adhesive thickness. These fluctuations are expected because of the statistical nature of interfacial imperfections. An additional percent variation should be associated with topographical differences among calcified tissue substrates (on dentin substrates, for example, an additional 30% fluctuation is observed). Therefore, at least a 50% variation coefficient (by no means an atypical coefficient in adhesion tests) should be kept in mind for interpretation of our results.

We tested a total of 111 bone specimens, and Table I shows the number of

TABLE 1- Bone Porosity

Type of bone: Metatarsal Vertebral Surface orientation: Transverse Longitudinal Transverse

Grade of porosity: Fine Average Average Average

Percent porosity 57 5 64 f 10 70f 11 60 f 9 66 i 6 70 i 9 64 f 8 64 f 14 5 2 i 10

Pore size 310 f 62 500 f 175 46s f I17 466 f 142 307 i 52 416 f 216 530 f 130 529 f 139 291 f 66


100 200 300 PLACEMENT TIYE (SECONDS)

Figure 7. Bar graphs showing average failure tensile stresses (horizontal line) of cement-bone joints prepared with transverse (T) and longitudinal (L) cuts of fixed tissues (see Fig. 2). All cement samples were allowed to solidify under a pressure of 0.11 MPa. Shaded areas cover the range of average values fl SD. Strain rate = 1.7-8.3 min-'.

samples in each category. Two-tailed statistical tests were performed in order to determine if significant differences existed between mean values of interfacial strength for joints prepared under various conditions. Tensile strengths for joints constructed with formalin-fixed vertebral bone are shown in Figure 7. An 0.11 MPa (16 psi) curing pressure was used and the effects of placement time on joint strength for two different bone orientations are presented.

Statistical tests were carried out to determine if significant differences exist between the strengths of joints constructed with longitudinal and transverse sections. The tests compared groups of samples at equal curing pressure and placement time. No statistically significant differences were found in this series of tests, which yielded standardized variates ranging from 0.15 to 1.27. Cement placement time appears to have had no effect upon joint failure stress as well. Again, statistical tests were performed, comparing placement times at a given curing pressure and combining longitudinally and transversely cut specimens as one group. No statistical difference was found, wilth standardized variates ranging from 0.15 to -0.67. Using data for low-pressure joints and not combining longitudinal and transverse specimens into one group, statistically significant differences were still not found among the various placement times (in this case standardized variates range from 1.61 to -1.10). Figure 8 shows tensile joint strengths for longitudinally cut vertebral bone for two different curing pressures. In this case, the data suggest that increased curing pressure does produce stronger joints. A statistical test was carried out to compare joint strength in fixed bone at 0.11 MPa curing pressure vs. 0.47 Mfa (incorporating both types of cut and all three placement


7 -

6 . - D P

E 2 5 - Y c t m 4 . Y a 3 d 3. E

5 2 -

I .

I00 200 300 PCACEYENT TIYE (SECONDS)

Figure 8. Bar graphs showing average failure tensile stresses (horizontal line) of cement-bone joints prepared from fixed tissues in longitudinal cuts and cement placement on the substrate after 100, 200, and 300 s following powder-liquid mixing. For each placement time, data are shown for cement samples allowed to solidify under a pressure of 0.47 MPa (Hi P) and 0.11 Mpa (Lo P). The shaded areas cover the range of average values f l SD. Strain rate = 1.7-8.3 min-’.

times in one group). This test indicates that the difference in strength between high- and low-pressure joints in fixed bone is statistically significant at the 5% level (standardized variate = 1.98). Tensile strengths of joints prepared with fresh human metatarsals are presented in Figure 9. For these samples, two curing pressures were also used, 0.1 1 MPa and 0.47 MPa (16 psi and 68 psi), as well as the three different placement times used for fixed bone. Like fixed bone substrates, joint strengths are also independent of cement placement time. Statistical tests did not reveal any significant differences due to placement time at each pressure used (standardized variates ranged from 1.21 to -0.60). In contrast to fixed bone samples, fresh metatarsal joint strengths appear to be independent of curing pressure. Combining all placement times into one group, the standardized variate for low-pressure joints is 0.43. When placement times are not combined into one group, there are also no statistically significant differences between low- and high-pressure joints.

Common to all sets of data presented above is the observation that, keeping all other variables constant; tensile adhesive joint strength is independent of cement placement time. This observation suggests that the rheological properties of the powder-liquid mixture does not change enough over the first 300 s to prevent cement flow into the bone and lead to inadequate mechanical retention. Actual measurements of commercial bone cement vis~osity’~ have


300

PLACEMENT TIYE (SECONDS)

Figure 9. Bar graphs showing average failure tensile stresses (horizontal line) of cement-bone joints prepared from freshly extracted human metatarsals and cement placement on the substrate after 100,200, and 300 s following powder-liquid mixing. For each placement time, data are shown for cement samples allowed to solidify under a pressure of 0.47 MPa (Hi P) and 0.11 Mpa (Lo P). The shaded areas cover the range of average values fl SD. Strain rate = 1.7-8.3 min-*.

shown that viscosity changes by about a factor of 2 over the first 5 min after mixing. We suspect that flow can easily occur through the large bone pores (ca. 500 pm) because powder particles are small (ca.'60 pm) and highly plasti- cized by monomer. Furthermore, our tests maintain a continuous pressure head on the viscous cement. The penetration of cement into bone samples was experimentally verified by sectioning some of the specimens, and depths of penetration were approximately 2-3 mm (roughly half the thickness of bone samples). Therefore, it appears as though continuous light pressure (0.11 m a ) , is sufficient to cause cement flow and mechanical interlock of bone and cement even at the comparatively late placement time of 300 s. Clinically, late cement placement might be desirable in order to reduce patient exposure to MMA monomer.I6 This suggestion should be specially valuable if, as suggested by our data, interfacial strengths of bone cement to cancellous bone do not benefit significantly from very early cement placements. Bone cement-metallic prosthesis interfaces, however, do appear to weaken because of late cement ~1acements . l~ In this context, alternative methods (other than early place-


- 20x I mm 20x 7s VERTEBRAL BONE, FORMALIN FIXED VERTEBRAL BONE, FORMALIN FEED

LONGITUDINAL SECTION LONGITUDINAL SECTION

Figure 10. Optical micrographs of bone cement left on the metal post of the adhesive joint (see Fig. 5) after tensile failure of the cement-weak bone interface. During joint formation, the bone cement was allowed to solidify under a pressure of 0.1 1 MPa.

ments) to generate improved joint strengths at cement-metal interfaces should be clinically significant (e.g., PMMA precoating of metallic prostheses).

A comparison between Figures 8 and 9 reveals a marked drop in tensile joint strengths for vertebral formalin-fixed bone substrates relative to freshly extracted metatarsal bone. As indicated by data in Table I, this difference is not likely to originate in varying porosity levels for both types of substrate. The difference does suggest, however, the occurrence of interfacial bone fracture under pure tension due to the mechanically weaker nature of fixed as opposed to fresh bone. In fact, during sectioning procedures, fixed bone was found to be markedly weaker than fresh metatarsal bone. The lower values measured for fixed bone point out the possibility of low-strength joints for interfaces between cement and pathologically weakened calcified tissues. Higher pressure, however, does raise the joint strength for weak bone substrates. This suggests that increased pressure might be clinically desirable for interfacial reinforcement in cases where weak bone is present. Our results on fresh metatarsal joints, however, suggest that little or no advantage might be gained clinically with pressure-enhanced cement penetration into sound cancellous bone. Adequate cement penetration for fixation might be achieved with low-pressure techniques such as finger packing, which studied9 show can achieve pressures of 17-31 psi. In this co-ntext, one could cite several disad- vantages of deep cement penetration, for example, increased monomer exposure to tissue with increased interfacial area of bone-acrylic cement contact, greater probability of a higher temperature rise, and more extensive necrosis owing to increased cement layer thicknesses. The use of pressure gun-packing techniques, which produce pressure in the range of 38-59 psi, should be


c-.-.l O.5mm 3 0 X

VERTEBRAL BONE.FORYALIN FIXED VERTEBRAL BONE, FORMALIN FIXED LONGITUDINAL SECTION LONGITUDINAL SECTION

M 3OX 0.5fMlI

Figure 11. Optical micrographs of bone cement left on the metal post of the adhesive joint (see Fig. 5) after tensile failure of the cement-weak bone interface. During joint formation, the bone cement was allowed to solidify under a pressure of 0.47 MPa.

carefully considered for strong tissues since its interfacial strength advantage is not readily obvious. It is possible that under shearing forces tissueIgun- packed cement interfaces might be stronger. Nonetheless, areas of the cement bed under pure tension (the nature of our studies), being the weakest links of cement-tissue mechanical interaction, would probably fail first.

The following is a possible interpretation of our results on joint strength versus curing pressure in fixed and fresh bone substrates. Optical micrographs of failed interfaces on the aluminum post of the testing apparatus show cement tag pullout and bone fracture in low-pressure joints with fixed bone (Fig. lo), and generally a mixture of cement pullout and cement fracture (Fig. 11) in high-pressure joints. The micrographs for low-pressure cases reveal cement tags that have pulled out of cancellous bone porosities as well as fractured bone. The smoother failed surface in Figure 11 (high pressure) is representative of cement fracture. In the case of fresh metatarsal substrates, cohesive cement fracture was more common (60% of the samples) than either bone fracture or tag pullout. Joint failure by bone fracture alone only occurred in 29% of the samples. Tensile strength data on cancellous bone1* (wet, fresh lumbar vertebral body cancellous bone in the longitudinal direction) indicated values in the order of.3.42 MPa.

For acrylic cements we measured an approximate 30-min tensile strength value (adhesive test joints are stressed 30 min after fabrication) as a 1% tensile yield stress (Fig. 12). These values are of the order of 10.2 MPa and suggest that in areas of the cement bed exposed to pure tension, acrylic cement is stronger than cancellous bone. In this context, one may interpret joint strength results in the following way. The curing pressure causes cement to

4 54 KUSLEIKA AND STUPP

Q

P

5 1 I . 1 r *

113 1 2 4 24 72 120 3 Q Q N

CURING TIME , HOURS

Figure 12. the onset of powder-liquid mixing. Strain rate = 0.4 min-I.

Dependence of bone cement 19% yield stress on time elapsed from

flow into a roughly conical zone of cement penetration within bone porosities. Hence, a certain volume of bone is reinforced while simultaneously creating a surface area of unreinforced bone in contact with reinforced bone. In- creasing the curing pressure increases the area of unreinforced-reinforced bone contact. The reinforced bone should act as a composite unit with the cement layer. The interface between these two types of bone is probably the one that fails in low-pressure vertebral body tests. This is evidenced by pullout of cement plugs and infrequent cohesive cement failure. In high-pressure vertebral body tests, however, this interface is more extensive and stronger than that obtained when low pressures are used. In this case, the trend is slightly biased toward cohesive cement failure. Since the cement is probably stronger than fixed bone, we expect a large amount of bone reinforcement to be necessary before much cohesive cement failure occurs. Tests on metatarsal bone show little effect of curing pressure because the strength of fresh cancellous bone is greater than that of fixed vertebral bone, suggesting that metatarsal and acrylic cement strengths are close in value (this way the bone is strong enough to retain shallow cement). Under pure tension, cement penetrations fracture easily at the surface rather than at the reinforced-unreinforced bone interface. At the latter interface, cement failure is not necessarily resisted by tensile strength alone but also by its shear strength which has a much higher value. This tvpe of failure mode would explain why we do not observe a curing pressure dependence of interfacial strength when fresh metatarsal bone is used.

Using a sample of 38 joints constructed with fresh cancellous bone, failed


- t mm 20x

MElATARSAl.FRESH CANCELLOUS BONE TRANSVERSE SECTION, 0.47 MPa

Figure 13. tensile failure of the cement-tissue interface.

Optical micrograph of cement left on a fresh bone substrate after

interfaces were grouped according to fracture morphology. Three types of fractures were identified. Cement failures were the most common ones (observed in 23 out of the 38 joints) and consisted predominantly of cohesive cement failure at the level of the bone section surface (a typical case is shown in Fig. 13). The surface has been lightly polished in order to illustrate the character of cement penetration. Bone failures predominantly consisted of fracture of bone underlying the penetrating cement mass, and were observed in 11 joints. Mixed fractures (four joints) were simply a combination of these two fracture types. The average failure stress associated with each type of fracture has a characteristic value (Fig. 14). A statistical test on values obtained for the 38 specimens indicates that the higher joint strength observed for cement failures is significant at the 0.5% level (standardized variate = 2.90), and one may attribute nearly all the strength values of cement failure joints to cohesive cement fracture alone. In this type of failure mode, the cement penetrates about 65% of the surface area as shown by porosity measurements. Assuming that all cement penetrations fail cohesively, we only need to reduce the cohesive tensile cement strength by 35% (10.2-6.6 MPa) to find the cohesive cement failure contribution. An interfacial strength of about 6.6 MPa repre- sents nearly 85% of the average joint strength when cement cohesive failure is observed.

The discrepancy might arise from a very small contribution from secondary chemical bonding between cement and bone surfaces. This is unlikely, or else a minor effect, given the hydrophilic nature of bone surfaces. The discrepancy is more likely to arise from uncertainty in the actual 30-min tensile strength of the bone cement tested. Bone fractures comprise the lowest failure stress group encountered. Large variations in cancellous bone strength allow for the possibility that these fractures occurred in weak bone samples which


CEMENT MIXED BONE - FRACTURE FRACTURE FRACTURE

Figure 14. Summary of average values (horizontal line) for failure tensile stresses of cement / bone interfaces according to observed fracture morphology: (1) largely cement fracture; (2) mixed fracture; and (3) largely bone fracture. The shaded areas cover the range of average values fl SD.

did not have extensive cement penetration. Considering the cancellous bone tensile strength to be approximately 3.42 MPa,'* a 30% increase in cement in- terpenetrating area over cross-sectional area will account for the tensile joint stress of about 4.4 MPa in terms of bone failure alone. Adhesive contributions need not occur to achieve failure stresses of these magnitudes. Therefore, one may conclude that the strength of bone fracture type failures is indicative of the strength of underlying cancellous bone and that little or no adhesion occurs between tissue and cement.

An accurate assessment of whether or not presently available cements meet interfacial strength requirements in joint surgeries is beyond the scope of our work. This assessment would require a detailed consideration of the me- chanics of implant support. The data suggest, however, that some chemical adhesion and /or greater values of cement tensile strength would be beneficial modifications for the interfacial bond. Based on our results, cortical bone surfaces in contact with acrylic cements contribute negligibly or not at all to interfacial strength, In this context, some adhesion would be desirable given the fact that complete anchorage of cement in cancellous bone is not always possible, particularly at the upper end of the femur.

Our observations also identify tensile fracture of cement as a common event in macroscopic interfacial failure. For this reason, higher ultimate tensile strengths are suggested as a desirable improvement. One can also infer from the macroscopic observations of the study that the tensile faiiures of cement could occur rn v i m within microscopic zones of the interface because of localized stress concentrations. In fact, previous research on retrieved'inter- facesz0 has suggested the in v i m occurrence of cement fracture. The evidence is based on the frequent presence of cement inclusions within underlying


tissues (single beads or bead clusters containing matrix material). Generally speaking, any localized fractures and their resulting debris are not only un- desirable from a biological standpoint but also as nucleation sites for implant loosening. Perhaps a more sensible and realistic approach toward improvement of interfacial load-bearing capacity in PMMA formulations is to seek toughness enhancement.21 A tougher cement under stress or upon impact would absorb more energy prior to fracture at interfacial regions. This could be accomplished through slightly higher moduli or yielding at stress levels where current cements fracture.

CONCLUSIONS

The mechanical strengths of interfaces formed by acrylic bone cement and cancellous human bone have been characterized for joints stressed under the critical condition of pure tension. The average interfacial strength was found to vary depending on the origin of the tissue substrate, 3.6 MPa for formalin- fixed bone and 5.6 MPa for freshly extracted bone. The time elapsed between powder-liquid contact and cement placement on tissue surfaces did not appear to affect mechanical strength. Pressure applied during cement setting was only found to reinforce interfaces formed with fixed tissues, which were mechanically weaker than freshly extracted bone. This effect is viewed as tissue reinforcement by cement penetration and suggests a possible enhancement of interfacial strength with increasing placement pressure in pathologically weakened tissue. Based on our measurements for freshly extracted bone substrates, the advantage of enhanced pressure is not readily obvious. In this context, i t is suggested that high-pressure surgical injection of cement should be carefully evaluated given its possible deleterious side effects, e.g., greater degrees of thermally induced necrosis. The morphology of the strongest failed interfaces revealed largely cohesive failure of cement, whereas bone cohesive failure was the most prominent feature of the weakest tissue-cement joints. In cases where cement failure occurs, the values of joint strength can be ac- counted for by cement tensile strength, with little or no contribution from adhesive forces. Considering the random occurrence of tissue and cement failure, as well as the apparent tissue reinforcing ability of cement penetrations, higher tensile strength and/or toughness in the PMMA-based cements should be valuable improvements for these implant materials.

The authors are grateful to the Department of Biological Materials, Northwestern Uni- vemity, for providing its laboratory facilities as well as support for R. Kusleika through NIH training grant DE 07042. The authors are also indebted to personnel in the Department of Orthopaedic Surgery, Northwestern University Medical School, for their cooperation in obtaining tissue samples for the experimental work. Finally, partial support for the study from NIH grant DE 05152-02 is greatly appreciated.

References

1. C. Hirsch, “Clinical Problems in Total Hip Replacements,” J. Biomed. Mater. Res. Symp., 5 (part 2), 227-244 (1974).

2. J. A. Dipiso, G. S. Sih, and A. T. Berman, “The Temperature Problem at


3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

the Bone-Acrylic Cement Interface of the Total Hip Replacement," Clin.

R. E. McLaughlin, C. A. DiFazio, M. Hakala, B. Abbott, J. F. MacPhail, W. P. Mack, and D. E. Sweet, ''Blood Clearance and Acute Pulmonary Toxicity of Methyl Methacrylate in Dogs After Simulated Orthoplasty and Intra- venous Injection," I. Bone It. Sirrg., 55A, 1621-1628 (1973). E. P. Lautenschlager, B. K. Moore, and C. M. Schoenfeld, "Physical Char- actenstics of Setting of Acrylic Bone Cements," ]. Btomed. Mater. Res. Symp.,

R. W. Treharne and N. Brown, "Factors Influencing the Creep Behavior of Poly(methy1 Methacrylate) Cements,'' I . Biomed. Mater. Res. Symp., 6,

J. 0. Galante, W. Rostoker, and J. M. Doyle, "Failed Femoral Stems in Total Hip Protheses," /. Bone It. Surg., 57A, 230-236 (1975). R. D. Beckenbaugh and D. M. Ilstrap, "Total Hip Arthroplasty," 1. Bone

F. A. Weber and J. Charnley, "A Radiological Study of Fractures of Acrylic Cement in Relation to the Stem of a Femoral Head Prosthesis," 1. Bone I t . Surg., 57B, 297-301 (1975). I. S. Leinbach and F. A. Barlow, "700 Total Hip Replacements," Clin. Orthop.,

P. S. Walker, "Tribial Fixation in Condylar Replacement Knee Prostheses," in Biocompatibility of Implant Materials, David Williams, Ed., Sector Pub- lishing, London, 1976, p. 93. R. Kolbel, E. Bergmann and U. Boenick, "Advances in Artificial Hip and Knee Joint Technology," in Engineering in Medicine 2, M. Schaldach and D. Hohman, Eds., Springer-Verlag, New York, 1976, p. 347. R. W. Bryant, "Strength of Adhesive Joints," Nature, 202, 1087-1088 ( 1964). Meissner and Baldauf, "Strength Behaviour of Adhesive Bonds," Trans.

J. J. Bikerman, "The Fundamentals of Tackiness and Adhesion," 1. Colloid

W. R. Krause, J. E. Miller, and P. Ng, "Rheology of Acrylic Bone Cements," Trans. 21th Inf . Biomaf. Symp., 3,66 (1979). S. Bayne, E. P. Lautenschlager, E. H. Greener, and P. Meyer, "Clinical In- fluences on Bone Cement Monomer Release," J. Biomed. Mater. Res., 11,

J. C. Keller, E. P. Lautenschlager, G. W. Marshall, and P. R. Meyer, "Factors Affecting Surgical Alloy-Bone Cement Interface Adhesion," Trans. I Ith Biomatrr. Symp., 3,92 (1979). T. Sonoda, "Studies on the Strength for Compression, Tension and Torsion

O~thop., 121,95-98 (1976).

5,185-196 (1974).

81-88 (1975).

It. S U T ~ . , 60A, 306-313 (1978).

95,174-192 (1973).

ASME, 73,697-704 (1951).

S C ~ . , 2,163-175 (1947).

859-869 (1977).

of the Human Vertebral Colu&-t," 1. Kyoto Prej. Med. Univ., 71(9), 659 (1962). I. Oh,'R. B. Bourne, and W. H. H a m s ; "Closed Pressurization of Medullarv Cement in Total Hip Replacement," Trans. 12th Znf. Biomater. Symp., 3,67 (1979). H. G. Willert, K. Mueller, and M. Semlitsch, 'The Morphology of Poly- (methyl Methacrylate) (PMMA) Bone Cement Surface Structures and Causes of Their Origin," Arch. Orthop. Traumat. Surs., 94,265 (1979). S. I. Stupp, "Characterization of Bone Cements and R e s i p of New For- mulations by Thermally Stimulated Discharge Analysis," Book 01 Abstracts, Frrsf World Biomatertals Congress. Baden, Austrla, April 1980

Received February 1982 Accepted August 6,1982

mechanical strength of poly(methyl methacrylate) cement-human bone interfaces

Documents