There have been various investigations into the utilization of cement for total hip and knee arthroplasties1-7. A number of factors can affect the cement-bone interface, including lavage and hemostasis as well as cement-mixing, viscosity, pressurization, and penetration depth. The strongest cancellous bone is found close to the corticocancellous bone junction, within 3 mm of the cortex8. A deeper penetration of 5 mm weakens the failure load and increases the risk of thermal damage to tissue. In four patients undergoing resurfacing arthroplasty, Gill et al. recorded a maximum temperature at the cement-bone interface of approximately 68°C, which was higher than the temperature at which osteocytes can be killed9. An ideal cement penetration depth would at least engage one level of transverse trabeculae with sufficient filling of the vertical channels and avoid large cement masses or thick cement layers that could cause thermal necrosis of bone. A 2 to 5-mm depth of cement penetration into bone is generally accepted as optimal10,11.
The available surface replacement components have different inner geometries; some components are line-to-line with the bone surface, while others have a cement mantle. Additionally, there are differences in the cementing technique recommended by the manufacturers. One technique recommends filling the femoral component with low-viscosity cement and then immediately applying it to the prepared femoral head12. Another technique recommends pressurizing higher-viscosity cement directly to the prepared bone, with component insertion when the cement is in an early dough stage (high viscosity)13.
In two retrieval studies, improper cementing technique and insertion trauma have been implicated in early femoral-side resurfacing failures14,15. Cement penetration was actually greater in loosened components15. There have been limited investigations into cementing techniques for hip resurfacing. A finite element model predicted that, depending on the mass of cement under the femoral shell, the curing temperature could exceed the threshold for bone necrosis15. The criteria for a satisfactory laboratory model include bone specimens that are exactly the same (a controlled variable). There is obvious variability in natural femoral head specimens, which might preclude their use in such a study. For studies of cement technique of femoral-side resurfacing, a surrogate material needs to be identified that has measurably similar properties to femoral-head bone16.
The goal of this study was to develop an in vitro model of femoral cementing with use of a surrogate material and to compare the cement penetration, pressure, and temperature to those of fresh-frozen femoral-head specimens with use of two different cementing techniques.
High-density open-cell reticulated carbon foam with sixty pores per inch (PPI) (2.4 pores per millimeter, 6% density) reticulated vitreous carbon (RVC) foam (ERG Materials and Aerospace, Oakland, California) and low-density 30-PPI foam material (1.2 pores per millimeter, 6% density, RVC foam; ERG Materials and Aerospace) were custom made for this study and compressed to emulate human femoral heads prepared for resurfacing. These materials have a density, porosity, and pore interconnectivity (three-dimensional structure) that emulate the trabecular bone of the human femoral head (Fig. 1). Fourteen paired fresh-frozen human femoral heads were used for comparison. The mean age of the six male and two female bone donors was seventy-eight years (range, sixty-eight to eighty-two years). The oldest female donor was excluded because of low bone density. The foam specimens were manufactured with the same geometry as the femoral heads prepared for resurfacing with use of a 49-mm ASR component (DePuy, Leeds, England).
The high-density foam material was investigated without use of any interporous material. To emulate bone marrow, commercially available vegetable fat (Crisco; J.M. Smucker, Orrville, Ohio) was used in the low-density foam. The preparation involved melting vegetable fat and pouring it into foam specimens wrapped in aluminum foil. The fat-filled foam specimens were cooled down to body temperature and unwrapped. All specimens (foam and bone) were prepared with high-pressure pulsatile saline solution lavage and dried with gauze sponges, as is commonly done in surgery prior to cementing.
Custom aluminum shells were manufactured with the same inner geometry as femoral resurfacing components (49-mm ASR; DePuy). This component has a 3° taper of the side walls and essentially fits line-to-line with the prepared bone surface when fully seated. Threads to fix pressure probes (XPM5/XAM; FGP Sensors and Instrumentation, Les Clayes sous Bois, France) were tapped at three different locations in the shells. One pressure probe was fixed at the top, one at the chamfer, and one at the outer wall of the aluminum shells. Polymerization temperature was measured at 5 and 15 mm beneath the foam surface with use of two catheter-shaped temperature probes (D-F1345A; EXACON Scientific, Roskilde, Denmark) (Fig. 2). Pressure and temperature data were recorded in real time with use of data-logging software (VirtualBench version 2.1; National Instruments, Austin, Texas) through a PCMCIA data acquisition card (DAQCard-1200; National Instruments) for a total of twenty minutes after the start of mixing the cement. We determined average pressures by computing the area of the pressure compared with time curves by numerical integration and then dividing that area by the time from seating of the aluminum shells to the temperature rise.
Two cementing techniques were performed with use of fourteen specimens in each of the three material groups. A manual application technique with high viscosity cement (HVC) (SmartSet GHV; DePuy, Leeds) or a half component-filling technique with low-viscosity cement (LVC) (SmartSet MV Endurance; DePuy, Blackpool, England) was performed on seven specimens of each foam material and on one side of each of the paired femoral-head specimens. For the manual application technique, the cement was applied to the foam or bone specimens with use of a syringe at 120 seconds after the start of mixing. The cement was then digitally pressurized into the pores of the specimen with use of a rolling movement of the fingertip over the entire surface of the foam model or bone while keeping the central pin hole clear. For the half component-filling technique, the cement was poured into the aluminum shells at 120 seconds after the start of mixing to the half-full point.
The mean room temperature was 21.6° ± 1.3° C with a humidity of 72.3% ± 0.8%. The aluminum shells were pressed onto the foam specimens with a 150-N weight 240 seconds after the start of mixing. The foam specimens were held in a stabilizing fixture that was equipped with ports for the temperature probes.
After twenty minutes of polymerization, the specimen was removed from the aluminum stand and cut into quarters with use of a band saw fitted with a fixture to ensure reproducibility. Digital pictures were made of each cut face with use of a lighted stand and markers for a consistent focus and picture size. The digital pictures were stored as color JPEG files. To quantify the distribution of cement in these pictures, the image data were digitally processed with use of custom-written MATLAB routines (MATLAB R2006a; The MathWorks, Natick, Massachusetts). Images were transformed into gray values. Two-dimensional, edge-preserving median filtering was applied to remove fine-scale inhomogeneities. Next, fuzzy k-means clustering was used for classifying gray values into two classes (cement versus other materials). This step resulted in binary images where "on" pixels indicated cement and "off" pixels corresponded to other materials. Finally, the total area of "on" pixels was calculated and scaled to square millimeters. Each step was stored in separate JPEG files to allow for manually checking the results of each particular processing step (Fig. 3).
Cement penetration was analyzed by defining a 3-mm wide fixation area of the interface and a deeper area of more than 3 mm beneath the foam or bone surface. Cement mantle thickness (cement layer between foam surface and aluminum shell) and cement penetration depth (maximal cement penetration into foam surface at the top area) were measured at the pole with use of a digital caliper.
Statistical Methods
Data were presented descriptively as means ± standard deviation of the evaluated values. Statistical analysis was performed in two steps. First we tested the distribution of data. This was done with use of the Shapiro-Wilk test for standard deviation and the Levene test for homogeneity of variances. In a second step, the effect of the three materials was examined in each cement technique group with use of the Kruskal-Wallis test. The effect of these two cementing techniques was examined with use of the Mann-Whitney U test in each material group. All tests were two-sided, and a p value of 0.05 was considered significant.
We found no significant differences between the human femoral heads and the fat-filled 30-PPI foam models in all measured variables (pressures at the top, chamfer, and outer wall and subsurface temperature at 5 mm and 15 mm). There was no significant difference in the depth of cement penetration into the top, chamfer, or outer wall of the human femoral heads and the fat-filled 30-PPI foam models. There were a number of significant differences between the human femoral heads and the 60-PPI foam models with the half component-filling technique (Figs. 4, 5, and 6 and Table I).
Average cement pressures at the top (Ptop) and at the chamfer (Pcha) for the manual application technique in bone were 30.2 kPa (range, 9.5 to 45.7 kPa; standard deviation = 18.7 kPa) and 32.8 kPa (range, 12.0 to 46.0 kPa; standard deviation = 18.3 kPa) (Ptop manual application technique compared with half component-filling technique, p = 0.02) (Pcha manual application technique compared with the half component-filling technique, p = 0.02). Average pressures with use of the 30-PPI fat-filled foam material were 28.6 kPa (range, 17.4 to 36.5 kPa; standard deviation = 8.8 kPa) and 25.6 kPa (range, 22.4 to 32.5 kPa; standard deviation = 4.1 kPa) (Ptop manual application technique compared with half component-filling technique, p = 0.01) (Pcha manual application technique compared with half component-filling technique, p = 0.01) and, for the 60-PPI foam, 30.5 kPa (range, 18.7 to 50.8 kPa; standard deviation = 17.7 kPa) and 28.9 kPa (range, 16.7 to 48.5 kPa; standard deviation = 17.1 kPa) (Ptop manual application technique compared with half component-filling technique, p = 0.89) (Pcha manual application technique compared with half component-filling technique, p = 1.0).
The average cement pressures at the top and at the chamfer for the half component-filling technique in bone were 96.2 kPa (range, 87.4 to 105.0 kPa; standard deviation = 12.4 kPa) and 99.0 kPa (range, 89.3 to 108.7 kPa; standard deviation = 13.7 kPa). The average pressures with use of the 30-PPI fat-filled foam material were 96.9 kPa (range, 88.7 to 104.7 kPa; standard deviation = 8.0 kPa) and 89.9 kPa (range, 83.7 to 94.4 kPa; standard deviation = 5.6 kPa) and, for the 60-PPI foam, 36.0 kPa (range, 28.1 to 47.7 kPa; standard deviation = 8.6 kPa) and 31.0 kPa (range, 23.8 to 39.8 kPa; standard deviation = 6.7 kPa) (Ptop bone compared with 60 PPI, p = 0.03) (Pcha bone compared with 60 PPI, p = 0.03).
There was no significant difference for the average cement pressure at the outer wall between the two cementing techniques (bone [p = 0.56]; 30 PPI [p = 0.29]; 60 PPI [p = 0.73]) or the materials (such as bone compared with 30 PPI [p = 0.19] for the half component-filling technique).
The maximal temperature 5 mm under the foam surface reached a similarly higher level for the half component-filling technique in bone and in the 30-PPI fat-filled foam. Temperature with the manual application technique was also similar in bone and in the 30-PPI fat-filled foam. The 60-PPI foam had higher temperatures with both cement techniques.
There was no significant difference in the maximal temperature 15 mm beneath the foam surface for the cementing techniques (bone [p = 0.90]; 30 PPI [p = 0.22]) and materials except for the half component-filling technique with the 60-PPI foam (manual application compared with half component-filling technique [p = 0.02], bone compared with 60 PPI [p = 0.02] for the half component-filling technique).
For both cement techniques, cement penetration depths (manual application: bone versus 30 PPI, p = 0.22; half component-filling technique: bone compared with 30 PPI, p = 0.67), cement content in the 3-mm wide fixation area (manual application: bone compared with 30 PPI, p = 0.55; half component-filling technique: bone compared with 30 PPI, p = 0.42) and cement penetration in deeper areas (manual application: bone compared with 30 PPI, p = 0.55; half component-filling technique: bone compared with 30 PPI, p = 0.29) showed no significant differences between bone and the 30-PPI fat-filled foam specimens (Table I). With the half component-filling technique, 60-PPI foam showed significantly higher cement penetration depths (half component-filling technique: bone compared with 60 PPI, p = 0.05) and cement penetration in deeper areas (half component-filling technique: bone compared with 60 PPI, p = 0.01) than bone.
There was no significant difference in the amount of incomplete seating of the components (polar cement cap) with the manual application technique between bone (1.5 mm [range, 0.6 to 1.7 mm; standard deviation = 0.3 mm]), 30-PPI fat-filled foam (0.9 mm [range, 0.0 to 2.3 mm; standard deviation = 1.2 mm]) and 60-PPI foam (1.8 mm [range, 1.5 to 2.1 mm; standard deviation = 0.3 mm]) (manual application: bone compared with 30 PPI, p = 0.06; manual application: bone compared with 60 PPI, p = 0.06). The risk of incomplete seating of the components with use of the half component-filling technique was similarly higher for bone and the 30-PPI fat-filled foam specimens than it was for the 60-PPI foam (half component-filling technique: bone compared with 60 PPI, p = 0.03).
This study demonstrates similarity in the measured parameters (pressure, temperature, and depth of penetration) between fresh-frozen human femoral heads and the 30-PPI fat-filled carbon foam model. Fat-filling of the lower density material (to emulate bone marrow) more closely simulates the resistance to cement penetration and thermal characteristics of human bone than does the denser (but unfilled) carbon foam material. The differences between the cementing techniques were greater than those between the three materials for most of the measurements, indicating a high sensitivity of the foam model for the comparison of cementing techniques.
The maximal temperature 5 mm beneath the bone surface of 54.5° ± 14.2°C for the half component-filling technique is associated with a risk of thermal damage of interfacial bone. With use of tibial metaphyses of rabbits, Eriksson and Albrektsson showed that heating dividable titanium implants to 47° or 50°C for one minute caused significantly reduced bone formation in the implants, while no significant effects were observed after heating to 44°C for one minute. The results reflect the importance of controlling the heat produced during surgery or cementing to avoid impaired bone regeneration17.
Some aspects of our model differ from conditions found during surgery. Bleeding of the femoral head was not emulated. However, most contemporary resurfacings utilize proximal femoral suction to temporarily minimize the circulation during cementation. Our custom test components were made of aluminum and had a higher wall thickness than actual resurfacing components do. This may have caused a slight reduction of the temperatures in vitro compared with those in vivo, but our temperature data were still quite similar to those in other reports9. Human femoral heads have variable bone density, not only from specimen to specimen, but also within the same femoral head. The foam standard makes relative comparisons between different cementing techniques possible, even if the absolute values might differ from some clinical conditions. We were limited by the number of human femoral heads available to us. With a larger number of specimens, it may have been possible to show significant differences for some of the measurements between the femoral heads and the 30-PPI fat-filled reticulated carbon foam model.
In summary, the 30-PPI fat-filled reticulated carbon foam model closely simulates the human femoral head when prepared for resurfacing. Emulation of bone marrow by fat-filling gave more similar penetration resistance and thermal properties. The differences between the cementing techniques were greater than those between the three materials for most of the measurements. 