The specific aim of this study was to examine the relative contributions to the implant insertion torque value (ITV) by cortical and trabecular components of an in vitro bone model. Simulated bone blocks of polyurethane were used with 2 densities of foam (0.08 g/cm3 to mimic trabecular bone and 0.64 g/cm3 to mimic cortical bone). We have developed a new platform technology to collect data that enables quantitative evaluation of ITV at different implant locations. Seven groups were used to model varying thicknesses of cortical bone over a lower-quality trabecular bone that have clinical significance: a solid 0.08 g/cm3 block; 1 mm, 2 mm, and 3 mm thick 0.64 g/cm3 sheets with no underlayer; and 1 mm, 2 mm, and 3 mm thick 0.64 g/cm3 sheets laminated on top of a 4 cm thick 0.08 g/cm3 block. The ITVs were recorded as a function of insertion displacement distance. Relative contributions of ITV ranged from 3% to 18% from trabecular bone, and 62% to 74% from cortical bone depending on the thickness of the cortical layer. Inserting an implant into 2-mm and 3-mm cortical layers laminated atop trabecular blocks had a synergistic effect on ITVs. Finally, an implant with a reverse bevel design near the abutment showed final average torque values that were 14% to 34% less than their maximum torque values. This work provides basic quantitative information for clinicians to understand the influence of composite layers of bone in relation to mechanical torque resistances during implant insertion in order to obtain desired primary implant stability.
Long-term success of dental implants depends on quality of the implant-bone interface.1–3 Primary implant stability is a prerequisite for implant survival, as it prevents micromovement and subsequent formation of connective tissue layers between implant and bone, thus ensuring bone healing and osseointegration. Biomechanically, 3 factors determine primary implant stability: quality of bone, drilling condition, and implant screw design.
The outer wall of human jawbones consists of a thin cortical layer in the occlusal direction, ranging from 0.5 mm to 2.5 mm.4–6 Total implant engagement of the cortical bone in the occlusal aspect is often <10% of the implant length. Based on this small contact area, it is often assumed that the cortical layer contributes little to the strength of the bone-implant interface. However, there is no qualitative or quantitative information regarding the contribution of overall mechanical strength from trabecular and cortical components of human jawbones. Mechanically, cortical and trabecular layers are significantly different materials. These differences critically impact clinically relevant parameters like resistance to implant insertion torque and final placement torque of implants.
In validation experiments of computer models and in vitro experiments for bone biomechanics and implant testing, human cadaver bone is the material of choice. However, challenges exist in using cadaveric bone to conduct biomechanical experiments, including availability, storage requirements, high cost, possibility of infection, and large variability in sample mechanics. These challenges make synthetic bone analogs an attractive alternative. The uniformity and consistency of rigid polyurethane is ideal for comparative testing of various medical devices and implants. An example for clinically relevant testing can be found in studies of osteoporotic bone. In searching for an in vitro osteoporotic model, mechanical properties of osteoporotic cadaver bone and synthetic bone (Sawbones, Pacific Research Laboratories Inc, Vashon, Wash) made of polyurethane foam with densities of 0.08 g/cm3 and 0.16 g/cm3, respectively, were compared. The mechanical fracture properties of the Sawbones and cadaver bone were found to be similar.6,7
The specific aim of this work was to study the relative contributions of implant insertion torque values from cortical and trabecular components of an osteoporotic bone model in vitro as a function of implant displacement during insertion. Our hypothesis was that insertion torque values are significantly different from the cortical and trabecular components when inserting dental implants into a jawbone. We also tested and analyzed how variables of implant diameter and thickness of simulated cortical layer affected insertion torque values.
Materials and Methods
Simulated bone block construction
Simulated bone blocks of polyurethane rigid foams (Sawbones) were used: 0.08 g/cm3 to mimic trabecular bone and 0.64 g/cm3 to mimic cortical bone. Both materials meet ASTM F-1839-08 specifications,8 and their nominal properties are presented in Table 1. We constructed 7 models of varying thickness of cortical bone (0.64 g/cm3) over a lower-quality trabecular bone (0.08 g/cm3): a solid 0.08 g/cm3 block ∼4 cm thick; 1 mm, 2 mm, and 3 mm thick 0.64 g/cm3 sheets with no underlayer; and 1 mm, 2 mm, and 3 mm thick 0.64 g/cm3 sheets laminated on top of a 4 cm thick 0.08 g/cm3 block. To ensure that the thin sheets with no underlayer did not bend during drilling and implant insertion, a low-density block was fabricated with ∼1 cm diameter holes drilled in it. We laminated the thin sheets over this block and then did our drilling and implant insertion over the holes.
Two types of self-tapping titanium implants were used (NobelActive, Nobel Biocare AB, Goteborg, Sweden) for insertion torque tests: narrow platform (NP) 3.5 mm in diameter and regular platform (RP) 4.3 mm in diameter. Both implants are sold as 13 mm-long devices. Actual measured lengths showed the implants to be 12.5 mm long.
Torque vs displacement setup
Samples were tested in a modified commercial torque measurement unit (Figure 1) (MARK-10 Corporation, Copiague, NY). A flat reflective surface (Figure 1d) was attached to a part of the system that translated axially during implant insertion, and was used with a laser displacement sensor (model RS422, Micro-epsilon Inc, Raleigh, NC) to determine axial displacement as a function of time (Figure 1f). An axial compensator provided a uniform dead load, 256 g ± 11 g (N = 8), to drive insertion of implants (Figure 1e). A torque sensor (Model M7-500, MARK-10 Corporation) was used to record torque as a function of time.
Manufacturer-recommended protocols for surgical placement of implants were followed. The Sawbones model material provided good mechanical reproducibility for comparison purposes among implants. Drill sequences across all 7 models were standardized with final drills of 2.8/3.2 mm diameter for NP implants and 3.2/3.6 mm for RP implants. These are manufacturer recommended final drill sizes for placement into dense cortical bone. Since the 3-mm overlayers required the dense bone protocol, we decided to use the same drill sizes for all the other samples to eliminate drill diameter as a variable in this study. Implants were aligned with a predrilled hole; the sample was rotated at 1 Hz; recording of torque and axial displacement at 1-millisecond intervals was started; and the axial actuator handle (Figure 1h) was used to move the implant down until the compensator dead load was applied. Insertion was stopped when the top of the implant was flush with the surface of the test piece. Each combination of Sawbones block, drill size, and implant type was tested multiple times (5 ≤ N ≤ 30)
Analysis of data
Torque vs axial displacement curves were plotted and used to obtain means and standard deviations of maximum torque, final torque, and dropoff between maximum and final torque. All values are reported as means ± standard deviations. The difference between groups in Table 2 was determined using the Wilcoxon rank-sum (2-sample) test. Because the maximum torque is guaranteed to be greater than or equal to the final torque, statistical significance of the dropoff between the torque at an axial displacement of 10.5 mm and the final torque (Table 3) was calculated using both the Wilcoxon signed-rank test for paired data and studentized bootstrap resampling.9,10 The P values reported in Table 3 were computed by the bootstrap analysis. The data summarized in Table 4 required a test statistic that included 3 means and 3 variances, and so was not amenable to a nonparametric test. For this analysis, we used a t-like statistic with
along with studentized bootstrap resampling9,10 to perform a 2-tailed test that disproved the hypothesis that there was no synergy, and to estimated the amount of synergy. Here, , , and represent the sample mean of the samples with under/over layer, just the under layer, and just the over layer, respectively, while , , , , and are the respective sample variances and sample sizes. Finally, the family-wise error rate was controlled for using the conservative Bonferroni correction. Statistical significance was assigned with α = 0.05/32 . Here, 32 is the total number of comparisons made throughout the study: 12 in Table 2, 14 in Table 3, and 6 in Table 4.
All torque vs displacement plots shown below are averages of multiple curves. Figure 2 shows a typical situation, Figure 2a shows 18 individual curves, and Figure 2b shows a single curve representing the average of these 18 curves with error bars indicating standard deviations of the individual curves about the mean. Table 2 summarizes all the conditions tested giving means and standard deviations of final insertion torques for each group. The following sections present the three main findings of this study.
Regular platform implants can have a final insertion torque less than the maximum torque
Figure 3 shows torque vs displacement curves for implants inserted into a 1-mm simulated cortical bone layer laminated atop a thick simulated trabecular block. The regular platform implant peak occurs near 10.5-mm axial displacement and is ∼20 N-cm higher than the torque at final placement near 12.5 mm. This falloff is absent with narrow platform implants, which show monotonic increase to maximum torque (Figure 3, lower curve). In addition, this falloff is absent with regular platform implants inserted into homogenous samples (see, for example, the monotonically increasing curve, Figure 3 inset). Table 3 shows mean falloffs for all tested groups. The RP and NP implants were inserted into overlayers of 1 mm to 3 mm thickness. These overlayers either had a low-density block beneath them (Y) or an air-fill space beneath them (N). Implants placed into low-density blocks with no overlayer show a zero in the overlayer thickness column. Mean falloffs, computed as the difference between the maximum insertion torque and the final torque, are reported in both N-cm and percentage of maximum torque where ± corresponds to standard deviation. This is an accurate reflection of what the data looks like for curves, such as those shown in Figure 2. However, statistical significance is not robustly calculable using this measurement because the value is guaranteed to be nonnegative. For this reason, to compute P values, we used the insertion torque at 10.5 mm. This is the displacement at which the retrograde slope of the RP implants just began to insert into the samples (see, for example, Figure 4). Statistically significant falloffs (P < .05/32, where 32 is the total number of tests in this article) have a dagger in the P column. Implants placed into low-density blocks with no overlayer (rows 1 and 8 in Table 3) showed no falloff. These curves simply rose until the final value, similar to all the NP implants. Thus, these rows show a statistically significant difference between 10.5 mm and the final torques at 12.5 mm. However, this difference was not a falloff, it was a rise to the final torque. There was one outlier with respect to statistical significance: the RP implant with a 3-mm overlayer atop a low-density underlayer (row 4 in Table 3). This condition did not show a statistically significant falloff at the Bonferroni corrected alpha value of 0.0016 using the bootstrap analysis. This particular case had a small sample size of 7. In addition, the Wilcoxen signed rank test did show this case to be significant with a P value of 6.3 × 10−5. The average falloff for RP implants was 34% ± 14%. The average falloff for the NP implants was 6% ± 6%.
Thickness of cortical overlayer between 1 and 3 mm affects insertion torque
Figure 5 shows torque vs axial displacement curves for a regular platform implant placed into layered constructs having 1–3 mm dense overlayers laminated on top of a low-density block. The general curve shape is similar to that shown in Figure 3. After 3 mm of axial insertion, torque as a function of axial insertion distance clearly increases with overlayer thickness. The drop in torque from its maximum value to the final position of the implant is ∼19 N-cm and does not change much as a function of overlayer thickness. Table 2 is organized in order of increasing overlayer thickness for each implant type (NP and RP), and each condition of having a low-density underlayer or not (Y and N). For each group, the mean final torque values clearly increase with increasing overlayer thickness. Statistical significance of this was determined using a Wilcoxon rank-sum test. Similar results were also obtained using a Welch's t-test (data not shown). All but one of the rows (the NP 2-mm overlayer with underlayer present) showed a statistically significantly higher mean final torque compared with the row above. This one condition likely failed to show statistical significance because both the 1-mm and 2-mm measures in this group had a small number of measurements.
The synergistic increase in final torque is due to the low-density underlayer
Figure 6 shows torque vs displacement curves for a regular platform implant inserted into 3 different constructs. Insertion into a pure low-density block #5 shows a final insertion torque of 3.8 ± 0.9 N-cm (bottom curve). Insertion into a 2-mm-thick overlayer supported over an air-filled space shows a final insertion torque of 34.7 ± 5.2 N-cm (middle curve). Insertion into a 2-mm-thick overlayer laminated on top of a low-density block shows a final insertion torque of 48.6 ± 3.9 N-cm (top curve). The sum of final torques into the 2 elements taken alone was 38.5 N-cm (indicated by the top of the vertical arrow shown near the end of the middle curve in Figure 6) or 10.1 N-cm less than the final torque into the laminated construct.
Table 4 summarizes relative contributions of final insertion torque values from simulated trabecular bone underlayers (U) and 3 thicknesses of cortical bone overlayers (O). The numbers are reported in N-cm. In every case, the final torque of the laminated construct is greater than the sum of the individual components of the laminate measured separately. In addition, scanning across each row, the magnitude of this synergistic effect due to layering increases with overlayer thickness. A bootstrap resampling approach9,10 showed that synergistic increase in final torque exists for both NP and RP implants placed into constructs having 2 mm or 3 mm thick overlayers. In these cases, the P values for a 2-tailed test (Table 4) were much smaller than the Bonferroni corrected alpha level. Constructs with 1-mm overlayers did not show a similar strong effect and had P values of .09.
The measurement system developed for this study allows for quantitative evaluation of primary implant stability for different implant designs. Most published studies on insertion torque of dental implants are based on final insertion torque generated by a manual torque wrench.11–22 The manual torque method gives clinicians only 1 data point during implant insertion. Other studies use torque vs time data to study primary implant stability. While providing more information than the single-point final insertion torque method, using time as the independent variable is problematic. The rate at which implants progress into samples varies for multiple reasons, including the amount of manual compressive loading forces applied by clinicians, complexity of implant thread design, uncontrollable fracture of bone and bonelike synthetic materials leading to rapid advancement of the implant, and inhomogeneity in the sample material causing occasional slippage at the implant-sample interface resulting in rotation of the implant without subsequent axial advancement. In contrast, plots of insertion torque vs axial displacement, as presented in this study, allow for characterization and quantification of mechanical behavior at any location of an implant during insertion. This allows for a kind of “mechanical spectroscopy” where the shape of the torque vs displacement curve correlates with the physical situation during the insertion process.
A rudimentary example of this mechanical spectroscopy is seen in data showing RP implants having final insertion torques less than the maximum torque. Final torques are assumed to be the maximum torques in many published studies. Our data show this assumption can be incorrect if there is a reversed bevel at the top of an implant. Figure 4 shows that RP implants have a retrograde slope near the top. Our study shows this back-tapered coronal design compromises mechanical strength during implant insertion. The drop-off from maximum to final torque ranged from 17% to 49% (rows 2–7 in Table 3) depending on the cortical bone construct. The average drop-off of all conditions is 34%, excluding the trabecular block alone for the RP implant groups. The NP implant lacks this retrograde slope and does not show the falloff in torque during insertion. In addition to this drop in torque, the complex thread design shows up in the data as 1 or 2 plateaus between 3 and 8 mm as well as several smaller features in the curves. This may provide a structure-function link between implant design and mechanics of implantation.
In addition to providing mechanics related to geometry, the measurement system provides automated, reproducible quantitative measures of implant behavior. For example, Table 4 shows the percentage of total final insertion torque contributed from the trabecular underlayer component clearly decreasing as thickness of the cortical overlayer increases. The final insertion torque of an RP implant inserted into a 1-mm-thick cortical overlayer showed a contribution from the trabecular component of 18%. This is remarkable given that 92% of the surface contact between implant and sample was accounted for by this trabecular underlayer. Once the cortical layer reached 3 mm for the RP implant, the trabecular component contributed only 5% to the final insertion torque, an almost negligible quantity. Clinical significance of this result rests on the fact that human bones are anisotropic layered composite materials exhibiting a great variation of their mechanical properties. The moduli of cortical bone ranges from 8.6 to 25.9 GPa and of trabecular bone from 4.3 to 6.8 GPa.23–25 Tables 2 and 3 provide quantitative measures demonstrating the important contribution of a cortical layer to the overall final insertion torque. This correlates with primary implant stability. Therefore, for a patient with osteoporosis, the layer of cortical bone may be the key factor for a clinician to be able to predict potential risks or successes of obtaining an acceptable final insertion torque necessary for primary implant stability.
A secondary but significant effect to cortical layer thickness is seen in Table 4. A statistically significant increase (P < .05/32) in final insertion torque is seen with 2-mm and 3-mm cortical layers when the low-density underlayer is present beneath the cortical layer. Even though the underlayer taken alone accounts for <10% of the final insertion torque, there is an increase of 14%–47% in overall final insertion torque when the underlayer is present beneath the cortical overlayer. Thus, even though final insertion torque is dominated by the thin cortical overlayer, it is clinically important to pay attention to the lower-quality bone beneath this layer. The difference between fluid-filled space vs low-quality trabecular bone space can impact the final insertion torque by as much as 47%.
The synergistic effect may be explained by considering that when an implant is rotated into a predrilled hole, elastic and plastic deformation of the bone material occurs and forms crack openings as the implant advances. Crack-opening stress is often higher when the crack advances in the wake of plastically deformed materials; eventually, some of the material is sheared off and compressed between threads and bone. Insertion torque is measured from the combined resistance of friction, shear, and compression stresses of trabecular and cortical bone materials at the implant-bone interface. Smear effects and cavity expansion phenomena could explain the observed synergistic effect26,27; the final insertion torque into layered constructs is larger than the summation of final insertion torque into overlayer alone plus underlayer alone. The fact that this effect was not strongly apparent with 1-mm overlayers was likely due to the fact that the standard deviations for the 1-mm samples were substantially larger as a percentage of mean torque than their 2-mm and 3-mm counterparts. This likely occurred because these thin overlayers bowed under the axial insertion pressure, causing fluctuations in the measured torques.
The falloff and synergistic effect phenomena observed in this study are intriguing. Torque vs displacement data allow for a better understanding of the characteristics of insertion torque as a function of implant displacement. Results from this in vitro model provide a good foundation for future animal-model studies. Ultimately, information derived from this and future work will be beneficial for clinicians to improve primary implant stability.
A new measurement system was presented enabling quantitative evaluation of insertion torque at different implant locations. The system showed that retrograde slopes negatively impact primary implant stability when inserting into thin dense cortical layers. For Nobel Biocare Regular and NP implants, final torque values depend primarily on the thickness of cortical bone when this layer is 1–3 mm thick. Finally, there is a synergistic effect when inserting implants into thin dense layers over lower-density blocks. The sum of the final insertion torque into the 2 elements taken alone is less than the final torque into the laminated construct.
This study received financial support from the American Academy of Implant Dentistry Research Foundation, and Nobel Biocare Services AG provided components (research grant 2011-1050).