Since the late 1970s, distal femoral fractures, especially unstable or intra-articular patterns, have been managed with surgery. Open reduction and internal fixation with a blade plate, as recommended by the AO/ASIF (Association for the Study of Internal Fixation) in the mid-1970s, soon became the gold standard. However, over the past decades, intramedullary devices have been increasingly used in the management of distal femoral fractures1,2. The distal third of the femur is involved in 6% of femoral fractures3. In these cases, there are apparently two different patient groups. The first group consists of younger people—predominantly men who are less than forty years old—with high-velocity trauma. The second group consists of older people—predominantly women—with osteoporotic fractures as a result of low-velocity trauma4-6.
The AO/ASIF-defined goals7 of restoration of the anatomy, stable fracture fixation, preservation of blood supply, and early mobilization are demanding and difficult to attain in osteoporotic bone stock, which often does not allow stable device fixation8-11. In response to this challenge, novel implants (intramedullary nails with special distal interlocking features, and angular stable devices) and modified (less invasive) surgical techniques have been developed.
This study was performed to investigate the biomechanical stability of different fixation devices for the treatment of comminuted distal femoral fractures. We evaluated the effect of three intramedullary distal locking methods in comparison with a locking plate on fracture fixation stability in osteoporotic bone, using both a cadaveric model and a synthetic osteoporotic bone surrogate model.
Implants
Three titanium-alloy intramedullary nails with different interlocking configurations in the distal fragment were evaluated. They included (1) a 320-mm-long and 12-mm-diameter femoral nail (T2 Femoral Nail [T2]; Stryker, Schönkirchen, Germany), with two lateral-to-medial 5-mm-diameter partially threaded locking screws (Fig. 1, a); (2) a 320-mm-long and 12-mm-diameter supracondylar nail (T2 Supracondylar Nail [SCN]; Stryker), with one anteromedial-to-posterolateral and one anterolateral-to-posteromedial 5-mm interlocking screw in addition to two lateral-to-medial condylar bolts (Fig. 1, b); and (3) a 320-mm-long and 12-mm-diameter distal femoral nail (DFN; Synthes, Umkirch, Germany), with a 6-mm fully threaded screw and a lateral-to-medial spiral blade (Fig. 1, c). Proximally, the three intramedullary devices were locked in the femoral shaft with two anterior-to-posterior screws. The fourth device was a steel-alloy 310-mm-long fourteen-hole angular stable plate (AxSOS; Stryker, Duisburg, Germany) anchored in the distal fragment with two lag screws and five angular stable screws. Proximal plate fixation to the femoral shaft was by means of two angular stable screws and two cortical screws (AxSOS) (Fig. 1, d). All constructs were performed as recommended by the implant manufacturer.
Fracture Model
The biomechanical properties of the constructs were investigated with a simulated AO/ASIF type 33-C2 fracture12 (Fig. 2). The pattern was produced with use of a transverse osteotomy, which created a 1.5-cm fracture gap at a point 6.5 cm proximal to the joint line, and a sagittal intercondylar osteotomy.
Synthetic Bone Model
Third-generation composite femora (Sawbones Europe, Malmö, Sweden) were modified by the replacement of their distal, condylar portion with anatomic polyurethane foam condyles fabricated at our laboratory with use of H200-AT two-component structural foam (Vosschemie, Uetersen, Germany). A mean foam density (and standard deviation) of 150 ± 5 kg/m3 was chosen to simulate osteoporotic bone stock13 and to more realistically model construct failure conditions in the condylar region. Each implant was tested under torsional and axial loading with use of five synthetic bone specimens, for a total of forty synthetic bone specimens. Separate specimens were used for torsional and axial loading tests.
Human Cadaveric Bone Model
Eight pairs of fresh-frozen human femora, harvested post mortem with consent, were used for a matched-pair comparison of the axial cycles to failure of the SCN and DFN constructs. The mean age of the donors (four male and four female) at the time of death was 74 ± 9 years. The femora were completely stripped of soft tissues. The bones underwent biplanar radiography to rule out bone disease, and bone mineral density was determined with dual x-ray absorptiometry (Lunar Prodigy Advance; GE Healthcare, Madison, Wisconsin). The specimens were stored in vacuum packs at —27°C and were thawed at room temperature for twenty-four hours before testing. The specimen lengths of the femora were standardized to 455 mm by cutting the shaft if required.
Test Setup
The test setup, in particular the principle of eccentric axial loading, was adapted from Milne and Latta14. The specimens were mounted, by means of two universal joints, on a biaxial electrohydraulic material testing machine (model 8874; Instron, High Wycombe, Bucks, United Kingdom) in such a way so as to align the mechanical axis of the femur with the machine axis (eccentric loading) (Fig. 3). The upper joint allowed tilting in the coronal and sagittal planes, and the lower joint was fixed to constrain the specimens to coronal plane movement such that only medial-lateral translation was possible. The distal end of the specimen was mounted on an ultra-high molecular weight polyethylene tibial insert of a total knee replacement. During the torsion tests, the distal, condylar portion was additionally restricted from rotation by a metal cage with four set screws. The screws secured the distal portion by clamping the condyles from the front and back. The set screw positions were arranged such that they would not contact or otherwise interact with the implant. The proximal portions of the synthetic femora were embedded, over a distance of 10 cm, in a split aluminum-polyurethane resin shell. The human cadaveric specimens were proximally potted in a two-component cast resin (RenCast FC 52; Huntsman Advanced Materials, Monthey, Switzerland) inside a split aluminum shell.
Time, displacement, load, angle, moment, and number of cycles were acquired and plotted with use of MAX software (version 9.2; Instron, Canton, Massachusetts).
Test Protocol
Synthetic Bone and Human Cadaveric Specimens
For torsional loading, a static compressive axial load of 20 N was applied and the specimens were cyclically torqued for ten cycles, in external and internal rotation around the mechanical axis of the femoral shaft, to 10 Nm, at a loading rate of 1 Nm/s. Torsional stiffness, range of motion, and neutral zone, as defined by Wilke et al.15, were obtained. The torsional stiffness was calculated with regression analysis of the linear portion of the torque-angular rotation curve. The range of motion reflects the deformation, in degrees, under a given maximum torque (in this study, at ±10 Nm), and the neutral zone represents the laxity of the construct, which we calculated in a corridor of ±0.5 Nm around 0 Nm. For these parameters, the mean of the ten cycles was calculated.
Following torsional testing, static external rotation was applied to failure, at a rate of 1 Nm/s, to determine the ultimate torsional strength of the constructs. With the test setup used in this study, ultimate static torsional strength could be determined for the nailed constructs only. During torsional loading of the constructs with the AxSOS plate, contact occurred between the plate and the metal cage of the antirotation fixture, inhibiting the calculation of the ultimate torsional strength.
Cyclic axial loading was performed incrementally, starting with an axial load oscillating between 20 N and 200 N, at a rate of 2 Hz, and raising the lower limit by 10 N and the upper limit by 100 N after each 500-cycle step. After completing a loading level, the machine was manually set to the mean of the subsequent load level. The necessary readjustment required a pause of less than thirty seconds.
The initial axial stiffness of the constructs was obtained from cycles 50 through 59 of the first cyclic loading step. For that purpose, the slope of the linear area of the load-displacement curve was calculated for each of these cycles by linear regression and then we determined the mean and standard deviation for each construct.
Construct failure as a result of cyclic loading was defined as deformation of =5 mm at an axial load of 190 N. The deformation was determined from machine displacement. The number of cycles when 5 mm of deformation was achieved was recorded.
Human Cadaveric Specimens
The SCN and the DFN constructs were also evaluated in eight pairs of human cadaveric femora to study the cyclic axial stiffness and strength of the constructs in a matched pair comparison. The cyclic testing protocol was the same as that for the synthetic femora.
For the human cadaveric bone constructs, failure was defined as deformation of =2.5 mm because the SCN constructs only reached a mean of 3.15 mm of deformation after 12,500 cycles. To compare the results of the human and the synthetic bone, we used the 2.5-mm criterion for both.
Statistical Analysis
Data analysis was conducted with use of Excel (Microsoft 2003; Microsoft, Unterschleissheim, Germany). Regression analyses were performed for calculation of the mean and standard deviation of torsional and axial stiffness. For statistical analysis, SPSS for Windows (version 13.00; SPSS, Chicago, Illinois) was used. Torsional stiffness and strength and axial stiffness and strength were analyzed for normal distribution with the Kolmogorov-Smirnov and the Shapiro-Wilk test. For our data, no property demonstrated a normal distribution in any group. Thus, the groups were tested for differences with use of the Mann-Whitney test. Significance was set at p < 0.05.
Source of Funding
This study received outside funding from Stryker Trauma (Schönkirchen-Kiel, Germany). They supported us by paying for all implants and synthetic bone models. Also, a donation for the human cadaveric bones was given to the Institute of Anatomy.
Synthetic Bone Model
Torsional Tests
The AxSOS plate construct achieved the greatest torsional stiffness (p < 0.001). The SCN construct had 88% of the torsional stiffness of the AxSOS plate construct; the T2 construct, 70%; and the DFN construct, 34% (Fig. 4). Of the intramedullary nails, the SCN and the T2 femoral nail produced significantly stiffer constructs than did the DFN nail (p < 0.001; Table I).
The AxSOS plate construct had the smallest range of motion (4.17°; p < 0.001) of all of the constructs tested. Of the intramedullary nails, the SCN construct had the smallest range of motion (p < 0.001), which was, however, 21% greater than that of the AxSOS construct. The mean range of motion for the T2 construct was 52%, and that for the DFN construct was 217%, of that of the AxSOS plate construct. All differences were significant, with p < 0.001 (Fig. 5, Table I).
The AxSOS plate construct had the smallest neutral zone (p < 0.001), whereas the neutral zone of the DFN construct was 6.7 times larger than that of the plate constructs. The SCN nail and T2 constructs had a significantly smaller neutral zone than the DFN construct (p < 0.001; Fig. 5; Table I).
With regard to static ultimate torsional strength, failure first occurred in the T2 construct (38 ± 14 Nm), followed by the DFN construct (40 ± 5 Nm). The SCN construct did not fail at the torque cell limit of 100 Nm. With use of the torque cell limit, the SCN construct had the highest static ultimate torsional strength, which was significantly higher than that for the other two constructs (p = 0.008). The T2 and DFN constructs failed by disruption of the distal bone model in the frontal plane. All specimens failed by splitting first from medial to lateral and then from distal to proximal.
Incremental Cyclic Axial Tests
The SCN construct achieved the greatest initial axial stiffness (280 ± 10 N/mm; p < 0.001, Fig. 6) followed by the DFN construct (257 ± 8 N/mm) and then the T2 construct (235 ± 15 N/mm). The AxSOS plate construct achieved 111 ± 11 N/mm, which was only 40% of the stiffness of the SCN constructs, and thus had the least axial stiffness (Fig. 6). All differences were significant (p < 0.001).
With regard to axial cycles to failure, the SCN construct had the highest axial cyclic deformation limit, with a mean of 6212 ± 909 cycles to failure and a mean failure load of 1400 ± 122 N (p = 0.008, Fig. 7). The AxSOS plate construct withstood 3504 ± 750 cycles, thus achieving 56% of the axial cycles to failure of the supracondylar nail construct (p = 0.008). This made the AxSOS plate construct significantly stronger in this testing protocol compared with the other two nails (p = 0.032 for the DFN construct, and p = 0.008 for the T2 construct). The DFN and the T2 constructs had the lowest number of axial cycles to failure, failing after 2504 ± 366 and 2090 ± 196 cycles, respectively; the difference between these two constructs was not significant (p = 0.056). The mean failure loads were 900 ± 160 N for the AxSOS plate, 700 ± 80 N for the DFN constructs, and 600 ± 49 N for the T2 constructs.
The failure modes in the incremental cyclic axial fatigue test differed markedly among the implants used. The T2 (Fig. 8, a) and the DFN (Fig. 8, b) constructs showed distal-to-proximal widening of the intercondylar fracture gap, with subsequent proximal displacement of the medial condyle. With both devices, this condyle was also disrupted in the coronal plane. The SCN and the AxSOS plate constructs failed by medial screw loosening in the bone model (Fig. 8, c and d).
Human Cadaveric Bone Model
The mean bone density (and standard deviation) of the specimens was 0.822 ± 0.15 g/cm2 in the SCN group and 0.831 ± 0.16 g/cm2 in the DFN group; the difference was not significant (p = 0.725). According to the literature, our specimens can be characterized as mildly osteoporotic16. The mildly osteoporotic bone quality was also confirmed by the obtained median T score of —217.
The initial axial stiffness was 382 ± 58 N/mm for the SCN construct and 358 ± 37 N/mm for the DFN construct. The 6% difference was significant (p < 0.001) (Fig. 6).
With regard to axial cycles to failure, the end point of =2.5 mm of deformation was reached after a mean of 10,715 ± 1499 cycles by the SCN construct and after a mean of 6210 ± 1690 cycles for the DFN construct (Fig. 9). The 42% difference was significant (p = 0.001).
With regard to failure modes, the SCN constructs had minimal observed loss of reduction in the intercondylar fracture gap; screw loosening was seen in only one specimen. The DFN constructs behaved as in the synthetic bone model, with rapid loss of reduction in the intercondylar fracture gap and distal gap widening. In one specimen, this was followed by medial screw loosening and coronal plane disruption of the medial condyle.
Synthetic Bone Compared with Human Cadaveric Bone Constructs
Both the initial axial stiffness (Fig. 6) and the axial cycles to failure (Fig. 9) were markedly higher, in absolute terms, in the human cadaveric bone constructs. In the human bone model, the SCN constructs had 36% greater stiffness and the DFN constructs had 39% greater stiffness than did the constructs in the synthetic bone model. These differences were significant (p < 0.001). The human bone constructs withstood 2.2 times (SCN constructs) to 2.7 times (DFN constructs) the number of cycles survived by the synthetic bone constructs (p = 0.002). However, the differences between the DFN and the SCN constructs were of a similar magnitude in the two models. In the synthetic bone model, the SCN construct had 8.9% greater stiffness than the DFN construct; in the human bone model, the difference was 6.7%. The number of cycles to failure of the SCN construct was 1.7 times (human bone model) to 2.1 times (synthetic bone model) greater than that of the DFN construct.
This study investigated the biomechanical characteristics of four fixation devices, which mainly differed in their distal interlocking patterns, for the treatment of comminuted distal femoral fractures in osteoporotic bone. This study supports the concept that distal locking has an influence on biomechanical behavior for this fracture model. The magnitude of the differences found in this study may be clinically important.
Distal femoral fractures may be, and commonly are, managed with intramedullary nailing or with plating9,18-22. Implant selection is guided by factors such as fracture pattern, joint involvement, bone stock quality, and soft-tissue condition21. Biomechanical studies have provided guidance as to the advantages and drawbacks of different fracture fixation23,24. Especially in osteoporotic bone, sufficient implant fixation of distal femoral fractures is a challenge. Fixation failures are common in osteoporotic distal femoral fractures, and the more stable the construct, the more likely a good clinical result25.
A newly developed intramedullary implant for distal femoral fractures featuring distal condylar interlocking bolts, the Stryker SCN, has not, to date, been the subject of many biomechanical studies. Höhle et al.26 did not find any significant difference with regard to axial and torsional stiffness among the SCN supracondylar nail, the LISS-DF (Less Invasive Stabilization System-Distal Femur; Synthes) nail, and the DFN (Synthes). In contrast to our study, Höhle et al. used an A3 fracture model, a fracture without joint involvement, although the distal interlocking was the same as that in our investigation. The results of Höhle et al. showed no significant difference between the SCN and DFN construct for torsional and axial stiffness.
The enhanced torsional stability and strength of the SCN compared with the DFN construct may be attributed to the compressive forces exerted on the fracture fragments by the condylar bolts, as well as to the three-plane configuration of the distal interlocking devices, which minimizes the neutral zone and thus provides enhanced stability. The DFN construct had the least stability, with the lowest torsional stiffness and the largest neutral zone. Since this device does not exert any interfragmentary compression, the fragments were able to move and displace along the intercondylar fracture site. Grass et al.9 compared the DFN construct, which has a spiral blade, with the Green-Seligson-Henry nail (Smith and Nephew, Memphis, Tennessee), which has two distal locking screws, and with an angled blade plate. They found the DFN construct to have significantly poorer torsional stability.
There is agreement, throughout a range of studies of distal femoral fracture management9-11,27-29, that plating provides greater torsional stiffness, whereas intramedullary devices have greater axial strength. In our study, all intramedullary devices showed similar axial stiffness compared with each other. Since the nails tested were of the same material, length, and diameter, the differences in their biomechanical behavior must have been due to the different distal interlocking patterns. The advantage of the spiral-blade design of the DFN device over a conventional two-screw interlocking design was demonstrated by Ito et al.30, who found the spiral blade to confer a significant increase of 41% in axial stiffness. In our study, the stiffness achieved with the DFN construct was 10% greater than that obtained with the T2 construct. In their comparison of the DFN device and the Green-Seligson-Henry nail, Grass et al.9 also found spiral-blade interlocking to provide better axial stiffness. The plated constructs in our study achieved 40% to 50% of the axial stiffness provided by the intramedullary nails. Zlowodzki et al.11 found the angled blade plate constructs to confer only 25% of the axial stiffness achieved with an intramedullary nail. Using a Synthes condylar plate, Ito et al.31 found the device to provide only 10% of the axial stiffness achieved with a Synthes intramedullary nail. These results may be accounted for by the design principle of the plate (a laterally eccentric applied device) and by the design features (material, thickness, and size) of the plate. The axial stiffness of plated constructs is a clinical concern: delayed union and even nonunion of plated distal femoral fractures have been reported, especially in cases where there was no medial support19,32,33. These adverse events have been attributed to poor axial stiffness permitting movement in the fracture gap34.
The axial strength of the implants in our study was influenced by the distal interlocking pattern, with a larger implant-bone contact area of the distal interlocking devices being associated with greater strength. Thus, failure occurred first in the T2 construct, followed by the DFN construct, the AxSOS plate, and finally the SCN construct. The implant-bone contact area of the spiral blade is 75% greater than that of a conventional interlocking screw. The combination of a spiral blade with an interlocking screw provides a 38% larger contact area than two conventional interlocking screws30. In our study, this construct appeared to result in enhanced stiffness and strength.
Ito et al.30 investigated the biomechanical properties of the spiral blade in comparison with conventional locking bolts in static compression testing. They found a 13% to 21% increase in axial strength of the spiral blade compared with conventional locking bolts. These results agree with the findings in our study, in which the DFN (spiral blade) constructs withstood 21% more cycles to failure than did the T2 (conventional locking bolt) constructs. The highest cyclic axial strength was achieved with the SCN construct, which has a three-plane configuration of four distal interlocking devices, two of which are condylar bolts with washers and nuts. The diagonal bolt arrangement of the SCN construct provides more secure fixation in the denser bone of the posterior condyles, a feature described as beneficial by Ingman35.
The failure modes in the incremental cyclic axial test observed by us have been reported by other authors, both in synthetic bone models and in human femora. Trough formation in the bone was found by several authors28-31; plate failure was found by Ito et al.31 and by Zlowodzki et al.11.
For the validation of our synthetic bone model, the cyclic axial test was repeated in human cadaveric bone with use of the DFN and SCN intramedullary nails. The synthetic bone model results were found to be qualitatively reproducible in the cadaveric bones. While the cadaveric bone constructs had higher axial stiffness and numbers of cycles to failure, the pattern of the results observed in the human cadaveric bone model was similar to that seen in the synthetic bone model. For the SCN and DFN implants, the failure modes under cyclic axial loading were comparable between the synthetic and human bones. The differences observed between the models are likely attributable to the comparatively high bone density of the human specimens. While the density of the structural foam condyles had been specially chosen to simulate osteoporotic bone stock, the mean bone density of the cadaveric bones available for our study was 0.826 g/cm2, which is well above the severe osteoporotic bone densities in the literature. Koval et al.28,36 used specimens with a bone mass density of 0.3 to 0.5 g/cm2 and of 0.35 to 0.675 g/cm2 for biomechanical testing on three different plate constructs for osteosynthesis. Our findings suggest that our synthetic bone model lends itself to a first-line comparative biomechanical evaluation of a range of implants. However, the results obtained would need to be validated, in each case, with human bone.
The synthetic bone model (Sawbones third-generation composite femoral shafts and polyurethane foam condyles) used in our study was derived from earlier studies with use of foam cortical shell (Sawbones Europe) and Sawbones third-generation composite femora. In those studies, the two commercially available synthetic bone models had shown atypical failure modes1,4,37. Therefore, we decided to customize the distal part of our bone model using a self-made anatomical polyurethane foam model without a cortical shell.
Our test setup was adapted from the one used by Milne and Latta14; similar setups have been used in biomechanical evaluations of constructs for the management of distal femoral fractures11,26,38 and of proximal tibial fractures39. The loading regimes chosen for our study were based on the results of in vivo telemetry studies40,41, as well as on one biomechanical study42, and thus, were reasonably representative of physiologic postoperative loading of the distal end of the femur. Incremental axial loading was established as a testing methodology by Marti et al.24 and has been used by several authors11,21,43.
Our biomechanical study, however, had some limitations. (1) The simplified test setup did not simulate soft-tissue forces (capsule, ligaments, and muscles); (2) 500 cycles per loading step was not truly representative of a postoperative loading pattern; and (3) we were unable to repeat the full range of tests in the human cadaveric bone model.
In this study, we demonstrated that, with intramedullary nails, the distal interlocking pattern is important to the stabilization of distal femoral fractures, in terms of both the primary biomechanical stiffness and the strength of the constructs.
The results obtained in the synthetic bone model for the SCN and DFN constructs were also found in a human cadaveric bone model, allowing for differences in bone density.
Because the SCN construct provided significantly higher axial stability than the other constructs and torsional stability that was nearly comparable with that of the AxSOS plate, the SCN construct had the best combined biomechanical stability. Clinically, we suggest that, if predominantly torsional stability is needed, e.g., for patients who are confined to bed, the AxSOS plate will be sufficient, since treatment of distal femoral fractures by plating is fast and straightforward. In contrast, the findings of our study support the concept that the SCN construct should be considered for mobile patients for whom early postoperative mobilization for rehabilitation is desired.