Load Sharing in the Femur Using Strut Allografts: A Biomechanical Study.

Background: Femoral strut allografts are used in revision hip arthroplasty for management of bone loss associated with implant failure or periprosthetic fractures. They have also been used to treat unremitting thigh pain in well-fixed cementless femoral stems, to address the differential in structural stiffness between the stem and femoral shaft. Our study used an in vitro biomechanical model to measure the effect of placement of allografts on femoral strains, to determine their load-sharing capacity. Material and methods: Three rosette strain gauges were applied to the femoral surface of each of 6 cadaveric femurs, at the stem tip level on anterior, medial, and lateral cortices. After stem implantation, cortical strut allografts were applied to the lateral femoral shaft and secured with 4 Dall-Miles cables. A fourth gauge was placed on the midpoint of the allograft. Strains were recorded in the intact femur, then the implanted femur with and without the allograft under simulated physiologic loading in a load frame.
Results: Reduction in distal femoral principal strains, between 12% and 59%, was seen in all cortices following placement of the allograft. Under axial loading, 30% of the strain in the lateral cortex was borne by the allograft. Greater reductions in strain, by as much as 59%, occurred under axial load and torque.
Conclusion: The results of this biomechanical model indicate that by placement of an allograft, cortical strains can be reduced to levels approaching those in an intact femur, supporting this technique for treatment of unremitting thigh pain in well-fixed prostheses.

Introduction

Strut allografts are in common use in revision hip arthroplasty in the setting of bone loss or in the treatment of periprosthetic fractures [[1], [2], [3]]. In addition to these uses, some authors have advocated application of a strut allograft in the treatment of enigmatic thigh pain, that is, thigh pain present in a well-fixed, noninfected femoral stem [4,5].

Some of the clinical success of cementless femoral stems has been marred by the incidence of thigh pain, which has ranged in recent literature between 4% and 18% [[6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16]]. Often the thigh pain resolves or is mild to moderate in nature and can be treated with analgesics. However, there are a subset of patients who have unremitting thigh pain requiring more aggressive management [4,5,7].

Although the mechanism of thigh pain in a well-fixed stem is not completely understood, some theories have been proposed, suggesting a mechanical etiology. Some authors suggest that thigh pain arises from focal stress transfer to the diaphysis, consistent with cortical hypertrophy seen at the stem tip [[17], [18], [19]]. Micromotion at the stem tip of proximally fixed systems is another proposed mechanism [20,21]. Others propose that a mismatch in structural stiffness between the implant and the femoral shaft causes increased strain at the implant bone interface and within the cortical bone, resulting in thigh pain [8,15,16,[22], [23], [24], [25], [26], [27], [28]].

Numerous femoral stem designs have been introduced to optimize periprosthetic strain distribution. The majority of these designs use variations of stem shapes to maximize load transfer to the proximal femur and minimize stress shielding in this region, with the intention of mitigating bone resorption due to stress shielding [29]. At the same time, the distal stem is tapered or sometimes even split to avoid higher bone stresses near the stem tip. However, a number of designs successfully achieve press-fit fixation using precise diaphyseal fit, which may be necessary depending on patient anatomy and bone quality. In addition to using titanium alloy, which has a lower modulus of elasticity, 3D-printed porous titanium femoral stems have been introduced with the intention of minimizing stress shielding [30]. Collectively, these design changes also help mitigate higher femoral stresses near the stem tip. While these improvements in implant design may reduce the overall incidence of thigh pain in the future, the question still remains of how to treat patients with unremitting thigh pain in the presence of a well-fixed femoral component.

To address the patient with unremitting thigh pain while retaining the femoral stem, a technique of using a laterally placed femoral strut allograft cabled to the femoral shaft was used by Dr. Anthony Hedley. In principle, it is designed to increase the sectional modulus of the bone in order to decrease the discrepancy in structural stiffness between the stem and the femoral shaft. Over the last few years, there have been a few small case series that have reported on this technique [4,5].

The purpose of this study was (1) to determine the effect of femoral strain at the stem tip by placement of a cabled allograft and (2) to determine the load-sharing capacity of the allograft cabled to the femur.

Material and methods

Six cadaveric femurs were selected based on radiographs in order to obtain disease-free bone with adequate bone stock. Radiographs with 15% magnification were used to template for the porous-coated anatomic femoral stems (Howmedica, Rutherford, NJ; now Stryker, Kalamazoo, MI). This femoral stem design was selected based on prior clinical series showing a relatively higher rate of enigmatic thigh pain [23,31]. The bones were stripped of soft tissues, and the femoral templates were used to determine the strain gauge position at the anticipated stem tip location. Three rosette strain gauges (Micro-Measurements, Inc., Raleigh, NC) were placed on the prepared cortical surfaces using M-Bond 200 adhesive. The distal femur was cut at the proximal end of the distal metaphysis, and the femurs were potted to a 10-cm depth in epoxy potting material. Specimens were brought to room temperature. Using an MTS-812 servohydraulic load frame, the specimens were loaded 5 times to peak axial load prior to recording measurement. The femurs were then ramp-loaded to 500 N with the load directed through the femoral head and parallel to the shaft of the femur. A linear bearing restricted motion of the femoral head to the frontal plane. While maintaining axial load, internal and external torsional loads of 10 Nm were applied about the femoral shaft. Strains were recorded continuously during axial and torsional loading. This was performed twice for each gauge in order to obtain individual load strain curves.

The intact potted femurs were then implanted with appropriately templated porous-coated anatomic stems by Dr. Anthony Hedley using standard surgical technique. Polymethyl methacrylate (PMMA) cement was used around the proximal porous coating to simulate bone ingrowth. Since fixation with PMMA is established in the arthroplasty literature to provide secure initial fixation of femoral stems, it was deemed to adequately secure the proximal femoral stem in the cadaveric experimentation in the present study to simulate an ingrown implant. Adequate fixation was expected around the porous-coated surface where cement would interdigitate in the pores, particularly for the few cycles of testing and measurement. Radiographs of the implanted femurs were then obtained to confirm strain gauge location relative to the stem tip. In some cases, the strain gauges had to be replaced to a more proximal position in order to keep the gauge within 0.5 cm of the stem tip (n = 2). Load was then applied, and strain was measured using the same protocol as for the intact femur.

Next, lateral strut allografts measuring 12-14 cm in length were contoured to fit the lateral cortex of each femur. A fourth rosette strain gauge was applied to the midpoint of the allograft. The strut was then centered over the lateral cortex such that the strain gauge on the allograft was centered over the stem tip. It was then carefully affixed to the femur using four 2.0-mm stainless steel Dall-Miles cables. Load was applied, and strains were recorded using the same protocol as with the intact femur. Additional recordings were made for the rosette on the allograft.

Data analysis and statistical methods

The primary outcomes in this study were principal femoral cortical strains near the stem tip and principal strains measured on the allograft. From the readings of each rosette strain gauge, the maximum and minimum principal strains, maximum shear strains, and the directions of these strains were calculated. This facilitated assessing the greatest strains at each gauge location, regardless of the orientation of the gauge. In the present study, all maximum principal strains were tensile strains, and all minimum principal strains were compressive strains.

The most linear portion of each load-strain curve was used to calculate the slope of the curve. For each gauge, the mean of 2 runs was calculated. Paired t-tests were used to compare the principal strains between the intact and implanted specimens and between the implanted and implanted-with-allograft specimens. The use of paired t-tests ensured that variations in bone shape and quality were minimized by using the differences in strain among the loading conditions for each femur.

Bar graphs were used to present the mean and standard error of principal strain and shear strain among the cadaver femurs tested with each condition: intact, implanted, and implanted with allograft. Separate P values were calculated for comparing tensile strains (maximum principal strains) and compressive strains (minimum principal strains).

Results

Intact bones

As anticipated, under axial load, principal strains in the lateral cortex were predominantly tensile, and those on the medial side were predominantly compressive. Anteriorly, principal strains were predominantly compressive. Under axial load, the peak strain was 1080 με tensile on the lateral cortex, 620 με compressive on the anterior cortex, and 1700 με compressive on the medial cortex (Fig. 1a and b).

Figure 1

Principal and shear strains in intact specimens, implanted specimens, and implanted specimens with allografts, under axial load alone. Maximum principal strains were all tensile, and minimum principal strains were all compressive. Specimens were loaded to 500 N of axial load, but strains presented here were extrapolated to 1000 N to facilitate comparison with reports using higher loading. (a) Maximum principal strain with axial load; (b) minimum principal strain with axial load; (c) maximum shear with axial load.

With the condition of axial load combined with external rotation, maximal and minimum principal strains were approximately equal, around 600 με on the lateral cortex, suggesting strain was affected predominately by torsional loads rather than bending. The relative equivalence of maximum and minimum principal strains in the intact bone under conditions of internal and external rotation was seen in all cortices, again suggesting primarily torsional effects (Figure 2, Figure 3a, b).

Figure 2

Principal and shear strains in intact specimens, implanted specimens, and implanted specimens with allografts, under combined axial load of 500 N and external rotation. Maximum principal strains were all tensile, and minimum principal strains were all compressive. (a) Maximum principal strain with axial load and external rotation; (b) minimum principal strain with axial load and external rotation; (c) maximum shear with axial load and external rotation.

Figure 3

Principal and shear strains in intact specimens, implanted specimens, and implanted specimens with allografts under combined axial load of 500 N and internal rotation. Maximum principal strains were all tensile, and minimum principal strains were all compressive. (a) Maximum principal strain with axial load and internal rotation; (b) minimum principal strain with axial load and internal rotation; (c) maximum shear with axial load and internal rotation.

Shear strain was highest in the medial cortex under axial load, with strains around 2300 με. Under torsional loads, shear strain was highest in the anterior cortex, around 1390 με (Figure 1, Figure 2, Figure 3c).

Implanted bones

Following implantation, under axial load, all gauge measurements remained essentially the same, with a slight decrease in principal strain of 50-100 με. Shear strains also remained similar, decreasing by about 50-200 με (Fig. 1a-c). However, under axial load combined with torsional loads, there was a significant increase in strain, most notably in the anterior cortex, where a nearly threefold increase in tensile strain was observed with external rotation (800 με to 2320 με) (P = .02) (Fig. 2a). Similarly, under axial load and internal rotation, compressive strains on the anterior cortex increased more than 1000 με from 775 με to 1850 με (P = .01) (Fig. 3b). Shear strain increased on all cortices by 1.5- to 2-fold compared with those measured with the intact femur (Figure 2, Figure 3c).

Addition of allograft

Following placement of the laterally cabled allograft, a reduction in strain under axial and torsional loads was observed on all cortices. Under axial load, the allograft reduced strain mainly in the lateral and medial gauges by 15%-20%. The strain gauge on the allograft experienced approximately 350 με tensile.

A greater reduction in strain was observed under combined axial and torsional loads, especially on the anterior cortex. With external rotation, strain on the anterior cortex was reduced by approximately 25%, from 2320 με tensile to 1600 με (P = .05). With internal rotation, strain on the anterior cortex was reduced by approximately 25%, from 1850 με compressive to 1340 με (P = .04) (Figure 2, Figure 3a, b). Again, similar reductions were observed in shear strain, with the allograft experiencing approximately 40% of the shear strain observed on the anterior cortex and 70%-75% of that observed on the lateral cortex (Figure 1, Figure 2, Figure 3c).

Discussion

In the present study, a biomechanical model was constructed using cadaveric femurs to measure the effect of a strut allograft to reduce femoral strains near the distal femoral stem tip under load. The application of a strut allograft using stainless steel cables reduced strains on the femoral shaft, by as much as 59%. The findings of this laboratory study support the load-carrying capacity of the strut allograft. This finding has potential implications for reducing enigmatic thigh pain, as well as applications in revision hip arthroplasty and in the treatment of bone loss. Strut allografts have been used clinically to avoid revision of well-fixed stems, for the treatment of unremitting thigh pain [4,5], and also in the treatment of periprosthetic fractures [[1], [2], [3]].

In the present biomechanical study, cadaveric femurs were loaded under 500 N axial load combined with ±10 Nm internal and external rotational torsion. The axial load produced bending on the shaft and tensile strains on the lateral cortex. This axial load was applied parallel to the femoral shaft to maximize the load’s bending moment on the femur. The magnitude of torsion, ±10 Nm, more closely approximated the physiological range applied during normal gait, 12 to 18 Nm [32]. Torsional loads produce shear strains longitudinally and transversely, as observed in our study. Both axial loads and torsional loads were intentionally kept sub-physiological in order to avoid cadaveric femur fractures in this laboratory model. The risk of fracture in this type of biomechanical model is high due to the absence of the supporting soft tissue and muscle forces. Since strain is approximately a linear function of load until failure, the measured values were extrapolated to estimate physiological strains.

Accordingly, tensile and compressive strains measured on the femoral shaft in the present biomechanical study were well below strains required for failure of cortical bone in the femoral shaft. Specifically, mean tensile strains (estimated for 1000 N axial load) for intact femurs were well below 1500 με, increasing to a mean strain below 2500 με with an implant under axial load combined with torsion. Even considering a more physiological axial load, the strains would still not approach the failure strength of femoral cortical bone, reported as 10,700 με for older patients and 13,200 με for younger patients [33]. Presuming a linear progression of strain with axial load, physiological axial load with an implant would not produce sufficient strain in this simulated model to lead to failure even with repetitive loading.

On the other hand, shear strains in the present study increased with an implant under axial load and external rotation to a mean of 3239 με, corresponding to 94% of longitudinal shear strength reported by Turner et al. for patients aged 63 to 83 years [34]. Shear strains measured in the present simulated model are consistent with an increased risk of bone failure, potentially leading to pathologic periprosthetic fractures, particularly with repetitive loading. Although this was not the focus of the present study, it may represent an important finding with implications for predicting implant periprosthetic fractures in older patients and should be potentially considered in future implant designs.

Numerous femoral stem designs have been introduced with the goal of reducing both proximal strain shielding and strain at the bone-implant interface, particularly in the diaphysis. In general, tapered femoral stems are intended to reduce femoral strains near the distal stem tip. However, cylindrical femoral stems, such as the one used in the present study, have been shown to produce large cortical strains near the stem tip, [16,35,36] likely contributing to thigh pain reported more commonly with this type of implant [7,9,23,24,26,31,[37], [38], [39], [40], [41], [42], [43]]. The results of the present simulated cadaveric model suggest that the use of a strut allograft with stainless steel cables may substantially reduce the strains on all aspects of the femur around the stem tip. In our study, strains were reduced by as much as 59% by the application of a lateral strut allograft, such that they approached strains in the intact femur.

In all simulated loading conditions of the present cadaveric femurs, the strains in the anterior, medial, and lateral cortices were reduced by the application of the femoral allograft, especially under torsional loads, suggesting that this technique can mechanically increase the apparent structural stiffness of bone and that the use of the tensioned stainless steel cables allows load transfer to bone. A possible reason for greater reduction under torsional loads may be an improved interlock of the bone to the allograft under this condition, with improved load transmission to the allograft. Greater reductions in torsional strains may be possible when using the allograft in more osteoporotic bone [44]. However, the number of specimens in this biomechanical study is too small to draw definite conclusions.

Although reports of enigmatic thigh pain have decreased in recent literature, [5,6,9,14,15,20,42,[45], [46], [47], [48]] more than 500,000 total hip replacements are being performed each year in a younger population with increased life expectancy [[49], [50], [51]]. In the United States population, these are predominantly noncemented femoral components. In this setting, even a small percentage of patients with thigh pain represent a significant clinical problem.

The technique used by Dr. Anthony Hedley, with a lateral strut allograft cabled to the femur, addressed the theory that thigh pain results from a difference in structural stiffness of the femur relative to the stem. With this technique, the sectional modulus (or flexural rigidity) of the bone is increased by the allograft, thereby better matching structural stiffness of the bone to the stem.

This study was not without limitations. The laboratory cadaver biomechanical model in this study used a simplified loading protocol and did not incorporate abductor or extensor muscles or other physiological forces acting on the femur. Other limitations include a small sample size, simulated ingrowth of the implant using cement, and an implant design that is no longer in use. Also, although this study measured reductions in strain with a cabled allograft in the femoral cortex and established that loads can be transferred to the allograft in vitro, it did not address the intermediate and long-term fate of the allograft in vivo. The model also did not address micromotion of the stem tip relative to the cortical shaft, as this may also be related to thigh pain. The model tested the initial condition of the cables after tightening, but the tension may change as the patient begins to mobilize and bear weight. Previous studies have reported gradual incorporation of cortical strut allografts, ranging from 80% to 94.7% [3,[52], [53], [54]].

If the allograft bone follows Wolff’s law, it should remodel over time as it is experiencing strain in the loaded femur, at least in its initial phases. Further clinical studies may be required to quantify this potential mechanism. Should the theory of discrepancy in the flexural rigidity of the stem as compared with the bone indeed be a source of thigh pain, then the technique used in the present study has the potential to increase the flexural rigidity of bone in vivo. Within the limitations of the biomechanical simulation in this study, the technique described may provide a useful method to treat thigh pain without the disadvantages of removing a well-fixed femoral component. The data generated in this study raise concerns about risks of periprosthetic fractures, which should be further investigated with current stem designs and materials.