Chemotherapy Curtails Bone Formation From Compliant Compression Fixation of Distal Femoral Endoprostheses

Abstract

Background

Modulated compliant compressive forces may contribute to durable fixation of implant stems in patients with cancer who undergo endoprosthetic reconstruction after tumor resection. Chemotherapy effects on bone hypertrophy and osteointegration have rarely been studied, and no accepted radiologic method exists to evaluate compression-associated hypertrophy.

Questions/purposes

(1) What was the effect of chemotherapy on the newly formed bone geometry (area) at 1 year and the presumed osteointegration? (2) What clinical factors were associated with the degree of hypertrophy? (3) Did the amount of bone formation correlate with implant fixation durability? (4) Was the amount of new bone generation or chemotherapy administration correlated with Musculoskeletal Tumor Society (MSTS) score?

Methods

Between 1999 and 2013, we performed 245 distal femoral reconstructions for primary or revision oncologic indications. We evaluated 105 patients who received this implant. Ten were excluded because they lacked 2 years of followup and two were lost to followup, leaving 93 patients for review. All underwent distal femur reconstruction with the compliant compressive fixation prosthesis; 49 received postoperative chemotherapy and 44 did not. During this period, the implant was used for oncology patients < 60 years of age without metastases and with > 8 cm of intact, nonirradiated bone distal to the lesser trochanter and ≥ 2.5 mm of cortex. Our cohort included patients with painful loosening of cemented or uncemented stemmed femoral megaprostheses when revision with the compliant compressive device was feasible. Patients with high-grade sarcomas all received chemotherapy, per active Children’s Oncology Group protocols, for their tumor diagnosis. At each imaging time point (3, 6, 9, 12, 18, 24 months), we measured the radiographic area of the bone under compression using National Institutes of Health open-access software, any shortening of the spindle-anchor plug segment distance as reflected by the exposed traction bar length, and prosthesis survivorship. Clinical and functional status and MSTS scores were recorded at each followup visit. Duration of prosthesis retention without aseptic loosening or mechanical failure was evaluated using Kaplan-Meier analysis, censoring patients at last followup.

Results

Chemotherapy was associated with the amount of overall bone formation in a time-dependent fashion. In the 12 months after surgery there was more bone formation in patients who did not receive postoperative chemotherapy than those who did (60.2 mm², confidence interval [CI] 49.3-71.1 versus 39.1, CI 33.3-44.9; p = 0.001). Chemotherapy was not associated with prosthesis survival. Ten-year implant survival was 85% with chemotherapy and 88% without chemotherapy (p = 0.74). With the number of patients we had, we did not identify any clinical factors that were associated with the amount (area) of hypertrophy. The hypertrophied area was not associated with the durability of implant fixation. MSTS scores were lower in patients treated with chemotherapy (25 versus 28; p = 0.023), but were not correlated with new bone formation.

Conclusions

The relationships among chemotherapy, bone formation, and prosthetic survivorship are complex. Because bone formation is less in the first year when the patient is being treated with chemotherapy, it is not clear if the rehabilitation schedule should be different for those patients receiving chemotherapy compared with those who do not. The relationship between early bone formation and the timing of weightbearing rehabilitation should be evaluated in a multicenter study.

Level of Evidence

Level III, therapeutic study.

Introduction

Preservation of bone stock and promotion of osteointegration are goals to extend the durability of endoprosthetic reconstructions in oncology. Modular rotating-hinge knee endoprostheses are the most frequently used reconstructive approach after bone tumor resection because of their convenience, versatility, and good function [21], but their durability is limited by progressive stem fixation failure of the implant [1, 10, 20, 23]. An alternative fixation approach using a compliant compression mechanism is associated with bone formation in the fixation segment under compression decreasing stress-bypass osteoporosis and promoting osteointegration at the bone-prosthetic interface [2]. The periprosthetic bone under compressive forces hypertrophies as expected in accordance with Wolff’s Law [25]. These anatomic skeletal changes are associated with clinical stability of the implants and have been associated with 5- and 10-year survival of the distal femur implant of at least 80% [15, 17].

Bone homeostasis is also influenced by biologic conditions. Cytotoxic chemotherapy, commonly used in patients who receive megaprostheses, has negative effects on calcium balance, bone formation, and fracture healing and could affect the bone hypertrophy and osteointegration of the compliant compressive fixation device. Osteointegration at the surface of the spindle has been documented experimentally in a single clinical patient but not commented on in the only other retrieval study reported [7, 18]. The effect of chemotherapy on bone hypertrophy seen with adaptive compression force has also not been assessed quantitatively. Avedian et al. [5], in a limited study, suggested that cortical hypertrophy at the implant interface in patients who received chemotherapy was delayed compared with the rate in patients who did not receive chemotherapy.

The radiographic assessment of bone hypertrophy and osteointegration is challenging as a result of (1) the asymmetric shape of the newly generated bone throughout the length of the compression segment; (2) variations in the bone maturation and apparent bone density [19]; and (3) the potential for bone resorption-induced alterations in the shape of the distal femur in response to avascular necrosis or excessive compressive forces. The quantity of bone formed under these conditions is of great interest to understand the response of bone to compression as well as to guide the clinical management of the recipient patients. The cortical width, bone ingrowth at the bone-implant interface, and associated changes in the measured two-dimensional area between the anchor plug and bone-spindle interface should be quantified to gain more insight into this process. Furthermore, the relevance of remodeling or hypertrophy of the newly generated bone to prosthetic survival needs clarification.

We therefore asked: (1) What was the effect of chemotherapy on the newly formed bone geometry (area) at 3 months, 1 and 2 years, and presumed osteointegration? (2) What clinical factors were associated with the degree of hypertrophy? (3) Did the amount of bone formation correlate with the durability of implant fixation? (4) Was the amount of new bone generation or the administration of chemotherapy correlated with Musculoskeletal Tumor Society (MSTS) score?

Patients and Methods

Between 1999 and 2013, we performed 245 distal femoral reconstructions for primary or revision oncologic indications. We evaluated 105 patients who received this implant. Ten were excluded because they lacked 2 years of followup and two were lost to followup, leaving 93 patients for review. Under a protocol approved by our institutional review board, we conducted a retrospective study and reviewed the records of these patients who underwent distal femur resection and reconstruction with the compliant compressive fixation prosthesis (Compress® Device; Biomet Orthopedics, Warsaw, IN, USA) between April 2000 and July 2014 at our tertiary care cancer hospital. During the period in question, this kind of implant was the default choice of a distal femoral prosthesis. It was used whenever the cortical bone was sufficiently thick (2.5 mm minimum), the proximal femoral segment was a minimum of 8 cm below the midlesser trochanter, there was no proven metastatic disease, radiotherapy was not going to be needed, and the patient was thought to be compliant enough to follow the 12-week protected weightbearing regimen. During this time, a total of 71 patients at our institution underwent revision surgery for a failed distal femoral megaprosthesis. Twenty-eight met the clinical and anatomic criteria to receive a compliant compressive fixation-based implant. Indications for revision to a compliant compression fixation implant also included a minimum 2.5-mm cortical thickness of remaining proximal femur without significant areas of attenuation, cortical defect, or osteolysis. Intraoperative confirmation was used to meet this indication.

Contraindications to this procedure were inadequate cortical thickness (< 2.5 mm), pre- or postoperative bone irradiation, and metastatic disease. We excluded 10 patients who, as a result of tumor recurrence, death, or infection, lacked the 2-year followup required to measure the primary outcome of the evolution of bone hypertrophy. The remaining 93 patients were divided into two groups based on timing of chemotherapy in relation to time of surgery. The first group comprised 44 patients who received no chemotherapy or had a history of receiving chemotherapy > 1 year before surgery, and the second group comprised 49 patients who received pre- and postoperative chemotherapy, bracketing implantation of the compliant compressive fixation device. A majority of patients in both groups had a diagnosis of osteogenic sarcoma. Other diagnoses included chondrosarcoma, Ewing sarcoma, malignant fibrous histiocytoma, and giant cell tumor, among others, with a wider array of diagnoses in the group that did not receive chemotherapy. During the period in question, chemotherapy was used to treat all patients with high-grade cancers, per active Children’s Oncology Group protocols. Revision of a failed traditional, cemented, or uncemented stemmed implant was more frequent in those who did not receive chemotherapy (23 of 44) than among those who did receive chemotherapy (1 of 49 patients). The cohorts differed in demographic and surgical characteristics (Table 1). All reconstructions were done at our institution. Briefly, reconstruction after distal femur resection incorporates the following steps. The canal was reamed to accommodate the smallest 12-mm anchor in the medullary canal. The anchor plug and traction bar were inserted and secured with five pins. The conical dome reamer then was used to prepare the surface for application of the spindle over the traction bar. The spindle size was determined by the bone outer diameter, and the compressive force (400, 600, or 800 pounds) was selected based on the cortical thickness. The manufacturer recommended 400 pounds for cortices 2.5 to 4 mm, 600 pounds for those 4 to 5.4 mm, and 800 pounds for those ≥ 5.5 mm. The compression nut was tightened, approximately a half-turn beyond the point that initial resistance was felt, to compress the Belleville washers within the implant’s compression chamber [15]. The remaining components were assembled in accordance with the manufacturer’s recommendations. Derotation pins were only used in six obese patients.

Table 1.

Patient demographic and clinical characteristics (N = 93)

Chemotherapy was started 2 to 3 weeks postoperatively. Weightbearing was limited to 10 pounds, foot flat for 6 weeks, then 50% weightbearing for 6 weeks. Subsequently, patients were allowed to bear weight as tolerated, weaning from one crutch to the patient’s maximum achievable level of unassisted ambulation without limping. The time to independence was identified from the orthopaedic or physical therapy records.

Followup was for a minimum 24 months or until implant removal. We considered a patient lost to followup if he or she did not attend the clinic visits and could not be reached by phone for 2 years after the last followup date. There were three such patients censored at last followup. The median followup was 95 months (range, 24-194 months) for the nonchemotherapy group and 94 months (range, 24-193 months) for the chemotherapy group. Twenty-six patients (28%) had followup of < 5 years, 61 patients (65.6%) were followed for 5 to 15 years, and six patients (6.5%) had > 15 years of followup. Fifteen patients (16%) died during the prosthetic followup period.

We defined the endpoint of the prosthesis as revision resulting from mechanical failure, osteointegration failure, or aseptic loosening, and other causes of implant removal were censored. We defined compliant compressive fixation failure as any revision for complications related to the anchor plug, traction bar, spindle, sleeve, or fixation pins because of mechanical reasons. Failure of interface stability was associated with a loss of compression, reflected by a progressive decrease in the distance between the anchor plug and spindle at the bone-prosthetic interface. This effectively converted the traction bar into an inadequate thin stem that could bend or break (Fig. 1). Symptoms of periprosthetic bone failure were noted and typically manifested as patient complaints of thigh pain and tenderness at the spindle-bone junction or reports of thigh pain during examination when the hip was rotated in the 90–90 position (supine, hip flexed 90°, knee flexed 90°). We evaluated function based on the time it took to achieve complete weightbearing without any assistive device and recorded the MSTS score at the last visit. AP and lateral radiographs were obtained postoperatively and at 3, 6, 9, 12, 18, and 24 months after implantation during scheduled followup visits using a routine clinical technique. Radiographic image analysis was performed with ImageJ software, Version 1.15i (US National Institutes of Health, Bethesda, MD, USA; https://imagej.nih.gov/), which converts radiographs to digitized images for quantitative analysis [9, 13, 22]. ImageJ can measure distances and angles and calculate area. The software also supports image processing functions such as sharpening, smoothing, and edge detection. These features were used to accentuate the image of the newly generated bone. These are two-dimensional (2-D) measurements that are used as surrogates for the desired three-dimensional (3-D) information that is impossible to measure from the radiographs or even with advanced 3-D imaging with the metallic devices in place. Enhanced radiographs showed the bone formation (Fig. 2). This enabled us to quantify the new bone formation of the femur between the anchor plug and the spindle. All measurements were standardized by correcting for magnification using the centering sleeve diameter as an internal calibration. We outlined the border of the femoral cortex in the area between the anchor plug and the spindle, and the software calculated the whole area in square millimeters as well as mean width and mean height; we also measured the length of the visible traction bar (Fig. 3). Areal measurements were made on the AP and lateral radiographs and summed to generate estimates of total bone formation at each time point. To quantify net bone hypertrophy and resorption, (1) we subtracted the compression segment area calculated in the first postoperative radiograph from each subsequent radiographic calculation to define the magnitude of hypertrophy; (2) we calculated the change in length of the compressed bone segment using the radiograph obtained immediately postoperatively and the last available radiograph and assumed that any reduction was indicative of the bone resorption at the spindle-bone interface; and (3) we calculated the ratio between the traction bar length in the earliest postoperative radiograph and its length in the endpoint radiograph, whether at the last date of followup or before subsequent revision. Shortening of the traction bar occurs when there is inadequate osteointegration at the spindle-bone interface and the compliant spring-loaded compression continues to squeeze the spindle and anchor plug closer. Ultimately, the entire potential compression dissipates if the typical 7 to 9 mm of exposed traction bow is used up and the collar and anchor plug impinge. This occurs when the bone fails to integrate in the spindle. Next, the traction bar becomes a functional stem that breaks, heralding prosthetic failure. Clinically relevant time points were selected for the analyses: after the first 3 months, corresponding to the time of protected weightbearing; after 12 months, corresponding to the completion of chemotherapy; and at 2 years, the time when bone hypertrophy appears to plateau.

Fig. 1 A-B.

Radiographs illustrating the two modes of aseptic implant failure. In Type 1 failure (A), there is breakage of the traction bar as well as periprosthetic fracture at the bone-spindle interface (arrow). In Type 2 failure (B), there is a loss of compression with notable collapse of the traction bar but without concomitant bone fracture (arrow).

Fig. 2 A-F.

Serial postoperative radiographs obtained at 0, 3, 6, 9, 12, and 18 months in a patient whose implant developed mechanical failure as a result of loss of compression are shown. ImageJ was used to calculate all measurements. (A) (postoperative, 0 months) There is an increase in bone hypertrophy, as evidenced by increases in the cortical area (162 mm²) and width (5 mm). An increase in bone resorption was found, as demonstrated by the 1.4-mm reduction in cortical length found when we subtracted the measures between the first and last radiographs. (B) (3-month followup) A marked reduction in the length of the traction bar can be seen between A and F (yellow circles). (C) (6-month followup) ImageJ software allowed detection of the gradual mineralization in the callus (the gray circles in A and F) and accurate detection of cortical margins to achieve a more accurate analysis. (D) Nine-month followup. (E) Twelve-month followup. (F) Eighteen-month followup.

Fig. 3 A-F.

Example of the method to quantify new bone formation of the femur (mm²) between the anchor plug and spindle is shown (A). Values from the AP and lateral views were summed. Differences were measured between serial sets of radiographs to quantify the bone formation. Spatial calibration of the radiographic images was achieved by using the central sleeve diameter (B) as the calibrating value. The Edges tool in ImageJ detects the edge of the femur (C). The segment of interest is marked for measurement (D, yellow lines) with the Polygon Selection tool in the toolbar (E). With the Measure tool, accessible within the Analyze dropdown menu, values are calculated and displayed (F). In this example, the mean cortical width was 31.3 mm, the mean height was 50.5 mm, and the entire area was 1384.9 mm².

Statistical Analysis

All statistical analyses were performed with SPSS Version 14.0 (SPSS Inc, Chicago, IL, USA). We used the independent t-test to compare the 2-D projections as surrogates of volumetric measures of bone formation, traction bar distance, and demographic features between the chemotherapy and nonchemotherapy groups, the different types of fixation failure, and other clinical and surgical factors.

Linear regression analysis assessed the relationship between the increase in cortical width and area and the functional restoration. New bone formation was measured every 3 months and summed. The variance of each of these measurements was summed to generate the total variance for the bone formation. Survival of the prosthesis was analyzed using the Kaplan-Meier method, defining survival as the time from surgical implantation to the time of revision resulting from mechanical failure only or until the latest followup. We used Cox regression analysis to generate hazard ratios and 95% confidence intervals for factors associated with implant failure. Nonparametric measures were compared by Wilcoxon rank test.

Results

Chemotherapy delayed and diminished the amount of newly formed bone in a time-dependent fashion in three ways. Bone area increased more if the patient did not receive chemotherapy. In the 12 months after surgery there was more bone formation in patients who did not receive postoperative chemotherapy than those who did (60.2 mm², confidence interval [CI] 49.3-71.1 versus 39.1 mm², CI 33.3-44.9; p = 0.001) (Fig. 4). By the 18-month time point, the differential narrowed. In the second year, the amount of bone formation was the same between the chemotherapy and nonchemotherapy groups. Shortening of the spindle-anchor plug segment distance, reflecting the amount of bone resorption that occurred before stability and presumably osteointegration occurred, was equal in the patients receiving chemotherapy and those not receiving chemotherapy. Regardless of chemotherapy status, there was much more shortening in failed implants (mean, 4.3 mm; range, 0-9.9 mm) than successful ones (mean, 0.2; range, 0-2 mm; p < 0.009) (Table 2). There was no prosthetic survival advantage in the nonchemotherapy group (Fig. 5). Thus, chemotherapy was associated with curtailed early bone formation under compression but not the final amount of postoperatively generated bone. Bone formation was not correlated to patient, nutritional, surgical resection, or implant variables that could have been influenced by chemotherapy and might have been influential. The small frequency of prosthetic failure events was low enough that small size bias was a problem so further statistical analysis (eg, multivariate analysis) was not appropriate. No failures occurred after 1 year in the nonchemotherapy group, but occasional failures after 1 year did occur in those treated with chemotherapy (Fig. 1).

Fig. 4.

This figure shows the bone area increase in femoral segment under compression with or without postoperative chemotherapy. Data points are means and standard error bars are shown. Notably, changes after 1 year postoperatively were not correlated with postoperative chemotherapy administration.

Table 2.

Measurements of newly generated bone at 3 months postsurgery

Fig. 5.

The Kaplan-Meier survival analysis of the compliant compressive fixation prosthesis in both chemotherapy and nonchemotherapy patient groups is shown. There were ongoing failures in patients receiving chemotherapy during the first 3 years, whereas the patients not receiving chemotherapy only experienced failures in the first year. Notably, no late failure occurred in either group.

The second goal identified no other clinical factors that correlated with the cortical bone hypertrophy. Clinical factors did not correlate with the amount of bone formation. The factors analyzed were age, sex, body weight, height, body mass index, baseline and 6-month hemoglobin, absolute neutrophil count, absolute lymphocyte count, albumin, resection percentage of the femur, compression force, and spindle size.

Postoperative chemotherapy was associated with worse MSTS scores at the last clinic visit. We found a median MSTS score of 25 in the chemotherapy group and a median score of 28 in the nonchemotherapy group (p = 0.023). No consistent relationship was seen among the potentially interrelated factors of bone formation, weightbearing (Fig. 6), and MSTS score.

Fig. 6 A-B.

(A) Linear regression correlated the cortical area after 9 months and the time at which patients walked independently. There was an inverse correlation. Greater increases in cortical area correlated with faster restoration of independent walking. (B) This correlation was stronger in patients receiving chemotherapy than in patients not receiving chemotherapy.

Discussion

Bone formation associated with compression fixation of an implant to bone is potentially important for several reasons. The rationale behind the prosthesis studied in this article is that compliant compression provides long-term mechanical stimulation to the underlying cortical segment and should lead to bone hypertrophy in the compressed segment. Preservation of this bone stock is considered important, especially in young patients who will likely need revision of the prosthesis during their lifetime. Chemotherapy potentially alters this mechanism and reduces bone hypertrophy. We investigated this hypothesis using a validated methodology with NIH ImageJ software. Chemotherapy after the operation could adversely affect both elements of bone formation and be associated with poorer function and prosthetic survival. Thus, it was important to define the differences in bone formation, with and without chemotherapy. We found that chemotherapy delayed and diminished the amount of new bone formation, but we could not show an association with prosthetic failure and only a modest difference in function as assessed by MSTS score.

Our analysis has several limitations. First, this is a retrospective study. Although the data capture was comprehensive, we lacked enough patients to explore the effect of confounding variables such as differences in chemotherapy intensity. The data acquisition was rigorous to minimize selection bias; the Compress Device was always used when patients met the entry criteria, eg, no metastatic disease, no radiation, sufficient subtrochanteric bone, and sufficient cortical thickness, as noted in the Patients and Methods. Imaging and other data were available for all patients in the radiographic followup period, and only three patients were lost to followup in the long-term prosthetic survivorship analysis. An additional 10 patients had their prostheses removed prematurely to address tumor recurrence or infection. There may be other competing risks that were not captured in our analysis. A second concern is the appropriateness of the bone formation measures, theoretically and practically. Theoretically, what indicator of bone formation is important? No clinical measure of osteointegration exists. Geometric changes in the bone segment under compression may not reflect osteointegration at the spindle. New bone volume and ultimately bone strength are the most relevant for potential revision surgery. Areal measurements in two planes serve as a surrogate for volume. There was no direct measure of bone quality under compression or how it may be affected by chemotherapy. Technically, this study is limited by the use of nonstandardized clinical radiographs. We minimized this problem through internal calibration using the known size of the implant’s centering sleeve and use of ImageJ’s capacity to detect the true borders of the bone, facilitating accurate measurements and data calculation.

The small population size limited our ability to identify factors associated with bone formation. A larger population in each group would have increased the power of the analysis. Nevertheless, there was sufficient power to establish that chemotherapy was associated with an early reduction of total bone formation and at each measurement site. Another limitation is that there were modest differences in the Children’s Oncology Group chemotherapy regimens for the different cancer diagnoses, and these treatment differences could have influenced the results. Because the use of chemotherapy was a dichotomous variable (yes/no), variations in the drugs and doses were not captured in the analysis. However, chemotherapeutic agents used for common bone sarcomas have similar antiosteous effects [3, 6, 12, 24, 26]. Thus, the differences in chemotherapy are unlikely to have an influence on prosthetic fixation and the histopathologic diagnosis was not associated with differences in bone formation or prosthetic survivorship. Additionally, our two patient groups differed in several known and possibly other unidentified aspects. The chemotherapy group was younger and newly diagnosed with a high-grade bone sarcoma, whereas the nonchemotherapy group contained patients with revision prostheses or low-grade or benign neoplasms. It is possible that differences such as the magnitude of the resection of muscle and other soft tissues influenced our observations, but it would be difficult to design a study to assess the effect of chemotherapy differently than we have done here and we believe our findings are relevant to the treating surgeons.

Chemotherapy was associated with reduced bone hypertrophy in a time-dependent fashion. This was most profound in the first year, and the difference persisted. To our knowledge, this is the first article to address this concern and we used a new technique to assess bone formation at the interface as accurately as possible using plane radiographs. This study is the first to evaluate the phenomenon using a validated methodology: NIH ImageJ software. Second, permanent fixation requires stable osteointegration of host bone into the porous surface of the prosthetic spindle. Based on preclinical data, limited human retrieval analysis, and clinical and radiographic implant stability, compressive fixation is thought to promote osteointegration [7]. Unfortunately, there is no noninvasive method to assess bone formation in this surface. There are no published comparable data. Avedian et al. [5] described the geometry of new bone formation with respect to cortical width at the bone-spindle junction. They addressed a single measure as the definitive surrogate for the relevant osteointegration and overall bone formation. Our analysis was more comprehensive in a much larger population. Our premise was consistent with the findings of other investigators—namely, that chemotherapy exerts adverse effects on bone turnover [3, 4, 6]. Thus, we expected decreases in the digitally measured radiographic area of the bone generated in the compression segment. This was confirmed in our chemotherapy-treated patients compared with those who did not receive chemotherapy.

With our patient numbers, we failed to identify any clinical factors that were associated with the amount of hypertrophy.

Chemotherapy was not associated with a reduction in prosthesis survival (chemotherapy 85%, no chemotherapy 88% at 10 years). The explanation for this unexpected finding is obscure. Overall, failures occurred most frequently during the first postoperative year, but among patients treated with chemotherapy, the failures were more broadly distributed over the first 3 postoperative years. This finding is hypothesis-generating for future studies, suggesting possible rehabilitation and surveillance strategies that should be investigated. Other factors must be at least as important; for example, the quality of bone formation may be more important than the quantity or presumed osteointegration at the interface was not functionally affected by chemotherapy. Although osteointegration could not be proven in this study, the durability of clinical success for this form of fixation is encouraging and is compatible with the hypothesis that compression promotes osteointegration. Several studies have defined chemotherapy as an independent variable in statistical analyses of the occurrence of aseptic failure [8, 14, 16, 21]. Studies that failed to identify an association among chemotherapy, bone formation, and prosthetic survival used significantly different patient populations and scientific methodology. For example, Kagan et al. [16] included patients with different reconstructive sites (proximal femur, distal femur, tibia), a high percentage of reconstructions for benign disease, and failed conventional arthroplasty. Although ostensibly examining a larger patient cohort, they actually included 47 proximal femur reconstructions so there are fewer patients with reconstructions at the relevant location. One study that commented on the effect of chemotherapy has shorter followup [21]. Bone healing and osteointegration have been shown to be reduced substantially during chemotherapy, consistent with our findings of retarded bone formation [24, 26].

MSTS scores were slightly lower in patients treated with chemotherapy as expected, but were not correlated with new bone formation. The significance of this observation is uncertain, but probably related to differences in the patient populations such as age, diagnosis, and extent of the resected muscle between the two groups. The difference in scores was fairly minimal and possibly not clinically important or it could be that this particular functional tool is not sufficiently sensitive to detect functional or quality-of-life differences between the two groups. We believe the score differences are attributable to these issues rather than anything related to new bone formation. Higher MSTS scores could correlate with more weightbearing activity that could stimulate bone formation. However, the differences we saw in MSTS scores were small and unlikely to make a clinical difference in the bone hypertrophy found in the compressed segment.

In conclusion, compliant compression fixation is associated with less bone formation in the first year during chemotherapy administration, but the use of chemotherapy alone does not appear to be a contraindication to use of compliant compressive fixation. We also believe that this type of fixation has the advantage of preserving bone stock for future revisions despite the observation that there may be less bone preserved in patients receiving chemotherapy. This form of fixation is also particularly well suited for revision situations if there is sufficient residual bone stock. Emerging systemic chemotherapy and immunotherapy regimens may also affect issues like bone formation and prosthetic survivorship. Such orthopaedic concerns should be an investigative priority in cooperative group trials because sufficient patient numbers will be needed to address the effect of these therapies on bone formation. The relationship between early bone formation and the timing of weightbearing rehabilitation should also be evaluated in a multicenter study. These followup studies should include patient-reported outcomes using instruments more sensitive than the MSTS score. We suggest that patients with metastases, poor oncologic prognosis, or poor local bone quality should have other forms of cemented or uncemented fixation of their implants [11].