Increased Number of Medical Comorbidities Associated With Increased Risk of Presenting With Pathological Femur Fracture in Metastatic Bone Disease

Abstract

Background

Many cancers metastasize to bone and may lead to pathologic fracture or impending pathologic fracture. Prophylactically stabilizing bones before fracture has been shown to be more cost-effective with improved outcomes. Many studies have examined risk factors for pathological fracture, with radiographic and functional pain data serving as predominant indicators for surgery. Conditions associated with poor bone health and increased risk of fracture in the non-oncologic population, including diabetes mellitus, chronic obstructive pulmonary disease (COPD), cardiovascular disease, renal disease, smoking, corticosteroid use, and osteoporosis, have not been studied in the context of metastatic disease. Characterization of these factors could help providers identify candidates for prophylactic stabilization thereby reducing the number of completed pathological fractures.

Methods

298 patients over the age of 40 with metastatic bone disease of the femur treated between 2010-2021 were retrospectively identified. Patients without complete medical documentation or with non-metastatic diagnoses were excluded. 186 patients met inclusion and exclusion criteria, including 74 patients who presented with pathological femur fracture and 112 patients who presented for prophylactic stabilization. Patient demographics and comorbidities including diabetes mellitus, COPD, cardiovascular disease, renal disease, osteoporosis, active tobacco or corticosteroid use, and use of anti-resorptive therapy were collected. Descriptive statistics were compiled, with univariable analysis by Mann-Whitney or chisquared testing. Multiple logistic regression was then performed to identify the most significant patient variables for presenting with completed fracture.

Results

On univariable analysis, patients with COPD were more likely to present with pathologic fracture (19/32 [59%] compared to 55/154 [36%], p = 0.02). A trend emerged for patients with an increasing number of comorbidities (28/55 [51%] for 2+ comorbidities compared to 18/61 [29%] with zero comorbidities, p = 0.06). On multivariable analysis, patients with two or more comorbidities (OR: 2.49; p=0.02) were more likely to present with a femur fracture.

Conclusion

This analysis suggests that those with an increasing number of comorbidities may be at increased risk for pathologic fracture. This study raises the possibility that patient factors and/ or comorbidities alter bone strength and/or pain experiences and may guide orthopaedic oncologists weighing prophylactic stabilization of femur lesions.

Level of Evidence: III

Introduction

Many cancers metastasize to bone and lead to completed pathologic fracture or impending pathologic fracture. Determining whether to prophylactically surgically stabilize cancer patients with metastatic bone lesions or treat them conservatively through radiation, chemotherapy, or lifestyle modification therapy has been extensively studied.¹ Completed pathological fracture is associated with increased mortality in cancer patients, with estimated one year survival rates as low as 22-40% following a hospitalization due to fracture.² Prophylactically stabilizing bones before fracture has been shown to be more cost-effective with improved patient outcomes.^3,4 Given the recognition of improved patient outcomes and cost effectiveness of prophylactic stabilization, the identification of cancer patients who are likely to fracture has become increasingly important.

Many studies have examined risk factors for pathological fracture and identified radiographic findings and patient perception of functional pain as predominant indicators for surgery.⁵ These indicators are not all-encompassing nor are they associated with a strong predictive value. Furthermore, much of the identification and characterization of these risk factors are subjective data points such as the patient’s perceived pain or degree of cortical bone involvement rather than reproducible and objective measures. Given the relative subjectiveness and limited scope of these fracture risk indicators, identification of more objective risk factors could supplement the surgeon’s clinical decision making. Common medical comorbidities that impact bone health have been extensively characterized in the context of the non-oncological population but have not been assessed in cancer patients with metastatic bone disease.

Common conditions that have been associated with poor bone health and increased risk of fracture in the non-oncologic population include gender, diabetes mellitus, chronic obstructive pulmonary disease (COPD), cardiovascular disease, renal disease, smoking, corticosteroid use, and osteoporosis.^6-12 However, these factors have not been studied in the context of metastatic bone disease. Thorough investigation of these risk factors could help providers identify candidates for prophylactic stabilization thereby reducing the number of completed pathological fractures and ultimate cancer related morbidity. Therefore, the objective of this study is to characterize the risk of fracture in cancer patients with identifiable comorbid conditions and disease.

Methods

Study Design

This is a retrospective cohort study using an institutional electronic medical record data source. Given the lack of direct patient interaction and minimal risk, this study was deemed to be exempt by an institutional review board.

Study Population

298 patients over the age of 40 with confirmed metastatic bone disease involving the femur and treated at an urban academic institution between 2010-2021 were retrospectively identified by CPT codes (27187, 27235, 27236, 27244, 27245, 27495, 27506, 27507, 27511). The presence of metastatic disease of the femur was then confirmed. Patients were then assigned one of two cohorts; those who presented with completed pathological fracture (fracture cohort) and those who were prophylactically stabilized (prophylactic cohort). Patients without complete medical documentation or with non-metastatic diagnoses were excluded. Given this study analyzes the risk factors prior to presentation, no minimum postoperative follow-up was required. 186 patients met our inclusion and exclusion criteria, including 74 patients who presented with completed pathological femur fractures and 112 patients who presented for prophylactic stabilization. Decision for prophylactic fixation due to high risk of fracture was judged by the treating surgeon primarily based on presence or absence of functional pain, but also considered factors such as location, size, and internal characteristics of the lesions, as well as patient preference.

Data Collected

Patient factors extracted from the medical record included: age at surgery, gender, date of surgery, cancer histology, and clinical diagnoses of particular interest, including diabetes mellitus, COPD, cardiovascular disease (including diagnoses of either coronary artery disease or peripheral arterial/vascular disease), osteoporosis, active tobacco or corticosteroid use (use within the preceding 3 months to fracture or fixation), and end-stage renal disease (ESRD; CKD Stage IV, GFR< 15ml/min, or use of hemodialysis). All diagnoses that were documented within three months pre-operatively were considered comorbidities and recorded. Use of bone-modifying agents such as Bisphosphonates or RANK-L inhibitor use within three months preoperatively were also recorded. Using anesthesia records from the date of service, patient weight and height were also collected to calculate body mass index (BMI).

Statistical Analysis

Descriptive statistics were compiled and univariable analysis performed by Mann-Whitney or chi-squared testing. A multivariable logistic regression analysis was performed after stepwise backward selection to identify the most significant patient variables for presenting with completed fracture. Statistical analyses were performed with the package DescTools in R (R Foundation; https://www.r-project.org/). Data were visualized in R with packages ggplot2 and forestplot. All statistical testing was two-sided, with a p value less than 0.05 considered significant.

Results

Out of 298 patients, 186 met our study criteria. The mean age was 64.9±10.6 years for the prophylactic fixation, and 65.7±12.0 years for the completed pathological fracture cohort. 94 female patients were included, of which 55.3% were prophylactically stabilized and 44.7% presented with a fracture. 92 male patients were included of which 65.2% were prophylactically stabilized, with 34.8% completing a pathological fracture. The most common histologies include multiple myeloma (40 patients), breast carcinoma (33), lung carcinoma (33), and renal cell carcinoma (25). Descriptive statistics on comorbidity and medication use between groups is shown in Table 1. Patient cancer histology is shown in Table 2.

Table 1.

Preoperative Characteristics and Comorbidities in Patients Presenting for Prophylactic Stabilization of Impending Pathologic Fractures, Compared to Those Presenting with Completed Pathologic Fracture

	Prophylactic Stabilization (N=116)	Completed Fracture (N=77)	p value			Prophylactic Stabilization (N=116)	Completed Fracture (N=77)	p value
Age (years)					End stage renal disease (ESRD)
Mean (SD)	64.9 (10.7)	65.7 (12.0)	0.566		No	110 (98.2%)	70 (94.6%)	0.217
Sex					Yes	2 (1.8%)	4 (5.4%)
Male	60 (53.6%)	32 (43.2%)	0.219		Osteoporosis
Female	52 (46.4%)	42 (56.8%)			No	110 (98.2%)	71 (95.9%)	0.388
Body mass index (BMI)					Yes	2 (1.8%)	3 (4.1%)
Mean (SD)	30.3 (15.7)	30.3 (8.99)	0.468		Steroid use
Chronic obstructive pulmonary disease (COPD)					No	90 (80.4%)	56 (75.7%)	0.563
No	99 (88.4%)	55 (74.3%)	0.0220		Yes	22 (19.6%)	18 (24.3%)
Yes	13 (11.6%)	19 (25.7%)			Number of comorbidities
Smoker					0	43 (38.4%)	18 (24.3%)	0.062
No	87 (77.7%)	60 (81.1%)	0.708		1	42 (37.5%)	28 (37.8%)
Yes	25 (22.3%)	14 (18.9%)			2+	27 (24.1%)	28 (37.8%)
Cardiovascular disease					Bisphosphonate use
No	92 (82.1%)	55 (74.3%)	0.272		No	104 (92.9%)	73 (98.6%)	0.0889
Yes	20 (17.9%)	19 (25.7%)			Yes	8 (7.1%)	1 (1.4%)
Diabetes					Denosumab use
No	88 (78.6%)	52 (70.3%)	0.266		No	111 (99.1%)	74 (100%)	1.00
Yes	24 (21.4%)	22 (29.7%)			Yes	1 (0.9%)	0 (0%)

p-value less than 0.05 is significant.

Table 2.

Biopsy Confirmed Metastatic Disease Cancer Histology of Patients who Presented with Completed Pathological Fracture vs. Prophylactic Fixation

Histology	Completed Fracture	Prophylactic Fixation	Total
B-Cell Lymphoma	5(6.7%)	3(2.7%)	8(4.3%)
Breast Carcinoma	18(24.3%)	15(13.4%)	33(17.7%)
Breast Spindle Cell Sarcoma	1(1.4%)	1(0.9%)	2(1.1%)
Breast Phyllodes Fibrosarcoma	0(0%)	1(0.9%)	1(0.5%)
Colonic Adenocarcinoma	0(0%)	3(2.7%)	2(1.1%)
Esophageal Carcinoma	2(2.7%)	2(1.8%)	4(2.2%)
Gastric Carcinoma	0(0%)	1(0.9%)	1(0.5%)
Lung Carcinoma	10(13.4%)	23(20.4%)	33(17.7%)
Melanoma	3(4.0%)	5(4.4%)	8 (4.3%)
Multiple Myeloma	15(20.2%)	25(22.3%)	40(21.5%)
Pancreatic Adenocarcinoma	0(0%)	2(1.8%)	2(1.1%)
Pheochromocytoma	0(0%)	1(0.9%)	1(0.5%)
Neuroendocrine- Poorly Differentiated	1(1.4%)	1(0.9%)	2(1.1%)
Carcinoma- Poorly Differentiated	1(1.4%)	3(2.7%)	3(1.6%)
Prostate Adenocarcinoma	5(6.8%)	4(3.5%)	8 (4.3%)
Renal Cell Carcinoma	11(14.9%)	14(12.5%)	25(13.4%)
Thyroid Carcinoma	1(1.4%)	4(3.5%)	4(2.2%)
Unknown Histology	0(0%)	1(0.9%)	1(0.5%)
Urothelial Carcinoma	1(1.4%)	4(3.5%)	5(2.7%)
TOTAL	74	112	186

Percentages rounded to nearest tenth.

On univariable analysis, patients with COPD were more likely to present with pathologic fracture (19/32 [59%] compared to 55/154 [36%], p = 0.02). There was also a trend for patients with an increasing number of comorbidities overall to present with pathological fracture 28/55 [51%] for 2+ comorbidities compared to 18/61 [29%] with 0 comorbidities, (p=0.06). On multivariable logistic regression analysis (Figure 1), patients with female sex (odds ratio [OR]: 1.65; 95% confidence interval [CI]: 0.9-3.06; p=0.106) trended towards increased fracture risk while those patients with or two or more comorbidities (OR: 2.49; 95% CI: 1.16-5.50; p=0.02) were more likely to present with a completed femur fracture. Bisphosphonate use trended towards a lower likelihood of presenting with a pathological fracture but did not reach statistical significance (OR: 0.19; 95% CI: 0.01-1.09; p=0.122).

Figure 1.

Logistic regression analysis after stepwise backward selection of factors associated with fracture on presentation. Unstandardized odds ratios are displayed on a logarithmic scale with 95% confidence intervals. CI indicates confidence interval. p-value less than 0.05 is significant.

Discussion

The goal of this study was to assess comorbidities commonly associated fracture in cancer patients with completed or impending pathologic fracture. Comorbidities associated with increased fracture risk in the non-oncologic population were studied. Female patients trended towards increased fracture risk while patients with at least two comorbidities were associated with increased fracture risk at presentation in this study.

The presence of metastatic bone disease is a negative prognostic marker for cancer patients, with the completion of a pathological fracture also being associated with worse survival.^13-17 Prophylactic fixation may prevent a patient from completing a pathological fracture, therapy reducing morbidity and costs. Still, it is essential to identify the patients that are most at risk of fracturing in order to avoid unnecessary surgical interventions in low fracture risk patients.^18,19 Identification of prophylactic fixation candidates has traditionally relied on radiographic indices and subjective patient pain.²⁰ Mirel’s risk stratification system, first developed in 1980’s, relies upon cortical bone involvement, lesion location, and perceived pain by the patient. These criteria, however, have been shown to exhibit low sensitivity and specificity in predicting pathologic fracture occurrence.²¹ Furthermore, these criteria fail to incorporate more comprehensive medical data, such as medical co-morbidities that are commonly associated with increased fracture risks and worsening bone health.

Our study identified common medical conditions that have previously been implicated in poor bone health. We found that COPD was the only medical comorbidity independently associated with fracture. Kulak et. al.²² previously suggested this could be due in part to a decrease in trabecular bone density resulting from decreased collagen cross-linking that is observed in females with COPD. Although corticosteroid use was not noted to be a significant risk factor for fracture, it is important to note that use was only assessed within the preceding 3 months prior to fixation or fracture. Patients with COPD are more likely to undergo repeated oral steroid courses in addition to potential long term inhaled corticosteroid usage. This may suggest that the increased fracture risk seen in the COPD cohort may be compounded by the prevalence of corticosteroid treatments. Kulak et. al also implicated diabetes as a contributor to decreased collagen cross-linking in female patients at higher risk for bone fragility. Notably, in our study, diabetes was not associated with completed fracture by itself with the numbers available, though diabetes was more prevalent in the fractured group. A multivariable logistic regression analysis showed that increasing number of comorbidities were associated with increased risk of presenting with a completed pathologic fracture of the femur. This suggests that the presence of multiple conditions known to alter bone quality may be a risk factor for completing a pathological fracture. Additionally, the incidence of completed fracture in the female cohort approached significance. Interestingly, breast carcinoma was the only histological cohort with more than 10 patients that exhibited more fractures than prophylactic stabilizations (54.5% vs. 45.4%). Given the association between females who have undergone menopause and increased fracture risk, further characterization of the role of hormonal therapies and fracture incidence may yield further insight into this observed trend.

Several trends were noted with specific comorbidities and antiresorptive therapies, but this analysis was likely underpowered to evaluate this comprehensively. Although not statistically significant, antiresorptive therapies likely would show decreased fracture rates in a higher-powered sample. This study did not account for socioeconomic differences that may predispose to both comorbidities and access to care before fracture. To aid in the feasibility and economy of this study, the metastatic disease characterized was limited to the femur. However, the axial skeleton and upper extremity are also common sites of metastatic disease and fracture. These anatomical sites necessitate further study to improve the generalizability of these findings. Future studies are needed, using larger patient cohorts and additional sites of metastatic bone disease to adequately address the question raised by and determine the potential role of estrogen deprivation in contributing to fracture risk in cancer patients.

Despite these limitations, this study raises the possibility that patient factors and/or comorbidities alter bone strength and/or pain experiences that may mislead orthopaedic oncologists weighing prophylactic stabilization of femur lesions in the setting of metastatic bone disease. The risk factors we assessed such as diabetes, tobacco use, corticosteroid use, and ESRD,^23-27 are associated with increased fracture risk in the non-oncologic population. However, none of these comorbidities were significant sole risk factors for fracture in our sample. This suggests that the presence of 2 or more of these comorbid conditions may increase the risk of fracture in metastatic bone disease of the femur. Mirel’s criteria and similar risk strata such as Harrington’s criteria do not account for the impact of non-oncologic factors on bone health. Together, the findings of this study may supplement existing prophylactic stabilization criteria to guide decision making in metastatic bone disease of the femur. Future studies examining the risk of pathologic fracture should include such patient factors and/ or comorbities in their analyses.