Abstract
Context:
Femoral shaft cortical thickening has been mentioned in reports of atypical subtrochanteric and diaphyseal (S/D) femur fractures, but it is unclear whether thickening precedes fracture or results from a preceding stress fracture and what role bisphosphonates might play in cortical thickening.
Objective:
Our objective was to examine the relationship of cortical thickness to S/D fracture risk as well as establish normal reference values for femoral cortical thickness in a large population-based cohort of older women.
Design:
Using pelvic radiographs obtained in 1986–1988, we measured femoral shaft cortical thickness 3 cm below the lesser trochanter in women in the Study of Osteoporotic Fractures. We measured this in a random sample and in those with S/D fractures and femoral neck and intertrochanteric fractures. Low-energy S/D fractures were identified from review of radiographic reports obtained between 1986 and 2010. Radiographs to evaluate atypia were not available. Analysis used case-cohort, proportional hazards models.
Outcomes:
Cortical thickness as a risk factor for low-energy S/D femur fractures as well as femoral neck and intertrochanteric fractures in the Study of Osteoporotic Fractures, adjusting for age and bone mineral density in proportional hazards models.
Results:
After age adjustment, women with thinner medial cortices were at a higher risk of S/D femur fracture, with a relative hazard of 3.94 (95% confidence interval = 1.23–12.6) in the lowest vs. highest quartile. Similar hazard ratios were seen for femoral neck and intertrochanteric fractures. Medial or total cortical thickness was more strongly related to fracture risk than lateral cortical thickness.
Conclusions:
In primarily bisphosphonate-naive women, we found no evidence that thick femoral cortices placed women at higher risk for low-energy S/D femur fractures; in fact, the opposite was true. Women with thin cortices were also at a higher risk for femoral neck and intertrochanteric fractures. Whether cortical thickness among bisphosphonate users plays a role in atypical S/D fractures remains to be determined.
Several case reports and case series have described the occurrence of low-energy subtrochanteric and diaphyseal (S/D) femoral fractures with specific radiographic features, which have been termed atypical fractures (1–7). Many of these cases have been described in patients taking bisphosphonates, particularly in those on long-term treatment, suggesting a possible association of these fractures to bisphosphonate use. One of the first papers to describe patients with atypical fractures noted a specific radiographic pattern characterized by beaking of the cortex on one side and focal cortical thickening around the site of fracture (8), whereas others have suggested more general cortical thickening (1, 9, 10). However, the reporting of thickened cortices in the radiographs has been incomplete and inconsistent between the various case reports and case series.
Subsequently, S/D fractures have been studied in a number of observational and randomized trials with focus on studying the association between these fractures and bisphosphonate use (11–16). However, none of these observational or randomized studies have measured or reported cortical thickness.
Recently, both the American Society of Bone and Mineral Research (ASBMR) and the International Osteoporosis Foundation (1, 17) have recommended additional studies to establish normal reference values for cortical thickness and to examine the relationship of cortical thickness to hip fractures of various types.
To address these questions as well as establish normal reference values for femoral cortical thickness in a population of older Caucasian women, we measured cortical thickness from pelvic radiographs obtained in 1986–1988 in the Study of Osteoporotic Fractures (SOF), an ongoing longitudinal National Institutes of Health-funded observational study of 9704 women over age 65 begun in 1986 to examine risk factors for hip and other fractures in older community-dwelling women. Furthermore, we were able to examine the relationship of these baseline cortical thickness values to subsequent development of S/D and other types of hip fractures.
Subjects and Methods
Subjects
The design of SOF has been described previously. Briefly, it is a multicenter, prospective cohort study of risk factors for fracture in 9704 community-dwelling ambulatory white women aged 65 yr or older, who were enrolled between September 1986 and October 1988 (18). Women were recruited from population-based listings in four regions of the United States: Baltimore County, MD; Minneapolis, MN; Portland, OR; and the Monongahela Valley near Pittsburgh, PA. African-Americans were initially excluded because of their lower rate of hip fracture. Other exclusion criteria included history of bilateral hip replacement and inability to walk without assistance. All women provided written informed consent. Approval was obtained from the Institutional Review Boards at each site. Funding for the study has been provided by the National Institutes of Health (primarily National Institute of Arthritis and Musculoskeletal and Skin Diseases and National Institute on Aging).
Baseline measurements
At baseline, all participants completed a questionnaire and interview and underwent a physical examination. In addition, baseline anterior-posterior (AP) pelvic radiographs were obtained on all women in the hopes that information available from these radiographs, such as bone size or hip geometry, might be predictive of fracture. Other potential fracture risk variables collected included age, height, weight, history of fracture, maternal history of hip fracture, history of falls, smoking status, and estrogen use. A more comprehensive listing of all variables is provided on the SOF Online website (http://sof.ucsf.edu). The period when SOF baseline visits were performed was about 8 years before the introduction of bisphosphonates for osteoporosis in the United States (1995).
Follow-up measurements
Participants were invited to participate in as many as seven follow-up clinic visits where weight, history of falls, history of fracture, smoking status, and medications were collected. All current prescription and over-the-counter medications that had been used in the last 30 d were entered into an electronic medication inventory system.
Dual-energy x-ray absorptiometry measurement of the hip was first performed between November 1988 and December 1990 (Hologic 1000; Hologic, Bedford, MA), about 2 yr after the baseline visit. Details of the measurement method and densitometry quality control procedures have been published elsewhere (18). The right proximal femur was scanned except in the case of right hip replacement or severe degenerative change. Dual-energy x-ray absorptiometry hip measurements were repeated at SOF visit 4 (in 1992–1994) and all subsequent visits for a median of two measurements per participant and a maximum of five measurements.
Hip fracture reporting and adjudication
During the SOF study
Participants were queried about fracture at each visit. In addition, they were contacted by postcard every 4 months, with telephone follow-up for nonresponders, to ask about any fractures. Follow-up was more than 95% complete. Copies of radiological reports were obtained for all nonvertebral fractures. Study physicians at the Coordinating Center adjudicated incident nonvertebral fractures from radiology reports or other materials if available. All hip and femur fractures were initially classified into one of six categories by anatomical region: femoral neck, intertrochanteric region, femur, hip, acetabular, or other (hip) region. There was originally no specific category for S/D femur fractures.
Hip fracture re-review in 2009–2010
All x-ray reports of hip and femur fractures that occurred in the intertrochanteric region or below, excluding femoral neck fracture, were initially re-reviewed by one of two orthopedic surgery residents (M.K. and R.W.) to examine whether any met the criteria for subtrochanteric or diaphyseal femoral fracture. Available for review were emergency department notes, radiology reports, and in some instances, orthopedic operative and clinic reports. Radiographs were not available for review. Subtrochanteric fractures were defined as those with fracture lines originating from the lesser trochanter, extending 5 cm distal to the lesser trochanter. Diaphyseal femur fractures were defined as those occurring 5 cm distal to the lesser trochanter to the distal metaphyseal flare. Pathological fractures, periprosthetic fractures, and fractures occurring in the context of major trauma were excluded. After the individual re-review had been completed, both surgeons again reviewed the specific fractures that had been reclassified as S/D and reached consensus on whether each fracture truly met the criteria for S/D fracture (Supplemental Fig. 1, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org).
Cortical thickness measurement
AP pelvic radiographs obtained on all participants at baseline were digitized in 2008–2009 and used to measure femoral cortical thickness at 3 cm below the lesser trochanter (Supplemental Fig. 2). Using Image J software, a script was written to collect coordinates from the digital images. Two points were marked on each lesser trochanter: a point at the top and a point at the bottom. Next, the outlines of the inner and outer margin of the four cortices (medial and lateral, left and right) were traced from the lesser trochanter to the bottom of the image. The software then calculated the imaginary midpoint for each lesser trochanter and drew a line between the two points, which was defined as horizontal. Using coordinates output by the software, bilateral medial and lateral cortical width were calculated at a level 3 cm vertically below (and perpendicular from) that line. We collected and averaged the width within 1.7 mm of the 3-cm mark. We also measured total femoral bone width at the same level. Some radiographs did not extend down far enough, and values for those cases were considered as missing.
All measurements were performed by a single, trained operator. To assess reproducibility of the measurements, a sample of 35 radiographs were randomly selected and resubmitted (blindly) for remeasurement. The correlation (Pearson) of the results from the two assessments was calculated as a measure of agreement. Test-retest reliability was excellent, with correlations between the first and second measurements ranging from 0.88–0.91.
Case cohort and fracture cohort selection
A random sample of 400 women in SOF who had baseline AP pelvis radiographs available was selected as a study population control group for normative cortical thickness measurements, of whom 378 had available data (Fig. 1). For analysis of the relationship of cortical width to hip fracture risk, three additional sets of fracture cases were selected. All women who sustained low-energy S/D fractures (n = 45) for whom we had cortical thickness measurements (n = 33) were included as one set of cases. A random sample of 100 women with available baseline AP pelvis radiographs who sustained intertrochanteric fractures was included as the intertrochanteric fracture case set. The same selection was performed for the femoral neck fracture cases. After exclusion of those for whom cortical thickness measurements could not be obtained from the radiographs, a total of 91 intertrochanteric and 95 femoral neck fractures were analyzed.
Fig. 1.
Flow chart of the design of the study. The randomly sampled cohort was used to define baseline epidemiological data on cortical thickness. The femoral neck, intertrochanteric, and S/D fracture cases were used as part of the case-cohort analysis to assess associations of cortical thickness with each fracture type. Note that because the randomly sampled cohort includes women with hip fractures, these fractures were also included in the case-cohort analysis, including one subtrochanteric (subtroch) fracture, 28 femoral neck fractures, and 33 intertrochanteric (intertroch) fractures.
Data analysis
The relationship of cortical thickness to risk of the three types of hip fracture was analyzed using proportional hazard models with modifications to account for the case-cohort nature of the study (19). The primary measures of cortical thickness included in the analysis were medial cortical thickness, lateral cortical thickness, total cortical thickness (medial plus lateral thickness), and the ratio of total cortical thickness to total femoral bone width. All were averaged across the left and right sides. As an exploratory analysis, we also measured the medullary width (mean total width minus cortical thickness).
Models were run with and without age adjustment. In addition, models were run in which age and total hip bone mineral density (BMD) (obtained 2 yr after the SOF baseline visit) were included as covariates.
Results
Cortical thickness within the random cohort of 378 women from the overall SOF population was normally distributed. The mean ± sd medial cortical thickness was 0.72 ± 0.14 cm, the lateral cortical thickness was 0.65 ± 0.11 cm, and the total cortical thickness was 1.36 ± 0.22 cm. The ratio of total cortical thickness to bone width was 0.37 ± 0.06.
Cortical thickness declined strongly with increasing age (Table 1). For example, for medial cortical thickness, the mean value was 0.75 ± 0.13 cm in the 65- to 69-yr age group compared with 0.68 ± 0.13 cm in the 80-yr and older age group. The decline was of similar magnitude for total cortical thickness and the ratio of total cortical thickness to total bone width. Lateral cortical thickness was less strongly age dependent. Percentiles for the four cortical thickness parameters are shown by age in Fig. 2. Total femoral bone width did not change significantly with age. Medullary diameter showed a marginally significant relationship to age (data not shown).
Table 1.
Femoral shaft cortical thickness variables by age
| Measurement (mm) | Cohort (n = 378) | Variables by age |
P value | |||
|---|---|---|---|---|---|---|
| Age 65–69 (n = 142) | Age 70–74 (n = 123) | Age 75–79 (n = 68) | Age 80+ (n = 45) | |||
| Medial cortical thickness | 0.72 ± 0.14 | 0.75 ± 0.13 | 0.72 ± 0.13 | 0.70 ± 0.16 | 0.68 ± 0.13 | 0.0002 |
| Lateral cortical thickness | 0.65 ± 0.11 | 0.65 ± 0.11 | 0.66 ± 0.11 | 0.63 ± 0.10 | 0.62 ± 0.11 | 0.1835 |
| Total cortical thickness | 1.36 ± 0.22 | 1.39 ± 0.21 | 1.37 ± 0.22 | 1.32 ± 0.23 | 1.29 ± 0.23 | 0.0029 |
| Total cortical thickness/bone width | 0.37 ± 0.06 | 0.38 ± 0.06 | 0.38 ± 0.06 | 0.36 ± 0.06 | 0.35 ± 0.06 | 0.0011 |
| Femoral shaft bone width | 3.67 ± 0.29 | 3.67 ± 0.30 | 3.65 ± 0.28 | 3.69 ± 0.28 | 3.67 ± 0.29 | 0.8013 |
P value is from an ANOVA test or from linear regression. All values reported as mean ± sd.
Fig. 2.
Age-adjusted normal percentiles of cortical thickness (CT).
The ratio of medial to lateral cortical thickness (data not shown) declined from 1.14 ± 0.18 in the 65- to 69-yr age group to 1.05 in the 80-yr and older age group, indicating a relative loss of medial cortical bone with age.
Cortical thickness and fracture risk
Selected baseline characteristics for the random cohort and sampled fracture cases for the three types of hip fractures are shown in Table 2. S/D cases had an average age of 75.1 yr compared with 72.1 yr for the cohort sample. All three groups of fracture cases had a lower BMD than the random cohort. Mean total hip BMD in the S/D fracture cases (T-score = −1.82 ± 0.96) was lower than the random cohort (−1.42 ± 1.04) but somewhat higher than intertrochanteric (−2.05 ± 0.89) and femoral neck (−2.20 ± 0.81) fracture cases, despite their older age.
Table 2.
Baseline characteristics of sample cohort and fracture cases
| Characteristic | Sample cohort (n = 378) | Type of fracture |
||
|---|---|---|---|---|
| S/D (n = 33) | Intertrochanter (n = 91) | Femoral neck (n = 95) | ||
| Age (yr) | ||||
| Mean ± sd | 72.08 ± 5.27 | 75.06 ± 6.95 | 73.63 ± 5.38 | 72.54 ± 4.72 |
| [n (%)] | ||||
| 65–69 | 142 (37.57) | 8 (24.24) | 18 (19.78) | 27 (28.42) |
| 70–74 | 123 (32.54) | 10 (30.3) | 38 (41.76) | 41 (43.16) |
| 75–79 | 68 (17.99) | 8 (24.24) | 24 (26.37) | 17 (17.89) |
| ≥80 | 45 (11.9) | 7 (21.21) | 11 (12.09) | 10 (10.53) |
| Height (cm), mean ± sd | 158.6 ± 6.16 | 157.0 ± 6.50 | 157.7 ± 6.72 | 158.9 ± 6.55 |
| BMI (kg/m2) | ||||
| Mean ±sd | 26.7 ± 4.75 | 27.48 ± 3.99 | 26.19 ± 4.96 | 24.99 ± 4.11 |
| [n (%)] | ||||
| <25 [n (%)] | 159 (42.06) | 8 (24.24) | 45 (49.45) | 52 (54.74) |
| 25–30 | 138 (36.51) | 16 (48.48) | 26 (28.57) | 29 (30.53) |
| >30 | 81 (21.43) | 9 (27.27) | 20 (21.98) | 14 (14.74) |
| Walks for exercise, yes [n (%)] | 189 (50) | 12 (36.36) | 47 (51.65) | 57 (60) |
| History of fracture after age 50, yes [n (%)] | 142 (37.77) | 16 (48.48) | 46 (50.55) | 49 (52.13) |
| Estrogen use [n (%)] | ||||
| Never | 227 (60.7) | 19 (57.58) | 56 (62.92) | 55 (57.89) |
| Past | 110 (29.41) | 5 (15.15) | 20 (22.47) | 26 (27.37) |
| Current | 37 (9.89) | 9 (27.27) | 13 (14.61) | 14 (14.74) |
| Total hip BMD (T-score) | ||||
| Mean ± sd | −1.42 ± 1.04 | −1.82 ± 0.96 | −2.05 ± 0.89 | −2.20 ± 0.81 |
| [n (%)] | ||||
| <−4 | 2 (0.63) | 0 (0) | 2 (2.50) | 1 (1.32) |
| <−3.5 to −4 | 7 (2.22) | 0 (0) | 3 (3.75) | 3 (3.95) |
| <−3 to −3.5 | 12 (3.81) | 2 (7.69) | 6 (7.50) | 10 (13.16) |
| <−2.5 to −3 | 25 (7.94) | 2 (7.69) | 11 (13.75) | 11 (14.47) |
| >−2.5 | 269 (85.40) | 22 (84.62) | 58 (72.50) | 51 (67.11) |
| Medial CT (cm), mean ± sd | 0.72 ± 0.14 | 0.67 ± 0.14 | 0.68 ± 0.13 | 0.66 ± 0.12 |
| Lateral CT (cm), mean ± sd | 0.65 ± 0.11 | 0.63 ± 0.12 | 0.62 ± 0.10 | 0.61 ± 0.11 |
| Total CT (cm), mean ± sd | 1.36 ± 0.22 | 1.29 ± 0.23 | 1.29 ± 0.21 | 1.26 ± 0.20 |
| Ratio total CT to bone width, mean ± sd | 0.37 ± 0.06 | 0.35 ± 0.05 | 0.36 ± 0.05 | 0.35 ± 0.06 |
The sample cohort is a random sample of the entire SOF population and includes participants with fractures. CT, Cortical thickness.
There was a strong relationship of cortical thickness to risk of all three categories of hip fractures, such that women with thinner cortices were at higher risk than women with thicker cortices (Table 3 and Fig. 3). For example, for femoral neck fractures, the risk in the lowest quartile of medial thickness was 3.49 times that in the highest (P < 0.001). The age-adjusted relative risk per sd increase was 0.65 (P < 0.001). The relationship of cortical thickness to intertrochanteric fractures was somewhat stronger than to femoral neck fractures (e.g. for medial cortical thickness, the lowest quartile was at 6.6 times the risk of the highest for intertrochanteric fractures vs. 3.49 for femoral neck fractures). The relationships of cortical thickness to S/D fractures were similar to those for femoral neck fractures, although the absolute fracture numbers are much smaller and therefore confidence intervals (CI) wider. The relationship of the ratio of total cortical thickness to total bone width with S/D fractures was statistically significant, and the relationship of medial thickness was marginally significant. Medullary width showed a similar relationship to fracture risk as the ratio of cortical to total bone width (data not shown). There is a suggestion that the relationship of lateral cortical thickness to S/D fractures is weaker, but numbers are too small to draw meaningful conclusions.
Table 3.
Age adjusted hazard ratio of cortical thickness (quartiles and as a continuous variable) to fracture risk for three types of fractures using case-cohort analysis
| Fracture type | Measures of CT |
|||
|---|---|---|---|---|
| Medial CT (mm) | Lateral CT (mm) | Total CT (mm) | Ratio total CT to bone width (mm) | |
| S/D | ||||
| Q1 (low) | 3.94 (1.23–12.62) | 1.81 (0.64–5.12) | 4.55 (1.23–16.81) | 3.25 (0.82–12.82) |
| Q2 | 1.29 (0.35–4.74) | 1.28 (0.43–3.88) | 3.06 (0.78–12.01) | 5.86 (1.58–21.77) |
| Q3 | 2.15 (0.62–7.49) | 1.31 (0.42–4.05) | 2.84 (0.72–11.23) | 1.31 (0.29–5.97) |
| Q4 (high) | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) |
| RR/sd | 0.67 (0.44–1.01) | 0.81 (0.54–1.21) | 0.71 (0.48–1.03) | 0.70 (0.50–0.98) |
| Femoral neck | ||||
| Q1 (low) | 3.49 (1.66–7.31) | 2.77 (1.37–5.58) | 2.93 (1.45–5.92) | 3.37 (1.68–6.73) |
| Q2 | 3.32 (1.62–6.79) | 2.33 (1.13–4.81) | 2.93 (1.46–5.90) | 2.23 (1.08–4.59) |
| Q3 | 1.79 (0.82–3.88) | 2.31 (1.13–4.70) | 1.58 (0.75–3.34) | 1.39 (0.66–2.93) |
| Q4 (high) | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) |
| RR/sd | 0.65 (0.50–0.85) | 0.67 (0.53–0.85) | 0.66 (0.53–0.83) | 0.66 (0.53–0.82) |
| Intertrochanteric | ||||
| Q1 (low) | 6.63 (2.77–15.87) | 2.85 (1.45–5.59) | 4.10 (1.96–8.58) | 4.49 (2.21–9.11) |
| Q2 | 4.98 (2.10–11.82) | 1.66 (0.82–3.36) | 2.37 (1.10–5.15) | 2.24 (1.07–4.70) |
| Q3 | 2.92 (1.18–7.21) | 1.66 (0.81–3.41) | 2.36 (1.11–5.03) | 1.40 (0.64–3.09) |
| Q4 (high) | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) | 1 (Ref.) |
| RR/sd | 0.57 (0.44–0.74) | 0.60 (0.46–0.78) | 0.59 (0.47–0.75) | 0.56 (0.44–0.72) |
Bold type indicates significant difference. CT, Cortical thickness; Q, quartile; Ref., reference; RR, relative risk.
Fig. 3.
Relative fracture risk is shown as stratified by quartile of cortical thickness (CT), where quartile 1 (Q1) is the lowest quartile and Q4 is the highest quartile. All hazard ratios are calculated relative to Q4. *, P < 0.05.
There was a high correlation between the measures of cortical thickness and areal BMD (measured about 2 yr after the radiographs were obtained). For example, the correlation of medial cortical thickness to total hip BMD was 0.56. The relationship of cortical thickness to risk of hip fracture was no longer statistically significant after adjustment for BMD. For example, for medial cortical thickness, the age-adjusted relative risk per sd values for femoral neck, intertrochanteric, and S/D fractures were 0.65 (95% CI = 0.50–0.85), 0.57 (95% CI = 0.44–0.74), and 0.67 (95% CI = 0.44–1.01), respectively, but with added adjustment for total hip BMD were attenuated to 1.09 (95% CI = 0.77–1.54), 1.07 (95% CI = 0.77–1.48), and 1.09 (95% CI = 0.71–1.68), respectively.
Discussion
We examined the relationship between cortical thickness and subsequent risk of hip fracture, particularly S/D fractures. Some case reports of atypical femur fractures suggest that thicker cortices in the femoral shaft would be associated with higher risk of S/D fractures. However, we found the opposite: women with thicker cortices, particularly the medial femoral cortex, were at lower risk of S/D fractures as well as femoral neck and intertrochanteric fractures. These results would be expected biomechanically because cortical bone contributes greatly to bone density measurements, which have been shown to be strongly predictive of future hip fracture risk in our study (20) and others (21). In addition, direct studies of the relationship of cortical thickness to hip fracture risk using various techniques for assessing cortical thickness have shown that thicker cortices predict lower risk of hip fracture (22). However, these studies did not examine the role of cortical thickness in predicting different types of hip fractures, and no studies to date have specifically examined the relationship between cortical thickness and low-energy S/D femur fracture.
Descriptions of cortical thickness in the literature surrounding atypical femur fractures have been inconsistent. Some reports mentioned focal thickening, as one might expect from stress fractures (23), whereas others have suggested a more diffuse pattern (8). Whether or not both medial and lateral sides or just the lateral side is thickened has also been inconsistent. These inconsistent results demonstrate the uncertainties about the relationship between cortical thickness and atypical fractures and support the recent suggestion of a need for additional information about the nature, extent, and bilaterality of cortical thickness in future case reports (1).
Our results showing that women with thicker cortices are at lower, not higher, risk of subtrochanteric fractures are seemingly at odds with those of the case reports that suggest an association of thicker cortices with risk, particularly in patients using bisphosphonates. There are a number of possible reasons for this discrepancy. First, in our study, all of the radiographs were obtained, and most of the fractures occurred, in bisphosphonate-naive patients. Therefore, this study cannot examine the impact of bisphosphonates on cortical thickness or study how bisphosphonates impact S/D fracture risk through alterations in cortical thickness. However, bisphosphonates have not been shown to have any significant effect on periosteal apposition or endosteal formation, and there is no evidence that bisphosphonates significantly alter cortical thickness. Furthermore, there is no known mechanism by which bisphosphonates could greatly increase cortical thickness while decreasing bone strength. Second, there are no published data on normal levels of cortical thickness so that it is possible that the thickening reported in the case reports is not beyond normal limits. Lastly, as suggested by the ASBMR report, cortical thickening may be a consequence of chronic stress response, and therefore, the thickened cortices may be a result of the pathology and not a risk factor for it and thus may be apparent only just before the fracture occurs. Moreover, the ASBMR report (1) and some case series (8) have pointed out the importance of focal thickening in both periosteal and endosteal surfaces of the lateral cortex, corresponding to the line of the fracture. This radiographic feature has been referred as cortical beaking or flaring and may be the result of local periosteal apposition. However, both local and general thickening have been indicated as features of atypia from the ASBMR task force (1) and therefore should be considered in the clinical characterization of patients with S/D fractures.
This study is the first to describe the normal distribution of cortical thickness in the subtrochanteric femoral shaft region in a representative sample of older postmenopausal Caucasian women. We confirmed the results in several other studies at various skeletal sites that cortical thickness decreases with age (24, 25). These epidemiological data may be used as a reference base for future studies to determine the extent to which a specific case deviates from normal values of cortical thickness in the femoral shaft. An important caveat is that SOF enrolled only ambulatory non-Black (primarily Caucasian) women at baseline, and thus, these data may not be generalizable to other ethnic groups or representative of the population as a whole. However, a previous study found that the SOF population is in general similar to Caucasian women over age 65 in a U.S. representative sample (26).
We used pelvic radiographs to measure cortical thickness at the femoral shaft, which has some advantages and disadvantages for studying cortical thickness. Although the pelvic radiographs were obtained in a standardized way, the use of radiographs to accurately measure cortical thickness has not been validated. Because radiographs are two-dimensional projections of a three-dimensional structure, the inner edge of cortical bone cannot be precisely determined. Other imaging modalities, such as quantitative computed tomography, might improve accuracy with three-dimensional images but may not be feasible for a prospective study such as ours. However, inclusion in our analysis of the ratio of cortical thickness to total bone width should allow more general use of our results, particularly in clinical settings where a radiograph may be the only image available.
There are several other limitations of our study. First, we did not have fracture radiographs available for review and therefore could not assess the atypical features of these fractures. Although we cannot determine the proportion of our S/D fractures that would have met criteria for atypia, a recent report (based on fractures in 2008) suggested that approximately 20% of fractures in the S/D region would meet criteria for atypia similar to those proposed by a recent task force from the ASBMR (15). Second, although it is a strength of our study that all of the radiographs were obtained in, and most of the fractures occurred in, bisphosphonate-naive women, allowing us to better understand the relationship of cortical thickness to S/D fractures before treatment, we do not have the ability to draw any conclusions about whether the relationship of cortical thickness to S/D fractures is altered by bisphosphonate treatment. Repeat routine pelvic radiographs collected later in SOF, which are currently being assessed, might yield insights into longitudinal changes with and without bisphosphonates. Third, due to limitations in the field size of the pelvic radiographs, we measured cortical thickness 3 cm below the lesser trochanter. Because the radiographs often did not extend further, it was not feasible for us to measure a more distal level. Although this is consistent with some other studies of femoral shaft cortical thickness from dual-energy x-ray absorptiometry or radiographs (22, 27, 28), most subtrochanteric fractures occur distal to this level. Furthermore, because we could not measure all along the femoral shaft, we had limited ability to assess focal cortical thickness, and thus, our study is more relevant to hypotheses about generalized cortical thickness.
In conclusion, our data in primarily bisphosphonate-naive women do not support the findings from case reports that thicker femoral shaft cortices compromised bone strength in the femoral shaft. In fact, we showed the opposite: women with thicker cortices in the femoral shaft were less likely to fracture over the next 20 yr in the S/D region as well as other regions of the hip. Other studies are needed to more clearly elucidate any effect bisphosphonates may have on the relationship between baseline cortical thickness, cortical stress reaction, and fracture risk.
Supplementary Material
Acknowledgments
We acknowledge Thuy Le and Lucy Wu (University of California, San Francisco) for editorial assistance on the manuscript.
The SOF study is supported by grants from the National Institutes of Health (National Institute of Arthritis and Musculoskeletal and Skin Diseases and National Institute on Aging): AG05407, AR35582, AG05394, AR35584, AR35583, AR46238, AG005407, AG08415, AG027576-22, AG005394-22A1, AG027574-22A1, and AG030474. The adjudication of subtrochanteric fractures and measurement of cortical width was supported by a grant from Merck and Co.
K.P. and D.M.B. accept responsibility for the integrity of the data analysis. D.M.B., J.C., and K.E. were responsible for study design; K.P., R.W., A.C., J.C., K.E., M.K., and D.M.B. for study conduct; D.M.B., R.W., M.K., A.C., J.C., and K.E. for data collection; K.P., D.M.B., and J.J. for data analysis; N.N., J.J., K.P., J.C., and K.E. for data interpretation; and N.N., J.J., K.P., R.W., and D.M.B. for drafting the manuscript and revising contents. All authors were responsible for approving the final version of manuscript.
Disclosure Summary: N.N. received consulting fees from MSD; S.B. received grant support from Eli Lilly.; J.C. received consulting fees from Horizon Steering Committee and Novartis and grant support from Novartis.; and D.M.B. received grant support from Novartis, Merck, and Roche and consulting or advisory board fees from Eli Lilly, Amgen, Zosano, Radius, and Nycomed. J.J., K.P., R.W., A.C., K.E., and M.K. report no conflicts of interest.
Footnotes
- AP
- Anterior-posterior
- BMD
- bone mineral density
- CI
- confidence interval
- S/D
- subtrochanteric and diaphyseal
- SOF
- Study of Osteoporotic Fractures.
References
- 1. Shane E, Burr D, Ebeling PR, Abrahamsen B, Adler RA, Brown TD, Cheung AM, Cosman F, Curtis JR, Dell R, Dempster D, Einhorn TA, Genant HK, Geusens P, Klaushofer K, Koval K, Lane JM, McKiernan F, McKinney R, Ng A, Nieves J, O'Keefe R, Papapoulos S, Sen HT, van der Meulen MC, Weinstein RS, Whyte M, American Society for Bone and Mineral Research 2010. Atypical subtrochanteric and diaphyseal femoral fractures: report of a task force of the American Society for Bone and Mineral Research. J Bone Miner Res 25:2267–2294 [DOI] [PubMed] [Google Scholar]
- 2. Goh SK, Yang KY, Koh JS, Wong MK, Chua SY, Chua DT, Howe TS. 2007. Subtrochanteric insufficiency fractures in patients on alendronate therapy: a caution. J Bone Joint Surg Br 89:349–353 [DOI] [PubMed] [Google Scholar]
- 3. Neviaser AS, Lane JM, Lenart BA, Edobor-Osula F, Lorich DG. 2008. Low-energy femoral shaft fractures associated with alendronate use. J Orthop Trauma 22:346–350 [DOI] [PubMed] [Google Scholar]
- 4. Odvina CV, Zerwekh JE, Rao DS, Maalouf N, Gottschalk FA, Pak CY. 2005. Severely suppressed bone turnover: a potential complication of alendronate therapy. J Clin Endocrinol Metab 90:1294–1301 [DOI] [PubMed] [Google Scholar]
- 5. Giusti A, Hamdy NA, Papapoulos SE. 2010. Atypical fractures of the femur and bisphosphonate therapy. A systematic review of case/case series studies. Bone 47:169–180 [DOI] [PubMed] [Google Scholar]
- 6. Armamento-Villareal R, Napoli N, Diemer K, Watkins M, Civitelli R, Teitelbaum S, Novack D. 2009. Bone turnover in bone biopsies of patients with low-energy cortical fractures receiving bisphosphonates: a case series. Calcif Tissue Int 85:37–44 [DOI] [PubMed] [Google Scholar]
- 7. Armamento-Villareal R, Napoli N, Panwar V, Novack D. 2006. Suppressed bone turnover during alendronate therapy for high-turnover osteoporosis. N Engl J Med 355:2048–2050 [DOI] [PubMed] [Google Scholar]
- 8. Lenart BA, Neviaser AS, Lyman S, Chang CC, Edobor-Osula F, Steele B, van der Meulen MC, Lorich DG, Lane JM. 2009. Association of low-energy femoral fractures with prolonged bisphosphonate use: a case control study. Osteoporos Int 20:1353–1362 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Giusti A, Hamdy NA, Dekkers OM, Ramautar SR, Dijkstra S, Papapoulos SE. 2011. Atypical fractures and bisphosphonate therapy: a cohort study of patients with femoral fracture with radiographic adjudication of fracture site and features. Bone 48:966–971 [DOI] [PubMed] [Google Scholar]
- 10. Lenart BA, Lorich DG, Lane JM. 2008. Atypical fractures of the femoral diaphysis in postmenopausal women taking alendronate. N Engl J Med 358:1304–1306 [DOI] [PubMed] [Google Scholar]
- 11. Abrahamsen B, Eiken P, Eastell R. 2009. Subtrochanteric and diaphyseal femur fractures in patients treated with alendronate: a register-based national cohort study. J Bone Miner Res 24:1095–1102 [DOI] [PubMed] [Google Scholar]
- 12. Nieves JW, Bilezikian JP, Lane JM, Einhorn TA, Wang Y, Steinbuch M, Cosman F. 2010. Fragility fractures of the hip and femur: incidence and patient characteristics. Osteoporos Int 21:399–408 [DOI] [PubMed] [Google Scholar]
- 13. Park-Wyllie LY, Mamdani MM, Juurlink DN, Hawker GA, Gunraj N, Austin PC, Whelan DB, Weiler PJ, Laupacis A. 2011. Bisphosphonate use and the risk of subtrochanteric or femoral shaft fractures in older women. JAMA 305:783–789 [DOI] [PubMed] [Google Scholar]
- 14. Abrahamsen B, Eiken P, Eastell R. 2010. Cumulative alendronate dose and the long-term absolute risk of subtrochanteric and diaphyseal femur fractures: a register-based national cohort analysis. J Clin Endocrinol Metab 95:5258–5265 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Schilcher J, Michaëlsson K, Aspenberg P. 2011. Bisphosphonate use and atypical fractures of the femoral shaft. N Engl J Med 364:1728–1737 [DOI] [PubMed] [Google Scholar]
- 16. Black DM, Kelly MP, Genant HK, Palermo L, Eastell R, Bucci-Rechtweg C, Cauley J, Leung PC, Boonen S, Santora A, de Papp A, Bauer DC, Fracture Intervention Trial Steering C, Committee HPFTS 2010. Bisphosphonates and fractures of the subtrochanteric or diaphyseal femur. N Engl J Med 362:1761–1771 [DOI] [PubMed] [Google Scholar]
- 17. Rizzoli R, Akesson K, Bouxsein M, Kanis JA, Napoli N, Papapoulos S, Reginster JY, Cooper C. 2011. Subtrochanteric fractures after long-term treatment with bisphosphonates: a European Society on Clinical and Economic Aspects of Osteoporosis and Osteoarthritis, and International Osteoporosis Foundation Working Group Report. Osteoporos Int 22:373–390 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Cummings SR, Nevitt MC, Browner WS, Stone K, Fox KM, Ensrud KE, Cauley J, Black D, Vogt TM. 1995. Risk factors for hip fracture in white women. Study of Osteoporotic Fractures Research Group. N Engl J Med 332:767–773 [DOI] [PubMed] [Google Scholar]
- 19. Prentice RL. 1986. A case-cohort design for epidemiologic cohort studies and disease prevention trials. Biometrika 73:1–11 [Google Scholar]
- 20. Cummings SR, Black DM, Nevitt MC, Browner W, Cauley J, Ensrud K, Genant HK, Palermo L, Scott J, Vogt TM. 1993. Bone density at various sites for prediction of hip fractures. The Study of Osteoporotic Fractures Research Group. Lancet 341:72–75 [DOI] [PubMed] [Google Scholar]
- 21. Kanis JA, Johnell O, Oden A, Johansson H, Eisman JA, Fujiwara S, Kroger H, Honkanen R, Melton LJ, 3rd, O'Neill T, Reeve J, Silman A, Tenenhouse A. 2006. The use of multiple sites for the diagnosis of osteoporosis. Osteoporos Int 17:527–534 [DOI] [PubMed] [Google Scholar]
- 22. Glüer CC, Cummings SR, Pressman A, Li J, Glüer K, Faulkner KG, Grampp S, Genant HK. 1994. Prediction of hip fractures from pelvic radiographs: the study of osteoporotic fractures. The Study of Osteoporotic Fractures Research Group. J Bone Miner Res 9:671–677 [DOI] [PubMed] [Google Scholar]
- 23. Koh JS, Goh SK, Png MA, Kwek EB, Howe TS. 2010. Femoral cortical stress lesions in long-term bisphosphonate therapy: a herald of impending fracture? J Orthop Trauma 24:75–81 [DOI] [PubMed] [Google Scholar]
- 24. Riggs BL, Melton LJ, 3rd, Robb RA, Camp JJ, Atkinson EJ, Peterson JM, Rouleau PA, McCollough CH, Bouxsein ML, Khosla S. 2004. Population-based study of age and sex differences in bone volumetric density, size, geometry, and structure at different skeletal sites. J Bone Miner Res 19:1945–1954 [DOI] [PubMed] [Google Scholar]
- 25. Greenspan SL, Beck TJ, Resnick NM, Bhattacharya R, Parker RA. 2005. Effect of hormone replacement, alendronate, or combination therapy on hip structural geometry: a 3-year, double-blind, placebo-controlled clinical trial. J Bone Miner Res 20:1525–1532 [DOI] [PubMed] [Google Scholar]
- 26. Donaldson MG, Cawthon PM, Schousboe JT, Ensrud KE, Lui LY, Cauley JA, Hillier TA, Taylor BC, Hochberg MC, Bauer DC, Cummings SR. 2011. Novel methods to evaluate fracture risk models. J Bone Miner Res 26:1767–1773 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Theobald TM, Cauley JA, Gluer CC, Bunker CH, Ukoli FA, Genant HK. 1998. Black-white differences in hip geometry. Study of Osteoporotic Fractures Research Group. Osteoporos Int 8:61–67 [DOI] [PubMed] [Google Scholar]
- 28. Beck TJ, Oreskovic TL, Stone KL, Ruff CB, Ensrud K, Nevitt MC, Genant HK, Cummings SR. 2001. Structural adaptation to changing skeletal load in the progression toward hip fragility: the study of osteoporotic fractures. J Bone Miner Res 16:1108–1119 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
