British Journal of Radiology (2004) 77, 219-223
© 2004 British Institute of Radiology
doi: 10.1259/bjr/79551075
Differences in proximal femur geometry distinguish vertebral from femoral neck fractures in osteoporotic women
S Gnudi, MD
1
N Malavolta, MD
2
D Testi, MEng
3 and
M Viceconti, MEng
3
1 Modulo Dipartimentale di Medicina Generale, Istituto Ortopedico Rizzoli, Via Pupilli, 1, 40100 Bologna, 2 Servizio di Reumatologia U.O. di Medicina Interna, Dipartimento di Medicina Interna e dell'Invecchiamento, Azienda Ospedaliera di Bologna, Policlinico S. Orsola Malpighi, Via Massarenti, 9, 40100 Bologna and 3 Laboratorio di Tecnologia Medica – Istituto Ortopedico Rizzoli, Via Pupilli, 1, 40100 Bologna, Italy
 |
Abstract
|
|---|
Bone mineral density (BMD) is generally used to predict the risk of fracture in osteoporotic subjects. However, femoral neck BMD and spine BMD have been reported not to be significantly different among patients with hip or vertebral fractures, suggesting that other risk factors are needed to determine the different fracture types. Proximal femur geometry (PFG) parameters, such as hip axis length (HAL), femoral neck-shaft angle (NSA) and femoral neck diameter (FND) have also been shown to predict the risk of hip fracture. These parameters are statistically different in spine fractures compared with both types of hip fractures (trochanteric and femoral neck) when considered together. We wanted to assess the difference in these parameters by comparing spine fractures with a homogeneous group of hip fractures, i.e. femoral neck fractures. 807 post-menopausal women were divided into three groups; those with vertebral fractures (182), those with femoral neck fractures (134) and a control group without fractures (491). Dual X-ray absorptiometry (DXA) scans of the spine and hip were carried out to measure BMD and define the PFG parameters of the hip. Data were statistically analysed. In agreement with other authors, we found that women with femoral neck fractures had longer HAL, wider FND and larger NSA than controls, whereas there were no statistically significant differences in PFG between women with spine fractures and controls. Logistic regression showed HAL and NSA could predict the risk of femoral neck but not vertebral fracture. These data indicate specificity of some PFG parameters for hip fracture risk.
|
Introduction
|
|---|
Different pathogenetic factors [1] cause post-menopausal (type one) and senile (type two) osteoporosis [2] differentiating them into two distinct syndromes, respectively characterized by the prevalent clinical outcome of vertebral or hip fracture and by the prevalence of trabecular or cortical bone loss. These differences in bone loss might explain why bone mineral density (BMD) measurements at the site of the fracture [3–5] have been shown to best predict the fracture risk. Nevertheless, BMD also discriminates osteoporotic fractures at non-specific skeletal sites [3, 6–8]. In addition, femoral neck BMD [9, 10] and spine BMD [11] have been recently reported not to be significantly different for hip or vertebral fractures. This indicates that in both types of osteoporosis bone loss is a generalized phenomenon and supports the concept that other risk factors are needed to determine the different types of fracture. This is also strengthened by the fact that some proximal femur geometry parameters (PFG) have also been shown to be associated to hip fracture risk [12–14] in osteoporotic subjects. Quite interestingly some of these parameters, such as hip axis length (HAL) [9, 10] and femoral neck-shaft angle (NSA) [9], have been recently reported to be significantly different in subjects that have suffered vertebral fractures compared with those that have suffered hip fractures. This suggests heterogeneity among osteoporotic fractures and specificity for the pathogenesis of hip fracture. Few studies of this kind are available [12–16], and in addition they have simultaneously considered both types of hip fracture, although femoral neck and trochanteric fractures have been reported to be distinguished entities not only anatomically but also for hip BMD values and for other risk factors [17, 18]. So the ability of each PFG parameter to discriminate femoral neck and/or trochanteric fractures in comparison with other osteoporotic fractures should be further investigated to better understand their specificity for these kinds of fractures.
As a contribution to this topic, we retrospectively studied three groups of post-menopausal women; one group with osteoporotic fracture of the spine, one group of osteoporotic fracture of the femoral neck and one group without fracture. Dual X-ray absorptiometry (DXA) scans of the spine and hip were carried out to measure BMD, and DXA images were used to define the PFG parameters of the hip.
|
Material and methods
|
|---|
For the study 807 post-menopausal white women older than 65 years were selected (mean age 75.4±6.6 years (±1 standard deviation, SD)). They consisted of 316 consecutive inpatients having had femoral neck (134) or vertebral (182) fracture and a control group of 491 consecutive outpatients without fracture (it was not known if they had fallen previously), who attended our centre for routine BMD measurement. The exclusion criteria were: endocrine diseases, lung, heart, kidney or liver failure, long term immobilization, malignancy, major rheumatic diseases, long-term users of steroids, Paget's disease of bone or severe osteoarthritis of the hip or of the spine, mental deterioration and inability to keep the correct scanning position. Only fractures due to minor trauma such as falling from a standing position or while walking were considered. Fractures were diagnosed from radiographs, covering T5–L5, by a radiologist. Vertebral fractures were diagnosed by radiographs of the spine by looking for a reduction in anterior, middle or posterior vertebral height of more than 25% [19]. Age, age at menopause, height, and weight were recorded for each woman. Lumbar spine and left hip (or non-fractured hip in fractured women) were scanned by a pencil beam XR 36 densitometer by Norland (Norland Corp, Fort Atkinson, WI) that does not give image magnification. All scans (both spine and hip) of the fractured subjects were taken within 4 weeks from the fracture. BMD measurements were taken at lumbar spine (L2–L4), femoral neck, Ward's triangle and trochanter. PFG measurements were taken, using the instrument's dedicated software, for HAL, femoral neck diameter (FND) and NSA. HAL was measured from the inner pelvic brim to the lateral side of the trochanter along the hip axis automatically placed by the software. The same software automatically measures the angle between the hip axis and the line orthogonal to the shaft axis from which NSA can be easily calculated by adding 90 degrees. Using the dedicated software FND was manually measured perpendicularly to the HAL in its narrowest portion. Patient positioning for DXA scan was standardized by using the leg rotating fixture provided by Norland. The internal rotation provided by this instrument was tested, at the level of the lateral condyle, on 33 consecutive examinations, which gave a mean angle of 17.7±2.5° (±1 SD), which is acceptable according to the instrument and method used. The precision of the technique for PFG measurements was determined on 24 subjects (11 fractured and 13 controls) by three repeated measurements by different operators after repositioning [13]. All research methods were conducted according to the declaration of Helsinki.
Statistics
Univariate analysis of variance (ANOVA) was used for testing differences between the groups of hip, vertebral and controls. Relationships between PFG parameters and the other variables were tested by linear regression analysis by calculating the Pearson correlation coefficient (R). The association of DXA measurements to fracture was estimated by a logistic regression model. Odds ratios (OR) were standardized for 1 standard deviation from the means of controls and corrected for confounders. The ability of PFG measurements to correctly separate hip fractured women from controls was compared by receiver operating characteristic (ROC) curves generated with their standard errors (SE).
|
Results
|
|---|
The averaged coefficient of variation (CV) for the PFG measurement technique was 1.08% for the NSA, 1.07% for HAL and 1.56% for FND. The ANOVA test did not show any statistically significant differences in CV between fractured patients and controls.
Biological characteristics of the three groups of women are reported in Table 1
. Univariate ANOVA showed statistically significant differences among groups for each considered variable. The post hoc test of univariate ANOVA showed that subjects with vertebral fracture were significantly shorter, lighter, and started menopause at a younger age than controls. They also had a significantly lower BMD at each measurement site but there were no statistically significant differences in PFG compared with controls. A significantly larger NSA, longer HAL, wider FND together with an older age and lower BMD at each measurement site distinguished femoral neck fractures from controls. Subjects with vertebral fracture were younger, shorter and had significantly lower BMD at lumbar spine compared with those with femoral neck fracture. They also had significantly smaller NSA and shorter HAL. The differences between groups persisted after correcting the data for age (Table 1).
Linear correlations of PFG parameters are reported in Table 2. NSA did not correlate with age and proximal femur BMD measurements; neither did HAL (except for Ward's Triangle) whereas FND significantly correlated with all these variables. HAL is significantly correlated to FND and NSA, whereas these do not correlate with each other (Table 2).
View this table:
[in this window]
[in a new window]
|
Table 2. Correlations between variables and proximal femur geometry parameters (linear regression), carried out in the whole population
|
|
Logistic regression after correcting only for age, age of menopause, height and weight, showed there was a statistically significant association between vertebral fracture and: lumbar spine BMD (OR: 1.89, 95% CI: 1.54–2.32), femoral neck BMD (OR: 1.84, 95% CI: 1.48–2.27), trochanter BMD (OR: 1.78, 95% CI: 1.44–2.20), Ward's triangle BMD (OR: 1.84, 95% CI: 1.47–2.29), and between femoral neck fracture and: spine BMD (OR: 1.28, 95% CI: 1.04–1.57), femoral neck BMD (OR: 2.33, 95% CI: 1.79–3.03), trochanter BMD (OR:1.84, 95%CI: 1.45–2.34), Ward's triangle BMD (OR: 2.47, 95% CI: 1.86–3.29). The association of PFG parameters to fracture is shown in Table 3. FND was associated to the vertebral fracture risk after simple adjusting for age, age at menopause, height and weight, but no longer predicted fracture after adjusting also for BMD. NSA and HAL did not distinguish any case of vertebral fracture. Femoral neck fractures were significantly distinguished by all PFG parameters after simple adjusting for age, age at menopause, height and weight. NSA and HAL were also good at distinguishing femoral neck fracture also after correcting for BMD.
View this table:
[in this window]
[in a new window]
|
Table 3. Logistic regression measuring the effect of PFG parameters on the risk of femoral neck and spine fractures after simple adjusting for age, age at menopause, weight, height and for the same variables plus BMD
|
|
ROC analysis was performed to compare the ability of PFG parameters to discriminate hip fractures. The area under the curve was 0.666 (SE: 0.027) for NSA, 0.586 (SE: 0.028) for HAL and 0.600 (SE: 0.027) for FND. There were no statistically significant differences between NSA and HAL, whereas there was statistical significance between NSA and FND (p<0.01), according to the Hanley McNeil test.
The areas under the ROC curves for BMD measurements were: 0.722 (SE: 0.024) for femoral neck BMD, 0.690 (SE: 0.025) for trochanter BMD, 0.708 (SE: 0.023) for Ward's triangle BMD, and 0.587 (SE: 0.029) for lumbar spine BMD, which was significantly lower (p<0.01) than the other ROC curve areas for BMD measurements of the hip. These were not significantly different among each other.
|
Discussion
|
|---|
In this study we evaluated the ability of some PFG parameters to differentiate spine from femoral neck fractures in osteoporotic women to test their specificity for the latter type of fracture. We focused our attention on a single type of hip fracture to have a homogeneous group, as risk factors differ between femoral neck and trochanteric fractures [17, 18].
As expected we found that BMD was lower in women with osteoporotic fracture than in controls and that BMD was effective in discriminating controls from both spine and hip fracture [3–5] after correcting for age, age of menopause, height and weight. Comparing the two groups of fractured women, spine BMD was significantly lower in vertebral fracture [10], whereas a wide overlap in BMD values was found at hip measurement sites [9, 10] where bone densities were almost identical (also after correcting for age) thus confirming that other risk factors should be considered in hip fracture pathogenesis [20–22]. With regards to PFG, our data show, in agreement with other authors, that hip fractured women have a longer HAL [10, 14, 19], a wider FND [24–26] and a larger NSA [14, 23] compared with controls and that there are no statistically significant differences in PFG between women with vertebral fracture and controls [9]. Logistic regression after simple adjusting for age, age of menopause, weight and height showed that FND was significantly able to discriminate women with vertebral and femoral neck fractures from controls, while HAL and NSA discriminated only those with femoral neck fracture. The ability of FND to discriminate both vertebral and femoral neck fractures precisely when not corrected for BMD and its statistically significant correlation to BMD at hip measurement sites and age [14] seems to indicate that its association with the fracture risk probably depends not only on the geometric factor itself, but also on the osteoporosis-related modifications of this parameter [26]. Because NSA and HAL discriminate only femoral neck fractured women independently of BMD and lack linear correlation to age, they are more likely to be associated with the fracture risk as hip geometric variables [27] and for this reason enter exclusively in femoral neck fracture risk prediction. This is also supported by the results of the ANOVA test showing that HAL and NSA are significantly different in women with vertebral or femoral neck osteoporotic fractures. NSA has been shown to be the best discriminator of femoral neck fracture although not significantly better than HAL according to the ROC curve area comparison. We therefore agree that some PFG parameters are independent risk factors for hip fracture [22, 23] and that their characteristics might indicate the hip as the site of the fracture in subjects at risk [9]. Our study has some limitations as it is not longitudinal, the sample studied is not population based, results are related to the software of only one type of densitometric device and trochanteric fractures were not considered. In addition NSA and HAL measurements from DXA scan image, being linear projections of the real anatomic forms, can be biased by inaccurate correction of femoral anteversion by the intrarotation of the patient's legs. As all subjects had the same angle of internal rotation of the scanned femur, those with wider femoral anteversion might result in an overestimation of the NSA by DXA image. So it cannot be excluded that the difference we found among groups concerning NSA could derive from wider anteverted angles in those with femoral neck fracture. If this were the case the same subjects would also have had an underestimation of HAL. Specific studies are needed to investigate whether our finding of a larger NSA in femoral neck fracture depends on the effect in the planar projection of greater femoral anteversion.
Nevertheless, with these limitations the PFG measurements from DXA images of the hip are consistent with the specificity of the apparent NSA and HAL for femoral neck fracture risk and consistent with the hypothesis of heterogeneity of factors leading to femoral neck or vertebral osteoporotic fracture, whereas FND is not, since it seems to be influenced by the osteoporotic bone rearrangement. Further longitudinal studies will be needed to confirm these data and to verify their utility in improving the specificity of risk assessment in women with low femoral density.
Received for publication September 11, 2002.
Revision received August 13, 2003.
Accepted for publication September 10, 2003.
|
References
|
|---|
- Callager JC. The pathogenesis of osteoporosis. Bone Miner 1990;9:215–27.[CrossRef][Medline]
- Albright F, Smith PH, Richardson AM. Postmenopausal osteoporosis. JAMA 1941;116:2465–74.[Abstract/Free Full Text]
- Marshall D, Johnell O, Wedel H. Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures. BMJ 1996;312:1254–9.[Abstract/Free Full Text]
- Sugimoto T, Kanbara Y, Shiraishi H, Kawakatsu M, Negishi H, Fukase M, et al. Femoral and spinal bone mineral density in Japanese osteoporotics with hip fracture. Osteoporos Int 1994;4:144–8.[CrossRef][Medline]
- Aguat P, Fan B, Lane NE, Lang TF, LeHir P, Lu Y, et al. Assessement of bone mineral at appendicular sites in females with fractures of the proximal femur. Bone 1998;22:395–402.[Medline]
- Melton III LJ, Atkinson EJ, O'Fallon WM, Wahner HW, Riggs BL. Long-term fracture prediction by bone mineral assessed at different skeletal sites. J Bone Miner Res 1993;8:1227–33.[Medline]
- Black DM, Cummings SR, Genanat HK, Nevitt MC, Palermo L, Browner W. Axial and appendicular bone density predict fractures in older women. J Bone Miner Res 1992;7:633–8.[Medline]
- Jergas M, Genant HK. Spinal and femoral DXA for the assessment of spinal osteoporosis. Calcif Tissue Int 1997;61:351–7.[CrossRef][Medline]
- Boonen S, Koutri R, Dequeker J, Aerssens J, Lowet G, Nijs J, et al. Measurement of femoral geometry in type I and type II osteoporosis: differences in hip axis length consistent with heterogeneity in the pathogenesis of osteoporotic fractures. J Bone Miner Res 1995;10:1908–12.[Medline]
- Rosso R, Minisola S. Hip axis length in an Italian osteoporotic population. Br J Radiol 2000;73:969–72.[Abstract]
- Takahashi M, Kushida K, Naitou K. The degree of osteoporosis in patients with vertebral fracture and patients with hip fracture: relationship to incidence of vertebral fracture. J Bone Miner Metab 1999;17:187–94.[CrossRef][Medline]
- Faulkner KG, Cummings SR, Black D, Palermo L, Gluer CC, Genant HK. Simple measurement of femoral geometry predicts hip fracture: the study of osteoporotic fractures. J Bone Miner Res 1993;8:1211–7.[Medline]
- Gnudi S, Ripamonti C, Gualtieri G, Malavolta N. Geometry of proximal femur in the prediction of hip fracture in osteoporotic women. Br J Radiol 1999;72:729–33.[Abstract]
- Alonso CG, Curiel MD, Carranza FH, Cano RP, Perez AD. Femoral bone mineral density, neck-shaft angle and mean femoral neck width as predictors of hip fracture in men and women. Osteoporos Int 2000;11:714–20.[CrossRef][Medline]
- Gluer CC, Cummings SR, Pressman A, Li J, Gluer K, Faulkner KG, et al and The study of osteoporotic fractures research group. Prediction of hip fractures from pelvic radiographs: the study of osteoporitic fractures. J Bone Miner Res 1994;9:671–7.[Medline]
- Center JR, Nguyen TV, Pocock NA, Noakes KA, Kelly PJ, Eisman JA, et al. Femoral neck axis length, height loss and risk of hip fracture in males and females. Osteoporos Int 1998;8:75–81.[CrossRef][Medline]
- Greenspan SL, Myers ER, Maitland LA, Kido TH, Krasnow MB, Hayes WC. Trochanteric bone mineral density is associated with type of hip fracture in the elderly. J Bone Miner Res 1994;9:1889–94.[Medline]
- Michaelsson K, Weiderpass E, Farahmand BY, Baron JA, Persson PG, Zidén L, et al on behalf of the Swedish Hip Fracture study Group. Differences in risk factor patterns between cervical and trochanteric hip fractures. Osteoporos Int 1999;10:487–94.[CrossRef][Medline]
- Genant HK, Wu CY, Van Kuijk C, Nevitt MC. Vertebral fracture assessment using a semiquantitative technique. J Bone Miner Res 1993;8:1137–48.[Medline]
- Pinilla TP, Boardman KC, Bouxsein ML, Myers ER, Hayes WC. Impact direction from a fall influences the failure load of the proximal femur as much as age-related bone loss. Calcif Tissue Int 1996;58:231–5.[Medline]
- Luukinen H, Koski K, Laippala P, Kivela SL. Factors predicting fractures during falling impacts among home-dwelling older adults. J Am Geriatr Soc 1997;45:1302–9.[Medline]
- Peacock M, Turner CH, Liu G, Manatunga AK, Timmerman L, Johnston CC Jr. Better discrimination of hip fracture using bone density, geometry and architecture. Osteoporos Int 1955;5:167–73.
- Partanen J, Jamsa T, Jalovaara P. Influence of the upper femur and pelvic geometry on the risk and type of hip fractures. J Bone Miner Res 2001;16:1540–6.[Medline]
- Michelotti J, Clark J. Femoral neck length and hip fracture risk. J Bone Miner Res 1999;14:1714–20.[CrossRef][Medline]
- Cody DD, Hou FJ, Divine GW, Fyhrie DP. Femoral structure and stiffness in patients with femoral neck fracture. J Orthop Res 2000;18:443–8.[Medline]
- Karlsson KM, Sernbo I, Obrant KJ, Redlund-Johnell I, Johnell O. Femoral neck geometry and radiographic signs of osteoporosis as predictors of hip fracture. Bone 1996;18:327–30.[Medline]
- Peacock M, Liu G, Carey M, Ambrosius W, Turner CH, Hui S, et al. Bone mass and structure at the hip in men and women over the age of 60 years. Osteoporos Int 1998;8:231–9.[CrossRef][Medline]
Top
Abstract
Introduction
