Interest in the measurement of the dimensions and geometry of the proximal femur as part of the assessment of fracture risk was spurred by the initial recognition of hip axis length as an independent predictor of hip fracture risk. Other measures have come under scrutiny as predictors of hip fracture risk as well. These are measures such as the neck-shaft angle and femoral neck width as well as segments of the hip axis length.
Hip axis length (HAL) has been demonstrated to be a predictor of hip fracture risk that is independent of BMD (67). As part of the Study of Osteoporotic Fractures, 8074 women aged 65 and older were evaluated with DXA (Hologic QDR-1000) measurements of the proximal femur. HAL was measured in 134 women without fractures and in 64 women who experienced hip fractures during 1.6 years of follow-up. A goniometer was used to measure HAL from the computer printout. HAL was defined as the distance from the inner pelvic brim to the outer edge of the greater trochanter along the femoral neck axis as shown in Fig. 10-8. Odds ratios for BMD and the risk of hip fracture and for HAL and the risk of hip fracture were calculated. For femoral neck BMD, each SD decline in BMD resulted in a 2.7-fold increase in the risk of hip fracture. For HAL, each SD increase in length resulted in a 1.9-fold increase in the risk of femoral neck fracture and a 1.6-fold increase in the risk of trochanteric fracture.
In 1994, Faulkner et al. (68) proposed automating the measurement of HAL on proximal femur bone density studies. Faulkner et al. re-analyzed the 198 proximal femur studies for which HAL had been determined manually (67), with modified proximal femur analysis software using a QDR-1000/W (Hologic, Inc.). The correlation between the manual and automated measurements was 0.98. The precision of the automated HAL measurement was determined as the average SD from three measurements on each of 33 women. This value was 0.07 cm or 0.68%.
Duboeuf et al. (69) also found that HAL was a significant predictor of femoral neck fracture risk with an odds ratio of 1.64 based on an analysis of proximal femur studies from the EPIDOS4 study. Unlike Faulkner et al. (67), however, these authors concluded that HAL was not a significant predictor of trochanteric fracture. Other authors have attempted to partition the HAL into segments, to determine if any particular segment of HAL was more predictive of hip fracture than HAL itself. One such segment is the femoral neck axis length (FNAL), which is a segment of HAL that spans the base of the greater trochanter to the apex of the femoral head as shown in Fig. 10-9. Center et al. (70) measured the FNAL in 260 men and women from the Dubbo Osteoporosis Epidemiology Study5 from proximal femur DXA studies in which the analysis software included a ruler that could be a manipulated for the measurement. In this study, FNAL was significantly correlated with current and maximum height. When FNAL was adjusted for current and maximum height, however, no difference was seen between hip fracture patients and nonfractured patients in the FNAL. Center et al. concluded that the FNAL had very little utility in improving hip fracture risk prediction.
4 EPIDOS is a multicenter prospective study of 7575 women living at home, age 75 to 95 years, in France. The data in this study are from 320 women from the Lyon and Montpellier centers.
5 The Dubbo Osteoporosis Epidemiology Study is a case-control study involving 1902 men and women who were recruited between 1989 and 1993.
Bergot et al. (71) evaluated HAL and seven different segments of HAL, including the FNAL, from proximal femur DXA scans acquired on a Hologic QDR 1000/W. Comparisons of measurements in 49 women with nontraumatic hip fractures to measurements in 49 age-matched women and 49 age- and BMD-matched women were made to determine which HAL segments best discriminated between the three groups of women. The best discriminators between women with hip fractures and all nonfractured women were femoral neck BMD and HAL. Femoral neck BMD and HAL were also the best discriminators between the women with hip fracture and the nonfractured, age-matched women. The best discriminator between fractured women and low BMD-matched, nonfractured women was a different segment of HAL called the intertrochanter-head center distance, defined by the line that begins at the center of the femoral head and ends at the intertrochanteric line, shown in Fig. 10-9. FNAL was not useful in discriminating between the groups. The authors noted that the intertrochanter-head center distance is more difficult to measure than HAL because of the imprecision in marking the center of the femoral head. Normal values, as well as the magnitude of risk imparted by increases in the intertrochanter-head center distance, must be established. This particular segment of HAL remains a focus of research interest.
HAL has been suggested as a possible explanation for the difference in hip fracture rates among various races. Cummings et al. (72) demonstrated that the mean HAL of predominantly US-born Asian and Black women was significantly shorter than that of Caucasian women. Both Asian and Black women are known to have a lower risk of hip fracture than Caucasian women. Nakamura et al. (73) demonstrated that the FNAL, a component of HAL, was significantly shorter in Japanese women than in Caucasian women, noting that hip fracture rates in Japanese are approximately half that seen in Caucasians. HAL is simple to measure and is now offered as an automated measurement commercially on proximal femur bone density studies. Figure 10-10 is a DualFemur® DXA study performed on a Lunar Prodigy showing the HAL measurement. The scale indicates the patient's value in comparison to the mean value predicted for height.
Although this is clearly a nonmodifiable risk factor, HAL can be useful for hip fracture risk stratification. Using the Lunar Prodigy, Bonnick and Lewis (74) reported RMS-SD and RMS-%CV precision values for HAL measurements of the left femur of 0.7 mm or 0.67% in women age 20 to 49 and 0.6 mm or 0.53% in women age 50 to 70.
The femoral neck-shaft angle is another geometric measure that has been studied as a predictor of hip fracture risk. This angle is indicated by the letter "E" in Fig. 10-8. The studies to date have provided mixed results on the utility of this measure. Several studies (75-78) have found significantly greater femoral neck-shaft angles in hip fracture patients than in controls while others (67,71,79) have not. In the study from Bergot et al. (71), the neck-shaft angle was not different in the hip fracture patients compared to the low-BMD nonfractured patients or the age-matched controls. The neck-shaft angle was 127.1° in the fracture patients compared to 127.2° in the low-BMD nonfractured patients and 125.6° in the age-matched control group. In contrast, in a study from Partanen et al. (78), the neck-shaft angle was statistically significantly greater in a group of 70 postmenopausal women with hip fractures (mean age 74.9 years) compared to a control group of 40 postmenopausal women (mean age 73.7) without hip fractures. In the fracture group, the mean neck-shaft angle was 133.7° compared to a mean of 128.3° in the control group. Partanen et al. also found that the neck-shaft angle was significantly greater in the 46 women with cervical fractures than in the 24 women with trochanteric fractures. The neck-shaft angle in the cervical fracture group was 135.7° compared to 130.3° in the trochanteric fracture group.
Femoral neck width is measured at the narrowest part of the femoral neck as indicated by line C-D in Fig. 10-8. An increase in neck width from periosteal bone apposition has been postulated as a response to a decrease in bone density. The result of the width increase should be an increase in the cross-sectional moment of inertia (CSMI). This would potentially compensate for the reduction in endosteal bone and theoretically reduce the risk of hip fracture. If this is so, the increase in femoral neck width in the presence of a low bone density should indicate a reduction in fracture risk compared to individuals with an average or reduced neck width and the same low bone density. Nevertheless, Bergot et al. (71) did not find a statistically significant difference in neck width between hip fracture patients and age- and BMD-matched nonfractured controls. The absolute neck width was slightly greater in the age- and BMD-matched nonfractured control group compared to the hip fracture patients at 3.20 cm compared to 3.17 cm. In this study, the femoral neck width was greater in both the fracture group and age- and BMD-matched nonfractured group compared to the age-matched controls but again, not statistically significantly so. Other investigators (75,76) have found a significant increase in femoral neck width in fracture patients compared to controls.
The upper femoral neck is a relatively new region of interest in the proximal femur. It is the superior half of the traditional femoral neck region of interest as shown in Fig. 10-11. Yoshikawa et al. (80) suggested that there was a greater decrease in bone density in the superior region of the femoral neck in women. They suggested that this would
cause the center of mass to move in such a way as to place greater stress on the femoral neck, increasing the risk of fracture. The upper femoral neck region of interest was compared to the entire femoral neck region and to the lower femoral neck region for the prediction of neck and trochanteric hip fracture in the study from Duboeuf et al. (69). In this study, upper femoral neck BMD was highly predictive of femoral neck fracture with an odds ratio of 2.8 for each SD decline in bone density and outperformed the more traditional total femoral neck measurement. The lower femoral neck BMD was not predictive of femoral neck fracture. All three regions were predictive of trochanteric fracture and hip fracture in general.
Yoshikawa et al. (80) developed algorithms to calculate the CSMI as well as other measures of hip strength. Yoshikawa et al. noted that although BMD was an important predictor of hip fracture risk, BMD accounted for only 50% of the bone strength estimated from the CSMI. This suggested that the CSMI reflected elements of bone strength not captured in the measurement of BMD. In 2002, Crabtree et al. (81) reported the application of a test version of proprietary DXA software (GE Medical Systems) designed to assess hip strength. The subjects for this study were 68 women age 60 and older who had recently suffered a hip fracture and 800 women age 60 and older without a hip fracture originally recruited as part of the EPOS study. All subjects underwent standard proximal femur DXA studies (Lunar DPX). The hip strength analysis (HSA) software uses the proximal femur DXA study to calculate measures reflecting the geometry and bone distribution within the proximal femur. In addition to standard measurements of proximal femur BMD, the program calculated the upper and lower femoral neck BMD, HAL, Cstress, and Fall Index (FI). Cstress reflects the compressive stress from a fall on the greater trochanter. Typical units are N/mm2. The calculation of the CSMI is necessary to calculate Cstress. The FI is a dimensionless quantity that reflects the resistance to fracture from forces generated during a fall on the greater trochanter. In this study, HAL was significantly longer in the fracture patients than in the controls. Cstress was also significantly greater in the fracture patients than in the controls. The FI was significantly lower in the fracture patients than in the controls. Femoral neck BMD, whether measured as a total, upper, or lower neck value, was significantly lower in the fracture patients than in the controls. Unlike the earlier study from Duboeuf et al. (69), Crabtree et al. could not show that the upper femoral neck BMD was a better predictor of femoral neck fracture than total femoral neck BMD. The authors then attempted to develop a statistical model for the prediction of hip fracture status. They found that the combination of Cstress, age, and BMI in the model resulted in an area under the curve (AUC) of 0.875. The use of femoral neck BMD alone resulted in an AUC of 0.827. The difference between these two areas was statistically significant. Femoral neck BMD alone was better than Cstress alone, but the addition of femoral neck BMD to the model containing Cstress, age, and BMI did not improve the model's ability to predict hip fracture status. The authors concluded that HSA could enhance the prediction of hip fracture risk.
Was this article helpful?