| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Experimental Studies |
1 From the Laboratories of Experimental Radiology and Synovial Pathology (V.B., C.B., F.B., C.P.C., A.M.L.J., J.D.L.) and Orthopedic Research (Pr Laurent Sedel) (A.M.), Faculty of Medicine, Lariboisière-Saint-Louis, 10, avenue de Verdun, 75010 Paris, France; and the Department of Musculoskeletal Radiology, Assistance Publique-Hôpitaux de Paris, Hôpital Lariboisière, France (V.B., C.P.C., J.D.L.). Received September 15, 1999; revision requested October 21; revision received January 4, 2000; accepted February 11. Supported in part by the Assistance Publique-Hôpitaux de Paris, France, and the Centre National de la Recherche Scientifique, Paris, France. Address correspondence to V.B. (e-mail: vbousson@free.fr).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Cortical bone specimens from 163 femurs were studied with CT and microradiography. Femurs were from 77 males and 86 females in a white anthropologic collection covering a broad age spectrum. In each sample, the cBMD was measured in the entire cortical width and in periosteal, midcortical, and endosteal subregions of interest. Age- and site-related changes in cBMD were tested for significance by using a two-way analysis of variance for both sexes. By using linear regression, cBMD was compared with porosity in the entire cortical width and in each subregion.
RESULTS: There were significant age-related differences in cBMD (P < .001 in females, P = .008 in males). In addition, cBMD values were significantly different between the three cortical subregions (P < .001 for both sexes), decreasing from the periosteum to the midcortex to the endosteum. The cBMD values were closely related to porosity, and porosity contributed to 71.6% of the variance in cBMD in the overall population.
CONCLUSION: CT is effective in the measurement of age- and site-related changes in cBMD. Decreases in cBMD are closely correlated with increased cortical porosity.
Index terms: Bones, mineralization, 444.1295, 444.56 • Bones, radiography, 444.123, 444.1292 • Computed tomography (CT), quantitative, 444.1211, 444.123 • Femur, CT, 444.1211 • Osteoporosis, 444.56
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
An important parameter of cortical bone strength is the percentage of void volume or porosity (11–13). McCalden et al (13) showed that age-related deterioration of the mechanical properties in the proximal femoral cortex were highly related to increased porosity and that changes in porosity accounted for 76% of the reduction of strength in the proximal femur. In addition, several authors have found greater cortical porosity in patients with femoral neck fractures compared with age-matched control subjects without femoral neck fractures (14–16).
To better define the contribution of CT in the evaluation of femoral cortical bone, we conducted an in vitro study. We first determined whether CT allowed us to measure age- and site-related changes in cortical bone mineral density (cBMD) values, and then we looked for a correlation between cBMD values and intracortical porosity in 163 specimens of human femoral cortex.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
Image Acquisition
CT.—All embedded specimens were imaged in air by using a fourth-generation CT scanner (Somatom Plus 4; Siemens, Erlangen, Germany) with the following operating parameters: small target, 80 kV, 225 mA, 1-second scanning time, and 1-mm section thickness. By using a scout image as a guide, a section passing through the center of the oriented specimen and parallel to the main axis of the femoral shaft was obtained in each specimen. To check for apparatus stability and to provide a calibration standard for CT number conversion, specimens were simultaneously imaged with a mineral reference phantom comprising two components, water-equivalent plastic and bone-equivalent plastic containing 200 mg calcium hydroxyapatite per milliliter (17). The phantom was placed under the specimen. A 1-cm-thick foam pad was placed between the specimen and phantom to avoid an air gap and, thereby, to limit beam hardening and radiation scattering. After acquisition, the images were reconstructed by using a pixel size of 97 µm (field of view, 50 x 50 mm; matrix, 512 x 512) and the SP 50 algorithm (Ramachandran-Lakshminarayanan kernel [18]). Calibration of the apparatus was performed before the first acquisition and then at intervals of 20 acquisitions.
Microradiography and microscopic measurement.—The 100-µm-thick sections were examined at microradiography by using an x-ray tube (CGR Sigma 2060; CGR-GE, Buc, France) operating with 12 kV, 5 mA, and a 15-minute exposition time. A high-spatial-resolution x-ray film (SO-343; Eastman Kodak, Rochester, NY) was placed 10 cm from the target. No aluminum step wedge was used. The exposed film was mounted on glass slides, and the microradiographs were digitized by using a camera (3CCD-Objective, Sony, Tokyo, Japan; AF micro NIKKOR 60 mm, Nikon, Champigny-Sur-Marne, France) linked to an image analysis system (Qwin; Leica Mikrosysteme Vertrieb; Bensheim, Germany). The full width of the specimen from the periosteum to the endosteum was contained within a field of 736 x 574 pixels. The system was calibrated by acquiring an image of a scale with 10 divisions at 1-mm intervals. Pixel size was 1.35 µm.
Image Analysis
CT.—Within each section, density was measured in three spherical subregions of interest, namely, the periosteal, the midcortical, and the endosteal subregions (Fig 1). In all specimens, each subregion of interest was marked once by the same author (V.B.). The midcortical subregion was in the center of the specimen, and the other two subregions were on either side of the midcortical subregion. Each subregion measured 1 mm2 in area and contained 83 pixels. The mean and SD of the pixel values were converted from Hounsfield units to bone units (milligrams of hydroxyapatite per milliliter) on each image and were, respectively, the cBMD and cBMD SD. In addition, the cBMD and cBMD SD were measured within a semiautomatically marked rectangular area that corresponded to the entire cortical width and that was placed inside the boundaries of the cortex. Accuracy in the placement of regions of interest was enhanced by using a threefold magnification of the image, and care was taken to avoid approaching the cortical rim.
|
Microradiography and microscopic measurement.—Microradiographs were analyzed by using an image analysis system (Qwin; Leica) and a homemade program (Visual Basic; Microsoft, Redmond, Wash). The procedure was automated. To allow contour digitalization, the operator (V.B.) used the mouse to place several points on the periosteal and endosteal boundaries. Before the study began, a consensus was developed (by C.B., A.M., and V.B.) regarding placement of the endosteal boundary in specimens with an endosteal aspect that had trabeculae; bone areas with trabeculae at the endosteal aspect were not considered part of the cortex.
Measurements obtained within traced boundaries were cortical thickness (micrometers), area of cortex (square micrometer), area of cavities (square micrometer), and cavity to periosteum distance (micrometers). According to Laval-Jeantet et al (19), bone porosity is defined as the ratio of total cavity area (Haversian and vascular canals, lacunae) to entire area of the cortex. Separation of bone tissue from void was achieved with automatic thresholds. The operator could control the procedure and remove artifacts such as cracks that otherwise would have been considered cavities. All results were entered into a spreadsheet (EXCEL version 5.0; Microsoft). The percentage of void in the periosteal, midcortical, and endosteal subregions was calculated by dividing cortical thickness into three regions of equal width.
The reproducibility of these measurements was tested by one author (V.B.), who examined a series of 23 specimens twice, with an interval of 3 weeks.
Statistical Analysis
Coefficients of variation for cBMD values measured at CT and for microscopic parameters were calculated according to the procedures of Glüer et al (20).
All experimental data were separately compiled in females and males as the mean and SDs in 10-year age groups. Because there were only four men older than 80 years and three women older than 90 years, we used 70–99 years and 80–99 years as our oldest age groups in men and women, respectively.
The cBMD values were studied by using analysis of variance (ANOVA). For each subregion (periosteal, midcortical, and endosteal), two-way analysis of variance for two between factors (sex and 10-year age group) was used to investigate sex- and age- related differences in cBMD. In addition, for each sex, two-way analysis of variance for one within factor (cortical subregion) and one between factor (10-year age group) was used to investigate age- and site-related differences in cBMD and different age-dependent profiles among the three sites. For each sex, two-way analysis of variance for one within factor (cortical subregion) and one between factor (10-year age group) was also used to study age- and site-related differences in porosity.
Relationships between porosity were microscopically assessed in total cortex and in the endosteal, periosteal, and midcortical subregions; cBMD values within each of these four regions were assessed by means of linear regression.
Linear regression analysis was also performed to assess the relationship between cBMD and cBMD SD and to determine age trends in cBMD, cBMD SD, cortical thickness, and intracortical porosity. Values of r quoted for the regression analysis refer to the coefficient of correlation. All tests were two-sided with a significance level equal to .05, and analyses were performed by using STATVIEW 5.0 (Abacus Concepts, Berkeley, Calif).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
|
|
In addition, when the age-dependent changes in the three sites were analyzed for each sex, analysis of variance in males demonstrated (a) age-related differences in cBMD (P = .008), (b) site-related differences in cBMD (P < .001), (c) no significantly different age-related changes in cBMD across the three subregions (P = .7 for the interaction). In females, age- and site-related differences were also found (P < .001 for both effects), but the age-related difference in cBMD varied significantly across the three subregions (P = .002 for the interaction); the decrease in cBMD with advancing age was larger in the endosteal subregion.
Relationships assessed by means of linear regression between age and cBMD values in the regions of interest are listed in Table 3. In both sexes in all regions of interest, cBMD values decreased significantly with advancing age. This decrease was less significant in males than in females. The relationship between cBMD and age was closest in the endosteal subregion in females (r = -0.65) (Fig 4) and in the midcortical subregion in males (r = -0.29).
|
|
Intracortical Porosity and Cortical Thickness
Coefficients of variation for measurements of microscopic parameters were 0.86% for cortical thickness, 0.64% for the bone fraction (ie, fraction free of empty space), 5.32% for total cortical porosity, 12.55% for periosteal porosity, 13.06% for midcortical porosity, and 8.15% for endosteal porosity.
Mean total intracortical porosity and cortical thickness in each 10-year age group in females and males are listed in Table 4. Periosteal, midcortical, and endosteal porosity are shown in Figures 5 and 6.
|
|
|
Total cortical porosity increased significantly with advancing age in females (r = 0.72; P < .001). A similar but weaker correlation was found in males (r = 0.37; P = .001).
An age-related decrease in cortical thickness was observed in females (r = -0.46; P < .001). In males, cortical thickness did not decrease significantly with advancing age (r = -0.19; P = .09). Coefficients of correlation are listed in Table 3.
Relation Between Density and Porosity
Correlations between age, microscopically assessed porosity, and cBMD values measured at CT in males and females are summarized in Table 3. With linear regression, all cBMD were found to be negatively correlated with porosity in both males and females (Table 3, Figs 7–9); correlations were closest in males. Total intracortical porosity accounted for 71.6% of the variance in cBMD values within the entire cortical width in the overall population. All cBMD SD values significantly increased with porosity in both females and males (Table 3).
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Among the strengths of our study is that these results were obtained in a large number of specimens (n = 163) from a homogeneous population of subjects with a broad age range. In addition, to our knowledge, our study findings are the first to show that cBMD values measured at CT reflect intracortical porosity.
Our study has several limitations. First, the source of the study specimens was an anthropologic collection that may not reflect the current situation. However, the specimens were from a relatively recent and well-defined rural population; for each member, age, sex, and cause of death were available.
Second, the study specimens were removed from the anterior cortex at the middle of the femoral shafts, which is not the site at which hip fractures occur and which is not examined at dual-energy x-ray absorptiometry. Therefore, data concerning these specimens cannot offer new insights into the values established at dual-energy x-ray absorptiometry. We chose this site because it is free of muscular insertions and because it is easy to define (allowing reproducible extraction). In addition, the cortex is thick at this site, providing good conditions for CT measurements. The same site was also used in many earlier studies of femoral cortical bone (11,12,21–25).
Third, the calibration phantom used in our study was not specifically designed for cBMD measurements, and the concentrations of bone equivalent in its two components were 0 and 200 mg/mL, which are different from cBMD values. However, similar phantoms have been widely used in the evaluation of cortical bone (7,22,26,27).
Fourth, coefficients of variation for microscopically assessed porosity were high. However, they corresponded to a relative precision error. The corresponding absolute precision errors were less than 0.6 % and did not affect the relationship between density and porosity.
The first point addressed by our study concerned the CT technique. Previous studies (28,29) focused on inaccuracies of cBMD values measured at CT; these inaccuracies were found to be caused mainly by limited spatial resolution and partial volume effects. These inaccuracies have limited the use of CT in the evaluation of the proximal femur, particularly the neck, where the cortex is relatively thin. At the proximal femur, there are large areas of trabecular bone, and it is relatively easy to select an adequate region of interest for obtaining a mean CT value in cancellous bone. The situation is different in cortical bone, which is subject to partial volume averaging effects. Partial volume effects tend to decrease CT numbers to the values corresponding to those of surrounding tissues (ie, cancellous bone and soft tissue). These effects are the main causes of inaccuracies in cBMD values measured at CT.
Partial volume effects explain the discrepancies among published data on the influence of age on cBMD. Studies of cross sections of vertebral bodies revealed that cBMD decreased with advancing age (7,30–32). Similarly, Fujii et al (33) reported that the radius showed a statistically significant decrease in cBMD with advancing age in both females and males. Conversely, other study findings revealed no significant effect of age on cBMD at the femur or radius (28,34) and suggested that previously reported age-related decreases in cBMD were artifacts produced by increased partial volume effects due to cortical thinning (28,29). Hangartner and Gilsanz (28) demonstrated that a cortical thickness of at least 2.0–2.5 mm was necessary to allow accurate evaluation of cBMD. Below this threshold, cBMD values decreased linearly with cortical thickness. These results were obtained by using pixel sizes of 0.3 and 0.6 mm.
We optimized our CT protocol on the basis of the results of previous experimental studies (29,35–37). We used a small target, a pixel size of about 100 µm, an appropriate algorithm, a correction for scanner drift, and frequent calibrations of the apparatus. Cortical thickness at the femoral middiaphysis was greater than 2 mm in all but seven subjects. In addition, the regions of interest, which contained more than 80 pixels, was placed entirely within the boundaries of the cortex. Placement of the midcortical subregion was entirely reliable and was not subject to partial volume effects. The external boundary of the cortex was well defined, and the periosteal subregion was also placed entirely within the cortex. Definition of the internal boundary of the cortex was less easy, since, with advancing age, formation of trabeculae in the internal aspect of the cortex occurs as a result of endosteal resorption. However, correlations between endosteal porosity and endosteal cBMD values in our study were similar to those of the periosteal and midcortical subregions.
Our technique was sensitive enough to demonstrate statistically significant age- and site-related variations in cBMD values. In females, cBMD values decreased significantly with advancing age. A smaller age-related decrease was found in the males; this finding was consistent with previous microscopic data indicating that, compared with females, males lose less cBMD as they age (21). In addition, cBMD decreased significantly from the periosteum to the endosteum; this effect was seen in both sexes but was most marked in females older than 50 years.
In longitudinal studies, techniques used to measure bone mineral density must be sufficiently sensitive to demonstrate continuing bone loss and effects of therapy (6,20). The sensitivity of techniques used to measure bone mineral density can be affected by precision errors (20). Short-term precision errors reflect the reproducibility of the techniques (20). In our study, the short-term precision error in cBMD measurements was low in the total cortex (2.50%), midcortical subregion (1.84%), and periosteal subregion (2.73%) and was higher in the endosteal subregion (3.32%).
With aging, the endosteal boundary of the cortex becomes less well defined as a result of endosteal resorption. A small displacement of the subregion of interest toward the endosteum or midcortex leads to variations in CT values. Although reproducibility problems related to placement of CT sections and regions of interest have been solved with trabecular spinal CT, they remain a challenge in the assessment of trabecular and cBMD in the hip because of the complex three-dimensional architecture of the proximal femur and the considerable variations in cBMD that occur across femoral sites (6). Development of simple automated methods to improve reproducibility of the acquisition of CT images and reproducibility of image analysis (22,34,38,39) is needed to allow reliable monitoring of changes in cortical bone status with CT. Whether the conclusions reached in the present study concerning the value of CT scanning in the evaluation of cBMD can be extended to examination of the proximal part of the femur (where cortical bone is thinner than it is at the middiaphysis) remains to be determined.
The second problem addressed by our study concerned the meaning of variations in cBMD. As measured at CT, cBMD is a mass per unit volume of whole bone (including empty spaces such as Haversian canals, blood vessels, and resorption cavities) (19,40). Thus, cBMD is an apparent mineral density that should be distinguished from true mineral density, which is defined as the weight of ash per unit volume of bone that is free of empty spaces. Therefore, cBMD reflects both the amount of bone (bone volume fraction) and the degree of bone mineralization. Since the degree of mineralization in adult cortical bone tissue seems to change little with age (13,19,41), changes in cBMD probably reflect changes in the bone volume fraction (13), which can be equated to bone porosity (13). In the present study, we found a close relationship between cBMD and porosity in the total cortex and in all three cortical subregions studied. In addition, the ranges of porosity and the age-related changes in porosity in males and females were in agreement with findings of previous microscopic studies (13,21).
To our knowledge, cBMD SD has not yet been studied. Concerning trabecular bone, Dougherty (42) suggested that the SD and coefficient of variation for trabecular CT numbers may reflect bone texture, since they help in the discrimination between subjects with and those without spinal fractures. On the contrary, Engelke et al (43) found that the trabecular SD provides no additional information, compared with bone mineral density values used in the distinction of subjects with osteoporosis and those without osteoporosis. In our study, the cBMD SDs within each of the three cortical subregions increased with age in females (r = 0.41–0.48). The cBMD SDs were moderately correlated with cBMD, since the SDs were larger with lower cBMD values, especially in females (r = -0.68 to -0.80). In addition, the cBMD SD increased with porosity (r = 0.65 in females for the entire cortical width). These data suggest that increased porosity, which resulted in a less homogeneous bone, was responsible for dispersion of pixel values.Practical application: Our in vitro data show that (a) age- and site-related changes in femoral cBMD can be detected by using a current CT scanner with an optimized protocol and (b) changes in cBMD values are closely correlated with cortical porosity, as measured microscopically. These findings further our understanding of the relationship between cortical bone strength and cBMD. Further studies are needed to demonstrate the clinical utility of this technique in the evaluation of hip fracture risk, particularly by establishing that (a) cBMD CT measurements can be used in the proximal part of the femur and (b) separate assessment of cortical and trabecular bone mineral density is better for the prediction of hip fracture than assessment of integral bone mineral density at dual-energy x-ray absorptiometry.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, V.B., J.D.L.; study concepts, V.B., J.D.L., F.B., C.P.C.; study design, V.B., C.B., A.M., J.D.L.; definition of intellectual content, V.B., J.D.L.; literature research, F.B., C.P.C., V.B.; experimental studies, V.B., C.B., A.M.; data acquisition, V.B., C.B., A.M.; data analysis, A.M., V.B.; statistical analysis, V.B., A.M.; manuscript preparation, V.B., J.D.L.; manuscript editing, C.B., J.D.L.; manuscript review, A.M.L.J., J.D.L., C.B., A.M.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. M. Link In Vivo Quantification of Cortical Bone Water with Ultrashort Echo-Time MR Imaging: A New Parameter to Measure Bone Quality? Radiology, September 1, 2008; 248(3): 705 - 706. [Full Text] [PDF]
|
|
|
|
|
|
A. Techawiboonwong, H. K. Song, M. B. Leonard, and F. W. Wehrli Cortical Bone Water: In Vivo Quantification with Ultrashort Echo-Time MR Imaging Radiology, September 1, 2008; 248(3): 824 - 833. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D M L Cooper, Y Ahamed, H M Macdonald, and H A McKay Characterising cortical density in the mid-tibia: intra-individual variation in adolescent girls and boys Br. J. Sports Med., August 1, 2008; 42(8): 690 - 695. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. A. Kontulainen, H. M. Macdonald, and H. A. McKay Change in Cortical Bone Density and Its Distribution Differs between Boys and Girls during Puberty J. Clin. Endocrinol. Metab., July 1, 2006; 91(7): 2555 - 2561. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Marcotte and E. Wang Replicative Senescence Revisited J. Gerontol. A Biol. Sci. Med. Sci., July 1, 2002; 57(7): B257 - 269. [Abstract] [Full Text]
|
|
|
|
|
|
C.S. Klein, B.L. Allman, G.D. Marsh, and C.L. Rice Muscle Size, Strength, and Bone Geometry in the Upper Limbs of Young and Old Men J. Gerontol. A Biol. Sci. Med. Sci., July 1, 2002; 57(7): M455 - 459. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
), midcortical (
), and endosteal (
) subregions in females in each age group. SDs are listed in 
