Interpretation of whole body dual energy X-ray absorptiometry measures in children: comparison with peripheral quantitative computed tomography

Bone 34 (2004) 1044 – 1052 www.elsevier.com/locate/bone

Interpretation of whole body dual energy X-ray absorptiometry measures in children: comparison with peripheral quantitative computed tomography Mary B. Leonard, a,b,* Justine Shults, b Dawn M. Elliott, c,d Virginia A. Stallings, a and Babette S. Zemel a a

Children’s Hospital of Philadelphia and the Department of Pediatrics, University of Pennsylvania, Philadelphia, PA 19104, USA b Department of Epidemiology and Biostatistics, University of Pennsylvania, Philadelphia, PA 19104, USA c Department of Orthopaedics, University of Pennsylvania, Philadelphia, PA 19104, USA d Department of Bioengineering, University of Pennsylvania, Philadelphia, PA 19104, USA Received 20 January 2003; revised 15 October 2003; accepted 3 December 2003

Abstract The assessment of bone health in children requires strategies to minimize the confounding effects of bone size on dual energy X-ray absorptiometry (DXA) areal bone mineral density (BMD) results. Cortical bone composes 80% of the total skeletal bone mass. The objective of this study was to develop analytic strategies for the assessment of whole body DXA that describe the biomechanical characteristics of cortical bone across a wide range of body sizes using peripheral quantitative computed tomography (pQCT) measures of cortical geometry, density (mg/mm3), and strength as the gold standard. Whole body DXA (Hologic QDR 4500) and pQCT (Stratec XCT-2000) of the tibia diaphysis were completed in 150 healthy children 6 – 21 years of age. To assess DXA and pQCT measures relative to age, body size, and bone size, gender-specific regression models were used to establish z scores for DXA bone mineral content (BMC) for age, areal BMD for age, bone area for height, bone area for lean mass, BMC for height, BMC for lean mass, and BMC for bone area; and for pQCT, bone crosssectional area (CSA) for tibia length and bone strength (stress-strain index, SSI) for tibia length. DXA bone area for height and BMC for height were both strongly and positively associated with pQCT CSA for length and with SSI for length (all P < 0.0001), suggesting that decreases in DXA bone area for height or DXA BMC for height represent narrower bones with less resistance to bending. DXA BMC for age (P < 0.01) and areal BMD (P < 0.05) for age were moderately correlated with strength. Neither DXA bone area for lean mass nor BMC for lean mass correlated with pQCT CSA for length or SSI for length. DXA BMC for bone area was weakly associated with pQCT SSI for length, in females only. Therefore, normalizing whole body DXA bone area for height and BMC for height provided the best measures of bone dimensions and strength. DXA BMC normalized for bone area and lean mass were poor indicators of bone strength. D 2003 Elsevier Inc. All rights reserved. Keywords: Quantitative computed tomography; DXA; Bone mineral density; Child; Cortical bone

Introduction The increase in bone strength throughout growth and development is the result of changes in bone geometry and bone mineral density (BMD) [30,35]. As recently reviewed [30], BMD can be expressed within three different levels of biological organization. Material density (BMDmaterial) is the * Corresponding author. Children’s Hospital of Philadelphia and the Department of Pediatrics, University of Pennsylvania, CHOP North, Room 1564, 34th Street and Civic Center Boulevard, Philadelphia, PA 19104. Fax: +1-215-590-0604. E-mail address: [email protected] (M.B. Leonard). 8756-3282/$ - see front matter D 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.bone.2003.12.003

amount of mineral divided by the volume of the bone matrix, excluding marrow spaces, canals, and lacunae. Compartment density (BMDcompartment), also known as apparent density, is the amount of mineral divided by the volume of the trabecular or cortical compartment, including marrow spaces, canals, and lacunae. Total density (BMDtotal) is the amount of mineral divided by the volume enclosed by the periosteal bone surface. Studies of bone mineral accretion in children should recognize the strengths and limitations of imaging techniques in characterizing changes in bone size and the different components of BMD during growth. BMDmaterial cannot be assessed with currently available noninvasive

M.B. Leonard et al. / Bone 34 (2004) 1044–1052

techniques. Trabecular and cortical BMDcompartment and BMDtotal can be measured with quantitative computed tomography (QCT) [30,35]. Numerous recent QCT studies have been reported in the vertebrae [14,15] and in long bones, such as the femur [14,15], radius [25,26,29,33,34], and tibia [5,24] in healthy children. These studies have characterized gender, maturation, and ethnic-specific differences in cortical dimensions, trabecular, and cortical BMDcompartment and BMDtotal. QCT studies have yielded important insights into the distinct components of bone mineral accretion in children; however, dual energy X-ray absorptiometry (DXA) remains, by far, the most common and convenient method for the assessment of bone mass in research studies and in clinical applications in children and adults. Due to the inherent limitations of a two-dimensional projection technique, BMDcompartment and BMDtotal cannot be measured by DXA. Trabecular BMDcompartment cannot be measured by DXA due to inability to isolate the trabecular compartment from the surrounding cortical bone. Cortical BMDcompartment and BMDtotal cannot be measured by DXA due to inability to measure the cortical compartment volume and total bone volume, respectively. Geometric algorithms have been developed to estimate BMDtotal and cortical BMDcompartment in skeletal sites with relatively simple geometry [36]. For example, BMDtotal in the vertebrae may be estimated from the projected dimensions using formulas based on the assumption that the vertebrae represent an elliptical cylinder or cube [7,19]. Similarly, BMDtotal and cortical BMDcompartment may be estimated in a discrete length of the femoral midshaft, assuming the shaft represents a hollow cylinder of constant cortical wall thickness [3,36]. Unfortunately, these approaches care not readily applied to the complex shape of the whole skeleton and the biomechanical significance of BMC relative to bone area across the entire skeleton is not known. Multiple sources of pediatric DXA reference data are now available for the calculation of whole body bone z

1045

scores (Table 1). These include gender-specific centile curves, age- and height-specific means and standard deviations, and z score prediction models [5,10,11,18,22,23,41]. Despite the recent widespread availability of whole body reference data, there is lack of consensus regarding the most appropriate strategy for the interpretation of two-dimensional whole body DXA BMC and bone area results across children of differing body size and body composition. Proposed strategies include assessing bone area relative to height and BMC relative to bone area [23], assessing BMC relative to height and age [10], or assessing BMC relative to body weight or lean mass [8,16,39]. No studies have compared these approaches to a three-dimensional imaging technique. Cortical bone composes 80% of the skeletal bone mass; therefore, whole body DXA BMC and area reflect predominantly cortical bone mass and dimensions. The primary function of cortical bone is mechanical strength. We hypothesize that measures of DXA whole body BMC and bone area relative to height will correlate with bone strength, while DXA whole body BMC and bone area relative to body weight and BMC relative to bone area will provide poor estimates of bone strength. The objective of this study was to develop analytic strategies for the assessment of whole body DXA that describe the biomechanical characteristics of cortical bone across a wide range of body sizes. Peripheral QCT measures of stress-strain index (SSI) in the tibia will be used as the gold standard measure of bone strength. SSI incorporates cortical BMDcompartment and cortical endosteal and periosteal dimensions in a summary measure of strength that has been validated in human cadaver studies and animal studies [1,12]. The comparison of DXA measures with peripheral quantitative computed tomography (pQCT) will provide a criterion for identifying the optimal approach for interpreting whole body DXA scans as measures of cortical bone strength in children.

Table 1 Whole body DXA reference data in children Reference

Location

n

Age (year)

Faulkner et al. [11]

Canada

Molgaard et al. [23]

Denmark

Maynard et al. [22]

USA

Ellis et al. [10]

USA

Hannan et al. [17]

Scotland

506 471 201 142 228 237 537 445 216

F, M F, M F, M F, M F

Binkley et al. [5]

USA

van der Sluis et al. [41]

Netherlands

124 107 256 188

F, M F, M

8 – 17 5 – 19 8 – 18 5 – 18 11 – 18 5 – 22 4 – 20

DXA

Reference data

Hologic QDR 2000 Hologic QDR 1000 Lunar DPX

Gender-specific mean and SD for BMC for age and areal BMD for age

Hologic QDR 2000 Hologic QDR 1000 Hologic QDR 4500 Lunar DPX

Gender-specific mean and SD for bone area for age, BMC for age, bone area for height, and BMC for bone area Gender-specific mean and SD for BMC for age and areal BMD for age Gender-specific prediction model for BMC for height, age, and ethnicity Female prediction equations for BMC for age; BMC for weight, height, age, and shoulder width; and BMC for weight, height, and age Gender-specific LMS centile curves for BMC for age and areal BMD for age Gender-specific mean and SD for BMC for age and areal BMD for age

1046

M.B. Leonard et al. / Bone 34 (2004) 1044–1052

Methods Study subjects Children and adolescents, ages 6 –21 years, were enrolled as healthy controls for ongoing bone studies in the Nutrition and Growth Laboratory at the Children’s Hospital of Philadelphia (CHOP). Subjects were recruited from the CHOP general pediatric clinics, as well as from the surrounding community using newspaper advertisements and flyers. Subjects with chronic medical conditions or medications possibly affecting growth, pubertal development, nutritional status, or dietary intake were excluded. The Institutional Review Board approved the protocol. All subjects and parents provided written informed consent. Anthropometry All measurements were performed in the Nutrition and Growth Laboratory. The measurements consisted of weight (kg), using a digital electronic stand-on scale, and height (cm), using a wall-mounted stadiometer following the methods described by Lohman et al. [21]. Age- and gender-specific z scores (standard deviation scores) for height, weight, and body mass index (BMI, kg/m2) were calculated using the National Center for Health Statistics (NCHS) 2000 Center for Disease Control growth data [28]. Pubertal status was determined by physical examination at the time of the study visit and classified according to the method of Tanner [37]. DXA whole body scans Whole body DXA (Hologic Inc., Bedford, MA) was performed with a fan beam in the array mode using standard positioning techniques. All subjects were wearing hospital scrubs and any interfering objects such as jewelry were removed. Quality control scans were performed daily using a simulated L1 – 4 lumbar spine made of hydroxyapatite encased in epoxy resin. In our institution, the in vitro coefficient of variation was less than 0.6% and the in vivo coefficient of variation in adults was less than 1%. The DXA whole body scans were analyzed to generate measures of whole body projected bone area (cm2), BMC (gm), and areal BMD (gm/cm2). Previous studies reported that pediatric whole body data were confounded by variability in relative skull size [38]. Therefore, all whole body DXA results presented in this study represent the bone area, BMC, and areal BMD excluding the skull. Estimates of lean mass (kg) and fat mass (kg) were obtained from the whole body DXA scan excluding the skull. Peripheral QCT of the Tibia The pQCT (XCT 2000; Stratec, Inc., Pforzheim, Germany) measures of cortical bone were performed in the

midshaft of the left tibia. A customized segmometer (Rosscraft, Blain, WA) was used to measure tibia length [21]. A scout view was completed to position the scanner at the site on the tibia at which the distance to distal tibia endplate corresponded to 20% of tibia length. A single tomographic slice of 2.3 mm transectional thickness was taken at a voxel size of 0.4 mm. Image processing and the calculation of numerical values were performed using the manufacturer’s software package, version 5.40, with the following analysis parameters: contour mode 1, peel mode 1, and bone threshold 711 mg/mm3. The following measures were obtained at the 20% site: BMDtotal (bone mineral content divided by the volume enclosed within the periosteal envelope, mg/mm3), total cross-sectional area (total CSA, mm2, the area contained within the periosteal border), cortical CSA (mm2, the area between the periosteal and endosteal border), cortical thickness (mm), and BMC per mm slice (mg/mm). To minimize the impact of partial volume effect on measures of cortical BMDcompartment, a discrete region of interest (2
2 voxels in size) was manually defined between the cortical periosteal and endosteal bone envelope. Cortical BMDcompartment (mg/mm3) was measured within this region of interest. The geometric variable section modulus (mm3) was calculated as [S(d2
A) / dmax], where d is the distance of the voxel from the center of gravity, A is the cross-sectional area of a voxel, and dmax is the maximum distance of any of the voxels in the cortical cross section from the center of gravity. The integrated product of the section modulus and cortical BMDcompartment provided the stress-strain index (SSI, mm3), a measure of fracture load in bending or torsion [31]. These measures have been validated in biomechanical studies, demonstrating very high correlations (R > 0.90) between the pQCT measures and fracture load [1,2,13,31]. Fig. 1 summarizes the pQCT measurements.

Fig. 1. pQCT cortical bone measures in the diaphysis of the tibia. Total cross-sectional area = area contained within the outer periosteal envelope; Cortical cross-sectional area = area contained between the outer periosteal envelope and inner endosteal envelope; a = area of one voxel in the cortex (mm2); d = distance of one voxel to the center of gravity (mm); dmax = maximum distance (eccentricity) to the center of gravity (mm); CD = Cortical BMDcompartment in one voxel (mg/mm3) ND = normal physiological density (1200 mg/mm 3 ); Stress Strain Index (SSI, P mm3) = i ½ðai
di2 ÞðCDi =NDÞ=dmax .

M.B. Leonard et al. / Bone 34 (2004) 1044–1052 Table 2 Subject characteristics (mean F SD) n

150

Age (year) Range Gender Tanner distribution (n at stages 1, 2, 3, 4, 5) Ethnicity

12.5 F 3.5 6.0 – 21.0 75 M/75 F (39, 16, 21, 50, 24) 54% Caucasian 36% African American 7% Hispanic 3% Other 151.9 F 17.7 111 – 185 0.17 F 1.00  1.8 – 2.5 48.7 F 17.2 17.1 – 88.0 0.43 F 0.95  2.3 – 2.3 20.3 F 4.0 13.0 – 24.9 0.4 F 1.0  2.5 – 2.2 34.3 F 12.5 12.7 – 62.8

Height (cm) Range Height z score Range Weight (kg) Range Weight z score Range BMI (kg/m2) Range BMI z score Range Lean mass (kg) Range

1047

pQCT measures were intentionally not adjusted for age, Tanner stage, or race. This was done to allow direct comparison of the DXA and QCT measures without adjusting away sources of variability. That is, if a particular DXA z score (e.g., BMC for height) is increased in a given child, does that child have stronger bones, regardless of his race, Tanner stage, muscle mass, or other characteristics? In fact, adjusting for other determinants of bone mass may mask these relationships. For example, a Tanner stage 5 child would be expected to have broader bones than a prepubertal child, potentially manifested as a greater DXA bone area for height and a greater pQCT total CSA for tibia length. If the regression models used to generate the z scores were adjusted for Tanner stage, the absolute greater bone breadth in the more mature children would be attenuated. His bone area for height z score would be less because he would be assessed relative to other Tanner stage 5 children of the same height. This would obviate our capability to determine if this child has stronger bones than the prepubertal child, defined according to the pQCT gold standard.

Results Analysis Subject characteristics Analyses were conducted using STATA 7.0 (Stata Corporation, College Station, TX). Two-sided tests of hypotheses were used and a P value <0.05 was considered to be statistically significant. Initial descriptive analyses included means and standard deviations of subject characteristics (Tables 1 and 2). The correlations between bone and anthropometric measures were assessed using Pearson product – moment estimates and identifying significant departures from zero. The relationships between bone measures were explored graphically. Transformations of the outcomes or explanatory variables were used to improve the linearity of relationships and fit of models, as assessed by the R2 values for each model. All DXA and pQCT bone measures were (natural) log transformed, as were tibia length, lean body mass, and height. For models based on age, the use of a second-degree polynomial provided the best fit. To determine if size-adjusted DXA measures within a subject corresponded to similar size-adjusted pQCT measures, regression models were constructed for each relationship and z scores (standardized residuals) from the models were computed. The z scores represent the residual divided by its standard error. The assumptions of linearity and constant variance for all regression models were assessed via graphical checks; the assumption of normality of the residuals was assessed via the Shapiro –Wilk test for normality. The pQCT measure of bone strength was normalized by length (e.g., SSI for tibia length) to take into account the moment arm for bending or shaft length for torsion. The regression models were gender specific, consistent with all the published reference data sources summarized in Table 1. The univariate regression models for the DXA and

A total of 150 healthy children and adolescents (75 female), 6– 21 years of age, were enrolled. The females were older than the males (females 13.4 F 3.1, males 11.6 F 3.5; P < 0.01) and were more mature ( P < 0.01). The subjects were similar in height to the national reference data. The mean weight and BMI z scores were significantly greater than zero, consistent with the high prevalence of overweight in the general population. The subject characteristics are summarized in Table 2. DXA and pQCT results The DXA and pQCT results are summarized in Table 3. All body size-dependent variables (height, weight, fat mass, Table 3 Whole body DXA and Tibia pQCT results in 150 healthy children and adolescents

Whole body DXA Bone area (cm2) BMC (gm) Areal BMD (gm/cm2) Tibia pQCT Total CSA (mm2) Cortical CSA (mm2) Cortical thickness (mm) BMC (mg/mm) BMDcompartment (mg/mm3) BMDtotal (mg/mm3) SSI (mm3)

Mean

SD

Median

Range

1354 1153 0.81

434 530 0.14

1408 1149 0.81

474 – 2332 271 – 2599 0.54 – 1.22

315 165 3.12 179 1182 631 973

87 49 0.66 57 61 83 410

318 170 3.15 182 1189 628 967

150 – 571 75 – 323 1.58 – 5.36 71 – 319 1001 – 1304 415 – 876 275 – 1970

1048

M.B. Leonard et al. / Bone 34 (2004) 1044–1052

Fig. 2. DXA whole body bone area-for-height is positively associated with bone strength in males and females.

lean mass, tibia length, DXA bone area, BMC and areal BMD, pQCT total CSA, cortical CSA, BMC, and SSI) were highly and positively correlated (all R > 0.80; P < 0.0001). The confounding effect of bone size on DXA areal BMD measures was demonstrated by the highly significant associations of DXA whole body BMD with height (R = 0.86; P < 0.0001) and with bone breadth (pQCT tibia crosssectional area) (R = 0.80; P < 0.0001). In contrast, the size independence of pQCT density measures was illustrated but the lack of correlation between cortical BMDcompartment and tibia total CSA and between cortical BMDtotal and tibia total CSA. Bone size relative to body size Regression models were generated for DXA bone area relative to height (females: R2 0.94; males: R2 0.96; both P < 0.0001), for pQCT tibia total CSA relative to tibia length (females: R2 0.67; males: R2 0.79; both P < 0.0001), and for pQCT tibia SSI relative to tibia length (females: R2 0.73; males: R2 0.85; both P < 0.0001). To determine if increased DXA bone area for height represented greater cortical bone width (i.e., broad bones), z scores for DXA bone area for height and pQCT total CSA for bone length were generated from the regression equations. Greater DXA bone area for

Fig. 3. DXA whole body BMC-for-height is positively associated with bone strength in males and females.

height was significantly and positively correlated with greater total CSA for tibia length (R = 0.41 in males and 0.61 in females). Fig. 2 illustrates that greater DXA bone area for height was significantly and positively correlated with greater bone SSI for tibia length (R = 0.56 in males and 0.74 in females). Therefore, increased whole body bone area for height represented increased bone breadth and strength relative to bone length. Prior authors have proposed using body weight or lean mass to normalize DXA whole body bone area for body size [16]. Z scores for whole body bone area for body weight and for whole body bone area for lean mass were generated and did not correlate with pQCT measures of bone breadth or strength (Table 4). Measures of BMC relative to bone size DXA bone area and patient height both provide estimates of bone size. DXA BMC was evaluated relative to each of these measures of bone size. Regression models were generated for DXA BMC relative to height (females: R2 0.90; males: R2 0.95; both P < 0.0001) and DXA BMC relative to bone area (females: R2 0.98; males: R2 0.99; both P < 0.0001).

Table 4 Univariate correlation coefficients (R) between DXA z scores and pQCT measures of bone strength pQCT SSI for length z score

DXA whole body z scores BMC for age

BMD for age

Bone area for height

BMC for height

Bone area for weight

BMC for weight

Bone area for lean mass

BMC for lean mass

BMC for bone area

Males

0.31, P < 0.01 0.31, P < 0.01

0.30, P < 0.05 0.36, P < 0.01

0.56, P < 0.0001 0.74, P < 0.0001

0.57, P < 0.0001 0.70, P < 0.0001

NS

NS

NS

NS

NS

NS

NS

NS

NS

0.28, P < 0.05

Females NS, nonsignificant.

M.B. Leonard et al. / Bone 34 (2004) 1044–1052

Greater DXA BMC for height was significantly and positively correlated with greater total CSA for tibia length (R = 0.36 in males and 0.46 in females). Fig. 3 illustrates that DXA BMC for height was significantly and positively correlated with bone SSI for tibia length (R = 0.57 in males and 0.70 in females). Therefore, increased whole body BMC for height represented increased bone breadth and strength relative to bone length. In contrast, DXA whole body BMC for bone area was weakly correlated with pQCT SSI for tibia length in females (R = 0.28) and was not correlated with pQCT SSI for tibia length in males, as shown in Fig. 4. DXA whole body BMC for bone area was not associated with pQCT BMDcompartment. Whole body BMC for bone area was correlated with BMDtotal (males R = 0.46, females R = 0.51; both P < 0.0001); however, BMDtotal was not correlated with SSI for tibia length. Again, investigators have proposed using body weight or lean mass to normalize DXA whole body BMC for body size [8,16,39]. Z scores for whole body BMC for body weight and for whole body BMC for lean mass were generated and did not correlate with pQCT measures of bone breadth or strength (Table 4). DXA BMC and areal BMD relative to age and gender Most available whole body DXA reference data sets generate percentiles or z scores of BMC and areal BMD for age and gender (Table 1). Regression models were generated for DXA BMC relative to age (females: R2 0.82; males: R2 0.87; both P < 0.0001) and for DXA areal BMD relative to age (females: R2 0.71; males: R2 0.77; both P < 0.0001). Greater DXA BMC and greater areal BMD for age were both significantly correlated with greater bone SSI for tibia length (BMC: R = 0.31 in males and 0.33 in females; BMD: R = 0.30 in males and 0.36 in females).

Fig. 4. DXA whole body BMC-for-bone area is weakly associated with bone strength-for-length in females only.

1049

However, these measures assessed relative to age were not as highly correlated with bone strength as DXA measures expressed relative to height, as summarized in Table 4.

Discussion The new data presented here demonstrated that whole body DXA measures of bone area for height and BMC for height provide the best estimates of cortical bone dimensions and strength in males and females across a broad range of body sizes. These measures provide better estimates of bone strength than measures expressed relative to age. Furthermore, DXA whole body BMC for bone area was weakly correlated with bone strength in females, and no correlation was observed in males. Cortical bone fulfills a predominantly mechanical function. Cortical bone strength depends on three factors: bone geometry, bone material (density and elastic modulus), and the location and direction of applied loads [40]. The geometric parameter section modulus is a function of the endosteal and periosteal bone dimensions and is proportional to bone strength [13]. The schematic in Fig. 5 illustrates the importance of bone dimensions in establishing bone strength. Two bones of identical mass and BMDcompartment demonstrate marked differences in strength as a result of the redistribution of the bone mass relative to the central axis. The larger bone is three times as strong in bending despite lower areal BMD and lower BMDtotal. BMDtotal is known to be a poor correlate of bone strength [32]. This is consistent with our observation that BMDtotal did not correlate with SSI for tibia length and that BMDtotal did not correlate with bone size. Of note, our primary outcome measure, SSI for tibia length, reflects long bone strength but does not reflect the strength of cancellous bone near joints. Trabecular bone BMDcompartment is a better measure of cancellous bone strength. Gender-specific reference data for the calculation of whole body bone area for height and whole body BMC for bone area have been published [23] and applied to healthy children and children with chronic diseases [20,27,42]. The authors proposed that decreased bone area relative to height represented ‘‘narrow’’ bones; this interpretation was confirmed here. However, the authors also proposed that decreased whole body bone BMC for bone area represented ‘‘light’’ bones. Our data demonstrated that decreased BMC for bone area was correlated with decreased BMDtotal; however, BMDtotal was not correlated with strength and provided a poor estimate of bone mass and bone size relationships as related to strength. Therefore, we do not believe that whole body bone BMC for bone area provides additional information regarding bone health. The assessment of whole body BMC relative to bone area is problematic for several reasons. First, as noted above, bone area determined by DXA is a two-dimensional projection and is not a measure of the depth or thickness of

1050

M.B. Leonard et al. / Bone 34 (2004) 1044–1052

Fig. 5. Effect of bone geometry on strength and BMD. In two bones with identical bone mineral content per unit length and identical BMDcompartment, the bone with the larger diameter has three-fold greater resistance to bendng. Note that BMDtotal is smaller in the stronger bone.

bone. This is a particular problem for the whole body DXA measure because it is a composite of many tubelike (e.g., limbs) and broad (e.g., pelvis) structures that vary in depth and thickness. Therefore, bone area is a poor measure of the volume of bone over which the BMC is distributed. Second, because of the two-dimensional nature of the measure, bone area is a function of subject height and bone width. In a child with narrow bones for height, assessment of BMC for bone area will result in the comparison of that child with a significantly shorter child of comparable bone area, reflecting differences in bone length, not width. Thirdly, in the event of increased cortical bone dimensions, the section modulus (and hence, bone strength) may increase while the BMC relative to the larger projected area decreases (Fig. 5). Bone strength increases as periosteal dimension to the third power [40]. Even very small increases in bone dimensions may result in profound increases in strength without manifesting an increased DXA areal BMD, BMC for bone area, or BMDtotal [4]. Other investigators have advocated the assessment of bone measures relative to muscle mass to distinguish between primary bone disorders and those that occur secondary to decreased muscle loading. We also found a strong correlation between height and lean body mass (R = 0.94). While DXA measures of BMC for height correlated with bone dimensions and strength, BMC relative to lean mass did not. This apparent discrepancy is due to the confounding effect of subject height. In two subjects of comparable height, the subject with greater BMC likely has broader bones and therefore greater bone strength. In two subjects of comparable lean body mass, the subject with greater BMC could be taller or could have broader bones. These would have different consequences for bone strength. Therefore, BMC adjusted for height is a better indicator of bone health. While it is unusual to compare whole body bone measures with tibia measures, these results provide important

insight into the interpretation of whole body DXA scans with regards to cortical bone health. The use of z scores provided measures of bone dimensions that accounted for the confounding effects of body size. Analysis of these residuals demonstrated that DXA BMC for height and DXA bone area for height were most strongly associated with the independent measure bone strength, as measured by pQCT. In addition, the values for cortical BMDcompartment reported here are comparable with prior QCT and pQCT studies in the femur, tibia, and radius and with autopsy studies of the relative weight of ashen femoral midshaft cortex [5,14,15,24– 26,29,30,33,34]. Therefore, DXA whole body measures provide valuable insight into overall cortical bone mass and structure. A potential limitation of the DXA analyses proposed here is magnification error. DXA manufacturers have largely converted to fan-beam techniques that introduce magnification errors in measures of bone area and BMC [9]. In the case of Hologic scanners, the apex of the fan beam is under the patient. Therefore, as the bone is elevated a greater distance off the table, bone area and BMC are underestimated to a comparable degree [6]. These effects may have important implications for the assessment of bone area and BMC relative to height, as proposed in our study. Projection error will result in the impression of smaller BMC and area for height in those subjects with greater ‘thickness.’ Furthermore, within our data set, regression analyses demonstrated that whole body BMC increased as whole body bone area to the 1.4 power. Therefore, while projection errors that are comparable in magnitude in BMC and area will have a negligible impact on the ratio of BMC or bone area, BMC relative to bone area will appear decreased. In conclusion, we propose that whole body bone area and BMC should be assessed relative to height in children. DXA BMC and areal BMD relative to age provide less information about bone strength in healthy children. This

M.B. Leonard et al. / Bone 34 (2004) 1044–1052

would likely be compounded in chronically ill children with decreased stature for age. In addition, the assessment of DXA data normalized to body weight or lean mass alone does not capture differences in bone density, dimensions, or resistance to bending. Moreover, DXA BMC normalized to bone area provides a poor estimate of cortical bone strength. These findings underscore the importance of identifying appropriate analytic approaches for assessment of bone health in children. Further studies are required to confirm the sensitivity and specificity of this approach through other independent assessments such as fracture risk.

Acknowledgments We greatly appreciate the dedication and enthusiasm of the children and their families who participated in this study. Special thanks to Margarita Gomelsky, Susan Kaup, and the staff of the General Clinical Research Center for their efforts collecting these data. This protocol was supported by NIH grants K08-DK02523 (MBL) and 1-R03-DK058200 (MBL), the General Clinical Research Center (M01RR00240), and the Nutrition Center, The Children’s Hospital of Philadelphia.

References [1] Augat P, Reeb H, Claes LE. Prediction of fracture load at different skeletal sites by geometric properties of the cortical shell. J Bone Miner Res 1996;11:1356 – 63. [2] Augat P, Fuerst T, Genant HK. Quantitative bone mineral assessment at the forearm: a review. Osteoporos Int 1998;8:299 – 310. [3] Bass S, Delmas PD, Pearce G, Hendrich E, Tabensky A, Seeman E. The differing tempo of growth in bone size, mass, and density in girls is region-specific. J Clin Invest 1999;104:795 – 804. [4] Beck TJ, Stone KL, Oreskovic TL, Hochberg MC, Nevitt MC, Genant HK, et al. Effects of current and discontinued estrogen replacement therapy on hip structural geometry: the study of osteoporotic fractures. J Bone Miner Res 2001;16:2103 – 10. [5] Binkley TL, Specker BL, Wittig TA. Centile curves for bone densitometry measurements in healthy males and females ages 5 – 22 yr. J Clin Densitom 2002;5:343 – 53. [6] Blake GM. Dual X-ray absorptiometry: a comparison between fan beam and pencil beam scans. Br J Radiol 1993;66:902 – 6. [7] Carter DR, Bouxsein ML, Marcus R. New approaches for interpreting projected bone densitometry data. J Bone Miner Res 1992; 7:137 – 45. [8] Crabtree NJ, Boivin CM, Shaw NJ. Can body size related normative data improve the diagnosis of osteoporosis in children? (Abstract). J Bone Min Res 2001;16(Suppl. 1):S560. [9] Ellis KJ, Shypailo RJ. Bone mineral and body composition measurements: cross-calibration of pencil-beam and fan-beam dual-energy Xray absorptiometers. J Bone Miner Res 1998;13:1613 – 8. [10] Ellis KJ, Shypailo RJ, Hardin DS, Perez MD, Motil KJ, Wong WW, et al. Z score prediction model for assessment of bone mineral content in pediatric diseases. J Bone Miner Res 2001;16:1658 – 64. [11] Faulkner RA, Bailey DA, Drinkwater DT, McKay HA, Arnold C, Wilkinson AA. Bone densitometry in Canadian children 8 – 17 years of age. Calcif Tissue Int 1996;59:344 – 51.

1051

[12] Ferretti JL. Noninvasive assessment of bone architecture and biomechanical properties in animals and humans employing pQCT technology. J Jpn Soc Bone Morphom 1997;7:1 – 10. [13] Ferretti JL. Perspectives of pQCT technology associated to biomechanical studies in skeletal research employing rat models. Bone 1995;17:353S – 64S. [14] Gilsanz V, Kovanlikaya A, Costin G, Roe TF, Sayre J, Kaufman F. Differential effect of gender on the sizes of the bones in the axial and appendicular skeletons. J Clin Endocrinol Metab 1997;82:1603 – 7. [15] Gilsanz V, Skaggs DL, Kovanlikaya A, Sayre J, Loro ML, Kaufman F, et al. Differential effect of race on the axial and appendicular skeletons of children. J Clin Endocrinol Metab 1998;83:1420 – 7. [16] Goulding A, Taylor RW, Jones IE, McAuley KA, Manning PJ, Williams SM. Overweight and obese children have low bone mass and area for their weight. Int J Obes Relat Metab Disord 2000;24: 627 – 32. [17] Hannan WJ, Cowen SJ, Wrate RM, Barton J. Improved prediction of bone mineral content and density. Arch Dis Child 1995;72:147 – 9. [18] Hannan WJ, Tothill P, Cowen SJ, Wrate RM. Whole body bone mineral content in healthy children and adolescents. Arch Dis Child 1998;78:396 – 7. [19] Kroger H, Vainio P, Nieminen J, Kotaniemi A. Comparison of different models for interpreting bone mineral density measurements using DXA and MRI technology. Bone 1995;17:157 – 9. [20] Laursen E, Molgaard C, Michaelsen K, et al. Bone mineral status in 134 patients with cystic fibrosis. Arch Dis Child 1999;81:235 – 40. [21] Lohman TG, Roche AF, Martorell R. Anthropometric standardization reference manual. Champaign (IL): Human Kinetics Books; 1988. [22] Maynard LM, Guo SS, Chumlea WC, Roche AF, Wisemandle WA, Zeller CM, et al. Total-body and regional bone mineral content and areal bone mineral density in children aged 8 – 18 y: the Fels Longitudinal Study. Am J Clin Nutr 1998;68:1111 – 7. [23] Molgaard C, Thomsen BL, Prentice A, Cole TJ, Michaelsen KF. Whole body bone mineral content in healthy children and adolescents. Arch Dis Child 1997;76:9 – 15. [24] Moyer-Mileur L, Xie B, Ball S, Bainbridge C, Stadler D, Jee WS. Predictors of bone mass by peripheral quantitative computed tomography in early adolescent girls. J Clin Densitom 2001;4:313 – 23. [25] Neu CM, Manz F, Rauch F, Merkel A, Schoenau E. Bone densities and bone size at the distal radius in healthy children and adoloescents: a study using peripheral quantitative computed tomography. Bone 2001;28:227 – 32. [26] Neu CM, Rauch F, Manz F, Schoenau E. Modeling of cross-sectional bone size, mass and geometry at the proximal radius: a study of normal bone development using peripheral quantitative computed tomography. Osteoporos Int 2001;12:538 – 47. [27] Nysom K, Molgaard C, Michaelsen KF. Bone mineral density in the lumbar spine as determined by dual-energy X-ray absorptiometry. Comparison of whole-body scans and dedicated regional scans. Acta Radiol 1998;39:632 – 6. [28] Ogden CL, Kuczmarski RJ, Flegal KM, Mei Z, Guo S, Wei R, et al. Centers for Disease Control and Prevention 2000 growth charts for the United States: improvements to the 1977 National Center for Health Statistics version. Pediatrics 2002;109:45 – 60. [29] Rauch F, Neu C, Manz F, Schoenau E. The development of metaphyseal cortex – implications for distal radius fractures during growth. J Bone Miner Res 2001;16:1547 – 55. [30] Rauch F, Schoenau E. Changes in bone density during childhood and adolescence: an approach based on bone’s biological organization. J Bone Miner Res 2001;16:597 – 604. [31] Schiessl H, Ferretti JL, Tsarczyk-Niemeyer G, Willnecker J. Noninvasive bone strength index as analyzed by peripheral quantitative computed tomography. In: Schonau E, editor. Pediatric osteology: new developments in diagnostics and therapy. Amsterdam: Elsevier; 1996. p. 141 – 6. [32] Schoenau E, Neu CM, Beck B, Manz F, Rauch F. Bone mineral

1052

[33]

[34]

[35] [36]

[37]

M.B. Leonard et al. / Bone 34 (2004) 1044–1052 content per muscle cross-sectional area as an index of the functional muscle-bone unit. J Bone Miner Res 2002;17:1095 – 101. Schoenau E, Neu CM, Rauch F, Mantz F. The development of bone strength at the proximal radius during childhood and adolescence. J Clin Endocrinol Metab 2001;86:613 – 8. Schonau E, Neu M, Mokov E, Wassmer G, Manz F. Influence of puberty on muscle area and cortical bone area of the forearm in boys and girls. J Clin Endocrinol Metab 2000;85:1095 – 8. Seeman E. From density to structure: growing up and growing old on the surfaces of bone. J Bone Miner Res 1997;12:509 – 21. Sievanen H, Kannus P, Nieminen V, Heinonen A, Oja P, Vuori I. Estimation of various mechanical characteristics of human bones using dual energy X-ray absorptiometry: methodology and precision. Bone 1996;18:17S – 27S. Tanner JM. Growth at adolescence. Oxford: Blackwell Scientific Publication; 1962.

[38] Taylor A, Konrad PT, Norman ME, Harcke HT. Total body bone mineral density in young children: influence of head bone mineral density. J Bone Miner Res 1997;12:652 – 5. [39] Tothill P, Hannan WJ. Bone mineral and soft tissue measurements by dual-energy X-ray absorptiometry during growth. Bone 2002;31: 492 – 6. [40] Turner CH. Basic biomechanical measurements of bone: a tutorial. Bone 1993;14:595 – 608. [41] van der Sluis IM, de Ridder MA, Boot AM, Krenning EP, de Muinck Keizer-Schrama SM. Reference data for bone density and body composition measured with dual energy X-ray absorptiometry in white children and young adults. Arch Dis Child 2002; 87:341 – 7. [42] Weiler HA, Janzen L, Green K, Grabowski J, Seshia MM, Yuen KC. Percent body fat and bone mass in healthy Canadian females 10 to 19 years of age. Bone 2000;27:203 – 7.