Osteoporosis International

, Volume 20, Issue 8, pp 1337–1346

A reference database for the Stratec XCT-2000 peripheral quantitative computed tomography (pQCT) scanner in healthy children and young adults aged 6–19 years


  • R. L. Ashby
    • Clinical Radiology, Imaging Science & Cancer Studies, Stopford BuildingUniversity of Manchester
  • K. A. Ward
    • Clinical Radiology, Imaging Science & Cancer Studies, Stopford BuildingUniversity of Manchester
  • S. A. Roberts
    • Biostatistics Group, School of Epidemiology and Health Sciences, Stopford BuildingUniversity of Manchester
  • L. Edwards
    • Clinical Radiology, Imaging Science & Cancer Studies, Stopford BuildingUniversity of Manchester
    • Faculty of Medicine, School of Health SciencesUniversity of Liverpool
  • M. Z. Mughal
    • Department of Paediatric Medicine, Saint Mary’s Hospital for Women & ChildrenCentral Manchester & Manchester Children’s Hospitals NHS Trust
    • Clinical Radiology, Imaging Science & Cancer Studies, Stopford BuildingUniversity of Manchester
    • Clinical RadiologyManchester Royal Infirmary
Original Article

DOI: 10.1007/s00198-008-0800-2

Cite this article as:
Ashby, R.L., Ward, K.A., Roberts, S.A. et al. Osteoporos Int (2009) 20: 1337. doi:10.1007/s00198-008-0800-2



We have produced paediatric reference data for forearm sites using the Stratec XCT-2000 peripheral quantitative computed tomography scanner. These data are intended for clinical and research use and will assist in the interpretation of bone mineral density and bone geometric parameters at the distal and mid-shaft radius in children and young adults aged between 6–19 years.


Peripheral quantitative computed tomography (pQCT) provides measurements of bone mineral content (BMC), density (BMD) and bone geometry. There is a lack of reference data available for the interpretation of pQCT measurements in children and young adults. The aim of this study was to provide reference data at the distal and mid-shaft radius.


pQCT was used to measure the 4% and 50% sites of the non-dominant radius in a cohort of healthy white Caucasian children and young adults aged between 5 and 25 years. The lambda, mu, sigma (LMS) technique was used to produce gender-specific reference centile curves and LMS tables for calculating individual standard deviations scores.


The study population consisted of 629 participants (380 males). Reference centile curves were produced; total and trabecular BMD for age (distal radius) and for age and height, bone area (distal and mid-shaft radius), cortical area, cortical thickness, BMC, axial moment of inertia, stress–strain index and muscle area (mid-shaft radius).


We present gender-specific databases for the assessment of the distal and mid-shaft radius by pQCT. These data can be used as control data for research studies and allow the clinical interpretation of pQCT measurements in children and young adults by age and height.


AdolescentBone densityBone developmentChildReference valuesTomographyX-ray computed


Peripheral quantitative computed tomography (pQCT) is being increasingly recognised as a valuable tool for the study of healthy skeletal development [110] and for the assessment of paediatric disorders associated with skeletal fragility [1121]. pQCT is designed to measure peripheral skeletal sites such as the radius, tibia and femur. The main advantages of pQCT are that the technique provides volumetric bone mineral density (BMD; milligrams per cubic centimetre) [22] so is not size dependent, in contrast to dual energy X-ray absorptiometry (DXA; grams per square centimetre), which provides areal BMD (BMDa); it also uniquely enables separate measurement of trabecular and cortical bone, thus providing a unique insight into how disease or treatment affects selective bone compartments [23]. At the diaphyses of long bones, measurements such as total area, cortical thickness and cross-sectional muscle area can be obtained, providing information about bone and muscle geometry and allowing assessment of the functional muscle-bone unit [24]. Biomechanical properties of long bones, such as axial moment of inertia (AMI) and strength strain indices (SSI) [25], which are related to bending and torsional strength respectively, of these tubular bones [26], can be estimated. Furthermore, pQCT scans utilise a very low radiation dose (0.43 μSv per slice [27], which is less than 2 h natural background radiation (NBR) in the United Kingdom (UK) [28]) and avoid systemic radiation [29]. These features make pQCT an ideal densitometric method for assessing peripheral (appendicular) skeletal and muscle parameters in children.

With increasing use of pQCT in the assessment of children with bone disorders, there is a need for reference data derived from healthy children [30]. Currently, there is a paucity of such reference data using this technique in children and young adults. The scanner manufacturer supplies reference data for the distal radius metaphysis (distal radius) taken from a population of healthy white Caucasian German children and adults (n ≈ 371, aged 5–23 years) [5, 6, 8]. Reference data for the distal tibia are available from a cohort of children and young adults [1, 31]. There are currently no published reference data available for cortical bone geometric parameters of the mid-radial diaphysis (mid-shaft radius). Therefore, the aim of the current study was to provide gender-specific reference centile curves for bone parameters at the distal and mid-shaft sites of the radius in healthy white Caucasian children and young adults.

Materials and methods


The total study population consisted of 629 white Caucasian children and young adults aged between 5 and 25 years. This population was an extended cohort of children described previously and from which reference data were provided for the Hologic QDR 4500 Discovery DXA scanners; participant recruitment, inclusion and exclusion criteria were similar [32]. In brief, participants were recruited by means of advertisement from general practitioner surgeries and educational establishments within Greater Manchester in the UK. Children and young adults aged between 5 and 25 years and who were of white Caucasian ethnic origin were eligible to take part in the study. Those taking low-dose inhaled glucocorticoids were eligible to be included within the study [18]. Children who suffered from systemic disease, were taking medications known to affect bone health, had suffered recurrent low trauma fractures or long periods of immobilisation over the past 12 months, or had been pregnant, were not included in the study. Other exclusion criteria are described elsewhere [32].

Ethical approval was obtained from the North West Multi-centre Research Ethics Committee (MREC 04/8/006). This study was conducted according to the Declaration of Helsinki. Verbal consent to participate in this study was obtained from each participant. Informed written consent was obtained from each participant over 16 years of age and from parents or guardians of participants under 16 years.

Socio-economic status

In a subset of children, the Townsend Material Deprivation score (TMDS) [33] was used as an indicator of socio-economic status. The TMDS was determined from research participants’ postcodes and data from the UK 2001 Census [34].

Anthropometric measurements

Participant’s height (centimetre) was measured to the nearest 0.1 cm using a wall-mounted stadiometer (Leicester Height Measure, Child Growth Foundation, UK). Weight (kilogram) was measured to the nearest 0.1 kg using Seca digital scales (Autoweigh Scales, UK). These measurements were taken with participants in light clothes, without shoes and coat. Body mass index (BMI; kilograms per square metre) was calculated (weight/height2). Standard deviation scores (SDS; also referred to as Z scores) for height, weight and BMI for each participant were calculated from UK reference data [35, 36] (Table 1). Forearm length (millimetre) was measured using a flexible tape measure (Sunlight Medical, Tel Aviv, Israel). Measurements were undertaken with the participant seated, with the elbow resting on a flat surface (table top) and flexed at 90° with forearm in mid-position. Forearm length was defined as the distance from the distal tip of the ulna styloid process to the olecranon (table top).
Table 1

Descriptive characteristics of the study population for height, weight and body mass index (BMI)

Descriptive characteristics



Mean ± SD


Mean ± SD


Age (years)

14.9 ± 4.9

12.7 ± 3.8

Height (cm)

161.5 ± 19.8

0.39 ± 0.92

150.0 ± 16.5

0.23 ± 0.97

Weight (kg)

56.4 ± 20.5

0.59 ± 0.99

46.4 ± 15.8

0.39 ± 1.01

BMI (kg/m2)

20.7 ± 4.0

0.47 ± 1.05

19.9 ± 3.9

0.36 ± 1.03

Standard deviation scores (SDS) are calculated from United Kingdom reference centile curves for height, weight and BMI [35, 36]

Bone densitometry assessment

Two XCT-2000 pQCT scanners (Stratec Medizintechnik GmbH, Pforzheim, Germany) were used to collect these data, one older static scanner located in the department and a more modern scanner located in a mobile research unit. The latter also houses a Hologic QDR 4500 Discovery DXA scanner, enabling scanning to be performed off site at schools, which enhanced recruitment to the study. Two adjacent computed tomography sections were performed at the distal metaphysis (4% distal radius) and one at the mid-shaft diaphysis (50% mid-shaft radius). One scanner, which was purchased in the early days of pQCT, has a narrower slice width (1.2 mm) than the more modern scanner with a slice width of 2.0 mm, which is now the routine slice width of such pQCT scanners. Consequently, each section was taken at a voxel size of either 0.4 × 0.4 × 1.2 mm or 0.4 mm × 0.4 mm × 2 mm. Scan speed was 25 mm/s−1. The resulting cross-sectional images were analysed using manufacturer’s software (version 5.5d). We tested differences between the two scanners by taking measurements of the European Forearm Phantom and healthy volunteers (n = 29), and the root mean square error between the scanners was compared to the precision error of the scanners. The differences between the scanners were less than the precision error for all the variables presented in this manuscript, so no adjustments were judged to be necessary between scanners to pool the results. We verified this decision with the scanner manufacturer Stratec Medizintechnik GmbH, Pforzheim, Germany (Dr Johannes Willnecker—personal communication) and also independently with Dr Klaus Engelke, a CT expert from the University of Erlangen (Germany).

Distal radius

Scans at the distal radius were performed according to manufacturer’s instructions and the method described for measurements in children [5, 8]. The distal forearm was positioned pronated in the scanner, and a scout scan was performed. If the growth plate was visible, the reference line was positioned to bisect the medial border of the distal metaphysis (Fig. 1A). If the growth plate had fused, the reference line was placed to bisect the medial border of the articular surface of the radius (Fig. 1B). Scans were performed at sites which corresponded to 3.6% and 4% of forearm length from the reference line, proximal to the distal radius metaphysis. Scans which were performed at the 3.6% site were not included in this study. The CALCBD analysis algorithm, contour mode 2, peel mode 1, was used to obtain measurements of total bone mineral density (total BMD; milligrams per cubic centimetre), trabecular bone mineral density (trabecular BMD; milligrams per cubic centimetre) and cross-sectional bone area (bone area; square millimetre).
Fig. 1

Scout scan showing the correct positioning of the reference line (A) and measurement line (B) when the growth plate is (i) open (bisecting the medial border) and (ii) closed (bisecting the medial border of the distal endplate of the distal metaphysic)

Mid-shaft radius

The arm was positioned manually in the scanner at the site which corresponded to 50% of forearm length. The resulting cross-sectional slice was analysed using the CORTBD analysis algorithm, separation mode 1, analysis threshold 710 mg/cm3 to determine cortical bone area (cortical area; square millimetre), cortical thickness (millimetre), cortical bone mineral content (cortical BMC; milligrams per millimetre), axial moment of inertia (AMI; mm4) and cross-sectional bone area (bone area; square millimetre); separation mode 1, threshold 480 mg/cm3 was used to determine the polar SSI (cubic millimetre). The SSI is calculated from the section modulus and cortical BMD and provides an estimate of bone torsional and bending strength [25]. The polar SSI provides an estimate of torsional bone strength, and the axial SSI can be used to predict bending bone strength in relation to the X or Y axis. As the axial SSI measurements may be influenced by forearm rotation, we provide measurements for polar SSI, as this measurement is independent of rotation and has better reproducibility than axial SSI [37]. For bending strength we have provided a measure of axial moment of inertia.

The total time taken to perform all three pQCT scans (two distal radius and one mid-shaft radius) was between 6 to 8 min. Quality assurance was performed daily using standard manufacturer’s phantoms. The short-term precision (CV%) for the XCT-2000 pQCT scanner in Clinical Radiology for adults (n = 22, two repeat measures) was trabecular BMD 1.27%, total BMD 2.1% and bone area 1.7% at the distal radius and cortical area 2.4% at the mid-shaft radius. The total estimated effective dose equivalent for all pQCT scans (0.43 μSv per slice, 2 at distal 4% site and 1 at 50% site) [27] performed in the current study was 1.29 μSv, which is less than 6 h of average NBR for the UK [28].

Statistical analysis

The LMS method of Cole and Green [38] was used to produce LMS tables and reference centile plots showing the fifth, 25th, 50th, 75th and 95th centiles for all measured parameters in both females and males by age and, where appropriate, by height. At the 4% distal radius site, total BMD, trabecular BMD and bone area were plotted for age, and bone area was also plotted for height. At the 50% mid-shaft radius site cortical area, thickness and BMC, radius and muscle area and polar SSI and AMI were plotted for both age and height. We have described comprehensive methodology for obtaining reference centile curves and associated LMS tables [32]. As there were insufficient participant numbers to fit values at the extremes of the data, the present centile curves and LMS tables cover the age range of 6–19 years only. Similarly, height charts cover the range 120–190 cm. Spearman’s rank-order correlation coefficient was used to investigate correlations between the SDS for height, weight and BMI, age and TMDS. Spearman correlation coefficients are shown together with their associated significance levels.

Calculation of SDS

As an accompaniment to the reference centile curves, LMS tables are provided, which allow calculation of a child’s SDS for distal and mid-shaft radius parameters according to age or height. The SDS can be calculated by obtaining the L, M and S values according to the child’s age/height and obtaining the value of y for the child (e.g. total BMD or cortical thickness according to the child’s age). The SDS corresponds to the y value is given by Eq. 1:
$${\text{SDS}} = \frac{{\left[ {{y \mathord{\left/{\vphantom {y M}} \right.\kern-\nulldelimiterspace} M}} \right]^L - 1}}{{L \times S}}$$


From the original cohort of 636 participants, seven participants were excluded prior to data analysis as they had low distal radial BMD for age or abnormal mid-radial bone geometry and were consequently referred to the Manchester Paediatric Metabolic Bone Disease Clinic for further investigation. From the remaining study population of 629 participants, a further six distal and nine mid-shaft radius scans were excluded due to degradation of scan quality by movement artefacts. Therefore, 623 (376 male) scans at the distal site and 620 (374 male) scans at the mid-shaft site of the radius were included in the chart derivation. Four hundred ninety-nine (270 male) were within the 6–19-year age range presented.

Descriptive characteristics for the study population are shown in Table 1. Males and females were taller, heavier and had greater BMI when compared to UK reference data [35, 36]. The standard deviations of the SDS scores for males and females were close to the expected value of one. Weight (r = 0.10, p = 0.015) and BMI (r = 0.07, p = 0.08) SDS increased with age, with younger children having a mean score close to zero, but height showed no significant association.

The TMDS was determined for the locality of 380 participants. The median score was −0.38, close to the population average, with an interquartile range of −2.5 to 2.7. Forty-three percent of the participants had TMDS greater than zero, and the distribution was skewed towards higher deprivation scores (range −4.8 to 13.9).

Reference centile curves are presented for males and females according to age and height where appropriate. Reference centile curves have been produced for total BMD and trabecular BMD of the distal radius for age (Fig. 2 a and b, respectively) and bone area and polar SSI of the mid-shaft radius (Fig. 3 a and b, respectively) for height. The LMS tables for all the parameters are provided in the Electronic supplementary material, together with plots showing the raw data and bootstrap estimates of the accuracy of the charts.
Fig. 2

Reference centile curves for 4% radius: (a) total bone mineral density (BMD; milligrams per cubic centimetre) and (b) trabecular bone mineral density (BMD; milligrams per cubic centimetre) according to age in males (i) and females (ii)

Fig. 3

Reference centile curves for 50% radius: (a) bone area (square centimetre) and (b) polar stress strain index (SSI; cubic millimetre) in relation to height in males (i) and females (ii)


Interpretation of pQCT measurements in children is hampered by a lack of reference data [30]. Whilst reference data are available for the distal radius gathered from a cohort of German children [5, 6, 8], these data may not be appropriate for the interpretation of pQCT measurements in children from other countries due to differences in bone and body size [39].

For instance, we have informally compared our plots for total and trabecular BMD and bone area of the distal radius to corresponding plots of Rauch et al. [8]. Whilst the age-related changes in these parameters are similar in both datasets and total and trabecular BMD are of similar values, German children have a larger bone size at the distal radius, in comparison to our data.

There are currently no published reference data available for interpreting pQCT measurements at the radial diaphysis. Such reference data are essential for interpreting geometric parameters of cortical bone, which are important determinants of bone strength [26] and provide greater insight into healthy development and the effects of disease than solely measuring the distal radius site [13, 16].

We present gender-specific reference data for the XCT-2000 pQCT scanner in a cohort of healthy white Caucasian children aged between 6 and 19 years from the UK. These reference curves will assist in the interpretation of a child’s pQCT measurement in relation to his/her age or height by enabling calculation of an individual child’s SDS. We have provided reference curves to allow the calculation of SDS for geometric, strength and muscle parameters by height. Height is strongly correlated with cortical bone parameters [40] and is a powerful determinant of cortical bone geometry and strength [31, 41]. However, height has little or no correlation with radius or tibia BMD [4, 6] as these measurements are independent of bone and body size.

There are few centres which currently utilise pQCT for clinical diagnosis. The recent International Society for Clinical Densitometry (ISCD) guidelines notes the lack of research data to inform guidelines [30]. The ISCD recommendations that are based upon a review of the available literature suggest that pQCT measurements should be made at both metaphyseal and diaphyseal sites. Measurements should be made of trabecular and total BMD at the metaphyseal (distal) site and cortical BMC, total bone area, BMD, cortical thickness, AMI and SSI at a diaphyseal site. In our centre (Clinical Radiology and Manchester Royal Infirmary, Manchester, UK) for clinical reports, we currently report total and trabecular BMD from the 4% distal radius site only. Our practice is to use these pQCT data in conjunction with a clinical history, thorough clinical examination, anthropometric (height, weight and body mass index SDS) and pubertal stage assessment, relevant biochemical tests, postero-anterior hand radiograph for the assessment of bone age and DXA measurements of the lumbar spine and proximal femur. It is vital to consider the whole clinical assessment and not to treat single bone densitometry techniques as the sole diagnostic tool (Appendix 1) [27]. Although measures at diaphyseal sites are advocated for use in clinical practice by the ISCD [30], to date, these measures have been reported in research studies, and their clinical application and utilisation are still to be defined. The data presented in our study will permit the wider application of the pQCT measures obtained at the 50% radius site in clinical practice in children with bone disorders. We do not recommend measuring multiple metaphyseal and diaphyseal sites; in order to fully utilise the reference data we present, measurements must be made at corresponding distal (4%) and mid-shaft (50%) radius sites.

We did not provide the curves for cortical BMD for two reasons: firstly, the issue of partial volume effect in the growing bone and, secondly, that the majority of research at the diaphyseal site generally concentrates on bone geometry and strength measurements and shows alterations in these parameters rather than ‘true’ changes in cortical BMD. Although correction factors for the partial volume effect have been suggested for cortical BMD analysis in adults, the issue has yet to be resolved in children [42].

Whilst the direct clinical applicability of pQCT is yet to be fully determined in terms of fracture risk in children, the importance of these measures for gaining greater understanding of how the developing skeleton is affected by disease and treatment is being increasingly recognised [30]. There are a growing number of research studies using pQCT as a tool to understand disease aetiology in various patient groups [1121]. For example, in pre-pubertal asthmatic children treated with high doses of inhaled glucocorticoids, we showed that distal radial trabecular BMD was not reduced [18], but that these children had smaller periosteal and endosteal circumferences and thicker cortices in comparison to controls [16]. We believe that the increased risk of non-vertebral fracture reported in asthmatic children [43] might be due to their smaller bones. We also found that childhood survivors of acute lymphoblastic leukaemia (ALL) had reduced distal radial trabecular BMD but normal total BMD at this site [13]. At the mid-shaft radius, ALL survivors had larger endosteal and periosteal circumferences with thinner cortices compared to healthy controls. From these data, we speculated that ALL or its treatment resulted in endosteal bone loss and cortical bone thinning, but the axial moment of inertia and, hence bone strength, was maintained as a result of bone gain at the periosteal surface post disease and treatment. These studies illustrate how the measurements of bone parameters by pQCT provide an insight into the impact of paediatric diseases upon the skeleton.

In adults, the SSI has strong predictive power to estimate the fracture strength and failure load of human radii in vitro [44, 45]; in vivo studies in adult females indicate a decline in SSI of the femoral neck [46] and reduced SSI at the radius in post-menopausal women compared to pre-menopausal controls [47]. In healthy children, the age-related increase and gender differences in SSI have been shown at the radius [7, 9, 48] and tibia [49]. The SSI has been described in some paediatric patient groups [12, 16, 50] and in gymnasts versus controls [51]. Data from normal children [7] suggest that bone strength at the distal radius (as measured by SSI) may develop slowly and until 13 and 15 years, in females and males, respectively, may be inadequate to meet the mechanical challenges which threaten bone stability during the event of a fall. This may contribute to the increased incidence of distal forearm fractures in children during growth [7]. Whilst SSI is a powerful research tool for estimating bone strength, at present, there are no data available regarding the use of SSI in predicting fractures in children [30].

It is important when using these reference data that comparisons are made with measurements obtained from the same anatomical sites, using the same type of pQCT scanner, scan speed, software version, voxel size, analysis modes and thresholds for analysis of BMD and bone geometry. Small variations in the site at which a scan is performed can have a significant effect upon BMD and bone geometry measurements [52, 53]. In children, if measures at the distal radius site include the growth plate, then BMD will be falsely elevated [5]. A similar problem can occur if the section scanned is through sclerotic metaphyseal lines [54], which occur as a result of cyclical intravenous therapy with pamidronate [55]. There is currently no consensus as to which skeletal sites and scan modes should be used to carry out and analyse pQCT scans. We therefore provide detailed methodology to assist the inexperienced new user to successfully perform pQCT scans at the radius. The accurate and consistent placement of the reference line is essential, particularly in longitudinal studies.

There are some limitations of this study. Males and females in the study cohort were taller and heavier with greater BMI compared to UK reference data [35, 36]. This may reflect secular trends, particularly for increasing weight [56]. Secondly, although we attempted to gain a representative sample of the Manchester population, we did not use a systematic population-based sampling [32]. However, the TMDS, although a measure of the participants’ place of residence rather than the participant directly, do indicate that the sample is reasonably representative of the wide range of socio-economic status present in the Manchester population, with a slight bias towards the more deprived areas. Finally, pQCT measurements do not allow assessment of the spine [32], which is an important site affected in primary bone problems such as osteogenesis imperfecta, in children with chronic inflammatory conditions and those treated with glucocorticoids [27]. This can be obtained from scans on conventional CT body scanners with an appropriate bone equivalent phantom and software programme.

In conclusion, we have presented reference data from pQCT measurements of the distal and mid-shaft radius in children and young adults. The LMS data provide a tool for clinicians to assess the bone status of children. pQCT is a quantitative method, which provides information about BMD, bone geometry, parameters related to bone strength and muscle. These pQCT data may be used in conjunction with measurements derived from DXA [32] to provide an enhanced evaluation of the child who may have a bone disorder.


The authors would like to thank the research participants and their families, the schools where recruitment was undertaken, our dedicated team of radiographers and database manager Mr. Mike Machin.

We would like to gratefully acknowledge financial support from the Central Manchester and Manchester Children’s University Hospital NHS Research Endowment Fund and the support of the National Osteoporosis Society (Camerton, Bath, UK), which awarded Rebecca Ashby a Linda Edwards Memorial Studentship in 2003 and funded the initial part of the study (1997–1998).

Conflicts of interest


Supplementary material

198_2008_800_MOESM1_ESM.doc (154 kb)
ESM 1(153 KB DOC)
198_2008_800_MOESM2_ESM.pdf (6 mb)

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2009