Bone density in children: a review of the available techniques and indications

Abstract

The recent development of methods for measuring bone mineral content in children has markedly improved our ability to determine changes in bone mass during growth. Currently, the three most generally accepted techniques for measuring the bones of children are dual-energy X-ray absorbtiometry (DXA), quantitative computed tomography (QCT) and quantitative ultrasound (QUS). These techniques vary considerably in their acquisition of data and comparisons between them are difficult and, more often than not, judgment regarding their value has been, at least partially, subjective. DXA is, by far, the most widely used technique for bone measurements. It is low in cost, accessible, easy to use, and provides an accurate and precise quantitation of bone mass in adults. Unfortunately, DXA is unable to account for the large changes in body and skeletal size that occur during growth, limiting its use in longitudinal studies in children. QCT can asses both the volume and the density of bone in the axial and appendicular skeletons, without influence from body or skeletal size, giving it a major advantage over other modalities for bone measurements in children. The cost and inaccessibility of CT scanners, however, has significantly limited its use for bone measurements. Measuring the bones of children by QUS is appealing because ultrasound is low in cost, portable, easy to use and does not emit radiation. In adults, this technique is able to predict fracture risk independent of bone mass determinations in patients with osteoporosis and, therefore, its measurements must be related to certain aspects of bone strength. However, ultrasound values are dependent on so many structural properties not yet fully understood, that it is difficult to use the information meaningfully in children.

Introduction

Bone is a composite material of organic (mostly collagen) and inorganic components (consisting of crystals of hydroxyapatite) [1]. Because of its higher attenuation coefficient, the mineral fraction accounts for the radiographic depiction of bone. The subjective evaluation of bone mineral in the skeleton by conventional radiography is relatively insensitive, as bone mass may have already decreased by 30–40% by the time osteoporosis is appreciated [2]. Thus, the recent development of more accurate and precise methods for measuring bone mineral content has significantly improved our ability to assess changes in bone mass. It should be stressed that bone measurement modalities were developed to study the adult skeleton and that, frequently, special modifications, either in equipment, software or technique, are required for accurate and precise measurements in children.

There are two main reasons for measuring bone mineral content in children. The first is to diagnose and quantify the loss of bone mineral associated with the various disorders that cause osteopenia [3]. Osteopenia, literally meaning poverty of bone, is a non-specific term used to indicate a reduction of calcified bone, which may be the result of many skeletal disorders [4]. The term arose from the frequent inability of conventional skeletal radiographs to differentiate between conditions that can cause a reduction in mineralized bone. Osteopenia in children is usually the result of one of three major types of disorders. Most frequently, it is a consequence of osteoporosis, a condition in which there is a parallel loss of bone mineral and matrix. Less commonly, it is the result of rickets, a pathological loss of mineralized bone due to a reduction of calcium-phosphate levels with resultant accumulation of nonmineralized matrix (osteoid). Lastly, osteopenia can result from a defect in bone formation associated to a congenital or developmental disease such as osteogenesis imperfecta (OI), homocystinuria, galactosemia, various chondrodysplasias, Menkes syndrome, etc. Many of these disorders are due to mutations in collagen genes, but for others the mechanism for the reduction in calcified bone is not clearly understood and may vary with the individual condition [3].

The second reason to measure bone mass in children is to improve our understanding of the childhood antecedents of a condition that happens to manifest in elderly subjects: osteoporosis. Since skeletal mass in adulthood is the result of both the amount of bone gained during growth and its subsequent rate of loss, it is clear that the factors affecting bone density during growth are important determinants of future skeletal resistance to fracture [5]. Large increases in bone mass occur in children over a relatively brief period of time during puberty and careful evaluation of the factors responsible for this increase may ultimately be of value in the prevention of osteoporosis in later life.

Several techniques are available for bone mass measurements for both research and clinical purposes in pediatrics. Single-photon absorptiometry (SPA) and dual-photon absorptiometry (DPA) have been used extensively in the past, but have been superseded. Magnetic resonance imaging is a promising technique for the analysis of cancellous bone, but it is still under investigation. In this chapter, we will describe the advantages and disadvantages of the three modalities most commonly employed in pediatrics: dual-energy X-ray absorptiometry (DXA), quantitative computed tomography (QCT) and quantitative ultrasound (QUS). Comparison between these techniques, which are totally dissimilar in the way they acquire data, is difficult, and, more often than not, judgment regarding their value has been, at least partially, subjective.

Ideally, the technique employed should be easy to perform, inexpensive, noninvasive, harmless, accurate and reproducible. Furthermore, it should be able to assess cancellous and cortical bone in the axial and appendicular skeletons and also to separately assess the two components of bone mass, bone volume and bone density. Lastly, measurements should not be influenced by the size of the bones nor should they be influenced by body size or soft tissues.

DXA is the most commonly employed technique for bone mass determinations world-wide. It represents an evolution of another projection method, dual photon absorptiometry, that used a radionuclide source [6]. With DXA, an X-ray tube is the source of photons resulting in a higher photon flux, better edge detection and precision. The photons exiting from the region of interest are detected, measured by a computer, and, with the use of calibration materials, the attenuation values are converted into determinations of bone mineral content or BMC (in g). Frequently, BMC values are divided by the projected area of the bone analyzed and referred to as bone mineral density or BMD (g/cm²).

DXA has been used for bone measurements in the whole body, the lumbar spine, the proximal femur, the forearm and the hand, and, in all sites, DXA reflects a composite of cancellous and cortical bone. Depending on the equipment used, the examination may take between 2 and 15 min. Radiation exposure from DXA is very low, the effective dose being approximately 1 μSv for lumbar spine measurements [7]and about 4 μSv for whole skeleton scans [8]. The in vivo precision of posteroanterior DXA measurements has been calculated to range from 0.8 to 2.5% in children 9, 10, 11, 12, 13, 14, however, different manufacturers display substantial variation in BMD values of the same bone. These variations are mainly attributable to differences between devices, scanning protocols and body size of the subjects studied. Although normative data for DXA values in pediatrics is available in the literature and is included in most DXA software packages, caution is advised before using published normative data for clinical use and institutional and device-specific norms are preferable to published references. Recently, leading manufacturers of DXA equipment have proposed a standardization for lumbar spine measurements [15]. Unfortunately, however, to provide similar standardization at other sites is not currently possible because of substantial differences in the regions of interest that the various manufacturers have incorporated into the design of their devices.

Bone measurements by DXA have been obtained from children of all ages 12, 13, 14, 15, 16, 17, 18. Generally, the values increase from infancy to adulthood 9, 11, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28. In the lumbar spine 9, 11, 22, 24, 25, the femoral neck 11, 21, 22, 25and the entire skeleton 11, 22, 23, 25the association between chronological age and BMC is observed as a segmented polynomial curve; the rapid increase seen during childhood is followed by an even greater increase during puberty that ends in the third decade of life 9, 10, 11, 20, 21, 24, 25, 29, 30. In contrast, DXA values at the radius have not been found to be affected by puberty [11].

Studies using DXA to assess gender differences in bone mass indicate that, in the lumbar spine, girls have greater BMD than boys by the age of 12 or 13 years 9, 19, 21, 24, 26, while in the femoral neck, some investigators found greater BMC in boys than in girls 11, 19, 21, 22and some found no gender differences 25, 31. Studies of the radius also yielded discrepant results; boys had greater values than girls in one study [26], while no gender differences were found in another [11]. Similarly, studies using DXA to examine racial differences in bone mass in children have conflicted. Some investigators found no differences in bone measurements between African–American girls and Caucasian girls 20, 27, yet others report higher values for African–American girls at all ages [32]. A recent study found that Asian–American children have lower spine, femur and whole body DXA values when compared to Caucasian children and suggested the difference to be size related [33].

The marked differences described in the literature regarding normal DXA values in children are, in part, attributable not only to variations in equipment, but also to differences in the population studied. It should be stressed that the major changes in body and skeletal size occuring during growth influence DXA measurements and limits its utility in children. DXA measurements are based on the two-dimensional projection of a three-dimensional structure and, therefore, are a function of both the mass and the size of the bone being examined [34]. While the inability of DXA to account for bone size is not of great concern when studying the mature skeleton, in growing children, longitudinal DXA values are subject to considerable error, as they reflect both the changes in skeletal size and in bone mass. Moreover, DXA values are also influenced by the unknown composition of soft tissues in the beam path of the region of interest. Because corrections for the soft tissues are based on a uniform distribution of fat around the bone, longitudinal DXA values in children may reflect the changes in body size and composition that occur with growth more than true changes in bone mineral content. It has been determined that inhomogeneous fat distribution in soft tissues, resulting in a difference of 2 cm fat layer between soft tissue area and bone area, will influence DXA measurements by 10% [35]. While this is not a disadvantage when studying subjects whose weight and body size remain constant, it especially limits studies in children with eating disorders, such as obesity or anorexia nervosa.

Lastly, motion artifacts are another limitation of DXA studies in very young and uncooperative children. Recent studies have shown that motion artifacts may increase the values of the projected bone area by 9%, of BMC by 13% and of BMD by 4% [36]. In an attempt to overcome these errors, children have been restrained 9, 13, sedated 12, 16, or studied while asleep [17].

In summary, the relatively low cost, availability and ease of use are the main advantages of DXA. This technique provides an accurate and precise quantitation of bone mass in the mature skeleton. Unfortunately, the inability of DXA to account for the large changes in body and skeletal size that occur during growth limits its use in longitudinal studies in children.

Computed tomography introduced the technique of digital imaging to diagnostic radiology. The transverse anatomic sections afforded by this digital technique provide a three-dimensional image unobscured by overlying structures. However, the pictorial display has overshadowed the very basis of that image—the ability to distinguish subtle differences in the absorption or attentuation of X-rays. The data displayed as the CT image are actually a representation of attenuation values or CT numbers of the object scanned. These numerical values are stored in a computer in digital form and are accessible for future study, as needed. The use of this digital information to provide quantitative information has been labeled QCT.

QCT bone measurements can be obtained at any skeletal site with a standard clinical CT scanner using an external bone mineral reference phantom for calibration and specially developed software. Five different bone measurements can be obtained using QCT to study skeletal development in children; the density of cancellous bone, the density of cortical bone, the size of the axial skeleton, the size of the appendicular skeleton and the volume of cortical bone in the appendicular skeleton. The coefficients of variation for these QCT measurements in children range between 0.6 and 2% 3, 37. It should be noted that, precision and accuracy of QCT bone density determinations are based on bone marrow fat content, which is much lower in children, and therefore, the measurements are much more precise and accurate than in adults.

Differences in morphology of cancellous and cortical bone and several technical issues regarding QCT measurements of bone density in these two compartments must be considered for the appropriate interpretation of QCT data. Cancellous bone exists as a three dimensional lattice of plates and columns (trabeculae). The trabeculae divide interior volume of the bone into intercommunicating pores, which are filled with a variable mixture of red and yellow marrow [38]. Because of the relatively small size of the trabeculae when compared to the pixel, the QCT unit of measurement, QCT values for cancellous bone density reflect not only the amount of mineralized bone and osteoid, but also the amount of marrow per pixel [39]. Similar limitations apply to in vitro determinations of the volumetric density of trabecular bone which are obtained by washing the marrow from the pores of a specimen of cancellous bone, weighing it, and dividing the weight by the volume of the specimen, including the pores [38]. Both CT and in vitro bone density determinations of cancellous bone are, therefore, directly proportional to the bone volume fraction and inversely proportional to the porosity of the bone. The relatively large coefficient of variation for values of cancellous bone density reflects the considerable variations in the dimensions of the pores throughout the vertebral body.

In contrast, CT measurements of cortical bone reflect the material density of the bone if the cortex is sufficiently thick (above 2–2.5 mm) to circumvent volume averaging errors [40]. The measured pixel represents the linear combination of the attenuation coefficients defined by the densities and concentrations of osteoid and mineral. While the nonmineral fraction may contribute to minor fluctuations in measurements of cortical bone density, the CT numbers are primarily based on the calcified bone fraction which has a high attenuation coefficient [40]. These measurements are analogous to in vitro determinations of the intrinsic mineral density of bone, which are commonly expressed as the ash weight per unit volume of bone [41]. On average, CT values for cortical bone density are eight times higher than those for cancellous bone density, a finding consistent with histomorphometric studies indicating an equivalent difference in the porosity of these two forms of structural organization of bone tissue [41]. Otherwise, cancellous bone can be viewed as a porous structure comprised of bone tissue with the same mechanical properties and composition as cortical bone.

The radiation exposure from QCT bone measurements depends on the technique employed and can be as low as 100–200 mrem (1.5 mSv) localized to the region of interest in the appendicular or axial skeleton. The total body equivalent dose of radiation is approximately 4–9 mrems (40–90 μSv), and this figure includes the radiation associated with screening digital radiographs used to localize the site of measurement 7, 42. This amount of radiation is far lower than that associated with other CT imaging procedures, accounting for the wide range of published figures for the radiation dose associated with CT measurements. It is also less than many other commonly used radiographic diagnostic tests. By comparison, a round-trip transcontinental flight in North America exposes a child to roughly 6–8 mrems (60–80 μSv) of ionizing radiation 7, 42.

Studies in healthy children using QCT have shown that the material density of cortical bone in the appendicular skeleton remains fairly constant (2.00±0.065 gm/cm³) and is not influenced by age, anthropometric parameters, puberty, gender or race [40]. Events during puberty are, however, the major determinants of the increases in cancellous bone density [43], which reaches its peak around the time of cessation of longitudinal growth and epiphyseal closure [44]. QCT studies have also shown a greater increase in vertebral cancellous bone density in African–American girls than in Caucasian girls during the later stages of puberty [45].

Recent QCT studies have also analyzed the influence of gender on the amount of bone that is gained during childhood and adolescence. QCT has helped establish that the lower vertebral bone mass of females, when compared to males, results from early gender differences in the size of the bones rather than differences in cancellous bone density 46, 47. Even after accounting for differences in body size, the cross-sectional area of the vertebral body is approximately 20% smaller in girls than in boys [46]. On the other hand, there are no gender differences in the size of the appendicular bones beyond those attributable to differences in body size [37]. QCT values for the size and the amount of bone in the appendicular skeleton do, however, correlate strongly with all anthropometric indices, suggesting that weight bearing and mechanical stresses are the major determinants of the increases in the size and the volume of cortical bone during growth 37, 40.

The recent development of the smaller, peripheral QCT scanners, designed exclusively for bone measurements, have the advantage of being relatively inexpensive, easy to use and mobile. Unfortunately, however, these scanners can only assess the bones of the appendicular skeleton.

In summary, the ability of QCT to assess both the size and the density of bone in the axial and appendicular skeletons, without influence from body size, is the major advantage of this modality when used to measure bone in children. Unfortunately, CT scanners are expensive and require costly maintenance and considerable technological expertise for proper function. Moreover, this equipment is usually located in the radiology department and is under constant clinical demand, creating a lack of accessibility. These disadvantages have partially been overcome by the development of less expensive peripheral QCT scanners designed exclusively for bone measurements of the appendicular skeleton.

QUS has recently been used to assess appendicular bone by measuring the changes that occur in the velocity and in the energy of ultrasound waves as they go through bone. The ultrasound transmission velocity (also known as speed of sound, or SOS) is obtained by dividing the width of the region of interest by the transit time and is expressed in meters per s (m/s). The loss of acoustic energy that occurs when the ultrasound wave is absorbed or scattered by the medium through which it is being propagated results in a reduction in the amplitude of the wave and is referred to as broadband ultrasound attenuation (BUA). The ultrasound attenuation is a function of frequency and for biologic soft tissues this relation is linear over the range of 200–600 kHz. BUA is defined as the slope of attenuation versus the frequency in this range and is expressed in decibels per megahertz (dB/MHz).

QUS bone measurements have been obtained mainly in the calcaneus and, less frequently, in the patella, tibia and phalanges. It is generally believed that, in the calcaneus, both BUA and SOS are determined by the number, the thickness, the mineral content and the three-dimensional arrangement of the trabeculae. BUA values are known to vary as much as 50% depending upon the principle orientation of the trabeculae [39]. Moreover, ultrasound measurements are influenced by the amount and composition of marrow in the bone and of other soft tissues in the path of the ultrasound waves [39]. Unfortunately, numerous studies in adults have failed to find a strong correlation between ultrasound measurements of bone and values obtained using other bone measurement methods and, therefore, ultrasound cannot be used as a surrogate for other modalities [39].

Currently, all commercially available ultrasound equipment for bone measurements are designed for adults and use relatively large transducers. Smaller transducers and special foot pads or calipers are needed when studying children to avoid interference from air 48, 49, 50, 51. Like in adults, ultrasound studies in children require a coupling medium (a water bath or gel) between the transducers and the skin overlying the bone. The temperature of the coupling medium must be maintained constant to avoid inaccuracies in the velocity and attenuation of ultrasound measurements. In children, SOS and BUA values increase with age but they exhibit considerable variability 48, 51. The influence of puberty on ultrasound indices has been investigated with the findings that SOS values increase substantially between the second and third Tanner stages of sexual development in girls, while the increase is constant throughout all stages of puberty in boys [50]. Reported intra-observer coefficients of variation in children range from 0.5 to 1.2% for SOS 48, 49, 51and from 3 to 5% for BUA 48, 50.

The attractiveness of QUS for bone measurements in children lies in its low cost, portability, ease of use and lack of ionizing radiation. Because, in adults, ultrasound is able to adequately predict fracture risk independent of bone mass determinations in patients with osteoporosis, QUS measurements must be related to some aspects of bone strength [52]. Unfortunately, ultrasound values are dependent on so many structural parameters, yet to be fully defined, that it is difficult to use this information in a meaningful way in children.