John Hopkins University has licensed the HSA software to Hologic, Inc. Dr Beck has received research funding and speakers fees from Merck, Eli Lilly, Amgen, and NPS Pharmaceuticals. All other authors state that they have no conflicts of interest.
Proximal femur geometry was assessed in children and young adults treated with chronic GCs for CD or SSNS. Subperiosteal width and section modulus were significantly lower in CD and greater in SSNS compared with controls, highlighting the importance of the underlying disease, persistent inflammation, and alterations in lean mass.
Introduction: The impact of glucocorticoid (GC) therapy on bone structure during growth is unknown. Our objective was to characterize proximal femur geometry in children and young adults with Crohn disease (CD) or steroid-sensitive nephrotic syndrome (SSNS) compared with controls and to evaluate the influence of lean mass and GC therapy on bone parameters.
Materials and Methods: DXA scans of the hip and whole body were obtained in 88 subjects with CD, 65 subjects with SSNS, and 128 controls (4–26 years of age). Hip structural analysis parameters (subperiosteal width, cross-sectional area [CSA], and section modulus in the narrow neck [NN], intertrochanteric region [IT], and femoral shaft [FS]), areal BMD, and whole body lean mass were expressed as Z scores compared with controls. Multivariable linear regression was used to adjust outcomes for group differences in age, sex, race, and height.
Results: Mean lean mass Z scores were lower in CD (−0.63, p < 0.001) and greater in SSNS (0.36, p = 0.03) compared with controls. Hip areal BMD Z scores were lower in CD (−0.73, p < 0.001) but not SSNS (−0.02, p > 0.2) compared with controls. In CD, Z scores for subperiosteal width (NN: −1.66, p < 0.001; FS: −0.86, p < 0.001) and section modulus (NN: −0.60, p = 0.003; FS: –0.36, p = 0.03) were significantly lower than controls. In contrast, in SSNS, Z scores were greater for IT subperiosteal width (0.39, p = 0.02), FS CSA (0.47, p = 0.005), and FS section modulus (0.49, p = 0.004). Alterations in section modulus in CD and SSNS were eliminated after adjustment for lean mass. Cumulative GC dose was inversely associated with FS subperiosteal width and section modulus only in CD.
Conclusions: These data show that the effects of GC on proximal femur geometry during growth are influenced by the underlying disease, persistent inflammation, and alterations in lean mass. These data also provide insight into the structural basis of hip fragility in CD.
Glucocorticoid (gc) medications are highly effective in the treatment of allergic and inflammatory conditions in children and adults. However, GC therapy is associated with potentially severe adverse side effects, including osteoporotic fractures.(1–3) The bone loss in GC-induced osteoporosis is mediated by transient increases in bone resorption(4) and sustained reductions in bone formation caused by inhibition of osteoblastogenesis and increased osteoblast apoptosis.(5) The growing skeleton may be particularly vulnerable to the effects of chronic GC therapy on bone formation. A population-based study revealed that children prescribed four or more courses of oral GC therapy had significantly greater fracture risk compared with nonusers.(1)
Inflammation, the target of GC therapy, has similar deleterious effects on bone cells. For example, pro-inflammatory cytokines, such as TNF-α and interleukin-6 (IL-6), inhibit osteoblast differentiation, function, and survival and promote osteoclastogenesis.(6–8) Crohn disease (CD), an inflammatory disorder of the gastrointestinal tract that is associated with high serum TNF-α and IL-6 levels, results in malnutrition, malabsorption, growth failure, pubertal delay, and osteoporosis. Recent studies in adults documented that CD was associated with significantly increased risk of vertebral and hip fractures(9,10); however, the relative contributions of the inflammation and the GC therapy to the fracture risk are not known.
We recently examined DXA measures of whole body BMC in children and young adults treated with GC for CD.(11) We reported that CD was associated with significantly lower whole body BMC relative to height compared with healthy controls.(11) In contrast, we showed that steroid-sensitive nephrotic syndrome (SSNS) in children and adolescents was associated with significantly greater whole body BMC relative to height compared with healthy controls.(12) Whereas CD results in sustained inflammation, SSNS is characterized by a rapid and complete response to high-dose GC therapy, although relapses are common. We hypothesized that the preserved whole body BMC in SSNS, despite greater cumulative GC exposure, was caused by the lack of sustained systemic inflammation and the markedly greater body mass index (BMI) compared with controls. To our knowledge, no studies have shown that SSNS is associated with increased fracture risk.
The bone structural implications of the observed lower and greater whole body BMC observed in CD and SSNS, respectively, are not known. Bone strength is determined by both material and geometric properties. In addition to BMD, the cross-sectional dimensions, length of the femoral neck, and the angle formed between the femoral neck and shaft are critical determinants of age-, sex-, and race-related differences in hip fracture risk.(13–16) The purpose of this study was to use the hip structural analysis (HSA)(17) technique to characterize proximal femur geometry and estimates of bone strength in children and young adults exposed to chronic, high-dose GC therapy in the presence (CD) or absence (SSNS) of persistent systemic inflammation compared with healthy controls.
MATERIALS AND METHODS
Subjects with SSNS, as defined by the International Society of Kidney Disease in Children, were identified at The Children's Hospital of Philadelphia (CHOP) and St Christopher's Hospital for Children. Subjects were deemed sensitive to steroids if a urinalysis documented trace or no proteinuria within 8 weeks of initiating daily GC therapy. Study visits were performed at least 14 days after the most recent remission to ensure that edema had resolved completely. Subjects were excluded from the study if renal insufficiency was present (glomerular filtration rate <90 ml/minute/1.73 m2). Whole body and lumbar spine BMC and body composition measures have been reported in this cohort of subjects with SSNS.(12,18)
Children and young adults with CD were recruited from CHOP and the Hospital of the University of Pennsylvania. The diagnosis of CD was confirmed by clinical, radiographic, and histological criteria. Prior publications have addressed whole body BMC, lumbar spine areal BMD, growth, body composition, and vitamin D status in this cohort.(11,19–22)
Subjects with SSNS and CD were excluded if they were younger than 4 years of age, had medical conditions unrelated to the main diagnosis and known to affect bone accrual, or had hip DXA scans that did not meet quality control standards.
Healthy control subjects were recruited from general pediatric clinics in the Philadelphia region and through newspaper advertisements. Control subjects with conditions known to affect growth, bone accrual, nutritional status, or pubertal development were excluded from the study. The control subject population, which has been used in prior publications,(11,12,18–22) consists of all available control subjects with quality-controlled hip DXA scans.
This protocol was approved by the Institutional Review Board of each participating hospital. Written informed consent was obtained from all young adult subjects and the parent or guardian of those younger than 18 years of age. Assent was obtained from children ≤7 years of age.
Disease characteristics and GC exposure
Medical records were reviewed to determine the date of diagnosis of CD or SSNS. Disease characteristics and medication exposures were abstracted from the medical record, as described previously.(11,12) All doses of GC were converted to prednisone equivalents. Cumulative GC exposure at the time of the study visits was summarized as milligram, milligram per kilogram, and milligram per kilogram per day.
Anthropometry and pubertal development
Weight and height were measured using a digital scale to the nearest 0.1 kg (Scaltronix, White Plains, NY, USA) and a stadiometer to the nearest 0.1 cm (Holtain, Croswell, Crymych, UK), respectively. Age- and sex-specific Z scores (SD scores) for height and BMI were calculated using the National Center for Health Statistics 2000 Centers for Disease Control growth data using the LMS method.(23) Pubertal stage was assessed according to the method of Tanner et al.(24)
Conventional DXA measurements
DXA scans of the left proximal femur and whole body were performed using a Hologic QDR 2000 bone densitometer (Hologic, Bedford, MA, USA) with a fan beam in the array mode. Proximal femur scans were analyzed to generate estimates of BMC (g) and areal BMD (g/cm2) of the total hip and femoral neck. Whole body DXA scans were used to estimate lean mass (kg), excluding the skull (postcranial). The instrument was calibrated daily with a hydroxyapatite phantom. A recent pediatric study showed that the in vivo CVs of conventional DXA measures in the proximal femur were <1.8%.(25)
Table Table 1.. Characteristics of Subjects With CD, Subjects With SSNS, and Healthy Controls
Areal BMD is a composite measure of BMD and dimensions that does not provide discrete measures of bone structure or strength. The HSA program (version 2.1) uses conventional DXA images to measure the cross-sectional geometry and strength of the narrow neck (NN), intertrochanteric (IT) region, and proximal femoral shaft (FS).(26) The femoral neck was measured at its narrowest point (i.e., NN). The IT region was measured along the bisector of the axes of the femoral neck and shaft. The FS was measured at a distance of 1.5 times the width of the femoral neck distal to the intersection of the femoral neck and shaft axes. To create a cross-sectional profile, HSA software generates lines of pixel values (g/cm2 hydroxyapatite) perpendicular to the bone axis at each cross-sectional region. Hence, BMD is computed as the average value in a region including only values above a threshold. Outer diameter is computed as the distance between the profile margins (subperiosteal width, cm) after correcting for image blur. Pixel values in the profile are converted to an equivalent linear thickness of cortical bone by dividing by the average mineral density of adult cortical bone. The surface area of bone in the cross-section (CSA, cm2) is computed from the integral of the thickness profile. After computing the center of mass, the cross-sectional moment of inertia (CSMI, cm4) is computed from the integral weighted by the pixel distance from the center of mass. The section modulus (cm3) is computed as the CSMI divided by the maximum distance from the center of mass to the medial or lateral outer margin. The section moduli are relevant only for bending in the image plane. The femoral neck length (cm) represents the distance from the center of the femoral head to the intersection of the neck and shaft axes. The neck shaft angle (degrees) represents the angle between the femoral neck and FS.
Unlike pencil beam scanners, fan beam DXA models cause a systematic magnification error: as the distance of the bone from the X-ray source decreases, the width of the projection increases.(27) This characteristic of fan beam technology is important to recognize in the study of obese or malnourished populations. As previously described,(26,28) a calibration phantom was used to quantify the height effect and provide a correction for pixel spacing as a function of the bone height for use in the HSA algorithms.
Analyses were conducted using Stata 8.2 (Stata Corp., College Station, TX, USA). Two sided tests of hypotheses were used, and p < 0.05 was considered to be statistically significant. The results of descriptive analyses are provided in Table 1. Differences in means were assessed using Student's t-test or the Wilcoxon rank sum test if the data were not normally distributed. Group differences in categorical variables were assessed using the χ2 or Fisher's exact test.
Calculation of Z scores
Bone outcomes and lean mass results were converted to Z scores to facilitate the comparisons among CD, SSNS, and controls. Growth is characterized by sex-specific increases in BMD, bone dimensions, and body composition(29); therefore, all Z scores are sex specific. Measures of areal BMD of the total hip and femoral neck were expressed as Z scores relative to sex and age, consistent with conventional practice. Z scores for total hip and femoral neck BMC, HSA measures of bone dimensions and strength, and whole body lean mass were generated relative to sex and height.(11,26) Data from the control subjects were used to derive the sex-specific predicted value of each bone measure relative to height or age using linear regression. Natural log transformations of bone variables and height were used to create linear relations. The residuals from the regression analysis were not heteroscedastic; therefore, the root mean square error from the regression analysis served as the SD at all levels of the predicted variable. The Z scores were calculated as follows: [(observed – predicted)/SD].
Table Table 2.. Conventional Hip DXA Z Scores in CD and SSNS
Analysis of conventional DXA and HSA measures
Separate multivariable linear regression models were used to compare the point estimates for the Z scores for areal BMD, BMC, or HSA parameters in CD versus controls and in SSNS versus controls, adjusting for age, sex, and race (black versus all others).(30) Conventional DXA femoral neck areal BMD-for-age Z scores were assessed in CD and SSNS, relative to controls and each other (Table 2). Height Z score was subsequently added to the models to assess the confounding effect of short stature on conventional areal BMD Z scores.(31) Because total hip and femoral neck BMC Z scores and HSA Z scores were generated relative to height, subsequent adjustment for height Z score was not needed. The p values for the comparison of Z scores in CD versus SSNS for each outcome represents the test of whether the observed difference in the regression coefficient for CD and the regression coefficient for SSNS differs significantly from zero. This difference in regression coefficients was obtained from the regression model by subtracting the expected Z scores for CD versus SSNS patients of the same sex and height. A significant difference indicates that expected Z scores differ significantly between CD versus SSNS patients of the same sex and height. Multivariable linear regression models were subsequently adjusted for lean mass-for-height Z score to determine the influence of differences in muscle mass on bone structure. All regression models were assessed for adequacy of model assumptions.
Analysis of the relation between GC exposure and bone outcomes
Analyses of the relation between total GC dose and bone outcomes were conducted using Pearson product-moment estimates. Subsequently, separate multivariable regression models were used within the CD and SSNS subjects to adjust for sex and race (Table 3). Models assessing areal BMD-for-age Z score were adjusted for height Z score. Models assessing BMC- and bone structure-for-height Z scores were adjusted for age but not height Z score. Mean values of conventional DXA and HSA Z scores in subjects with CD or SSNS on and off GC therapy at the time of the study visit were compared using Student's t-test.
Characteristics of the study population
Demographic and anthropometric characteristics of the study population are summarized in Table 1. There were significant differences in the age, sex, and racial distribution of study subjects. Subjects with CD were older and subjects with SSNS were younger than controls; the pubertal distribution of the three groups reflects these age differences. Subjects with CD had significantly lower height, BMI, and lean mass-for-height Z scores than healthy controls. Subjects with SSNS also had significantly lower height Z scores but had higher BMI and lean mass-for-height Z scores compared with controls. The median (range) disease duration in subjects with CD and SSNS was 3.7 (0.1–16.0) and 3.7 (0.6–23.8) years, respectively.
A summary of the cumulative GC dose and duration of exposure is provided in Table 1. All subjects with SSNS and 85% of those with CD had received GC therapy (p = 0.001). Cumulative GC dose (mg/kg) was inversely correlated with the height Z score in CD (r = −0.32, p = 0.003) and SSNS (r = −0.37, p = 0.004).
Conventional DXA assessment in CD and SSNS
Areal BMD and BMC estimates at the total hip and femoral neck in CD and SSNS are presented in Table 2. In CD, the sex-specific total hip and femoral neck areal BMD-for-age Z scores were significantly lower compared with healthy controls or subjects with SSNS. These deficits in CD were attenuated after adjustment for height Z score, but the Z scores remained significantly lower compared with controls or SSNS. For example, the mean total hip areal-BMD-for-age Z score was −0.99 (95% CI, –1.34 to –0.64; p < 0.001 versus controls) and increased to −0.73 (95% CI, –1.07 to –0.40; p < 0.001 versus controls) when also adjusted for height Z score. Analysis of total hip and femoral neck BMC-for-height Z score showed similar significant deficits in CD compared with controls. In contrast, no differences in areal BMD and BMC in the total hip and femoral neck between SSNS subjects and controls were detected.
Table Table 3.. Assessment of the Relation Between Total GC Exposure (mg/kg) and Bone Outcome Z Scores in CD and SSNS
Assessment of hip geometry and strength using HSA
A summary of the sex-specific Z scores for the hip structure parameters in CD and SSNS, compared with healthy controls and adjusted for age, sex, and race, is provided in Fig. 1. In CD compared with controls, significant deficits in subperiosteal width were noted at the NN and FS (p < 0.001), leading to lower section moduli (p < 0.05) at both sites. Compared with controls, a significantly greater proportion of subjects with CD had Z scores less than –2 at the NN (subperiosteal width: 33% versus 2%, p < 0.001; section modulus: 14% versus 2%, p < 0.001) and FS (subperiosteal width: 13% versus 2%). CSA at all sites was preserved. In contrast, in SSNS, subperiosteal width was significantly higher than controls at the IT region (p = 0.02) only, whereas CSA and section modulus were higher in the FS (p < 0.001). Compared with subjects with CD, subjects with SSNS had significantly higher subperiosteal width and section modulus at all sites and a higher CSA at the FS. Similar results were obtained in multivariable models that included Tanner stage as an indicator variable. Neck length- and neck-shaft angle-for-height Z scores were similar between subjects with CD, with SSNS, and controls (data not shown).
To determine the influence of muscle mass on altered bone structure in CD and SSNS, we added lean mass-for-height Z score to the multivariable regression models. After adjustment for the lower lean mass-for-height Z score in CD, NN and FS subperiosteal width deficits were minimally attenuated but still statistically significant (p < 0.001). However, the estimates for NN section modulus (Z = −0.36; 95% CI, –0.74 to –0.02; p = 0.06) and FS section modulus (Z = −0.04; 95% CI, –0.34 to 0.26; p > 0.2) were no longer lower than controls.
Adjustment for lean mass also eliminated group differences between SSNS and controls. After adjustment for the higher lean mass-for-height Z scores, IT subperiosteal width (Z = 0.21; 95% CI, –0.13 to 0.56; p > 0.2), FS CSA (Z = 0.15; 95% CI, –0.14 to 0.45; p > 0.2), and FS section modulus (Z = 0.20; 95% CI, –0.11 to 0.53; p > 0.2) were no longer significantly higher in SSNS compared with controls.
Relation between GC cumulative dose and DXA measures within subjects with CD or SSNS
Cumulative GC dose (mg/kg) was significantly correlated with areal BMD-for-age Z score of the total hip and femoral neck in CD (r = −0.23 to –0.29, p ≤ 0.04) and SSNS (r = −0.27, p = 0.04 for both measures). However, these results may be confounded by GC-induced reductions in linear growth. Therefore, as detailed in Table 3, multivariable linear regression models were adjusted for height Z score. After adjustment for height Z score, only the relation between GC dose and femoral neck areal BMD-for-age Z score in CD was maintained. No association between GC dose and areal BMD was detected when GC exposure was expressed as total milligrams or milligram per kilogram per day. There were no significant relations between GC dose and total hip or femoral neck BMC-for-height in CD or SSNS. When subjects with CD or SSNS on GC therapy at the time of the study visit were compared with those off GC therapy, no differences in conventional DXA measures were observed.
In CD, cumulative GC dose (mg/kg) was correlated with FS subperiosteal width and section modulus (both r = −0.26, p = 0.02) but did not correlate with any HSA measures in the IT region or the NN. In multivariable models, the relations between GC dose and FS HSA measures were maintained (Table 3). Similar relationships were noted when cumulative dose was expressed in total milligrams and milligram per kilogram per day. No relation between cumulative GC dose and HSA measures was observed at any site in SSNS. When subjects with CD or SSNS on GC therapy at the time of the study visit were compared with those off GC therapy, no differences in HSA measures were observed.
This study in children and young adults with CD and SSNS showed altered hip structure in both conditions. In CD, the lower subperiosteal width at the NN and FS resulted in lower bone strength, as measured by the section modulus. At the NN and FS, bone CSA was preserved, which we interpret to be the result of a relative endosteal contraction. In contrast, SSNS was characterized by an overall protective structural profile in the IT region, where subperiosteal width was higher, and at the FS, where the greater cortical CSA resulted in a greater section modulus. The differences in section modulus observed in CD and SSNS were no longer significant after adjustment for lean mass-for-height Z scores.
Recent studies have revealed that bone abnormalities in adults with CD result in clinically important fractures. Bernstein et al.(9) documented that the vertebral and nonvertebral fracture risk in a population of subjects with inflammatory bowel disease (CD and ulcerative colitis) was ∼40% higher than a healthy reference population. Fractures of the hip were 60% more common. In two population-based studies using the United Kingdom General Practice Research Database, the risk of hip fracture in CD was 68%(32) and 86%(10) higher than controls and was higher than subjects with ulcerative colitis. The authors hypothesized that greater disease severity and GC burden in CD contributed to higher fracture rates.(10) We reported a series of five children with CD who experienced painful vertebral compression fractures.(33) At our institution, there have been no known instances of such fractures in children with SSNS. If there are underlying differences in fracture risk between children with CD and SSNS, our data suggest a structural basis for these differences.
The optimal parameter to use in the identification of individuals at high risk of fracture at any age remains controversial, and few data exist in children. A case-control study of girls with forearm fractures revealed no group differences in cortical or trabecular volumetric BMD, as assessed by pQCT.(34) However, the CSA of the distal radius was 8% lower in children with fractures. Similarly, a lower metacarpal width and metacarpal index (cortical thickness divided by metacarpal width) were independently associated with wrist and forearm fractures in children.(35) The markedly lower subperiosteal width in the subjects with CD may represent an important structural threat, particularly without the protective effect of a shorter femoral neck.
The structural profile observed in CD stands in contrast to that observed in SSNS. We documented a protective structural profile in the proximal femur in SSNS, consistent with our previous findings of greater whole body BMC relative to height.(12) The differences noted using HSA software were not observed with conventional hip DXA analyses. Areal BMD-for-age and BMC-for-height Z scores in SSNS were not different from controls at either the total hip or femoral neck. This may be because of the failure of conventional DXA projection techniques to distinguish between trabecular and cortical bone.
There are at least two explanations for the augmented cortical bone geometry in children with SSNS. First, obesity may protect against bone loss. We showed that obesity in healthy children was associated with a higher section modulus at the NN and FS after adjusting for sex, pubertal stage, and height.(26) No group differences were observed when lean mass was added to the multivariable models, which suggests an appropriate skeletal adaptation to the higher muscle mass observed in obesity. Similarly, the whole body BMC surplus in obese children with SSNS was completely attenuated after adjustment for BMI Z score.(12) In this study, the elimination of differences in section modulus after adjustment for greater muscle mass in SSNS suggests that greater biomechanical loading may play a role in the structural alterations we observed and that GC-induced obesity may have a protective effect. Accordingly, the low muscle mass in CD may contribute to low bone strength. We evaluated the relation between exposure to selected GC-sparing agents (e.g., 6-mercaptopurine in CD and cyclophosphamide in SSNS) and section moduli.(11,12) 6-mercaptopurine was associated with lower FS section modulus in CD. Consistent with the medication's use in those with greater disease severity, this effect was not independent of lean mass deficits. No significant associations between GC-sparing agents and section moduli were observed in SSNS (data not shown).
Obesity may impact bone structure indirectly by hastening pubertal development.(36) In young men, cortical thickness in the tibia and forearm were inversely related to the age at peak height velocity, a measure of pubertal timing.(37) Complex hormonal changes occur in obesity, such as elevated insulin, sex hormone, and leptin concentrations, which may increase osteoblast activity and decrease osteoclast activity.(38,39) Second, inflammation may sensitize osteoblasts to the effects of GC on bone metabolism. Pro-inflammatory cytokines induce deleterious changes in osteoblast GC metabolism by increasing the activity of 11β-hydroxysteroid dehydrogenase 1, which converts inactive GC metabolites into their active form.(40) Conversely, 11β-hydroxysteroid dehydrogenase 2, which converts active GC to inactive metabolites, was disabled by TNF-α and IL-1β. Therefore, inflammation profoundly and negatively impacts the autocrine regulation of intracellular GC action. Whereas children with SSNS frequently enter prolonged remissions on GC therapy, children with CD often suffer sustained elevations in inflammatory cytokines. We found that cumulative GC dose was associated with bone deficits only in CD. However, because a greater GC dose implies greater disease severity, a causal relationship cannot be inferred in this cross-sectional study.
Our study has limitations that should be considered. Although HSA has been used to characterize structural bone changes during childhood,(28,41–44) it has not been validated against gold-standard imaging techniques in children such as QCT. However, compared with DXA areal BMD, HSA-derived estimates of femur structure and strength showed a greater ability to predict the breaking strength of cadaveric bone.(17) The HSA program fixes the mineral density of bone at 1.05 g/cm3 in the derivation of CSA and section modulus. Because pediatric bone tissue tends to be less completely mineralized than adult bone, the CSA and section modulus may underestimate these geometric properties in children. Chronic illness during childhood may have further effects on tissue mineralization, which we are unable to assess by noninvasive means. Therefore, a higher or lower mineralization density in a disease group could influence estimates of CSA and section modulus, but would not impact the marked differences in subperiosteal width that we observed between children with CD and SSNS. This study mainly focuses on alterations in cortical bone structure and does not provide additional insight on potential trabecular deficits in SSNS. A recent report suggested that young adults with a history of childhood SSNS were at risk for low trabecular volumetric BMD at the distal radius.(45) Interestingly, they found no deficits in total BMD at the distal radius. We hypothesize that greater cortical mass and dimensions (similar to those observed here) may have counterbalanced the trabecular deficits. We observed modest deficits in spine BMC-for-bone area in children with SSNS only after accounting for their degree of obesity.(12) The effect of long-term high dose GC exposure on volumetric BMD in children with CD and SSNS has not been thoroughly studied.
In conclusion, we showed that children with CD and SSNS have striking disparities in hip structure despite chronic high-dose GC therapy. We hypothesize that the impact of GC on cortical bone structure is amplified by concomitant inflammation in CD and that obesity and the absence of persistent inflammation protect children and young adults with SSNS against the deleterious effects of GC on bone metabolism. Furthermore, this study shows the usefulness of novel imaging techniques, such as HSA, to delineate the structural manifestations of the greater and lesser bone mass in CD and NS, respectively. Future longitudinal studies using QCT and MRI are needed to determine the impact of inflammation and long-term GC therapy in children on bone quality. These studies will identify appropriate high-risk groups to target in treatment trials.
We appreciate the support of David Piccoli, MD, Robert Baldassano, MD, and the Center for Pediatric Inflammatory Bowel Disease, as well as Bernard Kaplan, MBBCh, Jorge Baluarte, MD, and the Division of Nephrology at the Children's Hospital of Philadelphia. We also appreciate the dedication of the children and their families who participated in this study. This work was supported by the General Clinical Research Center (M01RR00240) and the Nutrition and Growth Center at the Children's Hospital of Philadelphia, and the National Institutes of Health (JMB, 1K23 RR021969).