| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ORIGINAL RESEARCH COMMUNICATION |
1 From the Department of Pediatrics, Division of Nephrology, The Children's Hospital of Philadelphia, University of Pennsylvania School of Medicine, Philadelphia (BJF and MBL); the Department of Pediatrics, Division of Gastroenterology and Nutrition, The Children's Hospital of Philadelphia, University of Pennsylvania School of Medicine, Philadelphia (BZ); and the Department of Epidemiology and Biostatistics, University of Pennsylvania School of Medicine, Philadelphia (BJF, JS, and MBL)
2 Supported by NIH grants K08-DK02523 (MBL), F32DK62637-01 (BAF), The Children's Hospital of Philadelphia General Clinical Research Center (M01RR00240), and a Duncan L Gordon Fellowship from the Hospital for Sick Children Foundation, Toronto, Canada (BJF). 3 Reprints not available. Address correspondence to BJ Foster, Montreal Children's Hospital, Division of Nephrology, 2300 Tupper Avenue, Montreal, QC, Canada H3H 1P3. E-mail: beth.foster{at}muhc.mcgill.ca.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONSUMMARYREFERENCES |
|---|
Objective: We sought to assess the effect of glucocorticoids on height and body composition by comparing children with SSNS with concurrent healthy reference children. We hypothesized that chronic glucocorticoid therapy leads to obesity, decreased lean mass, and distorted distributions of fat and lean.
Design: We performed a cross-sectional study of 52 subjects with SSNS (4–21 y) and 259 reference subjects. The evaluation included height, weight, and pubertal status. Fat and lean masses were assessed by dual-energy X-ray absorptiometry in all subjects. Lifetime glucocorticoid exposure was recorded for subjects with SSNS. Outcomes were expressed as SD scores (SDS).
Results: Forty-one percent of subjects with SSNS were obese [body mass index (BMI) > 95th percentile], but regional fat distribution was normal. Mean total lean mass–for-height was 0.43 SD (95% CI: 0.15, 0.72) higher and mean appendicular lean mass–for-total-lean-mass was lower (–0.39 SD; 95% CI: –0.64, –0.14) in SSNS compared with reference children. The mean height-SDS in SSNS was –0.08 SD (95% CI: –0.37, 0.21) relative to national reference data, but height-SDS was significantly decreased given the degree of obesity. Height-SDS was positively associated with BMI-SDS among subjects with SSNS.
Conclusion: Glucocorticoid therapy for SSNS is complicated by obesity and relatively low appendicular lean mass. Overall height-SDS is normal because of a mitigating effect of elevated BMI on glucocorticoid-induced growth retardation.
Key Words: Obesity • glucocorticoids • growth • body composition • steroids • nephrotic syndrome • body habitus
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONSUMMARYREFERENCES |
|---|
Numerous disorders are treated with prolonged glucocorticoid therapy. Most are associated with systemic inflammation, which can independently influence growth and body composition. In contrast, childhood steroid-sensitive nephrotic syndrome (SSNS) remits completely in response to glucocorticoid therapy. Although protracted courses of glucocorticoids are usually required to maintain remission, prolonged systemic inflammation is not a feature (10). For these reasons, SSNS is an ideal model to study the influence of exogenous glucocorticoids on growth and body composition.
Body composition was evaluated in several studies of individuals exposed to exogenous glucocorticoids, including individuals with giant cell arteritis (11), systemic lupus erythematosus (12), and renal transplant recipients (13). However, none addressed the confounding influence of underlying disease activity.
Previous studies of children with SSNS suggested minimal height deficits (2, 3, 14). However, those studies did not consider the influence of obesity on height. Obesity results in accelerated growth and skeletal maturation in otherwise healthy children (15, 16). The interactions between glucocorticoid-induced obesity and growth deficits were not examined.
The aim of this cross-sectional study of children and adolescents with SSNS was to assess height and body composition in children treated with glucocorticoids for SSNS compared with concurrent healthy reference children. We hypothesized that chronic high-dose glucocorticoid therapy would lead to greater adiposity, relative deficits in lean mass, and distorted distributions of fat and lean. Furthermore, we hypothesized that obesity would mitigate glucocorticoid-related growth retardation among children with SSNS.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONSUMMARYREFERENCES |
|---|
Reference subjects were identified through 2 mechanisms. First, 186 healthy subjects, aged 4–21 y, were recruited from general pediatric clinics and the surrounding community through newspaper advertisements and letters. Subjects with chronic medical conditions potentially affecting growth, pubertal development, or body composition were excluded. Obesity was not an exclusion criterion. These subjects are referred to as the "community-based reference group."
Second, additional obese subjects were identified through a systematic review of all clinic charts of patients seen in the Children's Hospital of Philadelphia Weight Management Program over a 3-y interval. All children seen in the Weight Management Program underwent a whole-body dual-energy X-ray absorptiometry (DXA) at the time of the initial evaluation. Approval from the institutional review board was obtained to analyze the DXA scans and to review the medical records. The same exclusion criteria were applied to these subjects as described earlier. Seventy-three eligible subjects were identified and are referred to as the "weight-management reference group." This oversampling of obese children was intentional. The weight-management reference group and the community-based reference group were combined to create a "complete reference group" of 259 children. By oversampling obese children we ensured that the complete reference group included individuals with body sizes spanning the entire range present in the SSNS group.
Data collection
All study visits were scheduled after at least 14 d of urinary remission (defined as negative or trace proteinuria by dipstick) to allow complete resolution of edema. Negative or trace proteinuria by dipstick was documented at the study visit. The absence of edema was confirmed by physical examination in all subjects with SSNS.
SSNS disease characteristics
Medical charts were reviewed for date of diagnosis of SSNS and total number of relapses. All doses of prednisone and methylprednisolone in the interval between the date of diagnosis and the study visit were documented. Methylprednisolone doses were converted to prednisone equivalents (0.8 mg methylprednisolone = 1.0 mg prednisone), and the total cumulative glucocorticoid doses were calculated. Glucocorticoid exposure was summarized as the cumulative amount (in mg, mg/kg, and mg · kg–1 · d–1) between the first dose and the last dose. Doses of glucocorticoid per unit body weight were calculated by using the child's weight at the time the glucocorticoid doses were taken. Insufficient data on height were available at each dosing interval to express glucocorticoid per body surface area.
Anthropometry and Tanner staging
Evaluation included height to the nearest 0.1 cm (wall-mounted stadiometer; Holtain, Crosswell, Wales) and weight to the nearest 0.1 kg (digital scale; Scaletronix, White Plains, NJ). Pubertal status was determined by physical examination and classified according to the method of Tanner (19).
DXA whole-body scans
Body composition was measured by DXA (Hologic QDR 2000, Waltham, MA) with a fan beam in the array mode. All subjects were assessed on the same machine using standard positioning techniques. In keeping with the hypotheses, the primary body composition outcome variables were total fat mass, total lean mass, trunk fat mass, and appendicular lean mass. Lean mass measures excluded bone mineral mass. All results excluded the head. The lean and fat mass measures were divided into trunk and appendicular components by using standardized landmarks, as illustrated in Figure 1
. These landmarks differed from the landmarks routinely used to assess bone. A single operator (BF) performed all analyses.
|
0.05 was considered to be statistically significant for all a priori questions. When >2 groups were compared within an outcome variable, the Bonferroni correction was applied. Mean differences between SSNS and reference subjects were assessed by using the t test. The race, sex, and pubertal status distributions were compared by using the chi-square test.
SD scores
All height and body composition outcome measures were expressed as SD scores (SDS). The SDS values were generated in 2 ways. First, sex-specific height-for-age- and body mass index (BMI)–for-age-SDS values were generated by using National Center for Health Statistics (NCHS) 2000 reference data in all subjects (20). Second, SDS values were generated for each body composition parameter as follows. Data from the community-based reference group were used to generate sex-specific regression equations to predict expected values of the following: total-fat-mass-for-height, total-lean-mass-for-height, appendicular-lean-mass-for-total-lean-mass, and trunk-fat-mass-for-total-fat-mass. The latter 2 parameters provided information about the proportion of the total lean mass found in the limbs and the proportion of the total fat mass distributed to the trunk, respectively. Variables were log transformed where appropriate to improve the fit of the models. The residuals from each of the models were evaluated for evidence of heteroskedasticity. When this was present, the method of Altman (21) was used to calculate the SD at each level of the predictor variable. When residuals were uniform at all levels of the predictor variable, the root mean square error served as the SD at all levels of the predictor variable. The SDS values were calculated as follows: [(observed -expected)/SD]. A fat-mass-for-height-SDS of 2.0, for example, indicated that the subject had a fat mass that was 2 SDs above that expected in community-based reference subjects of the same height and sex.
All SSNS and reference subjects were assigned a SDS for each outcome. Because the community-based reference group provided the data from which the SDS values were calculated, the mean ± SD SDS values for each outcome within this group was 0 ± 1.0, by definition.
Group comparisons
To allow comparisons between subjects with SSNS and the complete reference group and to permit adjustment for covariates, the body composition SDS values were compared between SSNS and reference subjects by using multivariable regression. Table 1 summarizes the models used for these comparisons, including the reference group with whom SSNS subjects were compared and the covariates in each model.
|
For 2 reasons it was important to use the complete reference group, which oversampled obese children, for these comparisons. The first reason is that regional body composition in thechildren with SSNS must be compared with that in reference children across a comparable range of total fat and lean masses Second, because lean mass increases with increasing adiposity (22, 23), lean masses must be compared in children of similar adiposity.
Verification of analyses comparing SDS values
Other investigators used ratios to compare DXA parameters between a target group and a healthy reference group (23, 24). This method uses multivariable regression analysis to compare log transformed body composition parameters between 2 groups, rather than comparing SDS values. All body composition analyses in this report were repeated by using the ratio method. For example, log trunk fat was regressed against log total fat, including terms for group (SSNS compared with reference) and potential confounders (race, sex, pubertal status, height). All conclusions were unchanged.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONSUMMARYREFERENCES |
|---|
. A greater proportion of subjects with SSNS than reference subjects were male, reflecting the known male predominance of this disorder.
|
The distributions of height-for-age-SDS and BMI-for-age-SDS for SSNS and reference subjects are summarized in Figure 2. The figure illustrates the mean height and BMI in subjects with SSNS and reference subjects compared with the NCHS reference group, on whom standard growth charts are based. By definition the NCHS population has mean height-for-age-SDS and BMI-for-age-SDS of 0 ± 1.0 SD (50th percentile). Because comparisons in height-for-age-SDS values were made for 6 pairs (community-based reference compared with NCHS, weight management reference compared with NCHS, complete reference compared with NCHS, SSNS compared with NCHS, SSNS compared with community-based reference, and SSNS compared with complete reference), the Bonferroni correction was used: corrected P values are reported (P value multiplied by 6). The community-based reference group was slightly taller (
height-for-age-SDS: 0.34 SD; 95% CI: 0.19, 0.49) and heavier ( BMI-for-age-SDS: 0.35 SD; 95% CI: 0.19, 0.51) than the NCHS population (both P < 0.0006). These SDS values represent the 63rd and the 64th percentiles, respectively. The trend to an increased prevalence of childhood obesity (26) likely accounts for these findings. As expected, height was markedly increased among the 73 children in the weight-management reference group compared with the NCHS population: mean height-for-age-SDS was 1.27 SD (95% CI: 1.03, 1.51; P < 0.0006), equivalent to the 90th percentile. The complete reference group was also tall ( height-for-age-SDS: 0.60 SD; 95% CI: 0.46, 0.74; equivalent to the 73rd percentile) because of the inclusion of a large number of obese children in this group (P < 0.0006). Subjects with SSNS had a mean height-for-age-SDS of –0.08 SD (95% CI: –0.37, 0.21). This represents the 47th percentile and was not significantly different from the 50th percentile.
|
height-for-age-SDS: –0.33 SD; 95% CI: –0.65, –0.0018; P = 0.29). However, in comparison to the complete reference group, in whom the prevalence of obesity was not significantly different (39% compared with 41%, P = 0.81), children with SSNS had a statistically significant height deficit (difference in height-for-age-SDS: –0.62 SD; 95% CI: –0.95, –0.29; P < 0.0006).
The relations between height-for-age-SDS and BMI-for-age-SDS in the SSNS and complete reference groups are summarized in Figure 3. Height-for-age-SDS was not correlated with BMI-for-age-SDS in reference subjects except when BMI-for-age-SDS was greater than 
1.0. In contrast, among subjects with SSNS, height-for-age-SDS was positively associated with BMI-for-age-SDS across the range of BMI-for-age-SDS until a BMI-for-age-SDS of 2.0, at which point greater BMI-for-age-SDS did not appear to be associated with greater height. The relation between BMI-for-age-SDS and height-for-age-SDS was significantly different for children with SSNS compared with reference children (P = 0.008 for the group by BMI-for-age-SDS interaction terms, and P = 0.009 for the group by BMI-for-age-SDS-squared interaction terms). The greatest height deficits were seen among the subjects with SSNS with BMI-for-age-SDS < 0 (ie, BMI-for-age < 50th percentile); the mean height-for-age-SDS in this subgroup was markedly decreased at –1.41 SD (95% CI: –2.04, –0.79; P = 0.0002). When these analyses were limited to the prepubertal subjects (n = 44 SSNS and 192 reference subjects), results were almost identical; the only difference was that there appeared to be a slight positive association between BMI-for-age-SDS and height-for-age-SDS even at BMI-for-age-SDS > 2.0. However, the relation still differed significantly in SSNS compared with reference subjects (interaction P < 0.01). No significant relation was observed between BMI-for-age-SDS and height-for-age-SDS in pubertal subjects with SSNS; however, numbers were small in this subgroup.
|
. This figure illustrates differences in mean body composition SDS, with 95% CI around the estimated difference. Note that Figure 4 is different from Figure 2
in that Figure 4 shows differences in SDS, whereas Figure 2 shows mean SDS.
|
: 0.08 SD; 95% CI: –0.13, 0.30; P = 0.46).
Lean mass
Lean-mass-for-height was first assessed among the subjects in the weight-management reference group. Consistent with earlier studies (22, 23), otherwise healthy obese subjects exhibited high lean-mass-for-height, with a mean lean-mass-for-height-SDS 0.52 SD (95% CI: 0.23, 0.81) higher than that in the community-based reference group (P < 0.0001). Among subjects with SSNS there was no evidence for reduced total-lean-mass-for-height, even after adjusting for the degree of adiposity. In fact, the mean total-lean-mass-for-height-SDS was 0.43 SD (95% CI: 0.15, 0.72) higher in subjects with SSNS compared with all reference subjects (P = 0.003), after adjusting for race, sex, pubertal status, height, and total fat mass.
The amount of appendicular lean mass, relative to total lean mass, was expressed as appendicular-lean-mass-for-total-lean-mass-SDS. Children with SSNS had a significantly lower mean appendicular-lean-mass-for-total-lean-mass-SDS than the complete reference group (difference in : –0.39 SD; 95% CI: –0.64, –0.14: P = 0.002), after adjustment for race, sex, pubertal status, and height, indicating a deficit in this appendicular lean.
Relation between glucocorticoid exposure, height, and body composition
Details of the SSNS course and exposure to glucocorticoids are presented in Table 3. Height-for-age-SDS was inversely correlated with cumulative mg glucocorticoid/kg (r = –0.44, P = 0.001). None of the glucocorticoid exposure measures was associated with fat-mass-for-height, lean-mass-for-height, trunk-fat-mass-for-total-fat-mass, or appendicular-lean-mass-for-total-lean-mass.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONSUMMARYREFERENCES |
|---|
Previous work has shown that otherwise healthy obese children have tall stature for age (15, 16, 23); this phenomenon was also observed in the obese children in our weight-management reference group. This finding could be due to decreased insulin-like binding protein-I, which leads to an elevation of insulin-like growth factor-I (IGF-I) (15). An influence of obesity on IGF-I concentrations has also been postulated in Cushing disease (27). IGF-I stimulates linear growth. Obesity is associated with increased growth velocity in the prepubertal period only and is accompanied by accelerated sexual and skeletal maturation (16). Growth then slows during puberty, leading to a convergence of the growth curves for obese and nonobese children by the age of 18 y (16). So, although obese children are tall during childhood, their final height is not increased (16).
Despite the high prevalence of obesity among children treated with glucocorticoids for SSNS, there is no evidence to date for accelerated maturation. In fact, Rees et al (6) described delayed sexual maturation, accompanied by blunted pulsatility of growth hormone and gonadotrophins in a group of boys on glucocorticoids for SSNS. The researchers concluded that glucocorticoids can interfere with the onset and progression of puberty (6). No comments were made about obesity in these children.
Children with SSNS in our study had a mean height for age that was not significantly different from the 50th percentile but substantial height deficits given the degree of obesity. This finding is consistent with a mitigating effect of glucocorticoid-induced overweight and obesity on glucocorticoid-related growth retardation. Among subjects with SSNS, a higher BMI-SDS was associated with a higher height-SDS. It was not necessary that children be obese, or even overweight, to reap the statural benefits of a higher BMI. In fact, no relation between BMI-SDS and height-SDS was evident for children with SSNS with a BMI-SDS in the obese range (
2). This flattening of the curve appears to be driven, in part, by pubertal subjects, although the relation also leveled off in the subset of prepubertal subjects. The relation was different in healthy obese children, in whom a correlation between BMI-SDS and height-SDS was only evident when BMI-SDS was above 1.0. It is not clear why the relation between BMI and height should be different for children with SSNS and healthy children. We hypothesize that obesity attenuates glucocorticoid-related growth retardation without concomitant acceleration of sexual maturation, resulting in a fairly normal final adult height.
The obesity seen in children with SSNS is part of what is often described as a Cushingoid body habitus. Although several studies on the influence of exogenous glucocorticoids on body composition were performed in adults (11-13), all were conducted in individuals with disorders associated with systemic inflammation or with fluid imbalances. Because systemic inflammation can independently influence body composition, disentangling the effects of glucocorticoids from those of inflammation is difficult. Fluid deficit or overload affects lean mass. Our study is strengthened by enrollment of subjects in remission with no demonstrable edema.
To our knowledge, only one study has evaluated the influence of exogenous glucocorticoids on regional body composition (11). That longitudinal study of 24 adults treated with glucocorticoids for giant cell arteritis showed that total fat mass (by DXA) increased with exposure to glucocorticoids and was positively associated with cumulative glucocorticoid dose. Total lean mass also increased; however, the researchers did not assess whether the increase was consistent with the degree of obesity. An increase in trunk fat mass as a percentage of body weight was also reported. However, when abnormalities in both fat and lean are possible, expressing measures relative to total weight makes interpretation difficult. In contrast, our study did not show a Cushingoid fat distribution in subjects exposed to glucocorticoids.
Contrary to our hypothesis, there was no evidence of a total lean mass deficit. Because obesity is associated with greater lean mass, we believed that relative deficits in lean mass were masked by the obesity seen in SSNS. However, even after controlling for degree of adiposity, lean mass was greater in the SSNS group. This was likely due to the known salt and water-retaining effects of glucocorticoids (28).
Appendicular lean mass was lower in the children with SSNS. There are 3 possible explanations for this finding: (1) there was muscle wasting in the limbs, (2) there was a preferential distribution of retained fluid to the trunk, or (3) growth retardation was more severe in the appendicular skeleton and relatively preserved in the axial skeleton. It seems unlikely that muscle wasting in the limbs would be associated with greater overall lean mass. Although it is possible that excess fluid is distributed preferentially to the trunk, there is no physiologic basis for this theory. We were unable to evaluate the possibility that limb growth was more severely affected by glucocorticoids than trunk growth because sitting height was not measured. However, we believe this to be the most likely explanation for this finding.
There was a significant relation between glucocorticoid exposure and height, but we did not demonstrate a relation between cumulative dose of glucocorticoids and body composition. Subjects with SSNS were studied at varied time points in the course of their disease, so in some cases the greatest glucocorticoid exposure was years before the study visit. Unlike height, body composition indicators can increase or decrease relatively rapidly, so it is not surprising that body composition correlates poorly with lifetime glucocorticoid exposure. Inability to evaluate for a dose-response relation between glucocorticoids and changes in body composition is a limitation of this cross-sectional study.
The most important limitation of this study is that children with prevalent, rather than incident, SSNS were studied. Because the children were at different stages in their disease, the pattern of glucocorticoid exposure varied from subject to subject. It is possible that regional body composition abnormalities would be more evident in individuals with uniform daily glucocorticoid exposure. Height deficits would almost certainly be more striking in children with long periods of daily glucocorticoid exposure. In addition, reliable height and weight data before starting glucocorticoid therapy were not available for subjects with SSNS. Therefore, it is impossible to establish a temporal relation between glucocorticoid exposure and incident obesity.
|
SUMMARY |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONSUMMARYREFERENCES |
|---|
We also demonstrated a clear relation between BMI and height in children exposed to glucocorticoids. Overweight and obesity appear to protect against glucocorticoid-related growth retardation in SSNS. Further studies are required to clarify the mechanisms responsible for the interactions between obesity, growth, and glucocorticoids.
|
ACKNOWLEDGMENTS |
|---|
BJF designed the study and did the background research, most of the data analysis, and the writing. JS provided statistical expertise and helped with the analyses. BSZ provided assistance with the study design and with data analysis and contributed valuable body composition expertise. MBL was instrumental in the study design and data collection, aided with data analyses, and provided guidance in the writing of the manuscript. The authors had no conflicts of interest to report.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONSUMMARYREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. Mok, C. To, and K. Ma Changes in body composition after glucocorticoid therapy in patients with systemic lupus erythematosus Lupus, November 1, 2008; 17(11): 1018 - 1022. [Abstract] [PDF]
|
|
|
|
 |
|
 M. B. Leonard Glucocorticoid-Induced Osteoporosis in Children: Impact of the Underlying Disease Pediatrics, March 1, 2007; 119(Supplement_2): S166 - S174. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M Burnham, J. Shults, E. Semeao, B. J Foster, B. S Zemel, V. A Stallings, and M. B Leonard Body-composition alterations consistent with cachexia in children and young adults with Crohn disease Am. J. Clinical Nutrition, August 1, 2005; 82(2): 413 - 420. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
