| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Pediatric Endocrinology (N.M.M.L.S., B.J.O.), Dental Statistics (B.A.E.v.t.H.-G., M.A.v.t.H.), and Endocrinology (A.R.M.M.H.), University Medical Center Nijmegen, 6500 HB Nijmegen, The Netherlands
Address all correspondence and requests for reprints to: Dr. B. J. Otten, 435 Department of Pediatric Endocrinology, University Medical Center Nijmegen, P.O. Box 9101, 6500 HB Nijmegen, The Netherlands. E-mail address: b.otten{at}cukz.umcn.nl.
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
Reduced adult height in CAH may result from adrenal androgen excess. Androgen excess leads to increased growth velocity, advanced skeletal maturation, and early epiphyseal fusion. In this respect the effect of poor compliance has not unambiguously been demonstrated (6, 7, 8), but the effect of late diagnosis is more evident; in patients in whom the diagnosis of CAH is made after 1 yr of age, androgen excess may cause substantial bone age advancement and loss of adult height potential (5, 6). It is obvious that this concerns predominantly classic simple virilizing and nonclassic patients.
In both classic (salt-wasting and simple virilizing) and nonclassic patients, reduced adult height may result from growth inhibition by glucocorticoid treatment. Glucocorticoids inhibit linear growth by interfering with the activity of the GH-IGF-I axis, by disturbing calcium absorption in intestine and kidney, and by impairing the function of the growth plate (9). Previous studies in CAH patients did not show any significant correlation between adult height and glucocorticoid doses at certain ages or age intervals (1, 2, 5, 10). However, in a prospective randomized cross-over trial significant negative correlations were found between growth velocity and glucocorticoid doses: growth velocity was demonstrated to be significantly decreased during treatment with 25 mg/m2 hydrocortisone compared with 15 mg/m2 in 26 children in the age range of 4 months to 15 yr (11). In addition, retrospective studies showed negative correlations between glucocorticoid dose and growth velocity in the first year(s) of life (3, 12) and between glucocorticoid dose and height at 2 yr of age (5).
These previous studies suggest that glucocorticoids have a dose-dependent negative effect on linear growth in CAH patients, and that this effect is age dependent. This prompted us to study the correlation between glucocorticoid dose and growth in CAH patients in more detail. We hypothesized that there is a dose-dependent negative effect of glucocorticoids on height and a positive effect on weight and weight for height, and that these effects are age dependent (e.g. more prominent in infancy). A retrospective study was carried out on growth data from 48 patients with classic salt-wasting 21-hydroxylase deficiency, who all had been diagnosed in the first year of life.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
The growth data of all patients with salt-wasting 21-hydroxylase deficiency, who were born between 1960 and 1999, diagnosed in the first year of life, and regularly followed up at our center (University Medical Center Nijmegen), were collected for this study. The diagnosis of salt-wasting 21-hydroxylase deficiency was made in patients who had salt-wasting crisis (dehydration, hyponatremia, and hyperkalemia), ambiguous external genitalia in females, elevated levels of serum 17-hydroxyprogesterone or urinary pregnanetriol, and elevated plasma renin activity or plasma renin concentration. All patients used glucocorticoids and mineralocorticoids from the time of diagnosis. Of a total of 64 patients, the following patients were excluded: patients who had been treated with GnRH analogs (n = 3), patients who were referred to our center after the age of 8 yr (n = 6), patients for whom growth data were incomplete (n = 5), and patients with comorbidity that can affect growth (n = 2, 1 patient with encephalopathy and muscular contractures and 1 patient with a renal disease). The remaining study population consisted of 48 patients. In 33 of the 48 patients, DNA analysis had confirmed the diagnosis of 21-hydroxylase deficiency. In the other 15 cases, DNA analysis had not been performed.
Measurements
Longitudinal data until August 2001 were obtained from the patients’ medical records in our center. Measurements of height (until 2 yr of age, supine length with supine measuring table; after 2 yr of age, standing height with stadiometer) and weight (in underwear) and the prescribed daily glucocorticoid dose had been recorded at clinic visits (commonly every 3–4 months). Measurements that had been performed at other centers (i.e. before referral to our center) were not used. In 33 patients, measurements in our center were available since the age of 0–1 yr, in eight patients since the age of 1–3 yr, and in seven patients since the age of 3–8 yr. In 16 patients, complete follow-up from less than 1 yr until 18 yr of age was available, and at the time of the analysis, 21 patients were still younger than 18 yr. Therefore, we used age intervals for data analysis. To asses the onset of pubertal development, the age at breast stage M2 (girls) or testicular volume of 4 ml (boys) was obtained from the patients’ records. The median age at M2 was 10.5 yr (range, 8.5–12.4 yr), the median age at testicular volume of 4 ml was 11.7 yr [range, 8.0–13.4 yr; Dutch reference values: median age at M2, 10.7 yr, 9.4–12.5 yr (P10–P90); median age at testicular volume of 4 ml, 11.5 yr, 13.0 yr P90, P10 not given (13)], suggesting that, in general, the timing of the start of puberty was normal.
The glucocorticoid doses were related to body surface (expressed in milligrams per square meter). For the sake of comparability, glucocorticoid formulations were transformed into an equivalent dose of hydrocortisone using dosage equivalents based on growth-suppressing effects: 30 mg hydrocortisone = 37.5 mg cortisone acetate = 6 mg prednisone = 0.375 mg dexamethasone (14). The surface area (SA) of the total body was calculated as: ln(SA) = -3.751 + 0.422 x ln(H) + 0.515 x ln(W), where ln is the natural logarithm, SA is the surface area in square meters, H is height in centimeters, and W is weight in kilograms (15). The midparental height was calculated as: (father’s height + mother’s height)/2. It was also expressed as a z-score: (father’s height z-score + mother’s height z-score)/2, based on the Dutch national growth studies of 1955 (16), 1965 (17), or 1980 (18), dependent on the year of birth of the parents.
Calculation of z-scores
Four successive Dutch national growth studies (in 1955, 1965, 1980, and 1997) had shown a clear secular trend in height and weight (13, 16, 17, 18). The patient population in the present study covered 4 decades (1960–1999), and therefore, the growth data were corrected for secular trend. The correction consisted of adding a correction factor (Hcorr and Wcorr, see below) to the actually measured height and weight, respectively. After this correction, height for age z-scores (HAZ) and weight for age z-scores (WAZ) were calculated based on the Dutch National Growth Study of 1997 (13). Obesity would have been preferably expressed as body mass index (BMI) (19), but since the Dutch National Growth Study of 1997 (13) does not provide z-scores for BMI, the weight for height z-score (WHZ) was used. For the sake of simplicity, no correction for secular trend in WHZ was applied, and in all patients WHZ values were calculated based on the 1997 Dutch National Growth Study.
Correction for secular trend: correction factors for height (Hcorr) and weight (Wcorr)
The P50 values at all ages for height and weight from the Dutch National Growth Studies of 1965, 1980, and 1997 were used to calculate correction factors for height (Hcorr) and weight (Wcorr). The difference between the P50 values in 1965 and 1980 compared with the P50 values in the 1997 standard was explained by age and year of birth. With polynomial regression, the best-fitting model was found, using a third degree polynomial in age and year of birth, including interaction between age and year of birth (product variable). The formulas for the correction factors Hcorr and Wcorr are given in Table 1
. The correction factors were added to the actually measured height and weight, respectively, to correct for secular trend. For the ease of analysis, linearly interpolated anthropometric and medication values were calculated at integer and half-year ages.
|
The effect of glucocorticoid dose on the dependent variables HAZ, WAZ, and WHZ was studied by ANCOVA. The dependent variables at 10 selected ages (1, 2, 4, 6, 8, 10, 12, 14, 16, and 18 yr) were explained by the mean daily glucocorticoid dose per body surface in the preceding time interval. This preceding interval was 0.5 yr before the age of 1 yr, 1 yr before the age of 2 yr, and 2 yr before the ages of 4, 6, 8, 10, 12, 14, 16, and 18 yr, respectively. The HAZ, WAZ, or WHZ value at the beginning of the age interval (initial value) was included as a covariable in the ANCOVA, instead of an analysis of growth increments (change scores), which is less powerful (20). As additional covariables, the z-score of midparental height (only in ANCOVA for HAZ) was included to correct for genetic influences, and the value 1 yr before the beginning of the considered age interval was included to correct for the past growth history (e.g. catch-up growth). The ANCOVA for height was redone to control for sex influences. In the ANCOVA for WAZ and WHZ, midparental height was not included, because it was assumed to be not relevant for WAZ and WHZ. Finally, the ANCOVA for height was also performed on the absolute height measurements (in centimeters), without correction for secular trend, as a check for the sensitivity to the secular trend. In this ANCOVA, sex and midparental height were also included as covariables.
The structure of all applied ANCOVAs is presented in Table 2. For example, the formula for HAZ at a certain age (HAZage) is: HAZage = constant + B1 x (glucocorticoid dose in the preceding age interval) + B2 x (HAZ at beginning of the preceding age interval) + B3 x (HAZ at 1 yr before the concerning age interval) + B4 x (midparental height z-score). The statistical analyses were performed with SAS computer software (SAS Institute, Inc., Cary, NC). P < 0.05 was considered significant. As this study was an audit, no informed consent was required.
|
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
. In general, patients born between 1960 and 1975 had higher glucocorticoid doses than patients born later. Until the early eighties, all patients used cortisone acetate. In the early eighties, cortisone acetate was replaced by hydrocortisone. Dexamethasone was used in combination with hydrocortisone or as a single preparation (twice a day) by 14 patients (median age at start dexamethasone, 16.3 yr; range, 13–19 yr).
|
. Therefore, in all ANCOVA tables, the number of included cases is given separately. The lowest number of cases is in the ANCOVA for HAZage, because parental heights were missing in some patients. The covariables HAZ, WAZ, and WHZ 1 yr before the age interval (to correct for the past growth history) had no relevant or significant contribution in the ANCOVA for HAZage, WAZage, and WHZage, respectively. Therefore, these covariables were left out of the final ANCOVAs for efficiency reasons.
The ANCOVA analysis for the height z-score (Table 4) showed that the mean daily glucocorticoid dose between 0.5 and 1 yr of age had a significant negative effect on height z-score at the age of 1 yr (HAZ1 yr). The mean daily glucocorticoid dose between 8–10 yr of age and between 12–14 yr of age had significant negative effects on HAZ at the age of 10 yr (HAZ10 yr) and 14 yr (HAZ14 yr), respectively. For the age interval 10–12 yr, there was a trend toward significance. There was no significant effect of the mean daily glucocorticoid dose on HAZ at the other ages. The initial values (HAZ at the beginning of each interval) had a significant positive contribution to HAZ at all ages (P < 0.0001). The midparental height z-score had a significant positive effect at HAZ1 yr (P < 0.02), but no significant effect on HAZ at the other ages. When correction for sex was applied, similar results were found. The covariable sex itself also had a significant contribution to HAZ, but only at the age of 10 yr (HAZ10 yr) and 14 yr (HAZ14 yr). Interestingly, the sex effects were opposite in the two age intervals. At 10 yr of age, female patients showed a lower HAZ of 0.2 compared with male patients (P = 0.01), and at 14 yr of age, male patients showed a lower HAZ of 0.6 compared with female patients (P = 0.002).
|
). The midparental height was not significant in any of the age intervals, but showed a trend toward significance in the period between 6 months and 1 yr (P = 0.053). The initial values (height at the beginning of each interval) had a significant contribution to height at all ages (P < 0.0005).
|
). The influence of the initial values (WAZ at the beginning of each interval) was significant in all cases (P < 0.0001).
|
). The influence of the initial values (WHZ at the beginning of each interval) was significant in all cases, except WHZ1 yr.
|
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
For the analysis of growth data, increment analysis (or change score analysis) is often used. In increment analysis, the difference between successive measurements (or the growth velocity) is the dependent variable, and the glucocorticoid dose is the independent variable. In the present study, however, we used ANCOVA, which means that we have analyzed the effect of glucocorticoid dose in a certain age interval (independent variable) on height z-score at the end of that interval (dependent variable), including the height z-score at the beginning of the interval (baseline) as a covariable (among other covariables). For the purpose of this study, ANCOVA analysis is superior to increment analysis for two equivalent reasons. The increment is the difference between two observations and contains the measuring error twice. In addition, ANCOVA also assigns a regression coefficient to the baseline z-score, which will give a better fit and a smaller residual variance (20).
In this study glucocorticoid dose was found to have growth-suppressing effects during the age intervals of 6 months to 1 yr, 8–10 yr, and 12–14 yr. These intervals correspond to the periods of high growth velocity in healthy children (21) and to the periods in which growth velocity in CAH patients is compromised (1, 4, 12, 22, 23). It is likely that the loss of growth potential in these periods can hardly be corrected later on (2, 3). The height correction for secular trend appeared to be not crucial, as analysis of the absolute height data without correction for secular trend yielded similar results in terms of both statistical significance and size of the glucocorticoid effect. Nevertheless, it was maintained in the analyses.
In the first year(s) of life, decreased growth velocity in CAH patients has been described previously (1, 4, 12, 23), and significant negative correlations have been found between the mean daily glucocorticoid dose and height velocity in the first year of life (increment analysis) (3, 12) and between the glucocorticoid dose in the first 2 yr and height at 2 yr of age (5). The results of our study confirm that glucocorticoid treatment is critical for linear growth in the first year of life. In this period overdosing should be avoided to allow optimal height gain. The risk that androgen excess in this period will compromise growth is thought to be small, as it has been suggested that growth during the first 1.5 yr is not very sensitive to androgens (24). After the age of 1 yr and before the age of 8 yr, we found no significant negative effect of glucocorticoid dose on height z-score. This implies that in this period, glucocorticoid dosing is less critical for linear growth, and that a higher dose is allowed to obtain adequate androgen suppression.
In puberty, reduced height velocity or reduced height gain have often been described in CAH patients (1, 5, 22, 23), but only a few reports have studied the correlation with glucocorticoid dose in this period. One study showed a weak negative, although significant, correlation between glucocorticoid dose and height velocity in puberty, but only when height velocity was considered for bone age (not for chronological age) (25). The results of our study show that the negative effect of the glucocorticoid dose on height z-score between the age of 12–14 yr was highly significant and even as large as in the age period between 6 months and 1 yr. In the age period 8–10 yr, the negative effect was smaller, but still significant, and between the age of 10–12 yr, there was a trend toward significance. These results suggest that also in puberty, glucocorticoid dosing is critical for height gain, and that from the age of 8 yr, overtreatment should be avoided to allow optimal height gain. After the age of 14 yr, there was no significant effect of glucocorticoid dose on height z-scores.
Correction for sex in the ANCOVA for HAZage showed that sex had a significant influence on HAZ at the age intervals of 8–10 and 12–14 yr. As we did not investigate pubertal development in this study, interpretation of this effect (which was opposite in the two age intervals) remains speculative. Despite the significant effect of sex, the glucocorticoid effects remained significant.
At all ages, the HAZ were significantly influenced by the z-score at the beginning of the interval, as can be expected given the fact that growth usually occurs along an established growth channel. The midparental height z-score had a significant contribution at the age of 1 yr, which probably reflects target seeking; until birth, growth is predominantly determined by intrauterine conditions, but after birth, growth is increasingly determined by the genetic potential; catch-up or catch-down growth toward the genetic growth channel occurs until the age of 20 months (26). After the age of 1 yr, the midparental height z-score had no significant contribution. The latter finding might be explained by the highly significant contribution of the covariable height z-score at the beginning of the interval, which represents, in fact, the genetic growth channel.
There were no significant effects of glucocorticoid dose on weight for age z-scores and weight for height z-scores in all age intervals. This corresponds with previous reports of CAH patients, which showed no significant correlations between glucocorticoid dose and BMI z-scores in successive developmental periods (5) or no significant correlations between glucocorticoid dose and BMI in early and later childhood (27). Also, no significant correlations were found between adult BMI and mean long-term glucocorticoid dose (28) or mean glucocorticoid dose in successive age intervals (2). Obviously, the hypothesis that there is a dose-dependent positive effect of glucocorticoid dose on weight and weight for height is not true in these children and adolescents with CAH. In one study, however, it was shown that treatment with more than 30 mg/m2·d in the first 2 yr of life was associated with obesity at school age in 75% of CAH patients, whereas treatment with less than 30 mg/m2·d was associated with obesity in 11% (29). This suggests that glucocorticoid-induced weight gain and obesity is probably only relevant above a certain, relatively high, glucocorticoid dose.
One of the limitations of our study was that insufficient data on birth weight and height were available to perform an ANCOVA in the age interval 0–6 months. In addition, the number of patients in the last age interval (16–18 yr) was relatively small. Glucocorticoid doses were collected in detail, based on patient records (prescribed doses), but no systematic information was available about compliance or about temporary overdosing in the case of intercurrent illnesses. In addition, we could not include other treatment characteristics, such as mineralocorticoid replacement and salt supplementation, which could also affect growth, especially during the first year of life. The glucocorticoid dose effects on growth were found in an observational study and not in a randomized clinical trial. Therefore, it cannot be excluded that the observed effects are biased due to confounding by indication. This is, however, unlikely, because the glucocorticoid dose differences between patients corresponded mainly with the year of birth (e.g. patients who were born between 1960 and 1975 had higher doses than the patients who were born later).
In conclusion, this study showed that in salt-wasting CAH patients, daily glucocorticoid doses had significant negative effects on linear growth between the age of 6–12 months, 8–10 yr, and 12–14 yr. The negative dose effect on height z-score in the age interval of 12–14 yr was as large as that in the interval between 6–12 months of life, which had already been identified as a critical period for growth and glucocorticoid dose. Thus, in the first year of life and between the age of 8–14 yr, the daily glucocorticoid dose should be sufficient to avoid androgen excess, but as low as possible to allow optimal linear growth and adult height. Weight and weight for height were not significantly influenced by glucocorticoid dose in any of the age intervals.
|
Footnotes |
|---|
Received December 30, 2002.
Accepted May 6, 2003.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  W. Bonfig, S. B. Dalla Pozza, H. Schmidt, P. Pagel, D. Knorr, and H. P. Schwarz Hydrocortisone Dosing during Puberty in Patients with Classical Congenital Adrenal Hyperplasia: An Evidence-Based Recommendation J. Clin. Endocrinol. Metab., October 1, 2009; 94(10): 3882 - 3888. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z. Chakhtoura, A. Bachelot, D. Samara-Boustani, J.-C. Ruiz, B. Donadille, J. Dulon, S. Christin-Maitre, C. Bouvattier, M.-C. Raux-Demay, P. Bouchard, et al. Impact of total cumulative glucocorticoid dose on bone mineral density in patients with 21-hydroxylase deficiency. Eur. J. Endocrinol., June 1, 2008; 158(6): 879 - 887. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. Bonfig, S. Bechtold, H. Schmidt, D. Knorr, and H. P. Schwarz Reduced Final Height Outcome in Congenital Adrenal Hyperplasia under Prednisone Treatment: Deceleration of Growth Velocity during Puberty J. Clin. Endocrinol. Metab., May 1, 2007; 92(5): 1635 - 1639. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. L. Claahsen-van der Grinten, K. Noordam, G. F. Borm, and B. J. Otten Absence of Increased Height Velocity in the First Year of Life in Untreated Children with Simple Virilizing Congenital Adrenal Hyperplasia J. Clin. Endocrinol. Metab., April 1, 2006; 91(4): 1205 - 1209. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. G. Forest Recent advances in the diagnosis and management of congenital adrenal hyperplasia due to 21-hydroxylase deficiency Hum. Reprod. Update, November 1, 2004; 10(6): 469 - 485. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
