Glucocorticoid Excess During Adolescence Leads to a Major Persistent Deficit in Bone Mass and an Increase in Central Body Fat

Abstract

Endogenous Cushing's syndrome (CS) in children causes growth retardation, decreased bone mass, and increased total body fat. No prospective controlled studies have been performed in children to determine the long-term sequelae of CS on peak bone mass and body composition. A 15-year-old girl with Cushing disease (CD), and her healthy identical co-twin, were followed for 6 years after the CD was cured. At the 6-year follow-up both twins had areal bone mineral density (BMD) and body composition determined by dual-energy X-ray absorptiometry (DXA) and three-dimensional quantitative computed tomography (3DQCT). Z scores for height, weight, and body mass index (BMI) were −2.3, −0.8 and 0.2, and 1.2, 0.2, and −0.6, in the twin with CD and her co-twin, respectively. In the twin with CD, areal BMD and bone mineral apparent density (BMAD) at different sites varied from 0.7 to 3 SD below her co-twin. Volumetric lumbar spine bone density Z score was −0.75 and 1.0, and total body, abdominal visceral, and subcutaneous fat (%) was 42, 10, and 41 versus 26, 4, and 17 in the twin with CD and her co-twin, respectively. The relationship between total body fat and L2-L4 BMAD was inverse in the twin with CD (p < 0.05), which by contrast in her co-twin was opposite and direct (p < 0.001). In the twin with CD, despite cure, there was a persistent deficit in bone mass and increase in total and visceral body fat. These observations suggest that hypercortisolism (exogenous or endogenous) during adolescence may have persistent adverse effects on bone and fat mass.

INTRODUCTION

ENDOGENOUS CUSHING's syndrome (CS) causes osteoporosis and marked increases in total body fat with the classical central pattern of deposition.¹ Growth failure, weight gain, and pubertal arrest are the hallmarks of this condition in children and growing adolescents.^2-4 Little is known about the effects of endogenous hypercortisolism during puberty, a critical period during which bone and fat mass increase and growth is accelerated.⁵ Peak bone mass is attained in early adulthood and is a major determinant of the risk of osteoporosis later in life.⁶ Multiple factors influence its accrual, including genetic factors,^{7, 8} sex steroids, physical activity,⁹ dietary calcium intake,^{10, 11} and the presence of chronic disease such as rheumatoid arthritis or exposure to drugs that can alter calcium homeostasis such as glucocorticoids.^{12, 13} Hypercortisolism during childhood and adolescence may have significant effects on peak bone mass accrual and body composition.¹⁴ Only a small number of relatively short-term prospective observations on areal bone mineral density (BMD) have been reported after the surgical cure of CS in children^{14, 15} or in adults.^15-19 Changes in total body and visceral abdominal fat in CS have been described in adults^{20, 21} though the body composition changes in children after cure of CS have not been described.

To determine the extent of recovery in BMD and the long-term changes in body composition in children with CS, we followed an identical a pair of female twins for 6 years until age 21 years. One of the twins was cured from Cushing disease (CD) at the age of 15 years; her normal, healthy identical twin sister served as the genetic control. This unique pediatric case study revealed that in comparison with her identical twin sister, marked differences in bone mass and body fat persisted in the affected twin 6 years after cure of CD.

CASE REPORTS

Monozygosity of this twin pair was based on parental reporting at birth of a clear identical phenotype and on substantiating photographic evidence and identical human leukocyte antigen (HLA)-typing. The first 27 months of this case study have been reported previously.¹⁴ Despite continued weight gain and growth failure from age 10 years, the affected twin was not evaluated until 14 years of age when CD was diagnosed based on an elevated 24-h urine-free cortisol (24h-UFC) of 1509 nmol, which was suppressed 90% after high-dose dexamethasone treatment. Transsphenoidal pituitary surgery (TSS) was performed with no clinical improvement. At the age of 15 years the patient was referred to the National Institutes of Health (NIH) and underwent repeat TSS during which an ACTH-positive secreting pituitary microadenoma was removed successfully. 24h-UFC was undetectable from a week after surgery until 6 months postsurgery and then remained in the normal range during further follow-up (Table 1). Hydrocortisone therapy (initial and maximum dose at 15 mg/m2 day) was continued in decreasing stepwise doses (∼5 mg/m² day every 3 months) until cessation at 9 months when the hypothalamic-pituitary-adrenal axis was normalized as assessed by an ACTH stimulation test. In the normal co-twin, 24h-UFCs remained in the normal range throughout follow-up. Seventy-eight months after curative surgery, the patient has remained clinically well with no evidence of tumor recurrence or any other illness and is not on any medication. Her menses commenced 13 months after surgery, which from 19 months after surgery remained regular, while her co-twin has had normal menstrual periods since age 14 years. Both twins exercise 30 minutes two to three times per week and had a normal and similar calorific intake. Neither twin smokes nor drinks alcohol excessively. Their daily calcium intake was similar, estimated to be a mean of 800 mg/day during follow-up as assessed at each visit. Anthropometric data of the twins at the 6-year follow-up showed the post-CD twin was severely growth retarded with a higher body mass index (BMI) in comparison to her co-twin (Table 1). Both twins were Tanner stage V and normal on physical examination.

Table Table 1.. Growth, Body Composition, Bone Metabolism, and Bone Mass

MATERIALS AND METHODS

Clinical and endocrine assessment

Both twins were reviewed before the twin with CD had her second TSS at the age of 15 years. The twin with CD was seen at 3 months and 6 months and then together with her normal co-twin at 12, 27, and 78 months after surgery. At each review a physical, biochemical, endocrine, and radiological examination was performed. Dietary calcium intake was assessed by a dietitian using a standard self-reporting questionnaire. Height, weight, and BMI Z scores were derived from a normal North American childhood population.²² Growth hormone (GH), insulin-like growth factor (IGF-1), and 3-day first-morning pooled urinary GH were measured at Endocrine Sciences (Calabasas Hills, CA, USA)^{23, 24}; serum osteocalcin (OC) and 24-h urine pyridinium cross-links (pyridinoline) were measured at Corning-Nichols Institute (San Juan Capistrano, CA, USA)^{25, 26}; and 25-hydroxyvitamin D₃ [25(OH)D₃], 1,25-dihydroxyvitamin D₃ [1,25(OH)₂D₃], and intact parathyroid hormone (PTH) were measured at Covance Laboratories (Vienna, VA, USA), all by standard techniques.

GH stimulation test

Arginine injection (10%, 0.5 g/kg iv) was started after blood was taken at time zero. Human regular insulin, (0.1 U/kg iv) was given after the 60-minute bloods. GH, glucose, and cortisol were drawn at 0, 30, 45, 60, 75, and 90 minutes.

Bone mass measurements

Bone densitometry and whole body composition were determined by dual-energy X-ray absorptiometry (DXA) using the Hologic QDR-2000 densitometer using the enhanced whole body software version 5.71 (Hologic, Waltham, MA, USA). Bone mass was expressed as bone mineral content (BMC; g) and areal BMD (g/cm²). These values were compared with a normal North American healthy age- and sex-matched control population whose bone mass DXA variables were uncorrected for height or weight.⁶ The estimation of BMC and areal BMD are influenced by bone size, so it may be underestimated in smaller individuals.²⁷ Because the height measurement was significantly less in the twin with CD than in her co-twin, bone mineral apparent density (BMAD) was calculated as an estimation of volumetric bone mass (g/cm³).^{27, 28} For whole body bone mass, BMC was divided also by height (BMC/height) to correct partially for differences in height.⁶ Volumetric BMD of the L1 and L2 vertebral bodies was performed by single-energy three-dimensional quantitative computed tomography (3DQCT; 80 V, 200 MA) scan using a high-speed advantage CT scanner (GE Medical Systems, Milwaukee, WI, USA) and analyzed with the QCT Pro program from Mind Waves Software (San Francisco, CA, USA). Quantitative analysis of body fat was obtained from the data obtained from the BMD QCT using an image analysis mask tool. All studies were approved by the National Institute of Child Health and Human Development's (NICHD's) institutional review board. Informed written consent to participate in the study was obtained from both twins and their parents.

Statistical analysis

Correlation analysis was performed using the Statview and Super ANOVA software programs (Abacus Concepts Inc., Berkley, CA, USA).

RESULTS

Endocrine data

At the 78-month follow-up, in both twins serum and urinary biochemistry and endocrine function were all normal, including serum calcium, phosphorus, 1,25(OH)₂D₃, PTH, free T4, thyroid-stimulating hormone (TSH), estradiol, 24hUFC, and adrenal androgens (Table 1). To assess the GH axis of the twin with CD, an arginine-insulin test (AIT) was performed at her 78-month visit. The fasting glucose was 78 mg/dl (4.3 mmol/liter) and after iv insulin it decreased, with concomitant symptoms, to 17 mg/dl (0.9 mmol/liter) with a GH peak of 6.5 ng/ml (6.5 μg/liter) and cortisol of 26.5 μg/dl (731 nmol/liter). As previously reported,¹⁴ in the twin with CD IGF-1 levels preoperatively and during the first 27 months postoperatively remained slightly below normal between 182 and 224 ng/ml (24-29 nmol/liter; NR 242-550 ng/ml; 32-73 nmol/liter), whereas at the 78-month follow-up IGF-1 levels in both twins were low (Table 1).

Bone metabolism, bone density, and body composition studies

At 78 months, bone turnover markers were normal (Table 1). Whole body DXA scan of both twins at 0 years and 6 years in the twin with CD and at 6 years in her co-twin (Fig. 1) show the marked effects of hypercortisolism on skeletal size and body composition. The difference in total body fat was especially persistent and noticeable around the lower trunk and hip. The distribution of fat estimated by 3DQCT, showed marked differences in both subcutaneous and visceral fat between the twins (Fig. 1B and Table 1). Over the 6-year follow-up period lumbar spine (L2-L4); femoral neck areal BMD and BMAD; and whole body BMC, BMC/height, areal BMD, and body composition were followed in both twins. In the twin with CD, all bone mass variables measured (raw data and Z scores) at diagnosis and throughout follow-up were significantly and markedly lower than that of her normal co-twin, even after correction for bone size and height (Figs. 2 and 3). 3DQCT studies confirmed that true volumetric lumbar spine bone density also was low in the twin with CD compared with her normal co-twin (Table 1). In the twin with CD the percentage of total fat decreased significantly during the first year after cure but remained elevated at the 78-month follow-up despite a normal BMI, while in the normal twin there was little change over time (Table 1 and Fig. 3). Changes in percent lean body mass were related inversely to changes in body fat and were significantly lower in the twin with CD than in her co-twin (Fig. 3).

Figure FIG. 1

(A) Whole body DXA of affected twin at diagnosis of CD and at 6 years after follow-up when aged 21 years (left panels) compared with her normal, healthy, co-twin at 15 years and 21 years (right panels). Note the marked effects of CD on body fat composition, bone size, and bone mass between the twins. (B) QCT transverse images of the twin with CD (upper panel) and her healthy co-twin (lower panel) obtained at the 6-year follow-up at the level of lumbar 1/2 vertebrae, showing differences in subcutaneous fat (left-sided scans) and visceral body fat (right-sided scans). Fat is shown as white areas in scans (arrows). Note the marked increase in both subcutaneous and visceral fat in the twin with CD compared with her normal co-twin.

Figure FIG. 2

(A-D) Lumbar spine and femoral neck areal BMD and BMAD in twins over 78 months of follow-up after the twin with CD was cured. The twin with CD is shown by the open circles and her healthy, normal co-twin is shown by solid symbols. Z scores are shown at 0, 24, and 78 months.

Figure FIG. 3

(A-E) Whole body BMC, BMC corrected for height, areal BMD, whole body fat, and lean body mass (%) obtained by DXA in twins over 78 months of follow-up after the twin with CD was cured surgically. The twin with CD is shown by open circles and her healthy, normal co-twin is shown by solid symbols. Z scores are shown at 0, 24, and 78 months.

A correlation analysis between total body fat versus areal BMD and BMAD showed a significant inverse relationship between percentage whole body total fat and L2-4 BMAD (r² = 0.7; p < 0.05) and a nonsignificant relationship with L2-L4 BMD (p = 0.08) in the twin with CD (Fig. 4). This was in marked contrast to that in the normal twin, in whom there was a significant opposite and direct relationship present between percentage whole body total fat and L2-L4 areal BMD (r² = 1.0; p < 0.001) and a nonsignificant relationship between percent fat and L2-L4 BMAD (p = 0.12).

Figure FIG. 4

(A and B) Correlation between total body fat (%) measured by whole body DXA against lumbar spine areal BMD and BMAD in both twins. Although the relationship between fat and bone mass was inverse in the twin with CD, it was opposite and direct in the normal co-twin. The correlation analysis only reached statistical significance in the twin with CD for BMAD (p < 0.05) and in the normal co-twin for areal BMD (p < 0.001), presumably because of the small sample number.

DISCUSSION

Despite almost 70 years since the original description of CS, the long-term effects on bone mass and body composition in children with this condition remain unclear. In this prospective study, we followed an identical twin pair for 6 years to examine these long-term effects independent of genetic factors. The twin with CD was hypercortisolemic from age 10 to 15 years, a critical period during which bone mass is acquired.^{6, 29} At the 6-year follow-up, in the twin with CD, though areal BMD had increased significantly after correction of the hypercortisolism, it remained from 0.7 to 3.0 SD below the values of her normal, healthy co-twin at different sites. Furthermore, volumetric bone density as estimated by BMAD and measured by 3DQCT suggested that peak bone mass in the affected twin was compromised severely.^{6, 27} Because peak bone mass is a major determinant of long-term risk of osteoporosis,^{30, 31} these observations strongly suggest that the twin with CD is at increased risk of osteoporosis. Additionally, in the affected twin the greater body fat, especially visceral fat, did not normalize after correction of hypercortisolism, remaining abnormally high despite normalization of BMI.

This is the first report suggesting that CS may cause persistent and long-term changes in body fat composition. CS is characterized by redistribution of fat to central parts of the body, including visceral and subcutaneous fat.²⁰ The association between visceral obesity and the risk profile for cardiovascular diseases is recognized widely.³² Thus, the increased percentage body and visceral fat in the affected twin 6 years after correction of hypercortisolemia suggests her long-term risk for cardiovascular and metabolic disease also is significantly worsened.³³

In the twin with CD, one possible explanation for the incomplete recovery of bone mass was the suboptimal daily calcium intake of 800 mg. However, because both twins had a similar calcium intake, this suggests that it had only a minor influence on her rate of bone mass recovery. A more likely explanation for the severe loss in bone mass was her delayed diagnosis and 5-year history of uncontrolled hypercortisolism, which also was associated with delayed puberty and menarche and hypothalamic hypogonadism.¹⁴

Growth retardation caused during CS is followed by incomplete catch-up growth, which may be exacerbated by associated pituitary and hormonal deficiencies including abnormalities of the GH-IGF-1 and gonadal axes.^{3, 34} The syndrome of GH deficiency characteristically presents with alterations in body composition including reduced BMD, lean body mass, and increased fat mass.^35-37 The possibility that the changes in body composition and bone mass observed in the affected twin were secondary to severe GH deficiency was excluded by a relatively normal GH response to insulin-induced hypoglycemia.³⁸ Nevertheless, the twin with CD did have slightly low IGF-1 levels during follow-up, though the normal twin also had a slightly low IGF-1 level at the last assessment. However, because IGF-1 levels are unreliable predictors of GH deficiency³⁹ and the degree of bone loss correlates with the severity of GH deficiency,⁴⁰ the possibility that a mild deficiency in GH secretion caused such marked effects on bone mass or body composition in the affected twin seems unlikely. Nevertheless, the suppressive effect of severe prolonged hypercortisolism on the GH-IGF-1 and gonadal axes³⁴ and a transient state of mild GH deficiency postoperatively may have contributed partly to the observed changes in body fat and bone mass in the twin with CD. Whatever the exact mechanisms for the observed changes in bone and fat mass in the affected twin (i.e., direct or indirect via effects on the GH and/or gonadal axes), the presence of a unique genetic control in this prospective 6-year study strongly suggests that prolonged hypercortisolism during adolescence may have long-term adverse effects on both peak bone mass and body composition.

Acknowledgements

The authors thank Laura Bachrach, John A. Eisman, and Ken Ho for their valuable advice and assistance with different aspects of this study.