Clinical Endocrinology (1992) 36,411-416
Changes in growth hormone concentrations during puberty in adolescents with insulin dependent diabetes Jennifer A. Batch and George A. Werther Department of Endocrinology, Royal Children’s Hospital, Melbourne, Victoria, Australia. (Received 19 August 1991; returned for revision 23 September 1991; finally revised 31 October 1991; accepted 13 November 1991)
Summary OBJECTIVE To document the changes in pulsatile growth
hormone secretion in diabetic adolescents during puberty, and to investigate their relationship to both metabolic control and stature. DESIGN Auxological parameters, overnight growth hormone secretion, fasting IGF-I, hourly glucose and metabolic control were assessed In a group of adolescentswith diabetes. PATIENTS Fifty-two diabetic adolescents (28 males and 24 females) at different pubertal stages and with varying degrees of metabolic control were studied. Ten of those with poor diabetic control were studied on two occasions. MEASUREMENTS Height and weight measurements, pubertal staging, growth velocity data and bone age estimation were obtained on all the patients. Overnight growth hormone profiles (Pulsar program analysis), glycosylated haemoglobin and fasting IGF-I were performed on all the subjects. Hourly overnight glucose measurements were also obtained on the ten subjects who had two overnight growth hormone studies. RESULTS For the whole diabetic growth, OH area under curve (AUC) was maximal in late puberty (pubertal stage 4), and was paralleled by maximal GH peak amplitude. No relationship between GH-AUC and metabolic control was demonstrated. No difference in OH parameters was demonstrated between the male and female subgroups. The relationship between growth hormone secretory parameters and stature was not significant. However, GH-AUC was significantly correlated with growth velocity in the males but not the females. CONCLUSIONS The pattern of GH secretion in adolescents with diabetes parallels that seen in normal adolescents during puberty, with increases In GH concentraflon Correspondence and current address: Dr J. A. Batch, Department of Paediatrics, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 2QQ.
associated with increased OH pulse amplitude. The degree of metabolic control had no effect on this pattern and there was no relationship between GH secretory parameters and stature.
Abnormal growth hormone secretion in diabetes has been recognized in both children and adults for many years (Hansen, 1970; Vigneri et al., 1976; Hayford et al., 1980; Press et al., 1984). Growth hormone (GH) has been demonstrated to be elevated in subjects with newly diagnosed diabetes (Johansen & Hansen, 1971), in those with poor metabolic control (Molnar et al., 1972) and those with stable improved control (Hermansen et al., 1987). Exogenously administered G H has also been shown to worsen diabetic control (Press et al., 1984). The potential role of growth hormone in the pathogenesis of complications of diabetes is being increasingly recognized (Ingerslev et af., 1989; Moller et al., 1989; Christiansen et al., 1982). Until recently, studies of growth hormone secretion in adolescents with diabetes have measured integrated concentrations of growth hormone secretion (Hayford et al., 1980),or have used two-hour sampling (Homer et al., 1981). The recent development of validated methods of pulsatile hormone analysis such as the Pulsar program (Merriam & Watcher, 1982) has enabled more precise evaluation of the secretory profile of growth hormone in diabetes and other conditions. The aim of this study was to characterize changes in growth hormone secretory profiles in a large group of adolescents with insulindependent diabetes mellitus (IDDM) during puberty, and to investigate the relationship between growth hormone secretion, metabolic control and stature in this group. Subjects, protocol and methods Subjects
Fifty-two diabetic adolescents (28 males and 24 females) aged between 12 and 17 years were enrolled in the study. The puberty stage distribution of the diabetic adolescents is outlined in Table 1. All subjects had been diabetic for more than 12 months, were free from other illnesses and were taking insulin as their only medication. Each subject and his/ her parents gave informed consent to the study, which had been approved by the Ethics Committee of the Royal Children’s Hospital Research Foundation, Melbourne, Australia. 411
412
J. A. Batch & G. A. Werther
Table 1
Clinical Endocrinology (1992) 36
Puberty stage and sex of the diabetic adolescents
Puberty stage
Males
Methods
Females
3 2 11
9 3
Protocol
All subjectswere admitted overnight for assessment. Clinical assessment consisted of physical examination, auxology and assessment of pubertal status. Pubertal status was assessed by the method of Tanner (1962), and for purposes of analysis genital staging was used in boys and breast stage in girls. To compare growth parameters at different ages and pubertal stages, height for chronological and skeletal age was expressed as a standard deviation score (SDS) based on population data (Tanner et al., 1966). Height velocity was determined for each subject from 3-monthly clinic visit measurements made over the preceding 12 months. Following this clinical assessment, each of the 52 subjects had an overnight growth hormone profile performed from 2000 until 0800 h with growth hormone samples obtained every 20 minutes. Ten of the subjects with poor diabetic control had a second overnight growth hormone profile performed after 3 months. Samples for the overnight profiles were obtained using a combination of the KowarskiCormed thromboresistant blood withdrawal needle with 3-foot tubing set and the Cormed ML-6-5 withdrawal pump (Kowarski et al., 1971). All subjects had fasting 0800 insulinlike growth factor-I (IGF-I) levels performed. Metabolic control was assessed in three ways. In all 52 subjects, mean 12-month glycosylated haemoglobin (HbAlC) in the 12 months preceding assessment and current HbAlC at the time of study were obtained. Mean 12-month HbAlC was obtained from the four HbAl C assessments performed at 3-monthly clinic visits in the preceding year. In the ten poorly controlled subjects who had two overnight growth hormone studies, hourly measures of glucose were obtained as a measure of ambient metabolic control. On the basis of mean 12-month metabolic control, the 52 subjects were further divided into two subgroups. Group A (33 subjects) consisted of subjects with poorly controlled diabetes defined as mean 12-month HbAlC> 8.5%; Group B (19 subjects) consisted of subjects with well controlled diabetes defined as mean 12month HbAlC < 8.5%. Non-diabetic populations have HbAlC ~ 6 . 2 % .
Growth hormone was assayed using the Pharmacia human growth hormone RIA 11 Kit (Pharmacia Diagnostics, Fairfield, NJ). All samples from an overnight profile were analysed in the same assay run. The intra-assay coefficients of variation at GH concentrations of 30 and 1.2 mU/1 were 6 and 16% respectively, and the interassay coefficients of variation at 30 and 7*2mU/lwere 7-5 and 10.1% respectively. HbAlC was measured using the Bio-rad Haemoglobin AIC Mini Column Test (Hercules, CA) (Jaynes et al., 1985). Whole blood glucose was measured using a Yellow Springs Analyzer Model 23AM-230087026 S.N. 2023 (Yellow Springs, Ohio). IGF-I was measured as previously described (Baxter et al., 1982). Bone age was determined by the same radiologist using the method of Greulich and Pyle (1959). Data analysis
The pattern of pulses in the 12-hour profiles was analysed using a modification (Thomson et al., 1990) of the Pulsar program (Merriam & Wachter, 1982). The program identifies secretory peaks by height and duration from a smoothed baseline, using the assay SD as a scale factor. The program calculations were performed on a desk-top computer (EPSOM PC-AX2) using the modified Pulsar program translated into Fortran. The assay coefficients were calculated from the duplicates of the 12-hour growth hormone curve. The cut-off parameters G1-5 of the Pulsar program were set to 3.9,2.4, 1.7, 1.2 and 0.9 times the intra-assay SD as criteria for accepting peaks 1,2,3,4 and 5 points wide, respectively. The smoothing time, a window used to calculate a running mean value, was set to 12 hours. The splitting cutoff parameter was set to 2.7, and the weight assigned to peaks was 0.05. From the Pulsar analysis of the 12-hour growth hormone profile, the following parameters were obtained for each subject and used for analysis: 12-hour growth hormone area under the curve (GH-AUC), mean G H peak amplitude, number of G H pulses per 12 hours and mean G H peak area. Glucose area under the curve (glucose-AUC) was calculated by triangulation. Statistical analysis
Statistical analysis was performed on untransformed data using parametric and non-parametric methods, consisting of the Mann-Whitney Test (difference between groups), the Kruskall-Wallis test (difference across pubertal stages), the non-parametric equivalent of the Newman-Keuls test (multiple comparisons between pubertal groups) (Conover, 1980) and Pearson correlations. These calculations were performed
Growth hormone in adolescents with diabetes
Clinical Endocrinology (1992) 36
"r
.
0 .
04-8
Table 2 Median and confidence intervals for growth hormone
parameters for the whole diabetic group, according to pubertal stage
+
. 0
GH-AUC (mU/1/12 h) Pubertal stage 1
2 3 U
0
II
413
4
I
I
I
1
I
1
1
12
13
14
15
16
17
18
5
GH-peaks (peaksll2 h)
GH-amplitude (mU/I)
n
Median
CI
Median
CI
Median
CI
4 3 20 17 8
2683 5792 4745 6424 2296
25384941 1951-8138 3529-6981 5714-7961 1966-4064
6.0 5.0
5.0-8.0 5.0-5.0
7.0 6.0 6.0
6.0-7.0 5.0-6.0 5.9-8.0
8.5 15.0 12.5 18.0 7.1
6.0-14.0 9.0-24.0 9.2-17'0 14.1-21.0 4.9-15.3
Age [years)
Flg. 1 Twelve-hour growth hormone-area under curve (GHAUC) during puberty in 52 diabetic adolescents. 0, Males;
+,
females. using the Minitab Statistical Package Version 7.0 (Minitab Inc., Pennsylvania, USA). P
Changes in growth hormone concentrations during puberty in adolescents with insulin dependent diabetes Jennifer A. Batch and George A. Werther Department of Endocrinology, Royal Children’s Hospital, Melbourne, Victoria, Australia. (Received 19 August 1991; returned for revision 23 September 1991; finally revised 31 October 1991; accepted 13 November 1991)
Summary OBJECTIVE To document the changes in pulsatile growth
hormone secretion in diabetic adolescents during puberty, and to investigate their relationship to both metabolic control and stature. DESIGN Auxological parameters, overnight growth hormone secretion, fasting IGF-I, hourly glucose and metabolic control were assessed In a group of adolescentswith diabetes. PATIENTS Fifty-two diabetic adolescents (28 males and 24 females) at different pubertal stages and with varying degrees of metabolic control were studied. Ten of those with poor diabetic control were studied on two occasions. MEASUREMENTS Height and weight measurements, pubertal staging, growth velocity data and bone age estimation were obtained on all the patients. Overnight growth hormone profiles (Pulsar program analysis), glycosylated haemoglobin and fasting IGF-I were performed on all the subjects. Hourly overnight glucose measurements were also obtained on the ten subjects who had two overnight growth hormone studies. RESULTS For the whole diabetic growth, OH area under curve (AUC) was maximal in late puberty (pubertal stage 4), and was paralleled by maximal GH peak amplitude. No relationship between GH-AUC and metabolic control was demonstrated. No difference in OH parameters was demonstrated between the male and female subgroups. The relationship between growth hormone secretory parameters and stature was not significant. However, GH-AUC was significantly correlated with growth velocity in the males but not the females. CONCLUSIONS The pattern of GH secretion in adolescents with diabetes parallels that seen in normal adolescents during puberty, with increases In GH concentraflon Correspondence and current address: Dr J. A. Batch, Department of Paediatrics, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 2QQ.
associated with increased OH pulse amplitude. The degree of metabolic control had no effect on this pattern and there was no relationship between GH secretory parameters and stature.
Abnormal growth hormone secretion in diabetes has been recognized in both children and adults for many years (Hansen, 1970; Vigneri et al., 1976; Hayford et al., 1980; Press et al., 1984). Growth hormone (GH) has been demonstrated to be elevated in subjects with newly diagnosed diabetes (Johansen & Hansen, 1971), in those with poor metabolic control (Molnar et al., 1972) and those with stable improved control (Hermansen et al., 1987). Exogenously administered G H has also been shown to worsen diabetic control (Press et al., 1984). The potential role of growth hormone in the pathogenesis of complications of diabetes is being increasingly recognized (Ingerslev et af., 1989; Moller et al., 1989; Christiansen et al., 1982). Until recently, studies of growth hormone secretion in adolescents with diabetes have measured integrated concentrations of growth hormone secretion (Hayford et al., 1980),or have used two-hour sampling (Homer et al., 1981). The recent development of validated methods of pulsatile hormone analysis such as the Pulsar program (Merriam & Watcher, 1982) has enabled more precise evaluation of the secretory profile of growth hormone in diabetes and other conditions. The aim of this study was to characterize changes in growth hormone secretory profiles in a large group of adolescents with insulindependent diabetes mellitus (IDDM) during puberty, and to investigate the relationship between growth hormone secretion, metabolic control and stature in this group. Subjects, protocol and methods Subjects
Fifty-two diabetic adolescents (28 males and 24 females) aged between 12 and 17 years were enrolled in the study. The puberty stage distribution of the diabetic adolescents is outlined in Table 1. All subjects had been diabetic for more than 12 months, were free from other illnesses and were taking insulin as their only medication. Each subject and his/ her parents gave informed consent to the study, which had been approved by the Ethics Committee of the Royal Children’s Hospital Research Foundation, Melbourne, Australia. 411
412
J. A. Batch & G. A. Werther
Table 1
Clinical Endocrinology (1992) 36
Puberty stage and sex of the diabetic adolescents
Puberty stage
Males
Methods
Females
3 2 11
9 3
Protocol
All subjectswere admitted overnight for assessment. Clinical assessment consisted of physical examination, auxology and assessment of pubertal status. Pubertal status was assessed by the method of Tanner (1962), and for purposes of analysis genital staging was used in boys and breast stage in girls. To compare growth parameters at different ages and pubertal stages, height for chronological and skeletal age was expressed as a standard deviation score (SDS) based on population data (Tanner et al., 1966). Height velocity was determined for each subject from 3-monthly clinic visit measurements made over the preceding 12 months. Following this clinical assessment, each of the 52 subjects had an overnight growth hormone profile performed from 2000 until 0800 h with growth hormone samples obtained every 20 minutes. Ten of the subjects with poor diabetic control had a second overnight growth hormone profile performed after 3 months. Samples for the overnight profiles were obtained using a combination of the KowarskiCormed thromboresistant blood withdrawal needle with 3-foot tubing set and the Cormed ML-6-5 withdrawal pump (Kowarski et al., 1971). All subjects had fasting 0800 insulinlike growth factor-I (IGF-I) levels performed. Metabolic control was assessed in three ways. In all 52 subjects, mean 12-month glycosylated haemoglobin (HbAlC) in the 12 months preceding assessment and current HbAlC at the time of study were obtained. Mean 12-month HbAlC was obtained from the four HbAl C assessments performed at 3-monthly clinic visits in the preceding year. In the ten poorly controlled subjects who had two overnight growth hormone studies, hourly measures of glucose were obtained as a measure of ambient metabolic control. On the basis of mean 12-month metabolic control, the 52 subjects were further divided into two subgroups. Group A (33 subjects) consisted of subjects with poorly controlled diabetes defined as mean 12-month HbAlC> 8.5%; Group B (19 subjects) consisted of subjects with well controlled diabetes defined as mean 12month HbAlC < 8.5%. Non-diabetic populations have HbAlC ~ 6 . 2 % .
Growth hormone was assayed using the Pharmacia human growth hormone RIA 11 Kit (Pharmacia Diagnostics, Fairfield, NJ). All samples from an overnight profile were analysed in the same assay run. The intra-assay coefficients of variation at GH concentrations of 30 and 1.2 mU/1 were 6 and 16% respectively, and the interassay coefficients of variation at 30 and 7*2mU/lwere 7-5 and 10.1% respectively. HbAlC was measured using the Bio-rad Haemoglobin AIC Mini Column Test (Hercules, CA) (Jaynes et al., 1985). Whole blood glucose was measured using a Yellow Springs Analyzer Model 23AM-230087026 S.N. 2023 (Yellow Springs, Ohio). IGF-I was measured as previously described (Baxter et al., 1982). Bone age was determined by the same radiologist using the method of Greulich and Pyle (1959). Data analysis
The pattern of pulses in the 12-hour profiles was analysed using a modification (Thomson et al., 1990) of the Pulsar program (Merriam & Wachter, 1982). The program identifies secretory peaks by height and duration from a smoothed baseline, using the assay SD as a scale factor. The program calculations were performed on a desk-top computer (EPSOM PC-AX2) using the modified Pulsar program translated into Fortran. The assay coefficients were calculated from the duplicates of the 12-hour growth hormone curve. The cut-off parameters G1-5 of the Pulsar program were set to 3.9,2.4, 1.7, 1.2 and 0.9 times the intra-assay SD as criteria for accepting peaks 1,2,3,4 and 5 points wide, respectively. The smoothing time, a window used to calculate a running mean value, was set to 12 hours. The splitting cutoff parameter was set to 2.7, and the weight assigned to peaks was 0.05. From the Pulsar analysis of the 12-hour growth hormone profile, the following parameters were obtained for each subject and used for analysis: 12-hour growth hormone area under the curve (GH-AUC), mean G H peak amplitude, number of G H pulses per 12 hours and mean G H peak area. Glucose area under the curve (glucose-AUC) was calculated by triangulation. Statistical analysis
Statistical analysis was performed on untransformed data using parametric and non-parametric methods, consisting of the Mann-Whitney Test (difference between groups), the Kruskall-Wallis test (difference across pubertal stages), the non-parametric equivalent of the Newman-Keuls test (multiple comparisons between pubertal groups) (Conover, 1980) and Pearson correlations. These calculations were performed
Growth hormone in adolescents with diabetes
Clinical Endocrinology (1992) 36
"r
.
0 .
04-8
Table 2 Median and confidence intervals for growth hormone
parameters for the whole diabetic group, according to pubertal stage
+
. 0
GH-AUC (mU/1/12 h) Pubertal stage 1
2 3 U
0
II
413
4
I
I
I
1
I
1
1
12
13
14
15
16
17
18
5
GH-peaks (peaksll2 h)
GH-amplitude (mU/I)
n
Median
CI
Median
CI
Median
CI
4 3 20 17 8
2683 5792 4745 6424 2296
25384941 1951-8138 3529-6981 5714-7961 1966-4064
6.0 5.0
5.0-8.0 5.0-5.0
7.0 6.0 6.0
6.0-7.0 5.0-6.0 5.9-8.0
8.5 15.0 12.5 18.0 7.1
6.0-14.0 9.0-24.0 9.2-17'0 14.1-21.0 4.9-15.3
Age [years)
Flg. 1 Twelve-hour growth hormone-area under curve (GHAUC) during puberty in 52 diabetic adolescents. 0, Males;
+,
females. using the Minitab Statistical Package Version 7.0 (Minitab Inc., Pennsylvania, USA). P
