Main

Achondroplasia is the most common, genetic form of disproportionate short stature and results from an arginine substitution for a glycine at amino acid 380 in the fibroblast growth factor receptor 3(1). Despite the abnormal skeletal growth in achondroplasia(2), GH levels are presumed to be normal because subjects with achondroplasia have normal GH responses to stimulatory tests during wakefulness(3). However, GH profiles have not been studied in this disorder, even though some GH secretion abnormalities may be seen only during overnight studies and despite normal daytime stimulatory tests(4, 5).

Previous reports of overnight GH secretion in achondroplasia have been limited(3) with a single case report demonstrating recovery of low GH levels after treatment of OSA(6). OSA and obesity are known to influence GH secretion(7, 8), and these disorders are common in subjects with achondroplasia(9, 10). We undertook the current study to assess overnight GH secretion in achondroplasia. The effects of age and sleep state, obesity, and OSA were analyzed to determine their influence on GH secretion in achondroplasia.

METHODS

Subjects were selected on the basis of an x-ray-confirmed diagnosis of achondroplasia and were enrolled at the time of their attendance at the skeletal dysplasia clinic of the Children's Hospital. Five adult subjects were contacted through the Little People's Association. Informed consent was obtained either from the subject themselves, or their parents, for sleep studies and concurrent collection of venous blood samples for the purpose of GH analysis. The study was approved by the Ethics Committees of the Children's Hospital Camperdown and the Royal Prince Alfred Hospital Camperdown.

Sleep studies. Full-night polysomnographic sleep studies were performed throughout the usual sleep period, with continuous monitoring of EEG, electro-oculogram, and submental electromyogram). Respiratory monitoring incorporated nasal airflow by thermistor or pressure transducer (Validyne DP-45, Validyne Corp., Northridge CA), thoracic and abdominal respiratory movement by inductance plethysmograph (Respitrace), surface electromyogram of the diaphragm, transcutaneous Sao2) by pulse oximeter (Ohmeda Biox 3700e) and carbon dioxide (Tcpco2) (Radiometer, Cophenhagen).

Sleep stages (wakefulness, sleep stages one to four, and REM sleep) were scored using Rechtschaffen and Kales(11) criteria with an epoch length of 1 min. Respiratory events were scored according to the standard methods of our unit(9). Briefly, apneic events were scored when there was no oral or nasal airflow. Hypopneas were scored when there was a greater than 50% reduction in the amplitude of the respiratory signals (airflow and/or respitrace sum). Events were considered significant if two or more successive respiratory cycles were perturbed in association with oxygen desaturation of more than 3%, and/or terminated by an arousal. Using this definition, apnea duration changes with age, and reaches 10 s duration when the respiratory rate is 12 breaths min-1,i.e. the normal adult respiratory rate and standard criteria for scoring adult OSA(13). Partial airway obstruction was treated if it was associated with hypoventilation and CO2 retention of 10 mm Hg or more. Carbon dioxide retention was confirmed with arterial blood gases, where possible. Values Sao2 and CO2 were recorded continuously, and maximum and minimum values were noted.

Sleep states were subdivided into three categories for correlation with GH data; light sleep (stages one and two = SI-II), SWS (stages three and four = SWS), and REM. The severity of OSA was correlated according to respiratory disturbance indices, i.e. number of apneas per h. The severity of OSA was also classified according to the index of obstructive respiratory events (ORDI = mixed and obstructive, apnea, and hypopnea combined). Categories: 1 = ORDI < 5 events h-1 (normal, no treatment); 2 = ORDI 5-10 h-1 (mild); 3 = ORDI 10-20 h-1 (moderate); 4 = ORDI > 20 h-1 (severe OSA, treatment definitely indicated.

Growth hormone sampling and assays. A Teflon (Intracath) needle was inserted into a peripheral vein. Blood collection commenced at 2000 h, and 2-mL samples were collected at 20-min intervals for the following 12-h period. Samples were placed in heparinized collection tubes and centrifuged within 1 h. Plasma IGF-I was measured on the 0800 h sample of the morning after the overnight collection. Plasma samples were frozen at -80°C until the time of analysis.

All of the growth hormone assays were performed within the endocrine laboratory of the Childrens Hospital by the same investigator (M.J.). Plasma GH was determined by time-resolved immunofluorometric assay (Kabi-Pharmacia, DELFIA). All samples from one subject were assayed together and in duplicate, and the mean of the two results was used in subsequent analyses. Where there was significant discrepancy between the duplicated results, the assay was repeated. The distribution of measurement error was determined by concurrent measurement of known hormone concentration, including blank samples. The assay profile was measured, and a distribution of measurement error was produced. The sensitivity of the assay as determined by 2 SD from zero (n = 16) was 0.018 mIU/L. The intraassay coefficient of variation was less than 6%. Plasma IGF-I was determined by a commercial kit (Bioclone Australia) with acid ethanol extraction before RIA.

Deconvolution. Deconvolution analysis(14) was used to estimate the growth hormone secretion with the two following assumptions: 1) constant GH half lives throughout the study period for each subject and 2) a constant rate of GH secretion between assay points. Deconvolution analysis was performed based on the formula where Ca andCb are two consecutively measured plasma GH concentrations,ta and tb are the respective times of the measurements, S is the input secretion during the episode, andE(t) is the GH elimination function.

To estimate the elimination of GH a two-component model with two half-lives of elimination (HL1 and HL2), and fractional distributions(F) and (1 - F), respectively, was used. Equation For estimation of two half-lives and fractional distributions of excretion, a single secretion peak with a clear exponential decay was identified in each subject. Four data points in the late part of this peak were used to determine the constants of the excretion formula with computer-based curve fitting. If a Gaussian form of secretion distribution is assumed(15), the secretion between these data points is negligible.

Overnight profiles were generated from the time of sleep onset to allow analyses of the group as a whole and comparison between subjects on different study nights. For comparison of GH secretion rates in different sleep states, the GH profile (after deconvolution) was matched with sleep state throughout the period of sleep.

Data analysis. A time-matched profile of GH raw data and sleep state was produced for each study. A continuous profile of GH secretion was calculated by deconvolution analysis of the raw GH data. Sleep state data from the time-matched analysis was superimposed on the (continuous) GH secretion curve (after deconvolution) to determine GH secretion for the different sleep states. See Figure 1.

Figure 1
figure 1

GH secretion by sleep stage for an individual.(A) Measured GH levels (•, scale on the left y axis) and calculated GH secretion rate (-), scale on right y axis) plotted against time from study onset. (B) Sleep hyponogram on same time axis as in A, indicating the sleep states that correlate with the GH data. Sleep stages are: wake, wakefulness; I and II, stages one and two(SI-II); III and IV, stages three and four (SWS); and REM.

Full size image

Analysis of GH secretion characteristics and correlations with sleep state, age, body weight (percent of ideal), and respiratory disturbance indices were made. Comparative analyses were performed between those studies with better RDI scores and those with worse OSA, for all five subjects who underwent repeat studies after treatment for OSA. In one child the study sequence was reversed because her OSA worsened between the initial and follow-up studies(due to regrowth of adenoidal tissue).

Statistical analyses. Results are presented as mean ± SEM (range) unless otherwise stated. Paired t tests were used to compare the differences in GH secretion in subjects studied twice. Linear correlations were calculated by using the BMDP 1R program(16). Comparative analyses of respiratory variables were performed using log-transformed data and SEM values are not provided. Factors affecting GH or IGF-I secretion were correlated using multiple linear regression. Statistical significance was assumed at p values less than 0.05.

RESULTS

Subjects. Twenty-three subjects with achondroplasia completed overnight sleep studies, but data are reported for only the 19 subjects with successful collection of GH samples, including 10 female and 9 male subjects. Mean age was 11.3 ± 2.3 y (1.8-30.9, median 6.7), including 13 children(<10 y), two adolescents (10-20 y), and four adults (>20 y). Percentage of ideal body weight, calculated from the 3rd percentile for age (on normal stature growth charts), was 111.5 ± 5% (80-169)(10); 5 of the 19 (26%) subjects were >120% of ideal body weight, and 9 (47%) were >110% ideal weight.

Five subjects with OSA had repeat overnight studies. This group included four male and one female subject, aged 8.1 ± 3.4 y (2.3-21.3, median 5.9) at the time of their first study. The second study was 6.4 mo from the first (5.1-7.9).

Sleep studies. Sleep-study duration was 9.2 ± 0.1 h(8.4-10.2), and sleep time was 6.8 ± 0.3 h (4.4-9.5)(Table 1). The degree of obesity (percentage of ideal body weight) correlated with the frequency of obstructive events in the baseline study group (r2 = 0.40 p < 0.005). The severity of OSA did not correlate with the degree of sleep fragmentation, whether this was measured as an arousal index or the number of sleep state transitions.

Table 1 Mean (SEM) for age, growth parameters, GH secretion, and sleep study results
Full size table

OSA was more severe in those treated for OSA than in the baseline group, whether measured by total number of respiratory events, obstructive events alone, or by clinical categories. For the OSA group, total RDI was 37.0± 5.1 versus baseline study group at 15.9 ± 3.5 h-1. Obstructive events totaled 29.2 ± 7.7 h-1 (median 31.1) in the OSA group versus 11.0 ± 1.7 h-1 (median 4.4) in the baseline studies. These respiratory event scores corresponded to clinical categories of 3.8 in the OSA group versus 2.4 for baseline studies (p < 0.01).

Growth hormone levels and secretion rates. The estimated half-lives and their fractional distributions are presented for each of the 19 subjects in Table 2. The total amount of GH secreted during the overnight study decreased with increasing age (r2= 0.22, p = 0.04). Figure 2A. The decrease in GH secretion rate, with age, was not significant in this group, although mean GH secretion rates correlated with total GH secreted (r2 = 0.93 p < 0.001). The secretion rates in SWS and REM were not significantly different from each other, but were greater during both of these sleep stages compared with SI-II sleep (Table 1). A clear peak of GH secretion occurred in SWS, during the first 2 h after sleep onset(Fig. 2B).

Table 2 Parameters of growth hormone secretion for each subject
Full size table
Figure 2
figure 2

Patterns of GH secretion in 19 subjects with achondroplasia. (A) Total overnight GH decreases with increasing age(years). r = 0.47, p < 0.05. (B) Mean GH secretion according to sleep state, for 19 subjects with achondroplasia. Values are plotted according to the time from sleep onset, for three sleep states: (▿- - -) GH secreted during SWS (stage III-IV sleep), demonstrating a clear peak of GH secretion during the first 2 h after sleep onset, (•- - -) GH secreted during light sleep (SI-II) sleep, and(○-) GH secreted during REM sleep.

Full size image

During the baseline studies, none of the indices of OSA severity(respiratory disturbance indices, minimum Sao2, maximum Tcpco2, or number of sleep state transitions) correlated with indices of GH secretion(total GH, GH secretion rates, or GH half-lives). IGF-I levels were generally within the normal range (Fig. 3), and correlated independently with age, percentage ideal body weight, and GH secretion rate(r2 = 0.68, p < 0.001).

Figure 3
figure 3

IGF-I values. Parameters for normal range are marked(perpendicular lines). Separate graphs are shown for male and female subjects.

Full size image

Subjects with OSA. Five subjects were treated for OSA (RDI 37.0± 5.1 events h-1); three underwent adenotonsillectomy and two were treated with nasal continuous positive airway pressure. Distribution and progression of sleep stages did not change with improved OSA (RDI 12.5± 4.9 events h-1), and there was no change in the number of arousals. There was no change in total overnight GH secretion(Fig. 4A), but secretion of GH during the first 2 h of SWS was lower during the night with OSA (0.37 ± 0.09 mIU/L/min) than the study night after treatment (1.06 ± 0.43 mIU/L/min, p< 0.05) (Fig. 4B). During the first 2 h of sleep, the mean number of sleep state transitions showed no change from 12.3 ± 2.7 to 7.8 ± 1.3 (NS). Levels of IGF-I showed no change after treatment of OSA (Table 3).

Figure 4
figure 4

(A) Total, overnight growth hormone secretion in five subjects treated for OSA. Values are shown for each subject during studies with OSA (▪) and after treatment (□). The subject with the highest baseline secretion had a decrease in GH secretion after treatment.(B) GH secretion rates during SWS, before and after OSA treatment. A clear peak of GH secretion can be seen during the first 1.5 h of sleep after treatment of OSA (open symbols), compared with the study night with OSA(filled symbols). Values are shown according to the time after sleep onset. The number of subjects represented by each symbol is illustrated by the different symbol shapes. ○, 1 subject; □, 2 subjects; ▵, 3 subjects; ⋄, 4 subjects; or ○, 5 subjects

Full size image
Table 3 Mean (SEM) before and after treatment of OSA, for five subjects with achondroplasia
Full size table

DISCUSSION

This study documents a profile of overnight GH secretion in 19 subjects with achondroplasia. An analysis of factors that influence GH measurements is included; age, sleep state, obesity, and the severity of OSA. Deconvolution analysis was used because, at this time, it provides the most accurate description of GH secretory profiles(17). In five subjects with OSA, we documented recovery of GH secretion after treatment; a GH secretion peak during the first period of SWS recovered to the normal pattern, as seen in the baseline group study.

The overnight GH secretion profiles, shown in this study, match previous descriptions in populations of normal stature. Total GH levels decreased with age(18, 19), and the highest GH secretion rates occurred during SWS and REM sleep. Maximal GH secretion occurred during the first 2 h of SWS in our study group(19, 20). We were not able to show any influence of puberty(21, 22) as only two teenagers were included in the study group.

In this study group, we were unable to show any effect of obesity (measured as percentage of ideal body weight) on GH secretion(8, 23). The severity of OSA increased with the degree of obesity (percent of ideal body weight), but neither parameter correlated with the measurements of GH secretion (secretion rate, half-life, or total GH secreted). There are no “normal weight” charts available for achondroplasia, but obesity is prevalent in this disorder(10, 24). It is possible that the difficulty in determining body weight accurately in this group has prevented us from showing the influence of obesity on GH secretion in our study group.

After treatment of OSA, we documented recovery of the early peak of GH secretion in SWS, even though total GH levels were not demonstrably affected. No previous studies have shown reduced GH secretion with OSA, in the presence of preserved sleep structure. OSA results in repetitive, sleep-associated, upper-airway obstruction and marked sleep disruption(13). Low GH levels have been associated with OSA(25, 26) and improved after treatment in both adult male subjects(7, 27) and, anecdotally, in a child with achondroplasia(6). Indices of OSA did not correlate with GH secretion in our baseline group measurements. Growth velocity was not evaluated, and the effects of these limited GH abnormalities on growth or height could not be determined in the current study. We note that linear growth has been seen in response to high doses of exogenous GH in achondroplasia(28), but the responses have been variable(29).

The levels of IGF-I presented here were generally normal(3, 30), and IGF-I levels also correlated with the GH secretion rate for our baseline study group; IGF-I is the messenger product of GH responsible for many of its growth-promoting effects(31). However, there was no correlation between IGF-I levels and markers of OSA severity(32), and IGF-I levels did not change after treatment of OSA. The clinical usefulness of IGF-I levels as a screen for GH deficiency has been limited by the overlap which exists between IGF-I levels in subjects with and without adequate GH secretion(32). It is possible that the changes which we observed in GH secretion may have effects which are independent of IGF-I(33), and a positive response to GH therapy may be seen despite normal IGF-I levels(34).

The secretion rates for GH that we have demonstrated are consistent with previous reports(22). However, the half-lives for GH are longer than those reported for normal stature populations(8, 22, 35). It is possible that GH half-lives are longer in achondroplasts, but potential sources of error exist in our study, such as the variation in the age of our subjects, the use of the last four points of a single GH peak to determine the GH half-lives for each subject, and the variability between sampling times(35). Clear GH peaks were present in each subject, and these would provide accurate half-life estimates if that period at the end of the peak was not associated with ongoing GH secretion(14). Positive evidence for a Gaussian wave form of secretion has been obtained by direct measurements of pituitary effluent blood in sheep(15), and measured profiles of blood levels conform to computer-simulated models of “expected” serum levels using this wave form of secretion(14). In addition, the ideal rate for collecting serum samples for GH analysis is unknown. Factors that dictated the sampling frequency in this study included the amount of sleep disturbance caused by the sampling and the relative blood volumes to be collected from our subjects who were otherwise normal infants and children. Sample collections at 20-min intervals disrupted the sleep of our subjects, reducing the mean sleep efficiency to 75%, compared with an overall mean of 95% for children studied in our unit.

We have demonstrated effects of age, sleep state, and OSA on the GH secretion in achondroplasia. The patterns of GH secretion were largely normal in our baseline studies. The abnormalities demonstrated in association with OSA are consistent with previous studies in normal stature populations, but it is unknown, at this time, whether the level of GH abnormalities we have demonstrated would result in any growth failure. We suggest that the high prevalence of obesity and of OSA in achondroplasia would predispose to abnormalities of GH secretion, but further studies of the normal GH secretion profile in achondroplasia are required.