Main

Seasonal variation in height velocity has been described in both normal children and in children on GH treatment(1). In healthy children the peak height velocity occurred during the period of longest day length in both the northern and southern hemispheres(2, 3). Year-to-year changes in the pattern of seasonal height velocity have also been shown to correspond to differences in hours of sunlight(3). The mechanism behind this variation is unknown, but the fact that GH-deficient children on GH therapy show seasonal variation in growth implies a peripheral regulatory mechanism(1), with variation in GH sensitivity in the target tissue. A possible indirect measure of GH sensitivity is the concentration of GHBP in the blood. GHBP corresponds to the extracellular domain of the GH receptor(4–6). Its regulation may therefore be similar to that of the GH receptor(7). Sensitivity to GH also involves post-GH receptor effects. Both IGF-I and its major binding protein, IGFBP-3, are regulated by GH. Although sensitive to changes in nutritional status, IGF-I and IGFBP-3 have been shown to be useful for the evaluation of GH status(8).

In this study, we have measured GHBP concentrations repeatedly to investigate intraindividual variation and to study whether any such variation is related to changes in IGF-I, IGFBP-3, or urinary excretion of GH. The relationship between changes in GHBP concentrations, short-term height velocity, and changes in body composition was also studied.

METHODS

Subjects. Twenty-nine healthy and well nourished children (13 girls and 16 boys) were investigated. At the start of the study their mean age was 9.3 y; auxologic and biochemical data for these children are given in Table 1. Height, weight, and weight-for-height were expressed in SD scores compared with Swedish reference values(9, 10). All children were prepubertal throughout the study period, as assessed according to Tanner regarding breast and pubic hair development(11) and according to Prader regarding testicular volume as measured using an orchidometer(12).

Table 1 Auxologic and biochemical characteristics of the prepubertal children
Full size table

Study protocol. Height and weight measurements were recorded monthly by the same person and at the same time of day (0800-1000 h). The children collected morning urine samples for 3 d preceding each follow-up visit and recorded the exact urine collection time and the urine volume. At each follow-up visit, blood samples were taken, and the children were interviewed and completed a questionnaire. A physical examination was performed at the start of the study, at 6 mo, and at completion.

The study was approved by the Ethical Committee of the Medical Faculty, University of Göteborg. Informed consent was obtained from each child and his/her parents.

Questionnaire. At each follow-up visit, the examiner questioned the child about recent illnesses, food intake, and physical activity the day before the follow-up visit. A questionnaire was also given to the parents in which they were asked about illnesses, psychosocial events, changes in the child's appetite, and changes in the child's activity during the period from the last follow-up visit. The information from these questionnaires was evaluated using a scoring system, with values between 1 and 5. In a questionnaire based on visual analog scales, the children answered questions about meal size, overall appetite, and appetite for sweets. For example, the question “Do you have a strong desire for sweets” was answered by putting a cross on the 100-mm line between the phrases “not at all” and “very much.”

Biochemical measurements . GHBP. A total of 354 serum samples were analyzed from the 29 children, i.e. 10-14 samples for each child. Blood samples were kept at room temperature and centrifuged within 8 h of withdrawal. After centrifugation the serum was stored at -20°C until assayed. GHBP was measured by a ligand-mediated immunofuctional assay as previously described(13). The detection range in the ligandmediated immunofunctional assay was 15.6-1000 pmol/L, and all samples from each child were measured in the same assay. The intraassay C.V. was 7.3%, and the interassay C.V. was 11.3%.

Urinary GH. Urinary GH was measured using an ELISA(NordiTest™; Novo Nordisk AS, Gentofte, Denmark). The urine samples were stored in the dark at 4°C and analyzed within 3 wk, as previously described(14). Repeated analysis showed that samples were unaffected by the storage. The intraassay CVs were 7.9, 4.1, and 3.8% and the interassay C.V.s were 12.4, 9.5, and 9.9% at GH concentrations of 3, 9, and 26 ng/L, respectively. The urinary excretion of GH was evaluated using an SD score reference that eliminates the influence of urinary volume on urinary GH measurements(15).

IGF-I. IGF-I was measured by an IGFBP-blocked RIA without extraction and in the presence of an approximately 250-fold excess of IGF-II(Mediagnost GmbH, Tübingen, Germany)(16). All samples from each individual child were analyzed in the same assay. The intraassay C.V. at 219 μg/L was 4.4%.

IGFBP-3. IGFBP-3 was measured by RIA (Mediagnost GmbH)(16). All samples from one child were analyzed in the same assay. The intraassay C.V. at 2927 μg/L was 5.6%.

Statistical analysis. Values are given as means (SD), if not stated otherwise. For correlation analysis a Spearman correlation coefficient(rs) was used.

In each child, monthly GHBP concentrations were normalized by expressing the concentration as a ratio to the mean GHBP concentration for the child. ANOVA together with Tukey's test was used to analyze the normalized GHBP values over months.

In the interindividual statistical analyses, the mean value for each child over the year was used. Individual curve fitting for GHBP concentrations was performed using the least square method for the cosine function. To study correlations between changes in GHBP and changes in other variables during the same period, the maximum difference for each child between two samples of GHBP taken 1 mo apart, regardless of sign, were selected. The GHBP changes were then correlated to changes in other variables during the same period. Sign test was used to evaluate the distribution of the maximum point of the cosine function over the year. The individual variation was estimated by the intraindividual SD and intraindividual C.V. using all measurements for each child.

RESULTS

There was marked variation in GHBP levels between individuals and between monthly samples in the same child (Fig. 1).

Figure 1
figure 1

GHBP concentrations in samples taken monthly over 1 y from 29 prepubertal children. GHBP concentrations from the same child are connected with a line. To illustrate individual changes, the longitudinal measurements are highlighted in three children.

Full size image

Characterization of individual mean level of GHBP. The mean serum GHBP level for the 29 children (calculated as the mean of the mean GHBP levels during the year for each child) was 320 ± 123.5 pmol/L (SD). The mean intraindividual variation around the individual child's mean GHBP concentration was 60.5 pmol/L. There was no significant gender difference in mean serum GHBP levels (girls, 361 ± 133.2 pmol/L; boys, 286 ± 107.8 pmol/L). The mean serum GHBP level during the year in the 29 children correlated significantly with height velocity (rs = 0.37,p < 0.05, Fig. 2A). This relationship was not evident for height or height SD score (Fig. 2B,Table 2). All weight parameters, including the children's weight gain(rs = 0.41, p < 0.05) and BMI(rs = 0.62, p < 0.0005, Fig. 2C) were positively correlated with mean GHBP levels. Correlations were also found between the mean GHBP level and IGF-I (rs = 0.42, 0 < 0.05, Fig. 2D) and IGFBP-3 (rs=0.60, p < 0.001, Fig. 2E), whereas the urinary GH SD score was negatively correlated with the mean GHBP(rs = -0.50, p < 0.01, Fig. 2F). In a multiple linear regression model, the child's weight or weight changes did not significantly contribute to the correlation between the mean concentrations of IGFBP-3 and GHBP.

Figure 2
figure 2

Correlations between the GHBP levels and (A) height velocity (spearman correlation coefficient (rs) = 0.37, p < 0.05), (B) height SD score(rs = 0.28, NS), (C) BMI (rs = 0.62, p < 0.0005), (D) IGF-I (rs = 0.42, p < 0.05), (E) IGFBP-3 (rs = 0.62,p < 0.005), and (F) urinary GH excretion(rs = -0.50, p < 0.01) in 29 prepubertal children. For GHBP, IGF-I, IGFBP-3, and urinary GH the mean level during the year was used.

Full size image

The magnitude and pattern of variation in GHBP. In each child, the GHBP concentration varied during the year, with a mean intraindividual C.V. of 18.0% (range, 6.7-33.0%). Normalized GHBP concentrations varied significantly during this period (ANOVA: 11 d.f.; p < 0.0001, Fig. 3), with the greatest difference between January and August of 22%. The changes seen in Figure 3 resembled a cosine function. To study further the individual pattern of changes in GHBP over the year, each child's monthly GHBP concentrations were fitted to a cosine function. In 16 of the 29 children, a significant fit was found, and all but two of the children with significant curve fitting had their highest GHBP concentration between September and March, i.e. during the winter (Fig. 4,A and B). Compared with the children with significant cosine variation in GHBP, the children without significant cosine fitting had a greater increase in BMI, during the sampling period (0.87 kg/m2 versus 0.25 kg/m2, p < 0.005). The mean concentration of GHBP was not different between the two groups (340 and 295 pmol/L, groups with and without significant cosine fitting, respectively, p = 0.29). The corresponding mean SD of the individual mean GHBP concentrations did not differ between the two groups (57versus 54 pmol/L, p = 0.65, respectively). Further, the children with significant cosine fit of GHBP concentrations did not differ in age, sex, height velocity, or weight changes compared with the children without significant cosine fit.

Figure 3
figure 3

The mean (±SEM) monthly GHBP level in 29 prepubertal children during 1 y. A significant monthly variation (ANOVA) was found (p < 0.0001). Significant decreases compared with December and January in the relative GHBP levels are seen in August and September(**p < 0.01; ***p < 0.095; Tukey's test).

Full size image
Figure 4
figure 4

Longitudinal GHBP measurements in two children. The values were fitted to a cosine function. Out of 29 prepubertal children there was a significant fit (p < 0.05) in 16 children, and 14 of these had their minimum concentration during the summer (A), whereas two had the opposite pattern (B) (p < 0.005, sign test).

Full size image

Short-term individual variation in GHBP. The maximal individual changes in GHBP concentrations between samples taken 1 mo apart correlated with the child's weight change (rs = 0.38, p < 0.05) and changes in IGF-I concentrations (rs = 0.38,p < 0.05) during the same period. Furthermore, maximal individual changes in GHBP concentrations were inversely correlated with changes in illness score (rs = -0.37, p < 0.05) and changes in urinary GH SD score (rs = -0.41, p < 0.05). No correlation was seen between changes in appetite and changes in GHBP concentrations.

DISCUSSION

In this study we demonstrate that, although healthy children have a certain level of GHBP in their serum, there is considerable variation between samples collected at monthly intervals. Furthermore, mean individual GHBP levels throughout the year, estimated from repeated monthly serum samples, were related to the child's height velocity and weight gain and not only to static auxologic characteristics such as weight and BMI. There was also a relationship between GHBP and a child's IGF-I and IGFBP-3 levels as well as between GHBP and the child's excretion of GH in urine.

There was a 22% difference between the highest mean GHBP concentrations for the group found in January, and the lowest mean value found in August. The magnitude of this variation exceeds the diurnal variation in GHBP concentrations(17). The variation in GHBP was largest in those children who had the greatest variation in BMI over the year. Further, the variation had a seasonal pattern, with the highest GHBP concentrations occurring during the winter. In addition, short-term changes in a child's GHBP concentration correlated with the child's weight gain and increases in IGF-I serum concentrations, whereas a lower GHBP concentration was found after illness. This means that single values of GHBP should be interpreted with caution in clinical situations, as the normal range of GHBP concentrations is wide and there is considerable intraindividual variation in GHBP concentrations throughout the year.

As GHBP is closely related to the GH receptor(4), it has been postulated that its regulation is similar to that of the receptor, and that measurements of GHBP may provide an indirect measure of the GH receptor(7). As changes in GHBP concentrations coincided with changes in IGF-I and IGFBP-3 in individual children, the individual variation in GHBP concentrations may reflect changes in GH sensitivity. This has previously been suggested in reports based on seasonal variation in the growth response during GH treatment in children with GH deficiency(1).

In the present study, we monitored GH secretion by measuring urinary GH excretion. To minimize the measurement problems caused by the large day-to-day variability in excreted GH(18), the evaluation of urinary GH was performed using a reference chart that compensates for variations in urine volume(15) and the mean excretion of 3 consecutive nights was measured. A negative correlation was found between GHBP and urinary GH excretion, both for individual levels of GHBP versus urinary GH and for intraindividual changes. This is in agreement with Martha et al.(19) who found an inverse relationship between concentrations of GHBP and spontaneous GH secretion in normally growing boys and young adult men. Another possible explanation for the correlation between GHBP and urinary GH is that serum concentrations of GHBP influences on GH excretion in urine, i.e. high serum concentrations of GHBP may decrease urinary GH excretion.

Body composition has been shown to influence the concentration of GHBP(20). This is supported by our finding that weight and BMI are related to GHBP and that a child's weight changes correlate with changes in GHBP. A questionnaire about appetite did not reveal any dietary changes, and we have no indications of major changes in body composition. Thus, these results are in line with Rasmussen et al.(21) that a short-term hypocaloric diet, without significant weight loss, does not affect GHBP concentrations in obese adults.

Apart from being influenced by nutritional factors, both IGF-I and IGFBP-3 are regulated by GH secretion. We found that both IGF-I and IGFBP-3 were correlated with GHBP levels and that changes in IGF-I and IGFBP-3 were also related to changes in GHBP. Although weight changes influence all these variables, weight changes do not explain these correlations. The most plausible explanation for the parallel changes in GHBP, IGF-I, and IGFBP-3 is a variation in GH sensitivity, because no parallel changes in urinary GH excretion were observed, suggesting that GH secretion did not change.

We conclude that GHBP levels in healthy children are characteristic for the child but vary within individuals over the year, and that this must be considered when interpreting measurements from single samples in clinical situations. Furthermore, the synchronization between these changes in serum concentrations of GHBP, IGF-I, and IGFBP-3, and height velocity, despite an inverse relationship with urinary GH excretion, implies that GHBP is a marker for changes in GH sensitivity.

Table 2 Correlations between the individual mean GHBP levels versus auxologic and biochemical measurements
Full size table