The GH peak response to pharmacological stimulation and the plasma IGF I concentration in young adults with GHD of childhood onset depend on the presence of additional pituitary deficiencies, reflecting a more severe defect of the hypothalamic-pituitary axis. Sex steroids cannot increase the IGF I if the GH secretion is zero.
The key features of this study are: 1) all the patients had GHD due to PSIS of prepubertal onset; 2) each of them was evaluated longitudinally before puberty and as a young adult and the data from these evaluations were compared.
1. Effect of the association with other pituitary deficiencies
The group 1 patients with isolated GHD had significantly greater pituitary height and GH peaks at diagnosis before puberty than did the group 2 patients with multiple deficiencies, despite their similar ages, heights and BMI. This may be due to differences in the way PSIS occurred in the 2 groups. We reported a positive correlation between the GH peak after GHRH and the anterior pituitary height, GH peak after no GHRH stimulation and the spontaneous GH peak in patients with GHD (and PSIS in 22/28) . While the GH peak after GHRH was greater in group 1 than in group 2 in the present study, the difference was not significant, possibly because only three group 1 patients underwent the GHRH test.
This study confirms that the GH peak response to pharmacological stimulation in GHD during childhood and in young adults depends on the presence of additional deficiencies. Maghnie et al  reported that all the 13 patients with PSIS tested as young adults had GH peak responses to arginine, insulin and sequential tests that were below 3 ng/ml, while 4 out of 21 patients of the present study, all with isolated GHD, had a GH peak greater than 3 ng/ml. Their findings and our results may differ because more of our patients had isolated GHD. Hartman et al  found that 41% of their adult patients without pituitary hormone deficiencies other than GH had a GH peak < 2.5 ng/ml, while 67% lacking one other pituitary hormone, 83% of those lacking two pituitary hormones, 96% of those lacking three pituitary hormones, and 99% of those lacking four pituitary hormones had a GH peak < 2.5 ng/ml. Their 3 patients with idiopathic GHD and a GH peak > 2.5 (2.9, 10 and 18) ng/ml despite three deficiencies had GHD of adult onset and the 2 women were given oral estrogen.
2. Effect of sex steroids on GH-IGF I
The group 1 patients with spontaneous puberty showed no increase in the GH peak in response to sex steroid secretion, but the plasma IGF I concentrations (ng/ml) did increase. The variations in the plasma estradiol or testosterone concentrations in the group 2 patients at the last evaluation probably partly reflect differences in the interval between the last administration of ethinyl estradiol or testosterone and the GH evaluation in those with gonadotropin deficiency. Among the three group 2 patients with spontaneous puberty, two had plasma IGF I concentrations similar to those of group 1 patients, while the third (case 12) had a very low plasma IGF I despite a plasma testosterone concentration of 5.6 ng/ml. They differed in that their GH peak was greater than zero in the first two (cases 18 and 19) and undetectable in the third patient (case 12). This suggests that some residual GH secretion is necessary for the sex steroids to increase IGF I. This probably explains the data reported by Aguiar-Oliveira et al , who studied patients with a mutated GHRH receptor that was responsible for a GH peak of 0.01–0.2 ng/ml. They were surprised to find that there was no significant pubertal rise in IGF I, IGF II, IGF binding protein-3, or acid-labile subunit concentrations, as pubertal development was normal, although slightly delayed.
The administration of high doses of sex steroids before the last evaluation to the patients with gonadotropin deficiency, to obtain levels similar to those of group 1, would help to confirm the absence of a direct affect of sex steroids on IGF I. However, one patient from group 1 (case 3) and two from group 2 (cases 10 and 15) with similar plasma testosterone concentrations (spontaneously or after administration) but different GH peaks (3 and 0 ng/ml) had different plasma IGF I concentrations (296 vs 59 and 78 ng/ml). Martinez et al  evaluated the effect of estradiol priming on the GH-IGF axis in 15 patients with GHD and radiological findings after magnetic resonance imaging. They were given a daily dose of 1 or 2 mg micronized estradiol or placebo for 3 days before a sequential arginine-clonidine test. Estradiol did not significantly stimulate GH secretion (3.1 ± 2.4 vs 4.5 ± 2.7 ng/ml). The IGF I concentrations of 14/15 patients on placebo were below normal and estradiol did not change the mean of the group (28 ± 48 vs 25 ± 29 ng/ml).
3. Diagnosis of GHD in adults
The key feature of this study is that all the patients had GHD due to PSIS of prepubertal onset. The majority of the patients with adult-onset GHD had had a hypothalamic-pituitary lesion and had been treated by surgery and/or irradiation. Their results are consistent with previous finding that patients with craniopharyngioma [8, 14] or given low dose cranial irradiation  may have a normal plasma IGF I concentration. This may partly explain why Hoffman et al  found that 70% of the IGF I and 72% of the IGF binding protein-3 concentrations were within the normal range in adults with a GH peak below 5 ng/ml after an insulin test (pituitary adenoma or cranial irradiation) while de Boer et al  found only 4% of the IGF I and 8% of the IGF binding protein-3 concentrations were normal in young adults with idiopathic GHD of childhood onset.
Our results partly explain the difficulty of defining a limiting plasma IGF I concentration for diagnosing GHD in adults. Hartman et al  concluded that patients with an appropriate clinical history and having 3 or 4 additional hormonal deficiencies or a serum IGF I less than 84 ng/ml do not require a GH stimulation test for the diagnosis of adult GHD. We found that IGF I was very low in all patients with gonadotropin deficiency, and in the sole patient with spontaneous puberty but a GH peak of 0 ng/ml.
4. Analysis of the limitations of the study
The number of subjects studied is limited, but there are no reported data on the longitudinal evolution in GH and IGF I in patients with PSIS and only limited data on patients with PSIS in the transition period . These authors used the GHRH plus arginine test. In the present study, the GH peaks for a given patient obtained during childhood and as a young adult were not obtained using similar stimulations, but arginine insulin or glucagon were used as stimulus in the majority of patients. The short time between stopping GH and testing our patients may partly explain the concentrations of IGF I. Thus, Maghnie et al  reported that the IGF I concentrations decreased significantly 6 and 12 months after stopping GH in patients with PSIS. However, the interval was greater than one month in all. Group 2 patients had low plasma IGF I concentrations, whatever the interval. A recent Endocrine Society Clinical Practice Guideline  suggested that the interval between the reevaluation and the discontinuation of GH treatment should not be less than one month, and because of the irreversible nature of GHD in children with PSIS and multiple hormonal deficiencies, a low IGF I measured at least one month after discontinuing treatment is sufficient documentation of persistant GHD without additional provocative testing. Our data confirm this statement.