In the present study, different thresholds for tissue and metabolic responses to GH treatment were found in short children who had varying GH secretion capacities, as well as varying responsiveness to GH.
Cardiac response to GH
Cardiac tissue, estimated by LVDd, was found to be the most GH-sensitive of the variables evaluated (effects seen from a dose of 33 μg/kg/d). This is in line with the findings of Capalbo et al.  who found that LVDd increased during treatment with a dose of 30 μg/kg/d. From the available data, it is difficult to draw conclusions concerning the mechanism responsible for GH effects on heart size. We have, however, previously shown the presence of GH receptors in cardiac tissue in children, indicating that a direct effect of GH on the heart is likely . Even though GH responsiveness was high, there were no adverse effects seen on any measurement of cardiac function or on blood pressure during the study .
GH dose-effect on body composition
It has been shown that GH treatment at a dose of 57 μg/kg/d given to children born small for gestational age (SGA) can lead to an increase in muscle mass, and a concomitant decrease in fat mass . However, data showing a dose-dependency increase in LST are lacking. We found that LST mass increased in a dose-dependent manner for the six GH doses administered. This effect was most marked at GH doses above 33 μg/kg/d, and the ED 50% for gain in LST mass was in the mid-range (47 μg/kg/d). This dose is in the same range as that found to promote longitudinal bone growth (51 μg/kg/d), confirming our previous findings of anabolic GH effects.
As demonstrated previously using principal component analysis , a strong lipolytic effect was found for all GH doses, seen by changes in fat mass and fat mass index, but no GH dose–response effect was seen. A possible explanation is that in the dose range studied the lipolytic effect had already reached its maximum, making lipolytic variables the most sensitive to GH. Early leptin reduction after the start of GH treatment was found previously to be positively correlated with first year growth response in a group of short children treated with 33 μg/kg/day; however, there was a wide range in Δ leptin levels as for growth response; at that time no individual responsiveness was possible to estimate . It is well known that fat mass decreases in GH-deficient adults when they are treated with GH . Further studies are needed to determine the dose response for the lipolytic effects at lower doses in children.
GH effect on alkaline phosphatase
Increased serum bone-specific ALP is known to be a reliable and early sign of increased bone metabolism and correlates to first year growth response in GH-deficient children . In the present study, ALP activity was found to be more responsive than longitudinal bone growth to a given GH dose.
GH dose-effect on insulin and insulin-sensitivity
GH doses above the common dose range of 25–35 μg/kg/d used for GH-deficient children , resulted in greater insulin increases than lower doses, although insulin levels did not exceed the normal range. This confirms previous studies that insulin levels are lower than normal at baseline in GH-deficient individuals . In short children born SGA, no impaired GH dose–response effect on insulin has been reported in a dose range between 33 and 66 μg/kg/d .
GH exerts both insulin-like and insulin-antagonistic effects in vitro. An insulin-like effect has been reported in some in vivo studies , but not in all . In the present study the ED 50% for the insulin enhancement was 48 μg/kg/d, which is very close to the ED 50% of 51 μg/kg/d for height gain found in the present study. Thus, the insulin antagonist effect of GH seems to be equally or more responsive to GH than the effect on IGF-I. An explanation for this may be that we used SDS for IGF-I, but not for fasting insulin, where mU/L was used. Insulin may be required as a growth factor during the catch-up growth phase, so the observation of a dose-dependent increase should be viewed as more than just compensation for GH-induced insulin resistance.
GH dose-effect on IGF
We demonstrated a marked dose–response effect on IGF-ISDS. Thus, as for LST mass, the prediction of growth response is not valid for IGF-ISDS levels. This is in line with findings in groups where individual responsiveness was not addressed, and GH therapy resulted in increasing IGF-ISDS in a dose-dependent manner in prepubertal children, with more dramatic changes being observed at higher doses (50 and 100 μg/kg/day vs 25 μg/kg/d) , and 100 μg/kg/d compared to 43 μg/kg/d in pubertal GH-deficient patients .
In the present study, the ED 50% for IGF-ISDS was 57 μg/kg/d, which was the highest dose observed for any metabolic variable. Therefore, liver response to IGF-I secretion was found to have a higher threshold than both height gain and muscle growth in the present study.
GH dose-effect on catch-up growth
In order to be able to compare metabolism and growth, height gain was studied in the present study and not the height target of the study, i.e. the diff MPHSDS. The growth effect associated with individualized GH doses was quantified and presented in Figure 1. Thus, a GH dose of 51 (47, 56) μg/kg/d was necessary in order to achieve half of the height gain. In previously performed studies in children with ISS, GH had a dose-related effect on longitudinal bone growth in the dose ranges of 33–67 μg/kg/d  and 31–47 μg/kg/d . In GH-deficient children, an increase in GH dose from 25 to 50 μg/kg/d resulted in a sustained increase in growth velocity, whereas in this study no additional effect was observed with a further increase in dose to 100 μg/kg/d, the highest GH dose in the trial given only to the most non-responsive children . However, in a randomized GH dose study during puberty in GH-deficient children, a dose dependent effect (33 μg/kg/d vs. 67 μg/kg/d) was found . This is in line with prior evidence for a dose-dependent effect of higher doses on adult height in children with ISS . To summarize, the adult height achieved in GH-deficient children treated with GH replacement therapy has been found to be dose-dependent [34, 42], as has adult height in children with ISS [21, 41, 43].
Effects on metabolism
There are, however, only a few studies on the metabolic consequences of GH therapy in children. Ciresi et al. studied metabolic parameters in GH-deficient children, but the question of dose-dependency was not investigated. Mauras et al.  compared IGF-I levels in two different GH dose groups in adolescents with GHD. Cohen et al.  analyzed the response of IGF-I, IGFBP-3, fasting glucose, fasting insulin, HbA1c and height to three different GH doses (25, 50, and 100 μg/kg/day) in prepubertal GH-deficient children. However, the question of whether there were different metabolic thresholds was neither addressed nor individual responsiveness.
In a recent report on our study group, we found dose-dependent effects on height gain, body composition and metabolism . There are no comparable prospective randomized studies. Nevertheless, our data can be compared with reported qualitative effects in response to different GH doses, which suggest that dose–response effects exist [21, 39].
Our study included children classified as both GHD and ISS as our study also includes aspects on tissue responsiveness, ranging from high to low within both diagnostic groups. Without a wide range, it would not have been possible to perform our study on the hypothesis of varying thresholds for different tissues and metabolic markers [19, 20].
A fact that influences the interpretation of the results is, that the individual GH dose in the trial was selected based on GH responsiveness according to estimated/expected growth response and adapted so that the child would reach MPHSDS within 2 years, although limited by the set maximal GH dose of 100 μg/kg/d. None of the metabolic effects presented here was found to be related to the dose selection procedure in the trial.
The strengths of the present study are that we were able to compare many different effects within the same individual, as well as assessing inter-individual variations. When monitoring GH treatment in the clinical setting, it is essential to know which processes will be affected and which marker will be the first to react to treatment based on responsiveness. The aim with estimating individual responsiveness is to set a target for treatment effect – it may be growth response for which we now have prediction models. In the future, prediction models may also be constructed for the metabolic markers studied in this paper.