Insulin-Like Growth Factor Binding Protein-3 Generation: An Index of Growth Hormone Insensitivity

Abstract

GH insensitivity may be an inherited condition or may arise as a consequence of disease or malnutrition. Laron syndrome is the most severe form of GH insensitivity, arising from an absent or defective GH receptor. Less severe forms of GH insensitivity, however, may exist, resulting in short stature but in few other features of Laron syndrome. We have identified a heterogeneous group of children with short stature and either high basal(>10 mU/L) or high peak GH levels (>40 mU/L) on GH provocation testing, to examine biochemical markers of GH sensitivity. These children received 4 d of GH (0.1 U/kg) and the increment in IGF-I, IGF binding protein (BP)-3, and GHBP was determined. Eight GHD children, commencing GH therapy, were recruited as positive controls. The two groups could not be differentiated by age, height SDS (SD score), height velocity SDS, or body mass index. IGF-I and IGFBP-3 generation were correlated in all children (ΔSDS IGF-Iversus ΔSDS IGFBP-3, r = 0.49, p = 0.03). Neither basal GHBP levels or the increment in GHBP were predictive of the IGF-I or IGFBP-3 response to GH. The GHI group had a significantly reduced IGFBP-3 response to stimulation with 4 d of GH (median percent increment in IGFBP-3, 26%, versus 72% in the GHD group, p = 0.03); their IGF-I response to GH was also reduced (median% increment in IGF-I 75%versus 144% in the GH deficient group), but this did not achieve significance, p = 0.06. In all children, the percentage rise orΔSDS in both IGF-I and IGFBP-3 inversely correlated with the GH peak obtained on provocation testing, the latter being the most significant determinant of GH peak. We propose that the “IGF generation test,” in particular IGFBP-3 generation, can be used in the investigation of partial GH insensitivity. Further work, however, is required to establish diagnostic criteria for partial GH insensitivity.

Main

GH insensitivity is characterized by short stature associated with high levels of GH and low levels of IGF-I and is found in a variety of pathologic states. GH receptor deficiency (Laron syndrome) represents the most extreme form of GH insensitivity and results in very severe growth retardation⁽¹⁾. GH insensitivity is also recognized to occur in poorly controlled diabetes^{(2, 3)}, chronic renal failure^{(4, 5)}, chronic cholestatic disorders^{(6, 7)}, hepatic cirrhosis^{(8, 9)} and in hypercatabolic states⁽¹⁰⁾. Laron syndrome itself is a genetically diverse condition⁽¹¹⁾, with variable phenotypic expression⁽¹²⁾. Recognition of this phenotypic diversity has led to speculation that less severe forms of GH insensitivity may exist^{(13, 14)}, producing a correspondingly less severe phenotype. Within the Laron syndrome population, the degree of short stature correlates with both basal IGF-I and IGFBP-3 levels and their increment in response to GH^{(15, 16)}, which indirectly lends support to this hypothesis.

Determination of GH sensitivity has been considered to rest upon IGF-I generation in response to GH; the “IGF generation test.” The standard regime involves four daily injections of GH with blood samples taken before and after treatment to determine the increment in IGF-I. This has been modified in recent years to include estimation of IGFBP-3^{(13, 15, 16)}. The generation test was originally developed as a dynamic test of the GH axis for use in determining GH responsiveness in GH deficiency and in non-GHD children with short stature^(17–22). GHD individuals produce a large increment in IGF-I, typically of the order of 400%^{(7, 17–19)}. In the Laron syndrome, the increment in IGF-I is usually less than 20%⁽¹⁶⁾.

We have therefore investigated a heterogeneous group of short children with high basal or provoked GH levels as a marker of GH insensitivity, compared with eight GHD controls, with an IGF generation test. In this study, we examined the response to exogenous GH of the GH receptor-derived GHBP and the GH-dependent proteins IGF-I and IGFBP-3. We compare the IGF-I and IGFBP-3 responses to GH stimulation with auxologic criteria (height SDS, height velocity SDS, and BMI), basal GHBP, and basal and peak GH levels achieved on a GH provocation test. From this analysis we seek to determine whether the IGF generation test is useful in the determination of partial GH insensitivity, and to define which biochemical marker may best serve as an indicator of the degree of GH insensitivity.

METHODS

Subjects. Eleven children aged 2.2-18.8 y were identified with possible GH insensitivity, on the basis of short stature for target height (-2 SD below mid-parental height), a high basal GH (>10 mU/L) and/or high peak GH (>40 mU/L) on a standard GH provocation test (arginine, n = 6; glucagon, n = 2; sleep, n = 2; clonidine, n = 1)(GHI group). We have previously found no significant difference in median GH peak attained in response to arginine (24.3 mU/L), glucagon (21.8 mU/L), or sleep (19.7 mU/L) in short normal children (our unpublished observations). Eight children with GH deficiency (peak GH response to arginine ≤10 mU/L) embarking on GH therapy were recruited as GH-responsive controls (GHD group).

Auxologic data for all children are presented in Table 1. Body mass indices were within our previously published age- and sex-matched norms⁽²³⁾ for all children except subject 14 with GH deficiency.

Table 1 Auxological parameters for all subjects

Protocol. Each child received recombinant human GH (0.1 U/kg) by daily s.c. injection for 4 d. Fasting blood samples for GH, IGF-I, IGFBP-3, and GHBP were drawn at the beginning of the test and 12 h after the last injection. Ethical approval was granted by the Salford Health Authority Ethics Committee.

Laboratory analyses. GH was measured by immunoradiometric assay(Netria, London, UK). Cellulose-coupled sheep polyclonal anti-GH antibody and¹²⁵ I-labeled mouse monoclonal anti-GH antibody were added sequentially to 50 μL of serum, before centrifugation. The resultant pellet was washed, and radioactivity was counted. GH standards were calibrated against reference standard (International Standard, IS) 80/805 (UK EQAS, Edinburgh, UK). Assay sensitivity was 0.5 mU/L. The interassay C.V. values at 4.9, 11.6, and 45.9 mU/L were 6.7, 7.2, and 7.2%, respectively. IGF-I and IGFBP-3 were measured by RIA (Nichols Institute Diagnostics, Nieuweweg, Wijchen, Holland). IGF-I was extracted from its binding proteins by addition of acid-ethanol solution(12.5%/87.5%, vol/vol). Phosphate buffer was added to neutralize and dilute the serum before addition of rabbit anti-IGF-I antibody and ¹²⁵I-IGF-I followed by overnight incubation. The antigen complexes were then precipitated by addition of goat anti-rabbit antibody with rabbit serum before centrifugation, and the radioactivity was counted. Assay sensitivity was 8.4μg/L. Interassay C.V. values at 60.8 and 184.5 μg/L were 5.2 and 8.4%, respectively. For IGFBP-3 analysis, serum samples were diluted 100-fold with assay buffer before addition of ¹²⁵I-IGFBP-3 and anti-IGFBP-3 antibody. Anti-rabbit precipitant was then added after overnight incubation. Samples were then counted for radioactivity after centrifugation and aspiration of the supernatant. Assay sensitivity was 0.06 mg/L. Intraassay C.V. values at 0.17, 0.56, 1.72, and 3.08 mg/L were 7.3, 8.0, 3.4, and 3.8%, respectively. Age- and sex-matched normal ranges (mean ± 2 SD) for IGF-I and IGFBP-3 were provided by Nichol's Institute. GHBP was measured by the dextran-coated charcoal separation technique⁽²⁴⁾. In brief,¹²⁵ I-labeled GH was prepared by the chloramine-T method. Purification was achieved by column chromatography. One nanogram of labeled GH was incubated with 50 μL of serum in the absence (total binding) or presence(nonspecific binding) of excess unlabeled GH (1 μg), for 24 h. Free and bound hormone were separated by the addition of dextran-coated charcoal followed by centrifugation. The supernatant solution was decanted, and radioactivity was counted. Specific binding by GHBP was determined by subtracting nonspecific binding. Results were expressed as specific binding(%) of the total counts per 50 μL of serum. Endogenous GH in the sample leads to underestimation of GHBP in a concentration dependent manner. GHBP values were therefore corrected for this effect by reference to a GH standard curve. Assay sensitivity was 1.5%. Intraassay C.V. at 2 and 10% GH binding in 50 μL of serum was 3.4%. GHBP and IGFBP-3 analyses were performed in single assays.

Data analysis. Basal and peak concentrations of IGF-I and IGFBP-3 were analyzed according to the absolute value and SDS. Changes in IGF-I and IGFBP-3 were expressed as absolute increments, ΔSDS, and percent increment. The absolute increments in IGF-I and IGFBP-3 were compared with the values proposed by the Pharmacia Study Group on IGF-I treatment in GH insensitivity⁽¹⁶⁾, namely that an increment in IGF-I of<15 μg/L and in IGFBP-3 of <0.4 mg/L was indicative of GH insensitivity. For the purposes of examining partial GH insensitivity, we have arbitrarily taken an increment in IGF-I or IGFBP-3 of <25% or <0.5 SD as being indicative of GH insensitivity (see Fig. 3).

Figure 3

Relationship between IGF-I and IGFBP-3 generation, expressed as percentage increment. Shaded regions indicate IGF-I or IGFBP-3 increments below 25%. GHD children are indicated by closed circles and GH insensitive children by open circles.

Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) program. Pearson's correlation coefficient was applied for data correlation. Stepwise regression was used to determine the relative significance of different variables (auxologic variables: age, height SDS, height velocity SDS, BMI; biochemical variables: increments in IGF-I, IGFBP-3 (as ΔSDS or percent increment), and GHBP) in predicting peak GH response during a provocation test. Group comparisons were undertaken using the Mann-Whitney U test.

RESULTS

Auxology. The two groups of children could not be differentiated by age, BMI, stature, or height velocity. Additionally, the subset of four GHI children with a defined diagnosis were not significantly different from the remaining children in the GHI group for any auxologic or biochemical parameter.

Biochemistry. Biochemical parameters for all children are summarized in Table 2. The IGF-I, IGFBP-3, and GHBP responses to GH are shown in Figure 1.

Table 2 Median and range of biochemical parameters for control group and study group

Figure 1

Individual basal and peak IGF-I, IGFBP-3 SDS, and GHBP(percent binding in 50 μL of serum) in response to GH for GH insensitive and deficient subjects. Normal ranges for basal IGF-I and IGFBP-3 are indicated by the shaded areas (±2 SD). Mean ± 1 SD are shown as bars.

IGF-I. Basal levels of IGF-I fell 2 SD below the mean for four subjects with GH deficiency and three children with GH insensitivity (Fig. 1). Basal IGF-I concentration was significantly correlated with basal IGFBP-3 concentration (r = 0.86 p< 0.001) and with basal GHBP (r = 0.60, p = 0.01), but was most strongly correlated with age (r = 0.89, p = < 0.001). There was no significant difference in basal IGF-I concentrations or basal IGF-I SDS between the two groups. IGF-I generation, expressed as percentage increment or ΔSDS, was reduced in the GHI group, but did not achieve significance (p = 0.06).

Three children failed to show an IGF-I increment of >15 μg/L(Table 3, a and b). One child was in the GHD group(subject 19) and two, the GHI group (subjects 5 and 10). Four children(subjects 2, 5, 8, and 10), all from the GHI group, showed an increment of<25% (Table 3, c and d) or < 0.5 SD(Table 3, e and f). The absolute rise in IGF-I was inversely correlated with the basal IGF-I level (r = -0.57,p = 0.009), and positively correlated with age (r = 0.78,p < 0.001) and also BMI (r = 0.57, p = 0.009). The correlations with age and BMI were abolished when the increment was expressed as ΔSDS or percent increment in IGF-I, but the inverse relationship between increment and basal IGF-I levels was preserved(ΔSDS IGF-I versus basal IGF-I SDS, r = -0.65,p = 0.002).

Table 3 Classification of GHD and GHI groups according to different modes of expression of IGF-I and IGFBP-3 generation

A significant inverse correlation was determined between peak GH achieved on provocation testing and IGF-I generation (Fig. 2,a and b).

Figure 2

Percentage increment and change in SDS in response to GH for IGF-I (a and b) and IGFBP-3 (c andd) vs GH peak (mU/L) in all children undergoing a generation test. GH deficient children are indicated by closed circles and GH insensitive children by open circles.

IGFBP-3. Basal IGFBP-3 levels fell more than 2 SD below the mean for one GHI subject and four GHD subjects (Fig. 1). Elevated levels (IGFBP-3 SDS > 2.0) were found in three GHI subjects(subjects 2, 8, and 10).

Basal IGFBP-3 concentration was correlated with age (r = 0.69,p = 0.001) but, in contrast to IGF-I, not with basal GHBP(r = 0.42, NS). However, a significant relationship was found between basal IGFBP-3 levels and peak GHBP (r = 0.53, p = 0.02). When expressed as SDS, the relationship with age was abolished, but with peak GHBP enhanced (r = 0.60, p = 0.005). The absolute increment in IGFBP-3 was inversely correlated with the basal level of IGFBP-3 (r = -0.85, p < 0.001) and also with BMI(r = 0.58, p = 0.008), but not with age. The relationship with BMI disappeared when the increment was expressed as ΔSDS or percentage increment in IGFBP-3. Basal IGFBP-3 SDS was significantly lower in the GHD children (p = 0.02) and increased more in response to GH than the GHI children (Table 2).

Four children had an increment in IGFBP-3 of <0.4 mg/L(Table 3, a and b) or ΔSDS < 0.5(Table 3, e and f), including three children from the GHI group and one child with GH deficiency (subject 16). However, eight children had an IGFBP-3 increment of < 25%, of whom six were from the GHI group(subjects 2, 3, 4, 8, 9, and 10) and two from the GHD group (subjects 12 and 16) (Table 3, c and d). As with IGF-I generation, a significant inverse relationship between GH peak on a provocation test and the rise in IGFBP-3 was seen (Fig. 2,c and d). Stepwise multiple regression showed that the most significant parameter to predict GH peak was percentage increment in IGFBP-3.

Significant relationships were found between IGF-I and IGFBP-3 generation expressed in absolute terms (r = 0.46, p = 0.04), asΔSDS (r = 0.49, p = 0.03), or as percentage increment (r = 0.77, p < 0.001) (Fig. 3).

GHBP. GHBP was detectable in all subjects. There was no difference between the two groups in basal or peak GHBP levels. Moreover, there was no consistent response to GH (Fig. 1). Basal GHBP showed a significant relationship with age (r = 0.63,p = 0.008) and with basal IGF-I. Peak GHBP showed significant relationships with basal GH during the provocation test (r = 0.67,p = 0.001) and basal IGFBP-3 SDS.

DISCUSSION

The IGF-I generation test was initially developed as a dynamic test of the GH axis. However, IGF-I generation in the majority of studies either did not reliably distinguish short stature of different etiologies, or predict the growth response to GH therapy^(18–21). However, the IGF generation test remains the mainstay of diagnosis of Laron syndrome^{(15, 16, 25)}.

We have applied the diagnostic criteria for IGF-I and IGFBP-3 generation, advocated for Laron syndrome to a very heterogeneous group of patients. Nevertheless, they represent a distinct group within our clinical practice, who have both a higher median basal GH level (mean for GHI group 16.5 mU/L) and a higher median peak GH level (mean peak 53.8 mU/L) than short normal children attending our clinic (mean basal GH 7 mU/L, mean peak GH 22 mU/L)⁽²⁶⁾. Using the IGF generation test, four children were identified with a poor IGF-I response (increment <15 μg/L, mean age 3.3 y), including one child with multiple pituitary hormone deficiency (Table 3a). Three had low basal levels of IGF-I. The IGFBP-3 generation criterion (increment <0.4 mg/L) identified four children, mean age 6.6 y, with a poor response, including one with GH deficiency. No child fulfilled the Pharmacia Study Group generation test criteria for Laron syndrome. Seven children, however, had discordant responses, with three showing a poor IGF-I response but a satisfactory IGFBP-3 response, and four, the converse (Table 3). Discordance was seen not only in children with possible GH insensitivity (5/11), but also in two with GH deficiency.

Laron et al.⁽²⁷⁾ have recently described three siblings with marked short stature, raised basal GH, and a negligible IGF-I response to GH, but normal IGFBP-3 generation, and have proposed that a selective block to IGF-I generation may be responsible. Subject 5 in this study showed such a response. We now also describe children with short stature, raised peak GH on provocation testing, high IGFBP-3 levels, but normal basal IGF-I levels, who have normal IGF-I generation (increment > 15μg/L) but impaired IGFBP-3 generation.

This study has highlighted certain problems in expressing change in IGF-I or IGFBP-3 as absolute increments, in the assessment of possible GH insensitivity. We found that IGF-I and IGFBP-3 generation were inversely correlated with the basal concentration. Furthermore, IGF-I and IGFBP-3 generation were correlated with BMI and IGF-I generation with age. Thus, young children are much more likely to fulfill the criteria than older children. This may result in spurious identification of impaired IGF-I generation: subject 19, Table 1, showed an increment in IGF-I of 8μg/L, but expressed as ΔSDS or percentage increment, the rise was satisfactory. We therefore propose that the percent increment above the basal level or ΔSDS should be used in the determination of partial GHI.

Using the alternative criteria of an increment in IGF-I or IGFBP-3 <25%, three children (Table 1, subjects 2, 8, and 10) showed both impaired IGF-I and IGFBP-3 generation (Table 3d). These children, all girls, had elevated basal IGFBP-3 levels (basal IGFBP-3 SDS 2.1 to 2.9) and normal IGF-I levels coupled with marked short stature (height SDS -3.9 to -2.8) and high peak GH (GH 80 to 87 mU/L). Two of these children were also identified by changes in IGF-I and IGFBP-3 SDS < 0.5 (Table 3f). Therefore, despite high basal IGFBP-3 levels, these children do appear to be genuinely GHI. One of these children has subsequently received GH on clinical grounds, but has shown a poor response. We suggest that these children have a degree of IGF-I insensitivity which may be related to high levels of IGFBP-3 reducing IGF-I bioavailability. Feedback mechanisms may partly overcome the resistance by maximal stimulation of the GH-IGF axis. Consequently additional exogenous GH cannot produce a further significant rise in IGF-I or IGFBP-3.

All children exhibited detectable GHBP in serum and were therefore unlikely to be GH receptor deficient. GHBP has been shown to be a major determinant of GH responsiveness⁽²⁸⁾. However, there was no significant change in GHBP in response to GH in these children and GHBP did not contribute to the determination of GH insensitivity. There was a strong relationship noted between basal GHBP and age, as previously reported⁽²⁹⁾.

In Laron syndrome, IGFBP-3 generation is a more sensitive and specific determinant of the diagnosis than IGF-I generation⁽¹⁶⁾. IGFBP-3 is also a sensitive and specific marker of GH deficiency⁽³⁰⁾, and pretreatment IGFBP-3 levels predict response to GH in GH deficiency⁽³¹⁾. In addition, in unselected short children undergoing an IGF generation test, Cotterill et al.⁽¹³⁾ showed a significant positive relationship between height SDS and IGFBP-3 concentration. In contrast, IGF-I is a relatively poor marker of GH sensitivity. We have noted that the inverse correlation between GH peak and percentage increment or ΔSDS in IGFBP-3 in response to GH is more significant than the correlation for IGF-I generation. This relationship was not present for absolute increments in IGF-I or IGFBP-3, confirming the findings of Cotterill et al.⁽¹³⁾. We propose that these correlations suggest that there is a range of GH insensitivity and that cut-offs in IGF-I and IGFBP-3 generation to determine GH insensitivity may not be appropriate. We do, however, feel that IGF-I and IGFBP-3 generation(expressed as percentage increment or ΔSDS) provide useful information. In this study, only five of the eleven children from the putative GHI group showed “normal” responses to GH by all of the above criteria (Fig. 3). Stricter selection criteria (ie basal GH >15 mU/L and peak GH >50 mU/L on provocation testing) would have excluded three of these five without also excluding any of the children with either abnormal IGF-I or IGFBP-3 generation. This further implies that peak GH concentrations are an important marker of GH insensitivity.

The identification of differential responsiveness of IGF-I and IGFBP-3 during an IGF generation test has shown that children may have selective resistance in the GH-IGF-I versus GH-IGFBP-3 axes. Such children may well provide an insight into post-GH receptor events and it is to this group of short children that studies of GH receptor sequence should be directed. Mutations of the GH receptor may lead to expression of a functional GH receptor, in terms of binding, but impaired transmission of the immediate postreceptor signal events. IGF-I generation alone, therefore, is inadequate as an index of partial GH insensitivity and it should be used in conjunction with IGFBP-3 generation.