The Effect of Growth Hormone on IGF-1R Activation in Overweight/Obese Women

Abstract

Objective:

Overweight/obesity is characterized by decreased growth hormone (GH) secretion whereas circulating IGF-I levels are less severely reduced. Yet, the activity of the circulating IGF-system appears to be normal in overweight/obese subjects, as estimated by the ability of serum to activate the IGF-I receptor in vitro (bioactive IGF or IGF-1R activation). We hypothesized that preservation of IGF-1R activation in overweight/obese women is regulated by an insulin-mediated suppression of IGF-binding protein-1 (IGFBP-1) and IGFBP-2, and by suppression of IGFBP-3, mediated by low GH. We additionally hypothesized that increases in IGF-1R activation would drive changes in body composition with low-dose GH administration.

Design:

Cross-sectional analysis and 6-month randomized, placebo-controlled study of GH administration in 50 overweight/obese women without diabetes mellitus. IGF-1R activation (kinase receptor activation assay) and body composition (DXA) were measured.

Results:

Prior to treatment, IGFBP-3 (r=−0.33, p=0.02), but neither IGFBP-1 nor IGFBP-2 associated inversely with IGF-1R activation. In multivariate analysis, lower IGFBP-3 correlated with lower peak stimulated GH (r=0.45, p=0.05) and higher insulin sensitivity (r=−0.74, p=0.003). GH administration resulted in an increase in mean serum IGF-I concentrations (144±56 to 269±66μg/L, p<0.0001) and IGF-1R activation (1.29±0.39 to 2.60±1.12μg/L, p<0.0001). The treatment-related increase in IGF-1R activation, but not total IGF-I concentration, predicted an increase in lean mass (r=0.31, p=0.03) and decrease in total adipose tissue/BMI (r=−0.43, p=0.003).

Conclusions:

Our data suggest that in overweight/obesity, insulin sensitivity and GH have opposing effects on IGF bioactivity, as measured by IGF-1R activation, through effects on IGFBP-3. Furthermore, increases in IGF-1R activation, rather than IGF-I concentration, predicted GH administration-related body composition changes.

Introduction

Obesity is characterized by a reduction in growth hormone (GH) secretion, which has been well demonstrated in multiple studies starting with Beck et al. [1]. Both peak-stimulated GH levels and spontaneous GH secretion measured by frequent sampling over 24 hours decrease substantially with increasing BMI [2, 3]. In contrast, serum insulin-like growth factor 1 (IGF-I) concentrations in obese individuals are less severely reduced [4]. While some studies show decreased serum IGF-I concentrations in obesity, the decline in serum IGF-I with increasing BMI is not as marked as the reduction in GH [2, 5–7]. Studies have demonstrated that obese women generate a greater IGF-I response than lean women to a standard dose of GH, suggesting that an increased sensitivity of the liver to GH contributes to the relative preservation of IGF-I [8].

Our group has previously shown that serum bioactive IGF, estimated as the ability of serum IGFs to activate the IGF-I receptor (IGF-IR) in vitro [9, 10], appears to be relatively preserved in obesity, with no significant difference in mean levels between normal and obese individuals [5]. Another study in a population of over 1,000 elderly individuals demonstrated that increased insulin resistance was associated with higher IGF bioactivity, but only in euglycemic subjects. In hyperglycemic subjects with type 2 diabetes (T2D), serum bioactive IGF declined [11]. Thus, it may be speculated that the preservation of IGF bioactivity in obesity is a compensatory mechanism for the preservation of insulin sensitivity and that this mechanism fails in overt T2D.

The regulation of IGF bioactivity is incompletely understood, particularly in obesity. Ninety-nine percent of circulating IGF-1 is bound to IGFBPs, which reduce IGF-1 bioactivity by limiting the ability of IGF-1 to bind the IGF-1 receptor [12]. GH is the primary regulator of IGF-I and of IGF-binding protein-3 (IGFBP-3), which is the main carrier of IGF-I and IGF-II. IGFBP-3 levels are known to be low in individuals with GHD due to hypopituitarism [13], and GH administration increases IGFBP-3 levels in hypopituitarism [14] and in obese subjects without pituitary disorders [15]. Prior studies have also established that states of chronic insulin resistance are associated with suppression of IGFBP-1 and IGFBP-2, which may at least theoretically result in a further increase in IGF-1 bioactivity and insulin sensitivity [16]. In contrast, serum IGFBP-3 levels are higher in obese subjects and in obese subjects with Type 2 diabetes than in lean subjects, suggesting that IGFBP-3 is higher in states of insulin resistance [12]. The regulation of these IGFBPs has complex implications for IGF-1 bioactivity in obesity, which is characterized by insulin resistance and reduced endogenous GH secretion.

Additionally, little is known about the effects of GH administration on the relative proportions of bioactive and total IGF-1 in obesity. There has been an interest in studying the impact of GH administration in obese subjects because the suppressed secretion of GH has been associated with an unfavorable body composition and cardiovascular risk marker profile [17, 18]. Indeed, it appears possible to improve some cardiovascular risk markers, body composition, and intramyocellular fat by low-dose GH therapy, even though insulin sensitivity may decline [19-22]. One could speculate that GH treatment in obese individuals would lead to a relatively larger increase in bioactive IGF than total IGF-1, as was shown for ultrafiltered levels of free IGF-I in healthy, normal-weight males [23]. In contrast, in GH-deficient subjects, GH treatment has been reported to cause a relatively smaller increase in bioactive IGF than total IGF-I [24].

In this study, we first investigated the regulation of IGF-1R activation (termed “IGF bioactivity” in the above-cited papers) in obesity relative to IGFBPs, insulin sensitivity and GH. We hypothesized that the relative preservation of IGF-1R activation in overweight/obese women would be associated with suppression of IGFBP-1 and IGFBP-2 in association with high insulin levels and with reduction in IGFBP-3 levels in association with low endogenous GH. Additionally, to increase our understanding of the complex effects of GH on the IGF-system in obesity, we compared the effects of 3 months of GH treatment on changes in IGF concentrations and IGF-1R activation in obese women characterized by a circulating IGF-I level within or below the lower reference range. Furthermore, we hypothesized that increases in IGF-1R activation, not total IGF-I, would drive changes in body composition with GH administration, reflecting a more sensitive measure of IGF action in vivo.

Methods

Subjects

The study was approved by the Partners Healthcare Inc. (Boston, Massachusetts) Institutional Review Board, and written informed consent was obtained from all subjects. A total of 50 overweight or obese women (BMI ≥25 kg/m²) with waist circumference of >88 cm, and an age between 18 and 45 years were included. All study participants were generally healthy, eumenorrheic without oral contraceptive use and free of diabetes mellitus. Additional inclusion criteria included an IGF-I level at or below the lower two quartiles of the age-appropriate normal range; however, no otherwise-eligible subject had an IGF-1 above this level to require exclusion based on these criteria [22]. Pre-treatment characteristics as well as total lean mass and total IGF-I levels at baseline have previously been reported [2, 5, 18, 22, 25-28].

Study Design

This is a post-hoc analysis of baseline and three-month data from a six-month randomized, placebo-controlled study of GH vs. placebo in women with abdominal obesity. Batched IGF-1R activation and IGFBP measurements were available pre-treatment and at three months, so these time points were selected for the current analysis.

Cross-sectional analysis was performed on data from the subjects prior to randomization and treatment (at “baseline”). Subjects underwent GH-releasing hormone (GHRH)-arginine stimulation testing at baseline. Additionally, serum IGF-I, IGF-II, IGF-1R activation and IGFBPs were measured. Subjects underwent fasting morning oral glucose tolerance testing (OGTT) with calculation of the Matsuda index, a validated index of insulin sensitivity that incorporates both fasting and OGTT stimulated glucose and insulin levels (http://mmatsuda.diabetes-smc.jp/english.html) [29]. Dual energy x-ray absorptiometry (DXA) was also performed for body composition.

Subjects were then randomized to receive GH or placebo. GH was initiated at a dose of 4 μg/kg daily and titrated under the direction of an unblinded study monitor to a target IGF-I level in the upper quartile for age. Sham dose titrations were also made in the placebo group to maintain the double-blinded study design. All baseline testing detailed above, with the exception of GHRH-arginine stimulation testing, was repeated after the three-month treatment period. Of note, computed tomography (CT) was used to assess body composition at baseline and six months in the parent study. However, CT was not performed at the three-month time point. Thus, DXA, which was performed at baseline and three months, was used for assessment of body composition in the current analysis. Three- and six-month effects of GH on body composition and cardiovascular markers have been published for the full six-month longitudinal trial [22]. However, visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT) and total abdominal adipose tissue (TAT) by DXA and effects on the primary endpoint in this study (IGF-1R activation) have not been previously published.

Assays and Testing Protocols

IGF-1R activation was measured by a cell-based kinase receptor activation (KIRA) assay as previously published [9, 10] with modifications [30]. In brief, the bioassay is based on human embryonic kidney cells stably transfected with cDNA of the human IGF-I receptor (IGF-IR) gene. To measure IGF-1R activation, serum is incubated with the cells for 15 min at 37 degrees Celsius to mimic physiologic conditions. When activated, the IGF-IR undergoes tyrosine auto-phosphorylation. After removal of samples the KIRA cells are lysed, and the crude cell-lysates are assayed for activated (tyrosine phosphorylated) IGF-IRs by an ELISA based on reagents from R&D Systems (Abingdon, UK). The signal obtained in serum is compared to signals obtained using a serial dilution of rhIGF-I (WHO 02/254) and accordingly, we can report levels of IGF-1R activation in μg/L. To acknowledge that IGF-II also is able to activate the IGF-IR, we have designated the output of the KIRA assay “IGF-1R activation” (usually termed “bioactive IGF”). Relative IGF-1R activation represents the proportion of bioactive IGF, as measured by IGF-1R activation, to total IGF-I. IGFBP-1 and IGFBP-2 were measured by a validated, in-house TF-IFMA as previously described [31, 32]. IGF-II was measured after acid-ethanol extraction of IGFBPs as previously described [33] with modifications [34]. IGFBP-3 was measured by a commercial chemiluminescence immunoassay on the IDS-iSYS multi-discipline automated analyzer (Immunodiagnostic Systems Nordic SA, Denmark) [35]. Total IGF-I was measured by the Immulite 2000 automated immunoanalyzer (Diagnostic Products Corp., CA, USA). Care was taken to asses all samples from the same individual within the same assay batch.

DXA for Body Composition Analysis

Lean body mass was measured by DXA (Hologic QDR-4500; Hologic, Inc., Waltham, MA, USA) at baseline and three months. The precision error was 2.4% for lean mass [22]. Estimated VAT and SAT were assessed using Hologic APEX 4.0 software (Hologic), as previously described [27, 36].

Statistical Analysis

Statistical analysis was performed using JMP Pro Statistical Database Software (version 11.0.0; SAS Institute, Cary, NC). Results are reported as mean ± standard deviation unless otherwise noted. Variables were log-transformed and means compared by ANOVA. Additionally, univariate and standard least squares multivariable regression were performed after log transformation of variables. Pearson’s correlation coefficients were reported for univariate regressions. Specific variables included in each multivariable model are noted in the Results section. Percent change of longitudinal variables was calculated as follows: Percent Change = log(three month time point data) - log(baseline time point data). Percent change between the GH and placebo groups were compared by ANOVA. Differences with a p-value ≤0.05 were considered statistically significant.

Results

Baseline Characteristics

Baseline characteristics of the whole cohort as well as those randomized to GH and placebo are reported in Table 1. The mean age of the cohort was 36.8±6.5 years and mean BMI was 34±6 kg/m². GH/IGF-I axis testing, OGTT results and body composition variables are also reported in this table. Of note, no baseline variables differed between subjects randomized to GH and placebo.

Table 1. Baseline Characteristics.

Data are presented as mean ± SD. No values were significantly different between the GH and placebo groups at baseline by ANOVA

	All	GH	Placebo
Demographic
Age, y	37 ± 7	36 ± 6	38 ± 7
GH/IGF-I Axis
GHRH-arginine Stimulation
Peak GH, μg/L	14.0 ± 9.15	12.6 ± 10.0	15.7 ± 8.2
IGF-I, μg/L	135 ± 55	144 ± 56	125 ± 53
IGF-I Z-score	−1.7±0.5	−1.6±0.5	−1.8 ± 0.5
IGF-1R Activation, μg/L	1.28 ± 0.45	1.29 ± 0.39	1.26 ± 0.52
Relative IGF-1R Activation (%)*	1.04 ± 0.40	1.01 ± 0.45	1.07 ± 0.34
IGF-II, μg/L	521 ± 92	509 ± 85	530 ± 97
IGF Binding Proteins
IGFBP-1, μg/L	26.4±17.1	22.9 ±13.9	30.7 ±19.8
IGFBP-2, μg/L	102.8 ± 55.7	97.8 ± 59.9	109.0 ± 50.9
IGFBP-3, μg/L	4733 ± 840	4852 ± 912	4586 ± 737
Body Composition
BMI, kg/m2	34 ± 6	34 ± 5	35 ± 7
VAT, g	665 ± 238	629 ± 200	706 ± 274
SAT, g	2575 ± 627	2587 ± 608	2561 ± 663
TAT, g	3240 ± 760	3217 ± 678	3267 ± 863
TAT/BMI, g/kg/m2	93 ± 12	94 ± 12	93 ± 12
Total Lean Mass, kg	51.9 ± 5.8	52.2 ± 6.0	51.5 ± 5.7
Glucose Homeostasis
Fasting Glucose, mg/dL	86 ± 7.7	87 ± 8.8	85 ± 6.1
OGTT 2 Hour Glucose, mg/dL	124 ± 30	122 ± 29	127 ± 31
HOMA-IR	1.73 ± 0.95	1.78 ± 1.03	1.66 ± 0.85
Matsuda Index	6.5 ± 3.6	6.5 ± 3.0	6.6 ± 4.3

^*Relative IGF bioactivity is defined as bioactive IGF/Total IGF-1.

Determinants of IGF Bioactivity in Overweight/Obese Women

Determinants of IGFBPs

Insulin sensitivity (as measured by the Matsuda insulin sensitivity index) was positively associated with IGFBP-1 and IGFBP-2 levels in univariate analysis (r=0.62, p<0.0001 and r=0.53, p=0.0001) (Figures 1A and 1B), indicating that reduced insulin sensitivity was associated with lower IGFBP-1 and IGFBP-2 levels. In contrast, IGFBP-3 was inversely associated with the Matsuda insulin sensitivity index (r=−0.38, p=0.01) (Figure 1C) on univariate analysis, indicating that reduced insulin sensitivity was associated with higher IGFBP-3 levels.

Figure 1:

The Matsuda insulin sensitivity index was a negative predictor of IGFBP-3, indicating that higher levels of insulin resistance are correlated with higher levels of IGFBP-3 (Panel A). The Matsuda index was a positive predictor of IGFBP-1 (Panel B) and IGFBP-2 (Panel C), indicating an association between high insulin resistance and low levels of IGFBP-1 and IGFBP-2. IGFBP-3 was a significant negative predictor of bioactive IGF (Panel D). However, neither IGFBP-1 and IGFBP-2 were correlated with IGF-1R activation (Panels E and F, respectively). Bioactive IGF and IGFBP measurements are in μg/L and Matsuda Insulin Sensitivity Index is unitless.

Multivariable models were then constructed to evaluate determinants of IGFBP-1 and IGFBP-2 levels. Variables including VAT and the Matsuda insulin sensitivity index were entered into the model based on significant univariate correlations. In these models, the Matusda index remained the only significant predictor of IGFBP-1 (r=0.52, p=0.0003) and IGFBP-2 (r=0.41, p=0.006).

Determinants of IGFBP-3 levels were evaluated in a similar manner. As above, VAT and the Matsuda insulin sensitivity index were entered into the model based on significant univariate correlations. Because IGFBP-3 is a GH-dependent binding protein, peak-stimulated GH was also included as an independent variable. In this model, peak-stimulated GH (partial correlation coefficient r=0.45, p=0.05) was a significant positive determinant and Matusda index (partial correlation coefficient r=−0.74, p=0.003) a significant negative determinant of IGFBP-3. VAT was not a significant determinant of IGFBP-3 in this model.

Determinants of IGF-1R Activation

IGFBP-3 was negatively associated with measures of IGF bioactivity, including IGF-1R activation (r=−0.41, p=0.004) (Figure 1F) and relative IGF-1R activation (r=−0.39, p=0.006), indicating an increased IGF-IR activation in the presence of low IGFBP-3. Contrary to our hypothesis, serum IGFBP-1 and IGFBP-2 levels were not associated with serum IGF-1R activation or relative IGF-1R activation (r=−0.04, p=NS and r=0.1, p=NS, respectively) (Figure 1D-E), suggesting that these binding proteins were not the major regulators of IGF-1R activation in individuals with overweight/obesity.

VAT was an additionally negatively associated with IGF-1R activation (r=−0.39, p=0.006) and total IGF-I (r=−0.53, p=0.0001), but not relative IGF-1R activation (r=0.22, p=NS). There was a trend toward an association between IGF-1R activation and the Matsuda index (r=0.28, p=0.06), whereas the Matsuda index correlated with neither total IGF-I or relative IGF-1R activation. Notably, peak-stimulated GH did not correlate with any of the IGF variables measured.

Multivariable models were constructed for the following dependent variables: IGF-I, IGF-1R activation and relative IGF-1R activation. IGFBP-3, VAT, age and Matsuda index were entered as independent variables. Similar to the univariate models, IGFBP-3 was a negative determinant of relative IGF-1R activation (partial r=−0.49 and p=0.001 for relative IGF-1R activation), suggesting that low IGFBP-3 is associated with relative preservation of IGF bioactivity. IGFBP-3 trended toward being a positive determinant of total IGF-I (partial r=0.29, p=0.07) and negative determinant of IGF-1R activation (partial r=−0.27, p=0.08), suggesting an association between lower IGFBP-3, lower total IGF-I and higher IGF bioactivity.

In these models, VAT was also a weak negative predictor of both total IGF-I (partial r=−0.29, p=0.06, trend only) and IGF-1R activation (partial r=−0.33, p=0.03) but was not a significant determinant of relative IGF-1R activation. Age was a significant positive predictor of relative IGF-1R activation (partial r=0.32, p=0.04 for IGF-1R activation/total IGF-I), but was not a significant determinant of absolute levels of total IGF-I or IGF-1R activation. Matsuda index was not a significant predictor of any IGF-1 variable. Finally, peak-stimulated GH was not a significant predictor of IGF-1R activation, total IGF-I or relative IGF-1R activation when added to the models above.

Effects of GH Administration

Changes in IGF-related Variables and Insulin Sensitivity

The mean GH dose for subjects in the treatment group was 1.3±0.5 mg/day. IGF-1R activation and total IGF-I both increased with GH treatment versus placebo (Figure 2A and B, respectively). The mean IGF-I z-score increased from −1.6±0.5 to 0.0±1.0 (Figure 2C) within the GH-treated group, demonstrating an increase of IGF-I levels to within the normal range as per protocol design. The Matsuda index decreased significantly in the GH versus the placebo group, reflecting higher insulin resistance (Table 2).

Figure 2:

IGF-1R activation (Panel A), total IGF-I (Panel B), IGF-I Z-score (Panel C) increased in response to GH versus placebo administration over three months. * Denotes p<0.0001.

Table 2. Effects of GH Administration Over 3 Months.

Data presented as mean ± SD. Percent change in GH versus placebo groups were compared by ANOVA.

	GH % Change 0 to 3 Mo	Placebo % Change 0 to 3 Mo	P-value GH vs PL
GH/IGF-I Axis
IGF-I, μg/L	67 ± 45.0	−2.1 ± 30.1	<0.0001
IGF-1R Activation, μg/L	67.1 ± 35.8	−2.6 ± 22.0	<0.0001
Relative IGF-1R Activation (%)*	0.02 ± 0.39	0.00 ± 0.30	NS
IGF-II, μg/L	6 ± 14	9 ± 10	NS
IGF Binding Proteins
IGFBP-1, μg/L	5.6 ± 49.2	−7.7 ± 38.8	NS
IGFBP-2, μg/L	10.6 ± 20.9	4.6 ± 26.8	0.04
IGFBP-3, μg/L	4.7 ± 12.8	−1.2 ± 7.8	0.06
Body Composition by DXA
BMI, kg/m2	0.61 ± 3.2	−0.070 ± 3.2	NS
VAT, g	−3.3 ± 11.4	0.08 ± 14.4	NS
SAT, g	0.04 ± 5.6	1.6 ± 6.9	NS
TAT, g	−0.73 ± 5.6	1.2 ± 6.3	NS
TAT/BMI, g/kg/m2	−2.1 ± 5.0	1.3 ± 4.5	0.02
Total Lean Mass, g	3.7 ± 4.3	0.6 ± 3.5	0.007
Glucose Homeostasis
Fasting Glucose, mg/dL	4.3 ± 10.2	0.67 ± 9.7	NS
OGTT 2-Hour Glucose, mg/dL	8.2 ± 22.7	−6.6 ± 18.7	0.02
HOMA-IR	33.4 ± 87.9	−7.0 ± 57.8	0.07
Matsuda Index	−0.37 ± 0.6	0.08 ± 0.4	0.006

^*Relative IGF bioactivity is defined as bioactive IGF/Total IGF-1.

Changes in IGFBPs and Associations with IGF Variables

There was a significant increase in IGFBP-2 (10.6 ± 20.9 vs 4.6 ± 26.8, p=0.04) and a trend towards an increase in IGFBP-3 (4.7 ± 12.8 vs −1.2 ± 7.8, p=0.06) in the GH versus placebo groups. There was no difference in the change in IGFBP-1 between the GH and the placebo groups (Table 1).

After GH treatment, total IGF-1 and IGF-1R activation were not associated with any IGFBP variable (IGFBP-1, IGFBP-2 or IGFBP-3). Relative IGF-1R activation was negatively correlated with IGFBP-3 (R=−0.38, p=0.05) but not IGFBP-1 or IGFBP-2.

Changes in Body Composition and Associations with IGF Variables

Lean mass significantly increased in the GH versus placebo group over three months (3.7± 4.3% vs. 0.6±3.5%, p=0.007, Figure 3A). Additionally, the TAT/BMI ratio, a measure of relative truncal adiposity, decreased significantly in the GH versus placebo group over three months (−2.5±5% vs. 1.3±4.5%, respectively p=0.02, Figure 3B). There was no change in BMI or other measures of adiposity in the GH vs. placebo group over this three-month period. An increase in IGF-1R activation over the three-month period predicted both the increase in lean mass and decrease in TAT/BMI ratio (Figure 3C and 3D, respectively). In contrast, there was no correlation between change in total IGF-I over three months and these or any other body composition variables.

Figure 3:

There was a greater increase in lean mass (Panel A, p=0.007) and greater decrease in TAT/BMI (Panel B, p=0.02) in the GH versus placebo group over three months. Percent change in lean mass (Panel C) and TAT/BMI ratio (Panel D) were both significantly correlated with change in IGF-1R activation but not total IGF-I (not shown).

Determinants of IGF-II and Effects of GH Administration

IGF-II levels were not significantly different between the GH and placebo groups prior to GH treatment (Table 1). Pre-treatment IGF-II levels were highly correlated with IGFBP-3 levels (r=0.77, p<0.0001) as well as the Matsuda index (r=0.42, p=0.004). IGF-II levels did not change with low-dose GH treatment compared with placebo administration (563±90 vs. 558±89 μg/L, p=NS) (Table 2).

Discussion

This study had two main findings. First, we demonstrated that in overweight/obese women without diabetes mellitus, higher insulin resistance predicted higher levels of IGFBP-3 while low GH predicted lower levels of IGFBP-3. Lower IGFBP-3 in turn was associated with upregulation of relative IGF-1R activation (usually termed “IGF bioactivity”). This suggests that in obesity, relative GH deficiency and insulin resistance may have opposing effects on IGFBP-3, resulting in variable effects on IGF bioactivity in obesity depending on the degree of GH suppression and insulin resistance. We also showed that increases in IGF-1R activation, but not total IGF-I concentrations, over three months of GH treatment were associated with favorable changes in body composition. This suggests that bioactive IGF is physiologically more closely linked to metabolic and body composition changes induced by GH than total IGF-I when measured by immunoassay.

It is well known that GH secretion decreases with increasing weight, with significant decreases noted above a BMI of 37.5 kg/m² [1-3, 5, 37, 38]. This may explain why serum IGF-I levels decline when BMI approaches 40 kg/m² [37]. In contrast, data on serum total IGF-I in obesity are controversial, with studies reporting both low and normal IGF-I levels, perhaps due to heterogeneity of IGF-I assays as well as differences in age, insulin levels and body composition, including VAT, among subjects. There is evidence that obese individuals have a greater response in total IGF-1 to a standard dose of GH compared with lean individuals, suggesting that increased sensitivity to GH in obesity is a potential mechanism for a relative preservation of total IGF-1 in this population [8, 39].

However, there is even less known about endogenous IGF-I action in obesity, particularly across varying degrees of insulin resistance. Studies suggest an inverse U-shaped curve, with higher levels of IGF bioactivity in obese individuals with mild insulin resistance and a significant drop in IGF bioactivity in individuals with more severe metabolic derangements [5, 40, 41]. One study demonstrated higher free IGF-I in non-diabetic obese subjects versus lean controls, whereas obese diabetic subjects had similar free IGF-I levels compared to lean controls [41]. Data from our group demonstrated that in the absence of diabetes mellitus, obese women had relatively preserved bioactive IGF compared to lean women and overweight women, despite a marked suppression in 24-hour GH secretion [5]. Moreover, Brugts et al. showed that there was a significant reduction of bioactive IGF in elderly obese men, but this reduction was only evident in those who had overt diabetes mellitus or who met all five metabolic syndrome criteria with relatively preserved IGF bioactivity under more moderate conditions [40].

Our data are concordant with these studies, suggesting that in the state of obesity, GH deficiency contributes to low IGFBP-3 while insulin resistance is associated with high IGFBP-3. The present data also demonstrated a negative association between IGFBP-3 and IGF bioactivity in this population. This is in contrast to the published data that IGFBP-3 is directly related to IGF bioactivity in states of normal or elevated GH [5, 40, 42]. Taken together, these data could explain why in obesity without diabetes mellitus, relative GHD may act to lower IGFBP-3 liberating relatively more IGF-I to its unbound and active state. This could be a potential compensatory mechanism in obesity, as an increase in IGF bioactivity would serve to enhance insulin sensitivity. However, as insulin resistance develops, this could act as an opposing force to GH, raising or restoring IGFBP-3 and thus explaining why increases in IGF bioactivity are lost at the extremes of insulin resistance and the metabolic syndrome.

Contrary to our hypothesis, our data did not demonstrate an association between IGFBP-1 or IGFBP-2 and IGF bioactivity in otherwise healthy overweight/obese women. Our hypothesis that regulation of IGF-I bioactivity would be dependent on insulin-mediated decreases in IGFBP-1 and IGFBP-2 was based on prior literature demonstrating negative regulation of IGFBP-1 by hyperinsulinemia in both in vitro and human in vivo experiments [43-47]. Although our data showed that IGFBP-1 and IGFBP-2 were indeed negatively associated with insulin resistance, these IGFBPs did not appear to be significant drivers of IGF bioactivity itself. We speculate that the changes in IGFBP-3, which are present in much higher concentrations than IGFBP-1 and −2, have obscured any impact of the two other IGFBPs on IGF-bioactivity. GH administration was associated with an increase in IGFBP-3, as expected. However, IGFBP-3 was not associated with changes in total IGF-1 or absolute IGF-1R activation. Of note, these results apply only to serum measurements of IGF-1R activation and IGFBPs, and not to the interstitial fluids of target tissues, which warrant further study.

Our second main finding is that the IGF-IR activating potential of serum bioactive IGF appears to be a more accurate and sensitive measure of IGF-I action in obesity than the circulating IGF-I concentration. Our data demonstrate that the increase in IGF-1R activation, not total IGF-I, correlated with an increase in lean mass and decrease in relative VAT with three months of GH administration. When this study was extended to 6 months, our group showed that the increase in total IGF-I with GH treatment correlated with a decrease in VAT as measured by CT (r=−0.56, p=0.002), suggesting that with more time and larger changes in body composition, total IGF-I levels are able to predict body composition changes as well [22]. Given that total IGF-I assays are relatively inexpensive and easily accessible, they will likely remain the mainstay of assessment of the GH/IGF-I axis in clinical research and clinical care. However, our data demonstrate that IGF-1R activation may be a more sensitive measure of IGF-I action, and it will remain an important research tool to further elucidate the regulation of IGF-I action in states of altered IGFBP levels, such as obesity. Of note, the lack of any augmentation in IGF-II with GH administration suggests that changes in IGF-1R activation are directly related to IGF-I.

Limitations of our study include that the examination of the regulation of IGF-1R activation is cross-sectional in nature, which precludes the definitive determination of causality. Our study is also limited in that it involves post-hoc analysis within a larger trial. In addition, this study examined otherwise healthy overweight and obese menstruating women without type 2 diabetes (T2D), and the findings might not be fully generalizable to all populations, such as men, individuals with T2D, or older men and women, as these groups all represent states that could further alter IGF-I, IGF-II and IGFBP dynamics. The study is additionally limited to the examination of serum measurements of total IGF, IGF-1R activation and IGFBPs, which do not reflect levels in interstitial fluids of target tissues. Additionally, peak-stimulated GH levels were only available in a subset of subjects, limiting the power for this analysis. The present study was based on a placebo-controlled RCT, but still we cannot conclude that the correlation between serum IGF-IR activation and changes in lean body mass is causal. Nor can we conclude that measurement of total IGF-I is of less biological significance than the cell-based activity measurement. Such conclusions require further studies. In addition, IGFBP-cleaving proteases play an important role in liberating bound IGFs from the IGFBPs, and in this way, the proteases are believed to regulate the activity of the IGF-system in vivo [48]. In the present study, we did not measure degradation of IGFBPs, and therefore, we cannot exclude that this may impact the observed changes in body composition or the ability of serum to activate the IGF-IR in vitro. However, we have no evidence that GH treatment alters the degradation of IGFBP-3, which was the only IGFBP that showed a correlation with the IGF-1R activation measurement [49].

In conclusion, this study suggests that the relative IGF-1R activation in obesity may be upregulated by low levels of IGFBP-3 driven by relative GH deficiency and insulin resistance. These data also suggest that IGF-1R activation is a more sensitive measure of IGF-I action than total IGF-I. This work demonstrates the importance of IGF-1R activation as an important research tool, though it is likely that total IGF-I will remain the main measurement of IGF-I activity in the clinical setting due to cost and accessibility. Moreover, this work will help to further define the action and regulation of the IGF system in states of altered IGFBPs, including obesity and insulin resistance, in which alterations of the GH/IGF-I axis have broad pathophysiologic and therapeutic implications. Future areas of investigation remain, including the potential for sex differences in IGF-1R activation as well as the regulation of IGF-1R activation at the end-organ or tissue level.

Highlights.

• In obesity, insulin sensitivity and GH have opposing effects on IGF bioactivity, as measured by IGF-1R activation.

• These effects on IGF-1R activation appear to be mediated through IGFBP-3.

• Increase in IGF-1R activation predicted body composition change with GH administration.

• In contrast, total IGF-I was not predictive of any body-composition changes.