Growth hormone (GH) and insulin-like growth factor I (IGF-I) deficiencies have been associated with osteopenia in both children and adults. To examine the effects of growth hormone resistance on bone mineral and body composition, we studied 11 adults (mean age 30 years) with growth hormone receptor deficiency (GHRD, Laron syndrome) and 11 age- and gender-matched controls from Southern Ecuador. Bone mineral and body composition were determined by dual-energy X-ray absorptiometry. Bone physiology was assessed with biochemical markers of bone turnover and dynamic bone histomorphometry. Bone size and body composition differed markedly between subjects with GHRD and controls. Affected adults were 40 cm shorter than controls, had significantly less lean body mass, and had increased percent body fat. Bone mineral content and density (BMD) at the spine, femoral neck, and whole body were significantly lower in adults with GHRD than in controls. Mean BMD Z scores were −1.5 to −1.6 at all sites in affected women and −2.2 to −2.3 in men with GHRD. Estimated volumetric bone density (BMAD) at the spine and femoral neck, however, was not reduced in GHRD. Spine BMAD was 0.210 ± 0.025 versus 0.177 ± 0.021 for affected women versus controls (p < 0.05) and 0.173 ± 0.018 versus 0.191 ± 0.025 for men with GHRD versus normals (p = 0.31). Urinary pyridinoline concentrations were significantly greater in adults with GHRD than in controls, while type I collagen C-telopeptide breakdown products and markers of bone formation did not differ. Differences in histomorphometry were limited to a reduction in trabecular connectivity; bone volume and formation rate were similar to controls. These data confirm the importance of the GH/IGF axis in regulating bone size and body composition. The contribution of these peptides to the acquisition and maintenance of bone mineral is less certain since volumetric bone density was preserved despite low levels of IGF-I and IGFBP-3 associated with GH resistance.
Growth hormone (GH) and the insulin-like growth factors (IGFs) are essential for normal bone growth, but the contribution of these hormones to bone mineral acquisition remains controversial. Both GH and IGF-I have been shown to stimulate osteoblast proliferation and differentiated function in vitro.(1–4) Children(5,6) and adults(5,7) with GH deficiency have reduced bone mass which increases with recombinant human growth hormone (hGH) therapy.(6) Osteopenia has also been observed in adults who have low circulating levels of IGF-I without classical GH deficiency.(8,9) Following IGF-I therapy in these patients(10) or in elderly women,(11) markers of bone formation and resorption increase. In children with GH resistance, IGF-I therapy has been shown to increase bone mineral.(12)
In most settings, it is not possible to separate the direct effects of GH on bone from those mediated indirectly by generation of IGF-I from liver and local tissues. The identification of many GH receptor deficient (GHRD) subjects from Southern Ecuador permits the study of the biologic effects of IGF-I independent of GH.(13,14) GHRD is a rare autosomal recessive disorder characterized by clinical features of GH deficiency, but with normal or elevated serum GH concentrations. Mutation of the GHR gene(15) results in reduced GHR protein or decreased affinity of the GHR for its ligand. Biological nonresponsiveness to endogenous or exogenous GH occurs as a result. Marked deficiencies of IGF-I, IGF-II, and IGF-binding protein 3 (IGFBP-3) occur under these conditions. Initial studies found premenopausal Ecuadorian women with GHRD to have significantly lower bone densities than local controls,(16) despite normal sex steroid concentrations.
The present study was designed to examine bone mineral and body composition in Ecuadorian adults with GHRD preparatory to a 24-month open trial of recombinant human IGF-I (rhIGF-I) therapy. Dual-energy X-ray absorptiometry (DXA) and studies of biochemical bone markers were completed in 11 affected adults and 11 normal controls. Bone histomorphometry was also performed to examine bone physiology and to address the problems of interpreting areal bone density in the short adults with GHRD.
MATERIALS AND METHODS
Twenty-two adults were recruited from the affected population originating from Southern Ecuador. Eleven (8 women and 3 men) had phenotypic and biochemical findings of GHRD including profound short stature, baseline serum GH concentration above 10 ng/ml, and serum IGF-I, IGF-II, IGFBP-3, and GHBP concentrations more than 2 SD below control age-matched levels. All subjects with GHRD were homozygous for an alternative splice mutation site of codon 180 in exon 6 of the GHR gene.(15) Eleven unaffected, healthy adult relatives (6 women) of similar age served as controls; 3 were homozygous normal and 8 were heterozygous for the abnormal GHR gene. Five of the controls were siblings of subjects with GHRD.
The study protocol was approved by the Human Subjects Institutional Review Boards of Stanford University (Stanford, CA, U.S.A.) and the University of Florida (Gainesville, FL, U.S.A.), and the Ethics Committee of the Institute of Endocrinology, Metabolism, and Reproduction (IEMIR, Quito, Ecuador) in compliance with the laws and regulations of the United States and Ecuador. All subjects signed informed consent forms provided in Spanish.
Bone density of lumbar spine (L2–L4), proximal hip, and whole body was determined by DXA (Lunar, Madison, WI, U.S.A.) in the pencil beam mode. In the IEMIR laboratory, the in vivo coefficient of variation for replicate bone mineral density (BMD) measurements is 0.8% for whole body and lumbar spine BMD and 1.5% for femoral neck; the precision error for bone mineral content (BMC) is 1.6% for whole body, 1.4% for spine, and 2.7% for femoral neck.
Bone mass was expressed in conventional terms of BMC (grams) and BMD (g/cm2). Age-specific BMD Z scores were provided by Lunar. Both BMC and BMD are influenced by bone size and may underestimate the bone mineral properties in smaller individuals.(17) The subjects with GHRD had profound short stature which could potentially bias BMC and BMD results. To compensate for differences in body size between controls and affected adults, we used an approximation of volumetric bone mass (g/cm3), BMAD, which minimizes the influence of bone size.(17,18) At the spine, the expression for BMAD was spine BMC ÷ (area)1.5; femoral neck BMAD was calculated from the formula BMC ÷ (area)2. Whole body BMC ÷ height was used as a partial correction for body size for whole body bone mass. Bone densitometry was performed within 3 months prior to bone biopsy.
To permit dynamic histomorphometry, subjects were given tetracycline according to a standardized schedule before biopsy. On days 1 and 2, they received demeclocycline (300 mg twice daily); on days 14–16, they took tetracycline (500 mg twice daily). The biopsy was obtained on day 20. Bone biopsy specimens were taken transversely from the anterior iliac crest using an 8 mm wide Meunier trephine. The biopsy cores were placed immediately into 70% ethanol for transport to Seattle, WA, U.S.A. (Dr. Susan Ott). On arrival, specimens were embedded in methacrylate for nondecalcified sectioning and staining. Samples were sectioned with a microtome to a thickness of 8 μm and stained with Goldner's stain. Two sections, separated by 100 μm, were examined. The cancellous tissue area measured was 28 ± 8 mm2. The Osteometrics (Atlanta, GA, U.S.A.) digitizer and program were used to quantitate sections. Cancellous bone was measured beginning 0.2 mm from the endocortical surface. Osteoid width was directly measured, at equidistant intervals along osteoid-covered surfaces, if it was at least 3 μm wide. Wall width was measured under polarized light, using a grid. Mineralizing surface was the sum of the double labels and half the single labels. The following were directly measured: tissue area, bone area, osteoid perimeter, eroded perimeter, quiescent perimeter, osteoid width, wall width, cortical width, number of acid phosphatase–staining osteoclasts, tetracycline label length, and distance between tetracycline labels. Using the computer-generated tracings, the average marrow-space diameter was measured by using a grid measuring every 60° from the marrow-space grid points to the nearest bone surfaces. The numbers of trabecular nodes and ends were directly counted from the tracings. Histomorphometric parameters were calculated according to the American Society of Bone and Mineral Research nomenclature.(19) The formation period was calculated by dividing the wall thickness by the mineral apposition rate (unadjusted). Trabecular thickness calculations assumed a plate model. Specimens were identified only by code number, and measurements were made without knowledge of the clinical status of the study subject.
The major structural parameters considered were cortical width, cancellous bone volume, trabecular thickness, marrow diameter, and node/end ratio. Well-connected bone has a small marrow diameter and a high node/end ratio. Resorption is related to eroded surfaces and osteoclast count. Parameters that reflect aspects of bone formation include the osteoid surface and thickness, bone formation rate per bone volume (the product of the mineralizing surface and the mineral apposition rate), and the formation period, which is the time required to fill an individual resorption cavity. The wall thickness reflects the amount of bone filled by one team of osteoblasts.
Fasting morning samples of blood and urine were collected within 3 months of bone biopsy for measurement of hormone levels and markers of bone metabolism. Samples were frozen immediately and stored at −70°C until assayed. Specimens from all subjects were assayed within the same batch to eliminate interassay variability. Commercial kits were obtained from the following sources: intact parathyroid hormone (PTH) from Nichols Institute (San Juan Capistrano, CA, U.S.A.); C-terminal propeptide of type I collagen (CICP) and total pyridinolines from Metra Biosystems (Mt. View, CA, U.S.A.); osteocalcin (IRMA), type I collagen C-telopeptide breakdown products (CTX, CrossLaps™), luteinizing hormone (LH) (DSL-4600 Active LH IRMA), follicle stimulating hormone (FSH), and estradiol from Diagnostic Systems Laboratories (Webster, TX, U.S.A.). Serum concentrations of IGF-I (DSL-5600) and IGFBP-3 (DSL-6600 ACTIVE IGFBP-3) were determined using immunochemiluminometric assays kits provided by Diagnostic Systems Laboratories.
Statistical analyses was performed using StatView 2 (Abacus Concepts, Inc., Berkeley, CA, U.S.A.). Data are reported as the mean ± SD. Regression analysis was used to assess associations between bone marker results and measures of bone mineral. Student's t-test for unpaired samples was used to compare GHRD with control subjects. A single group t-test was used to determine if BMD Z score data of the controls was significantly different from zero. Significance was based on two-tailed tests at the p = 0.05 alpha level.
Characteristics of the study participants are presented in Table 1. The subjects ranged in age from 19 to 44 with a mean of 29.6 years for both groups. Patients with GHRD were well matched to controls by age. The mean heights of the women and men with GHRD were 37 and 44 cm less those of than controls, respectively. Weight was also significantly lower in patients than in unaffected adults. Mean body mass index tended to be greater in women with GHRD than in controls, while affected men had significantly lower body mass index than male controls. DXA-derived body composition data indicated a marked reduction in lean body mass, but not fat mass, in the patients. All subjects with GHRD had significantly greater percent body fat and markedly less lean body mass than did controls.
Table Table 1. Bone Mineral and Body Composition of Study Subjects
As expected, IGF-I and IGFBP-3 concentrations were significantly lower in adults with GHRD than in control subjects (IGF-I, 25.3 ± 14 vs. 253.8 ± 92.2 ng/ml; IGFBP-3, 698 ± 331 vs. 3330 ± 818 ng/ml; p < 0.0001). IGF-I and IGFBP-3 concentrations were more than 2 SD below the expected range for age and gender.
Estradiol, LH, and FSH measurements in 20 of 22 subjects were within the expected range for age and gender. Two 44-year-old women, one with GHRD and one control, had low serum estradiol (11.8 and 18.4 pg/ml, respectively) and elevated LH (54 and 63 mIU/ml, respectively) and FSH (33 and 34 mIU/ml, respectively) concentrations, consistent with a perimenopausal state. Analyses of bone turnover markers, bone mineral and histomorphometry conducted without and with these two subjects did not alter the results; therefore, their data were included in the results reported below.
Intact PTH concentrations were normal in all but one 30-year-old woman with GHRD whose serum concentration was 77.9 pg/ml, just above the upper limit of 72 for normal adult women. Serum CICP concentrations did not differ significantly between affected and unaffected subjects (158 ± 35 vs. 135 ± 59 ng/ml, respectively; p = 0.28). Osteocalcin concentrations were also not different between the two groups (17.8 ± 6.8 vs. 15.8 ± 9.9 ng/ml for GHRD and controls, respectively).
Urinary pyridinoline and CTX values, markers of bone resorption, were divided by creatinine concentrations to correct for differences in water excretion. Urinary pyridinoline concentrations were significantly higher in subjects with GHRD than in controls (51.6 ± 19.9 vs. 35.1 ± 16.6 nM/mM creatinine, respectively; p < 0.05). When the two perimenopausal subjects were excluded from the analysis, the difference between pyridinolines in GHRD and controls was more marked (52.9 ± 10.5 vs. 30.5 ± 7.5, respectively; p < 0.005). Mean urinary CTX concentration also tended to be greater in subjects with GHRD versus unaffected adults (505 ± 280 vs. 286 ± 226 μg/l/mM creatinine, respectively; p = 0.06); the difference between groups remained nonsignificant after exclusion of the perimenopausal women (p = 0.10).
Since adults with GHRD had significantly lower lean body mass than controls, urinary creatinine excretion would be reduced. Therefore, adjustment of urinary markers of bone resorption for creatinine concentrations could lead to spurious elevations in these peptides. We addressed this problem by estimating the 24-h urinary creatinine excretion using formulae to relate this to the lean body mass,(20) and then calculating the pyridinoline excretion in millimoles per day. This value was divided by the whole body BMC to express pyridinoline per unit of bone mass, which remained elevated in the GHRD subjects (0.37 ± 0.14 vs. 0.24 ± 0.04, respectively; p = 0.04). Markers of bone formation and resorption were not correlated with the corrected bone mass (BMAD) at femoral neck or spine in the 22 subjects.
Bone mineral status
As shown in Table 1, lumbar spine, femoral neck, and whole body BMC values was markedly lower in the subjects with GHRD than in controls (p < 0.005). This observation was expected in light of the significantly smaller bone area in the adults with GHRD. Areal bone density, expressed both as BMD and BMD Z score, was also reduced in those with GHRD. When data from men and women were combined, differences in BMD between patients and controls were highly significant at lumbar spine (p < 0.005), at left femoral neck (p < 0.0001), and for whole body (p < 0.0001). The BMD Z scores for controls were not significantly different from zero, indicating that the BMDs of the unaffected men and women fell within the expected range for healthy adults.
Because areal bone density fails to correct for differences in bone thickness, we also estimated BMAD at the spine and femoral neck.(17) Spine BMAD (g/cm3) was significantly greater in women with GHRD when compared with controls, while spine BMAD did not differ significantly between men with and those without GHRD (Table 1 and Fig. 1). At the femoral neck, BMAD was not significantly different in subjects with GHRD when compared with controls. Whole body BMC/height was used to correct partially for differences in body (and bone size). Men and women with GHRD had significantly lower values for this measurement than did controls.
Bone biopsy results from 10 subjects with GHRD and 11 adult controls are compared in Table 2. The biopsy specimen from one woman with GHRD consisted entirely of cortical bone and could not be analyzed. Of the 16 quantitative indices examined, only the node/end ratio showed significant differences between subjects with GHRD and controls. Subjects with GHRD had a significantly smaller mean node/end ratio (0.55 ± 0.46 vs. 1.19 ± 0.75, respectively; p < 0.05); a node/end ratio of >1.0 indicates better bone connectivity. The qualitative determination of trabecular connectivity also differed between the GHRD and controls. Poor trabecular connectivity was observed in biopsies from 5 of 10 subjects with GHRD but only 1 of 11 control subjects.
Table Table 2. Iliac Crest Histomorphometry
At the outset of this study, we hypothesized that subjects with GH resistance would have short stature, decreased lean body mass and osteopenia, findings described in patients with GH deficiency.(21) We observed profound short stature, decreased lean tissue, and deficits in BMC and BMD in adults with GHRD. Estimates of BMAD and bone histomorphometry, however, showed few differences between affected adults and controls. Thus, BMAD appeared to be preserved within the smaller bones of subjects with GHRD.
The influence of GH/IGF-I on bone growth and body composition is largely undisputed. Growth failure is observed in GH-deficient children,(5,6) and decreased lean body mass is characteristic of GH-deficient individuals of all ages.(21,22) The GH-resistant adults in our study had marked changes in body size and body composition, similar to those seen in severe, isolated GH deficiency. Their mean height was approximately 40 cm less than that of the controls. Subjects with GHRD also had significantly lower body weight, which reflected their smaller size and a selective reduction in lean body mass.
The contribution of the GH/IGF-I axis to the acquisition and maintenance of bone mineral remains more controversial. GH has been proposed to act directly or indirectly(2) through generation of IGF-I or IGFBP-3. Alternatively, GH may effect bone mineral by increasing muscle mass, which in turn increases the mechanical stress on the bone.(23) In animal models, GH and IGF stimulate proliferation and differentiated function of osteoblasts.(1–3) Administration of GH or IGF-I in normal adults and those with GH deficiency results in increased levels of bone formation and resorption markers.(10,11,22,24) It remains uncertain whether these effects translate to net gains in bone mass. Rosen and colleagues demonstrated that rats treated with human GH and IGF-I alone or in combination had greater bone length, area, and volume than untreated animals.(25) Although bone size was increased, volumetric bone density was lower in the treated animals.
Osteopenia has been described in children and adults with GH deficiency(1–7) and in patients with GH resistance.(16) Most of these studies have relied upon areal bone densities (BMD), however, without adjusting for reduced bone size.(5,6,16) DeBoer et al.(26) measured adults who had childhood onset GH deficiency and found that areal BMD of the spine was 18.3% lower in GH-deficient patients, whereas volumetric BMD was only 10.3% lower. Low BMD has also been reported in patients with adult-onset GH deficiency and normal stature, whose osteopenia cannot be an artifact of bone dimensions.(27) However, this condition is frequently associated with other pituitary hormone abnormalities which could contribute to low bone mineral.(27) Osteopenia has also been observed in adults with isolated deficits in IGF-I.(8,9,28,29) Histomorphometry in these patients has shown reductions in bone volume, cortical thickness, and osteoblastic surface.(29) In summary, considerable data suggest that abnormalities in the GH/IGF axis are associated with reduced bone mass, notwithstanding the use of areal bone mineral measurements for most of this work.
Our data indicated that childhood onset of GH resistance resulted in smaller bones, but the ratio of mineralized bone volume to marrow volume within these small bones was normal. Despite marked deficits in BMC and BMD at all sites, estimated BMAD was not significantly reduced compared with controls.(17) Women with GHRD actually had greater spine BMAD than controls. These results suggest that the apparent osteopenia, based upon BMC and BMD, is explained by the smaller bone dimensions of adults with GHRD.
Since BMAD is an estimate rather than a direct measurement of volumetric BMD, we considered the possibility that the model overcompensated for the extremely small bone size of subjects with GHRD. Bone biopsies were performed to examine the skeletal status more directly. Histomorphometry data confirmed preservation of volumetric bone density and cortical widths in adults with GHRD. The only difference observed in the biopsies was a smaller node/end ratio in affected adults, suggesting a reduction in bone connectivity. Poor trabecular connectivity was also observed qualitatively in 5 of 10 affected adults but in only 1 of 11 controls. These results are generally consistent with findings reported for adults with childhood onset GH deficiency.(30) Bone formation rates and bone volumes were normal, but trabecular bone volume was increased in a third of those patients.(30)
We found that most measurements of bone physiology were not different in controls and adults with GHRD. Markers of bone formation (osteocalcin and CICP) did not differ for the two groups. Bone formation rate, measured directly using tetracycline labeling of bone biopsy specimens, was also similar. Concentrations of urinary pyridinolines, but not CTX, were significantly greater in adults with GHRD. However, the normal bone mass and resorption surfaces by histomorphometry and the nomal BMAD suggest that bone resorption was not appreciably increased in the adults with GHRD.
We considered possible mechanisms to explain how volumetric bone density could be preserved despite GH resistance and reduced circulating IGFs and IGFBP-3. Although adults with GHRD lacked a normal GH/IGF-I axis, we hypothesized that local production of IGF-I, IGFBPs, or other growth factors in the bone microenvironment could have been sufficient to stimulate adequate bone mineral acquisition per unit of bone. Mechanical loading of the skeleton by the increased body mass (particularly in the women with GHRD) might provide a stimulus for bone mineral acquisition. Since bone biopsies were taken only from the iliac crest, we could not exclude the possibility of bone mineral deficits at other skeletal sites. This seems unlikely, however, since the molecular defect in GHRD affects receptors in all cells and would likely have an equivalent impact on bone mineral at all sites. Because of the small sample size, we could not exclude the possibility of a Type II error.
In conclusion, we found that bone size and lean body mass were markedly reduced in adults with GH resistance, consistent with the current view that the GH/IGF axis is important in the regulation of skeletal growth and body composition. BMC and areal BMD were also profoundly reduced in adults with GHRD, but these deficits could be attributed to small bone size. BMAD and histomorphometry results indicated that bone volume was largely preserved. This study exemplifies errors that may arise with the uncritical use of BMD as a measure of bone mineral status.(31,32) Our findings also raise questions about the contribution of GH/IGFs to the acquisition and maintenance of bone mineral.
We thank David Guido, Kyla Kent, Leah Holloway, Nancy Davila, Zully Moreno, Mary Ann Berrie, Linda Allen, Don Wanek, and Dr. Brad H. Pollock for their invaluable technical support. This work was supported by National Institutes of Health grants DK 45830 (A.R.) and DK 45226 (L.K.B.).