Effect of Long-Term Growth Hormone Treatment on Bone Mass and Bone Metabolism in Growth Hormone-Deficient Men



Long-term GH treatment in GH-deficient men resulted in a continuous increase in bone turnover as shown by histomorphometry. BMD continuously increased in all regions of interest, but more in the regions with predominantly cortical bone.

Introduction: Adults with growth hormone (GH) deficiency have reduced rates of bone turnover and subnormal BMD. GH treatment is effective in enhancing bone turnover as shown by biochemical markers and bone histomorphometric studies. However, it is uncertain whether long-term treatment will result in higher bone mass. In this study, we present BMD and histomorphometric data on 5 years of GH treatment in GH-deficient men.

Materials and Methods: Thirty-eight adult men with childhood onset GH deficiency (20-35 years) were included in the study. Twenty-six of these had multiple pituitary hormone deficiencies and were on stable conventional hormone replacement. BMC (total body) and BMD (lumbar spine and hip) were measured before and after 1, 2, 3, 4, and 5 years of treatment. BMD in various regions of the total body was calculated by computer software (head, trunk, arms, and legs). Transiliac bone biopsies were obtained before and after 1 and 5 years of GH treatment.

Results: Total body BMC increased 18% after 5 years of treatment. This increase was observed in all regions of interest: head, 13.7%; trunk, 27.8%; arms, 24.4%; legs, 13.8%. BMD also increased in all separately measured regions: lumbar spine, 9%; femoral neck, 11%; femoral trochanter, 16%. Lumbar spine area significantly increased (p = 0.0002). Histomorphometric data showed increased osteoid surface (p < 0.02), osteoid volume (p < 0.01), and activation frequency (p < 0.006), but trabecular bone volume did not increase significantly. Qualitative assessment of the cortical bone showed endosteal and periosteal bone formation.

Conclusions: In conclusion, GH considerably increases BMC after long-term treatment. The combination of BMD and histomorphometric data suggests that GH has a greater effect on cortical than on trabecular bone.


GROWTH HORMONE (GH) deficiency in adult patients is associated with reduced rates of bone turnover(1, 2) and subnormal BMD even after correction for body height. (3) This results in an increased fracture risk. (4) GH treatment is effective in enhancing bone turnover, as has been shown in several studies using biochemical markers. (2, 5–8) However, the enhanced osteoblastic activity did not always result in an increase of bone mass. The results of many short-term studies have been inconsistent. (7, 9, 10) A few years ago, several investigators published results on the effect of long-term GH treatment in adult GH-deficient patients on bone metabolism and bone mass. These studies also showed an increase of bone turnover and showed promising results on bone mass, (11–13) even when a low physiological dose was administered. (14) Results of 3-year GH treatment showed increases of BMD at the lumbar spine and the forearm as well as an increase of quantitative ultrasound parameters after an initial decrease. Our previously published data on BMD after 5-year treatment with GH showed an increase of 8–10% in the total body, lumbar spine, and femoral neck. (15)

The inconsistent results have been attributed to the different responses of cortical versus trabecular bone. Thus far, these studies used absorptiometric techniques in which a mixture of cortical and trabecular bone is tested. Histomorphometry enables the study of the different response of cortical and trabecular bone at the tissue level.

In an earlier study, we presented bone histomorphometric data after 1-year treatment with GH. (16) We concluded that GH increased the indices of bone turnover, but trabecular bone volume did not increase. Histomorphometric data of bone biopsies from GH-deficient patients after long-term GH treatment were not available until now.

In this study, we present biochemical, absorptiometric, and histomorphometric results concerning bone mass and bone turnover before and after 5 years of GH treatment in GH-deficient men.



Fifty adult men with childhood onset GH deficiency were initially included in the study as described previously. (17) Previous GH treatment for short stature had been discontinued for at least 1 year. GH deficiency was confirmed before the study by a serum IGF-1 concentration of at least 2 SD below the age-related normal mean, and a maximal GH response to 100 μg GH-releasing hormone or insulin-induced hypoglycemia of <7 μg/liter, in all subjects. Twelve patients were excluded from analysis because of incomplete follow-up. Two of these 12 patients showed poor compliance, 1 was diagnosed as having Crohn's disease, 1 died because of generalized convulsive seizures despite antiepileptic treatment, and 8 patients withdrew because of lack of motivation. The mean age of the remaining 38 patients at baseline was 28 ± 4 years (range, 20–35 years). Twelve of them had isolated GH deficiency. The 26 patients with multiple pituitary hormone deficiencies received stable, conventional hormone replacement: thyroxin (n = 24), testosterone undecanoate (n = 20), human gonadotropin (n = 2), hydrocortisone (n = 19), and antidiuretic hormone (n = 4). The cause of the pituitary failure was idiopathic or related to perinatal hypoxia in 33 patients and related to treatment for craniopharyngioma in 5 patients.

GH replacement therapy

The study was approved by the review board of the VU University Medical Center. All patients gave their informed consent before entering the study. The study protocol dictated that during the first 2 years, the patients were randomized to receive GH (Norditropin; Novo Nordisk, Gentofte, Denmark) in a dose of 0.33, 0.67, or 1 mg/m2/day, respectively. In the third year, the dose would be 0.67 mg/m2/day for all patients. This was believed to be the physiological dose at that time. Already in the first year, the dosages were reduced in one-third of the patients because of clinically relevant side effects. Thereafter, the GH dose was titrated, based on individual serum IGF-1 values. This resulted in average GH dosages of 0.55 ± 0.2, 0.63 ± 0.1, 0.51 ± 0.1, 0.46 ± 0.1, and 0.43 ± 0.1 mg/m2/day in each year follow-up.

Assessment of BMD

BMD was measured in 30 patients by DXA (Norland XR-26) as described previously. The long-term precision of the method (CV) was 2.4% for the lumbar spine and 2.3% for the femoral neck. (3) Consistency in measurements over 6 years was guaranteed by daily routine quality control procedures according to the manufacturer's guidelines. The Norland machine was replaced by a new device during the course of the study, and therefore, BMD could no longer be measured in some patients. For this reason, the measurement is missing in 7 patients after 4 and 5 years and in 13 patients after 5 years. BMD was calculated by the software program and is presented as the areal density (g/cm2), as the Z score calculated by the Norland software, or as the percentage increase from baseline. The total body BMC (TBBMC) was measured using the same equipment. The computer software calculated various regions of interest from the total body scan: head, trunk, left arm, right arm, and legs.


The serum IGF-1 concentration was measured by radioimmunoassay (RIA; Medgenics Diagnostics, Fleurus, Belgium), after acid-ethanol extraction of IGF-binding proteins. Intra- and interassay CVs were 6% and 10%, respectively.

Serum osteocalcin (OC) concentrations were measured by RIA (DiaSorin, Saluggia, Italy). The interassay CV was 13%. With this assay, the intact molecule and the 1–43 fragment is measured.

Procollagen 1C-terminal propeptide (P1CP) concentration was measured by RIA (Orion Diagnostica, Espoo, Finland). The interassay CV was 6%.

Serum and urinary calcium, creatinine, acid phosphatase, and alkaline phosphatase (ALP) were measured with routine laboratory methods. Urinary hydroxyproline was measured as described previously. (18)

Bone biopsies

Transiliac bone biopsies were obtained from 26 patients at baseline and after 1 year of GH treatment and from 11 patients after 5 years of GH treatment. Before the bone biopsy, patients received tetracycline double labeling with a dose of 250 mg four times a day on days 1, 2, 13, and 14 (2-10-2). Between 2 and 7 days later, the patients underwent a transiliac bone biopsy.

The bone biopsies were fixed overnight in 4% phosphate-buffered formaldehyde and transferred to alcohol 70%. After dehydration, the bone specimens were embedded without prior decalcification in methylmethacrylate supplemented with 20% plastoid-N and 0.13 g/ml percadox. Sections of 5 μm were prepared using a Jung K microtome. Bone mass indices, osteoid surface, and the eroded surface were measured in Goldner's trichrome-stained sections. TRACP staining was performed to visualize osteoclasts. Measurements were performed semiautomatically using a drawing tube, a digitizer, and image analysis software (Videoplan and Osteoplan, Kontron; Zeiss, Oberkochen, Germany). Nomenclature is used according to the ASBMR nomenclature committee. (19)


Results are expressed as mean ± SD. DXA values and serum IGF-1 levels are also expressed as Z scores. The primary outcome measures are the regional BMD values. Differences as a result of treatment were tested using ANOVA for repeated measurements, without the use of the intention-to-treat principle. For DXA measurements, the last value was carried forward to deal with missing values. This approach was only used one visit ahead, so if a patient missed two or more visits in a row, the data set was excluded from the analysis. Posthoc tests were performed in case of significance in the ANOVA.



For serum IGF-1, Z scores were calculated using a reference population of healthy controls between 20 and 40 years. (20) At baseline, IGF-1 Z scores were negative, but Z scores became positive after treatment at 1 and 5 years of follow-up (Table 1).

Table Table 1.. Biochemical Data on Bone Turnover in 38 GH-Deficient Patients Before and During 5 years of Treatment With GH
original image

Both the markers for bone formation and bone resorption increased during GH treatment (Table 1). Most markers showed a peak after 1 year of treatment, but serum OC reached a peak value after 3 years of treatment.


All BMC and BMD results are shown in Table 2. TBBMC increased 17.7% after 5 years of treatment compared with baseline. This increase was observed in all regions of interest: head, 13.7%; trunk, 27.8%; arms, 24.4%; legs, 13.8% (Fig. 1). The BMD also increased in all separately measured regions: lumbar spine, 9%; femoral neck, 11%; femoral trochanter, 16% (Fig. 2). The total area of L2-L4 increased significantly from 42.95 ± 3.16 to 44.23 ± 3.04 cm2 (p = 0.0002). The Z scores for BMD lumbar spine and BMD femoral neck did not significantly correlate with the Z score for IGF-1 at any time-point during the study.

Table Table 2.. BMC and BMD Results of GH-Deficient Patients Before and During GH Treatment
original image
Figure FIG. 1..

Percentage increase in total body BMC and in BMC of different regions of interest (head, trunk, arms, and legs) in GH-deficient patients before and during GH treatment. The regions of interest were determined by Norland computer software.20

Figure FIG. 2..

Percentage increase in BMD in different regions (lumbar spine, femoral neck, and femoral trochanter) in GH-deficient patients before and during GH treatment. The regions of interest were determined by Norland computer software.20

Significant correlations between the bone markers in serum or urine and GH dose or serum IGF-1 were not observed at any time-point. Correlations between GH dose and bone histomorphometric indices were not conducted because of the low number of patients. A correlation between TBBMC and GH dose was observed at t = 104 (p = 0.031).

Bone histomorphometry

From 11 patients, we were able to collect data on three time-points (baseline, after 1 year of treatment, and after 5 years of treatment). These patients are a reliable subset because age, serum IGF-1, and DXA variables (total body, lumbar spine, femoral neck, and femoral trochanter) were not statistically different from the whole group at baseline and at all measured time-points throughout the follow-up. Bone structure indices were not different at those time-points, as is shown in Table 3. Bone formation indices increased after GH treatment (Table 3; OS/BS, p < 0.02; OV/BV, p < 0.01). Although osteoid surface remained elevated after 5 years, osteoid volume was not significantly different from baseline at this time-point. Bone resorption indices did not significantly change during GH treatment (Table 3). Eroded surface (ES/BS) was high at all time-points compared with normal reference values. The number of osteoclasts increased from 0.68/mm2 to 1.10/mm2, but this was only borderline significant. The dynamic histomorphometric values showed significant increases of mineral apposition rate (MAR), mineralization lag time (Mlt), and activation frequency (AcF) (p < 0.006) after 5 years of treatment (Table 4).

Table Table 3.. Histomorphometric Data on Bone Mass and Bone Turnover in 11 GH-Deficient Patients Before and After 1 and 5 Years of Treatment With GH
original image
Table Table 4.. Dynamic Histomorphometric Results of 11 GH-Deficient Patients Before and During GH Treatment
original image

Qualitative assessment of the cortical bone showed a lamellar pattern at the periosteum in 3 of the 11 bone biopsies after 5 years of GH treatment (Fig. 3), suggesting endosteal and periosteal bone formation.

Figure FIG. 3..

Cortical bone of the iliac crest in a GH-deficient patient after 5 years of treatment with GH. This typical image of a parallel lamellar cortical bone tissue with bone apposition was observed in 3 of 11 treated patients. (A) Goldner-stained section through polarized light. (B) Tetracycline double label, showing active mineralization at the time of biopsy. Magnification, x200.20


This study compares histomorphometric data with regional BMD data after 5 years of GH treatment. This creates a unique opportunity to study the response of different bone types and regions.

The study shows that 5 years of GH treatment in adults with GH deficiency was, after the first year, associated with a continuous linear increase of BMD and BMC, which did not level off at 5 years for all measured regions of interest, including total body. Other long-term GH studies point in the same direction, indicating that GH could serve as an anabolic agent for bone. The lumbar spine area significantly increased after 5 years, indicating an expansion of the lumbar vertebrae.

Although BMD increased in all measured regions, the increases appeared to be larger in the regions with more cortical bone. TBBMC, comprised of 85% cortical bone, increased 14% in comparison with a 9% increase of lumbar spine BMD, which contains 50% trabecular bone. This suggests that GH may especially stimulate cortical bone formation, which is in contrast with a few previous studies concerning GH treatment in GH-deficient patients, which suggested larger increments in areas with predominantly trabecular bone. (9, 11) The treatment period in those studies was 12 months. Because the remodeling in trabecular bone is at a higher speed, the response to medication is expected earlier than in cortical bone, where a decrease was seen after the first year, as is visible in the TBBMC. It does not exclude larger increments in cortical BMD after treatment periods longer than 12 months.

After 5 years of GH treatment, trabecular bone volume in the iliac crest had not significantly changed, but a trend was observed toward a lower trabecular number and an increment of trabecular separation. Bone formation was higher after 1 and 5 years of treatment than at baseline. Although osteoid surface remained high after 5 years of treatment, the osteoid volume had decreased to baseline levels after 5 years, indicating thinning of osteoid seams. The wall thickness of trabecular osteons did not change during 5 years of GH treatment. Cortical thickness increased, but this was not significant. Qualitative assessment of the biopsies clearly showed thicker cortices with periosteal and endosteal bone apposition in three patients. These data correlate well with the finding of bone apposition in the lumbar vertebrae, as shown by the increased area with DXA. This bone expansion has consequences for the mechanical properties of the bones, because bone strength strongly improves as the diameter increases.

The difference in response to GH between trabecular and cortical bone has been reported in the literature. In rats, (21) IGF-1 increased longitudinal growth rate and inhibited trabecular bone formation in intact animals and was associated with a doubling of periosteal bone formation at the tibial diaphysis. Johansson et al. (13) studied the effect of 2 years of GH treatment in patients with adult onset GH deficiency. They showed a BMD increase in all regions, without differences between regions. However, the observed discrepancy between total body BMC and BMD was explained by an increase in bone area, suggesting bone apposition. In this study, the suggestion of a differential response of cortical versus trabecular bone is not only based on absorptiometric data but also on histomorphometric data, showing that the trabecular bone volume had not increased after 5 years of GH treatment

Overexpression of the IGF-1 gene in osteoblasts in transgenic mice results in increased trabecular bone volume without increased cortical thickness. (22) In the IGF-1-deficient mouse, cortical thickness of the proximal tibia was reduced by 17%. (23) However, trabecular bone volume of the tibia was increased, associated with increased connectivity, increased trabecular number, and decreased spacing. These differences were less evident in the lumbar vertebrae. Data from Jockenhovel et al., (24) concerning acromegalic patients, showed that cortical BMD of the distal radius was significantly higher in acromegalic men and women than in control men and women. In contrast, the trabecular BMD in this region did not differ between patients and controls.

A differential effect on trabecular and cortical bone was also observed during treatment with PTH. PTH or PTH(1-34) (teriparatide) is a promising anabolic agent for bone, with consistent improvements in BMD of the lumbar spine. The results on femoral BMD, however, vary across studies and clinical diseases. (25)

Almost all long-term GH treatment studies show a temporary decrease in BMD after 6 months to 1 year, which is most likely caused by an increase in remodeling space, as a result of increased turnover. The increased turnover is reflected in the biochemical bone markers (serum OC, P1CP, and ALP and urine hydroxyproline excretion) and the histomorphometric data. Our data on BMD also showed this temporary fall in bone mass after 1 year of treatment at various sites, including the total body, head, arms, and legs. The trunk, lumbar spine, and the femur did not show this temporary fall, which most likely should have occurred within the first treatment year.

In this trial, the study population was relatively young, ranging from 20 to 35 years of age. GH-deficient patients may have delayed timing of peak bone mass. (26) This peak, measured as lumbar spine BMD, occurs at a mean age of 19.8 years and mostly occurs 1 or 2 years after reaching final height. We do not believe that the patients of this trial were in the accrual phase of peak bone mass, suggested by the linear increases of BMD. GH treatment started at 1.4-15.5 years after discontinuation of GH therapy that they had received for short stature, and thus after reaching final height. Therefore, we assume that all patients had reached peak bone mass at the time the study started.

The increased bone formation and resorption is in agreement with the literature, describing a temporary increase in biochemical markers of bone turnover. Van de Weghe et al. (7) showed an OC peak at 1 year and a decrease to baseline values after the peak. However, in our study, serum OC increased during 3 years of treatment, after which it decreased, but it remained above baseline levels. This was confirmed by histomorphometric data on bone formation, because osteoid surface remained high throughout the 5 years.

We observed the peak rates for bone resorption markers at 1 year, although the levels remained elevated thereafter. This is also in agreement with the literature. However, these data on biochemical markers for bone resorption were not confirmed by the histomorphometric data on eroded surface, which increased after 1 year and stayed at the same level after 5 years of treatment.

Various studies on the treatment of GH-deficient patients with recombinant human GH have used different dosages. In this study, the starting dosage was 0.67 mg/m2/day on average in the first 3 years, because several side effects occurred at the highest dose (1 mg/m2/day). Only in the fourth and fifth year was the dose adjusted to obtain normal serum IGF-1 levels, resulting in average GH dosages of 0.55 ± 0.2, 0.63 ± 0.1, 0.51 ± 0.1, 0.46 ± 0.1, and 0.43 ± 0.1 mg/m2/day in each year of follow-up. The Consensus guidelines published in 1998(27) recommend that therapy should start with a low dose (0.15-0.3 mg) and should be increased gradually on the basis of clinical and biochemical response. The maintenance dose may vary considerably from person to person and will seldom exceed 1 mg. In this study, the initial dose was 1 mg in one group of patients, probably resulting in a temporary situation of GH excess, which could explain the histological results in bone after 1 year.

In conclusion, GH considerably increases BMC after long-term treatment. The combination of BMD and histomorphometric data suggest that GH especially stimulates cortical bone apposition. Histomorphometry did not show an increase in trabecular bone volume.