Pituitary secretion of growth hormone and circulating levels of insulin-like growth factor 1 decrease with aging [7]; these events are called somatopause. Growth hormone is both anabolic and lipolytic [8-10], and the action of growth hormone on peripheral tissues is mediated, in part, by circulating insulin-like growth factor 1 [11]. Growth hormone deficiency results in body composition changes that are similar to the changes seen with aging [1, 7]. Growth hormone replacement in patients with hypopituitarism and in older men reverses some of the body composition changes associated with both growth hormone deficiency and aging [12-14]. In a study by Rudman and colleagues [14] on the effects of growth hormone replacement therapy on body composition, the administration of growth hormone to 12 men 61 to 73 years of age for 6 months increased lean body mass by 9% and decreased adipose tissue mass by 15% (P < 0.005). Circulating insulin-like growth factor 1 levels increase similarly in young and old men after exogenous administration of either growth hormone or growth hormone-releasing hormone [7, 11]. Therefore, low insulin-like growth factor 1 levels reflect decreased growth hormone secretion rather than a loss of hepatic responsiveness to the hormone. On the basis of these data, it has been proposed that diminished secretion of growth hormone is responsible, in part, for the somatic changes of aging [14].

Growth hormone replacement has been shown to improve muscle volume, isometric strength, and exercise capacity in young adults with hypopituitarism [12, 13]. In one study of growth hormone-deficient young adults treated with growth hormone [12], thigh muscle volume increased by 6% (P < 0.01), and isometric strength increased by 8% (P = 0.08) compared with controls. Increases in isometric strength were associated with increases in muscle volume (r = 0.75; P < 0.001). Exercise capacity also increased by 12% (P < 0.05). Such data on the effects of growth hormone replacement therapy on strength or function are not available for older persons.

To test the hypothesis that the muscle weakness and functional decline associated with aging are in part due to decreased growth hormone secretion, we conducted a 6-month randomized, controlled, double-blind trial comparing recombinant human growth hormone therapy with placebo in 56 elderly men whose baseline insulin-like growth factor 1 levels were less than the tenth percentile found in younger healthy adults.

Methods

Top Methods Results Discussion Author & Article Info References

Participants

We recruited healthy, ambulatory, community-dwelling men older than 69 years of age. Because growth hormone secretion is pulsatile and diurnal, a more convenient measure of growth hormone secretion is insulin-like growth factor 1, which has constant and nonpulsatile plasma levels [1, 7, 15, 16]. Inclusion criteria included two fasting morning plasma insulin-like growth factor 1 levels (assayed after plasma was extracted with acid ethanol by the Nichols Institute [17]) less than 161 ng/mL that were obtained at least 2 weeks apart, a fasting blood glucose level less than 8.4 mmol/L (140 mg/dL), stable body weight between 80% and 120% of ideal body weight [18] during the previous year, and a normal complete blood count at baseline. Participants had to be able to follow the study instructions and perform the strength measurements. Exclusion criteria included diseases that might contraindicate the use of growth hormone (such as carcinoma within the previous 5 years, diabetes mellitus, and the carpal tunnel syndrome), uncontrolled hypertension (blood pressure more than 180/100 mm Hg), oral corticosteroid use within the previous year, or a recent change in ambulatory status (for example, as the result of a stroke).

Design

After the participants gave informed written consent, the following baseline measurements were obtained: questionnaires (demographic characteristics and medical conditions); fasting blood chemistry (levels of electrolytes, urea nitrogen, creatinine, glucose, calcium, phosphate, and albumin); complete blood count; serum androgen levels (total and free testosterone and dehydroepiandrosterone); fasting lipoprotein levels (total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and triglycerides) and thyroid-stimulating hormone and free thyroxine levels; 24-hour urinary calcium excretion; and creatinine clearance. At baseline, we also measured the body composition, strength, and functional ability of all participants. We then used a computer-generated randomization Table to assign each participant to receive either growth hormone or placebo. Participants, investigators who had contact with the participants, and laboratory personnel were blinded to treatment status.

Treatment consisted of recombinant human growth hormone (somatropin [Nutropin, Genentech]), 0.03 mg/kg of body weight, or an equivalent placebo volume (supplied in identical vials by Genentech) injected subcutaneously three times per week in the morning for 6 months. We evaluated the participants after 2, 4, 8, 12, 18, and 24 weeks of treatment to determine whether side effects developed and to ensure compliance. Participants were examined and specifically asked about the presence of the following side effects: malaise, fatigue, edema, arthralgias, arthritis, the carpal tunnel syndrome, muscle discomfort, dyspnea, depression, polydipsia, polyuria, nocturia, and tender or enlarged breasts. At each visit, we measured fasting morning plasma insulin-like growth factor 1 levels 48 hours after the last treatment dose and repeated most of the baseline blood and urinary tests.

We adjusted the dose of recombinant human growth hormone to maintain serum insulin-like growth factor 1 levels between 190 and 350 ng/mL (the 25th to 75th percentiles of a young adult population) 48 hours after a dose. To preserve blinding, each growth hormone recipient was paired with a placebo recipient. An unblinded physician who had not had contact with the participants reviewed insulin-like growth factor 1 levels during the trial. If the levels were less than 190 ng/mL or greater than 350 ng/mL, the unblinded physician instructed study personnel to make identical dose adjustments for both participants in the pair. The dose was also decreased in any participant who reported troublesome side effects that were potentially attributable to growth hormone.

We repeated all study outcome measurements after 6 months of treatment. We also measured fasting plasma renin activity, cortisol and aldosterone levels, and 24-hour urine aldosterone excretion.

The investigational review boards of the University of California, San Francisco, and the Department of Veterans Affairs Medical Center, San Francisco, approved the study protocol.

Measurements

Body Composition

We used dual-energy x-ray absorptiometry (Lunar DPX-Plus, Madison, Wisconsin) total-body composition scans to measure lean tissue mass and fat mass, and we used anterior-posterior lumbar scans to measure bone mineral content. We measured skin thickness using ultrasonography with a 10-MHz transducer (Diasonics, Milpitas, California) that was positioned transversely 5 cm below the umbilicus. Using the same transducer, we made two ultrasonographic measurements on each participant. The average of the two measurements was used for analysis. The study radiologist, who was blinded to the participants' treatment status, interpreted all test results.

Muscle Strength

The study's physical therapist measured the muscle strength of the participants' knee flexor and extensor muscles using an isokinetic dynamometer (Cybex 340, Lumex, Bay Shore, New York). Using a standard protocol, we measured peak torque of knee flexion and extension at joint speeds of 90, 120, and 180 degrees per second. We measured hand grip strength according to a standardized protocol using a grip dynamometer (Smedley Grip Dynamometer, JA Preston, Jackson, Mississippi). The greater of the two measurements was used for analysis.

Systemic Endurance

To assess systemic endurance, we measured maximal oxygen consumption (VO₂ max) during exercise cycle testing. Upright exercise testing was done using an electronically braked cycle ergometer (Erich Jaeger, Rockford, Illinois). Exercise was initially done unloaded and was then increased by 25 watts every 2 minutes. Participants continued to exercise until they were exhausted or unable to maintain critical pedal frequency (greater than 50 revolutions per minute). During exercise cycle testing, we continuously measured respiratory gas exchange to determine VO₂ max (Ergopneumotest, Erich Jaeger).

Physical Performance

The study's physical therapist used the Physical Performance Test [19] to assess the participants' performance on nine physical functions. Participants were asked to write a prescribed sentence, transfer kidney beans using a teaspoon, place a heavy book on a shelf, remove a jacket, pick up a penny from the floor, turn 360 degrees, walk a 50-foot walking test course, and climb stairs to determine speed and the number of flights climbed before the development of fatigue. The test score is based on the time required to complete each task. Higher scores (the best performance is a maximum score of 36) are associated with the shortest time required to complete a task and with better functional status.

Cognitive Function and Mood

Each participant completed the Trails B Test, the Mini-Mental State Examination, the Digit Symbol Substitution Test, and the Geriatric Depression Scale [20-23]. The Trails B Test (from the Halstead-Reitan Neuropsychological Test Battery) [20] assesses visual and motor tracking and attention. The participant connects a line on a page between sequential numbers and letters (for example, 1-A, 2-B). Rating is based on the time required to complete the task. Higher scores (maximum, 300) are associated with cognitive impairment. The Mini-Mental State Examination [21] assesses orientation, attention, calculation, language, and memory. Lower scores (range, 0 to 30) are associated with cognitive impairment. The Digit Symbol Substitution Test [22] requires the participant to substitute specific symbols for digits; lower scores are associated with cognitive impairment (range, 0 to 90). The Geriatric Depression Scale [23] consists of 15 questions about mood; scores greater than 5 suggest depression (range, 0 to 15).

Statistical Analysis

We analyzed the statistical significance of differences between the treatment groups using t-tests for continuous variables. For continuous variables with distributions that were not normal, we confirmed the t-test results using the Mann-Whitney-Wilcoxon unpaired test, a nonparametric analog of the t-test. We used the Fisher exact test in the 2 x 2 tables.

We compared absolute changes in the growth hormone and placebo groups (values at 6 months – the baseline values) for all variables except body composition and strength. For the latter variables, we compared percentage changes ([value at 6 months – the baseline value]/[baseline x 100]).

We used Pearson correlation coefficients (r) to describe the association between changes in strength and changes in body composition. First, we did two sets of analyses to assess the possibility of an interaction between treatment and age, baseline insulin-like growth factor 1, or baseline testosterone levels. We stratified the sample into tertiles of age (70 to 74 years, 75 to 79 years, and more than equal 80 years), baseline insulin-like growth factor 1 levels, and testosterone levels. We then examined the effect of treatment on strength within these strata for each of the three variables. Second, to assess the statistical significance of this interaction, we did a linear regression analysis using strength as the outcome. We included treatment, the potential interacting variable (age, baseline insulin-like growth factor 1, or baseline testosterone level), and the interaction between these two [24]. We then tested the significance of the interaction term in the model.

All analyses were done using the Statistical Analysis System (SAS, Inc., Cary, North Carolina). The study was planned to have a power of at least 0.8 to detect a 15% difference in the primary outcome variable, change in knee extension strength.

Results

Top Methods Results Discussion Author & Article Info References

Fifty-two of the 56 men enrolled in the trial completed the protocol. Twenty-six men were assigned to receive growth hormone, and 26 were assigned to receive placebo (Table 1). The mean age of the participants was 75.0 years (range, 70 to 85 years). At baseline, the participants' functional ability was intact, as determined by their high scores on the Physical Performance Test. Most of the multiple baseline measurements did not differ between groups (P > 0.05), but dehydroepiandrosterone levels, triglyceride levels, and creatinine clearance did differ somewhat (Table 1). Two participants assigned to receive growth hormone and 2 participants assigned to receive placebo did not complete the study. One growth hormone recipient had coronary artery bypass graft surgery during the study, and the second was concerned that his preexisting back pain might worsen. One of the placebo recipients had bowel resection, and the other had recurrent gastrointestinal bleeding. Table 2 shows that insulin-like growth factor 1 levels in the growth hormone group markedly increased after 6 months of therapy compared with the placebo group (P < 0.001).

Effect of Growth Hormone on Blood and Urine Variables

The two groups had similar mean changes in serum and urine outcome variables from baseline to 6 months (P > 0.20) (Table 2). A trend toward decreased triglyceride (P = 0.10) and free thyroxine (P = 0.06) levels was seen in the growth hormone group.

Plasma renin activity, serum and urine aldosterone levels, and plasma cortisol levels did not differ between the two groups after 6 months of treatment (P > 0.20) (Table 3).

Effect of Growth Hormone on Body Composition

Six months of growth hormone treatment caused changes in body composition but not body weight (Table 4). Lean mass increased by an average of 4.3% in the growth hormone group and decreased by an average of 0.1% in the placebo group, a difference of 4.4 percentage points (CI, 2.1 to 6.8 percentage points; P < 0.001). Fat mass decreased by averages of 13.1% in the growth hormone group and 0.3% in the placebo group, a difference of 12.8 percentage points (CI, 8.6 to 17.0 percentage points; P < 0.001). Bone mineral content increased by an average of 0.9% in the growth hormone group and decreased by an average 0.1% in the placebo group, a difference of 1.0 percentage points (CI, 0.02 to 2.0 percentage points; P = 0.05). We noted a trend toward a 12.3% greater increase in skin thickness in the growth hormone group than in the placebo group (CI, – 1.0% to 25.6%; P = 0.09).

Effect of Growth Hormone on Muscle Strength, Systemic Endurance, and Physical Performance Test Outcomes

At 6 months, the changes in muscle strength were similar in the two groups (Table 4). Knee extension strength at a joint speed of 120 degrees per second increased by averages of 3.8% in the growth hormone group and 1.3% in the placebo group, a difference of 2.5 percentage points (CI, – 11.1 to 16.1 percentage points; P > 0.2). Knee flexion strength at a joint speed of 120 degrees per second increased by averages of 10% in the growth hormone group and 8.2% in the placebo group, a difference of 1.8 percentage points (CI, – 9.99 to 13.5 percentage points; P > 0.2). No significant differences were seen between the changes in hand grip and VO₂ max in the two treatment groups at the end of the 6 months (Table 3). The change in lean tissue mass and knee muscle strength at joint speeds of 120 degrees per second in extension (r equals – 0.005; P > 0.2) or flexion (r = 0.03; P > 0.2) did not significantly differ. To determine whether treatment with growth hormone improved muscle strength in any subgroup, we compared changes in the outcomes by age (70 to 74 years, 75 to 79 years, or more than equals 80 years), baseline plasma levels of insulin-like growth factor 1 (by tertiles), and baseline serum levels of total testosterone (by tertiles). A significant difference was seen in the changes in muscle strength between the two treatment groups in any subgroup in only 1 of the 24 comparisons. A linear regression analysis confirmed that there was no significant interaction between any of these three baseline variables and the treatment groups. Changes in the results of the Physical Performance Test did not differ between groups.

Effect of Growth Hormone on Cognitive Function or Mood

Absolute changes in cognitive function and mood variables after 6 months of treatment are shown in Table 5. The mean Trails B score improved by 8.5 seconds in the growth hormone group and deteriorated by 5.0 seconds in the placebo group, a difference of 13.5 seconds (CI, 3.1 seconds to 23.9 seconds; P = 0.01) However, the growth hormone group's score on the Mini-Mental Status Examination deteriorated by 0.4, whereas the placebo group's score improved by 0.2, a difference of 0.6 (P = 0.11). Results of the Digit Symbol Substitution Test did not change in either group; the difference in scores between the two treatment groups was 0.1 (CI, – 2.2 to 2.4; P > 0.2). Performance on the Geriatric Depression Scale also remained unchanged; the difference in scores between the two treatment groups was – 0.3 (CI, – 1.8 to 1.2; P > 0.2).

Side Effects

During the 6-month trial, 77% of the growth hormone recipients and 46% of the placebo recipients reported one or more of the side effects about which they had been asked (P < 0.001) (Table 6). Twenty-six percent of the growth hormone recipients and none of the placebo recipients required a dose decrease because of side effects. The frequency of side effects in the growth hormone group was similar in the three age subgroups (70 to 74 years, 75 to 79 years, and more than equals 80 years) (P > 0.20). In most growth hormone recipients, the side effects developed within the first month of treatment, but about one quarter of the side effects first appeared midway through the study. The most common side effects were pitting lower-extremity edema and diffuse arthralgias. In the growth hormone recipients who noted arthralgias, both small and large joints were affected, but no clinical evidence of inflammation developed. Palpable gynecomastia did not occur, but two growth hormone recipients and two placebo recipients reported tender breasts. The side effects disappeared or remitted markedly within 2 weeks after the growth hormone dose was decreased by 25% to 50%. No participants developed the carpal tunnel syndrome during the study.

Other side effects included angina or myocardial infarction in one growth hormone recipient and three placebo recipients. In one growth hormone recipient, the growth hormone dose was reduced because of hypertension.

Dose Adjustments To Maintain Target Insulin-like Growth Factor 1 Levels

Nineteen asymptomatic treatment pairs had 33 dosage adjustments to maintain the insulin-like growth factor 1 level within the protocol target range. The dosage was decreased in one third of the adjustments and increased in two thirds. The need for dosage adjustments was fairly evenly distributed throughout the 6 months of the trial.

Discussion

Top Methods Results Discussion Author & Article Info References

In an unblinded trial, Rudman and colleagues [14] showed that in older men, treatment with growth hormone increased lean body mass by 9% and decreased fat mass by 15%. This finding led to the hypothesis that treatment with growth hormone might also improve strength, endurance, and physical ability in the elderly. Our study of the effects of growth hormone replacement in healthy older men with low insulin-like growth factor 1 levels confirmed a modest beneficial effect of growth hormone on body composition: Lean tissue mass increased by 4% and fat mass decreased by 12% in the growth hormone group compared with the placebo group. Despite these improvements in body composition, treatment with growth hormone was not associated with a statistically significant improvement in muscle strength, systemic endurance, or cognitive function. This finding is consistent with cross-sectional data showing that insulin-like growth factor 1 is not associated with functional ability independent of age [25].

Interpretation of our study results includes four possibilities. First, despite improvement in body composition, treatment with growth hormone may not improve functional ability in healthy older men with low insulin-like growth factor 1 levels. Second, the physiologic growth hormone dose we used may have been too low to improve function. We cannot use our trial results to address this issue, but we believe that, given the frequency of side effects in the elderly, higher growth hormone doses will be associated with unacceptable side effects. Third, the duration of therapy may have been too short. Growth hormone therapy might prevent loss of strength and endurance without dramatically reversing this age-related process. In this case, therapy would need to last much longer to show a clinically and statistically significant benefit. Fourth, free thyroxine levels were lower in the growth hormone group than in the control group (P = 0.06). However, most of the difference between the two groups was due to an increase in free thyroxine levels in the controls. In addition, thyroid-stimulating hormone levels, a reliable measure of thyroid function, did not differ between the two groups.

The statistically insignificant 2% increase in knee muscle strength that resulted from growth hormone treatment should be compared with known improvements in muscle strength of 24% to 97% in the elderly who begin exercise [26, 27]. Exercise in the elderly has been found to produce substantial muscle hypertrophy and a 60% increase in muscle mass [27] compared with the 4% increase in lean tissue mass with growth hormone therapy that we found. The pharmacy cost for a year's supply of recombinant human growth hormone is approximately $12 000. The long-term consequences of growth hormone therapy remain unknown.

In the growth hormone group, we found improved Trails B scores, a trend toward worsened outcome in the Mini-Mental State Examination, and no difference in scores on the Digit Symbol Substitution Test. The multiple statistical tests indicate that the improved Trails B score was probably caused by chance.

No participants dropped out of the study because of side effects, but troublesome side effects were common and required a dose decrease in one quarter of the growth hormone recipients. Side effects generally resolved after the doses were reduced, and we cannot determine whether such side effects would have been self-limited. Although it is scientifically interesting to determine whether higher doses of growth hormone or longer duration of therapy might be useful, we believe that the side effects limit the clinical usefulness of growth hormone therapy. What also remains unknown is whether long-term administration of physiologic growth hormone has a protective effect or whether any salutary effect reaches a plateau, as is seen with resistance exercise training in elderly persons who do or do not receive growth hormone supplementation [28]. The diabetogenic effect of long-term growth hormone therapy remains a concern. Of note, growth hormone therapy did not alter blood glucose levels in the elderly men in our study.

Growth hormone may increase extracellular fluid volume, an effect that could confound the measurement of lean tissue mass [29]. However, both growth hormone and insulin-like growth factor 1 do increase lean tissue [30]. Although we used growth hormone doses that were lower than those in the above trials, edema developed in 65% of the participants. More subtle fluid abnormalities could have been present in the other participants. The observation that renin and aldosterone levels were not significantly lower in men given growth hormone, however, suggests that fluid was not primarily retained in the effective extracellular fluid compartment. A similar lack of effect of growth hormone on renin and aldosterone was noted in patients with human immunodeficiency virus infection who were given much larger doses of growth hormone [31].

Participants responded to growth hormone similarly regardless of their baseline testosterone level. Consistent with previous studies in younger adults, growth hormone administration did not affect serum androgen levels [32].

A limitation of our study is that we only used insulin-like growth factor 1 levels to identify growth hormone deficiency. Insulin-like growth factor 1 levels in persons in the seventh decade of life are approximately one half those found in persons in the third decade [33]. Insulin-like growth factor 1 is highly correlated with spontaneous 24-hour growth hormone secretion in young adults, but the correlation is not as strong in the elderly [7, 11]. Other factors independent of growth hormone secretion, such as diminished physical activity and adiposity, may contribute to the age-related decline in insulin-like growth factor 1 levels [34, 35]. We do not believe that these factors were important in our study participants, who were healthy, active older men within 20% of their ideal body weight. Some investigators [11, 36] have proposed that growth hormone stimulatory testing or circadian monitoring of growth hormone most reliably identifies growth hormone deficiency. However, other investigators [14, 16] have also used insulin-like growth factor 1 levels to identify patients with growth hormone deficiency and to monitor therapy. Because our participants had almost perfect baseline scores on the Physical Performance Test, our results may not be generalizable to more functionally impaired men or to women. The optimum dose and dosage interval of growth hormone for elderly persons is unclear. In studies of younger adults with growth hormone deficiency, physical performance improved after as little as 6 months of treatment with daily growth hormone administration; these improvements persisted through 3 years of daily growth hormone administration [37, 38]. We chose to mimic physiologic levels so that side effects were minimized and to test the hypothesis that decreased growth hormone secretion in part causes the weakness and functional decline associated with aging.

The importance of increased lean tissue mass, decreased fat mass, and increased bone mineral content induced by growth hormone treatment remains unknown.

In our study, the difference in muscle strength between the growth hormone and the placebo groups was not statistically significant. Compared with the placebo group, the knee muscle extensor strength of the growth hormone group improved by 2.5 percentage points, knee muscle flexor strength improved by 1.8 percentage points, and hand grip strength decreased by 5.3 percentage points. In addition, no correlation was seen between changes in lean tissue mass and muscle strength. At the study onset, we assumed that a 15% greater increase in strength from growth hormone treatment would be clinically significant. The width of the CIs does not preclude a small improvement in strength caused by growth hormone therapy (CI, 1% to 16%). This modest gain must be balanced against the side effects and cost of the drug. Moreover, exercise alone in the elderly can achieve much greater increases in muscle strength [26, 27]. Whether higher-dose therapy would be more effective or whether longer-term therapy would prevent age-related decline in functional outcome is unknown.

Our data do not support the hypothesis that an age-related decline in growth hormone secretion is responsible for the functional decline of aging. The lack of demonstrable efficacy in our study sample, coupled with the frequent side effects and substantial expense, leads us to conclude that growth hormone should not be used to preserve or improve functional ability in healthy, functionally intact older men. Controlled studies in women and functionally impaired elderly persons would enhance our understanding of the clinical use of growth hormone.

Presented in part at the Society of General Internal Medicine national meeting, 6 May 1995.

Dr. Black: Box 0886, 74 New Montgomery #600, San Francisco, CA 94105.

Dr. Schambelan: San Francisco General Hospital, Building 100, Room 286, 1001 Potrero Avenue, San Francisco, CA 94110.