Growth hormone secretion among HIV+ patients: effects of gender, race and fat distribution
AIDS: Volume 20(6) 4 April 2006 p 855-862
Koutkia, Polyxenia; Eaton, Kristinaa; You, Sung Mina; Breu, Jeffb; Grinspoon, Stevena
From the aMassachusetts General Hospital Program in Nutritional Metabolism and Neuroendocrine Unit, Harvard Medical School, Boston, Massachusetts, USA
bGeneral Clinical Research Center, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA.
"....In summary, our data suggest that relative GH deficiency is very common among HIV-infected men with fat redistribution, even those with primary limb fat loss and an increased WHR by this mechanism. Our data further indicate an important effect of gender on GH responses to GHRH + arginine, such that fewer HIV-infected women would be expected to fail this test, because of higher GH levels. Race is also shown to affect GH responses, with differential effects among male and female HIV-infected subjects. Studies are now underway to determine the effects of physiologic GH augmentation among HIV-infected patients, and whether such a strategy will reduce cardiovascular risk in this population, as has been suggested among non HIV-infected patients with relative truncal adiposity."
Objective: To determine the effects of gender, race, and fat redistribution on growth hormone (GH) secretory patterns in HIV-infected patients.
Design: We investigated GH responses to growth hormone releasing hormone (GHRH) + arginine stimulation testing in HIV-infected subjects with fat redistribution, comparing HIV-infected males (n = 139) and females (n = 25) to non HIV-infected male (n = 25) and female (n = 26) control subjects similar in age, body mass index and race.
Methods: A standard GHRH + arginine stimulation test [GHRH 1 mg/kg and arginine 0.5 g/kg (maximum dose 30 g)] was performed, and fat redistribution was assessed by anthropometry.
HIV-infected women had significantly higher peak GH in response to GHRH + arginine (36.4 ± 7.3 versus 18.9 ± 2.0 ng/ml; P = 0.003) and GH area under curve (AUC) (2679 ± 593 versus 1284 ± 133 (mg-min)/dl, P < 0.001) compared to HIV-infected men.
Among men, a cutoff of 7.5 ng/ml for peak GH response on the GHRH + arginine test achieved good specificity and sensitivity and optimally separated the HIV and control groups (e.g., the failure rates were 37% versus 8%; P = 0.004, respectively).
Among women, no specific cutoff could be determined to separate the HIV-infected and control subjects.
Non-Caucasians demonstrated a higher GH AUC response compared to Caucasians, among the HIV-infected male subjects.
In stepwise regression modeling waist-to-hip ratio was most significantly related to peak GH in response to GHRH + arginine in HIV-infected men.
Conclusions: HIV-infected men with fat redistribution have significantly reduced GH peak responses and increased failure rates to standardized GH stimulation testing in comparison to healthy male control subjects and to HIV-infected women of similar age and body mass index. GH secretion is related to gender and race in HIV-infected patients.
Growth hormone (GH) and growth hormone secretagogues have been used to treat patients with HIV lipodystrophy and fat redistribution [1-4] but little is known about GH secretory patterns in this population. Initial studies suggested that GH is reduced in association with truncal adiposity in HIV-infected patients . Although absolute increases in visceral fat have not been shown in all cross-sectional studies , a number of longitudinal studies demonstrate relative truncal adiposity developing in association with HAART [7,8], which may predispose HIV-infected patients to GH deficiency. Furthermore, gender is known to have significant effects on GH secretion in non HIV-infected patients [9-13], but the effects of gender on GH secretion have not been investigated among HIV-infected patients. In this study we examined gender and racial differences in GH responses to standard stimulation testing in HIV-infected patients and matched control subjects. Peak GH responses were significantly greater among HIV-infected women than men. The specific cutoff to define GH deficiency in response to the growth hormone releasing hormone (GHRH) + arginine test is controversial, and ranges from 1.5 to 9.0 ng/ml in different studies, depending on desired specificity [14,15]. We compared failure rates to the GHRH + arginine test in HIV-infected patients versus control subjects across a range of cutoffs to determine the optimal separation points for male and female subjects. To our knowledge, the current study is the largest to date to evaluate the prevalence of GH deficiency in the HIV population (n = 164), using standardized stimulation testing and controlling for age, body mass index (BMI) and waist-to-hip ratio (WHR).
Preliminary studies in HIV-infected men suggest reduced overnight GH secretion  and an increased failure rate to standardized GH stimulation testing  in association with increased visceral adiposity. Recent studies suggest that physiologic GH replacement in non HIV-infected GH deficient patients may improve cardiovascular indices, including lipid levels, truncal adiposity and inflammatory indices . Identification and treatment of HIV-infected patients with relative GH deficiency, using strategies to increase GH within the physiologic range, may therefore improve cardiovascular risk indices. In contrast, strategies using pharmacologic dosing of GH may result in reduced visceral fat, but worsened hyperglycemia and insulin resistance, without a clear benefit on overall cardiovascular risk [1,2,20].
In the present study we investigated the following questions: (i) is there a gender or race effect on GH secretion in the HIV population that may influence targeted treatment strategies with GH or GHRH; (ii) is there an optimal cutoff on standardized GH testing to identify GH deficiency among male and female HIV-infected patients with fat distribution; (iii) how does fat distribution itself relate to GH secretion. Characterization of GH deficient subjects may be important to avoid side effects associated with use of pharmacological GH administration in HIV-infected patients with normal GH secretion.
Our data demonstrate that HIV-infected women demonstrate a much higher peak and AUC GH response, to GHRH + arginine compared to HIV-infected men. As a result, a much lower percentage of HIV-infected women with fat redistribution fail the GHRH + arginine test across a range of cutoffs (Fig. 2). Among non HIV-infected patients, higher GH levels in women have been ascribed to an effect of estrogen, which may reduce IGF-I and thereby increase GH via a reduction in feedback inhibition [21,22]. In our patients, a lower IGF-I was seen in women, in agreement with this potential mechanism. Veldhuis et al. reported that estrogen augments GH secretory burst mass by amplifying GHRH secretion and by attenuating feedback by somatostatin . Other explanations for the lower IGF-I among women include an effect of higher sex hormone binding globulin in women which has been reported to correlate inversely with GH/IGF-I axis in GH deficient subjects  and in elderly healthy patients . Our population of women was relatively young, and further reductions in GH might be seen in an older, postmenopausal population .
Our data are in accordance with data from other populations suggesting a significant effect of gender on GH regulation. In this regard, increased endogenous testosterone production in males may affect the GH-IGF-I axis, and contribute to gender differences. Our data suggest that a smaller population of HIV-infected women than similarly aged young HIV-infected men would fail a standardized GHRH + arginine test. Further studies with larger numbers of women will be necessary to investigate GH responses in HIV-infected women and the need for GH augmentation in this population.
In this study, we demonstrate differences by race in peak GH and GH AUC response to GHRH + arginine, with higher responses in non-Caucasian than Caucasian male HIV-infected subjects. In contrast lower GH responses were seen among non-Caucasian compared to Caucasian female HIV-infected subjects. Our data in male HIV-subjects are in accord with data from Wright et al. suggesting increased overall GH secretion in black than white men . The mechanism of this effect is thought to relate to the increased circulating estrogen among non-Caucasian men. Detailed studies on the hormonal mechanisms of racial differences in GH secretion among the HIV-infected population were beyond the scope of this study. Further investigation of this important question is needed as therapeutic strategies involving GH and GH secretagogues are being considered for the racially diverse population of patients with HIV disease.
In the current study, we assessed GH deficiency using the GHRH + arginine test in 164 HIV infected patients with fat redistribution. Numerous tests exist to determine GH deficiency. The gold standard test to determine GH deficiency is the insulin tolerance test (ITT) (in which insulin is given intravenously, and hypoglycemia provokes a GH response), but this test is associated with risk, and may not be appropriate for the HIV-infected population. The GHRH + arginine test has been shown to be a reasonable alternative test to determine GH deficiency [15,28]. The GHRH + arginine tests works by simultaneously stimulating GH directly with GHRH and reducing somatostatin, an inhibitory hormone to GH, via arginine.
What is an appropriate cutoff to identify GH deficiency among HIV-infected patients with fat redistribution? The optimal cutoff to characterize GH deficiency is controversial in the literature for normal controls or subjects with hypopituitarism. European groups have proposed a higher cutoff at 9 ng/ml, using the GHRH + arginine test, below which severe GH deficiency has been documented in patients with hypopituitarism . Aimaretti et al. showed that 92.5% of adults with hypopituitarism have peak GH lower than 9.0 ng/ml . Biller et al. studied the specificity and sensitivity of six different stimulation tests in 33 subjects with hypopituitarism in comparison to control subjects. A cutoff of 4.1 mg/l on the GHRH + arginine test provided 96% sensitivity and 92% specificity, comparing well to the ITT, but lower cutoffs provided increased specificity .
We assessed failure rates across a range of established cutoffs for the GHRH + arginine test. The cutoff of 7.5 ng/ml provided the optimal separation between the HIV and control groups, while maintaining good specificity, i.e., < 10% of the controls failed. Using this cutoff approximately one-third with fat redistribution fail the GHRH + arginine test, and thus can be considered at least relatively GH deficient.
In contrast to patients with true GH deficiency, e.g. due to a pituitary tumor or radiation, the patients in this study were presumed to have otherwise normal pituitary function, based on a careful history and screening process. The reduced response to GHRH + arginine was most strongly related to increased WHR ratio in regression modeling. Thus although the patients did not have increased waist circumference per se, but rather a reduced hip circumference due to predominant lipoatrophy, the increased WHR or relative fat redistribution was nonetheless associated with reduced GH secretion. This is an important point not made in prior studies, in which reduced GH responses to stimulation paradigms, were related to an absolute excess of visceral fat [5,17]. Among non HIV-infected patients, reduced GH secretion is also associated with central adiposity , and reverses with weight loss [30,31]. It remains to be seen whether reduced GH secretion would normalize with improvement in fat redistribution among HIV-infected patients, but this will be an important area of investigation in the future.
In contrast to the peak GH level on a GH stimulation test, the IGF-I level is a poor test to determine GH deficiency. Among non HIV-infected patients, there is a large overlap in IGF-I levels between GH deficient and non-GH deficient and IGF-I is a low sensitivity test in the diagnosis of GH deficiency in adults . The poor sensitivity of IGF-I to diagnose GH deficiency may result from increased IGF-I production for a given GH level in GH deficiency, the opposite phenomenon to that of GH resistance seen in malnutrition or starvation.
In summary, our data suggest that relative GH deficiency is very common among HIV-infected men with fat redistribution, even those with primary limb fat loss and an increased WHR by this mechanism. Our data further indicate an important effect of gender on GH responses to GHRH + arginine, such that fewer HIV-infected women would be expected to fail this test, because of higher GH levels. Race is also shown to affect GH responses, with differential effects among male and female HIV-infected subjects. Studies are now underway to determine the effects of physiologic GH augmentation among HIV-infected patients, and whether such a strategy will reduce cardiovascular risk in this population, as has been suggested among non HIV-infected patients with relative truncal adiposity.
Clinical and demographic characteristics and comparison within each gender between HIV-infected and control subjects are shown in Table 1 and Table 2. HIV-infected men and women were similar in age, BMI, and race compared to respective gender-matched healthy control subjects. The HIV-infected subjects, selected on the basis of fat redistribution, demonstrated significantly increased WHR compared to control subjects. Among males, the difference in WHR was largely the result of a significant reduction in hip circumference (Table 1). In contrast, the waist circumference was similar between HIV-infected male subjects and healthy control subjects. In women, the waist circumference tended to be higher in the HIV-infected patients than in the control subjects (Table 2).
HIV infected men and women were similar in age (43.8 ± 0.6 versus 43.6 ± 1.2 years; P = 0.92) and BMI (26.3 ± 0.3 versus 27.5 ± 1.1 kg/m2, P = 0.18), but WHR was greater in males than in females (0.99 ± 0.01 versus 0.94 ± 0.02; P = 0.002). Viral load (3044 ± 1212 versus 8241 ± 6651 copies/ml; P = 0.20), CD4 cell count (585 ± 29 versus 456 ± 60 cells/ml; P = 0.10), protease inhibitor (PI) and nucleoside reverse transcriptase inhibitor (NRTI) use (51% versus 57%; P = 0.61, and 91% versus 83%; P = 0.25, respectively) were not significantly different between the male and female HIV-infected subjects.
Comparison by gender
HIV-infected women had significantly higher peak GH in response to GHRH + arginine (36.4 ± 7.3 versus 18.9 ± 2.0 ng/mL; P = 0.003) and GH area under the curve (AUC) (2679 ± 593 versus 1284 ± 133 mg/dl/min; P < 0.0001) compared to HIV-infected men. In contrast, IGF-I levels were significantly reduced in the HIV-infected women compared to men (356 ± 13 versus 265 ± 27 ng/ml, male versus female respectively, P = 0.004) (Fig. 1).
Comparison by Race
In response to GHRH + arginine stimulation, GH AUC was higher (1831 ± 403 versus 1146±130 (mg-min)/dl; P = 0.04) and peak GH response tended to be higher (25.3 ± 5.1 versus 17.4 ± 2.2 ng/ml; P = 0.13) in non-Caucasian versus Caucasian male HIV-infected subjects, respectively. In contrast, peak GH (24.5 ± 5.9 versus 49.3 ± 10.1 ng/ml; P = 0.09) and GH AUC response to GHRH + arginine (1740 ± 457 versus 3617 ± 1046 mg/dl/min; P = 0.12) tended to be lower among non-Caucasian versus Caucasian HIV-infected women, respectively. IGF-I levels were not significantly different between non-Caucasian and Caucasian HIV-infected males (346 ± 32 versus 360 ± 14 ng/ml; P = 0.65) nor between non-Caucasian and Caucasian HIV-infected females (273 ± 36 versus 257 ± 41 ng/ml; P = 0.77).
Within gender comparisons
Among males, the failure rates to GHRH + arginine were higher among the HIV-infected patients compared to the control subjects across a range of cutoffs, and the optimal cutoff that provided the greatest separation between the groups was at 7.5 ng/ml (37% versus 8%; P = 0.004, male HIV versus control groups, respectively (Fig. 2a). In contrast IGF-I levels were increased in male HIV-infected versus control subjects (356 ± 13 versus 277 ± 31 ng/ml; P = 0.01, male HIV versus control groups, respectively). Among females, there was no difference between HIV and control subjects across a range of cutoffs (Fig. 2a). IGF-I levels were not different between female HIV-infected and control subjects (265 ± 27 versus 263 ± 27 ng/ml; P = 0.94, female HIV versus control groups, respectively).
Comparison by anitretroviral medication class and individual medications
Among male HIV infected patients, no significant differences were seen by PI or NRTI use (data not shown). Similarly, among female HIV infected patients, no significant differences were seen by PI or NRTI use (data not shown). Comparisons of GH responses were made for individual agents used by more than 20% of male (Epivir, Viread and Sustiva) and female (Epivir and Viread) HIV-infected patients. Peak GH and GH AUC responses did not differ by use of these individual agents either among men or women (data not shown).
Comparison by menstrual function
Among female HIV-infected patients, 54% were eumenorrheic and 45% were oligoamenorrheic. Peak GH (36.5 ± 7.5 versus 37.1 ± 14.6 ng/ml; P = 0.97) and GH AUC (2774 ± 653 versus 2692 ± 1058 mg/dl/min; P = 0.95) did not differ by menstrual status.
Relationship between fat distribution and GH
Among the group of HIV-infected male subjects, age (r, -0.23; P = 0.007), BMI (r, -0.35; P < 0.0001), waist circumference (r, -0.39; P < 0.0001) and WHR (r, -0.47; P < 0.0001) correlated with peak GH response to GHRH + arginine (Table 3). Among HIV infected female subjects BMI (r, -0.63; P < 0.001), waist circumference (r, -0.61; P = 0.001) and WHR (r, -0.43; P = 0.03) correlated with peak GH response to GHRH and arginine (Table 3). In stepwise regression analysis, age (B = -0.68; P = 0.009) and WHR (B, -146.8; P < 0.001) remained the most significant predictors of peak GH response to GHRH and arginine, controlling for BMI, and waist circumference among male HIV-infected subjects. None of the measured variables (age, BMI, WHR, waist circumference) were significant in the stepwise model for HIV-infected women.
Materials and methods
Clinical research protocol
A total of 164 male subjects [HIV-infected (n = 139), healthy controls (n = 25) similar in age and BMI] and 51 female subjects [HIV-infected (n = 25), and healthy female controls (n = 26) similar in age and BMI] were enrolled in the study between October 2001 and September 2005. The HIV-infected subjects were evaluated for participation in a research protocol for low dose GH replacement in HIV-infected patients. In this study, we report the screening data for the study, and also additional data from control subjects recruited for comparison with HIV-infected patients. All HIV-infected subjects were required to be on a stable antiretroviral regimen for at least 12 weeks prior to entrance into the study. HIV-infected subjects were characterized with fat redistribution at study entry, based on a WHR ≥ 0.90 and evidence of increased fat under the chin, at the back of the neck, or in the abdominal, chest, or breast areas, or decreased fat in the arms, legs, or face. Similar criteria were used in prior studies to identify patients with evidence of fat redistribution [3,16,17]. Subjects with diabetes mellitus, BMI < 20 kg/m2, hemoglobin < 9 g/dl or use of GH, GHRH, oral or parenteral glucocorticoids, oral contraceptive estrogen, megesterol acetate or antidiabetic agents within the 3 months prior to testing were excluded. Normal control subjects were healthy men and women 18-60 years old without history of anemia, diabetes mellitus, renal failure, carpal tunnel syndrome or history of any malignancy. Subjects with symptoms or history of pituitary disease or radiation treatment were excluded. Written informed consent was obtained from each subject before testing, in accordance with the Committee on the Use of Humans as Experimental Subjects of the Massachusetts Institute of Technology and the Subcommittee on Human Studies at the Massachusetts General Hospital.
Screening outpatient visit
After a 12-h overnight fast, subjects reported to the General Clinical Research Center for a screening visit during which a detailed medical history was obtained. Past and current medications, including the subject's history of antiviral therapy, were recorded. Subjects underwent GHRH + arginine stimulation testing in the morning following a 12-h fast. A complete blood count, blood glucose and creatinine levels were measured. Urine pregnancy tests were performed for female subjects. Height, weight and BMI were determined in the fasting state during the screening visit by bionutrition staff at the Massachusetts General Hospital General Clinical Research Center. Measurement of neck, mid-arm, waist (at the iliac crest), and hip circumferences were made to determine the WHR. Female patients were characterized as eumenorrheic or not based on regularity of menses in the 3 months prior to screening.
Growth hormone assessment
Subjects received standardized stimulation testing with GHRH + arginine [GHRH 1-29 (Geref, Serono, Inc., Norwell, Massachusetts, USA) 1 mg/kg intravenous bolus along with simultaneous administration of arginine hydrochloride 0.5 g/kg (maximum dose 30 g) given intravenously over 30 min]. Growth hormone levels were collected at -15, 0, 15, 30, 45, 60, 90, and 120 min after GHRH administration.
Biochemical and immunological function
Fasting glucose and insulin like growth factor (IGF-I) level were determined for all subjects after a 12-h fast prior to any stimulation testing. The CD4 cell count and viral load were determined for HIV-infected subjects.
GH was measured by a two-site radioimmunometric assay with an intra-assay coefficient of variation of 4.4% (Corning Inc., Nichols Institute Diagnostics, San Juan Capistrano, California, USA). The inter-assay coefficient (CV) was 6.6%. The sensitivity of the assay was determined to be 0.05 ng/mL. IGF-I was measured by a two-site radioimmunometric assay with an intra-assay CV of 4.9% (Diagnostics Systems Laboratory Inc., Webster, Texas, USA). The inter-assay coefficient was 5.1%. The sensitivity of the assay was determined to be 2.6 ng/ml.
The CD4 cell count was determined by flow cytometry (Becton Dickinson Immunocytochemistry Systems, San Jose, California, USA), and the HIV viral load was determined by ultra-sensitive assay (Amplicor HIV-1 Monitor Assay; Roche Molecular Systems, Indianapolis, Indiana, USA) with limits of detection of 50-75 000 copies/ml. Glucose, creatinine and complete blood count (CBC) were measured by standard techniques .
Demographic, body composition, and biochemical indices were compared by ANOVA for continuous variables and likelihood ratio for dichotomized variables. For race, a bivariate analysis was performed comparing percentage Caucasian and non-Caucasian by likelihood ratio. In HIV infected men and women and normal male and female control groups, separate correlation analyses were performed. Within the groups, separate univariate and stepwise regression analyses were performed to relate age, BMI, waist circumference, and WHR to peak GH response to GHRH + arginine stimulation testing. In stepwise regression modeling, peak GH response was the dependent variable, and age, BMI, waist circumference, and WHR were tested for entry into the model as dependent variables at P = 0.10. Final P values and parameter estimates were determined for each model. All statistical analyses were made using SAS JMP Statistical Database Software (version 5; SAS Institute, Inc., Cary, North Carolina, USA). Statistical significance was defined as a two tailed a value of P ≦ 0.05. Results are mean ± SEM unless otherwise indicated.