Growth Hormone Scientific Study Results

0021-972x/92/7404-0757$03.00/0 Journal of Clinical Endocrinology Copyright 0 1992 by The Endocrine

and

Metabolism Society

Vol. 74, No. 4 Printed in U.S.A.

Augmented Growth Hormone (GH) Secretory Frequency and Amplitude Mediate Enhanced Secretion during a Two-Day Fast in Normal MARK L. HARTMAN, JOHANNES D. VELDHUIS, MARY M. LEE?, K. G. M. M. ALBERTI, EUGENE MICHAEL 0. THORNER

MICHAEL SAMOJLIK,

Burst GH Men*

L. JOHNSON, AND

Departments of Medicine (M.L.H., J.D. V., M.O.T.) and Pharmacology (M.L.J.), University of Virginia National Science Foundation Science and Technology Center for Biological Timing (J.D. V., M.L.J., M.O.T.), University of Virginia Health Sciences Center, Charlottesville, Virginia 22908; the Department of Pediatrics, University of Pennsylvania School of Medicine (M.M.L.), Philadelphia, Pennsylvania 19104; the Department of Medicine, University of Newcastle Upon Tyne (K.G.M.M.A.), Newcastle Upon Tyne, Great Britain; and the Department of Medicine, Newark Beth Israel Medical Center, University of Medicine and Dentistry of New Jersey, New Jersey Medical School (E.S.), Newark, New Jersey 07112

ABSTRACT.

Serum GH concentrations are increased in fasted or malnourished human subjects. We investigated the dynamic mechanisms underlying this phenomenon in nine normal men by analyzing serum GH concentrations measured in blood obtained at 5-min intervals over 24 h on a control (fed) day and on the second day of a fast with a multiple-parameter deconvolution method to simultaneously resolve endogenous GH secretory and clearance rates. Two days of fasting induced a 5fold increase in the 24-h endogenous GH production rate [78 + 12 us. 371 + 57 uelL, volume) or 0.24 f -, (L,. liter of distribution 0.038 vs. 1.1 f 0.16 mg/m’ (assuming a distribution volume of 7.9% body weight), P = O.OOOl]. This enhanced GH production rate was accounted for by 2-fold increases in the number of GH secretory bursts per 24 h (14 + 2.3 vs. 32 f 2.4, P = 0.0006) and the mass of GH secreted per burst (6.3 + 1.2 vs. 11 + 1.6 pg/Lv, P = 0.002). The latter was a result of increased secretory-event amplitudes (maximal rates of GH release attained within a burst)

with unchanged secretory burst durations. GH was secreted in complex volleys composed of multiple discrete secretory bursts. These secretory volleys were separated by shorter intervals of secretory quiescence in the fasted than fed state (respectively, 88 f 4.2 vs. 143 + 14 min, P = 0.0001). Similarly, within volleys of GH release, constituent individual secretory bursts occurred more frequently during the fast [every 33 f 0.64 (fasted) vs. every 44 + 2.0 min (fed), P = O.OOOl]. The tllz of endogenous GH was not significantly altered by fasting [18 + 2.2 (fasted) vs. 20 f 1.5 min (fed), P = 0.471. Serum insulin-like growth factor I concentrations were unchanged after 56 h of fasting. In conclusion, the present data suggest that starvation-induced enhancement of GH secretion is mediated by an increased frequency of GHRH release, and longer and more pronounced periods of somatostatin withdrawal. (J Clin Endocrinol Metub 74: 757-765, 1992)

G

ROWTH failure occurs in the setting of increased serum GH and decreased insulin-like growth factor I (IGF-I) concentrations in patients with various forms

of nutrient deprivation, such as kwashiorkor and marasmus (1, 2). In healthy subjects, serum IGF-I levels decrease significantly after 5 days of fasting (2). Early studies of fasting observed inconsistent changes in serum GH concentrations and thus concluded that GH was unimportant in the regulation of altered metabolism in starvation (3-5). However, frequent blood sampling of fasted normal subjects indicates that pulsatile GH release is enhanced by fasting (6). This dissociation of the normal relationship between circulating levels of GH and IGF-I suggests that impaired somatic growth in malnourished patients is related to reduced IGF-I synthesis or action while GH, with its known actions to promote hepatic glucose production, lipolysis, and nitrogen conservation, may mediate, at least in part, the metabolic adaptation to starvation (7). Pulsatile GH release is

Received April 8, 1991. Address requests for reprints to: Michael 0. Thorner, Division of Endocrinology and Metabolism, Department of Medicine, Box 511, University of Virginia Health Sciences Center, Charlottesville, Virginia 22908. * Presented in part at the Second International Pituitary Congress, Palm Desert, CA, June 25-28,1989. This work was supported by NIH Grants RR-00847 (to the University of Virginia General Clinical Research Center and CLINFO laboratory), Clinical Investigator Award l-K08-HD-00860 (to M.L.H.), Research Career Development Award lK04-HD-00634 (to J.D.V.), GM-28928 (to M.L.J.), DK-32632 (to M.O.T.), DK-38942 (to the University of Virginia Diabetes Endocrinology Research Center), grants from the University of Virginia Computer Services, Pratt Fund, and Academic Enhancement Program, and the NSF Science and Technology Center for Biological Timing. t Current address: Department of Pediatrics, Massachusetts General Hospital, Boston, MA 02114.

757 Downloaded from jcem.endojournals.org by on October 30, 2008

758

HARTMAN

decreased in obese subjects and increased in patients with insulin-dependent diabetes mellitus, providing further evidence that nutrition and the metabolic milieu are important determinants of spontaneous GH release (8, 9). The importance of frequent blood sampling over an extended period of time to characterize the pulsatile release of hormones is now well known. Approximately twice as many GH pulses are detected in normal subjects when blood samples are obtained every 5 min us. every 20 min, most likely because of the relatively short t1/2 of GH disappearance from serum (-19 min) (10,ll). However, changes in GH secretory events cannot be directly inferred from such data because of the confounding influence of ongoing metabolic clearance which is variable among individuals (11, la), and in different pathophysiological contexts (8, 13). Thus, the increase in circulating GH concentrations reported in fasted subjects may be attributable to either an increase in GH secretion or a decrease in the metabolic clearance of GH or both. In this study we have measured GH in blood collected at 5-min intervals for 24 h and used deconvolution analysis to demonstrate that fasting increases the frequency and amplitude of GH secretory bursts without changing the GH MCR (14). Materials

and Methods

ET AL.

JCE & M. 1992 Vol74.No4

I. Volunteers were permitted to ambulate but were not allowed to nap or sleepuntil 2200 h. All serum sampleswere frozen at -20 C until analyzed. Assays

Serum GH concentrations were measuredin duplicate by immunoradiometric assay(IRMA; Nichols Institute, San Juan Capistrano, CA) using standardsdiluted in equine serum. Our results were multiplied by a factor of 0.5 to correct for the parallel shift in the IRMA standard curve causedby equine serum matrix relative to results obtained with human serum (12, 15, 16). The sensitivity of the assaywas taken as 0.25 pg/ L; undetectable sampleswere assignedthis value. The median intraassay coefficient of variation calculated from all 289 duplicated samplesin each subject averaged 7.4% (range 2.926.4%). IGF-I was measuredin unextracted serumin the presence of heparin which causes IGF-I to dissociate from its binding proteins (17). Commercially available RIAs, adapted by E. Samojlik, were usedto measuretestosterone (RSL, Los Angeles, CA) and estradiol (DiagnosticsProducts Corporation, Los Angeles, CA) (Samojlik, E., Kirschner, M. A., Ribot, S., and Szmal, E., submitted for publication). Serum free testosterone and estradiol were measuredby ultracentrifugation dialysis (18). Urine free cortisol was measuredby RIA after methylene chloride extraction (y-Coat Cortisol, Clinical Assays, Cambridge,MA). Serum concentrations of Tq, LH, FSH, PRL, BOH, AcAc, FFA were measuredby previously described methods(16, 19,20).

Subjects and study design

Deconvolution

analysis

The study was approved by the Human Investigation and General Clinical ResearchCenter Advisory Committeesof the University of Virginia. Nine healthy men (ages24-28) of normal body weight [body massindices (BMI) 21-25 kg/m2) were studied after written informed consent. All were nonsmokers, weretaking no medications,had not undertaken transmeridian travel for at least 4 weeks, and had unremarkable clinical histories and physical examinations. All had normal biochemical indices of renal, hepatic, and hematologic function and normal fasting serumconcentrations of glucose,Tq, TSH, PRL, testosterone, IGF-I, and immunoactive LH and FSH. The subjectswere studied on the General Clinical ResearchCenter on two occasions:1) a control fed day during which a weightmaintenancediet was served at 0900, 1300,and 1800 h; and 2) day 2 of a fast (32-56 h after the last meal) during which the subjectsingestedonly water, potassiumchloride (20 meq/day), and a multivitamin tablet. Compliancewith the fast was monitored by daily weights and measurementof urine ketones. The two admissionswere separatedby at least 30 days and their order was randomized. On each study day a cannula was inserted into a forearm vein at 0700 h; blood sampleswere obtained from 0800-0800h at 5-min intervals for measurement of GH and at 6-h intervals for measurementof gonadalsteroids. A 24-h urine was collected for measurement of urine free cortisol. Daily blood samples(0800 h) were obtained for complete blood count, serum chemistries, hepatic enzymes, serum /3-hydroxybutyrate (BOH), acetoacetate(AcAc), FFA, and IGF-

Multiple-parameter deconvolution derives quantitative estimatesof attributes of hormone secretory events from measured peripheral serumhormoneconcentrations while simultaneously estimating the endogenoussubject-specifichormone MCR (14). SerumGH concentrations were consideredto arisefrom a series of discretesecretory bursts of determinablelocations, durations and amplitudes, acted upon by metabolic clearancekinetics. A distinct secretory burst was defined as a random (Gaussian) distribution of instantaneous molecular secretory rates, whose fitted amplitude could be distinguished from zero (i.e. pure noise)with 95% statistical certainty. A tonic secretionfunction was not required to model the present data. Clearanceof GH was modeledasa monoexponential function with a unique rate constant for each subject. Serum GH concentrations were assumedto decay to the sensitivity of the IRMA. The convolution integrals relating the secretion and elimination functions were solvedby a numerical deconvolution technique (12, 14). The following independent parameters were calculated for eachsubject:amplitudes (maximal secretory rate) and temporal positions of all GH secretory bursts; GH secretory burst halfduration (duration at half-maximal amplitude); and the tllz of GH disappearance.The latter two parameters and the GH distribution volume were assumedto be constant throughout the 24-h period for each individual (12). The massof GH secreted per burst was estimated as the area of the resolved secretory burst [in units of pg/L” (L,, L of distribution volume)]. The 24-h endogenousGH production rate was esti-

Downloaded from jcem.endojournals.org by on October 30, 2008

GH SECRETORY

BURSTS

DURING

FASTING

759

mated as the product of the number of secretory bursts and the mean GH secretory burst mass.

occurred in serum levels of IGF-I enzymes.

Statistical analyses

Effects

Results are expressed as means + SE, except where noted otherwise. Comparisons among mean parameter estimates were made by Duncan’s multiple range test after analysis of variance or by paired two-tailed t tests. Non-Gaussian distributed parameters were logarithmically transformed before analysis. The randomness of individual subjects’ serial interburst intervals and secretory burst mass estimates was examined by autocorrelation analysis with one time lag. The autocorrelation was considered significant for the group of subjects if the distribution of Z-scores (r/SE) was nonrandom by the KolmogorovSmirnov test (12). Correlations between properties of GH secretion and BMI, and serum levels of IGF-I, BOH, AcAc, and gonadal steroids were sought with linear regression analysis. Statistical significance was assumed for P less than or equal to 0.05. If multiple comparisons were made the overall per-study error rate was limited by restricting the per-comparison P value to less than or equal to 0.01 (16).

Results Biochemical

effects of fasting

Fasting-induced changes in serum concentrations of IGF-I and selected metabolites are shown in Table 1. After 32 h of fasting (start of frequent blood sampling) significant increases in serum concentrations of BOH (4fold), AcAc (6-fold), FFA (3-fold), bilirubin (3-fold), and uric acid were observed. After 56 h of fasting, further increases above basal levels were observed for BOH (9fold), AcAc (&fold), FFA, uric acid, and creatinine, and significant decreases in serum levels of glucose and bicarbonate occurred. These metabolic alterations were reversed with refeeding. All subjects were ketonuric by the evening of day 2 of the fast. No significant changes TABLE

1. Effect of a P-day fast on the serum concentrations

or activities

of hepatic

on gonadal and adrenal steroids

Pooled specimens were analyzed for gonadal steroid concentrations. Serum free testosterone concentrations decreased significantly from 0.42 f 0.066 to 0.34 + 0.046 nmol/L after 2 days of fasting (P < 0.05). Fastingassociated decreases in serum concentrations of total testosterone (20 + 1.4 us. 18 + 1.9 nmol/L), total estradiol (94 f 45 us. 69 + 14 pmol/L) and free estradiol (4.8 +2.4 us. 3.9 + 1.0 pmol/L) were not statistically significant. Urinary excretion of free cortisol was not significantly increased by the 2-day fast (230 f 40 us. 370 f 150 nmol/ day). Mean serum GH concentrations The 2-day fast resulted in a more than 3-fold increase in 24-h mean serum GH concentrations (2.0 Z!Z0.29 vs. 6.7 + 1.1 r.Lg/L, P = 0.0004). The percent of samples with undetectable GH concentrations was 29 + 8.5% (range O-74%) on the control day, and 3.0 f 2.0% (range O18%) on the fasting day (P = 0.01). Deconvolution and clearance

analysis

of endogenous

GH secretion

Twenty-four hour profiles of pulsatile serum GH concentrations and deconvolution-resolved GH secretory rates from two normal men on the control and fasting days are shown in Fig. 1. The quantitative changes in specific attributes of endogenous GH secretory bursts and the tIlz of GH disappearance are illustrated in Fig. 2. Twenty-four hour endogenous GH production rates were increased 5-fold by 2 days of fasting (78 f 12 us.

of IGF-I and selected metabolites in normal men

Day 04 of fast

Glucose (mmol/L)

fi-Hydroxybutyrate (mmol/L)

Acetoacetate (mmol/L)

FFA (mmol/L)

IGF-I W/ml)

Day 1 (8 h) Day 2 (32 h) Day 3 (56 h)

4.3 + 0.22 4.1 + 0.20 3.2 + 0.17"

0.21 f 0.11 0.80 zk 0.28" 1.8 f 0.39

0.075 5 0.023 0.43 + 0.16" 0.58 + 0.10

0.37 2 0.10 1.0 f 0.25” 1.2 + 0.14”

1.7 * 0.21 1.6 + 0.21 1.5 + 0.25

Uric Bicarbonate Creatinine Bilirubin acid (mmol/L) (mWL) GmWL) (moW 26 + 0.80 92 f 5.1 8.0 2~ 0.80 Day 1 (8 h) 38Ok 19 22 f 1.2" Day 2 (32 h) 450f15 24 k 1.2 99 _+ 6.7 Day 3 (56 h) 550 + 18" 20 + 1.3" 116 + 6.0" 24 + 1.9” Blood samples were taken at 0800 h on each day. “P < 0.05 us. day 1 by analysis of variance and Duncan’s multiple range test; non-Gaussian distributed parameters were logarithmically transformed before analysis. Control day values are not shown since these were not different from those obtained on day 1 after 8 h of fasting. Serum concentrations of BOH, AcAc, and FFA were not measured on the control day.

Downloaded from jcem.endojournals.org by on October 30, 2008

HARTMAN

760

TIME (ttwa) FAST 7

1

1. Twenty-four hour profiles of serum GH concentrations and deconvolution-resolved GH secretory rates in two normal men, aged 26 (A) and 25 (B), on a control fed day (l&panels) and the second day of a fast (right panels). For each individual, the upper panels depict serial serum GH concentrations measured in blood collected at 5-min intervals over 24 h. The intrasample SDS are denoted by vertical marks. The continuous line through the data represents the calculated reconvolution curve predicted by the multiple-parameter convolution model (see Materials and Methods). In the louver panels, the calculated GH secretory rate is plotted US. time. The secretory rate is derived by removing the influence of subject-specific endogenous GH clearance on the GH concentration profile. Serum GH concentrations and secretory rates were increased on the fasting day. Differing scales for the vertical axes are used because of the wide range of serum GH concentrations. Note that the resolved GH secretory pattern consists of volleys of multiple secretory bursts. L, = L of distribution volume.

FIG.

371 + 57 pg/LV, P = 0.0001; Fig. 2A). These 24-h GH production rates correspond to approximately 0.24 + 0.038 and 1.1 + 0.16 mg/m’, respectively, assuming a mean GH distribution volume of 7.9% body weight (13). This enhanced GH production rate was accounted for by a 2-fold increase in the number of GH secretory bursts/ 24 h (14 f 2.3 us. 32 f 2.4, P = 0.0006; Fig. 2B) and the mass of GH secreted per burst (6.3 f 1.2 vs. 11 + 1.6 pg/ L,, P = 0.002; Fig. 2C). The latter was a result of

ET AL.

JCE & M. 1992 Vol74.No4

increased secretory burst amplitudes (0.24 + 0.053 us. 0.45 + 0.052 pg. L,-’ . min-‘, P = 0.0004; Fig. 2D) with unchanged secretory burst half-durations [27 + 2.5 (fed) US. 24 + 1.4 (fasted) min, P = 0.36; Fig. 2E]. Based on the GH secretory burst frequencies and half-durations, we estimated that 95% of daily GH secretion occurred during 45% (11 h) of the control days and 91% (22 h) of the fasting days. The mean tllz of endogenous GH was not significantly altered by fasting [20 + 1.5 (fed) vs. 18 f 2.2 (fasted) min, P = 0.47; Fig. 2F), although based on the individual statistical confidence limits of their tllz estimates, three subjects had individually significant decreases and two subjects had individually significant increases. The increased number of GH secretory bursts/24 h on the fasted day was also reflected by a decrease in the mean interval between secretory burst centers (106 f 17 us. 45 f 3.9 min, P = 0.0004). On both days of study, two types of interburst intervals were apparent: 1) those that separated individual secretory bursts within a cluster or volley of multiple secretory events (intravolley intervals); and 2) those that spanned periods of secretory quiescence during which secretory rates approached zero (intervolley intervals). A histogram of the two types of interburst intervals on the control and fasting days is shown in Fig. 3. On both study days, the distributions of these two types of interburst intervals were significantly different. On the control day, volleys of GH secretion were separated by a mean interburst interval of 143 f 14 (median = 106) min, whereas constituent individual secretory events within secretory volleys occurred every 44 f 2.0 (median = 41) min (P = 0.0001). On the fasting day, mean inter- and intravolley interburst intervals were 88 + 4.2 (median = 80) and 33 + 0.64 (median = 32) min, respectively (P = 0.001). The decrease in these two types of interburst intervals with fasting was significant (P = 0.0001). Intravolley interburst intervals accounted for 60% and 82% of the total number of interburst intervals on the control and fasting days, respectively. A significantly positive autocorrelation existed between the mass of GH secreted in successive secretory bursts on both the control and fasting days (P < 0.01). The individual autocorrelation coefficients (r values) in the nine subjects ranged from -0.31 to 0.70 (median = 0.32) on the control days and from 0.10 to 0.54 (median = 0.41) on the fasting days. Successive interburst intervals were not significantly autocorrelated. Relationship between GH secretion and BMI (kg/m’), and serum concentrations of IGF-I and gonadal steroids

The correlation of BMI and the endogenous GH production rate on the control and fasting days is shown in Fig. 4. The amount of GH secreted on the fasting day

Downloaded from jcem.endojournals.org by on October 30, 2008

GH SECRETORY

CONlROL

BURSTS

DURING

FASTING

FAFT

30

761

CONT’WL

FAST

CONTROL

FAST

1.0 0.83

20

M&S

GH SECRl3ORY BURST AMPLITUDE

OF GH

SECRETED/WRSl

,o

(pg*L,-l

bdb)

0.0 CONll?lX

FM

so

40 IE

-

P-0.3l

1F

1

4a

SECRETORY

o-n-Y.-I ‘7

r

30..

3.

GH ti4.F~LIFE

BURST HALF-oURAnON

o.4

.minql)O.Z

0

‘3

0.6

EE

2.

‘,

ZE

FE

hi4

2o

4

OL

01 CONTROL

FEZ

CONTROL

FAST

resulting from 2 days of FIG. 2. Quantitative changes in specific attributes of endogenous GH secretory bursts and t iI2 of GH disappearance fasting as derived from deconvolution analysis of serum GH concentrations measured at 5-min intervals for 24 h in 9 normal men (see Materials and Methods). For each individual, the 24-h endogenous GH production rate (A) was estimated as the product of the total number of GH secretory bursts (B) and the mean mass of GH secreted per burst (C). The mean 24-h GH production rates correspond to approximately 0.24 + 0.038 (control) and 1.1 f 0.16 (fast) mg/m’ assuming a mean GH distribution volume of 7.9% body weight (13) and correcting for the subjects’ body surface areas. The GH secretory burst amplitude (D) is the maximal secretory rate attained during a secretory burst and the GH secretory burst half-duration (E) is the duration of a secretory burst at half-maximal amplitude. The GH tljz (F) for each subject was derived from a singlecomponent disappearance rate constant. In each panel, the changes for each individual are shown by connecting lines and the mean + SE for each groupare shown adjacent to the individual data. Differences between attributes on the control and fasting days were tested by paired two-tailed t tests. L, = L of distribution volume.

was inversely correlated with BMI (r = -0.90, P < 0.001). This relationship was not observed on the control day in these subjects. Serum concentrations of IGF-I, and total and free concentrations of testosterone and estradiol were not significantly correlated with GH production rates, the mass of GH secreted per burst, or the number of GH secretory bursts/24 h on either the control or fasting days. Discussion

We have investigated the mechanisms underlying augmented serum GH concentrations during nutrient deprivation using a multiple-parameter deconvolution technique and frequent blood sampling. Our results demonstrate that endogenous GH secretion rates are enhanced &fold by a a-day fast in normal young men. A doubling of both GH secretory burst frequency and the mass of GH secreted per burst fully accounted for the observed increase in serum GH concentrations as the GH MCR was not significantly altered by fasting. Fasting increased

the amplitude but not the half-duration (width) of GH secretory bursts. It is unlikely that GH distribution volumes were significantly altered since plasma volumes estimated by infusions of lz51-labeled albumin in six other normal men were unchanged by a &day fast (Vance, M. L., Thorner, M. O., and Veldhuis, J. D., unpublished data). Although GH secretion became virtually continuous with fasting, serum GH concentrations were modeled adequately by pulsatile secretion without measurable intervening tonic secretion (12). Since serum GH concentrations below the limit of detection of conventional GH immunoassays are also pulsatile (21), it is possible that we have underestimated the number of GH secretory bursts in the fed state. However, the GH secretory burst frequency was still increased 2-fold by fasting in three subjects in whom GH was measurable in 94-100% of samples on the control day. Food restriction increases GH secretion in sheep (22), steers (23), pigs (24), dogs (25), rabbits (26), and chickens (27), but decreases GH release in rats (28). In dogs (25) both serum GH concentration pulse frequency and am-

Downloaded from jcem.endojournals.org by on October 30, 2008

762

HARTMAN

FAST

0

20 40 60

.a0

350

loil50

550

INTERBURST INTERVALS (mini 3. Histograms of the GH interburst intervals derived from deconvolution analysis. An interburst interval is a general term defined as the time in minutes between designated secretory burst centers. A volley is defined as a cluster of two or more secretory bursts, between which the secretory rate does not fall to 0. Accordingly, an intravolley interval separates secretory bursts within a volley. Intervolley intervals separate consecutive volleys or solitary secretory bursts; during these intervals, the secretory rate approaches zero asymptotically. The distributions of these two types of structurally distinct interburst intervals were significantly different on both study days by the Wilcoxon rank sum test (P 5 0.001). Both types of interburst intervals were significantly shorter on the fasting day compared to the control day (P = 0.0001). Thus, fasting prolonged the duration of GH secretory volleys, shortened the intervening periods of secretory quiescence, and increased the frequency of individual secretory bursts within the volleys. The mean + SE (median) is shown above each distribution.

FIG.

-S <

800

CONTROL I?- -0.22.

cn 4

.I5 Y 600.. $

O0

0.56

R- -0.90. PC 0.001

0

400..

0

i? 200.. 6 G.4

P-

F*sT0

0

8

r

*

0. 20

:-‘h

21 BODY

22 MASS

0

A A 23

24

INDEX

(kg/m2)

25

26

FIG. 4. Correlations of the 24-h endogenous GH production rate with body mass index on the control and fasting days. A significantly negative correlation was observed on the fasting day indicating that subjects with higher body mass indices secreted less GH at this time. No correlation was observed on the control days. The correlation coefficients and P values are shown in the figure. L, = L of distribution volume.

plitude are increased, whereas in sheep (22) and steers (23) only the latter is increased. Most investigators have assumed that increased serum GH concentrations in fasting reflect enhanced GH secretion and have not

ET AL.

JCE & M. 1992 Vol’74.No4

calculated the GH MCR. This assumption may not always be valid. Nutritional restriction has been reported to increase the GH tl,n in sheep and calves (29) but not in chickens (30). In our study, the individual increases and decreases in GH tljz observed with fasting are attributable to the variability inherent in such estimates since no significant overall change was observed. All of the calculated GH tl,z were within the range of previously reported values for exogenous and endogenous GH tliz in normal subjects (reviewed in 11, 12). In contrast, deconvolution estimates of GH tliz are significantly shorter in obese subjects (8). Pulses of GH secretion arise from the interactions of GHRH and SRIH released into the hypophyseal-portal circulation (31). In rats, GHRH peaks in hypophysealportal. blood occur during periods of decreased SRIH concentrations (32) whereas this relationship has not been evident in sheep (22). Since such measurements are impossible in humans, we have analyzed the intervals between GH secretory bursts to make inferences about the mechanisms by which fasting enhances GH secretion in man. On both study days, the distributions of intraand intervolley intervals were significantly different, suggesting that these two types of intervals may relate to distinct physiological phenomena. We hypothesize that intravolley intervals reflect the frequency of bursts of GHRH secretion, whereas intervolley intervals represent the time between nadirs of SRIH secretion (12). According to this model (Fig. 5), volleys of GH secretory activity might arise from secretion of multiple discrete bursts of GHRH into the hypophyseal-portal blood during a sustained period of decreased SRIH secretion. Fasting significantly decreased both types of intervals so that volleys of GH secretion were prolonged and the frequency of GH secretory bursts within volleys was increased. Since serum GH concentrations do not decay to undetectable levels within secretory volleys (in either the fed or fasted state), the increased frequency of GH release episodes within such volleys suggests that the GH pulse frequency is truly accelerated by fasting. Based on these findings, we speculate that in fasting men the frequency of GHRH release is increased and nadirs of SRIH secretion are prolonged, resulting in increased GH secretory burst amplitude and frequency. Nutritional restriction increases the amplitude but not the frequency of GH pulses in sheep and this is associated with a 50% decrease in hypophyseal-portal blood concentrations of SRIH with no change in GHRH pulse amplitude or frequency (22). These data support our human model of modulation of GH pulse amplitude by SRIH and a coupling of GHRH and GH secretory bursts. However, our analysis suggests that in man fasting alters both SRIH and GHRH secretion. The metabolic and hormonal mechanisms by which

Downloaded from jcem.endojournals.org by on October 30, 2008

GH SECRETORY FED AAA

--

AAA -

0

0

AAA

BURSTS

FASTED B IP-?

AM LI

HYPOTHETICAL GHRH SECRETORY RATE

FIG. 5. Hypothetical

model for the physiological basis of fasting-induced increases in burst-like volleys of GH secretion in man. The top panels depict typical patterns of GH secretory rates derived by deconvolution analysis on a fed day (left) and on the second day of a fast (right). Intervals between GH secretory bursts are defined as either intravolley (denoted by “A”) or intervolley (denoted as “B”) intervals, as defined in Fig. 3. The middle and bottom panels illustrate hypothetical patterns of somatostatin (SRIH) and GHRH secretion, derived from analysis of the GH interburst intervals. Intravolley interburst intervals are considered to reflect the frequency of bursts of GHRH secretion, whereas intervolley interburst intervals represent periods of time separating nadirs of SRIH secretion. Thus multiple GHRH bursts during an interval of decreased SRIH secretion may give rise to volleys of GH secretion. During periods of increased SRIH secretion, GH response to GHRH is inhibited. We hypothesize that decreased mean intra- and intervolley interburst intervals observed in fasting subjects (Fig. 3) reflect an increased frequency of GHRH release and prolonged nadirs of SRIH secretion. The frequency of GHRH is illustrated here as constant, although some physiologic variability probably occurs based on mean intravolley interburst interval coefficients of variation of 23% (control) and 26% (fasting).

nutritional deprivation signals the hypothalamic-somatotroph axis are not known although it is likely that IGF-I and insulin are involved. Since IGF-I directly inhibits pituitary GH secretion and also stimulates hypothalamic SRIH release (33), fasting-associated decreases in IGF-I concentrations may mediate increased GH secretion (2). However, as previously shown (6), pulsatile GH secretion increased before significant decreases in total serum IGF-I concentrations occurred. This may reflect the stability of IGF-I bound to its binding proteins since the bound fraction accounts for the vast majority of circulating IGF-I (17). Nevertheless, fasting may decrease free IGF-I concentrations more rapidly by increasing concentrations of IGF-I binding proteins. Preliminary results suggest that serum concentrations of GH and IGF binding protein-l (IGFBP-1) increase with fasting and decrease with refeeding with similar time courses (34, 35). Furthermore, small in-

DURING

FASTING

763

creases in free IGF-I concentrations achieved by a lowdose infusion of recombinant human IGF-I rapidly suppressed fasting-enhanced pulsatile GH secretion (36). Fasting may also decrease the direct inhibitory effect of insulin on somatotroph secretion (37). Decreased IGF-I synthesis in the hypothalamus or pituitary may increase GH secretion by paracrine mechanisms (38). Finally, inhibitors of IGF-I bioactivity, identified in the serum of malnourished children and diabetics, may attenuate IGFI negative feedback on the somatotrophic axis (1). Several other hormones and metabolites must be considered as possible mediators of the effect of fasting to stimulate GH secretion. Although infusions of sodium phydroxybutyrate modestly stimulate GH secretion, this effect is blocked by increases in serum FFA which inhibit the GH response to GHRH (39, 40). Thus, it is unlikely that fasting-associated increases in serum concentrations of BOH, AcAc, and FFA stimulate GH secretion. In this study, fasting significantly decreased serum concentrations of free testosterone but total testosterone, total and free estradiol, and urine free cortisol were unchanged. Fasting has previously been reported to decrease serum TB and increase serum cortisol concentrations (5, 41). However, it is unlikely that these hormones regulate the GH response to fasting since the observed changes are in the opposite direction of what would be expected if they were responsible for the stimulation of GH secretion (42). Finally, glucagon and branched-chain amino acids, which are capable of stimulating GH release, increase during the first few days of fasting (5, 42-44). Linear regression analysis revealed that the amount of GH secreted during the fast was negatively correlated with the degree of adiposity, as estimated by the BMI. The lack of correlation in the fed state probably reflects the narrow range of BMIs in these men since a negative correlation has previously been observed in obese and nonobese fed subjects (8, 45). A negative correlation of BMI and the GH response to GHRH, measured with the same GH IRMA, has also been evident in nonobese men (46). More sensitive GH assays and accurate measurements of fat mass may strengthen this correlation in fed normal-weight subjects. In summary, fasting increases the amplitude and frequency of GH secretory bursts in normal men without altering their duration or the GH MCR. Somatotroph secretion becomes virtually continuous during fasting but the pattern remains pulsatile without measurable tonic secretion. Volleys of GH secretion are prolonged and constituent individual secretory bursts occur more frequently than in the fed state. This pattern is consistent with an increased frequency of GHRH release and prolonged periods of reduced SRIH secretion. Increases in GH secretion with fasting are inversely related to the degree of adiposity of normal-weight subjects. Although

Downloaded from jcem.endojournals.org by on October 30, 2008

764

HARTMAN

enhanced GH release during starvation may promote lipolysis and nitrogen conservation, further investigation is required to elucidate the metabolic and hormonal responses to, and mechanisms which mediate, increased GH secretion during fasting.

ET AL.

17.

18.

Acknowledgments We thank Ms. Sandra W. Jackson and the staff of the General Clinical Research Center of the University of Virginia for their expert assistance. We also thank Ms. Ginger Bauler, Ms. Catherine Kern, and Ms. Marjorie Gingell for performing the GH assays, Ms. Suzan Pezzoli for technical and artistic assistance, and Mr. David G. Boyd for assistance with CLINFO.

19.

20.

21.

References 1. Phillips LS, Unterman TG. Somatomedin activity in disorders of nutrition and metabolism. Clin Endocrinol Metab. 1984;13:145-89. 2. Clemmons DR, Van Wyk JJ. Factors controlling blood concentration of somatomedin C. Clin Endocrinol Metab. 1984;13:113-43. 3. Cahill Jr GF, Herrera MG, Morgan AP, et al. Hormone-fuel interrelationships during fasting. J Clin Invest. 1966;45:1751-69. 4. Quabbe HJ, Schilling E, Helge H. Pattern of growth hormone secretion during a 24-hour fast in normal adults. J Clin Endocrinol Metab. 1966;26:1173-7. 5. Aoki TT. Hormone-fuel interrelationships in normal, fasting, and diabetic man. In: Marble A. Krael LP. Bradlev RF. Christlieb AR. eds. Joslin’s diabetes mellitus. Philadelphia: Lea and Febiger; 1985:138-57. 6. Ho KY, Veldhuis JD, Johnson ML, et al. Fasting enhances growth hormone secretion and amplifies the complex rhythms of growth hormone secretion in man. J Clin Invest. 1988;81:968-75. -‘I. Thorner MO, Vance ML. Growth hormone, 1988. J Clin Invest 1988;82:745-7. 8. Veldhuis JD, Iranmanesh A, Ho KKY, Waters MJ, Johnson ML, Lizarralde G. Dual defects in pulsatile growth hormone secretion and clearance subserve the hyposomatotropism of obesity in man. J Clin Endocrinol Metab. 1991;72:51-9. 9. Asplin CM, Faria ACS, Carlsen EC, et al. Alterations in the pulsatile mode of growth hormone release in men and women with insulin-dependent diabetes mellitus. J Clin Endocrinol Metab. 1989;69:239-45. 10. Evans WS, Faria ACS, Christiansen E, et al. Impact of intensive venous sampling on characterization of pulsatile GH release. Am J Physiol 1987;252:E549-56. 11. Faria ACS, Veldhuis JD, Thorner MO, Vance ML. Half-time of endogenous growth hormone (GH) disappearance in normal man after stimulation of GH secretion by GH-releasing hormone and suppression with somatostatin. J Clin Endocrinol Metab. 1989;68:535-41. 12. Hartman ML, Faria ACS, Vance ML, Johnson ML, Thorner MO, Veldhuis JD. Temporal structure of in uiuo growth hormone secretory events in humans. Am J Physiol 1991;260:ElOl-10. 13. Owens DM, Srivastava C, Tompkins CV, Nabarro JDN, Sonksen PH. Studies on the metabolic clearance rate, apparent distribution space and plasma half-disappearance time of unlabeled human growth hormone in normal subjects and in patients with liver disease, renal disease, thyroid disease and diabetes mellitus. Europ J Clin Invest. 1973;3:284-94. 14. Veldhuis JD, Carlson ML, Johnson ML. The pituitary secretes in bursts: appraising the nature of glandular secretory impulses by simultaneous multiple-parameter deconvolution of plasma hormone concentrations. Proc Nat1 Acad Sci USA. 1987;84:7686-90. 15. Felder RA, Ho11 RW, Martha Jr PM, et al. Influence of matrix on concentrations of somatotropin measured in serum with commercial immunoradiometric assays. Clin Chem. 1989;35:1423-6. 16. Hartman ML, Veldhuis JD, Vance ML, Faria ACS, Furlanetto RW, Thorner MO. Somatotropin pulse frequency and basal con”

I

22.

23.

24.

25.

26. 27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

JCE & M. 1992 Voll4.No4

centrations are increased in acromegaly and are reduced by successful therapy. J Clin Endocrinol Metab. 1990;70:1375-84. Furlanetto RW, Marino JM. Radioimmunoassay of somatomedinC/insulin-like growth factor I. Peptide Growth Factors Part A. In: Barnes D, Sirbasku DA, eds. Methods in enzymology. Orlando, FL: Academic Press Inc; 1987;146:216-27. Kirschner MA, Samojlik E, Drejka M, Szmal E, Schneider G, Ertel N. Androgen-estrogen metabolism in women with upper body versus lower body obesity. J Clin Endocrinol Metab. 1990;70:4739. Harrison J, Hodson AW, Skillen AW, Stappenbeck R, Agius L, Alberti KGMM. Blood glucose, lactate, pyruvate, 3-hydroxybutyrate and acetoacetate measurements in man using a centrifugal analyzer with fluorimetric attachment. J Clin Chem Clin Biochem. 1988;26:141-6. Knox DP, Jones DG. Automated enzymatic determination of plasma free fatty acids by centrifugal analysis. J Automatic Chem. 1984;6:152-4. Winer LM, Shaw MA, Baumann G. Basal plasma growth hormone levels in man: new evidence for rhythmicity of growth hormone secretion. J Clin Endocrinol Metab. 1990;70:1678-86. Thomas GB, Cummins JT, Francis H, Sudbury AW, McCloud PI, Clarke IJ. Effect of restricted feeding on the relationship between hypophysial portal concentrations of growth hormone (GH)-releasing factor and somatostatin, and iugular concentrations of GH in ovariectomized ewes. Endocrinology~1991;128:1151-8. Breier BH. Bass JJ. Butler JH. Gluckman PD. The somatotronhic axis in young steers: influence of nutritional status on pulsatile release of growth hormone and circulating concentrations of insulin-like growth factor 1. J Endocrinol. 1986;111:209-15. Atinmo T. Baldiiao C. Pond WG. Barnes RH. Immunoreactive growth hormone ievels’in pigs fed protein or energy restricted diets during the postweaning period. J Nutr. 1976;106:947-51. Cella SG, Moiraghi V, Minuto F, et al. Prolonged fasting or clonidine can restore the defective growth hormone secretion in old dogs. Acta Endocrinol (Copenh). 1989;121:177-84. McIntyre HB, Ode11 WD. Physiological control of growth hormone in the rabbit. Neuroendocrinology.-1974;16:8-21. Lauterio TJ. Scanes CG. Time course of changes in nlasma concentrations of the growth related hormones durmg protein restriction in the domestic fowl (Gallus domesticus). Proc Sot Exp Biol Med. 1987;185:420-6. Tannenbaum GS, Rorstad 0, Brazeau P. Effects of prolonged food deprivation on the ultradian growth hormone rhythm and immunoreactive somatostatin tissue levels in the rat. Endocrinology. 1979;104:1733-8. Trenkle A. Estimates of the kinetic parameters of growth hormone metabolism in fed and fasted calves and sheep. J Anim Sci. 1976;43:1035-43. Lauterio TJ, Scanes CG. Effect of age and protein restriction on the clearance and secretion of growth hormone in the domestic fowl. Poultry Sci. 1988;67:120-5.Thormer MO. Vance ML. Hartman ML. et al. Phvsioloeical role of somatostatin on growth hormone regulation in human; Metabolism. 1990;39[Suppl2]:40-2. Plotsky PM, Vale W. Patterns of growth hormone-releasing factor and somatostatin into the hypophysial-portal circulation of the rat. Science. 1985;230:461-3. Berelowitz M, Szabo M, Frohman LA, Firestone S, Chu L, Hintz RL. Somatomedin-C mediates growth hormone negative feedback by effects on both the hypothalamus and the pituitary. Science. 1981;212:1279-81. Hartman ML, Thorner MO. Fasting-induced enhancement of pulsatile growth hormone (GH) secretion is rapidly abolished by refeeding. Proz 72nd Meeting of the Endocrine Sot. 1990:123. Wass JAH, Coiterill AM. The control of insulin-like growth factor binding protein-l (IGFBP-1) during prolonged fasting. Prog 72nd Meeting of the Endocrine Sot. 1990;1141. Hartman ML, Clayton PE, Perlman AJ, Thorner MO. A low-dose euglycemic IGF-I infusion rapidly suppresses fasting-enhanced pulsatile GH secretion in man. Prog 73rd Meeting of the Endocrine Sot. 1991;1609.

Downloaded from jcem.endojournals.org by on October 30, 2008

GH SECRETORY

BURSTS DURING

37. Yamashita S, Melmed S. Effects of insulin on rat anterior pituitary cells: inhibition of growth hormone secretion and mRNA levels. Diabetes. 1986;35:440-7. 38. Fagin JA, Fernandez-Mejia C, Melmed S. Pituitary insulin-like growth factor-I gene expression: regulation by triiodothyronine and growth hormone. Endocrinology. 1989;125:2385-91. 39. Quabbe HJ, Trompke M, Luyckx AS. Influence of ketone body infusion on plasma growth hormone and glucagon in man. J Clin Endocrinol Metab. 1983;57:613-8. 40. Imaki T, Shibasaki T, Shizume K, et al. The effect of free fatty acids on growth hormone (GH)-releasing hormone-mediated GH secretion in man. J Clin Endocrinol Metab. 1985;60:290-3. 41. Vance ML, Thorner MO. Fasting alters pulsatile and rhythmic cortisol release in normal men. J Clin Endocrinol Metab. 1989;68:1013-8. 42. Reichlin S. Regulation of somatotrophic hormone secretion. In:

43. 44. 45

46.

FASTING

Knobil E, Wawyer WH, eds. Handbook of Physiology, Section 7, Volume IV. Washington, DC: American Physiological Society; 1974;405-47. Mitchell ML, Byrne MJ, Sanchez Y, Sawin CT. Detection of growth hormone deficiency: the glucagon stimulation test. N Engl J Med. 1970:282:539-41. Felig P, Owen OE, Wahren J, Cahill Jr GF. Amino acid metabolism durine nrolonaed fastine. J Clin Invest. 1969:48:584-94. Iranmynesh A’, Lizarrahe G, Veldhuis JD. Age and relative adiposity are specitic negative determinants of the frequency and amplitude of growth hormone (GH) secretory bursts and the halflife of endogenous GH in healthy men. J Clin Endocrinol Metab. 1991;73:1081-8. Hartman ML, Pezzoli SS, Thorner MO. Inverse correlation of adiposity and the growth hormone response to growth hormone releasing factor (GRF) (l-29)-NHz in nonobese men. Clin Res. 1990;3&528A.

Downloaded from jcem.endojournals.org by on October 30, 2008