.Journal of Clinical Endocrmology and Metabolism
Copyright (* 199‘2 by The Endocrine Society
Vol.15, No. 1
Prrnted in U.S.A.
Effect of Low and High Intensity Exercise on Circulating
Growth Hormone in Men*
NANCY E. FELSING, JO ANNE BRASEL,
AND
DAN M. COOPER?
Divisions
of
Respiratory and Critical Care, and Endocrinology, Department
of
Pediatrics, Harbor- UCLA
Medical Center, UCLA School
of
Medicine, Torrance, California 90509
ABSTRACT
We hypothesized that circulating GH would increase only if a
threshold of work intensity [corresponding to the anerobic or lactate
threshold (LT)] was exceeded. Ten healthy male volunteers (18-35 yr)
first performed ramp-type progressive cycle-ergometer exercise to de-
termine the LT and the maximal oxygen uptake. On subsequent morn-
ings after an overnight fast, each subject performed bouts of 1, 5, and
10 min constant work rate exercise of either high intensity (above LT)
or low intensity (below LT). A l-h interval separated exercise bouts.
Gas exchange (breath-by-breath), GH, immunoreactive insulin, glu-
cose, lactate, pyruvate, and epinephrine and norepinephrine were meas-
ured at regular intervals. After the lo-min bouts of high compared with
low intensity exercise, lactate was 7.2 f 3.7 mmol/L us. 1.4 + 1.3,
P <
0.05; epinephrine was 1,113 & 519 pmol/L us. 496 + 273,
P
< 0.05; and
norepinephrine was 7.89 & 3.45 nmole/L us. 2.83 * 1.34,
P < 0.05).
GH
did not increase significantly from preexercise baseline during low
intensity exercise (e.g., GH after lo-min low intensity exercise changed
from baseline values by 1.5 & 2.0 fig/L, NS). Although lactate was
elevated after 5-min of high intensity exercise, peak GH was signifi-
cantly elevated (mean increase above baseline of 7.7 + 2.4 pg/L,
P <
0.05) only after 10 min of high intensity exercise (increases in 9 of 10
subjects). The GH increase occurred despite simultaneous increases in
both IRI and glucose. A minimum duration of 10 min, high intensity
exercise consistently increased circulating GH in adult males. (J Clin
Endocrinol Metab
75: 157-162, 1992)
E
XERCISE AFFECTS GH secretory patterns, and this GH
effect likely plays a role in growth and in the training
phenomenon (l-4). The exercise duration or intensity that
will elicit a reproducible and substantial GH pulse in humans
is not precisely known. There are reasons to believe that the
GH response to increasing work intensity is nonlinear in
nature and may be characterized by a threshold: The purpose
of this study was to test whether or not a work-intensity
threshold could be identified for the GH response.
Although the magnitude of the GH response is related to
work intensity (5-8), there is great variability in the reported
amplitude and duration of exercise-induced GH response in
humans. The variability might be explained by the fact that
in previous studies, some subjects exercised below, whereas
others worked above a threshold for GH release. The GH
response may be correlated to the lactate response to exercise
(9-11) which increases in a nonlinear manner with work
intensity (12). Finally, other stress responses, e.g. circulating
catecholamines,
are disproportionately elevated during
heavy compared to light exercise (13, 14). We hypothesized
that exercise would elicit a significant GH pulse only when
the work rate exceeded the anaerobic or lactic acid threshold
(high intensity exercise). The latter refers to the work rate
above which circulating lactic acid increases (15, 16).
Table 1 reviews 10 studies of the acute GH response to
exercise (5, 17-25). It highlights the diversity of approaches
Received September 23, 1991.
Address requests for reprints to: Dan M. Cooper, M.D., N-4, 1000
West Carson Street, Torrance, California 90509.
* This work was supported by USPHS Grants HD-26939 and HL-
11907 and by General Clinical Research Grant RR-00425.
t Recipient of the Career Investigator Award of the American Lung
Association.
used to determine exercise intensity. In most studies, the
work rate chosen is equivalent to 60-70% of the subject’s
maximal oxygen uptake (irO,max). By and large, these pro-
tocols represent submaximal high intensity exercise since the
lactate threshold (LT) occurs between 40-60% of vO,max in
healthy subjects (26). But some of the subjects may have
exercised above and others below their LT for the following
reasons. Investigators often used predicted rather than meas-
ured vOzmax, and most did not measure the LT in individual
subjects. This, combined with the large intersubject variabil-
ity of ir02max and the LT, may have confounded the inves-
tigators’ ability to precisely determine work rate intensity.
The duration of the exercise input in previous studies
tended to be long (the mean of Table 1 is 44 min). In contrast,
naturally occurring patterns of activity in adults and children
are shorter. Thus, we specifically examined the effects of
short bursts of exercise more likely to mimic physiologically
significant patterns of activity.
Subjects and Methods
Subjects
(Table 2)
Ten healthy adult male volunteers participated in the study. They
ranged in age from 18-35 yr old (mean 27 + 5 yr). None of the subjects
were smokers, suffered from chronic diseases, or took drugs or medica-
tions, None of these individuals trained as competitive athletes, but
most participated in some form of regular exercise. The study was
approved by the institutional Human Subjects’ Committee, and each
participant granted informed consent.
Protocol
The protocol consisted of three exercise sessions each performed on
different days, separated by at least 1 week. On day one each volunteer
157
158
FELSING
ET AL.
JCE & M. 1992
Vol75.Nol
TABLE 1. Representative previous studies of the effect of exercise on circulating GH
Lactate thresh-
old
Work intensity
Duration
(min)
(17)
1990 Predicted Not measured
(18)
1990 Predicted Not measured
(19)
1990 Measured Not measured
(20)
1990 Predicted Not measured
(21)
1989
Measured Measured
(22)
1988 Measured Not measured
(23)
1987 Not measured
Not measured
(24)
1986
Measured Not measured
(5) 1985 Measured Not measured
(25)
1981 Not measured Not measured
66% irO,max
70% VOzmax
65% $‘O,max
60% vO,max
Equal to LT
70% V02max
1, 1.5, and 2 watt/kg for
5-min periods
60% v02max
63 (10 min), 86 (10
min), 100 (5-7 min)
55 watt/m*
6
15
-90
30
-80
60
15
60
-26
60
TABLE 2. Subject age, weight, height, and exercise gas exchange characteristics
Subject
Age (rd Wt (Kg)
Ht (cm)
LT iiO,~ClX LT Low WR High WR
(% VO,max)
(ml/min kg) (ml/min kg) (% VOZm3X) (% VO*max)
1 35 80 178 47 38 18 18 68
2 24 75 180 59 37 22
24
14
3 28 86 185 48 49 23 25 75
4 27 89 185 64 41 26
23 58
5 32 81 178 70 49 35 26 78
6 21 81 190 50 43 21
20 69
7 29 93 196 80 38 30 32 82
8 18 14 187 54 40 22
24 74
9 27 58 173 45 46 21
17 68
10 29 58 170 43 48 21
23 73
Mean 27 78 182 56 43 24 23
72
SD 5 11 8 12 5 5
4 6
performed progressive ramp-type cycle ergometry to determine the
VO>max and the LT (see below). The next two sessions consisted of l-,
5-, and lo-min of constant work rate exercise on the cycle ergometer.
An hour of rest separated each exercise burst. For one session, the work
rate chosen corresponded to 50% of the difference between the subject’s
LT and VOzmax (high intensity exercise), whereas for the other session
all work rates corresponded to 50% of the subject’s lactate threshold
(low intensity).
For a given session, subjects always performed a fixed sequence of
exercise duration (l-, 5-, IO-min), but the order of the exercise intensity
(i.e.
high intensity day, low intensity day) was randomized. We chose
this protocol because we anticipated that lo-min exercise bouts would
be the most likely to elicit GH responses and it was performed last to
minimize possible confounding effects of a prior GH pulse on subse-
quent pulses. We chose a l-h interval between exercise bouts. This was
a balance between, on the one hand, allowing sufficient time for gas
exchange parameters, lactate, and hormones levels to return to baseline,
and, on the other hand, minimizing the possibility of spontaneous GH
pulses and too prolonged a fasting period.
combined dead space of 90 mL. O2 and CO* tensions were determined
by mass spectrometry from a sample drawn continuously from the
mouthpiece at 1 mL/s. The inspired and expired volumes and gas tension
signals upderwent analog-to-digital conversion, from which oxygen
uptake (V02) (standard temperature pressure dry), CO, production
(VCO*) (standard temperature pressure dry), and minute ventilation (v,)
(body temperature pressure saturated) were calculated on-line, breath-
by-breath as previously described (27). The breath-by-breath data were
then interpolated to l-s time intervals.
Noninvasive determination of LT and vO,max
The subjects arrived at approximately 0800 h on the morning of a
low or high intensity exercise study. We chose the morning to perform
the study because fewer naturally occurring GH pulses are seen during
the morning hours (after about 0800-0900 h). They were instructed to
refrain from any exercise and to remain fasted for at least 12 h before
the study. An antecubital venous catheter was placed for intermittent
blood sampling. Baseline blood samples were taken at 10 and 5 min
before the first (1 min) exercise burst. The subject performed the exercise,
then samples were taken every 10 min during the rest period. This was
repeated after the 5- and lo-min bursts. An additional sample was
obtained during exercise (at 5 min) during the last (lo-min) exercise
period. Breath-by-breath gas exchange measurements were made 5 min
before, during, and 10 min after each exercise period.
The LT and 7j02max were measured noninvasively from the gas
exchange data obtained during the progressive exercise. The lactate
threshold was defined as the V02 at which the ventilatory equivalent
for O2 ($,/$O,) and the end tidal O2 (PetOz),in<reased without an
increase in the ventilatory qquivalent for CO* (VE/VC02) and the end
tidal CO2 (PetCO>) (28). VOzmax was defined as the highest VOZ
achieved by the subject.
GH,
glucose,
insulin, lactate, and catecholamines
Gas exchange measurements
The subjects breathed through a mouthpiece connected to a low
impedance turbine volume transducer and a breathing valve with a
An in-house RIA was used to measure GH using WHO standard no.
66/217, antisera generated in-house, and hGH from NIDDK for iodi-
nation purposes. The GH intraassay variability is less than lo%, inter-
assay variability is 12.6%, and the sensitivity is 0.5 pg/L. Insulin was
also measured using an in-house RIA using standard from Wellcome
equated to first IRP 66/304, antiporcine antibody from ICN and porcine
insulin from Lilly for iodination. The insulin intraassay variability is less
than lo%, interassay variability is 11.5%, and the sensitivity 7 pmol/L.
Glucose was measured using the Abbott bichromatic analyzer using the
Abbott UV glucose kit. The glucose intraassay variability is 2.1% and
the interassay variability is 2.4%. Lactate was measured spectrophoto-
metrically using the Behring Stat-pack rapid lactate test. The lactate
intraassay variability is 2.8%, the interassay variability is 3.5%, and the
GH AFTER LOW- AND HIGH-INTENSITY EXERCISE
sensitivity is 0.55 mmol/L. Pyruvate was measured enzymatically using
the Perkin Elmer luminescence spectrophotometer. Pyruvate intraassay
variability is 4% and interassay variability is 12%. Catecholamines
[norepinephrine (NE), epinephrine (E)] were measured by the radioen-
zymatic method. For NE the intraassay variability is less than 10% and
the interassay variability is 10.6%, and the sensitivity is 0.12 nmol/L.
For E the intraassay variability is less than lo%, interassay is 14.6%,
and the sensitivity is 109 pmol/L.
Statistical analysis
Repeated measures analysis of variance was used to describe the
patterns of the multiple samples of GH, other hormones, and substrates.
Separate analyses were performed for the preexercise values, the peak
value, and the A (i.e. the difference between peak and preexercise
values). The preexercise values were taken as the level of hormone or
substrate immediately before the l-, 5-, or lo-min exercise bout. When
analysis of variance (ANOVA) was found to be significant, Duncan’s
Multiple Range test was used to determine intergroup significance.
Unless otherwise stated, values are presented as mean +_ SD. A P value
less than 0.05 was considered significant.
Results
Gas exchange parameters
The individual subject age, weight, height, and exercise
gas exchange characteristics are shown in Table 2. As ex-
pected, the VOz reached a steady state during the 5- and lo-
min low intensity exercise bouts but not for high intensity
exercise (Fig. 1). V02, of course, increased significantly with
exercise duration for both low and high intensity exercise.
With 10 min of exercise, the peak VO, was 4 and 9 times
greater, on average, than baseline for low and high intensity
exercise, respectively.
Lactate and pyruvate (Fig. 2)
As expected, lactate concentrations were increased during
all durations of high intensity exercise. There were no sub-
stantial differences between the two baseline lactate values;
however, the lactate levels immediately before the lo-min
high intensity exercise bout were slightly but significantly
higher (P < 0.01) than the baseline levels and than those
before the 5-min exercise bouts. The coefficient of variation
3.0 7
Time (min)
FIG. 1. VO, during low and high intensity exercise in a representative
subject. Note that VO, did not achieve a steady state during high
intensity (above LT) exercise even though the work rate was constant.
7
$6
E 5
-54
z 3
;2
4 1
0
1 min 5 min 10 min
FIG. 2. Mean peak serum lactate levels (minus baseline) and SE after
1, 5, and 10 min of low (hollow
bars)
and high (hatched
bars)
intensity
exercise. * indicates P < 0.05.
Time (min)
FIG. 3. Mean GH and SE (closed circles) and VO, (lines) after 1, 5,
and 10 min of low (left panel) and high (right panel) intensity exercise.
* indicates
P
<
0.05.
Clear vertical
bars
represent the exercise bouts.
GH significantly increased only after 10 min of high intensity exercise.
for peak lactate among the subjects was 57%. Qualitatively
similar results were found for the lactate-to-pyruvate ratios.
GH (Fig. 3)
Spontaneous GH pulses (judged by inordinately elevated
preexercise GH with patterns suggesting upward or down-
ward slopes) occurred in
only
one subject. There were no
significant differences between the two baseline GH levels.
GH responses to exercise were quite variable in magnitude
among the subjects; the ratio of peak pulse to baseline ranged
from 0.7 to 14 after low intensity exercise and from 0.5 to
51 after high intensity exercise. After low intensity exercise,
exercise-associated increases in GH for the group as a whole
were not statistically significant. Two of the 10 subjects had
disproportionately greater GH pulses than the others and
their GH pulses account by and large for the increase in the
mean GH concentration after 5 and 10 min of exercise. After
the lo-min high intensity bout there was a significant GH
pulse in 9 of 10 subjects (mean peak of 7.7 & 2.4 pg/L ZIS.
a
mean baseline of 1.7 f 2.4 pg/L, P < 0.05). The small increase
in GH after 5-min of high intensity exercise was not statis-
tically significant. The coefficient of variation for the peak
GH pulse after 10 min in the high intensity range was 85%.
After the lo-min high intensity exercise bout, the peak GH
pulse occurred at a mean of 29 + 12 min after the onset of
exercise and occurred significantly (P < 0.05) later than those
FELSING
ET AL.
JCE & M. 1992
Voll5.Nol
160
of lactate (15 min), E (
min).
10 min), NE ( 11 min), and insulin (21
Glucose and insulin (Fig. 4)
After low intensity exercise, glucose increased by 0.39 +
0.11 mmol/L (P < 0.05) for 5-min exercise and by 0.28 +
0.11 mmol/L for lo-min bouts. With high intensity exercise,
significant increases (P < 0.05) were found after the I-min
(0.28 f 0.06 mmol/L), 5-min (0.39 + 0.11 mmol/L), and lo-
min (0.61 + 0.11 mmol/L) protocols. Low intensity exercise
had no significant effect on insulin. During high intensity
exercise, insulin increased significantly (P < 0.05) after the
1-min (29 -C 7 pmol/L), 5-min (50 + 14 pmol/L), and lo-
min (29 + 7 pmol/L) protocols.
Catecholamines (Fig. 5)
During low intensity protocols, significant increases were
found for E after the lo-min exercise bout and for NE after
the l- and IO-min bouts. Both E and NE increased with the
high intensity protocols after the 5- and IO-min bouts. The
magnitude of the increase in catecholamines after low inten-
sity was substantially less than that observed after the high
intensity protocols. The coefficient of variation for peak NE
among the subjects was 45% and for epinephrine was 49%.
2 200 -
2 s
E
.3
150-
$6
E
.&
2
100 -
-5
z
c 8
50.5,
cJi!...jii.1 Ji.,&L
*
. l
0 40 60 120 160 200
0 40 60 120 160 200
Time (min)
FIG. 4. Mean glucose and SE (open circles) and mean insulin and SE
(closed circles) after 1, 5, and 10 min of low (left panel) and high (right
panel) intensity exercise.
* indicates P < 0.05 for peak over baseline
values. Clear vertical bars represent the exercise bouts. Both insulin
and glucose increased consistently after high intensity exercise.
‘t
1000
c i
I-min 5-min IO-min
I-min 5-min lo-min
FIG. 5. Mean peak values (minus baseline) t SE for NE (left panel)
and E
(right
panel) after 1, 5, and 10 min of low (hollow bars) and high
(hatched bars) intensity exercise. * indicates P < 0.05 compared to low
intensity exercise.
Discussion
The gas exchange and lactate data demonstrate that we
achieved the goal of identifying low- and high-intensity
exercise in the subjects. The lo-min period of high-intensity
(above the lactate threshold) exercise consistently resulted in
bursts of GH secretion in adult males. In contrast, low
intensity exercise, including the lo-min protocols, did not
elicit significant GH responses. Despite rigorous control over
the work rate, our data, like most previous studies, revealed
great subject-to-subject variability in the peak GH response
achieved (nb., the coefficient of variation among the subjects
was higher for GH than for lactate and catecholamines).
Thus, whereas there does appear to be a minimum threshold
of exercise duration and intensity necessary for a GH pulse,
exercise intensity alone cannot entirely predict the amplitude
and duration of the subsequent GH response.
The onset of the GH response to exercise was later and
less consistent than the gas exchange, lactate, and other
hormonal responses. The elevation in GH induced by exer-
cise lasted far longer than both the lo-min exercise bout
itself and the other substrate and hormonal responses stud-
ied. The time required to achieve a GH response in our study,
between 5 and 10 min, was similar to that observed by
Sutton and Lazarus (9) who used a 20-min protocol. The
peak GH in their study also occurred at about 30 min after
the onset of exercise,
i.e.
after the exercise bout was com-
pleted. It appears that 10 min of high intensity exercise are
necessary to reliably stimulate pituitary secretion of GH.
The disappearance of GH from the circulation follows a
first-order exponential decay (29-32), and reported half-
times range from 8.9 min (30) to as high as 27 min (32). A
variety of techniques, including deconvolutional analysis (29,
33), has been used to quantify pituitary GH secretion during
spontaneous pulses. We constructed a simple single com-
partment model in which GH is transported from the pitui-
tary to the circulation with first-order kinetics, and follows a
first-order disappearance from the plasma. Iterative nonlin-
ear curve-fitting techniques (34) were used to calculate the
time constants and the amount of GH secreted from the
mean GH levels during and after the IO-min high intensity
protocol. The following equation was used:
where A represents the GH released consequent to the
exercise stimulus, 71 corresponds to the disappearance
time constant (equivalent to the half-time divided by 0.69)
for GH from the plasma compartment, r2 corresponds to
the time constant of GH release from the pituitary com-
partment to the plasma, and t is the time in min. (A 5-
min delay was included in the model).
A reasonably good fit was seen when 27 min was used
as the value for the half-time of plasma GH disappearance
(Fig. 6). The analysis predicted a total GH pulse of 0.061
mg when using the mean weight of our subjects (78 kg)
and a GH volume of distribution of 4.4% (35). The model
suggests a t% of the GH release into the plasma of 11 f
3 min, and a mean GH secretory rate over four half-lives
GH AFTER LOW- AND HIGH-INTENSITY EXERCISE
I -_I
0 10
20 JO 40 50 60 70 60
Time (min)
FIG. 6. Mean + SE for GH (closed circles) during and after 10 min of
high intensity exercise. The dashed line is a cubic splines best-fit curve
for the GH data. The solid line is the best fit curve for the exponential
model as described in the text.
of 0.41 pg/L. min, a value comparable in magnitude to
those found from spontaneous pulses (36).
The data show that the GH response to exercise likely
has a different mechanism than other known physiolog-
ical stimuli. Hypoglycemia and/or rapid falls in glucose
concentration, for example, cause GH release (37), but, as
shown in Fig. 4, subjects remained euglycemic and glu-
cose concentrations tended to increase. Classically, insulin
can be used to stimulate GH by inducing hypoglycemia.
In our studies, insulin increased significantly following
short bursts of high intensity exercise with no evidence
of hypoglycemia. It has been reported that insulin de-
creases during long term exercise (e.g. greater than 40
min) (38), but recent data in human subjects during and
after short term high intensity exercise demonstrate either
no change or, consistent with our results, increases in
plasma insulin (39-41). In addition, both insulin and GH
were increasing simultaneously following exercise (peak
insulin occurred at 21 min; peak GH at 29 min). It would
be difficult to conclude from these observations that in-
sulin either directly or indirectly stimulated the exercise-
induced GH pulses.
Both E and NE were significantly elevated after both
5- and lo-min of high intensity exercise, while GH was
significantly elevated only after lo-min high-intensity
exercise. The E and NE responses reflect alterations in
neuroadrenergic control known to occur during graded
exercise (42, 43); namely, a reduction in parasympathetic
tone, an increase in sympathetic tone, and stimulus of
adrenal production of epinephrine. The discrepancy be-
tween the GH and catecholamine responses, therefore,
suggests that neuroadrenergic inputs to the hypothalamus
or the pituitary, i.e. stress, are not the only modulators of
the exercise-associated GH response.
There is mounting evidence to support the idea that
exercise-induced GH is important in somatic growth and
in the effect of physical training on muscles. Borer et
al.
(44) demonstrated that exercising hamsters grew at faster
rates than did sedentary animals, and the exercising ani-
mals had greater frequency and amplitude of spontaneous
GH pulses. More recently, Grindeland et al. (1) studied
the effect of exercise and administration of exogenous
GH on muscle growth in hypophysectomized rats re-
covering from hindlimb suspension. Their preliminary
data show that the most marked increases in muscle mass
occurred with the combination of exercise and GH. In
contrast, the training effect in humans can occur with
exercise protocols that are above or below the subject’s
lactate threshold (45); and DeVol and co-workers (46)
showed that training induced muscle hypertrophy with
increases in muscle tissue insulin-like growth factor-I
messenger RNA even in hypophysectomized animals.
Apparently, the growth and hypertrophy observed in
response to exercise is modulated by both GH-dependent
and GH-independent processes.
Ten minutes of constant work rate, high intensity ex-
ercise is a minimum stimulus for consistent GH release in
adult males. Whether or not such patterns of activity
represent naturally occurring, physiologically important
GH stimuli remains unknown. It is intriguing that the
character of the exercise stimulus may be as important as
the total work done in eliciting the GH response. Van-
helder et
al.
(7) demonstrated that a series of 1-min bursts
of very high intensity exercise resulted in a greater GH
response than constant work rate exercise (20 min) in
which the work expenditure and duration of the two
protocols were the same. The importance of the pattern
of exercise is highlighted by recent findings that the
pulsatile nature of GH release may optimize its overall
effect on growth (47). Finally, our data add to the body
of evidence that GH pulses in response to exercise, unlike
spontaneous GH pulses, are accompanied by increases in
other tissue growth mediators (insulin, catecholamines).
Perhaps this hormonal “milieu” is as important to growth
and training as is the elevation in GH itself.
1.
2.
3.
4.
5.
6.
7.
References
Grindeland RE, Roy R, Edgerton VR, et al. 1991 Exercise and
growth hormone have synergistic effects on skeletal muscle and
tibias of suspended rats [Abstract]. FASEB J. 5:A1037.
Borer KT. 1980 Characteristics of growth-inducing exercise. Physiol
Behav. 24:713-720.
Kelly PJ, Eisman JA, Stuart MC, Pocock NA, Sambrook PN,
Gwinn TH. 1990 Somatomedin-C, physical fitness, and bone den-
sity. J Clin Endocrinol Metab. 70:718-23.
Poehlman ET, Copeland KC. 1990 Influence of physical activity
on insulin-like growth factor-I in healthy younger and older men. J
Clin Endocrinol Metab. 71:1468-73.
Naveri H. 1985 Blood hormone and metabolic levels during graded
cycle ergometer exercise. Stand J Lab Clin Invest. 45:599-603.
VanHelder WP, Casey K, Goode RC, Radomski WM. 1986 Growth
hormone regulation in two types of aerobic exercise of equal oxygen
uptake. Eur J Appl Physiol. 55:236-9.
VanHelder WP, Goode RC, Radomski MW. 1984 Effect of anaer-