| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Department of Intensive Care Medicine, University Hospital Gasthuisberg, University of Leuven (G.V.d.B., F.W., P.W.), B-3000 Leuven, Belgium; Kolling Institute of Medical Research, Royal North Shore Hospital, Sydney University (R.C.B.), St. Leonards, New South Wales 2065, Australia; the Department of Medicine, Division of Endocrinology, Tulane University Medical Center (C.Y.B.), New Orleans, Louisiana 70112-2699; and the General Clinical Research Center and Department of Medicine, Division of Endocrinology, University of Virginia Health Sciences Center (J.D.V.), Charlottesville, Virginia 29908
Address all correspondence and requests for reprints to: Greet Van den Berghe, M.D., Ph.D., Department of Intensive Care Medicine University Hospital Gasthuisberg, University of Leuven, B-3000 Leuven, Belgium. E-mail: greta.vandenberghe{at}uz.kuleuven.ac.be
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Despite identical mean serum GH concentrations, total GH output, GH half-life, and number of GH pulses, critically ill men paradoxically presented with less pulsatile (mean ± SD pulsatile GH fraction, 39 ± 14% vs. 67 ± 20%; P = 0.002) and more disorderly (approximate entropy, 0.946 ± 0.113 vs. 0.805 ± 0.147; P = 0.02) GH secretion than women. Serum IGF-I, IGFBP-3, and acid-labile subunit (ALS) levels were low in patients compared with controls, with male patients revealing lower IGF-I (P = 0.01) and ALS (P = 0.005) concentrations than female patients. Correspondingly, circulating IGF-I and ALS levels correlated positively with pulsatile (but not with nonpulsatile) GH secretion. Circulating levels of GH-binding protein and IGFBP-1, -2, and -6 were higher in patients than controls, without a detectable gender difference. In female patients, PRL levels were 3-fold higher, and TSH and cortisol tended to be higher than levels in males. In both genders, estrogen levels were more than 3-fold higher than normal, and testosterone (2.25 ± 1.94 vs. 0.97 ± 0.39 nmol/L; P = 0.03) and dehydroepiandrosterone sulfate concentrations were low. In male patients, low testosterone levels were related to reduced GH pulse amplitude (r = 0.91; P = 0.0008). GH responses to GHRH were relatively low and equal in critically ill men and women (7.3 ± 9.4 vs. 7.8 ± 4.1 µg/L; P = 0.99). GH responses to GHRP-2 in women (93 ± 38 µg/L) were supranormal and higher (P < 0.0001) than those in men (28 ± 16 µg/L). Combining GHRH with GHRP-2 nullified this gender difference (77 ± 58 in men vs. 120 ± 69 µg/L in women; P = 0.4).
In conclusion, a paradoxical gender dissociation within the GH/IGF-I axis is evident in protracted critical illness, with men showing greater loss of pulsatility and regularity within the GH secretory pattern than women (despite indistinguishable total GH output) and concomitantly lower IGF-I and ALS levels. Less endogenous GHRH action in severely ill men compared with women, possibly due to profound hypoandrogenism, accompanying loss of the putative endogenous GHRP-like ligand action with prolonged stress in both genders may explain these novel findings.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Stress responses within the immune system appear to differ in male and female experimental animals (9, 10), which has been attributed in part to distinct alterations within the hypothalamic-pituitary-adrenal axis and/or effects of sex steroids (10, 11). Moreover, in healthy rodents and humans, a sexual dimorphism in the GH secretory pattern is thought to determine gender differences in metabolism. Female rats exhibit a more irregular and less pulsatile GH secretory pattern compared with males (12, 13), a difference that may be related to lower insulin-like growth factor I (IGF-I) gene expression and serum levels in the female, in turn contributing to the gender differences in growth (13, 14, 15, 16). Women, both premenopausal and elderly, also display more disorderliness in the GH secretory pattern compared with men of the same age, which may determine differences in body composition (17, 18, 19, 20, 21). In both genders, aging is associated with a progressive loss of regular, pulsatile GH secretion (19). Sexual dimorphism appears largely dependent upon priming effects of androgens on hypothalamic signaling and on the modifying influences of available estrogens and androgens during adult life.
In view of the specific metabolic dysfunction in protracted critical illness and the apparent survival benefit associated with female gender, we here investigated whether a gender dissociation is evident within the GH/IGF-I axis in this particular condition of chronic severe stress. The hypothesis was tested by comparison of nocturnal GH secretory patterns and levels of GH-binding protein (GHBP), IGF-I, and IGF-binding proteins (IGFBPs) in matched male and female patients with protracted critical illness. To aid in interpretation, we also measured sex steroids and other anterior pituitary hormones, and we assessed GH secretory responses to the single and combined administration of GHRH and GH-releasing peptide-2 (GHRP-2).
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In a first dynamic study, patterns of nocturnal GH secretion
were compared in nine male and nine female patients experiencing
prolonged critical illness, matched for age (mean ±
SD age, 67 ± 11 and 67 ± 15 yr), body mass
index (25 ± 4 and 29 ± 7 kg/m2),
type, severity (Apache II score, 16 ± 7 and 17 ± 5), and
duration (35 ± 24 and 21 ± 6 day) of illness, type and
caloric load (27 ± 6 and 27 ± 7 Cal/kg·24 h) of
continuously administered feeding, blood glucose (7.1 ± 0.7 and
7.7 ± 2.2 mmol/L), and triglycerides (2.1 ± 0.8 and
2.3 ± 1.8 mmol/L) as well as any concomitant medications (Table 1
). Patients depending on intensive care
(including mechanical ventilatory support) for at least 2 weeks and
with an expected stay in the intensive care unit of at least another 2
weeks were eligible for participation in this study. Further inclusion
criteria were a stable condition without dopamine treatment for at
least 72 h, in view of the pronounced suppressive effect dopamine
exerts on GH secretion in such patients (22). Exclusion criteria were
age less than 18 yr; preexisting neurological, psychiatric, metabolic,
or endocrine disease; intracranial lesions; clinically significant
liver failure (prothrombin time, <30%); renal failure requiring
replacement therapy; concomitant treatment with glucocorticoids,
estrogens, somatostatin, thyroid hormones, Ca2+
entry blockers, clonidine, amiodarone, etomidate, dopamine agonists, or
antagonists; and the use of iodine in antiseptic dressings or as iv
contrast agents.
|
For measures of GHBP, IGF-I, IGFBPs, and PRL, patients were compared
with a healthy, community-living, age- and body mass index-matched
control group consisting of 50 men and 50 women (Table 1
). Exclusion
criteria for participation as a healthy control were acute and chronic
illnesses, including neurological, psychiatric, metabolical, or
endocrine diseases, and use of the drugs mentioned above as exclusive
for the patients.
In a second study, we investigated GH, PRL, TSH, and cortisol responses
to GH secretagogues [random iv administration of a bolus of GHRH (1
µg/kg), GHRP-2 (1 µg/kg), or GHRH plus GHRP-2 (1 plus 1 µg/kg)]
in a comparable group of prolonged critically ill (15 male and 15
female) patients (age, 67 ± 9 yr), with each secretagogue group
consisting of 5 men and 5 women carefully matched as described above
(Table 1). GH, PRL, TSH and cortisol concentrations were measured
before and 20, 40, 60, and 120 min after administration of the GH
secretagogues.
Ethical aspects
Informed consent was obtained from volunteers and from a first degree relative of patients before inclusion. The study protocols were approved by the ethical review board of the University of Leuven School of Medicine (Leuven, Belgium).
Blood sampling
Blood samples were collected through an arterial line inserted in patients for clinical purposes independently of this study. The Edwards VAMP system (Baxter Healthcare Corp., Irvine, CA) was used, which permitted withdrawal of undiluted blood samples from an indwelling catheter without undue blood loss. Healthy volunteers were sampled by venous puncture. Blood was collected into Vacutainer (Becton Dickinson Vacutainer Systems Eur.; Meylan Cedex, France) tubes; after clotting and centrifugation, the serum was kept frozen at -20 C until assay.
Assays
Samples from the same patient were always processed within one assay run. In the nocturnal profiles of the first study, the serum concentrations of GH were measured by immunoradiometric assay (IRMA), using the Nichols Institute Diagnostics hGH immunoassay 100T kit 40–2155(40–2155, San Juan Capistrano, CA). The intraassay coefficient of variation was 4.2% at 1.4 µg/L and 2.8% at 12.2 µg/L. The detection limit of this immunoradiometric assay was 0.2 µg/L. In the second study (GH responses to GH secretagogues), GH was measured by RIA, using a polyclonal antibody (25). The intraassay coefficient of variation was 7.3% at 6.7 µg/L and 4.6% at 14.4 µg/L. The detection limit was 1.2 µg/L. All processed samples had detectable GH values using these assays.
The serum concentrations of total IGF-I were measured by RIA, after acid-ethanol extraction. The intraassay coefficient of variation was 10.1% at 95 µg/L and 5.5% at 474 µg/L. The between-assay coefficient of variation was 14.8% at 109 µg/L and 10.1% at 389 µg/L.
Serum concentrations of IGFBP-1, IGFBP-2, IGFBP-3, IGFBP-6, and ALS were determined by RIA, as previously described (26).
The serum GHBP levels were measured by enzyme-linked immunosorbent assay (DSL Kit, Diagnostics Systems Laboratories, Inc., Webster, TX). The intraassay coefficient of variation was 5.4% at 0.82 µg/L, 1.8% at 3.83 µg/L, and 5.1% at 10.4 µg/L.
The serum TSH concentrations were measured by IRMA using the TSH Riabead II (Abbott Laboratories, North Chicago, IL). The intraassay coefficient of variation was 4.3% at 1.2 mIU/L and 2.2% at 7.0 mIU/L. The detection limit was less than 0.02 mIU/L. Normal nocturnal values ranged between 1–7 mIU/L.
The serum concentrations of PRL were measured by immunoradiometric assay using the PRL IRMA kit (Medgenix, Fleurus, Belgium). The intraassay coefficient of variation was 6.2% at 6.6 µg/L and 4.7% at 46.4 µg/L. Normal values are less than 25 µg/L.
The serum concentrations of cortisol were measured by RIA after extraction with dichloromethane. The intraassay coefficient of variation was 3.1% at 417 nmol/L. Normal morning values are 200–700 nmol/L, and normal nocturnal values when asleep are less than 50 nmol/L.
Serum concentrations of E2 were determined by a heterogeneous competitive magnetic separation assay (Technicon Immuno 1 System, Bayer Corp., Tarrytown, NY). The intraassay coefficient of variation was 3.9% at 268 pmol/L (n = 114). The detection limit was 37 pmol/L. Normal mean ± SD (median and ranges) values are 74.5 ± 28.1 (72.4 and 8.2–146) pmol/L for age-matched men and 71.8 ± 30.1 (57.2 and 10.5–104) pmol/L for age-matched women (21).
Serum concentrations of E1 were determined by RIA (DSL-8700 Estrone RIA kit, Diagnostics Systems Laboratories, Inc.). The intraassay coefficient of variation was 5.6% at 376 pmol/L (n = 10). The detection limit was 4.4 pmol/L. Normal values for age-matched men and women range between 52–379 pmol/L, with a median of 120 pmol/L.
The serum DHEAS concentrations were measured by RIA using the DSL DHEAS RIA Kit (Diagnostics Systems Laboratories, Inc.). The intraassay coefficient of variation was 5.1% at 0.6 µmol/L and 4.1% at 6.1 µmol/L. The detection limit was 0.05 µmol/L. Normal values in men range between 2–10 µmol/L and in women between 3–12 µmol/L.
Serum concentrations of androstenedione and testosterone were determined by RIA, as previously described (27). Normal values for testosterone in men range between 9–35 nmol/L (mean, 16.2 nmol/L) (21) and in women between 0.5–3 nmol/L (mean, 1.3 nmol/L) (21).
Data analysis
The time series of sequential serum GH concentrations measured overnight was transformed into pituitary secretion profiles by adjusting for endogenous GH half-life using multiple parameter deconvolution analysis (23). This method is designed to compute hormonal half-life and the number, amplitude, and mass of underlying pituitary secretory bursts and to estimate tonic (basal or nonpulsatile) secretion (23). Besides mean serum concentrations, the following parameters were calculated for each hormonal profile in each subject: basal secretion rate (estimated as the amount of hormone continuously released from the pituitary to achieve serum concentrations approximating the mean of the lowest 5% of all serum GH values observed [micrograms per L of distribution volume (Lv) and per min], amplitudes (maximal secretory rate; micrograms per Lv and per min) and temporal positions of all secretory bursts, mass of hormone secreted per burst (estimated as the area of the resolved secretion burst; micrograms per Lv), and pulsatile production (calculated as the product of the number of secretory bursts and the mean secretory burst mass over the time interval considered; micrograms per Lv over 9 h). Total nocturnal GH production (micrograms per Lv) was calculated as the sum of pulsatile GH production (micrograms per Lv) over 9 h and the nonpulsatile GH secretion over 9 h [basal GH secretion rate (micrograms per Lv/min) x 540 min]. The proportion of pulsatile secretion (percentage) was calculated as the pulsatile production over 9 h divided by the total nocturnal GH production, multiplied by 100. In healthy humans, both males and females, more than 90% of GH is released in a pulsatile fashion (28, 29).
As an approximate measure of IGF-I and ALS responsiveness to GH, the IGF-I/GH and ALS/GH ratios [either IGF-I (ALS)/total GH output ratio or IGF-I (ALS)/pulsatile GH ratio) were calculated, while recognizing that the dynamics of the two analytes, and the volume base on which their concentrations were calculated, are different.
The regularity in the GH secretory pattern was quantified by the ApEn statistic (24). ApEn measures the logarithmic likelihood that runs of patterns that are similar remain similar on next incremental comparisons. ApEn assigns a single nonnegative number to a time series with larger values corresponding to greater irregularity. ApEn is stable to small changes in noise characteristics and to infrequent and significant artifacts. It detects variations in episodic behavior not necessarily reflected in changes in peak occurrences or amplitudes. Additionally, ApEn provides a direct barometer of feedback system changes in many coupled systems. The calculation of ApEn was performed as previously reported (24). As ApEn will generally increase with increasing process noise (and increasing intraassay variation), it is important to compare datasets with similar assay coefficients of variation, as performed here. To this end, we used a tolerance/threshold for ApEn of 0.2 times the individual series between-sample SD and a window/range of m = 1 consecutive samples over which to test for pattern reproducibility, as is appropriate for time series of less than 150 samples (30). A normal mean ± SD ApEn scores of 0.60 ± 0.20 has been calculated for age-matched men, and a score of 0.81 ± 0.23 has been calculated for age-matched women (21).
Results are presented as the mean ± SD unless indicated otherwise. Single measurements were compared between males and females using unpaired Student’s t test, performed after log transformation if values were nonnormally distributed or with a one-way ANOVA with post-hoc Fisher’s protected least significant difference testing when appropriate. Analysis of the effects of gender and critical illness (and the interaction between these two factors) was performed using a two-way ANOVA. Linear regression analysis was used to evaluate relationships between paired measures of interest. The changes over time (increments above baseline) in response to GH secretagogues were compared using repeated measures ANOVA. P < 0.05 was construed as significant.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
and
Figs. 1–6)
GH secretory pattern. Critically ill men and women presented
with comparable mean nocturnal GH concentrations, GH half-life, total
nocturnal pituitary GH output, and number of GH pulses over the studied
9 h (Table 2). In contrast, men had a lower fraction of GH
released in a pulsatile fashion compared with women (and conversely a
higher nonpulsatile component; Figs. 1 and 2). In addition, men released GH with
more irregularity than women, as indicated by a higher ApEn (Table 2).
|
|
|
|
|
|
|
), with a gender difference in IGF-I and
ALS (men had lower values than women) in the face of critical illness
(Table 2 and Fig. 3). The serum concentrations of IGF-I and ALS
normalized per total amount of GH released in male patients were only
half those observed in female patients, whereas IGF-I and ALS
normalized per amount of pulsatile GH secretion were identical in both
genders (Fig. 2). Correspondingly, there was a positive correlation
between circulating IGF-I and ALS levels and pulsatile (but not
nonpulsatile) GH secretion (Fig. 4). Age was negatively correlated, in both patients and controls, with serum levels of IGF-I (r = -0.81; P < 0.0001 and r = -0.32; P = 0.001, respectively), IGFBP-3 (r = -0.85; P < 0.0001, and r = -0.21; P = 0.03, respectively) and ALS (r = -0.81; P < 0.0001 and r = -0.38; P = 0.0001, respectively).
Other IGFBPs. Mean nocturnal serum IGFBP-1 concentrations (measured with blood glucose clamped below 9 mmol/L) were not detectably different (37 ± 24 vs. 30 ± 21 µg/L; P = 0.5) in male and female patients and correlated inversely with circulating levels of IGF-I (r = -0.63; P = 0.005), IGFBP-3 (r = -0.58; P = 0.01), and ALS (r = -0.62; P = 0.006) and tended to relate positively to ApEn (r = 0.45; P = 0.06).
Serum IGFBP-2 concentrations in patients were more than double the
control values (P < 0.0001) without an effect of
gender (Fig. 3). Serum IGFBP-2 was positively correlated to IGFBP-1
(r = 0.57; P = 0.01) and inversely correlated to
IGF-I (r = -0.50; P = 0.03), IGFBP-3 (r =
-0.51; P = 0.03), ALS (r = -0.56;
P = 0.01).
Serum IGFBP-6 concentrations in patients were higher than values in
matched controls (P < 0.0001) without an effect of
gender (Fig. 3). Serum IGFBP-6 levels were unrelated to any other
studied endocrine parameter.
In healthy controls, IGFBP-2 (r = 0.43; P < 0.0001) and IGFBP-6 (r = 0.30; P = 0.003) correlate to age, as did mean nocturnal IGFBP-1 levels in the patients (r = 0.67; P = 0.002).
GHBP. Serum GHBP levels were higher in patients than controls
(2.67 ± 1.97 vs. 1.95 ± 1.16 µg/L;
P = 0.03) without a detectable gender difference (Table 2 and Fig. 3). In critically ill female patients, GHBP levels were
positively correlated to mean nocturnal GH concentrations and to
nonpulsatile GH release, but not to pulsatile GH secretion (Fig. 5).
Other anterior pituitary hormones and steroids. Measured in a
single daytime sample, PRL levels were 5-fold higher in patients than
in controls (mean ± SD, 22 ± 23 vs.
4 ± 3 µg/L; medians, 15 vs. 3 µg/L;
P < 0.0001) without a gender difference in healthy
subjects, but with critically ill men revealing lower mean nocturnal
(Table 2) and single daytime PRL concentrations (median, 10
vs. 21 µg/L; P = 0.01) than sick women. In
patients, PRL was positively correlated to maximal nocturnal GH
concentration (r = 0.77; P = 0.0002), GH pulse
amplitude (r = 0.21; P = 0.05), and mass (r =
0.61; P = 0.008) and inversely correlated to GH ApEn
(r = -0.60; P = 0.009). Mean nocturnal cortisol
and TSH levels also tended to be lower in sick men than women (Table 2).
Serum levels of E2 and E1
were positively interrelated (r = 0.64; P = 0.006)
and more than 3-fold elevated in male and female patients (Table 2)
compared to age-matched healthy subjects (mean ±
SD E2 value, 74.5 ±
28.1 pmol/L for healthy age-matched men and 71.8 ± 30.1 pmol/L
for age-matched women) (21). Serum androstenedione concentrations were
high normal and comparable in men and women, whereas serum levels of
DHEAS and testosterone were low in both genders (Table 2). Serum
testosterone levels in male patients, although statistically slightly
higher than those in females, were only 13.8% of the normal mean for
age-matched men (21), reflecting pronounced hypoandrogenism.
Circulating E2 and GH half-lives were positively
interrelated (r = 0.49; P = 0.04). In men only,
serum testosterone levels (r = 0.91; P = 0.0008)
and the molar ratio between testosterone and E2
(r = 0.76; P = 0.017) correlated positively with
GH pulse amplitude (Fig. 6).
Endocrine parameters and outcome. Four of 9 male and 3 of 9 female patients died. Among all of the studied endocrine parameters, serum IGFBP-1 concentration and GH half-life were the only parameters related to outcome; IGFBP-1 concentrations were higher (P = 0.01) in nonsurvivors (49 ± 21 µg/L) compared to survivors (23 ± 18 µg/L) in the presence of identical (clamped) blood glucose levels. GH half-life was longer (P = 0.01) in nonsurvivors (20 ± 5 min) compared to survivors (15 ± 3 min).
Study 2 (Fig. 7)
The peak serum GH concentration responses (increments above
baseline) to GHRH were relatively low (31) and statistically identical
(P = 0.99) in men and women (mean peak GH increment,
7.3 ± 9.4 vs. 7.8 ± 4.1 µg/L; Fig. 7). In
contrast, the GH responses to GHRP-2 in women (mean peak GH increment,
93 ± 38 µg/L) were supranormal (31) and much higher
(P < 0.0001, respectively) than those in men (mean
peak GH increment, 28 ± 16 µg/L). Combining GHRH with GHRP-2
eliminated this gender difference (P = 0.4); both men
(mean peak GH increment, 77 ± 58 µg/L) and women (mean peak GH
increment, 120 ± 69 µg/L) revealed substantially elevated peak
GH responses compared to healthy subjects (31).
|
). |
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The less pulsatile and more irregular, and therefore more ‘feminized’ (13, 14, 21, 33), pattern of GH secretion paradoxically observed in male patients was striking. It could be hypothesized that the exaggerated nonpulsatile (vs. pulsatile) GH secretion in critically ill men is due to reduced inhibitory somatostatin tone between pulses. Indeed, in experimental animals intermittent somatostatin suppression of the somatotropes determines the depth of the troughs and the frequency of GH pulses (34). In the human, however, somatostatin seems to primarily blunt GH pulse amplitude rather than to determine nonpulsatile GH release (35, 36). Therefore, in the face of reduced somatostatin tone, higher GH pulses would be anticipated (35) as well as a more pronounced GH response to the injection of GHRH (37), whereas the latter was found to be relatively low and equal in male and female patients. The tendency for lower TSH concentrations in critically ill men compared with women was indeed not consistent with a reduced somatostatin effect. Alternatively, as exogenous dopamine profoundly suppresses PRL, TSH, and GH pulses in critically ill patients (22), the substantially lower PRL levels in sick men compared to women and the positive correlation between PRL and the size of the GH pulses as well as the regularity with which they occur may point to a dopaminergic mechanism. The results for critically ill women strongly indicated an endogenous GHRP-like ligand deficiency, with GHRH secretion being maintained, as evidenced by the extremely high GH responses to exogenous GHRP, but not to GHRH, when administered alone (31). The foregoing and the observation that adding a high dose of GHRH to the GHRP injection nullified the gender difference in the GHRP response, allowing men also to release supranormal amounts of GH during combined stimuli, could indicate that 1) critically ill men have less endogenous GHRH activity than women; and 2) both genders have an inferred reduced GHRP-like ligand availability with a potentially even greater deficiency in women. This subtle difference in hypothalamic regulation may explain the relatively smaller amount of GH released in pulses in critically ill men than women.
The higher GH ApEn scores in critically ill men indicate a more irregular GH secretory pattern and, inferentially, less synchrony between somatostatin withdrawal and GHRH release (12). This also is a feminine characteristic of the GH axis, as females, both rodents and healthy young and elderly humans, display greater irregularity in GH secretion than males (21, 12). In both genders, disorderliness increases with age (38).
Sex steroids are known to influence GH secretion. Estrogens stimulate GH release (39) directly through modulation of both GH pulse frequency and amplitude in lower doses (40) as well as indirectly in higher doses, the latter by reducing IGF-I negative feedback due to suppression of hepatic IGF-I synthesis (41). Serum testosterone levels correlate with measures of GH secretion in healthy men (42, 43), and administration of androgens selectively stimulates GH pulse amplitude and mass (43). The effects of testosterone are thought to be at least partially mediated by aromatization to E2 (20, 44) and have been interpreted as reflecting a relatively greater ratio of GHRH vs. somatostatin actions on responsive somatotropes (45). However, the tight correlation observed in the current study between GH pulse amplitude and levels of testosterone, but not E2, in men favor a direct suppressive effect of hypoandrogenism on pulsatile GH secretion. The latter is in line with enhancement of GHRH-induced GH release by testosterone in vitro (46). Estrogens exert a negative effect on the regularity of GH secretion in the human, inferentially by impairing the synchrony of firing between somatostatin and GHRH neuronal networks (12). To account for less orderly GH release in the ill men than women, reduced testosterone antagonism of estrogen and/or higher sensitivity of the male hypothalamus to the estrogenized sex steroid environment during protracted critical illness could be hypothesized.
As high estrogen and low androgen levels were present in both genders, however, an alternative explanation for the observed gender dissociation in the GH/IGF-I axis could be an intrinsic underlying difference in hypothalamic signaling between men and women (e.g. lower endogenous GHRH effect in the male), which is only unveiled when the impact of sex steroids is identical. A tendency to equalize sex steroids in men and women occurs in protracted critical illness, unlike in normal aging, where circulating testosterone levels remain substantially higher in men (21).
The mechanisms underlying the profound alterations in circulating sex steroids during critical illness remain incompletely understood. Leydig cell dysfunction and hypogonadotropism in critically ill men have repeatedly been reported (2, 47, 48, 49, 50, 51, 52, 53). Elevated levels of E2 and E1 have been previously documented in sepsis (52), in burn injury (54), and after myocardial infarction (55) in both male and female patients. In view of the hypogonadotropic hypogonadism in protracted critical illness and the postmenopausal age of most of the studied women here, it is unlikely that the elevated estrogen levels originate from the gonads. Indeed, the finding that serum concentrations of E2 and E1 are elevated proportionately is consistent with increased aromatase activity either in adipose tissue or in muscle. The latter could be explained by the concomitantly elevated cortisol (56) and PRL (57) levels, by effects of endotoxin (58), endogenous catecholamines (59), and/or increased fat storage as a result of feeding during critical illness (1, 60). Conversely, altered pulsatility of GH secretion could contribute to changes in hepatic steroid metabolism (61, 62, 63).
A less pulsatile and more disorderly GH secretory pattern for a comparable mean serum GH concentration appeared to determine more severely impaired somatotropic GH effects, as evidenced by lower serum IGF-I and ALS concentrations in male compared to female patients. Indeed, IGF-I and ALS correlated positively with pulsatile, but not with nonpulsatile, GH production. These data support the hypothesis that in the human distinct components of the pulsatile signal are involved in diverse biological GH effects, as shown in rodents (15).
In contrast with the low GHBP levels observed in acute stress (64), the prolonged critically ill patients studied here displayed 37% higher GHBP levels than matched controls. Assuming that serum concentrations of GHBP reflect GH receptor expression in peripheral tissues, as has been demonstrated in the acute catabolic state evoked by elective surgery (64), this observation is consonant with our earlier finding of recovered responsiveness of IGF-I and ALS to reactivated pulsatile GH secretion in prolonged critically ill patients treated with GH secretagogue infusion (3, 4, 5). In addition, the equally elevated GHBP concentrations in male and female patients may indicate that the evident gender difference in IGF-I and ALS concentrations is more likely explained by the lower pulsatile GH fraction in the males than by a difference in peripheral GH responsiveness.
Circulating levels of the small (inhibitory) IGFBP-2 and -6 were clearly elevated in patients compared with healthy controls, and a suggested link between a high serum IGFBP-1 concentration in critical illness and poor prognosis (5) and/or hypercatabolism (5, 65) was confirmed. The inverse correlation of inhibitory binding proteins with IGF-I and the GH-dependent proteins ALS and IGFBP-3 is consistent with their inverse regulation by GH, as previously suggested (66, 67, 68, 69). In unfavorable metabolic conditions, the hepatocyte appears to alter the production of IGF regulatory proteins, for which the trigger might be reduced hepatocyte substrate availability leading to increased cAMP production, which would both suppress IGF-I and ALS (70) and stimulate IGFBP-1 (71). It is unclear to what extent loss of GH pulsatility may also contribute to this switch, but our data suggest that activation of hepatic IGF-I and ALS expression may require pulsatile GH, and animal studies similarly suggest that suppression of hepatic IGFBP-1 expression by insulin requires acute, rather than prolonged or nonpulsatile, GH action (72).
In summary, a gender dissociation within the GH/IGF-I axis is evident in protracted critical illness, with men being more affected than women by illness-induced loss of pulsatility and regularity within the GH secretory pattern despite equivalent total GH output. Reduced endogenous GHRH action in chronically ill men, possibly in part due to profound hypoandrogenism, and, inferentially, a reduced availability of the putative endogenous GHRP-like ligand in both critically ill men and women may explain these novel findings. Lower IGF-I and ALS levels in male compared to female patients were statistically concomitants of the lack of pulsatility and regularity of GH secretion, pointing to distinct biological effects of the pulsatile GH signal in the human. Whether this sexual dimorphism within the somatotropic axis is a factor contributing to the apparent survival benefit of female critically ill patients or only reflects an association cannot be concluded from these data.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 From 1996 to 1998, only 36% of the 562
protracted critically ill patients (with an age of 33 yr or more and an
ICU stay of at least 21 days) treated in our intensive care department
were women, revealing a 6% survival benefit (20% vs.
26%; P = 0.057, by 
2 test) over the
protracted critically ill men (unpublished data).
Received July 29, 1999.
Revised September 22, 1999.
Accepted October 15, 1999.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. Maggio, F. Lauretani, G. P. Ceda, S. Bandinelli, S. M. Ling, E. J. Metter, A. Artoni, L. Carassale, A. Cazzato, G. Ceresini, et al. Relationship Between Low Levels of Anabolic Hormones and 6-Year Mortality in Older Men: The Aging in the Chianti Area (InCHIANTI) Study Arch Intern Med, November 12, 2007; 167(20): 2249 - 2254. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. H. Lang, R. A. Frost, and T. C. Vary Skeletal muscle protein synthesis and degradation exhibit sexual dimorphism after chronic alcohol consumption but not acute intoxication Am J Physiol Endocrinol Metab, June 1, 2007; 292(6): E1497 - E1506. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. K Gleeson and S. M Shalet GH responsiveness varies during the menstrual cycle Eur. J. Endocrinol., December 1, 2005; 153(6): 775 - 779. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M A Spruit, M J Thomeer, R Gosselink, T Troosters, A Kasran, A J T Debrock, M G Demedts, and M Decramer Skeletal muscle weakness in patients with sarcoidosis and its relationship with exercise intolerance and reduced health status Thorax, January 1, 2005; 60(1): 32 - 38. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. D. Veldhuis, J. T. Patrie, K. T. Brill, J. Y. Weltman, E. E. Mueller, C. Y. Bowers, and A. Weltman Contributions of Gender and Systemic Estradiol and Testosterone Concentrations to Maximal Secretagogue Drive of Burst-Like Growth Hormone Secretion in Healthy Middle-Aged and Older Adults J. Clin. Endocrinol. Metab., December 1, 2004; 89(12): 6291 - 6296. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
