Sex steroid control of gonadotropin secretion in the human male. I. Effects of testosterone administration in normal and gonadotropin-releasing hormone-deficient men.

0021-972X/91/7303-0609$03.00/0 Journal of Clinical Endocrinology and Metabolism Copyright © 1991 by The Endocrine Society

Vol. 73, No. 3 Printed in U.S.A.

Sex Steroid Control of Gonadotropin Secretion in the Human Male. I. Effects of Testosterone Administration in Normal and Gonadotropin-Releasing HormoneDeficient Men* JOEL S. FINKELSTEINf, RANDALL W. WHITCOMB, LOUIS ST. L. O'DEA$, CHRISTOPHER LONGCOPE§, DAVID A. SCHOENFELD, AND WILLIAM F. CROWLEY, JR. Reproductive Endocrine Unit, Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts 02114; and the Departments of Obstetrics and Gynecology and Medicine, University of Massachusetts Medical Center, Worcester, Massachusetts 01655

A

ABSTRACT. The precise sites of action of the negative feedback effects of gonadal steroids in men remain unclear. To determine whether testosterone (T) administration can suppress gonadotropin secretion directly at the level of the pituitary, the pituitary responses to physiological doses of GnRH were assessed in six men with complete GnRH deficiency, whose pituitary-gonadal function had been normalized with long term pulsatile GnRH delivery, before and during a 4-day continuous T infusion (15 mg/day). Their responses were compared with the effects of identical T infusions on spontaneous gonadotropin secretion and the response to a lOO-jtg GnRH bolus in six normal men. Both groups were monitored with 15 h of frequent blood sampling before and during the last day of the T infusion. In the GnRH-deficient men, the first three GnRH doses were identical and were chosen to produce LH pulses with amplitudes in the midphysiological range of our normal men (i.e. a physiological dose), while the last four doses spanned 1.5 log orders (7.5, 25, 75, and 250 ng/kg). The 250 ng/kg dose was always administered last because it is known to be pharmacological. In the GnRH-deficient men, mean LH (P < 0.02) and FSH (P < 0.01) levels as well as LH pulse amplitude (P < 0.05) decreased significantly during T infusion, demonstrating a direct pituitary-suppressive effect of T and/or its metabolites. Mean LH levels were suppressed to a greater extent in the normal than in the GnRH-deficient men (58 ± 15% vs. 28 ± 7%; P < 0.05). In addition, LH frequency decreased significantly (P < 0.01) during T administration in the normal men. These latter two findings suggest that T administration also suppresses hy-

pothalamic GnRH release. T was unable to suppress gonadotropin secretion in one GnRH-deficient and one normal man. In both groups, the suppressive effect of T administration was present only in response to physiological doses of GnRH. Because the pituitary- and hypothalamus-suppressive effects of T could be mediated by its aromatization to estrogens, five GnRH-deficient and five normal men underwent identical T infusions with concomitant administration of the aromatase inhibitor testolactone (TL; 500 mg, orally, every 6 h). As an additional control, four GnRH-deficient and four normal men received TL alone. TL administration completely prevented the effect of T administration to suppress gonadotropin secretion in both the normal and GnRH-deficient men, and mean LH levels increased significantly in both the GnRH-deficient (P < 0.01) and the normal (P < 0.001) men who received TL alone. The increase in mean LH levels was greater (P < 0.01) in the normal men who received TL alone than in the normal men who received T plus TL, thus revealing a direct effect of androgens in normal men. Measurements of T and estradiol production rates in three men demonstrated that TL effectively blocked aromatization. These results demonstrate that 1) T or one of its metabolites inhibits gonadotropin secretion at both pituitary and hypothalamic levels in men; 2) aromatization of androgens to estrogens is required for part of this suppressive effect, while part is due to androgens themselves; 3) there is some heterogeneity in the susceptibility to the inhibitory effects of T administration; and 4) it is important to use physiological stimuli to assess the effects of neuromodulators on the pituitary. (J Clin Endocrinol Metab 73: 609-620,1991)

NDROGEN administration inhibits gonadotropin secretion in normal men (1-9), although its precise

site(s) of action on the hypothalamic-pituitary axis remains unclear. One major reason that this question remains unanswered is the inability of prior investigative efforts to dissect hypothalamic from pituitary feedback effects when studying men whose hypothalamic-pituitary

Received October 29,1990. Address requests for reprints to: William F. Crouley, Jr., M.D., Reproductive Endocrine Unit, Bartlett Hall Extension 5, 40 Rear Blossom Street, Boston, Massachusetts 02114. * This work was supported by NIH Grants HD-15788, HD-18169, and RR-1066; FDA Grant FD-U-000523-1; and an NIH Clinical Associate Physician Award (to J.S.F.). t Current address: Endocrine Unit, Bulfinch 3, Massachusetts General Hospital, Boston, Massachusetts 02114.

X Current address: Division of Endocrinology and Metabolism, Montreal General Hospital, Montreal, Quebec, Canada H3G 1A4. § Current address: Departments of Obstetrics and Gynecology and Medicine, University of Massachusetts Medical School, Worcester, Massachusetts 01605.

609

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 13 July 2015. at 05:01 For personal use only. No other uses without permission. . All rights reserved.

610

FINKELSTEIN ET AL.

axes are intact. In normal men, investigators have generally assumed that LH pulse frequency is determined solely by the frequency of GnRH release from the hypothalamus, whereas LH pulse amplitude is determined by the pituitary responsiveness to GnRH. Because androgen administration slows LH pulse frequency without altering pulse amplitude or the pituitary response to pharmacological doses of GnRH (2-7), most investigators have concluded that androgens exert their negative feedback effect solely at the hypothalamic level. Some observations, however, suggest that androgens also may inhibit gonadotropin secretion directly at the pituitary. First, androgens can suppress pituitary gonadotropin secretion directly in vitro (10-15). Second, in men with the androgen insensitivity syndrome or in normal men treated with an androgen receptor-blocking drug, both LH pulse frequency and amplitude increase, suggesting that androgens may exert suppressive effects at both the pituitary and hypothalamic levels (16, 17). However, the experimental models used in these studies may be misleading, because 1) decreases in LH pulse frequency could reflect GnRH pulses that are secreted from the hypothalamus but are blocked at the pituitary level (18, 19); 2) changes in LH pulse amplitude may reflect changes in the amount of GnRH released in each bolus from the hypothalamus (19, 20); 3) changes in LH pulse amplitude may alter the ability to detect LH pulses, leading to apparent changes in LH pulse frequency; 4) alterations in GnRH pulse frequency produce reciprocal changes in LH pulse amplitude, such that isolated decreases in LH pulse frequency alone without any change in the dose of GnRH should increase LH pulse amplitude, and increases in GnRH pulse frequency should decrease LH pulse amplitude (18, 21, 22); and 5) in vitro models may not reflect function in vivo accurately, because they depend on static cell cultures or perifusion systems in which GnRH delivery and/or gonadal steroid levels are not physiological. Therefore, an alternative in vivo model that allows direct assessment of either pituitary gonadotropin or hypothalamic GnRH secretion would be useful. Most men with idiopathic hypogonadotropic hypogonadism (IHH) have a complete deficiency of hypothalamic GnRH secretion (23-28), and their pituitary gonadotropin and gonadal steroid secretion can be normalized during long term pulsatile GnRH replacement. Because GnRH stimulation of the pituitary can be experimentally controlled in these men (28-30), they provide a powerful human model to assess directly any pituitary effect of gonadal steroids. In GnRH-deficient men receiving exogenous GnRH replacement, any suppression of gonadotropin secretion during T administration must reflect a direct effect of T or one of its metabolites on the anterior pituitary. In contrast, gonadal steroids can inhibit go-

JCE & M • 1991 Vol 73 • No 3

nadotropin secretion in normal men either by inhibiting LH or FSH secretion from the pituitary or by inhibiting GnRH secretion from the hypothalamus. Therefore, to determine the site(s) of action of T or its metabolites in men, these two models are best studied in tandem. Accordingly, exogenous T was infused into GnRH-deficient men to determine its effect directly on the pituitary and into normal men to assess its effect on the entire hypothalamic-pituitary axis. By comparing the responses of the two groups, the hypothalamic and pituitary components or sex steroid feedback were deduced. Additionally, to determine whether the suppressive effects of T administration on gonadotropin secretion required aromatization of androgens to estrogens, T was administered together with an aromatase inhibitor or the aromatase inhibitor was given alone to both groups, and the effects were compared with those in the patients who received only T. Materials and Methods Patient

populations

GnRH-deficient men Twelve men with IHH between the ages of 22-44 yr were selected on the basis of the following criteria: 1) failure to undergo spontaneous puberty by the age of 18 yr; 2) serum T concentrations below 3.5 nmol/L (normal T, 10.4-34.7 nmol/ L) in the presence of serum LH and FSH concentrations below 5.0 IU/L (normal LH and FSH, 3-19 IU/L); 3) the absence of any endogenous gonadotropin pulsations during a 16- to 24-h period of blood sampling at 10- to 20-min intervals before GnRH therapy; 4) normal baseline thyroid function tests; 5) normal serum cortisol, GH, TSH, and PRL responses to insulin-induced hypoglycemia and TRH injection; 6) normal computerized tomography of the hypothalamic-pituitary region; and 7) the maintenance of normal serum gonadotropin and gonadal steroid concentrations for at least 3 months while receiving a previously reported regimen of low dose, sc GnRH administered at 2-h intervals (29, 30). Five men participated in two different studies separated by at least 6 months. Normal men Sixteen normal men between the ages of 18 and 40 yr were selected on the basis of the following criteria: 1) normal pubertal development, sexual function, and general health; 2) a normal physical examination, with testicular volumes greater than 20 mL; 3) normal serum LH, FSH, T, and estradiol (E2) concentrations; 4) normal serum PRL, T4, and TSH concentrations; and 5) a normal semen analysis (>20 X 106 sperm/mL, >60% motility, and >2.0 mL volume). One man participated in two different studies separated by 8 months. GnRH regimen for the IHH men Each GnRH-deficient man underwent an iv GnRH doseresponse study employing 4-5 GnRH doses, ranging from 2.5-


SITES OF T FEEDBACK IN MEN 250 ng/kg-bolus, at least 6 weeks before taking part in the clinical protocol (20). An individual dose-response curve was then constructed, and the resulting LH pulses were compared to the range of LH pulse amplitudes from 20 normal men whom we studied previously (31) to select a GnRH dose that would produce LH pulses with amplitudes within the midrange of those in the normal men. On the basis of these dose-response studies, the physiological GnRH doses selected for use during the sex steroid infusions ranged from 7.5-200 ng/kg bolus. GnRH was administered iv throughout these studies, because iv GnRH injections produce pituitary responses that mimic the spontaneous pulses of normal men more closely than do sc injections (32), and at 2-h intervals, because this is the average LH pulse frequency in normal men (31, 33). Preparation of T infusions Fifteen milligrams of T USP (kindly provided by the Upjohn Co., Kalamazoo, MI) were dissolved in 5 mL sterile ethanol and then diluted in 1 L normal saline. This solution was infused over 24 h by an infusion pump (IMED, San Diego, CA), using Nitrostat tubing to minimize nonspecific T adsorption (data not shown). A similar procedure, including addition of ethanol, was followed to prepare normal saline control infusions. Experimental protocol GnRH-deficient men Each GnRH-deficient man was admitted to our Clinical Research Center, and an iv catheter was inserted to convert patients from long term sc GnRH administration by an autoinfusion pump (Ferring Laboratories, Inc., Bloomfield, NJ) to iv GnRH administration by rapid manual delivery. The physiological GnRH dose, as determined above, was administered every 2 h for at least three doses before any blood sampling. Each study began at 2400 h, with 15 h of frequent blood sampling. During the first 6 h the physiological GnRH dose was given three times, and blood was drawn for gonadotropin determinations every 10 min. During the next 9 h patients received doses of 7.5, 25, 75, and 250 ng/kg GnRH every 2 h. The first three doses were administered in a randomized order, whereas the 250 ng/kg dose was always given last and monitored for 3 h, as previous studies have demonstrated that this dose produces a pharmacological profile of pituitary responses (20). Blood was sampled at 0, 1, 3, 5, 10,15, 20, 30, and 40 min and then every 20 min after each of these four GnRH doses. Patients then received one of the following regimens for 4 days: 1) iv T (15 mg/day) (n = 6); 2) iv T (15 mg/day) plus oral testolactone (TL; 500 mg every 6 h; n = 5); 3) oral TL alone (500 mg every 6 h; n = 4); or 4) normal saline (1 L/day; n = 2). The physiological GnRH dose was administered every 2 h throughout the 4-day experimental period until the last 15 h, when GnRH administration and blood sampling were repeated in a manner identical to that performed before the infusion. To determine the production rate of E2 (PRE2) and the peripheral aromatization of T to E2 ([P]B]VI2) and confirm the extent of their blockade during TL administration, [3H]T and [14C]E2 were infused at a constant rate for 3.5 h into one man

611

before and during T plus TL administration according to a previously described protocol (34). Three blood samples were collected over the last hour of the infusion and analyzed for [3H]T and [14C]E2 (34), and PRE2 was calculated (35). All urine was collected for 96 h after the infusion and analyzed for [3H] and [14C]estradiol glucuronide (36), and [P]B1W2 was calculated (37). Normal men In the normal men each study began at 2400 h with 15 h of frequent blood sampling. During the first 12 h blood was drawn every 10 min to assess spontaneous gonadotropin secretion. Each man then received a 100-^g iv GnRH bolus, followed by blood sampling at 0, 15, 30, 45, 60, 90, 150, and 180 min. They then received 1) iv T alone (n = 6), 2) iv T plus oral TL (n = 5), 3) oral TL alone (n = 4), or 4) normal saline (n = 2) for 4 days. Blood was sampled as described above during the last 15 h of the infusion. PRE2 and [P]BM2 were determined in two men who received T plus TL as described above. Evaluation of pituitary gonadotropin and gonadal steroid secretion Mean LH and FSH levels, LH pulse amplitude, and the percent decrease in mean LH and FSH levels were determined to evaluate the effects of T, T plus TL, or TL alone on gonadotropin secretion in the GnRH-deficient men. Mean LH levels and LH amplitude were determined both for the entire 15-h blood-sampling period (i.e. including the 250 ng/kg dose) and for the first 12 h of each blood-sampling period. Mean serum LH concentrations were determined using a computer program designed to weight values appropriately for differences in sampling intervals. LH pulse amplitude was defined as the difference between peak and nadir LH concentrations. Mean FSH, T, and E2 levels were determined in serum pools composed of equal aliquots of each sample obtained during the first 6 h and the last 9 h of each blood-sampling period, and these values were then averaged. To evaluate the effects of T, TL, or T plus TL administration on gonadotropin secretion in the normal men, mean LH and FSH levels, LH pulse amplitude, LH pulse frequency, and the percent decrease in mean LH and FSH levels were determined during the first 12 h of each blood-sampling period. The amplitude of the LH response to 100 /ng GnRH was assessed during the last 3 h of each blood-sampling period. Mean FSH, T, and E2 levels were determined in equal aliquots of each sample obtained during the first 12 h of each blood-sampling period. The presence of LH pulses was determined by a modification of the method of Santen and Bardin (33), requiring a 20% increase from nadir to peak LH concentrations for pulse detection. The false positive rate of pulse detection, based on multiple measurements of serum pools with mean LH levels of 10.5-37.8 IU/L and assay coefficients of variation of 4.0-11.9%, was 1.6%. To decrease the incidence of false positive pulse detection, we also required that each pulse contain at least two points and have a minimum amplitude of 2 IU/L (24, 31). LH pulse frequency was defined as the number of pulses during the 12 h of frequent blood sampling. When the serum LH level was


FINKELSTEIN ET AL.

612

below the detection limit of the RIA, the value was set equal to the assay detection limit. Missing samples were omitted from the pulse analysis. Assays Serum LH and FSH levels were determined by previously described RIAs with a sensitivity of 0.8 IU/L, using the Second International Reference Preparation as the reference standard (38). The intraassay coefficients of variation for the LH and FSH RIAs were 5.9% and less than 10%, respectively, and the interassay coefficients of variation for both assays were less than 10% (38). The cross-reactivities of a-subunit in the LH and FSH RIAs were 4.9% and 1.0%, respectively. Serum T and E2 levels were determined by previously described RIAs with sensitivities of 0.3 nmol/L and 37 pmol/L and interassay coefficients of variation less than 15% and 8%, respectively (39, 40).

Data analysis For the GnRH-deficient men, differences in mean LH levels and LH pulse amplitude were compared before and during the experimental manipulations using a paired sample t test (which gives the same results as a two-way analysis of variance) after log transformation of the data, and mean FSH, T, and E2 levels were compared by paired t tests. Because the 250 ng/kg GnRH dose routinely produces LH pulses with amplitudes outside of the 95% confidence limits for our normal men (20, 31), the paired sample t tests were performed both with and without this pharmacological GnRH dose. To determine whether the magnitude of the change in mean LH levels and LH pulse amplitude during androgen administration was different for the physiological and the pharmacological GnRH doses, differences in the amount of change for all GnRH doses except the 250 ng/ kg dose were compared with the change in response to the 250 ng/kg dose using paired sample t tests. For the normal men, all parameters of hormone secretion were compared before and during the experimental manipulations by paired t tests. Because no LH pulses could be detected in two of the normal men who received T alone, LH pulse amplitudes were only analyzed in the men who received T plus TL or TL alone. Parameters of hormone secretion were compared between groups using nonpaired t tests. Because our goal was to make physiologically relevant comparisons between normal and GnRH-deficient men, data from the 250-ng/kg and the 100-/ig GnRH doses were excluded whenever the two groups were compared. All data are presented as the mean ± SE, and statistical significance is construed for P < 0.05.

Results Responses to T administration GnRH-deficient men The serum LH responses to GnRH administration before and during T administration in a representative

JCE & M • 1991 Vol 73 • No 3

GnRH-deficient man are shown in Fig. 1. For the group of six GnRH-deficient men, mean serum LH (P < 0.02 without the 250-ng/kg dose and P < 0.05 with the 250ng/kg dose) and FSH (P < 0.01) levels and LH amplitude (P < 0.05 both with and without the 250-ng/kg dose) all decreased significantly (Figs. 2 and 3 and Table 1). However, serum LH levels did not change during T administration in one GnRH-deficient man despite the expected rise in serum T and E2 levels (Fig. 2). Mean LH levels fell more (P < 0.05) during T administration in response to the physiological GnRH doses than in response to the pharmacological 250-ng/kg dose. Serum T (P < 0.001) and E2 (P < 0.001) levels both rose significantly during T administration (Table 1). Normal men

The serum LH secretory pattern in a representative normal man before and during T administration is shown in Fig. 1. For the group of six normal men, mean LH and FSH levels and LH frequency all decreased significantly (P < 0.01; Figs. 2 and 3 and Table 1). As with the GnRHdeficient men, there was one normal man in whom LH secretion did not change during T infusion despite a marked rise in his serum T level (Fig. 2). In response to a 100-/ig bolus of GnRH, LH pulse amplitude increased from 57.7 ± 18.1 to 78.1 ± 15.2 IU/L during T delivery (P < 0.02). Serum T and E2 levels both rose significantly (P < 0.05) during T delivery (Table 1). GnRH-deficient vs. normal men

Mean LH levels, FSH levels (P < 0.1), LH amplitude (P < 0.1), T, and E2 levels were similar in the normal and GnRH-deficient men before T administration (Table 1). Mean LH levels fell 58 ± 15% in the normal men and 28 ± 7% in the GnRH-deficient men during T infusion (P < 0.05; Fig. 3). Mean FSH levels fell 40 ± 14% in the normal men and 20 ± 5% in the GnRH-deficient men (P < 0.05; Fig. 3). Responses to T plus TL administration GnRH-deficient men The serum LH responses to GnRH administration before and during T plus TL administration in a representative GnRH-deficient man are shown in Fig. 4. For the group of five GnRH-deficient men, mean LH levels and LH amplitude increased, although these changes were not statistically significant (P < 0.1 both with and without inclusion of the 250-ng/kg dose; Figs. 3 and 5 and Table 2). Mean LH levels rose 77 ± 49% in the GnRH-deficient men who received T plus TL, whereas they fell 28 ± 7% in the GnRH-deficient men who


SITES OF T FEEDBACK IN MEN 7.5

40 35

7.S

7.5

7.5

7i

25

230

G n R H

D o s e

I I 1 I 1 I 1 T - 30.7

40 35 30

E , = 62

0 0

20

1

15

J

10

613

^ 1 -

s 0

1

T - 28.2

E,-77

T - 47.8

I , - 161

20 15 10 S 0

I I 1 I I 1I 67.7

6 TIME

8

6

(hours)

B

TIME (hours)

FlG. 1. Left panels, Serum LH levels determined at frequent intervals for 15 h in a GnRH-deficient man before (top) and during (bottom) iv T infusion. The arrows indicate each GnRH bolus and the GnRH doses (in nanograms per kg) are shown above the arrows. The dashed line denotes the 95% confidence limit for peak LH levels in our normal men. Right panels, Serum LH levels determined at frequent intervals for 12 h in a normal man before (top) and during {bottom) iv T infusion. LH pulses are indicated by asterisks. Mean serum T (nanomoles per L) and E2 (picomoles per L) levels are indicated on each panel. To convert T to nanograms per dL divide by 0.03467. To convert E2 to picograms per mL, divide by 3.671.

received T alone (P < 0.05; Fig. 3). The large amount of variability in mean LH levels and LH amplitude during T plus TL administration in these patients was caused by a single patient whose LH levels increased nearly 300% while these values either remained constant or increased by no more than 45% in the other four patients (Fig. 5). The amount of change in both mean LH levels and LH amplitude was similar in response to the physiological and pharmacological GnRH doses. Mean serum FSH (P < 0.02) and T (P < 0.001) levels rose significantly during T plus TL administration, while E2 levels fell (P < 0.05; Fig. 5 and Table 2). PRE2 decreased from 44 to 34 Mg/day, and the percent aromatization decreased by 58% in the man who had these parameters measured. Normal men

The serum LH secretory pattern in a representative normal man before and during T plus TL administration is shown in Fig. 4. For the group of five normal men, there was no significant change in mean LH and FSH levels, LH amplitude, LH frequency (P < 0.1), or the LH response to a 100-/*g bolus of GnRH (45.9 ± 6.8 to 58.0 ± 10.3 IU/L; P < 0.1; Figs. 3 and 5 and Table 2). Mean LH levels rose 6 ± 9% in the normal men who received T plus TL, whereas they fell 58 ± 15% in the normal men who received T alone (P < 0.01). Serum T levels rose significantly (P < 0.001), and serum E2 levels did not change during T plus TL administration (Table 2). In the two men who had production rates determined,

PRE2 increased from 46 to 57 and from 54 to 74 Although the PRE2 increased slightly, the percent aromatization decreased by 58% and 70%, respectively, in these two men. GnRH-deficient us. normal men

Mean LH levels, FSH levels (P < 0.1), LH pulse amplitude (P < 0.1), T, and E2 levels were similar in the normal and GnRH-deficient men before T plus TL administration (Table 2). Mean LH levels increased 76 ± 47% in the GnRH-deficient men vs. 6 ± 10% in the normal men. However, this difference was not statistically significant, possibly because of the large degree of variability among the GnRH-deficient men (Fig. 3). Mean FSH levels increased 55 ± 10% in the GnRHdeficient men vs. 18 ± 19% in the normal men (P < 0.02; Fig. 3). Responses to TL administration GnRH-deficient men The serum LH responses to GnRH administration before and during TL administration in a representative GnRH-deficient man are shown in Fig. 6. For the group of four GnRH-deficient men, mean LH levels (P < 0.01 with and without the 250-ng/kg dose) and LH amplitude (P < 0.02 without and P < 0.05 with the 250-ng/kg dose) both increased significantly (Figs. 3 and 7 and Table 3). The increases in mean LH levels and LH amplitude were


FINKELSTEIN ET AL.

614 GnflH-DEFICIENT MEN

JCE&M«1991 Vol 73 • No 3 TESTOSTERONE

NORMAL MEN

TESTOSTERONE PLUS TESTOLACTONE

TESTOLACTONE

+ 140 + 120 + 100 +80 +60 +40 +20 0 -20

[-40 -60 -80 -100

L

GnRH- NORMAL DEFICIENT

II G n R H - NORMAL DEFICIENT

II G n R H - NORMAL DEFICIENT

FIG. 3. Percent change in mean LH (•) and FSH (M) levels in GnRHdeficient and normal men during administration of T alone, T plus TL, or TL alone. *, P < 0.05 vs. mean LH and FSH levels in GnRHdeficient men receiving T alone. **, P < 0.02 vs. mean FSH levels in GnRH-deficient men receiving T plus TL. +, P < 0.05 vs. mean LH levels in GnRH-deficient men receiving T alone. ++, P < 0.01 vs. mean LH levels in normal men receiving T alone. +++, P < 0.01 vs. mean LH levels in normal men receiving T plus TL.

PRE

DURING PRE TESTOSTERONE

DURING

FIG. 2. Individual mean LH levels, LH amplitude, and LH frequency in six GnRH-deficient {left panels) and six normal (right panels) men before and during iv T infusion. The group mean ± SE for each parameter is indicated to the side of each graph. *, P < 0.02; +, P < 0.05; #, P < 0.01.

greater than in the GnRH-deficient men who received T plus TL (137 ± 24% vs. 76 ± 47% for mean LH and 218 ± 50% us. 83 ± 54% for LH amplitude), although these differences failed to reach statistical significance (P < 0.12) because of the single GnRH-deficient man whose LH levels rose nearly 300% in response to T plus TL administration. Mean serum FSH, T, and E2 levels did not change significantly during TL administration in these men (Table 3). Normal men

levels increased significantly, while E2 levels did not change (Table 3). GnRH-deficient us. normal men

Mean LH levels, LH pulse amplitude, and T levels were similar in the normal and GnRH-deficient men before TL administration (Table 3). Baseline mean FSH (P < 0.01) and E2 (P < 0.05) levels were significantly higher in the GnRH-deficient than in the normal men (Table 3). Mean LH levels increased 137 ± 24% in the GnRH-deficient men and 83 ± 17% in the normal men, although this difference was not statistically significant (P < 0.12; Fig. 3). Mean FSH levels increased 32 ± 20% in the GnRH-deficient men and 83 ± 25% in the normal men (P < 0.2; Fig. 3). Responses to saline administration

The serum LH secretory pattern in a representative normal man before and during TL administration is shown in Fig. 6. For the group of four normal men, mean LH (P < 0.001) and FSH (P < 0.01) levels both increased significantly (Figs. 3 and 7 and Table 3), and the increases in mean LH levels were greater (83 ± 17% vs. 6 ± 9 % ) than those in the normal men who received T plus TL (P < 0.01; Fig. 3). Although LH pulse amplitude and frequency both rose, these increases failed to reach statistical significance, possibly because of the small sample size (Fig. 7 and Table 3). Serum T (P < 0.01)

All parameters of LH and FSH secretion were similar before and during saline administration in the two GnRH-deficient and two normal men who received control infusions (data not shown). Discussion This study demonstrates that exogenous T administration, in mildly supraphysiological doses, suppresses gonadotropin secretion in GnRH-deficient men receiving a physiological regimen of GnRH replacement. This find-


SITES OF T FEEDBACK IN MEN

615

TABLE 1. Hormone levels in GnRH-deficient and normal men before and during T Normal men (n = 6)

GnRH-deficient men (n = 6)

Mean LH (IU/L) Mean FSH (IU/L) LH amplitude (IU/L) LH frequency (pulses/12 h) T (nmol/L) E2 (pmol/L)

Baseline

During T

14.4 ± 2.1 14.9 ± 2.9 12.6 ± 2.7 EC 18.1 ± 2.9 122 ± 17

10.2 ± 1.3° 12.2 ± 2.7* 8.2 ± 1.3e EC 51.6 ± 6.0d 183 ± 25d

Baseline 10.7 ± 8.0 ± 10.7 ± 4.5 ± 21.5 ± 107 ±

During T

1.5 1.3 2.5 0.4 2.7 15

3.7 ± 0.9* 4.2 ± 0.8* 2.9 ± 1.0 1.5 ± 0.7* 50.3 ± 8.2C 148 ± 23C

EC, Experimentally controlled. ° P < 0.02 us. baseline. 6 P < 0.01 us. baseline. c P < 0.05 us. baseline. d P< 0.001 vs. baseline. 25

23

23

75

7.5

25

250

G n R H

D o s e

. 1 1 1 1 1 1 1

1 1 1 1 1 11

T = 20.9

E , = 73

T = 87.4

E , = 92

20 r

E , = 128 15 10

0L 4

6

TIME

8

10

(hours)

r4

6

TIME

8

10

(hours)

FiG. 4. Left panels, Serum LH levels determined at frequent intervals for 15 h in a GnRH-deficient man before (top) and during (bottom) iv T plus oral TL administration. The arrows indicate each GnRH bolus, and the GnRH doses (in nanograms per kg) are shown above the arrows. The dashed line denotes the 95% confidence limit for peak LH levels in our normal men. Right panels, Serum LH levels determined at frequent intervals for 12 h in a normal man before (top) and during (bottom) T plus TL administration. LH pulses are indicated by asterisks. Mean serum T (nanomoles per L) and E2 (picomoles per L) levels are indicated on each panel.

ing clearly demonstrates that androgens exert a negative feedback effect at the pituitary level in the human. These data also demonstrate that T administration suppresses gonadotropin secretion to a considerably greater degree in normal than in GnRH-deficient men, suggesting an additional inhibitory effect of T or one of its metabolites on hypothalamic GnRH secretion. This conclusion is supported by the marked decrease in LH frequency during T administration in the normal men. The findings that gonadotropin levels were not suppressed by T when aromatization was inhibited by TL, but rose significantly in response to TL alone, suggest that part of the suppressive effect of T is mediated by its conversion into estrogens, while part is due to T itself or one of its nonestrogenic metabolites. Prior studies in males have demonstrated that andro-

gen administration decreases, while androgen deficiency or blockade increases, the frequency of spontaneous LH pulsations (2, 3, 5-7, 16, 17, 41-44). Because the frequency of LH pulses reflects the frequency of antecedent GnRH secretion from the hypothalamus (19, 45), most investigators have concluded that androgens exert their negative feedback effects principally at the hypothalamic level in men. While the current data are consistent with the hypothesis that the hypothalamus is a major site of action for the feedback effects of T, it is possible that the changes in LH pulse frequency seen in previous studies reflect alterations in the modulation of GnRH stimulation at the level of the anterior pituitary (18,19). Thus, the effect of T administration on hypothalamic GnRH secretion was deduced by comparing its impact in GnRH-deficient men, in whom GnRH input could be


FINKELSTEIN ET AL.

616

GnRH-DEFICIENT MEN 40

•

NORMAL MEN 25

PRE DURING PRE DURING TESTOSTERONE PLUS TESTOLACTONE FlG. 5. Individual mean LH levels, LH amplitude, and LH frequency in five GnRH-deficient (leftpanels) and five normal {right panels) men before and during T plus TL administration. The group mean ± SE for each parameter is indicated to the side of each graph.

controlled, and normal men, in whom both pituitary and hypothalamic effects could occur. Because T administration suppressed gonadotropin secretion to a greater extent in normal than in GnRH-deficient men, the studies demonstrate that T and/or one of its metabolites inhibit gonadotropin secretion at both the hypothalamic and pituitary levels in men. T plus TL administration also suppressed serum LH and FSH concentrations to a greater extent in the normal than in the GnRH-deficient

JCE & M • 1991 Vol 73 • No 3

men, although the large amount of variability in the LH levels (due entirely to one GnRH-deficient man whose serum LH levels increased nearly 300%) prevented this difference from being statistically significant. This finding demonstrates that the additional hypothalamic inhibitory effect of T administration in normal men is independent of aromatization and is, therefore, due to androgens themselves. Whereas prior studies in men with intact hypothalamic-pituitary axes have demonstrated that androgen administration slows LH pulse frequency, these studies have failed to demonstrate a suppressive effect of androgens on LH pulse amplitude, so they have concluded that androgens do not exert significant feedback effects directly on the pituitary (2-4, 6). However, because LH amplitude increases when GnRH pulse frequency decreases (18, 21, 22), the fall in LH pulse frequency without an accompanying rise in LH amplitude during androgen administration in past studies of normal men probably reflects either a decrease in the amount of GnRH secreted per pulse or a pituitary site of negative feedback. Thus, the failure of androgens to suppress LH pulse amplitude in normal men may be due to the limitations of studying men with intact hypothalamic-pituitary axes rather than the inability of androgens to exert feedback effects directly at the level of the pituitary. Studies of the effects of androgens on LH and FSH release in animals and from pituitary cells in vitro are consistent with the notion that androgens can exert negative feedback effects directly at both the pituitary and hypothalamic levels. The ability of androgens to suppress the LH response of cultured pituicytes to GnRH in vitro (10-15), the pituitary LH response to GnRH in vivo (46, 47), and LH/? and a-subunit mRNA levels in GnRH antagonist-treated rats (48) demonstrates a direct negative feedback effect of androgens on the pituitary. Similarly, the ability of hypothalamic androgen implants to suppress LH secretion (49) and of castration to lower hypothalamic GnRH content and release (50, 51) demonstrates that androgens can act directly on the hypo-

TABLE 2. Hormone levels in GnRH-deficient and normal men before and during T plus TL GnRH-deficient men (n = 5)


Normal men (n = 5)

Baseline

During T + TL

Baseline

During T + TL

12.1 ± 1.4 8.5 ± 1.0 9.4 ± 1.3 EC 20.5 ± 3.8 161 ± 32

20.7 ± 5.2 13.2 ± 2.0° 16.6 ± 4.9 EC 83.6 ± 8.0" 94 ± 13C

8.7 ± 1.5 4.9 ± 1.3 6.2 ± 1.0 4.2 ± 1.0 20.3 ± 2.9 99 ±14

9.7 ± 2.5 5.1 ± 1.2 10.1 ± 2.9 3.2 ± 1.0 83.2 ± 10.9" 90 ± 1 1

EC, Experimentally controlled. 0 P < 0.02 vs. baseline. b P < 0.001 vs. baseline. c P < 0.05 vs. baseline.


617

SITES OF T FEEDBACK IN MEN SO

60

50

50

25

75

7.5

250

GnRH

Dose

.I I I I I I I

50 "

E ,=

T = 29.6

88

^

40 35 30

O

20

X -1

« 10 5 0

E , = 77

40 30 20 10 0

1 I I I I l I

6

TIME

8

10

(hours)

4

6

TIME

6

10

(hours)

FlG. 6. Left panels, Serum LH levels determined at frequent intervals for 15 h in a GnRH-deficient man before (top) and during (bottom) oral TL administration. The arrows indicate each GnRH bolus, and the GnRH doses (in nanograms per kg) are shown above the arrows. The dashed line denotes the 95% confidence limit for peak LH levels in our normal men. Right panels, Serum LH levels determined at frequent intervals for 12 h in a normal man before (top) and during (bottom) TL administration. LH pulses are indicated by asterisks. Mean serum T (nanomoles per L) and E2 (picomoles per L) levels are indicated on each panel.

thalamus. Although these studies provide evidence for both pitutary and hypothalamic feedback effects of androgens, they share the common limitations that the models employed may not be physiologically relevant or applicable to the human. Two other studies in GnRH-deficient men have attempted to demonstrate that T administration has a direct pituitary effect (52, 53). However, E2 levels increased significantly in each of these studies, making it impossible to determine whether the suppressive effects of T were due to its androgenic properties or its aromatization into estrogens. Thus, a direct pituitary negative feedback effect of androgens on gonadotropin secretion in men has not been previously established. Gonadotropin levels fell in men who received T alone, were unchanged in men who received T plus TL, and rose in men who received TL alone, suggesting that part of the suppressive effect of T on gonadotropin secretion in men requires aromatization of androgens to estrogens, while part is mediated directly by androgens. Because TL eliminated the suppressive effect of T on gonadotropin secretion, we might have concluded that the suppressive effects of T are entirely due to its aromatization to estrogens had we only examined the effects of T alone and T plus TL on gonadotropin secretion. However, such a conclusion would not account for the possibility that TL, when given alone, might increase LH and FSH concentrations above the levels observed during T plus TL administration, thereby revealing a suppressive effect of androgens themselves on gonadotropin secretion. These results are similar to those previously reported by

Marynick et al. (54), who found that mean LH and FSH levels decreased in normal men given T alone, were similar to controls in men given T plus TL, and increased in men given TL alone. However, these investigators only studied normal men, and they did not report detailed patterns of pulsatile LH secretion. Several other lines of evidence suggest that estrogens play a vital role in the feedback regulation of gonadotropin secretion in men. First, estrogen administration alone, in only microgram quantities, suppresses gonadotropin secretion in normal and GnRH-deficient men despite a concomitant fall in serum T levels and consequent lessening of the inhibitory effect of androgens (1, 2, 6, 7, 55, 56). Second, estrogen receptor antagonists increase gonadotropin levels despite a concomitant increase in T levels in normal men (57-60) and prevent the ability of exogenous androgens to inhibit gonadotropin secretion (6). Finally, both gonadectomy and administration of clomiphene citrate elicit a marked rise in gonadotropin levels in patients with the complete androgen insensitivity syndrome, clearly demonstrating that both androgens and estrogens exert negative feedback effects on gonadotropins (61, 62). Thus, both the present as well as prior studies are consistent with the notion that both androgens and estrogens play important roles in the regulation of gonadotropin secretion in men. Of interest, not all patients responded to T administration in a similar fashion, as one patient in each group had no discernible inhibitory effect of T on LH secretion. Similar observations have been made by Plant (43), who noted that LH secretion was suppressed by T replace-


JCE&M«1991 Vol73-No3

FINKELSTEIN ET AL.

618 GnRH-DEFICIENT MEN

NORMAL MEN

fc X

PRE

DURING

PRE

DURING

TESTOLACTONE

FIG. 7. Individual mean LH levels, LH amplitude, and LH frequency in four GnRH-deficient (left panels) and four normal (right panels) men before and during TL administration. The group mean ± SE for each parameter is indicated to the side of each graph. *, P < 0.01; #, P < 0.02; +, P < 0.001.

ment in some, but not all, orchidectomized monkeys. Similarly, Sisk and Desjardins (63) found that LH pulse frequency was faster than that of castrates in some intact male ferrets and suggested that some animals were relatively insensitive to negative feedback by T. Taken to-

gether, these findings suggest that the role of gonadal steroids in the neuroendocrine control of gonadotropin secretion may vary considerably among individuals. Prior studies in normal men, using pharmacological doses of GnRH, have suggested that androgens do not inhibit, and may even enhance, the pituitary response to GnRH (2, 4, 6, 9, 64, 65). The present data also demonstrate that serum LH levels rise in response to a pharmacological dose of GnRH during T administration in normal men. In contrast, this study unequivocally shows an inhibitory effect of T on the pituitary responsiveness to physiological doses of GnRH. It has also been suggested that pharmacological doses of GnRH may release LH from all pituitary LH pools, whereas physiological GnRH doses elicit a selective release of LH from a highly bioactive LH pool, so that pharmacological GnRH doses mask the ability of GnRH to preferentially increase the secretion of bioactive LH (66). These findings underscore the importance of using a physiological GnRH regimen to assess pituitary gonadotroph responsiveness, a principle that may also be important when evaluating the effects of TRH, CRH, and GHRH on thyrotroph, corticotroph, and somatotroph responsiveness. In summary, T administration suppresses gonadotropin secretion at both the pituitary and hypothalamic levels in men, although there is some variability between individual men in their susceptibility to the negative feedback effects of gonadal steroids. The suppressive effects of T on gonadotropin secretion are mediated by both its aromatization into estrogens and its intrinsic androgenic actions. Finally, the studies underscore the importance of using physiological stimuli to assess the importance of factors that act on the pituitary.

Acknowledgments We gratefully acknowledge Ms. Denise Musket and Ms. Donna Peltier-Saxe for their devoted care of the patients; the staff of the Mallinckrodt General Clinical Research Center for

TABLE 3. Hormone levels in GnRH-deficient and normal men before and during TL GnRH-deficient men (n = 4)


Normal men ( n = 4)

Baseline

During TL

Baseline

During TL

13.2 ± 2.0 9.7 ± 1.2 9.6 ± 1.5 EC 21.7 ±3.9 98 ± 5

30.6 ± 4.8° 13.5 ± 3.8 28.8 ± 4.2d EC 27.2 ± 7.0 74 ±16

11.6 ±1.7 3.6 ± 0.7c 9.5 ± 1.8 5.5 ± 1.7 20.6 ± 2.9 53 ± 16e

20.3 ± 1.2" 6.0 ± 0.7° 13.6 ± 3.3 7.3 ± 0.8 55.2 ± 6.5° 54 ±17

EC, Experimentally controlled. 0 P < 0.01 vs. baseline. b P< 0.001 us. baseline. c P < 0.01 vs. GnRH-deficient men baseline. d P

Sex steroid control of gonadotropin secretion in the human male. II. Effects of estradiol administration in normal and gonadotropin-releasing hormone-deficient men.

Steroid control of gonadotropin secretion.

Gonadotropin gene expression and secretion in gonadotropin-releasing hormone antagonist-treated male rats: effect of sex steroid replacement.

The effects of testosterone administration and visual erotic stimuli on nocturnal penile tumescence in normal men.

Control of gonadotropin secretion in the ovine fetus. IV. Male-specific entrainment of the hypothalamic control of luteinizing hormone secretion by testosterone in the female ovine fetus.

Acute effects of testosterone infusion and naloxone on luteinizing hormone secretion in normal men.

Effects of chronic bromocriptine-induced hypoprolactinemia on plasma testosterone responses to human chorionic gonadotropin stimulation in normal men.

Prolactin-related testosterone secretion in normal adult men.

Effects of recombinant human inhibin and testosterone on gonadotropin secretion and subunit mRNA in superfused male rat pituitary cell cultures stimulated with pulsatile gonadotropin-releasing hormone.

Effect of testosterone on bone mineral gain: observations of male patients with growth hormone deficiency and normal gonadotropin secretion.

Gonadotropin and testosterone measurements after estrogen administration to adult men, prepubertal and pubertal boys, and men with hypogonadotropism: evidence for maturation of positive feedback in the male.

Steroid control of gonadotropin release.

Biochemical changes of the human semen produced by chlorambucil, testosterone propionate and human chorionic gonadotropin administration.

Restraint inhibits luteinizing hormone and testosterone secretion in intact male rhesus macaques: effects of concurrent naloxone administration.

Effects of experimentally-induced chronic hyperprolactinemia on testosterone and gonadotropin levels in male rats and mice.

Circulating prolactin and its response to TRH following administration of testosterone undecanoate in normal men.

Plasma testosterone response to short-term human chorionic gonadotropin administration in men with follicle-stimulating hormone suppressed by exogenous estrogen.

Effect of long-term testosterone oenanthate administration on male reproductive function: clinical evaluation, serum FSH, LH, testosterone, and seminal fluid analyses in normal men.

Gonadotropin and testosterone secretion in normal human males after stimulation with gonadotropin-releasing hormone (GnRH) or potent GnRH analogs using different modes of application.

Gonadotropin receptors, cAMP and testosterone in estrogenized male rats.

Steroid induction of gonadotropin surges in the immature rat. I. Priming effects of androgens.

Gonadotropin, prolactin and thyrotropin pituitary secretion after exogenous dehydroepiandrosterone-sulfate administration in normal women.

Combined administration of a gonadotropin-releasing hormone antagonist and testosterone in men induces reversible azoospermia without loss of libido.

Effects of testosterone metabolites on serum gonadotropin concentrations in immature male rats.