0021.9'72W96/$03.00/0 Vol. 81, No.

9
Journal of Clinical Endocrinology and Metabolism Printed m U.S.A.
Copyright 0 1996 by The Endocrine Society

Mode of Pulsatile Follicle-Stimulating Hormone

Secretion in Gonadal Hormone-Sufficient and
-Deficient Women-A Clinical Research Center Study*
R. A. BOOTH, JR., J. Y. WELTMAN, V. I. YANKOV, J. MURRAY?, T. S. DAVISON,
A. D. ROGOL, C. M. ASPLIN, M. L. JOHNSON, J. D. VELDHUIS, AND W. S. EVANS
Departments of Medicine, Pharmacology, and Pediatrics, Health Sciences Center, and National Science
Foundation Center for Biological Timing, University of Virginia, Charlottesville, Virginia 22908

ABSTRACT Cross-correlation analysis showed strongly positive associations be-

To test the hypothesis that FSH is secreted at least in part within tween successively paired serum FSH and LH concentrations in all
discrete secretory bursts in women and that the characteristics of four groups of women. Deconvolution of simultaneously obtained LH
episodic FSH secretion are altered within differing gonadal hormone concentration-time series revealed statistically significant concor-
environments, we measured FSH by immunoradiometric assay every dance (13-25%) between FSH and LH secretory episodes at a lag time
10 min for 24 h in premenopausal women during the early follicular of 0 min in EF, LF, and PM women and when LH secretory bursts led
(EF), late follicular (LF), and midluteal (ML) phases of the menstrual FSH secretory bursts by 10 min in ML phase women. However, as
cycle and in postmenopausal (PM) women (n = 8 in each group). 75-87% of FSH and LH secretory pulses were discordant, we infer the
Secretory events were evaluated using multiparameter deconvolu- operation of distinct control mechanisms in the generation of FSH and
tion. FSH was secreted in an episodic manner, with the number of LH release episodes.
secretory bursts (per 24 h; mean 2 SEM) detected in LF (20 2 0.79) and In summary, these results suggest that FSH is secreted within
PM (20 + 0.90) women being greater than that in EF (16 2 0.88) and discrete secretory bursts in women, that the mass and frequency of
ML (14 z 0.93) women. FSH secretory burst mass (milliinternational FSH secretory bursts differ in women exhibiting various gonadal
units per mL) was significantly higher in PM (12 t- 1.6) than in EF hormone environments, and that FSH and LH secretory bursts occur
(1.8 + 0.21), LF (3.1 2 1.3), or ML (0.8 ? 0.11) women and primarily coincidentally at a higher rate than expected on the basis of chance
reflected a relative increase in the maximal secretory rate rather than alone, but at such a low overall rate of concordance that distinct
increased burst half-duration. The estimated half-life (minutes) of mechanisms probably operate to direct episodic FSH and LH secre-
endogenous FSH in LF women (155 f 18) was shorter than those tory activity. (J Clin Endocrinol Metab 81: 3208-3214, 1996)
calculated in EF (2512 24), ML (277 2 38), and PM (231-+ 18) women.

A LTHOUGH our understanding of the physiological

tions of FSH at the level of the ovary has increased
substantially within the past several decades, knowledge
ac-

of the
ported by recent studies using animal models (4) and in the
human (5-15). Technical limitations
identification and characterization
with regard to the specific
of FSH pulses have, how-
dynamics of FSH secretion and clearance remains incomplete. ever, hampered these studies. Indeed, on mathematical
That FSH release is episodic at least in part was appreciated by grounds, FSHs longer half-life is predicted to make peak de-
investigators undertaking initial studies assessing the pulsatile tection considerably more difficult (16, 17).
nature of LH release (l-3). However, earlier reports empha- To test the hypothesis that FSH is secreted within discrete
sized that FSH pulses appeared minor and of low magnitude bursts in normal women, we applied a multiparameter de-
compared to those documented for LH. Even so, these early convolution procedure to serum FSH concentration-time se-
studies observed possible concordance between the episodic ries. This methodology simultaneously estimates the FSH
release of LH and FSH. The idea that LH and FSH molecules half-life and discrete FSH secretory episodes, thus minimiz-
are discharged concomitantly by gonadotropes has been sup- ing the otherwise strongly confounding effect of a prolonged
FSH half-life on the serum FSH concentration profile. To
Received February 5,1996. Revision received April 24,1996. Accepted examine whether the endogenous gonadal hormone envi-
April 26, 1996. ronment influences FSH burst frequency or amplitude char-
Address all correspondence and requests for reprints to: Dr. William acteristics, or the estimated half-life of secreted FSH, both
S. Evans, Box 511, University of Virginia Health Sciences Center, Char-
lottesville, Virginia 22908. pre- and postmenopausal women were studied. In addition,
* This work was supported in part by NIH Grant RR-00847 to the to address the issue of concordance between FSH and LH
General Clinical Research Center of the University of Virginia, the NIH- secretory burst activity, matched serum LH and FSH con-
sponsored CLINFO Data Reduction Systems, NIH Grant DK-38942 to centration-time series were also appraised using cross-cor-
the University of Virginia Diabetes and Endocrinology Research Center,
relation analysis and coincidence probability testing.
the NSF Center for Biological Timing, Baxter Healthcare Corp. (Round
Lake, IL), Research Career Development Award l-K04-HD-00634 (to
J.D.V.), and Research Career Development Award l-K04-HD-00711 and Subjects and Methods
NIH Grant ROl-AG-05977 (to W.S.E.). Presented in part at the 76th Subjects
Annual Meeting of The Endocrine Society, Anaheim, CA, 1994.
t Supported by a supplement from the Education Human Resources Twenty-four healthy premenopausal women (mean age, 27 yr; range,
Directorate by the NSF Center for Biological Timing. 19-38 yr) and eight healthy postmenopausal women (PM; mean age, 59

3208

Downloaded from https://academic.oup.com/jcem/article-abstract/81/9/3208/2651030

by guest
on 27 November 2017
 FSH SECRETION IN WOMEN 3209

yr; range, 50-70 yr) were studied at the University of Virginia General phases of the menstrual cycle and in PM women was evaluated by the
Clinical Research Center after having provided written informed con- hypergeometric probability distribution (13). The temporal location of
sent approved by the University of Virginia human investigation com- each calculated FSH or LH secretory burst was assigned to the nearest
mittee. Each woman had a normal medical history and physical exam- 10 min blood sample. The expected numbers of purely random coinci-
ination and normal serum concentrations of T, TSH, PRL, and total dences were calculated for the individual pulse frequencies in any given
testosterone. The premenopausal women each gave a history of normal FSH and LH paired series (29). Corresponding P values were computed
menstrual cycles, and ovulation was presumed to have occurred during for the hypothesis that at least the number of observed coincidences
the study cycle when the development of a normal preovulatory follicle occurred on the basis of chance alone. Analysis was performed at several
(mean diameter, 20 mm; range, 16-24 mm) was followed by its disap- lag times (i.e. times separating the LH and FSH burst centers).
pearance, as characterized using transvaginal ovarian ultrasonography.
The premenopausal women were admitted to the General Clinical Re-
search Center during the early follicular phase (EF; days 2-5 of menses; Cross-correlation analysis
n = 8), the late follicular phase (LF; 1-4 days before ovulation; n = 8), As distinct from enumerating whether individual discrete secretory
or the midluteal phase (ML; days 5-8 after ovulation; n = 8) of their events coincide in time, we used cross-correlation analysis to correlate
menstrual cycles. All older women had experienced menopause at least successively paired FSH and LH concentrations over time, as described
1 yr before the study and were admitted only after discontinuing any previously (14). Cross-correlation analysis quantifies the correlation be-
hormone replacement at least 6 weeks earlier. One hour after placement tween two series of equally spaced measurements at various lag times,
of an indwelling heparin cannula into a forearm vein, blood samples The lag is the time in minutes separating the measurements of interest,
were obtained at lo-min intervals for 24 h. In addition, blood samples e.g. a lag of zero denotes that the cross-correlation coefficient is calcu-
were obtained at 6-h intervals and pooled for the measurement of 17/3- lated using all simultaneous (same sample) FSH and LH pairs. A lag of
estradiol and progesterone. On the day of sampling, breakfast, lunch, +lO or -10 min indicates that all FSH and LH pairs are considered, in
and dinner were provided, and water was allowed ad libitum. which the first series leads ( + 10 min) or lags (- 10 min) the second series
by 10 min. The SE of the cross-correlation value was computed as ran =
Assays l/(Jn - k), where n is the number of samples per series (e.g. 145) and
k is the number of sample lags [e.g. zero, one (for a lo-min lag), etc.]. The
Serum concentrations of T, estradiol, progesterone, TSH, PRL, and
significance of all eight cross-correlation values in any given group of
testosterone were determined by RIA as previously described (18, 19).
women (e.g. EF) at any given lag (e.g. FSH follows LH by a lo-min lag)
Serum FSH and LH concentrations were measured in duplicate by
was then tested by the Kolmogorov-Smirnov statistic assuming the null
two-site monoclonal immunoradiometric assay (IRMA; Nichols Labo-
hypothesis that z score transformation of the observed r values (z =
ratories, San Juan Capistrano, CA), using as standards, respectively, the
r/'s~) yielded a random normal distribution of unit SD and zero mean
Second and First International Reference Preparations (20, 21). Dose-
(29). Because multiple comparisons were of interest (cross-correlations
dependent within-assay coefficients of variation (3-13%) were calcu-
at various lags), protected 01 values of P 5 0.01 were used to infer
lated from all 145 samples in each FSH or LH series via a power-function
statistical significance. For graphical purposes, the median and range of
fit (11) to variance-weight the deconvolution fit (22). These assays were
cross-correlation r values at various lags are shown for each group of
shown to correlate well with corresponding Sertoli and Leydig cell
women.
bioassays (21, 23).

Deconvolution analysis Statistical analysis

The biomathematical basis of multiparameter deconvolution analysis ANOVA with Duncans new multiple range test was used to identify
(DECONV) has been previously presented in detail (24, 25). Pulsatile differences in mean concentrations and deconvolution-resolved secre-
FSH and LH release profiles are deconvolved into underlying secretory tory burst attributes among the groups. P < 0.05 was construed as
bursts with a finite duration, maximal secretory rate (amplitude), tem- statistically significant.
poral location, and associated subject-specific half-life dispersed with a
finite SD. Although basal secretion rates can also be modeled mathe-
matically (26), a simple model of purely burst-like gonadotropin release Results
was evaluated here to obviate potentially confounding parameter cross- Serum gonadotropin and gonadal hormone concentrations
correlations (22,27). Prefitting estimates were obtained via a waveform-
independent deconvolution technique (PULSE), as previously described The mean ( ~SEM) serum concentrations of FSH, LH, 17p-
(28). DECONV then successively estimated 1) only the starting (T,) estradiol, and progesterone in the pre- and post-menopausal
concentration and the amplitudes of all putative secretory bursts, 2) step women are shown in Table 1. As expected, mean serum
1 plus burst positions, 3) step 2 plus hormone t,,, and 4) step 3 plus
half-duration (time elapsing in minutes at half-maximal amplitude) concentrations of FSH and LH were higher in the PM women
simultaneously. Peaks were retained only when their amplitudes had than in the premenopausal women during any phase of the
nonzero lower bounds based on joint 95% confidence limits (22, 26). menstrual cycle. In addition, the mean serum LH concen-
tration was higher in the LF phasethan the EF and ML phases
Discrete coincidence testing of the cycle. Mean serum 17P-estradiol concentrations were
The probability of significant nonrandom associations between FSH similar and minimal in EF phase and PM women, maximal
and LH secretory bursts in premenopausal women during the three in LF women, and intermediate in ML women. Mean serum

TABLE 1. Mean (2 SEM) serum concentrations of FSH, LH, 176-estradiol, and progesterone in premenopausal women during the early
follicular phase (EF), late follicular phase (LF), and midluteal (ML) phase of the menstrual cycle and in postmenopausal (PM) women

EF LF ML PM
FSH (W/L)* 6.7 2 0.60 5.2 k 1.00" 2.6 k 0.28 56 2 6.6
LH (W/L)** 3.4 t- 0.54 10.0 2 2.94b 1.8 2 0.30" 25 k 3.3
176-Estradiol (pmol/L) 95 k 23 540 If- 88 246 k 19 70 ir 10
Progesterone (nmol/L) 1.49 IT 0.29 1.75 t 0.54 28.6 2 5.34 1.65 k 0.25
For each hormone, values identified with different superscripts (e.g., b, ) differ significantly (P < 0.05).
* Second International Reference Preparations of human menopausal gonadotropin.
** First International Reference Preparation of human menopausal gonadotropin.

Downloaded from https://academic.oup.com/jcem/article-abstract/81/9/3208/2651030

by guest
on 27 November 2017
 3210 BOOTH ET AL. JCE & M . 1996
Vol81 . No 9

progesterone concentrations were similar in EF, LF, and PM

women and significantly higher in ML women.

Deconvolution of FSH and LH concentration-time series

Representative serum FSH and LH concentration-time
profiles and the deconvolution-resolved discrete FSH and
LH secretory bursts observed in the pre- and postmeno-
pausal women are shown in Fig. 1. The mean (SEM) number
of FSH and LH secretory episodesresolved from the respec-
tive concentration-time curves obtained in EF, LF, and ML
women and in PM women are depicted in Fig. 2. We ob-
served no differences in the mean numbers of FSH US.LH EF LF ML PM
secretory bursts within any phase of the cycle or in PM
FIG. 2. Mean ( ?SEM) numbers of FSH (W) and LH (@I secretory bursts
women. Mean gonadotropin secretory burst frequencies resolved with multiparameter deconvolution from 24-h serum FSH
were similar in EF/ ML women and in LF / PM women. How- and LH concentration-time series obtained from eight premenopausal
ever, mean gonadotropin secretory burst frequencies were women studied in the EF, LF, and ML phases of the menstrual cycle
significantly higher in the LF/PM groups than those in the and from eight PM women. No differences in FSH us. LH burst fre-
quencies were found within any of the four groups. For each exper-
EF/ ML groups. imental group, values identified with different superscripts differ
The characteristics of the FSH and LH secretory bursts are significantly.
shown in Fig. 3. As can be seen,the massof FSH secreted per
burst (integral of the calculated secretory event) did not vary FSH LH
across the menstrual cycle, but was significantly greater in
PM women. The larger mass per secretory episode in PM
women reflected an enhanced maximal secretory rate rather

EARLY FOLLICULAR MID-LUTEAL

Y
TIME (HOURS) TIME (HOURS)

LATE FOLLICULAR POST MENOPAUSAL

EF LF ML PM EF LF ML PM

FIG. 3. Mean (SEM) calculated mass of gonadotropic hormone se-

creted per burst (upper panels), maximal secretory rate (mid&e pan-
els) and burst half-duration (lower panels) characterizing deconvolu-
tion-resolved FSH (left panel) and LH (right panel) secretory events.
Serum FSH and LH concentration-time series were obtained from
premenopausal women during the EF, LF, and ML phases of the
menstrual cycle and from PM women. For each experimental group,
values identified with different superscripts differ significantly (P <
0.05).

TIME (HOURS) TIME (HOURS) than prolonged duration of the secretory bursts. The latter
characteristic did not vary in either the pre- or postmeno-
FIG. 1. Representative serum FSH concentration US. time profiles
(upperpanels) and deconvolution-resolved FSH secretory bursts (Zow- pausal groups. The mass of LH secreted within secretory
erpanels) obtained from EF, LF, and ML premenopausal women and bursts was similar in EF and ML women, greater in LF
PM women. women, and maximal in PM women. The relatively greater

Downloaded from https://academic.oup.com/jcem/article-abstract/81/9/3208/2651030

by guest
on 27 November 2017
 FSH SECRETION IN WOMEN 3211

LH mass per burst in the latter two groups primarily re- a lag time of 0 min (P = 0.0024). Table 2 shows the percent-
flected increased maximal secretory rates rather than a pro- agesof FSH and LH secretory bursts that were coincident in
longed burst half-duration. The LH secretory burst half-du- pre- and postmenopausal women.
ration was greater in ML women than in EF and LF women
or in PM women. Cross-correlation analysis
The deconvolution-estimated half-lives for endogenous
As shown in Fig. 6, cross-correlation (r) values for serum
FSH and LH are shown in Fig. 4. The results suggest that the
FSH and LH concentration-time seriesdisclosedsignificantly
calculated half-life of circulating FSH is relatively shorter
nonrandom and positive correlations between paired FSH
during the LF phase and longer in EF, ML, and PM women.
and LH concentrations over time at one or more lags in all
In contrast, the estimated LH half-life is shortest during the
four groups. In contrast, negative cross-correlations were not
luteal phase, longest in PM women, and intermediate in EF
observed consistently. Maximal correlations typically oc-
and LF women.
curred at a zero or nearly zero lag (e.g. -t 10 min). As shown,
Discrete secretory pulse coincidence there was a sustained cross-correlation over a 30- to 50-min
interval in each of the premenopausal groups. In contrast, a
The results obtained with application of the hypergeomet- significant nonrandom and positive cross-correlation was
ric probability-testing algorithm to FSH and LH time-con- observed only at a lag time of 0 in the PM group.
centration seriesare shown in Fig. 5. During the EF phase,the
number of coincident FSH and LH secretory bursts was Discussion
significantly higher than expected on the basis of chance
alone at a lag time of 0 min, i.e. for simultaneously occurring Although FSH secretion may in part reflect a constitutive,
pulses (P = 0.000001). During the LF phase, coincidence perhaps non-GnRH-dependent, process (30), FSH is also re-
between FSH and LH secretory bursts was found both when leasedwithin pulses, presumably reflecting a GnRH-depen-
the lag time was 0 min (P = 0.000011)and when LH secretory dent regulated secretory pathway. However, the identifica-
bursts led the FSH secretory bursts by 10 min (P = 0.04). tion of discrete FSH pulses is a formidable task (11, 17), in
During the midluteal phase, coincidence was documented great part due to the slow metabolic clearance of the hor-
only when LH led FSH secretory bursts by 10 min (P = mone. For example, whether evaluating true FSH pulses
0.0024), although there was a strong trend toward signifi- identified by electrophysiological monitoring of the medio-
cance at a lag time of 0 min (P = 0.057). In PM women, basal hypothalamus in the rhesus monkey (31) or computer-
significant gonadotropin burst coincidence was observed at simulated FSH pulses, a prolonged half-life impairs peak
detection sensitivity and positive accuracy (17). Certain bi-
omathematical modeling techniques, including multiparam-
FSH eter deconvolution (24), have been designed to decompose
hormone concentration profiles into separable secretion
events and clearance, thus addressing one major factor that
has limited the appraisal of episodic FSH release. To our
knowledge, the current data represent the first systematic
application of deconvolution methodology to jointly esti-
mate secretion and half-life from spontaneous serum FSH
concentration-time seriesin women.
Our results are consistent with the idea that FSH is se-
creted in part within discrete bursts in both pre- and post-
menopausal women. Deconvolution analysis of the simul-
taneously obtained serum LH concentration-time series
LH confirmed our previous observation that this hormone is
likewise secreted within discrete bursts (32) and allowed us
to test the hypothesis that FSH and LH secretory events are
concordant. Moreover, we could estimate the apparent am-
plitude and half-duration of the secretory bursts and the
half-life of secreted hormone. By evaluating women at three
phasesof the menstrual cycle and within the setting of pri-
mary gonadal failure, we were able to assesspotential effects
of the endogenous gonadal hormone environment on pul-
u - satile FSH secretion.
PM
EF LF ML
The observations that mean LH and FSH secretory burst
FIG. 4. Mean ( ~SEM) endogenous gonadotropin half-lives estimated frequencies are similar, and that cross-correlations between
from deconvolution-resolved FSH (upperpanel) and LH (lowerpanel) their respective hormone concentrations over 24 h are
pulse profiles. Serum FSH and LH concentration-time series were strongly positive are consistent with but do not prove that
obtained from premenopausal women during the EF, LF, and ML
phases of the menstrual cycle and from PM women. For each exper-
FSH and LH secretory bursts actually occur simultaneously.
imental group, values identified with different superscripts differ In addition, it is necessaryto show that the observed number
significantly (P < 0.05). of coincident pulses exceedsthat due to chance associations

Downloaded from https://academic.oup.com/jcem/article-abstract/81/9/3208/2651030

by guest
on 27 November 2017
 3212 BOOTH ET AL. JCE & M . 1996
Vol81 . No 9

EARLY FOLLICULAR MID-LUTEAL

X=OBSEREDNMBER
0 = EXPECTED NUMBER

X = OBSERVED NUMBER
0 = EXPECTED NUMBER

FIG. 5. Observed and expected num-

bers of pulse coincidences between FSH
and LH secretory bursts in EF, LF, ML, -20 -10 0 +10 +20
and PM women. X denotes the total (LH leads FSH) (LH lags FSH) (LH leads FSH) (LH lags FSH)

number of coincident bursts found per Lag Time (min) Lag Time (min)
group when FSH and LH secretory
bursts were allowed to lag each other by
0, 10, or 20 min. The expected numbers
of coincidences (based on chance alone)
are denoted by the line of open circles,
LATE FOLLICULAR Post Menopausal
and P values, indicated by interrupted
lines, were computed using a hypergeo- X-0BSEREDNMBER X=OSSEREDNMBER
metric probability density function ap- 0 = EXPECTED NUMBER ; 50 I 0 = EXPECTED NUMBER

plied to each set of FSH and LH secre-

tory bursts.

10

0 -2b -10 0 +10 +20

(LH leads FSH) (LH lags FSH 1 (LH leads FSH) (LH IagsFSH)

Lag Time (min) Lag Time (min)

TABLE 2. Percentage of coincident FSH and LH secretory bursts sitivity and/or specificity of LH and FSH peak detection
in premenopausal women during the early follicular phase (EF), could contribute to discordance.
late follicular phase (LF), and midluteal phase (ML) of the
menstrual cycle and in postmenopausal (PM) women
Certain FSH secretory burst characteristics differed at var-
ious stages of the menstrual cycle and in PM women. FSH
secretory burst frequency was maximal in LF women and in
FSH (%) 23.2 25.3 12.6 21.1 the PM group. Numerous studies have inferred that the
LH (%) 24.4 25.2 13.1 21.0 GnRH pulse generator is firing frequently when under max-
FSH (%) denotes the fraction (expressed as a percentage) of FSH imal estrogen positive feedback (e.g. during the LF phase) (5,
secretory events accompanied by (i.e., coincident with) a concomitant 6, 8, 9, 32) and when unrestrained by estrogen (e.g. in the
LH secretory burst. LH (%) denotes the fraction (expressed as a gonadal hormone-deficient PM state) (33,34). However, the
percentage) of LH secretory events accompanied by (i.e. coincident
with) a concomitant FSH secretory burst. observation that similar numbers of FSH (and LH) secretory
events were identified during the EF and ML phases of the
menstrual cycle is intriguing, given the long-held belief that
alone (13,29). We documented that the number of coincident the GnRH pulse generator slows markedly under the influ-
FSH and LH secretory episodes in PM women and the EF and ence of luteal phase concentrations of serum progesterone.
LF phases of premenopausal women significantly exceeds Interestingly, Gennazani and colleagues reported similar
that expected by chance alone. To the extent that LH and FSH findings when Detect (35) was used to identify FSH and LH
bursts are nonrandomly coupled over time, we infer that the pulses in follicular and luteal phase women sampled over a
secretion of both gonadotropins occurs in part in response to 4-h interval (15). Moreover, we (32) and others (7, 8) noted
a common releasing factor (e.g. GnRH) or via temporally that enhanced pulse detection methodologies can resolve a
linked mechanisms, in contrast with entirely independent larger number of LH pulses/ secretory bursts in luteal phase
control mechanisms. Of considerable interest, the absolute women, primarily reflecting lower amplitude (but statisti-
percentages of coincident FSH and LH secretory episodes cally significant) pulses/bursts accounting for up to 70% of
were 13-25%, indicating that 75-87% of secretory bursts the events. These data raise the possibility that GnRH pulse
were nonconcordant. Discordance would probably arise generator activity is not suppressed during the luteal phase
from one or more plausible mechanisms, including 1) inde- to the degree previously thought, but that smaller quantities
pendent neurally derived FSH-releasing activity distinct of GnRH and/or partial inhibition of the gonadotroph re-
from GnRH; 2) intrapituitary para- and autocrine mecha- sponse characteristics may result in a combination of larger
nisms that differentially control FSH and LH secretion via and smaller gonadotropin bursts. Of interest, Filicori and
activin, i&bin, and follistatin; and/or 3) differences in sen- colleagues reported that low amplitude (rather than high
sitivity of LH and FSH secretory mechanisms to feedback amplitude) LH pulses correlated best with progesterone
effects of the gonadal hormones. In addition, unequal sen- release (7).

Downloaded from https://academic.oup.com/jcem/article-abstract/81/9/3208/2651030

by guest
on 27 November 2017
 FSH SECRETION IN WOMEN 3213

FIG. 6. Cross-correlation (r) values for

paired serum FSH and LH concentra-
tion-time series in EF, LF, ML, and PM 0
women. The data in each panel are me-
Lag
dian (and range) values of cross-corre-
lation coeffkients from eight women at
different lags (time in minutes separat-
ing samples of interest). Asterisks de-
note group r values that are significant
at P < 0.01.

The mass of FSH released within secretory bursts was in women, chromatofocusing studies revealed that FSH ob-
similar during the three phases of the menstrual cycle tested, tained at midcycle in the face of high estradiol levels was
but was considerably higher in PM women. The inverse considerably more basic and manifested a shorter half-life
relationship between estradiol levels and FSH secretory than that in the EF or ML phase (40). These investigators
burst amplitude and mass (but not half-duration), inferred concluded that the estrogen milieu and pituitary responsiv-
here for the first time in PM women, is consistent with the ity to GnRH can modulate the release of FSH molecules with
ability of antiestrogen to increase and estrogen infusion to differing biological activities and kinetics. Padmanabhan
decrease FSH secretory burst mass in men (36). In contrast, and colleagues (41), employing a rat Sertoli cell bioassay,
the mass of LH released per secretory burst was considerably reported that women at midcycle secreted FSH of higher
higher in both the LF phase and PM women, due to a greater biological activity, with more basic isoforms represented. In
maximal LH secretory rate (burst amplitude) rather than a contrast, Jia and colleagues (42), using a rat granulosa cell
prolonged burst half-duration. The enhanced LF phase LH assay, found a trend toward a decreased bioactive/immu-
secretory burst amplitude (maximal rate) observed here noactive FSH ratio only at midcycle.
agrees with our earlier findings (32). However, in our pre- Although our FSH half-life estimates are, in fact, similar to
vious study of a different group of women, we used a RIA others determined independently in the literature, caution
rather than an IRMA for LH, and a less stringent version of must be exercised in their interpretation. In addition to iden-
the multiparameter deconvolution algorithm, which sug-
tifying and characterizing individual hormone secretory
gested that the LF phase mean LH secretory burst mass was
bursts, the deconvolution strategy used here in principle
not increased significantly, because increased burst ampli-
could also estimate hormone release, assuming an admixture
tude was accompanied by a decrease in the calculated se-
of nonpulsatile (i.e. basal) and pulsatile secretion (25). How-
cretory burst half-duration. Also of note, although there was
ever, because half-life and secretion parameters are highly
a strong but nonsignificant trend toward a shorter ML LH
correlated, they are difficult to estimate uniquely (22, 27).
half-life in the earlier study, here we confirmed a ML LH
Given that we have no independently derived information
half-life significantly shorter than that observed in the EF and
about basal secretion or a priori FSH half-life in our particular
LF groups.
Our deconvolution-derived monocomponent estimates of groups of women studied, we evaluated the serum FSH
endogenous GnRH-driven FSH half-life in pre- and post- concentration profiles assuming a purely pulsatile model, in
menopausal women are within the range obtained by infu- which basal FSH secretion is negligible. Our simulation stud-
sion of human (urine-derived) FSH in men or women and ies indicate that the presence of a significant basal secretory
subsequent measurement with RIA, IRMA, or bioassay (21, component would reduce the monocomponent half-life es-
37,38) as well as values recently reported by Zambrano and timates somewhat (26). Indeed, we favor the possibility that
colleagues (39). These investigators administered GnRH to the apparently longer half-life of LH in PM (compared to
women at three phases of the menstrual cycle, separated and premenopausal) women studied here reflects the emergence
characterized FSH isoforms by chromatofocusing, and esti- of significantly nonzero basal LH secretion in this endocrine
mated a monocomponent endogenous FSH half-life of ex- milieu. Whether both constitutive (putatively basal) and reg-
ogenous GnRH-stimulated FSH peaks with resultant values ulated (presumptively pulsatile) secretory pathways are, in
of: EF, 222 min; LF, 108 min; and ML, 244 min. Other studies fact, involved in the physiological secretion of FSH in the
suggest that steady state iv estradiol infusions significantly human, and, if so, whether the amount of FSH sorted into
reduced FSH half-lives in healthy young men (36). Moreover, each pathway is dependent on neural and/or gonadal hor-

Downloaded from https://academic.oup.com/jcem/article-abstract/81/9/3208/2651030

by guest
on 27 November 2017
 BOOTH ET AL. JCE & M . 1996
Volt31 . No 9

mone input to the gonadotrope remain to be determined Attenuation of luteinizing hormone secretory burst amplitude as a proximate
basis for the hypoandrogenism of healthy aging in men. J Clin Endocrinol
from additional physiological experiments. Metab. 75:52-58.
21. Urban RJ, Padmanabhan V, Beitins I, Veldhuis JD. 1991 Metabolic clearance
Acknowledgments of human follicle-stimulating hormone assessed by radioimmunoassay, im-
munoradiometric assay, and in vitro Sertoli cell bioassay. J Clin Endocrinol
We thank Sandra Jackson and the nursing staff at the General Clinical Metab. 73:818-823.
Research Center for their expert patient care, Catherine Kern and Ginger 22. Veldhuis JD, Moorman J, Johnson ML. 1994 Deconvolution analysis of neu-
Bauler for technical assistance with the assays, David Boyd for help with roendocrine data: waveform-specific and waveform-independent methods
the data analysis, Paula Azimi for preparation of the artwork, and Dale and applications. Methods Neurosci. 20:279-325.
23. Veldhuis JD, Wilkowski MJ, Urban RJ, Lizarralde G, Iranmanesh A, Bolton
Durrer and Delores Saunders for preparation of the manuscript. WK. 1993 Evidence for attenuation of hypothalamic gonadotropin-releasing
hormone (GnRH) impulse strength with preservation of GnRH pulse fre-
References quency in men with chronic renal failure. J Clin Endocrinol Metab. 76:648-654.
24. Veldhuis JD, Carlson ML, Johnson ML. 1987 The pituitary gland secretes in
1. Midgley Jr AR, Jaffe RB. 1971 Regulation of human gonadotropins. X. Epi- bursts: appraising the nature of glandular secretory impulses by simultaneous
sodic fluctuation of LH during the menstrual cycle. J Clin Endocrinol Metab. multiple-parameter deconvolution of plasma hormone concentrations. Proc
33:962-969. Nat1 Acad Sci USA. 84:7686-7690.
2. Yen SSC, Tsai CC, Naftolin F, Vandenberg G, Ajabor L. 1972 P&a& 25. Veldhuis JD, Johnson ML. 1992 Deconvolution analysis of hormone data.
patterns of gonadotropin release in subjects with and without ovarian function. Methods Enzymol. 210:539-575.
J Clin Endocrinol Metab. 34671-675. 26. Veldhuis JD, Johnson ML. 1995 Specific methodological approaches to se-
3. Santen RJ, Bardin CW. 1973 Episodic luteinizing hormone secretion in man: lected contemporary issues in deconvolution analysis of pulsatile neuroen-
pulse analysis, clinical interpretation, and pathological mechanisms. J Clin docrine data. Methods Neurosci. 28:25-92.
Invest. 522617-2628. 27. Veldhuis JD, Evans WS, Johnson ML. 1995 Complicating effects of highly
4. Irvine CHG, Alexander SL. 1993 Secretory patterns and rates of gonadotropin- correlated model variables on nonlinear least-squares estimates of unique
releasing hormone, follicle-stimulating hormone, and luteinizing hormone parameter values and their statistical confidence intervals: estimating basal
revealed by intensive sampling of pituitary venous blood in the luteal phase secretion and neurohormone half-life by deconvolution analysis. Methods
mare. Endocrinology. 132:212-218. Neurosci. 28:130-138.
5. Backstrom CT, McNeilly AS, Leask RM, Baird DT. 1982 P&a& secretion 28. Veldhuis JD, Iranmanesh A, Johnson ML, Lizarralde G. 1990 Twenty-four
of LH, FSH, prolactin, oestradiol and progesterone during the human men- hour rhythms in plasma concentrations of adenohypophyseal hormones are
strual cycle. Clin Endocrinol (Oxf). 1729-42. generated by distinct amplitude and/or frequency modulation of underlying
6. Soules MR, Steiner RA, Clifton DK, Cohen NL, Aksel S, Bremner WJ. 1984
pituitary secretory bursts. J Clin Endocrinol Metab. 71:1616-1623.
Progesterone modulation of pulsatile luteinizing hormone secretion in normal 29. Veldhuis JD, Johnson ML, Faunt LM, Seneta E. 1994 Assessing temporal
women. J Clin Endocrinol Metab. 58:378-383.
coupling between two, or among three or more, neuroendocrine pulse trains:
7. Filicori M, Butler JP, Crowley Jr WF. 1984 Neuroendocrine regulation of the
cross-correlation analysis, simulation methods, and conditional probability
corpus luteum in the human. Evidence for pulsatile progesterone secretion.
testing. Methods Neurosci. 20:336-376.
J Clin Invest. 731638-1647.
30. Muyan M, Ryzmkiewicz DM, Boime I. 1994 Secretion of lutropin and folli-
8. Reame N, Sauder SE, Kelch Rl, Marshall JC. 1984 Pulsatile gonadotropin
secretion during the human menstrual cycle: evidence of altered frequency of tropin from transfected GH, cells: evidence for separate secretory pathways.
gonadotropin-releasing hormone secretion. J Clin Endocrinol Metab. Mol Endocrinol. 8:1789-1797.
31. Wilson RC, Kesner JS, Kaufman J-M, Uemura T, Akema T, Knobil E. 1984
59:328-337.
9. Filicori M, Santoro N, Merriam GR, Crowley Jr WF. 1986 Characterization of Central electrophysiologic correlates of pulsatile luteinizing hormone secre-
the physiologic pattern of episodic gonadotropin secretion throughout the tion in the rhesus monkey. Neuroendocrinology. 39:256-260.
human menstrual cycle. J Clin Endocrinol Metab. 621136-1144. 32. Sollenberger ML, Carlson EC, Johnson ML, Veldhuis JD, Evans WS. 1990
10. Samuels MH, Veldhuis JD, Henry I, Ridgway EC. 1990 Iathophysiology of Specific physiological regulation of LH secretory events throughout the human
pulsatile and copulsatile release of thyroid-stimulating hormone, luteinizing menstrual cycle: new insights into the pulsatile mode of gonadotropin release.
hormone, follicle-stimulating hormone, and a-subunit. J Clin Endocrinol J Neuroendocrinol. 2:845-852.
Metab. 71:425-432. 33. Rossmanith WG, Liu CH, Laughlin GA, Mortola JF, Suh BY, Yen SSC. 1990
11. Urban RJ, Evans WS, Rogol AD, Kaiser DL, Johnson ML, Veldhuis JD. 1988 Relative changes in LH pulsatility during the menstrual cycle: using data from
Contemporary aspects of discrete peak-detection algorithms. I. The paradigm hypogonadal women as a reference point. Clin Endocrinol (Oxf). 32647-660.
of the luteinizing hormone pulse signal in men. Endocr Rev. 9:337. 34. Veldhuis JD, Evans WS, Rogol AD, Rogol AD, Thorner MO, Stumpf I. 1987
12. Veldhuis JD, Iranmanesh A, Clarke I, Kaiser DL, Johnson ML. 1989 Random Actions of estradiol on discrete attributes of the luteinizing hormone pulse
and non-random coincidence between luteinizing hormone peaks and follicle- signal in man. Studies in postmenopausal women treated with pure estradiol.
stimulating hormone, alpha subunit, prolactin, and gonadotropin-releasing J Clin Invest. 79:769-776.
hormone pulsations. J Neuroendocrinol. 1:185-194. 35. Oerter KE, Guardabasso V, Rodbard D. 1986 Detection and characterization
13. Veldhuis JD, Johnson ML, Seneta E. 1991 Analysis of the copulsatility of of peaks and estimation of instantaneous secretory rate for episodic p&at&
anterior pituitary hormones. J Clin Endocrinol Metab. 73:569-576. hormone secretion. Comput Biomed Res. 19:170.
14. Veldhuis JD, King JC, Urban RJ, et al. 1987 Operating characteristics of the 36. Urban RJ, Dahl KD, Lippert MC, Veldhuis JD. 1992 Modulation of immu-
male hypothalamo-pituitarygonadal axis: pulsatile release of testosterone and noradiometric and bioactive follicle stimulating hormone secretion and clear-
follicle-stimulating hormone and their temporal coupling with luteinizing ance in young and elderly men during treatment with tamoxifen or flutamide.
hormone. J Clin Endocrinol Metab. 65:929-941. J Androl. 13:579-586.
15. Genazzani AD, Petraglia F, Volpogni C, DAmbrogio G, Facchinetti F, 37. Coble Jr YD, Kohler PO, Cargille CM, Ross GT. 1969 Production rates and
Genazzani AR. 1993 FSH secretory pattern and degree of concordance with metabolic clearance rates of human follicle-stimulating hormone in premeno-
LH in amenorrheic, fertile, and postmenopausal women. Am J Physiol. pausal and postmenopausal women. J Clin Invest. 48359-363.
264:E776-E781. 38. Gemzell CA. 1965 Induction of ovulation with human gonadotropins. Recent
16. Veldhuis JD, Lassiter AE, Johnson ML. 1990 Operating behavior of dual or Prog Hoi-m Res. 21:179-204.
multiple endocrine pulse generators. Am J Physiol. 259:E351-E361. 39. Zambrano E, Olivares A, Mendez JP, et al. 1995 Dynamics of basal and
17. Urban RJ, Johnson ML, Veldhuis JD. 1990 In Vim biological validation and gonadotropin-releasing hormone-releasable serum follicle-stimulating hor-
biophysical modeling of the sensitivity and positive accuracy of endocrine mone charge isoform distribution throughout the human menstrual cycle.
peak detection. II. The follicle-stimulating hormone pulse signal. Endocrinol- J Clin Endocrinol Metab. 80:1647-1656.
ogy. 128:2008-2014. 40. Chappel SC, Ulloa-Aguirre A, Coutifaris C. Biosynthesis and secretion of
18. Thomer MO, Schran HF, Evans WS, Rogol AD, Morris JL, MacLeod RM. follicle-stimulating hormone. Endocr Rev. 4:179-211.
1980 A broad spectrum of prolactin suppression by bromocriptine in hyper- 41. Padmanabhan V, Lang LL, Sonstein J, Kelch RP, Beitins IZ. 1988 Modulation
prolactinemic women: a study of serum prolactin and bromocriptine levels of serum follicle-stimulating hormone bioactivity and isoform distribution by
after acute and chronic administration of bromocriptine. J Clin Endocrinol estrogenic steroids in normal women and in gonadal dysgenesis. J Clin En-
Metab. 50:1026-1033. docrinol Metab. 67465-473.
19. Thomer MO, Martin WH, Rogol AD, et al. 1980 Rapid regression of pituitary 42. Jia X-C, Kessel B, Yen SSC, Tucker EM, Hsueh AJW. 1986 Serum bioactive
prolactinomas during bromocriptine treatment. J Clin Endocrinol Metab. follicle-stimulating hormone during the human menstrual cycle and in hyper-
51:438-445. and hypogonadotropic states: application of a sensitive granulosa cell aro-
20. Veldhuis JD, Urban RJ, Lizarralde G, Johnson ML, Iranmanesh A. 1992 matase bioassay. J Clin Endocrinol Metab. 62:1243-1249.

Downloaded from https://academic.oup.com/jcem/article-abstract/81/9/3208/2651030

by guest
on 27 November 2017