0021-972X/92/7403-0600$03.00/0
Journal of Clinical Endocrinology and Metabolism Copyright 0 1992 by The Endocrine Society
Vol. 74, No. 3 Printed
in U.S.A.
Hypothalamic Gonadotropin-Releasing Hormone Secretion and Follicle-Stimulating Hormone Dynamics during the Luteal-Follicular Transition* J. E. HALL?,
D. A. SCHOENFELD,
Reproductive Endocrine Unit, Department Boston, Massachusetts 02114
K. A. MARTIN, of
Medicine,
AND W. F. CROWLEY,
Massachusetts
ABSTRACT. To define the precise neuroendocrine characteristics of the luteal-follicular transition, 11 normal women underwent 12 frequent sampling studies at lo-min intervals for 48 h at various points during the transition from one cycle to the next. Daily blood samples captured both the preceding and subsequent LH surges, so that studies could be characterized in relation to the preceding LH peak (LH+), the subsequent LH peak (LH-), and menses (M). In the frequent sampling study, LH and FSH were measured in all samples, and estradiol (EZ) and progesterone (P) were measured in 2-h pools. The frequency of pulsatile LH secretion increased 4.5-fold over an d-day period spanning the luteal-follicular transition. This increase in LH pulse frequency was strongly related to the preceding LH peak (r = 0.82; P < O.OOOOl),but was not at all related to the onset of menses. When the temporal markers (i.e. LH+, LH-, and M) were removed from the analysis, LH pulse frequency was inversely related to the log of serum P (r = -0.50; P < 0.005), but not EP. FSH levels increased both within the individual studies (P c 0.005) and in the group as a whole over the duration of the luteal-follicular transition. Mean FSH rose 3.5-fold compared to less than a 2-fold increase in mean LH. As with LH pulse frequency, the increase in FSH was most strongly
T
HE FREQUENCY of pulsatile LH secretion has been shown to accurately reflect the frequency of episodic GnRH secretion measured in portal blood in a variety of animal species (l-3). During the luteal-follicular transition of the normal human menstrual cycle, an abrupt change in the frequency of pulsatile GnRH secretion has been inferred from studies demonstrating that the LH pulse frequency is every 4-6 h in the late luteal phase, but increases to every 90 min by the early follicular phase (4). This time of the cycle is also characterized by a dramatic change in the secretory dynamics of FSH, as first described by Ross et al. (5), who noted that the rise in FSH is greater than that in LH and begins before the clinical marker of menses. This rise in FSH is thought to be one of the key determinants in the selection Received March 14, 1991. *This work was supported by NIH Grants HD-15080 and RR01066. t To whom requests for reprints should be addressed.
JR.
General Hospital,
related to the preceding LH peak, but was also significantly associated with the subsequent LH peak and the onset of menses. The relationship between FSH and the number of days from the preceding LH peak is even better tit by a second degree polynomial, which revealed an abrupt increase in LH beginning at LH+ll. With the temporal markers excluded, the increase in FSH related only to LH pulse frequency (r = 0.62; P < 0.001). FSH was not statistically related to the decreases in P or EZr which are also key variables at this stage of the menstrual cycle. We reached the following conclusions. 1) A dramatic increase in LH pulse frequency, and by inference GnRH pulse frequency, accompanies the selective rise in FSH levels during the lutealfollicular transition of the normal menstrual cycle. 2) Both the increase in GnRH pulse frequency and the rise in FSH levels during this transition are strongly related to the preceding LH peak, while the clinical marker of menses is a relatively poor indicator of these events. 3) These studies indicate a high degree of neuroendocrine regulation of this critical transition and suggest that the abrupt increase in GnRH pulse frequency per se may play a significant role in the etiology of the selective rise in FSH that is critical to normal folliculogenesis. (J Clin Endocrinol Metab
74:600-607,1992)
of the subsequent dominant follicle (6). It has been demonstrated that antral follicles responsive to gonadotropins are present in the human ovary in the late luteal phase (7). It has also been shown that the time required for follicular growth in the primate can be condensed by premature removal of the corpus luteum, suggesting that folliculogenesis is restrained by a factor(s) relating to the function of the corpus luteum (8). While it is possible that the increase in FSH during this key transition of the menstrual cycle is due entirely to the waning negative feedback effects of estradiol (E2) and/or inhibin operating directly at the pituitary, it is also possible that the increase in GnRH pulse frequency, resulting from removal of the restraining effect of progesterone (P) on the hypothalamic pulse generator (9, lo), plays a significant role. Previous studies of the luteal-follicular transition in normally cycling women (11-14) have employed short sampling windows and/or suboptimal sampling frequen600
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 10 February 2015. at 02:32 For personal use only. No other uses without permission. . All rights reserved.
LUTEAL-FOLLICULAR ties and, thus, have not permitted a detailed study to be made of the change in GnRH pulse frequency during this transition and its relationship to the rise in FSH. In addition, the prior studies have only referenced the neuroendocrine and gonadal events of this critical transition to menses, which is likely to be an imprecise marker of neuroendocrine changes. We have used prolonged frequent sampling in a combined longitudinal and cross-sectional approach and have related these studies to both preceding and subsequent LH surges in addition to menses. These studies demonstrate a remarkable coordination of the changes in GnRH pulse frequency and levels of FSH and sex steroids which bespeaks a high degree of neuroendocrine regulation in this stage of the menstrual cycle and points to a critical role for both the hypothalamus and the pituitary during the luteal-follicular transition.
Materials Experimental
and Methods
protocol
Twelve studies were conducted in 11 healthy women, aged 29 + 6.1 yr (mean f SD). The women were all euthyroid, normoprolactinemic, and of normal body weight, had a normal exercise history, and were taking no medication. All had regular menstrual cycles of 25-35 days and had ovulated in the cycle before the study, as indicated by a midluteal phase plasma P level above 19 nmol/L and/or a biphasic basal body temperature chart. One subject was studied on two occasions, 10 months apart. The study was approved by the Subcommittee on Human Studies of the Massachusetts General Hospital, and signed informed consent was obtained from each subject. All subjects were required to take ferrous gluconate (300 mg, twice daily) during the 2 study months. Basal body temperature charts were kept by each subject for two full cycles. Daily blood samples were drawn from day 10 after the onset of menses of the first cycle until approximately day 17 of the subsequent cycle. Urinary LH measurements were used to determine the LH peak in both cycles and to indicate when daily blood sampling could cease in the second cycle. Subjects were admitted to the Clinical Research Center of the Massachusetts General Hospital for a 48-h period of frequent sampling on selected days spanning the luteal-follicular transition. Blood was sampled every 10 min through an antecubital iv catheter, and the line was kept patent with a volume of 10 U heparin USP/mL normal saline equal to the volume of blood withdrawn. Vital signs were monitored routinely, and sleep was noted by trained observers. Hemoglobin and hematocrit levels were assessed at 12-h intervals. Sampling was discontinued in four studies (at 240 h) due to a fall in hemoglobin levels below that approved for these studies by the Subcommittee on Human Studies. Daily blood samples were assayed for LH, FSH, EP, and P. Plasma LH and FSH levels were measured in all samples from the frequent sampling study. Two-hour pools were constituted from equal aliquots of the IO-min samples and assayed for EP and P.
TRANSITION
601
Assays
PlasmaLH, FSH, EZ, and P concentrations were measured by RIA, as previously described (9, 15). All sampleswere analyzed in duplicate, and all samplesfrom an individual study were measuredin the sameassay.The intraassay coefficients of variation were obtained from a pool of equal aliquots of the lo-min samplesfor an individual’s study run 20 times throughout that assay and were 6.5% and 7.3% for LH and FSH, respectively. Gonadotropin values are expressedin international units per L, as equivalents of the Second International ReferencePreparation of human menopausalgonadotropin. Data analysis
Pulsatile LH and FSH secretion was analyzed using an objective computer-assistedmodification of the Santen and Bardin method (4) with the amplitude calculated asthe differencebetweenthe peak and the precedingnadir for eachpulse. Undetectablevalueswere assignedthe lowestmeasurableassay value, and missingvalues were ignored. To minimize the false positive rate, each pulse was required to have one point in which the difference from nadir to peak wasgreater than 20% of the nadir and greater than 1 IU/L and a secondpoint that met at least one of these two criteria. In the assaysdescribed, 20% correspondedto 3-4 times the intraassay coefficient of variation. The frequency of pulsesis expressedper 24 h. Pulses of LH and FSH were consideredto be concordant if the FSH peak fell within *20 min of the LH peak. The day of the LH surgewas determined from daily blood samplesin the first and secondcycles, as previously described (4). The frequent samplingstudieswere related to the number of days from the prior LH surge(LH+), the number of days to the subsequentLH surge(LH-), and the onsetof menses(M). The data were alsocombinedwith the 24-h frequent sampling studiesusedto compilethe normative data previously published for the late luteal (LLP; n = 7) and early follicular (EFP; n = 8) phasesby our group (4) and two further 24-h studiesat the transition, which were not included in the previously published series.These previous studies were analyzed in an identical mannerto the current studies.However, only one LH peak was captured in the majority of the previous studies,and therefore, the data are related to mensesand LH+ or LH-. To examine gonadotropin and sex steroid dynamics across the duration of the luteal-follicular transition, the LH pulse frequency, mean LH, mean FSH, mean and log mean EP,and meanand log meanP were correlated with M, LH+, and LHby regressionanalysis,using first and seconddegreepolynomial models.This analysiswas performed using 1) the 48-h studies divided into two 24-h time periods which permitted examination of within- as well as between-subjectchangesin the key variables; 2) the 48-h studiesconsideredasa singleobservation to satisfy strict statistical requirements for independenceof observations; and 3) the 48-h studiescombined with the previously published 24-h studies to determine how well these studiesfit within the context of these data in the LLP and EFP. The significancelevel and r values referred to in the text apply to the initial analysis, but only those relationships that were significant in all analyses are presented. Relationships between various factors were further investigated using step-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 10 February 2015. at 02:32 For personal use only. No other uses without permission. . All rights reserved.
602
HALL
ET AL.
JCE & M. 1992 Vol74.No3
wise regression. Paired t tests were employed to examine changes in pulse frequency and mean FSH between the first and second 24 h of the 48-h studies. To investigate the relationship of FSH to LH, EP, and P within the 48-h studies, a univariate analysis was applied to the 2-h pools for Ez and P and the mean of the lomin time points for the equivalent 2-h period for LH and FSH for each study. For each pair of variables, the significance of the correlations was tested amongthe subjectsusing the Wilcoxon signedrank test (16). Values are expressedas the mean f SEM unlessspecified,and the two-sided0.05 level is construed assignificant unlessotherwisenoted.
Results Daily samples confirmed the relatively selective rise in FSH compared to LH during the luteal-follicular transition (Fig. 1) and permitted precise timing of the frequent sampling study in relation to preceding and subsequent LH peaks and menses (Fig. 2). The cycles preceding and following the frequent sampling study were ovulatory in all studies, with the exception of one subject in whom an Ez peak was noted in the second cycle, but there was no elevation of LH or P before termination of daily sampling. In the first cycle, the mean cycle length was 28.6 f 0.5 days (~sEM), the luteal phase length was 13.7 f 0.5 days (range, lo-16 days), and the peak luteal phase P was 22.3 nmol/L in all subjects. Retrospective 1
150 ?2
I
1
’
8
16
20
24
28
32
36
40
44
TIME (HOURS)
100 0
4
8
50
500
12
150
‘7-l
4i
-
4
1
I
100
0
I
12
16
20
24
28
32
36
40
44
TIME (HOURS)
LH and FSH during the frequent sampling studies (bottom in two individuals (A and B) studied at different points in the luteal-follicular transition. The timing of the frequent sampling study is indicated by the open bar in the toppanel in relation to daily levels of LH (solid line) and FSH (dotted line) and to menses (solid bar) for each subject. The inuerted triangles indicate statistically identified pulses of LH and FSH in the lower panels, and the hat&d bars indicate observed sleep. Note the sleep-related slowing of LH pulses that is typical of the EFP in patient B. FIG.
0
2.
panels) 30
;
-
&
J!
00//o
T
30
100
1000 750
75
4E
500
50
M
250
25
? 5 E 5 o
0
0 DAYS FROM
PRIOR
LH PEAM
1. Mean (SEMI of daily LH, FSH, EO (0), and P (0) levels in the 12 subjects studied, graphed in relation to the preceding and subsequent LH surges. The timing of the 48-h frequent sampling studies in relation to the preceding LH peak is indicated by the open boxes in the top panel. Note the difference in scale between the LH and FSH graphs.
analysis determined that the frequent sampling studies were conducted between 8-17 days from the preceding LH peak (Fig. l), between 4 days before and 1 day after the onset of menses, and between 6-20 days from the subsequent LH peak. Frequent sampling studies
FIG.
Pulsefrequency
The frequency of LH pulses increased dramatically over the luteal-follicular transition. This increase was evident within the individual studies (P < 0.05; Fig. 2)
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 10 February 2015. at 02:32 For personal use only. No other uses without permission. . All rights reserved.
LUTEAL-FOLLICULAR
as well as in the group as a whole (Fig. 3A). Linear regression analysis indicated significant positive correlations of increasing LH pulse frequency with days from the previous LH peak (LH+; r = 0.82; P < 0.00001) and days to the subsequent LH peak (LH-; r = 0.63; P < 0.005), but not with M, and negative correlations with mean Ez (r = -0.57; P < O.Ol), the log of serum Ee (r = -0.58; P < O.Ol), and the log of serum P (r = -0.50); P < 0.005). However, stepwise regression analysis revealed LH+ to be the most important predictor of LH pulse frequency. There was no independent significant relationship of LH pulse frequency with LH- or M or with declining levels of ES or P when LH+ was accounted for. LH pulse frequency increased from approximately 3 IA 15 -
Y-1.17X-6.12 R-0.624 p
Journal of Clinical Endocrinology and Metabolism Copyright 0 1992 by The Endocrine Society
Vol. 74, No. 3 Printed
in U.S.A.
Hypothalamic Gonadotropin-Releasing Hormone Secretion and Follicle-Stimulating Hormone Dynamics during the Luteal-Follicular Transition* J. E. HALL?,
D. A. SCHOENFELD,
Reproductive Endocrine Unit, Department Boston, Massachusetts 02114
K. A. MARTIN, of
Medicine,
AND W. F. CROWLEY,
Massachusetts
ABSTRACT. To define the precise neuroendocrine characteristics of the luteal-follicular transition, 11 normal women underwent 12 frequent sampling studies at lo-min intervals for 48 h at various points during the transition from one cycle to the next. Daily blood samples captured both the preceding and subsequent LH surges, so that studies could be characterized in relation to the preceding LH peak (LH+), the subsequent LH peak (LH-), and menses (M). In the frequent sampling study, LH and FSH were measured in all samples, and estradiol (EZ) and progesterone (P) were measured in 2-h pools. The frequency of pulsatile LH secretion increased 4.5-fold over an d-day period spanning the luteal-follicular transition. This increase in LH pulse frequency was strongly related to the preceding LH peak (r = 0.82; P < O.OOOOl),but was not at all related to the onset of menses. When the temporal markers (i.e. LH+, LH-, and M) were removed from the analysis, LH pulse frequency was inversely related to the log of serum P (r = -0.50; P < 0.005), but not EP. FSH levels increased both within the individual studies (P c 0.005) and in the group as a whole over the duration of the luteal-follicular transition. Mean FSH rose 3.5-fold compared to less than a 2-fold increase in mean LH. As with LH pulse frequency, the increase in FSH was most strongly
T
HE FREQUENCY of pulsatile LH secretion has been shown to accurately reflect the frequency of episodic GnRH secretion measured in portal blood in a variety of animal species (l-3). During the luteal-follicular transition of the normal human menstrual cycle, an abrupt change in the frequency of pulsatile GnRH secretion has been inferred from studies demonstrating that the LH pulse frequency is every 4-6 h in the late luteal phase, but increases to every 90 min by the early follicular phase (4). This time of the cycle is also characterized by a dramatic change in the secretory dynamics of FSH, as first described by Ross et al. (5), who noted that the rise in FSH is greater than that in LH and begins before the clinical marker of menses. This rise in FSH is thought to be one of the key determinants in the selection Received March 14, 1991. *This work was supported by NIH Grants HD-15080 and RR01066. t To whom requests for reprints should be addressed.
JR.
General Hospital,
related to the preceding LH peak, but was also significantly associated with the subsequent LH peak and the onset of menses. The relationship between FSH and the number of days from the preceding LH peak is even better tit by a second degree polynomial, which revealed an abrupt increase in LH beginning at LH+ll. With the temporal markers excluded, the increase in FSH related only to LH pulse frequency (r = 0.62; P < 0.001). FSH was not statistically related to the decreases in P or EZr which are also key variables at this stage of the menstrual cycle. We reached the following conclusions. 1) A dramatic increase in LH pulse frequency, and by inference GnRH pulse frequency, accompanies the selective rise in FSH levels during the lutealfollicular transition of the normal menstrual cycle. 2) Both the increase in GnRH pulse frequency and the rise in FSH levels during this transition are strongly related to the preceding LH peak, while the clinical marker of menses is a relatively poor indicator of these events. 3) These studies indicate a high degree of neuroendocrine regulation of this critical transition and suggest that the abrupt increase in GnRH pulse frequency per se may play a significant role in the etiology of the selective rise in FSH that is critical to normal folliculogenesis. (J Clin Endocrinol Metab
74:600-607,1992)
of the subsequent dominant follicle (6). It has been demonstrated that antral follicles responsive to gonadotropins are present in the human ovary in the late luteal phase (7). It has also been shown that the time required for follicular growth in the primate can be condensed by premature removal of the corpus luteum, suggesting that folliculogenesis is restrained by a factor(s) relating to the function of the corpus luteum (8). While it is possible that the increase in FSH during this key transition of the menstrual cycle is due entirely to the waning negative feedback effects of estradiol (E2) and/or inhibin operating directly at the pituitary, it is also possible that the increase in GnRH pulse frequency, resulting from removal of the restraining effect of progesterone (P) on the hypothalamic pulse generator (9, lo), plays a significant role. Previous studies of the luteal-follicular transition in normally cycling women (11-14) have employed short sampling windows and/or suboptimal sampling frequen600
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 10 February 2015. at 02:32 For personal use only. No other uses without permission. . All rights reserved.
LUTEAL-FOLLICULAR ties and, thus, have not permitted a detailed study to be made of the change in GnRH pulse frequency during this transition and its relationship to the rise in FSH. In addition, the prior studies have only referenced the neuroendocrine and gonadal events of this critical transition to menses, which is likely to be an imprecise marker of neuroendocrine changes. We have used prolonged frequent sampling in a combined longitudinal and cross-sectional approach and have related these studies to both preceding and subsequent LH surges in addition to menses. These studies demonstrate a remarkable coordination of the changes in GnRH pulse frequency and levels of FSH and sex steroids which bespeaks a high degree of neuroendocrine regulation in this stage of the menstrual cycle and points to a critical role for both the hypothalamus and the pituitary during the luteal-follicular transition.
Materials Experimental
and Methods
protocol
Twelve studies were conducted in 11 healthy women, aged 29 + 6.1 yr (mean f SD). The women were all euthyroid, normoprolactinemic, and of normal body weight, had a normal exercise history, and were taking no medication. All had regular menstrual cycles of 25-35 days and had ovulated in the cycle before the study, as indicated by a midluteal phase plasma P level above 19 nmol/L and/or a biphasic basal body temperature chart. One subject was studied on two occasions, 10 months apart. The study was approved by the Subcommittee on Human Studies of the Massachusetts General Hospital, and signed informed consent was obtained from each subject. All subjects were required to take ferrous gluconate (300 mg, twice daily) during the 2 study months. Basal body temperature charts were kept by each subject for two full cycles. Daily blood samples were drawn from day 10 after the onset of menses of the first cycle until approximately day 17 of the subsequent cycle. Urinary LH measurements were used to determine the LH peak in both cycles and to indicate when daily blood sampling could cease in the second cycle. Subjects were admitted to the Clinical Research Center of the Massachusetts General Hospital for a 48-h period of frequent sampling on selected days spanning the luteal-follicular transition. Blood was sampled every 10 min through an antecubital iv catheter, and the line was kept patent with a volume of 10 U heparin USP/mL normal saline equal to the volume of blood withdrawn. Vital signs were monitored routinely, and sleep was noted by trained observers. Hemoglobin and hematocrit levels were assessed at 12-h intervals. Sampling was discontinued in four studies (at 240 h) due to a fall in hemoglobin levels below that approved for these studies by the Subcommittee on Human Studies. Daily blood samples were assayed for LH, FSH, EP, and P. Plasma LH and FSH levels were measured in all samples from the frequent sampling study. Two-hour pools were constituted from equal aliquots of the IO-min samples and assayed for EP and P.
TRANSITION
601
Assays
PlasmaLH, FSH, EZ, and P concentrations were measured by RIA, as previously described (9, 15). All sampleswere analyzed in duplicate, and all samplesfrom an individual study were measuredin the sameassay.The intraassay coefficients of variation were obtained from a pool of equal aliquots of the lo-min samplesfor an individual’s study run 20 times throughout that assay and were 6.5% and 7.3% for LH and FSH, respectively. Gonadotropin values are expressedin international units per L, as equivalents of the Second International ReferencePreparation of human menopausalgonadotropin. Data analysis
Pulsatile LH and FSH secretion was analyzed using an objective computer-assistedmodification of the Santen and Bardin method (4) with the amplitude calculated asthe differencebetweenthe peak and the precedingnadir for eachpulse. Undetectablevalueswere assignedthe lowestmeasurableassay value, and missingvalues were ignored. To minimize the false positive rate, each pulse was required to have one point in which the difference from nadir to peak wasgreater than 20% of the nadir and greater than 1 IU/L and a secondpoint that met at least one of these two criteria. In the assaysdescribed, 20% correspondedto 3-4 times the intraassay coefficient of variation. The frequency of pulsesis expressedper 24 h. Pulses of LH and FSH were consideredto be concordant if the FSH peak fell within *20 min of the LH peak. The day of the LH surgewas determined from daily blood samplesin the first and secondcycles, as previously described (4). The frequent samplingstudieswere related to the number of days from the prior LH surge(LH+), the number of days to the subsequentLH surge(LH-), and the onsetof menses(M). The data were alsocombinedwith the 24-h frequent sampling studiesusedto compilethe normative data previously published for the late luteal (LLP; n = 7) and early follicular (EFP; n = 8) phasesby our group (4) and two further 24-h studiesat the transition, which were not included in the previously published series.These previous studies were analyzed in an identical mannerto the current studies.However, only one LH peak was captured in the majority of the previous studies,and therefore, the data are related to mensesand LH+ or LH-. To examine gonadotropin and sex steroid dynamics across the duration of the luteal-follicular transition, the LH pulse frequency, mean LH, mean FSH, mean and log mean EP,and meanand log meanP were correlated with M, LH+, and LHby regressionanalysis,using first and seconddegreepolynomial models.This analysiswas performed using 1) the 48-h studies divided into two 24-h time periods which permitted examination of within- as well as between-subjectchangesin the key variables; 2) the 48-h studiesconsideredasa singleobservation to satisfy strict statistical requirements for independenceof observations; and 3) the 48-h studiescombined with the previously published 24-h studies to determine how well these studiesfit within the context of these data in the LLP and EFP. The significancelevel and r values referred to in the text apply to the initial analysis, but only those relationships that were significant in all analyses are presented. Relationships between various factors were further investigated using step-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 10 February 2015. at 02:32 For personal use only. No other uses without permission. . All rights reserved.
602
HALL
ET AL.
JCE & M. 1992 Vol74.No3
wise regression. Paired t tests were employed to examine changes in pulse frequency and mean FSH between the first and second 24 h of the 48-h studies. To investigate the relationship of FSH to LH, EP, and P within the 48-h studies, a univariate analysis was applied to the 2-h pools for Ez and P and the mean of the lomin time points for the equivalent 2-h period for LH and FSH for each study. For each pair of variables, the significance of the correlations was tested amongthe subjectsusing the Wilcoxon signedrank test (16). Values are expressedas the mean f SEM unlessspecified,and the two-sided0.05 level is construed assignificant unlessotherwisenoted.
Results Daily samples confirmed the relatively selective rise in FSH compared to LH during the luteal-follicular transition (Fig. 1) and permitted precise timing of the frequent sampling study in relation to preceding and subsequent LH peaks and menses (Fig. 2). The cycles preceding and following the frequent sampling study were ovulatory in all studies, with the exception of one subject in whom an Ez peak was noted in the second cycle, but there was no elevation of LH or P before termination of daily sampling. In the first cycle, the mean cycle length was 28.6 f 0.5 days (~sEM), the luteal phase length was 13.7 f 0.5 days (range, lo-16 days), and the peak luteal phase P was 22.3 nmol/L in all subjects. Retrospective 1
150 ?2
I
1
’
8
16
20
24
28
32
36
40
44
TIME (HOURS)
100 0
4
8
50
500
12
150
‘7-l
4i
-
4
1
I
100
0
I
12
16
20
24
28
32
36
40
44
TIME (HOURS)
LH and FSH during the frequent sampling studies (bottom in two individuals (A and B) studied at different points in the luteal-follicular transition. The timing of the frequent sampling study is indicated by the open bar in the toppanel in relation to daily levels of LH (solid line) and FSH (dotted line) and to menses (solid bar) for each subject. The inuerted triangles indicate statistically identified pulses of LH and FSH in the lower panels, and the hat&d bars indicate observed sleep. Note the sleep-related slowing of LH pulses that is typical of the EFP in patient B. FIG.
0
2.
panels) 30
;
-
&
J!
00//o
T
30
100
1000 750
75
4E
500
50
M
250
25
? 5 E 5 o
0
0 DAYS FROM
PRIOR
LH PEAM
1. Mean (SEMI of daily LH, FSH, EO (0), and P (0) levels in the 12 subjects studied, graphed in relation to the preceding and subsequent LH surges. The timing of the 48-h frequent sampling studies in relation to the preceding LH peak is indicated by the open boxes in the top panel. Note the difference in scale between the LH and FSH graphs.
analysis determined that the frequent sampling studies were conducted between 8-17 days from the preceding LH peak (Fig. l), between 4 days before and 1 day after the onset of menses, and between 6-20 days from the subsequent LH peak. Frequent sampling studies
FIG.
Pulsefrequency
The frequency of LH pulses increased dramatically over the luteal-follicular transition. This increase was evident within the individual studies (P < 0.05; Fig. 2)
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 10 February 2015. at 02:32 For personal use only. No other uses without permission. . All rights reserved.
LUTEAL-FOLLICULAR
as well as in the group as a whole (Fig. 3A). Linear regression analysis indicated significant positive correlations of increasing LH pulse frequency with days from the previous LH peak (LH+; r = 0.82; P < 0.00001) and days to the subsequent LH peak (LH-; r = 0.63; P < 0.005), but not with M, and negative correlations with mean Ez (r = -0.57; P < O.Ol), the log of serum Ee (r = -0.58; P < O.Ol), and the log of serum P (r = -0.50); P < 0.005). However, stepwise regression analysis revealed LH+ to be the most important predictor of LH pulse frequency. There was no independent significant relationship of LH pulse frequency with LH- or M or with declining levels of ES or P when LH+ was accounted for. LH pulse frequency increased from approximately 3 IA 15 -
Y-1.17X-6.12 R-0.624 p
