Luteal Phase Dynamics of Follicle Stimulating and Luteinizing Hormones in Obese and Normal Weight Women

Summary

Objectives

Female obesity is a state of relative hypogonadotrophic hypogonadism. The aim of this study is to examine gonadotrophin secretion and response to GnRH in the luteal phase of the menstrual cycle and to investigate the pharmacodynamics and pharmacokinetics of endogenous and exogenous LH in obese women.

Design

Participants underwent a luteal phase frequent blood sampling study. Endogenous LH pulsatility was observed, gonadotrophin releasing hormone (GnRH) was given in 2 weight based doses, and GnRH antagonist was administered followed by recombinant LH.

Patients

Regularly menstruating obese (n=10) and normal weight (n=10) women

Measurements

Endogenous hypothalamic-pituitary function (as measured by LH pulsatility), pituitary sensitivity (GnRH induced LH secretion), pharmacodynamics of endogenous LH, and pharmacokinetics of exogenous LH were compared between the obese and normal weight groups.

Results

There were no statistically significant differences in endogenous LH pulsatility or pituitary responses to two weight-based doses of GnRH between the obese and normal weight women. There were no differences in the pharmacodynamics of endogenous LH or the pharmacokinetics of exogenous LH between the groups. FSH dynamics did not differ between the groups throughout the study.

Conclusions

The relative hypogonadotrophic hypogonadism of obesity cannot be explained by differences in LH and FSH luteal phase dynamics or differences in endogenous LH pharmacodynamics or exogenous LH pharmacokinetics.

Clinical trial registration number

NCT01457703, www.clinicaltrials.gov

Introduction

Approximately 20% of reproductive-aged women are obese (1). Obesity has a multitude of negative effects on health (2), including a number of negative consequences on reproduction (3).

Excess body weight is associated with a state of relative hypogonadotrophic hypogonadism in both sexes (4–11). Luteinizing hormone (LH) pulse amplitude is lower in obese men (4) and ovulatory obese women in the follicular phase (5) compared to normal weight controls. Additionally, whole menstrual cycle LH (6, 10), whole cycle and follicular phase follicle stimulating hormone (FSH) (6, 7, 10), and whole cycle progesterone (5, 6, 10) are significantly lower in ovulatory obese versus normal weight women. Obesity has also been associated with decreased levels of sex steroids in both sexes (7, 9).

The physiology behind the relative hypogonadotrophic hypogonadism of obesity is not fully understood. Although previous studies point to a defect in LH pulsatility (5), other areas of the hypothalamic-pituitary axis remain underexplored. Obesity is associated with an increase in blood volume (doubling of BMI results in a 30% increase in blood volume) that could lead to a relative dilution of hormone concentrations (12, 13). Pharmacokinetics of gonadotrophins have not been explored in ovulatory obese women, however obese men have been shown to exhibit increased clearance of LH compared to normal weight men (14). Although it is unlikely that gonadotrophins are sequestered in fat, lipophilic sex steroids (i.e. progesterone) might be (13, 15, 16), and this sequestration could be a source of sustained negative feedback on the hypothalamus and pituitary.

Decreased LH pulsatility has been illustrated in the follicular phase in ovulatory obese women (5) but gonadotrophin dynamics have not been evaluated in the luteal phase. We chose to investigate the luteal phase because it has not been characterized in ovulatory obese women. Additionally, low luteal phase pregnanediol glucuronide (Pdg, a urinary progesterone metabolite) excretion seen in ovulatory obese women (5) may be secondary to inadequate luteal phase LH pulsatility. The slower LH pulses in the luteal phase allow for GnRH stimulation testing with a lower chance of endogenous LH pulsatility interfering with the results.

The aims of this investigation are to (a) examine the pattern of gonadotrophin secretion and response to GnRH in the luteal phase of the menstrual cycle of obese women and (b) investigate the pharmacodynamics and pharmacokinetics of endogenous and exogenous LH in obesity. Endogenous LH pulsatility was observed as an indicator of endogenous hypothalamic-pituitary function of obese compared to normal weight women. Gonadotrophin releasing hormone (GnRH) was given in two weight-based doses spanning the physiologic range (17) to compare pituitary sensitivity (GnRH-induced LH secretion) between obese and normal weight women. Finally, pharmacodynamics of endogenous LH were evaluated during the unstimulated study and after GnRH administration. Pharmacokinetic differences were evaluated after GnRH antagonist followed by recombinant LH administration to evaluate possible differences in clearance of exogenous LH. We hypothesized that the relative hypogonadotrophic hypogonadism previously seen in ovulatory obese women in the follicular phase of the menstrual cycle would hold true in the luteal phase.

Materials and Methods

Participants

Regularly menstruating obese (n=10) and normal weight (n=10) women were recruited from the community through campus-wide advertisement from August 2011 through September 2012. Inclusion criteria were: (a) age 18–40 years; (b) obese (≥30kg/m²) or normal (18–25kg/m²) BMI; (c) history of regular menses every 25–40 days; (d) normal baseline prolactin, thyroid stimulating hormone (TSH), and blood count. Participants were excluded if they had a chronic disease or used medication known to affect reproductive hormones, used exogenous sex steroids within the last three months, exercised more than four hours weekly, or were attempting pregnancy. All participants had a baseline physical exam by study personnel and underwent all blood tests at the Clinical and Translational Research Center (CTRC) of the University of Colorado School of Medicine’s Clinical and Translational Sciences Institute (CCTSI). A comprehensive metabolic panel (CMP) and serum pregnancy test were performed, with the CMP repeated at the end of the study.

Two obese participants were excluded from further analysis as outliers. Their LH values throughout the frequent blood sampling were found to be greater than 2 standard deviations above the mean for all participants. Both had elevated serum testosterone levels, indicative of polycystic ovary syndrome. The study was approved by Colorado Multiple Institutional Review Board, and signed informed consent was obtained from each participant prior to participation.

Protocol

A pictorial overview of the protocol is shown in Figure 1. A two-day frequent blood sampling study was scheduled 6 to 10 days after a commercially available urinary LH kit indicated that an ovulatory LH surge was about to occur. On the day of their frequent sampling study, all participants underwent a transvaginal ultrasound to assess antral follicle count and check for the presence a corpus luteum. FSH, LH, and anti-müllerian hormone (AMH) were also checked the day of the frequent sampling study. Day 1 of the study consisted of 12 hours of unstimulated, frequent blood sampling at 10 minute intervals. This was followed by administration of GnRH 25 ng/kg intravenously (IND 7420). Two hours later, GnRH 150 ng/kg was given followed by 2 more hours of frequent blood sampling. GnRH antagonist (cetrorelix 3mg subcutaneously, Cetrotide® EMD Serono) was given at the end of day 1 and the participant slept undisturbed in the inpatient CTRC of the CCTSI until 8am of the following morning. Day 2 consisted of a 6-hour frequent blood sampling study after intravenous administration of a physiologic dose of recombinant LH (lutropin alfa 12.5 IU, Luveris® EMD Serono). All participants also underwent a dual-energy X-ray absorptiometry scan (DXA) (Hologic Discovery W, Apex 4.0.1) after completing the frequent blood sampling study to evaluate body composition.

Figure 1.

Study protocol

Hormone assays

LH and FSH were measured with immunofluorometric assays (DELFIA, Perkin-Elmer) that have been used previously in the authors’ laboratory. The LH intra-assay coefficient of variation (CV) ranged from 2.86–4.05% and the inter-assay CV ranged from 2.62–4.68%. The FSH intra-assay CV range was 4.70–5.28% and the inter-assay CV range was 4.01–8.22%.

Oestradiol, oestrone, progesterone, testosterone were measured with immunoassay (Siemens, Centaur XP). Intra-assay and inter-assay CVs are as follows: estradiol 3.7%, 10.6%, estrone 6.4%, 11.7%, testosterone 1.6%, 3.7%, progesterone 2.6%, 3.6%.

Anti-müllerian hormone was measured with AMH Gen-2 ELISA (Beckman Coulter). Intra-assay CVs ranged from 4.7–6.0% and inter-assay CVs ranged from 5.2–6.3%.

Pulsatile characterization

LH pulsatility was evaluated using a modified Santen-Bardin method as described previously (5, 18). A blinded set of 72 samples of the same serum has been previously run for LH and FSH and subjected to pulsatile hormone analysis using the same gonadotrophin assay and pulse detection method. One false positive, low amplitude LH pulse was detected (0.8 IU/ml) and no false positive FSH pulses were detected.

Pharmacokinetic Analysis

LH data was evaluated by non-compartmental analysis with Phoenix WinNonlin (version 6.2.1, Pharsight). Exposure was determined by calculating the area under the LH concentration-time curve (AUC_0→t) by the trapezoidal rule and calculated for given time intervals: 0–710 minutes for baseline; 720–830 minutes for GnRH 25ng/kg; 840–960 minutes for GnRH 150ng/kg; and 1440–1670 minutes for Luveris. The elimination half-life (t_½) of LH was determined from the elimination phase following Luveris administration.

Statistical Methods

An a priori sample size estimate was performed using follicular phase LH pulse amplitude from a prior study (5) as the measure of interest. With 10 patients in each group, 90% power was present to detect a difference of 0.59 IU/L in LH pulse amplitude using a two-sample t-test and alpha of 0.05.

Endogenous LH was modeled over time by group using a linear mixed effects model in order to use every observation from each participant while accounting for similarities within-person. Patient-level characteristics of endogenous LH pulsatility (patient pulse and amplitude), patient-average LH and FSH, patient-level pharmacokinetic parameters (AUC, t1/2); and DXA measures were compared using t tests or Mann-Whitney tests. Biometric parameters (DXA and anthropometric measurements) and patient-level hormone values (baseline LH, total AFC, and AUC within each phase) were compared graphically and using Pearson’s correlation coefficient. Results of statistical analysis are reported as mean ± standard deviation if a t test was used and as median (25%ile, 75%ile) if a Mann-Whitney test was used. P<0.05 was considered statistically significant. Analysis was conducted using SAS software (v9.2 × 64 platform).

Results

Participant Sample Characteristics

Demographic data is shown in Table 1. The obese women were significantly older than the normal weight women (32.5 ± 4.7 vs. 27.3 ± 2.6 years, p=0.006). FSH, anti-müllerian hormone levels (AMH), and antral follicle counts (AFC), all markers of ovarian reserve (19), did not differ between the two groups. By design, the obese group had a significantly greater BMI than the normal weight group (34.3 (31.8, 38.9) vs. 22.3 (21.1, 22.8) kg/m², p<0.001). As expected, obese women had a significantly greater waist and hip circumference than the normal weight women. The groups did not differ in terms of race or ethnicity, with the majority of participants being Caucasian and non-Hispanic.

Table 1.

Demographic information

	Obese n=10	Normal weight n=10	p
Age (years)	32.5 ± 4.7#	27.3 ± 2.6	0.006
Race			0.08
Caucasian	4 (40)^	9 (90)
African American	3 (30)	0 (0)
Other/ not reported	3 (30)	1 (10)
Ethnicity			1.0
Hispanic	1 (10)	2 (20)
Non-Hispanic	9 (90)	8 (80)
Body mass index (kg/m2)	34.3 (31.8, 38.9)$	22.3(21.1, 22.8)	<0.001
Waist (cm)	104 ± 11	78.2 ± 6.3	<0.001
Hip (cm)	114 (104.3, 128)	92.5 (90, 98)	<0.001
FSH	3.8 (2, 4.2)	3.3 (3, 4.9)	0.7
Anti-müllerian hormone (ng/dl)	1.6 (0.6, 6.2)	5.4 (1.8, 10.3)	0.1
Antral follicle count	16.5 (12, 41.4)	23 (15.7, 50.7)	0.2
Oestradiol (pg/ml)*	374 ± 146	388 ± 113	0.8
Oestrone (pmol/L)*	1301.9 ± 913.6	1642.2 ± 525.2	0.3
Progesterone (nmol/L)*	25.76 ± 11.45	18.76 ± 16.85	0.3
Testosterone (nmol/L)*	1.21 ± 0.70	1.1 ± 0.46	0.5
Pre-prandial Insulin (mU/L)	5.8 ± 3.0	3.5 ± 1.6	0.05
Random glucose (mg/dl)	6.1 ± 1.2	5.2 ± 1.09	0.02

^#mean ± standard deviation,

^{^}frequency (percentage),

^$median (25%ile, 75%ile),

^*serum pooled from unstimulated frequent sampling study

Endogenous LH and FSH Secretion

Figure 2a is a composite graph showing mean circulating LH for the unstimulated portion of the frequent blood sampling study, representing endogenous luteal phase LH pulsatility. Figure 2b is a raw and linear mixed effects model of endogenous LH; age was considered for inclusion in modeling, however was not itself significant, and did not alter conclusions. A linear mixed effects model allows us to use every observation from each participant while accounting for similarities within-person. Although the obese group had a lower average LH at every point in time, LH was not statistically significantly different between BMI groups. Additionally, the groups did not differ with respect to mean LH, pulse frequency, or pulse amplitude (Table 2a). Mean FSH, measured hourly, did not differ between the groups (Table 2a). There was a moderate correlation between AFC and baseline LH, ρ=0.49, p=0.03.

Figure 2.

Figure 2a and b.

a. Composite of mean endogenous LH (± SEM)

b. Raw and linear mixed effects model of endogenous LH; normal weight: blue, obese: red. The dashed lines show each participant’s results and the bold line was estimated using the linear mixed effects model.

Table 2.

2a: Characteristics of endogenous LH pulsatility
	Obese	Normal weight	p
Mean LH	4.1 (2.9, 5.2)#	3.6 (2.7, 9.9)	0.8
LH Pulses per hour	0.3 ± 0.1$	0.3 ± 0.2	0.6
LH Pulse amplitude	4.4 (2.8, 7.6)	5 (3.5, 7.1)	0.5
Mean FSH	3.8 (2.1, 4.2)	3.3 (3, 4.9)	0.7

2b: DXA Results
Whole body fat mass (gm)	39,667 ± 6988	17,199 ± 2540	<0.001
Percent whole body fat (%)	41.8 ± 3.1	27.5 ± 2.3	<0.001
Whole body lean mass (gm)	55,001 ± 7427	45,346 ± 4986	0.004
Percent whole body lean (%)	58.2 ± 2.9	72.5 ± 2.2	<0.001
Trunk fat mass (gm)	19,862 ± 3909	7,037 ± 1516	<0.001
Percent trunk fat (%)	42.3 ± 3	24 ± 3.6	<0.001
Percent visceral fat (%)	50.1 ± 4.5	40.7 ± 3.9	<0.001

^#median (25%ile, 75%ile),

^$mean ± standard deviation

GnRH-Stimulated LH Secretion

Figure 3 shows the composite LH responses to 2 weight-based doses of GnRH. The first dose of GnRH (25 ng/kg) is considered slightly sub-physiologic (17) and the second dose of GnRH (150 ng/kg) is considered slightly supra-physiologic (17). There were no significant differences in the LH concentration-time curves (AUC_0→t) following either dose of GnRH in obese compared to normal weight women. Peak LH, LH increment, and time to peak LH did not differ between the groups after either dose of GnRH. There was a moderate correlation between AFC and response to GnRH 150ng/kg, ρ=0.47, p=0.04.

Figure 3.

Composite LH response to GnRH 25ng/kg (small arrow) and GnRH 150ng/kg (large arrow), mean ± standard error

Exogenous LH Disappearance

Figure 4 shows the composite LH levels 8 hours after suppression with GnRH antagonist (cetrorelix 3mg) and administration of recombinant LH (Luveris 12.5mg). There were no differences between the groups with respect to mean LH, peak LH, or time to peak LH.

Figure 4.

Composite mean LH after GnRH suppression and administration of recombinant LH (purple arrow), mean ± standard error

Pharmacodynamics of Endogenous or Exogenous LH

Pharmacodynamics of endogenous and exogenous LH, as measured by LH concentration-time curve (AUC_0→t), did not differ between the groups. The half-life of exogenous LH, calculated by linear regression, did not differ between the two groups.

Body Composition Analyses

DXA data are shown in Table 2b. As expected, whole body and trunk fat mass and percent whole body fat, trunk fat, and visceral fat are significantly higher in the obese versus normal weight group (Table 2b). Additionally, whole body and percent whole body lean mass are significantly lower in the obese versus normal weight women. There was no correlation between any DXA measurement and unstimulated LH, GnRH-stimulated LH response, or exogenous LH pharmacokinetics.

Discussion

We used luteal phase sampling to examine LH dynamics to take advantage of a slowed endogenous GnRH pulse generator (20), because we intended to administer exogenous GnRH as part of our experimental characterization. However, in contrast to our previous observations in the follicular phase (5), luteal phase differences in mean LH and LH pulse amplitude were not seen. Response to two exogenous GnRH doses spanning the physiologic range (17) were not significantly different by weight. The obese group had relatively consistent responses to GnRH and exogenous LH and their luteal phase LH secretory patterns were similarly consistent. However, the normal weight group displayed wide variation in their endogenous LH pulsatility and in their responses to exogenous GnRH and LH, and thus accounted for a great deal of the variability that obscured an ability to distinguish between the two groups. This degree of variation was somewhat surprising, as we had not seen it in our prior studies (5). There were no correlations between DXA measurements and endogenous LH, response to GnRH, or response to exogenous LH.

Importantly, we did not observe any differences between obese and normal weight women in pharmacodynamics or pharmacokinetics of either endogenous or exogenous recombinant LH. This finding implies that obesity per se does not affect post-translational processing of the LH molecule, nor does obesity appear to cause circulating LH to be lower because of factors such as volume of distribution. While there is no reason to expect sequestration of LH into adipose tissue, it is possible that progesterone may be taken up by the fat tissues of obese women, thereby lowering circulating progesterone (12, 13). Taken together, the data suggest that if circulating LH is distributed in a larger plasma volume in obese women, secretion keeps pace with this increased volume of distribution to maintain the reproductive system in equilibrium. The lack of difference in pharmacodynamics and pharmacokinetics of LH contrasts with a study in obese men that found the endogenous LH half-life to be significantly shorter in obese versus normal weight men, and implies that LH clearance may differ between the sexes (14). Srouji, et al. investigated endogenous and recombinant LH pharmacokinetics in women with PCOS and, similar to our results, found no differences in recombinant LH pharmacokinetics based on BMI (21). However, the obese women with PCOS had accelerated clearance of endogenous LH as evidenced by a decreased half-life (21). It is also possible that the former studies were performed against a background of relatively rapid LH pulse frequency making it difficult to follow individual endogenous LH pulses long enough to calculate robust LH disappearance curves. Thus, our luteal phase sampling paradigm is uniquely valuable for this purpose.

In this study GnRH was given to investigate pharmacokinetics of endogenous LH, thereby bypassing the typical kisspeptin controlled pituitary GnRH secretion (22). There is some evidence that obesity decreases kisspeptin via decreased leptin (23–26). Decreased kisspeptin would decrease GnRH resulting in decreased gonadotropin secretion as seen in the follicular phase (5, 10). However, decreased gonadotrophins in the obese group were not illustrated in this study.

The lack of difference between obese and normal weight women with respect to luteal LH pulse amplitude was unexpected, as a difference has been repeatedly shown in the follicular phase (5, 10). The subfertility associated with the relative hypogonadotrophic hypogonadism of obesity is believed to be secondary to luteal phase deficiency (27) and to originate from a relative deficiency of follicular stimulation in the first half of the menstrual cycle such that a poorly cultivated follicle leads to a poorly functioning corpus luteum. However, most overweight and obese women are fertile and although there are hormonal alterations in the follicular phase (5, 10) our results suggest that those differences do not carry over into the luteal phase. This makes sense biologically as the corpus luteum is necessary for pregnancy (28) and therefore for carrying on reproduction of the human race. It is possible that the defect in LH secretion and pituitary response to GnRH is limited to the follicular phase. This is supported by Legro et al., who showed that the most notable change in menstrual function and hormone parameters in the setting of extreme weight loss after bariatric surgery was a shortening of the follicular phase (29). Additionally, our group previously found that obese women had a lower mean LH in the follicular phase compared with normal weight women but no difference in mean LH was seen over an entire menstrual cycle (5).

It is noteworthy that the obese group was significantly older than the normal weight group (mean age obese 32.5 vs. normal weight 27.3 years) and this could impact their gonadotropin parameters. Despite their age difference, the obese groups’ ovarian reserve parameters (FSH, ohestradiol, AMH, and antral follicle counts) did not differ in comparison to the normal weight group making it less likely that diminished ovarian function played a role in the results of this study. AMH is the most useful ovarian reserve parameter (32–34) and did not differ significantly between the groups. This is consistent with a large study investigating age-specific AMH values for over 17,000 women showing the average decrease in AMH between ages 27 (the mean age of our normal weight group) and 32 (the mean age of our obese group) was 1.0 ng/ml (35). Additionally, the study findings did not change when age was taken into account for statistical modeling.

It is important to note that the obese group of women that we studied had neither clinical nor biochemical evidence of polycystic ovary syndrome (PCOS). Aside from having regular cycles, their antral follicle counts, circulating testosterone and AMH levels were normal (36). In fact, the obese group had a lower AMH than the normal weight group (though not statistically significantly so). Women with PCOS have been reported to have elevated AMH levels (37, 38), whereas recent observations of ovulatory obese women without PCOS indicate that AMH is lower in this state (as seen in this study) (39–41). The obese women had higher random glucose levels and pre-prandial insulin levels compared to the normal weight women. Insulin levels increase with body weight (42) but the obese women in this study are not insulin resistant (43).

It is also noteworthy that the obese group’s oestradiol, oestrone, and progesterone levels did not differ from the normal weight group. Similar oestradiol and oestrone levels indicate that negative feedback from increased circulating oestrogens is not the cause for the relative hypogonadotrophic hypogonadism of obesity. Many observations linking obesity to increased circulating oestrogens are derived from data in postmenopausal women (44, 45), thus not taking into account the role of ovarian oestradiol production. In regularly cycling women, it appears that the ovary produces much higher levels of estrogens than does adipose tissue, such that any adipose contribution to the total estrogen pool is minor. Although prior studies report poor Pdg excretion in obese women (5, 10), no difference in serum progesterone levels between the groups was seen in the present study.

Hypothalamic dysfunction does not appear to play a role in the hypogonadotrophic hypogonadism of obesity, since LH pulse frequency is preserved and luteal phase LH pulse amplitude may not be affected by obesity. Additionally, the pharmacodynamics and pharmacokinetics of endogenous and exogenous LH are not different between obese and normal weight women. It may be that, by the time ovulation has occurred, the pathophysiologic events responsible for subfertility have already taken effect, and thus it is more difficult to locate the hypothalamic-pituitary-ovarian axis defect(s) responsible for the problem in the luteal phase. There may also be inadequate LH to progesterone throughput, such that the ovarian response to an equivalent LH pulse produces less progesterone in an obese compared to a normal weight woman. Alternatively, other, non-steroidogenic factors secreted by the corpus luteum may contribute to the adverse reproductive phenotype of obesity.

In summary, the relative hypogonadotrophic hypogonadism associated with obesity may not be caused by differences in luteal phase LH pulsatility, pituitary response to GnRH, or differences in endogenous and exogenous LH pharmacokinetics.