Abstract
Objective
To examine the effects of common treatments for polycystic ovary syndrome (PCOS) on a panel of hormones (reproductive/metabolic).
Design
Secondary analysis of blood from a randomized controlled trial of three 16-week preconception interventions designed to improve PCOS-related abnormalities: continuous oral contraceptive pills (OCPs, N = 34 subjects), intensive lifestyle modification (Lifestyle, N = 31), or a combination of both (Combined, N = 29).
Materials and Methods
Post-treatment levels of activin A and B, inhibin B, and follistatin (FST), as well as Insulin-like growth factor 1 (IGF-1), insulin-like growth factor binding protein 2 (IGFBP-2), glucagon, glucagon-like peptide 1 (GLP-1) and 2, and oxyntomodulin were compared to baseline, and the change from baseline in these parameters were correlated with outcomes.
Results
Oral contraceptive pill use was associated with a significant suppression in activin A, inhibin A, and anti-mullerian hormone (AMH), but a significant increase in FST. IGF-1, IGFBP-2, glucagon, and GLP-2 levels were significantly decreased. Oxyntomodulin was profoundly suppressed by OCPs (ratio of geometric means: 0.09, 95% confidence interval [CI]: 0.05, 0.18, P < 0.001). None of the analytes were significantly affected by Lifestyle, whereas the effects of Combined were similar to OCPs alone, although attenuated. Oxyntomodulin was significantly positively associated with the change in total ovarian volume (rs = 0.27; 95% CI: 0.03, 0.48; P = 0.03) and insulin sensitivity index (rs = 0.48; 95% CI: 0.27, 0.64; P < 0.001), and it was inversely correlated with change in area under the curve (AUC) glucose [rs = -0.38; 95% CI: -0.57, -0.16; P = 0.001]. None of the hormonal changes were associated with live birth, only Activin A was associated with ovulation (risk ratio per 1 ng/mL increase in change in Activin A: 6.0 [2.2, 16.2]; P < 0.001).
Conclusions
In women with PCOS, OCPs (and not Lifestyle) affect a wide variety of reproductive/metabolic hormones, but their treatment response does not correlate with live birth.
Keywords: obesity, inflammation, incretins, hyperandrogenism, insulin resistance
Polycystic ovary syndrome (PCOS) is described as both a reproductive and a metabolic condition (1). The metabolic aspect typically presents as increased weight gain and insulin resistance, with an increased likelihood of diabetes, hyperlipidemia, and overall elevated cardiovascular disease risk. The reproductive disorder, which defines the syndrome, typically manifests as hyperandrogenism, irregular ovulation, and polycystic-appearing ovaries, which in turn may result in decreased fertility. Common current chronic treatments that target the metabolic and reproductive aspects of the syndrome include, respectively, metformin or oral contraceptives. These treatments have differential impacts on both reproductive and metabolic function. An ideal treatment would preferentially improve both, but given the lack of a concrete single therapy, women with PCOS are often treated with combined, presumably complementary therapies. These treatments have varying effects alone or in combination on circulating hormones, which are linked to both reproductive and metabolic functions. There has been great interest in hormones of the TGF-Beta superfamily as well as the incretin family given their association with both the pathophysiology of PCOS as well as potential genetic links (2–5).
One such incretin hormone that may contribute to metabolic dysfunction in PCOS is oxyntomodulin (OXM). Oxyntomodulin is a peptide hormone, which is a byproduct of proglucagon processing by the L-cells in the small intestines (6). It has weak agonist activity at both the glucagon and GLP-1 receptors, though its current mechanism of action is unknown, as no specific receptor for the hormone has been identified. When administered to humans, OXM leads to appetite suppression, decreased energy utilization, and weight loss (7, 8). Furthermore, OXM is believed to reduce gastric acid secretion through an unknown mechanism. No specific reproductive function for OXM has been identified to date.
Follistatin (FST) is thought primarily to be a reproductive hormone that inhibits secretion of members of the TGF-β family of proteins, including activin and inhibin. Furthermore, FST seems to function in folliculogenesis as well as the development of the corpus luteum (9). Follistatin is believed to play a metabolic role in peripheral tissue in mouse models through the inhibition of myostatin, leading to increased muscle mass (10, 11). Recent data shows that FST is primarily produced in the liver and is tightly regulated by a glucagon-to-insulin ratio, suggesting that FST has a role in energy metabolism (12).
While there have been multiple reports of the association of these hormones with characteristics of women with PCOS at baseline (13, 14), there are a paucity of studies that have examined the effects of common treatments for PCOS on these hormones. In order to identify possible reproductive and metabolic functions of these newer hormones, we examined the effects of treatments specifically targeting either reproductive (ie, oral contraceptive pills [OCPs]) or metabolic (ie, through weight loss and intensive life style modification) or both (OCPs and weight loss) on a selection of these hormones in women with PCOS. This is a secondary analysis of changes in hormone levels from a randomized, controlled trial that was conducted in women with PCOS (OWL-PCOS, NCT00704912) (15).
Methods
The OWL-PCOS study randomized patients into 1 of 3 treatment arms: lifestyle modification for weight loss (Lifestyle), oral contraceptive only (OCP), and combined oral contraceptive and lifestyle (Combined) (15). The women enrolled in the study had a diagnosis of PCOS, a body mass index (BMI) of 27 to 42 kg/m2, and were 18–40 years old. Polycystic ovary syndrome was defined using the Rotterdam criteria, which requires 2 of the following symptoms: anovulation, signs of hyperandrogenism, or a polycystic ovary on ultrasound. Hyperandrogenism was determined by the Ferrimen-Galwey score or elevated testosterone levels.
The 3 treatment groups received their designated preconception intervention for 16 weeks. The Lifestyle group had a goal of 7% weight loss. Participants were prescribed a 500 kcal/d deficit diet, encouraged to increase physical activity, and received behavioral modification counseling regarding healthy eating and nutrition per the Diabetes Prevention Program guidelines (16). The Lifestyle group was also given sibutramine at an initial 5 mg/d dose, which, over time, was increased to 15 mg/d. When sibutramine was removed from the market by the Food and Drug Administration (FDA) during the course of the study, it was replaced by orlistat 60 mg/tid. The OCP group received continuous daily treatment with 20 mcg ethinyl estradiol per 1 mg norethindrone acetate. The Combined group was given both treatments (Lifestyle and oral contraceptives).
The primary outcome of the OWL-PCOS study was live birth. At the end of the 16-week treatment phase, all participants were given clomiphene to increase fertility, with the goals of pregnancy and live birth. Other secondary outcomes included the effect of the treatment on the PCOS phenotype, ovulation rates, glucose tolerance, pregnancy weight, dual-energy x-ray absorptiometry (DXA) parameters (including fat composition, total body mass, and lean muscle composition), insulin resistance, and lipid profile, as described previously (15).
The data for this secondary analysis were collected from patients at 1 of the 2 study sites (Penn State M.S. Hershey Medical Center). Of the 112 patients randomized at this site, 94 had samples at baseline and end of intervention and were included in this secondary analysis.
Assays
We utilized thawed serum and plasma samples frozen at -80°C for this secondary analysis. All testing was performed by the Ansh Esoteric Laboratory (CLIA #45D2081313), which is located at Ansh Labs (Webster, Texas). Activins A and AB, inhibins A and B, anti-Müllerian hormone (AMH), insulin-like growth factor I (IGF-I), and insulin-like growth factor binding protein 2 (IGFBP-2) were measured in serum using ELISA kits manufactured by Ansh Labs (Webster, Texas) in conformity with ISO13485:2016 quality guidelines. Follistatin, oxyntomodulin, glucagon, and glucagon-like peptides 1 (GLP-1) and 2 (GLP-2) were measured in plasma using ELISA kits manufactured by Ansh Labs.
Anti-Müllerian hormone was measured using an FDA-cleared ELISA (Ansh Labs Product #AL0.124), which recognizes the noncovalently associated prohormone and mature AMH, which is the major form of AMH in human serum. Values obtained using this method are traceable to an Ansh Labs in-house recombinant human AMH. The limit of detection determined according to Clinical and Laboratory Standards Institute (CLSI) guideline EP17 is 3 pg/mL (17). The calibrated range of the assay is 6 to 1150 pg/mL (0.003–1.15 ng/mL), and the extended range has been validated to 23 000 pg/mL (23 ng/mL) (18). Intra-assay coefficient of variation (CV) was <4% and inter-assay CVs for concentrations of approximately 20, 90, and 400 pg/ml were 5.8%, 3.2%, and 4.3%, respectively (19). Masked replicate quality control aliquots tested across study specimens have <9% CV (18, 19). The values reported in this study are based on the ANSH assay that was run specifically for this study, as we previously reported AMH levels in the primary outcome paper based on the Beckman Coulter assay (15).
Inhibin A was measured using an FDA-cleared ELISA (Ansh Labs Product #AL0.123). The inhibin A calibrators supplied are traceable to the World Health Organization (WHO) International preparation NIBSC code 91/624 version 3.0. The limit of detection (95% confidence) is 6 pg/mL, and the calibrated assay range is from the limit of detection (6 pg/mL) to 1100 pg/mL. The intra-assay CV is <6% and the inter-assay CV <8% (20).
Inhibin B was measured using a highly specific ELISA for the measurement of inhibin B in human serum, plasma, or follicular fluid (Ansh Labs Product #AL0.107). Inhibin B values generated by this method are traceable to human inhibin B WHO 96/784 reference material. The limit of detection (95% confidence) is 1.6 pg/mL. The calibrated range of the assay is roughly 2 to 1300 pg/mL. The intra-assay CV was <4% and the inter-assay CV <6.3% (20, 21).
Activin A was measured using a three step double monoclonal antibody ELISA (Ansh Product #AL0.110). The limit of quantitation as established according to CLSI EP17 guideline is 0.07 ng/mL (17). The calibrated range of the assay is roughly 0.6 to 20 ng/mL. The intra-assay CV is <5% and the inter-assay CV <7% (22).
Activin AB was measured using a 3-step double monoclonal antibody ELISA (Ansh Product #AL0.153). The limit of detection is 0.12 pg/mL. The calibrated range of the assay is roughly 1.4 to 100 pg/mL. The intra-assay CV was <5% and the inter-assay CV <7%.
Ansh FST ELISA (Product #AL0.117) was used to measure FST. The monoclonal antibodies used in this assay are directed at domains I and III, thus the assay measures total FST in circulation. Values (ng/mL) obtained are traceable to an in-house recombinant FST reference material. The limit of detection (95% confidence) is 0.18 ng/mL. The calibrated range of the assay is roughly 0.6 to 20 ng/mL. The intra-assay CV was <5% and the inter-assay CV <8% (23, 24).
Insulin-like growth factor-I, also known as somatomedin C, was measured using a quantitative 1-step sandwich type immunoassay (Ansh Product #AL0.121) (25). The assay monoclonal antibodies used specifically measure only free IGF-I. The reagent kits supplied by the manufacturer contain preanalytical buffers for the release of IGF-I from all binding proteins and components to prevent rebinding. The preanalytical steps result in a 25-fold dilution of specimen prior to assay. Values were reported as ng/mL of total serum IGF-I. The limit of detection is 2 ng/mL. The calibrated range of the assay is 12 to 800 ng/mL. The intra-assay CV was <4% and the inter-assay CV <9%. The between extraction inter-assay CV was <11%. Values obtained using this assay are traceable to the 1st WHO International Standard for Insulin-like Growth Factor-I, recombinant, human, for immunoassay (NIBSC code: 02/254) (21, 22, 24).
IGFBP-2 was measured using a 3-step double monoclonal antibody ELISA (Ansh Product #AL0.140). The limit of detection is 0.08 ng/mL. The calibrated range of the assay is roughly 0.45 to 16 pg/mL. The intra-assay CV was <5% and the inter-assay CV <6%.
Oxyntomodulin was measured using a highly sensitive quantitative 2-step sandwich type ELISA (Ansh Product #AL0.139) (26). Oxyntomodulin concentration in the calibrators is traceable to the manufacturer’s working calibrators (peptide weight/volume). The limit of detection (95% confidence) is 0.25 pg/mL. The calibrated range of the assay is roughly 3 to 300 pg/mL. An extended range of up 1500 pg/mL has been validated for specimens diluted 5-fold preanalytically. The intra-assay CV was <4% and the inter-assay CV <5%.
Glucagon was measured using a quantitative 2-step sandwich-type ELISA (Ansh Product #AL0.157). Glucagon concentration in the calibrators is traceable to the WHO International Standard Glucagon, Porcine NIBSC 69/194. The limit of detection (95% confidence) is 2.4 pg/mL. The calibrated range of the assay is roughly 7 to 300 pg/mL. The intra-assay CV was <5.0% and the inter-assay CV <6%.
Glucagon-like peptides 1 was measured using a quantitative 2-step sandwich-type ELISA (Ansh Product #AL0.172). Values obtained using this method are traceable to the manufacturer’s working calibrators (peptide weight/volume). The limit of detection (95% confidence) is 5.6 pg/mL. The calibrated range of the assay is roughly 28 to 425 pg/mL. The intra-assay CV was <3% and the interassay CV <4%.
Glucagon-like peptides 2 was measured using a quantitative 3-step sandwich-type ELISA (Ansh Product #AL0.174). Glucagon-like peptide 2 concentration in the calibrators is traceable to the manufacturer’s working calibrators (peptide weight/volume). The limit of detection (95% confidence) is 11 pg/mL. The calibrated range of the assay is roughly 140 to 7500 pg/mL. An extended range of up 1500 pg/mL has been validated for specimens diluted 5-fold preanalytically. The intra-assay CV was <6% and the interassay CV <8%.
The quality control parameters of the assays are summarized in Supplemental Table 1 (27).
Table 1.
Baseline characteristics of the study cohort
| OCP(N = 34) | Lifestyle(N = 31) | Combined(N = 29) | |
|---|---|---|---|
| Mean (SD) | Mean (SD) | Mean (SD) | |
| Demographics | |||
| Age (years) | 29.5 (3.2) | 28.6 (3.2) | 28.7 (3.5) |
| Hispanic: N (%) | 2 (5.9) | 1 (3.2%) | 3 (10.3) |
| Caucasian: N (%) | 31 (91.2) | 25 (80.) | 24 (82.8) |
| Black/African-American: N (%) | 1 (2.9) | 1 (3.2) | 2 (6.9) |
| Other/multiracial: N (%) | 2 (5.9) | 5 (16.1) | 3 (10.3) |
| Biometric | |||
| Weight (kg) | 92.9 (14.5) | 96.3 (15.1) | 94.4 (14.3) |
| BMI (kg/m2) | 34.7 (4.0) | 34.9 (4.7) | 34.8 (4.1) |
| Waist (cm) | 105.7 (10.8) | 107.5 (13.8) | 105.5 (10.9) |
| Ultrasound parameters | |||
| Antral follicle count (both ovaries) | 73.0 (38.3) | 57.6 (37.2) | 59.8 (35.4) |
| Total ovarian volume (cm3)a | 22.0 (17.6, 29.6) | 22.1 (13.5, 28.1) | 16.0 (12.8, 23.8) |
| Core laboratory results | |||
| Testosterone (ng/dL)a | 54.7 (41.2, 69.1) | 54.9 (33.4, 76.8) | 51.9 (37.9, 78.7) |
| SHBG (nmol/L)a | 27.4 (20.2, 34.7) | 29.3 (21.6, 38.1) | 28.9 (17.5, 45.0) |
| Cholesterol (mg/dL) | 183.5 (27.5) | 181.3 (34.6) | 183.3 (31.1) |
| HDL (mg/dL)a | 43.0 (39.0, 52.0) | 42.0 (36.0, 46.0) | 42.0 (39.0, 49.0) |
| LDL (mg/dL) | 110.8 (23.8) | 110.4 (32.3) | 109.8 (30.4) |
| Triglycerides (mg/dL)a | 124.0 (94.0, 161.0) | 116.0 (96.0, 171.0) | 120.0 (87.0, 175.0) |
| Fasting glucose (mg/dL) | 86.8 (9.2) | 87.3 (9.6) | 91.0 (13.8) |
| Fasting insulin (uU/mL)a | 22.5 (17.0, 30.0) | 25.0 (19.0, 32.0) | 24.0 (15.0, 28.0) |
| 2-hour glucose (mg/dL) | 113.0 (25.9) | 122.2 (36.8) | 124.3 (42.2) |
| 2-hour insulin (uU/mL)a | 111.0 (81.0, 160.0) | 166.0 (84.0, 248.0) | 120.0 (69.5, 217.0) |
| AUC glucose (mg/dLahour) | 242.9 (42.6) | 262.4 (54.5) | 259.6 (68.4) |
| AUC insulin (uU/mLahour)a | 197.5 (170.0, 295.3) | 269.8 (184.6, 361.4) | 212.3 (127.5, 362.5) |
| Matsuda’s Insulin Sensitivity Indexa | 2.2 (1.8, 3.0) | 1.7 (1.2, 2.4) | 2.0 (1.4, 3.3) |
| DXA parameters | |||
| Total BMD (g/cm2) | 1.18 (0.09) | 1.17 (0.08) | 1.14 (0.07) |
| Fat (kg) | 41.5 (8.6) | 42.2 (8.6) | 42.6 (9.2) |
| Lean (kg) | 49.5 (6.5) | 52.0 (7.7) | 49.8 (5.9) |
| Fat (%) | 45.3 (3.6) | 44.6 (3.8) | 45.8 (4.0) |
| Reproductive hormones | |||
| AMH (ng/mL) | 13.7 (6.3) | 10.8 (6.4) | 12.1 (5.6) |
| Activin A (ng/mL)a | 0.7 (0.5, 0.8) | 0.6 (0.5, 0.6) | 0.5 (0.5, 0.6) |
| Inhibin A (pg/mL)a | 8.5 (5.7, 15.0) | 8.0 (5.7, 11.0) | 9.5 (5.7, 13.0) |
| Inhibin B (pg/mL) | 109.1 (84.3) | 86.5 (44.8) | 89.3 (53.6) |
| Plasma follistatin (ng/mL) | 2.7 (1.4) | 2.7 (1.0) | 2.6 (0.8) |
| Metabolic hormones | |||
| IGF-1 (ng/mL) | 253.6 (99.2) | 220.4 (85.9) | 248.5 (113.4) |
| IGFBP-2 (ng/mL)a | 7.1 (3.8, 11.8) | 6.0 (3.4, 12.3) | 4.4 (3.0, 13.3) |
| Glucagon (pg/mL)a | 8.5 (3.5, 15.5) | 10.0 (3.5, 15.0) | 7.0 (3.5, 20.5) |
| GLP-1 (pg/mL) | 157.3 (90.7) | 158.3 (69.4) | 154.1 (75.4) |
| GLP-2 (pg/mL) | 1.6 (0.8) | 1.9 (0.9) | 1.8 (0.7) |
| Plasma oxyntomodulin (pg/mL)a | 103.5 (15.5, 216.5) | 80.0 (4.0, 205.0) | 53.0 (13.0, 179.0) |
Abbreviations: AMH, anti-mullerian hormone; AUC, area under the curve; BMD, bone mineral density; BMI, body mass index; DXA, dual energy x-ray absorptiometry; GLP-1, glucagon-like peptide 1; GLP-2, glucagon-like peptide 2; HDL, high density lipoprotein; IGF-1, Insulin-like growth factor 1; IGFBP-2; insulin-like growth factor 2; LDL, low desnity lipoprotein; OCP, oral contraceptive pills; SD, standard deviation; SHBG, sex hormone binding globulin.
a Median (25th percentile, 75th percentile) are reported.
Statistical analysis plan
Calculation of the insulin sensitivity index (ISI) followed the methods outlined by Matsuda and DeFronzo (28). The area under the curve (AUC) for glucose and insulin, obtained from oral glucose challenge test (OGTT) measurements, were calculated per patient using the trapezoidal rule. Ovarian volume was calculated using the formula for a prolate ellipsoid (length × width × height × [π/6]).
Contrasts were constructed from linear mixed-effects models (29), which account for the correlation of multiple visits for each patient, to assess change from baseline to end of intervention within each treatment group, as well as differences between the treatment groups, with respect to continuous variables (eg, assay measurements). Effect sizes are reported as the difference in means and 95% confidence intervals (CI). If the data were not normally distributed, a logarithmic transformation was applied prior to analysis and then back-transformed after analysis to their original scale for ease of interpretation. This back-transformation results in the data being reported as the ratio of geometric means, with associated 95% CI.
Spearman correlation coefficients, which dampen the impact of outlying observations, were used to assess the association between the change in hormone levels being studied and the change in biometric and biochemical parameters previously reported in our primary outcome paper. Results are reported as the Spearman correlation coefficient (rs) and 95% CI. A binary regression model with a complementary log-log link function was used to estimate the relative risk (RR) of the change in reproductive and metabolic hormones with respect to the outcomes of live birth and ovulation. All hypothesis tests were 2-sided, and all analyses were performed using SAS software, version 9.4 (SAS Institute Inc., Cary, North Carolina). All supplementary material and figures are located in a digital research materials repository (27).
Results
Ninety-four participants (OCP n = 34, Lifestyle n = 31, Combined n = 29) were included in this analysis. Baseline characteristics for the key parameters are found in Table 1. We do not report activin AB levels, as they were nondetectable in the majority of samples. The effects of the 16-week interventions on the reproductive and metabolic hormones, as well as weight and BMI, are found in Table 2. Box and whisker plots of baseline and end of intervention hormonal distributions are provided in Supplemental Figs. 1 and 2 (27). Additionally, we have included a supplementary table (Supplemental Table 2) (27) to identify the changes in the other parameters in this subset of participants that we previously reported in our primary outcome paper from this trial, as there were no substantive differences from the effect size and direction of the interventions on these parameters.
Table 2.
Change in biometrics, reproductive, and metabolic hormones after treatment
| OCP | Lifestyle | Combined | Lifestyle vs OCP | Combined vs OCP | Combined vs Lifestyle | ||||
|---|---|---|---|---|---|---|---|---|---|
| Mean Change from Baseline (95% CI) | P-value | Mean Change from Baseline (95% CI) | P-value | Mean Change from Baseline (95% CI) | P-value | P-value | P-value | P-value | |
| Biometrics | |||||||||
| Change in weight (lbs) | -2.2 (-5.1, 0.4) | 0.11 | -13.4 (-16.3, -10.6) | <0.001 | -16.1 (-19.0, -13.0) | <0.001 | <0.001 | <0.001 | 0.24 |
| Change in BMI (kg/m2) | -0.4 (-0.8, 0.1) | 0.10 | -2.2 (-2.7, -1.7) | <0.001 | -2.7 (-3.2, -2.2) | <0.001 | <0.001 | <0.001 | 0.15 |
| Reproductive hormones | |||||||||
| AMH (ng/mL) | -4.46 (-5.83, -3.10) | <0.001 | -0.81 (-2.24, 0.61) | 0.26 | -3.76 (-5.26, -2.25) | <0.001 | <0.001 | 0.49 | 0.01 |
| Activin A (ng/mL)a | 0.70 (0.63, 0.79) | <0.001 | 1.00 (0.89, 1.12) | 0.97 | 0.84 (0.75, 0.96) | 0.01 | <0.001 | 0.03 | 0.06 |
| Inhibin A (pg/mL)a | 0.59 (0.42, 0.84) | 0.003 | 1.22 (0.85, 1.76) | 0.27 | 0.70 (0.48, 1.02) | 0.07 | 0.005 | 0.51 | 0.04 |
| Inhibin B (pg/mL) | -73.82 (-95.88, -51.75) | <0.001 | -4.26 (-27.37, 18.85) | 0.72 | -48.91 (-73.19, -24.63) | <0.001 | <0.001 | 0.14 | 0.01 |
| Plasma follistatin (ng/mL) | 3.27 (2.69, 3.86) | 0.001 | -0.25 (-0.85, 0.36) | 0.42 | 2.93 (2.30, 3.56) | <0.001 | <0.001 | 0.43 | <0.001 |
| Metabolic hormones | |||||||||
| IGF-1 (ng/mL) | -43.95 (-79.99, -7.91) | 0.02 | 42.13 (4.39, 79.87) | 0.03 | -3.84 (-43.34, 35.66) | 0.85 | 0.001 | 0.14 | 0.10 |
| IGFBP-2 (ng/mL)a | 0.67 (0.51, 0.87) | 0.003 | 1.26 (0.96, 1.66) | 0.10 | 0.72 (0.54, 0.97) | 0.03 | 0.001 | 0.70 | 0.01 |
| Glucagon (pg/mL)a | 0.51 (0.37, 0.71) | <0.001 | 0.76 (0.54, 1.06) | 0.11 | 0.51 (0.36, 0.73) | <0.001 | 0.11 | 0.99 | 0.11 |
| GLP-1 (pg/mL) | 12.96 (-15.17, 41.10) | 0.36 | -7.06 (-36.29, 22.16) | 0.63 | 5.87 (-24.85, 36.60) | 0.70 | 0.33 | 0.74 | 0.55 |
| GLP-2 (pg/mL) | 0.47 (0.18, 0.75) | 0.002 | -0.13 (-0.43, 0.16) | 0.36 | 0.18 (-0.13, 0.49) | 0.26 | 0.004 | 0.17 | 0.15 |
| Plasma oxyntomodulin (pg/mL)a | 0.09 (0.05, 0.18) | <0.001 | 0.90 (0.45, 1.81) | 0.77 | 0.11 (0.05, 0.22) | <0.001 | <0.001 | 0.70 | <0.001 |
Abbreviations: AMH, anti-mullerian hormone; BMI, body mass index; CI, confidence intervals; DXA, dual energy x-ray absorptiometry; GLP-1, glucagon-like peptide 1; GLP-2, glucagon-like peptide 2; IGF-1, Insulin-like growth factor 1; IGFBP-2, insulin-like growth factor 2; OCP, oral contraceptive pills.
a Data were log transformed and change from baseline represents a ratio of geometric means
Figure 1.
Scatterplots of change in follistatin with change in testosterone, total ovarian volume, area under the curve (AUC) glucose, and insulin sensitivity index. Spearman correlations are noted on each figure.
Figure 2.
Scatterplots of change in oxyntomodulin with change in total ovarian volume, area under the curve (AUC) glucose, and insulin sensitivity index. Spearman correlations are noted on each figure.
Both BMI and weight significantly declined after Lifestyle modification and Combined treatment. Of note, almost all hormones with the exception of GLP-1 were significantly altered by the OCP intervention. In the Lifestyle group, almost all of the hormones remained unaffected with the exception of IGF-1, which was significantly increased. In the Combined group, these hormonal parameters more closely followed the effects of OCPs alone; however, unlike the OCP intervention, there were no significant changes in inhibin A, IGF-1, and GLP-2.
The correlation of change in OXM and FST with other biometric and biochemical parameters affected by treatment are reported in Table 3. We focused on these 2 hormones, as they experienced the most significant change from the interventions as well as the limited information of treatment effects on these parameters. Figure 1 presents the significant inverse correlation between the change in FST levels and the change in testosterone levels (rs = -0.47; 95% CI: [-0.62, -0.29]; P < 0.001) as well as the change in ovarian volume (rs = -0.41; 95% CI: [-0.59, -0.19]; P < 0.001). Additionally, a change in FST levels was positively correlated with a change in AUC glucose levels (rs = 0.40; 95% CI: [0.18, 0.59]; P < 0.001) and inversely correlated with the ISI (rs = -0.36; 95% CI: [-0.55, -0.14]; P = 0.002). Figure 2 depicts the significant positive correlation between the change in OXM levels with change in total ovarian volume (rs = 0.27; 95% CI: [0.03, 0.48]; P = 0.03) and the ISI (rs = 0.48; 95% CI: [0.27, 0.64]; P < 0.001) and inverse correlation with change in AUC glucose (rs = -0.38; 95% CI: [-0.57, -0.16]; P = 0.001).
Table 3.
Correlation between change in key parameters and change in assays
| Change in Oxyntomodulin | Change in Follistatin | |||
|---|---|---|---|---|
| Spearman Correlation Coefficient (95% CI) |
P-value | Spearman Correlation Coefficient (95% CI) |
P-value | |
| Change in weight | -0.21 (-0.41, 0.00) | 0.05 | 0.29 (0.08, 0.47) | 0.01 |
| Change in waist circumference | -0.21 (-0.41, 0.00) | 0.05 | 0.26 (0.05, 0.45) | 0.02 |
| Change in antral follicle count | 0.11 (-0.14, 0.34) | 0.40 | -0.36 (-0.55, -0.13) | 0.003 |
| Change in total ovarian volume | 0.27 (0.03, 0.48) | 0.03 | -0.41 (-0.59, -0.19) | <0.001 |
| Change in testosterone | 0.17 (-0.05, 0.37) | 0.12 | -0.47 (-0.62, -0.29) | <0.001 |
| Change in SHBG | -0.21 (-0.40, 0.00) | 0.05 | 0.48 (0.30, 0.63) | <0.001 |
| Change in triglycerides | -0.22 (-0.41, 0.00) | 0.05 | 0.24 (0.03, 0.43) | 0.03 |
| Change in fasting glucose | -0.07 (-0.28, 0.15) | 0.53 | 0.15 (-0.06, 0.35) | 0.17 |
| Change in fasting insulin | -0.17 (-0.37, 0.04) | 0.12 | 0.31 (0.11, 0.50) | 0.003 |
| Change in 2-hour glucose | -0.26 (-0.45, -0.05) | 0.02 | 0.27 (0.06, 0.46) | 0.01 |
| Change in 2-hour insulin | -0.35 (-0.53, -0.14) | 0.001 | 0.21 (-0.01, 0.41) | 0.06 |
| Change in AUC glucose | -0.38 (-0.57, -0.16) | 0.001 | 0.40 (0.18, 0.59) | <0.001 |
| Change in AUC insulin | -0.32 (-0.52, -0.08) | 0.01 | 0.06 (-0.18, 0.30) | 0.63 |
| Change in insulin sensitivity index | 0.48 (0.27, 0.64) | <0.001 | -0.36 (-0.55, -0.14) | 0.002 |
| Change in DXA total BMD | -0.04 (-0.26, 0.19) | 0.73 | 0.03 (-0.20, 0.25) | 0.81 |
| Change in DXA fat | -0.17 (-0.38, 0.06) | 0.15 | 0.29 (0.07, 0.48) | 0.01 |
| Change in DXA % fat | -0.13 (-0.34, 0.10) | 0.28 | 0.27 (0.05, 0.47) | 0.02 |
Abbreviations: AUC, area under the curve; BMD, bone mineral density; CI, confidence intervals; DXA, dual energy x-ray absorptiometry; SHBG, sex hormone binding globulin.
Associations between the change in hormones and reproductive outcomes were examined (Supplemental Table 3) (27). There were no significant associations between change in hormones and live birth. Only change in Activin A was strongly associated with ovulation: risk ratio per 1 ng/mL increase in change in Activin A: 6.0; 95% CI: (2.2, 16.2); P < 0.001.
Discussion
In this study, we examined several reproductive and metabolic hormones that have been implicated in the pathology associated with PCOS to examine their change during common treatments for PCOS, including treatments with both a reproductive and metabolic focus. We were particularly interested in members of the TGF-Beta superfamily as well as incretins. We also explored the correlations of these hormones with change in other measured biometric parameters as well as reproductive outcomes in a hypothesis-generating study. These hormones have previously been studied in untreated women with PCOS, but the effects of common treatments have been only superficially studied. We were surprised by the lack of effect of lifestyle modification with significant weight loss on most of the incretin family, as well as by the extent of oral contraceptive pills on altering both reproductive and metabolic hormones. Finally, with the exception of Activin A and ovulation, we found no association, and by extrapolation, little predictive value of change in these hormones on reproductive outcomes.
Oral contraceptive pill use in women with PCOS resulted in widespread significant changes in almost all of the reproductive and metabolic hormones evaluated, whereas a lifestyle modification intervention designed to produce a modest, yet clinically significant weight loss resulted in minimal changes. The combination of the 2 largely mirrored OCPs alone. However, many changes in the combination group were significantly attenuated, suggesting that the addition of lifestyle modifications may have blunted some of the OCP’s effects. These findings largely support the findings of our primary study, ie, that OCPs can profoundly suppress the hypothalamic-pituitary-ovarian axis in PCOS as well as perturb and challenge the metabolic load in women with obesity and PCOS. Further, it underscores that lifestyle alone has minimal effect on the hormones of the reproductive axis (15). In our original study we did see favorable changes of lifestyle modification with improved insulin sensitivity and body composition, but these favorable changes were not reflected in the metabolic hormones we measured in this study (15).
Members of the TGF-Beta family of hormones share many structural similarities and have diverse functions beyond reproduction in the human. Interestingly, all of the members of this family were significantly impacted by OCPs and none were affected by our lifestyle intervention. The significant suppression of AMH and inhibin B levels has previously been documented in women with PCOS on OCPs (30, 31). This is likely due to follicular suppression. A recent study has also shown a significant increase in FST levels in women on OCPs compared with nonusers similar to our findings (32). Oral contraceptive pill effects on activin A have also been less studied, and the decrease in activin A we noted may be a counterregulatory response to increased activin A binding by the increased FST levels (33). The effects of OCPs on inhibin A are also not well documented, and the significant decrease is counterintuitive given the increase in inhibin A previously shown in the luteal phase in ovulatory women as well as the significant increase that occurs in the follicular phase during gonadotropin ovarian stimulation (33).
The interesting, strong association of change in activin A levels with ovulation provides a potential mechanism for the improved ovulation we noticed in our overall trial in the lifestyle groups (with or without OCPs) compared with OCPs alone (15). That mechanism would be an increase in circulating FSH through increased activin A levels (with the lesser ovulatory response after OCPs alone exacerbated by the significant suppression of activin A). Insufficient FSH relative to luteinizing hormone (LH) has long been thought to be a primary defect leading to anovulation in women with PCOS (34). Furthermore a relative deficit of circulating Activin A relative to FST has been implicated in the development of PCOS (14). We did not measure FSH and LH levels in our study, given the poor correlation of single measures with gonadotropin pulsatility, so we cannot directly address this mechanism in our study.
We found that OCPs led to the suppression of IGF-1 levels, findings that have been replicated with OCPs in premenopausal women (35, 36) as well as combination hormone replacement therapy in postmenopausal women, indicating a primarily hepatic effect (37). Insulin-like growth factor I binding proteins have been shown to increase with OCP use, including IGFBP-1 and -3 (38, 39). In our study, however, IGFBP-2 decreased. Suppression of IGF-1 and decreased bioavailability through increased levels of IGF binding proteins has been associated with increased bone turnover and decreased bone mineral density (40). However in our study we noted increases in bone mineral density as a result of OCP use (15). We attribute the increase in GLP-2 and glucagon as counterregulatory responses to improve insulin sensitivity in response to the adverse effects of OCPs on glycemic control (41).
The most perplexing finding of our study is the profound suppression of oxyntomodulin levels by OCPs. Oxyntomodulin is produced from the glucagon precursor proglucagon and is secreted by the L-cells of the gut. It has orexogenic effects, possibly through binding to the glucagon and the GLP-1 receptor, though its native receptor is unknown. Levels increase following the significant weight loss seen with bariatric surgery, and administration to humans (along with other incretins) has been associated with weight loss and improved glycemia (7, 8, 41). In our study, we found that changes in OXM correlated best with parameters of glycemic control and ovarian volume. The mixed effects of OCP on incretins with orexogenic effects (ie, glucagon and GLP-2 increased, with GLP-1 trending upwards) and OXM markedly suppressed may lead to no significant change in appetite and the weight neutral effect of OCP in our study (15) and as reported previously (39, 42, 43).
The weight loss seen with the lifestyle intervention had few effects on reproductive hormones; this was perhaps anticipated based on the similar lack of changes seen in the reproductive hormones from our main study (15). The lack of significant changes in the incretins may be related to the modest weight loss in our participants, which has been shown to have inconsistent effects (44), compared with changes associated with anatomic disruption and comparatively large weight loss, typically 25% to 35% of total body weight, seen within the first 2 years after undergoing bariatric surgery (41).
The strengths of our study are that this is a secondary analysis of a well-constructed randomized, controlled trial in which subjects were studied at baseline free of, and again after 16 weeks of, 2 first line treatments for PCOS, according to a structured protocol. An additional strength is that assays were obtained on prealiquoted and frozen specimens specifically obtained for ancillary studies and run in a lab under outstanding quality control, with many of them validated in other clinical studies (20–22, 24, 26). Weaknesses include that this is a post hoc study without an a priori hypothesis. Another weakness is that some of the assays are new (eg, oxytomodulin) and still undergoing evaluation of their clinical significance. These may limit the utility of the information obtained at the present time. Additionally, we examined associations between treatment-induced change in circulating levels of these hormones with other parameters we measured in the study, so we can make no conclusions about cause and effect. Our small sample size for reproductive outcomes such as live birth and to a lesser extent ovulation limit the power to detect small, but significant differences in hormonal changes after treatment. Finally, our findings may also only be applicable to women with PCOS, as that was our study population.
Further investigation into these circulating hormones from the incretins and TGF beta superfamily and especially their changes after treatment may shed light on potential mechanisms of treatment response in women with PCOS. There is evidence from our associations that hormones thought primarily to have a reproductive function, may also have a secondary role in regulating metabolic aspects of PCOS, though cause and effect must be established in future studies. For now, based on the reproductive outcomes of our study, we can derive little clinical prognostic value to the measure of these analytes than to surmise that treatments yielding decreased activin A levels would not favor ovulation.
Acknowledgments
Funding Support: This project was supported by the Eunice Kennedy Shriver National Institutes of Child Health and Human Development, National Center for Research Resources, and the National Center for Advancing Translational Sciences at the National Institutes of Health, through Grants R01 HD056510, UL1 TR000127 and U54 HD29834 (UVA Core Ligand Assay Core of the Specialized Cooperative Centers Program in Reproduction). Assays in this article were provided by Ansh Labs, Webster, Texas. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Clinical Trial Registration: clinicaltrials.gov: NCT00704912.
Additional Information
Disclosure Summary: S, P.K., C.S., C.G., K.A., C.C., and A.D. have nothing to disclose. W.D. is a consultant for Boehringer Laboratories. A.K. reports stock ownership in Merck. S.E. is a consultant with AbbVie and is receiving research funds from AbbVie, Ferring, and Obseva. D.S. has consulting relationships with Ethicon, Merz, and NovoNordisk. P.S. is a paid consultant for ANSH Labs. R.L. was a consultant for Ferring, Bayer, Fractyl, and AbbVie during the conduct of the study and received funding from Guerbet. Assays in this article were provided by Ansh Labs, Webster, TX.
Data Availability
All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.
References
- 1. Legro RS, Arslanian SA, Ehrmann DA, et al. ; Endocrine Society . Diagnosis and treatment of polycystic ovary syndrome: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2013;98(12):4565–4592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Tata B, Mimouni NEH, Barbotin AL, et al. Elevated prenatal anti-Müllerian hormone reprograms the fetus and induces polycystic ovary syndrome in adulthood. Nat Med. 2018;24(6):834–846. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Moffett RC, Naughton V. Emerging role of GIP and related gut hormones in fertility and PCOS. Peptides. 2020;125:170233. [DOI] [PubMed] [Google Scholar]
- 4. Gorsic LK, Kosova G, Werstein B, et al. Pathogenic anti-Müllerian hormone variants in polycystic ovary syndrome. J Clin Endocrinol Metab. 2017;102(8):2862–2872. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Urbanek M, Legro RS, Driscoll DA, et al. Thirty-seven candidate genes for polycystic ovary syndrome: strongest evidence for linkage is with follistatin. Proc Natl Acad Sci U S A. 1999;96(15):8573–8578. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Pocai A. Unraveling oxyntomodulin, GLP1’s enigmatic brother. J Endocrinol. 2012;215(3):335–346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Wynne K, Park AJ, Small CJ, et al. Oxyntomodulin increases energy expenditure in addition to decreasing energy intake in overweight and obese humans: a randomised controlled trial. Int J Obes (Lond). 2006;30(12):1729–1736. [DOI] [PubMed] [Google Scholar]
- 8. Wynne K, Park AJ, Small CJ, et al. Subcutaneous oxyntomodulin reduces body weight in overweight and obese subjects: a double-blind, randomized, controlled trial. Diabetes. 2005;54(8):2390–2395. [DOI] [PubMed] [Google Scholar]
- 9. Roberts VJ, Barth S, el-Roeiy A, Yen SS. Expression of inhibin/activin subunits and follistatin messenger ribonucleic acids and proteins in ovarian follicles and the corpus luteum during the human menstrual cycle. J Clin Endocrinol Metab. 1993;77(5):1402–1410. [DOI] [PubMed] [Google Scholar]
- 10. Nakatani M, Takehara Y, Sugino H, et al. Transgenic expression of a myostatin inhibitor derived from follistatin increases skeletal muscle mass and ameliorates dystrophic pathology in mdx mice. Faseb J. 2008;22(2):477–487. [DOI] [PubMed] [Google Scholar]
- 11. Gilson H, Schakman O, Kalista S, Lause P, Tsuchida K, Thissen JP. Follistatin induces muscle hypertrophy through satellite cell proliferation and inhibition of both myostatin and activin. Am J Physiol Endocrinol Metab. 2009;297(1):E157–E164. [DOI] [PubMed] [Google Scholar]
- 12. Hansen JS, Rutti S, Arous C, et al. Circulating follistatin is liver-derived and regulated by the glucagon-to-insulin ratio. J Clin Endocrinol Metab. 2016;101(2):550–560. [DOI] [PubMed] [Google Scholar]
- 13. Vrbikova J, Hill M, Bendlova B, et al. Incretin levels in polycystic ovary syndrome. Eur J Endocrinol. 2008;159(2):121–F 127. [DOI] [PubMed] [Google Scholar]
- 14. Eldar-Geva T, Spitz IM, Groome NP, Margalioth EJ, Homburg R. Follistatin and activin A serum concentrations in obese and non-obese patients with polycystic ovary syndrome. Hum Reprod. 2001;16(12):2552–2556. [DOI] [PubMed] [Google Scholar]
- 15. Legro RS, Dodson WC, Kris-Etherton PM, et al. Randomized controlled trial of preconception interventions in infertile women with polycystic ovary syndrome. J Clin Endocrinol Metab. 2015;100(11):4048–4058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Knowler WC, Barrett-Connor E, Fowler SE, et al. ; Diabetes Prevention Program Research Group . Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002;346(6):393–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Bertone-Johnson ER, Manson JE, Purdue-Smithe AC, et al. Anti-Müllerian hormone levels and incidence of early natural menopause in a prospective study. Hum Reprod. 2018;33(6):1175–1182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Eliassen AH, Zeleniuch-Jacquotte A, Rosner B, Hankinson SE. Plasma anti-Müllerian hormone concentrations and risk of breast cancer among premenopausal women in the nurses’ health studies. Cancer Epidemiol Biomarkers Prev. 2016;25(5):854–860. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Perakakis N, Mougios V, Fatouros I, et al. Physiology of activins/follistatins: associations with metabolic and anthropometric variables and response to exercise. J Clin Endocrinol Metab. 2018;103(10):3890–3899. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Perakakis N, Upadhyay J, Ghaly W, et al. Regulation of the activins-follistatins-inhibins axis by energy status: Impact on reproductive function. Metabolism. 2018;85:240–249. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Andrew M, Liao L, Fujimoto M, Khoury J, Hwa V, Dauber A. PAPPA2 as a therapeutic modulator of IGF-I bioavailability: in vivo and in vitro evidence. J Endocr Soc. 2018;2(7):646–656. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Cabrera-Salcedo C, Mizuno T, Tyzinski L, et al. Pharmacokinetics of IGF-1 in PAPP-A2-deficient patients, growth response, and effects on glucose and bone density. J Clin Endocrinol Metab. 2017;102(12):4568–4577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Panagiotou G, Papakonstantinou E, Vagionas A, Polyzos SA, Mantzoros CS. Serum levels of activins, follistatins, and growth factors in neoplasms of the breast: a case-control study. J Clin Endocrinol Metab. 2019;104(2):349–358. [DOI] [PubMed] [Google Scholar]
- 24. Dauber A, Muñoz-Calvo MT, Barrios V, et al. Mutations in pregnancy-associated plasma protein A2 cause short stature due to low IGF-I availability. EMBO Mol Med. 2016;8(4):363–374. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Anastasilakis AD, Koulaxis D, Upadhyay J, et al. Free IGF-1, intact IGFBP-4, and PicoPAPP-A are altered in acute myocardial infarction compared to stable coronary artery disease and healthy controls. Horm Metab Res. 2019;51(2):112–119. [DOI] [PubMed] [Google Scholar]
- 26. Fothergill LJ, Ringuet MT, Sioras E, et al. Cellular and sub-cellular localisation of oxyntomodulin-like immunoreactivity in enteroendocrine cells of human, mouse, pig and rat. Cell Tissue Res. 2019;375(2):359–369. [DOI] [PubMed] [Google Scholar]
- 27. Shah A, Dodson WC, Kris-Etherton P, et al. Data from: effects of oral contraception and lifestyle modification on incretins and TGF-ß superfamily hormones in PCOS 2020. Penn State Scholarsphere. Deposited 18 August 2020. https://scholarsphere.psu.edu/concern/generic_works/ftt44pn07q. [Google Scholar]
- 28. Matsuda M, DeFronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care. 1999;22(9):1462–1470. [DOI] [PubMed] [Google Scholar]
- 29. Fitzmaurice GM, Laird NM, Ware JH. Applied Longitudinal Analysis. Hoboken, N.J.: John Wiley & Sons, Inc.; 2004. [Google Scholar]
- 30. Kallio S, Puurunen J, Ruokonen A, Vaskivuo T, Piltonen T, Tapanainen JS. Antimüllerian hormone levels decrease in women using combined contraception independently of administration route. Fertil Steril. 2013;99(5):1305–1310. [DOI] [PubMed] [Google Scholar]
- 31. Renier MA, Vereecken A, Van Herck E, et al. Dimeric inhibin serum values as markers of ovarian activity in pill-free intervals. Contraception. 1998;57(1):45–48. [DOI] [PubMed] [Google Scholar]
- 32. Wallner C, Rausch A, Drysch M, et al. Regulatory aspects of myogenic factors GDF-8 and follistatin on the intake of combined oral contraceptives. Gynecol Endocrinol. 2020;36(5):406–412. [DOI] [PubMed] [Google Scholar]
- 33. Bloise E, Ciarmela P, Dela Cruz C, Luisi S, Petraglia F, Reis FM. Activin A in mammalian physiology. Physiol Rev. 2019;99(1):739–780. [DOI] [PubMed] [Google Scholar]
- 34. Rebar R, Judd HL, Yen SS, Rakoff J, Vandenberg G, Naftolin F. Characterization of the inappropriate gonadotropin secretion in polycystic ovary syndrome. J Clin Invest. 1976;57(5):1320–1329. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Blackmore KM, Wong J, Knight JA. A cross-sectional study of different patterns of oral contraceptive use among premenopausal women and circulating IGF-1: implications for disease risk. BMC Womens Health. 2011;11:15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Elkazaz AY, Salama K. The effect of oral contraceptive different patterns of use on circulating IGF-1 and bone mineral density in healthy premenopausal women. Endocrine. 2015;48(1):272–278. [DOI] [PubMed] [Google Scholar]
- 37. de Kraker AT, Kenemans P, Smolders RG, Kroeks MV, van der Mooren MJ. Short-term effects of two continuous combined oestrogen-progestogen therapies on several cardiovascular risk markers in healthy postmenopausal women: a randomised controlled trial. Eur J Obstet Gynecol Reprod Biol. 2009;142(2):139–144. [DOI] [PubMed] [Google Scholar]
- 38. Suikkari AM, Tiitinen A, Stenman UH, Seppälä M, Laatikainen T. Oral contraceptives increase insulin-like growth factor binding protein-1 concentration in women with polycystic ovarian disease. Fertil Steril. 1991;55(5):895–899. [DOI] [PubMed] [Google Scholar]
- 39. Balogh A, Kauf E, Vollanth R, Gräser G, Klinger G, Oettel M. Effects of two oral contraceptives on plasma levels of insulin-like growth factor I (IGF-I) and growth hormone (hGH). Contraception. 2000;62(5):259–269. [DOI] [PubMed] [Google Scholar]
- 40. Amin S, Riggs BL, Melton LJ 3rd, Achenbach SJ, Atkinson EJ, Khosla S. High serum IGFBP-2 is predictive of increased bone turnover in aging men and women. J Bone Miner Res. 2007;22(6):799–807. [DOI] [PubMed] [Google Scholar]
- 41. Perakakis N, Kokkinos A, Peradze N, et al. Circulating levels of gastrointestinal hormones in response to the most common types of bariatric surgery and predictive value for weight loss over one year: Evidence from two independent trials. Metabolism. 2019;101:153997. [DOI] [PubMed] [Google Scholar]
- 42. Legro RS, Pauli JG, Kunselman AR, et al. Effects of continuous versus cyclical oral contraception: a randomized controlled trial. J Clin Endocrinol Metab. 2008;93(2):420–429. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Morin-Papunen LC, Vauhkonen I, Koivunen RM, Ruokonen A, Martikainen HK, Tapanainen JS. Endocrine and metabolic effects of metformin versus ethinyl estradiol-cyproterone acetate in obese women with polycystic ovary syndrome: a randomized study. J Clin Endocrinol Metab. 2000;85(9): 3161–3168. [DOI] [PubMed] [Google Scholar]
- 44. Zouhal H, Sellami M, Saeidi A, et al. Effect of physical exercise and training on gastrointestinal hormones in populations with different weight statuses. Nutr Rev. 2019;77(7): 455–477. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.