Comparison of resumption of ovulation after cessation of oral contraceptives and medroxyprogesterone acetate in women with polycystic ovary syndrome

Abstract

Objective

Both oral contraceptive pills (OCPs) and cyclic medroxyprogesterone acetate (MPA) are widely used to control menstrual abnormalities in women with polycystic ovary syndrome (PCOS). We aimed to evaluate the chance of ovulation resumption after cessation of OCPs and MPA in women with PCOS.

Methods

A retrospective study was conducted of women with PCOS who were treated with OCPs or cyclic MPA from September 2015 to March 2019. After cessation of medication, ovulation was assessed using basal body temperature and/or measurement of serum progesterone. The odds ratio for ovulation resumption was assessed with multivariable logistic regression. Additionally, doubly robust analysis was performed with inverse-probability-weighted analysis and regression adjustment based on the covariate balancing propensity score to adjust for the effect of covariates on the treatment assignment.

Results

Among 272 women with PCOS, 136 were prescribed OCPs and 136 were prescribed cyclic MPA. Ovulation resumed in 18.4% of women (n = 25) after cessation of MPA and in 24.3% of women (n = 33) after cessation of OCPs. The odds of ovulation resumption in MPA users were comparable with those in OCP users (adjusted odds ratio (aOR) 1.00, 95% confidence interval (CI) 0.89–1.12). After multiple imputation due to missing values, the results did not change substantially (aOR 0.99, 95% CI 0.89–1.10).

Conclusions

Among women with PCOS, MPA users have a similar chance of ovulation resumption as OCP users after cessation of medication. Cyclic MPA can be a good alternative to OCPs in women for whom OCPs are contraindicated or who decline to take OCPs.

Keywords:

• Polycystic ovary syndrome
• ovulation
• oral contraceptive pills
• medroxyprogesterone acetate

Introduction

Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders in women with reproductive age [1,2]. The key feature of PCOS is ovulatory dysfunction, which about two-thirds of women with PCOS complain of [3]. Ovulatory dysfunction can manifest as amenorrhea, oligomenorrhea, or abnormal uterine bleeding [4,5]. In addition, it can result in infertility, endometrial hyperplasia, and endometrial carcinoma [6–8], and menstrual abnormalities caused by ovulatory dysfunction have a negative impact on quality of life and cause psychological morbidity [9]. Therefore, treatment of menstrual abnormalities and resumption of ovulation are crucial in the management of women with PCOS. For women who do not want to get pregnant, oral contraceptive pills (OCPs) are commonly used to treat menstrual abnormalities [4,10–12]. Also, cyclic medroxyprogesterone acetate (MPA) provides good menstrual control in women with PCOS [13] and can therefore be considered in women for whom OCPs are contraindicated, or who decline to take OCPs [13,14].

Women with PCOS often experience resumption of ovulation after cessation of OCPs [15,16] because OCPs suppress the level of circulating androgen and increase the level of sex hormone-binding globulin (SHBG) [17]. Considering that administration of MPA elicits beneficial changes associated with androgen [18], ovulation might also be expected to resume after cessation of MPA. Although a few studies have compared hormonal changes and metabolic parameters after administration of OCPs and MPA in women with PCOS [18], resumption of ovulation after medication cessation has not been compared between OCP and MPA users. Therefore, we aimed to compare the resumption of ovulation after cessation of OCPs and MPA in women with PCOS using a quasi-experimental method.

Materials and methods

Study design and subjects

A total of 1,077 women with a presumptive diagnosis of PCOS visited Reproductive Endocrinology Outpatient Services at Seoul National University Hospital from September 2015 to March 2019. According to the Rotterdam criteria, PCOS was diagnosed when two or three of the following criteria were met: ovulatory dysfunction including oligo-ovulation, anovulation, or anovulatory bleeding, clinical and/or biochemical hyperandrogenism and polycystic ovaries on ultrasound examination [19]. Menstrual intervals longer than 35 days and shorter than 3 months were defined as oligo-ovulation, and amenorrhea for longer than 3 months was defined as anovulation [10]. An irregular period with an interval shorter than 21 days was defined as anovulatory bleeding. The following patients were excluded from the study population (Figure 1): (1) patients who had diseases that might have caused ovulatory dysfunction, including chronic kidney disease, psychiatric disease, pituitary disease, thyroid disease, and Cushing’s syndrome (n = 244); (2) patients who did not fulfill the diagnostic criteria for PCOS (n = 124); (3) patients younger than 15 years (n = 20); (4) patients with suspicious hypothalamic amenorrhea (n = 4); (5) patients who were in the middle of ovulation assessment (n = 125); (6) patients who did not take any medication (n = 203); (7) patients who were administered medication except for OCPs or MPA (n = 66); and (8) patients who had taken OCPs or MPA for less than 2 months (n = 11). Among 280 women with definite PCOS, 136 OCP users and 136 MPA users were randomly selected based on the sample size calculation. The estimated number of study participants was determined using G-power 3.1 [20]. We assumed an expected proportion of patients with ovulation resumption of 0.15 in the MPA group and 0.30 in the OCP group and an R squared statistic of 0.1 by covariates other than medication. Finally, 272 women were included in this retrospective cohort study to obtain 80% power to detect a difference between two groups and type I error at 0.05 (two-sided). The entire study population was Asian.

Figure 1. Flow chart of the study population.

The attending physician determined which medication to prescribe (OCPs or MPA) by considering general treatment principles such as contraindications for OCPs, age, obesity, underlying diseases, smoking, migraines with aura, and the patient’s preference. Oral MPA (Provera®, Pfizer Italia S.R.L, Ascoli Piceno, Italy) was administered at a dosage of 10 mg daily for 14 days every 4 weeks. OCPs were used according to the approved dosage and administration for each drug. Yaz® (Bayer Weimar GmbH & Co KG, Weimar, Germany), Yasmin® (Bayer Weimar GmbH & Co KG), Minivlar® (Dong-A Pharmaceutical, Seoul, South Korea), Alesse® (Pfizer Ireland Pharmaceuticals, Little Connell, Ireland), and Qlaira® (Bayer Weimar GmbH & Co KG) were prescribed in this study.

Study parameters and assay methods

Medical records were reviewed retrospectively for age, body mass index (BMI), medical history, family history of diabetes mellitus, modified Ferriman-Gallwey score, and clinical hyperandrogenism of the patients assessed at the first evaluation. A patient with acne, androgenic alopecia, or hirsutism with a modified Ferriman-Gallwey score ≥6 was defined as having clinical hyperandrogenism [21]. Transvaginal or transrectal ultrasound was also performed at the first evaluation, and polycystic ovaries were defined in patients with 12 or more follicles with a diameter of 2–9 mm or an ovarian volume of 10 ml or more [22].

Serum anti-Müllerian hormone (AMH) levels were measured using an enzyme-linked immunosorbent assay (ELISA) until December 2018 (AMH Gen II, Beckman Coulter, La Brea, CA, USA), and a chemiluminescence immunoassay (CLIA) after December 2018 (Access AMH, Beckman Coulter). The correlation coefficient between these two measures was 0.996 (AMH (CLIA) = 0.868 × AMH (ELISA) − 0.039). Total testosterone levels were measured using a radioimmunoassay (RIA) (Cisbio Bioassays, Codolet, France). Dehydroepiandrosterone sulfate (DHEAS) and 17α-hydroxyprogesterone levels were measured using an RIA (Asbach Medical Products, Obrigheim, Germany). SHBG levels were measured using an electrochemiluminescence immunoassay using a Roche Cobas e411 analyzer (Roche Diagnostics, Mannheim, Germany). The free androgen index (FAI) was calculated using the following formula of Vermeulen et al.: FAI = 100 × (total testosterone/SHBG) [23].

Assessment of ovulation

Ovulation was assessed using basal body temperature (BBT) and/or measurement of the serum progesterone level up to 3 months immediately after cessation of OCPs or MPA. A period during which the body temperature was elevated prior to the onset of menstruation was regarded as a biphasic BBT pattern. An electronic thermometer (MC-172L, Omron Healthcare Co., Kyoto, Japan) was used to ensure consistency in recording BBT. Patients were educated with a protocol, which included placing the thermometer next to their bed and, ideally, waking up every day. They were advised to measure their BBT immediately upon waking, without getting out of bed, by placing the thermometer under their tongue. BBT was then plotted on graph paper. Ovulation was defined in patients with a biphasic BBT pattern. If BBT was questionable, the serum progesterone level was estimated weekly. If the serum progesterone level was equal to or higher than 3 ng/ml, ovulation was also diagnosed.

Ethical approval

This study was approved by the Institutional Review Board (IRB) of Seoul National University Hospital (IRB Number: H-2104-027-1209) and was performed according to the principles set out in the Declaration of Helsinki. The IRB waived the need for informed consent because this was a retrospective study and analyses were conducted using anonymized data.

Statistical analysis

For continuous variables, medians and interquartile ranges were compared between OCP and MPA users using the Wilcoxon rank-sum test. For categorical variables, counts and proportions were calculated and the distribution of the variables was assessed using the chi-square test.

The crude and adjusted odds ratios (ORs) for resumption of ovulation were determined using logistic regression analysis and presented as point estimates with 95% confidence interval (CI). For multivariable logistic regression, variables were included through descriptive analysis or based on prior clinical and biological knowledge. They included demographic variables (age and BMI at baseline), and laboratory tests (AMH and 17α-hydroxyprogesterone levels). To evaluate if the estimated probability of ovulation resumption and the observed proportions of ovulation resumption agreed, goodness of fit test statistics were computed. Furthermore, to assess the effect of covariates on treatment assignment, the covariate balancing propensity score (CBPS) introduced by Imai and Ratkovic was calculated [24]. The CBPS is a generalized method of moments estimate and enhances the performance of propensity score analysis [24]. Based on the CBPS, we obtained the estimates of the average treatment effect on the treated and doubly robust analysis was conducted with inverse-probability-weighted analysis (IPW) and regression adjustment (RA) [25]. To test the balance of covariates before and after IPW-RA with the CBPS, the standardized mean differences and variance ratios of covariates were estimated. Well-balanced weighting was considered if standardized mean differences were less than 0.1 and variance ratios were less than 2.

For all statistical analyses, complete case analyses were primarily conducted. Approximately 20% of participants had missing data for variables of interest and multiple imputation by chained equations was conducted. After multiple imputation, the CBPS was not available; therefore, only IPW with RA was performed to compare the odds of ovulation resumption between the two groups. All hypotheses were tested based on a two-sided p value of .05. Stata 17.0 (College Station, TX, USA) was used for all statistical analyses. IPW-RA with CBPS was performed using KMATCH [26], a community-contributed Stata command.

Results

Patient demographics are demonstrated in Table 1. MPA users were younger than OCP users (20 vs. 23 years, p < .001). MPA users were more likely to have a higher BMI than OCP users; however, this did not reach statistical significance (21.9 vs. 21.3 kg/m², p = .15). Other clinical and laboratory parameters, including the modified Ferriman-Gallwey score, the antral follicle count, and levels of AMH, total testosterone, SHBG, DHEAS, 17α-hydroxyprogesterone, and blood glucose, were similar in the two groups. The median value of the duration of medication was comparable, but the interquartile range was significantly different between the two groups (p < .001).

Table 1. Demographic and clinical characteristics of the study population.

	Oral contraceptive (n = 136)	Medroxyprogesterone acetate (n = 136)	p
Age (years)	23 (19–26)	20 (17–23)	<0.001
Body mass index (kg/m^2)	21.3 (18.8–24.1)	21.9 (19.7–25.2)	0.15
Modified Ferriman-Gallwey score	5 (3–8)	6 (3–9)	0.45
Clinical hyperandrogenism	73 (54.7%)	88 (64.7%)	0.09
Antral follicle count	37 (29–50)	36 (28–50)	0.82
Anti-Müllerian hormone (ng/ml)	13.6 (8.6–18.1)	14.1 (8.2–18.7)	0.84
Total testosterone (ng/ml)	0.5 (0.4–0.7)	0.6 (0.4–0.9)	0.10
Sex hormone-binding globulin (nmol/L)	52.0 (33.5–82.7)	46.9 (29.9–73.9)	0.15
Free androgen index	3.5 (1.9–5.9)	3.6 (2.2–7.5)	0.14
17α-Hydroxyprogesterone (ng/ml)	1.5 (1.1–1.9)	1.4 (1.1–1.8)	0.51
Dehydroepiandrosterone sulfate (ng/ml)	2611 (1756–3580)	2501 (1715–3604)	0.66
Fasting blood sugar (mg/dl)	93 (88–95)	92 (87–95)	0.32
2-h Postprandial blood sugar (mg/dl)	104 (94–120)	111 (93–128)	0.17
Family history of diabetes mellitus	58 (42.7%)	57 (41.9%)	0.96
Duration of medication (months)	3 (3–3)	3 (3–6)	<0.001
Recovery of ovulation	33 (24.3%)	25 (18.4%)	0.24

Values are presented as median (interquartile range) or number (%).

p Values were calculated using the Wilcoxon rank-sum test or chi-square test.

After cessation of medication, 25 women (18.4%) in the MPA group experienced resumption of ovulation and this percentage was not significantly different from that in the OCP group (33 women, 24.3%, p = .24). The OR for resumption of ovulation was slightly lower in MPA users than in OCP users (OR 0.70, 95% CI, 0.39–1.26, p = .24), and the difference remained insignificant after multivariable logistic regression (adjusted OR (aOR) 0.98, 95% CI 0.49–1.98, p = .96) (Table 2). Absolute values of standardized mean differences of age, and BMI were higher than 0.1 (Table 3), and the cumulative density function of probability differed between the two groups (Figure 2), which means the balance of covariates significantly differed between the two groups.

Figure 2. Cumulative density function of probability of treatment assignment. ATT: average treatment effect on the treated.

Table 2. Odds ratio for resumption of ovulation after cessation of medication.

	Univariable logistic regression			Multivariable logistic regression			IPW-CBPS and RA
	OR	95% CI	p Value	Adjusted OR	95% CI	p Value	Adjusted OR	95% CI	p Value
Oral contraceptive	Ref			Ref			Ref
Medroxyprogesterone acetate	0.70	0.39–1.26	0.24	0.98	0.49–1.98	0.96	1.00	0.89–1.12	0.94
p Value by the goodness of fit test						0.19

IPW-CBPS: inverse probability weighting with weights estimated from the covariate balance propensity score; RA: regression adjustment; OR: odds ratio; CI: confidence interval.

The covariate balancing propensity score was based on age, body mass index, and anti-Müllerian hormone and 17α-hydroxyprogesterone levels.

Regression adjustment for age, body mass index, anti-Müllerian hormone and 17α-hydroxyprogesterone levels, and duration of medication.

Table 3. Summary of the balance of variables before and after IPW-CBPS.

	Standardized mean difference		Variance ratio
	Before IPW-CBPS	After IPW-CBPS	Before IPW-CBPS	After IPW-CBPS
Age	−0.47	−1.2 × 10^-8	0.82	1.07
Body mass index	0.18	−7.0 × 10^-8	1.43	1.20
Anti-Müllerian hormone	0.11	−1.3 × 10^-7	2.13	1.59
17α-Hydroxyprogesterone	−0.06	1.4 × 10^-7	0.86	1.05

IPW-CBPS: inverse probability weighting with weights estimated from the covariate balance propensity score.

Therefore, we performed adjustment with IPW-RA with the CBPS based on age, BMI, and the levels of AMH and 17α-hydroxyprogesterone. The OR of ovulation resumption did not significantly differ according to the medication (aOR 1.00, 95% CI 0.89–1.12, p = .94) (Table 2). After IPW-RA with the CBPS, the covariates were well balanced between the two groups (all standardized mean differences <0.1 and variance ratios <2) (Table 3) and the cumulative density function of probability showed a similar pattern in the two groups (Figure 2).

Because 20% of participants had at least one missing value of the variables (Supplementary Table 1), multiple imputation was used with IPW-RA. However, the OR of ovulation resumption did not change substantially (aOR 0.99, 95% CI 0.89–1.10, p = .83) (Supplementary Table 2).

Discussion

In this study, we compared the resumption of ovulation in women with PCOS after cessation of OCPs and MPA. The odds of ovulation resumption were similar after cessation of OCPs and cyclic MPA. This is the first study to objectively evaluate resumption of ovulation after cessation of OCPs and MPA, which are widely used to manage menstrual disturbances in women with PCOS.

MPA is a derivative of progesterone that has an α-hydroxy group and a methyl group at positions 17 and 6, respectively. It is widely used for contraception, postmenopausal hormone therapy, and management of amenorrhea [18], abnormal uterine bleeding [27], and endometrial hyperplasia and cancer [28]. Although OCPs are commonly prescribed for women with PCOS, cyclic MPA can be offered to women who cannot use OCPs. For example, OCPs are not recommended for women aged ≥35 years who smoke ≥15 cigarettes/day and women who have migraines with aura [29]. In Korea, OCPs are approved for women aged 14 years and older. In Asian countries, women are reluctant to use OCPs due to misconceptions and a fear of adverse events [30]. The proportion of ever-users of OCPs in Korea is less than 15% [31]. Cyclic MPA may be a good alternative to OCPs in women for whom OCPs are contraindicated, or who decline to take OCPs.

In this study, the odds of ovulation resumption after cessation of MPA were similar to those after cessation of OCPs. The efficacy of MPA can be explained as follows. Administration of MPA decreases the luteinizing hormone (LH) level [32–34], induces a greater reduction of LH than that of follicle-stimulating hormone, and reduces pituitary sensitivity to gonadotropin-releasing hormone [34–36]. By contrast, OCPs inhibit production of ovarian androgen and promote the synthesis of SHBG in the liver [37–39]. Therefore, MPA may be as effective for treatment of hyperandrogenism as OCPs. Indeed, several studies demonstrated that serum testosterone levels significantly decrease after MPA treatment and this effect is not mediated by changes of the SHBG level [18,40,41]. Moreover, oral MPA treatment improves insulin sensitivity and decreases LH and total testosterone levels [42]. Considering that hyperinsulinemia is associated with production of androgen in the ovary [43,44], ovulation may resume after cessation of MPA due to the improvement of insulin sensitivity upon MPA treatment. Taken together, these findings demonstrate that ovulation can resume after cessation of MPA in women with PCOS.

In the present study, ovulation resumed in 18.4 and 24.3% of women with PCOS after cessation of MPA and OCPs, respectively. These are lower than the values reported in previous studies, which had different study designs or whose study populations had different characteristics. For example, ovulation was reported in 45.5% (15/33) of overweight women with PCOS (mean BMI 32.1 kg/m²) after body weight reduction of 5–10% [45], whereas our study population had a relatively normal BMI (median BMI 21.5 kg/m²). Sharpe et al. reported that ovulation was observed in 37.1–54.5% of women with PCOS during metformin treatment [46], whereas resumption of ovulation was observed after cessation of medication in our study. Therefore, the proportions of patients with ovulation resumption cannot be directly compared between studies.

Several limitations of the present study need to be addressed. First, this was a retrospective observational study. OCPs or MPA were prescribed based on the preference of clinicians. In addition, the baseline characteristics of the patients might have affected clinicians’ decisions. For example, MPA users were more likely to be younger than OCP users and they might have visited clinicians with their parents who did not want their daughters to use OCPs. To overcome this shortcoming, we used IPW-RA to adjust for confounders and the CBPS to enhance the performance of propensity score weighting [24]. Indeed, the balance between the treatment groups in our study was improved after IPW-RA with the CBPS (Table 3 and Figure 2). Second, the present study was performed at a single center and the generalizability of the results may be limited. However, this enabled us to thoroughly review the medical records of the study participants and minimize measurement errors. Third, various formulations of OCPs were prescribed in the present study. Most OCPs used in this study contain ethinyl estradiol, a synthetic estrogen with improved bioavailability [47], while one regimen contains estradiol valerate (EV), a natural estrogen. Specific types of OCPs are not recommended in women with PCOS [10] and treatment with OCPs containing EV may improve acne and increase the level of SHBG [48].

There are several strengths of this study. First, this is the first study to investigate the resumption of ovulation after cessation of cyclic MPA in women with PCOS. MPA is widely used to control menstrual cycles in women with PCOS, but resumption of ovulation after cessation of cyclic MPA has not been investigated. Second, we confirmed ovulation in a reproducible and clear manner. Resumption of the menstrual cycle cannot be used as a proxy for resumption of ovulation [49]. Therefore, our institution performed an objective assessment of ovulation, and two experienced clinicians (H.K. and S.J.H) reviewed all BBT charts and serum progesterone measurements. Finally, missing data are common in observational studies. In most studies, listwise deletion, which means discarding an observation with at least one missing variable, is performed, but this may lead to bias. Therefore, we additionally performed analysis with repeated simulated datasets by multiple imputation, which can attenuate biased results.

Conclusion

Our results demonstrated that women with PCOS taking MPA and OCPs have a similar chance of ovulation resumption after cessation of medication. Therefore, in patients who are unable to take OCPs because of contraindications or concerns about adverse events, cyclic MPA can effectively induce regular menstrual cycles and ovulation can resume after cessation of MPA, similar to OCPs.

Author contributions

H.K. contributed to the conception and design. H.K. and S.J.H. collected data. H.K. performed the statistical analyses. H.K. and S.J.H. drafted the manuscript. S.Y.K. and C.S.S contributed to the interpretation of results and editing of the manuscript. All authors read and approved the final manuscript.

Disclosure statement

H.K. has received honoraria for participation on the advisory board of Bayer, and lectures for Roche Diagnostics and Organon, which are unrelated to the subjects addressed in this paper. No potential conflict of interest was reported by the author(s).

Data availability statement

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Additional information

Funding

This study was funded by the SNUH Research Fund [grant no. 04-2015-0910]. The funding source had no role in the study design, collection, analysis, or interpretation of the data, writing the manuscript, or decision to submit the paper for publication.