Study design, ethics approval and participants
This study is a crossover randomized controlled 6-arm trial designed based on the requirements of the Consolidated Standards of Reporting Trials (CONSORT)  and initiated in February 2016. This clinical trial is registered in the Iran Registry of Clinical Trials (Registration number: IRCT201702071281N2). Ethics approval was obtained from the ethics committee of Shahid Beheshti University of Medical Sciences, Tehran, Iran. (Ethics Code: IR.SBMU.ENDOCRINE.REC.1396.425). We obtained written informed consent from all eligible participants, following study content being clearly explained to the subjects by the trial assistant; all research tools, including questionnaires, were completely anonymous.
Patients with PCOS diagnosed based on the Androgen Excess Society (AES) criteria (age 18–45 years) were recruited at the endocrine out-patients clinic of the Research Institute for Endocrine Sciences (RIES) of the Shahid Beheshti University of Medical Sciences, Tehran, Iran. According to the 2006 criteria of the AES, PCOS was diagnosed for patients with oligo-anovulation and/or polycystic ovaries (PCO) and clinical and/or biochemical signs of hyperandrogenism (HA) . We ruled out all secondary etiologies, including hyperprolactinemia, thyroid dysfunction, Cushing’s syndrome, congenital adrenal hyperplasia, and virilizing tumors, in all patients using appropriate tests. Pregnant women or those of willingness for pregnancy, patients with contraindications of OC therapy, and those using medications related to PCOS such as hormonal, insulin sensitizers, or antiandrogens drugs within the previous 3 months were excluded from the study.
All participants were alternatively treated for two 6-month treatment periods with OCs. This study had 6 treatment arms with different sequences. All participants were randomly assigned to one of the following treatment groups: Group 1: First treated with Ethinyl estradiol (EE) 30 μg + LNG 0.15 mg and then received the second treatment with OCs containing EE 30 μg + DSG 150 μg. Group 2: First treated with EE 30 μg + LNG 0.15 mg and then received the second treatment with OCs containing EE 35 μg + CPA 2 mg. Group 3: First treated with OCs containing (EE) 30 μg + LNG 0.15 mg and then received the second treatment with OCs containing EE 30 μg + DRSP 3 mg. Group 4: First treated with EE 30 μg + DSG 150 μg and then received the second treatment as EE 30 μg + LNG 0.15 mg. Group 5: First treated with EE 35 μg + CPA 2 mg and then received the second treatment with EE 30 μg + LNG 0.15 mg. Group 6: First treated with EE 30 μg + DRSP 3 mg and then received the second treatment with EE 30 μg + LNG 0.15 mg.
Accordingly, all participants received EE 30 μg + LNG 0.15 mg as the standard treatment. A washout period of 6–8 weeks was considered between the two treatment periods for the elimination of carry-over effects of treatment. Interventions were performed by a trained midwife with the assistance of another individual, who was aware of the type of intervention.
Randomization and blinding
A blocking or stratification random allocation (block size = 6), using a computer-based random number generator was applied to assign participants to treatment groups. The randomization sequence was prepared before the trial, initiated by an independent statistician. For those patients meeting inclusion criteria and having given written informed consent, the next randomization sequence was assigned by the research assistant according to the schedule. Both the clinical examiner and data analyst were blinded to participants during the trial.
For each participant, outcome measures, including clinical, biochemical (androgenic and metabolic), and QOL were collected at 6 time points, i.e. first baseline, at the end of third and sixth months of the first treatment period, after the washout period (second baseline), and at the end of the third and sixth months of the second treatment period, respectively.
Clinical assessments of participants were conducted by only one person who was blinded to groups to minimize any assessor effects. We evaluated all patients for regularity of menstrual cycles; oligomenorrhea was defined as vaginal bleeding episodes at intervals ≥35 days. Menstrual cycle intervals less than 22 days were defined as polymenorrhea. Patients who had no menstrual bleeding for 6 months or longer were considered as amenorrhea [18,19,20]. Clinical hyperandrogenism was defined by the presence of hirsutism, acne, or androgenic alopecia. The standardized scoring system of m-FG score was used for determining the density of terminal hair at 9 different body sites, i.e., upper lip, chin, chest, upper back, lower back, upper abdomen, lower abdomen, arm, and thigh; a total score ≥ 8 was considered as hirsutism [21, 22]. Acne was diagnosed based on the grading system on the basis of the number of lesions and their spread on the face, back, and chest, and was classified to mild, moderate, moderate to severe, and severe . The Ludwig classification system was used to diagnose androgenic alopecia in patients .
Biochemical measurements were performed by an expert laboratory technician under the supervision of a specialist in laboratory sciences. At baseline and follow-ups, fasting (at least 9 h) blood samples were collected between days 3 and 5 of the spontaneous menstrual cycle or progesterone-induced menstrual bleeding. All sera were stored at − 80 °C until the time of testing. Follicle stimulating hormonetotal (FSH) and luteinizing Hormone (LH) were measured by Immunoradiometric assay (IRMA). Androgenic profiels, including testosterone (TT), and DHEAS, were measured by the enzyme immunoassay (EIA), (DRG Diagnostics, GmbH, Germany); SHBG was measured by immune enzymometric assay (IEMA), (Diagnostic biochem Canada Co. Ontario, Canada). FAI was calculated using the formula [TT (nmol/L) × 100/SHBG (nmol/L)] . Metabolic parameters including fasting blood sugar (FBS), TG, TC, LDL cholesterol and HDL cholesterol were measured by colorimetric enzymatic assay (Pars Azmun Co. Tehran, Iran). Insulin was measured by ECLIA (Roche Diagnostics GmbH, Mannheim, Germany). HOMA-IR was calculated by the following formula: [glucose (nmol/L) × insulin (μU/mL)/22.5]. We considered IR as HOMA-IR ≥ 2.63.
We perfomed sonography only at the beginning of the study; it was performed on the same day as the blood samples were collected. Endometrial thickness, ovarian volume, number, diameter, and distribution of the follicles were recorded. The ovaries were considered polycystic when observed as having an ovarian size more than 10 ml) and/or at least 12 follicular cysts measuring 2–9 mm .
According to the World Health Organization (WHO) report, QOL is defined as “the individual’s perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals” . In this study, we assessed QOL using a specific and valid health-related quality of life questionnaire for PCOS, developed by Iran, which is consisted of 43 items in six domains, including psychosocial–emotional, self-image, fertility, sexual function, hirsutism, and obesity– menstrual disorders. Items were scored based on the 5-point Likert scale (always, often, sometimes, rarely, never) .
In this study, we applied the non-inferiority assumption and intention-to-treat the principle to compare the intervention treatments. EE 30 μg + LNG 0.15 mg was considered as a reference group.
Continuous and categorical variables at the baseline are reported as median (IQR) and percentage, respectively. Effect of oral contraceptives containing EE + DSG, EE + CPA, and EE + DRSP was compared to EE + LNG for the outcomes of interests using the generalized estimating equations (GEE) which are appropriate for longitudinal data. Regression models were adjusted for period, sequence and baseline measure. Confidence intervals (CIs) for effect sizes were estimated through the robust variance estimators obtained on the assumption that the correlation structure was exchangeable. Carryover effect for this 2 × 2 (two sequence-two period) cross over analysis were tested via the first-order-interaction effect of period and treatment, and in case of significant results, the independent working correlation structure was considered to omit intra subject variability [29,30,31]. Statistical analysis was performed using STATA software (version 13; STATA, INC., College Station, TX, USA). A significant level was considered < 0.05.
To show that the OCs containing LNG are clinically as effective as those containing antiandrogenic progestins including DSG, CPA, and DRSP, CPA, we used a noninferiority hypothesis, where μT is the mean of the test drug, μS is the mean of standard therapy, and δ is a difference of clinical importance.
By rejecting the null hypothesis, we conclude that the difference between the test drug and the standard therapy is less than a clinically meaningful difference (ie, δ), and therefore the test drug is as effective as the standard therapy .
We considered 80% power, 0.05 type I error, and ϴ = 0.52, where ϴ is defined as the difference of mean of test drug and standard therapy minus difference of clinical importance divided by standard division. Sample size was calculated from the table introduced by sample size calculations in clinical research . We estimated 25 samples were needed for each group, with 150 total samples needed.