We conducted a randomized controlled trial (RCT) study on PCOS patients with vitamin D deficiency or insufficiency at the First Affiliated Hospital of Xi’an Jiaotong University from January 2021 to June 2023. All participants gave written informed consent on the basis of procedures granted by the Ethics Committee of The First Affiliated Hospital of Medical College of Xi’an Jiaotong University (XJTU1AF2021LSK-207). This study has been registered on China’s clinical trials registration: www.chictr.org.cn (ChiCTR2100048736).

Sixty-eight PCOS women were enrolled in this study. All patients were aged 21–34 years. The diagnostic criteria for PCOS we used were the Rotterdam criteria: chronic ovulatory dysfunction, clinical manifestations or biochemical evidence of hyperandrogenism, and polycystic ovarian morphology (PCOM) [20]. PCOS could be diagnosed when any two of these three criteria were presented. Hyperandrogenism was identified as either clinical manifestations or laboratory evidence. Laboratory evidence was defined as an abnormally increased testosterone level. Clinical manifestations included hirsutism or acne. Hirsutism was defined as a modified Ferriman-Gallwey score of more than 3 at the time of physical examination. PCOM was identified as the presence of 12 or more follicles in unilateral ovary or bilateral ovaries measuring 2–9 mm in diameter, and/or ovarian volume ≥ 10 ml. Ovarian volume = 0.5 × length diameter × transverse diameter × anteroposterior diameter. Participants who were diagnosed with tumor secreting androgens, thyroid dysfunction, hyperprolactinemia, hypothalamic amenorrhea, Cushing’s syndrome, congenital adrenal hyperplasia (CAH), premature ovarian insufficiency (POI), premature ovarian failure (POF), hypercalcemia, malabsorption disorders and diabetes mellitus were excluded. In addition, participants who used corticosteroids, hypolipidemic agents, calcium supplements, or any other drugs known to affect vitamin D metabolism were also excluded from the study.

The baseline data of PCOS women were collected by questionnaire in a face-to-face interview, such as age, body mass index (BMI), waist-to-hip ratio (WHR), blood pressure, marital status, gravidity, parity, employment status, family history and outdoor exercise. BMI ≥ 30.0 kg/m2 was defined as obesity [21].

Sixty-eight PCOS women were enrolled initially in this RCT study, 8 women with normal serum vitamin D level were excluded. Then, sixty PCOS women with vitamin D deficiency or insufficiency were randomly divided into vitamin D group or control group, and 30 participants in each group. Two participants in the vitamin D group and 1 participant in the control group were lost to follow-up. Finally, 57 participants (vitamin D group, n = 28. control group, n = 29) completed the 12 weeks study period (Fig. 1).

Flow chart of study participation

25-hydroxyvitamin D3 [25(OH)D] is the primary circulating form of vitamin D, which is the most abundant vitamin D metabolite and considered as the best parameter of vitamin D status within the human body. Therefore, serum 25(OH)D was detected by chemiluminescence method. According to the Institute of Medicine (IOM) and World Health Organization (WHO), vitamin D deficiency was defined as serum 25 (OH) D concentration lower than 20 ng/mL, and vitamin D insufficiency was defined as a concentration of 20–<30 ng/mL. Serum 25(OH)D concentration of 30–50 ng/mL were regarded as normal [22].

All biochemical indicators in this study were detected in the clinical laboratory of our hospital. The basic sex hormone concentrations and anti-Müllerian hormone (AMH) concentration were tested during 2–4 days of natural menstrual cycle or progesterone withdrawal bleeding using chemiluminescence method. Sex hormone included follicle-stimulating hormone (FSH), luteinizing hormone (LH), prolactin (PRL), estradiol (E2), progesterone (P) and testosterone (T). All participants were given 83 g oral glucose for oral glucose tolerance test (OGTT) and insulin release test. The blood glucose was detected by hexokinase method. Serum insulin level was detected through radioimmunoassay. HOMA-IR was calculated to evaluate IR. HOMA-IR = [fasting glucose (mmol/L) × fasting insulin (mIU/L)]/22.5. A HOMA-IR > 3.0 was defined as IR, and an insulin level after oral glucose powder intake 10 times higher than the fasting level was also defined as IR [23]. At the same time, the blood lipids were detected using automatic biochemical analyzer, including TC, TG, LDL-C and high-density lipoprotein cholesterol (HDL-C).

Sample size was calculated provided that serum 25(OH)D concentration and glucose metabolism and lipid metabolism indicators were the primary outcomes. According to the method reported in previous literatures, sample size was calculated with the following parameters: probability of type one error (α) of 0.05 and type two error (β) of 0.20 (power = 80%), difference between two means to be detected was 0.52, expected background standard deviation was 1. Based on this, we needed 27 participants in each group. Considering a loss to follow-up of 3 participants per group, we needed to have 30 participants in each group.

All participants were randomly divided into vitamin D group or control group by means of computer-generated random numbers. Participants in vitamin D group received basic treatment combined with vitamin D supplementation (2000 IU/day). Basic treatment included a proper diet and aerobic exercise outdoors at least three times per week. Each aerobic exercise lasted at least 30 min. Participants in control group were only given basic treatment.

The primary outcomes were the serum 25(OH)D concentration, glucose metabolism, insulin concentration and lipid metabolism indicators. The secondary outcomes were general metabolic parameters, including BMI, WHR and blood pressure.

Statistical analyses were performed using SPSS version 20.0 (IBM, Armonk, NY, USA). The Kolmogorov–Smirnov test was used to check the normal distribution prior to statistical tests. For normally distributed variables, the continuous variables were represented as mean ± standard deviation and were analyzed by Student’s t test, whereas the Mann–Whitney U test was used to analyze non-normally distributed data. Differences in dichotomous outcomes were given as number and percentage (%), which were compared by chi-square test. P < 0.05 was considered statistically significant.