Association Between Polycystic Ovary Syndrome and the Vaginal Microbiome

A Case-Control Study

Xiang Hong; Pengfei Qin; Kaiping Huang; Xiaoling Ding; Jun Ma; Yan Xuan; Xiaoyue Zhu; Danhong Peng; Bei Wang


Clin Endocrinol. 2020;93(1):52-60. 

In This Article

Subjects and Methods

The Study Population

This case-control study was designed using Kelly's method,[19] which was used to estimate sample size; therefore, at least 20 samples were required in each study group. From March to May 2019, at the reproductive clinic of Zhongda Hospital (Nanjing, China), women with newly diagnosed PCOS were consented and included in the case group. PCOS diagnosis followed the Rotterdam criteria[20] and referred to the current Chinese Guidelines for the Diagnosis of PCOS,[21] meeting at least two out of three symptoms: oligo- and/or anovulation (fewer than eight cycles per year or more than 3 months without menstruation); clinical and/or biochemical signs of hyperandrogenism (hirsutism with a modified Ferriman-Gallwey score of more than 6 or total testosterone level more than 1.77 nmol/L); and polycystic ovaries (there were 12 or more follicles in each ovary measuring 2–9 mm in diameter and/or increased ovarian volume (>10 mL) by ultrasound examination). The testosterone level was detected by the immunoassay (DXI800 Immune Analyzer of Beckman Coulter Inc,). During this period, women who participated in the prepregnancy health examination programme in the Maternal and Child Health Center of Gulou District were consented and screened as the healthy control group. The inclusion criteria for this group were as follows: (a) women aged between 20–45 years old; (b) no history of PCOS, infertility and other endocrine diseases and (c) B-ultrasound examination revealing no polycystic ovary images. To avoid interference with the vaginal microbiome, exclusion criteria for both groups were as follows: (a) women who took antibiotics in the previous three days; (b) women who were sexually active or had vaginal irrigation in the previous 24 hours; (c) women who were menstruating and (d) women diagnosed with Neisseria gonorrhoeae, Chlamydia or Trichomonad. All participants signed an informed consent sheet, and the Ethics Committee of Zhongda Hospital approved this study (2018ZDSYLL072-P01). In total, 39 PCOS women and 40 healthy control women were included.

Basic Information

Basic demographic characteristics were collected using basic questionnaires, including age, educational level, pregnancy history and history of vaginitis. Body mass index (BMI) was calculated based on the following formula: BMI = weight/height2. Then, BMI metrics were categorized into four groups based on Chinese standard: underweight (<18.5 kg/m2), normal weight (18.5–23.9 kg/m2), overweight (24.0–27.9 kg/m2) and obese (≥28.0 kg/m2).[22]

Vaginal Swab Collection

Gynaecologists collected two vaginal swabs from each woman using standard operating procedures. The women were placed in a lithotomy position. The gynaecologist used a sterile swab to scrape secretions at the posterior fornix, assisted by a speculum. Swabs were rotated three times to uniformly scrape any discharge. One swab was used for smear microscopy examination, and vaginal cleanliness grading (grades; I–IV) was assessed according to Chinese general standard.[23,24] These standards are outlined (Table S1). A second swab was stored in a dry tube with a unique identification number and immediately placed in a 4°C collection box. Within 8 hours, swabs were transferred to a −80°C refrigerator and stored for nucleic acid extraction.

DNA Extraction and 16S rRNA Gene Sequencing

The swab was placed in a 2 mL sterile centrifuge tube, to which 1 mL sterile PBS buffer was added to submerge the swab. After 10 min shaking at room temperature, PBS eluates were centrifuged for 2 min (12 000rpm, Centrifuge 5424 R, Eppendorf Co., Ltd). The sediment was stored, and DNA extractions were performed using a TIANamp Bacteria DNA Kit (Tiangen Biochemical Technology), following manufacturer's instructions. Purified nucleic acids were eluted in TE buffer.

Universal primers, 338F (5'-ACTCCTACGGGAGGCAGCA-3') and 806R (5'-GGACTACHVGGGTWTCTAAT-3'), were used to amplify the V3-V4 region of the 16S rRNA gene, using polymerase chain reaction (PCR). PCR reactions contained 13.25 μL H2O, 2.0 μL 10 × PCR Ex Taq Buffer, 0.5 μL DNA template (100 ng/mL), 1.0 μL primer 1 (10 mmol/L), 1.0 μL primer 2 (10 mmol/L), 2.0 μL dNTP, 0.25 μL Ex Taq (5 U/mL) in a final volume of 20 μL. Standard PCR parameters were performed as follows: initial denaturation at 95°C for 5 minutes, then 30 amplification cycles (30 seconds at 95°C, 20 seconds at 58°C and 6 seconds at 72°C), followed by a final extension at 72°C for 7 minutes. Amplified products were purified and observed on a 1.0% agarose gel. Finally, a library was constructed and PCR products were sequenced on an Illumina HiSeq 2500 platform (Beijing Biomarker Technologies Co. Ltd.).

Sequence Processing

To obtain raw tags, paired-end reads were merged by FLASH (v1.2.7, Some tags of low quality (with more than six mismatches when compared to the primers, or with an average quality score < 20 in a 50 bp sliding window, or shorter than 350 bp) were removed using Trimmomatic (v0.33). We identified possible chimaeras by employing UCHIME, a tool included in mothur software ( Denoised sequences were clustered using USEARCH (version 10.0), and tags with similarity of> = 97% were regarded as an operational taxonomic unit (OTU). For each representative sequence, the silva database (Release128, was used to annotate taxonomic information by QIIME. OTU relative abundance information was normalized using a standard of sequence numbers corresponding to the sample with the least sequence. To identify specific Lactobacillus species, representative sequences of the Lactobacillus genus were queried against NCBI's 16S rRNA gene database using BLAST. OTU abundance was normalized using a standard of sequence numbers corresponding to the sample with the least sequence. All bioinformatics analyses were completed on the Biomarker BioCloud platform (

Statistical Analysis

To determine comparability between the two groups and address potential bias sources, we compared differences in demographic distribution between groups. T tests (normally distributed) or Mann-Whitney tests were used to compare continuous variable differences between groups. Categorical variables were presented as frequencies (percentages) and compared between groups using the chi-square or Fisher's exact test. Ranked data, such as vagina cleanliness levels, were tested by the non-parametric Mann-Whitney method for between-group variance. Linear covariance adjustment was used to adjust age, BMI and vaginal cleanliness grading between groups for some continuous variables comparisons between groups. These analyses were performed on SAS software (version 9.2; SAS Institute, Inc).

For the OTU data, alpha diversity indices, including Simpson, Shannon and Chao1 indices, were assessed by mothur software. It is generally recognized that if the Simpson index is lower, or Shannon and Chao1 indices are higher, the vaginal microbiome will be more diverse and rich.[18] β diversity focuses on the difference in taxonomic abundance profiles from different samples, which can be measured by different distances, including binary Jaccard, unweighted UniFrac and weighted UniFrac distances.[25] The β diversity comparisons between groups were generated by principal coordinate analyses (PCoAs) and permutational multivariate analysis of variance (PERMANOVA). Linear discriminant analysis (LDA) effect size (LEfSe) was calculated to determine biomarkers between groups. The relative abundance of specific species was logarithmically transformed and displayed in scatter diagrams; differences between groups were tested by the Mann-Whitney method. The community structure type (CST) of the vaginal microbiome was defined using Jensen-Shannon divergence clustering.[26] We then named these classifications based on the dominant bacteria: CST I was dominated by L. crispatus, CST II was dominated by L. gasseri, CST III was dominated by L. iners, CST IV was highly diverse, and CST V was dominated by L. jensenii.[27] Receiver operating characteristic (ROC) curves were used to assess the diagnostic value of specific bacteria on PCOS, using Stata (version 13.0). To determine whether changes in the vaginal microbiome in PCOS could be confounded by vaginal cleanliness grading, BMI and androgen levels of PCOS women, we performed four sensitivity analyses. A two-sided P value ≤ .05 was deemed statistically significant.