A total of 684 individuals, including 386 PCOS patients and 298 non-PCOS control women (some of them with one child or more) with normal menstrual cycles (< 32 days) and without obesity, hirsutism, cystic acne, overmuch sebum, and insulin resistance, were studied. All the participants recruited for our study were of Chinese Han origin, a predominant Chinese ethnic population. The study was approved by Medical School of Nanjing University, and informed consent was obtained from each study participant.
PCOS diagnostic criteria and hormone measurements
Patients with PCOS were diagnosed by the 2003 Rotterdam Criteria  (The Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group, 2004). The Rotterdam Criteria requires at least two of the following indicators for diagnosis of PCOS: clinical or biochemical signs of hyperandrogenism, oligomenorrhea or amenorrhea, and presence of polycystic ovarian (PCO) morphology on ultrasound, with the exclusion of other causes of hyperandrogenism such as hyperprolactinemia, androgen-secreting tumors, Cushing's syndrome and nonclassic congenital adrenal hyperplasia.
We obtained the participants' age at menarche (AAM) through inquiry and calculated the body mass index (BMI = body weight in kilograms divided by square of height in meters) to assess obesity. Peripheral blood was obtained by a single venipuncture during the 3rd to the 5th day of the menstrual cycle for those who had menstruation and at any time for those who had amenorrhea. All peripheral blood samples were obtained between 8 AM and 9 AM after a 12-hour overnight fast. None of the study participants had been taking hormonal medications, e.g. contraceptive pills, for the previous three months before the hormone measurement. Blood samples were immediately centrifuged and then serum was separated and frozen at -80°C until assayed. Levels of total testosterone (T), follicle-stimulating hormone (FSH), luteinizing hormone (LH), and estradiol (E2) in the sera were measured by RIA (Beijing North Institute of Biological Technology of China and the CIS Company of France). The E2/T ratio was used as an index of the aromatase activity. All assays had intra- and inter-assay coefficients of variation less than 10%.
Polymorphism genotyping analysis
Genomic DNA was isolated from human leukocytes by using Chelex®-100 as a medium (Promega, Madison, WI, USA). Genotyping of the rs2414096 polymorphism of the CYP19 gene was performed with the polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) method. The sequences of the primers were 5'-TCT GGA AAC TTT TGG TTT GAG TG-3' (forward primer) and 5'-GAT TTA GCT TAA GAG CCT TTT CTT ACA-3' (reverse primer). PCR amplification was carried out in a total volume of 25 μL containing 50 ng of genomic DNA, 6.25 pmol of each primer, 2.5 μL STR (short tandem repeat), 10×buffer (STR 10×buffer, Promega, Madison, WI, USA) and 1.5 U of Taq DNA polymerase (Promega, Madison, WI, USA). The PCR was performed in a PTC-100 (MJ Research™, Incorporated) thermocycler as follows: 30 cycles consisting of 1 minute of denaturation at 94°C, 1 minute of annealing at 60°C, and 1 minute of extension at 72°C. An initial denaturation step of 5 minutes at 94°C and a final extension of 10 minutes at 72°C were used. The PCR products of 189-bp were then digested with HSP92 II at 37°C overnight. A single 189 bp band corresponds to the wild-type G homozygote; bands of 189, 161, and 28 bp stand for the AG heterozygotes; and 161 and 28 bp for the A homozygote. The DNA fragments were separated by electrophoresis on a 2% agarose gel, and visualised by staining with ethidium bromide.
Fisher's Exact Test was used to compare the CYP19 gene genotype distributions in the case-control study. The analysis was performed using the SAS system software (SAS Institute Inc., Cary, NC 27513-2414 USA). The results of serum hormone levels are reported as means ± SD. Clinical variables such as age and BMI, and AAM were compared using one-way analysis of variance (ANOVA). Differences in serum hormone levels among different genotypic individuals were assessed using analysis of covariance (ANCOVA) to correct for age and BMI. P < 0.05 was considered statistically significant. Hardy-Weinberg distribution of genotypes in the PCOS and control groups was assessed.