Correlation Between Glucose Metabolism and Serum Steroid Hormones in Patients With Polycystic Ovary Syndrome

Xuelin Li; Tianyue Zhang; Shengxian Li; Yuying Deng; Lihua Wang; Tao Tao; Shujie Wang; Yanyun Gu; Weiqiong Gu; Jie Hong; Wei Liu; Weiqing Wang; Yifei Zhang


Clin Endocrinol. 2020;92(4):350-357. 

In This Article

Materials and Methods


A total of 1044 women with PCOS were selected from patients who visited Ruijin Hospital and Renji Hospital in Shanghai from January 2006 to October 2016. We further excluded 130 patients who lacked either fasting or 2-hour postload plasma glucose measurements, and ultimately, we included 914 PCOS patients with complete glucose metabolism data (including both fasting and 2-hour postload plasma glucose information) in this analysis. PCOS was diagnosed according to the 2003 Rotterdam criteria,[15] with patients exhibiting two or more conditions (clinical/biochemical hyperandrogenism, oligo-ovulation and polycystic ovaries), while excluding other diseases that may cause hyperandrogenism and ovulation disorders. The exclusion criteria included pregnancy, other related endocrine diseases, previous ovarian surgery and medication history that may affect steroid hormones within the previous 3 months. The detailed inclusion and exclusion criteria can be found in our previous study.[16]

The study protocol was approved by the Ethics Committee of Ruijin Hospital and Renji Hospital, which are both affiliated with Shanghai Jiao Tong University School of Medicine. All subjects had a thorough understanding of the research procedures and signed an informed consent form.


Baseline data were collected, including medical history, menstrual condition and information from a physical examination (such as weight, height, waist circumference, hip circumference, blood pressure, acne, acanthosis, galactorrhoea and hirsutism). All patients received a 75-g fasting oral glucose tolerance test, and fasting plasma glucose (FPG) and 2-hour postload plasma glucose (2hPG) were detected for the diagnosis of glucose metabolism status. Clinical sex hormone testing and gynaecological ultrasound were also performed.

The waist-to-hip ratio (WHR) was calculated as waist circumference (cm)/hip circumference (cm). BMI was determined by weight (kg)/height (m)2. The free androgen index (FAI) was calculated using the following formula: total testosterone (measured by LC-MS/MS) (nmol/L) × 100/sex hormone-binding globulin (SHBG) (nmol/L). The insulin resistance index (HOMA-IR) was calculated using the following formula: FPG (mmol/L) × fasting serum insulin (uIU/mL)/22.5.

According to the 1999 WHO criteria,[17] prediabetes or impaired glucose regulation is defined as either impaired fasting glucose (IFG), impaired glucose tolerance (IGT) or both. Abnormal glucose metabolism status includes prediabetes and T2DM. The levels for the above are as follows: IFG: 6.1 mmol/L ≤ FPG < 7.0 mmol/L and 2hPG < 7.8 mmol/L, IGT: 7.8 mmol/L ≤ 2hPG < 11.1 mmol/L and FPG < 7.0 mmol/L, T2DM: FPG ≥ 7.0 mmol/L or 2hPG ≥ 11.1 mmol/L.

In addition, serum samples from each participant were collected to measure steroid hormones by using LC-MS/MS. All patients maintained a normal diet and the same lifestyle for 3 days before blood drawing. Venous blood was collected by vacuum tubes from the elbow vein of each patient while in a seated position after 10–12 hours of overnight fasting. The serum was separated and stored at −80°C for further analysis. Through a series of sample preparations as previously described,[16] a total of 14 steroid hormones were extracted and measured, including 17-hydroxyprogesterone, 17-hydroxypregnenolone, progesterone, pregnenolone, cortisol, cortisone, 11-deoxycortisol, 11-deoxycorticorsterone, aldosterone, corticosterone, oestrone, androstenedione, DHEAS and testosterone. The details of LC/MS-MS assay and clinical biochemistry methods are shown in the supplementary materials.

Statistical Method

Data were analysed by SPSS version 22.0. Continuous variables with a normal distribution are reported as the mean ± standard deviation (SD), whereas those with a nonnormal distribution are reported as the median and interquartile range (IQR). Nonnormal variables were log-transformed when analysing differences between the two groups of patients with different glucose metabolism statuses. We applied Student's t test to compare the two groups with age and BMI as covariates. Correlation was assessed by Spearman's test, and the results are reported as a heat map. The relationship between steroid hormone profiles and abnormal glucose metabolism was determined by a binary logistic regression model with fasting insulin levels, age and BMI as covariates. P value < .05 was considered statistically significant.