Serum level of sex hormone-binding globulin is a good independent predictor of metabolic syndrome in women with polycystic ovary syndrome

Abstract

Background: Polycystic ovary syndrome (PCOS), being a common reproductive endocrine disorder resulting in oligo-amenorrhoea and anovulatory subfertility, is also associated with the metabolic syndrome (MetS) which is an important long term health problem. Hyperandrogenism is a cardinal feature of PCOS. This study analysed the predictive value of the androgen indices on the occurrence of MetS in women with PCOS.

Methods:

We prospectively recruited 462 Hong Kong Chinese women diagnosed with PCOS by the Rotterdam criteria during April 2011 to March 2013. All subjects attended the clinic within the first 5 days of spontaneous or induced menstruation. Fasting serum was assayed for total testosterone, sex hormone-binding globulin (SHBG), anti-Mullerian hormone (AMH), glucose, insulin and lipid profile. Cross-sectional analyses of the clinical and biochemical parameters at recruitment were performed. MetS was defined according to the Joint Interim Statement (Alberti et al, 2009). Prediction of the occurrence of MetS by SHBG and free androgen index (FAI = 100 X total testosterone / SHBG) was determined using the receiver-operator characteristic (ROC) curve as well as the univariate and multivariate logistic regression analyses after controlling for body mass index (BMI) and HOMA-IR.

Results:

Out of the 462 recruited PCOS subjects, 72 (15.6%) had MetS at recruitment. Compared to those without MetS, subjects having MetS had significantly higher BMI, waist circumference, SHBG, FAI and HOMA-IR. There was no significant difference in age, total testosterone, antral follicle count (AFC) and AMH among the two groups. Area under the ROC curve of SHBG for prediction of MetS (0.863, 95% CI 0.828–0.893 was higher than that of FAI (0.804, 95% CI 0.765–0840, p=0.005). A cut-off SHBG level at 26 nmol/L gave a sensitivity of 89% and specificity of 75% in predicting MetS. Univariate binary logistic regression revealed that SHBG, FAI, BMI and HOMA-IR, but not total testosterone, mean AFC nor AMH, were significant predictors of MetS. Putting SHBG, BMI and HOMA-IR into a multivariate logistic regression model showed that SHBG (p=0.001), BMI (p<0.001) and HOMA-IR (p=0.004) all remained as significant independent predictors of MetS.

Conclusions:

Serum level of SHBG was a good independent predictor of MetS in women with PCOS after controlling for adiposity and insulin resistance measures.