Anthropometric, Metabolic, and Endocrine Parameters as Predictors of Estimated Average Glucose and Other Biomarkers of Dysglycemia in Women with Different Phenotypes of Polycystic Ovary Syndrome

Abstract

AbstractThe aim of the study was to evaluate the efficacy of anthropometric, metabolic, and endocrine abnormalities as predictors of estimated average glucose and other biomarkers of dysglycemia in women with different phenotypes of polycystic ovary syndrome (PCOS). This cross-sectional study included 648 women with PCOS and 330 controls. A single protocol of investigation was applied for all subjects. PCOS women were divided by phenotypes according to the Rotterdam criteria. Biomarkers of dysglycemia were considered dependent variables and anthropometric, lipid, and hormone alterations as independent variables using univariate and multivariate logistic regressions. Univariate logistic regression analysis, controlled for age and BMI, showed that many biomarkers of dysglycemia could be predicted by anthropometric, lipid, and endocrine variables. Multivariate logistic models showed that in non-PCOS women estimated average glucose (eAG) was predicted by lower TSH levels (OR=0.39; p=0.045); fasting glucose was predicted by increased T (OR=2.3). For PCOS, phenotype A, eAG was predicted by decreased HDL-C (OR=0.17, p=0.023) and high levels of free estradiol (OR=7.1, p<0.001). Otherwise, in PCOS, phenotype D, eAG was predicted by higher levels of HDL-C. The current study demonstrated that eAG was poorly predicted by anthropometric, lipid, and hormone parameters. Nevertheless, without adding significant benefits, it was comparable with other established markers of dysglycemia in women with different PCOS phenotypes.