Data-driven endocrine-metabolic phenotypes in young women with polycystic ovary syndrome and associations with cardiometabolic risk markers.

Abstract

Polycystic ovary syndrome (PCOS) is a heterogeneous endocrine disorder associated with reproductive dysfunction and increased cardiometabolic risk. Conventional diagnostic phenotypes may not adequately capture endocrine-metabolic variability, particularly in young women. We aimed to identify data-driven endocrine-metabolic phenotypes in young women with PCOS and assess their associations with selected cardiometabolic risk markers. This cross-sectional study included 1032 women aged 16-25 years with PCOS diagnosed according to the Rotterdam criteria, selected from 1300 consecutive patients evaluated at a tertiary gynecological endocrinology center in Poland between 2018 and 2025. Principal component analysis (PCA) followed by Gaussian mixture model clustering was used to identify latent endocrine-metabolic subgroups. Cardiometabolic outcomes included impaired glucose tolerance (2-hour plasma glucose ≥ 140 mg/dL), elevated triglyceride/high-density lipoprotein cholesterol (TG/HDL) ratio, elevated non-high-density lipoprotein cholesterol (≥ 130 mg/dL), and a composite endpoint. PCA retained 10 components explaining 81.9% of total variance. The optimal clustering solution identified two subgroups: a predominant mixed endocrine-metabolic subgroup (n = 954; 92.4%) and a smaller thyroid/autoimmune-enriched subgroup (n = 78; 7.6%) characterized by higher anti-thyroid peroxidase antibody and thyroid-stimulating hormone levels. Cluster assignment certainty within the selected model was high, although sensitivity analyses suggested that the latent structure was partially driven by thyroid/autoimmune-related variables. The thyroid/autoimmune-enriched subgroup showed a higher prevalence of an elevated TG/HDL ratio, whereas differences in glucose tolerance and non-high-density lipoprotein cholesterol were less pronounced. These findings suggest the presence of exploratory endocrine-metabolic subgroup structure in young women with PCOS, including a thyroid/autoimmune-enriched subgroup associated with modest differences in lipid-related cardiometabolic risk markers. The observed heterogeneity extends beyond conventional diagnostic classifications, although external validation and longitudinal studies are required to determine clinical relevance and reproducibility. CLINICAL TRIAL REGISTRATION: This study was not registered as a clinical trial, as it was an observational cross-sectional analysis.