Machine learning-based steroid metabolome analysis reveals three distinct subtypes of polycystic ovary syndrome and implicates 11-oxygenated androgens as major drivers of metabolic risk

Abstract

Introduction: Polycystic ovary syndrome affects 10% of women and comes with a 2-3fold increased risk of type 2 diabetes, hypertension, and fatty liver disease. Androgen excess, a cardinal feature of PCOS, has been implicated as a major contributor to metabolic risk. Adrenal-derived 11-oxygenated androgens represent an important component of PCOS-related androgen excess and are preferentially activated in adipose tissue. We aimed to identify PCOS sub-types with distinct androgen profiles and compare their cardiometabolic risk parameters. Methods: We cross-sectionally studied 488 treatment-naïve women with PCOS diagnosed according to Rotterdam criteria [median age 28 (IQR 24-32) years; BMI 27.5 (22.4-34.6) kg/m 2 ] prospectively recruited at eight centres in the UK & Ireland (n=208), Austria (n=242) and Brazil (n=38). All participants underwent a standardised assessment including clinical history, anthropometric measurements, fasting bloods and a 2-hour oral glucose tolerance test. We quantified 11 androgenic serum steroids, including classic and 11-oxygenated androgens, using a validated multi-steroid profiling tandem mass spectrometry assay. We measured serum insulin to calculate HOMA-IR and the Matsuda insulin sensitivity index (ISI). Steroid data were analysed by unsupervised k-means clustering, followed by statistical analysis of differences in clinical phenotype and metabolic parameters. Results: Machine learning analysis identified three stable subgroups of women with PCOS with minimal overlap and distinct steroid metabolomes: a cluster characterised by mainly gonadal-derived androgen excess (testosterone, dihydrotestosterone; GAE cluster; 21.5% of women), a cluster with predominantly adrenal-derived androgen excess (11-oxygenated androgens; AAE cluster; 21.7%), and a cluster with comparably mild androgen excess (MAE cluster; 56.8%). Age and BMI were similar between groups. As compared to GAE and MAE, the AAE cluster had the highest rates of hirsutism (76.4% vs 67.6% vs 59.9%) and female pattern hair loss (32.1% vs 14.3% vs 21.7%). The AAE cluster had significantly increased insulin resistance as indicated by higher values for fasting insulin, 120min insulin and HOMA-IR, and lower ISI than GAE and MAE clusters (all p<0.01). The AAE cluster also had a 2-3fold higher prevalence of impaired glucose tolerance and newly diagnosed type 2 diabetes. Conclusion: Unsupervised cluster analysis revealed three PCOS subtypes with distinct androgen excess profiles. Women within the adrenal androgen excess cluster had a significantly higher prevalence of insulin resistance, impaired glucose tolerance and type 2 diabetes. These results implicate 11-oxygenated androgens as major drivers of metabolic risk in PCOS and provide proof-of- principle for an androgen-based stratification tool that could guide future preventative and therapeutic strategies in women with PCOS.