| Abstract Title: | A combination of liquid chromatography tandem-mass spectrometry androgen profiling and machine learning identifies three distinct androgen metabolism subtypes in women with polycystic ovary syndrome |
| Presenter Name: | Dr Angela E Taylor |
| Co-authors: | Veronika Tandl Lida Abdi Tara Mcdonnell Roland Veen Eka Melson Thais Rocha James M Hawley Laura Wittemans Amarah Anthony Lorna C Gilligan Fozia Shaheen Punith Kempegowda Caroline D.T Gillett Leanne Cussen Cornelia Missbrenner Fannie Lajeunesse-Trempe Helena Gleeson Aled Rees Lynne Robinson Channa Jayasena Harpal Randeva Georgios K. Dimitriadis Larissa Gomes Alice J Sitch Elini Vradi Michael O'reilly Barbara Obermayer-Pietsch Michael Biehl Wiebke Arlt |
| Company/Organisation: | University of Birmingham |
| Country: | United Kingdom |
Abstract Information :
Introduction
Polycystic ovary syndrome (PCOS) affects 10% of women. Androgen excess is a cardinal feature of PCOS. Recent studies have demonstrated two classes of androgens may be responsible for the phenotype associated with androgen excess, the classic androgens, and 11-oxygenated androgens. We aimed to identify PCOS sub-types with distinct androgen profiles using liquid chromatography tandem-mass spectrometry (LC-MS/MS) combined with machine learning.
Experimental and Results
We quantified eleven androgens from classic and 11-oxygenated pathways, LC-MS/MS (Waters TQ-XS with Acquity). Calibrants and samples were spiked with isotopically labelled internal standards and extracted via liquid/liquid extraction using methyl-tertbutyl ether. Steroids were separated on a Phenomenex Luna Omega C18 column (1.6 µm, 100 Å, 2.1 x 50 mm), using a methanol and water (0.1% formic acid) gradient.
We recruited 488 women with PCOS [median age 28 (IQR 24-32) years; BMI 27.5 (22.4-34.6) kg/m2] from UK & Ireland (n=208), Austria (n=242) and Brazil (n=38). Steroid data was analysed by unsupervised k-means clustering, followed by statistical analysis of differences in clinical phenotype and metabolic parameters.
Machine learning analysis identified three subgroups of women with PCOS with distinct steroid metabolomes: the first characterised by mainly gonadal-derived androgen excess (GAE) (testosterone, dihydrotestosterone; 21.5%), the second with predominantly adrenal-derived androgen excess (AAE) (11-oxygenated androgens; 21.7%), and a third with comparably mild androgen excess (MAE; 56.8%). Groups demonstrated different rates of hirsutism, 76.4% (AAE) vs 67.6% (GAE) vs 59.9% (MAE) and female pattern hair loss 32.1% (AAE) vs 14.3% (GAE) vs 21.7% (MAE).
Conclusions
We discovered three subtypes of PCOS, each with a distinct androgen profile. The AAE group 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.
