DAISy-PCOS Phenome Study - Dissecting Androgen Excess and Metabolic Dysfunction in Polycystic Ovary Syndrome

Polycystic Ovary Syndrome

polycystic ovary syndrome, androgens, steroids, metabolic risk, prediction, stratified medicine

Polycystic ovary syndrome (PCOS) affects 10% of all women and usually presents with irregular menstrual periods and difficulties conceiving. However, PCOS is also a lifelong metabolic disorder and affected women have an increased risk of type 2 diabetes, high blood pressure, and heart disease. Increased blood levels of male hormones, also termed androgens, are found in most PCOS patients. Androgen excess appears to impair the ability of the body to respond to the sugar-regulating hormone insulin (=insulin resistance). The investigator has found that fat tissue of PCOS patients overproduces androgens and that this can result in a build-up of toxic fat, which increases insulin resistance and could cause liver damage. In a large cohort of women registered in a GP database, the study team have found that androgen excess increases the risk of fatty liver disease. The aim is to identify those women with PCOS who are at the highest risk of developing metabolic disease, which would allow for early detection and potentially prevention of type 2 diabetes, high blood pressure, fatty liver and cardiovascular disease. The investigator will assess clinical presentation, androgen production and metabolic function in women with PCOS to use similarities and differences in these parameters for the identification of subsets (=clusters) of women who are at the highest risk of metabolic disease. The investigator will do this by using a standardised set of questions to scope PCOS-related signs and symptoms and the patient's medical history and measure body composition and blood pressure. This standardised recording of a patient's clinical presentation (=clinical phenotype) is called Phenome analysis. The investigator will collect blood and urine samples for the systematic measurement of steroid hormones including a very detailed androgen profile (=steroid metabolome analysis) and of thousands of substances produced by human metabolism (=global metabolome analysis). Phenome and metabolome data will then undergo integrated computational analysis for the detection of clusters predictive of metabolic risk.

The investigator propose an innovative approach to solving the clinical problem at hand, the lack of identified measurable parameters one can use to predict the risk of future metabolic disease in women diagnosed with PCOS.The chosen approach is the standardised collection of phenome and metabolome data and their unbiased integration by machine learning analysis. Utilising the detailed results of the clinical phenome and metabolome analysis in the DAISy-PCOS Phenome Study cohort, The study will aim to identify distinct subsets (=clusters) of PCOS patients that share similar characteristics. This approach has previously been used by the team to successfully identify distinct steroid markers that can serve as a malignant steroid fingerprint in urine to distinguish benign from malignant tumours in patients with incidentally discovered adrenal masses. Similarly, The investigator have used unbiased analysis of steroid metabolome data to reveal that patients with aldosterone excess also overproduce glucocorticoids and that the latter explains the majority of metabolic disease risk observed in affected patients. In the integrated analysis of the DAISy-PCOS phenome and metabolome data, The investigator will apply a variety of methods in the context of connectivity or centroid-based clustering and density estimation. Supervised relevance learning will give insight into markers, e.g. steroids, that are most decisive for the determination of cluster memberships. In addition, The investigator will use state-of-the-art visualisation and machine learning techniques based on adaptive similarity measures.the investigator will use integrative approaches, addressing the heterogeneous data from different sources as a whole, whilst considering data-driven adaptation of generative models for the underlying biological processes. The investigator will employ these approaches to characterise central phenotype clusters affecting large numbers of patients as the basis of personalised management including outcome prediction.

No related papers found yet.

United Kingdom