PCOS (polycystic ovary syndrome) is a highly heritable, complex reproductive and metabolic disorder affecting up to 15% of reproductive-age women worldwide. The etiology of PCOS remains unknown so the diagnostic criteria, e.g. NIH and Rotter-dam, are based on expert opinion rather than on knowledge of disease mechanisms. Our recent meta-analysis of genomewide association studies (GWAS) of European (EA) ancestry cases found that the genetic architecture of PCOS defined by the different diagnostic criteria was generally similar. This finding suggests that these criteria do not identify biologically distinct disease subtypes. In contrast, using unsupervised hierarchical cluster analysis in EA PCOS, we identified two PCOS subtypes: a “reproductive” group characterized by higher luteinizing hor-mone (LH) and sex hormone binding globulin (SHBG) levels with relatively low BMI and insulin levels; and a “metabolic” group characterized by higher BMI as well as glucose and insulin levels with relatively low SHBG and LH levels. We replicated these subtypes in an additional EA PCOS cohort. We performed GWAS with the subtypes and found six novel loci at genomewide significance, five loci associated with the reproductive subtype and one locus associated with the metabolic subtype. Further, the effect sizes for these loci were substantially greater than those GWAS loci associated with PCOS diagnosis by the existing criteria. We have exciting pre-liminary data that these subtypes are present in PCOS cases of African American (AA), Hispanic (HA) and East Asian (Korean, KA) ancestry. Our overarching hypothesis is that there are phenotypic subtypes of PCOS with distinct genetic architecture. We will: (1) Test the hypothesis that there are subtypes of PCOS in additional EA cohorts of phenotypically diverse PCOS and assess the genetic architecture of these subtypes. We will perform unsupervised hierarchical cluster analysis of reproductive and metabolic quantitative traits in additional EA an-cestry PCOS case-control cohorts. We will formally assess differences in genetic architecture and conduct fine-mapping of GWAS data to select variants for Aim 3 functional studies. (2) Test the hypothesis that subtypes are present in PCOS of African, Hispanic and East Asian ancestry and assess the genetic architecture of these subtypes. Cluster analysis, GWAS with subtypes, assessment of genetic architecture and fine-mapping will be performed as in Aim 1 in AA, HA and KA PCOS case-control cohorts. Transethnic meta-analysis will be con-ducted to leverage differences in ancestry for gene discovery. (3) Test the hypothesis that high priority variants associated with PCOS subtypes are functional in tissues relevant to disease pathogenesis. We will identify the noncoding genetic variants from Aims 1 and 2 causing the genetic association signals with a high-throughput reporter assay we developed in human theca, granulosa and preadipocyte cell lines. The genes impacted will be investigated using CRISPR/Cas9-based assays. This research will have a sustained and lasting impact on the field by defining biologically relevant subtypes of PCOS and identifying causal variants in key pathways related to PCOS pathogenesis in diverse populations.

PCOS is a common genetic disorder affecting up to 15% of reproductive-age women worldwide. PCOS is a leading cause of infertility and type 2 diabetes in young women. Our studies promise to define biologically relevant subtypes of PCOS and to identify genes contributing to PCOS in diverse populations.