I don’t have an answer, but Corbett and Morin-Papunin have some ideas in their new article at Molecular and Cellular Endocrinology. I’ve not read it, but here’s a quote from the abstract:
The Polycystic Ovary Syndrome (PCOS) is a complex endocrine disorder characterised both by reproductive and metabolic disturbance, and is the most common cause globally of ovarian infertility. It is also a familial polygenic condition, linked genetically to both Type 2 diabetes and the metabolic syndrome. The striking evolutionary paradox of this prominent genetically-based condition, which impairs fertility, is that not only should it have diminished in prevalence, but it should have done so rapidly – unless there has been some form of balancing selection.