Like any other syndrome, PCOS is a complex of symptoms of unknown etiology. Such conditions are heterogeneous by nature, and the identification of useful diagnostic criteria poses a major challenge. A novel test aiming to identify a disease is usually assessed by its capacity to ascertain the absence or presence of this illness. In any syndrome, a gold standard for the condition to be identified is absent, rendering the usual assessment of sensitivity and specificity of clinical tests elusive. Hence, the debate regarding criteria for PCOS diagnosis seems to come down to which end point of the syndrome is considered most relevant and whether to be inclusive or exclusive.
In the current era of evidence-based medicine, features upon initial screening are being identified that influence the extent and prognosis of the disease or modify the most appropriate treatment strategy. In the context of PCOS, such a strategy may help to identify features (i.e., diagnostic criteria) relevant for health issues during different phases of life.
It was widely recognized at the turn of the millennium that PCOS defined by the 1992 National Institutes of Health (NIH) criteria (with hyperandrogenemia and cycle abnormalities both as mandatory features) was too narrow. Consequently, a consensus meeting was organized in Rotterdam in 2003. At the end of the 2-day closed workshop, it was agreed that diagnosis for PCOS should be widened, with the involvement of three rather than two criteria (the occurrence of polycystic ovaries was added). Thus, the Rotterdam criteria represent an extension rather than a replacement of the NIH PCOS criteria.
The Rotterdam consensus meeting was endorsed by the two largest global reproduction societies, and consequently, two identical papers were published simultaneously in the January 2004 issue of both Fertility and Sterility and Human Reproduction. The Rotterdam PCOS consensus paper became a citation classic in the field of reproductive medicine, with currently close to 2,000 citations for both papers. This probably reflects that the Rotterdam consensus criteria have been widely accepted in most parts of the world, although controversy remains. In 2009, the Androgen Excess and PCOS Society formulated a position statement in between the Rotterdam and the NIH criteria (but closer to the first), still persisting on hyperandrogenemia being the hallmark of PCOS diagnosis.
Since this society mainly focuses on future health risks associated with PCOS, it is no surprise that hyperandrogenemia was considered the most relevant feature. Certainly, hyperandrogenemia is clearly associated with metabolic dysfunction (and presumably long-term health outcomes) regardless of how PCOS is defined. As stated repeatedly, the assessment of hyperandrogenemia is notoriously inaccurate, both clinically and biochemically. Precise, accurate, and sensitive androgen assays should be developed and introduced into the clinic, and agreement should be reached on how to best identify the unbound (biologically active) fraction of total testosterone. In addition, evidence is accumulating that the ultrasound diagnosis of polycystic ovaries (added in the Rotterdam consensus criteria) can be replaced by serum concentration of anti-Müllerian hormone, although this assay too is not without problems.
Over the last decade, much relevant information has been generated in well-phenotyped women with PCOS regarding the validity of using various diagnostic criteria and implications for the clinical management of complaints such as hirsutism, cycle abnormalities, infertility, and pregnancy complications, for genetic and ethnic studies, and for the long-term health of the woman herself as well as her offspring. Clinical investigators should continue to focus on the adequate phenotyping of women with PCOS, which may allow prioritizing relevant diagnostic features dependent on the phase of life of the woman and on specific complaints.
