Abstract
Context: Polycystic ovary syndrome (PCOS) is a very common endocrinopathy with a heterogeneous presentation whose etiology is still uncertain. Not surprisingly, therefore, the definition of, and diagnostic criteria for, PCOS remain controversial.
Objective: The objective of the study was to review and justify the basis for the recently revised definition of PCOS arising from the joint European Society for Human Reproduction & Embryology/American Society for Reproductive Medicine (ESHRE/ASRM) Rotterdam workshop in 2003.
Intervention(s): The Rotterdam criteria take account of the broad spectrum of presenting features of PCOS, including women with hyperandrogenism but regular menses and, more controversially, those with menstrual disturbance without overt androgen excess.
Positions: The Rotterdam criteria for definition and diagnosis of PCOS, in the opinion of this author, represent a significant advance in recognizing the broad spectrum of presentation of the syndrome and acknowledge that the clinical and biochemical features may vary with time within individuals. The important refinements when compared to the 1990 NIH definition of PCOS are: 1) inclusion of polycystic ovarian morphology; and 2) inclusion of subjects with hirsutism and regular menses.
Conclusions: These new diagnostic criteria for PCOS reflect the significant advances, particularly from studies of familial PCOS, in understanding of the etiology of the syndrome and the basis for its heterogeneity. Under the revised diagnostic criteria, the inclusion of women with hyperandrogenism and regular cycles has met with general agreement. The inclusion of women with oligomenorrhea and polycystic ovaries who do not have clear evidence of androgen excess is, in the opinion of this author, also justified but remains a contentious issue and one that requires further investigation.
IN MAY 2003, a group of experts in the field of polycystic ovary syndrome (PCOS) gathered in Rotterdam for a conference sponsored jointly by the European Society for Human Reproduction and Embryology and the American Society for Reproductive Medicine. The objective of the meeting was to evaluate recent advances in the understanding of PCOS and included the express task of revisiting and, if necessary, redefining the diagnostic criteria for PCOS (1) in light of information resulting from those advances. Thus arose the so-called revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS) (2, 3). Two of three criteria had to be met to fit the definition of PCOS: chronic anovulation, clinical and/or biochemical evidence of hyperandrogenism, and polycystic ovaries (Table 1 and Fig. 1). A consensus may well have been reached by the participants of that meeting, but the resulting statement has not exactly settled the controversy surrounding the definition of PCOS. The main issues that continue to promote passionate discussion are, first, whether hyperandrogenemic women with polycystic ovaries and regular cycles should be included in the definition, and, second, and perhaps more contentiously, whether women with polycystic ovaries and chronic anovulation without evidence of androgen excess should be defined as part of the syndrome (Fig. 1).
Venn diagram illustrating the three elements of diagnosis according to the 2003 Rotterdam criteria, their interaction, and their relationship to the NIH criteria of 1990. Area 1 recognizes the now widely accepted entity of hyperandrogenism (clinical and/or biochemical) in women with regular cycles and polycystic ovaries. Area 2 illustrates the more contentious entity of anovulation in women with polycystic ovaries and no clear evidence of androgen excess. Area 3 reprises the 1990 NIH criteria.
| NIH 1990 | Rotterdam 2003 |
|---|---|
| 1) Chronic anovulation | 1) Oligo- and/or anovulation |
| 2) Clinical and/or biochemical signs of hyperandrogenism | 2) Clinical and/or biochemical signs of hyperandrogenism |
| 3) Polycystic ovaries |
| NIH 1990 | Rotterdam 2003 |
|---|---|
| 1) Chronic anovulation | 1) Oligo- and/or anovulation |
| 2) Clinical and/or biochemical signs of hyperandrogenism | 2) Clinical and/or biochemical signs of hyperandrogenism |
| 3) Polycystic ovaries |
The NIH definition requires both criteria to make the diagnosis; the ESHRE/ASRM requires two of three. In both definitions, the diagnosis assumes exclusion of other diagnoses that may replicate symptoms of PCOS (4 ) such as nonclassical congenital adrenal hyperplasia due to 21-hydroxylase deficiency.
| NIH 1990 | Rotterdam 2003 |
|---|---|
| 1) Chronic anovulation | 1) Oligo- and/or anovulation |
| 2) Clinical and/or biochemical signs of hyperandrogenism | 2) Clinical and/or biochemical signs of hyperandrogenism |
| 3) Polycystic ovaries |
| NIH 1990 | Rotterdam 2003 |
|---|---|
| 1) Chronic anovulation | 1) Oligo- and/or anovulation |
| 2) Clinical and/or biochemical signs of hyperandrogenism | 2) Clinical and/or biochemical signs of hyperandrogenism |
| 3) Polycystic ovaries |
The NIH definition requires both criteria to make the diagnosis; the ESHRE/ASRM requires two of three. In both definitions, the diagnosis assumes exclusion of other diagnoses that may replicate symptoms of PCOS (4 ) such as nonclassical congenital adrenal hyperplasia due to 21-hydroxylase deficiency.
PCOS is, by its nature, a concurrent set of symptoms, signs, and biochemical features that can occur in various combinations (4). As yet, we do not have a clear view of the etiology of PCOS, although both genetic and environmental influences are implicated (5, 6). Consequently, there is no single diagnostic marker, whether biochemical or genetic, to provide a gold standard for reference. We therefore rely upon a definition that incorporates a critical number of clinical and/or biochemical abnormalities. References to what we now recognize as PCOS can be found in scientific publications dating back to the 19th century, but in modern times, the syndrome was first clearly described in the classic paper by Stein and Leventhal in 1935 (7). In their series, the typical polycystic ovarian morphology was the common finding in a group of patients who had amenorrhea and clinical evidence of androgen excess. Nevertheless, even in this classic reference series, heterogeneity of presentation was apparent; not all patients were hirsute, and three showed evidence of recent ovulation (7). The combination of hyperandrogenism and chronic anovulation remained the mainstay of the definition of PCOS for the next half-century. This, indeed, was the basis of the widely accepted definition that was published in the proceedings of an international workshop on PCOS held at the National Institutes of Health (NIH) in 1990. Here PCOS was defined as chronic anovulation associated with clinical and/or biochemical evidence of androgen excess after exclusion of other pathologies (such as late-onset 21-hydroxylase deficiency) that might masquerade as PCOS (1). Notably, ovarian morphology did not form part of this definition.
It had, however, been recognized for many years that anovulation was not necessarily chronic, and that intermittent, or even prolonged, episodes of regular, ovulatory cycles could punctuate the pattern of anovulatory vaginal bleeding or amenorrhea (7–9). It was really the application of high definition ultrasound imaging of the ovaries in the 1980s that first highlighted the finding that polycystic ovaries were very commonly associated with hirsutism and hyperandrogenemia in women with regular, ovulatory cycles (10, 11). The question that arose was are such women part of the same spectrum of presentation as classic PCOS, as defined by the NIH criteria? There is much evidence to suggest that the answer is yes (12). The results of more recently published data have confirmed that women with regular cycles and polycystic ovaries share many of the biochemical features of PCOS (13). Such evidence includes data from studies of familial PCOS in the search for the genetic basis of the syndrome. Probands presenting with classic features of PCOS, anovulation and hyperandrogenism, may have an affected sister who is equally hyperandrogenemic, but has regular cycles (14, 15) and polycystic ovaries (6).
Intriguingly, women with polycystic ovaries, hyperandrogenism, and regular cycles are less likely to have insulin resistance and hyperinsulinemia than those with chronic anovulation (16–18). The link between metabolic disturbance and anovulation has implications for understanding the mechanism of arrested antral follicle development in anovulatory women with PCOS (19). Obese women with PCOS are more likely to be anovulatory than lean, hyperandrogenemic subjects (20), and weight gain is associated with an increasing chance of chronic anovulation. Conversely, diet and lifestyle changes in obese women with PCOS or insulin-lowering/sensitizing medication have beneficial effects on ovulation and fertility (21–23); this supports the association between metabolic abnormalities and ovulatory function. Importantly, the ability to move between ovulatory and anovulatory cycles as a result of dietary factors or advancing reproductive age (women with PCOS tend to have more regular cycles as they enter their late thirties and forties) (24) provides persuasive evidence that PCOS represents a wide, but single, spectrum of disordered ovarian and metabolic function. The burgeoning evidence for including ovulatory women with polycystic ovaries and hyperandrogenism in the definition of PCOS was a major factor in motivating the European Society for Human Reproduction and Embryology/American Society for Reproductive Medicine workshop.
We now come to the issue of whether women with polycystic ovaries and chronic anovulation, but without clinical or biochemical evidence of androgen excess, can be said to have polycystic ovary syndrome. There is some concern that by using the Rotterdam criteria women who had polycystic ovaries, but were amenorrheic because of either a disorder of gonadotropin regulation [i.e. hypothalamic amenorrhea (HA)] or primary ovarian failure (POF) might be misdiagnosed as having PCOS. However, this should not be an issue, because these disorders come under the heading of other endocrine disorders that should be excluded by both the NIH and Rotterdam definitions. Thus, evaluation of estrogen status (PCOS is characterized by estrogen-replete amenorrhea or oligomenorrhea, whereas women with HA or POF are estrogen deficient) and gonadotropins (LH and FSH are low or normal in HA, but elevated in POF) will identify these other pathologies. In the context of clinical practice, it is unlikely, for example, that an underweight woman with hypogonadotropic, estrogen-deficient amenorrhea who happens to have polycystic ovaries will be diagnosed as having PCOS.
Of more concern are women with polycystic ovaries and estrogen-replete amenorrhea or oligomenorrhea who have neither clinical nor biochemical evidence of androgen excess. There are two questions that need to be answered with regard to such subjects. The first question is how common is this? i.e. what proportion of women with chronic anovulation and polycystic ovaries have no indices of androgen excess? There have been few systematic studies that have attempted to address this question, but the limited evidence to date favors the inclusion of this subgroup within the definition of PCOS. In a series of 84 patients with anovulation and polycystic ovaries, it was found that although less than half the patients were hirsute, only 3% had no biochemical features to support the diagnosis of PCOS, i.e. elevated serum concentrations of testosterone, androstenedione, and/or LH (10). The second question is are anovulatory patients without clear evidence of androgen excess a distinct subgroup or do they fit into the same spectrum as hyperandrogenic women with chronic anovulation? In terms of response to induction of ovulation, the baseline serum total testosterone concentration in anovulatory women with polycystic ovaries has little impact on the outcome of treatment (25–29). Women with elevated serum total testosterone required a higher threshold dose of FSH to induce unifollicular ovulation, but there was no effect of testosterone on ovulation or pregnancy rate (29). In contrast, body mass index had a significant adverse effect on ovulation and ongoing pregnancy rate (25, 29).
Another issue to be considered is the definition of biochemical hyperandrogenism. Conventionally, this implies raised serum concentrations of testosterone (or elevated free androgen index) or androstenedione. However, women with asymptomatic polycystic ovaries who have normal baseline levels of serum androgens on repeated testing may have occult androgen excess. Those with polycystic morphology alone demonstrate abnormal production of androstenedione and 17α-hydroxyprogesterone after hCG stimulation (30). Therefore, while acknowledging that this is an area that would benefit from additional research to provide definitive answers, the available data appear to support the inclusion of what appears to be a small subsection of patients with polycystic ovaries and anovulation under the definition outlined in Rotterdam.
In any syndrome in which the primary pathological abnormality remains unclear, there will inevitably be discussion about its definition and diagnosis. The Rotterdam criteria are necessarily more inclusive than the NIH criteria, but that does not mean that they are imprecise. In my view, the Rotterdam definition is much more appropriate in this era in which the broader spectrum of clinical and biochemical presentation of women with polycystic ovaries has been widely acknowledged. As the studies of familial PCOS suggest, it would seem unwise to consider the anovulatory hirsute woman to have a fundamentally different disorder from the woman with hirsutism and regular cycles or the nonhirsute woman with anovulatory infertility. This is not just a matter of semantics; it is as important in clinical management as it is in devising and interpreting research programs. It may be useful to subcategorize patients (e.g. those with the classic NIH criteria) to provide a more homogeneous group for research, but in terms of clinical management, the broader definition has much merit and should be allowed to hold sway.
Acknowledgments
Received November 16, 2005. Accepted December 23, 2005.
