New Phenotypes Proposed by the Rotterdam 2003 Criteria as PCOS

Phenotypes of PCOS according to the NIH 1990 and Rotterdam 2003 criteria¹

Features	Phenotypes
	A	B	C	D
Ovulatory dysfunction	X	X	X
Hirsutism and/or hyperandrogenemia	X	X		X
Polycystic ovaries	X		X	X
PCOS by NIH 1990	✓	✓
PCOS by Rotterdam 2003	✓	✓	✓	✓

X, Features included in phenotypes; ✓, phenotypes included in definitions.

¹

Assuming exclusion of other androgen excess or related disorders.

View Large

Defining a Syndrome

A syndrome is a symptom complex of unknown etiology, characteristic of a particular abnormality. There are a number of approaches to defining a syndrome, including 1) historical usage in medical practice and/or literature, 2) expert knowledge and consensus processes, and 3) analysis of hard data by an independent body, i.e. evidence-based (e.g. the process generally undertaken by the NIH Consensus Development Program; http://consensus.nih.gov/). Considering that the current definitions for PCOS have been arrived at by a consensus process of experts and advocates in the field, it is instructional to review the comments of author Michael Crichton, who argued that “… . the work of science has nothing whatever to do with consensus. Consensus is the business of politics. Science, on the contrary, requires only one investigator who happens to be right, which means that he or she has results that are verifiable by reference to the real world. In science consensus is irrelevant. What is relevant is reproducible results. The greatest scientists in history are great precisely because they broke with the consensus. … . There is no such thing as consensus science. If it’s consensus, it isn’t science. If it’s science, it isn’t consensus. Period” (8). He concludes his argument by noting that “consensus is invoked only in situations where the science is not solid enough”. In light of this argument we should be compelled to examine what scientific evidence we have to support the available definitions of PCOS.

First, we should note that PCOS is not the only disorder facing the dilemma of definition. Irritable bowel syndrome, systemic lupus erythematosus, antiphospholipid syndrome, fibromyalgia, chronic fatigue syndrome, and, more closely related to PCOS, the metabolic syndrome, are among the many disorders that are experiencing unclear and competing definitions. In fact, although the metabolic or insulin resistance syndrome was initially described in the late 1960s (9), today there are still at least six different diagnostic criteria in use (10–15). Even the name of this syndrome has not been agreed upon. In comparison, the controversy surrounding the definition of PCOS appears almost trivial.

How are hard data used to define a syndrome and its phenotypic limits? By definition, a syndrome is of unknown etiology, at least when initially described, and this appears to be the case for PCOS. Consequently, we cannot define a syndrome as the collection of those phenotypes with a common underlying etiology. Alternatively, we can define a syndrome by its consequences or long-term impact or morbidity. For example, we define the metabolic syndrome as a symptom complex that results in an increased risk for cardiovascular disease (15). In this analysis, we should note that the features used in defining a syndrome should not also be used to determine its phenotypic limits. Consistent with this concept, in a joint statement from the American Diabetes Association and the European Association for the Study of Diabetes, Kahn et al. (15) did not consider previous publications whose focus was on the ability of the metabolic syndrome to predict type 2 diabetes mellitus (DM), because the definition of the syndrome already included glucose intolerance, a strong predictor of diabetes.

Defining the Phenotypic Limits of the Syndrome of PCOS

The phenotypic limits of a syndrome can be defined by 1) identifying the specific phenotypes that may compose or be included in the syndrome, and 2) as discussed above, determining the long-term morbidity(s) that will be used as the phenotypic anchor. So, what long-term morbidity can we use to define PCOS? In a recent analysis of economic burden, we noted there is good evidence that PCOS is associated with increased risks for 1) menstrual dysfunction and abnormal uterine bleeding, 2) infertility, 3) type 2 DM, and 4) hirsutism (16). However, because hirsutism (i.e. clinical evidence of hyperandrogenism) and oligoanovulation (a strong predictor of menstrual dysfunction and infertility) are part of all current definitions of PCOS, these morbidities should probably not be considered when attempting to define the phenotypic limits and consequently the criterion for the definition. Alternatively, it may be possible to define PCOS by its increased risk for type 2 DM. It should be noted that the use of the risk of type 2 diabetes to identify the phenotypic limits of PCOS does not imply that that all women with PCOS, as defined, will develop type 2 DM, but that the phenotypic group as a whole will be at higher risk. It also is not meant to denote that this is the only or necessarily the most important of the morbidities, because these patients clearly suffer reproductive and dermatological consequences. Rather, type 2 DM is being used as a phenotypic anchor against which to measure each of the phenotypes.

How do the proposed phenotypes of PCOS relate to this morbidity? Considering the features of oligoanovulation, hirsutism and/or hyperandrogenemia, and polycystic ovaries, both the NIH 1990 and the Rotterdam 2003 criteria agree on two phenotypes (phenotypes A and B; Table 2), whereas Rotterdam 2003 defines an additional two phenotypes (phenotypes C and D; Table 2). Ample evidence is available to support the association of the two phenotypes included in both the NIH 1990 and the Rotterdam 2003 criteria (i.e. oligoanovulation, and hirsutism and/or hyperandrogenemia, with or without polycystic ovaries) with an increased risk for type 2 DM (17–19).

Alternatively, what is the evidence to support consideration of the two additional phenotypes proposed by the Rotterdam 2003 criteria (phenotypes C and D; Table 1) as PCOS? First, there is evidence that the sole presence of polycystic ovaries in women, observable in 20–30% of the population of reproductive-age women (20–25), is associated with features reminiscent of those observed in patients with PCOS. The presence of polycystic ovaries is associated with mild elevations in circulating LH (21, 24, 26) and androgen (21, 24, 26, 27) levels and a higher prevalence of obesity and hirsutism (20, 21). Alternatively, there are conflicting data regarding the impact of polycystic ovaries on measures of insulin action (25, 27–29) and their association with impaired fertility (20, 26). Nonetheless, we should recognize that the majority of these studies were performed using older criteria for polycystic ovaries and often transabdominal ultrasonography, and the prevalence, specificity, and relevance of polycystic ovaries diagnosed by newer criteria (30) remain to be determined.

A limited number of investigators have specifically studied ovulatory women with polycystic ovaries and hyperandrogenism. Carmina and colleagues (31, 32) studied women with polycystic ovaries, clinical and/or biochemical hyperandrogenism, and normal ovulation and observed that these patients had higher circulating androgen and insulin levels and a greater 17-hydroxyprogesterone response to a long-acting GnRH analog compared with controls. However, the degrees of hyperinsulinism and hyperandrogenemia were significantly less than those observed in patients with PCOS defined by the NIH 1990 criteria. Furthermore, whether these individuals have an increased risk for developing metabolic complications, including type 2 DM, is not known.

There are even less data showing that women with polycystic ovaries and ovulatory dysfunction, but without clinical or biochemical evidence of hyperandrogenism, are at increased risk for type 2 DM, as are those patients with PCOS defined by the NIH 1990 criteria. Norman and colleagues (33) compared six women with polycystic ovaries, irregular menstrual cycles, and no hirsutism or hyperandrogenemia with 54 women with polycystic ovaries, irregular cycles, and hyperandrogenism. They did not observe significant differences between the groups in age, body mass, waist to hip ratio, total cholesterol, or integrated glucose or insulin responses to an oral glucose challenge, although this could be a reflection of the small number of subjects studied. Whether this group of patients is at risk for the development of metabolic complications, including DM, is also unknown.

Finally, we should note that many patients with ovulatory disorders other than PCOS also demonstrate polycystic ovarian morphology, including 21-hydroxylase-deficient nonclassical CAH (34, 35), bulimia and other eating disorders (36, 37), hyperprolactinemia (38), or functional hypothalamic amenorrhea (FHA) (38, 39). In addition, many adolescents transiently demonstrate polycystic-appearing ovaries (40, 41). Nonclassical CAH can be differentiated from PCOS by basal or stimulated 17-hydroxyprogesterone (42), and the medical history may be able to discern some of the patients with bulimia and other eating disorders. However, FHA is usually defined by a history of amenorrhea of at least 6-month duration; negative urinary pregnancy test; serum LH, FSH, TSH, prolactin, and androgen levels within the normal range; and LH to FSH ratio less than 2 (43). So how will FHA be differentiated from women with nonhyperandrogenic PCOS? Although it may be argued that patients with FHA and girls undergoing the pubertal transition actually demonstrate multicystic and not polycystic ovaries, comprehensive studies and distinguishing ultrasonographic criteria are generally lacking.

Risks and Benefits of Considering the Two New Phenotypes Proposed by the Rotterdam 2003 Criteria as PCOS

What are the risks and benefits of accepting these two phenotypes and consequently the Rotterdam 2003 criteria as the new definition of PCOS? The negative impact of labeling the two new phenotypes proposed by the Rotterdam 2003 criteria (i.e. ovulatory women with polycystic ovaries and hyperandrogenism and, particularly, oligoanovulatory women with polycystic ovaries, but without hyperandrogenism) as PCOS can be significant. Although the proceedings of the Rotterdam 2003 conference acknowledged that the criteria were possibly not suitable for trials focusing on clinical outcomes in women with PCOS, the reality is that they are being used as such. Because these criteria increase the phenotypic heterogeneity of the disorder, their use will decrease the ability of genetic and other molecular studies to detect a common underlying abnormality. The use of these criteria also implies that we already understand the health risks of these newly proposed phenotypes far more than we do, potentially stifling future investigation. Importantly, the adoption of these criteria suggests to patients and their physicians that women with these new phenotypes are at increased risk for metabolic and cardiovascular consequences, as are patients with more classic forms of PCOS, despite the absence of data in this regard. Finally, these criteria have the potential of negatively affecting the insurability of patients who may have been prematurely labeled as having PCOS.

Alternatively, we should recognize that a finding of polycystic ovaries can predict the response to ovulation induction, because women with this ovarian morphology are more sensitive to gonadotropin stimulation than spontaneously cycling women, possibly as a result of the larger pool of small antral follicles available for recruitment (44). However, patient management need only be tailored to the ovarian morphology observed at the time of treatment and, consequently, does not require a previous diagnosis of PCOS.