C. K. Welt,* J. A. Gudmundsson,* G. Arason,* J. Adams, H. Palsdottir, G. Gudlaugsdottir, G. Ingadottir, and W. F. Crowley

Transcription

1 X/06/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 91(12): Printed in U.S.A. Copyright 2006 by The Endocrine Society doi: /jc Characterizing Discrete Subsets of Polycystic Ovary Syndrome as Defined by the Rotterdam Criteria: The Impact of Weight on Phenotype and Metabolic Features C. K. Welt,* J. A. Gudmundsson,* G. Arason,* J. Adams, H. Palsdottir, G. Gudlaugsdottir, G. Ingadottir, and W. F. Crowley Reproductive Endocrine Unit (C.K.W., J.A., W.F.C.), Department of Medicine, Massachusetts General Hospital, Boston Massachusetts 02114; ÞJÓNUSTUMIiÐSTÖÐ RANNSÓKNAVERKEFNA (H.P., G.G., G.I.), NÓATÚN 17, 105 Reykjavík, Iceland; and Department of Obstetrics and Gynecology (J.A.G.), Landspitali University Hospital, and ARTmedica IVF (G.A.), 101 Reykjavík, Iceland Context: The Rotterdam criteria for polycystic ovary syndrome (PCOS) defines discrete subgroups whose phenotypes are not yet clear. Objective: The phenotypic characteristics of women in the PCOS subgroups defined by the Rotterdam criteria were compared. Design: The study was observational. Setting: Subjects were studied in an outpatient setting in Boston and Reykjavik. Patients: Four subgroups of subjects with PCOS defined by 1) irregular menses (IM), hyperandrogenism (HA), and polycystic ovary morphology (PCOM, n 298); 2) (n 7); 3) (n 77); and 4) and a group of controls, aged yr, were examined. Intervention: Subjects underwent a physical exam; fasting blood samples for androgens, gonadotropins, and metabolic parameters; and a transvaginal ultrasound. COMPARING THE PHENOTYPES of polycystic ovary syndrome (PCOS) around the world has been extremely difficult because of the lack of standardized criteria to make this diagnosis. A recent workshop in Rotterdam attempted to reconcile these differences by expanding the diagnostic criteria for PCOS with the addition of ultrasound evaluations (1, 2). According to these new criteria, PCOS can be defined when two of three of the following symptoms or signs are present: 1) irregular menses (IM); 2) hyperandrogenism (HA), either clinical or biochemical; and/or 3) PCO morphology (PCOM) on pelvic ultrasound; all in the absence of another disorder that can cause the same symptoms. Using these Rotterdam criteria, there are four possible diagnostic subcategories of First Published Online September 26, 2006 * C.K.W., J.A.G., and G.A. contributed equally to the work. Abbreviations: ANCOVA, Analysis of covariance; BMI, body mass index; DHEAS, dehydroepiandrosterone sulfate; HA, hyperandrogenism; HOMA, homeostasis model assessment; IM, irregular menses; LSD, least significant difference; PCOM, polycystic ovary morphology; PCOS, polycystic ovary syndrome. JCEM is published monthly by The Endocrine Society ( endo-society.org), the foremost professional society serving the endocrine community. Main Outcome Measures: The phenotype was compared between groups. Results: Ninety-seven percent of women with had PCOM. Therefore, the groups with and without PCOM were combined. The Ferriman-Gallwey score and androgen levels were highest in the hyperandrogenic groups ( and ), whereas ovarian volume was higher in all PCOS subgroups compared with controls, as expected based on the definitions of the PCOS subgroups. Body mass index and insulin levels were highest in the subgroup. Conclusions: Subjects with PCOS defined by are the most severely affected women on the basis of androgen levels, ovarian volumes, and insulin levels. Their higher body mass index partially accounts for the increased insulin levels, suggesting that weight gain exacerbates the symptoms of PCOS. (J Clin Endocrinol Metab 91: , 2006) PCOS, i.e. IM/,,, and. Documenting differences in subjects within these four PCOS groupings is now critical to determine whether their phenotypes are distinct or overlapping with each other or with unaffected women (3, 4). The definition of these PCOS subsets is especially timely because the resulting phenotypes may have implications for the genetic searches that they presage now that the tools are in place to determine the gene(s) underlying this polygenic condition (5, 6). Although three subgroups have been evaluated (7 9), there have been no studies comparing the differences between all four of the subgroups defined by the Rotterdam workshop, nor have there been comparisons in different ethnic groups. In addition, the existing subgroup evaluations have differed. One study demonstrated differences in the body mass index (BMI) and insulin levels in women with compared with those with (8, 9), whereas the other failed to demonstrate such differences between the three groups (7). We have studied a large population of women with PCOS in Iceland and Boston. We therefore set out to contrast women with PCOS defined by 1) IM/; 2) with normal ovarian morphology; 3) with regular menses; and 4) in the absence of HA. 4842

2 Welt et al. PCOS Phenotypes Defined by the Rotterdam Criteria J Clin Endocrinol Metab, December 2006, 91(12): PCOS subjects Subjects and Methods Four-hundred eighteen subjects with PCOS, aged yr, were recruited in Iceland (n 154) and Boston (n 264). Icelandic subjects were recruited from the practice of two gynecologists (J.A.G. and G.A.; n 88), an obesity study (n 19), self-referral (n 12), and advertisements or family members of PCOS subjects (n 35). Boston subjects were recruited from endocrinology or medicine practices and advertisements (supplemental Table 5, published as supplemental data on The Endocrine Society s Journals Online web site at PCOS subjects in Boston were recruited based on the National Institutes of Health criteria (2). However, a subset of subjects presenting as controls with regular menstrual cycles were discovered to have HA and PCOM on ultrasound (n 30) and were therefore included with the PCOS group defined as (see below). subjects subjects from Boston (n 32) and Iceland (n 32) were recruited simultaneously. Subjects aged yr had menstrual cycle lengths between 25 and 35 d and subjects aged yr had menstrual cycle lengths between 21 and 35 d based on age-related follicular-phase shortening (10). Subjects had no self-reported or physical exam evidence of hirsutism. subjects were not sisters of subjects fulfilling the Rotterdam criteria (1) for the diagnosis of PCOS. All subjects in Iceland were examined by one of two research nurses (H.P. and G.I.), trained by a Boston physician (C.K.W.). All subjects in Boston were examined by one of two physician assistants and a Boston physician (C.K.W.). Subjects were divided into four PCOS subgroups based on the criteria outlined in the Rotterdam PCOS consensus workshop (1). The groupings included 1) IM (fewer than nine menstrual periods per year), clinical and/or biochemical evidence of HA (elevated Ferriman-Gallwey score or androgen level; see below), and PCOM on ultrasound (11, 12) (IM/); 2) IM and HA without PCOM (); 3) HA and PCOM with regular menstrual cycles of d (); and 4) IM and PCOM with no HA (). Clinical HA was defined as a Ferriman-Gallwey score greater than the upper 95% confidence limit for control populations in Iceland ( 6) (13) or Boston ( 9) (14). Biochemical HA was defined as an androgen level greater than the 95% confidence limits in control subjects with regular, ovulatory menstrual cycles [testosterone 63 ng/ml (2.8 nmol/liter), dehydroepiandrosterone sulfate (DHEAS) 430 g/dl (1.16 mol/liter), or androstenedione levels 3.8 ng/ml (13.3 nmol/liter)] (14). All subjects were healthy, not on hormonal medication, and had normal thyroid function, prolactin levels, and a premenopausal follicular phase FSH level. Late-onset congenital adrenal hyperplasia was excluded with a follicular-phase 17OH-progesterone of no more than 300 g/dl (15). Protocol The study was approved by the Institutional Review Board of the Massachusetts General Hospital, the Data Protection Commission of Iceland, and the National Bioethics Committee of Iceland. All subjects gave written informed consent. PCOS subjects were studied at least 10 d after their last menstrual period to avoid the time frame in which LH and androgen levels are suppressed after spontaneous ovulation (14). subjects were studied in the follicular phase. All subjects arrived after a 12-h fast and underwent a detailed history and physical exam, including an assessment of hirsutism by the method of Ferriman and Gallwey (16) and measurement of waist circumference at the umbilicus and hip circumference at the widest diameter. Subjects had blood samples drawn for fasting lipids, glucose, insulin, gonadotropin, and sex-steroid levels. Additional blood samples were drawn at 10 and 20 min for LH and FSH to obtain an average gonadotropin concentration, as previously described (17). Subjects also underwent a transvaginal ultrasound or, rarely, a transabdominal ultrasound (ATL HDI 1500, 5-MHz convex array transducer) to assess ovarian morphology. Ultrasound scans were read and scored as PCOM, independently, by one of two experienced and blinded reviewers (C.K.W. or J.M.A.). Assays All blood samples were assayed in the Reproductive Endocrine Unit Laboratory at Massachusetts General Hospital. Serum LH and FSH were measured using a two-site monoclonal nonisotopic system (Axsym; Abbott Laboratories, Abbott Park, IL) (18) and are expressed in international units per liter as equivalents of the Second International Reference Preparation 71/223 of human menopausal gonadotropins. Serum testosterone and androstenedione were measured using a RIA (Coata-Count; Diagnostic Products Corp., Los Angeles, CA) (19). 17OH-Progesterone was measured by RIA (20). SHBG and insulin were measured using an immunochemiluminescent immunoassay (Immulite; Diagnostic Products Corp.). Data analysis A progesterone level greater than 1 ng/ml ( 3.2 nmol/liter) was taken as evidence of presumed recent ovulation. In those subjects, gonadotropins and androgens were not analyzed to avoid the known effect of progesterone to lower these laboratory results (14). Ovarian volume was calculated as length width height in centimeters multiplied by (21). PCOM was defined as at least one ovary with 10 or more follicles of 2 10 mm in a single plane (11, 12) or an ovarian volume of more than 10 ml in the absence of a dominant follicle more than 10 mm, a corpus luteum, or a cyst (1, 11). A peripheral pattern of follicles surrounding increased stroma was not required. The maximum ovarian volume and follicle number in both ovaries was used for comparisons, excluding the volume of an ovary with a dominant follicle, corpus luteum, or cyst. Free testosterone was calculated using total testosterone and SHBG (22). The homeostasis model assessment (HOMA) was used to estimate insulin resistance (23). Laboratory and ultrasound data were compared between groups using analysis of covariance (ANCOVA) to control for age and BMI, or age only, for anthropomorphic measurements. Fisher s least significant difference (LSD) post hoc test was used to determine significant differences between groups. Categorical variables were compared using 2 tests. All analyses were also performed between Icelandic subjects and Boston Caucasian subjects (supplemental Tables 1 4, published as supplemental data on The Endocrine Society s Journals Online web site at A z test was used to compare the proportions of subjects with metabolic syndrome in relation to data from the NHANES III study (24). Spearman correlations were used to examine relationships between variables. Data are expressed as mean sd, unless indicated. A P value of 0.05 was considered significant. Results Of the total number of subjects, 298 (71%) fulfilled the criteria for IM/, seven (2%) for, 77 (18%) for, and 36 (9%) for (Table 1). Virtually all subjects with also had PCOM (see Table 4). The only differences between the seven subjects with and normal ovarian morphology and those with PCOM were older ages ( vs yr; P 0.01) and higher FSH levels ( vs IU/liter; P 0.001). Because follicle count and the prevalence of PCOM decrease with age in women with PCOS (25, 26), the groups / PCOM and were analyzed together and designated. Anthropomorphic and physical exam features Subjects in each subset differed slightly in age (Table 1). Weight, BMI, waist and hip measurements, and waist-to-hip ratio were increased in women with compared with those with, those with, and controls (Table 1 and Fig. 1). Systolic and diastolic blood pressures were not different between the groups when controlled for

3 4844 J Clin Endocrinol Metab, December 2006, 91(12): Welt et al. PCOS Phenotypes Defined by the Rotterdam Criteria TABLE 1. Comparison of the PCOS phenotype in subjects with,, and P value, ANCOVA Age (yr) Height (m) Weight (kg) a b b b BMI (kg/m 2 ) a b b b Waist circumference (cm) a b b b Hip circumference (cm) a b b b Waist/hip ratio a b b b SBP (mm Hg) DBP (mm Hg) FG Score a b c c No. with acne (%) 236 (80.3) 48 (66.7) 31 (88.6) 34 (57.6) No. with acanthosis (%) 191 (64.5) 23 (31.5) 15 (42.9) 13 (22.0) P values are indicated for the differences in groups as analyzed by a one-way ANOVA for age and BMI and by ANCOVA controlling for age for anthropomorphic measures and age and BMI for systolic blood pressure (SBP), diastolic blood pressure (DBP), and Ferriman-Gallwey (FG) score. age and BMI. As expected based on the criteria used to define the categories of PCOS, there was a gradient in Ferriman- Gallwey scores with the highest score in the group and the lowest score in and controls. The prevalence of acne was highest in the and groups. The prevalence of acanthosis was greatest in PCOS subjects with, intermediate in subjects with IM/ PCOM, and lowest in subjects with and controls. When the Caucasian Icelandic women with PCOS were analyzed separately, women with and had similar weight, BMI, waist and hip circumference, and waist-to-hip ratio (supplemental Table 1). s in Iceland were also well matched for these parameters. The Ferriman- Gallwey scores in the Icelandic women with and were also similar. The prevalence of acne and acanthosis did not differ in the Icelandic PCOS subjects and controls. Gonadotropin and androgen levels LH and the LH/FSH ratio were higher in women with and compared with controls and were higher in women with than women with HA/ PCOM (Table 2 and Fig. 1), but FSH levels did not differ among groups. The predominantly ovarian androgen testosterone exhibited decreasing levels with the apparent severity of PCOS (Table 2 and Fig. 1). Testosterone and free testosterone levels were highest in women with, intermediate in women with, and lowest in women with IM/ PCOM and controls. The opposite pattern was demonstrated for SHBG. The hormones secreted at least partially by the adrenal, androstenedione, DHEAS, and 17OH-progesterone exhibited higher levels in the HA groups ( and HA/ PCOM) and lower levels in women with and controls. When analyzed separately, there was no difference in LH or in the LH/FSH ratio between PCOS subgroups and controls in Iceland (supplemental Table 2). In addition, Icelandic women with PCOS in the subgroups and HA/ PCOM had similar testosterone, SHBG, and free testosterone, whereas DHEAS was highest in women with. Metabolic hormones There were no differences in the fasting glucose or glycosylated hemoglobin levels among the four groups, although insulin resistance varied significantly. Insulin levels and HOMA for insulin resistance were highest in women with compared with women with and controls when controlled for age and BMI (Table 3). There were no differences in cholesterol, high-density lipoprotein, low-density lipoprotein, or triglyceride levels between groups when controlled for age and BMI. When analyzed separately, there was no difference in insulin levels or HOMA in Icelandic PCOS subjects and controls (supplemental Table 3). High-density lipoprotein levels were different among the PCOS subgroups in Iceland, with the lowest levels in women with and highest in women with. There was no difference in the prevalence of impaired fasting glucose or type 2 diabetes in the four groups. Although the prevalence of metabolic syndrome was highest in women with and lowest in women with, particularly in the 30- to 39-yr-old age group, there was no difference in the prevalence of metabolic syndrome when women with a BMI greater than 30 kg/m 2 were compared. The prevalence of metabolic syndrome was also compared with that in the U.S. Third National Health and Nutrition Examination Survey (NHANES III) (24). The proportion of subjects with metabolic syndrome was higher in PCOS women with when compared with women in the NHANES III study aged yr (17 vs. 6%; vs. NHANES III; P 0.001; Table 3) and yr (30 vs. 15%; P 0.001). However, there was no difference in the proportion of subjects with metabolic syndrome when subjects with a BMI greater than 30 kg/m 2 were compared between the two groups (20 29 yr, 30 vs. 28%; and yr, 39 vs. 43%; vs. NHANES III, respectively; both P values were not significant). The proportion of subjects with metabolic syndrome was not different in 20- to 29- or 30- to 39-yr-old women with PCOS defined as or or controls compared with NHANES III subjects regardless of BMI.

4 Welt et al. PCOS Phenotypes Defined by the Rotterdam Criteria J Clin Endocrinol Metab, December 2006, 91(12): Correlations LH correlated with testosterone best in the group (r 0.254, P ; r 0.444, P ; r 0.367, P 0.05 ), as did the androstenedione level (r 0.241, P ; r 0.585, P ; r 0.413, P 0.05 ). Insulin correlated with androstenedione in women with PCOS defined as and in controls (r and r 0.256, respectively; both P 0.05). Insulin correlated with DHEAS in and groups (r and r 0.318, respectively; both P 0.01) and inversely with SHBG in all groups (r 0.498, P ; r 0.395, P 0.01 ; r 0.344, P 0.01 control). There was no correlation between insulin and testosterone in any group. Ovarian volume correlated with testosterone (r 0.349; P 0.001) and androstenedione (r 0.145; P 0.05) in women with and with testosterone in control women (r 0.35, P 0.05). There was no correlation between ovarian volume and LH, FSH, or insulin. FIG. 1. BMI, testosterone (T), androstenedione (AD), and the LH/ FSH ratio in Boston and Icelandic women with PCOS in subsets identified by the Rotterdam workshop (1) and controls. Groups include women with PCOS identified by 1) IM and HA regardless of PCOM (), 2) HA and regular menses with PCOM (), 3) IM and PCOM (), and 4) controls (CTL). Data are shown in box and whisker plots. The box represents the 25 and 75% confidence limits, and the line represents the median. Bars indicate the SE, and circles are 5th and 95th percentiles. A bar with an asterisk indicates significant differences in pairwise comparisons between the groups indicated using the Fisher s LSD post hoc test. To convert to SI units, multiply testosterone by and androstenedione by Ultrasounds Subjects with PCOS defined as and had the largest ovarian volumes and the greatest number of follicles in a single plane on ultrasound, whereas the subjects with had intermediate values, and all groups had greater volume and follicle number than controls (Table 4). All subjects with and had PCOM, by definition, whereas the majority of subjects with had PCOM. Interestingly, slightly more than one half of the control women had PCOM. When analyzed separately, ovarian volume and follicle number were similar in all subgroups of Icelandic women with PCOS but slightly lower in women with compared with other PCOS groups in Boston (supplemental Table 4), although all PCOS subgroups were higher than controls. There were more ultrasounds demonstrating PCOM in the control group in Boston than in Iceland and more indeterminate ultrasounds in Iceland than in Boston, although the numbers were small. Discussion The four groups of PCOS empanelled by the Rotterdam criteria differ based on the presence or absence of IM, HA, and PCOM (1). Therefore, the relationships between the higher Ferriman-Gallwey scores, androgen levels, and presence of PCOM between the groups are expected based on the definitions used. What is surprising, however, is that the anthropomorphic features of PCOS, i.e. higher weight, BMI, and waist/hip ratios distinguish women with HA, particularly, from other women with PCOS. Thus, the most severe physical features and their metabolic consequences define the groups with HA, whereas the group with IM/ PCOM is distinguished from control women predominantly by menstrual cycle abnormalities and ovarian morphology. The current study and others suggest that weight and/or insulin resistance modify the severity of PCOS symptoms. Similar to the current data, BMI, insulin levels, and insulin resistance were higher in women with PCOS defined as compared with those with, although there was no non-ha group in these studies (8, 9, 27). Consistent with these findings, sisters of women with PCOS who have HA but regular cycles are thinner and have a tendency toward less insulin resistance than PCOS probands and sisters with PCOS, although ovarian morphologies were not examined and the Rotterdam criteria could not be applied (28, 29). Only one previous study has examined the relationship between weight, insulin resistance, and PCOS using all three of the Rotterdam criteria (7). This study demonstrated higher insulin levels in subjects with IM regardless of their androgenic status. Similarly, in the current study, which is substantial in size, subjects with PCOS and IM in the absence of HA () had similar high insulin levels when compared with women with, although the difference did not reach significance when compared with women with. Finally, women with had the highest rate of metabolic syndrome related to their increased weight. Taken together, weight and/or hyperinsulinemia may be the inciting factors in the IM and

5 4846 J Clin Endocrinol Metab, December 2006, 91(12): Welt et al. PCOS Phenotypes Defined by the Rotterdam Criteria TABLE 2. Comparison of the gonadotropins and androgens in PCOS subjects with,, and P value, ANCOVA LH (IU/liter) a b a,b c FSH (IU/liter) LH/FSH a b a,b c Testosterone (ng/dl) a b c c SHBG (nmol/liter) a b b b Free testosterone a b c c Adione (ng/ml) a a b b DHEAS ( g/dl) a a b b OH-progesterone (ng/ml) a b a,c c P values are indicated for the differences in groups as analyzed by ANCOVA, controlling for age and BMI. To convert to SI units, multiply testosterone by , androstenedione by 3.492, DHEAS by , and 17OH-progesterone by metabolic risk factors associated with PCOS, and IM may in turn exacerbate existing HA. The current study examines a large number of subjects recruited from two different countries and from medical vs. gynecology practices, along with common recruitment from general advertising. Therefore, bias based on recruitment source should be analyzed, particularly because endocrinologists and gynecologists can define PCOS differently (30), and we have previously identified differences in Boston and Icelandic PCOS subjects (17). The proportion of subjects recruited from practices vs. advertising was similar in Boston and Iceland, as was the proportion of PCOS subjects falling into the categories and. The only difference between Boston and Iceland was that subjects with were not recruited from medical practices in Boston (supplemental Table 5). When the Icelandic and Boston Caucasian groups were examined separately, the weight, BMI, and waist/hip ratio of women with in Iceland was similar to that in women with. The lower weight in the Boston HA/ PCOM subjects may reflect the fact that they presented as normal controls but were found to be hirsute with elevated androgen levels, perhaps because they were motivated to volunteer based on a concern over these abnormalities. As a possible consequence of the weight similarity in the HA/ PCOM and groups in Iceland, the Ferriman-Gallwey scores and testosterone levels were also similar, whereas in Boston the subjects exhibited a more intermediate phenotype. Furthermore, the insulin levels and HOMA in the group in Boston, but not Iceland, were similar to those in the group. An Australian study also demonstrated similar BMI and insulin levels in women with and (7). Taken together, separate analyses of Boston and Icelandic subjects also support the concept that weight and/or increased insulin resistance modify the PCOS phenotype, contributing to either HA or IM. Both LH and insulin stimulate androgen production in women with PCOS (31), and both have been suggested as the primary factors related to weight and anovulation that stimulate ovarian testosterone and androstenedione levels. The correlation between testosterone, androstenedione, and LH, TABLE 3. Comparison of the metabolic parameters in PCOS subjects with,, and P value, ANCOVA Glucose (mg/dl) HbA1C Insulin ( IU/ml) a b a,b b 0.02 HOMA a b a,b b 0.02 Cholesterol (mg/dl) HDL (mg/dl) Calculated LDL (mg/dl) Triglycerides (mg/dl) No. with IFG (%) 13 (4.5) 0 (0) 1 (2.9) 2 (3.2) 0.3 No. with type 2 diabetes mellitus (%) 1 (0.3) (1.6) 0.5 Metabolic syndrome Total no. (%) 64 (22.2) 8 (10.5) 2 (5.6) 7 (11.1) yr (%) 28 (17.3) 2 (5.7) 2 (15.4) 3 (11.5) yr (%) 31 (29.5) 5 (16.7) 0 (0) 4 (11.4) 0.01 BMI 30 kg/m yr (%) 24 (30.4) 1 (10.0) 2 (25.0) yr (%) 26 (39.4) 4 (30.8) 4 (36.4) 0.8 In the group with BMI greater than 30 kg/m 2,n 6. Therefore, there were too few subjects to make any statement about metabolic syndrome. P values are indicated for the differences in groups as analyzed by ANCOVA, controlling for age and BMI. To convert to SI units, multiply insulin by 7.175; glucose by ; cholesterol, high-density lipoprotein (HDL), and low-density lipoprotein (LDL) by ; and triglycerides by HbA1C, Glycosylated hemoglobin; IFG, impaired fasting glucose.

6 Welt et al. PCOS Phenotypes Defined by the Rotterdam Criteria J Clin Endocrinol Metab, December 2006, 91(12): TABLE 4. Ultrasound findings in PCOS subjects with,, and Ovarian volume (cm 3 ) a b a,b c Maximum follicle number a b a,b c PCOM, n (%) 254 (97.3) 76 (100) 34 (100) 24 (52.2) Normal morphology, n (%) 7 (2.7) 0 (0) 0 (0) 22 (47.9) Indeterminate morphology, n (%) 25 (8.7) 0 (0) 0 (0) 12 (20.7) 0.05 P values are indicated for the differences in groups as analyzed by a one-way ANOVA. P value but not insulin, as in some previous studies (32 34), suggests that LH is the major factor driving ovarian androgen production in women with PCOS defined by. However, women with but no HA exhibited LH levels and LH/FSH ratios indistinguishable from those of the women with. Furthermore, insulin correlated with androstenedione and DHEAS in women with, suggesting a role for insulin in this subgroup. Taken together, these findings provide support for the concept that both LH and insulin drive hyperandrogenemia in women with PCOS. The findings on ovarian ultrasound most closely paralleled those of LH, as seen previously (32, 35), with the highest ovarian volume and follicle number in the group, intermediate levels in the group, and lowest levels in the group. Although there was a correlation between testosterone and androstenedione and ovarian volume in the group, the order of ovarian volume and follicle number did not parallel the dose-response levels of testosterone in which the highest levels were demonstrated in the group and the lowest in. There was also no relationship between ovarian volume and FSH or insulin. Thus, it is not clear whether the ovarian volume is a primary abnormality or is exacerbated by persistent anovulation or another related hormonal factor. The group has been the most controversial of the four groups identified in the Rotterdam workshop (3, 4, 36) because HA has been considered a defining feature of PCOS (2). The current data suggest that the PCOS group defined by is distinguished from control women by IM and the greater ovarian volume and follicle number on ultrasound, expanding previous findings (8). The LH/FSH ratio, a test that has specificity for the diagnosis of PCOS (37), was also high in the group and was similar to the ratio in the group. Of note, to make PCOS subcategories consistent with the recommendations of the Rotterdam workshop in the current study, acne alone was not used as an indicator of HA because it is listed only as a potential marker in the Rotterdam criteria (1), and it can occur in the absence of systemic HA (38). Thus, within the definitional limits of this study, women with have phenotypic features similar to those in other groups with PCOS despite a weight similar to controls, suggesting they are the least affected PCOS subjects. These data provide evidence to support three subsets of subjects with PCOS defined by the Rotterdam criteria for potential use in genetics studies. Subjects with PCOS defined by HA are the most severely affected, with the highest androgen levels. They also have the highest BMI of the Rotterdam subgroups, suggesting that weight gain or intrinsic insulin resistance may exacerbate symptoms of PCOS or cause cycle irregularity in women with HA at baseline. Women with are clearly distinguished from controls by their ovarian volume, follicle number, and IM. Longitudinal studies using well-defined phenotypes in these three groups will be needed to determine whether the differences in phenotype result in differences in long-term consequences such as infertility and cardiovascular disease. Acknowledgments We thank Patrick Sluss, Ph.D., and Joseph Moy for their assay expertise. We also thank Kristleifur Kristjansson, M.D., Struan Grant, Ph.D., Larus Gudmundsson, and Gyda Bjornsdottir for their expert assistance with the study design, data management, and retrieval. Received June 21, Accepted September 19, Address all correspondence and requests for reprints to: Corrine Welt, Reproductive Endocrine Unit, BHX 511, Massachusetts General Hospital, 55 Fruit Street, Boston, Massachusetts cwelt@ partners.org. This work was supported by the National Institutes of Health Grant U01 HD 4417 and National Center for Research Resources General Clinical Research Centers Program Grant M01-RR Disclosure summary: The authors have nothing to disclose. References 1. The Rotterdam ESHRE/ASRM-Sponsored PCOS consensus workshop group 2004 Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Hum Reprod 19: Zawadzki JK, Dunaif A 1992 Diagnostic criteria for polycystic ovary syndrome: towards a rational approach. In: Dunaif A, Givens JR, Haseltine FP, Merriam GR, eds. Polycystic ovary syndrome. Boston: Blackwell Scientific; Azziz R 2006 Controversy in clinical endocrinology: diagnosis of polycystic ovarian syndrome: the Rotterdam criteria are premature. J Clin Endocrinol Metab 91: Franks S 2006 Controversy in clinical endocrinology: diagnosis of polycystic ovarian syndrome: in defense of the Rotterdam criteria. J Clin Endocrinol Metab 91: Vink JM, Sadrzadeh S, Lambalk CB, Boomsma DI 2005 Heritability of polycystic ovary syndrome (PCOS) in a Dutch twin-family study. J Clin Endocrinol Metab 91: Grant SF, Thorleifsson G, Reynisdottir I, Benediktsson R, Manolescu A, Sainz J, Helgason A, Stefansson H, Emilsson V, Helgadottir A, Styrkarsdottir U, Magnusson KP, Walters GB, Palsdottir E, Jonsdottir T, Gudmundsdottir T, Gylfason A, Saemundsdottir J, Wilensky RL, Reilly MP, Rader DJ, Bagger Y, Christiansen C, Gudnason V, Sigurdsson G, Thorsteinsdottir U, Gulcher JR, Kong A, Stefansson K 2006 Variant of transcription factor 7-like 2 (TCF7L2) gene confers risk of type 2 diabetes. Nat Genet 38: Norman RJ, Hague WM, Masters SC, Wang XJ 1995 Subjects with polycystic ovaries without hyperandrogenaemia exhibit similar disturbances in insulin and lipid profiles as those with polycystic ovary syndrome. Hum Reprod 10: Carmina E, Chu MC, Longo RA, Rini GB, Lobo RA 2005 Phenotypic variation in hyperandrogenic women influences the findings of abnormal metabolic and cardiovascular risk parameters. J Clin Endocrinol Metab 90:

7 4848 J Clin Endocrinol Metab, December 2006, 91(12): Welt et al. PCOS Phenotypes Defined by the Rotterdam Criteria 9. Carmina E, Rosato F, Janni A, Rizzo M, Longo RA 2006 Extensive clinical experience: relative prevalence of different androgen excess disorders in 950 women referred because of clinical hyperandrogenism. J Clin Endocrinol Metab 91: Treloar AE, Boynton RE, Behn BG, Brown BW 1967 Variation of the human menstrual cycle through reproductive life. Int J Fertil 12: Jonard S, Robert Y, Cortet-Rudelli C, Pigny P, Decanter C, Dewailly D 2003 Ultrasound examination of polycystic ovaries: is it worth counting the follicles? Hum Reprod 18: Adams J, Franks S, Polson DW, Mason HD, Abdulwahid N, Tucker M, Morris DV, Price J, Jacobs HS 1985 Multifollicular ovaries: clinical and endocrine features and response to pulsatile gonadotropin releasing hormone. Lancet 2: Holte J, Bergh T, Berne C, Wide L, Lithell H 1995 Restored insulin sensitivity but persistently increased early insulin secretion after weight loss in obese women with polycystic ovary syndrome. J Clin Endocrinol Metab 80: Taylor AE, McCourt B, Martin KA, Anderson EJ, Adams JM, Schoenfeld DA, Hall JE 1997 Determinants of abnormal gonadotropin secretion in clinically defined women with polycystic ovary syndrome. J Clin Endocrinol Metab 82: Azziz R, Hincapie LA, Knochenhauer ES, Dewailly D, Fox L, Boots LR 1999 Screening for 21-hydroxylase-deficient nonclassic adrenal hyperplasia among hyperandrogenic women: a prospective study. Fertil Steril 72: Ferriman D, Gallwey JD 1961 Clinical assessment of body hair growth in women. J Clin Endocrinol Metab 21: Welt CK, Arason G, Gudmundsson JA, Adams JM, Palsdottir H, Gudlaugsdottir G, Ingadottir G, Crowley Jr WF, Defining constant versus variable phenotypic features of women with polycystic ovary syndrome using different ethnic groups and populations. J Clin Endocrinol Metab, in press 18. Welt CK, Adams JM, Sluss PM, Hall JE 1999 Inhibin A and inhibin B responses to gonadotropin withdrawal depends on stage of follicle development. J Clin Endocrinol Metab 84: Adams JM, Taylor AE, Crowley Jr WF, Hall JE 2004 Polycystic ovarian morphology with regular ovulatory cycles: insights into the pathophysiology of polycystic ovarian syndrome. J Clin Endocrinol Metab 89: Ibanez L, Hall JE, Potau H, Carrascosa A, Prat N, Taylor AE 1996 Ovarian 17-hydroxyprogesterone hyperresponsiveness to GnRH agonist challenge in women with polycystic ovary syndrome is not mediated by gonadotropin hypersecretion : evidence from GnRH agonist and human chorionic gonadotropin stimulation testing. J Clin Endocrinol Metab 81: Sample WF, Lippe BM, Gyepes MT 1977 Gray-scale ultrasonography of the normal female pelvis. Radiology 125: Vermeulen A, Verdonck L, Kaufman JM 1999 A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 84: Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC 1985 Homeostasis model assessment: insulin resistance and -cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 28: Ford ES, Giles WH, Dietz WH 2002 Prevalence of the metabolic syndrome among US adults: findings from the third National Health and Nutrition Examination Survey. JAMA 287: Elting MW, Korsen TJ, Schoemaker J 2001 Obesity, rather than menstrual cycle pattern or follicle cohort size, determines hyperinsulinaemia, dyslipidaemia and hypertension in ageing women with polycystic ovary syndrome. Clin Endocrinol (Oxf) 55: Bili H, Laven J, Imani B, Eijkemans MJ, Fauser BC 2001 Age-related differences in features associated with polycystic ovary syndrome in normogonadotrophic oligo-amenorrhoeic infertile women of reproductive years. Eur J Endocrinol 145: Robinson S, Kiddy D, Gelding SV, Willis D, Niththyananthan R, Bush A, Johnston DG, Franks S 1993 The relationship of insulin insensitivity to menstrual pattern in women with hyperandrogenism and polycystic ovaries. Clin Endocrinol (Oxf) 39: Legro RS, Bentley-Lewis R, Driscoll D, Wang SC, Dunaif A 2002 Insulin resistance in the sisters of women with polycystic ovary syndrome: association with hyperandrogenemia rather than menstrual irregularity. J Clin Endocrinol Metab 87: Sam S, Legro RS, Bentley-Lewis R, Dunaif A 2005 Dyslipidemia and metabolic syndrome in the sisters of women with polycystic ovary syndrome. J Clin Endocrinol Metab 90: Cussons AJ, Stuckey BG, Walsh JP, Burke V, Norman RJ 2005 Polycystic ovarian syndrome: marked differences between endocrinologists and gynaecologists in diagnosis and management. Clin Endocrinol (Oxf) 62: Barbieri RL, Makris A, Ryan KJ 1984 Insulin stimulates androgen accumulation in incubations of human ovarian stroma and theca. Obstet Gynecol 64: Pache TD, de Jong FH, Hop WC, Fauser BC 1993 Association between ovarian changes assessed by transvaginal sonography and clinical and endocrine signs of the polycystic ovary syndrome. Fertil Steril 59: Sharp PS, Kiddy DS, Reed MJ, Anyaoku V, Johnston DG, Franks S 1991 Correlation of plasma insulin and insulin-like growth factor-i with indices of androgen transport and metabolism in women with polycystic ovary syndrome. Clin Endocrinol (Oxf) 35: Franks S 1989 Polycystic ovary syndrome: a changing perspective. Clin Endocrinol (Oxf) 31: Conway GS, Honour JW, Jacobs HS 1989 Heterogeneity of the polycystic ovary syndrome: clinical, endocrine and ultrasound features in 556 patients. Clin Endocrinol (Oxf) 30: Azziz R 2005 Diagnostic criteria for polycystic ovary syndrome: a reappraisal. Fertil Steril 83: Fox R, Corrigan E, Thomas PA, Hull MG 1991 The diagnosis of polycystic ovaries in women with oligo-amenorrhoea: predictive power of endocrine tests. Clin Endocrinol (Oxf) 34: Lookingbill DP, Horton R, Demers LM, Egan N, Marks Jr JG, Santen RJ 1985 Tissue production of androgens in women with acne. J Am Acad Dermatol 12: JCEM is published monthly by The Endocrine Society ( the foremost professional society serving the endocrine community.