DEFINING THE POLYCYSTIC ovary syndrome (PCOS)

Transcription

1 X/06/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 91(10): Printed in U.S.A. Copyright 2006 by The Endocrine Society doi: /jc Oligoanovulation with Polycystic Ovaries But Not Overt Hyperandrogenism Didier Dewailly, Sophie Catteau-Jonard, Anne-Céline Reyss, Maryse Leroy, and Pascal Pigny Department of Endocrine Gynecology and Reproductive Medicine (D.D., S.C.-J., A.-C.R., M.L.), Hôpital Jeanne de Flandre, and Faculty of Medicine of Lille, Université de Lille II and Laboratory of Endocrinology (P.P.), Parc Eurasanté, Centre Hospitalier Régional Universitaire, Lille, France Objectives: By requiring a minimum of two of three items [hyperandrogenism (HA), oligoanovulation (OA), and polycystic ovaries (PCO) at ultrasound], the Rotterdam definition recognizes four PCO syndrome (PCOS) phenotypes: HA OA PCO (full-blown syndrome), HA OA (former National Institutes of Health definition), HA PCO (ovulatory PCOS), and OA PCO. However, the latter phenotype is controversial, and it is not known to what extent it shares similarities with the others. Design: The study was a comparative analysis of hormonal, metabolic, and ultrasound parameters obtained from patients and controls that were consecutively included in a database. First Published Online July 18, 2006 Abbreviations: A, Androstenedione; BMI, body mass index; FHA, functional hypothalamic anovulation; FN, follicle number; HA, hyperandrogenism; I, fasting serum insulin; OA, oligoanovulation; OvA, ovarian area; PCO, polycystic ovaries; PCOS, polycystic ovary syndrome; T, testosterone; U/S, ultrasound; WC, waist circumference. JCEM is published monthly by The Endocrine Society ( endo-society.org), the foremost professional society serving the endocrine community. Patients and Methods: Sixty-six patients having OA PCO without hirsutism or elevated serum androstenedione and testosterone levels were compared with 118 normally cycling nonhyperandrogenic age-matched women without PCO (controls). These patients (phenotype D) were also compared with patients with HA OA PCO (phenotype A, n 246), HA OA (phenotype B, n 27), and HA PCO (phenotype C, n 67). Results: Patients with phenotype D had higher mean values of waist circumference and higher mean levels of serum testosterone, androstenedione, and LH than controls. Conversely, they had lower mean serum levels of FSH and SHBG (P 0.05 for each parameter). Variance analysis disclosed significant group effects between the different patients phenotypes for all parameters, except age, BMI, and FSH. After multiple comparisons with post hoc analysis, phenotype D had milder endocrine and metabolic abnormalities than phenotype A, although it did not differ from phenotype C, except for androgen data, by definition. Phenotypes A and B were statistically similar, except for the ultrasound data, by definition. Conclusion: Oligoanovulatory patients with PCO but without HA have mild endocrine and metabolic features of PCOS. (J Clin Endocrinol Metab 91: , 2006) DEFINING THE POLYCYSTIC ovary syndrome (PCOS) has generated a lot of original studies and reviews, and it keeps stimulating passionate debates, particularly intense after an ASRM/ESHRE-sponsored consensus conference was held in Rotterdam, The Netherlands, in 2003 (1, 2). By adding the ultrasound (U/S) criteria for polycystic ovaries (PCO) to the items of the former so-called National Institutes of Health (NIH) definition (3), i.e. hyperandrogenism (HA) and oligoanovulation (OA), and by requiring a minimum of two items of three, the so-called Rotterdam definition has extended the frame of PCOS. By giving more flexibility, this definition recognizes four PCOS phenotypes: HA OA PCO (full-blown syndrome), HA OA (former NIH definition), HA PCO (so-called ovulatory PCOS), and OA PCO. The absence of a label for the last phenotype is symptomatic of the perplexity that it induces. Some authors (4, 5) question whether some forms of PCOS can present as a nonhyperandrogenic disorder because HA is considered the cornerstone of the syndrome (6). Therefore, in their opinion, the absence of HA precludes the diagnosis of PCOS. In addition, the U/S criteria for PCO are said to be nonspecific enough because they were seemingly observed in other situations than PCOS, in particular functional hypothalamic anovulation (FHA). For others (7, 8), the risk of mistakenly diagnosing PCOS in non-pcos oligoanovulatory disease is minimal provided that exclusion criteria are properly checked before applying the Rotterdam definition. Studies about the significance of PCO at U/S in women not having the classical symptoms of PCOS have yielded conflicting results about their endocrine and metabolic profiles (9 12). This is mainly explained by the different modes of patient recruitment, which were normal volunteering (10), self-reported symptoms in cohort follow-up (11), or clinicbased patients (9, 12). So far, however, no study has specifically focused on the OA PCO phenotype as strictly defined by the Rotterdam criteria to verify whether it shares endocrine and metabolic similarities with the other PCOS phenotypes. The aim of the present study was therefore to isolate this phenotype from a cohort of patients consecutively referred to an endocrine and infertility unit for HA and/or OA and to compare this population to ovulatory normo-androgenic women without PCO (controls) and to patients having the other PCOS phenotypes. Patients and Methods Control and patient populations This study was approved by the Institutional Review Board of the Lille University Hospital. Data were obtained from a database including clinical, hormonal, and U/S features that were consecutively and pro- 3922

2 Dewailly et al. Nonhyperandrogenic PCOS J Clin Endocrinol Metab, October 2006, 91(10): TABLE 1. Definition and frequencies of the four patients phenotypes HA OA PCO at U/S Phenotype A (n 246; 60.6%) Yes Yes Yes Phenotype B (n 27; 6.7%) Yes Yes No Phenotype C (n 67; 16.5%) Yes No Yes Phenotype D (n 66; 16.3%) No Yes Yes Hormonal immunoassays Blood sampling was performed in the early follicular phase (i.e. between d 2 and 5 after the last menstrual period) both in patients and control women, as previously described (14). In oligo- or amenorrheic patients, the last menstrual period was either spontaneous or induced by the administration of didrogesterone (10 mg/d for 7 d). SHBG, A, T, LH, FSH and fasting serum insulin (I) levels were measured by immunoassays as described previously (14). spectively recorded between 2000 and 2005 in patients referred to our department, after obtaining their informed consent. The following exclusion criteria were applied to select both patients and controls for this study: age less than 18 yr, presence of premature ovarian failure (FSH 12 IU/liter), hyperprolactinemia (prolactin 20 ng/ml), or nonclassic 21-hydroxylase deficiency (basal 17-hydroxyprogesterone 2 ng/ml and/or post-acth stimulated value 12 ng/ml) (13). Controls. The control population consisted of 118 healthy women whose age ( yr.) was matched to patients. Their mean body mass index (BMI) ranged from kg/m 2. They were referred for in vitro fertilization because of tubal and/or male infertility. Exclusion criteria were a history of menstrual disturbances (i.e. cycle length either 25dor 35 d), hirsutism (i.e. modified Ferriman-Gallwey score 6), abnormal serum level of androgens [i.e. 95th percentile of our previous control group (14), which was 0.6 ng/ml for serum testosterone (T) and 2.2 ng/ml for serum androstenedione (A)], PCO at U/S (see below), and hormonal treatment during the 3 months before the study. Patients. Data from 406 women aged yr were used for this study. BMI ranged from kg/m 2. All patients were referred to our department for HA and/or OA. At least two of the following three items were required for patients inclusion in this study, according to the Rotterdam classification (1, 2): 1) HA, defined clinically as the presence of hirsutism (modified Ferriman-Gallwey score 6) and/or biologically by serum testosterone (T) and/or androstenedione (A) level greater than 0.6 ng/ml and 2.2 ng/ml, respectively; 2) OA, defined as the presence of oligomenorrhea (i.e. less than eight menstrual bleedings during the last year) or amenorrhea (i.e. no menstrual bleeding during the last 3 months); 3) presence of PCO at U/S, according to the Rotterdam criteria (15), except that we used an ovarian area (OvA) of more than 6 cm 2 instead of an ovarian volume of more than 10 cm 3 because only OvA was recorded in our database until late Amenorrheic patients having a history of food restriction, intensive exercise, no progestin-induced withdrawal bleeding, and/or a basal serum LH less than 1 IU/liter were considered as primarily having FHA. They were therefore excluded, even though PCO were seen at U/S in some of them. No patient took hormonal treatment during the 3 months before the study, except didrogesterone (see above) for some of them. Pelvic U/S examination In every patient and control, U/S examination was performed the same day as blood sampling, between cycle d 2 and 5, with a 7-MHz transvaginal transducer (Logic 400; General Electric, Milwaukee, WI). U/S measurements were taken in real time, according to a standardized protocol as previously reported (16). Patients in whom transvaginal U/S was inappropriate (virgin or refusing patients) were excluded from the analysis as well as those in whom the sum of follicles from both ovaries was not at least five and/or in whom the ovarian area was below the lower normal limit, i.e. 2.5 cm 2. Any patient or control with at least one follicle with a diameter greater than 9 mm at U/S or a serum estradiol level above 80 pg/ml was excluded from the study so as not to confound the hormonal data with the presence of a dominant follicle. Statistical methods A P value 0.05 was considered significant. The 2 and Student s t tests were used to compare two independent groups, where appropriate. All variables that had a log-normal distribution were log-transformed before statistical calculations. For comparison of three or more groups, variance analysis (ANOVA) was first performed to search for a group effect. When this effect was significant, differences between subgroups were searched through two by two comparisons with post hoc Bonferroni s correction for multiple comparisons. All statistic procedures were run on SPSS 11.5 (SPSS Inc., Chicago, IL). Results Distribution of patients according to the phenotypic classification Table 1 shows that OA with PCO but without HA (phenotype D) accounted for only 16% of the patients, a proportion similar to that of phenotype C (HA PCO). About two thirds of our patients met the former NIH definition (i.e. OA and HA, phenotypes A B). Twenty-seven (10%) of these 273 patients had no evidence of PCO (phenotype B) (Table 2). TABLE 2. Characteristics of patients and controls Phenotypes A(n 246) B (n 27) C (n 67) D (n 66) Controls (n 118) Age (yr) 28.0 ( ) 28 ( ) 28.0 ( ) 28.0 ( ) 28.0 ( ) BMI (kg/m 2 ) 27.0 ( ) 25.7 ( ) 25.2 ( ) 23.5 ( ) 23.0 ( ) I (miu/liter) 6.2 ( ) 4.7 ( ) 5.0 ( ) 3.8 ( ) b 3.4 ( ) WC (cm) 86.0 ( ) 81 ( ) 82.5 ( ) 76.0 ( ) a,b 73.0 ( ) SHBG (mmol/liter) 30 ( ) 28 ( ) 37.3 ( ) 47.3 ( ) a,b 51 ( ) T (ng/ml) 0.50 ( ) 0.54 ( ) 0.41 ( ) 0.31 ( ) a,b,c,d 0.25 ( ) A (ng/ml) 2.69 ( ) 2.63 ( ) 2.47 ( ) 1.66 ( ) a,b,c,d 1.40 ( ) LH (IU/liter) 6.8 ( ) 6.0 ( ) 4.3 ( ) 4.4 ( ) a,b 3.8 ( ) FSH (IU/liter) 5.2 ( ) 5.2 ( ) 5.2 ( ) 5.1 ( ) a 6.1 ( ) 2 9 mm FN 20.0 ( ) 9.75 ( ) 17.5 ( ) ( ) a,c 7.0 ( ) OvA (cm 2 ) 6.3 ( ) 4.7 ( ) 5.75 ( ) 5.6 ( ) a,b,c 3.9 ( ) Values represent median (5 95th percentile). To convert T to nmol/liter, multiply by 3.467; to convert A to nmol/liter, multiply by a Phenotype D significantly different from controls (P 0.05). b, c, d Phenotype D significantly different from phenotypes A, B, and C, respectively (P 0.05 after post hoc Bonferroni s correction for multiple comparisons).

3 3924 J Clin Endocrinol Metab, October 2006, 91(10): Dewailly et al. Nonhyperandrogenic PCOS Comparison between phenotype D and controls All the recorded parameters except age (P 0.74), BMI (P 0.068), and I (P 0.13) were significantly different between patients with phenotype D and controls (Table 2). Notably, the serum T and A mean levels were significantly higher in phenotype D than in controls, although all individual values were within the normal range, by definition. Comparison between phenotype D and other phenotypes When applied to the whole patients population, variance analysis (ANOVA) disclosed a significant group effect between phenotypes A, B, C, and D for all parameters except age (P 0.82), BMI (P 0.22), and FSH (P 0.82). Figures 1 and 2 show a trend from phenotype A to phenotype D toward progressively lower mean values of waist circumference (WC) (P 0.03); toward lower mean levels of serum T(P ), A (P ), LH (P ), and I (P 0.008); and conversely toward progressively higher mean serum levels of SHBG (P 0.003). The ANOVA was also significant for the follicle number (FN) (P ) and the OvA (P ). After two by two comparisons with post hoc Bonferroni s correction for multiple comparisons, no significant difference was found between phenotypes C and D, except for androgen data, by definition (Table 2). On the other hand, beside androgens, phenotype D had significantly lower mean WC, lower mean LH and I levels, and higher mean SHBG level than phenotype A as well as significantly smaller mean OvA but similar FN (Table 2). Phenotypes A and B were statistically similar, except for U/S data, by definition (Table 2). Phenotype C had significantly lower mean LH and T levels as well as mean OvA than phenotype A (Table 2). FIG. 1. Box-and-whisker plots showing the distribution of individual values for serum T (A), A (B), LH (C), and FSH (D) in controls and in patients with phenotypes A D. Horizontal small bars represent the 5 95th percentile range, and the boxes indicate the 25 75th percentile range. The horizontal line in each box corresponds to the median. E, Values beyond the 95th percentile; *, values beyond the 99th percentile. See Table 2 for comparisons between groups. To convert T to nmol/liter, multiply by To convert A to nmol/liter, multiply by

4 Dewailly et al. Nonhyperandrogenic PCOS J Clin Endocrinol Metab, October 2006, 91(10): FIG. 2. Box-and-whisker plots showing the distribution of individual values for BMI (A), waist circumference (B), and SHBG (C) in controls and in patients with phenotypes A D. Horizontal small bars represent the 5 95th percentile range, and the boxes indicate the 25 75th percentile range. The horizontal line in each box corresponds to the median. E, Values beyond the 95th percentile; *, values beyond the 99th percentile. See Table 2 for comparisons between groups. To convert T to nmol/liter, multiply by To convert A to nmol/liter, multiply by Discussion In this population carefully selected by stringent exclusion criteria, two thirds of the patients met the former NIH definition for PCOS (i.e. HA and OA, phenotypes A and B), and most of them (90%) had PCO at U/S (phenotype A). Presumably, phenotype B corresponded to false-negative results of U/S (i.e. FN 12 and OvA 6.0 cm 2 ). Indeed, except for U/S features, both groups were statistically similar. Such a low false-negative rate (10%) of U/S in the group of patients having a full-blown PCOS confirms the validity of the U/S criteria that we used. They were those proposed by the Rotterdam consensus conference (15), except that we used the OvA instead of the ovarian volume (see Patients and Methods). According to our recent study (17), an OvA of 6.0 cm 2 corresponds to an ovarian volume of 10 ml, and this threshold has similar sensitivity and specificity to an ovarian volume of more than 10 ml to define PCO. Providing the presence of PCO, patients presenting with only either HA or OA have other PCOS phenotypes (phenotypes C and D, respectively), according to the Rotterdam classification (1, 2). Each phenotype accounted for only one sixth of our patients population. However, such prevalence data have limited value because they may vary substantially depending on whether the clinical setting is oriented to endocrinology or to infertility. Because of the dual orientation of our unit, we had the opportunity to recruit equally both phenotypes. Phenotype C, so-called ovulatory PCOS in the literature, does not suffer from controversy anymore (18). As shown previously by others (19, 20), our patients with this phenotype had indices of both gonadotropin dysregulation and insulin resistance. However, their mean LH and T levels as well as their mean OvA were lower than in the full-blown PCOS, in agreement with others findings (19, 20). Our data indicate that patients with nonhyperandrogenic OA (phenotype D) had in fact slightly but significantly higher mean androgen levels than controls, although by definition, all individual values were within the normal range. This raises the question about the validity of using an upper normal threshold for the androgen data as a yes or no answer to the question of whether this patient is normo- or hyperandrogenic? This holds true for the Ferriman-Gallwey score, which suffers from a high subjectivity, as well as for the serum T and A assays, which have weak sensitivities, lower than those of U/S criteria for PCO (21). For this reason, we think that the absence of overt HA might simply represent a false-negative finding in many of our patients with phenotype D and that it is not sufficient by itself to preclude the diagnosis of PCOS. Moreover, in support of our opinion that the absence of overt HA cannot be exclusive, our patients with phenotype D had respectively higher and lower LH and FSH mean levels than controls, a typical figure of the gonadotropin derangement of PCOS. Notably, the degree of FSH suppression was similar to other phenotypes. In addition, their mean WC was higher, a parameter that has been recently shown to be one of the most sensitive markers of the metabolic syndrome in patients with typical PCOS (22). Accordingly, their mean SHBG level was significantly lower than controls, a finding being considered as a marker of hyperinsulinism (23). On the other hand, their mean I level was not different from controls, whereas their mean WC and I and SHBG levels were significantly less than in patients with phenotype A. These data fit with previous reports indicating a trend toward higher insulin levels along with a higher degree of symptoms in women with PCO recruited either as normal volunteers (10) or via hyperandrogenic symptoms (24). Nonetheless, the presence of OA in the face of minimal insulin resistance implies that the ovulation disorder of PCOS is not exclusively driven by insulin resistance and that other still unknown factors are likely to be involved. That OA patients with PCO but without overt HA constitute a PCOS phenotype is a highly disputed issue. Some authors argue that such patients cannot be differentiated

5 3926 J Clin Endocrinol Metab, October 2006, 91(10): Dewailly et al. Nonhyperandrogenic PCOS from patients suffering from other causes of OA because multifollicular ovaries or even PCO have been described in patients with FHA (25, 26) or with bulimia nervosa (27). However, the finding of genuine PCO in such patients is questionable. In the above mentioned studies that were published before the Rotterdam conference, no consensual threshold for the number of small (2 9 mm) follicles was used to define the follicle excess. Such data should be revisited now that this threshold is consensually set at 12 follicles per whole ovary, because it yielded a very low false-positive rate (1%) in our previously reported experience comparing patients with full-blown PCOS with controls (16). Nevertheless, if we leave aside the unsolved issue about U/S, we admit that an artificial PCOS phenotype could be built by applying too carelessly the Rotterdam definition to normo-androgenic women with OA and PCO. There is therefore some fear that the Rotterdam classification leads to an overspill of the PCOS population, and this raises medical, psychological, and economic concerns (28). However, in the setting of an infertility unit, the risk exists mainly for women with FHA. In this situation, amenorrhea is not reversible after progestin withdrawal, and although BMI can be normal, patients have a history of food restriction, with indices of negative energy balance and LH deficiency (29, 30). Such features were not observed in our women with phenotype D who were slightly more overweight than controls and who had a mildly elevated mean LH level. Therefore, on condition that FHA is carefully excluded before applying the Rotterdam criteria, as we did in the present study, we and others (7) think that the risk is low for the Rotterdam definition to include erroneously non-pcos ovulatory disorders. Lastly, one should also consider the possibility that both FHA and PCOS could coexist in the same patient. It has been recently reported that some women with FHA caused by anorexia nervosa had genuine PCO at U/S that were previously associated with hyperandrogenic symptoms before the patients became amenorrheic and turned off their LH secretion because of food restriction (31). In conclusion, our data indicate that OA with PCO but without overt HA constitutes a phenotype that has subtle PCOS endocrine and metabolic features and is presumably the mildest form of PCOS, with minimal insulin resistance. Whether this phenotype shares the same long-term risks as the classical PCOS remains to be elucidated to inform the patients appropriately. Acknowledgments We thank the staffs of the Laboratoire de Biochimie et Hormonologie, Parc Eurasanté, and of the Service de Radiologie, Hôpital Jeanne de Flandre, Centre Hospitalier Régional Universitaire de Lille, for their excellent technical help. We also thank Ms. Sophie Delva and Céline Vandaele for collecting the clinical data and Mrs. Lydie Lombardo and Sylvie Vanoverschelde for collecting the blood samples. We thank Dr. Alain Duhamel, Centre d études et de recherches en informatique médicale, University of Lille 2, for his help with the statistics. Received May 16, Accepted July 12, Address all correspondence and requests for reprints to: Didier Dewailly, Department of Endocrine Gynecology and Reproductive Medicine, Hôpital Jeanne de Flandre, Centre Hospitalier Régional Universitaire, Lille, France. ddewailly@chru-lille.fr. This work was supported by the Délégation à la Recherche du Centre Hospitalier Universitaire de Lille (France) and the Direction Régionale des Etudes Doctorales, Université de Lille II, France. Disclosure statement: The authors have nothing to disclose. References 1. 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. Fertil Steril 81: 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: Azziz R 2005 Diagnostic criteria for polycystic ovary syndrome: a reappraisal. Fertil Steril 83: Azziz R 2003 Androgen excess is the key element in polycystic ovary syndrome. Fertil Steril 80: Franks S 2006 Controversy in clinical endocrinology: diagnosis of polycystic ovarian syndrome: in defense of the Rotterdam criteria. J Clin Endocrinol Metab 91: Dewailly D 2006 How many of the items in the polycystic ovary syndrome can be validated statistically? Fertil Steril 85: 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: Michelmore K, Ong K, Mason S, Bennett S, Perry L, Vessey M, Balen A, Dunger D 2001 Clinical features in women with polycystic ovaries: relationships to insulin sensitivity, insulin gene VNTR and birth weight. Clin Endocrinol (Oxf) 55: Taponen S, Ahonkallio S, Martikainen H, Koivunen R, Ruokonen A, Sovio U, Hartikainen AL, Pouta A, Laitinen J, King V, Franks S, McCarthy MI, Jarvelin MR 2004 Prevalence of polycystic ovaries in women with self-reported symptoms of oligomenorrhoea and/or hirsutism: Northern Finland Birth Cohort 1966 Study. Hum Reprod 19: Ng EH, Chan CC, Ho PC 2006 Are there differences in ultrasound parameters between Chinese women with polycystic ovaries only and with polycystic ovary syndrome? Eur J Obstet Gynecol Reprod Biol 125: Azziz R, Dewailly D, Owerbach D 1994 Nonclassic adrenal hyperplasia: current concepts. J Clin Endocrinol Metab 78: Pigny P, Merlen E, Robert Y, Cortet-Rudelli C, Decanter C, Jonard S, Dewailly D 2003 Elevated serum level of anti-mullerian hormone in patients with polycystic ovary syndrome: relationship to the ovarian follicle excess and to the follicular arrest. J Clin Endocrinol Metab 88: Balen AH, Laven JS, Tan SL, Dewailly D 2003 Ultrasound assessment of the polycystic ovary: international consensus definitions. Hum Reprod Update 9: 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: Jonard S, Robert Y, Dewailly D 2005 Revisiting the ovarian volume as a diagnostic criterion for polycystic ovaries. Hum Reprod 20: Carmina E, Lobo RA 1999 Do hyperandrogenic women with normal menses have polycystic ovary syndrome? Fertil Steril 71: Carmina E, Lobo RA 2001 Polycystic ovaries in hirsute women with normal menses. Am J Med 111: 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: van Santbrink EJ, Hop WC, Fauser BC 1997 Classification of normogonadotropic infertility: polycystic ovaries diagnosed by ultrasound versus endocrine characteristics of polycystic ovary syndrome. Fertil Steril 67: Ehrmann DA, Liljenquist DR, Kasza K, Azziz R, Legro RS, Ghazzi MN 2006 PCOS/Troglitazone Study Group. Prevalence and predictors of the metabolic syndrome in women with polycystic ovary syndrome. J Clin Endocrinol Metab 91: Pugeat M, Crave JC, Elmidani M, Nicolas MH, Garoscio-Cholet M, Lejeune H, Dechaud H, Tourniaire J 1991 Pathophysiology of sex hormone binding globulin (SHBG): relation to insulin. J Steroid Biochem Mol Biol 40: 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:

6 Dewailly et al. Nonhyperandrogenic PCOS J Clin Endocrinol Metab, October 2006, 91(10): Adams J, Franks S, Polson DW, Mason HD, Abdulwahid N, Tucker M 1985 Multifollicular ovaries: clinical and endocrine features and response to pulsatile gonadotropin releasing hormone. Lancet 2: Futterweit W, Yeh HC, Mechanick JI 1988 Ultrasonographic study of ovaries of 19 women with weight loss-related hypothalamic oligo-amenorrhea. Biomed Pharmacother 42: Morgan JF, McCluskey SE, Brunton JN, Hubert Lacey J 2002 Polycystic ovarian morphology and bulimia nervosa: a 9-year follow-up study. Fertil Steril 77: Azziz R, Marin C, Hoq L, Badamgarav E, Song P 2005 Health care-related economic burden of the polycystic ovary syndrome during the reproductive life span. J Clin Endocrinol Metab 90: Warren MP 1996 Evaluation of secondary amenorrhea. J Clin Endocrinol Metab 81: Couzinet B, Young J, Brailly S, Le Bouc Y, Chanson P, Schaison G 1999 Functional hypothalamic amenorrhoea: a partial and reversible gonadotrophin deficiency of nutritional origin. Clin Endocrinol (Oxf) 50: Pinhas-Hamiel O, Pilpel N, Carel C, Singer S 2006 Clinical and laboratory characteristics of adolescents with both polycystic ovary disease and anorexia nervosa. Fertil Steril 85: JCEM is published monthly by The Endocrine Society ( the foremost professional society serving the endocrine community.