Polycystic ovaries at ultrasound: normal variant or silent polycystic ovary syndrome?

Authors

  • S. Catteau-Jonard,

    Corresponding author
    1. Department of Endocrine Gynaecology and Reproductive Medicine, Hôpital Jeanne de Flandre, Centre Hospitalier Régional Universitaire de Lille, Lille, France
    2. Faculty of Medicine of Lille, Université de Lille II, Lille, France
    • Department of Endocrine Gynaecology and Reproductive Medecine, Hôpital Jeanne de Flandre, Centre Hospitalier Régional Universitaire de Lille, 59037 Lille, France
    Search for more papers by this author
  • J. Bancquart,

    1. Department of Endocrine Gynaecology and Reproductive Medicine, Hôpital Jeanne de Flandre, Centre Hospitalier Régional Universitaire de Lille, Lille, France
    2. Faculty of Medicine of Lille, Université de Lille II, Lille, France
    Search for more papers by this author
  • E. Poncelet,

    1. Faculty of Medicine of Lille, Université de Lille II, Lille, France
    2. Department of Radiology, Hôpital Jeanne de Flandre, Lille, France
    Search for more papers by this author
  • C. Lefebvre-Maunoury,

    1. Department of Endocrine Gynaecology and Reproductive Medicine, Hôpital Jeanne de Flandre, Centre Hospitalier Régional Universitaire de Lille, Lille, France
    2. Faculty of Medicine of Lille, Université de Lille II, Lille, France
    Search for more papers by this author
  • G. Robin,

    1. Department of Endocrine Gynaecology and Reproductive Medicine, Hôpital Jeanne de Flandre, Centre Hospitalier Régional Universitaire de Lille, Lille, France
    2. Faculty of Medicine of Lille, Université de Lille II, Lille, France
    Search for more papers by this author
  • D. Dewailly

    1. Department of Endocrine Gynaecology and Reproductive Medicine, Hôpital Jeanne de Flandre, Centre Hospitalier Régional Universitaire de Lille, Lille, France
    2. Faculty of Medicine of Lille, Université de Lille II, Lille, France
    Search for more papers by this author

Abstract

Objective

It is not known whether polycystic ovaries (PCO) are an ovarian appearance without pathological meaning or whether they share with polycystic ovary syndrome (PCOS) the same ovarian follicle abnormality. There are few studies including strictly selected women with PCO but without other criteria of PCOS. In order to address these issues, we compared hormonal, metabolic and ultrasound parameters obtained from patients with PCO only, patients with PCOS and controls.

Methods

This was a comparative analysis including three age-matched groups of 95 patients, who were included consecutively in a database: controls, patients with sonographic PCO but no symptoms (PCO group) and patients with PCOS. A clinical examination, fasting serum sampling and pelvic ultrasound examination were performed between cycle days 2 and 5 and results were compared between groups.

Results

The median serum anti-Mullerian hormone (AMH) level in the PCO group was intermediate between that in controls and that in the PCOS group (33.6 pmol/L, 19.8 pmol/L and 63.3 pmol/L, respectively), the differences being significant after adjustment for follicle number (P < 0.05), while the mean androgen serum level in the PCO group was similar to that in the control group and significantly lower than that in the PCOS group (P < 0.05) (median serum testosterone levels: 0.90 nmol/L, 0.79 nmol/L and 1.39 nmol/L; median androstenedione levels: 5.25 nmol/L, 4.37 nmol/L and 6.09 nmol/L, respectively). Body mass index, waist circumference and insulin levels had no effect on these differences.

Conclusion

PCO is an abnormal condition, affected women showing no evidence of hyperandrogenism but having higher AMH serum levels compared with controls, suggesting a granulosa cell abnormality in PCO similar to that observed in PCOS. The absence of hyperandrogenism in PCO does not seem linked to the metabolic status. Copyright © 2012 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Polycystic ovary syndrome (PCOS) is the most common cause of anovulation, infertility and hyperandrogenism in women, affecting between 5 and 10% of women of reproductive age worldwide1. Since the ASRM/ESHRE (American Society for Reproductive Medicine/European Society of Human Reproduction and Embryology) sponsored consensus conference in Rotterdam in 20032, the ultrasound criteria for polycystic ovaries (PCO) have now been added to the criteria in the National Institutes of Health (NIH)3 definition, i.e. hyperandrogenism and oligoanovulation.

The inclusion of ultrasound criteria is the subject of controversy because of its apparent lack of specificity; ultrasound features of PCO are observed in 21–63% of apparently normal women4–10. However, these numbers need to be interpreted carefully. Indeed, in these studies, a large proportion of women with PCO on ultrasound also had irregular menstrual cycles, supranormal serum androgen level or clinical hyperandrogenism, thus meeting the Rotterdam criteria for PCOS. Moreover, in the recent studies8, 9, advances in ultrasound technology allowing accurate detection of small antral follicles may have contributed to an artificial increase in prevalence of PCO. Therefore, the distinction between PCO and PCOS was not clear in any of these reports and the criteria defining PCOS were heterogeneous and not consensual. At present, there is a lack of studies with a strict selection of women with PCO but without other PCOS criteria.

The aim of our study was to select a pure population of women with PCO but without clinical symptoms or biological hyperandrogenism and to determine whether PCO are an ovarian state without pathological meaning or reflect the same abnormality of ovarian folliculogenesis as in PCOS.

Patients and Methods

Patients

This study was approved by the institutional review board of Lille University Hospital. Informed consent was obtained from patients before their inclusion in the study, on the day of examination.

The study included three groups of women referred to our department between 2004 and 2008 (at which point the ultrasound equipment was changed). Groups included women with normal ovaries on ultrasound and no symptoms (controls), women with sonographic PCO but no symptoms (PCO group) and patients with PCOS (PCOS group). We initially recruited 112 women with PCO who were included consecutively in our database. From this group, 95 patients were eligible for the study, i.e. all required data were available. We then selected randomly 95 controls and 95 patients with PCOS, who were included in our database during the same time period and were matched by age to the women with PCO. We used the same sample size as in our previous study in which we detected significant difference in mean anti-Mullerian hormone (AMH) serum levels between PCOS women and controls11.

The control population and the PCO population had been referred for tubal or male infertility. Exclusion criteria were a history of menstrual irregularity (cycle length < 25 days or > 35 days), hirsutism (as assessed by a modified Ferriman and Gallwey score of > 6) or severe acne/seborrhea and/or abnormal serum level of prolactin, testosterone and/or androstenedione (≥ 25, ≥ 0.6 or ≥ 2.2 ng/mL, respectively). For the control population, we excluded women with PCO on ultrasound examination. PCO patients were defined as having more than 11 small-sized (2–9 mm in mean diameter) ovarian follicles per ovary, unilaterally or bilaterally. Most women had more than 23 follicles in total. Six women had fewer than 23 follicles because they had an excess of follicles only unilaterally, but they met the Rotterdam criteria. We did not use ovarian volume for the definition of PCO because this criterion is less specific and less sensitive than is antral follicle count. Moreover, the threshold of 10 mL is not consensual12.

For the 95 patients with PCOS, the diagnosis was based on the definition of the consensus conference in Rotterdam in 20032, except that, as for PCO, ovarian volume was not considered. The diagnosis of PCOS was made in the presence of at least two of the three following criteria: (1) ovulatory irregularity (i.e. cycle length < 25 days or > 35 days); (2) hyperandrogenism, defined clinically by hirsutism (modified Ferriman and Gallwey score > 6) or severe acne/seborrhea and/or biologically by elevated testosterone (≥ 0.6 ng/mL) and/or androstenedione (≥ 2.2 ng/mL) serum levels; and (3) more than 11 follicles measuring 2–9 mm in mean diameter per ovary, unilaterally or bilaterally, at ultrasound examination. All women with PCOS had more than 23 follicles in total.

Any woman with at least one follicle > 9 mm in diameter at ultrasound examination, or a serum estradiol level > 300 pmol/L, was excluded from the study so as not to confound the data with the presence of a dominant follicle.

Methods

Each of the 285 patients underwent a clinical examination, fasting serum sampling at 08.00 h and a pelvic ultrasound examination, all performed on the same day, between cycle days 2 and 5. In PCOS patients presenting with amenorrhea, menses were induced by the administration of didrogesterone (10 mg/day for 7 days).

Clinical examination included an assessment of hirsutism by the method of Ferriman and Gallwey13 and measurement of waist circumference (WC). The mean body mass index (BMI) was calculated and systolic and diastolic blood pressure (mmHg) were measured.

Serum sampling involved measurement by immunoassay of estradiol, androstenedione, testosterone, 17-hydroxyprogesterone (17-OH-progesterone), dehydroepiandrosterone sulfate (DHEAS), luteinizing hormone (LH), follicle-stimulating hormone (FSH), sex-hormone-binding globulin (SHBG) and fasting serum insulin levels as described previously14. Serum AMH levels were assessed using the second-generation enzyme immunoassay AMH-EIA (ref A16507) provided by Beckman Coulter Immunotech (Villepinte, France) as described previously15. The free androgen index (FAI) was expressed as a number without units, using both total testosterone and SHBG molar concentrations according to the formula FAI = total testosterone × 100/SHBG. The quantitative insulin sensitivity check index (QUICKI) was also expressed as a number without units according to the formula: QUICKI = 1/log insulin + log glycemia (in mg/dL). High-density lipoprotein (HDL) cholesterol and triglycerides were measured in our laboratory. The concentration of HDL cholesterol was determined using cholesterol oxidase enzymatic assay (Thermo Fisher Scientific, Clinical Diagnostics, Vantaa, Vantaa, Finland). Serum triglyceride concentration was determined using a Konelab analyser with a calibrating probe (Thermo Fisher Scientific).

Pelvic ultrasound examination was performed with a Sonoline Elegra (Siemens Ultrasound Division, Issaquah, WA, USA) machine equipped with a 7-MHz transvaginal transducer. The same machine was used for patients from all three groups throughout the inclusion period but different ultrasonographers performed the examinations. The interobserver coefficient of variation (SD/mean) was < 10% for the antral follicle count (determined by the independent measures of two observers on 140 patients). Ultrasound measurements were taken in real time, according to a standardized protocol. Only two-dimensional ultrasound was used. The highest possible magnification was used to examine the ovaries. After the longest medial axis of the ovary had been determined, the length and thickness were measured and the area was calculated using a manual or automatic ellipse to outline the ovary. Ovarian area was used as a variable because its specificity and sensitivity are higher than those of ovarian volume12. All follicles < 9 mm but > 2 mm in diameter were counted. The diameter of several follicles was calculated by taking the mean of the longitudinal and anteroposterior diameters, then the number of follicles measuring > 2 mm and < 9 mm was established by scanning each ovary from the inner margin to the outer margin in longitudinal cross-section.

Statistical analysis

The ovarian area and follicle number data were obtained by summing the values for left and right ovaries. Since all parameters were non-normally distributed, all comparisons between the three groups were performed on log values by ANOVA and covariance analysis, both with Bonferroni correction. All statistical procedures were carried out using SPSS 15.0 (SPSS Inc., Chicago, IL, USA) and P < 0.05 was considered statistically significant.

Results

Clinical, hormonal and ultrasound data

The clinical, hormonal and ultrasound data of controls and patients with PCO or PCOS are presented in Tables 1 and 2. There were no differences for any metabolic markers (BMI, blood pressure, WC, SHBG, insulin, HDL cholesterol, low-density lipoprotein (LDL) cholesterol, triglycerides, glycemia/insulin, QUICKI) between the control and PCO groups (Table 1). In contrast, in PCOS women, BMI, WC, fasting serum insulin, LDL cholesterol and triglyceride levels were significantly increased (P < 0.05), whereas SHBG, QUICKI and HDL cholesterol levels were significantly decreased (P < 0.05), compared with the control group and with the PCO group (Table 1). There was no difference in blood pressure between the three groups.

Table 1. Clinical and metabolic characteristics of controls, women with polycystic ovaries (PCO) and women with polycystic ovary syndrome (PCOS)
CharacteristicControls (n = 95)PCO group (n = 95)PCOS group (n = 95)P§
  • Data are expressed as median (range).

  • *

    To convert glycemia to mmol/L, multiply by 5.551.

  • To convert insulin to pmol/L, multiply by 7.175.

  • To convert cholesterol and triglycerides to mmol/L, multiply by 2.586.

  • §

    ANOVA with Bonferroni correction on log values: bPCO group vs PCOS group, P < 0.05; ccontrols vs PCOS group, P < 0.05.

  • BMI, body mass index; BP, blood pressure; HDL, high-density lipoprotein; LDL, low-density lipoprotein; NS, not significant; QUICKI, quantitative insulin sensitivity check index; SHBG, sex-hormone-binding globulin.

Age (years)29 (17–39)29 (17–39)29 (17–39)NS
BMI (kg/m2)22.9 (17.2–44.0)23.0 (17.0–43.4)26.5 (15.0–53.0)b,c
Systolic BP (mm/Hg)110 (95–150)110 (90–140)110 (90–150)NS
Diastolic BP (mm/Hg)70 (50–93)70 (50–97)70 (50–90)NS
Waist circumference (cm)75 (59–131)76 (61–123)88 (61–142)b,c
SHBG (nmol/L)53.6 (15.6–132.0)43.9 (8.3–143.0)33.8 (6.9–97.9)b,c
Glycemia (g/L)*0.86 (0.59–1.14)0.84 (0.62–1.23)0.81 (0.62–1.06)c
Insulin (mIU/L)3.6 (0.7–19.7)4.5 (0.7–20.6)6 (0.7–20.9)b,c
QUICKI0.69 (0.44–1.28)0.63 (0.43–1.33)0.61 (0.43–1.34)b,c
Glycemia/insulin0.24 (0.05–1.29)0.18 (0.04–1.11)0.14 (0.04–1.13)b,c
HDL cholesterol (g/L)0.60 (0.31–1.31)0.60 (0.34–1.53)0.52 (0.29–0.81)b,c
LDL cholesterol (g/L)0.99 (0.38–2.34)0.99 (0.47–1.76)1.90 (0.49–2.48)b,c
Triglycerides (g/L)0.57 (0.22–2.72)0.58 (0.26–2.20)0.72 (0.29–3.53)b,c
Table 2.  Hormonal and ovarian characteristics of controls, women with polycystic ovaries (PCO) and women with polycystic ovary syndrome (PCOS)
CharacteristicControls (n = 95)PCO group (n = 95)PCOS group (n = 95)PP adjusted for BMI, WC or IP adjusted for 2–9 FN
  • Data are expressed as median (range).

  • *

    To convert 17-OH-P to nmol/L, multiply by 3.026.

  • ANOVA or

  • covariance analysis, both with Bonferroni correction on log values: aPCO group vs controls, P < 0.05; bPCO group vs PCOS group, P < 0.05; ccontrols vs PCOS group, P < 0.05. 17-OH-P, 17-hydroxyprogesterone; 2–9 FN, 2–9-mm follicle number (sum of both ovaries); AMH, anti-Mullerian hormone; BMI, body mass index; DHEAS, dehydroepiandrosterone sulfate; E2, estradiol; FAI, free androgen index; FSH, follicle-stimulating hormone; I, insulin; LH, luteinizing hormone; NA, not applicable; NS, not significant; WC, waist circumference.

Testosterone (nmol/L)0.79 (0.17–1.84)0.90 (0.14–1.80)1.39 (0.17–3.57)b,cb,cb,c
FAI5.41 (1.09–14)5.99 (1.12–20)13.3 (2.8–51)b,cb,cb,c
Androstenedione (nmol/L)4.37 (0.63–7.66)5.25 (1.57–7.7)6.09 (2.1–17.36)b,cb,cb,c
17-OH-P (ng/mL)*0.48 (0.21–1.86)0.52 (0.17–1.46)0.64 (0.24–1.49)b,cb,cb,c
DHEAS (µmol/L)4.0 (0.8–12)4.6 (1.5–10.4)5.1 (0.6–14)NSNSNS
LH (IU/L)3.8 (1.3–12.5)4.3 (1.2–9.4)5.7 (1.3–19.1)b,cb,cb,c
FSH (IU/L)6.5 (3.8–12)5.9 (2.9–10.9)5.5 (3.1–11.1)a,ccc
E2 (pg/mL)33 (19–80)34 (13–79)35 (15–69)NSNSNS
AMH (pmol/L)19.8 (6.6–69)33.6 (6.9–75.9)63.3 (15.4–239.6)a,b,ca,b,ca,b,c
Ovarian area (cm2)7.4 (3.6–16.6)9.6 (6.2–15.8)10.8 (6–23.4)a,b,ca,b,ca,b,c
2–9 FN15 (4–23)35 (14–67)45 (25–140)a,b,ca,b,cNA

Serum androgen data did not differ between controls and patients with PCO (Table 2), but, with the exception of DHEAS, they were significantly higher in PCOS women compared with PCO and control groups (Table 2). These results did not change after adjustment for BMI, WC or fasting serum insulin by covariance analysis (Table 2).

The number of 2–9-mm follicles and ovarian area were found to be intermediate in the PCO group, being significantly higher than in controls and significantly lower than in the PCOS group (Table 2, Figure 1a and b, respectively). The same pattern was observed for AMH serum level (Table 2, Figure 1c). This last result remained significant after adjustment for the number of 2–9-mm follicles (Table 2).

Figure 1.

Box -and-whisker plots showing distribution of individual values for number of 2–9-mm follicles (FN, sum of both ovaries) (a), ovarian area (sum of both ovaries) (b), anti-Mullerian hormone (AMH) (c) and follicle-stimulating hormone (FSH) (d) in 95 controls, 95 women with polycystic ovaries (PCO) and 95 patients with polycystic ovary syndrome (PCOS). Horizontal small bars represent the 5th–95th percentile range, boxes indicate the 25th–75th percentile range, the horizontal line in each box corresponds to the median and open circles represents values beyond the 95th percentile. Comparisons between groups were performed by ANOVA with Bonferroni correction on log values, *P < 0.05.

To study the potential role of AMH in cycle regularity and to determine why women with PCO did not have menstrual irregularity despite higher AMH serum levels, we divided the PCOS group into three sub-groups according to menstrual cycle irregularity: 13 (13.5%) patients presented with amenorrhea, defined by absence of menses > 3 months; 70 (74%) patients had oligomenorrhea, defined by fewer than eight menses in the preceding year; and 12 (12.5%) patients had eumenorrhea (regular cycles, with length generally 25–35 days). The mean AMH serum level appeared similar between women with PCO and eumenorrheic women with PCOS. In both groups, it was significantly higher than in controls and significantly lower than in oligo- or amenorrheic women with PCOS (Figure 2).

Figure 2.

Box-and-whisker plot showing distribution of individual values for anti-Mullerian hormone in 95 controls, 95 women with polycystic ovaries (PCO) and 95 patients with polycystic ovary syndrome (PCOS), according to menstrual status. Horizontal small bars represent the 5th–95th percentile range, boxes indicate the 25th–75th percentile range, the horizontal line in each box corresponds to the median and open circles represents values beyond the 95th percentile. Pairwise comparisons between groups were performed by ANOVA with Bonferroni correction on log values. All were significant (P < 0.0001), except comparisons between the PCO group and the eumenorrheic PCOS sub-group (Eum PCOS), and between the oligomenorrheic PCOS sub-group (Oligo PCOS) and the amenorrheic PCOS sub-group (Am PCOS). NS, not significant.

Finally, the FSH serum level was significantly higher in controls than in PCO and PCOS groups (Table 2, Figure 1d). However, the significance of the difference between controls and PCO was lost after adjustment for BMI and the number of 2–9-mm follicles (Table 2).

Discussion

In this study, we found that serum AMH level in the PCO group was intermediate between that in the control and PCOS groups, and significantly different from both. This result leads us to conclude that PCO is apparently not a normal variant, but rather a mild phenotype of PCOS. The difference in AMH serum level compared with controls suggests an abnormality of the granulosa cells similar to that in PCOS; the excess is apparently not caused by the higher number of follicles, since the difference from controls persisted after adjustment for the number of 2–9-mm follicles, suggesting excessive secretion of AMH per follicle. This could be a reflection of an increased number of granulosa cells within each follicle or over-expression of AMH by each granulosa cell, similar to women with PCOS16, 17.

A strength of our study was the rigorous selection of women with PCO, allowing us to explore this specific population, especially concerning their AMH levels. Women with PCO had no clinical or biological hyperandrogenism. Moreover, we preferred to include only patients who underwent ultrasound examination with the equipment used to establish the standards adopted at the Rotterdam Consensus18. We believe that the newer machines detect small follicles not visible with older ultrasound equipment, which increases the risk of overdiagnosing PCOS. We therefore believe that the threshold of 24 follicles (sum of both ovaries) should be reviewed when using newer machines and we have identified a new threshold for use at our center (38 follicles19). To meet the Rotterdam criteria (concerning the follicle number), we thus chose to include only patients examined with our ‘old’ ultrasound equipment, up to late 2008.

These strict selection criteria may explain the difference in our results from those of other studies. Through our selection method, no clinical or biological hyperandrogenism was observed in our sample of women with PCO. Conversely, in the other studies4–8, 10, a large proportion of women with PCO had supranormal serum androgen levels or clinical hyperandrogenism, thus meeting the Rotterdam criteria for PCOS. However, our data obtained at basal state, without hormonal stimulation, do not eliminate the possibility of a hidden dysregulation of theca cells in the PCO group. Indeed, Chang et al.20 found that women with PCO had an abnormal androgen response to a gonadotropin-releasing hormone agonist test, whereas their baseline androgen levels were similar to those of controls. Lastly, we found no metabolic abnormality in women with PCO, as opposed to those with PCOS. These data confirm those obtained in most previous studies including series of women with PCO4, 8, 10, 20–22.

On the basis of our present data, we hypothesize that, like women with PCOS, those with PCO also have a granulosa cell abnormality. Presumably, in these patients, this phenomenon is too mild to affect the ovulatory process. Interestingly, in our study, the mean AMH serum level was similar between the PCO and eumenorrheic PCOS groups. Conversely, both groups had significantly lower mean AMH serum levels compared with the PCOS patients with amenorrhea or oligomenorrhea. We have previously reported such a gradation of AMH level according to menstrual cycle irregularity in women with PCOS11. In agreement, Piouka et al.23 found an association of AMH serum levels with the degree of severity of PCOS. The reproductive consequences of moderately elevated serum AMH level need to be documented further. Clayton et al.7 and Hassan and Killick24 did not report deleterious effects of PCO on fertility, probably due to the absence of dysovulation in this group. Nevertheless, in the context of assisted reproduction, Sahu et al.25 reported that women with PCO had similar characteristics to women with PCOS in terms of ovarian response to human menopausal gonadotropin stimulation, oocyte and embryo quality and pregnancy rate. The clinically silent granulosa cell abnormality in women with PCO may be part of the reason for such findings.

In conclusion, our study suggests that PCO is not a normal variant but rather a mild phenotype of PCOS. This is based on clinical, hormonal and ultrasound data. To extend this hypothesis, further molecular studies on granulosa cells from women with PCO are required. We speculate that PCO and PCOS share some common genetic background but it remains to be determined which genetic and/or environmental influence(s) contribute(s) to the phenotypic evolution towards either PCO or PCOS.

Acknowledgements

We thank Carole Santoro for her contribution to the English translation.

Ancillary