Abstract and Introduction
Objectives: Although the Rotterdam 2003 polycystic ovarian syndrome (PCOS) diagnostic criteria is widely used, the need to consider multiple variables makes it unwieldy in clinical practice. We propose a simplified PCOS criteria wherein diagnosis is made if two of the following three items were present: (i) oligomenorrhoea, (ii) anti–mullerian hormone (AMH) above threshold and/or (iii) hyperandrogenism defined as either testosterone above threshold and/or the presence of hirsutism.
Design setting and participants: This prospective cross–sectional study consists of healthy women (n = 157) recruited at an annual hospital health screen for staff and volunteers from the university community, and a patient cohort (n = 174) comprising women referred for suspected PCOS.
Main outcome measures: We used the healthy cohort to establish threshold values for serum testosterone, antral follicle counts (AFC), ovarian volume (OV) and AMH. Women from the patient cohort, classified as PCOS by simplified PCOS criteria, AMH alone and Rotterdam 2003, were compared with respect to prevalence of oligomenorrhoea, hyperandrogenism and metabolic indices.
Results: In healthy women, testosterone ≥1.89 nmol/L, AFC ≥22 follicles and OV ≥8.44 mL, best predicted oligomenorrhoea and were used as threshold values for PCOS criteria. An AMH level ≥37.0 pmol/L best predicted polycystic ovarian morphology. AMH alone as a single biomarker demonstrated poor specificity (58.9%) for PCOS compared to Rotterdam 2003. In contrast, there was a 94% overlap in women selected as PCOS by the simplified PCOS criteria and Rotterdam 2003. The population characteristics of these two groups of PCOS women showed no significant mean differences in androgenic, ovarian, AMH and metabolic (BMI, HOMA–IR) variables.
Conclusions: Our data recommend the simplified PCOS criteria with population–specific thresholds for diagnosis of PCOS. Its ability to replace ovarian ultrasound biometry with the highly correlated variable AMH, and use of testosterone as a single marker for hyperandrogenaemia alongside the key symptoms of oligomenorrhoea and hirsutism confers significant clinical potential for the diagnosis of PCOS.
More than 80 years since its description, polycystic ovary syndrome (PCOS) remains a diagnostic conundrum. The 1990 National Institutes of Health criteria require oligomenorrhoea, and clinical or biochemical hyperandrogenism for the diagnosis of PCOS. The Rotterdam 2003 criteria broadened the definition to include two of the three following criteria: (i) oligomenorrhoea indicating chronic anovulation, (ii) clinical and/or biochemical signs of hyperandrogenism, or (iii) polycystic ovarian morphology (PCOM) defined as antral follicle count (AFC) of ≥12 follicles per ovary and/or ovarian volume (OV) ≥10 mL in any ovary. The widely used Rotterdam 2003 criteria in its complete manifestation require assessment of multiple variables including average menstrual cycle length; transvaginal ultrasound measurement of the volume and number of ovarian antral follicles in each ovary; clinical assessment for hyperandrogenism including hirsutism, virilism, acne and male pattern baldness; biochemical measurement for androgens including testosterone, androstenedione, dehydroepiandrosterone sulphate (DHEAS) and sex hormone binding globulin (SHBG) to calculate free androgen index (FAI). Although widely used, the need to consider so many variables makes the Rotterdam 2003 criteria rather cumbersome and unwieldy for routine clinical use. The inconsistent inclusion or exclusion of variables result in heterogeneity in PCOS phenotypes, complicating diagnoses and assessment of therapeutic outcomes. It is not surprising that a recent study among U.S. gynaecologists indicate that more than a quarter of respondents admitted to not knowing which criteria they used to diagnose PCOS. There are growing calls for revised definitions of PCOS, amidst increasing concern that current diagnostic criteria for PCOS introduces variability and poses barriers for accurate pathophysiological characterization of PCOS.[7–9]
In order to pinpoint the essential elements of PCOS for designing simplified diagnostic criteria, we performed a prospective cross–sectional study to identify, in an unbiased manner by computer algorithm, the critical predictors of oligomenorrhea. Oligomenorrhoea was chosen as the reference variable as it is the most frequent criteria used for the diagnosis of PCOS and is an essential pathological feature that has particular clinical relevance. Some 23 potential variables (clinical, hormonal and biochemical) known to be associated with PCOS were subjected to factor analysis with respect to their ability to predict oligomenorrhoea. Remarkably, only one factor group, comprising five variables (anti–mullerian hormone [AMH], AFC, OV, testosterone and luteinizing hormone [LH]) was significantly associated with oligomenorrhea. Leaving aside LH, which has wide variations due to pulse frequency, wide amplitude and short half–life,[3,13] there is a possibility that the items identified by factor analysis (AMH, AFC, OV and testosterone) may be core variables to diagnose PCOS.
Anti–mullerian hormone is a glycoprotein of the transforming growth factor family secreted by ovarian antral follicles. Anti–mullerian hormone levels are highly correlated with AFC and OV regardless of the phase of the menstrual cycle,[14,15] whether in PCOS patients or in normal controls. It would be advantageous if AMH can replace the use of transvaginal ultrasound given the complexity, cost, inter–observer variation involved in the determination of AFC and OV.[6,17,18] The strong correlation that has been reported between PCOS and AMH by a multitude of studies,[19–21] has even lead to the view that AMH alone, may be a suitable as a single biomarker for PCOS. Nevertheless, there is a lack of studies examining the performance of various proposed criteria compared to the Rotterdam 2003 criteria.
The overall aim of this study was to compare the performance of a simplified 4–item PCOS criteria and AMH alone as a single biomarker of PCOS to Rotterdam 2003. Threshold values for AMH, testosterone, AFC and OV were established using a healthy cohort recruited from an annual health screen. In a separate patient cohort comprising women referred for suspected PCOS, women identified as PCOS by the simplified 4–item PCOS criteria and AMH as a single biomarker were compared to Rotterdam 2003, with respect to menstrual cycle length, body mass index, androgenic and metabolic parameters.
Clin Endocrinol. 2018;89(2):202-211. © 2018 Blackwell Publishing