Anti‐Müllerian hormone measurement for the diagnosis of polycystic ovary syndrome

Anti‐Müllerian hormone (AMH) is derived from the small antral follicles, and an elevated level has been suggested to add value to the Rotterdam criteria for the diagnosis of PCOS in cases of diagnostic uncertainty. Therefore, the role of AMH in the classical phenotype of PCOS was defined within a Caucasian population.


| INTRODUCTION
Polycystic ovary syndrome (PCOS) is one of the most common endocrine disorders and affects 6%-20% of reproductive-aged women. 1,2 Currently, PCOS is diagnosed using the Rotterdam, or Androgen Excess Society or NIH criteria, but it still remains a diagnosis to be made only after the exclusion of other conditions. 3,4 A diagnostic test that would positively identify a patient with PCOS and add value to the current diagnostic criteria would be of great benefit, and anti-Müllerian hormone (AMH) measurement has been suggested to add value to the Rotterdam criteria. AMH is produced in the granulosa cells by the preantral and small antral follicles whose mechanism appears to be to inhibit the action of FSH on aromatase so assisting with the development of a single follicle for ovulation. 5 AMH is reported to be elevated in PCOS due to the increased small antral follicle count, where that increase in AMH may be a combination of the increased number of small antral follicles and the increased secretion of AMH per follicle. 6 It has been suggested that AMH levels are a good marker for the size of the antral follicle pool. Ultrasound is used to measure the antral pool size but concerns over operator dependency are its main limitation, and therefore, a single serum measurement of AMH would be attractive. 7 Several studies have been carried out on the utility of AMH in the diagnosis of PCOS 8,9 with differing diagnostic thresholds being proposed, although a meta-analysis suggested a cut-off of 35 pmol/L with a sensitivity of 79.4% and a specificity of 82.8% 8 ; however, it was noted that this may not be applicable to all PCOS subgroups and that the populations tended to be from fertility clinics rather than a general population and differing assays for AMH undertaken. 8 A raised AMH is particularly associated with PCOS patients with all 3 diagnostic criteria 10 ; therefore, the aim of this study was to look specifically at the utility of AMH for the diagnosis of PCOS within a well-defined cohort of PCOS patients from the general population that fulfilled all 3 of the Rotterdam criteria to determine if it could be complementary to accepted androgen markers.

| MATERIALS AND METHODS
This was a cross-sectional study involving 105 well-characterized women with PCOS and 65 women without PCOS who presented sequentially to the Department of Endocrinology and were recruited to the local PCOS biobank (ISRCTN70196169). All patients gave written informed consent. This study was approved by the Newcastle & North Tyneside Ethics committee. The diagnosis of PCOS was based on all 3 diagnostic criteria of the Rotterdam consensus, namely clinical and/or biochemical evidence of hyperandrogenaemia (Ferriman-Gallwey score >8; free androgen index >4, respectively), oligomenorrhea or amenorrhea and polycystic ovaries on transvaginal ultrasound, 11 described as the type A phenotype in which AMH levels are highest. Study participants had no concurrent illness, were not on any medication for the preceding 9 months except study medications and were not planning to conceive. Nonclassical 21-hydroxylase deficiency, hyperprolactinaemia, Cushing's disease and androgen-secreting tumours were excluded by appropriate tests. All of the control women had regular periods, no clinical or biochemical hyperandrogenaemia, no significant background medical history and none of them were on any medications including oral contraceptive pills or over the counter medications. All women underwent a 75-g oral glucose tolerance test to exclude impaired glucose tolerance and type 2 diabetes. All women with PCOS and control women were Caucasian. Height, weight and waist circumference and body mass index (BMI) were performed according to WHO guidelines. 12

| Collection and analysis of blood samples
Blood samples were collected immediately (within 5 minutes) and were stored frozen at −80°C pending analysis. Serum T was measured by LC/ MS/MS on an Acquity UPLC system coupled to a Quattro Premier XE mass spectrometer (Waters, Manchester, UK). Sex hormone-binding globulin (SHBG) was measured by an immunometric assay with fluorescence detection on the DPC Immulite 2000 analyser using the manufacturer's recommended protocol. The free androgen index (FAI) was calculated as the total testosterone × 100/SHBG. Serum insulin was assayed using a competitive chemiluminescent immunoassay performed on the manufacturer's DPC Immulite 2000 analyser (Euro/ DPC, Llanberis, UK). The analytical sensitivity of the insulin assay was 2 μU/mL, the coefficient of variation was 6%, and there was no stated cross-reactivity with proinsulin. Plasma glucose was measured using a Synchron LX 20 analyser (Beckman Coulter, High Wycombe, UK), using the manufacturer's recommended protocol. The coefficient of variation for the assay was 1.2% at a mean glucose value of 5.3 mmol/litre during the study period. The insulin resistance was calculated using the HOMA method [HOMA-IR=(insulin × glucose)/22.5]. Anti-Müllerian hormone was measured using a Beckman Coulter Access-automated immunoassay. Between-run precision was <3% across the range measured.

| Statistical analysis
Data trends were visually evaluated for AMH and each androgen, and nonparametric tests were applied on data that violated the assumptions of normality when tested using Kolmogorov-Smirnov test. Accordingly, comparative analysis evaluating androgen levels between PCOS cases and controls was performed using the nonparametric Mann-Whitney tests. Pearson's correlations were also estimated to assess any linear relationship between AMH and the different parameters. Finally, multivariable logistic regressions adjusting for age and BMI, serum testosterone or FAI were conducted to assess the effects of AMH in the diagnoses PCOS, firstly with an AMH categorical value of greater than 46 (based on the 95 th percentile sensitivity of the ROC) and secondly based on AMH with a categorical value of greater than 35 according to the literature. 8 Significance was defined at α = 0.05. All analyses were carried out using IBM-SPSS version 24.0. All values are given as (mean ± SD) unless specified.

| RESULTS
The baseline demographics of patients are given in Table 1. Of the 170 subjects with a serum AMH, 4 of 65 controls and 3 of 105 PCOS subjects did not have a FAI level available. PCOS women were significantly younger (27.7 ± 6.3 years) than the controls (30.2 ± 6.30 years).
Their body mass index (BMI), waist and hip circumferences were also significantly higher (P < .0001) than the controls (Table 1). Patients with PCOS showed greater insulin, HOMA-IR and 2-hour glucose post oral glucose tolerance test (OGTT) values (P < .001) ( Table 1). Serum T (1.5 ± 0.9 vs 1.0 ± 0.5 nmol/L) and FAI (6.3 ± 5.7 vs 2.5 ± 1.6) were significantly elevated in PCOS compared to control (P < .001). The PCOS groups with biochemical hyperandrogenaemia showed a more metabolic profile with increased BMI, insulin and HOMA values (P < .01) compared to those with clinical hyperandrogenism, but AMH levels did not differ between the 2 groups (Table 2). When the metabolic parameters were compared between those patients with an AMH of 35 or 46 pmol/L with those less than 35 pmol/L, there were no differences in any of the metabolic parameters (Table S1). AMH, as a quantitative measure, did not correlate with insulin, HOMA-IR or CRP in either the control or PCOS subjects, but it did correlate with serum T and FAI in controls (r = .3, P < .05; r = .35, P < .01, respectively), but only to serum testosterone (r = .2, P < .05), and not FAI in the PCOS subjects.
As individual tests, the age-and BMI-adjusted multiple logistic regression showed that for AMH at thresholds greater than either 35 or 46 pmol/L, added significantly to both serum T and FAI in the prediction of PCOS. When serum T and FAI were both included in the model AMH remained predictive of PCOS with an OR 3.8 and 4.5 (P < .01) for an AMH cut-off of 46 and 35 pmol/L, respectively (Table 3).

| DISCUSSION
It has been suggested that AMH may be a useful initial diagnostic test for PCOS with a cut-off value of around 35 pmol/L. 8,13 In accord with others, we found a raised AMH was strongly associated with a diagnosis of PCOS with a fourfold odds ratio. Thus, in an analogous way to making a diagnosis of diabetes-where a diagnosis can be made using either a raised fasting plasma glucose, a raised 2-hour glucose value or a raised HbA1c 14 -then a raised serum testosterone, or a raised FAI or a raised AMH will identify an additional 22% more PCOS patients positively with an AMH threshold of 35 (16% with an AMH cut-off of 46 pmol/L).
Therefore, there is additional discriminatory benefit for making the diagnosis of PCOS by measuring AMH in addition to the traditional androgen markers. The sensitivity of AMH for a diagnosis of PCOS was low with either cut-off (46 pmol/L; 41% and 35 pmol/L;55%) and lower than that reported of 79%, although with a similar specificity and likely reflected the general population of study rather than from fertility clinics. 8 The number of patients with clinical hyperandrogenism who had a raised AMH but had no biochemical evidence of hyperandrogenaemia (as evidenced by normal serum T and FAI) was 16/42 (38%) vs 23/57 (40%) using 46 pmol/L and 35 pmol/L thresholds, respectively.
If one then applies the AMH levels at either cut-off to that of the NIH There was a fourfold prediction of PCOS by AMH, and both cut-off values appeared equally predictive. study were in part the converse of the strengths. This was a narrow group of individuals all with the same 3 diagnostic criteria fulfilled for the diagnosis of PCOS, and other PCOS phenotypes not included. In addition, this was a narrow ethnic group, and therefore, more detailed studies including the other PCOS phenotypes with other ethnicities are required.
In conclusion, whilst an elevated AMH has poor sensitivity, it is fourfold more likely to be associated with a diagnosis of PCOS which, in this study, results in its measurement identifying an additional 16% or 22% of patients (using AMH cut-offs of 35 or 46 pmol/L, respectively) as having PCOS when routine measurement of androgen concentrations are not elevated. This means there may be merit in considering AMH as a marker which complements androgen measurement in the biochemical assessment of PCOS.