The diagnostic performance of antimullerian hormone for polycystic ovarian syndrome and polycystic ovarian morphology

The diagnosis of polycystic ovary syndrome (PCOS) remains a challenge to clinicians due to heterogeneous clinical presentation and diagnostic criteria. This study investigated the utilization of Anti-Müllerian hormone (AMH) alone or replacing polycystic ovarian morphology (PCOM) in the PCOS diagnostic criteria. A total of 401 women were categorised as PCOS (n:154), nonPCOS with polycystic ovarian morphology (PCOM) (n:105), and nonPCOS with normal ovarian morphology (NOM) (n:142). First, the diagnostic performance of AMH for PCOS diagnosis in Rotterdam, Androgen Excess Society, and National Institutes of Health (NIH) criteria was analyzed. Second, AMH was used instead of PCOM in Rotterdam criteria and we searched diagnostic performance for PCOS phenotypes. AMH levels were positively correlated with LH, testosterone, hirsutism score, menstrual cycle length, and antral follicle count (p < 0.05). AMH alone had specificity and sensitivity for PCOS diagnosis were 84.9% and 72.4% in Rotterdam (AUC: 0.866); 84.4% and 72% in Androgen Excess Society (AUC: 0.857); 83.3% and 66.4% in National Institute of Health criteria (AUC: 0.825). AMH alone had satisfactory diagnostic potential for phenotype A, but not other phenotypes. The replacement of PCOM with AMH in Rotterdam criteria had a high diagnostic potential for PCOS (AUC: 0.934, sensitivity:97.4%, specificity: 90.67%). Phenotype A and phenotype D were diagnosed with 100% sensitivity and 94.5% specificity. Phenotype C was recognised with 96.15% sensitivity and 94.5% specificity. AMH may be used with high diagnostic accuracy instead of PCOM in the Rotterdam PCOS criteria.


Introduction
Polycystic ovary syndrome (PCOS) is the most frequent endocrinopathy of reproductive age women and presents a variety of clinical findings such as oligomenorrhea, acne, hirsutism, alopecia, and infertility [1]. PCOS women also have long-term health risks as increased cardiovascular disease, metabolic syndrome, diabetes mellutes, and endometrial cancer [2]. Early diagnosis is an important task to prevent associated morbidity.
Although 80 years have passed since the first description [3], PCOS is still one of the most discussed endocrine disorders in the diagnosis. There are three conventional diagnostic criteria. National Institutes of Health (NIH) criteria define PCOS as hyperandrogenism (HA) and oligoanovulation (OA) [4]. Rotterdam criteria add polycystic ovary morphology (PCOM) to NIH criteria [5]. Androgen Excess Society AES accepts hyperandrogenism with OA or PCOM for diagnosis [6]. The prevalence of PCOS varied (6-21%) with ethnicity and diagnostic criteria used [7].
The Rotterdam criteria are widely used in diagnosis. However, the fact that hyperandrogenemia is not a mandatory criterion and the diagnosis of PCOM is subjective are the issues that have been criticized [8]. The ultrasonographic evaluation of ovarian morphology for PCOM diagnosis may not be convenient all the time. The insufficient imaging by abdominal ultrasonography in young obese girls, the large-multicystic appearance of ovaries in teenagers, and the unavailability of transvaginal ultrasonography due to virginity, the alterations of ovarian morphology during the menstrual cycle or oral contraceptives use are the limitations in PCOM diagnosis [9][10][11]. Recently, in 2014 AES-PCOS task force proposed an antral follicle count above 25 per ovary to diagnose PCOM by using a transducer frequency of ≥ 8 MHz [12]. Due to the limitations, we have listed, it is not possible to make this count with the basic ultrasonography used in normal gynecology practice. Further endeavors are required for the objective test.
Anti-Müllerian hormone (AMH) has an inhibitory effect on early follicular recruitment and regulates early follicular development till menopause [13][14][15]. AMH reflects the follicular pool both antral follicles (2-9 mm) seen on ultrasonography and also pre-antral follicles (< 2 mm) which are hardly seen in an ultrasound [16]. The AMH levels of PCOS women are much higher than healthy controls [17,18]. The AMH levels may be attributed to impaired folliculogenesis, accumulation of small antral follicles, or stimulatory effects of androgens in early follicular growth [14]. The relation of AMH with antral follicle count and clinical features of PCOS make the AMH test a promising method of PCOS diagnosis [19,20]. In addition, the fact that it does not change during the cycle, it is not affected by exogenous steroid intake, and has low interobserver variability brings AMH to the forefront as an objective biological marker [11].
Whether AMH alone is a reliable tool in diagnosing PCOS or an alternative to ultrasound still needs to be demonstrated. This study aimed to investigate the diagnostic potential of AMH as a single diagnostic tool and AMH replaced with PCOM in Rotterdam criteria.

Materials and methods
This cross-sectional study was performed in the Gynecology Clinic, University of Health Sciences, Haydarpaşa Numune Training and Research Hospital, Istanbul, Turkey. Ethical approval was obtained from the local Ethics Committee (HNEAH-KAEK, 2021/293-3343).
This study was carried out on PCOS patients followed in the gynecology clinic of our hospital between January 2018 and December 2020. All data extracted from files and electronic database filled detailly in the gynecology clinic, retrospectively. All patients had transvaginal ultrasonography (ovarian morphology) and hormone evaluation (cycle day (3). Of the 554 patients whose AMH was studied, 401 patients who fulfilled the working conditions were included in the study. Patients were divided into two age-matched groups: PCOS (n:154), nonPCOS (n:247). NonPCOS group was further stratified into nonPCOS with polycystic ovaries (PCOM, n:105), and normal ovarian morphology (NOM, n:142).
Patients with incomplete data, unavailable ovarian morphologic evaluation, FSH > 10 IU/l and patients who used hormonal drugs in the last six months, patients with known endocrine, collagen tissue, liver, kidney, and hematological diseases, previous ovarian surgery, ovarian cyst (> 14 mm), and unavailable ultrasound reports were excluded from the study.
The clinical features of PCOS were evaluated for each patient. Menstrual cycle days (Oligomenorrhea > 35 days), acne, alopecia, hirsutism (Ferriman Gallwey score > 7) were recorded. Biochemical hyperandrogenism (HA) was accepted as above 2.3 nmol/L of testosterone level [14]. Any of the ovaries with antral follicle count above 12 or ovarian volume > 10 cm 3 was accepted polycystic ovarian morphology (PCOM) [5]. The PCOS woman was diagnosed by Rotterdam Criteria (two of three features: OA, HA, PCOM) [5]. Phenotype A: OA-HA-PCOM, Phenotype B: OA-HA, Phenotype C: HA-PCOM, Phenotype D: OA-PCOM [21]. The patients were also evaluated for NIH criteria (OA and HA) and AES (HA and OA/PCOM) [4,6]. The primary outcome measure was the diagnostic performance of AMH alone or replaced PCOM in the Rotterdam criteria.
Each patient underwent transvaginal ultrasonography (Mindray DC-7 MX29003997 China, 5-8 mHz) for Antral follicle count (AFC) and ovarian volume (0.5 × length x width x thickness) measurements. All ultrasonographic measurements were performed by the same ultrasonographer (EK). Hormones such as follicle-stimulating hormone (FSH), luteinizing hormone (LH), estradiol, free testosterone, prolactin, and thyroid-stimulating hormone (TSH) were measured by an electrochemiluminescent immunometric assay. The serum AMH was analyzed by ECLIA (electrochemiluminescent immunometric assay) method using commercial kits (Elecsys ® Cobas AMH, Cobas 6000, e601, Roche Diagnostics). The blood required for hormone analysis (FSH, LH, estradiol, testosterone) was taken on the 2nd or 3rd day of menstruation. AMH hormone was measured from the blood taken on any day of the cycle. SPSS Statistics 22 programs were used. In addition to descriptive statistical methods (mean, standard deviation, frequency), Pearson or Spearman correlation analysis was used in correlation analysis following the data distribution. For statistical evaluation, tests such as the Oneway Anova test, Kruskal Wallis test, Mann Whitney U, Fisher's Exact test, Student t-test were used in appropriate analyzes. The ROC analysis was performed to measure cut-off value, AUC (area under the curve), sensitivity, and specificity of AMH. Significance was evaluated at the p < 0.05 level.
AMH levels were found to be strongly and positively associated to clinical and laboratory characteristics of PCOS. Menstrual cycle length (r: 0.513 p: 0.000), Antral Follicle Count (r: 0.300, p: 0.000) and PCOM (r: 0.242 p: 0.00), FGS (r: 0.228 p:0.00), LH (r: 0.426, p: 0.000), and testosterone levels (r: 0.181, p: 0.01) were positively correlated with AMH levels. AMH levels decreased by increasing age (r: − 0.480 p: 0.00), but no correlation with BMI (r: 0.037 p: 0.686). Table 3. demonstrates the utility of AMH alone for PCOS diagnosis with NIH, AES and Rotterdam criteria. On applying the receiver operating characteristics curve (ROC curve) analysis, the maximum diagnostic potency of AMH alone for PCOS was at a cut-off of 4 ng/ml in our study population (Fig. 1). The ROC analysis showed AUC value 0.866 (95% CI 0.83-0.89; p: 0.00) for Rotterdam, 857  (Fig. 1a), the sensitivity of 84.4% and specificity of 72% in AES (Fig. 1b), and sensitivity of 83.3% and specificity of 66. 4% in NIH criteria (Fig. 1c). Figure 2 shows the ROC analysis of AMH for PCOM. Figure 3 shows PCOS diagnosed with Rotterdam criteria were re-evaluated after the replacement of PCOM with AMH (> 4 ng/ ml) ( AMH, OA, HA). The diagnosis of phenotype A (OA + HA + AMH) was 100% sensitivity and 94.5% specificity. The diagnosis of phenotype C (HA + AMH) had 96.15% sensitivity and 94.5% specificity. The diagnosis of phenotype D (OA + AMH) had 100% sensitivity and 94.5% specificity. The diagnostic performance of AMH replacing PCOM in the Rotterdam criteria had high sensitivity (97.4%) and specificity (90.67%). The accuracy of AMH replaced PCOM for PCOS-Rotterdam was 90.67% (Table 4).

Discussion
The growing body of evidence suggests the key role of AMH in the pathogenesis of PCOS [14]. This study investigated the utilization of AMH alone and using AMH instead of PCOM in PCOS diagnosis. Briefly, this study reported that AMH alone had low sensitivity and specificity, but replacing PCOM with AMH in PCOS diagnostic criteria had high diagnostic accuracy.  Although many studies have been conducted on the diagnostic value of AMH and PCOS in recent years, the results of these studies are inconsistent. Sample size, diagnostic criteria, PCOS phenotypes, geography, ethnicity, data from the infertile group, age-related decline in AMH levels, and lack of standardization of laboratory assays for AMH may all play role in the results [22]. Apart from the others, the diagnostic validity of using AMH alone or in place of ultrasonography in PCOS criteria was investigated in different diagnostic criteria and phenotypes.
Women with PCOS had two to four times higher AMH levels than healthy women, according to studies [17,18]. Similarly, AMH levels of PCOS women were significantly higher than nonPCOS women, and phenotype A was the most common type with the highest AMH levels in our study. Pellat et al. found that AMH production is much higher in PCOS patients when the amount of AMH per granulosa cell is examined [23]. They stated that the high AMH cannot be explained only by the increased number of follicles, but AMH also reflects ovarian dysfunction [23]. Some studies showed the relevance of AMH levels with clinical features of PCOS such as hyperandrogenism, oligoovulation, and antral follicle count [19,20]. Consistent with the literature, our data showed that serum AMH level was related to AFC, PCOM, clinical/laboratory hyperandrogenism, oligoanovulation, menstrual cycle length, and LH levels. The correlation of AMH with LH and menstrual cycle length suggested to us the possible role of AMH on GnRH pulsatility of PCOS women [24].
Many studies have been conducted on the use of AMH alone in the diagnosis of PCOS. Unfortunately, the AMH values found are very diverse and far from being uniform value. Most of the literature on this topic is from studies with Diagnostic Systems Lab (DSL) or Immunotech (IOT) tests, which are no longer marketed [22]. At an AMH cut-off level of 8.4 ng/ml, Pingy et al. found a sensitivity of 67% and specificity of 92% [18]. In the study by Hart et al., the cut-off value of 4.2 ng/ml showed a low sensitivity (53.1%) and specificity (69.1%) [25]. Woo et al. observed 75.9% sensitivity and 86.8% specificity (cut-off: 7.82 ng/ml) [26]. The meta-analysis by Iliodromiti et al. [15] evaluated ten studies that used the Rotterdam criteria and IOT assays. They found a sensitivity of 82.8% and a specificity of 79.4% at a cut-off value of AMH of 4.7 ng/ml. Apart from the Dewally study (92% sensitivity and 97% specificity cut off: 4.9 ng/ml) [27], most studies found low sensitivity and specificity for the utility of AMH alone for PCOS diagnosis.
Despite using the same diagnostic criteria and assay, the study's sensitivity and specificity differed. Our findings are in line with those of another study conducted in our country by Sahmay et al. [9]. These results suggest that phenotypes and ethnicity are important factors. Most of the studies used Rotterdam diagnostic criteria, but few used NIH or AES criteria [25,31]. In our study, we investigated the diagnostic capacity of AMH alone in all three diagnostic criteria of PCOS (Rotterdam, AES, and NIH). The ROC analysis for AMH alone (4 ng/ml) showed satisfying AUC values for all three diagnostic criteria (ROT; 0.866, AES: 0.857, NIH: 0.825), but low sensitivity and specificity. In addition, when searching the AMH alone for PCOS phenotypes, only phenotype A had a satisfactory AUC value. In agreement with the study of Li et al. [32] we found that AMH alone was not suitable for the diagnosis of all types of PCOS patients.
There are few studies exploring PCOM replaced by AMH in Rotterdam criteria [8,13,31]. Sahmay et al. reported high diagnostic accuracy with 96% sensitivity and 100% specificity (3.8 ng/ml) [13]. Saxena et al. noted sensitivity of 86.67% and specificity of 71.11% at a cut-off value of 3.44 ng/ml [8]. Two studies used DSL assay [13,31], but Saxena et al. used third-generation assay. The serum AMH levels decline with increasing age. For studies on AMH to be comparable, the age range of patients is essential since AMH decreases with aging. Like Saxena et al. we studied under 35 years of age, but the Eliartson study had an older non-PCOS group [8,31]. In our study, replacing AMH (> 4 ng/ml) with PCOM in Rotterdam diagnostic criteria had a high sensitivity (97.4%) and specificity (90.67%). Also, AMH instead of PCOM in Rotterdam had a high diagnostic capacity for PCOS phenotypes.
The existence of so many different measurement methods is an obstacle to the determination of a uniform cut-off value. For this reason, authorities stay cautious until the standardization of laboratory assays [13]. A previous study showed a high correlation between the IOT and DSL assays, but the values in the DSL assay were about half that of the IOT assay [33]. Studies have revealed good correlations between new-generation assays, but the cut-off levels obtained from new assays need to be verified in the studies [34]. Replacing PCOM with AMH in Rotterma criteria seems a useful diagnostic tool, but studies with new-generation assays are needed for the comparability of results.
The limitation of this study is that it was designed retrospectively. Since the study was not an epidemiological study and was conducted only on patients presenting to our clinic, the results of PCOS prevalence do not reflect the general population. Our study has the advantage of same race and geography with women under 35 years. Moreover, we collected data from regular records and excluded women with missing data.

Conclusion
AMH cannot be used alone as a single diagnostic method, but it can be replaced with PCOM in conventional PCOS diagnostic criteria with high sensitivity and specificity. These results suggested that AMH could be used instead of PCOM when the ultrasound is not available. Although AMH is a promising diagnostic tool for PCOS, studies with different phenotypes, and age groups, as well as laboratory testing are needed.
Author contributions All authors contributed adequately to this study and approved its submission. FV: study design; concept; supervision of analyses; writing; read and approved the final version of the manuscript. BV: Conceptualization, Methodology, Supervision, critical revision of the manuscript for important intellectual content; read and approved the final version of the manuscript. EK: Data curation, software, read and approved the final version of the manuscript. ADEC: Writing and editing; read and approved the final version of the manuscript. IY: Data curation; read and approved the final version of the manuscript.
Funding No funding was received for this study.

Data availability
The datasets created and/or analyzed during the current study are subject to the permission of the relevant hospital and are therefore not available to the public, but are available from the corresponding author on reasonable request.

Conflict of interest
There are no conflicts of interests that are directly or indirectly related to the research including the following: research grants from funding agencies; honoraria for speaking at symposia; financial support for attending symposia; financial support for educational programs; employment or consultation; support from a project sponsor; position on advisory board or board of directors or other types of management relationships; multiple affiliations; financial relationships, for example equity ownership or investment interest; intellectual property rights (e.g., patents, copyrights and royalties from such rights); holdings of spouse and/or children that may have a financial interest in the work.
Ethical approval This study was conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from all patients. Ethical approval was obtained from the local Ethics Committee (HNEAH-KAEK, 2021/293-3343).