Abbreviations
Polycystic Ovary Syndrome (PCOS)
Follicle Number Per Ovary (FNPO)
Hyperandrogenism (HA)
Oligo-Anovulation (OA)
Polycystic ovarian morphology (PCOM)
National Institute of Health (NIH)
Androgen Excess and Polycystic Ovary syndrome societies (AE-PCOS)
European society of human reproduction and embryology (ESHRE)
American society of reproductive medicine (ASRM)
Ovarian volume (OV)
Transforming growth factor β (TGF-β)
Antral follicle count (AFC)
Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)
QUADAS2 tool (quality assessment for studies of diagnostic accuracy)
Ultrasound (USS)
Transvaginal scan (TVS)
Transabdominal scan (TAS)
Body mass index (BMI)
Immunotech (IOT) Diagnostic system Diagnostic Systems Laboratories (DSL)
Second-generation ELISA (Gen II)
Introduction
Every year, many women are informed that they have polycystic ovary syndrome (PCOS), a disorder linked to infertility and a possibly increased risk of type 2 diabetes. However, PCOS is frequently a questionable diagnosis since its features are like those of other conditions, and the diagnostic standards are still vague, leaving room for misinterpretation and inconsistent application [1,2]. Furthermore, clinician doubtfulness may prolong the diagnosis of affected women [2] and lead to the categorization of these women as having a chronic ailment they do not have [1], both of which have considerable effects on female psychology [3]. Also, with the realization that women with PCOS are susceptible to metabolic syndrome and its associated comorbidities, there has been in recent years a growth in awareness of this illness in the general population and the various medical sectors [3].
PCOS is an endocrine disorder characterized by hyperandrogenism (HA) and ovulatory dysfunction, including oligo-anovulation (OA) and polycystic ovarian morphology (PCOM) on ultrasound examination [4]. The syndrome was described for the first time in the literature by Stein and Leventhal in 1935 as a laparoscopic finding, seen in hirsute women. It is one of the most common endocrinopathies affecting females and the most common cause of ovulatory infertility. Worldwide it affects between 5–20% of women during reproductive age [5]. The prevalence varies according to the definition and reference population [6]. Based on the original 1990 US National Institutes of Health (NIH) diagnostic criteria, prevalence is between 5-9%. This rate increases to 5.5-19.9% when diagnostic based Rotterdam criteria is used [5].
The Rotterdam criteria recommend the presence of two out of the three characteristic PCOS features [7]: 1) Oligo-anovulation (OA); 2) Clinical and/or biochemical signs of HA and 3) Polycystic ovarian morphology (PCOM).
All the criteria proved the requirement of exclusion of related disorders such as thyroid dysfunction, hyperprolactinemia, non-classic adrenal hyperplasia, androgen-secreting neoplasia, and other conditions that could present with features of HA or OA [8].
Because of technological improvements, such as high-frequency vaginal probes and image-enhancing software, which have improved resolution and measurement capabilities, the definition of the diagnostic characteristics for polycystic ovaries by ultrasonography has generated debate. Illustrations from the past were based on transabdominal ultrasound [9]. Adams et al. [10] In 1985 described PCOM as an ovary with ≥ 10 follicles (measuring 2–8 mm in diameter) arranged at the periphery around a dense stroma. Many studies used this criterion to diagnose PCOS [10]. Revision of the requirements was made based on transvaginal techniques. The threshold of 12 follicles number per ovary (FNPO) has been used as a diagnostic threshold for PCOM since 2003 [9]. With the development of ultrasonography and development of high-resolution technology (>6 MHz), the following recommendation is advised by the AE-PCOS society in 2013 [11,12]: 1) The threshold for FNPO used to diagnose PCOM in women aged 18–35 years is raised to ≥ 25 follicles per ovary when using a transducer frequency ≥ 8 MHz; 2) ovarian volume (OV) > 10 ml is used instead of FNPO in case of poor image quality that does not allow a precise estimation of the follicular number; 3) OV has less diagnostic potential for PCOM compared with FNPO; 4) ovarian blood flow and measurement of the ovarian stroma have no diagnostic advantages; and 5) with the addition of automated software and offline analysis of saved data, three-dimensional ultrasound has shown only a modest reduction in variability, with the added drawback of more time and money.
The requirement of a transvaginal ultrasound scan for the diagnosis of PCOS has the following concerns:
- First, it is an expensive tool.
- It may not be available in some places, especially primary health care centres.
- Need experience and time to perform.
- Interobserver variations.
- Different standardization with advanced ultrasound equipment.
- It may not be acceptable to teenagers, while most patients who need diagnosis are teenagers.
- A specific period is needed after menarche to apply the criteria of PCOM.
- Using an abdominal ultrasound scan to diagnose PCOM is more complex,
- especially when there is a thick abdominal wall.
- Racial difference in defining PCOM.
All these concerns make finding another diagnostic tool an attractive idea.
Serum anti-Mullerian hormone (AMH), also known as Mullerian-inhibiting substance, is a dimeric glycoprotein and a member of the transforming growth factor β (TGF-β) family of growth and differentiation factors. It is exclusively produced by the granulosa cells in the gonadotropin resistant phase [13]. The production starts after recruitment from the primordial follicle pool until the early antral stage when the exhibition is the maximum [14]; when a follicle's diameter reaches between 8 and 10 mm, production stops [13]. This pattern in output suggests the use of AMH to assess the number of small follicles; this suggestion was confirmed by Pigny et al. [15] as they confirmed that there is a significant association between blood AMH and antral follicular count (AFC), which has been proven by numerous clinical trials [15].
It is seen that the recruited follicles in the follicle pool produce AMH, and this hormone may operate to prevent the recruitment of the remaining primordial follicles. This effect results from AMH's inhibitory action on FSH, suppressing the hormone's stimulatory effects on the growing follicles. The role of AMH in conditions linked to ovarian dysfunction is of particular interest since it may be implicated in first and cyclic recruitment [13]. In addition, serum AMH levels appear to be significantly elevated in women with PCOS [16]; this observation points to AMH as a factor involved in the mechanism of disruption of follicular growth [17].
The factors affecting serum AMH levels include [11]:
- Age: immediately after birth, there is a peak, followed by a steady increase through age 9, then another spike to 25 years. From then there is a continuous fall to undetectable levels at 50–51 years.
- High interindividual variation is related to the number of antral follicles.
- The level is relatively stable during the menstrual cycle as the dominant follicles and corpus luteum do not secrete AMH.
- The use of hormonal contraceptives may affect AMH levels.
- Pregnancy is associated with changes in AMH levels.
- It is related to body mass index.
PCOS cases have been found to have 2 to 3-fold greater serum AMH levels than women with typical menstrual periods [16] . In addition, AMH/AFC ratios were noticeably greater in PCOS-affected women than in non-PCOS-affected ones; this shows a more outstanding production of AMH per follicle [18] . This finding suggests that AMH could be used as a PCOS marker and diagnostic tool for PCOS [19].
AMH has the following features, which make it an attractive possibility [20-23]:
- First, it is not subjected to technical development.
- Not operator-dependent, there are stable, standardized assays for measuring AMH levels that reduce inter-laboratory variability, especially with newer automated tests.
- It is correlated to the AFC.
- There is a correlation between HA and oligomenorrhea.
- It can be measured independently of cycle day.
- The test can be used where ultrasound scans for diagnosis are unavailable or cannot be done among women who are obese or virgin cases.
Iliodromiti et al. [19] 2013 in a systematic review and meta-analysis determined that AMH is an adequate first-line examination in diagnosing women with PCOS. Still, different studies did not reach an agreeable cut-off point to be used instead of the ultrasonic criterion to discriminate PCOS women from normal ones. Therefore, the recommendation from the international evidence-based guideline for assessing and managing PCOS is that serum AMH levels shouldn't yet be used as a test to diagnose PCOS or as a substitute for detecting PCOM [8].
In the present review, the author will systematically review the previous literature to answer the following points:
- The evidence of using AMH for diagnosing PCOS and comparing it to ultrasound.
- The question of why AMH does not replace ultrasound for the diagnosis of PCOS, although it is a simple blood test, mostly accepted by the patients, in comparison to the vaginal ultrasound?
- How to achieve a diagnostic cut-off point for the diagnosis.
Methods and design
A systematic review was conducted to compare the ability of the serum AMH to ultrasound for diagnosing PCOS.
Search strategy
Literature search
This review was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [24] which includes a 16 item checklist to improve the quality of the literature review.
Approval
This is a systematic review of the primary studies; no individuals were involved in data collection. A low-risk ethical approval was taken from the Ethical Committee of The Faculty of Life Science and Education University of South Wales, United Kingdom.
Search question
Was developed following the PICOS framework.
- Population: Women having PCOS.
- Intervention: Serum AMH testing for PCOS diagnosis.
- Comparison: Ultrasound used for PCOS diagnosis.
- Outcome: The accuracy of AMH in diagnosing PCOS, tests for the accuracy are the sensitivity and the specificity.
- Study design: all types of studies were included.
Databases
Published literature on the subject was searched in PubMed and Google Scholar starting ten years back until 2022. The search was limited to studies done on humans and those published in English.
The keywords
The keywords for the search to cover the topic of interest were identified, and truncation was employed. The following were the searched keywords:
Polycystic ovary syndrome AND Anti-Mullerian hormone, Anti-Mullerian factor AND ultrasound AND diagnosis. In addition to these words, the basic Boolean search operators were used as the following:
(“Anti-Mullerian hormone” OR “Mullerian Inhibiting Substance” OR “Mullerian
Inhibitory Factor” OR "Anti-Mullerian Hormone"[Mesh]) AND
("Ultrasonography"[Mesh] OR Ultrasound OR Sonography OR Ultrasonography OR Echo) AND (polycystic ovarian syndrome OR polycystic ovary disease OR “Polycystic Ovary Syndrome"[Mesh]).
A data search was done, the investigation of PubMed resulted in 174 titles, and the Google Scholar examination resulted in 886 titles. A total of 1,060 articles were retrieved.
Results from each database were imported to the endnote reference management tool, and then to an excel sheet and duplicates were removed.
Inclusion and exclusion criteria
All articles were retrieved at Excel sheet, and the inclusion and exclusion criteria were applied to retrieve the needed reports. In the first step, titles were screened, and abstracts were screened in the second step. Finally, contact with the authors was performed to retrieve unavailable data.
The following were the inclusion criteria:
- All studies (Prospective, retrospective, and cross-sectional) reporting the use of AMH or ultrasound for diagnosing PCOS were included.
- The use of AMH and ultrasound for PCOS diagnosis.
- Use of Rotterdam criteria for PCOS diagnosis.
- All females of reproductive age.
- All age groups.
- Comparative studies in which each woman received both the AMH and ultrasound testing.
- Use of sensitivity and specificity testing for AMH diagnostic ability.
The studies were excluded for the following reasons:
- Studies not written in English.
- Review articles.
- Articles published as textbooks and tables.
- Use of a comparison group diagnosed as PCOS cases previously.
- Studies for specific groups.
- Studies that were not measuring the sensitivity and specificity of the AMH test for diagnosing PCOS.
Quality assessment
PRISMA statement
The mentioned databases were searched, and duplicate papers were removed. Titles were screened for their relevance, and then the abstracts were checked to ensure their relevance. Inclusion and exclusion criteria were applied. The following figure shows the literature inclusion and exclusion at each stage (PRISMA statement): The process of search and screening is summarized in Figure 1.
Figure 1. The PRISMA flow diagram for the comparison of AMH to ultrasound for the diagnosis of PCOS.
QUADAS2 toll for quality assessment
The quality of included studies was assessed by the QUADAS2 tool (quality assessment for studies of diagnostic accuracy). It is a quality assessment tool recommended to be used to determine the risk of bias and applicability of the primary diagnostic studies; it consists of four key domains [25]: patient selection, index test, reference standard and flow and timing.
Figure 2 shows the results of the assessment.
Figure 2. Quality assessment of the studies using the QUADAS2 tool.
Data extraction
One reviewer selected six eligible studies. A double check was done for eligibility. Data were extracted from eligible texts using a standardized piloted form.
The following data are extracted from the included articles using a predesigned Excel spreadsheet:
- Regarding the study characteristics:
- Title, first author, year of publication, type of study, study region, period of study, total cohort, and study group age.
- Regarding the ultrasound:
- The route of ultrasound (USS), the frequency of USS, diagnostic threshold, and criteria for
- AMH assay:
- The type of the used assay.
- Results:
- AMH cut-off point for PCOS diagnosis.
- Sensitivity and specificity of AMH for PCOS diagnosis.
- The receiver operator curve finding the ability of AMH to predict PCOS.
Results
Characteristics of the included studies:
Table 1. The characteristics of the included studies.
The different used AMH assays. In the following figure, the different used methods for immunoassay are displayed:
Figure 3. The different used AMH assays.
The accuracy of AMH measurements for the diagnosis of PCOS. AMH sensitivity for PCOS
The results of the papers regarding the sensitivity are shown in the following figure:
Figure 4. The sensitivity of AMH.
The highest sensitivity (94.6%) was seen in the study by Tina B. Eilertsen et al. [26]. Gurkan Bozdag's investigation revealed an AMH sensitivity of 39% for the diagnosis of PCOS [27].
AMH specificity for PCOS. Figure 5 explains the different specificity values of the studies.
Figure 5. The specificity of AMH.
The specificity of AMH for PCOS diagnosis was approximately similar for Dewailly et al. and Eilertsen et al. In both studies, AMH was 97% specific for PCOS diagnosis [26,28]. The lowest specificity (74.7%) was obtained in the study of et al. [29].
AMH cut-off point
Figure 6. The AMH cut-off point.
Different cut-off points were found; the range was between 2.52 and 5.03 ng/mL [26-31].
The accuracy of ultrasound for the diagnosis of PCOS.
The number of follicles used for PCOM definition:
The different values used to define PCOM are shown in the following figure:
Figure 7. The number of follicles used to define PCOM.
Different follicle numbers were used to describe the PCOM; three studies have used the threshold of 12 follicles [26,29,30]. Bozdag has used both thresholds of 12 and 20 follicles as a diagnostic threshold for PCOM [27].
Different ultrasound approaches and frequency:
Table 2. The different ultrasound (USS) routes and frequencies.
The association between AMH levels and the different PCOS phenotypes. The prevalence of each phenotype
Figures 8, 9 and 10 clarify the different ratios of PCOS phenotypes, from three studies.
Figure 8. Distribution of phenotypes according to Lauritsen et al. [30].
In the study by Lauritsen et al. [30], phenotype C (ovulatory) was the most prevalent form in this group. On the other hand, phenotypes A and B constitute 10% of the total group.
Figure 9. Distribution of phenotypes according to Bozdag et al. [27] 2019.
Bozdag et al. [27] also demonstrated the dominance of phenotype C (46%) and equal distribution between the phenotypes A and D.
Figure 10. Distribution of phenotypes according to Mahajan and Kaur [31].
Phenotype A was the most prevalent form in the study by Mahajan and Kaur [31], next to that are phenotypes C and D, with approximately equal distributions. Phenotype D showed the lowest prevalence (3%).
Discussion
After more than 80 years of Stein's description of PCOS, the most frequent endocrine disorder in women of reproductive age and the most common cause of anovulatory infertility, it is plagued with complex diagnostic issues, delayed diagnoses, and suboptimal treatment plans [32]. Since it is a syndrome, as its name implies, no single test would be able to diagnose it, leading to the development of several criteria throughout time. Also, many difficulties arise as its initial manifestations start early during adolescence [7]. The diagnosis is still a challenging issue and there are still knowledge gaps in various medical specialties, including reproductive medicine [20].
In this systematic review, data were extracted and summarized regarding the currently available evidence concerning the accuracy of AMH in diagnosing PCOS compared to the ultrasonic criterion used for the diagnosis. However, the previous results conflicted regarding AMH's diagnostic ability in discrimination between PCOS and normal women [19]. In our review, an analysis of the possible causes for the conflicting results regarding the diagnostic ability of AMH for PCOS will be discussed.
AMH is a surrogate marker of classical HA [33] and several other studies conclude that the concentration of AMH is associated with gonadotropin release disturbances seen in PCOS, and it is related to the disease severity and morphological and hormonal changes seen with it [33]. As AMH levels show a positive correlation to insulin resistance, LDL cholesterol levels, and a negative correlation to HDL cholesterol level [34], it is seen that AMH level also indicates the risk of cardiovascular disease in PCOS patients [35]. Due to its role in ovulatory dysfunction, AMH has also been hypothesized to be a diagnostic tool for the condition [36]. AMH acts paracrine to prevent granulosa cells from undergoing FSH-induced aromatase activity, which promotes the formation of a dominant follicle [37].
AMH serum levels changes with age. Girls experience a "mini puberty" of the neonate, which is confirmed by a peak soon after birth. There is after that a steady increase till around nine years of age. Then, during the pubertal years (9 to 15 years), there is a decline, followed by a second increase phase that reaches a high peak at age 25. Following this, there is a gradual fall until the levels are undetectable at an average age of 50 to 51 years, which corresponds to menopause [38].
During follıculogenesis, AMH is a glycoprotein produced by the granulosa cells of the pre-antral follicles and the small antral follicles. It is known to be a proxy for PCOM as both (AMH and PCOM) are related to the number of small follicles in the ovary [39]. Therefore, PCOM and AMH assessment should obtain similar results if both are done correctly. Hence, for this AMH is recognized as an attractive alternative diagnostic option [11,40]. Pigny et al. [15] also confirm that the blood levels of AMH are higher in the PCOS group compared to normal women.
As follicle development fluctuates over a person's life, increasing during adolescence and declining until menopause, when oocytes are depleted, obtaining diagnostic values, PCOM and AMH require age-specific cut-offs [15].
In a study by Laven et al. [16] regarding infertile cases, the PCOS group showed the highest AMH level compared to other subjects [16]. This finding of the elevated serum level of AMH associated with PCOS has attracted researchers to study its association with the disorder’s pathophysiology. Sahmay et al. [41] found a threefold increase in AMH levels in PCOS compared to non-PCOS. An association has been seen by Fanchin et al. [42] between AMH and ovarian follicular status. Also, a correlation between serum AMH levels and the clinical manifestations of PCOS has been confirmed [43]. AMH level is related to the severity of the syndrome, as the highest AMH level is found with type A, which is the most severe phenotype of PCOS compared to other phenotypes and women without PCOS [44].
The development of AMH assays started in 1990 [45], and after that, the companies continued to develop the products to achieve better results. Since 2004 and until 2010, two ELISA systems (first generation) have been developed; these are [11]:
- Immunotech (IOT) (Marseille, France).
- Diagnostic Systems Laboratories (DSL) (Webster, Texas, USA).
It is recognized that the results of these sera were correlated, but the results obtained from the IOT assay were 5.05 times higher than those from the DSL method [46].
After that, The Beckman Coulter Company owned both and developed a second-generation ELISA (the AMH Gen II assay), gradually replacing the two methods [38]. Then a protocol system for the AMH Gen II, developed in 2013, that includes a predilution step before adding the sample to the ELISA plate [47].
Also, new AMH ELISA assays were developed by Ansh labs [48]:
- An ultra-sensitive AMH ELISA.
- Pico AMH assay for samples with low AMH levels.
The assays mentioned above are manual, so the results show variability between the laboratories. Nowadays, fully automated measurements have become available, with a lower inter-laboratory variation [49]:
- Elecsys, AMH from Roche.
- Access AMH from Beckman Coulter.
Regarding the studies performed early, manual testing was the used method as shown in Table 1 [26,28,29,30]. After that, there was a recommendation from the international evidence-based guideline for assessing and managing PCOS to standardize the assays used to establish cut-off points of different ages and ethnicities [11]. As a result, the three recent studies have used automated assays [27,31].
The following steps are advised to get reliable results [8]:
- Sample handling.
- Efficient transport system.
- Optimum storage conditions.
- There is also a clear need for an international reference standard for AMH.
The results were reported in nanograms per milliliter or picomoles per liter. All data in this review were shown in nanograms per milliliter to avoid heterogeneity in the interpretation of the results (1 ng/mL corresponds to 7.14 pmol/L) [50].
The advantages of AMH measurements:
- It is a stable, standardized assay with the new automated assays [51].
- The use of AMH is sensitive and specific for diagnosing PCOM [28].
- Increased serum AMH levels of > 35 pmol/L (or > 5 ng/mL) has been proposed for the PCOM definition in diagnosing PCOS. It is a more sensitive and specific marker than follicle count in ultrasonographic examination [28].
- The level is associated with the severity of PCOS symptoms [52,53].
- There is growing proof that AMH tests will be more effective in detecting PCOM with enhanced assay standardization and established cut-off levels or thresholds based on extensive validation in populations of various ages and ethnicities [8,54].
By reviewing the databases, published articles regarding the use of AMH for PCOS diagnosis have increased in recent years; this is because of the heterogeneity of the ultrasound description of PCOM, which is included in the classification of PCOS and the awareness of the importance of AMH in the pathogenesis of PCOS and its diagnostic value. Many investigators are trying to find a replacement for the follicular count measurement, which has many obstacles [39].
This review includes comparative studies in which each woman was tested for both the AMH and ultrasound to decrease the risk of bias. The quality of included studies was assessed by the QUADAS2 tool. Three studies were scored as highly biased. In the survey by Eilertsen et al. [26] infertile women were excluded, which is an inappropriate exclusion that could bias the results. Lauritsen et al. [30] selected healthcare workers with the exclusion of contraceptive pills users, which is sometimes used for PCOS management. On the other hand, Mahajan et al. [31] excluded the fertile women from the study cohort and has generated his findings for all Indian women. Regarding the index test, all the articles were highly biased, as the used threshold of AMH was not pre-specified, also a high bias score was the result for all the studies, as the reference standard (ultrasound) did not correctly classify PCOS. Regarding flow and timing, the bias here was low for all the studies because both tests were performed at the near interval, all patients had received the reference test, and the cohort was included in the analysis.
Age of the included cases was different. For example, in the study by Dewailly et al. [28] the youngest age was 19.9, and for the Bozdag et al. [27] study it was 18 years. Including all women with different characteristics in one group seems nonlogical. Dewailly et al. [28] have clustered the whole cohort into other age groups and analyzed these bases. Also, Mahajan et al. [31] have adjusted the results of AMH measurements to the age and body mass index. There was a conclusion by Lauritsen [30] that the application of Rotterdam criteria for young women less than 30 years will result in a three-fold increase in PCOS.
Although several thresholds for elevated AMH readings (from 3 to 5 ng/mL) have been suggested, serum AMH alone has been thought to have relatively low diagnostic sensitivity, increasing in only 70 to 80% of PCOS cases (using Rotterdam or AE-PCOS criteria). A systematic review and meta-analysis of 10 studies concluded that a value of 4.7 ng/mL may be used to diagnose PCOS with a sensitivity and specificity of 76.7% and 79.4%, respectively (AUC = 0.829) [19]. An elevated AMH value had a 76% sensitivity for the diagnosis of PCOS documented by Cassar et al. [55].
Different thresholds with varied sensitivities and specificities have been proposed, ranging from 2.8 ng/mL to 8.4 ng/mL [54]. This result is comparable to the cut-off point for the studies included in the review, that ranges from 2.52 ng/mL to 5.03 ng/mL [9,28,38,57,58].
Lauritsen et al. [30] have shown the lowest cut-off point of 2.52 mg with 91.80% sensitivity, 98.10% specificity and an AUC of 0.994. This may be due to selection bias, as all the study participants were healthcare professionals. Furthermore, the results may have been skewed by the omission of research participants who used hormonal contraceptives, thus perhaps eliminating women with PCOS symptoms using hormonal contraceptives for the management of PCOS-associated symptoms. Also, the study by Eilertsen et al. [26] resulted in a cut-off point of 2.8 mg/dL, a sensitivity of 94.60% and specificity of 97.10% and an AUC of 0.992. The fact that all the women included had a history of childbirth may have caused a selection bias, as one of the main effects of PCOS is infertility. Also, the findings of the previous articles, particularly AMH, may show that PCOS women who become pregnant have a more favourable variant of the syndrome [59]. The other three studies show comparable results, as depicted in Table 1. These results also align with results provided by Pigny et al. [15] and others [27,29,31].
In the study by Dewailly et al. [28] the clustering of the control group was done not to include PCOM in the control group; the specificity of AMH was 97% and the sensitivity 92%. The AUC was 0.973, and the threshold was 4.9 ng/mL, which is lower than the previously reported threshold (8.4 ng/mL) by Pigny et al. [15] which may be due to the inclusion of PCOM in the control group. These results indicate that AMH cut-off values can be established with higher specificity and sensitivity and at lower levels than previously predicted [15,28]. The lowest sensitivity has been observed in the study of Bozdag et al. [27] which may be due to the use of frozen samples and the possibility of including women with PCOM in the control group. In addition, the study was performed on women working in one institution, which may be a factor for selection bias.
Most previous studies look for cut-offs at the 95th centile, which is not an appropriate biological cut-off. Cut-offs for long-term health concerns make more sense rather than relying on percentiles. Therefore, it is crucial to conduct more studies on how AMH clusters with other PCOS characteristics and how AMH affects long-term health effects [8].
Transvaginal scanning is the recommended approach for patients with PCOS, because it has a high resolution [8]. It was the most frequently used approach in the included studies. In two studies, the transabdominal scan was used when the transvaginal approach could not be used [27,29]. One of these two studies did not use ovarian volume as a diagnostic feature for PCOS. However, it is recommended for a transabdominal approach when the counting cannot be done previously [38].
The results of the included studies have demonstrated the following different criteria for the establishment of PCOM:
- Ovarian volume, with different sizes to define PCOS (7-10 mL). The cut-off point proposed at the ESHRE consensus was 10 mL [7], but after the researchers noticed different thresholds to diagnose PCOS, it appeared that other factors could affect the ovarian volume, such as age and ethnicity, body build and age. For these reasons, the cut-off point of 10 mL is accepted if group-specific measurements are unavailable [11].
- The mentioned numbers were 12-19-20 follicles/ovary). The ultrasound criteria for diagnosis have been continuously changed with the development and improvement of ultrasound equipment and the understanding of ovarian physiology. This was obvious as our included studies have used different threshold values for PCOS diagnosis; some have used the criteria set by the 2003 Rotterdam consensus [7], and others have taken into account more number of follicles. Mahajan et al. [31] should have mentioned the threshold used in their study. The conflict in the threshold number was due to ethnic variation, difficulty in defining diseased cases and the difficulty in determining normality, as the presence of PCOM alone, without other features of PCOS, has raised the suspicion of considering it as a typical feature or a mild form of PCOS. Dewailly et al. [28] excluded cases with PCOM in the control group and was considered a separate group.
- Size of the small follicle between 2-9 or 10 mL.
- The validity of using AMH to identify PCOM or PCOS with criteria of 12 and 20 AFC per ovary in an unselected patient group is being evaluated in the study by Bozdag et al. [27]. In the same study population, applying the 20 AFC per ovary threshold reduced the prevalence of the condition by 50%.
After the inclusion of ultrasound in the 2003 Rotterdam consensus [7], the studies have recognized the need to re-evaluate the FNPO ≥ 12 as threshold. As a result, some researchers advise not to consider the FNPO as a diagnostic criterion [60]. At the same time, others have suggested reviewing the cut-off value [56]. These findings make the ultrasound diagnostic criterion questionable.
The recommendations from the international evidence-based guideline for the assessment and management of PCOS concerning ultrasound have been established to ensure homogeneity of the results; the proposals include the last menstrual period, the used approach, transducer frequency, FNPO, ovarian volume, and any cysts [8]. Although it is recommended not to use ultrasound for PCOS diagnosis within eight years from the menarche, there should be an update on the cut-off point for diagnosing PCOM with advancing ultrasound techniques. Also, it is recommended to use vaginal ultrasound over the transabdominal one; the ovarian volume of ≥ 10 mL has been used when the resolution is low, and for transvaginal scan with a frequency of 8 MHz, the threshold for FNPO is ≥ 20.
The defining characteristics of PCOS result in a variety of phenotypes. In epidemiological research, in unselected (unbiased) groups, the phenotypic distribution of PCOS is 40–45% for phenotypes A and B combined, 35% for phenotype C, and 20% for phenotype D [58], this is in comparison to studies involving only PCOS patients who exhibit a higher percentage of phenotypes A and B, showing significant referral bias [61] .
Phenotyping of PCOS cases allows epidemiologists to study the different group outcomes in order to find specific correlations. It is also helpful for the clinician to differentiate the PCOS groups at high risk for metabolic disarrangements [58]. The observations show that the classical phenotypes (A and B) constitute 75% of cases presenting with metabolic disorders, highlighting that the two phenotypes are the most severe syndrome [62].
Lauritsen [30] and Bozdag [27] found phenotype C as the most common phenotype (83.7% and 46.1%, respectively), reflecting a selection bias in the chosen sample, as both studies have selected workers from one institution for study. As the group selected for the study done by Mahajan was infertile, the majority of the cases (48.1%) were of the phenotype A [31]. This is comparable to the published data which state that the classic PCOS phenotype A constitutes more than half of PCOS cases, while the other three share the other half with equal percentages [58]. A study performed in Indonesia has shown a higher prevalence of phenotype D, which points to the racial difference in the presentation of PCOS cases [44].
It is clear from the previously performed studies that AMH levels increase in accordance to PCOS severity. As type A is the most severe form of PCOS, it displays the highest AMH levels when compared to the other phenotypes [40]; this result is in correlation with the results from two studies that measure AMH in relation to the phenotypes [27,31]. Although phenotype D is considered the mild form of PCOS, the level of AMH was higher than that seen with phenotypes B and C in the study by Bozdag et al. [27], and the story was similar PCOS types C and D in one study [31]. Results from the earlier studies explain that the ovulatory PCOS display the lowest AMH levels, indicating the role for AMH in the ovulatory disorder [37], and this observation has opened the discussion for the position of AMH in PCOS pathogenesis [39].
An ovarian ultrasound is not needed to diagnose PCOS in patients with irregular menstrual cycles and hyperandrogenism. However, ultrasound will still reveal all the phenotypes of PCOS [63].
Histological examination of the PCOS ovaries shows the same number of primordial follicles and doubling of the subsequent developing follicles [64]. Inhibin levels were normal although the number of small follicles was increased, indicating that follicles are atretic [16]. In 2003, Fanchin et al. [42] confirmed that there is a strong correlation between AMH and early AFC, which is more than that observed with inhibin estradiol, FSH and LH on the third day of the cycle.
The strongest correlation is found between PCOM and AMH; the increase in AMH is due to many preantral and small antral follicles, which are the sole producers of AMH. The high level of AMH seen in PCOS is also due to increased AMH production by each follicle [23].
In the study done by Dewailly et al. [28], there was an association between AMH level and PCOM; it was demonstrated that AMH is superior to the AFC in defining PCOM. The result was obtained after excluding women with PCOM more than 12 follicles as compared to the control group [28]. The excluded women had high AMH blood levels, like those with HA or oligo-anovulation. This finding suggests that the presence of PCOM represents a milder form of PCOS [65].
In 2011, Dewailly et al. [28] suggested a sensitivity of 92% and a specificity of 97% (AUC 0.973) for the cut-off value of 4.9 ng/mL for AMH in the definition of PCOM in PCOS, which is higher than the results shown by Lauritsen [30]: AUC for AMH to recognize AFC ≥ 12 was 0.909 (95% CI: 0.882– 0.936) and the AUC for AMH to identify ovarian volume 10 ml was 0.804 (95% CI: 0.732–0.876). Both studies used different groups; the first one studied the symptomatic group, while the second one studied the fertile group [30]. For the detection of PCOM, the ROC analysis displayed an AUC for AMH levels of 0.87 (95% CI 0.83–0.90). In one study, the used AMH threshold level was 3.31 ng/mL with 53.1% of sensitivity and 93.2% of specificity, displaying the and the lowest sensitivity [27].
AMH levels correlate with the extent of ovarian dysfunction in these women, as represented by elevated LH or testosterone levels [16]. AMH has a strong correlation with the status of ovarian follicles than other hormones, as AMH levels were 12 times higher in anovulatory PCOS patients in comparison to the ovulatory PCOS ones [16,66].
Pigney et al. [15] confirmed a proportional increase in AMH in relation to the length of the menstrual cycle. Women with cycles of regular length exhibit increased AMH levels in correlation with increased cycle length, then the level of AMH increases steadily from oligomenorrhea PCOS women to be high for amenorrhoeic cases [67}.
AMH is a significant inhibitory factor for folliculogenesis, which results in anovulation [23] . Going back to our previous comment, amenorrhoeic PCOS patients display the highest AMH levels, which point out to its role in the pathogenesis of anovulation related to PCOS [68]. Furthermore, the AMH/FSH ratio differs when comparing the different PCOS phenotypes, with a clear association between the ratio and the ovulatory disorder. This confirms the hypothesis of intrinsic granulosa cell dysfunction [36}. A previous study confirmed the positive correlation between AMH and HA[16]. AMH has a positive relation with hirsutism but no correlation with the other clinical manifestation of HA such as hair loss and acne [69]. Several hypotheses address the role of androgens in stimulating granulosa cells to increase their AMH secretion [69].
Pigny et al. [15] confirmed that AMH could be used as a diagnostic criterion instead of the follicle count when accurate ultrasonographic data are unavailable. For the included studies in this systematic review, AMH can replace ultrasound with high sensitivity and specificity as AMH is closely related to HA and ovulatory disturbances, the characteristic features of PCOS [26,28,30]. Sahmay et al. [29] confirmed that the level of serum AMH is a helpful indication for PCOS diagnosis and correlates with established diagnostic standards. However, combining the blood AMH level with HA and/or OA significantly improves PCOS diagnosis and can be used as an objective, well-organized criterion [29]. According to the study by Mahajan [31], PCOS in women of Indian descent may be easier to diagnose when the serum AMH content is significantly higher than 5.03 ng/mL when performed using an automated test. The diagnostic criteria for women of reproductive age may eventually replace PCOM with serum AMH as an essential indication of PCOS [31]. On the contrary, Bozdag et al. [27] concluded that AMH has a low to fair accuracy in identifying PCOS in a general population of women, except for those with all three symptoms (Phenotype A). This conclusion is proper even after changing the definition of PCOM to an AFC of 20 follicles and applying the NIH, Rotterdam-2003, and AE-PCOS criteria [27]. Abbara et al. [67] found that a combination of serum AMH and testosterone identifies PCOS by the new international criteria with an area under the ROC of 0.77 (95% CI 0.67–0.85).
Pitfalls of the present study
The following points are essential pitfalls in this review:
- One reviewer made the search and the selection of the articles.
- Most of the studies were retrospective.
- There was a risk of language bias, as it was limited to studies published in English.
- The references were not tracked.
- Some information needed to be obtained due to the lack of response.
- The Rotterdam criteria were the only ones used to define PCOS in the included papers.
Conclusion
Although AMH is a useful tool, the recommendation from the international evidence-based guideline for the assessment and management of PCOS is that serum AMH levels should not yet be used as a substitute for other tests for the diagnosis of PCOS or definition of PCOM. There is growing evidence that AMH tests will be more reliable in identifying PCOM with enhanced assay standardization and specified cut-off levels or thresholds based on large-scale validation in populations of varied ages and ethnicities [8].
The available data from the included articles are inadequate to support or to decline the use of AMH as a diagnostic tool for PCOS instead of PCOM [15,33,28,70,71]:
- Most of the earlier studies are retrospective.
- The application of different criteria for PCOS diagnosis.
- The use of infertile women as controls.
- The overlap between PCOS and controls.
- Adolescence inclusion. The International guidelines recommend ultrasound for PCOS diagnosis eight years after the menarche.
- The use of low-resolution ultrasound for PCOM detection, as mentioned by some researchers [27,28,30].
- Different definitions for PCOM are used for comparison.
- Ultrasound equipment details were not mentioned in some reports.
- Androgen measurements used immunoassays that have low sensitivity and precision for low levels.
- Some studies use older manual immunoassays for AMH measurement.
- Inconsistent manipulation of AMH samples.
- Requirement of assay-specific interpretation.
The following key issues must be applied in order to establish AMH as a diagnostic test:
- To avoid bias, PCOS cases and controls must be consistently defined, with an appropriate exclusion.
- Adjustment of the Rotterdam criteria for particular age groups to avoid overdiagnosis.
- It is necessary to review the validity of ultrasonography for PCOM diagnosis and update the diagnostic cut-off following the new updates.
- International criteria need to be agreed upon for PCOM diagnosis.
- Strict adherence to the suggested PCOM criteria is necessary.
- International standardization of the assays.
- The new commercial enzyme-linked immunosorbent assay (automated) is used to measure AMH levels to avoid measurement variations.
- The cut-off values must be age-related and relevant to the health outcomes that reflect the clinical features.
- Additional research is required to evaluate the efficacy of AMH for diagnosing various phenotypes and subgroups of PCOS.
The International PCOS Guidelines recommend treating PCOS-related symptoms, including anovulation, hirsutism, and irregular periods, which are essential to diagnosis [32]. AMH-specific antibodies or antagonists, which try to block AMH function, may prove clinically helpful in treating various elements of this condition [72].
Acknowledgment
Thanks to Dr Janine Elson from the University of South Wales, for her assistance in preparing the manuscript.
Conflict of interest statement
The author declares having no conflicts of interest.
Funding
There was no specific funding for the elaboration of this manuscript.