Age‐Independent Serum AMH Levels in Women With PCOS Defined by the 2018 Evidence‐Based Guideline Diagnostic Criteria: A Cross‐Sectional Study

ABSTRACT

Objective

The 2018 evidence‐based guideline revised the follicle count threshold for polycystic ovary morphology (PCOM) from ≥ 12 to ≥ 20, thereby introducing a stricter definition than the Rotterdam criteria. In 2023, anti‐Müllerian hormone (AMH) was incorporated into defining PCOM. Although PCOS‐related symptoms often improve with age, some women continue to exhibit symptoms and meet the PCOS diagnostic criteria even as they age. This study examined AMH patterns across age groups in women who already met the PCOS diagnostic criteria according to either the Rotterdam or the stricter 2018 criteria.

Methods

This cross‐sectional study included 725 women diagnosed with PCOS according to the Rotterdam criteria, of whom 520 also fulfilled the 2018 criteria. Serum AMH levels were compared across age groups: < 25, 25–34.9, and 35–45 years.

Results

Among women meeting the Rotterdam criteria, AMH levels were significantly lower in the oldest group (9.0 ng/mL) than in those < 25 years (11.2 ng/mL, p = 0.032). Meanwhile, among women who met the 2018 criteria, mean AMH levels were 12.5, 12.0, and 10.0 ng/mL in < 25, 25–34.9, and 35–45 year groups, respectively (p = 0.077), with no correlation between age and AMH (r = −0.050, p = 0.178). Additionally, the oldest group showed worse metabolic profiles than the younger groups.

Conclusions

Women who continued to meet the stricter criteria at older reproductive ages showed AMH levels comparable to those of younger patients, and had worse metabolic profiles, supporting AMH as a stable diagnostic marker across reproductive ages in PCOS.

1. Introduction

Anti‐Müllerian hormone (AMH), a 140‐kDa homodimeric glycoprotein of the transforming growth factor‐ß family, plays an inhibitory role in follicular growth by counteracting the actions of follicle‐stimulating hormone (FSH) on aromatase activity [1, 2]. It is exclusively produced by granulosa cells of pre‐antral and small antral follicles in the ovary [3]. Therefore, AMH has been recognized as a useful marker for reflecting the antral follicle pool.

Polycystic ovary syndrome (PCOS), a common endocrine disorder in women of reproductive age, is characterized by irregular menstruation (IM) due to chronic anovulation, hyperandrogenism (HA), and the presence of increased antral follicle counts (AFC) or ovarian volumes. Serum AMH levels are significantly higher in women with PCOS due to an increase in the number of antral follicles or a greater production per follicle [4, 5, 6]. Considering an inhibitory role of AMH in folliculogenesis, its high production in women with PCOS might play a role in the pathophysiology of anovulation and follicular pattern observed in this syndrome [7, 8].

In 2018, the international evidence‐based guideline spanning all topics for assessment and management of PCOS has revised the Rotterdam criteria for the diagnosis of PCOS [9]. A noteworthy change was that an AFC threshold for polycystic ovary morphology (PCOM) was revised from 12 to 20 in adult women when using transducers with a high‐resolution frequency including 8 MHz, reflecting advances in ultrasound resolution. With the tightening of PCOM diagnostic requirements, approximately one‐fifth of adult women with PCOS who were recruited based on the original Rotterdam criteria no longer met the PCOS diagnosis in our study [10]. Subsequently, the guideline was updated in 2023. The most profound change was that both ultrasonography and AMH could be used for defining PCOM in adults [11].

Women with PCOS typically begin to exhibit hallmark symptoms such as hirsutism and IM during adolescence. Although there is a paucity of information regarding experiences of women with PCOS as they navigate through stages of reproductive aging, interestingly, menstrual cycles often become regular, with testosterone levels and the prevalence of acne and hirsutism decreasing in middle age in women with PCOS [12, 13, 14, 15]. Ovarian volume and follicle number are also decreased with age in women with PCOS [16]. Regarding AMH, its age‐related decrease is well‐known, with levels typically peaking between the ages of 20 and 25 years in the general population [11]. In women with PCOS, AMH also decreases with age [17, 18, 19, 20]. However, some studies have shown that the decline in AMH with age is less pronounced in women with PCOS than in controls [17, 21].

The current study examined AMH patterns across different age groups in women who already met the PCOS diagnostic criteria according to either the Rotterdam criteria or the stricter 2018 evidence‐based guideline. We were particularly interested in AMH levels in women who still met the PCOS diagnostic criteria although they were older. We also aimed to identify any correlations between serum AMH and PCOS related parameters and factors influencing AMH levels in women with PCOS.

2. Materials and Methods

2.1. Participants

A total of 725 women with PCOS were enrolled according to the 2003 Rotterdam criteria from 2004 to 2022 [22]. Because combined oral contraceptives may suppress AMH levels, women taking combined oral contraceptives were excluded. Oligomenorrhea was defined as fewer than eight periods per year or cycles longer than 35 days. Amenorrhea was defined as absence of menstruation for more than 3 months. Hirsutism was assessed using a modified Ferriman and Gallwey system, and clinical HA was defined as a score of 6 or greater as reported in our previous study [23]. Biochemical HA was defined if one of the following three criteria was met: total testosterone (T) > 0.68 ng/ml, free T > 1.72 pg/ml, and free androgen index >5.36 [24]. PCOM was defined as either 12 or more antral follicles measuring 2–9 mm in diameter or an increased ovarian volume (≥ 10 cm³). All ultrasound evaluations were performed via transvaginal or transrectal route with informed consent. Ultrasound examinations were carried out with wide band frequency (5–9 MHz) transducers (Voluson E8, GE Healthcare, Milwaukee, WI, USA) with automatic optimization and a center frequency of 8 MHz.

According to the 2018 evidence‐based guideline, PCOM in ultrasonography was defined as either having 20 or more antral follicles measuring 2–9 mm in diameter when using transducers with a high‐resolution frequency including 8 MHz or an ovarian volume greater than 10 cm³ [9, 11]. Additionally, the guideline group advises against evaluating PCOM in adolescents. In line with this update, we also identified women who met the 2018 evidence‐based guideline diagnostic criteria.

This study was approved by the Institutional Review Board of Seoul National University Hospital (Approval no. 1407‐129‐597). Written informed consent was obtained from each patient.

2.2. Assays

Serum AMH was measured using an enzyme‐linked immunosorbent assay (Beckman Coulter Immunotech, Marseille, France). Intra‐ and inter‐assay coefficients of variation were 5.4% and 5.6%, respectively.

Clinical variables including body weight, height, waist circumference, and blood pressure were assessed. Body mass index (BMI) was calculated as weight (kg) divided by height squared (m²). Serum lipid, fasting glucose, insulin, and hemoglobin A1c levels were measured. 2‐h glucose and insulin levels were evaluated using a 75‐g oral glucose tolerance test (OGTT). Serum levels of luteinizing hormone (LH), FSH, total T, free T, and sex hormone‐binding globulin were measured using radioimmunoassay (Siemens, Los Angeles, CA, USA).

2.3. Statistical Analysis

Data are presented as mean ± standard deviation. If a Gaussian distribution was achieved after a natural logarithmic or square root transformation, data are presented as geometric means and 95% confidence intervals. Continuous parameters were compared using analysis of variance. Correlations were assessed by calculating Spearman coefficient. Independent relationships were assessed by means of multiple regression analysis. All data analyses were performed using Statistical Package for the Social Sciences software version 29.0 (IBM SPSS, NY, USA). Statistical significance was considered when a two‐sided P‐value was less than 0.05.

3. Results

The mean age of a total of 725 women with PCOS who were diagnosed according to the Rotterdam criteria was 25.4 ± 6.5 years and mean AMH level was 10.9 ± 1.7 ng/mL. Considering AMH levels typically peaking between ages of 20 and 25 years in the general population and ovarian reserve usually declines after 35 years [11, 25], we compared mean AMH levels for the following age groups: < 25 years (n = 360), 25.0–34.9 years (n = 292), and 35.0–45.0 years (n = 73). Mean AMH levels in these three age groups were 11.2 (10.5, 12.0) ng/mL, 10.9 (10.2, 11.7) ng/mL, and 9.0 (7.7, 10.5) ng/mL, respectively, showing significant differences between age groups of < 25 years and 35.0–45.0 years (Table 1).

Table 1.

Serum anti‐Müllerian hormone (AMH) levels by age group in women with polycystic ovary syndrome (Rotterdam vs. 2018 criteria).

	Original Rotterdam criteria (n = 725)		2018 evidence ‐ based guideline (n = 520)
Age group (years)	AMH (ng/mL)	p value	AMH (ng/mL)	p value
< 25	11.2 (10.5, 12.0) (n = 360)*	0.032	12.5 (11.6, 13.5) (n = 264)	0.077
25.0–34.9	10.9 (10.2, 11.7) (n = 292)		12.0 (11.1, 13.0) (n = 212)
35.0–45.0	9.0 (7.7, 10.5) (n = 73)*		10.0 (8.0, 12.1) (n = 44)

^*This symbol indicates a significant difference between the groups.

p‐values are indicated for the differences between groups, as analyzed using the analysis of variance (ANOVA). A post hoc test was performed using Bonferroni's method.

A Gaussian distribution was achieved after square root transformation, and data are shown as geometric means and 95% confidence intervals.

We further analyzed the subset of women who also met the stricter diagnostic criteria as defined by the 2018 evidence‐based guideline. Thus, in adults, PCOM was defined by AFC ≥ 20 or an increased ovarian volume, and in adolescents, individuals with both HA and IM were included. In 520 women with PCOS diagnosed based on the 2018 guideline, mean AMH levels for three age groups were as follows: 12.5 (11.6, 13.5) ng/mL for the age group of < 25 years (n = 262), 12.0 (11.1, 13.0) ng/mL for the age group of 25.0–34.9 years (n = 212), and 10.0 (8.0, 12.1) ng/mL for the age group of 35.0–45.0 years (n = 44), showing no significant differences among groups (Table 1). Compared with the original Rotterdam criteria, a trend toward AMH elevation (approximately 1 ng/mL in each group) was observed when the stricter criterion was applied. In a scatter plot, we also found that elevated AMH levels maintained until late reproductive age, showing no correlation with age (Figure 1). Furthermore, the oldest group, who met the 2018 criteria, showed worse metabolic profiles, including higher BMI, blood pressure, triglyceride, low‐density lipoprotein cholesterol, fasting glucose, 75 g OGTT 2 h glucose and hemoglobin A1c and lower high‐density lipoprotein cholesterol levels than the other two groups (Table 2).

Figure 1.

Scatter plot depicting relationship between serum level of anti‐Müllerian hormone (AMH; ng/ml) and age of women with polycystic ovary syndrome diagnosed with the 2018 evidence‐based guideline. Correlations were assessed by calculating Spearman coefficient.

Table 2.

Comparison of clinical features according to age groups in women with polycystic ovary syndrome diagnosed with the 2018 evidence‐based guideline.

Age group (years)	< 25 (n = 264)	25.0–34.9 (n = 212)	35.0–45.0 (n = 44)	p value
Age (years)a	20.1 ± 2.7d	29.1 ± 2.6e	37.3 ± 2.3f	0.001
Body mass index (kg/m2)a	22.3 ± 4.3 d	22.6 ± 4.3d	24.9 ± 5.3e	0.005
Systolic blood pressure (mmHg)a	113.5 ± 12.1d	114.9 ± 12.4d	119.9 ± 11.3e	0.024
Diastolic blood pressure (mmHg)a	74.9 ± 8.7d	76.9 ± 9.5d	80.0 ± 10.2e	0.006
Phenotype A and B (%)	89.7% (233/259)	74.4% (154/207)	76.7% (33/43)	0.005
Mean AFC of both ovariesb	19 (14, 25)d	18 (13, 23)	15 (12, 19)e	0.009
Total testosterone (ng/mL)c	0.56 (0.51, 0.60)	0.51 (0.47, 0.55)	0.54 (0.44, 0.66)	0.317
Free androgen indexc	4.59 (3.99, 5.24)d	3.29 (2.90, 3.71)e	4.79 (3.00, 7.03)	0.002
LH (mIU/ml)c	10.0 (8.8, 11.4)	10.0 (8.5, 11.6)	7.7 (4.9, 11.2)	0.315
FSH (mIU/ml)c	5.1 (4.8, 5.4)	5.4 (4.8, 6.0)	4.7 (3.7, 5.7)	0.345
Triglyceride (mg/dl)b	82 (59, 114)d	76 (57, 109)d	111 (70, 169)e	< 0.001
HDL cholesterol (mg/dl)a	61.8 ± 16.4	63.7 ± 18.0d	54.3 ± 15.2e	0.022
LDL cholesterol (mg/dl)a	106.7 ± 30.2d	111.1 ± 34.2d	145.1 ± 38.1e	< 0.001
Fasting glucose (mg/dl)a	89.8 ± 8.7d	92.2 ± 11.6e	96.2 ± 11.3e	< 0.001
Fasting insulin (µU/ml)c	12.1 (10.7, 13.0)	10.3 (9.4, 11.3)	12.6 (8.7, 17.1)	0.118
Hemoglobin A1c (%)a	5.3 ± 0.5	5.2 ± 0.4d	5.5 ± 0.6e	0.030
75 g OGTT 2 h glucose (mg/dl)a	105.7 ± 27.1d	110.1 ± 35.1	123.2 ± 32.3e	0.014
75 g OGTT 2 h insulin (µU/ml)c	56.9 (50.4, 63.7)	48.7 (42.4, 55.5)	59.4 (51.9, 91.2)	0.229

^a Data are shown as the mean ± standard deviation.

^b median (interquartile range).

^c A Gaussian distribution was achieved after a natural logarithmic or square root transformation, and data are shown as geometric mean and 95% confidence intervals.

^d–f Values with different superscripts are significantly different.

p‐values are indicated for the differences between groups, as analyzed using the chi‐square test or analysis of variance (ANOVA). A post hoc test was performed using Bonferroni's method.

Phenotype A: ovulatory dysfunction + hyperandrogenism + polycystic ovary morphology; Phenotype B: ovulatory dysfunction + hyperandrogenism.

Abbreviations: AFC, antral follicle count; FSH, follicle stimulating hormone; HDL, high‐density lipoprotein; LDL, low‐density lipoprotein; LH, Luteinizing hormone; OGTT, oral glucose tolerance test.

PCOS can be categorized into four phenotypes based on combinations of diagnostic parameters: (1) IM + HA + PCOM (phenotype A); (2) IM + HA (phenotype B); (3) HA + PCOM (phenotype C); and (4) IM + PCOM (phenotype D). Among the 509 women with PCOS who were diagnosed according to the 2018 guideline and had clearly defined phenotype classifications, the majority (82.5%, n = 420) had severe phenotypes (phenotype A or B) (Table 2), suggesting that they were already clear in their diagnosis regardless of their PCOM (either by USG or AMH) status. Age‐stratified AMH distributions by phenotype are detailed in Supporting Table 1, and the comparison of mean AMH levels among the three age groups, when grouping phenotype C and D together – for which the presence of PCOM is diagnostically significant – still showed no differences in AMH levels among the groups.

Finally, to analyze determinants of serum AMH levels in women with PCOS, multiple linear regression models were constructed with AMH as a dependent variable (Table 3). In women who were diagnosed with PCOS according to the 2018 evidence‐based guideline, serum AMH levels were independently related (R² = 0.343) to mean antral follicle number of both ovaries (β coefficient = 0.497, p < 0.001), total T (β coefficient = 0.133, p = 0.039), and serum LH concentration (β coefficient = 0.127, p = 0.048) (using centered data to avoid high multicollinearity). However, AMH levels were not related to metabolic parameters such as BMI, lipid profile, glucose profile, or insulin levels (data not shown).

Table 3.

Multiple regression analysis for determinants of serum anti‐Müllerian hormone level in women with polycystic ovary syndrome diagnosed by the 2018 evidence‐based guideline.

	Adjusted R2	Partial regression coefficient (β)	Variance inflation factor	p value
Age (years)	0.343	0.081	1.02	0.208
Luteinizing hormone (IU/L)		0.127	1.07	0.048
Mean antral follicle number of both ovaries		0.497	1.23	< 0.001
Total testosterone (ng/ml)		0.133	1.16	0.039

4. Discussion

This study examined AMH patterns across age groups in women who met the PCOS diagnostic criteria according to either the Rotterdam or the stricter 2018 criteria, ensuring that the participants had a definite PCOS diagnosis independent of their AMH levels. Among women diagnosed according to the Rotterdam criteria, mean AMH levels were comparable between those aged < 25 and 25.0–34.9 years, but were significantly different between the < 25 years and 35.0–45.0 years age groups. In contrast, in women who were recruited by the more rigorous 2018 guideline diagnostic criteria, serum AMH levels showed no differences among age groups, with no correlation with age in a scatter plot analysis. Application of the stricter 2018 diagnostic criteria reduced the number of women in the oldest group classified as having PCOS due to the decline of PCOM features (77 according to the Rotterdam criteria vs. 43 according to the 2018 criteria). However, those older women who continued to exhibit symptoms and met the stricter diagnostic criteria demonstrated similarly elevated AMH levels compared to younger women with PCOS, while demonstrating worse metabolic profiles. Thus, the current study observed relatively stable AMH levels across age groups among women with PCOS who met the stricter 2018 guideline diagnostic criteria, suggesting that AMH may serve as a consistent diagnostic marker even in those over 35 years despite declining PCOM features.

Recent studies align with our findings regarding the age‐independent nature of AMH levels in women with PCOS. A Korean study has reported that AMH level does not negatively correlate with age in 175 women with PCOS under 40 years of age [26]. Similarly, a Chinese study found no significant differences in multiples of the median (MoM) AMH values among women aged 21–25, 26–30, and 31–35 years [27]. Interestingly, women aged 36–40 years had even higher MoM AMH levels than their younger counterparts. More recently, age‐related AMH curves were generated in a large cohort of 2725 women with PCOS aged 20 to 40 years [28]. The authors proposed a cutoff value of 5.5 μg/L to best distinguish PCOS from non‐PCOS women under 30 years of age and suggested a “single” cutoff of 5.0 μg/L for women aged 30 years or older.

AMH is now integrated into the diagnostic flow for PCOS in the updated 2023 international guideline. However, clinical experience with this approach remains limited. The guideline development group recommended population and assay specific cut‐offs for AMH. The aim of our study was to evaluate AMH distributions in women with PCOS who already met the established diagnostic criteria, rather than to define a specific cutoff value. Therefore, our study does not provide diagnostic sensitivity or specificity for any particular threshold, which is a limitation of the current study. Regarding AMH cutoffs specific to Korean women using the Beckman Coulter assay, one receiver operating characteristic analysis reported a cutoff of 7.82 ng/mL, with a sensitivity of 75.9% and a specificity of 86.8% for diagnosing PCOS [29]. Another Korean study suggested an optimal cutoff value of 10.0 ng/mL for diagnosing PCOS in women under the age of 40 years [30]. Investigation of AMH thresholds for diagnosing PCOM—rather than PCOS itself—in Korean women using the same assay platform may provide relevant context for interpreting our findings. Furthermore, studies that define PCOM based on the updated follicle count threshold of ≥ 20 (as opposed to the previous standard of ≥ 12) are more closely aligned with the objective of the present study. However, reference data that satisfy all these conditions remain limited.

According to the diagnostic flow for PCOS, in cases where both IM and HA are present, PCOM is not required to establish a PCOS diagnosis. However, when a woman presents with either IM or HA, the presence of PCOM—determined by either ultrasound or AMH levels—is assessed, suggesting the potential diagnostic relevance of AMH in such cases. In the current study, the majority (82.5%) of patients diagnosing using the 2018 guideline had both IM and HA. Thus, their PCOS diagnosis was already clear regardless of their AMH levels, and this represents a strength of the current study. Interestingly, when phenotype C and D were combined and analyzed together—for which the presence of PCOM is diagnostically significant— there were still no differences in AMH levels among the three age groups.

With aging, cardinal features of PCOS can be gradually lessened in women with PCOS, including ovarian volume and follicle numbers [12, 13, 14, 15, 16]. Because of its cross‐sectional design, our study could not evaluate longitudinal changes in AMH levels or determine the proportion of women who no longer met the diagnostic criteria over time. Further research is warranted to clarify why some women experience resolution of PCOS‐related symptoms, while others continue to fulfill the diagnostic criteria and maintain AMH levels similar to those seen in younger individuals. The oldest group who met the increased rigor in the diagnosis of PCOS by the 2018 guideline likely had an extreme phenotype in their reproductive lifespan, showing no significant difference in AMH level compared to young groups. Currently, there are no age‐ or reproductive‐stage–specific diagnostic criteria for PCOS. Even at older ages, should only women who meet the same diagnostic criteria as younger women be considered as having PCOS? This raises an important question of how to define PCOS in aging women.

In the current study, AMH level was not significantly influenced by age in women with PCOS. Serum LH level, serum T level, and AFC were significant determinants of serum AMH level, with AFC showing the highest regression coefficient as expected. Androgen can augment FSH responsiveness of granulosa cells during follicle development. However, in women with PCOS, androgen excess can cause dysregulated follicle development and lead to PCOM [31, 32]. Exogenous high‐dose androgen supply has been used as a validated method to produce PCOM in rat or monkey, and administration of androgens increased the production of AMH in mouse ovaries [33, 34, 35, 36, 37]. Although direct evidence of stimulatory effect of T on AMH production by granulosa cells was not obtained in our study, HA might be associated with an extra increase in AMH. Furthermore, LH concentration was independently associated with serum AMH level, consistent with previous studies reporting positive correlation between serum LH and AMH levels [17, 38, 39, 40]. Interestingly, Skałba et al. have reported that 34.4% of AMH level variability could be explained by LH and FSH in women with PCOS [41]. It is currently unclear whether a direct effect of LH on AMH secretion exists. High LH levels in women with PCOS could increase the secretion of androgen, and Qi et al. have reported that LH might have a synergistic role in androgen stimulation of AMH production [42]. We can assume that AMH levels might reflect interactions among LH, androgens, and disorders of follicular maturation.

Our study has strengths (large cohort and use of updated criteria), but there are some limitations. First, due to the relatively small number of women aged 35 years or older, caution is warranted when interpreting group comparisons involving this age category. Second, given the cross‐sectional design of our study, we were unable to assess longitudinal trends in AMH levels within individuals; prospective follow‐up studies are needed to elucidate age‐related AMH trajectories and PCOS status. No follow‐up data have been collected for this cohort at present. Third, AMH is measured using various assay platforms. Strong inter‐assay correlations have been reported [43, 44, 45, 46], and for reference, the regression equation from Beckman Coulter Access to Roche Elecsys®—the two widely used automated immunoassay methods—was reported to be as follows: Elecsys® = (0.97 X Access) + 0.003 [43]. However, the lack of standardization across different platforms remains a challenge, and the generalizability of our findings to other assay kits may be limited. Finally, AMH levels have been shown to vary by ethnicity and race [47]. Furthermore, a recent population study reported that age‐related changes in AMH levels were different between European and Chinese women—AMH levels were higher in 461 Chinese women before age 25, but then they had lower AMH concentrations than 887 European women—and the disparity widened with increasing age [48]. Considering these ethnicity‐specific variations, the 2023 guideline recommends population‐specific AMH thresholds. Since our study population was exclusively composed of Korean women, the applicability of our results to other ethnic groups may also be limited.

In conclusion, as individuals age, the clinical manifestations of PCOS generally tend to improve; however, older women who continue to exhibit symptoms and meet the stricter diagnostic criteria did not differ in AMH levels compared to younger women with PCOS, while demonstrating worse metabolic profiles. These findings underscore the potential of AMH as a stable and reliable diagnostic marker for PCOS throughout the reproductive lifespan. Further studies are needed to investigate why some women experience symptomatic resolution while others persistently meet diagnostic criteria beyond the early reproductive years, demonstrating elevated AMH levels similar to those observed in younger patients.

Supporting information

Supporting Table1 AMH PCOS (2025‐06‐24).