Clin Chem Lab Med 2020; aop
Osman Evliyaoglu, Mathias Imöhl, Ralf Weiskirchen and Josef van Helden*
Age-specific reference values improve the
diagnostic performance of AMH in polycystic
ovary syndrome
https://doi.org/10.1515/cclm-2019-1059
Received August 31, 2019; accepted January 8, 2020
Abstract
Background: The increased secretion of anti-Müllerian
hormone (AMH) by the growing follicles has been sup-
posed as a determinative feature of polycystic ovary syn-
drome (PCOS). The diagnostic performance of AMH in
PCOS is superior compared to the free androgen index
(FAI) and luteinizing hormone (LH)/follicle-stimulating
hormone (FSH) quotient. We established age-­dependent
reference ranges to further improve the diagnostic
­performance of AMH.
Methods: In a cross-sectional study, in samples of 4712
reproductive age patients, ranging from 14 to 50 years,
BMI, AMH and other reproductive hormones were deter-
mined by immunoassay or tandem mass spectrometry
(LC-MS/MS) to calculate age-specific reference ranges and
the diagnostic performance.
Results: Age-specific diagnostic performances for
Elecsys® AMH, FAI and LH/FSH ratio were established in
the reference group. No significant difference in BMI was
found between the groups. AMH values were significantly
negatively correlated with age (r = −0.628, p < 0.001) in
patients with normal ovarian function, but there was no
correlation between age and AMH levels in PCOS patients
(r = −0.041, p < 0.174). In all the study groups, AMH
showed a weak correlation between FAI and LH/FSH ratio
(r = 0.302, p < 0.001 and r = 0.434, p < 0.001, respectively).
The sensitivity/specificity for AMH, FAI and LH/FSH ratio
were 89/96%, 71/69% and 75/72%, respectively, according
to the Youden index.
*Corresponding author: Josef van Helden, Laboratory Diagnostic
Center, University Hospital RWTH Aachen, Pauwelsstr. 30, 52074
Aachen, Germany, E-mail: jvhelden@ukaachen.de
Osman Evliyaoglu and Ralf Weiskirchen: Institute of Molecular
Pathobiochemistry, Experimental Gene Therapy and Clinical
Chemistry, University Hospital RWTH, Aachen, Germany.
https://orcid.org/0000-0003-3888-0931 (R. Weiskirchen)
Mathias Imöhl: Laboratory Diagnostic Center, University Hospital
RWTH Aachen, Aachen, Germany
Conclusions: We determined the age-dependent reference
ranges for serum AMH levels in a large population-based
study and calculated the age-specific diagnostic perfor-
mance of FAI and LH/FSH ratio, which allows physicians
to evaluate patients with PCOS who have normal AMH lev-
els. AMH is suggested as the strongest diagnostic marker
in patients with PCOS compared to FAI and LH/FSH ratio.
Keywords: age-specific reference values; anti-Müllerian
hormone (AMH); diagnostic performance; polycystic
ovary syndrome (PCOS).
Introduction
Anti-Müllerian hormone (AMH) is secreted by ovarian
granulosa cells and is one of the transformed growth
factor-β superfamily members [1]. The main feature of AMH
in the ovary is the prevention of primordial follicle recruit-
ment, in addition to the regulation of follicle-­stimulating
hormone (FSH) activity in early follicular development
[2]. A higher level of AMH secreted by the growing folli-
cles has been supposed as a decisive feature of polycystic
ovary syndrome (PCOS) [3]. A 2–3 times increase in AMH
was found in PCOS patients when compared to women
with normal ovaries [4]. But AMH is not recommended
as a singular diagnostic test as per the 2018 international
guidelines because standardized AMH assays and cut-offs
based on long-term studies of diverse groups of women
are lacking [5] for PCOS. Because AMH levels were associ-
ated with the severity of PCOS [4], AMH is recommended
as an additional marker in the subsequent evaluation
of women actively trying to become pregnant. PCOS is
one of the most common endocrine diseases in women
of reproductive age, and affects an estimated 4–21% of
young women [6]. Insulin resistance, hyperandrogenism,
neuroendocrine cascade dysfunction and altered follicu-
logenesis play an important role in PCOS pathogenesis [7].
The important influence of PCOS on reproductive, obstet-
ric, metabolic and cardiovascular diseases makes the
early detection of this syndrome particularly important in
terms of a higher clinical and prognostic value [8]. In July
2018, the evaluation of guidelines on PCOS considered a
Unangemeldet
Heruntergeladen am | 19.02.20 07:35
2 Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS
single serum marker, such as serum AMH [9]. However, it
is questionable whether a single parameter is sufficient to
meet the diagnostic criteria in such a heterogeneous and
complex disease. As an alternative to AMH, the free andro-
gen index (FAI) and the sex hormone-binding globulin
(SHBG) were considered to be the strongest predictors of
PCOS when compared with a variety of hyperandrogen-
ism markers [10, 11]. Moreover, due to the dysfunction of
the neuroendocrine axis, luteinizing hormone (LH) and
the combination of LH/FSH were recommended in PCOS
diagnosis [12].
AMH assays lack an international standard, and con-
centrations as well as cut-off values are method depend-
ent. At the first line, the reference values were established
for the manual AMH Gen II assay from Beckman Coulter
and are still commonly used in clinical laboratory prac-
tice [13, 14]. At the second line, the reference values were
determined for a fully automated assay of AMH (Access
AMH) distributed by the same company [15, 16]. More-
over, the reference values for AMH measurements in the
fully automated AMH electrochemiluminescence assay
(ECLIA) measured on the Elecsys® system from Roche are
­presented in age clusters [17].
Here, we present the long-term performance data of
the Roche AMH-Elecsys Assay compared to the data of the
other hormone assays involved in the diagnosis of PCOS
to support the validity of the reference ranges identified
in this study. We have established the reference percen-
tiles of AMH levels by age interval in a large sample set
and clarified the diagnostic performance of parameters
used in PCOS diagnostics. Based on the fact that high
AMH values are associated with the hyperandrogenemic
or anovulatory cycle disorders and low AMH values with
reduced ovarian activity or insufficiency, we only included
patients with normal ovarian function.
Materials and methods
Clinical characteristics of patients
This cross-sectional study was conducted from September 2016
to April 2019 in a single reference laboratory in Germany. A total
of 4712 reproductive age patients, ranging from 14 to 50 years from
various gynecological practices and fertility centers located in North
Rhine-Westfalia, Germany, were included. Inclusion criteria were:
(i) patients having no history of gynecological surgery, (ii) normal
thyroid function and (iii) normoprolactinemia. Information about
body weight, height, cycle day and cycle length was obtained from
order forms. AMH, LH, FSH, estradiol (E2), prolactin, testosterone,
androstenedione, DHEA-S and SHBG were measured for diagnostic
purposes between the third and fifth day.
Assay description
Serum samples were measured directly after arriving in the laboratory.
Elecsys® AMH assay, LH, FSH, E2, prolactin, testosterone, DHEA-S and
SHBG measurements were performed on two Roche CCM automations
equipped with six cobas 8000 analyzers with e801 and e602 immuno-
assay modules according to the manufacturer’s instructions. Andros-
tenedione was measured by tandem mass spectrometry (LC-MS/MS).
Assay performance
To determine the assay performance for the parameters AMH, LH,
FSH, testosterone and SHBG, the Validation Manager Software (Finbi-
osoft, Espoo, Finland) was used to determine the accuracy (bias), the
quadratic mean of the measurement error (Δ), the extended measure-
ment uncertainty (U) as well as the data for the within-run, between-
run and between-day precision for the parameters AMH, LH, FSH,
testosterone and SHBG over the entire observation period by evaluat-
ing the internal quality controls performed in each run. The calcula-
tions were carried out according to the respective CLSI guidelines. In
the case of AMH, a comparison was made with the Gen II ELISA and
the automated AMH Access Assay of Beckman Coulter (Krefeld, Ger-
many) according to the method of Passing and Bablok [18].
Statistical methods
The data were analyzed using SPSS Version 18.0. Values were given
as means and standard deviations (SD) or percentiles (%). The distri-
bution of continuous variables was explored using the one-sample
Kolmogorov-Smirnov test. A one-way ANOVA or the Mann-Whitney
U-test was conducted where appropriate to detect the significance of
the difference. Correlations were determined using Spearman’s cor-
relation coefficients. p-values <0.05 were considered significant. The
Youden Index (sensitivity + specificity − 1) was calculated to deter-
mine the age-specific optimal limits for each parameter. Receiver
operating characteristics (ROC) were created to examine the diagnos-
tic value of AMH, FAI and LH/FSH ratio. Areas under the curve (AUC)
were calculated by recording the sensitivity to 1−specificity at each
threshold as a measure of the probability that correctly identified
controls and PCOS patients.
Ethical approval
Our study was approved by the Ethics Commission located at the
RWTH University Hospital Aachen (EK158/18).
Results
Patient characteristics
A total of 4712 patients were included in the study. To eval-
uate the status of the ovarian function in a second blood
sample taken about 8 days after the expected ovulation, E2
Unangemeldet
Heruntergeladen am | 19.02.20 07:35
 Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS 3

and progesterone were determined to distinguish between
women with normal ovulation, follicular disorder and
anovulation. The measured reproductive hormone levels
in the different study groups are summarized in Table 1. In
65% of the patients (n = 3055), the menstrual period was
regular. The group having normal ovarian activity was
defined as the reference group. The reference group was
formed at the same time when the patients suffering from
PCOS (24%, n = 1132) and primary ovarian insufficiency
(POI) (11%, n = 525) were grouped. Participants were diag-
nosed with PCOS according to Rotterdam criteria [19] and
those with POI according to criteria described previously
[20].
results, instrument comparisons were performed for the
methods measured on more than one instrument during
the study. The graphical representation is shown in
Figure 1. The Pearson coefficients of correlation were all
between 0.915 and 0.978 for the analytes LH, FSH, SHBG
and testosterone. In all methods, there was a significant
agreement between the analytes (p < 0.001). Figure 1E
and F shows the correlations of Elecsys® AMH with the
Access AMH assay and the AMH Gen II ELISA. Both
method comparisons showed a coefficient of correlation
of 1 and an intercept of 0 ng/mL. However, the Elecsys®
AMH assay measures 5% lower than the Access Test and
in this example about 19% lower than the AMH Gen II
ELISA.

Performance of hormone assays

The accuracy values for the five hormone tests examined
varied between 1.58 and 7.42%. The square mean of the
measurement error ranged between 3.15 and 10.4%. The
expanded uncertainty of measurement for four of the
five methods examined resulted in values between 5.49
and 15.1%. Only SHBG showed values above 20%. The
precision data for within-run, intra-day and between-
day precision were all below 8.15%, whereby the highest
coefficients of variation for SHBG were also measured
here. Overall the performance characteristics were suit-
able for studies across sites and of a longitudinal nature.
All performance data collected can be found in Table 2.
To further demonstrate the comparability of the study
Distribution of hormone values in the patient
groups
There was no significant difference in BMI between the
groups. LH was significantly higher in both the patient
groups (PCOS, POI) when compared to the reference
group (Table 1). AMH, testosterone, DHEA-S, FAI, andros-
tenedione and LH/FSH ratio were all significantly higher,
while FSH, E2 and SHBG were significantly lower in PCOS
patients than in the reference group. In patients with POI,
the values for AMH, testosterone, DHEA-S, androstenedi-
one and LH/FSH ratio were significantly lower. In addi-
tion, FSH and LH were significantly higher than in the
reference group.

Table 1: Descriptive properties and hormone levels of the study groups.

Reference group PCOS POI p1 p2

n 3055 1132 525

Age, years 34.1 ± 8.5 26.6 ± 6.2 37.6 ± 6.3 <0.001 <0.001
BMI, kg/m2 26.1 ± 11.7 27.0 ± 19.5 25.3 ± 5.5 0.36 0.72
E2, pg/mL 92.4 ± 196.3 68.2 ± 63.0 84.4 ± 147.1 <0.001 <0.05
Prolactin, ng/mL 17.5 ± 14.9 16.3 ± 11.3 14.9 ± 7.7 0.62 <0.05
LH, IU/L 7.8 ± 6.9 12.6 ± 9.0 24.4 ± 20.4 <0.001 <0.001
FSH, IU/L 6.8 ± 3.9 5.8 ± 2.4 38.3 ± 36.4 <0.05 <0.001
Testosterone, nmol/L 1.0 ± 1.6 1.8 ± 1.2 0.8 ± 0.6 <0.001 <0.05
DHEA-S, μg/mL 207 ± 102 259 ± 120 181 ± 103 <0.001 <0.001
Androstenedione, ng/mL 1.4 ± 0.8 2.7 ± 1.4 1.1 ± 0.6 <0.001 <0.001
AMH, ng/mL 2.2 ± 1.6 9.6 ± 4.6 0.1 ± 0.2 <0.001 <0.001
SHBG, mmol/L 82.2 ± 44.6 66.0 ± 44.0 85.0 ± 45.5 <0.001 0.69
FAI 1.7 ± 1.7 4.7 ± 4.7 1.4 ± 1.7 <0.001 0.33
LH/FSH 1.2 ± 0.9 2.2 ± 1.2 0.8 ± 0.6 <0.001 <0.001

p1, significance between patients in the reference group and PCOS group; p2, significance between patients in the reference group and
POI group. LH/FSH, luteinizing hormone/follicle-stimulating hormone; FAI, free androgen index; AMH, anti-Müllerian hormone; SHBG, sex
hormone-binding globulin; E2, estradiol; PCOS, polycystic ovary syndrome; POI, primary ovarian insufficiency.

Unangemeldet
Heruntergeladen am | 19.02.20 07:35
 4

Table 2: Performance data of long-term performance data of the analytes AMH, LH, FSH, testosterone and SHBG on the Roche CCM Automation, AMH was performed on one e602, and the
remaining assays on two e801 immunoassay analyzers each.

Assay Conc Unit n Mean Bias, % Δ, % U, % Within-run Between-run Between-day

SD CV, % SD CV, % SD CV, %

AMH 0.66 ng/mL 351 0.691 4.7 6.51 6.02 0.021 1.89 0.012 1.34 0.024 2.11
3.91−5.48 5.45−7.615 5.55−7.14 0. 012−0.024 1.63−2.23 0.009−1.24 0.44−15.4 0.016−0.042 1.64−3.77
5.18 ng/mL 351 5.08 −1.93 4.03 5.49 0.126 2.11 0.153 1.39 0.142 1.44
−2.21 to 1.77 4.75−4.51 4.16−7.92 0.097−0.165 1.98−2.43 0.0889−1.97 0.71−10.8 0.123−0.177 1.11−2.93
LH 10.9 IU/L 685 10.5 −3.83 4.9 9.96 0.209 1.99 0.119 1.14 0.237 2.21
−4.28 to 3.39 4.44−5.45 9.54−10.5 0.182−0.245 1.73−2.33 0.0513−6.84 0.489−65.2 0.162−0.405 1.54−3.873
10.9 IU/L 685 10.5 −3.81 4.67 9.49 0.226 2.15 0.163 1.55 0.102 0.974
−4.2 to 3.41 4.24−5.2 9.16−9.92 0.197−0.265 1.88−2.53 0.0789−1.87 0.735−17.8 N/A N/A
53.1 IU/L 687 51.8 −2.47 3.61 8.06 0.836 1.61 1.42 2.74 0.12 0.46
−2.92 to 2.03 3.25−4.02 7.52−8.8 0.729−0.98 1.41−1.89 0.925−3.02 1.79−5.84 N/A N/A
53.1 IU/L 685 51.9 −2.31 3.39 6.88 0.857 1.65 0.715 1.38 0.713 1.37
−2.67 to 1.95 3.08−3.77 6.49−7.4 0. 747−1 1.44−1.94 0.374−4.38 0.721−8.44 0. 37−4.56 0.714−8.78
FSH 19.2 IU/L 688 18.9 −1.38 3.57 7.26 0.493 2.6 0.0667 0.352 0.396 2.09
−1.86 to 0.903 3.25−3.98 6.65−8.03 0.429−0.578 2.27−3.05 N/A N/A 0.267−0.763 1.41−4.03
19.2 IU/L 684 19.1 −0.631 3.15 7.21 0.356 1.86 3.02 0.352 0.336 1.99
−1.14 to 0.118 2.86−351 6.43−8.21 0.31−0.417 1.63−2.19 1.87−7.595 N/A 0.287−0.712 1.33−3.763
49.4 IU/L 690 48.6 −1.58 3.87 7.88 1.25 2.56 0.302 0.62 1.2 2.46
−2.09 to 1.583 3.52−4.31 7.21−8.74 1.09−1.468 2.23−3 N/A N/A 0.873−1.9 1.8−3.91
49.4 IU/L 682 49.1 −0.695 3.3 6.67 0.923 1.88 0.992 2.02 0.853 1.74
−1. 17 to 0.22 2.99−3.68 5.98−7.57 0.803−1.098 1.64−2.21 0.558−3.84 1.14−7.82 0. 42−8.53 0.856−17.4
T 9.54 nmol/L 694 8.99 −5.8 7.25 14.9 0.245 2.73 0.0496 0.552 0.334 3.72
−6.42 to 5.18 6.6−8.05 14.2−15.7 0.214−0.287 2.38−3.19 N/A N/A 0. 273−0.43 3.04−4.79
9.54 nmol/L 700 9.03 −5.4 7.34 15.1 0.319 3.54 0.223 2.48 0.274 3.04
−6.09 to 4.7 6.68−814 14.4−16.1 0.278−0.374 3.08−4.14 0.13−0.742 1.44−8.22 0.182−0.551 2.02−6.11
21.1 nmol/L 700 20.3 −3.87 6.31 13.5 0.853 4.21 0.322 1.9 0.721 3.56
−4.61 to 3.13 5.75−7 12.6−14.6 0.744−1 3.67−4.93 0.21−0.452 1.18−6.58 0.563−1 2.78−4.95
21.1 nmol/L 694 20.4 −3.24 5.05 10.3 0.596 2.92 0.47 2.3 0.309 1.51
−3.79 to 2.69 4.59−5.6 9.71−11.1 0.52−0.698 2.55−3.42 0.273−1.54 1.34−7.56 0.127−39.8 0.624−1.95
Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS

SHBG 27.1 nmol/L 661 25.1 −7.42 10.4 21.8 0.596 2.27 0.439 1.75 1.86 7.42
−8.57 to 6.28 9.4−11.7 20.4−23.8 0.489−0.681 1.95−2.71 N/A N/A 1.55−2.34 6.17−9.32
27.1 nmol/L 671 25.5 −5.96 9.76 20.4 0.564 2.21 0.339 1.33 2 7.86
−7.14 to 4.79 8.83−10.9 18.8−22.7 0.487−0.67 1.91−2.63 0.143−27.6 0.56−80 1.72−2.39 6.75−9.4
53.2 nmol/L 661 49.3 −7.38 10.4 21.6 1.19 2.41 1.77 3.59 3.26 6.61
−8.52 to 6.253 9.35−11.6 20.2−23.6 1.02−1.42 2.07−2.89 0.81−40 1.64–81.3 2.29−5.66 4.64−11.5
53.2 nmol/L 679 49.9 −6.13 10.1 21.1 1.19 2.39 0.607 1.22 4.07 8.15

Heruntergeladen am | 19.02.20 07:35

Unangemeldet
−7.36 to 4.91 9.11−11.3 19.4−23.5 1.03−1.42 2.06−2.84 N/A N/A 3.49−4.87 6.99−9.75

Two control materials were evaluated. Bias, accuracy; Δ, quadratic mean of the measurement error; U, extended uncertainty of measurement; SD, standard deviation; CV, correlation coefficient;
AMH, anti-Müllerian hormone; LH, luteinizing hormone; FSH, follicle-stimulating hormone; SHBG, sex hormone-binding globulin; T, testosterone.
 Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS 5

Figure 1: Passing-Bablok method comparisons.

(A–D) Comparison of methods for luteinizing hormone (LH), follicle-stimulating hormone (FSH), testosterone and sex hormone-binding
globulin (SHBG) between both e801 analyzers used in the study. (E and F) Comparison of different anti-Müllerian hormone (AMH) methods.
The regression equations, the Pearson coefficients of correlation and the p-values for n = 50 measurements are given.

AMH values were significantly negatively correlated
with age (r = −0.628, p < 0.001) in patients with normal
ovarian function, but there was no correlation between
age and AMH levels in PCOS patients (r = −0.041, p < 0.174).
In all the study groups, AMH showed a weak correlation
between FAI and LH/FSH ratio (r = 0.302, p < 001 and
r = 0.434, p < 001), respectively. The age-specific refer-
ence values for the Elecsys® AMH test were calculated
and displayed as mean ± SD values and percentile (%)
values (Table 3). In the comparison of indexes, FAI and
LH/FSH ratio have a wide intersection band with the refer-
ence group between ages 19 and 35 years, but AMH values
demonstrated a reliable difference from the control group
(Figure 2).
Establishment of year-specific reference
values
For all patients in the reference group, the reference
ranges were determined as the median ± 2 SD range for
each age group from 14 to 51 years. The 5th, 10th, 25th,
50th, 75th, 90th and 95th percentiles were calculated and
are compiled in Table 4. Figure 3 shows the 10th, 50th and
90th percentile lines of the AMH for the reference group
compared to the PCOS and POI groups.
Diagnostic performance of biomarkers in
PCOS
In the ROC analysis, the AUC values of AMH, FAI and LH/
FSH ratio were 0.98, 0.78 and 0.78, respectively (Figure 3).
The diagnostic performance of FAI was not different
between obese and non-obese patients. For each para-
meter, cut-off levels and values having a 100% sensitivity
or 100% specificity were determined (Table 4). The diag-
nostic evaluation of all parameters was performed. All
showed AUC scores below 0.60 apart from AMH, FAI and
the LH/FSH ratio. The AUC value for SHBG was 0.37.
Furthermore, we calculated the diagnostic performance
of these parameters for each age group (Supplementary
Table 1). We observed the differential power of the AMH with
the change in AMH percentiles (Figure 4A) and distribution

Unangemeldet
Heruntergeladen am | 19.02.20 07:35
6 Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS

Table 3: Percentiles of AMH (ng/mL) in the reference group.

Age, n Mean ± SD Percentiles

years
5th 10th 25th 50th 75th 90th 95th

14 20 3.29 ± 1.32 1.07 1.47 2.21 3.27 4.48 5.09 5.41

15 27 3.60 ± 1.40 1.00 1.24 2.44 3.68 4.80 5.54 5.76
16 40 3.28 ± 1.46 0.76 1.25 2.12 3.14 4.32 5.31 5.63
17 24 3.43 ± 1.51 0.42 0.99 2.31 3.58 4.62 5.37 5.60
18 28 3.50 ± 1.30 0.89 1.07 2.78 3.65 4.47 5.22 5.72
19 31 3.27 ± 1.40 0.79 1.33 2.06 3.54 4.58 5.14 5.39
20 33 2.90 ± 1.29 1.01 1.45 1.82 2.67 3.99 4.54 5.35
21 40 3.77 ± 1.10 1.66 2.71 3.06 3.68 4.52 5.60 5.78
22 39 3.26 ± 1.41 1.11 1.28 2.17 3.10 4.50 5.29 5.76
23 29 3.40 ± 1.41 0.83 1.26 2.38 3.47 4.22 5.55 5.74
24 58 3.39 ± 1.36 1.27 1.57 2.28 3.41 4.37 5.33 5.67
25 76 3.59 ± 1.44 0.82 1.59 2.36 3.73 4.77 5.29 5.61
26 78 3.11 ± 1.51 0.70 0.92 1.85 3.08 4.24 5.39 5.72
27 76 3.01 ± 1.36 0.79 1.08 2.08 3.10 3.86 5.04 5.37
28 107 3.15 ± 1.46 1.01 1.23 1.89 2.96 4.31 5.18 5.56
29 119 3.06 ± 1.52 0.66 1.05 1.72 2.96 4.34 5.19 5.61
30 135 3.13 ± 1.47 0.75 1.17 2.00 3.00 4.30 5.21 5.63
31 144 3.01 ± 1.46 0.86 1.06 1.82 2.91 4.09 5.10 5.53
32 143 2.61 ± 1.43 0.55 0.79 1.33 2.42 3.83 4.46 4.90
33 143 2.35 ± 1.29 0.49 0.84 1.36 2.20 3.10 4.24 4.96
34 155 2.54 ± 1.56 0.43 0.79 1.20 2.14 3.98 4.88 5.19
35 141 2.20 ± 1.39 0.46 0.69 1.10 1.91 2.92 4.48 5.03
36 153 2.16 ± 1.32 0.42 0.58 1.06 1.94 3.05 4.19 4.77
37 148 2.01 ± 1.32 0.39 0.61 0.95 1.75 2.70 3.89 4.86
38 121 1.62 ± 1.19 0.32 0.44 0.74 1.32 2.29 3.41 4.20
39 131 1.77 ± 1.37 0.20 0.38 0.64 1.39 2.40 4.20 4.46
40 140 1.37 ± 1.09 0.19 0.26 0.62 1.13 1.81 2.64 3.58
41 112 1.12 ± 1.00 0.18 0.27 0.44 0.88 1.34 2.06 3.64
42 84 1.06 ± 0.90 0.17 0.23 0.41 0.76 1.58 2.28 2.97
43 66 1.05 ± 0.89 0.18 0.21 0.37 0.80 1.53 2.24 2.75
44 65 0.79 ± 0.79 0.13 0.16 0.32 0.52 1.03 1.70 2.29
45 55 0.70 ± 0.74 0.09 0.13 0.25 0.44 0.90 1.70 2.76
46 58 0.62 ± 0.66 0.08 0.15 0.20 0.33 0.78 1.49 2.36
47 47 0.39 ± 0.34 0.06 0.07 0.13 0.28 0.56 0.82 1.20
48 52 0.50 ± 0.81 0.07 0.08 0.16 0.30 0.48 1.12 1.90
49 30 0.37 ± 0.26 0.06 0.08 0.13 0.34 0.47 0.86 0.97
50 33 0.30 ± 0.27 0.04 0.05 0.09 0.18 0.52 0.77 0.88
51 19 0.24 ± 0.24 0.04 0.05 0.08 0.11 0.22 0.46 0.73

AMH levels are given in ng/mL. AMH, anti-Müllerian hormone; SD, standard deviation.

of AMH levels with an interpolation line in the three groups
(Figure 4B). The 10th percentile line of AMH in PCOS evi-
dently deviates from the 90th percentile line of the reference
group after the age of 18 years. The 90th percentile course of
AMH in POI patients was different from the 10th percentile
line of the reference group until the age of 38 years.
Discussion
One of the most important things for the determination
of reference ranges and the assessment of the diagnostic
performance of biomarkers is the long-term comparabil-
ity as well as the comparability of the data determined
between different laboratories. Here, we were able to
show that the assays we used showed excellent long-term
performance over the entire study period of more than
2 years, even over several lots used (Table 2). We were
also able to prove the good comparability between differ-
ent analyzers. With AMH we have once again confirmed
the known differences to the AMH tests of other manu-
facturers, which are also in routine use [21]. The Elecsys®
AMH test measures 5% lower than the Access AMH assay
and about 15–20% lower than the AMH Gen II ELISA [21].

Unangemeldet
Heruntergeladen am | 19.02.20 07:35
 Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS 7

Figure 2: Distribution and diagnostic performance of free androgen index (FAI), anti-Müllerian hormone (AMH) and luteinizing hormone
(LH)/follicle-stimulating hormone (FSH) ratio in patients with polycystic ovary syndrome (PCOS) and controls.
The lower and upper quartiles of AMH, FAI and LH/FSH ratio are depicted.

Therefore, it is necessary to determine age-specific refer- Because hyperandrogenism inhibiting follicular

ence ranges for each of these assays, even if the antibodies
used are the same [22], which expresses the coefficient of
correlation and the absence of intercept.
As PCOS is one of the most common endocrine disor-
ders in women of reproductive age, there are many efforts
to early diagnose the disease. In patients with PCOS, the
change in gonadotropin-releasing hormone secretion has
been observed to increase LH without any change in FSH
secretion, which is widely accepted as a specific endocrine
profile. This secretion pattern causes an abnormal LH/FSH
ratio in many patients, increasing the likelihood of being
a valuable diagnostic marker for PCOS [23]. The diagnostic
performance of the LH/FSH ratio was evaluated in previous
studies using the ROC analysis between the values of AUC
which were 0.73 and 0.87 [12, 24]. In our study, the value
of AUC was 0.79. The small difference may arise from the
unequal size of the studies. Unfortunately, it is obvious that
the LH/FSH ratio is not sufficient to diagnose PCOS alone.
development is considered to be the major pathophysi-
ological feature of PCOS [25, 26], the androgenic marker
FAI is also treated as a diagnostic marker for PCOS. In our
study, FAI cannot fulfil the diagnostic criteria with the AUC
value of 0.78, which is consistent with previous studies
with the AUC value of 0.79 [24]. SHBG, a component of the
FAI, is also suggested as a biomarker for PCOS. However,
we found that the AUC value was too low to suggest this
parameter as a positive or a negative diagnostic criterion
[11]. In our study, SHBG also showed the highest expanded
measurement uncertainty and the worst precision values
of all the parameters investigated.
After the observation of the 75 times higher AMH con-
centration in granulosa cells of anovulatory patients and
4 times higher concentrations in ovulatory patients with
PCOS when compared to normal women’s ovaries, AMH
is considered as one of the main criteria for the diagno-
sis of PCOS [27]. In line with this, several studies have

Unangemeldet
Heruntergeladen am | 19.02.20 07:35
 8 Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS

Table 4: Diagnostic characteristics of parameters with the cut-off can be obtained when AMH measurements were com-
levels determined by the Youden index, maximum sensitivity and bined with other indexes, androgens and anthropometric
maximum specificity.
and clinical characteristics [31].
Although BMI was presented as an anthropometric
Cut-off Sensitivity, Specificity,
value % % marker for the diagnosis of PCOS [12], we found no asso-
ciation between BMI and the study parameters as well
LH/FSH 1.41 74.64 72.15
as between PCOS and BMI as reported in other studies
0.04 99.99 1
7.1 1 100 [32]. Although BMI is an important parameter for dys-
FAI 1.95 70.54 69.44 lipidemia in patients with PCOS, it is a doubtful marker
17.2 3 100 in the diagnosis of PCOS [33]. The BMI was not suitable
0.25 99.99 5 to discriminate between the reference group, PCOS and
AMH 5.33 89.47 96.19
POI groups (Table 1). In line with this assumption, it
2.10 100 47.1
5.98 84.5 100
was observed that the AMH values were not changed in
weight loss programs [34]. The independence of AMH
Cut-off values for LH/FSH and FAI are given in %, and AMH values in from the BMI [35] suggests that the role of adiposity in
ng/mL. LH/FSH, luteinizing hormone/follicle-stimulating hormone;
PCOS is complicated and the influence of BMI on ano-
FAI, free androgen index; AMH, anti-Müllerian hormone.
vulation is different from that in AMH, a marker for the
ovarian reserve [36].
1.0 Upon detection of low levels of AMH as an indica-
tor of ovarian failure, this parameter is recommended
as a primary screening index for premature ovarian
0.8
failure (POF) patients [37], and is considered to be the
most reliable marker for ovarian function in POI [38]. We
0.6 established the AMH percentiles for the PCOS, POI and
Sensitivity

reference groups to illustrate the separation efficiency of

0.4
AMH
0.2
FAI
LH/FSH
0.0
0.0 0.2 0.4 0.6 0.8
1–Specificity
Figure 3: Receiver operating characteristic (ROC) curves for anti-
Müllerian hormone (AMH), free androgen index (FAI) and luteinizing
hormone (LH)/follicle-stimulating hormone (FSH) ratio for the
prediction of polycystic ovary syndrome (PCOS).
AMH, area under the curve (AUC) 0.98, confidence interval (CI)
0.94–1.00; FAI, AUC 0.78, CI 0.71–0.87; LH/FSH ratio, AUC 0.78, CI
0.72–0.89.
represented an acceptable diagnostic performance for
AMH, which have AUC values between 0.67% and 98%
[28]. Our findings support the recent studies that offered
AMH as a diagnostic criterion (Figure 3) [29]. The increase
in the diagnostic properties of AMH in recent studies may
be related to the improvement in the analytical perfor-
mance of the test [30]. When comparing the FAI, LH/FSH
ratio and AMH at different ages to predict PCOS, we found
an obvious priority for AMH (Figure 3). Moreover, the
increase in diagnostic validity in the prediction of PCOS
AMH (Figure 4A).
Because of lack of a well-defined population and
other aspects such as heterogeneity of circulating AMH,
inter-laboratory variabilities and different immunoassays,
serum AMH is a valuable tool for the diagnosis of PCOS. In
previous studies, the Elecsys® AMH assay exhibited high
1.0 levels of stability and sensitivity, lower analytical vari-
ability and superior technical and inter-laboratory per-
formance [17, 21]. This is also confirmed by the long-term
performance data we have collected here for the first time.
Age-stratified thresholds demonstrate the superior diag-
nostic performance compared with a single threshold in
the patients with PCOS by the aspect of other diagnostic
methods [39]. We observed the same superiority in AMH,
FAI and LH/FSH ratio. We established age-specific refer-
ence values for the Elecsys® AMH assay with population-
based data to reach better diagnostic performance for
the Elecsys® AMH assay (Table 3). Furthermore, an age-
related diagnostic performance of AMH, FAI and LH/FSH
ratio was also proposed (Table 4). In different ages, some
of the tests have superior sensitivity or specificity than the
others.
The strengths of this study are the high number of
participants, well-defined patients, the reference group
and proof of good long-term performance of the assays.
As for the limitations of the study, the study was not

Unangemeldet
Heruntergeladen am | 19.02.20 07:35
 Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS 9

Figure 4: Scatter plots of anti-Müllerian hormone (AMH) values determined with the Roche Elecsys AMH assay.
(A) AMH percentile lines in the reference group, polycystic ovary syndrome (PCOS) and primary ovarian insufficiency (POI) groups.
(B) Distribution of AMH levels in the study group.

population-based but consisted of women who had visited
an infertility clinic. Moreover, the given reference values
are normative data, but they do not have a r­eplication
pattern to test the suggested values.
In conclusion, AMH appears to be the best diagnostic
marker in patients with PCOS compared to FAI and LH/FSH
ratio. The age-related reference intervals are more impor-
tant for the clinical use of AMH as an indicator of PCOS. In
addition, the age-specific diagnostic performance of FAI
and LH/FSH ratios allows physicians to evaluate patients
with PCOS who have normal AMH levels. Therefore, we
suggest that our study may give clinicians a more reliable
reference for accurate interpretation of AMH levels and
facilitate the use of AMH as a diagnostic tool according to
age intervals.
Author contributions: All the authors have accepted
responsibility for the entire content of this submitted
manuscript and approved submission.
Research funding: None declared.
Employment or leadership: None declared.
Honorarium: None declared.
Competing interests: The funding organization(s) played
no role in the study design; in the collection, analysis, and
interpretation of data; in the writing of the report; or in the
decision to submit the report for publication.

Unangemeldet
Heruntergeladen am | 19.02.20 07:35
10 Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS
References
1. La Marca A, Broekmans FJ, Volpe A, Fauser BC, Macklon NS, on
behalf of the ESHRE Special Interest Group for Reproductive
Endocrinology-AHM Round table. Anti-Müllerian hormone: what
do we still need to know? Hum Reprod 2009;24:2264–75.
2. Visser JA, Themmen AP. Anti-Müllerian hormone and folliculo-
genesis. Mol Cell Endocrinol 2005;234:81–6.
3. Dumont A, Robin G, Catteau-Jonard S, Dewailly D. Role of anti-
Mullerian hormone in pathophysiology, diagnosis and treatment
of polycystic ovary syndrome: a review. Reprod Biol Endocrinol
2015;13:137–49.
4. Jacob SL, Field HP, Calder N, Picton HM, Balen AH, Barth JH.
Anti-Mullerian hormone reflects the severity of polycystic ovary
syndrome. Clin Endocrinol 2017;86:395–400.
5. McCartnez CR, Chang AZ, Straseki JA, Auchus RJ, Woodworth A.
Challenges in the assessment and diagnosis of polzczstic ovary
syndrome. Clin Chem 2019;65:370–7.
6. Belenkaia LV, Lazareva LM, Walker W, Lizneva DV, Suturina LV.
Criteria, phenotypes and prevalence of polycystic ovary syn-
drome. Minerva Ginecol 2019;71:211–23.
7. Crespo RP, Bachega TA, Mendonça BB, Gomes LG. An update
of genetic basis of PCOS pathogenesis. Arch Endocrinol Metab
2018;62:352–61.
8. Moreau KL, DuBose LE. The role of androgens in microvascular
endothelial dysfunction in PCOS: does size matter? J Physiol
2019;597:2829–30.
9. Teede HJ, Misso ML, Costello MF, Dokras A, Laven J, Moran L,
et al. on behalf of the International PCOS Network. Recommen-
dations from the international evidence-based guideline for the
assessment and management of polycystic ovary syndrome.
­Clin Endocrinol 2018;89:251–68.
10. Bui HN, Sluss PM, Hayes FJ, Blincko S, Knol DL, Blankenstein
MA, et al. Testosterone, free testosterone, and free androgen
index in women: reference intervals, biological variation, and
diagnostic value in polycystic ovary syndrome. Clin Chim Acta
2015;450:227–32.
11. Deswal R, Yadav A, Dang AS. Sex hormone binding globulin –
an important biomarker for predicting PCOS risk: a systematic
review and meta-analysis. Syst Biol Reprod Med 2018;64:12–24.
12. Le MT, Le VN, Le DD, Nguyen VQ, Chen C, Cao NT. Exploration of
the role of anti-Mullerian hormone and LH/FSH ratio in diagnosis
of polycystic ovary syndrome. Clin Endocrinol 2019;90:579–85.
13. Du X, Ding T, Zhang H, Zhang C, Ma W, Zhong Y, et al. Age-­
specific normal reference range for serum anti-Müllerian
hormone in healthy Chinese Han women: a nationwide popula-
tion-based study. Reprod Sci 2016;23:1019–27.
14. Heidar Z, Bakhtiyari M, Foroozanfard F, Mirzamoradi M. Age-
specific reference values and cut-off points for anti-müllerian
hormone in infertile women following a long agonist treatment
protocol for IVF. J Endocrinol Invest 2018;41:773–80.
15. Cheng X, Zhang Q, Liu M, Li S, Tao Z, Ichihara K, et al. Establish-
ing age-specific reference intervals for anti-Müllerian hormone
in adult Chinese women based on a multicenter population. Clin
Chim Acta 2017;474:70–5.
16. Helden JV, Weiskirchen R. Age-independent anti-Müllerian
hormone (AMH) standard deviation scores to estimate ovarian
function. Eur J Obstet Gynecol Reprod Biol 2017;213:64–70.
17. Anckaert E, Öktem M, Thies A, Cohen-Bacrie M, Daan NM, Schi-
ettecatte J, et al. Multicenter analytical performance evaluation
of a fully automated anti-Müllerian hormone assay and refer-
ence interval determination. Clin Biochem 2016;49:260–7.
18. Passing H, Bablok W. A new biometrical procedure for testing
the equality of measurements from two different analytical
methods. J Clin Chem Clin Biochem 1983;21:709–20.
19. Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop
Group. Revised 2003 consensus on diagnostic criteria and long-
term health risks related to polycystic ovary syndrome (PCOS).
Hum Reprod 2004;19:41–7.
20. Komorowska B. Autoimmune premature ovarian failure. Meno-
pause Rev 2016;15:210–4.
21. van Helden J, Weiskirchen R. Performance of the two new fully
automated anti-Müllerian hormone immunoassays compared
with the clinical standard assay. Hum Reprod 2015;30:1918–26.
22. Gassner D, Jung R. First fully automated immunoassay for anti-
Müllerian hormone. Clin Chem Lab Med 2014;52:1143–52.
23. Normal RJ, Deweilly D, Legro RS, Hickey TE. Polycystic ovary
syndrome. Lancet 2007;370:685–97.
24. Nadaraja RN, Sthaneshwar P, Razali N. Establishing the cut off
values of androgen markers in the assessment of polycystic
ovarian syndrome. Malays J Pathol 2018;40:33–9.
25. Rosenfield RL, Ehrmann DA. The pathogenesis of polycystic
ovary syndrome (PCOS): the hypothesis of PCOS as functional
ovarian hyperandrogenism revisited. Endocr Rev 2016;37:467–
520.
26. Walters KA, Handelsman DJ. Role of androgens in the ovary. Mol
Cell Endocrinol 2018;465:36–47.
27. Pellat L, Hanna L, Brincat M, Galea R, Brain H, Whitehead S,
et al. Granulosa cell production of anti-Müllerian hormone
is increased in polycystic ovaries. J Clin Endocrinol Metab
2007;92:240–5.
28. Teede H, Misso M, Tassone EC, Dewailly D, Ng EH, Azziz R, et al.
Anti-Müllerian hormone in PCOS: a review informing interna-
tional guidelines. Trends Endocrinol Metab 2019;30:467–78.
29. Yue CY, Lu LK, Li M, Zhang QL, Ying CM. Threshold value of
anti-Mullerian hormone for the diagnosis of polycystic ovary
syndrome in Chinese women. PLoS One 2018;13:e0203129.
30. Jacobs MH, Reuter LM, Baker VL, Craig LB, Sakkas D, Surrey E,
et al. A multicentre evaluation of the Elecsys(®) anti-Müllerian
hormone immunoassay for prediction of antral follicle count.
Reprod Biomed Online 2019;38:845–52.
31. Wissing ML, Mikkelsen AL, Kumar A, Kalra B, Pors SE, Flachs EM,
et al. Associations of different molecular forms of antimüllerian
hormone and biomarkers of polycystic ovary syndrome and
normal women. Fertil Steril 2019;112:149–55.
32. Casadei L, Fanisio F, Sorge RP, Collamarini M, Piccolo E, Piccione
E. The diagnosis of PCOS in young infertile women according
to different diagnostic criteria: the role of serum anti-Müllerian
hormone. Arch Gynecol Obstet 2018;298:207–15.
33. Kiranmayee D, Kavya K, Himabindu Y, Sriharibabu M, Madhuri
GL, Venu S. Correlations between anthropometry and lipid
­profile in women with PCOS. J Hum Reprod Sci 2017;10:167–72.
34. Moini A, Arabipoor A, Hemat M, Ahmadi J, Salman-Yazdi R,
Zolfaghari Z. The effect of weight loss program on serum anti-
Müllerian hormone level in obese and overweight infertile
women with polycystic ovary syndrome. Gynecol Endocrinol
2019;35:119–23.
35. Kominiarek MA, Jungheim ES, Hoeger KM, Rogers AM, Kahan
S, Kim JJ. American Society for Metabolic and Bariatric Surgery
position statement on the impact of obesity and obesity treat-
ment on fertility and fertility therapy Endorsed by the American
Unangemeldet
Heruntergeladen am | 19.02.20 07:35
 Evliyaoglu et al.: Age-specific reference values improve the diagnostic performance of AMH in PCOS
College of Obstetricians and Gynecologists and the Obesity
Society. Surg Obes Relat Dis 2017;13:750–7.
36. Simões-Pereira J, Nunes J, Aguiar A, Sousa S, Rodrigues C,
­Sampaio Matias J, et al. Influence of body mass index in
anti-Müllerian hormone levels in 951 non-polycystic ovarian
syndrome women followed at a reproductive medicine unit.
Endocrine 2018;61:144–8.
37. Sun S, Chen H, Zheng X, Ma C, Yue R. Analysis on the level
of IL-6, IL-21, AMH in patients with auto-immunity prema-
ture ovarian failure and study of correlation. Exp Ther Med
2018;16:3395–8.
11
38. Nyström A, Mörse H, Nordlöf H, Wiebe K, Artman M, Øra I, et al.
Anti-müllerian hormone compared with other ovarian markers
after childhood cancer treatment. Acta Oncol 2019;58:218–24.
39. Ahmad AK, Quinn M, Kao CN, Greenwood E, Cedars MI, Hud-
dleston HG. Improved diagnostic performance for the diagnosis
of polycystic ovary syndrome using age-stratified criteria. Fertil
Steril 2019;111:787–93.
Supplementary Material: The online version of this article offers
supplementary material (https://doi.org/10.1515/cclm-2019-1059).
Unangemeldet
Heruntergeladen am | 19.02.20 07:35