JOURNAL OF MEDICAL RESEARCH
COMPARING AMH, AFC AND FSH FOR PREDICTING HIGH
OVARIAN RESPONSE IN WOMEN UNDERGOING ANTAGONIST
PROTOCOL
Nguyen Xuan Hoi1, Nguyen Manh Ha2
1
National Obstetrics and Gynecology Hospital, 2Hanoi Medical Unviversity
The aim of this study was to assess the predictive values of AFC, AMH and FSH in predicting high
ovarian response during in - vitro fertilization (IVF). We recruited 600 IVF patients who were receiving GnRH
antagonist therapy and recombinant FSH for ovarian stimulation. High ovarian response during IVF was defined as > 15 oocytes retrieved. AMH, FSH and AFC levels were assessed on cycle day 2. We found that the
AMH threshold value for high ovarian response was 4.04 ng/ml with a sensitivity of 73% and a specificity of
61%. The AFC threshold value for high ovarian response was 10.5, with a sensitivity of 78.7% and a specificity of 52%. The FSH threshold value for high ovarian response was 6.14 (IU/L) with a sensitivity of 53.2%
and a specificity of 72.7%. The area under the curve (AUC) of AMH, AFC and FSH were 71%, 65%, 62.7%,
respectively. Conclusions: AMH was the best marker for predicting high ovarian response during IVF, followed by AFC and FSH.
Keywords: AMH, FSH, AFC, high ovarian response, GnRH antagonist
I. INTRODUCTION
individualization of the ovarian stimulation
A high ovarian response to ovarian stimula-
treatment regimen and to counsel patients
tion during in-vitro fertilization (IVF) has been
about the risk of OHSS. Factors used to pre-
associated with increased cancellation rates,
dict ovarian stimulation include markers of
compromised pregnancies, and live birth rates
ovarian reserve such as follicle stimulating
[1]. A high ovarian response also increases
hormone (FSH) and antral follicle count (AFC).
the risk for development of ovarian hyperstimulation syndrome (OHSS). OHSS is an
excessive response to ovarian stimulation,
characterized by increased vascular permeability and ovarian enlargement. Moderate and
severe forms of OHSS may occur in 3% to
10% of all IVF cycles and the incidence may
reach 25% among women undergoing IVF
treatment [2]. Thus, early identification of potential high responders is necessary to enable
Recently, anti - Müllerian hormone (AMH)
has been used as a reliable indicator of
ovarian reserve [3; 4]. Determining an AMH
threshold is important in order to identify
women who are at risk of high ovarian
response and OHSS [5]. Some studies have
shown that AMH is an accurate biomarker for
predicting OHSS [6; 7]. Others have compared
the predictive values of AMH, AFC and FSH
for ovarian response. In controlled ovarian
hyperstimulation, AFC has been found to be a
Corresponding author: Nguyen Xuan Hoi, National Obstetrics and Gynecology Hospital
E-mail:
Received: 20 October 2016
Accepted: 10 December 2016
JMR 105 E1 (7) - 2016
better predictor of ovarian response than AMH
[8; 9]. However, the predictive values of AMH,
FSH and AFC in IVF women undergoing the
antagonist protocol are not fully understood.
57
JOURNAL OF MEDICAL RESEARCH
This study was designed to assess the
there were ≥ 2 follicles of ≥ 18 mm. Oocyte
predictive values of AFC, AMH, and FSH in
retrieval was conducted 36 hours after hCG
predicting high ovarian response during IVF.
administration.The criteria for ovarian response was based on the number of oocytes retrie-
II. SUBJECTS AND METHODS
ved [10]. High ovarian response was defined
as more than 15 oocytes retrieved.
1. Subjects
Female
members
of
infertile
couples
Measurement of AFC, AMH, and FSH
undergoing IVF antagonist treatment at the
To determine AMH and FSH levels, eligible
National Assisted Reproductive Technology
subjects had 3 mL of blood drawn on day 2 of
Center were eligible to participate in this study.
their menstrual cycle and just prior to FSH
The research was conducted at the National
stimulation. Serum separation was done within
Assisted Reproductive Technology Center in
one hour after blood collection. Serum was
Vietnam. All patients in this study met the
stored at –20°C and then transferred to
selection criteria and voluntarily agreed to
testing laboratories within 24 hours after blood
participate.
sampling. Serum AMH levels were determined
using the AMH Gen II assay (Beckman
Inclusion criteria
Coulter, Texas, USA; lowest detection limit
Patients between the ages of 18 - 45 cur-
0.08 ng/mL) and the FSH level was deter-
rently receiving ovarian stimulation with a
mined using the electrochemiluminescence
gonadotropin - releasing hormone (GnRH)
method (Roche, Mannheim, Germany; assay
antagonist protocol and recombinant FSH at
sensitivity 0.100 mIU/mL). To determine AFC
the National Assisted Reproductive Technolo-
levels, eligible subjects underwent transvagi-
gy Center were included in the study.
nal 2-dimensional ultrasounds (7.5MHz, Aloka,
Japan) on day 2 of their cycle. Total AFC level
Exclusion criteria
was measured by including all follicles of 2 Patients who had undergone other stimula-
10 mm in both ovaries.
tion regimens, such as the long protocol and
the agonist protocol, or who had participated
3. Research ethics
in egg donation, were ineligible to participate.
Research subjects were informed about
the goals of the study and voluntarily agreed
2. Methods
to participate. All personal information was be
This prospective study was conducted at
the National Hospital of Obstetrics and Gynecology in Vietnam from October 2014 to June
kept confidential. The study protocol was
approved by National Hospital of Obstetrics
and Gynecology.
2015. The study included 600 IVF patients
receiving the GnRH antagonist protocol with
III. RESULTS
recombinant FSH. The starting dose of recombinant FSH was based on patient age, AMH
level,
and
AFC
level.
Human
chorionic
gonadotrophin (hCG) was administered when
58
1. Patient characteristics and ovarian
stimulation outcomes
600 patients were eligible to participate in
JMR 105 E1 (7) - 2016
JOURNAL OF MEDICAL RESEARCH
the study. Demographic and clinical data, in-
was 12.09 ± 5.74. Te lowest number of folli-
cluding basal AFC, AMH, and FSH, were as
cles was 3 follicles, while the highest number
follows:
of follicles was 30 follicles. Finally, oocytes
The average age of participants was 31.7 ±
retrieved per trigger averaged to 13.21 ± 6.66,
5.2, with the group of 30 - 34 year olds
with a range from 0 - 30 oocytes.
accounting for 42% of the patients. The
Results of ovarian response
youngest patient was 18 years old and the
oldest was 45 years old.
54.7% of patients had primary infertility,
while 45.3% had secondary infertility. The average duration of infertility was 5.0 ± 3.2
years, with 52.2% having less than 5 years of
Poor response accounted for 4.7% of
participants (28 patients in total), normal
response accounted for 62.3% of participants
(374 patients in total), and high response
accounted for 33% of participants (198 patients in total).
infertility. 33 patients had an infertility duration
of more than 10 years. 44.2% of patients had
unexplained infertility.
Characteristics of AFC, AMH, basal
FSH, and E2
2. Comparing the predictive value of
AMH, AFC, and FSH for predicting high
ovarian response
Our data showed that an AFC threshold of
The lowest AFC value was 1, the highest
8 had a sensitivity of 78.7% and a specificity of
AFC value was 30 and the average AFC value
52% for predicting high ovarian response. The
was 13.0 ± 10.8. In terms of AMH level, the
AFC value was highly correlated with the num-
lowest AMH level was 0.2, while the highest
ber of oocytes retrieved that reflex ovarian
AMH level was 23.6 and the average AMH
reserve, with a correlation coefficient of r =
level was 4.57 ± 3.25.
0.34 (p < 0.001). AFC had a weak correlation
The lowest basal FSH level was 0.09, the
highest FSH level was 15.00 and the average
FSH level was 5.97 ± 4.56.
with high ovarian response (r = 0.167,
p < 0.05), indicating their poor value as indicator for high ovarian response.
Finally, the lowest E2 level was 1.54, the
In terms of FSH, our study found that the
highest E2 level was 174.00 and the average
FSH threshold to predict high ovarian re-
E2 level was 36.22 ± 19.00.
sponse was 6.14 (IU/L), with 53.2% sensitivity
Ovarian stimulation and cycle outcomes
and 72.7% specificity.
The average number of total rFSH doses
We could not determine the predictive
was 1971,2 ± 753,4 IU, with the lowest dose at
value of E2, since there were no statistical
400 IU and the highest dose at 6750 IU.
differences between E2 concentration among
Duration of ovarian stimulation was 9.84 ±
the 3 groups. We also found no correlation
1.16 days. The shortest duration of ovarian
between the concentration of E2 and the num-
stimulation was eight days, while the longest
ber of oocytes retrieved.
was 15 days.
The average number of follicles ≥ 14mm
JMR 105 E1 (7) - 2016
* The predictive value of AMH for high
ovarian response
59
JOURNAL OF MEDICAL RESEARCH
Table 1. The predictive value of AMH for high ovarian response
High ovarian response (> 15 oocytes retrieved)
Threshold value
Sensitivity
Specificity
3.125
85%
45%
3.49
81%
51%
3.54
80%
52%
3.62
78%
54%
3.87
73%
58%
3.95
73%
59%
4.04
73%
61%
4.12
72%
61%
4.21
71%
61%
4.25
69%
62%
AMH (ng/ml)
The AMH threshold to predict high ovarian response was 4.04 ng/ml, with 73% sensitivity and
61% specificity.
The predictive value of AFC for high ovarian response
Table 2. The predictive value of AFC for high ovarian response
High ovarian response (> 15 oocytes retrieved)
Threshold value
Sensitivity
Specificity
5.5
96.4%
14.1%
6.5
92.9%
18.6%
7.5
88.3%
25%
8.5
85.3%
34.1%
9.5
82.8%
42.4%
10.5
78.7%
52.0%
11.5
70.7%
56.3%
12.5
61.6%
61.6%
13.5
50.0%
68.6%
14.5
44.4%
72.9%
AFC
60
JMR 105 E1 (7) - 2016
JOURNAL OF MEDICAL RESEARCH
An AFC threshold of 8 had a sensitivity of 78.7% and a specificity of 52% for predicting high
ovarian response.
The predictive value of FSH for high ovarian response
Table 3. The predictive value of FSH with high ovarian response
High ovarian response (> 15 oocytes retrieved)
FSH (IU/L)
Threshold value
Sensitivity
Specificity
6.07
56.0%
69.1%
6.09
55.8%
69.2%
6.10
53.5%
72.2%
6.12
53.2%
72.2%
6.14
53.2%
72.7%
6.15
53.0%
72.7%
6.17
52.7%
72.7%
6.18
52.0%
72,7%
6.19
51.7%
72.7%
The FSH threshold to predict high ovarian response was 6.14 (IU/L), with 53.2% sensitivity
and 72.7% specificity.
Comparing the predictive value of AMH, AFC and FSH for predicting high ovarian
response
Table 4. AMH, AFC and FSH thresholds to predict high ovarian response
High ovarian response (> 15 oocytes retrieved)
Threshold value
Sensitivity
Specificity
AUC
AMH (ng/mL)
4.04
73%
61%
71%
AFC
10.5
78.7%
52.0%
65%
FSH (IU/L)
6.14
53.2%
72.7%
62.7%
AMH had the best predictive value in determining which women would have high ovarian
response, followed by AFC and finally FSH, as demonstrated by each measurement’s sensitivity
and specificity.
JMR 105 E1 (7) - 2016
61
JOURNAL OF MEDICAL RESEARCH
Figure 1. Receiver operating characteristics (ROC) curves for AMH, FSH and AFC in
predicting high ovarian response
Multivariate analysis for predictive factors of high ovarian response
Table 5. Multivariate analysis for predictive factors of high ovarian response
High ovarian response (n = 198 patients)
Predictive factors
P
OR
CI 95%
AMH ≥ 4.04 (n = 307) < 4.04 (n = 293)
< 0.001
2.69
1.80 - 4.00
AFC ≥ 10.5 (n = 345) < 10.5 (n = 255)
< 0.001
2.67
1.76 - 4.07
FSH ≤ 6.14 (n = 332) > 6.14 (n = 268)
< 0.001
2.11
1.42 - 3.14
The adjusted odds ratio (OR) of having a high ovarian response based on AMH ≥ 4.04 ng/ml
was 2.69, as compared with AMH < 4.04 ng/ml (95% CI, p < 0.001). Conversely, the OR of having
a high ovarian response based on AFC ≥ 10.5 was 2.67, as compared with AFC < 10.5 (95% CI,
p < 0.001). Finally, the OR of having a high ovarian response based on FSH ≤ 6,14 IU/l was 2.11,
as compared with FSH > 6.14 IU/l (95% CI, p < 0.001).
IV. DISCUSSION
Our results showed that AMH and AFC are
larger than the area under the curve for AFC
good predictors of high ovarian response in
(AUC = 65%). AMH is more highly correlated
women undergoing the GnRH antagonist
to the number of oocytes retrieved at pick - up
protocol. AMH appears to be a superior
(r = 0.338) than AFC (r = 0.167). We found no
predictor to AFC, since we found that the area
correlation between FSH and E2 and the
under the curve for AMH (AUC = 71%) was
number of oocytes retrieved, indicating that
62
JMR 105 E1 (7) - 2016
JOURNAL OF MEDICAL RESEARCH
these factors are not predictors of high ovarian
as a predictive marker for ovarian response.
response. These results are in agreement with
They concluded that AMH is the most reliable
previous studies [11; 12].
marker
Recent studies have suggested that the
use of AMH as a marker of ovarian response
has
clinical
advantages
when
assessing
ovarian reserve. A meta-analysis with data
from more than 20 studies concluded that
AMH was a more accurate and robust
biomarker of ovarian response in IVF than
of ovarian reserve
[16].
Moreover,
AMH has a number of obvious clinical
advantages, since AMH levels vary less
across different menstrual cycles, within one
menstrual cycle, during a pregnancy period,
and when undergoing GnRH agonist treatment
[13]. This variation is often seen with other
ovarian
biomarkers
[13].
AMH
can
be
assessed at any time point during the
FSH, LH, E2 and inhibin B [13].
menstrual cycle, whereas AFC and other
Our
findings
are
in
agreement
with
previous studies which found that the combination of AFC and AMH enhances prediction
of ovarian response. However, there are
limited data and conflicting results in the
literature with regards to comparing AMH and
AFC to predict the number of oocytes
retrieved. Ficicioglu et al revealed that the
level of AMH, as an indicator of ovarian
reserve, is more sensitive and specific than
AFC, with an AUC for AMH of 92% and for
AFC of 78% [12]. On the contrary, Mutlu et al
measured basal levels of AMH, FSH and AFC
in 192 patients prior to IVF treatment and
demonstrated that AFC is better than AMH at
predicting poor ovarian response [14]. The
AUC values from this study were 93%, 86%
and 75% for AFC, AMH, FSH, respectively,
indicating that in our study, AFC was better at
predicting poor ovarian response. Similarly,
Kwee et al found the AUC for AFC and AMH
to
be
93%
and
85%,
respectively,
biomarkers have to be measured at the start
of the menstrual cycle. AFC can be used as a
prognostic indicator of ovarian response in
patients with a history of ovarian surgery, or in
patients with endometriosis in the ovaries. So
far, AMH has been found to be a useful,
convenient, and promising marker to assess
ovarian
reserve
and
to
predict
ovarian
response.
The real value of the above information lies
in its ability to help predict a female patient's
required dose of rFSH. In our study, the target
for ovarian stimulation was set at 7 – 15
oocytes at retrieval. Seven or more oocytes
are considered to give a reasonable chance
(∼25%) of pregnancy, and the risk of developing
moderate/severe
ovarian
hyperstimulation
syndrome (OHSS) is low in patients with ≤ 15
oocytes. Severe OHSS was most frequent in
patients with high ovarian reserve and who
were given high rFSH doses. In contrast, in
patients with low ovarian reserve and who
demonstrating that AFC seemed to perform
were treated with low or medium doses of
slightly
rFSH, fewer or no oocytes were retrieved,
better
than
AMH
for
predicting
hyperresponse [15].
cycles were cancelled, and the proportion of
Recently, Fleming et al reviewed the cur-
oocytes retrieved below the stimulation target
rent evidence evaluating individualized ovarian
was higher. In these patients, high doses of
stimulation protocols using AMH concentration
rFSH may be appropriate.
JMR 105 E1 (7) - 2016
63
JOURNAL OF MEDICAL RESEARCH
In summary, clinical practitioners should
antimullerian hormone levels in a population of
use AMH and AFC to assess ovarian reserve
infertile women: a multicenter study. Fertil
in ovarian stimulation, to both increase the
Steril, 95(7), 2359 - 2363 e1.
efficiency of the number of oocytes obtained at
4. La Marca A., Sighinolfi G., Radi D.
retrieval and to decrease the risk of develop-
(2010). Anti-Mullerian hormone (AMH) as a
ing OHSS in IVF.
predictive marker in assisted reproductive
technology (ART). Hum Reprod Update, 16(2),
V. CONCLUSION
113 - 130.
In conclusion, our study provides additional
5. Lee H., Liu H (2008). Serum anti-
data to support the clinical value of AMH and
mullerial hormone and estradiol levels as pre-
AFC in predicting high ovarian response in
dictions of ovatian hyperstimulation syndrome
women
in assisted reproduction technology cycles.
undergoing
the
IVF
antagonist
protocol. AMH seems to be a better predictor
Hum Reprod, 23, 160 - 167.
(AUC = 71%) than AFC (AUC = 65%). The
6. Nardo G., Gelbaya A et al (2009). Cir-
sensitivity and specificity for AMH in predicting
culating basal anti-mullerian hormone levels
high ovarian response were 73% and 61%,
as predictor of ovarian response in women
respectively, while the sensitivity and specific-
undergoing ovarian stimulation for in vitro fer-
ity for AFC were 78.7% and 52.0%, respec-
tilization. Fertil Steril, 92(5), 1586 - 1593.
no predictive value in
7. Broer L., Eijkemans J., Scheffe J et al
determining high ovarian response (r = 0.10
(2011). Anti-mullerian hormone predicts meno-
and p > 0.05).
pause: a long term follow up study in normoo-
tively. FSH has
vulatory women. J Clin Endocrinol Metab, 96
ACKNOWLEDGEMENTS
(8), 2532 - 2539.
We would like to express our deepest
8. Van Rooij A., Broekmans J et al
gratitude to all staff from the National ART
(2002). Serum anti-mullerian hormone levels:
center at the National Hospital of Obstetrics
a novel measure of ovarian reserve, Hum Re-
and Gynecology in Vietnam.
prod, 17, 3065 - 3071.
9. Himabindu Y., Sriharibaru M., Gopina-
REFERENCES
1. Sunkara SK., Rittenberg V., RaineFenning N et al (2011). Association between
the number of eggs and live birth in IVF treatment: an analysis of 400 135 treatment cycles.
Hum Reprod, 26(7), 1768 - 1774.
2. Nikolaou D., Templeton A (2004). Early
ovarian ageing. Eur J Obstet Gynecol Reprod
Biol, 113(2), 126 - 133.
than K et al (2013). Anti-Mullerian hormone
and antral follicle count as predictors of ovarian response in assisted reproduction. J Hum
Reprod Sci, 6, 27 - 31.
10. Ferraretti A., Fauser M., Tarlatzis B.,
Nargund G., Gianaroli L (2011). Ehsre
consensus on the definition of “poor response”
to ovarian stimulation for in vitro fertilization:
the Bologna criteria.
Hum Reprod, 26(7),
1616 - 1624.
3. Almog B., Shehata F., Suissa S. et al
11. Nardo G., Gelbaya A et al (2009). Cir-
(2011). Age-related normograms of serum
culating basal anti-mullerian hormone levels
64
JMR 105 E1 (7) - 2016
JOURNAL OF MEDICAL RESEARCH
as predictor of ovarian response in women
ovarian response better than anti-mullerian
undergoing ovarian stimulation for in vitro fer-
hormone but age is the only predictor for live
tilization. Fertil Steril, 92(5), 1586 - 1593.
birth in invitro fertilization cycles. J Assist Re-
12. Ficicioglu C., Kutlu T., Baglam E et
prod Genet, 30, 657 - 665.
al (2005). Early follicular anti-mulerian hor-
15. Kwee J., Schats R., McDonnell J et al
mone as an indicator of ovarian reserve. Fertil-
(2008). Evaluation of anti-mullerian hormone
ity and Steril, 85(3), 592 - 596.
as a test for the prediction of ovarian reserve.
13. La Marca A., Sighinolfi G., Radi D.
Fertility and Sterility, 90(3), 737 - 743.
(2010). Anti-Mullerian hormone (AMH) as a
16. Fleming R., Broelmans F., Calhaz-
predictive marker in assisted reproductive
Jorge C et al (2013). Can anti-mullerian hor-
technology (ART). Hum Reprod Update, 16(2),
mone concentrations be used to determine
113 - 130.
gonadotrophin dose and treatment protocol for
14. Mutlu F., Erdem M., Erdem A et al
(2013). Antral follicle count determines poor
JMR 105 E1 (7) - 2016
ovarian stimulation?. Reproductive BioMedicine Online, (26), 431 - 439.
65