Anda di halaman 1dari 12

Clinical utility of magnetic resonance

imaging and ultrasonography for

diagnosis of polycystic ovary
syndrome in adolescent girls
Lisa E. Kenigsberg, M.D.,a Chhavi Agarwal, M.D.,a Sanghun Sin, M.S.,b Keivan Shifteh, M.D.,c
Carmen R. Isasi, M.D., Ph.D.,d Rebecca Crespi, N.P.,a Janeta Ivanova,b Susan M. Coupey, M.D.,e
Rubina A. Heptulla, M.D.,a and Raanan Arens, M.D.b
Division of Pediatric Endocrinology, b Division of Respiratory and Sleep Medicine, e Division of Adolescent Medicine,
Childrens Hospital at Monteore, d Department of Epidemiology and Population Health, Albert Einstein College of
Medicine; and c Department of Radiology, Monteore Medical Center, Albert Einstein College of Medicine, Bronx,
New York

Objective: To evaluate ovarian morphology using three-dimensional magnetic resonance imaging (MRI) in adolescent girls with and
without polycystic ovary syndrome (PCOS). Also compare the utility of MRI versus ultrasonography (US) for diagnosis of PCOS.
Design: Cross-sectional study.
Setting: Urban academic tertiary-care childrens hospital.
Patient(s): Thirty-nine adolescent girls with untreated PCOS and 22 age/body mass index (BMI)-matched controls.
Intervention(s): Magnetic resonance imaging and/or transvaginal/transabdominal US.
Main Outcome Measure(s): Ovarian volume (OV); follicle number per section (FNPS); correlation between OV on MRI and US; pro-
portion of subjects with features of polycystic ovaries (PCOs) on MRI and US.
Result(s): Magnetic resonance imaging demonstrated larger OV and higher FNPS in subjects with PCOS compared with controls.
Within the PCOS group, median OV was 11.9 (7.7) cm3 by MRI compared with 8.8 (7.8) cm3 by US. Correlation coefcient between
OV by MRI and US was 0.701. Due to poor resolution, FNPS could not be determined by US or compared with MRI. The receiver oper-
ating characteristic curve analysis for MRI demonstrated that increasing volume cutoffs for PCOs from 1014 cm3 increased specicity
from 77%95%. For FNPS on MRI, specicity increased from 82%98% by increasing cutoffs from R12 to R17. Using Rotterdam cut-
offs, 91% of subjects with PCOS met PCO criteria on MRI, whereas only 52% met criteria by US.
Conclusion(s): Ultrasonography measures smaller OV than MRI, cannot accurately detect follicle number, and is a poor imaging mo-
dality for characterizing PCOs in adolescents with suspected PCOS. For adolescents in whom
diagnosis of PCOS remains uncertain after clinical and laboratory evaluation, MRI should be
considered as a diagnostic imaging modality. (Fertil Steril 2015;104:13029. 2015 by Amer- Use your smartphone
ican Society for Reproductive Medicine.) to scan this QR code
Key Words: PCOS, ovarian imaging, MRI, adolescent and connect to the
discussion forum for
this article now.*
Discuss: You can discuss this article with its authors and with other ASRM members at http:// * Download a free QR code scanner by searching for QR
scanner in your smartphones app store or app marketplace.

olycystic ovary syndrome ated with overweight and obesity, However, there continues to be signi-
(PCOS) is one of the most com- affecting 5%10% of adolescent girls cant debate over the criteria for accu-
mon endocrine disorders associ- and women of reproductive age (1). rate diagnosis, specically in
adolescents, in whom no age-specic
Received April 8, 2015; revised July 31, 2015; accepted August 1, 2015; published online September 3,
2015. diagnostic criteria have been estab-
L.E.K. has nothing to disclose. C.A. has nothing to disclose. S.S. has nothing to disclose. K.S. has nothing lished (2). The 1990 National Institutes
to disclose. C.R.I. has nothing to disclose. R.C. has nothing to disclose. J.I. has nothing to disclose.
S.M.C. has nothing to disclose. R.A.H. has nothing to disclose. R.A. has nothing to disclose.
of Health criteria dened PCOS as a dis-
Supported by National Institutes of Health funding (grant R01 HL-105212). order of hyperandrogenism and oligoo-
Reprint requests: Raanan Arens, M.D., Division of Respiratory and Sleep Medicine, The Childrens Hos- vulation. In 2003, the Rotterdam
pital at Monteore, 3415 Bainbridge Avenue, Bronx, New York 10467 (E-mail: rarens@ criteria were proposed, requiring two
of three features including: [1] oligoo-
Fertility and Sterility Vol. 104, No. 5, November 2015 0015-0282/$36.00
Copyright 2015 American Society for Reproductive Medicine, Published by Elsevier Inc.
vulation or anovulation, [2] hyperan- drogenism (clinical or biochemical),

1302 VOL. 104 NO. 5 / NOVEMBER 2015

Fertility and Sterility

and [3] polycystic ovaries (PCOs) (3). The 2006 Androgen and body mass index (BMI), had regular monthly menstrual
Excess-PCOS Society criteria, which encompass the National cycles, dened as R10 menses per year (14), and no clinical
Institutes of Health and the Rotterdam criteria, also include evidence of hyperandrogenism. Subjects were excluded if
the assessment of ovaries for polycystic morphology (4). they were taking any hormonally active medication including
Therefore, if the Rotterdam or the Androgen Excess-PCOS insulin sensitizers or oral contraceptive (OC) pills, were preg-
Society criterion is used, evaluation for PCOs should be per- nant, or had other endocrine disorders. Control subjects were
formed. As stated by the Rotterdam consensus, PCOs are imaged during the early follicular phase of their menstrual cy-
dened as including 12 or more follicles per ovary measuring cle (days 210). Due to irregular menses, the timing of imag-
29 mm in diameter and/or ovarian volume (OV) more than ing in subjects with PCOS was unrelated to menses. The MRI
10 cm3, with features in one ovary being sufcient for diag- and US imaging studies were performed within a 24-hour
nosis (3). These criteria were created based on studies evalu- time frame in the subjects with PCOS.
ating ovarian morphology by transvaginal ultrasound (US) Informed consent was obtained from the legal guardian
in adult women (57). However, in virginal adolescent girls, of each subject younger than 18 years and from 18-year-
transvaginal US is contraindicated and transabdominal US old subjects themselves before participation in the study.
is not optimal due to poor resolution and central adiposity. The study was approved by the Institutional Review Board
Therefore, transvaginal US data from adult women may be at Monteore Medical Center and Albert Einstein College of
inappropriate to derive cutoff values to apply to Medicine.
adolescents, who not only differ in age, but also in method
of ultrasound. In addition, follicle number per ovary and
Magnetic Resonance Imaging
follicle number per section (FNPS) are used interchangeably
in the PCOS literature (8, 9). Acquisition. All subjects with PCOS and controls underwent
Due to these limitations, ovarian imaging by US, particu- a pelvic MRI using a 3T Philips Achieva system (Philips Med-
larly in obese adolescent girls with PCOS, may preclude inter- ical Systems). Axial and coronal T2-weighted spin-echo se-
pretation and diagnosis and should be challenged. An quences were created for optimal visualization,
improved imaging modality for characterizing ovarian characterization, and distribution of ovarian follicles. Axial
morphology is needed to assist with diagnosis of PCOS in T2-weighted images were obtained (TR, 4,500 ms; TE,
this age group. In recent years, several small studies have 80 ms) with slice thickness of 4 mm and slice spacing of
used magnetic resonance imaging (MRI) to examine ovarian 5 mm. The eld of view during axial image acquisition was
morphology in adult women (10, 11) and adolescents (8, 12) 210  229 mm (frequency encoding  phase encoding direc-
with PCOS. Although these studies conrmed the presence tions. respectively). Coronal T2-weighted images were ob-
of large ovaries and numerous small follicles, volumetric tained (TR, 4,500 ms; TE, 80 ms) with a slice thickness of
calculations were based on estimated volumes derived from 3 mm and slice spacing of 4 mm. The eld of view during cor-
two-dimensional images and not true three-dimensional vol- onal image acquisition was 250  231 mm (frequency encod-
ume rendering. ing  phase encoding directions, respectively) (Fig. 1A). All
Thus, the aims for the present study were to perform a images were archived on a picture archiving and communica-
detailed comparison of ovarian ultrastructure including OV tion system (PACS) network.
and follicle count in adolescents with and without PCOS using Analysis. The DICOM les were analyzed using a DICOM soft-
three-dimensional volumetric analysis and to compare MRI ware reader, Amira 5.45 for image linear, area, and volumetric
versus US for identication of PCO features in a large group analysis. A board certied radiologist reviewed the images in a
of adolescent girls with PCOS. blinded fashion. In the axial and coronal planes, a region of
interest was determined around the ovary outer margin on
MATERIALS AND METHODS each slice to calculate each slice area. The OV was later deter-
mined by multiplying each area by slice thickness. Follicle
count was determined on the axial slice encompassing the
Adolescent females, aged 1318 years, with PCOS and control largest ovarian diameter, and the number of 2- to 9-mm folli-
girls without PCOS were recruited for this study from the pe- cles was counted, giving a total FNPS for each subject as out-
diatric endocrine and adolescent medicine clinics at the Chil- lined by Lujan et al. (9). Ovarian follicle distribution was noted
drens Hospital at Monteore in the Bronx, New York, from to be peripheral, with follicles conned to the outermost
September 2010 through August 2014, as part of a larger margin of the ovary, or random. The ovarian stromal area
study evaluating sleep disordered breathing, body composi- was calculated as a percentage of total ovarian area from an
tion, and metabolic parameters in adolescents. Subjects axial slice encompassing the largest ovarian diameter.
with PCOS were consecutively recruited during the study
period after a new diagnosis of PCOS was established by the
subjects physician, based on biochemical hyperandrogene- Ultrasonography
mia and irregular, infrequent menstrual bleeding, in accor- Subjects with PCOS also had a pelvic US, either transvaginal
dance with National Institutes of Health criteria (13). (n 12; Fig. 1B) if they were sexually active or transabdomi-
Hyperandrogenemia was dened as total T >41 ng/dL nal (n 21; Fig. 1C) if they were virginal. No pelvic US imag-
and/or free T >3.9 pg/mL, according to our clinical labora- ing was done on control subjects as it was not indicated
torys reference ranges. Control subjects, matched for age clinically and was not part of the larger study protocol.

VOL. 104 NO. 5 / NOVEMBER 2015 1303


Acquisition. Pelvic sonography was performed on a GE Logiq

FIGURE 1 E9 unit, using an IC5-9-D (310 MHz) or C1-5-RS (25 MHz)
transducer for transvaginal or transabdominal imaging,
respectively. Image acquisitions were performed by a certied
sonographer. The ovaries were imaged in the sagittal and
transverse planes, and three orthogonal measurements were
obtained in real time by the sonographer.
Analysis. The OV was calculated using the ellipse formula of
length (in centimeters)  width (in centimeters)  thickness
(in centimeters)  0.523 (8). Follicle count was ascertained
based on the clearest image frame, as determined by the radi-
ologist, as is currently done in clinical practice, giving a total
FNPS for each subject. Follicle distribution was categorized as
peripheral or random. Stromal area was unable to be deter-
mined by US.

Literature Review
A literature review was conducted to describe current litera-
ture on ovarian morphology in adolescents with PCOS by a
single investigator (L.E.K.). A PubMed search was performed
looking under Title/Abstract for the following search terms:
polycystic ovary syndrome, adolescent(s), ultrasound, ultra-
sonography, sonographic, MRI, imaging, ovarian volume,
and follicle number. Articles were excluded if the subjects
did not have PCOS, no imaging data was included (OV or fol-
licle count), or the article was not written in English.

Data Analysis
Descriptive analyses included calculation of the percentage
distribution and medians with their associated interquartile
range as well as means with SDs for categorical and contin-
uous variables. Before any analyses, variables were checked
for normality. Differences in the percentage distribution of
categorical variables were tested using c2 tests for differences
between categorical variables. When continuous variables
were not normally distributed, we used nonparametric tests
to examine associations. Group differences in means were
tested using the Wilcoxon-Mann test. The concordance corre-
lation coefcient was used to determine agreement between
MRI and US in subjects with PCOS (15) and a Bland Altman
was plotted. To assess the diagnostic performance of MRI
we calculated the sensitivity and specicity using different
thresholds for ovarian volume and follicle number, an overall
performance was calculated by receiver operating character-
istic curve analysis. P values < .05 were considered statisti-
cally signicant. Analyses were performed using STATA
(STATA/SE 14.0, StataCorp LP).
Ovarian morphology. (A) Magnetic resonance imaging; (B)
transvaginal ultrasound; (C) transabdominal ultrasound. (A) Coronal RESULTS
view by magnetic resonance imaging of an ovary in an adolescent
subject with polycystic ovary syndrome (PCOS). Follicles Subjects
(hyperintense) are clearly demarcated from stroma (hypointense). (B Thirty-nine subjects with PCOS and 22 controls without PCOS
and C) Ultrasound images from adolescent subjects with PCOS; (B)
transvaginal image; (C) transabdominal image. Follicles are were studied. Mean age of the PCOS group was similar to con-
visualized in black (hypoechoic) with stroma appearing more trols (16.68  1.56 years vs. 16.56  1.77 years; P .97), as
hyperechoic. Distinguishing individual follicles by ultrasound is was mean BMI z-score for subjects with PCOS and controls
difcult, precluding a follicle count.
(1.84  0.92 vs. 2.02  0.64; P .78). Within the PCOS group,
Kenigsberg. Ovarian imaging in adolescent PCOS. Fertil Steril 2015.
80% were overweight or obese, as were 91% of controls.

1304 VOL. 104 NO. 5 / NOVEMBER 2015

Fertility and Sterility

Comparison of Ovarian Morphology by MRI in the analysis of follicle count and distribution by US imaging
Subjects with PCOS versus Controls in all subjects.
MRI of the pelvis was performed in 39 subjects with PCOS and
in 22 controls. Subsequently, ovarian analysis was performed Application of Diagnostic Criteria
on 78 and 44 ovaries, respectively. As no clear diagnostic criteria are established for ovarian
Ovarian volume. By MRI, median OV and interquartile morphology by MRI in adolescents with PCOS, we compared
range was 11.8 (6.9) cm3 and 7.0 (4.4) cm3 in subjects our MRI measurements of OV and FNPS with the Rotterdam
with PCOS and controls, respectively (P< .001) (Table 1). criteria (OV >10 cm3, follicle number per ovary R12), as
Distribution of OV in subjects with PCOS and controls is well as with the newly proposed Lujan et al. (9) criteria (OV
shown in Figure 2A. There was no signicant difference 10 cm3, FNPS R9). In addition, in this study, we established
in mean stromal area between subjects with PCOS and MRI cutoffs for characterization of an ovary as polycystic in
controls (Table 1). adolescent subjects with PCOS by dening abnormal as more
than two SDs from the mean for our control subjects. We deter-
Follicle count and distribution. For ovaries in subjects with
mined the cutoff for polycystic ovarian volume to be >14 cm3
PCOS, the median FNPS was 12.0 (5.0) compared to 7.5
(based on a mean for controls of 7.4  3.4 cm3). For FNPS, this
(5.0) in controls (P< .001) (Table 1). The FNPS distribution is
cutoff was established as R17 (based on a mean for controls
shown in Figure 2B. When analyzing for follicle distribution,
of 7.7  4.4). Based on our receiver operating characteristic
62% of ovaries in the PCOS group compared with 37% in the
analysis curve (Supplemental Fig. 2, available online), these
control group had a peripheral distribution of follicles
new cutoffs allow for maximal specicity (Supplemental
(P .01) (Table 1).
Table 1, available online). Images were evaluated both in terms
of how many ovaries met criteria as well as how many individ-
Comparison of Ovarian Morphology in Subjects ual subjects met criteria, as ndings in one ovary are sufcient
with PCOS by MRI versus US to qualify as a criterion for diagnosis of PCOS (3).
We studied 39 subjects with PCOS. Of these, 33 underwent Table 1 shows the application of the three criteria listed to
both MRI and US imaging and subsequently 66 ovaries our MRI ndings in ovaries from adolescent subjects with and
were analyzed. without PCOS. Using an ovarian volume cutoff of >10 cm3 as
an indication of a PCO, 65% of ovaries of subjects with PCOS
Ovarian volume. The median OV (interquartile range) for the met this criterion; however, 23% of ovaries of subjects
66 ovaries in subjects with PCOS was 11.9 (7.7) cm3 by MRI without PCOS also met the criterion. By increasing the cutoff
compared with 8.8 (7.8) cm3 by US imaging (P .05) to 14 cm3, only 5% of ovaries from control subjects without
(Table 2). The concordance correlation coefcient (rho) in PCOS qualify as polycystic; however, in subjects known to
measuring OV by MRI and US imaging in subjects with have PCOS only 33% of ovaries now qualify as polycystic.
PCOS was 0.701 (P< .001). The Bland Altman plot illustrating When evaluating for FNPS meeting the criterion for a PCO,
the agreement between the two imaging modalities suggests 85%, 64%, and 17% of ovaries in subjects with PCOS met
moderate differences in ovary volume between tests, although the criteria of R9, R12, and R17, respectively. Ovaries in
a good agreement is maintained. This plot is presented in control subjects without PCOS have elevated FNPS in 41%,
Supplemental Figure 1, available online. 18%, and 2%, using cutoffs of R9, R12, and R17, respec-
Follicle count and distribution. Median FNPS was 13 (5.0) by tively, showing the increased specicity but decreased sensi-
MRI in subjects with PCOS (Table 2). Poor images precluded tivity of the more stringent criteria.


Comparison of ovarian morphology measured by magnetic resonance imaging in adolescent subjects with polycystic ovary syndrome (PCOS)
versus controls using Rotterdam, Lujan et al., and current study criteria.
PCOS (N [ 39) Controls (N [ 22) P value
Number ovaries 78 44
Median (IQR) OV (cm3) 11.8 (6.9) 7.0 (4.4) < .001
Mean  SD OV (cm3) 13.6  8.5 7.4  3.4 < .001
% (number) of ovaries with OV >10 cm3 (Rotterdam) 65% (51) 23% (10) < .001
% (number) of ovaries with OV >14 cm3 (current study) 33% (26) 5% (2) < .001
Median (IQR) FNPS (29 mm) 12.0 (5.0) 7.5 (5.0) < .001
Mean  SD FNPS (29 mm) 12.7  4.0 7.7  4.4 < .001
% (number) of ovaries with FNPS R12 (Rotterdam)a 64% (50) 18% (8) < .001
% (number) of ovaries with FNPS R9 (Lujan et al.) 85% (66) 41% (18) < .01
% (number) of ovaries with FNPS R17 (current study) 17% (13) 2% (1) .01
% (number) of ovaries with peripheral distribution of follicles 62% (48) 37% (15)b .01
% stromal area (stromal area/total area) 51  11.2 49  14.3 NS
Note: Data presented as median with interquartile range (IQR) or mean  SD. FNPS follicle number per section; NS not signicant; OV ovarian volume.
Based on follicle count from Rotterdam criteria.
Unable to evaluate on three ovaries thus denominator is 41 not 44.
Kenigsberg. Ovarian imaging in adolescent PCOS. Fertil Steril 2015.

VOL. 104 NO. 5 / NOVEMBER 2015 1305



10 Rotterdam Criterion
Current StudyCriterion
Number of Ovaries

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 70

Ovarian Volume Range (cm )
12 Lujan Criterion
Rotterdam Criterion
Current Study Criterion
Number of Ovaries

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

Follicle Number Per Section

Range of ovarian volume and follicle number per section by magnetic resonance imaging in adolescents with polycystic ovary syndrome (PCOS)
versus controls without PCOS. (A) Distribution of ovarian volume in subjects with PCOS and controls. The vertical dashed line delineates
polycystic ovarian volume (>10 cm3) from normal ovarian volume (%10 cm3) as per Rotterdam criteria and the solid line delineates polycystic
ovarian volume (>14 cm3) from normal ovarian volume (%14 cm3) as per the cutoff calculated in the present study. (B) Distribution of follicle
number per section in subjects with PCOS and controls. The vertical dashed line delineates follicle number per section characterizing a
polycystic ovary (PCO) (R12) versus normal (<12) per Rotterdam criteria, the dash/dot line per Lujan et al. criteria, and the solid line per the
cutoff calculated in the present study: PCO (R17) versus normal (<17).
Kenigsberg. Ovarian imaging in adolescent PCOS. Fertil Steril 2015.

In our adolescent subjects with PCOS who had both MRI statistically signicant. We were unable to determine individ-
and US performed, ovarian morphology was compared by im- ual follicle count by US in the present study. Therefore this
aging modality (Table 2). By MRI, 65% of ovaries had a vol- feature could not contribute to a determination of whether
ume of >10 cm3, whereas only 42% of these same ovaries or not an ovary was polycystic in our subjects.
were >10 cm3 by US (P< .001). When the volume cutoff for Within subject analysis was performed. Using Rotterdam
a PCO was increased to 14 cm3, 32% of ovaries met this crite- guidelines, 91% of subjects with PCOS met PCO criteria by
rion by MRI and 24% by US; however, this difference was not MRI, having at least one ovary characterized as polycystic,

1306 VOL. 104 NO. 5 / NOVEMBER 2015

Fertility and Sterility


Ovarian morphology in adolescent subjects with PCOS measured by magnetic resonance imaging versus ultrasound imaging modalities.
MRI (N [ 33) US (N [ 33) P value
Number of ovaries 66 66
Median (IQR) OV (cm3) 11.9 (7.7) 8.8 (7.8) .05
Mean  SD OV (cm3) 14.0  9.1 11.0  8.6 .05
% (number) of ovaries with OV >10 cm3 (Rotterdam)a 65% (43) 42% (28) < .001
% (number) of ovaries with OV >14 cm3 (current study) 32% (21) 24% (16) NS
Median (IQR) FNPS (29 mm) 13 (5) N/A
Mean  SD, FNPS (29 mm) 13.0  4.0 N/A
% (number) of ovaries with FNPS R12 (Rotterdam) 67% (44) N/A
% (number) of ovaries with FNPS R9 (Lujan et al.) 85% (56) N/A
% (number) of ovaries with FNPS R17 (current study) 18% (12) N/A
% (number) of ovaries with peripheral distribution of follicles 61% (40) 44% (28)b NS
% (number) of subjects with at least one PCO based on Rotterdam criteria 91% (30) 52% (17) .004
(OV >10 cm3 or FNPS R12)
% (number) of subjects with at least one PCO based on Lujan et al. criteria 94% (31) 52% (17) < .001
(OV >10 cm3 or FNPS R9)
% (number) of subjects with at least one PCO based on current study criteria 64% (21) 39% (13) .04
(OV>14 cm3 or FNPS R17)
Note: Data presented as median with interquartile range (IQR) or mean  SD. FNPS follicle number per section; MRI magnetic resonance imaging; N/A cannot be determined; NS not
signicant; OV ovarian volume; PCO polycystic ovary; US ultrasound.
Based on follicle count from Rotterdam criteria.
Unable to evaluate on three ovaries thus denominator is 63 not 66.
Kenigsberg. Ovarian imaging in adolescent PCOS. Fertil Steril 2015.

and 52% met criteria by US imaging (Table 2). Using the study criteria for PCOs by US but >90% had morphology consistent
cutoffs of Lujan et al. (9), 94% of subjects with PCOS met PCO with PCOs by MRI. Based on our results as well as previous
criteria by MRI, yet only 52% met criteria by US. With our studies in adolescents illustrating the difculty of obtaining
studys more stringent cutoffs, 64% of subjects with PCOS individual follicle counts by US (19), it seems that US is a
met PCO criteria by MRI, yet only 39% of the same subjects poor imaging modality for diagnosing PCOs in the adolescent
met criteria by US. age group by the current criteria.
Although MRI is more sensitive than US in characterizing
Literature Review PCOs using Rotterdam criteria, our study raises concern about
the appropriateness of using the current diagnostic criteria
A PubMed search of ovarian imaging in adolescents with that were developed based on transvaginal ultrasound imag-
PCOS revealed 54 articles. A total of nine studies (8, 12,16 ing of adult women to apply to MRI ndings in adolescent
22) met the inclusion criteria, in addition to the present girls. For adolescents, imaging is often used as a conrmatory
study. These studies were compared with the current study. test for PCOS or in an adolescent whose diagnosis is uncer-
Demographic data and pertinent results for each study are tain. Thus we believe that it is important to maximize speci-
presented in Supplemental Table 2, available online. city over sensitivity as we found that, using Rotterdam
criteria, nearly one quarter of our control girls without
DISCUSSION PCOS met criteria for PCOs. By using the new, more stringent
In the present study we have used MRI, a high precision radio- cutoffs that we derived from MRI ndings in our age- and
graphic modality, to characterize ovarian morphology in a BMI-matched adolescent control subjects without PCOS, an
large group of adolescent girls with PCOS, and compared it ovarian volume of 14 cm3 and FNPS of 17 maximizes speci-
with morphology in girls without PCOS. Using MRI, we rst city, albeit sacricing sensitivity. However, future large
conrmed that adolescents with PCOS had signicantly scale studies are needed to justify these cutoffs.
larger ovarian volume than adolescents without PCOS and a For adolescents who meet PCOS diagnostic criteria based
higher number of follicles with a peripheral distribution. In on clinical and laboratory ndings, an evaluation of ovarian
our secondary analysis, we demonstrated that MRI is an morphology may not be required. However, for adolescents
important imaging modality that allows for improved charac- who require radiologic evaluation due to uncertain diagnosis
terization of polycystic ovarian morphology when compared after a clinical evaluation, this study suggests that once
with US. We showed that ovarian volumes are smaller when appropriate diagnostic criteria are established for identifying
measured by US than when measured with three- PCOs in adolescent girls, MRI may be a useful imaging modal-
dimensional MRI. In addition, we found that US imaging ity. In addition, some sources now advocate the need for ad-
did not allow for individual follicle counts in our predomi- olescents to meet all three criteria of hyperandrogenism,
nantly overweight and obese adolescent girls, in contrast to PCOs, and irregular menses to be diagnosed with PCOS (23).
MRI, in which follicle counts and distribution were easily ob- If these criteria are adapted, the traditionally used modality
tained. Strikingly, in our adolescent subjects with known of US imaging will lead to underdiagnosis of PCOS in a signif-
PCOS, we found that nearly 50% did not meet Rotterdam icant number of adolescent girls, thereby precluding

VOL. 104 NO. 5 / NOVEMBER 2015 1307


appropriate early treatment and prevention of cardiovascular REFERENCES

and other morbidities associated with PCOS. 1. Azziz R, Woods KS, Reyna R, Key TJ, Knochenhauer ES, Yildiz BO. The prev-
Not only does MRI provide clearer, more precise images, alence and features of the polycystic ovary syndrome in an unselected pop-
which allows for identication of individual ovarian follicles, ulation. J Clin Endocrinol Metab 2004;89:27459.
this methodology is not altered by obesity. Magnetic reso- 2. Roe AH, Dokras A. The diagnosis of polycystic ovary syndrome in adoles-
nance imaging has the added advantage of allowing the clini- cents. Rev Obstet Gynecol 2011;4:4551.
cian to evaluate body composition simultaneously, which 3. Rotterdam ESHRE/ASRM-Sponsored PCOS consensus workshop group.
Revised 2003 consensus on diagnostic criteria and long-term health
may provide additional information to improve clinical
risks related to polycystic ovary syndrome (PCOS). Hum Reprod 2004;
care. Furthermore, ovarian volume calculated by MRI is com- 19:417.
puter generated based on volumes in continuous axial seg- 4. Azziz R, Carmina E, Dewailly D, Diamanti-Kandarakis E, Escobar-
ments, as opposed to volume estimation by US, which relies Morreale HF, Futterweit W, et al. The Androgen Excess and PCOS Society
on an assumption of a standardized ellipse formula. criteria for the polycystic ovary syndrome: the complete task force report.
A review of the literature supported the ndings in this Fertil Steril 2009;91:45688.
5. Pache TD, Wladimiroff JW, Hop WC, Fauser BC. How to discriminate be-
study that although ovarian volume may measure differently
tween normal and polycystic ovaries: transvaginal US study. Radiology
by MRI versus US, there is a correlation between the two mo- 1992;183:4213.
dalities (8), and in all studies OV was signicantly larger in 6. Van Santbrink EJ, Hop WC, Fauser BC. Classication of normogonado-
adolescents with PCOS compared with adolescents without tropic infertility: polycystic ovaries diagnosed by ultrasound versus endo-
PCOS, regardless of imaging modality (12, 16, 19, 20, 22) crine characteristics of polycystic ovary syndrome. Fertil Steril 1997;67:
(Supplemental Table 2). However, our study is unique in 4528.
that the OV is based on true three-dimensional volume 7. Jonard S, Robert Y, Cortet-Rudelli C, Pigny P, Decanter C, Dewailly D. Ultra-
sound examination of polycystic ovaries: is it worth counting the follicles?
rendering, as opposed to estimated volume derived from
Hum Reprod 2003;18:598603.
two-dimensional images. The lack of literature on follicle 8. Yoo RY, Sirlin CB, Gottschalk M, Chang RJ. Ovarian imaging by magnetic
number in adolescents with PCOS emphasizes the difculty resonance in obese adolescent girls with polycystic ovary syndrome: a pilot
of obtaining this measurement by US. In general, data on study. Fertil Steril 2005;84:98595.
ovarian morphology in adolescents with and without PCOS 9. Lujan ME, Jarrett BY, Brooks ED, Reines JK, Peppin AK, Muhn N, et al. Up-
remains inconsistent and scarce, making it difcult to reach dated ultrasound criteria for polycystic ovary syndrome: reliable thresholds
a consensus using the currently available literature. for elevated follicle population and ovarian volume. Hum Reprod 2013;
Only two studies have compared MRI and US ndings in
10. Barber TM, Alvey C, Greenslade T, Gooding M, Barber D, Smith R, et al. Pat-
adolescents with PCOS, including the current study and the terns of ovarian morphology in polycystic ovary syndrome: a study utilising
one by Yoo et al. (8), and ndings between these two studies magnetic resonance imaging. Eur Radiol 2010;20:120713.
are inconsistent. However, both studies found a signicant cor- 11. Hauth EA, Umutlu L, Libera H, Kimmig R, Forsting M. [Magnetic resonance
relation between OV by MRI and OV by US. Only Yoo et al. (8) imaging of the pelvis in patients with polycystic ovary syndrome]. Rofo 2009;
compared follicle count by different imaging modalities in ad- 181:5438.
12. Brown M, Park AS, Shayya RF, Wolfson T, Su HI, Chang RJ. Ovarian imaging
olescents with PCOS, nding a signicantly larger number of
by magnetic resonance in adolescent girls with polycystic ovary syndrome
follicles identied on MRI when compared with ultrasound. and age-matched controls. J Magn Reson Imaging 2013;38:68993.
Limitations of this study include a relatively small sample 13. Zawadski JK, Dunaif A. Diagnostic criteria for polycystic ovary syndrome: to-
size and inability to count follicles by US, preventing a com- wards a rational approach. Boston: Blackwell Scientic Publications; 1992:
parison of follicle number between MRI and US. As recom- 37784.
mended by Dewailly et al. (24), maximal resolution can be 14. Rieder J, Santoro N, Cohen HW, Marantz P, Coupey SM. Body shape and
obtained by US if a transducer frequency of R8 MHz is size and insulin resistance as early clinical predictors of hyperandrogenic
anovulation in ethnic minority adolescent girls. J Adolesc Health 2008;
used and this may allow for counting of individual follicles.
In addition, we evaluated ovarian morphology in adolescents 15. Lin LI. A concordance correlation coefcient to evaluate reproducibility. Bio-
with known PCOS. Additional studies are needed to conrm metrics 1989;45:25568.
our ndings in girls who are undergoing evaluation for 16. Chen Y, Yang D, Li L, Chen X. The role of ovarian volume as a diagnostic cri-
PCOS but have not yet been diagnosed. Our study did not terion for Chinese adolescents with polycystic ovary syndrome. J Pediatr
assess reproducibility, as only one radiologist reviewed the Adolesc Gynecol 2008;21:34750.
images. Future studies with multiple radiologists will allow 17. Dramusic V, Goh VH, Rajan U, Wong YC, Ratnam SS. Clinical, endocrino-
logic, and ultrasonographic features of polycystic ovary syndrome in Singa-
investigation of the reproducibility of MRI versus US.
porean adolescents. J Pediatr Adolesc Gynecol 1997;10:12532.
Overall, more research is needed to dene MRI character- 18. Pawelczak M, Kenigsberg L, Milla S, Liu YH, Shah B. Elevated serum
istics of ovarian morphology in adolescents with PCOS and to anti-Mullerian hormone in adolescents with polycystic ovary syndrome:
establish normal data. However, the current study demon- relationship to ultrasound features. J Pediatr Endocrinol Metab 2012;
strated that MRI allows for precise imaging of ovaries and 25:9839.
that US, due to underestimation of ovarian volume and 19. Rossi B, Sukalich S, Droz J, Grifn A, Cook S, Blumkin A, et al. Prevalence of
metabolic syndrome and related characteristics in obese adolescents with
inability to assess individual follicle number, may be insuf-
and without polycystic ovary syndrome. J Clin Endocrinol Metab 2008;93:
cient to properly characterize PCOs in a predominately obese 47806.
adolescent population. Although MRI is more expensive and 20. Villa P, Rossodivita A, Sagnella F, Moruzzi MC, Mariano N,
may require more resources, once normal values and cutoffs Lassandro AP, et al. Ovarian volume and gluco-insulinaemic markers in
are established, MRI will allow for improved detection of the diagnosis of PCOS during adolescence. Clin Endocrinol (Oxf) 2013;
ovaries with polycystic features. 78:28590.

1308 VOL. 104 NO. 5 / NOVEMBER 2015

Fertility and Sterility

21. Silfen ME, Denburg MR, Manibo AM, Lobo RA, Jaffe R, Ferin M, et al. Early 23. Carmina E, Obereld SE, Lobo RA. The diagnosis of polycystic
endocrine, metabolic, and sonographic characteristics of polycystic ovary ovary syndrome in adolescents. Am J Obstet Gynecol 2010;203:
syndrome (PCOS): comparison between nonobese and obese adolescents. 201.e15.
J Clin Endocrinol Metab 2003;88:46828. 24. Dewailly D, Lujan ME, Carmina E, Cedars MI, Laven J, Norman RJ, et al. De-
22. Youngster M, Ward VL, Blood EA, Barnewolt CE, Emans SJ, Divasta AD. Util- nition and signicance of polycystic ovarian morphology: a task force report
ity of ultrasound in the diagnosis of polycystic ovary syndrome in adoles- from the Androgen Excess and Polycystic Ovary Syndrome Society. Hum Re-
cents. Fertil Steril 2014;102:14328. prod Update 2014;20:33452.

VOL. 104 NO. 5 / NOVEMBER 2015 1309


Difference of MRI volume and US volume
-20 -10 0
-30 10 20

0 20 40 60
Mean of MRI volume and US volume

observed average agreement 95% limits of agreement

y=0 is line of perfect average agreement

Correlation of magnetic resonance imaging (MRI) and ultrasound (US)

ovarian volume by Bland Altman agreement analysis, used to analyze
the agreement between the two imaging modalities. Ovarian volume
on magnetic resonance imaging is highly correlated with ovarian
volume by ultrasound, with a concordance correlation coefcient of
0.701 (P<.001).
Kenigsberg. Ovarian imaging in adolescent PCOS. Fertil Steril 2015.

1309.e1 VOL. 104 NO. 5 / NOVEMBER 2015

Fertility and Sterility



0.00 0.25 0.50 0.75 1.00

1 - Specificity
Area under ROC curve = 0.8176


0.00 0.25 0.50 0.75 1.00

1 - Specificity
Area under ROC curve = 0.8036

Receiver operating characteristic (ROC) curve analysis for ovarian

volume (A) and follicle number per section (B) by magnetic
resonance imaging in adolescents with and without polycystic ovary
syndrome (PCOS). The ROC area under the curve was 0.82 for
ovarian volume and 0.80 for follicle number per section.
Kenigsberg. Ovarian imaging in adolescent PCOS. Fertil Steril 2015.

VOL. 104 NO. 5 / NOVEMBER 2015 1309.e2



Receiver operating characteristic curve analysis to evaluate the

sensitivity and specicity of magnetic resonance imaging for
characterization of polycystic ovarian morphology using different
cutoff thresholds for ovarian volume and follicle number per section.
Sensitivity (%) Specicity (%)
Ovarian volume
10 cm3 67 77
12 cm3 50 89
14 cm3 35 95
Follicle number per section
9 85 59
12 64 82
17 17 98
Kenigsberg. Ovarian imaging in adolescent PCOS. Fertil Steril 2015.

1309.e3 VOL. 104 NO. 5 / NOVEMBER 2015

Fertility and Sterility


Studies evaluating ovarian imaging in adolescent females with polycystic ovary syndrome.
N OV (in cm3) FNPS
Author (year) PCOS Control PCOS Control P value PCOS Control P value Image modality
Brown et al. (2013) 19 21 15.1  5.6 5.9  3.0 .0001 24.0  8.7 a
13.9  7.1 a
.0001 MRI
Chen et al. (2008) 69 26 9.25 (4.1715.12)b 4.44 (3.257.26)b < .01 TRUS
Dramusic et al. (1997) 150 Right: 11.4 TAUS
Left: 10.7c
Pawelczak et al. (2012) 23 12 9.65  4.58 TAUS
Rossi et al. (2008) 43 31 7.8  3.0 5.0  1.7 < .0001 TAUS
Silfen et al. (2003)
Nonobese 11 9.3  3.1 TAUS
Obese 20 13 8.2  4.4 5.9  3.9 NS TAUS
Villa et al. (2013) 86 48 9.63  4.3 4.6  1.9 < .001 TAUS
Yoo et al. (2005) 11 9.7  0.9 21.9  1.3 MRI
11.6  1.1 5.5  1.7 TAUS
Youngster et al. (2014) 54 98 8.2  3.9 5.96  2.4 < .0001 TAUS
Current study (2015) 39 22 11.8 (6.9)b 7.0 (4.4)b < .0001 12.0 (5.0)b 7.5 (5.0)b < .0001 MRI
33 8.8 (7.8)b TAUS and TVUS
Note: Data for all studies presented as mean  SD, unless otherwise indicated. FNPS follicle number per section; MRI magnetic resonance imaging; NS not signicant; OV ovarian volume;
PCOS polycystic ovary syndrome; TAUS transabdominal ultrasound; TRUS transrectal ultrasound; TVUS transvaginal ultrasound.
Follicle number per ovary.
Median (interquartile range).
Mean provided without SD.
Kenigsberg. Ovarian imaging in adolescent PCOS. Fertil Steril 2015.

VOL. 104 NO. 5 / NOVEMBER 2015 1309.e4