Clinical Opinion ajog.

org

Ovarian massedifferentiating benign from

malignant: the value of the International
Ovarian Tumor Analysis ultrasound rules
Jacques S. Abramowicz, MD; Dirk Timmerman, MD, PhD

various diagnostic modalities, of which

Ovarian cancer, the fifth most common cause of cancer death among women, has the ultrasound is an integral part.8
highest mortality rate of all gynecologic cancers. General survival rate is <50% but can The major reason to try and differen-
reach 90% if disease is detected early. Ultrasound is presently the best modality to tiate benign from malignant tumors is the
differentiate between benign and malignant status. The patient with a malignant mass recommendation to refer the patient with
should be referred to an oncology surgeon since results have been shown to be superior a malignant mass to a specialized center
to treatment by a specialist. Several ultrasound-based scoring systems exist for or an oncology surgeon since therapeutic
assessing the risk of an ovarian tumor to be malignant. The International Ovarian Tumor results have been shown to be superior to
Analysis group published 2 such systems: the ultrasound Simple Rules and the treatment by an obstetrician/gynecologist
Assessment of Different NEoplasias in the adneXa model. The Simple Rules classifies a specialist (previously called gener-
tumor as benign, malignant, or indeterminate and the Assessment of Different NEo- alist).9,10 Many scoring systems for
plasias in the adneXa model determines the risk for a tumor to be benign or malignant assessing the risk of an ovarian tumor to
and, if malignant, the risk of various stages. Sensitivity of the Simple Rules and be malignant, based on ultrasound find-
Assessment of Different NEoplasias in the adneXa model (using a cut-off of 10% to ings, have been proposed.11,12 Two,
predict malignancy) are 92% and 96.5%, respectively, and specificities are 96% and originating from the International
71.3%, respectively. These models are the best predictive tests for the preoperative Ovarian Tumor Analysis (IOTA) group,
classification of adnexal tumors. Their intent is to help the specialist make management are simple to use and extremely effective:
decisions when faced with a patient with a persistent ovarian mass. The models are the ultrasound Simple Rules13 and
simple, are easy to use, and have been validated in multiple reports but not in the United the assessment of different neoplasias in
States. We suggest they should be validated and widely introduced into medical practice the adnexa (ADNEX) model14 to predict
in the United States. the risk of malignancy of an ovarian mass.
The Simple Rules classifies a tumor as
Key words: ADNEX model, International Ovarian Tumor Analysis group, ovarian cancer, benign, malignant, or indeterminate and
scoring systems, Simple Rules, ultrasound the ADNEX model determines the risk
(in percent) for a tumor to be benign or
malignant and, if malignant, the risk of

T he numbers are known and

appalling: ovarian cancer is the fifth
most common cause of cancer death
new cases of ovarian cancer in the United
States, an incidence of >12 per 100,000
and 14,240 deaths from the disease.2
various stages (borderline tumor, stage I,
stage II-IV, or metastatic). The sensitivity
of the Simple Rules and ADNEX model
among women1 and has the highest Worldwide, the number of newly diag- (using a cut-off of 10% to predict ma-
mortality rate of all gynecologic cancers. nosed cases is presently approximately lignancy) are 92% and 96.5%, respec-
in 2016, there were an estimated 22,280 240,000.3 General survival rate is 46.2%, tively, and specificities are 96% and
but can reach 90% in stage I of the dis- 71.3%, respectively.15,16 The Simple Rules
ease.4 Unfortunately, only 15% of and the ADNEX model are not in-
From the University of Chicago, Chicago, IL (Dr
Abramowicz), and Leuven University, Leuven,
ovarian cancers are detected at this stage. struments for screening for ovarian can-
Belgium (Dr Timmerman). In later stages (III and IV), when cer but are the best predictive tests for the
Received April 22, 2017; revised July 10, 2017; approximately 70-75% of cancers are preoperative classification of adnexal tu-
accepted July 15, 2017. detected,5 the 5-year survival drops to mors. Why have they not penetrated the
Dr Abramowicz receives author royalties from 28%.6 United States?
UpToDate. Dr Timmerman reports no conflict of Ultrasound, alone or in combination
interest. with serum markers, appears to be the Screening for ovarian cancer or early
Corresponding author: Jacques S. best modality to detect the presence of diagnosis?
Abramowicz, MD. [email protected]. early ovarian cancer and to differentiate Screening for the risk of cancer is an
edu
between benign and malignant tumors obvious clinical goal, successful in
0002-9378/$36.00
when an ovarian mass is present.7 In fact, several cancers, such as the uterine cer-
ª 2017 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ajog.2017.07.019 there has been improvement in the vix. This, however, has not, so far, been
numbers quoted above with the use of very effective in ovarian cancer, where a

652 American Journal of Obstetrics & Gynecology DECEMBER 2017

ajog.org Clinical Opinion

better approach, at the moment, is an surveillance with transvaginal ultra- serum levels of CA-125 and human
attempt to diagnosis at early stages sound (TVS) and CA-125. However, epididymis factor 4.18 It classifies pa-
(preferably stage I). there are no current recommendations tients with a pelvic mass as being at a low
The only modality to determine who for less commonly identified mutations or high risk for malignant disease. It does
is at risk for ovarian cancer is patient and there is limited evidence to guide not, however seem to be much better
medical or family history: women with a them in expected cancer risk and than serum CA-125 alone,19 and is not,
strong family history of breast or ovarian appropriate risk reduction strategies.17 per se, a screening test. Similarly, other
cancer (1 first-degree relative [mother, Furthermore, these account for an esti- screening methods, employing various
daughter, sister] or inheritance of a mated 10-15% of ovarian cancers and serum markers, are also early diagnosis
particular faulty mutation in a gene population screening is currently not tools.20-22 A recent study23 reviewed and
associated with breast and ovarian can- recommended due to low prevalence of reanalyzed previous data from 4 studies:
cer susceptibility, eg, breast cancer gene the disease as well as the prohibitive cost the Prostate, Lung, Colorectal, and
[BRCA]1, BRCA2, and a mutation in the of such a large-scale screening policy. Ovarian Cancer Study24; the United
epithelial cell adhesion molecule gene Thus, a better concept is screening for Kingdom Collaborative Trial of Ovarian
[MLH1, MSH2, PMS1, PMS2, and early cancer, as opposed to screening for Cancer Screening Study8; the Kentucky
MSH6]). Risk is higher in women with 1 the risk of developing cancer. Screening Study25; and the Shizuoka
first-degree relative and 1 second-degree As stated by American Congress of Cohort Study of Ovarian Cancer
relative (grandmother or aunt) and even Obstetricians and Gynecologists Screening.26 The new study23 seems to
higher with
2 first-degree relatives. (ACOG)6 currently, there is no effective indicate that the prognosis for the
Risk is also very high in women with a strategy for ovarian cancer screening. In screening group is significantly better
personal history of breast cancer age <40 the United States, the only recom- than that of the control group (P ¼
years or who are diagnosed with hered- mended screening modality for high- .0017). Furthermore, the screening
itary nonpolyposis colorectal cancer or risk women is yearly serum testing for group contained significantly fewer stage
Lynch syndrome. CA-125 glycoprotein antigen and TVS, IV cases than the control group (P ¼
Based on this history, selective genetic although this method is far from ideal6 .005). It is important to specify that the
testing may be offered to women with an and, clearly, the ultrasound is used as a word “screening” in the various studies
estimated elevated risk. Women with tool for diagnosing ovarian pathology can be confusing: some subjects are
BRCA1 and BRCA2 mutations are and not precursor signs. The Risk of screened, some are not, but “screened”
at most increased risk of developing Ovarian Malignancy Algorithm is an means examined to detect early cancer,
ovarian cancer and are currently ovarian malignancy risk stratification not “undergoing a screening test to
advised to have a 6-monthly or yearly tool that includes a combination of predict cancer.” Furthermore, there is no

FIGURE 1
Simple Rules sonographic images for features of ovarian masses

Simple Rules: sonographic images for benign (B) and malignant (M) features of ovarian masses.
Abramowicz. US IOTA ovarian tumor rules. Am J Obstet Gynecol 2017.

DECEMBER 2017 American Journal of Obstetrics & Gynecology 653

Clinical Opinion ajog.org

ultrasound contrast agents.35 Color and

TABLE 1 spectral Doppler were originally
Simple Rules acclaimed as excellent differentiators
Benign features Malignant features between benign and malignant masses,
Unilocular cyst (B1) Irregular solid tumor (M1) specifically for decreasing the false-
positive rate of cancer diagnosis,
Solid component <7 mm in diameter (B2) Ascites (M2)
because of the neoangiogenesis present
Presence of acoustic shadows (B3)
4 Papillary structures (M3) in malignant masses, with creation of
Smooth multilocular tumor with largest Irregular multilocular mass >10 cm in low-resistance new vessels and, thus,
diameter <10 cm (B4) diameter (M4) clear Doppler patterns.36-38 Overlap be-
No detectable color Doppler signal (B5) Strong color Doppler signal (M5) tween normal and pathological values
Abramowicz. US IOTA ovarian tumor rules. Am J Obstet Gynecol 2017.
made this modality less successful than
originally thought,39 with false-positive
rates of almost 40%.40 It is possible that
the use of ultrasound contrast agents
evidence that screening for ovarian almost 6 years showed a specificity of may rekindle the excitement, particu-
cancer in the general and high-risk 98.5% for primary invasive epithelial larly with the use of agents targeted to
populations with any of the above ovarian cancer but a positive predictive tumor-growth factors.41
methods actually decreases mortality.27 value (PPV) of only 8.9%.25 In another
large (>202,000) United Kingdom Earlier ovarian masses scoring
Early diagnosis of ovarian cancer by study, asymptomatic postmenopausal systems and predictive models
ultrasound women, aged 50-74 years, were Together with simple observation and
Transvaginal sonography with Doppler, randomly assigned to no screening, description of the ovarian masses, many
described almost 40 years ago,28 gener- annual screening with TVS, or with CA- diagnostic algorithms have been gener-
ally has a sensitivity of <90% for early 125, followed up, if abnormal, by TVS.8 ated, based on scoring systems for ul-
ovarian cancer and a specificity of 94- Among screened women, the PPV was trasound findings,42-57 including IOTA
99%,29 with less than optimal general 5.3% for all primary ovarian and tubal ultrasound Simple Rules58 and, most
availability precluding annual screening. cancers and 2.8% for primary invasive recently, IOTA ADNEX model,14
The University of Kentucky study of cancer with no significant mortality detailed below. One of the earliest algo-
>37,000 asymptomatic women evalu- reduction compared to unscreened rithms considered whether an ovarian
ated annually with TVS in an expert women and with 10 women undergoing mass was unilocular, unilocular solid,
center and followed up for a mean of (arguably) avoidable surgery (for benign multilocular, multilocular solid, or
findings) for every ovarian cancer diag- solid.45 In addition, papillary pro-
nosed. In a study of 83,000 Japanese jections, if present, were counted.
TABLE 2 women,26 33 surgeries were performed Sensitivity and specificity for malignancy
ADNEX model criteria to detect each case of ovarian cancer, an was 82% and 92%, respectively. The Risk
Age unacceptable large number. In high-risk of Malignancy Index (RMI), incorpo-
women TVS did equally poorly with no rating ultrasound findings and levels of
Serum CA-125a
early disease identified.30 In a study of CA-125, was introduced in 199046 with
Type of center (oncology referral center or high-risk patients, however, PPV of CA- a specificity and PPV for primary inva-
general hospital) 125 plus TVS screening was 25.5%.31 sive cancer of 99.8% (95% confidence
Maximum diameter of lesion (mm) All tests do better in symptomatic interval, 99.7e99.9) and 19% (95%
Proportion of solid tissue (diameter of women,32 but, often, this is too late with confidence interval, 4.1e45.6), respec-
largest solid component divided by cancer diagnosed at advanced stages. At tively.59 It is beyond the scope of this
maximum diameter of lesion) the moment, the most important role of opinion piece to describe the specifics of
Presence of >10 cyst locules (yes/no) TVS is the differentiation between ma- each one. They are considered in detail in
lignancy and benignity when an ovarian several important publications.11,12,60,61
No. of intracavitary papillary projections
(0, 1, 2, 3, >3) mass is detected by ultrasound.10,33
Several ultrasound modalities alone or Performance of predictive models
Presence of acoustic shadows (yes/no)
in combination have been employed Previously, and before the publication of
Presence of ascites (yes/no) with various degrees of success to the IOTA Simple Rules, RMI was found
ADNEX, Assessment of Different NEoplasias in adneXa. attempt and precisely define the nature to outperform 82 other models.60 The
a
Although software will also calculate risk if no value for of ovarian masses: from 2-dimensional RMI I and the RMI II (product of the
CA-125 is available.
gray-scale, color Doppler,28 spectral serum CA-125 level, an ultrasound scan
Abramowicz. US IOTA ovarian tumor rules. Am J
Obstet Gynecol 2017. Doppler,34 3-dimensional ultrasound result, and menopausal state) had a
(with or without Doppler), and pooled estimate for sensitivity of 78% for

654 American Journal of Obstetrics & Gynecology DECEMBER 2017

ajog.org Clinical Opinion

FIGURE 2
Results of ADNEX model for benign tumor

Chance of tumor to be benign is 96.6%.

ADNEX, Assessment of Different NEoplasias in adneXa; IOTA, International Ovarian Tumor Analysis.
Abramowicz. US IOTA ovarian tumor rules. Am J Obstet Gynecol 2017.

a specificity of 87%. The RMI IV had a gynecologists, radiologists, and oncol- published recommendation on terms
sensitivity, specificity, PPV, and negative ogists, as well as many “basic” scientists: and definitions.63 Later, IOTA developed
predictive value (NPV) of 86.4%, 91%, physicists, clinical biologists, civil engi- the Simple Rules and mathematical
63.5%, and 97.5%, respectively.56 As neers, mathematicians, psychologists, models based on logistic regression 1-2,
opposed to the IOTA group’s recom- and biostatisticians. which are very easy to use in clinical
mendations, no standardized method of Large cohorts of patients with a practice to estimate the risk of malig-
describing sonographic findings or clas- persistent adnexal mass were recruited nancy.64 The Simple Rules15 consist of 2
sifying ovarian tumors based on these by the various centers. Research has sets of descriptors: benign and malignant
findings is implemented in the United focused on the development of predic- (Figure 1 and Table 1). Based on sono-
States, as recently pointed out by the tive models to estimate the risk of ma- graphic findings, 3 rules are applied: if
First International Consensus Report on lignancy (eg, various regression
1 malignant features are present, in the
Adnexal Masses.62 models64,65 and the use of artificial net- absence of benign features, the mass is
works51). These models are not intended classified as malignant; conversely, if
1
What is IOTA? to be used to screen for ovarian cancer benign features are present in the mass
The IOTA group was founded in 1999 but to determine, when an ovarian mass and there are no malignant features, it
by Dirk Timmerman, Lil Valentin, Tom is detected by ultrasound, the risk that can be confidently considered benign. If
Bourne, William Collins, Herman Ver- this tumor is malignant. This extensive both malignant and benign features are
relst, Sabine Van Huffel, and Ignace database and the close involvement of present or none is, the findings are
Vergote. Its first aim was to develop the civil engineering department at the inconclusive. This occurs in up to 24% of
standard terms, forms, definitions, and University of Leuven enabled previously cases.67 Other methods, including
measurements to describe the sono- developed prediction models to be tested various ultrasound criteria and mea-
graphic features of adnexal masses.63 and novel prediction models to be surements of CA-125 are not helpful68,69
IOTA is a network of approximately developed and externally validated. and the recommendation is to refer for
50 contributing clinical centers Because of lack of standard vocabulary an expert ultrasound.70 When this oc-
throughout the world (but mostly in and large variations between examiners curs, the sensitivity is 91% and the
Europe) coordinated from the Univer- in scoring sonographic findings specificity is 93%,15 yielding the best
sity of Leuven, Belgium. The group is as well as results less optimal results compared to other models.71
multidisciplinary and involves than initially suggested,66 IOTA first Application of the Simple Rules in

DECEMBER 2017 American Journal of Obstetrics & Gynecology 655

Clinical Opinion ajog.org

FIGURE 3
Results of ADNEX model for malignant tumor

Risk for tumor to be malignant is 92.6%. Within this group, risk of tumor to be borderline is 9.1%; stage I, 14.5%; stage II-IV, 61.6%; and metastatic, 7.4%.
ADNEX, Assessment of Different NEoplasias in adneXa; IOTA, International Ovarian Tumor Analysis.
Abramowicz. US IOTA ovarian tumor rules. Am J Obstet Gynecol 2017.

asymptomatic postmenopausal women benign, borderline, early-stage malignant receiver operating characteristic curve for
with an ovarian mass resulted in fewer (stage I), late-stage malignant (stage II- the discrimination between benign and
surgical interventions than when using IV). or metastatic.14 The model is based malignant tumors was 0.94 (0.93-0.95).16
the RMI, without added delay in diag- on clinical and ultrasound data from Figures 2 and 3 are examples of results, as
nosing malignancy.72 almost 6000 women recruited at 24 cen- generated by the software for a benign
The gold standard for diagnosis of ters in 10 countries14 and uses 9 pre- (chance of tumor to be benign: 96.6%)
malignancy is, naturally, histopatholog- dictors. There are 3 clinical criteria and 6 (Figure 2) and a malignant (chance of
ical confirmation. In a meta-analysis ultrasound descriptors (Table 2). All tumor to be malignant: 92.6%) (Figure 3)
comparing the power of 19 methods to subjects included had surgery, since, as tumor. It should be noted that the level of
discriminate between benign and ma- stated, the gold standard is pathological CA-125 was identical in both patients.
lignant adnexal masses before surgery, diagnosis. If only patients presumed to The intent of both the Simple Rules and
the Simple Rules had a sensitivity of 93% have a malignant tumor were operated the ADNEX is to help the specialist to
and a specificity of 81% when incon- on, this would clearly introduce a selec- improve triage and make more educated
clusive tumors were all considered ma- tion bias and prevent a false-negative decisions regarding management and/or
lignant.12 The model was prospectively assessment of the model. The model has referral to a subspecialist (ie, a gynecology
validated, and proved to have perfor- also been validated in several clinical oncologist) when faced with a persistent
mance close to that of subjective assess- studies.73-77 In clinical practice, when ovarian mass.
ment by an expert sonographer. using this model, the results are expressed As noted on the IOTA group website
Furthermore, users with different levels in percent probability that the tumor is (http://www.iotagroup.org/) ADNEX
of ultrasound experience were shown to benign or malignant and, if malignant, (or any model or software) cannot
obtain similar good results. The Simple what is the percentage risk of it being replace adequate training and experi-
Rules model is available at http://www. borderline, stage I, stage II-IV, or meta- ence in ultrasonography, nor can it
iotagroup.org/simplerules/. static. Results are presented both in compensate for poor-quality ultrasound
Recently, the ADNEX model was graphic form and numbers. This software equipment. The Simple Rules and the
introduced. This is the first predictive is also available to all on the IOTA website ADNEX model fare better than any
multiclass model, able to estimate the (http://www.iotagroup.org/adnexmodel/ other model, including the RMI.72 For
probability/risk that an adnexal tumor is site%20iota.html). The area under the a 1% risk cut-off, sensitivity was 99.7%,

656 American Journal of Obstetrics & Gynecology DECEMBER 2017

ajog.org Clinical Opinion

specificity 33.7%, likelihood ratioþ 1.5, committee opinion states that evaluation conservative treatment, including ex-
likelihood ratioe 0.010, PPV 44.8%, includes tumor size, morphology, wall pectancy and observation, if the diag-
and NPV 98.9%.13 A comparison of characteristics, and presence of septae, nosis can be made with a relatively high
several methods of analyzing ultra- offering a sensitivity and specificity of degree of confidence; (2) the patient who
sound finding is presented in Table 3. 86% and 99%, respectively, but very few is at higher surgical risk secondary to
technical details are given. When the associated cardiac, pulmonary, or other
Why IOTA algorithms are not widely mass is suspected to be malignant (in conditions; and (3) distance a patient
implemented in the United States and postmenopausal women: elevated CA- may have to travel to a specialized center,
why they should be 125 or presence of ascites, nodular or particularly in remote areas.
Medicine in the Unites States tends to be fixed pelvic mass, or evidence of In these 3 types of patients, a method
more defensive than in many other abdominal or distant metastasis; and in to classify ovarian tumors with a high
countries, mostly because of the liti- premenopausal women: very elevated degree of confidence would alleviate
giousness of the US society. In the vast CA-125, ascites or evidence of abdom- many of the uncertainties and avoid
majority of cases, the presence of a inal or distant metastasis), referral to a unnecessary referral to a sometimes
persistent ovarian mass, particularly in a specialist trained in oncologic surgery is distant center, which may be burden-
postmenopausal woman, but also in recommended.6 some with respect to travel, cost, and
younger patients, will be an indication to Using IOTA Simple Rules would yield anxiety. The type of surgery (eg, lapa-
recommend active intervention. Many better results. In a recent study, the roscopy vs laparotomy) may need to be
unnecessary procedures may be per- Simple Rules, followed by expert altered, based on the risk of a tumor to be
formed if the diagnostic test employed opinion, had the best NPV (94.9%).80 In malignant. No model or biochemical
cannot differentiate between benign and addition to the risk of morbidity and marker of ovarian malignancy has been
malignant conditions with a sufficient mortality present in any surgical inter- shown to be superior to assessment of
degree of certainty. vention (10% in 1 publication: 8% in gray-scale images and color Doppler
The prevalence of ovarian cancer in patients with benign pathologies and findings by an experienced ultrasound
postmenopausal women is approxi- 19% in patients with malignancy81), 3 examiner.80 There is, however, a rela-
mately 1 in 2500. To be effective, a test very specific conditions may be particu- tively low number of ultrasound experts
for diagnosing ovarian cancer should larly problematic: (1) finding of an (radiology or obstetrics/gynecology)
have a minimum PPV of 10%. With a ovarian mass in a young woman desirous able to screen a large number of patients
prevalence of 1 in 2500, for a PPV of of maintaining her fertility, without and, furthermore, experience of the
10%, a sensitivity of
75% for early- recourse to freezing eggs or embryose examiner plays an important role,84,85
stage disease and a specificity of 99.6% while fertility-sparing (less than radical) and this is not easily transmitted to
are needed. Furthermore, the final surgery is an option in some of these newer end-users.
diagnosis of ovarian cancer is direct patients with early-stage invasive disease This is precisely why a simple method,
histopathological examination of tissues or borderline ovarian tumor,82,83 a such as the IOTA Simple Rules, is so
obtained at surgery and a PPV of 10% number may opt for even more valid. The first reason why these should
implies 10 surgical interventions for
each case of cancer detected.78 This is an
accepted number in the United States,
although some report 19.5 surgeries per TABLE 3
diagnosed case.79 According to the IOTA Comparison of Simple Rules and ADNEX model with pattern recognition
data, this number is 5.9 surgeries per by expert and Risk of Malignancy Index
diagnosed cases in general centers and Sensitivity Specificity Negative predictive
2.3 in oncology centers.13 The RMI is a (95% CI) (95% CI) value (95% CI)
good test, although, to our knowledge, Pattern recognition by expert85 91 (88-94) 96 (94-97) 90.9 (81.3e96.6)75
also not commonly employed in the 12,15 a
Simple Rules 93 (89-94) 76-96 (94-97) 92.0 (87.8e95.1)75
United States. The ACOG/Society of
Gynecologic Oncology committee Subjective assessment after 91 (88-93) 93 (89-94) 91.7 (88.1e94.6)75
opinion “The role of the obstetrician- inconclusive Simple Rules70
gynecologist in the early detection of ADNEX model 96.5 71.3
ovarian epithelial cancer” is the recom- RMIb56,60 75-86.8 85-91 97.5
mended approach in the United States 5
RMI with CA-125 78 (76-80) 87 (85-89)
but is, in fact, somewhat unclear on how
ADNEX, Assessment of Different NEoplasias in adneXa; CI, confidence interval; RMI, Risk of Malignancy Index.
to screen.6 For low-risk women, no a
Different studies; b Four models, RMI1-4 RMI4 is best, with sensitivity of 86.8%, specificity of 91.0%, positive predictive value
strategy seems effective. For high-risk of 63.5%, negative predictive value of 97.5%, and accuracy of 90.4%.
women, CA-125 and TVS are recom- Abramowicz. US IOTA ovarian tumor rules. Am J Obstet Gynecol 2017.
mended every 6 months.6 The

DECEMBER 2017 American Journal of Obstetrics & Gynecology 657

Clinical Opinion ajog.org

be adopted in the United States is that 2. Howlader N, Noone A, Krapcho M, et al. tumors: prospective multicenter diagnostic
they are, well, simple. This may not be of SEER cancer statistics review 1975-2013. study. BMJ 2014;349:g5920.
Available at: http://seer.cancer.gov/csr/1975_ 17. Ring KL, Garcia C, Thomas MH,
tremendous help for the expert who may 2013. Accessed Oct. 8, 2016. Modesitt SC. Current and future role of genetic
do as well as various models by pattern 3. Ferlay J, Soerjomataram I, Dikshit R, et al. screening in gynecologic malignancies. Am J
recognition,86 but will definitely help the Cancer incidence and mortality worldwide: Obstet Gynecol 2017 Apr 12. pii: S0002-
less-expert examiner such as obstetri- sources, methods and major patterns in GLO- 9378(17)30512-4. http://dx.doi.org/10.1016/j.
cian/gynecologist specialists in office BOCAN 2012. Int J Cancer 2015;136:E359-86. ajog.2017.04.011. [Epub ahead of print].
4. National Cancer Institute. Cancer stat facts: 18. Moore RG, Jabre-Raughley M, Brown AK,
settings, radiologists, as well as other ovarian cancer. Available at: https://seer.cancer. et al. Comparison of a novel multiple marker
users of ultrasound without advanced gov/statfacts/html/ovary.html. Accessed Oct. assay vs the risk of malignancy index for the
training and specific expertise in gyne- 8, 2016. prediction of epithelial ovarian cancer in patients
cologic ultrasound. In fact, the intent is 5. Siegel RL, Miller KD, Jemal A. Cancer statis- with a pelvic mass. Am J Obstet Gynecol
very clear that these rules be used by the tics, 2017. CA Cancer J Clin 2017;67:7-30. 2010;203:228.e221-6.
6. American College of Obstetricians and Gy- 19. Van Gorp T, Cadron I, Despierre E, et al.
specialist in obstetrics and gynecology or necologists Committee on Gynecologic Prac- HE4 and CA125 as a diagnostic test in ovarian
other medical specialties to obtain an tice. The role of the obstetrician-gynecologist in cancer: prospective validation of the risk of
assessment of the chance that a tumor is the early detection of epithelial ovarian cancer. ovarian malignancy algorithm. Br J Cancer
benign or the risk of it being malignant. Committee opinion no. 477. Obstet Gynecol 2011;104:863-70.
It is not intended to be used by the 2011;117:742-6. 20. Simmons A, Baggerly K, Bast RJ. The
7. Fischerova D, Burgetova A. Imaging tech- emerging role of HE4 in the evaluation of advanced
subspecialist (although, obviously they niques for the evaluation of ovarian cancer. Best epithelial ovarian and endometrial carcinomas.
may). Likewise, the ADNEX model also Pract Res Clin Obstet Gynaecol 2014;28: Oncology (Williston Park) 2013;27:548-56.
permits better patient triage and optimal 697-720. 21. Urban N, Hawley S, Janes H, et al. Identi-
decisions about type of management and 8. Jacobs IJ, Menon U, Ryan A, et al. Ovarian fying post-menopausal women at elevated risk
referral to a specialized center, if neces- cancer screening and mortality in the UK for epithelial ovarian cancer. Gynecol Oncol
Collaborative Trial of Ovarian Cancer Screening 2015;139:253-60.
sary,58 as results are better if surgery is (UKCTOCS): a randomized controlled trial. 22. Yurkovetsky Z, Skates S, Lomakin A, et al.
performed by, for example, a gyneco- Lancet 2016;387:945-56. Development of a multimarker assay for early
logic oncologist.87 This is exactly the 9. Woo YL, Kyrgiou M, Bryant A, Everett T, detection of ovarian cancer. J Clin Oncol
approach that ACOG has long Dickinson HO. Centralization of services for gy- 2010;28:2159-66.
recommended.6,88 necological cancer. Cochrane Database Syst 23. Koshiyama M, Matsumura N, Konishi I.
Rev 2012;3:CD007945. Clinical efficacy of ovarian cancer screening.
One major reason why the IOTA rules 10. van Nagell JR Jr, Miller RW. Evaluation and J Cancer 2016;7:1311-6.
have not been widely implemented in the management of ultrasonographically detected 24. Buys SS, Partridge E, Greene MH, et al.
United States is a lack of validation in ovarian tumors in asymptomatic women. Obstet Ovarian cancer screening in the prostate, lung,
this country. All the IOTA studies origi- Gynecol 2016;127:848-58. colorectal, and ovarian (PLCO) cancer screening
nate from countries other than the 11. Dodge JE, Covens AL, Lacchetti C, et al. trial: findings from the initial screen of a random-
Preoperative identification of a suspicious ized trial. Am J Obstet Gynecol 2005;193:1630-9.
United States, both for the Simple Rules adnexal mass: a systematic review and meta- 25. van Nagell JR Jr, Miller RW, DeSimone CP,
(Belgium, France, Italy, Sweden, and the analysis. Gynecol Oncol 2012;126:157-66. et al. Long-term survival of women with epithelial
United Kingdom) and the ADNEX 12. Kaijser J, Sayasneh A, Van Hoorde K, et al. ovarian cancer detected by ultrasonographic
model (Belgium, Canada, China, Czech Presurgical diagnosis of adnexal tumors using screening. Obstet Gynecol 2011;118:1212-21.
Republic, France, Italy, Poland, Spain, mathematical models and scoring systems: a 26. Kobayashi H, Yamada Y, Sado T, et al.
systematic review and meta-analysis. Hum A randomized study of screening for ovarian
Sweden, and the United Kingdom). In Reprod Update 2014;20:449-62. cancer: a multicenter study in Japan. Int J
fact, as of the writing of this opinion, 13. Timmerman D, Van Calster B, Testa A, Gynecol Cancer 2008;18:414-20.
there are 3612 IOTA members, accepted et al. Predicting the risk of malignancy in 27. Schorge JO, Modesitt SC, Coleman RL,
after attending a course and passing an adnexal masses based on the Simple Rules et al. SGO white paper on ovarian cancer: eti-
exam and only 10 in the United States. from the International Ovarian Tumor Analysis ology, screening and surveillance. Gynecol
group. Am J Obstet Gynecol 2016;214: Oncol 2010;119:7-17.
Hopefully, this can easily be remedied by 424-37. 28. Campbell S, Goessens L, Goswamy R,
first performing, in a US population, 14. Sayasneh A, Ferrara L, De Cock B, et al. Whitehead M. Real-time ultrasonography for
retrospective analyses of ultrasound Evaluating the risk of ovarian cancer before determination of ovarian morphology and vol-
ovarian findings with the Simple Rules surgery using the ADNEX model: a multicenter ume. A possible early screening test for ovarian
and the ADNEX model and comparing external validation study. Br J Cancer 2016;115: cancer? Lancet 1982;1:425-6.
542-8. 29. Bourne TH, Campbell S, Reynolds KM, et al.
the results with pathology examination
15. Timmerman D, Ameye L, Fischerova D, et al. Screening for early familial ovarian cancer with
of the surgical specimens as well as a Simple ultrasound rules to distinguish between transvaginal ultrasonography and color blood
similar prospective, possibly multi- benign and malignant adnexal masses before flow imaging. BMJ 1993;306:1025-9.
center, study. - surgery: prospective validation by IOTA group. 30. Fishman DA, Cohen L, Blank SV, et al. The
BMJ 2010;341:c6839. role of ultrasound evaluation in the detection of
16. Van Calster B, Van Hoorde K, Valentin L, early-stage epithelial ovarian cancer. Am J
REFERENCES et al. Evaluating the risk of ovarian cancer before Obstet Gynecol 2005;192:1214-21.
1. Jayson GC, Kohn EC, Kitchener HC, surgery using the ADNEX model to differentiate 31. Rosenthal AN, Fraser L, Manchanda R, et al.
Ledermann JA. Ovarian cancer. Lancet between benign, borderline, early and advanced Results of annual screening in phase I of the
2014;384:1376-88. stage invasive, and secondary metastatic United Kingdom familial ovarian cancer

658 American Journal of Obstetrics & Gynecology DECEMBER 2017

ajog.org Clinical Opinion

screening study highlight the need for strict 45. Granberg S, Wikland M, Jansson I. Macro- 57. Weber G, Merz E, Bahlmann F, Leber AM.
adherence to screening schedule. J Clin Oncol scopic characterization of ovarian tumors and A new sonomorphologic scoring-system (Mainz
2013;31:49-57. the relation to the histological diagnosis: criteria score) for the assessment of ovarian tumors
32. Gilbert L, Basso O, Sampalis J, et al. to be used for ultrasound evaluation. Gynecol using transvaginal ultrasonography. Part II: a
Assessment of symptomatic women for early Oncol 1989;35:139-44. comparison between the scoring-system and
diagnosis of ovarian cancer: results from the 46. Jacobs I, Oram D, Fairbanks J, Turner J, the assessment by an experienced sonographer
prospective DOvE pilot project. Lancet Oncol Frost C, Grudzinskas JG. A risk of malignancy in postmenopausal women. Ultraschall Med
2012;13:285-91. index incorporating CA 125, ultrasound and 1999;20:2-8.
33. Suh-Burgmann E, Kinney W. The value of menopausal status for the accurate preopera- 58. Timmerman D, Testa AC, Bourne T, et al.
ultrasound monitoring of adnexal masses for tive diagnosis of ovarian cancer. Br J Obstet Simple ultrasound-based rules for the diagnosis
early detection of ovarian cancer. Front Oncol Gynaecol 1990;97:922-9. of ovarian cancer. Ultrasound Obstet Gynecol
2016;6:25. 47. Lerner JP, Timor-Tritsch IE, Federman A, 2008;31:681-90.
34. Sladkevicius P, Valentin L, Marsal K. Abramovich G. Transvaginal ultrasonographic 59. Menon U, Skates SJ, Lewis S, et al. Pro-
Transvaginal Doppler examination for the differ- characterization of ovarian masses with an spective study using the risk of ovarian cancer
ential diagnosis of solid pelvic tumors. improved, weighted scoring system. Am J algorithm to screen for ovarian cancer. J Clin
J Ultrasound Med 1995;14:377-80. Obstet Gynecol 1994;170:81-5. Oncol 2005;23:7919-26.
35. Marret H, Cayrol M. Sonographic diagnosis 48. Sassone AM, Timor-Tritsch IE, Artner A, 60. Geomini P, Kruitwagen R, Bremer GL,
of presumed benign ovarian tumors [in French]. Westhoff C, Warren WB. Transvaginal sono- Cnossen J, Mol BW. The accuracy of risk scores
J Gynecol Obstet Biol Reprod (Paris) 2013;42: graphic characterization of ovarian disease: in predicting ovarian malignancy: a systematic
730-43. evaluation of a new scoring system to predict review. Obstet Gynecol 2009;113:384-94.
36. Hamper UM, Sheth S, Abbas FM, ovarian malignancy. Obstet Gynecol 1991;78: 61. Campbell S. Ovarian cancer: role of ultra-
Rosenshein NB, Aronson D, Kurman RJ. 70-6. sound in preoperative diagnosis and population
Transvaginal color Doppler sonography of 49. Tailor A, Jurkovic D, Bourne TH, Collins WP, screening. Ultrasound Obstet Gynecol 2012;40:
adnexal masses: differences in blood flow Campbell S. Sonographic prediction of malig- 245-54.
impedance in benign and malignant lesions. AJR nancy in adnexal masses using multivariate lo- 62. Glanc P, Benacerraf B, Bourne T, et al. First
Am J Roentgenol 1993;160:1225-8. gistic regression analysis. Ultrasound Obstet international consensus report on adnexal
37. Valentin L. Prospective cross-validation of Gynecol 1997;10:41-7. masses: management recommendations.
Doppler ultrasound examination and gray-scale 50. Timmerman D, Bourne TH, Tailor A, et al. J Ultrasound Med 2017;36:849-63.
ultrasound imaging for discrimination of benign A comparison of methods for preoperative 63. Timmerman D, Valentin L, Bourne TH, et al.
and malignant pelvic masses. Ultrasound Obstet discrimination between malignant and benign Terms, definitions and measurements to
Gynecol 1999;14:273-83. adnexal masses: the development of a new lo- describe the sonographic features of adnexal
38. Wilson WD, Valet AS, Andreotti RF, Green- gistic regression model. Am J Obstet Gynecol tumors: a consensus opinion from the Interna-
Jarvis B, Lyshchik A, Fleischer AC. Sonographic 1999;181:57-65. tional Ovarian Tumor Analysis (IOTA) group. Ul-
quantification of ovarian tumor vascularity. 51. Timmerman D, Verrelst H, Bourne TH, et al. trasound Obstet Gynecol 2000;16:500-5.
J Ultrasound Med 2006;25:1577-81. Artificial neural network models for the preop- 64. Timmerman D. The use of mathematical
39. Carter J, Saltzman A, Hartenbach E, erative discrimination between malignant and models to evaluate pelvic masses; can they beat
Fowler J, Carson L, Twiggs LB. Flow charac- benign adnexal masses. Ultrasound Obstet an expert operator? Best Pract Res Clin Obstet
teristics in benign and malignant gynecologic Gynecol 1999;13:17-25. Gynaecol 2004;18:91-104.
tumors using transvaginal color flow Doppler. 52. Timor-Tritsch LE, Lerner JP, Monteagudo A, 65. Timmerman D, Testa AC, Bourne T, et al.
Obstet Gynecol 1994;83:125-30. Santos R. Transvaginal ultrasonographic char- Logistic regression model to distinguish be-
40. Valentin L. Gray scale sonography, subjec- acterization of ovarian masses by means of color tween the benign and malignant adnexal mass
tive evaluation of the color Doppler image and flow-directed Doppler measurements and a before surgery: a multicenter study by the In-
measurement of blood flow velocity for dis- morphologic scoring system. Am J Obstet ternational Ovarian Tumor Analysis group. J Clin
tinguishing benign and malignant tumors of Gynecol 1993;168:909-13. Oncol 2005;23:8794-801.
suspected adnexal origin. Eur J Obstet Gynecol 53. Tingulstad S, Hagen B, Skjeldestad FE, 66. Mol BW, Boll D, De Kanter M, et al. Dis-
Reprod Biol 1997;72:63-72. Halvorsen T, Nustad K, Onsrud M. The risk-of- tinguishing the benign and malignant adnexal
41. Barua A, Bitterman P, Bahr JM, et al. malignancy index to evaluate potential ovarian mass: an external validation of prognostic
Contrast-enhanced sonography depicts spon- cancers in local hospitals. Obstet Gynecol models. Gynecol Oncol 2001;80:162-7.
taneous ovarian cancer at early stages in a 1999;93:448-52. 67. Nunes N, Ambler G, Foo X, Naftalin J,
preclinical animal model. J Ultrasound Med 54. Tingulstad S, Hagen B, Skjeldestad FE, et al. Widschwendter M, Jurkovic D. Use of IOTA
2011;30:333-45. Evaluation of a risk of malignancy index based on Simple Rules for diagnosis of ovarian cancer:
42. DePriest PD, Shenson D, Fried A, et al. serum CA125, ultrasound findings and meno- meta-analysis. Ultrasound Obstet Gynecol
A morphology index based on sonographic pausal status in the pre-operative diagnosis of 2014;44:503-14.
findings in ovarian cancer. Gynecol Oncol pelvic masses. Br J Obstet Gynaecol 1996;103: 68. Valentin L, Ameye L, Testa A, et al. Ultra-
1993;51:7-11. 826-31. sound characteristics of different types of
43. Ferrazzi E, Zanetta G, Dordoni D, 55. Van Holsbeke C, Daemen A, Yazbek J, et al. adnexal malignancies. Gynecol Oncol
Berlanda N, Mezzopane R, Lissoni AA. Trans- Ultrasound methods to distinguish between 2006;102:41-8.
vaginal ultrasonographic characterization of malignant and benign adnexal masses in the 69. Valentin L, Jurkovic D, Van Calster B, et al.
ovarian masses: comparison of five scoring hands of examiners with different levels of Adding a single CA 125 measurement to ultra-
systems in a multicenter study. Ultrasound experience. Ultrasound Obstet Gynecol sound imaging performed by an experienced
Obstet Gynecol 1997;10:192-7. 2009;34:454-61. examiner does not improve preoperative
44. Finkler NJ, Benacerraf B, Lavin PT, 56. Yamamoto Y, Yamada R, Oguri H, discrimination between benign and malignant
Wojciechowski C, Knapp RC. Comparison of Maeda N, Fukaya T. Comparison of four malig- adnexal masses. Ultrasound Obstet Gynecol
serum CA 125, clinical impression, and ultra- nancy risk indices in the preoperative evaluation 2009;34:345-54.
sound in the preoperative evaluation of ovarian of patients with pelvic masses. Eur J Obstet 70. Fathallah K, Huchon C, Bats AS, et al.
masses. Obstet Gynecol 1988;72:659-64. Gynecol Reprod Biol 2009;144:163-7. External validation of simple ultrasound rules of

DECEMBER 2017 American Journal of Obstetrics & Gynecology 659

Clinical Opinion ajog.org

Timmerman on 122 ovarian tumors [in French]. 76. Meys EM, Jeelof LS, Achten NM, et al. stage IA or IC invasive epithelial ovarian cancer
Gynecol Obstet Fertil 2011;39:477-81. Estimating the risk of malignancy in adnexal treated with fertility-sparing therapy. Gynecol
71. Meys EM, Kaijser J, Kruitwagen RF, et al. masses: an external validation of the ADNEX Oncol 2002;87:1-7.
Subjective assessment versus ultrasound model and comparison with other frequently 84. Timmerman D, Schwarzler P, Collins WP,
models to diagnose ovarian cancer: a system- used ultrasound methods. Ultrasound Obstet et al. Subjective assessment of adnexal
atic review and meta-analysis. Eur J Cancer Gynecol 2017;49:784-92. masses with the use of ultrasonography: an
2016;58:17-29. 77. Szubert S, Wojtowicz A, Moszynski R, et al. analysis of interobserver variability and expe-
72. Nunes N, Ambler G, Foo X, et al. Compari- External validation of the IOTA ADNEX model rience. Ultrasound Obstet Gynecol 1999;13:
son of two protocols for the management of performed by two independent gynecologic 11-6.
asymptomatic postmenopausal women with centers. Gynecol Oncol 2016;142:490-5. 85. Yazbek J, Raju SK, Ben-Nagi J, Holland TK,
adnexal tumorsea randomized controlled trial of 78. Rauh-Hain JA, Krivak TC, Del Carmen MG, Hillaby K, Jurkovic D. Effect of quality of gyne-
RMI/RCOG vs Simple Rules. Br J Cancer Olawaiye AB. Ovarian cancer screening and cological ultrasonography on management of
2017;116:584-91. early detection in the general population. Rev patients with suspected ovarian cancer: a ran-
73. Araujo KG, Jales RM, Pereira PN, et al. Obstet Gynecol 2011;4:15-21. domized controlled trial. Lancet Oncol 2008;9:
Performance of the IOTA ADNEX model in the 79. Partridge E, Kreimer AR, Greenlee RT, et al. 124-31.
preoperative discrimination of adnexal masses in Results from four rounds of ovarian cancer 86. Valentin L, Hagen B, Tingulstad S, Eik-
a gynecologic oncology center. Ultrasound screening in a randomized trial. Obstet Gynecol Nes S. Comparison of “pattern recognition” and
Obstet Gynecol 2017;49:778-83. 2009;113:775-82. logistic regression models for discrimination
74. Epstein E, Van Calster B, Timmerman D, 80. Niemi RJ, Saarelainen SK, Luukkaala TH, between benign and malignant pelvic masses: a
Nikman S. Subjective ultrasound assessment, Maenpaa JU. Reliability of preoperative evalua- prospective cross validation. Ultrasound Obstet
the ADNEX model and ultrasound-guided tru- tion of postmenopausal ovarian tumors. Gynecol 2001;18:357-65.
cut biopsy to differentiate disseminated primary J Ovarian Res 2017;10:15. 87. du Bois A, Rochon J, Pfisterer J,
ovarian cancer from metastatic non-ovarian 81. Baldwin LA, Pavlik EJ, Ueland E, et al. Hoskins WJ. Variations in institutional infra-
cancer. Ultrasound Obstet Gynecol 2016;47: Complications from surgeries related to ovarian structure, physician specialization and experi-
110-6. cancer screening. Diagnostics (Basel) 2017;7(1). ence, and outcome in ovarian cancer: a
75. Joyeux E, Miras T, Masquin I, Duglet PE, 82. Rao GG, Skinner EN, Gehrig PA, Duska LR, systematic review. Gynecol Oncol 2009;112:
Astruc K, Douvier S. Before surgery predict- Miller DS, Schorge JO. Fertility-sparing surgery 422-36.
ability of malignant ovarian tumors based on for ovarian low malignant potential tumors. 88. Im SS, Gordon AN, Buttin BM, et al. Vali-
ADNEX model and its use in clinical practice Gynecol Oncol 2005;98:263-6. dation of referral guidelines for women with
[in French]. Gynecol Obstet Fertil 2016;44: 83. Schilder JM, Thompson AM, DePriest PD, pelvic masses. Obstet Gynecol 2005;105:
557-64. et al. Outcome of reproductive age women with 35-41.

660 American Journal of Obstetrics & Gynecology DECEMBER 2017