Int J Gynecol Cancer 2007, 17, 61–67

Differential diagnosis of adnexal masses: risk of

malignancy index, ultrasonography, magnetic
resonance imaging, and radioimmunoscintigraphy
P.O. VAN TRAPPEN*, B.D. RUFFORD*, T.D. MILLS**, S.A. SOHAIB**, J.A.W. WEBB**,
A. SAHDEV**, M.J. CARROLLy, K.E. BRITTONy, R.H. REZNEK* & I.J. JACOBSz
*Departments of Gynaecological Oncology, **Radiology and yNuclear Medicine, Queen Mary University
of London, St Bartholomew’s Hospital, West Smithfield, London, United Kingdom; and zDepartment
of Gynaecological Oncology, Institute of Women’s Health, University College, London, United Kingdom

Abstract. Van Trappen PO, Rufford BD, Mills TD, Sohaib SA, Webb JAW, Sahdev A, Carroll MJ, Britton
KE, Reznek RH, Jacobs IJ. Differential diagnosis of adnexal masses: risk of malignancy index, ultrasonography,
magnetic resonance imaging, and radioimmunoscintigraphy. Int J Gynecol Cancer 2007;17:61–67.

A risk of malignancy index (RMI), based on menopausal status, ultrasound (US) findings, and serum
CA125, has previously been described and validated in the primary evaluation of women with adnexal
masses and is widely used in selective referral of women from local cancer units to specialized cancer cen-
ters. Additional imaging modalities could be useful for further characterization of adnexal masses in this
group of women. A prospective cohort study was conducted of 196 women with an adnexal mass referred
to a teaching hospital for diagnosis and management. Follow-up data was obtained for 180 women; 119
women had benign and 61 women malignant adnexal masses. The sensitivity and specificity of specialist
US, magnetic resonance imaging (MRI), radioimmunoscintigraphy (RS), and the RMI were determined. We
identified a subgroup of women with RMI values of 25–1000 where the value of further specialist imaging
was evaluated. Sensitivity and specificity for specialist US were 100% and 57%, for MRI 92% and 86%, and
for RS 76% and 87%, respectively. Analysis of 123 patients managed sequentially, using RMI cutoff values
of
25 and ,1000 and then US and MRI provided a sensitivity of 94% and a specificity of 90%. Using this
RMI cutoff followed by specialist US and MRI, as opposed to the traditional RMI cutoff value of 250, can
increase the proportion of patients with cancer appropriately referred in to a cancer center, with no change
in the proportion of patients with benign disease being managed in a local unit.

KEYWORDS: adnexal mass, MRI, ovarian cancer, risk of malignancy index, ultrasonography.

Greater use of ultrasound (US) and other radiologic in-
vestigations has resulted in an increase in the number
of adnexal abnormalities coming to the attention of
general practitioners, gynecologists, and other physi-
cians. The prevalence is high both in the premeno-
pausal women when they are frequently physiologic
and in the postmenopausal age group where over 20%
Address correspondence and reprint requests to: Prof. Ian J. Jacobs,
MD, Department of Gynecological Oncology, Institute for Women’s
Health, University College London. Email: i.jacobs@ucl.ac.uk
The first and second author contributed equally to the work
presented in the manuscript.
doi:10.1111/j.1525-1438.2006.00753.x
# 2007, Copyright the Authors
Journal compilation # 2007, IGCS and ESGO
of asymptomatic women have abnormal ovarian mor-
phology(1). Adnexal abnormalities may be discovered
as a result of screening, be an incidental finding, or be
recognized as a result of investigations performed spe-
cifically for a suspected pelvic mass. The only defini-
tive way of determining whether a mass is benign or
malignant is removal at surgery. However, the major-
ity of women with adnexal masses will not have
malignant disease and many do not require surgery.
Those women who are likely to have ovarian cancer
should be referred to specialized oncology centers able
to perform optimal surgical staging and cyto-
reduction, and this is reflected in the new structure of
cancer care in the UK(2,3). This is important, as one of
62 P.O. Van Trappen et al.
the key prognostic factors in ovarian cancer is the
quality of cytoreductive surgery(4). A reliable method
for differentiating benign from malignant adnexal
masses would provide a basis for appropriate manage-
ment; either locally for benign masses or at a specialist
cancer center if the mass is thought to be malignant.
Initially, clinical, demographic, biochemical, and
ultrasonographic features were used to distinguish
benign from malignant adnexal masses but were
found to have poor diagnostic accuracy. The tumor
marker CA125 has limited specificity due to raised
levels in nonmalignant conditions such as endometri-
osis, and limited sensitivity, particularly in early-stage
disease. The sensitivity of US can be improved by
using transvaginal scanning and morphologic scoring
systems, but it has limited specificity(5,6). The use of
a risk of malignancy index (RMI)(7–9) incorporating
menopausal status, CA125 and US offers improve-
ments in sensitivity and specificity over US or tumor
markers used in isolation. This has recently been vali-
dated as a tool used to triage women with adnexal
masses for referral to a cancer center(10). More recently,
multivariate logistic regression models(11–13) and artifi-
cial neural networks(14,15) have been introduced using
a variable set of demographic, clinical, tumor marker,
and US characteristics. The accuracy of three regres-
sion models was recently assessed in a prospective
collaborative study(16). The models were not as suc-
cessful as originally reported, and subjective assess-
ment by an experienced sonographer was still as good
as the mathematical models or scoring systems
devised to date(17,18). The RMI remains the simplest
and best validated approach to differential diagnosis.
Although US remains the primary imaging modal-
ity for assessing adnexal masses, more sophisticated
imaging techniques such as magnetic resonance imag-
Table 1. Comparison of patient and US characteristics in benign and malignant adnexal masses
Characteristic Benign mean (range) (n ¼ 119)
Age (years) 46 (15–82)
CA125 (IU/mL) 76 (15–850)
USb
Ascites 5% (5/109c)
Metastases 0% (0/109)
Multilocular cysts 49% (53/109)
Solid areas 61% (67/109)
Bilateral 15% (16/109)
US score: 0, 1, 3 21, 43, 45 (n ¼ 109)
RMI 219 (0–5130)
NS, not significant.
a
P values were calculated with the t test for the means and the Chi-square test or when not possible the Fisher’s exact test.
b
Referral (first) US scan.
c
Number of patients with known US characteristics of referral US scan.
# 2007 IGCS and ESGO, International Journal of Gynecological Cancer 17, 61–67
ing (MRI) with gadolinium enhancement(19,20) and ra-
dioimmunoscintigraphy (RS)(21,22) have recently been
shown to have encouraging accuracy in the character-
ization of adnexal masses. However, the Royal College
of Obstetricians and Gynaecologists guideline on the
management of ovarian cysts in postmenopausal
women(23) includes the use of CA125 and ultrasonog-
raphy, but finds no routine role for MRI, computed
tomography (CT), or positron emission tomography
(PET) and does not mention RS. We undertook this
prospective cohort study to assess the diagnostic accu-
racy of US, MRI, and RS in the differential diagnosis
of adnexal masses. In particular, we aimed to deter-
mine whether any additional imaging technique pro-
vided additional information to the RMI.
Materials and methods
Patients
A consecutive group of 196 women between the ages
of 15 and 92 years (mean: 50 years) with suspected
adnexal masses were recruited from outpatient gyne-
cology clinics at St Bartholomew’s Hospital. On the
basis of menopausal status, serum CA125 level and
US characteristics, the RMI was calculated as
described previously(7): RMI ¼ US score 3 serum
CA125 3 menopausal score. An US score of 0, 1, or 3
was given when 0, 1, or 2–5 key US characteristics
were present (ascites, evidence of metastases, multi-
locular cysts, solid areas, and bilaterality; Table 1). A
menopausal score of 1 or 3 was given, respectively, to
pre- or postmenopausal women.
Women were offered additional imaging techniques:
specialist US (performed by a consultant radiologist
with an interest in gynecological scanning), MRI, and
Malignant mean (range) (n ¼ 61) Pa
57 (19–92) ,.001
1637 (15–17,800) ,.001
37% (19/52c) ,.001
13% (7/52) ,.001
79% (41/52) ,.001
87% (45/52) .001
25% (13/52) NS
1, 7, 44 (n ¼ 52) ,.001
5513 (0–83,700) ,.001

RS. Due to logistic as well as patient restrictions (eg,
claustrophobia and allergy), not all patients underwent
all additional imaging techniques. All additional imag-
ing techniques were performed at St Bartholomew’s
Hospital.
The local Research Ethics Committee for the ‘‘East
London and City Health Authority’’ approved the
study, and before patients were included, a full clinical
history and consent were taken in the clinics of
the Department of Gynaecology at St Bartholomew’s
Hospital.
Methods
Gray scale and Doppler ultrasonography were per-
formed in all patients. Transabdominal scans were
done with an Acuson Sequoia using a 2–4 MHz trans-
ducer or an Acuson 128 XP/10 using a 3.5 MHz trans-
ducer, and transvaginal scans were done with an
Acuson 128 XP/10 using a 7.5 MHz transducer. The
scans were all performed by one radiologist
(J.A.W.W.) who was unaware of the clinical history.
Based on morphologic features of the mass (eg, size,
bilaterality, consistency, the presence of septa and/or
nodules in cystic masses), blood flow in the mass (eg,
presence of flow, site and type of flow, pulsatility
index and/or resistance index) and other findings (eg,
ascites, liver masses, renal obstruction), an impression
score was allocated (1 ¼ benign, 2 ¼ probably benign,
3 ¼ possibly malignant, 4 ¼ probably malignant or
5 ¼ malignant). For further analysis, impression score
1 and 2 were grouped as nonsuspicious (probably
benign) and impression score 3, 4, and 5 as suspicious
(probably malignant).
MRI was performed on a GE Signa Henzen 1.5T MR
Unit (GE Medical System, Milwaukee, WI). All
women underwent axial T1-weighted spin echo, axial
and sagittal T2-weighted fast spin echo, and pre- and
post-gadolinium-enhanced fat suppressed spoiled gra-
dient echo T1-weighted images in the best plane. Simi-
lar morphologic features as for the US analysis were
assessed with MRI. The images were reviewed inde-
pendently, without knowledge of clinical history or
mass histology, by two radiologists (S.A.S. and
R.H.R.). Again, a final assessment of whether the mass
was benign or malignant was given using the same
criteria as for US.
RS was performed using SM3 (stripped mucin 3), an
IgG1 murine monoclonal antibody labelled with Tech-
netium (Tc-99m). This antibody, which has been devel-
oped by the Imperial Cancer Research Fund, binds to
malignant ovarian tissue 17 times more efficiently
than to benign lesions or normal tissue. The women
# 2007 IGCS and ESGO, International Journal of Gynecological Cancer 17, 61–67
Differential diagnosis of adnexal masses 63
were informed about the nature of the procedure and
questioned about possible allergies to proteins. The in-
jections were performed at the Department of Nuclear
Medicine at St Bartholomew’s Hospital, and consisted
of an intravenous injection of 600MBq of 99mTc
labelled SM3 (0.5 mg). A gamma camera was set up
over the patient’s pelvis and images were taken at 10
min, 4 h, and 24 h, respectively. Images with change
detection analysis and probability mapping were in-
terpreted on a Hermes workstation, and masses were
classified as benign or malignant.
Specialist gynecological oncology surgeons at St
Bartholomew’s Hospital performed all surgical pro-
cedures. All tumor specimens were examined histol-
ogically in the Department of Pathology at St
Bartholomew’s Hospital by a specialist in gynecological
pathology (Prof. D. Lowe). The tumors were classified
according to the World Health Organization classifica-
tion(24). Tumors were staged according to the FIGO
staging system. When no surgery was performed, fol-
low-up US scans and serial serum CA125 levels were
organized at subsequent outpatient clinic visits.
Statistical analysis
Statistical analysis was done with SPSS 10.0.7. The
t test for the means and the Chi-square test or when
not possible the Fisher exact test was used to compare
the demographic, biochemical, and ultrasonographic
data between patients with benign and malignant
adnexal masses. Sensitivity and specificity were deter-
mined for the RMI, specialist US, MRI, and RS in the
differentiation of adnexal masses. The Pearson Chi-
square test was used to assess the ability of the differ-
ent imaging techniques in distinguishing benign from
malignant adnexal masses.
Results
Of 196 women initially entered into the study, com-
plete follow-up data was available for 180. Of these,
158 underwent surgery and 22 had follow-up serial
scans and serum CA125 levels. In the 158 women who
underwent surgery, 214 adnexal masses were detected
and the results of histologic analysis are given in
Table 2. All of the 22 women who underwent follow-
up scans had unchanged or normalized CA125 levels
and regression or disappearance of the adnexal mass
within 1 year. These 22 women who were followed for
over a year were classified as having a benign adnexal
mass for the purposes of further analysis. Women
with a malignant adnexal mass were approximately
10 years older than those with a benign adnexal mass
64 P.O. Van Trappen et al.

Table 2. Histologic characteristics of benign and malignant (two benign cystadenomas, two endometriotic cysts,
adnexal masses one ovarian fibroma, one salpingo-oophoritis, and one
Histology n (%) infarcted ovarian tissue) were incorrectly classified
as having malignant disease. The sensitivity was
Benign (n ¼ 132)
Cystadenoma 31 (23.5) therefore 92% and the specificity was 86% for MRI.
Endometriomas 28 (21) Of the 69 women who underwent an RS scan, 39
Teratoma (dermoid) 20 (15) women were correctly classified as having benign dis-
Functional cyst 18 (14) ease and 19 women were correctly classified as having
Fibroma/thecoma 18 (14)
malignant disease. Six women with malignant disease
Leiomyoma 11 (8)
Pelvic inflammatory disease/abscess 3 (2) (three borderline ovarian tumors, two stage I mucin-
Brenner tumor 2 (1.5) ous cystadenocarcinomas, and one stage 2 clear cell
Infarcted Ovarian tissue 1 (1) carcinoma) were incorrectly classified as having
Malignant (n ¼ 82) benign disease and five women with benign disease
Borderline 14 (17)
(one endometriotic cyst, one mature teratoma, one
Serous cystadenoca 41 (50)
Mucinous cystadenoca 11 (13.5) ovarian fibroma, and two leiomyomas) were incor-
Endometrioid cystadenoca 5 (6) rectly classified as having malignant disease. The sen-
Clear cell ca 3 (4) sitivity was therefore 76% and the specificity was 87%
Transitional cell ca 1 (1) for RS scans.
Ovarian carcinoid 2 (2.5)
Metastatic breast ca 2 (2.5)
Sarcoma 1 (1) RMI and subsequent imaging
Peritoneal carcinomatosis 2 (2.5)
The RMI was calculated for the 174 women who had
both preoperative CA125 values and referral scan
data available. The sensitivity and specificity for the
and 71% of women with malignant adnexal masses different cutoff values for RMI are given in Table 3
were postmenopausal. CA125 levels were significantly and are consistent with previous reports. A cutoff
higher in women with malignant adnexal masses value of 25 achieved a sensitivity and specificity of
(P , .001). All of the 5 key US characteristics except 98% and 42%, respectively, meaning that 98% of
‘‘bilaterality’’ were reported significantly more fre- ovarian cancer cases have a RMI of more than 25
quently in the malignant than benign adnexal masses. (Fig. 1). In the subgroup with a RMI ,25 (n ¼ 55
Demographic and US characteristics of benign women), only one woman (with a borderline ovarian
and malignant adnexal masses are provided in tumor) was wrongly classified as having benign dis-
Table 1. ease. A cutoff value of 1000 gave a sensitivity and
specificity of 54% and 96%, respectively, meaning
that 96% of benign cases have a RMI less than 1000
Sensitivity and specificity of different
(Fig. 1). Considering the subgroup with a RMI .1000
imaging techniques
Of the 142 women who underwent a specialist US
scan, 58 women were correctly classified as having Table 3. Sensitivity and specificity of the different cutoff
benign disease and 40 women were correctly classified values for the RMI
as having malignant disease. The remaining 44 RMI Sensitivity, % Specificity, %
women, all with benign disease, were incorrectly clas-
25 98 42
sified as having malignant disease. No women with 50 92 59
malignant disease were incorrectly classified as having 75 88 67
benign disease. The sensitivity was therefore 100% 100 84 70
and the specificity was 57% for specialist US scans. 125 80 75
Of the 76 women who underwent an MRI scan, 44 150 78 80
175 78 81
women were correctly classified as having benign dis- 200 76 82
ease and 23 women were correctly classified as having 225 76 84
malignant disease. Two women with malignant dis- 250 73 85
ease (one borderline ovarian tumor and one metastatic 300 72 86
carcinoma) were incorrectly classified as having 500 68 91
1000 54 96
benign disease and seven women with benign disease
# 2007 IGCS and ESGO, International Journal of Gynecological Cancer 17, 61–67
 Differential diagnosis of adnexal masses 65

Figure 1. Receiver operator curve for the RMI. A RMI cutoff value of 25 achieved a sensitivity and specificity of 98% and 42%, respectively,
and a cutoff value of 1000 gave a sensitivity and specificity of 54% and 96%, respectively. The red bullet demonstrates the 123 patients for
whom a RMI was calculated, and additional imaging (sequential specialist US scan and MRI) was performed for those with a RMI between 25
and 1000. In the group with a RMI between 25 and 1000, the sequential use of MRI was only required when specialist US revealed suspicious
features in this subgroup. This strategy provided an overall sensitivity of 94% and a specificity of 90%.

(n ¼ 32 patients) as having malignant disease re- Discussion

sulted in just five women being incorrectly classified
as having malignant disease.
In the subgroup of women with a RMI of
25–
1000, we evaluated the sensitivity/specificity and
diagnostic accuracy of specialist US, MRI, and RS.
Specialist US was performed in 71 women in this sub-
group and achieved a sensitivity, specificity, and
diagnostic accuracy of 100%, 44%, and 56%, respec-
tively. MRI was performed in 37 women and ach-
ieved a sensitivity, specificity, and diagnostic
accuracy of 78%, 86%, and 84%, respectively. RS was
performed in 34 women and achieved a sensitivity,
specificity, and diagnostic accuracy of 38%, 88%, and
76%, respectively.
As specialist US achieved a 100% sensitivity, an
improvement in power may be achieved by using first
specialist US and then MRI or RIS in sequential fash-
ion. In this group of women with a RMI between 25
and 1000, 36 women had sequential specialist US scan
and then MRI, achieving a sensitivity of 88% and
a specificity of 86%. Twenty-four women had sequen-
tial specialist US and RS scan, which gave a sensitivity
of 33% and a specificity of 90%.
Analysis of the 123 patients who were managed
sequentially, first using RMI cutoff values of
25
and ,1000 and then using US and MRI in those
with a RMI between 25 and 1000 provided an over-
all sensitivity of 94% and a specificity of 90%
(Fig. 1).
The aim of this study was to compare three radiologic
investigations—specialist US, MRI, and RS, and ascer-
tain whether they could add to the performance of the
RMI in distinguishing benign from malignant adnexal
masses. Additional information would be valuable
both in determining whether surgery is necessary and
if surgery is required where this should take place
within the structure of cancer care in the UK. Cur-
rently, the RMI is advocated as the discriminatory
tool(23).
Specialist US scans achieved a sensitivity of 100%
and a specificity of 57%. While no cancers would be
missed, using this as a diagnostic tool would result in
many women either having unnecessary surgery or
having surgery in a cancer center where a unit would
be sufficient. False-positive results on US were mainly
benign teratomas and endometriomas/hemorrhagic
cysts, which often contain variable amounts of solid
tissue and have thick, irregular walls. MRI is superior
to US in its ability to detect fat and blood in teratomas
and endometriomas/hemorrhagic cysts, respec-
tively(19,25,26). In our study, MRI had a sensitivity and
specificity of 92% and 86%, respectively, comparable
to previous reports(20,27). RS had a sensitivity and
specificity of 76% and 87%, respectively. Using these
modalities as a diagnostic tool may reduce the level
of unnecessary surgery or referrals, but the lack
of sensitivity may lead to an unacceptable level of

# 2007 IGCS and ESGO, International Journal of Gynecological Cancer 17, 61–67
66 P.O. Van Trappen et al.

malignancies being missed. It would therefore seem This study suggests that an optimal approach is to
sensible to use the RMI as an initial screening tool and use the RMI followed by sequential specialist US and
then to perform further imaging to improve diagnostic MRI. In the study group of 123 patients, where cases
accuracy. with a RMI less than 25 (n ¼ 55) were classified as
A RMI value of 25 appears to be a suitable threshold having a benign lesion, those with a RMI greater than
for determining which women would benefit from 1000 (n ¼ 32) were classified as having a malignant
additional imaging. The sensitivity and specificity in lesion, and those with a RMI between 25 and 1000
the detection of malignancy with a threshold level of (n ¼ 36) were classified according to additional imag-
25 in our study was 98% and 42%, respectively, as pre- ing with specialist US and MRI, and an overall sensi-
dicted from previous studies, which means that tivity of 94% and a specificity of 90% was achieved
almost all patients with a RMI of less than 25 have (Fig. 1). Using this method, rather than a traditional
a benign adnexal mass. Only one woman with a RMI RMI cutoff of 250, in this group of women would
of less than 25 in this study had a malignancy— result in an increase in the percentage of malignant ca-
a serous borderline ovarian tumor. With sensitivities ses being managed in a center from 88% to 94% with
in other reports of 100%(7) at this cutoff, it would seem no change in the percentage of benign cases being
reasonable for this group of women to be managed in managed in a unit at 90%.
a cancer unit either surgically or conservatively. Using As US scan and CA125 are more readily available
a RMI threshold of 1000 gives a sensitivity and speci- and less expensive than MRI, the application of the
ficity of 54% and 96%, respectively, in this study, with RMI in our study is an efficient modality for first tri-
over 80% of women with a RMI above this value hav- age of adnexal masses. In the group with a RMI
ing malignant disease. These women are likely to between 25 and 1000 specialist US is very useful as
require surgery in a cancer center anyway, so charac- second triage, given its sensitivity of 100%. The
terization of the mass is less important and it may be sequential use of MRI, as third triage, is only required
more appropriate to perform a CT scan for staging when specialist US reveals suspicious features in this
purposes than an MRI or RS. subgroup (Fig. 2). In this study, the sequential use of
The key group for specialist imaging to refine the RMI, specialist US and MRI achieved a greater accu-
RMI are those with a RMI of 25–1000. The perfor- racy in differential diagnosis than the RMI alone.
mance of the imaging techniques falls off in this group
as the women with RMI’s outside these values are the
most straightforward to diagnose with either benign
or malignant disease. In this more challenging group
of women, specialist US has a sensitivity of 100% and
a specificity of 44%. MRI has a sensitivity of 78% and
a specificity of 86%, while RS has a sensitivity of 38%
and a specificity of 88%. The diagnostic accuracy of
specialist US, MRI, and RS in this subgroup was 56%,
84%, and 76%, respectively. These values indicate that
RS and to a lesser extent MRI lack the sensitivity
required to be used as a first-line test after the RMI in
this subgroup. Conversely, to use specialist US on its
own would still result in too many women with
benign masses being referred for surgery in a cancer
center. An improvement in diagnostic accuracy may
be achieved by using specialist US and then MRI or
RS in sequential fashion. In the group of women with
a RMI between 25 and 1000, sequential specialist US
and then MRI gave a sensitivity of 88% and a specific-
ity of 86%; sequential specialist US and then RS gave
a sensitivity of 33% and a specificity of 90%. RIS lacks
sensitivity in this context, but the combination of spe-
cialist US and MRI looks promising giving sensitivities
and specificities that better those available from the Figure 2. Strategy for differentiating and managing adnexal masses
RMI alone at any cutoff value. using sequentially the RMI, specialist US, and MRI.

# 2007 IGCS and ESGO, International Journal of Gynecological Cancer 17, 61–67

However, these findings require further validation by
additional prospective studies.
Acknowledgments
We are most grateful to Prof. D. Lowe for the histo-
logic analysis of all tumor specimens and Miss Emma
Maturin for collection of patient follow-up data.
References
1 Hartge P, Hayes R, Reding D et al. Complex ovarian cysts in post-
menopausal women are not associated with ovarian cancer risk fac-
tors: preliminary data from the prostate, lung, colon, and ovarian
cancer screening trial. Am J Obstet Gynecol 2000;183:1232–7.
2 Whitehouse M. A policy framework for commissioning cancer ser-
vices. BMJ 1995;310:1425–6.
3 Luesley D. Improving outcomes in gynaecological cancers. BJOG
2000;107:1061–3.
4 Kehoe S, Shafi M, Luesley D. Interval cytoreduction in ovarian can-
cer. N Engl J Med 1995;333:254; author reply 255.
5 DePriest PD, Varner E, Powell J et al. The efficacy of a sonographic
morphology index in identifying ovarian cancer: a multi-institutional
investigation. Gynecol Oncol 1994;55:174–8.
6 Leibman AJ, Kruse B, McSweeney MB. Transvaginal sonography:
comparison with transabdominal sonography in the diagnosis of
pelvic masses. AJR Am J Roentgenol 1988;151:89–92.
7 Jacobs I, Oram D, Fairbanks J, Turner J, Frost C, Grudzinskas JG. A
risk of malignancy index incorporating CA 125, ultrasound and
menopausal status for the accurate preoperative diagnosis of ovar-
ian cancer. Br J Obstet Gynaecol 1990;97:922–9.
8 Tingulstad S, Hagen B, Skjeldestad FE et al. Evaluation of a risk of
malignancy index based on serum CA125, ultrasound findings and
menopausal status in the pre-operative diagnosis of pelvic masses.
Br J Obstet Gynaecol 1996;103:826–31.
9 Tingulstad S, Hagen B, Skjeldestad FE, Halvorsen T, Nustad K,
Onsrud M. The risk-of-malignancy index to evaluate potential ovar-
ian cancers in local hospitals. Obstet Gynecol 1999;93:448–52.
10 Bailey J, Tailor A, Naik R et al. Risk of malignancy index for referral of
ovarian cancer cases to a tertiary center: does it identify the correct ca-
ses? Int J Gynecol Cancer 2006;16(Suppl. 1):30–4.
11 Tailor A, Jurkovic D, Bourne TH, Collins WP, Campbell S. Sonographic
prediction of malignancy in adnexal masses using multivariate logistic
regression analysis. Ultrasound Obstet Gynecol 1997;10:41–7.
12 Alcazar JL, Jurado M. Using a logistic model to predict malignancy
of adnexal masses based on menopausal status, ultrasound mor-
# 2007 IGCS and ESGO, International Journal of Gynecological Cancer 17, 61–67
Differential diagnosis of adnexal masses 67
phology, and color Doppler findings. Gynecol Oncol 1998;69:
146–50.
13 Timmerman D, Bourne TH, Tailor A et al. A comparison of methods
for preoperative discrimination between malignant and benign
adnexal masses: the development of a new logistic regression
model. Am J Obstet Gynecol 1999;181:57–65.
14 Tailor A, Jurkovic D, Bourne TH, Collins WP, Campbell S. Sono-
graphic prediction of malignancy in adnexal masses using an artifi-
cial neural network. Br J Obstet Gynaecol 1999;106:21–30.
15 Timmerman D, Verrelst H, Bourne TH et al. Artificial neural net-
work models for the preoperative discrimination between malig-
nant and benign adnexal masses. Ultrasound Obstet Gynecol 1999;13:
17–25.
16 Aslam N, Tailor A, Lawton F, Carr J, Savvas M, Jurkovic D. Pro-
spective evaluation of three different models for the pre-operative
diagnosis of ovarian cancer. BJOG 2000;107:1347–53.
17 Aslam N, Banerjee S, Carr JV, Savvas M, Hooper R, Jurkovic D.
Prospective evaluation of logistic regression models for the diagno-
sis of ovarian cancer. Obstet Gynecol 2000;96:75–80.
18 Valentin L, Sladkevicius P, Marsal K. Limited contribution of Dopp-
ler velocimetry to the differential diagnosis of extrauterine pelvic
tumors. Obstet Gynecol 1994;83:425–33.
19 Hricak H, Chen M, Coakley FV et al. Complex adnexal masses:
detection and characterization with MR imaging–multivariate anal-
ysis. Radiology 2000;214:39–46.
20 Sohaib SA, Sahdev A, Van Trappen P, Jacobs IJ, Reznek RH. Char-
acterization of adnexal mass lesions on MR imaging. AJR Am J
Roentgenol 2003;180:1297–304.
21 Granowska M, Britton KE, Mather SJ et al. Radioimmuno-
scintigraphy with technetium-99m-labelled monoclonal antibody,
SM3, in gynaecological cancer. Eur J Nucl Med 1993;20:483–9.
22 Ali N, Jan H, Van Trappen P et al. Radioimmunoscintigraphy with
Tc-99m-labelled SM3 in differentiating malignant from benign
adnexal masses. BJOG 2003;110:508–14.
23 Rufford BD, Jacobs IJ. Ovarian cysts in postmenopausal women.
RCOG Green Top Guideline 2003;34:1–8.
24 Serov SF, Scully RE, Robin IH. International Histological Classifica-
tion of Tumours. Histological Typing of Ovarian Tumours: No. 9.
Geneva: WHO, 1973.
25 Yamashita Y, Torashima M, Hatanaka Y et al. Adnexal masses:
accuracy of characterization with transvaginal US and precontrast
and postcontrast MR imaging. Radiology 1995;194:557–65.
26 Komatsu T, Konishi I, Mandai M et al. Adnexal masses: transvaginal
US and gadolinium-enhanced MR imaging assessment of intratumoral
structure. Radiology 1996;198:109–15.
27 Sohaib SA, Mills TD, Sahdev A et al. The role of magnetic reso-
nance imaging and ultrasound in patients with adnexal masses.
Clin Radiol 2005;60:340–8.
Accepted for publication July 5, 2006