AJR 2005; 184:999-1003
© American Roentgen Ray Society
Borderline Tumors of the Ovary: CT and MRI Features and Tumor Markers in Differentiation from Stage I Disease
Nandita M. deSouza1,2,
Richard O'Neill1,
G. Angus McIndoe3,
Roberto Dina4 and
W. Patrick Soutter3
1 Department of Imaging, Hammersmith Hospital, DuCane Rd., London W12 0HS,
England.
2 Present address: Section of Magnetic Resonance Imaging, Institute of Cancer
Research, Royal Marsden Hospital, Downs Rd., Sutton, Surrey SM2 5PT,
England.
3 Department of Gynaecological Oncology, Hammersmith Hospital, London W12 0HS,
England.
4 Department of Histopathology, Hammersmith Hospital, London W12 0HS,
England.
Received January 26, 2004;
accepted after revision June 30, 2004.
Address correspondence to N. M. deSouza.
Abstract
OBJECTIVE. We sought to describe MDCT and MRI features and tumor
marker levels that differentiate borderline ovarian tumors from stage I
ovarian tumors.
CONCLUSION. Borderline ovarian tumors are complex masses with
imaging features similar to stage I tumors. The thickness of septations and
the size of solid components are significantly larger in stage I tumors, and
these features may be helpful for predicting likelihood of invasive tumors.
However, neither feature allows confident differentiation of borderline tumors
from stage I disease.
Introduction
Ovarian tumors of borderline malignancy are a distinct histologic and
clinical entity diagnosed in up to 15% of patients presenting with an ovarian
neoplasm [1]. Compared with
frankly malignant tumors, borderline tumors have a much better prognosis and,
because they are noninvasive, may be treated less radically than invasive
ovarian cancer [1,
2]. They are common in younger
women and often appear clinically to be benign. Occasionally, they are
associated with deposits within the peritoneal cavity. The aim of treatment
for these tumors in young women is complete surgical removal of the tumor
either by oophorectomy or occasionally by ovarian cystectomy. In women who
have completed their families, the optimal treatment is total abdominal
hysterectomy, bilateral salpingo-oophorectomy, and omentectomy. However, the
prognosis is not improved by adjuvant chemotherapy. In comparison, early
invasive ovarian carcinomas can metastasize not only to the contralateral
ovary and peritoneal cavity but also to the retroperitoneal lymph nodes. Stage
I tumors that are confined to one ovary in young women are occasionally
treated conservatively with unilateral oophorectomy. However, complete staging
requires pelvic and paraaortic lymph node sampling with omentectomy. The
optimal treatment involves total abdominal hysterectomy and bilateral
salpingo-oophorectomy. In stage I tumors, adjuvant chemotherapy is important
for reducing the risk of recurrence. The ability to distinguish borderline
from stage I disease preoperatively therefore considerably influences surgical
treatment and allows improved counseling of patients.
Cross-sectional imaging is routinely used to assess pelvic masses. MDCT is
still the primary imaging technique used for staging ovarian tumors
[3,
4], with MRI reserved for more
detailed lesion characterization because morphologic features of the solid
component are clearly depicted on T2-weighted images
[5,
6]. The purpose of this study
therefore was to describe the MDCT and MRI features and the tumor marker (CA
125) levels of borderline ovarian tumors that differentiate them from stage I
ovarian tumors.
Materials and Methods
Patients
We retrospectively reviewed the preoperative MDCT and MRI findings of 38
ovarian tumors (in 30 consecutive patients) classified as either borderline or
stage I at subsequent surgical resection. We identified 19 borderline tumors
in 17 patients and 19 stage I tumors in 13 patients. In all patients, serum
levels of the CA 125 tumor marker had been measured preoperatively. In the
borderline group, patients' ages ranged from 25–81 years (mean, 50.5
± 18.2 [SD] years). Most (14/17 patients) had been referred for an
asymptomatic adnexal mass. One had been referred because of pelvic pain, and
two had been referred because of a previous ovarian cyst. Eight underwent MDCT
and nine underwent MRI preoperatively. In the stage I group, patients' ages
ranged from 43–82 years (mean, 63.1 ± 13.5 years), and all had
been referred because of an asymptomatic adnexal mass. Preoperative MDCT was
performed in six patients and preoperative MRI, in seven patients.
MDCT Method
Patients underwent bowel preparation with oral Gastrografin (meglumine
diatrizoate, Schering; 500 mL of a 2.5% solution) 2 hr before scanning. Images
were obtained using an MDCT scanner (Somatom Plus 4, Siemens Medical
Solutions). The patients were scanned from the symphysis pubis to the apices
of the lungs after IV injection of 100 mL of iopromide (Ultravist 300,
Berlex). Slices with 10-mm-thickness image reconstruction were obtained in two
acquisitions—the first covering the pelvis to the diaphragm and the
second, the chest. This protocol ensured that the liver was imaged during the
portal phase.
MRI Method
Images were obtained on a 1.5-T scanner (Intera, Philips Medical Systems)
using a pelvic phased-array receiver coil. Coronal T1-weighted spin-echo
(TR/TE, 700/20) and STIR (3,000/30; inversion time, 130 msec) images as well
as sagittal and axial T2-weighted fast spin-echo (4,500/90) images were
obtained with a 5- to 6-mm slice thickness, a 256 x 512 matrix, and a
30-cm field of view.
Data Analysis
Both MDCT and MR images were reviewed by two radiologists in conference.
They were blinded to the histopathologic findings and to patient identities.
On both MDCT and MR images, the size of each mass; the number and size of
septations, nodules, and vegetations; and the presence of ascites were noted.
The number of septations was scored as more than 10, between five and 10, or
less than five per mass. The maximum thickness of septations was recorded by
taking the mean of three separate measurements. Solid components such as
nodules and papilliform vegetations were scored as to the number (> 5,
5–10, >10) and the maximum diameter (mean of three separate
measurements). In addition, the presence of omental cake, peritoneal deposits,
mesenteric deposits, and lymphadenopathy also was documented.
Statistical Analysis
The difference in the numbers and in thickness or diameter of the
septations and solid components (nodules and vegetations) between the two
groups was computed using statistical software (Statistical Package for the
Social Sciences, SPSS Inc., version 10.1 for Windows [Microsoft]) using an
independent-samples Student's t test. Receiver operating
characteristic (ROC) curves also were constructed using SPSS (version 11.5 for
Windows). ROC curves were used to examine whether the radiologic measurements
could be used to predict the histologic diagnosis of the tumors. Discriminant
analysis was used to construct a model with the parameters as predictors.
Results
Nineteen borderline tumors were noted in 17 patients. The largest diameter
of each mass varied from 5–23 cm (mean, 10.5 ± 5.4 cm). All
tumors had septations, nodules, or both (Figs.
1A,
1B,
1C,
1D, and
2). None was a simple
unilocular cyst (Table 1).
Ascites was seen in three patients, a small amount of free fluid was
identified in a fourth, and no free fluid or ascites was seen in 13 patients.
In no patient was there evidence of omental cake, peritoneal nodules, or
lymphadenopathy. Thickened septations (> 2 mm thick) were seen in 16 of 19
tumors. The mean maximum thickness of these septations was 3.3 ± 1.5
mm. Solid components (nodules or vegetations) were seen in 17 tumors. The mean
maximum single dimension for such components was 15.0 ± 9.8 mm.
View larger version (161K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1A. —27-year-old woman with borderline ovarian cancer. Sagittal
(A) and transverse (B) T2-weighted fast spin-echo images
(TR/effective TE, 4,500/90 msec) show large multicystic mass (solid
arrows) with fine, nodular solid components (arrowheads). Normal
ovarian tissue (open arrow, A) is seen posteriorly. In
A, A = anterior, P = posterior.
|
|
View larger version (148K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1B. —27-year-old woman with borderline ovarian cancer. Sagittal
(A) and transverse (B) T2-weighted fast spin-echo images
(TR/effective TE, 4,500/90 msec) show large multicystic mass (solid
arrows) with fine, nodular solid components (arrowheads). Normal
ovarian tissue (open arrow, A) is seen posteriorly. In
A, A = anterior, P = posterior.
|
|
View larger version (143K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1C. —27-year-old woman with borderline ovarian cancer. Photomicrograph
shows histopathologic confirmation of borderline serous tumor, with
micropapillation, lined by moderately atypical epithelium (arrow). (H
and E, x100)
|
|
View larger version (150K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1D. —27-year-old woman with borderline ovarian cancer. Photomicrograph of
histopathologic specimen obtained at high power (x400) reveals areas of
cellular crowding with loss of cohesiveness. (H and E)
|
|
View larger version (170K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2. —MDCT scan obtained through mid pelvis of 81-year-old woman with
borderline ovarian tumor shows well-defined heterogeneous mass
(arrow) with multiple thickened septations (arrowhead).
Bladder (BL) is seen anteriorly.
|
|
Nineteen stage I tumors were noted in 13 patients. The largest diameter of
each mass varied from 5–27 cm (mean, 13.0 ± 5.3 cm). All tumors
had either septations, nodules, or both. None was a simple unilocular cyst
(Table 1). Ascites was seen in
four patients, a small amount of free fluid was identified in a fifth, and no
free fluid or ascites was seen in eight patients. In no patient was there
evidence of an omental cake, peritoneal nodules, or lymphadenopathy. Thickened
septations (> 2 mm thick) were seen in 11 tumors. The mean maximum
thickness of these septations was 5.1 ± 2.3 mm). Solid components
(nodules or vegetations) were seen in 17 of the 19 tumors (Figs.
3A,
3B,
3C,
3D, and
4). The mean maximum single
dimension for these components was 32.0 ± 19.0 mm.
View larger version (194K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3A. —77-year-old woman with stage I ovarian tumor. Sagittal (A)
and transverse (B) T2-weighted fast spin-echo images (TR/effective TE,
4,500/90 msec) show well-defined heterogeneous mass (arrow).
Hemorrhage within cystic component contributes to its low signal-intensity on
T2-weighting. Florid large nodules and vegetations (arrowheads) are
seen posteriorly. Uterus and cervix (open arrow, B) are
normal. In A, A = anterior, P = posterior.
|
|
View larger version (188K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3B. —77-year-old woman with stage I ovarian tumor. Sagittal (A)
and transverse (B) T2-weighted fast spin-echo images (TR/effective TE,
4,500/90 msec) show well-defined heterogeneous mass (arrow).
Hemorrhage within cystic component contributes to its low signal-intensity on
T2-weighting. Florid large nodules and vegetations (arrowheads) are
seen posteriorly. Uterus and cervix (open arrow, B) are
normal. In A, A = anterior, P = posterior.
|
|
View larger version (122K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3C. —77-year-old woman with stage I ovarian tumor. Photomicrograph of
histopathologic specimen shows serous papillary carcinoma with papillary
formations that form solid nodule (arrows) within a cystic area. (H
and E, x100)
|
|
View larger version (156K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3D. —77-year-old woman with stage I ovarian tumor. Photomicrograph of
histopathologic specimen obtained at high power (x400) shows tumor
islands within stroma (arrow) that confirm invasive nature of lesion.
(H and E)
|
|
View larger version (108K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4. —MDCT scan obtained through mid pelvis of 51-year-old woman with
stage I ovarian tumor shows well-defined mass (solid arrow) of
heterogeneous density with large nodular solid components
(arrowhead). Small amount of ascites (open arrow) is
seen.
|
|
There was a significant difference in age between the groups (p =
0.04), with the borderline tumor group tending to be younger. No significant
difference was found between the two groups in the largest diameter of each
mass (p = 0.16). Although the number of septations per mass was
similar in the borderline and stage I tumors, there was a significant
difference between two types of masses in the thickness of septations (mean,
3.3 ± 1.5 mm vs 5.1 ± 2.34 mm, p = 0.044). Also,
although the numbers of nodules and vegetations within the masses were
similar, the borderline tumors had significantly smaller solid components than
the stage I masses (mean, 15.0 ± 9.8 mm vs 32.0 ± 19.0 mm,
p = 0.003).
ROC curves were constructed for the data on the thickness of septations and
on the size of the nodules. Images obtained in patients who had no septations
but who did have large solid components were scored as "no
information" for septal thickness, because the solid components obscured
visualization and evaluation of any septation. The sensitivity for detecting
an invasive tumor using different values for the thickness of the septation
was plotted against the inverse of the sensitivity. The area under this ROC
curve (Az) indicates the accuracy of using the thickness
of the septation to identify invasive tumors; the closer the value is to 1.0,
the better the test. The Az for the thickness of
septations was 0.75. The cutoff value suggested by the ROC curve analysis was
a thickness of more than 4.5 mm. Using this value, we found that the
sensitivity and specificity for the identification of invasive tumors were
0.55 and 0.94, respectively. We analyzed the nodule sizes in the same way. The
Az was 0.68, and the sensitivity and specificity for the
identification of invasive tumors using a cutoff value of a nodule size of
greater than 28.5 mm were 0.55 and 0.82, respectively
Discriminant analysis was performed using septal thickness, maximum
diameter of solid components, and patient age. There was no significant
variability between groups (Box's test of equality of covariance matrices,
p = 0.16). Age did not contribute significantly to the model
(p = 0.07), so a discriminant model based only on septal thickness
and maximum diameter of solid components was constructed. Use of this model
resulted in overall correct classification of 78.6% of the tumors (82.4% of
borderline and 72.7% of stage I tumors).
CA 125 levels varied from 4–1,223 U/mL in the borderline group
(median, 45 U/mL; first quartile, 17.5 U/mL; third quartile, 176.5 U/mL) and
from 6–1,304 U/mL in the stage I group (median, 49 U/mL; first quartile,
25 U/mL; third quartile, 411 U/mL). No significant difference in CA 125 levels
was found between the two groups (p = 0.40).
Discussion
Our study shows that borderline tumors often present as complex adnexal
masses; none was a purely simple cystic lesion. Thus, although borderline
ovarian tumors may appear to be benign masses clinically, their radiologic
features require differentiation from early invasive disease
[3–6].
Both borderline and stage I tumors are usually multiloculated with significant
solid elements. The architecture of the solid component includes smooth round
nodules, plaquelike thickening, and papilliform projections, with no tendency
for any specific feature to predominate in malignancy. However, the thickness
of the septations and the size of the nodules are greater in invasive tumors
than in borderline tumors.
Other small series have described the complex architectural features in
borderline tumors: thickened walls, endocystic vegetations, focal cystic wall
masses, and branching papillary fronds
[7]. Large size also has been
described as a feature of these masses, with many being larger than 10 cm in
diameter. It was not possible to differentiate between borderline and invasive
tumors on the basis of size criteria in our study.
The complex architectural features were recognizable on both MDCT and MRI,
although the contrast between the solid and cystic components of the tumor was
greatest on T2-weighted MR images, making the solid components easiest to
define on these images. However, because patients underwent either MDCT or MRI
depending on the referral pattern and equipment availability, a comparison
between MDCT and MRI was not possible.
It was not possible to differentiate between borderline and invasive tumors
on the basis of MRI relaxation characteristics. The solid components were
usually intermediate in signal intensity on both T1- and T2-weighted images.
This finding reflects the composition of the solid components on histology in
both borderline and stage I cases: papillary fibrovascular cores are covered
by epithelium, with varying degrees of nuclear atypia
[8]. The stromal invasion
identified on histology in stage I tumors is difficult to define with the
resolution of current scanning techniques (pixel size, 
1 mm). Borderline
tumors have shown early enhancement of the walls and septations on dynamic
gadolinium-enhanced studies
[7]. The use of
contrast-enhanced MRI has proven to be unhelpful for the differentiation of
borderline from invasive tumors
[9].
The presence of ascites as a predictor of ovarian malignancy has been the
subject of much debate. A retrospective study of 125 patients with ovarian
masses found ascites in 58% of the borderline and in 73% of the malignant
cases; however, most cases in the latter group were stage III and IV disease
[10]. The presence of ascites
is therefore not a good feature with which to differentiate between borderline
and stage I disease. This is borne out by our cohort, in which ascites or free
fluid was seen in 23.5% (4/17) of the borderline tumors and 38.4% (5/13) of
the malignant tumors.
CA 125 levels were marginally raised in women with borderline tumors and
bear out the observations of others
[7]. However, the level of this
tumor marker does not provide a means of differentiating borderline from stage
I disease.
In conclusion, this study shows that the radiologic features of borderline
tumors are similar to those of stage I ovarian cancers. The solid components
are smaller and the septations are thinner in the borderline tumors, but
although these features may be helpful in predicting likelihood of invasive
tumors, neither feature allows confident differentiation of borderline from
stage I disease.
Acknowledgments
We thank Clare Peckitt for her statistical advice.
References
- Manchul LA, Simm J, Levin W, et al. Borderline epithelial ovarian
tumors: a review of 81 cases with an assessment of the impact of treatment.
Int J Radiat Oncol Biol Phys1992; 22:867
-874[Medline]
- Rice LW, Berkowitz RS, Mark SD, et al. Epithelial ovarian tumors of
borderline malignancy. Gynecol Oncol1990; 39:195
-198[Medline]
- Buy JN, Ghossain MA, Sciot C, et al. Epithelial tumors of the
ovary: CT findings and correlation with US. Radiology1991; 178:811
-818[Abstract/Free Full Text]
- Dobson M, Carrington BM, Radford JA, Buckley CH, Crowther D. The
role of computed tomography in the management of ovarian tumours of borderline
malignancy. Clin Radiol1997; 52:280
-283[Medline]
- Grab D, Flock F, Stohr I, et al. Classification of asymptomatic
adnexal masses by ultrasound, magnetic resonance imaging, and positron
emission tomography. Gynecol Oncol2000; 77:454
-459[Medline]
- Rieber A, Nussle K, Stohr I, et al. Preoperative diagnosis of
ovarian tumors with MR imaging: comparison with transvaginal sonography,
positron emission tomography, and histologic findings.
AJR 2001; 177:123
-129[Abstract/Free Full Text]
- van Vierzen PB, Massuger LF, Ruys SH, Barentsz JO. Borderline
ovarian malignancy: ultrasound and fast dynamic MR findings. Eur J
Radiol 1998;28:136
-142[Medline]
- Jung SE, Lee JM, Rha SE, Byun JY, Jung JI, Hahn ST. CT and MR
imaging of ovarian tumors with emphasis on differential diagnosis.
RadioGraphics2002; 22:1305
-1325
- Takemori M, Nishimura R, Hasegawa K. Clinical evaluation of MRI in
the diagnosis of borderline ovarian tumors. Acta Obstet Gynecol
Scand 2002;81:157
-161[Medline]
- Shen-Gunther J, Mannel RS. Ascites as a predictor of ovarian
malignancy. Gynecol Oncol 2002;87
: 77-83[Medline]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
Top
Abstract
Introduction
