CME. MRI of Ovarian Masses. Review. Hebert Alberto Vargas, MD, 1 * Tristan Barrett, MBBS, BSc, 2 and Evis Sala, MD, PhD 2

Transcription

1 CME JOURNAL OF MAGNETIC RESONANCE IMAGING 37: (2013) Review MRI of Ovarian Masses Hebert Alberto Vargas, MD, 1 * Tristan Barrett, MBBS, BSc, 2 and Evis Sala, MD, PhD 2 This article is accredited as a journal-based CME activity. If you wish to receive credit for this activity, please refer to the website: ACCREDITATION AND DESIGNATION STATEMENT Blackwell Futura Media Services designates this journalbased CME activity for a maximum of 1 AMA PRA Category 1 Credit TM. Physicians should only claim credit commensurate with the extent of their participation in the activity. Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. EDUCATIONAL OBJECTIVES Upon completion of this educational activity, participants will be better able to discuss the usefulness of MRI in differentiating benign and malignant ovarian tumors, describe the typical MRI features of the most common benign and malignant ovarian masses, and discuss the role of conventional and advanced MRI techniques in staging, treatment selection, and follow up of patients with ovarian cancer. ACTIVITY DISCLOSURES No commercial support has been accepted related to the development or publication of this activity. Faculty Disclosures: The following contributors have no conflicts of interest to disclose: Editor-in-Chief: C. Leon Partain, MD, PhD CME Editor: Scott B. Reeder, MD, PhD CME Committee: Scott Nagle, MD, PhD, Pratik Mukherjee, MD, PhD, Shreyas Vasanawala, MD, PhD, Bonnie Joe, MD, PhD, Tim Leiner, MD, PhD, Sabine Weckbach, MD, Frank Korosec, PhD Authors: Hebert Alberto Vargas, MD, Tristan Barrett, MBBS, BSc, and Evis Sala, MD, PhD This manuscript underwent peer review in line with the standards of editorial integrity and publication ethics maintained by Journal of Magnetic Resonance Imaging. The peer reviewers have no relevant financial relationships. The peer review process for Journal of Magnetic Resonance Imaging is double-blinded. As such, the identities of the reviewers are not disclosed in line with the standard accepted practices of medical journal peer review. Conflicts of interest have been identified and resolved in accordance with Blackwell Futura Media Services s Policy on Activity Disclosure and Conflict of Interest. No relevant financial relationships exist for any individual in control of the content and therefore there were no conflicts to resolve. INSTRUCTIONS ON RECEIVING CREDIT For information on applicability and acceptance of CME credit for this activity, please consult your professional licensing board. This activity is designed to be completed within an hour; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period. Follow these steps to earn credit: Log on to Read the target audience, educational objectives, and activity disclosures. Read the article in print or online format. Reflect on the article. Access the CME Exam, and choose the best answer to each question. Complete the required evaluation component of the activity. This activity will be available for CME credit for twelve months following its publication date. At that time, it will be reviewed and potentially updated and extended for an additional period. 1 Department of Radiology, Memorial Sloan-Kettering Cancer Center, New York, New York, USA. 2 Radiology, Addenbrooke s Hospital and University of Cambridge, Cambridge, United Kingdom. *Address reprint requests to: H.A.V., Radiology Department, Room C-278, Memorial Sloan-Kettering Cancer Center, 1275 York Av, New York, NY vargasah@mskcc.org Received May 2, 2011; Accepted May 2, DOI /jmri View this article online at wileyonlinelibrary.com. VC 2012 Wiley Periodicals, Inc. 265

2 266 Vargas et al. MRI provides exquisite views of the pelvic anatomy through its high spatial resolution and tissue contrast, and as such plays a key role in the work up of ovarian lesions, identifying features that distinguish benign and malignant lesions. In the case of primary tumors it enables local staging and detection of metastatic disease to help guide management options such as complex surgery or the consideration of neoadjuvant chemotherapy. Functional MRI techniques such as diffusion-weighted MRI (DW-MRI), dynamic contrast-enhanced MRI (DCE-MRI) and tumor-selective molecular imaging are currently being evaluated as possible predictive and prognostic biomarkers in the context of ovarian malignancy, and may play a larger role in routine clinical practice in the future. Herein we provide an overview of the conventional and advanced MRI techniques used to characterize ovarian masses and of the role that MR plays in the staging, treatment selection and follow up of patients with ovarian cancer. Key Words: ovarian cancer; magnetic resonance imaging; diffusion-weighted MRI; dynamic contrast-enhanced MRI; apparent diffusion coefficient J. Magn. Reson. Imaging 2013;37: VC 2012 Wiley Periodicals, Inc. THE OVARIES ARE paired nodular structures situated on either side of the uterus in relation to the lateral walls of the pelvis, and are responsible for the production of the female reproductive cells (ova) and the secretion of hormones (mainly estrogen and progesterone). Although their exact location is highly variable, they are typically found below the bifurcation of the common iliac vessels lateral to the uterine cornua. Histologically, the ovaries consist of several vesicular follicles imbedded in a mesh of stroma formed of connective tissue and spindle cells. The surface of the ovary is covered by a layer of columnar epithelium. Both benign and malignant tumors can arise from abnormal proliferation of any of the tissue types contained within the ovaries. The main role of imaging in clinical practice is to identify features distinguishing benign and malignant lesions, and to discriminate between ovarian pathology and pathologies from adjacent organs, such as the uterus and gastrointestinal tract. Ultrasound is generally used as first line examination for investigating ovarian lesions, mainly due to increased availability and reduced cost. MRI provides exquisite views of the pelvic anatomy through its superb spatial and tissue contrast resolution, and is particularly helpful for characterizing sonographically indeterminate lesions. Furthermore, MRI techniques such as diffusion-weighted MRI (DW-MRI) and dynamic contrast-enhanced MRI (DCE-MRI) are currently being evaluated as possible predictive and prognostic biomarkers in the context of ovarian malignancies, and could possibly play a larger role in routine clinical practice in the future. The purpose of this review is to discuss the usefulness of MRI in differentiating benign and malignant ovarian tumors, to describe the typical MRI features of the most common benign and malignant ovarian masses and to discuss the role of conventional and advanced MRI techniques in staging, treatment selection and follow up of patients with ovarian cancer. MRI TECHNIQUE Patient Preparation and Coil Selection Adequate patient preparation is extremely important to obtain optimal results. Patients are instructed to fast for 4-6 h before MRI examination to limit artifacts due to bowel peristalsis (1). Further reduction of motion artifacts can be achieved with the use of antiperistaltic agents such as hyoscine butyl bromide or glucagon. A full bladder may degrade images due to ghosting and motion artifacts, therefore, patients are asked to void before the examination. Imaging is performed with the patient in the supine position using a multichannel phase array coil (e.g., a cardiac coil usually offers the best image quality). Endoluminal coils (endorectal and endovaginal), although sometimes used in gynecological imaging to provide high resolution views of small cervical tumors and asses subtle parametrial invasion, have a limited field of view for the assessment of lateral pelvic structures, and are therefore not routinely used for imaging the ovaries. Conventional MRI Sequences A typical imaging protocol includes T1-weighted images (T1WI) in the axial plane and T2-weighted images (T2WI) in the axial, sagittal and coronal planes using a small field of view (e.g., 20 cm), a high resolution matrix (e.g., ) and thin section images (e.g., 3 mm) (Table 1). Fat-saturation T1-weighted images (T1W FS) are used to evaluate the presence of fat or hemorrhage, especially when high signal intensity is seen within the adnexa on T1WI and the diagnosis of a dermoid tumor or endometriosis is suspected. In patients undergoing ovarian cancer staging, axial T1WI or fast spin-echo T2WI with a large field-of-view are used to evaluate the abdomen for the presence of metastatic disease (e.g. suprarenal lymph nodes, liver or bone metastases). Contrast-enhanced MRI Sequences Gadolinium-enhanced images are useful for the evaluation of complex lesions as they may help differentiate solid components or papillary projections from clots and debris. Multiphase contrast-enhanced MRI is performed by using a 3-dimensional gradient echo T1-weighted sequence after administration of 0.1 mmol gadolinium per kg of body weight (preferably using an automatic injector). These enable acquisition of images at multiple phases of contrast medium enhancement (e.g., before contrast administration, 1 min and 5 min after contrast medium administration). Dynamic contrast-enhanced MRI (DCE-MRI) involves repeatedly imaging a predetermined volume of interest every few seconds over 5-10 min during the passage of the contrast agent through the tissue of interest, so that changes in signal intensity can be assessed as a function of time. Semiquantitative analysis can be performed through several parameters derived from contrast enhancement/time curves. Pharmacokinetic models are also used to characterize tissue permeability, which in tumors reflects perfusion because of high first pass extraction values. Several pharmacokinetic models have been proposed to characterize

3 MRI of Ovarian Masses 267 Table 1 Typical Imaging Protocol for Evaluating Ovarian Masses Sequence T1* FSE T2 FSE T2 FSE T2 T1 pre-gd T1 70s-post Gd* T1 5m-post Gd* DW-MRI Plane Axial Axial Coronal Sagittal Axial Axial Axial Axial Sequence 2D SE 2D FSE 2D FSE 2D FSE 3D SPGR 3D SPGR 3D SPGR EPI TR (ms) Default Default Default Flip angle ( ) N/A N/A N/A N/A N/A Field of view (mm 2 ) Slice thickness (mm) Gap (mm) Phase encoding steps Frequency steps Breathhold No No No No Yes Yes Yes No FSE ¼ fast spin-echo; SE ¼ spin-echo; fat-sat ¼ fat saturation; Gd ¼ Gadolinium-chelated low molecular weight contrast agent; DW-MRI ¼ diffusion-weighted MRI; EPI ¼ echo-planar imaging; TR ¼ repetition time. *T1-weighted post-contrast images can be conventional spin echo, but in the interest of time most centers use gradient echo sequences which are faster but yield images of lower quality. tissue perfusion on DCE-MRI. The most common is a two-compartmental model developed by Tofts (2), from which several variables can be quantitatively measured; these include variables reflective of tumor permeability such as the volume transfer constant (K trans ) from the vascular space (VS) to the extravascularextracellular space (EES), the rate constant (k ep ) from EES to VS and the fractional volume of EES (v e ). Diffusion-weighted MRI (DW-MRI) DW-MRI is a functional imaging technique that displays information about water mobility, tissue cellularity, and the integrity of the cellular membranes. No exogenous contrast medium administration is necessary as image contrast is derived from inherent differences in the restriction of water molecule movements, thus facilitating the incorporation of DW-MRI sequences in routine imaging protocols. Diffusion-weighted images are obtained by measuring signal loss after a series of symmetric pairs of motion-probing gradients (MPG) are added at both sides of a 180 refocusing radiofrequency pulse of a modified single or dual spin-echo T2-weighted sequence. The weighting of the MPGs is indicated by the b-value (measured in s/mm 2 ). DW- MRI is performed using two or more b-values, which include one or more low b-values ( s/mm 2 ) and a high b-value ( s/mm 2 ). The apparent diffusion coefficient (ADC) is measured in mm 2 /s and is calculated by the slope of the line of the natural algorithm of signal intensity versus b-values. The ADC maps are usually displayed as gray-scale images. Areas of restricted diffusion have lower ADC values and appear as a darker shade of gray on the ADC maps compared with areas of freely moving water such as cysts or bladder, which appear as a lighter shade of gray. Areas of restricted diffusion appear bright in the high b-value images (b-value ¼ s/mm 2 ). NORMAL MRI ANATOMY The entire pelvic anatomy is exquisitely depicted on pelvic MRI. On T1WI, the ovaries display homogeneous low to intermediate signal intensity, whereas on T2WI the follicles, if present, become hyperintense compared with the surrounding stroma (Fig. 1). Immature follicles are usually smaller than 1 cm in size however normal ovaries may contain follicles up to 3 cm in size (3). With the advent of high resolution imaging, the ovaries are now frequently identified in postmenopausal women (3). They appear as predominantly solid structures with a relative increase in stromal tissue, which is T2 hypointense, and may contain small T2 hyperintense follicles, although dysfunctional cysts up to several centimeters in size may also be encountered (4). The normal fallopian tubes are not routinely seen because of their small size and tortuous course. The size of the ovaries and the presence and size of follicles varies depending on the patient s age and menopausal status. The zonal anatomy of the uterus, cervix and vagina are also well demonstrated on T2WI. ROLE OF MRI IN CHARACTERIZING OVARIAN MASSES The greatest strength of MRI lies in its ability to characterize the physical and biochemical properties of different tissues (e.g., water, iron, fat, and extravascular blood and its breakdown products) through the use of multiple imaging sequences. It is important to recognize that as there are no MRI signal intensity characteristics that are specific for malignant epithelial tumor, such tumors must be distinguished based on morphologic criteria. The MRI features most predictive of malignancy are an enhancing solid component or vegetations within a cystic lesion, presence of necrosis within a solid lesion as well as presence of ascites and peritoneal deposits (5,6). The presence of at least one of the primary criteria coupled with a single criterion from the ancillary group correctly characterizes 95% of malignant lesions (5,6) (Table 2). Both transvaginal ultrasound and contrast-enhanced MRI have high sensitivity (97% and 100%, respectively) in the identification of solid components within an adnexal mass. MRI, also shows high specificity (92%) (5). Other MRI techniques also have a role in characterizing ovarian masses (7). Multiphase and dynamic contrast-enhanced MRI helps differentiate solid

4 268 Vargas et al. Figure 1. Normal ovaries at MRI. a c: Axial T1 (a), axial T2 (b) and coronal T2-weighted images (c) of a normal right ovary in a premenopausal woman. The ovary demonstrates a homogeneous low signal intensity on T1 (arrow in a). The stromal tissue appears as intermediate signal on T2 (arrow in b), whilst follicles are of high signal intensity (open arrows in b, c) and can normally measure up to 3 cm. d: Axial T2-weighted image in a post menopausal patient, the ovaries appear as atrophic solid structures with a relative increase in stromal tissue, which is T2 hypointense (arrows). components or papillary projections from clots and debris (8,9). In addition, DCE-MRI might play a role in preoperative evaluation of borderline ovarian malignancies (9). In a small prospective study, Van Vierzen et al found that the accuracy to detect a tumor of borderline malignancy with CA 125, ultrasound and DCE-MRI was 50, 63, and 75%, respectively (9). Another recent study (10) found that DCE-MRI can help distinguish between benign, borderline, and invasive tumors. The same authors have also shown that the addition of a postprocessed semi-quantitative DCE-MRI sequences (analyzing time-signal intensity curves) to conventional protocols significantly improved the diagnostic accuracy and reader confidence in characterizing complex adnexal masses as benign or malignant (11). The use of DW-MRI to differentiate benign from malignant ovarian tumors remains controversial. Nakayama et al (12) found no significant difference in the ADC values between the benign and malignant cystic ovarian lesions. There was a wide variation in the ADC values of malignant ovarian tumors which is related to their morphological heterogeneity. Recently, Thomassin-Naggara et al have also shown that the presence of low-signal intensity solid components on high b-value DW-MRI was the most significant criterion for prediction of a benign lesion (11,13). However, the ADC values did not contribute in differentiating benign from malignant adnexal lesions. False-positive results on DW-MRI arise from normal hypercellular tissues, such as small bowel mucosa, and noncancerous lesions, such as tubo-ovarian abscesses, inflammatory collections, lymphoceles and proteinaceous or hemorrhagic ovarian cysts (14); false-negative results may occur in malignant lesions of low cellularity, such as mucinous subtypes, and in predominantly cystic or calcified deposits (15). Preliminary results also show significantly lower ADC in peritoneal deposits compared with primary ovarian tumors and omental cake (16), which may relate to stromal reactivity at peritoneal sites. ALGORITHMIC APPROACH TO MRI EVALUATION OF OVARIAN MASSES A practical approach for the evaluation of ovarian masses on MRI is to divide them into three groups

5 MRI of Ovarian Masses 269 Table 2 Criteria for Differentiating Benign From Malignant Ovarian Masses Criteria Benign Malignant Primary criteria Cystic Simple Papillary projection Septa < 3 > 3 Wall/septa thickness < 3mm > 3mm Solid Homogeneous Heterogeneous with necrosis or hemorrhage *Size < 4cm > 4cm Ancillary criteria Ascites Absent Present Peritoneal/omental Absent Present deposits Lymph nodes (LN) No enlarged LN Enlarged LN *This is an US criteria, which has not been shown to be an independent predictor of malignancy on MRI (6,7). based of their features on T1- and T2-weighted sequences. The 3 groups are: (i) T1 bright masses (high T1 signal intensity), (ii) T2 solid masses (signal intensity similar to that on skeletal muscle) and (iii) mixed solid/cystic masses (3,17). A decision tree for MRI characterization of ultrasound indeterminate adnexal masses based on these groups is presented in Figure 2 (4,17). DIFFERENTIAL DIAGNOSIS OF T1 BRIGHT MASSES Endometriosis and Endometriomas Endometriosis refers to the presence of endometrial glandular tissue outside of the uterus. Common locations include ovary, uterine ligaments, Fallopian tube, recto-vaginal septum, pouch of Douglas, bladder wall, utero-vesical fold and umbilicus. On MRI, endometriotic implants may be seen as spiculated bands (or occasionally solid masses) of intermediate to low signal intensity on both T1W and T2WI due to fibrosis. Small hemorrhagic endometrial implants become more obvious on fat-saturated T1WI (18). Ovarian involvement results in the development of endometriomas which are cystic lesions containing blood products. Typical MRI findings of endometriomas include multiple lesions which are hyperintense on T1WI (due to the presence of methemoglobin) and hypointense on T2WI (due to high iron concentration from blood degradation products). Other possible appearances include lesions which are hyperintense on both T1WI and T2WI, or a range of low signal intensities within the lesion (shading) or in the wall of the lesion (due to hemosiderin deposition) on T2WI (Fig. 3). Endometriomas retain their high signal intensity on fat-suppressed T1WI. Mature Cystic Teratomas (Dermoid cysts) Mature cystic teratomas are the most common benign ovarian tumor in women younger than 45 years (19). They are derived from all three germ-cell layers, with the ectodermal component usually being most prominent. Typical MRI findings include a mass of high signal intensity on T1WI and intermediate signal intensity on T2WI, fat fluid and/or fluid fluid levels; layering debris; low-signal-intensity calcifications (e.g. teeth) and soft-tissue protuberances (Rokitansky nodules or dermoid plugs) attached to the cyst wall (20); (Fig. 4). Both mature cystic teratomas and endometriomas show high signal intensity on T1WI, however, on fat-suppressed T1WI, teratomas show drop in signal intensity whereas endometriomas retain their high signal. In addition, the fat in mature cystic teratomas results in chemical shift artifact at the fat fluid interface which appears as bright or dark bands along the frequency-encoding direction. Immature teratomas are malignant ovarian tumors which also demonstrate fat on MRI. They are usually very large at presentation and have significant solid components which contain coarse calcification and multiple small foci of fat. DW-MRI may also be useful particularly in the detection of mature cystic teratomas containing a small amount of fat Figure 2. Decision tree for MRI of the US indeterminate adnexal mass (**). FST1W: fat-suppressed T1- weighted images; T2W: T2-weighted images; CET1W: contrast-enhanced T1-weighted images. *Masses with MR features suggestive of malignancy should undergo CET1W imaging. **Reproduced with kind permission from Springer Science: Eur Radiol, ESUR guidelines for MR imaging of the sonographically indeterminate adnexal mass: an algorithmic approach, Spencer et al. 2010;20:25 35.

6 270 Vargas et al. that is not detectable by conventional MRI. The cystic components of mature cystic teratomas show significantly lower ADC values than endometriomas, and both benign and malignant cystic ovarian tumors. Endometriomas also show low ADC values due to their hemosiderin content which results in T1 shortening and decrease in ADC values (12,21,22). Teaching point: Both mature cystic teratomas and endometriomas show high signal intensity on T1WI, however, on fat-suppressed T1WI, teratomas show drop in signal intensity whereas endometriomas retain their high signal. DIFFERENTIAL DIAGNOSIS OF BENIGN T2 SOLID MASSES Fibrotic Tumors Fibromas, thecomas, and fibrothecomas are all fibrotic tumors of sex-cord stromal origin that account for approximately 5% of all ovarian tumors. They are usually asymptomatic and typically detected in middleaged women during routine gynecologic examination. Due to their solid appearance on imaging, they may mimic pedunculated uterine leiomyomas and even malignant ovarian tumors. They are associated with ascites in 15% of the cases and pleural effusion (Meigs syndrome) in 1%. On MRI, the fibrotic component tipycally demonstrates homogeneous low signal intensity on both T1WI and T2WI (Fig. 5). In addition, scattered areas of high signal intensity may be present on T2WI, representing cystic degeneration or edema. Thecomas may also have a prominent lipid content. Benign Epithelial Tumors Depending on their pathologic features and clinical behavior, epithelial ovarian tumors can be classified as benign, borderline (low malignant potential) or malignant. Benign types of serous and mucinous tumors are common (together accounting for approximately 50% of benign ovarian neoplasms) while benign forms of endometrioid and clear cell tumors are exceptionally rare. Serous cystadenomas are the most common benign epithelial tumors. They can occur at any age, with a peak incidence in the 4th and 5th decades, and are bilateral in approximately 20% of cases (23). They appear as thin-walled ovarian unilocular cysts and are usually indistinguishable from follicular cysts on MRI (Fig. 6). However, unlike follicular cysts, serous cystadenomas remain unchanged (and may even increase in size) over subsequent menstrual cycles. The signal intensity of cyst contents is variable but is usually low to intermediate on T1- weighted images and high on T2-weighted images. The cyst wall may contain small nodules which are of high signal intensity peripherally and have thin low-signal-intensity cores on T2-weighted images due to fibrosis and/or calcification. If these solid projections are prominent, a borderline tumor should be suspected. Figure 3. Endometriomas. a: Axial T1-weighted imaging shows two high signal intensity masses arising from the left ovary (arrows) and the two ovaries touching in the midline: the kissing ovaries sign. b: T2-weighted images show the masses to be of lower signal compared with T1 imaging, with T2 shading demonstrated. c: Fat saturated-t1 images confirm that the masses do not contain fat and actually increase in signal intensity, consistent with endometriomas.

7 MRI of Ovarian Masses 271 Figure 4. Benign germ cell tumor: mature cystic teratoma. a: Axial CT images demonstrates a left pelvic cystic lesion containing macroscopic fat (open arrow), fluid (*), an area of calcification (black arrow), and a mural nodule projecting into the cyst cavity (white arrow), known as a Rokitansky nodule. b d: Axial T1-weighted (b), T2 (c), and T2-fat suppressed images (c) in a different patient. There is a complex right adnexal lesion with a fluid component showing low T1 and high T2 signal intensity (open arrows in b, c) and a component with high signal intensity on T1 and T2-weighted imaging (arrows in b, c, d), which is confirmed to contain fat following signal drop-out on the fat-suppressed imaging. Mucinous cystadenomas are the second most common benign epithelial tumors. They are bilateral in only 2 3% of cases. Their typical MRI appearance is a large multilocular cyst without solid components. The signal intensities of the different locules often vary depending on the amount of proteinaceous or mucinous fluid and whether hemorrhage is present, giving rise to the so-called stained-glass appearance. Teaching Point: Lesions which are hypointense on T1 and T2WI are typically benign fibrotic tumors Benign epithelial tumors typically appear as unilocular or multilocular cysts and may contain thin, nonenhancing septations. Solid components are usually absent or small. DIFFERENTIAL DIAGNOSIS OF MALIGNANT T2 SOLID MASSES THE USEFULNESS OF CON- TRAST-ENHANCED MRI Approximately 90% of ovarian cancers originate from epithelial cells. Malignant histological subtypes include serous (50%), mucinous (20%), endometrioid (20%), clear cell (10%) and undifferentiated (1%). Each of these subtypes has a borderline counterpart, defined as having some or all of the features of a malignant tumor but no stromal invasion, behave clinically in a more aggressive way than a benign neoplasm but have a better prognosis than an invasive malignancy. Nonepithelial cancers include germ cell tumors and sex-cord stromal tumors. Germ cell tumors are made up of teratomas which can be benign (e.g., dermoid) or malignant (e.g., dysgerminomas, yolk sac tumors (Fig. 7), choriocarcinomas and gonadoblastomas). Sex-cord stromal tumors include granulosa cell tumors (Fig. 8), thecomas and Sertoli-Leydig cell tumors. Mixed histology tumors and undifferentiated carcinomas may also occur. They are usually metastatic at the time of presentation and are associated with very poor prognosis (Fig. 9). MRI techniques cannot confidently differentiate between specific surface epithelial, germ cell, stromal cell or metastatic tumors. Features that suggest the presence of malignant tumors include mixed solid and cystic appearance with irregular solid components. The solid components show avid enhancement and areas of necrosis. These features combined with a disproportionately large amount of ascites compared with the tumor

8 272 Vargas et al. Figure 5. Ovarian fibroma. a: Transabdominal ultrasound shows a hypo-echoic right adnexal mass with marked posterior acoustic attenuation due to homogeneous fibrous tissue. b,c: Axial T1 (b) and T2-weighted (c) MR images show a right adnexal mass of homogeneous low signal intensity (arrows in b, c) on both T1 and T2 imaging characteristic of an ovarian fibroma, which was confirmed at histopathology. size, presence of enlarged lymph nodes and peritoneal and/or serosal implants increase the likelihood of malignancy. Mucinous tumors may be of high signal intensity on T1WI due to high protein concentration within the mucoid material. They may also be associated with pseudomyxoma peritonei appearing on MRI as diffuse intraperitoneal material of low signal intensity on T1-weighted images and high signal intensity on T2WI that may contain septae and cause scalloping of the liver and spleen surfaces as well as displacement of the bowel loops due to pressure effects (19). Endometrioid and clear cell carcinomas are associated with endometriosis and endometriomas. Regardless of the histological subtype, ovarian cancer is staged according to the International Federation of Gynecology and Obstetrics (FIGO) or the American Joint Committee on Cancer TNM criteria (Table 3) (24,25). Teaching Points Features that suggest the presence of malignant tumors include mixed solid and cystic appearance with irregular solid components. The solid components show avid enhancement and areas of necrosis. These likelihood of malignancy is increased in the presence of a disproportionately large amount of ascites compared with the tumor size, presence of enlarged lymph nodes and peritoneal and/or serosal implants. DIFFERENTIAL DIAGNOSIS OF BENIGN CYSTIC ADNEXAL LESIONS ON MRI Functional Cysts These include follicular cysts that result from failure of the follicle to rupture or regress and

9 MRI of Ovarian Masses 273 Figure 6. Benign epithelial tumors: serous cystadenoma. a: Axial CT image shows a low attenuation cystic mass in the right flank (arrow). Axial T1 (b) and T2-weighted (c) MR imaging confirms homogeneous low signal intensity on T1 and high signal on T2-weighted imaging. d: Sagittal T2-weighted imaging confirms the presence of multiple locules (arrows), but with thinwalled septae and walls with no soft tissue mass. Histology confirmed a serous cystadenoma. Figure 7. Malignant germ cell tumor: yolk sac tumor. a,b: Axial (a) and sagittal (b) T2-weighted MR imaging in a 16-year-old female patient. There is a complex right adnexal mass with a thickened wall and both solid (black arrow in a) and cystic (white arrows in a, b) components and an enlarged right external iliac lymph node (open arrow in a). Histology confirmed a yolk sac tumor.

10 274 Vargas et al. Figure 8. Granulosa cell tumor. a: Transabdominal Doppler US image shows a solid adnexal lesion with blood flow within and a central cystic area. b: Axial T2-weighted MR shows the mass to be of intermediate signal intensity (arrow) and confirms the central fluid-filled cyst; an incidental uterine fibroid is noted (open arrow in b). c,d: Fat-suppressed axial T1-weighted images pre (c) and post (d) contrast show the mass to be of intermediate T1 signal intensity with homogeneous enhancement. Histology confirmed a granulosa cell tumor. corpus luteum cysts which form after the release of the ova from the follicle during ovulation. Both are common in women of reproductive age. Follicular cysts appear as a thin-walled unilocular nonenhancing structure ranging from 3 8 cm in size which demonstrates typical low signal intensity on T1WI and high signal intensity on T2WI (Fig. 10). Malignancy is highly unlikely in the absence of any solid components. The differential diagnosis includes serous cystadenoma of the ovary, which may sometimes contain thin, regular septae. Hemorrhage into a corpus luteum cyst can produce a complex appearance that changes over time depending on the stage of clot evolution. On MRI, it appears as a complex structure which may contain areas of increased T1 signal intensity (representing hemorrhage) as well as thick, irregular, avidly enhancing wall. It should regress spontaneously after 1 2 menstrual cycles (Fig. 11). Polycystic Ovary Disease (Stein-Leventhal Syndrome) Polycystic ovary disease is a clinical syndrome characterized by clinical or biochemical evidence of hyperandrogenism, chronic anovulation and polycystic ovaries on imaging (19). The ovaries are typically enlarged (volume 10 ml) with a prominent central stroma of low to intermediate signal intensity on T2WI and multiple small peripheral high T2 signal intensity follicles ( 12 follicles per ovary; each less than 10 mm in size) without a dominant follicle 10 mm in size or a corpus luteum cyst (26) (Fig. 12). Approximately 50%

11 MRI of Ovarian Masses 275 Figure 9. Ovarian cancer. a,b: Sagittal T2-weighted (A) and axial T2-fat suppressed images shows a large heterogeneous multi-lobulated mass arising from the pelvis (white arrows), displacing the uterus (black arrows in a), containing a cystic area with a soft tissue nodule (open arrow in a), and associated ascites (*). The right ovary abuts the mass (arrowhead in b). c,d: Sagittal T1 images pre (c) and post i.v. contrast medium (d) show the mass to be of intermediate to low signal intensity on T1 with heterogeneous enhancement of the solid components. Histology diagnosis: high grade mixed papillary serous tumor with clear cell elements. of patients are hirsute and many are obese. Failure of ovulation may lead to infertility. Tubo-ovarian Abscesses Tubo-ovarian abscess (TOA) is collection of frank pus or an inflammatory mass within the fallopian tube and ovary. These usually arise as a complication of pelvic inflammatory disease in young women, but can occasionally occur in post-menopausal patients. TOAs may be solid, cystic, or complex; and as a result, ultrasound features may be nonspecific. Ancillary MRI and CT features aid diagnosis, but the imaging findings are supported by the clinical picture of sepsis. At CT, the abscess will typically have thick enhancing walls with stranding and edema of the surrounding fat (Fig. 13). Internal septations may also be appreciated and whilst the presence of internal air locules is a strong indicator of the diagnosis, these are not always seen. TOAs typically have low signal intensity on T1WI and heterogeneously high signal intensity on T2WI; the other MRI features mimic those demonstrated by CT (Fig. 13). Teaching Points Follicular cysts appear as a thin-walled unilocular nonenhancing structure with the signal intensity of fluid on MRI. In polycystic ovary syndrome, the ovaries are enlarged with a prominent central stroma of low to intermediate signal intensity on T2WI and multiple small peripheral high T2 signal intensity follicles.

12 276 Vargas et al. Table 3 TNM and FIGO Staging of Primary Ovarian Cancer TNM FIGO Ovary T1 I Tumor confined to ovaries T1a Ia Tumor confined to one ovary, capsule intact T1b Ib Tumor confined to both ovaries, capsule intact T1c Ic Tumor present on the surface of one or both ovaries; capsule ruptured; malignant ascites (positive peritoneal washings) T2 II Tumor involving one or both ovaries with pelvic extension T2a IIa Extension and/metastasis to the uterus or tubes T2b IIb Extension to other pelvic tissue T2c IIc Tumor either stage IIa or IIb but present on surface of one or both ovaries; or capsule ruptured; or malignant ascites (positive peritoneal washings) T3 and/or N1 III Tumor involving one or both ovaries with peritoneal implants outside the pelvis or positive retroperitoneal or inguinal nodes. T3a IIIa Tumor limited to the true pelvis with negative lymph nodes but with histologically confirmed microscopic peritoneal metastases T3b IIIb Peritoneal metastases, none exceeding 2 cm in diameter; negative lymph nodes T3c and/or N1 IIIc Abdominal implants greater than 2 cm in diameter or positive retroperitoneal or inguinal nodes M1 IV Distant metastases Tubo-ovarian abscess should be suspected in a septic patient presenting with a mixed solid/cystic adnexal structure with associated inflammatory changes. MRI IN OVARIAN CANCER STAGING The role of imaging in the staging of ovarian cancer is better established and more widely accepted than in other gynecological malignancies. Although CT and MRI have similar accuracy when used for this purpose, CT is most commonly performed (6,27). MRI is particularly useful in patients with contraindications to enhanced CT, including pregnancy, renal impairment and iodinated contrast allergy (28). Intraperitoneal dissemination is the most common route of spread of ovarian cancer, therefore T1WI and T2WI of the whole abdomen and pelvis should be performed when staging ovarian carcinoma with MRI to evaluate for metastatic disease (e.g., liver) and peritoneal deposits. The most common sites of peritoneal metastases include the pouch of Douglas, paracolic gutters, surface of the small and large bowel, greater omentum, surface of the liver (perihepatic implants) and subphrenic space, although in essence peritoneal implants may occur anywhere in the peritoneal cavity. MRI is very sensitive (95%) for detection of peritoneal metastases, which appear as nodular or plaque-like enhancing soft tissue masses showing delayed enhancement following intravenous gadolinium (29). The detection of small implants is facilitated when surrounded by ascites. MRI is also useful in differentiating between subcapsular liver implants and parenchymal liver metastasis, which alters staging and therapy (30). These implants are best seen on the delayed (5 min) contrast-enhanced fat-suppressed T1- weighted images. Surface implants are usually welldefined, biconvex and peripheral, and they indent the liver. True intraparenchymal metastases are often illdefined, circular, and partially or completely surrounded by liver tissue. Preliminary reports have also shown that DW-MRI may be helpful for detecting the extent of peritoneal disease (Fig. 14). On conventional T1WI, T2WI and fat-suppressed T1WI, small serosal implants invaginated within peritoneal reflections are Figure 10. Simple follicular ovarian cyst. a,b: Axial T1 (a) T2-weighted (b) MR images show a unilocular thin-walled cyst arising from the right ovary (arrow in a, b) with homogeneous low signal intensity on T1 and high signal intensity on T2-weighted imaging. The cyst resolved on follow-up imaging.

13 MRI of Ovarian Masses 277 Figure 11. Hemorrhagic ovarian cyst. a: Axial CT image showing a hyperdense cystic lesion arising from the left ovary (arrow). b d: Axial T1 (b), T1-fat suppressed (c) and T2-weighted (d) MR images in the same patient demonstrating the cyst to be of homogeneous high signal intensity on T1, increasing signal intensity on T1-fat suppressed imaging and with layering on T2-weighted imaging, consistent with blood products. The cyst resolved on follow-up imaging. often obscured by adjacent structures. DW-MRI depicts deposits on the visceral peritoneum as foci of high signal intensity against a background of suppressed signal from surrounding ascites, bowel contents and fat, leading to greater conspicuity. However, there is considerable overlap of mean and lowest ADC values between benign and malignant ovarian tumors (13,21). This may be explained by tissue heterogeneity, where high ADCs typically associated with benign conditions may be observed in malignant tumors with abundance of desmoplastic stroma and low ADCs usually associated with malignant lesions may be present in benign lesions with compact matrix organization such as fibromas (13). Evaluation of the chest is also important in ovarian cancer staging, as lung parenchymal or pleural involvement constitutes advanced (Stage IV) disease. This is typically performed with CT, mainly due to the ability to acquire images rapidly and limit interference from breathing artifacts. MRI IN OVARIAN CANCER TREATMENT SELECTION The contemporary management of ovarian cancer is related to stage and extent of disease and includes primary surgery followed by adjuvant chemotherapy, neoadjuvant chemotherapy followed by interval debulking surgery or primary chemotherapy alone (stage IV disease). Optimal debulking refers to the reduction of all tumor sites to a maximal diameter of less than 1 cm. Imaging plays a central role in every stage of the management of patients with ovarian cancer. Before treatment, it is used for characterizing the primary tumor and identifying the presence of metastatic disease or disease features that may result in complex surgery or be an indication for neoadjuvant chemotherapy. Such features include the presence of tumor deposits greater than 1 cm in size in the lesser sac, gastro-splenic ligament, fissure for the ligamentum teres, porta hepatis, subphrenic space, small bowel mesentery, or retroperitoneal nodes above the renal hila, as well as invasion of the pelvic side wall, rectum, sigmoid colon or urinary bladder. MRI IN OVARIAN CANCER FOLLOW-UP The mainstay of ovarian cancer response assessment is based on serial measurements of serum 125 in combination morphologic evaluation with CT and/or MRI (31,32). The limitations of CA 125 as a tumor

14 278 Vargas et al. Figure 12. Polycystic ovaries syndrome (PCOS). Imaging features consistent with the clinical syndrome of PCOS include 12 or more ovarian cysts (Rotterdam criteria), of 2 9 mm diameter, predominantly located in the periphery of at least one ovary, with an ovarian volume of >10 ml. Axial (a,b) and sagittal (c) T2-weighted MR imaging showing multiple small peripheral cysts located in the right and left ovaries in the same patient (arrows). Incidental note is made of a didelphys uterus in this patient (open arrow in a); uterine anomalies are commonly associated with PCOS. marker are well known, namely the fact that normal values do not exclude the presence of disease, and elevated values cannot establish the location and extent of recurrence (33 35). Gadolinium-enhanced MRI is comparable (sensitivity 90% and specificity 88%) to laparotomy (sensitivity 88% and specificity 100%) but superior to serum CA-125 (sensitivity 65% and specificity 88%) in the detection of residual or recurrent peritoneal and serosal implants in women who have been treated for ovarian cancer (29,36). In the presence of recurrence, MRI is also used to evaluate for the presence of features that would preclude secondary cytoreduction, such as pelvic side wall invasion (suspected when the tumor lies within 3 mm of the pelvic side wall or when the iliac vessels are surrounded or distorted by tumor) and osseous invasion from the adjacent pelvic sidewall recurrence. This becomes extremely relevant given that the previously advocated second look surgery, which consisted of surgical re-exploration following initial surgery and completion of chemotherapy in patients without clinical evidence of recurrent tumor, is no longer part of standard clinical practice and imaging diagnosis of recurrence may obviate a second laparotomy because secondary cytoreduction is only justified if resection is possible with no residual tumor. MOVING FORWARD: MRI AS A PREDICTIVE AND PROGNOSTIC BIOMARKER IN OVARIAN CANCER The use of MRI as a predictive and prognostic biomarker requires a paradigm shift from an emphasis on anatomical and morphological features to functional imaging techniques that allow quantification of tumor function, metabolism and heterogeneity. It is well known that cancer response to therapy as measured by changes in tumor size often lags behind functional response in terms of changes in tumor microenvironment, apoptosis and necrosis. Furthermore, functional imaging offers important advantages over biochemical markers or tissue sampling for measuring biomarkers, as it has the potential to yield insights into in vivo tumor biology and the heterogeneity of both primary and metastatic lesions providing information about disease activity in real time (37). In addition, primary and metastatic sites can be assessed serially and noninvasively over the course of

15 MRI of Ovarian Masses 279 Figure 13. Tubo-ovarian abscess. a: Transabdominal US shows an enlarged ovary containing a complex fluid-filled adnexal mass (arrow) with low levels internal echoes consistent with debris and thickened wall. b: Axial CT in the same patient demonstrates a low attenuation left adnexal mass adherent to the uterus with a thickened irregular margin (arrow), with additional stranding in the surrounding fat. c,d: Axial T1 (a) and T2-weighted (d) MR imaging shows the mass to have low signal on T1 and high signal intensity on T2-weighted imaging, consistent with an abscess, the other described CT features are also demonstrated. the disease, without the need for repeated invasive procedures to obtain tissue samples necessary for in vitro assays. DCE-MRI provides an insight into tumor vascularity and angiogenesis, and is already being evaluated as a response assessment tool in multiple cancer sites including the kidney, bladder, rectum, prostate and cervix (38 41). DCE-MRI is also being used as a surrogate biomarker in early clinical trials of new antivascular drugs (42). Preliminary studies have highlighted the potential for this technique as a prognostic tool in ovarian cancer. Priest et al demonstrated the feasibility of DCE-MRI performed at 3T in multiple metastatic sites with improved temporal resolution (1.6 s) and extended spatial coverage (43). Another study showed that serological angiogenic markers VEGFR-1 and VEGFR-2 correlated negatively with fractional blood volume v e and positively with K trans from omental, peritoneal and nodal lesions in patients with low volume, asymptomatic relapsed disease, indicating a relationship between hypoxiainduced up-regulation of angiogenic receptors and increased vascular permeability (44). Semiquantitative analysis of enhancement curves derived from DCE- MRI showed that amplitude and maximal slope of enhancement achieved 100% sensitivity and 72% and 92% specificity in identifying histological invasion (10). There was a positive correlation between these parameters and vascular endothelial growth factor receptor-2 (VEGFR-2) expression but not with microvessel density, thus representing functional angiogenic properties, such as permeability, rather than structural hypervascularization. No significant changes in K trans were shown after induction chemotherapy, suggesting that the detection of antivascular effects of standard cytotoxic agents may require more rigorous acquisition and analysis techniques (45). It is also possible that targeted MRI imaging probes could be combined with blood-based markers of

16 280 Vargas et al. Figure 14. Ovarian cancer. A,B: Axial T2-weighted (a,c) and axial diffusion-weighted images (b,d) show bilateral solid lesion within the ovaries, with restricted diffusion (white arrows in a, b). Additionally an omental cake pattern is demonstrated, with anterior soft tissue peritoneal deposits which demonstrate restricted diffusion (open arrows in c, d). ovarian cancer for screening in high risk population (46). Macromolecular-Gadolinium based agents or iron based nanoparticles (as negative contrast agents) have the potential to incorporate specific targeting moieties on their surface. For example, folate receptors are over-expressed in 90 95% of epithelial ovarian cancers, and successful tumor-selective folate receptor targeting has been reported with a variety of radionuclides and optical probes (47,48). Targeting folic acid with MR-sensitive probes has also been achieved using compounds containing polymeric gadolinium chelates either as low molecular weight compounds, as polymers or dendrimers, incorporated into liposomes or microemulsions, conjugated to iron oxide nanoparticles or coupled to gadolinium-tetraazacyclododecane tetraacetic acid (Gd.DOTA) (48). Tumor-specific imaging may have the potential to improve preoperative and intraoperative staging, allow optimization of cytoreductive surgery, as well as facilitating post-treatment monitoring. These recent developments have resulted in increasing interest in functional MRI techniques as a biomarker of treatment response, however they also highlight the need for validation in rigorously designed prospective studies and for standardized, integrated multiparametric models that can maximize predictive and prognostic power of these techniques and bring us one step closer to the concept on individualized medicine. CONCLUSION MRI plays a key role in the characterization of ovarian masses. Ovarian lesions display a myriad of imaging findings which are dependent on the tissue type present and knowledge of such features can enable a definitive diagnosis in certain cases, or at least help to narrow the differential diagnosis. MRI additionally enables local staging of ovarian tumors, is complimentary to CT is detection of metastatic disease, helps determine the management plan for such lesions and is important in the detection of recurrent disease. Recent developments in the field of functional MRI show promise and may enable its use as a predictive and prognostic biomarker in patients with ovarian cancer. REFERENCES 1. Sala E. Magnetic resonance imaging of the female pelvis. Semin Roentgenol 2008;43: Tofts PS, Brix G, Buckley DL, et al. Estimating kinetic parameters from dynamic contrast-enhanced T-weighted MRI of a diffusable tracer: standardized quantities and symbols. J Magn Reson Imaging 1999;10: Spencer JA, Ghattamaneni S. MR imaging of the sonographically indeterminate adnexal mass. Radiology 2010;256: Outwater EK, Talerman A, Dunton C. Normal adnexa uteri specimens: anatomic basis of MR imaging features. Radiology 1996; 201: