MRI Features of Ovarian Fibroma and Fibrothecoma With Histopathologic Correlation

Women s Imaging Original Research Shinagare et al. MRI of Ovarian Fibroma and Fibrothecoma Women s Imaging Original Research Atul B. Shinagare 1 Liesbeth J. Meylaerts 1,2 Anna R. Laury 3 Koenraad J. Mortele 1 Shinagare AB, Meylaerts LJ, Laury AR, Mortele KJ Keywords: fibrothecoma, imaging features, MRI, ovarian fibroma DOI:10.2214/AJR.11.7221 Received May 16, 2011; accepted after revision August 5, 2011. 1 Division of Abdominal Imaging and Intervention, Department of Radiology, Brigham and Women s Hospital, Harvard Medical School, 75 Francis St, Boston, MA 02115. Address correspondence to A. B. Shinagare (ashinagare@partners.org). 2 Department of Radiology, Catholic University of Leuven, Leuven, Belgium. 3 Department of Pathology, Brigham and Women s Hospital, Harvard Medical School, Boston, MA. WEB This is a Web exclusive article. AJR 2012; 198:W296 W303 0361 803X/12/1983 W296 American Roentgen Ray Society MRI Features of Ovarian Fibroma and Fibrothecoma With Histopathologic Correlation OBJECTIVE. The purpose of this article is to evaluate MRI features of ovarian fibroma and fibrothecoma with histopathologic correlation. MATERIALS AND METHODS. In this retrospective study, preoperative MRI examinations of 35 women (mean age, 49 years; range, 24 86 years) with pathologically proven ovarian fibroma (n = 25) or fibrothecoma (n = 10) were reviewed by two radiologists in consensus. MRI features, including visibility of ovaries, presence of capsule, degeneration, T1 and T2 signal, and enhancement pattern, were recorded and correlated with histopathologic features. After administration of gadopentetate dimeglumine, the maximum percentages of enhancement of fibroma or fibrothecoma, myometrium, and, if present, uterine fibroids (11/35 patients) were compared. RESULTS. All fibromas and fibrothecomas appeared well defined, with a mean size of 6.36 4.81 cm. Ipsilateral and contralateral ovaries were each seen in 89% (31/35) of patients. Most fibromas and fibrothecomas were isointense to hypointense compared with myometrium on T1-weighted (91% [32/35]) and T2-weighted (77% [27/35]) images. Capsule was noted in 63% (22/35) and degenerative changes were noted in 66% (23/35) of patients. Fibromas and fibrothecomas larger than 6 cm more likely showed capsule (p < 0.0001, Fisher exact probability test), degenerative changes (p = 0.003), peripheral subcapsular cystic areas (p < 0.0001), heterogeneous T2 signal (p = 0.001), and heterogeneous enhancement (p = 0.005). At least four of the above five characteristics were present in 93% (14/15) of fibromas and fibrothecomas larger than 6 cm (p < 0.0001). The maximum percentage of enhancement for fibromas and fibrothecomas (63%) was significantly lower than those for myometrium (131%; p < 0.0001) and fibroids (103%; p < 0.0001), without a statistically significant difference between the maximum percentage enhancement of myometrium and fibroids. A maximum percentage of enhancement less than 75% yielded 92% positive predictive value in differentiating fibromas and fibrothecomas from fibroids. Fibrothecomas had a higher maximum percentage of enhancement than did fibromas (p = 0.01). CONCLUSION. MRI features of ovarian fibromas and fibrothecomas depend on size, with capsule and degenerative changes common with fibromas and fibrothecomas larger than 6 cm. Fibromas and fibrothecomas enhance less than myometrium and fibroids do, and less than 75% maximum percentage enhancement can help in differentiating fibromas and fibrothecomas from fibroids. F ibromas and fibrothecomas, which are benign ovarian stromal tumors, account for only 4% of all ovarian neoplasms but represent the most common solid primary ovarian tumors in asymptomatic women of all ages [1]. Pathologically, fibromas and fibrothecomas are benign tumors that belong to the sex cord stromal tumor category. Fibromas are composed of whorled fascicles of cytologically bland spindle cells embedded in a collagenous stroma [2]. Fibrothecomas differ slightly from fibromas; they have sheets and nests of plump spindle cells with lipid-rich cytoplasm (thecalike cells) in a background of bland fibromalike spindle cells. Moreover, they may have intermixed lutein cells and are frequently estrogenic, although up to 10% may have androgenic activity [2]. Ovarian fibromas and fibrothecomas may present with pleural effusions or ascites, known as the Meigs syndrome. Fibromas and fibrothecomas may also be associated with the basal cell nevus (Gorlin-Goltz) syndrome, which consists of bilat- W296 AJR:198, March 2012

MRI of Ovarian Fibroma and Fibrothecoma eral large multinodular ovarian fibromas, multiple basal cell carcinomas of the skin, odontogenic keratocysts, and other abnormalities [2]. The importance of correct diagnosis of fibromas and fibrothecomas lies in their benign nature. These are relatively common incidental solid ovarian tumors, and the ability to make a diagnosis of this benign tumor on imaging can greatly affect the patient management, especially in terms of avoiding unnecessary surgery, decreasing patient anxiety, and avoiding morbidity associated with invasive surgical procedures. It is also important to differentiate them from broad ligament fibroids (leiomyomata), because of differences in the management of these entities. Imaging is important to differentiate fibromas and fibrothecomas from fibroids, and from malignant ovarian tumors, especially when these present as a solid ovarian mass associated with ascites and pleural effusion. Several studies have previously reported imaging features of fibromas and fibrothecomas [1, 3 11]. However, many studies have evaluated ultrasound features of fibromas and fibrothecomas, which are often nonspecific [7], and when MRI features were studied, the studies either had small sample sizes or focused on specific features of fibromas and fibrothecomas. Even now, radiologists often face difficulty with accurate diagnosis of fibromas and fibrothecomas, especially with their differentiation from fibroids. Therefore, the purpose of this study was to evaluate MRI features of ovarian fibromas and fibrothecomas with histopathologic correlation, using state-of-the-art MRI on a relatively large number of patients. Materials and Methods Subjects In this institutional review board approved HIPAA-compliant retrospective study, using the surgical pathology and radiology information electronic databases, we identified 45 patients with surgically proven ovarian fibromas and fibrothecomas (35 fibromas and 10 fibrothecomas), who underwent preoperative pelvic MRI at our institution between August 2001 and January 2011. Ten patients were excluded either because the MRI examinations were not available for review (n = 4) or because fibromas and fibrothecomas were smaller than 1 cm (n = 6). The remaining 35 women (mean age, 49 years; range, 24 86 years; 25 fibromas and 10 fibrothecomas) were included in the study. The median time difference between pelvic MRI and surgery was 34 days (range, 1 1068 days). Imaging Technique MRI was performed on either a 3-T (Trio and Verio, Siemens Healthcare) or a 1.5-T (Echospeed, GE Healthcare) magnet with a pelvic phased-array multicoil. Twenty-nine patients were scanned on a 1.5-T magnet, and the remaining six were scanned using a 3-T magnet. The MRI protocol included T2-weighted fast spin-echo images in axial, coronal, and sagittal planes (TR/TE, 3417 6067/88 126.5); fat-suppressed T1-weighted spin-echo images in axial, coronal, and sagittal planes (TR/TE, 310 800/4.2 20); and multiphasic (30, 60, and 90 second) fat-suppressed T1-weighted images (TR/ TE, 260 450/1.6 6.4) after IV administration of gadopentetate dimeglumine (2 mmol/kg; Mag nevist, Berlex Laboratories) in all but two (33/35) patients. The slice thickness used for all acquisitions varied between 4 and 5 mm. Image Analysis The preoperative pelvic MRI examinations were reviewed by two radiologists in consensus. The location, size (largest dimension in two orthogonal planes), shape (round, oval, lobulated, or other) and outline (well or ill defined) of fibromas and fibrothecomas; visibility of bilateral ovaries; and presence of pelvic free fluid (graded as none, small, moderate, or large amount) were noted. T1 and T2 signals of fibromas and fibrothecomas were qualitatively evaluated. T1 signal was graded as homogeneous or heterogeneous, and T2 signal was graded as homogeneous, less than 25% cystic component, 26 50% cystic component, 51 75% cystic component, or more than 75% cystic component. The signal intensity value of fibromas and fibrothecomas was then quantitatively compared with uterine myometrium and iliopsoas muscle using the signal intensity value measured using a circular region of interest. Signal intensity values of fibromas and fibrothecomas, myometrium, and iliopsoas muscle were measured on the same image to maintain comparability. When the fibromas and fibrothecomas had heterogeneous signal intensity, the measurements were obtained on the predominantly solid component, in a relatively homogeneous portion. The presence or absence of uterine fibroids larger than 1 cm was noted; fibroids (> 1 cm) were present in 11 patients. The presence of capsule, degenerative changes (none, central, peripheral, or both central and peripheral), peripheral subcapsular cystic area (present or absent), T2 signal homogeneity (homogeneous or heterogeneous), and enhancement pattern (homogeneous or heterogeneous) were recorded and were correlated with histopathologic features. These imaging features were also correlated with size of the fibromas and fibrothecomas. Additionally, the presence or absence of well-defined cystic structures along the periphery of fibromas and fibrothecomas and ovarian cortical fibrosis, seen as a T2-hypointense rim around the ovaries, was recorded. The signal intensity values of fibromas and fibrothecomas, myometrium, and, whenever present, of fibroids (> 1 cm) were recorded on unenhanced T1-weighted images and at three time points after injection of IV gadopentetate dimeglumine. If multiple fibroids were present, the single largest, preferably subserosal, fibroid was selected for this purpose. Percentage enhancement values (ratio of the signal intensity value on an enhanced image to the signal intensity value at the same location on unenhanced T1-weighted image) and maximum percentage enhancement values (highest percentage enhancement value obtained for a particular lesion) of fibromas and fibrothecomas, myometrium, and uterine fibroids were calculated at each time point. Imaging characteristics, including presence of capsule, degenerative changes, peripheral subcapsular cystic area, T2 signal homogeneity, and enhancement pattern, including maximum percentage enhancement, of ovarian fibromas and fibrothecomas were compared. Statistical Analysis The presence of capsule, degenerative changes, homogeneity or heterogeneity of T2 signal, the presence of a peripheral subcapsular cystic area, and enhancement pattern were correlated with size of the lesion using the Mann-Whitney U test. Because larger fibromas and fibrothecomas (> 6 cm) have been reported to have higher prevalence of capsule [1], the presence or absence of these characteristics was compared in fibromas and fibrothecomas larger than 6 cm or 6 cm or smaller using the Fisher exact probability test. The percentage enhancement values of fibromas and fibrothecomas were compared with myometrium and fibroids, and those of fibroids were compared with myometrium using the Mann-Whitney U test. Various values of maximum percentage enhancement were used to determine a threshold that best differentiates between fibromas and fibrothecomas and fibroids. Imaging characteristics of ovarian fibromas and fibrothecomas were compared using the Fisher exact probability test. The maximum percentage of enhancement of fibromas and fibrothecomas was compared using the Mann- Whitney U test. Nonparametric tests were used for statistical analysis wherever feasible to avoid the influence of a few outlying values. Pathologic Examination and Analysis A single pathologist at our institution retrospectively reviewed the pathologic features of all fibromas and fibrothecomas. H and E stained slides of AJR:198, March 2012 W297

Shinagare et al. Fig. 1 52-year-old woman with left ovarian fibroma. A, T2-weighted MRI in axial plane shows left ovarian fibroma (straight arrow) that is hypointense compared with both uterine myometrium and iliopsoas muscle. Note multiple uterine fibroids (arrowheads) and minimal pelvic free fluid (curved arrow). B, T1-weighted MRI in axial plane again shows left ovarian fibroma (arrow) that is slightly hypointense compared with both myometrium and iliopsoas muscle. Uterine fibroids are not well visualized on this image. C, MRI in axial plane after administration of IV gadopentetate dimeglumine shows low-level homogeneous enhancement (maximum percentage enhancement, 40%) of fibroma (arrow), compared with more enhancing (maximum percentage enhancement, 118%) uterine fibroid (arrowhead). D, H and E stained slide of fibroma shows bland spindle cells arranged in fascicles and set within collagenous stroma. adnexal masses were retrieved from the departmental files in the Department of Pathology. The pathology report and H and E slides were reviewed for each case, and the following characteristics were recorded: size, laterality, cellularity, presence or absence of a capsule, presence of central or peripheral degeneration, and presence and characteristics of peripheral cysts. Additionally, when available, H and E slides were evaluated for the presence or absence of ovarian cortical fibrosis. Results Fibromas and fibrothecomas were located in the left ovary in 17 patients, the right ovary was involved in 16 patients, and in two women, the location of fibromas and fibrothecomas could not be identified on MRI. Their mean dimension was 6.36 4.81 cm, with size ranging from 1.2 to 13.2 cm. Fibromas and fibrothecomas appeared oval, lobulated, round, and bilobed in 18, 10, five, and two cases, respectively, and were well defined in all the cases, irrespective of size and C shape. Both ipsilateral and contralateral ovaries were each seen in 89% (31/35) of cases. Free fluid was present in the pelvis in 69% of cases (24/35; minimal in 21 and moderate in three cases), and no free fluid was seen in the remaining 31% (11/35) of cases. The majority of the fibromas and fi brothe co mas were isointense to hypointense compared with uterine myometrium on T1- weighted (91% [32/35]) and T2-weighted (77% [27/35]) images (Fig. 1). The imaging characteristics of fibromas and fibrothecomas are summarized in Table 1. Most fibromas and fibrothecomas had homogeneous T1 signal (63% [22/35]) and heterogeneous T2 signal (63% [22/35]) (Fig. 2). A thin T2-hypointense enhancing capsule was present in 63% (22/35) of cases (Figs. 2 and 3). Degenerative changes were seen in 66% (23/35) of cases (Figs. 2 and 4); both central and peripheral degeneration were present in 10 fibromas and fibrothecomas, peripheral degeneration was seen in seven fibromas and fibrothecomas, A and central degeneration was seen in six fibromas and fibrothecomas. Peripheral subcapsular high T2 signal nonenhancing cystic areas were present in 49% (17/35) of cases (Figs. 3 and 4). Of the 33 patients who received IV gadopentetate dimeglumine, 58% (19/33) showed homogeneous enhancement (Fig. 1), and 42% (14/33) showed heterogeneous enhancement (Fig. 4). We also noted peripheral small, usually subcentimeter, cysts in 29% (10/35) of cases (Fig. 5). In nine of 10 cases, fewer than five cysts were present, and more than five cysts were present in one case. The cysts were hemorrhagic in one patient. Superficial cortical fibrosis, seen as a continuous or discontinuous T1- and T2-hypointense band around the ovaries, was seen in 17% (6/35) of cases, and involved both ovaries in all but one case (Fig. 6). The imaging features of fibromas and fibrothecomas were found to vary significantly with size. Capsulated fibromas and fibrothecomas were larger than noncapsulated D B W298 AJR:198, March 2012

MRI of Ovarian Fibroma and Fibrothecoma fibromas and fibrothecomas (8.5 vs 2.7 cm; p = 0.0001). Similarly, degenerative changes (8.1 vs 3 cm; p = 0.0002), peripheral subcapsular cystic area (9.1 vs 3.8 cm; p = 0.0002), heterogeneous T2 signal (8.3 vs 3.1 cm; p = 0.0003), and heterogeneous enhancement TABLE 1: Imaging Characteristics of Fibromas and Fibrothecomas Imaging Finding Percentage (No./Total) T1 signal Homogeneous 63 (22/35) Heterogeneous 37 (13/35) T2 signal Homogeneous 37 (13/35) Heterogeneous 63 (22/35) Cystic change < 25% cystic component 34 (12/35) 26 50% cystic component 17 (6/35) 51 75% cystic component 6 (2/35) > 75% cystic component 6 (2/35) Comparison with myometrium, T1 signal Isointense 57 (20/35) Hypointense 34 (12/35) Hyperintense 9 (3/35) Comparison with myometrium, T2 signal Isointense 14 (5/35) Hypointense 63 (22/35) Hyperintense 23 (8/35) Comparison with iliopsoas muscle, T1 signal Isointense 31 (11/35) Hypointense 63 (22/35) Hyperintense 6 (2/35) Comparison with iliopsoas muscle, T2 signal Isointense 40 (14/35) Hypointense 14 (5/35) Hyperintense 46 (16/35) Capsule Present 63 (22/35) Absent 37 (13/35) Degenerative changes Present 66 (23/35) Absent 34 (12/35) Peripheral cystic areas Present 49 (17/35) Absent 51 (18/35) Enhancement pattern Homogeneous 58 (19/33) Heterogeneous 42 (14/33) Peripheral cysts 29 (10/35) Superficial cortical fibrosis 17 (6/35) (9.1 vs 4.3 cm; p = 0.0029) were common with larger fibromas and fibrothecomas. Conversely, fibromas and fibrothecomas larger than 6 cm were more likely to show the presence of capsule, degenerative changes, peripheral subcapsular cystic areas, heterogeneous T2 signal, and heterogeneous enhancement (Table 2). At least four of the above five characteristics were usually present in fibromas and fibrothecomas larger than 6 cm (93% [14/15]; mean, 4.4) in comparison with those 6 cm or smaller (20% [4/20]; mean, 1.6; p < 0.0001). The mean percentages of enhancement for fibromas and fibrothecomas, uterine myometrium, and fibroids at three time points are listed in Table 3. The mean percentage of enhancement for fibromas and fibrothecomas was significantly lower than those for myometrium and fibroids at all time points (Table 3). The average maximum percentage of enhancement for fibromas and fibrothecomas (63%) was also significantly lower than those for myometrium (131%; p < 0.0001) and fibroids (103%; p < 0.0001). There was no statistically significant difference between the percentage of enhancement of fibroids and myometrium (Table 3). Various values of maximum percentage of enhancement were used to determine a threshold that best differentiates between fibromas and fibrothecomas and fibroids (Table 4). We found that most (73% [24/33]) fibromas and fibrothecomas had maximum percentages of enhancement less than 75% and that most fibroids (82% [9/11]) had maximum percentages of enhancement greater than 75% (Fig. 1). A maximum percentage of enhancement cutoff of less than 75% yielded a 92% positive predictive value for diagnosis of fibromas and fibrothecomas (Table 4). There was no statistically significant difference between fibromas and fibrothecomas in terms of presence of capsule, degenerative changes, peripheral subcapsular cystic area, T2 signal homogeneity, and enhancement pattern. However, the maximum percentage of enhancement for fibrothecomas was higher than that for fibromas (90% vs 51%; p = 0.01). On histopathologic correlation, only two cellular fibromas were noted. Both were hyperintense to iliopsoas muscle on T2-weighted images. Compared with myometrium, one was hyperintense (Fig. 7) and the other was isointense on T2-weighted images. The apparent capsule seen on MRI was composed of compressed ovarian stroma or loose fibroconnective tissue, indicating that it repre- AJR:198, March 2012 W299

Shinagare et al. sents a pseudocapsule (Fig. 2 and 3). A true fibrous capsule was present in only one case. Degenerative changes noted on MRI consisted of ischemic or edematous changes (Fig. 4). The peripheral subcapsular cystic areas seen on MRI mainly consisted of edematous changes, and the peripheral cysts noted on MRI correlated with follicles or cortical inclusion cysts on pathologic examination. Discussion Ovarian fibromas and fibrothecomas are benign neoplasms that can mimic malignant ovarian tumors because they present as solid adnexal masses, sometimes associated with ascites and pleural effusions. Sonography is generally used as the first-line imaging technique for the evaluation of ovarian pathologic abnormalities. However, ultrasound features of fibromas and fibrothecomas are usually nonspecific [7], and MRI is often needed for further differentiation of ovarian fibromas and fibrothecomas from other solid ovarian masses, especially pedunculated or broad ligament leiomyomas. A finding of normal ovaries was initially thought to exclude ovarian fibromas and fibrothecomas [1], a report that has been contradicted by a more recent study in which the ipsilateral ovary was identified in 46% of cases [9]. However, in that study, only structures showing the presence of follicles were considered to be ovaries. In the present study, the ipsilateral ovary was seen in 89% of cases and could be identified as frequently as the contralateral ovary. Only a small amount of pelvic free fluid was seen in the majority of cases. MRI is an excellent imaging modality for the detection and characterization of ovarian fibromas and fibrothecomas. Troiano et al. [1] reported the MRI findings of ovarian fibromas and fibrothecomas in 11 patients. All the lesions, regardless of size, had homogeneous low signal intensity on T1-weighted images, and all but one had predominantly low signal intensity on T2-weighted images. Outwater et al. [6], in their report of five ovarian fibromas, also showed that tumor components representing fibrous tissue showed low signal intensity on T2-weighted images. Similarly, in our study, on T1-weighted images, the majority of fibromas and fibrothecomas were isointense to hypointense to both uterine myometrium and iliopsoas muscle. On T2-weighted images, the majority were isointense to hypointense compared with myometrium, and isointense to hyperintense compared with iliopsoas muscle. Fig. 2 47-year-old woman with right ovarian fibroma. A, T2-weighted MRI in coronal plane shows large heterogeneous pelvic mass (straight arrow) with predominantly low T2 signal and areas of higher signal in some regions. Note presence of thin T2-hypointense pseudocapsule (arrowheads). Pelvic free fluid (curved arrow) is also present. B, Gross specimen of resected fibroma shows predominantly solid mass. C, H and E stained slide shows pseudocapsule composed of loose fibroconnective tissue. In a report of 24 ovarian fibromas, Oh et al. [9] noted a T2-hypointense capsule in 67% of cases and the presence of degenerative changes in 33% of cases. We also noted the presence of capsule in 63% of cases but areas of degenerative changes in 66% of cases. The apparent capsule most commonly represented compressed ovarian stroma or loose fibroconnective tissue, indicating that it represented a pseudocapsule. True fibrous capsule is rare, seen in only one case A in our study. It has been previously reported that larger fibromas (> 6 cm) more often show degenerative changes and edematous changes [1]. However, there have been conflicting reports of the relationship between the size of fibromas and fibrothecomas and signal heterogeneity [11, 12]. In our opinion, the heterogeneous signal on T2-weighted images and heterogeneous enhancement pattern were related to the presence of edema or degenerative changes. In addition, peripheral TABLE 2: Comparison of Frequency of Imaging Features Between Fibromas and Fibrothecomas Larger Than 6 cm or 6 cm or Smaller Imaging Feature Size (cm) > 6 6 Capsule 100 (15/15) 35 (7/20) < 0.0001 Degenerative changes 93 (14/15) 45 (9/20) 0.003 Peripheral subcapsular cystic areas 87 (13/15) 20 (4/20) < 0.0001 Heterogeneous T2 signal 93 (14/15) 21 (4/19) 0.001 Heterogeneous enhancement b 71 (10/14) 21 (4/19) 0.005 Note Data are percentage (no. of fibromas and fibrothecomas/total). a Fisher exact probability test. b Two patients did not receive contrast agent. p a B C W300 AJR:198, March 2012

MRI of Ovarian Fibroma and Fibrothecoma Fig. 3 28-year-old woman with left ovarian fibrothecoma. A, T2-weighted MRI in axial plane shows large lobulated hypointense mass (short thick arrow) anterior to uterus (arrowhead). Note presence of T2-hypointense pseudocapsule (long thin arrows) and peripheral subcapsular cystic spaces (curved arrows). B, H and E stained slide shows pseudocapsule composed of compressed ovarian stroma. C, H and E stained slide of fibrothecoma at high power shows thecoma component, which is composed of nodules of plump round-to-spindled cells with abundant cytoplasm, admixed with fibromatous component, which consists of fascicles of bland spindle cells within collagenous stroma. TABLE 3: Mean Percentage of Enhancement and Statistical Significance of Difference Between Enhancement of Fibromas and Fibrothecomas, Uterine Myometrium, and Fibroids at Three Time Points After Injection of IV Contrast Agent (Gadopentetate Dimeglumine) Time Point After IV Administration of Contrast Agent (s) Mean Percentage of Enhancement Fibromas and Fibrothecomas subcapsular cystic areas were noted in 49% of fibromas and fibrothecomas; these pathologically represented edematous change. We found that the presence of pseudocapsule, degenerative changes, peripheral subcapsular cystic areas, heterogeneous T2 signal, and heterogeneous enhancement were more common with larger fibromas and fibrothecomas (mean size, > 8 cm), and at least four or more of these five imaging characteristics were present in 93% of fibromas and fi brothe co mas larger than 6 cm. There was no dif- Uterine Myometrium Fibroid TABLE 4: Performance of Various Cutoff Values for Maximum Percentage of Enhancement in Differentiation Between Fibromas or Fibrothecomas and Fibroids Fibromas and Fibrothecomas Versus Fibroids Fibromas and Fibrothecomas Versus Myometrium Maximum Percentage of Enhancement 65% 70% 75% 80% 85% Sensitivity 0.67 0.7 0.73 0.73 0.76 Specificity 0.82 0.82 0.82 0.73 0.64 Positive predictive value 0.92 0.92 0.92 0.89 0.86 Negative predictive value 0.45 0.47 0.5 0.47 0.47 p a Fibromas and Fibrothecomas Versus Fibroid Fibroid Versus Myometrium 30 37 100 71 < 0.0001 < 0.0001 0.0615 60 51 116 87 < 0.0001 < 0.0001 0.0512 90 63 131 103 < 0.0001 < 0.0001 0.1096 a Mann-Whitney U test. A B C AJR:198, March 2012 W301

Shinagare et al. ference between fibromas and fibrothecomas in terms of these imaging characteristics. About 30% of fibromas and fibrothecomas showed the presence of peripheral small (usually subcentimeter) cysts, which on pathologic examination, represented follicles or cortical inclusion cysts. This again correlates with the normal ovarian stroma stretched around the fibromas and fibrothecomas, sometimes seen on MRI as a pseudocapsule. Superficial cortical fibrosis was seen in 17% of cases and was usually bilateral. To the best of our knowledge, these features have not been described before. Despite these MRI features, it can still be difficult to distinguish fibromas and fi brothe co mas from fibroids, usually pedunculated subserosal or broad ligament fibroids [11, 13]. Dynamic contrast-enhanced MRI has been reported to aid in this distinction [11]. We found that enhancement of fibromas and fibrothecomas was significantly lower than that for myometrium and fibroids at all time points. The maximum percentage of enhancement for fibromas and fibrothecomas (63%) was also significantly lower than that for myometrium (131%) and fibroids (103%). On the other hand, there was no statistically significant difference between the enhancement of fibroids and myometrium (Table 3). Most (73% [24/33]) fibromas and fibrothecomas had a maximum percentage of enhancement greater than 75%, and most fibroids (82% [9/11]) had a maximum percentage of enhancement greater than 75%. A maximum Fig. 4 54-year-old woman with right ovarian fibrothecoma. A, T2-weighted MRI in coronal plane shows midline pelvic mass (short thick arrow) with peripheral hypointense component and central hyperintense, probably degenerative, portion (long thin arrow). Also note thin T2- hypointense pseudocapsule (arrowhead) and small subcapsular cystic area (curved arrow). B, MRI in axial plane after administration of IV gadopentetate dimeglumine shows heterogeneous enhancement of fibrothecoma, with peripheral mildly enhancing solid portion (arrow) and large central and smaller peripheral nonenhancing degenerative areas (arrowheads). C, H and E stained slide shows area of degeneration in fibrothecoma. A B C Fig. 5 65-year-old woman with left ovarian fibroma. T2-weighted MRI in axial plane shows midline pelvic mass (straight arrow) with heterogeneous T2 signal. Note thin T2-hypointense pseudocapsule (arrowhead) and small well-defined peripheral cystic lesion (curved arrow) that pathologically correlates with ovarian follicle (data not shown). Fig. 6 70-year-old woman with right ovarian fibroma. T2-weighted MRI in axial plane shows right ovarian fibroma (arrowhead). Note T2-hypointense bands around both ovaries (arrows) consistent with superficial cortical fibrosis. Fig. 7 50-year-old woman with right ovarian cellular fibroma. T2-weighted MRI in axial plane shows well-defined right ovarian mass (straight arrow) that is hyperintense compared with both uterine myometrium and iliopsoas muscle. Note thin T2- hypointense pseudocapsule (curved arrow). W302 AJR:198, March 2012

MRI of Ovarian Fibroma and Fibrothecoma percentage of enhancement cutoff of less than 75% yielded a 92% positive predictive value for diagnosis of fibromas and fibrothecomas (Table 4). However, the negative predictive value with this cutoff was only 50%, indicating that, although lesions showing less than 75% maximum enhancement are almost always fibromas and fibrothecomas, lesions enhancing more than 75% can be either fibromas and fibrothecomas or fibroids. Fibrothecomas showed higher enhancement than did fibromas, which may be related to their thecalike cell composition, as compared with fibromas, which consist predominantly of spindle cells embedded in collagenous stroma. The limitations of this study include its retrospective nature and small number of fibroids for comparison of enhancement pattern with fibromas and fibrothecomas. In our study, the number of patients with pedunculated subserosal fibroids was too small for meaningful comparison. Therefore, uterine fibroids larger than 1 cm were used for comparison of enhancement. In summary, most ovarian fibromas and fibrothecomas were isointense to hypointense to uterine myometrium on T1- and T2-weighted images. MRI features of fibromas and fibrothecomas depend on the size of the lesion. The presence of pseudocapsule, degenerative changes, peripheral subcapsular cystic areas, heterogeneous T2 signal, and heterogeneous enhancement are more common with larger fibromas and fibrothecomas, and at least four or more of these five features were present in 93% of fibromas and fibrothecomas larger than 6 cm. Peripheral small cysts can be seen, indicating ovarian stroma stretched around the fibromas and fibrothecomas, and may definitively indicate ovarian origin of the lesion, thus helping in differentiation from fibroids. Fibromas and fibrothecomas enhance significantly less than uterine myometrium and fibroids do, and maximum enhancement of less than 75% yields a 92% positive predictive value for diagnosis of fibromas and fi brothe co mas, which can also help in differentiating fibromas and fibrothecomas from fibroids. Correct diagnosis of these tumors on imaging can decrease patient anxiety and avoid unnecessary invasive procedures. References 1. Troiano RN, Lazzarini KM, Scoutt LM, Lange RC, Flynn SD, McCarthy S. Fibroma and fi brothe co ma of the ovary: MR imaging findings. Radiology 1997; 204:795 798 2. Scully RE, Youmg RH, Clement PB. Tumors of the ovary, maldeveloped gonads, fallopian tube, and broad ligament. Washington, DC: AFIP, 1996:189 201 3. Bazot M, Ghossain MA, Buy JN, et al. Fibrothecomas of the ovary: CT and US findings. J Comput Assist Tomogr 1993; 17:754 759 4. Schwartz RK, Levine D, Hatabu H, Edelman RR. Ovarian fibroma: findings by contrast-enhanced MRI. Abdom Imaging 1997; 22:535 537 5. Outwater EK, Wagner BJ, Mannion C, McLarney JK, Kim B. Sex cord-stromal and steroid cell tumors of the ovary. RadioGraphics 1998; 18:1523 1546 6. Outwater EK, Siegelman ES, Talerman A, Dunton C. Ovarian fibromas and cystadenofibromas: MRI features of the fibrous component. J Magn Reson Imaging 1997; 7:465 471 7. Athey PA, Malone RS. Sonography of ovarian fibromas/thecomas. J Ultrasound Med 1987; 6:431 436 8. Kitajima K, Kaji Y, Sugimura K. Usual and unusual MRI findings of ovarian fibroma: correlation with pathologic findings. Magn Reson Med Sci 2008; 7:43 48 9. Oh SN, Rha SE, Byun JY, et al. MRI features of ovarian fibromas: emphasis on their relationship to the ovary. Clin Radiol 2008; 63:529 535 10. Paladini D, Testa A, Van Holsbeke C, Mancari R, Timmerman D, Valentin L. Imaging in gynecological disease (5): clinical and ultrasound characteristics in fibroma and fibrothecoma of the ovary. Ultrasound Obstet Gynecol 2009; 34:188 195 11. Thomassin-Naggara I, Daraï E, Nassar-Slaba J, Cortez A, Marsault C, Bazot M. Value of dynamic enhanced magnetic resonance imaging for distinguishing between ovarian fibroma and subserous uterine leiomyoma. J Comput Assist Tomogr 2007; 31:236 242 12. Siegelman ES, Outwater EK. Tissue characterization in the female pelvis by means of MR imaging. Radiology 1999; 212:5 18 13. Weinreb JC, Barkoff ND, Megibow A, Demopoulos R. The value of MR imaging in distinguishing leiomyomas from other solid pelvic masses when sonography is indeterminate. AJR 1990; 154:295 299 AJR:198, March 2012 W303