MRI diagnosis of ovarian fibrothecomas: tumour appearances and oestrogenic effect features

Abstract

Objective:

To study MRI diagnosis of ovarian fibrothecomas.

Methods:

MRI appearances of 27 ovarian fibrothecomas 26 female patients confirmed by surgical pathology were retrospectively reviewed and correlated with clinical and histological findings.

Results:

Most patients were post-menopausal females 73.1% (19/26) of cases. 38.5% (10/26) of patients showed an elevated serum carbohydrate antigen 125 (CA-125) level (>35 U ml⁻¹). On MR T₂ weighted imaging, 3 distinct features were revealed. (1) Homogeneous hypointense masses in 25.9% (7/27) cases. (2) Heterogeneous tumours with mostly isointense and few patchy hyperintense areas in 51.9% (14/27) cases. (3) Heterogeneous tumours with predominantly hyperintense and few isointense parts in 22.2% (6/27) cases. On T₁ weighted imaging, all the fibrothecomas turned out to be homogeneous masses except six cystic degeneration cases. After contrast, 70.4% (19/27) tumours showed homogeneous mild enhancement in all phases. Beyond the tumours, endometrial abnormality and uterus enlargement were found in 38.5% (10/26) and 15.4% (4/26) patients, respectively. Ascites were seen in 53.8% (14/26) patients. There was a statistically significant difference among the 3 T₂ weighted image features (F = 7.024; p = 0.003) in terms of the size of fibrothecomas. The cystic tumours also had the tendency to show the ascite condition (p = 0.002) as well as elevated serum CA-125 levels (p = 0.014). Vimentin was positively stained in 10 (76.9%) of 13 cases who underwent the immunohistochemical analysis.

Conclusion:

MRI has the superiority to show the distinct appearances of tumours as well as their functional features according to oestrogenic effect.

Advances in knowledge:

This study describes the unique features of fibrothecomas on MRI on a relatively large series of patients with tumours and the indirect oestrogenic effect findings.

Fibrothecomas are rare benign ovarian tumours extending from solid fibromas to lipid-rich thecomas, making up approximately 4% of all ovarian neoplasms, while constituting the most common sex cord-stromal tumours of the ovary.¹ These tumours also account for the most common hormonally active ovarian neoplasms, which often show oestrogenic effects, and could present as uterine morphological changes and thickened endometrium, particularly in females of post-menopausal age.^2,3 Clinically, ovarian fibrothecomas are generally latent masses and might be detected occasionally during routine gynaecologic examinations. At gross inspection, fibrothecomas typically present as solid tumours with or without cystic degeneration. In pathology, fibrothecomas contain theca cells with abundant lipids in the cytoplasm and varying degrees of fibrous content, the theca cells may be responsible for the oestrogenic effects of the tumours. The intraperitoneal ascites or pleural effusions are often associated with fibrothecomas and might make them difficult to differentiate from malignant epithelial ovarian tumours. Correct diagnosis of these benign tumours can greatly affect the patient's management, especially avoiding unnecessary invasive surgical procedures. So far, most of the previous studies^4,5 have reported non-specific ultrasound and CT features of these tumours, while MR studies^6,7 had only small sample sizes. The purpose of our study was to describe the unique features of fibrothecomas on MRI in a relatively large series of patients with tumours as well as to report indirect oestrogenic effect findings, and to discuss the points of differential diagnosis.

METHODS AND MATERIALS

Patient population

There were 26 female patients with surgically and histopathologically proven ovarian fibrothecomas constituting this study cohort. Ages of the patients ranged from 24 to 80 years (median age, 54 years). 73.1% (19/26) patients were post-menopausal females. After MR examination, all the patients subsequently underwent surgical procedures (total abdominal hysterectomy and bilateral salpingo-oophorectomy in 16 patients, and unilateral oophorectomy in the other 10 patients). The interval time between MRI and surgery was less than 2 weeks. In the laboratory, serum carbohydrate antigen 125 (CA-125) was routinely tested. The institutional ethics board of our hospital approved this retrospective study and waived the requirement for informed consent.

Imaging protocol

Of these 26 patients, 16 were scanned using the 1.5-T magnet (Signa TwinSpeed Exite; GE Medical Systems, Millwaukee, WI), and the remaining 10 were scanned using the 3-T magnet (Signa HDx; GE Medical Systems). All MR examinations were performed according to the same imaging protocol as follows: after the three plane localization and the array spatial sensitivity encoding technique parallel imaging, sagittal T₂ weighted fast-spin echo images without fat suppression were obtained from the abdominal aortic bifurcation to the bottom of the buttocks [repetition time (TR)/echo time (TE) range, 3000/90 ms; echo-train length, 17; slice thickness, 5 mm; interslice gap, 1 mm; field of view, 22 cm; number of acquisitions, 2; matrix, 384 × 160]. Axial T₁ weighted spin echo images were obtained from the aortic bifurcation to the femoral neck (TR/TE range, 540/9.1 ms; slice thickness, 5–6 mm; interslice gap, 1–1.5 mm; field of view, 22 cm; number of acquisitions, 2; matrix, 320 × 224). Transverse T₂ weighted fast-spin echo images with fat suppression were obtained from the abdominal aortic bifurcation to the femoral neck (TR/TE range, 4000–5000/130–150 ms; echo-train length, 15–17; slice thickness, 5–6 mm; interslice gap, 1.5–2.5 mm; field of view, 33–39 cm; number of acquisitions, 2; matrix, 320 × 160). Axial and sagittal (or coronal) T₁ images with fat suppression were obtained immediately after a bolus infusion of 0.1 mmol kg⁻¹ of gadolinium-diethylene triamine pentaacetic acid (Gd-DTPA, Magnevist; Bayer Schering Pharma AG, Berlin, Germany) from the renal hila to the femoral neck (TR/TE range, 135/2.4 ms; slice thickness, 5–6 mm; interslice gap, 1–1.5 mm; field of view, 24 cm; number of acquisitions, 2; matrix, 256 × 128). The dual-phase contrast-enhanced dynamic imaging was performed by obtaining early phase images (30–60 s after starting the injection), followed by delayed phase images (120–180 s after starting the injection). All patients were imaged using a phased-array eight channel body coil for the pelvis.

Image analyses

All MR scans were reviewed by two experienced radiologists (BW and WJP), with knowledge of only the histological diagnosis. Imaging appearances were assessed for tumour location, morphology (round, oval, lobulated or irregular), size (the largest dimension in two orthogonal planes), signal intensity characteristics (emphasis on T₂ weighted imaging) and enhancement status, as well as the functional oestrogenic effect signs beyond the tumour, such as endometrial conditions, uterine morphological changes and intraperitoneal ascites. Visibility of contralateral ovary was also noted. The signal intensity value of fibrothecomas was qualitatively ranked on both T₂ and T₁ weighted images as low, intermediate or high compared with the adjacent skeletal muscle. All signal intensity values were measured on the same image to maintain comparability. The presence of heterogeneity propensity of the lesions was assessed on T₂ weighted imaging (homogeneous or heterogeneous). Cystic degeneration was identified if signal intensity of well-defined collections that were equal to the urine and never get enhancement, and were confirmed by surgery and pathology afterwards. The degree of enhancement of the tumour parenchyma was also qualitatively assessed as mild (less than), moderate (equal) and avid (greater than) compared with the uterine myometrium in each enhancement phase. Uterine morphological changes and endometrial conditions were evaluated for patient age, for example, the normal uterus for a post-menopausal female was considered a proper size (<6 cm in length with distinct junction zonal anatomy), endometrium was <5 mm thick and myometrium had lower signal intensity on T₂ weighted imaging; otherwise, it would be considered as abnormal. Ascites were identified on T₂ weighted images as high signal intensity peritoneal collections that were equal to the urine and never get enhancement, and were also confirmed during the surgical procedure. All the imaging findings were agreed by the two radiologists in consensus.

Pathological examination

Slides were retrospectively reviewed by a single professional pathologist (YFC) for each case, with an emphasis on the tumour cellularity, fibrous component and cystic degeneration. The histopathological analysis included both routine haematoxylin and eosin (HE) staining and immunohistochemical evaluation. For the latter, antibodies to vimentin, CD34, α-inhibin, smooth muscle actin, calretinin and S-100 protein were used. The diagnosis was confirmed both histologically and immunohistochemically.

Statistical analyses

The T₂ weighted images were correlated with the size of the lesion, serum CA-125 elevated conditions, ascites, endometrial status and uterine morphological changes, with the use of the Variance test and Fisher's exact χ² test. A p-value <0.05 was considered statistically significant. Statistical analyses were performed using SPSS® software v. 15.0 (SPSS Inc., Chicago, IL).

RESULTS

Clinical data

27 ovarian lesions were identified in 26 patients after the surgery, including 1 bilateral case. The frequencies of the individual patient's symptoms at presentation were as follows: routine screen (9/26, 34.6%), post-menopausal bleeding (7/26, 26.9%), acute abdominal pain (4/26, 15.4%), abdominal distention (2/26, 7.7%), irregular menstruation (2/26, 7.7%) and self-found pelvic mass (2/26, 7.7%). In fact, 77.8% (21/27) pelvic masses could typically be detected at palpation during clinical gynaecologic examination. Overall, 38.5% patients showed elevated serum CA-125 levels (>35 U ml⁻¹) in the laboratory test. 52.6% (14/26) patients had 100-400 ml of free intraperitoneal fluid confirmed in surgery. Ovarian torsion was found in 4 patients (4/26, 15.4%) complaining of acute abdominal pain. Among the 27 lesions, 77.8% were predominantly solid masses by pathology (21/27). For those 16 patients who had undergone bilateral salpingo-oophorectomy, normal contralateral ovaries were found in 11 cases (11/16, 68.8%), as well as 3 ovarian cysts, 1 serous cystic adenoma and 1 bilateral fibrothecoma.

Imaging manifestations with histopathological correlation

16 lesions were located in the left ovaries, and 11 in the right side. All lesions could be shown clearly with sharp demarcated contours and well-defined margins on MR images, 12 lesions (44.4%) manifested as ovid shape, 7 (25.9%) as lobulated shape, 5 (18.5%) as round shape and 3 as irregular shape. The median of the maximum diameters was 9.5 cm (range, 3.5–24 cm). On T₂ weighted images, three distinct features were revealed referring to the tumour component: (1) 25.9% (7/27) fibrothecomas with abundant fibrous tissue demonstrated homogeneous lower signal intensity than those of the adjacent skeletal muscle (Figure 1a). (2) 51.9% (14/27) lesions showed primarily isointensity with few patchy areas of slight hyperintensity consistent with the components of oedema or degeneration (Figure 2a); cellular tumour and less fibrous or collagen tissue could be found under the microscope. (3) Fibrothecomas with predominately cystic changes and litter solid components showed as obviously heterogeneous hyperintense in 6 large tumours (6/27, 22.2%) (Figures 3a and 6a). Regardless of the inconsistent appearances of T₂ weighted images, all the solid fibrothecomas turned out to show homogeneous isointense masses on T₁ weighted images (Figures 1b and 2b), certainly, hypointensity could be seen in the six cystic cases (Figure 3b). After intravenous gadopentetate dimeglumine administration, the enhancement performances were not very diverse. 70.4% (19/27) tumours showed almost homogeneous enhancement (Figures 1c and 2c), and the rest showed heterogeneous enhancement (Figure 3c), mild enhancement could be shown in almost all cases whether in early or delayed phase (Figure 6b) except for a young patient (Figure 3c). Because of the prevalence age, the majority of contralateral ovaries could actually not be observed on MRI. By contrast, for those 11 cases (11/26, 42.3%) whose contralateral ovaries could be seen, negative appearances (antral follicles present in the cortex) were found in 7 cases and various cystic structures could be found in 4 patients, including 3 ovarian cysts and 1 serous cystic adenoma, confirmed by surgery.

Figure 1.

MRI features of fibrothecoma with abundant fibrous tissue. Homogeneous characteristics in both T₂ weighted (a) and post-contrast T₁ weighted (b) images, and the tumour shows mild and homogeneous enhancement. Grossly sectioned surface shows rubbery hard solid components with bright yellow colour (c), haematoxylin and eosin staining (d) identifies rich spindle fibroblast cells (red arrow) arranged in fascicles within collagenous stroma (black arrow). For colour images please see online www.birpublications.org/doi/full/10.1259/bjr.20130634.

Figure 2.

MRI features of celluar fibrothecoma with less fibrous content and degeneration. Primarily isointense with varing degrees of patchy areas of slight hyperintensity can be seen in T₂ weighted images with fat-suppressed (a), still, almost homogeneous phenotype implying the characteristics of solid mass in T₁ weighted images before (b) and after (c) the administration of contrast agents, and mild enhancement was also detected. Haematoxylin and eosin staining consistent with the components of rich thecoma cell with lighter stained abundant cytoplasm (d).

Figure 3.

Fibrothecoma with cystic changes showed apparent heterogenous characteristics with diffuse high signal intensity (star) on T₂ weighted imaging (a) and low signal intensity in T₁ weighted imaging (b). After the administration of contrast agent, the tumour parenchyma can appear obviously enhanced (c). Abundant cystic area (star) could be found under the microscope (d).

Figure 6.

Ovarian torsion induced by fibrothecoma shows huge pelvic mass with distinct hetergenous propensity, and predominantly cystic changes. (a) Sagittal T₂ weighted image. (b, c) Sagittal and coronal post-contrast T₁ weighted images.

Abnormal uterus enlargement was only found in 4 patients of post-menopausal age (Figure 4b), 38.5% (10/26) patients presented thickened endometria that could be typically shown in sagittal T₂ weighted images (Figure 4d,e); among these, 70% (7/10) patients presented with vaginal bleeding symptoms. Ascites were found as high-signal intensity in the peritoneum or pleural cavity on T₂ weighted imaging in 15 patients (15/26, 57.7%), 3 pleural effusions were also noted on chest radiographs, which might have indicated Meigs' syndrome (Figure 4c,f). There was a statistically significant difference among the three features on T₂ weighted images (F = 7.024; p = 0.003) in terms of the size of fibrothecomas (Figure 5), namely, fibrothecomas showing hyperintensity with predominately cystic changes were significantly larger than those of the hypointense cases (13.29 ± 5.76 cm vs 5.37 ± 2.65 cm, respectively; p = 0.002) and the isointense cases (13.29 ± 5.76 cm vs 9.23 ± 3.86 cm, respectively; p = 0.04). This cystic pattern was also more likely to be associated with ascite conditions (6/6, 100%; p = 0.002), and elevated serum CA-125 level (5/6, 83.3%; p = 0.014). Among the three features on T₂ weighted images, although the statistic results were negative in difference, isointense tumours had the tendency for endometrial thickened statuses (6/13, 46.2%) (Table 1).

Figure 4.

Indirect MRI appearances beyond the fibrothecoma. (a) Normal uterus in post-menopausal female shows that the uterus is small and the cervix (white arrows) is longer than the corpus. (b) Abnormal uterus changes with enlargement of uterus and thickened endometrium in cervix (arrow), which indicates (d) endometrial hyperplasia or (e) carcinoma (star), can be typically seen in sagittal T₂ weighted images or post-contrast images, conveyed the consequences of excess of oestrogen and might present with vaginal bleeding clinical symptoms. (c) Intraperitoneal ascites or (f) pleural effusion can be indicated with high signal intense in fat-suppressed T₂ weighted imaging or the CT scan.

Figure 5.

Box-and-whisker plots show T₂ weighted imaging T2WI features of varying tumour sizes. Fibrothecomas showing hyperintensity with predominantly cystic changes are significantly larger than the other hypointense and the isointense cases. Boxes represent the interquartile range, whiskers indicate range of all values and the horizontal line within the boxes are the median.

Table 1.

Comparison of T₂ weighted tumour features with ascites, carbohydrate antigen 125 (CA-125) status and functional uterine changes

T2 weighted feature	Ascites		CA-125		Endometrium abnormal		Uterus enlargement
	+	−	+	−	+	−	+	−
Hypointense	1	6	0	7	2	5	2	5
Isointense	8	5	6	7	6	7	1	12
Hyperintense	6	0	5	1	2	4	2	4
p-value	0.002		0.014		0.771		0.533

+, positive; −, negative.

Immuno-histochemistry analysis

Most of the fibrothecomas could be easily diagnosed using HE stain, only nearly half of the tumours (13/27, 48%) underwent immuno-histochemistry study. Among these, vimentin was strongly positively stained in 10 cases (10/13, 76.9%) and coincides with 1 report,⁸ followed by α-inhibin (4/13, 30.8%) and calretinin (3/13, 23.1%), and negative staining could be found in smooth muscle actin and S-100 proteins.

DISCUSSION

The fibromas, fibrothecomas and thecomas form a spectrum of the most common sex cord-stromal tumours of the ovary, usually occurring in elderly post-menopausal females, but may affect patient's of any age. Because of their histological overlaps, most tumours actually appear as an admixture of fibroma and thecoma components with varied proportions, so-called fibrothecomas.³ Clinically, fibrothecomas used to be latent tumours and presented as palpable pelvic masses during routine examination or occasionally were perceived by the patients themselves; however, as oestrogen-producing tumours, manifestations like vaginal bleeding can be seen particularly in post-menopausal females (26.9% in our cases).⁴ On gross inspection, ovarian fibrothecomas often present as ovid or round solid masses with smooth margins. The proportion of solid tumours (77.8%) in our cases was almost similar to the reported detection in literature (79%).⁵ However, cystic changes are not uncommonly seen, especially when the ovarian torsion occurs.

Ultrasonography is generally used as the first-line modality to image this ovarian entity; however, the ultrasound features of fibrothecomas are primarily hypoechoic ovarian masses with posterior wall attenuation, which are similar to other pelvic masses.^5,9 Fibrothecomas typically appear as unilateral, well-defined ovarian tumours at CT scans, with homogeneous or slightly heterogeneous mild-contrast enhancements, thus the diagnosis of benign ovary lesions follows as matter of course. In other words, lack of differentiation from other benign ovarian tumours accounts for the poor characterization on CT imaging. MRI has been proven to be the preferred technique to depict the characteristic features of the neoplasm, for the reason of its exquisite soft-tissue resolution, T₂ weighted imaging is thought to be the most appropriate pulse sequence to give diagnoses. In previous studies,^6,7,10,11 the appearances of fibrothecomas were described as “unique” because they demonstrated low signal intensity on T₂ weighted MRI, but these unique signs might just be limited to the ovarian fibromas with abundant fibrous tissue. Owing to the complex and overlapping composition of the tumours, there are three different appearances that were actually found on T₂ weighted MRI in our study, this was thought to somewhat reveal the distinct tumour characters according to the pathological results. In fact, on T₂ weighted imaging, low signal intensity appearances are not the most common, whereas more than half the lesions were cellular tumours that were showing primarily as isointense with few patchy areas of slightly hyperintensity, with less fibrous or collagen tissue found under the microscope. Mild enhancement of the tumour was seen in almost all the tumours of our series whether in early or delayed phase, this also confirmed the lack of rich blood supply in fibrothecomas. Currently, controversy still remains about the origin of ovarian fibromas or thecomas,¹² these various MRI appearances seem to support this view to some degree and further studies using functional imaging like diffusion-weighted MRI might be needed.

Fibrothecomas are the most common functional ovarian neoplasm that can produce oestrogen. They mostly present as enlarged uteruses with or without thickened endometria for the patient's age. The theca components might be considered to be responsible for the oestrogenic activity.^2–4 MRI is advantageous to reveal these similar features of hyperoestrogenism manifestation especially in post-menopausal patients.³ In our study population, ten thickened endometria and five enlargement uteruses could be seen on MRI. Although there is no statistically significant difference, cellular fibrothecomas with isointense appearances on T₂ weighted imaging tended to have the highest proportion of thickened endometria (6/13, 46.2%), namely, the MR characteristic of the tumour can be somewhat used to determine the composition of the associated hyperestrogenism. However, because hysterectomy was only implemented in 16 patients in our cases it might not be accurate to evaluate the endometrial status and uterus enlargement condition just relying on MRI, these results need more cases to verify them.

Ascites are very popular in ovarian tumours whether benign or malignant. 13 cases in our series show ascites, and there is an apparent tendency of the fibrothecomas to appear with cystic degeneration (p = 0.026). The pathogenesis of ascites with tumours has not been known, vasoendothelial and fibroblast growth factors and cytokines may play a role in fluid accumulations. The pressure of the tumour on lymph vessels also may causes fluid.¹³

Occasionally, fibrothecomas might cause ovarian torsion, which presents as acute abdominal pain. The larger the tumour, the greater the chance of torsion. The four torsion lesions in our series were relatively large, and the largest lesion among all the tumours was 24 cm in diameter. At T₂ weighted images, all the torsion tumours show heterogeneous predominantly cystic degeneration probably related to internal oedema or necrosis. The absence of gadolinium enhancement might be the most reliable sign to indicate arterial compromise or infarction;¹⁴ still, there are no explicit signs to predict the condition of tumour torsion except the large size.

CA-125 is a non-specific tumour marker usually associated with ovarian malignant lesions, elevated CA-125 levels are not uncommon in our series of fibrothecomas (38.5%). Among them, the predominantly cystic lesions had the tendency of elevated CA-125 levels (83.3%; 5/6; p = 0.014), although the reason is not clear.

The main differential diagnosis is other predominantly solid pelvic masses, including Brenner tumour, granulose cell tumour and dysgerminoma, as well as non-degenerated subserous pedunculated uterine leiomyoma or broad ligament leiomyoma. (1) The Brenner tumour and dysgerminoma are also solid ovarian tumours, which can show low T₂ weighted signal intensity, but both are exceedingly rare. Granulosa cell tumour is also the common oestrogenic ovarian tumour-associated uterine and endometrial abnormality that may be demonstrated, in this sense, extensive intratumoral haemorrhage and avid enhancement might help to identify the difference.^2,15 (2) Pedunculated subserosal or broad ligament leiomyoma are well circumscribed pelvic masses that might show low signal intensity on T₂ weighted images, and also can produce variable signal intensity changes according to the degeneration. MRI is superior in detecting the relationship between the tumour and adjacent organs,^7,11 and normal ovary depiction might exclude the diagnosis of ipsilateral ovarian mass; however, it is not easy to find the normal ovary on MRI at a post-menopausal age,⁵ yet, most leiomyomas show early avid enhancement, whereas this is barely seen in fibrothecomas. Moreover, some authors mentioned that diffusion-weighted MRI might be useful in discrimination.¹⁶ (3) Occasionally, cystic ovarian fibrothecomas could be easily misdiagnosed as the ovarian cystadenocarcinoma if accompanied with severe ascties and significantly elevated CA-125 levels; however, the well circumscribed contour and mild enhancement might be the essential points to distinguish, whereas the cystadenocarcinoma tends to be a more irregular ovarian mass and might present with evident metastatic signs.

In conclusion, T₂ weighted MRI has the superiority to show the ovarian fibrothecoma characteristics as well as other oestrogenic functional appearances. Main differential diagnosis should include the other solid ovarian masses and subserous uterine or broad ligament leiomyomas. The affinity of fibrothecomas in vimentin staining might help in diagnosis of the immunohistochemical aspects.