Adnexal masses characterized on 3 tesla magnetic resonance imaging – added value of diffusion techniques

Incidental adnexal masses are commonly detected in daily medical practice due to the frequent lack of clinical manifestation.1 Approximately 9% to 10% of women undergoing ultrasound have ovarian lesions.2 Although most commonly used, ultrasound has some limitations including the small field of view, low resolution and interference by obesity or by gaseous bowel loops.3 Ultrasound indeterminate adnexal masses vary between 5% and 25%.4, 5 If furthermore examined with computed tomography (CT), distant metastases, respectively the staging of the disease could be assessed. Magnetic resonance (MR) has been considered as the most useful imaging technique for characterizing adnexal formations. This modality has a key role in the preoperative evaluation and their follow-up, identifying the origin of the mass and the different types of tissue contained in with accuracy of 88% to 93%.6 3 Tesla MRI is superior for examining female pelvis due to its higher resolution and the possibility of providing more detailed images.7, 8, 9 MRI techniques such as diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) are of an additional benefit differentiating malignant from benign lesions.10, 11, 12,

The aim of our study is to assess different types of adnexal masses as identified by 3T MRI and to discuss the added value of diffusion and perfusion techniques compared with conventional sequences.

174 women age between 13 and 87 underwent MRI examination of the pelvis for a period of three years. Indications were: sonographically detected pelvic mass; or gynecological complaints; or history of previous adnexal tumor; or family history of ovarian cancer. Six women were examined for other reasons (hips, sigma/colon or perianal abscess), nine for uterine pathology, but adnexal mass was detected and the complete gynecological MRI protocol was performed, too. At least one adnexal mass was diagnosed in 98 patients and they are subject of this study. 51% of them were reviewed retrospectively. Data from patients’ history, physical examination and laboratory tests were reviewed as well.

Patients were examined in two radiology departments, 135 of them on 3 Tesla MRI Siemens Verio and the 39 on 3 Tesla MRI Philips Ingenia. The Siemens MRI protocol included: coronal (COR) T1; sagittal (SAG) T2; paracoronal and paratransversal of the uterus T2 with and without fat saturation; SAG T1; transversal T1 Vibe Dixon; DWI and ADC. The Philips MRI protocol included: COR STIR; SAG T2; COR T2; COR T1; axial (AX) T2; AX T2 with fat saturation; DWI and ADC. (Table 1)

Measurement of the ADC value was carried out for all ovarian masses in our study. For each tumor a region of interest (ROI: 1 cm²) was manually defined. In the mixed malignant formations ROIs were placed on the solid component only. The ADC values are presented as numerical value x 10^-3 mm²/s representing quantitative metric.

Intravenous contrast administration was applied when needed and when there were no contraindications. A macrocyclic contrast agent Gadobutrol [1.0 mmol/ml] (Gadovist® 1.0, Bayerhealthcare, Berlin, Germany) was used at a dose of 0.1 mmol/ kg in all contrast-enhanced studies on both MR devices. Injection rate of 0.5 mL/sec was performed in order to achieve equimolar amounts of gadolinium. Saline flush (25–30 ml) at the same flow rate followed the contrast administration.

In part of our cases dynamic contrast enhanced – magnetic resonance imaging (DCE-MRI) was performed and time-signal intensity curve (TIC) was generated using the Mean Curve software package (Philips). A round region of interest (ROI: 1 cm²) was placed at target areas referring to T2W and contrast-enhanced images. Areas with hemorrhage and necrosis were avoided.

The following patterns were evaluated on MR images:

Following MR criteria of malignancy, as reported in the literature (by Jeong et al.13, Valentini et al.6 and El-Wekil et al.14), are used:

SPSS Statistics release 21 for Microsoft Windows was used to perform Kolmogorov-Smirnov (2-tailed) test for establishing correlations between malignancy and diffusion restriction and between malignancy and type of mass.

Approval was obtained from the Institutional Review Board of both University hospitals prior the initiation of the study. Informed written consent was obtained from each patient. Personal identity information of all patients was protected.

In 98 females of ave rage age 47.2 years, a total of 124 ovarian masses were detected. In 16 of the patients (16.3%) additional uterine pathology was found. One case considered as an ovarian cyst was histologically proven to be an inclusion peritoneal cyst. The results of all ovarian masses according to their MRI features are listed in Table 2.

Following the MRI criteria, 59.2% of the cases were considered benign, 30.6% malignant and 10.2% borderline. The results of DWI sequences show a statistically significant correlation with the assessment of masses as benign/borderline/malignant. 34.3% of all malignant cases were found in the age group 61–70. Of all patients 32 were tested for CA-125 tumor marker and 12 had elevated levels. Only half of those 12 cases were histologically proven malignant.

The biggest diameter of all 124 ovarian masses was measured – the largest one was 216 mm, the smallest one was under 10 mm. 54% of all tumors had diameter larger than 4 cm.

58.1% out of all mas ses were classified as cystic, 12.9% as solid and 29% as mixed. In four cases both solid and cystic masses were found in the same patient. Of all ovarian tumors 37 (29.8%) had wall thickness greater than 3 mm, 16 (12.9%) had papillary projections and 41 (33%) were septated. Only 6 masses of all contained fat, 5 of them were histologically proven to be mature teratomas. Kolmogorov-Smirnov test shows a statistically significant correlation between the type of mass and the assessment of masses as benign/borderline/malignant.

Оf histologically proven tumors 74.4% were benign and 25.6% were malignant. All masses classified on MRI as benign were identified correctly. Two masses, described as suspicious and malignant, turned out to be benign. All of the malignant tumors had restricted diffusion. The calculated ADC values of malignant adnexal masses are significantly lower than the ADC values of benign masses. Exceptions were found for endometrioma (1.01 ± 0.05 x 10^-3 mm²/s); mature teratoma (0.80 ± 0.04 x 10–3 mm2/s) and chronic abscess (0.61 ± 0.06 x 10^-3 mm²/s) – all of them presenting lower ADC values. (Figure 1) 72.7% of malignant neoplasms were mixed masses, 18.2% were solid and only one (9.1%) was cystic. Compared to them, 75% of benign tumors were cystic.

Figure 1

64 out of all 98 patients underwent contrast enhancement. 34 there were a subject of contraindications (history of previous allergic reactions to the contrast agent, elevated levels of serum creatinine or patient refusal). 39 (61%) of the masses showed enhancement. Three were classified as benign and four – as suspicious. 32 of the enhanced tumors were identified as malignant.

Ascites was found in 33 of the cases – in 15 of which is located only in the pouch of Douglas. In 15.3% of the cases, enlarged lymph nodes with diffusion restriction were found – all in patients with malignant masses and one with a proven chronic inflammatory process. In 15 cases enlarged metastatic locoregional lymph nodes were found. Eight patients had peritoneal deposits; four patients liver metastases; three patients bone metastases, two patients were with urinary bladder invasion and one patient had adrenal metastasis. In all cases with metastases three turned out to be from uterine cancer (ovarian masses in these cases were proven benign).

Assessing different types of adnexal lesions is important preoperatively. We find a number of reasons about the value of 3 Tesla MRI in such differentiation.

The MRI gynecological protocols we used concur the ESUR Quick Guide to Female Pelvis Imaging, 1.0 from 2019.15 Classical sequences (T1, T2) combined with post-gadolinium sequences and diffusion techniques provide reliable information on the nature of the adnexal masses.

It is known from previous studies that dynamic contrast enhanced MRI (DCE-MRI) is helpful in characterizing adnexal tumors. It could discriminate malignant from benign masses. According to the study of Thomassin-Naggara et al. there are three types of TIC showing benign, borderline and malignant ovarian tumors. Figure 2 demonstrates representative Type III curve of a malignant adnexal mass.

Figure 2

The number of patients (98) in our study exceeds those of similar ones known from the literature (30 in El-Wekil et al.14 and 58 in Koc et al.16). The average age of patients (47.2 years) as well differs respectably by seven and four years from the cited studies.14, 16

The WHO histological classification (according to Foti et al.17) divides primary ovarian masses into three main categories: epithelial, germ cell and sex cord-stromal tumors. Metastatic tumors are classified in a separate category. In 2016 Meinhold-Heerlein et al. revised the WHO classification introducing seromucinous tumors as a new entity.18 Our study includes 14 histologically different groups of ovarian masses – ten benign and four malignant. Diffusion MRI appearance of histologically different groups is shown in Table 3. Some of the benign formations have diffusion restriction – abscess, endometrioma, mature cystic teratoma and serous adenofibroma. In 88% of cases mature cystic teratomas are filled with sebaceous material and are lined with keratinized squamous epithelium19, compared to the most relevant feature – adipose tissue which is presented in only 67–75%.20 Diffusion restriction is caused by the presence of keratin or Rokitansky nodule and fat globules.21 Endometriomas as containing blood and hemosiderin can show diffusion restriction too.21, 22 Solid areas with similar changes can help the detection of malignant transformation. When it comes to an ovarian abscess, diffusion characteristics depend on the content – in more viscous one the signal intensity is higher on DWI and lower on ADC map.23 Diffusion techniques could differentiate abscess from cystic or necrotic neoplasm. Neoplasms usually show diffusion restriction peripherally and abscesses centrally.22, 24 According to cystic degeneration, some adenofibromas also could be characterized by restriction of the water molecules.22 ,25 ,26

In this study adnexal masses are classified based on their morphological appearance, similar to Foti et. al.17 and divided into three main groups – cystic, solid and mixed (cystic and solid).

Cystic adnexal masses could be unilocular or multilocular. Some of them have a non-ovarian origin. They are usually benign, with low signal intensity on T1-weighted images and high signal intensity on T2-weighted images.

Peritoneal inclusion cysts and hydrosalpinx are the most common extra ovarian lesions. They occur almost exclusively in premenopausal women and at imaging the ovaries are clearly separated from these cystic formations.27, 28

Functional ovarian cysts are the most common finding in women of reproductive age. Follicles are up to 20 mm as the dominant one could be 25 mm. Follicular cysts and corpus luteum cysts are larger and tend to increase if there is internal bleeding. This manifests with an increase of the signal on T1-weighted images.13, 17 They do not usually have diffusion restriction and does not change after contrast administration. Although, in corpus luteum cysts intense wall enhancement may be seen.

Serous cystadenoma and mucinous cystadenoma are benign tumors with thin walls (Figure 3). Mucinous type is usually larger, septated and has variable intensity on both T1- and T2-weighted images based on different mucin concentration. Some loculi are hyperintense on T1-weighted images, forming a pattern known as “honeycomb” or “stained glass”.17, 29, 30 Serous cystadenoma is more often bilateral and its wall could contain small nodules due to fibrosis or calcification.31, 32 Diffusion restriction could be detected in mucinous cystadenoma due to the dense mucinous material.12

Figure 3

Cystadenofibroma is usually a benign epithelial tumor that can present as a complex cystic mass with thick septa and solid component. It could present with plaques and nodules that have low signal intensity on T2-weighted images due to fibrous tissue (Figure 4).33, 34, 35

Figure 4

Endometriomas are part of the cystic lesions containing blood products. In addition to that, they characterize with hyperintensity on T1-weighted images and lower signal intensity on T2-weighted images, called “shading sign”. Sometimes these lesions could have high signal intensity on both T1- and T2-weighed images. They do not change their signal intensity on fat-suppressed sequences13, 31, 36, 37 Patients having endometriosis are at risk of developing ovarian malignancies.38 Endometriomas usually do not enhance after contrast administration but could have restricted diffusion.11, 22

Mixed ovarian masses containing both cystic and solid parts are always suspicious for malignant – surface epithelial tumors and metastases. The benign representative of this category is mature cystic teratoma.

Mature cystic teratoma is known as the most common ovarian neoplasm that arises from ovarian germ cells.13, 31 Usually part of this tumor has high signal intensity on T1WI and intermediate on T2WI, fat-fluid or fluid-fluid level, low signal calcification parts and floating debris. It could also have a soft-tissue protuberance called Rokitansky nodule. On fat-suppressed sequences the areas containing fat show drop in signal intensity. Malignant transformation of mature cystic teratoma is rare.19, 37, 39, 40 Enhancement after contrast application is not typical. They could represent with restricted diffusion in the areas with keratin and fat globules.22, 23

Serous and mucinous cystadenocarcinoma are the most common epithelial malignancies of the ovaries – 50% and 10% of malignant lesions.41 Mucinous tumors are larger, lobulated and may be hyperintense on T1WI in addition to the high protein concentration in mucoid material.30, 42 Cystadenocarcinomas have thick and irregular walls, septations, solid components and papillary projections that have low signal intensity on T2WI with contrast enhancement after contrast administration. Serious fluid part demonstrates with high signal intensity on T2WI (Figure 5). Peritoneal invasion is sometimes discovered.17, 41 In connection with their malignant nature, a pronounced diffusion restriction is observed.

Figure 5

Ovarian metastases most frequently originate from a primary process in the female genital tract, gastrointestinal tract (Krukenberg tumor) or breast. They are more commonly bilateral and multiloculated. Their solid parts are hypointense on T2WI and enhance after gadolinium administration. Distinguishing them from a primary ovarian process is not easy.17, 41, 43 Ovarian metastases have high signal intensity on DWI and low on ADC map (Figure 6).

Figure 6

Other less common representatives of mixed ovarian neoplasms are endometrioid tumors, yolk sac tumors and granulosa cell tumors.

Solid ovarian masses could have benign, borderline and malignant behavior. They include all three main histological types – epithelial, germ cell and sex cord tumors and metastases.

The Brenner tumor is a rare epithelial tumor and represents 2% of ovarian neoplasms.44 It is usually benign and has largely homogeneous low signal intensity on T1- and T2-weighed images. Its signal intensity is similar to those of fibromas but no cysts and necrosis are found in Brenner tumor. This ovarian tumor can occur in association with mucinous cystadenoma (Figure 7). Mild enhancement is observed after contrast application. Diffusion restriction is not characteristic of benign representatives of this tumor.39, 45

Figure 7

Fibromas encounter around 4% of all ovarian tumors. They could mimic malignant neoplasm as their size can vary and may be associated with ascites and pleural effusion (Meig syndrome). Another pathology they should be defined from is pedunculated uterine leiomyoma. These tumors demonstrate low signal intensity on both T1- and T2-weighted images. Scattered areas of high signal intensity could be present on T1WI due to cystic degeneration or edema.13, 17, 35 In this case diffusion restriction may be found. After contrast administration minimal enhancement is evident.

In this study 72.7% of histologically proven malignant neoplasms were mixed cystic and solid, 18.2% were solid and only one (9.1%) was cystic. That statement disagrees with El-Wekil et al.14 where no solid mass was found but cystic masses were 37.5% of their case series. However, 62.5% of tumors in their study were mixed cystic and solid which roughly coincides with our findings.

All of the histologically proven malignant lesions in this study show restricted diffusion. This confirms the literature data that an adnexal mass with higher signal intensity on DWI and lower on ADC map usually is a malignant lesion. Our results confirm the findings of previous studies in the literature that benign adnexal lesions have higher ADC values than the malignant once. We also found some exceptions of this statement concerning endometrioma, mature teratoma and chronic abscess presenting with lower ADC values despite of their benign origin.

Borderline ovarian tumors are usually complex masses that have some of the MR characteristics of the malignant one. They could show cellular proliferation and moderate nuclear atypia but without stromal invasion.46, 47 Similar to the study of Bent et. al.46 we identified 11 of the cases as suspicious. All of them demonstrated one or more MRI feature suggestive for malignancy – size more than 4 cm, solid part, cystic part with vegetations and septations, wall thickness more than 3 mm; contrast enhancement. In our study only one of the 10 suspicious cases were bilateral.

CA-125 is established tumor marker for ovarian cancer.13, 48 Limitation of this study is the small number of CA-125 tests performed before magnetic resonance imaging. Of these, elevated levels of CA-125 were found in 12 patients. Similarly, to other studies, over 60% of our patients with elevated CA-125 levels have proven malignant ovarian lesions.

Concerning unilateral or bilateral adnexal masses, we found malignant to be more often bilateral. Unilateral lesions are more often found in the right adnexa and in younger patients. This study as well as the Zhang et al. one49 suggests that large sizes and atypical signal intensity may influence the correct assessment of the type of ovarian lesions. The main limitations of our study include the retrospective reviewing of patients with some clinical missing, as well as surgical missing findings in patients who underwent surgery in another hospital.

Classifying adnexal masses is essential for the preoperative management of the patients. 3T MRI protocols, in particular diffusion techniques, increase significantly the accuracy of the diagnostic assessment. Further studies correlated with histological validation would support the role of MRI as a mandatory part of the patients’ management.