Correlation between ultrasound and Magnetic Resonance Imaging in diagnosis of Ovarian Tumors

INTRODUCTION

Ovarian cancer (OC) is an insidious disease that affects one out of every 74 women worldwide.¹ The mortality rate of OC is higher than any other female reproductive cancer and, accordingly, OC is the 5th leading cause of cancer deaths among women worldwide.²

In routine clinical practice, OC may be observed or identified indirectly in patients with asymptomatic disease. Characterization of OC provides a diagnostic challenge as it is essential to schedule effective therapeutic interventions in the preoperative setting, as well as to influence patient management.^3,4

OC patients usually present with an advanced extra-pelvic mass and this commonly results in a poor prognosis despite clinical progress and improved surgical techniques.^5,6

During the assessment of an adnexal mass, the primary purpose of imaging is to differentiate benign from malignant masses. Management strategies may include extreme staging operations for suspected ovarian malignancies, less intrusive operations, potentially benign neoplasms (i.e., laparoscopy), avoiding procedures, and following benign injuries.⁷

OC are classified based on tumor origin as metastatic, epithelial, germ cell, and sex cord-stromal cell tumors.⁸

Sonography is the first imaging test to examine women with suspected adnexal masses due to its widespread availability, relatively low cost, and high sensitivity in mass detection.⁹

 Nevertheless, it is restricted by its reduced specificity for benign tumors which can vary from 60 to 95%.¹⁰

The transvaginal ultrasound (TVUS) is considered a better instrument to detect normal ovaries and ovarian lesions than transabdominal US, as transabdominal US may not image close enough to the pelvis and bowel loops which may affect resolution, whereas TVUS directly contacts the ovaries and uterus.¹¹ The role of Doppler examination is to evaluate the presence of new vascularity inside the mass (central or peripheral).

Magnetic resonance imaging (MRI) offers valuable information for the classification of numerous ovarian masses as neoplastic or non-neoplastic, and benign or malignant.¹²

The use of MRI to diagnose ovarian masses involves a morphological consideration and signal intensity features on T1-weighted and T2-weighted images.¹³

In particular, improved MRI can help differentiate between benign and malignant tumors and may be useful to identify an adnexal mass.¹⁴

We focused this study on a discussion of the relative values of US/doppler and MRI in the evaluation of female patients with ovarian tumors.

PATIENT AND MATERIALS

1- Study Design

This retrospective study was performed on 250 patients who ranged from 13 to 70 years seen at the university hospital and the National Cancer Institute who were referred from the Obstetric and Gynecology department to the Radiodiagnosis department during the period from March 2017 to March 2018. All patients were subjected to detailed history with a particular focus on age, parity, menstrual history, previous history of gynecological disorders or operations, abdominal pelvic US, TVUS, color flow doppler, MRI, and histopathological review. No exclusion criteria except the claustrophobic patients.

2- US Imaging

US imaging was performed using a General Electric medical system logic P5, Siemens acuson X300, and Mindray DC3. All patients were examined in the supine position and, where possible, patients were asked to fast for at least eight hours before the study. The ultrasound scans, were reviewed by two experienced radiologists (MH, AO) and gynecological (SA) consultants based on the International Ovarian Tumor Analysis (IOTA) simple-rules. We examined the site of the tumors (right, left, or both), tumor size, tumor shape (rounded or oval), tumor echogenicity (hypo, iso, or hyperechoic), presence of vegetation or septation (if cystic), wall of the tumor (regular or not), Doppler vascularity of the content, containing cystic degeneration, soft tissue nodules, fat content (or not), compression of other organs (or not), suggestive benign or malignant staging of the tumor, presence of pelvic para iliac lymph node invasion (or not), and the presence of ascites or peritoneal nodules.

3- MRI

MRI was performed by a Phillips Achieva 1.5T system. Using a pelvic phased-array Torso coil, the all patients were imaged in the supine position. To minimize bowel peristalsis, intravenous administration of an antispasmodic drug (10 mg of visceralgine) was administered immediately prior to MRI. As feasible, patients were instructed to fast eight hours before the test. Non-contrast MRI protocol was applied T1-weighted axial (TR / TE, 500/10 ms), T2-weighted axial (TR / TE, 3300/100 ms), T2-weighted sagittal, and T2-weighted coronal. Before the administration of the contrast medium, MRI-DWI was acquired in the axial plane using a single-shot echo-planar imagery sequence. The MRI exam. were reviewed by three experienced radiolodists (TM, AO and MB) consultants.

4- Histopathological correlation

The histo-pathological evaluation is the gold standard for ovarian cancer diagnosis regardless of the findings on US or MRI. According to the post surgical histo-pathological results (180), lesions were divided into benign and malignant groups.

5- Image analysis for US, Doppler study and MRI-DWI

The ultrasound grey scale confirms the presence of cystic or solid ovarian mass and was analyzed for size, site, inner wall surface (smooth, irregular, papillae or solid projections), wall thickness (thin, thick or mostly solid) , septae (presence, thin or thick) and echogenicity of tumor sonolucent, low echogenicity, low echogenicity with echogenic core, mixed echogenicity, high echogenicity).

Following gray scale scanning, color flow signals were superimposed in real time and the regions of intratumoral neovascularisation were identified as location of vessels was central (solid areas, septa and papillae) or peripheral (around a cyst or a liquid area). The spectral wave forms were obtained from which the pulsatility and resistive indices were noted (PI and RI).  The range of values of PI for malignant masses was 0.48 to 1.33 with a mean of 0.85. PI of benign masses ranged from 1.3 to 7.4 with mean standing at 3.15, by using a PI value of 1.1 as the cut-off point, for differentiating benign and malignant lesions.

The range of values of RI for malignant masses was 0.4 to 0.98 with a mean of 0.5. RI of benign masses ranged from 0.4 to 0.9 with mean standing at 1.05, using an RI value of 0.6 as the cut-off point.

In MRI and diffusion three experienced radiologists observed the images (TM, AO and MB), The conventional MR images were analyzed and evaluated for the site (RT/LT), size, shape, and complexity of the tumor, T2 signal intensity of the solid part inside the mass and the presence of ascites and peritoneal deposits as well as lymph nodes. In MR-diffusion the lesions with low signal on diffusion images and high signal in the ADC maps assigned for free diffusion and diagnosed as benign masses, while Lesions with high signal on diffusion images with low signal in the corresponding ADC maps assigned for restricted diffusion and diagnosed as malignant masses, In ADC map we selected the region of interest (ROI) manually on the solid and the cystic component of the tumors, which was then automatically calculated on the work station to get the ADC values.

6- Statistical Analysis

All continuous data are presented as mean ± SD, median, and range while qualitative data are described in frequencies (number of cases) and percentages. Accuracy was demonstrated by the following terms: sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy. To define the optimum cut-off for the diagnostic markers studied, receiver operator characteristic (ROC) analysis was used. The student t-test for independent samples was used to compare numerical variables between study groups. Chi-square (c2) test was performed to compare categorical results. Alternatively, exact tests were used when the predicted frequency was below 5. A p-value of less than 0.05 was considered statistically significant. All statistical analyses have been conducted using Microsoft Windows version 22 of SPSS (Chicago, IL, USA).

RESULTS

This study enrolled 250 patients (98 patient post menopausal) with a mean age of 38.85 ± 17.1 years. Malignant tumors were more commonly found in older patients (38.85 years) compared to benign tumors (36 years). Neoplasms were classified according to histopathological examination into 120 benign and 130 malignant tumors.

Benign tumors included 15 serous cystadenomas, 15 mucinous cystadenomas, 25 dermoid cysts Figure (1), 15 benign dysgerminomas, 10 chocolate cyst, 5 granulosa cell tumor, 5 fibrothecoma Figure (2), 15 simple cysts, 5 tubo-ovarian abscess Figure (3), 5 spindle cell tumor (histo-pathological term of low grade sarcoma), and 5 hemorrhagic cyst.

Malignant tumors included 15 mucinous cystadenocarcinomas, 5 dysgerminoma, 15 serous cystadenocarcinomata, 5 metastatic tumor, 5 invasive high-grade mucinous carcinoma, 5 immature teratoma, 5 borderline tumors, 5 malignant hemorrhagic cyst, and 65 cases are classified pathologically as high grade ovarian carcinoma Table (1).

Analysis of ovarian tumors by the US

Lesions were classified pathologically by US into 105 benign cases, and 145 malignant cases Table (2). The sensitivity of the US was 43.75%, with 26.9% specificity. The positive predictive value (PPV) was 31.34%, and the negative predictive value was 38.64%. (Figure 4) shows the ROC curve of the US data obtained (Table 3). Twenty-six (26) were classified as complex tumors (52%), 17 (34%) as cystic tumors, and 7 (14%) as solid tumors. The majority (48) of tumors were unilateral (96%), and only two tumors were bilateral. In terms of Solid vegetation, 29 (58%) were equal or greater than one centimeter, and 21 (42%) were less than one centimeter. After dopper examination the sensitivity was 81.51%, with 93.9% specificity, The positive predictive value (PPV) was 91%, and the negative predictive value was 86%.

Analysis of ovarian tumors by MRI

The mean dimension of the benign neoplasms measured was 7.6 ± 1.3 cm, while the mean dimension of malignant tumors was 12.3 ± 2.4 cm, with a highly significant difference between both types (p<0.001). Tumor composition, laterality of the lesions, and solid vegetation of the lesions were similar to those observed using US. However, the sensitivity of MRI was 96.43%, specificity was 95.83%, PPV was 96.43%, and NPV 95.83% Figure (5), Table (4).

MRI were found 22 cases with the criteria of malignant lesions (tumors larger than 6 cm, septae more than 3 mm, and solid vegetation more than 1 cm with thick wall). Three cases (two mature teratomas and one serous cystadenoma) identified as benign by pathology were presumed to be malignant by MRI Table (5). Analysis of one patient with mucinogenic cystadenocarcinoma assumed that the tumor was benign because by MRI analysis there were no septations or solid nodules, but pathology later proved this to be a malignant tumor.

Analysis of ovarian tumors by DWI

In terms of diffusion-weighted imaging (DWI), the apparent diffusion coefficient (ADC) values were 1.85 ± 0.58 for benign tumors, and 0.84 ± 0.16 for malignant tumors (Table 6). MRI-DWI showed 100% sensitivity, 93% specificity, 95.5% PPV, and 98% NPV Figure (6), Table (7).

Regarding the difference between MRI-DWI and pathology, restriction by MRI-DWI refers to low ADC values. Specifically, a low signal in a corresponding ADC map and high signal in diffusion images was observed in 130 patients but tumors were proven to be malignant by pathology. Alternatively, 120 patients showed facilitated diffusion (high ADC values, high signal on corresponding ADC map, and low signal in diffusion images). These tumors were later proven to be benign by pathological examination.

35 cases of the 130 initially characterized as malignant were later proven to be false positives as 5 were pathologically classified as mature teratomas, 5 fibrothecoma, 5 tubo-ovarian abscess, and 10 cases of chocolate cyst. Each proved to have a benign pathology with high signal on diffusion images (restriction) but the low signal on corresponding ADC map and ADC values 1.1 to 0.8x10-3 mm Table (8).

The diagnosis of such lesions was depended on the combination of the other MRI sequences with the MRI-diffusion and the ADC map/ value.

Figure 1: Analysis of a 32 year old female patient complaining of lower abdominal pain. (A) Axial trans abdominal US scan shows a well defined mass lesion at the right adnexal region with hyper and hypo echoic pattern. The mass has cystic components and fluid level. (B) Axial T2-wighted image, and (C) Axial T1 fat saturation images show high T1 and T2 weighted images with loss of signal in and out of phase and fat suppression images consistent with fat content. (D) MRI-DWI and ADC shows high signal on both MRI-DWI and ADC (T2 shine-through effect). The lesion was diagnosed by histopathology as a mature cystic teratoma.

Figure 2: Analysis of a 60 year old female patient complaining of abdominal pain. (A) US images shows a large well-defined, encapsulated, isoechoic right adnexal soft tissue mass lesion with areas of small cystic degeneration with no invasion of the surrounding structure. MRI indicates a right adnexal mass, lobulated, well-defined and encapsulated with heterogeneous structure associated with signal intensities sugg. of internal fibrosis, small cystic changes, and peripheral hemorrhage. This mass was hypointense at T1 weighted image (B) and T2 weighted image (C) with peripheral enhancement on T1 fat saturation (D). The mass also shows no evidence of fat, minimal amount of free peritoneal fluid collection (arrow) on (D) sagittal T2. (E) The lesion appeared as a homogeneous hyperintense signal on the ADC map. The lesion was proved by histopathology to be a fibrothecoma of the right ovary with myxoid degeneration, intact capsule, and no malignancy.

Figure 3: Analysis of a 46 year old female patient who presented with lower abdominal pain and fever. (A) Trans abdominal and TVUS, revealed the right adnexa contains a complex, predominantly cystic mass lesion of thick wall, turbid contents, and fine septation associated with a small amount of free pelvic fluid collection at the Douglas pouch. The lesion also shows peripheral vascularization on color Doppler study. The right adnexa shows a well defined cystic lesion measuring about 5.7 x 6.7 cm in maximum dimensions. It displays hypointense signal with peripheral enhancement on post-contrast T1 weighted image (B), hyperintense signal on T2 weighted image (C). MRI-DWI (D) indicates the lesion shows high signal intensity with a low ADC map (E) (low ADC value indicates true diffusion restriction). The lesion was diagnosed by pathology as a right adnexal tubo-ovarian abscess with no malignant cells.

Figure 4: ROC curve indicating the diagnostic accuracy of US for detection of ovarian tumors.

Figure  5: ROC curve indicating the diagnostic accuracy of MRI for detection of ovarian tumors.

Figure 6: ROC curve indicating the diagnostic accuracy of MRI-DWI for detection of ovarian tumors.

Table 1: The different pathological types.

Table 2: Represents the pathological types according to US.

Table 3: ROC curve for Results of US compared to pathology.

Table 4: ROC curve for results of MRI compared to pathology.

Table 5: The pathological type according to MRI.

Table 6: Represents the pathological types according to DWI.

Table 7: ROC curve for results of DWI compared to pathology.

Table 8: Shows the comparison between the ADC values between benign and malignant tumors