Differentiation of an adnexal mass into benign and malignant is the primary role of imaging modalities. Proper diagnoses will direct patients to the appropriate treatment algorithm to reduce the number of women unnecessarily undergoing cancer surgery, to preserve fertility in young women (by allowing laparoscopy) [2].
MRI had helped in identifying malignant lesions before surgery, particularly if an indeterminate lesion was discovered by the US. MRI can reveal morphologic characteristics (e.g., papillary projections, nodularity, septa, solid portions, and signal intensity), but these criteria are unreliable in differentiating between benign and malignant tumors [5]. MRS can offer information concerning cellular integrity, cell proliferation or degradation, and energy metabolism of tissues [6]. MRS can detect metabolic and molecular features and precede morphologic alterations characteristic of malignancy, so sensitivity is expected to be improved by adding MRS [4].
The current study was performed on 57 patients having 65 adnexal masses (bilaterally 8 cases) at the US. The patient’s age ranged from 15 to 70 years old (mean age 34.80 ± 14.76 SD). Pelvic MRI with MRS was done for all patients. Fifty-two (52) patients underwent surgery with pathologic correlation; the other five cases had a clinical and the US follow-up. We classified the detected adnexal lesions into 37 benign, 4 borderline, and 24 malignant lesions (Table 1).
In this study, the detected adnexal lesions varied in their mean maximum diameter ± SD (benign lesions ranging about 2–17 cm (8.2 ± 3.5), borderline lesions 6–15 cm (7.5 ± 3.3), and malignant lesions 6–25 cm (13.6 ± 8.0) (Table 2).
Regarding their morphology, they were classified into predominantly multi-locular cystic lesions; 40 lesions (62%), mixed cystic/solid lesions; 15 lesions (23%), and solid lesions; 10 lesions (15%) (Table 2).
MRI was done to all cases with a variable T2 signal; a low signal in 3 lesions (5%), a mixed intermediate signal in 23 lesions (35%), and a hyper-intense T2 signal were seen in 39 lesions (60%). On post-contrast enhancement, the complex adnexal masses showed variable contrast uptake either avidly enhanced in 16 lesions (25%), moderately enhanced in 21 lesions (32%), mildly enhanced in 15 lesions (23%), or no appreciable enhancement in 13 lesions (20%) (Table 2).
The MRI features for the diagnosis of malignancy are a large solid component, wall thickness larger than 3 mm, a septal thickness over 3 mm, and/or nodularity and necrosis. Supportive criteria can help in differentiating a tumor as malignant one; this embraces the involvement of pelvic organs or sidewall; peritoneal, mesenteric, or omental disease; and ascites and adenopathy [7]. In this study, 20 cases associated with ascites, 7 cases associated with lymphadenopathy, and 3 cases with peritoneal implants (pathologically proved to be 2 cases of malignant serous cystadenocarcinomata and a case of malignant immature teratoma) (Table 2).
Borderline epithelial ovarian tumors (BEOTs) are tumors of low malignant potential. They are characterized by atypical epithelial proliferation, moderate nuclear atypia, however, without stromal invasion. Patients with BEOTs are discovered with an early stage disease, and have an excellent prognosis one surgical excision; however, they have a high recurrence rate. It is necessary to preoperatively differentiate BEOTs because a conservative fertility-sparing laparoscopic surgery can be done [8].
A study done by Lyer et al. (2010) revealed that the preoperative diagnosis of borderline tumors is rare because they lack diagnostic imaging features. On MRI, borderline tumors are predominantly cystic, with fluid ranging in T1 and T2 signal because of varying concentrations of protein and mucin. There are also multiple enhanced solid mural nodules or thick septa. There is no proof of lymphadenopathy, ascites, or peritoneal implants [9].
In this study, BOTs were a great challenge for us. Their morphological features can mimic benign or malignant lesions, yet they lack the invasive behavior of the malignant lesions. In functional imaging, some of them mimic benign lesions, which is consistent with their non-invasive behavior. The papillary projections were detected in three cases, MRI diagnosed two of them towards borderline to malignant lesions and one lesion to be sure malignant, while MRS showed lactate, lipid, and NAA peaks with no significant differences in the values between benign and borderline masses so MRS evaluated them as benign masses. The final MRI-MRS diagnosis considered them as borderline cases. So the difference between MRI and MRS data can suspect borderline tumors in which suspicious features can be detected in MRI while no metabolic levels indicate malignancy can be detected in MRS.
A study conducted by MA and colleagues in 2015 on 69 patients showed CHO peak was detected in all 69 cases (100%), NAA peak in 67 cases (97%, 25 benign and 42 malignant), lipid peak in 47 cases (17 benign and 30 malignant), and lactate peak in 8 cases (4 benign and 4 malignant) [10].
In this study, CHO peak was detected in 42 cases (64.5%) (17 benign (40%), 22 malignant (53%), and 3 borderline (7%) lesions), Cr peak in 19 cases (29%), NAA peak in 49 cases (75%) (26 benign (53%), 3 borderline (6%), and 20 malignant (41%) lesions), lipid peak in all cases (100%), and lactate peak in 53 cases (81.5%) (27 benign (51%), 3 borderline (6%), and 23 malignant (43%) lesions (Tables 3, 4, 5, and 6).
CHO is a cell membrane marker observed at 3.2 ppm. It reflects the biochemical activity of cell membranes. Increased CHO probably reflects the increased number of cells and/ or cell membrane synthesis [11]. El Sorogy et al. (2012) have reported that Cho peak was detected in all solid tumor or solid parts in cystic tumors. However, the Cho peak was also found in benign tumors so, it could not be used for the differentiation between benign and malignant tumors. Ovarian lesions have shown significantly higher levels of Cho in malignant lesions compared to the benign ones, may be due to tumor cell proliferation and increased cellularity and growth. However, it was also detected in some benign lesions such as mature cystic teratomas [4]. A similar study conducted by Malek et al (2015) on 23 ovarian masses showed the presence of Cho peak in 17 of 19 malignant masses (sensitivity 89%) and in 3 of 4 benign masses. So it could not be used in differentiation between benign and malignant tumors [12].
In this study, CHO peak was detected in 42 lesions (64.5%) including 17 (40%) benign lesions (5 abscesses, 3 mature teratomas, 2 mucinous and 6 serous cystadenomas (Fig. 2), 1 broad ligament fibroid), three borderline (7%) lesions. It was above the noise level but lower than twofold. In 22 malignant lesions (53%), sharp peaks were found in serous and mucinous cystadenocarcinomas. So CHO peak could not help in differentiation between benign and malignant tumors (Tables 3, 4, 5, and 6).
Regarding CHO/Cr ratio a study conducted by El Sorogy et al. (2012) showed that the mean ratios of the malignant ovarian masses were significantly higher than benign ovarian masses with a sensitivity 83%, and specificity 82% [4]. Another study conducted by Stanwell et al. (2008) showed a higher CHO/Cr ratio in malignant than in benign cystic ovarian tumors except for one patient who had a serous cystadenofibroma and the CHO/Cr ratio was 3.13. The benign tumors had a CHO/Cr ratio ≤ 1.15 for both solid and cystic components. All malignant tumors had the CHO/Cr ratio ≥ 3.09 for both solid and cystic tumors [13].
In the current study, mean CHO/Cr ratio was 1.29 ± 0.98SD for malignant lesions, while the mean value in borderline lesions was 0.63 ± 0.15 SD and the mean value for the benign lesions was 0.65 ± 0.34 (Fig. 2). The mean CHO/Cr ratio was much higher in malignant than in benign lesions; which was statistically significant (P ≤ 0.001) as well as between the borderline and invasive lesions (P = 0.05), but not between the benign and borderline lesions (Table 7).
NAA is found in high concentration within the nervous system [14]. NAA is reverberant at 2.0 ppm, it is called neuronal marker [15]. Some studies [16] detected NAA metabolites in both solid and cystic elements of the mucinous cystadenoma. A study conducted by El Sorogy et al. (2012) also detected the NAA signal in all cases of mature cystic teratomas, denoting the presence of a neural component (ectodermal tissue) [4]. Also, a study conducted by Stanwell et al. (2008) reported the presence of NAA in all of the teratomas and serous cyst-adenomas, as well some of the serous carcinomas. We agreed with this study, as sharp NAA peaks were found in mature cystic teratomas, serous and mucinous cystadenomas, and tubo-ovarian abscesses. NAA peak was also detected in 3 borderline tumors and most malignant tumors (20 lesions) [13] (Fig. 3).
Lipids are not detectable in the spectrum of a healthy human. Some pathologic conditions yield lipid signals at 0.9 and 1.3 ppm [9]. Both studies conducted by Okada et al. (2001) and Fiaschetti et al., (2012) agreed in observation of lipid peak in lesions with high serous content and teratoma which is reflecting a high content of the fatty component. It was in agreement with our study; a lipid peak was detected in all the cases (100%) but values were higher in cases of mature cystic teratoma which reflects a high fatty component of the tumor [17, 18] (Fig. 1).
Lactate is the end-product of anaerobic glycolysis and appears at 1.3 ppm [12]. Lactate signal is detected in the cystic portion of the ovarian tumors which were detected not only in all malignant tumors but also in some benign tumors as in dermoid cysts. A previous study was done by Hascalik et al. (2005) demonstrated that lactate signals were detected in some benign tumors; but, all cases of malignant tumors showed lactate and attended higher peaks than benign tumors [19]. In the current study, there was agreement with the previous studies. A sharp lactate peaks were found in serous benign lesions, tubo-ovarian abscesses, broad ligament fibroids, and solid portions of the borderline lesions, as well as the solid portions of the malignant serous and mucinous cystadenocarcinoma lesions (Fig. 1), but the value detected in the malignant case was much higher than the value detected in any of the benign cases. Therefore, when no lactate signals were detected; there is a very low probability of malignancy and the lactate signal might be used in differentiation between malignant and benign tumors.
In this study, c MRI suggested malignant pathology in 37 masses; 28 of them were true malignant (TP) (24 of them were true malignant and 4 of them were borderline tumors), while 9 were benign but were faulty diagnosed as being malignant (FP) (4 cases of benign serous cystadenoma (Fig. 2), 2 cases of broad ligament fibroid, 1 tubo-ovarian abscess, 1 endometrioma (Fig. 3), last case of granulosa cell tumor (Fig. 4). C MRI also suggested benign pathology in 28 cases; all of them were true benign (TN) (Table 8).
MRS suggested malignancy in 25 cases; all of them were TP, however, it suggested benign pathology in 40 cases; 37 of them were benign (TN), misdiagnosed 3 cases were diagnosed as being benign (FN); (2 cases of borderline serous cystadenoma and a case of malignant immature teratoma) (Table 8). Two cases of borderline papillary serous cystadenoma showed lactate, lipid, and NAA peaks like cases of benign serous cystadenoma but both of them showed suspicious features in MRI. A case of malignant immature teratoma showed only lipid peaks with its peak values were much lower than its value in our cases of benign teratoma. MRS could not detect any other metabolites reflect the malignant nature of the mass it might be due to contamination of the voxel by the necrosis and heterogeneous nature of the tumor.
In our study, c MRI diagnostic performance in the assessment of adnexal lesions was 100% sensitivity, 76% specificity, and accuracy 86%. A study carried out by Bekiesińska et al. (2007); the diagnostic accuracy of MRI in characterizing ovarian lesions; it was 83.3% which is slightly lower compared with ours in which the diagnostic accuracy of MRI was 86% [20]. This difference could be due to the different number and nature of cases included in both studies. Another study performed by Rieber et al. (2001) on preoperative diagnosis of ovarian tumors with MRI. MRI had shown 83% sensitivity, 84% specificity, and 83% diagnostic accuracy, while in our results, MRI had shown higher 100% sensitivity and diagnostic accuracy (being 100% and 86% respectively) and lower specificity 76% (Table 9) [21].
Regarding the diagnostic performance of MRS, a study conducted by MA and colleagues (2015) on MRS for differentiating benign from malignant solid adnexal tumors. The study included 69 patients with solid adnexal tumors (27 benign and 42 malignant) with a sensitivity of 94.1%, specificity of 97.1%, and an accuracy,91.2% respectively [10]. In our study, MRS had shown lower sensitivity 89%, however, higher specificity and accuracy of 100% and 95% respectively (Table 9).
MRS is optional cost-effective imaging. It provided good information about the different metabolites in the different ovarian lesions. MRS does not require the IV contrast administration, thus it can be used safely in pregnant patients and patients with renal dysfunction. However, MRS has few drawbacks; it depends on field homogeneity, so voxels should be selected carefully to avoid contamination from surrounding structures. Also, the respiratory movement in the pelvis contributes to the field inhomogeneity which may produce noisy spectra. Another major drawback is the longtime of examination (about 10 min), which is not practical with old and claustrophobic patients [22].
This study has many limitations, first of all is the limited sample size, which may affect the diagnostic values presented by the MRI and MRS. We recommend another extended broad analysis with more biologically variable range of tumors especially malignant cases for proper evaluation of metabolic differences in malignant and benign lesions, proper estimation of the metabolite cut off value, and to improve the ways of the noninvasive preoperative diagnosis. Another limitation is that the size of the voxel needed to attain an adequate SNR ratio to discover the presence of malignancy. In very small lesions, the sensitivity to differentiate benign from malignant lesions decrease due to partial volume effects. To overcome this limitation, MRS should be performed using higher field strengths with high SNR and smaller voxel sizes.