Choline/creatine ratio
Although proton MR spectroscopy (1H-MRS) has established its role as a non-invasive tool in the diagnosis and post-treatment follow-up of brain, breast, and prostate tumors, its role in ovarian tumors is still in its infancy [8]. A limited number of studies have been published with variable results probably due to the diversity of pathological subtypes of ovarian tumors and the technical challenges of MR spectroscopy [6,7,8]. A study conducted by Ma et al. [8] showed a higher choline-to-creatine ratio in malignant ovarian tumors compared to benign ones. They reported that a choline-to-creatine ratio greater than 7.46 indicated that a tumor was malignant. An earlier study by Stanwell et al. [7] reported that a choline-to-creatine ratio greater than 3.09 indicated a malignant tumor. In this study, the integral choline-to-creatine ratio was 1.73 ± 3.16 in benign adnexal lesions versus 5.46 ± 3.91 in malignant lesions, with a statistically significant difference and a threshold of 3.6 to differentiate between benign and malignant lesions, with a sensitivity, specificity, positive predictive value, and negative predictive value of 69.2%, 90.9%, 90%, and 71.4% respectively. This threshold is more in keeping with that of Stanwell et al. [7] probably due to the heterogeneous cystic and solid consistency of the majority of the tumors, as opposed to Ma et al. [8] who studied solid tumors in all cases. Moreover, the relatively low sensitivity in this study compared to that conducted by Ma et al. [8] who reported a sensitivity of 94.1% is also likely attributed to the heterogeneous consistency of the studied sample, with some malignant cases not showing a choline peak due to small solid component.
A striking observation in this study was a higher integral choline-to-creatine ratio in metastatic ovarian tumors compared to primary tumors with mean choline-to-creatine ratio 9.74 ± 2.11 in metastatic tumors versus 5.55 ± 1.65 for primary tumors with a threshold of 6 to differentiate between them. To our knowledge, this is the first study to compare choline-to-creatine ratio in primary and metastatic ovarian tumors. Further investigations with larger numbers of cases are needed to investigate this.
N-acetylaspartate
N-acetylaspartate (NAA) is a marker of neuronal elements and it resonates at 2.02 ppm. An unassigned prominent peak was previously shown on MRS of breast, cervical, prostate, and ovarian cancers. This peak could possibly be assigned to NAA or to the –CH3 moiety of salicylic acid [11,12,13,14]. In a study by Boss et al. [15], NAA signals were reported in cyst fluid of serous cystadenomas of the ovary. This was also reported by Kolwijck et al. [16] who concluded NAA and N-acetyl groups from glycoproteins or glycolipids contribute to the 2.0–2.1 peak in ovarian cyst fluid. It has been proposed that the NAA system is a molecular water pump [17]. The results of our study are in line with these previous studies, where we recorded sharp NAA peaks in our two cases of mucinous cystadenomas (Fig. 6). Stanwell et al. [7] reported a significant peak at 2.07 ppm in all teratomas, serous cystadenomas, and two of five serous carcinomas, with a higher peak in malignant tumors. In this study, NAA peak was detected in three benign cases and two malignant ones, yet we had one case of mature cystic teratoma which did not show NAA peak. This was unlike the results of Stanwell et al. [7], possibly due to the absence of neuronal elements in our lesion.
Lactate
Lactate is the end product of anaerobic glycolysis. It resonates at 1.3 ppm and usually indicates carbohydrate catabolism and the lack of oxygen in energy metabolism [18]. Stanwell et al. [7] reported higher lactate levels in malignant ovarian tumors compared to their benign counterparts. Hascalik et al. [19] found high lactate levels in pelvic abscess due to anaerobic glycolysis. Okada et al. [11] reported lactate peaks in some benign and all malignant tumors, higher in the latter group. In this study, lactate was only detected in one case of pelvic abscess, in line with Hascalik et al. [20], one case of mature cystic teratoma, and one case of serous cystadenofibroma complicated with torsion. This could be explained partly due to obtaining the MR spectra from only the solid portions of malignant tumors, unlike Okada et al. [10] who obtained spectra from both solid and cystic parts, with the highest lactate levels reported in the cystic parts of malignant tumors. Our results were more in keeping with Ma et al. [8] who found a lactate peak in four malignant and four benign tumors, which was also explained by the more solid tumor sample in their study.
Lipid
Previous studies of brain and cervical cancers suggested that lipids could possibly be a cancer marker [19, 21, 22]. In malignant brain tumors, this was assumed to be a result of necrosis because tumors proliferate without a vascular network to maintain their metabolic needs [19]. Cho et al. [13] had similar results in MR spectroscopy of ovarian masses. Their study reported a lipid peak only in malignant ovarian tumors and mature cystic teratomas. However, Ma et al. [8] reported lipid peaks in benign and malignant ovarian masses, with greater lipid-to-creatine ratios in malignant than benign masses, yet the difference was not statistically significant. This is in keeping with the results of this study, where lipid peaks were found in six benign and six malignant adnexal masses. Takeuchi et al. [23] suggested lipid as a marker for ovarian fibrothecoma, where they used it to differentiate them from other benign uterine lesions like subserous leiomyomas and from ovarian fibromas. In this study, lipid was reported in all fibrothecomas (2), in keeping with Takeuchi et al. [23], one mature cystic teratoma, one tubo-ovarian abscess, one of three broad ligament leiomyomas, and one benign serous cystadenofibroma complicated with torsion and with pathologically proven massive infarction necrosis. In the last case, the lipid peak could be due to the disruption of macromolecules in the cell membrane, a finding in line with El Sorogy et al. [24] who reported a case of fibroma with infarction necrosis also showing a high lipid peak. Moreover, in the case of broad ligamentary leiomyoma with high lipid peak, the voxel was also placed in the cystic portion, which showed degeneration on pathology.
Our study had several limitations. First is the relatively small number of cases which included only a limited pathological variety of adnexal lesions. Second, lactate was detected only in three lesions, which could be attributed to higher intensity of lipid peaks which resonate at the same frequency as lactate. Lipid peaks could have obscured the lactate peak. Third, a large number of cases were heterogeneous in consistency rendering the obtaining of adequate spectroscopic data technically challenging.