Differentiation between endometriosis-associated ovarian cancers and non- endometriosis-associated ovarian cancers based on magnetic resonance imaging

Objectives: Endometriosis-associated ovarian cancer (EAOC) patients show different clinical characteristics compared with non-EAOC patients. However, a few studies are focused on the imaging characteristics of EAOC until now. We assessed MRI characteristics in differentiating EAOC and non-EAOC.

We retrospectively analyzed clinical and MRI characteristics from 54 patients with 67 lesions diagnosed with primary epithelial ovarian carcinoma at the Third Affiliated Hospital of Guangzhou Medical University between January 2012 and October 2020. We studied MRI findings such as maximum diameter, morphology, configuration, locularity, features of mural nodules, lymphadenopathy, peritoneal implants, the presence of hyperintensity on T1WI, and hypointensity on T2WI. We also studied the clinical characteristics. Significant MRI variables in univariate analysis were selected for subsequent multivariate regression analysis. This study evaluated the diagnostic performance of the significant MRI variables in univariate analysis.

We found that the patients with EAOC, compared with those with non-EAOC, were younger, more unilateral, and had earlier FIGO stage. Univariate analysis revealed that morphology, locularity, growth pattern of mural nodules, and hypointensity on T2WI were factors that significantly differed between EAOC and non-EAOC. In the multivariate logistic regression analysis, locularity and hypointensity on T2WI were independent predictors to distinguish EAOC from non-EAOC.

EAOC typically presented as a unilocular mass with hypointensity on T2WI in cystic components. MRI could help distinguish EAOC from non-EAOC. Advances in knowledge: MRI is a promising tool for preoperative diagnosis of EAOC.

Epithelial ovarian cancer (EOC) is the seventh most common malignant tumor and one of the most lethal gynecological malignancies in females patients worldwide.1,2 The incidence of endometriosis in EOC patients ranges from 4.2 to 14.5% in different studies.2 The second and most common pathological types of endometriosis-associated ovarian cancers (EAOC) are clear cell carcinoma (CCC) and endometrioid carcinoma, respectively.3 Compared with western countries, the incidence of clear cell carcinoma is higher in Asia.2 EAOC patients show younger age, earlier clinical symptoms, and better survival than that of non-endometriosis- associated ovarian cancer (non-EAOC) patients.4 At present, primary debulking surgery followed by platinum chemotherapy is considered to be the primary treatment for EOC and EAOC.5 Unfortunately, the prognosis of high-grade tumors is generally poor, especially CCC.6 Personalized treatment might be necessary and targeted therapy and immunotherapy should be further studied. Imaging plays an important role in diagnosing, differentiating various types of ovarian tumors.7–9 The imaging findings of typical epithelial ovarian cancer include cystic solid mass, thick septum, mural nodule or papillary projection, necrosis, solid-enhanced components, ascites, lymphadenopathy, peritoneal, mesenteric, and omental metastasis.10,11 To our knowledge, a few studies are focused on the imaging characteristics of EAOC until now, mainly including the comparison between ovarian cancer with endometriosis and benign endometrioma,12 longitudinal changes of malignant transformation of endometriosis, and CT imaging findings to identify EAOC.13–15 Multilocular, cystic-solid mass with the loss of T2 shading, enhanced mural nodules, the maximum diameter of mural nodules > 3 cm, restricted diffusion of solid components are considered to be the imaging findings of ovarian cancer with endometriosis.15,16 However, these imaging findings overlapped with traditional ovarian cancer, and the MRI imaging disparities between EAOC and non-EAOC have not been discussed before. Although surgical staging is the gold standard, the MRI-based pre-operative staging assessment could be helpful in the initial patient management to identify unresectable disease. Pre-surgical MRI examination has the potential to diagnose EAOC and evaluate tumor staging. Therefore, we explored potential MRI findings that can differentiate EAOC from non-EAOC. And the diagnostic performance of MRI findings of EAOC was explored in this study. This study may help to distinguish EAOC from non-EAOC.

and materials Population Patients suspected of ovarian cancer without confirmation of benignity or malignancy were undergone an MRI examination. Besides, MRI was more widespread selective use in clinical practice for its superiority of high soft-tissue resolution. This is a retrospective study of consecutive 54 patients of EOC with pre-surgical MR examination at the Third Affiliated Hospital of Guangzhou Medical University between January 2012 and October 2020. The study was approved by our institutional research ethics board and waived for each patient. Sixty-seven lesions in 54 patients confirmed pathologically to be primary epithelial ovarian carcinoma were analyzed. These patients underwent MRI scanning within one week preoperatively. Inclusion criteria were as follows: (1) preoperative MR scans were performed at the Third Affiliated Hospital of Guangzhou Medical University; (2) surgical treatment and a diagnosis of primary EOC confirmed by pathology, and (3) with or without pathological evidence of endometriosis. Exclusive criteria were as follows: (1) ovarian tumor was invaded or metastasized from other organs (five cases); (2) patients without pre-surgical MRI (53 cases); (3) lesions were too small to assessed on MRI (three cases); (4) MR images with poor image quality (0 cases); and (5) patients without full MR imaging information (six cases). Fifty-four patients diagnosed with EOC in our institute were divided into EAOC group and non-EAOC group. Those who met Sampson’s conditions were divided into EAOC group: (1) coexistence of ovarian cancer and endometriosis in the same ovary; (2) a similar histological pattern; (3) secondary tumors metastatic to the ovary were excluded; and (4) histopathological evidence demonstrating the transition from benign endometriosis to malignancy. And those patients who did not meet the above conditions were identified as the non-EAOC group. MRI protocol MRI scans were performed using Philips 3.0 T superconducting scanner (Achieve TX, Best, the Netherland) with a 16-element Sense-XL-Torso phased array coil. Routine scan sequences contained the following sequences and corresponding parameters : axial T2W images (T2WI, turbo spin-echo (TSE); repetition time (TR)/echo time (TE) = 1275/70 ms; matrix = 268/163; FOV = 300 × 300 ), sagittal T2WI (TSE; TR/TE = 1275/70 ms; Matrix = 268/163; FOV = 320 × 320 ); axial or coronal T2WI SPAIR sequence (TR/TE = 1250/50; matrix = 268/163), axial T1W images (TSE; T1WI; TR/TE = 600/8 ms; matrix = 268/163; FOV = 300 × 300 ) and diffused-weighted imaging (DWI, b value = 0,500,800 s/mm2; TR/TE = 4708/62 ms; matrix = 132/94; FOV = 400 × 279 ). 53 of 54 cases underwent dynamic contrast-enhanced MRI(DCE-MRI) scans (TR/TE = shortest/shortest ms; FOV = 250 × 250 ; Matrix = 224/224). During DCE MRI acquisition, pre-contrast T1 mapping was acquired, followed by a dynamic scan with 15 consecutive phases. The contrast agent gadopentetate meglumine was i.v injected with a dose of 0.1 ~ 0.2 mmol/kg, followed by a 20 ml flush of 0.9% sodium chloride solution using an automated injector at a rate of 2 ml s−1. All MRIs were obtained with a slice thickness of 4 to 6 mm. Clinical characteristics Clinical and histological records (HIS) including age at diagnosis, menopausal status, preoperative serum levels of CA125, presence or absence of endometrial cancer, FIGO stages, and histopathological results were collected by a resident doctor from the HIS system. Preoperative serum CA125 were divided into low (600 U ml−1) level. MRI image analysis MRI characteristics were retrospectively assessed by two radiologists with 3 -year and more than 10-year experience in female genital system MRI, respectively. All MRI materials of this study were extracted from the picture archiving and communication system (PACS). Both radiologists were allowed to be aware of the clinical data including laboratory results. But they were blinded to the pathological results. The MRI findings, such as the maximum diameter of the tumor, laterality, morphology of tumor, configuration, locularity, features of mural nodules, ascites, lymphadenopathy, and peritoneal implants, were assessed on both T1WI and T2WI. Also, the presence or absence of hyperintensity on T1WI and hypointensity on T2WI in the cystic components of the mass were assessed. The morphology of the tumor was divided into an oval and irregular shape. The configuration of tumors was divided into cystic (cystic components > 70%), solid (solid components > 70%), and cystic-solid type (the proportion of cystic components was 30–70%). Features of mural nodules include growth pattern, surface, and height of mural nodules. A focal growth pattern was defined as one or two papillary nodules protrude from the inner wall of masses. And the growth pattern was classified as multifocal growth pattern when more than two papillary nodules or wide-base nodular structures consist in the inner surface of tumors. The surface of mural nodules was divided into two parts: smooth and irregular. And the height of mural nodules was defined as the maximum length from the top to the bottom of nodules. Hyperintensity on T1WI was defined as the high signal intensity in cystic components of the mass on T1WI. Hypointensity on T2WI was defined as the low signal intensity in cystic components of the mass on T2WI, and the signal intensity is at least as dark as the skeletal muscle on the T2WI. Statistical analysis Statistical software SPSS25 was used for statistical analysis. The clinical characteristics of the two groups were analyzed using the independent sample t-test, Pearson chi-square test, or Fisher’s exact test. MRI findings such as the maximum diameter of the tumor, laterality, morphology of tumor, configuration, locularity, features of mural nodules, ascites, lymphadenopathy, peritoneal implants, and the presence of hyperintensity on T1WI or hypointensity on T2WI were analyzed using Pearson chi-square test or Fisher’s exact test. Significant variables of MRI findings in univariate analysis were subjected to multivariate logistic regression analysis to evaluate independent predictive factors to differentiate EAOC. This study evaluated the diagnostic performance of the significant MRI variables in univariate analysis. Diagnostic performance, including sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated. A p-value less than 0.05 (p < 0.05) signified that groups differed statistically.

Comparison of clinical characteristics between EAOC and non-EAOC A total of 54 patients with 67 lesions were included in the study. Three lesions of EOC patients with bilateral lesions were excluded from MRI assessment because the diameter of these lesions was too small to evaluate their MRI findings. The pathological types of EAOC and non-EAOC were shown in Table 1. There were 14 patients with 14 lesions in the EAOC group and 40 patients with 53 lesions in the non-EAOC group. The most common pathological types of EAOC were ovarian clear cell carcinoma and endometrioid carcinoma (11/14, 78.6%), and the most common histological type of non-EAOC was ovarian serous carcinoma (38 cases, 71.7%). Comparison of the mean age between the two groups was shown in Figure 1. Patients with non-EAOC were about 9 years older than those patients with EAOC in this study. And the mean age ±SD of the two groups were 47.1 ± 11.0 and 56.1 ± 11.9, respectively, (p = 0.017). Besides, the clinical characteristics of the two groups such as laterality of tumor, menopausal state, the coexistence of uterine endometrial carcinoma, serum level of CA125, and FIGO stages were summarized in Table 2. Laterality of tumor showed statistical significance when comparing EAOC and non-EAOC. No bilateral lesions were found in the EAOC group, while found in 16 patients in the non-EAOC groups. And the FIGO stage between the two groups was statistically different (p = 0.027). There are 9 of 14 patients with FIGO stage I EAOC and 11 of 40 patients with FIGO stage I non-EAOC. In the matter of tumor markers among 53 patients, the proportion of CA125 in the low- and medium-level group of the EAOC group was higher than that of the non-EAOC group, but there was no significant difference between the two groups (p = 0.067). No significant difference was found between the two groups in variables such as menopausal status and coexistence of endometrial cancer (p = 0.188, p = 1.000, respectively). Table 1. | Pathological diagnosis | Lesion number | Percentage (%) | |---|---|---| | EAOC | 14 | | | Serous carcinoma | 1 | 7.1 | | Endometrioid carcinoma | 5 | 35.7 | | Clear cell carcinoma | 6 | 42.9 | | Mucinous carcinoma | 1 | 7.1 | | Mixed | 1 | 7.1 | | Non-EAOC | 53 | | | Serous carcinoma | 38 | 71.7 | | Endometrioid carcinoma | 6 | 11.3 | | Clear cell carcinoma | 2 | 3.8 | | Mucinous carcinoma | 7 | 13.2 | | Mixed | 0 | 0 | EAOC, Endometriosis-associated ovarian cancer. Table 2. | Clinical characteristics | EAOC (n = 14） | Non-EAOC (n = 40） | p value | |---|---|---|---| | Laterality | 0.005 | || | Unilateral | 14 | 24 | | | Bilateral | 0 | 16 | | | Menopausal status | 0.188 | || | Postmenopausal | 7 | 29 | | | Premenopausal | 7 | 11 | | | Endometrial carcinoma | 1.000 | || | Yes | 2 | 5 | | | No | 12 | 35 | | | Ca125 | 0.067 | || | ＜35 | 5 | 5 | | | 35–600 | 6 | 14 | | | ＞600 | 3 | 21 | | | FIGO stage | 0.027 | || | I | 9 | 11 | | | II | 4 | 10 | | | III | 1 | 18 | | | IV | 0 | 1 | FIGO, International federation of gynecology and obstetrics. Comparison of MRI findings between EAOC and non-EAOC MRI findings between the two groups were shown in Table 3. The morphology of tumors in the EAOC group frequently appeared to be oval-shaped in comparison with that of the non-EAOC group (11/14, 78.6% vs 23/53, 43.4%, p = 0.033). Hypointensity on T2WI in the cystic components was seen in five lesions of both the EAOC group and non-EAOC group (35.7%, 9.4%, respectively, p = 0.027). Unilocular mass were common in 42.9% (6/14) of EAOC group versus 3.9% (2/53) of non-EAOC group (p = 0.001). Although cystic components predominated in EAOC than non-EAOC (9/14, 64.3% vs 23/53, 43.4%) and solid components common in non-EAOC than EAOC (2/14, 14.3% vs 15/53, 28.3%), no statistical difference on configuration (p = 0.396) occurred between two groups. Focal growth pattern of mural nodules more often occurred in EAOC (p = 0.015). Among 53 lesions of non-EAOC, only four lesions showed focal growth pattern (7.5%). The surface and height of mural nodules did not achieve statistical significance. (p = 0.477, 1.000, respectively). Also, there was no significant difference in the maximum diameter of the tumor, hyperintensity on T1WI, lymphadenopathy, and peritoneal implants between the EAOC group and non-EAOC group. MRI findings of EAOC and non-EAOC were shown in Figures 2 and 3, respectively. Table 3. | MRI features | EAOC | Non-EAOC | p value | |---|---|---|---| | Maximum diameter | 0.370 | || | ≤9 cm | 6 | 31 | | | >9 cm | 8 | 22 | | | Morphology | 0.033 | || | Oral | 11 | 23 | | | Irregular | 3 | 30 | | | Configuration | 0.396 | || | Cystic | 9 | 23 | | | Cystic-solid | 3 | 15 | | | Solid | 2 | 15 | | | Locularity | 0.001 | || | Unilocular | 6 | 2 | | | Multilocular | 8 | 51 | | | Growth pattern of mural nodules | 0.015 | || | Focal | 5 | 4 | | | Multifocal | 9 | 49 | | | Surface of mural nodules | 0.477 | || | smooth | 4 | 9 | | | irregular | 10 | 44 | | | Height of mural nodules | 1.000 | || | 4 cm | 2 | 10 | | | Hyperintensity on T1WI | 0.133 | || | Presence | 8 | 18 | | | Absence | 6 | 35 | | | Hypointensity on T2WI | 0.027 | || | Presence | 5 | 5 | | | Absence | 9 | 48 | | | Lymphadenopathy | 0.664 | || | Presence | 1 | 7 | | | Absence | 13 | 33 | | | Peritoneal Implants | 0.064 | || | Presence | 2 | 17 | | | Absence | 12 | 23 | Significant MRI variables in univariate analysis were included in the multivariate logistic analysis. The results of the multivariate regression analysis were shown in Table 4. As the results showed, locularity was an effective predictive factor of MRI differences for discriminating EAOC (Odds ratio [OR]=21.069, 95% confidence interval [CI]: 3.289–134.947). And hypointensity on T2WI was similarly an effective predictive factor. (Odds ratio [OR]=6.120, 95% confidence interval [CI]: 1.200–31.227). Morphology and growth pattern of mural nodules were eliminated from logistic regression (p = 0.091, 0.403, respectively). Table 4. | MRI features | Standard error | Wald | OR | 95% CI | p value | |---|---|---|---|---|---| | Locularity | 0.948 | 10.347 | 21.069 | 3.289–134.947 | 0.001 | | Hypointensity on T2WI | 0.831 | 4.747 | 6.120 | 1.200–31.227 | 0.029 | CI, Confidence interval; OR, Odds ratio. The diagnostic performance of the unilocular mass, growth pattern of mural nodules, hypointensity on T2WI, and morphology were calculated to distinguish EAOC from non-EAOC. Significant MRI variables were evaluated using sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), and the diagnostic performance was demonstrated in Table 5. Table 5. | MRI features | Sensitivity | Specificity | PPV | NPV | |---|---|---|---|---| | Locularity | 0.429 | 0.962 | 0.750 | 0.864 | | Hypointensity on T2WI | 0.357 | 0.906 | 0.500 | 0.842 | | Growth pattern of mural nodules | 0.555 | 0.925 | 0.555 | 0.845 | | Morphology | 0.786 | 0.566 | 0.324 | 0.909 | NPV, Negative predictive value; PPV, Positive predictive value.

Epithelial ovarian carcinoma is one of the most common and aggressive tumors, which cause more than 151,000 deaths per year.2 Management of EAOCs remains a significant challenge at this stage. Treatments of EAOCs are varied according to histological type and stage of tumor.17 In recent years, primary debulking surgery and chemotherapy are considered to be the first-line treatment for patients with EAOC.6 MRI examination is a reliable non-invasive imaging tool for diagnosis and pre-surgical evaluation of EAOC, which might contribute to the management of EAOC. To our knowledge, this is a novel study in differentiating EAOC from non-EAOC using MRI findings. Several studies focused on the imaging characteristics of the malignant transformation of pelvic endometriosis18 but rarely emphasized the MRI differences between EAOC and non-EAOC. Our results showed that there were some differences in the clinical characteristics and imaging findings between the EAOC and non-EAOC. Clinically, EAOC patients were younger than non-EAOC patients and had an early stage of the tumor (FIGO I/II 13, 92.9%), which is consistent with the results of the study by Li et al.19 Consistent with Scarfone, G et al,20 most EAOC were unilateral. CA125 is a tumor marker for the screening and diagnosis of EOC. No significant difference was found in the CA125 levels between EAOC and non-EAOC.19 However, this result was different from the result of Wang et al, who found that the serum ca125 level of EAOC patients is significantly lower than that of non-EAOC.21,22 The reason might be that serum CA125 is easily interfered with by other factors. CA125 is elevated in both endometriosis and EOC,23,24 and is affected by pathological types of EOC.25 According to a previous study, nearly 10% of ovarian cancer patients have synchronous endometrial cancer.26 Nevertheless, the coexistence of endometrial cancer was found to be not significantly different between the two groups in this study. Although EAOC cannot be directly diagnosed by MRI morphological findings, some imaging findings were more suggestive of EAOC. The typical morphologic characteristics of EAOC appear a unilateral large cystic mass containing hemorrhagic fluid and mural nodules.18 Compared with non-EAOC, EAOC especially clear cell carcinoma appears to be more often unilocular,14 although both of them generally manifest multilocular mass. Non-EAOC rarely showed unilocular appearance. EAOC should be considered if ovarian mass manifests a unilocular appearance. In the current study, only two cases of non-EAOC appeared as unilocular masses. Yasuhito et al reported that papillary projections can be seen in part of endometrioma,12 and we assume that this is related to a focal growth pattern in EAOC. However, non-EAOC, especially HGSC, evolve rapidly and are highly aggressive. This might result in the non-EAOC multifocal growth pattern. Although morphology and growth pattern of mural nodules were statistically significant in univariable analysis, no significant difference was found concerning these two factors in multivariate logistic regression analysis. On multivariate analysis, locularity and hypointensity on T2WI were important independent predictors. Our study showed that MRI findings such as locularity, and hypointensity on T2WI have high specificity and low sensitivity. This probably because typical epithelial ovarian cancer usually presents as a multilocular cystic-solid mass, while EAOC especially ovarian clear cell carcinoma usually presents as unilocular cystic masses.14 Despite its low sensitivity, unilocular masses may be helpful to indicate EAOC. Hypointensity on T2WI of the cyst fluid may be due to bleeding and densely concentrated fluid.27 Studies have pointed out that the loss of T2 shading is a sign of the malignant transformation of endometriosis. The possible cause is that the fluid secreted by the tumor dilutes the bleeding components.27 However, there were five cases (35.7%) that presented hypointensity on T2WI. We considered this signal characteristic may be an independent indicator that helps to distinguish EAOC from non-EAOC. And we speculated that the possible reason is that these cysts still contain bleeding components. Alizadeh et al demonstrated that serum iron level in patients with endometriosis was higher than those without endometriosis.28 Contents of endometriotic cysts, especially iron, lead to persistent oxidative stress and therefore induced carcinogenesis.2 The iron level might differ between EAOC and non-EAOC. Ma et al29 pointed out that high intensity on T1WI is one of the MRI findings for distinguishing CCC from HGSC. This signal feature may be caused by bleeding associated with endometriosis.30 The most frequent pathological type of EAOC is CCC and the most common pathological type of traditional EOC is high-grade serous carcinoma. In our study, hyperintensity on T1WI was seen in 57.1% of EAOC, while only 34.0% of non-EAOC showed this signal characteristic. However, T1WI hyperintensity could not distinguish EAOC from non-EAOC (p = 0.133). The reason might be mucus components can appear high signal on T1WI besides bleeding components.11 The mucinous tumors in non-EAOC contain thick mucinous material, which may increase the intensity of the T1 signal in the cyst fluid. Studies show that malignant tumors related to endometriosis rarely occur in peritoneal implants.16 And clear cell carcinoma and endometroid carcinoma arising from endometriosis tend to present with lower percentages of lymph node metastasis compared with tumors without endometriosis.22 Only one case presented with lymphadenopathy in this EAOC group, no significant difference was found between the two groups in lymphadenopathy and peritoneal implants. This might because lymph node size is not a good indicator in determining lymph node metastasis of epithelial ovarian cancer.31 Kobayashi et al show that the size of the endometriomas (>9 cm) is an independent predictor of the development of ovarian cancer.32 However, no significant difference in the maximum diameter of the tumor between the two groups. The mean size of both groups was larger than 9 cm in this study. The clinical characteristics and prognosis between EAOC and non-EAOC are different. Thus, it is important to make an accurate diagnosis of EAOC. In our study, age at diagnosis, laterality and FIGO stages were different between EAOC and non-EAOC. Based on MRI imaging, locularity and hypointensity on T2WI played an important role in the differential diagnosis. When the mass presented as a unilateral, unilocular mass with hypointensity on T2WI, it tended to be EAOC. And the typical appearance of non-EAOC was unilateral or bilateral multilocular irregular mass without hypointensity on T2WI. Notably, the lack of mural nodule in an early-stage tumor can cause a great challenge in the diagnosis of ovarian tumors. One of our cases pathologically diagnosed with EAOC showed no obvious mural nodule. And we must acknowledge that it is difficult to diagnose EAOC or even ovarian tumor based on morphological MRI features in patients without the presence of mural nodules within a cystic mass. Therefore, it is important to find an alternative imaging method for recognizing early-stage EAOC. Chiharu et al found that EAOC exhibited lower in vivo R2 values and total iron levels compared to benign ovarian endometriosis.33 MR relaxometry may be a useful non-invasive imaging method to predict the malignant change of ovarian endometriosis. In our study, Hypointensity on T2WI was an effective predictive factor for discriminating EAOC from non-EAOC. We assume this MRI signal feature might be associated with blood material. Iron levels of cyst fluid may differ between EAOC and non-EAOC. Therefore, prospectively study about differentiating EAOC from non-EAOC using MR transverse relaxation rate technique seems promising. Also, this method might help to identify early-stage EAOC without obvious papillary projection. Optimized MRI protocol contributes to the assessment of adnexal masses. For masses with T1 high signal intensity, the fat-suppressed sequence helps identify fat and blood. DWI has clinical value in differentiating the benign and malignant adnexal masses. Malignant tumors show restricted diffusion. A rapid rate of enhancement in contrast-enhanced MRI of adnexal masses has a high likelihood of malignant lesions. Thus, a dedicated MR protocol is important to analyze and record the imaging features. DWI and DCE MRI are recommended for MR imaging of adnexal cystic-solid masses.34 Although the use of anti-peristaltic agents can decrease bowel movement artifacts and improve the image quality of pelvic MRI, it was mainly used in gastrointestinal MR examinations. Some European Society of Urogenital Radiology (ESUR) MRI experts did not use it.35 Hence, it was not generally applied in pelvic MR scanning and was not used in our study. Our research had four main limitations. Firstly, this is a retrospective study, and only patients who underwent MRI were included in our study. The selective use of MRI leads to selective bias. Secondly, we did not subgroup EAOC by histological subtypes due to its distinct clinical characteristics and lack of sufficient sample size. Further investigations according to histological subtypes should be conducted. Compared with non-EAOC, EAOC is relatively rare. An adequate sample capacity is required. Thirdly, differences in MRI findings might be caused by diverse stages between the two groups, thus further study is required with the study population limited to FIGO stage I/II. Fourthly, dynamic contrast-enhanced MRI and functional imaging such as DWI was not discussed in our study. Prospective investigation on the differentiation between EAOC and non-EAOC should be conducted, using DWI-based ADC value and semiquantitative DCE-MRI parameters.

Although several MRI findings were overlapped, MRI morphological and signal characteristics may help to distinguish EAOC from non-EAOC. EAOC typically presented as a unilocular mass with hypointensity on T2WI in cystic components. MRI could help distinguish EAOC from non-EAOC. Contributor Information Ximing Zhang, Email: [email protected]. Min Li, Email: [email protected]. Zhuopeng Tang, Email: [email protected]. Xinyi Li, Email: [email protected]. Ting Song, Email: [email protected].

- 1.Webb PM, Jordan SJ. Epidemiology of epithelial ovarian cancer. Best Pract Res Clin Obstet Gynaecol 2017; 41: 3–14. doi: 10.1016/j.bpobgyn.2016.08.006 [DOI] [PubMed] [Google Scholar] - 2.Kajiyama H, Suzuki S, Yoshihara M, Tamauchi S, Yoshikawa N, Niimi K, et al. Endometriosis and cancer. Free Radic Biol Med 2019; 133: 186–92. doi: 10.1016/j.freeradbiomed.2018.12.015 [DOI] [PubMed] [Google Scholar] - 3.Wilbur MA, Shih I-M, Segars JH, Fader AN. Cancer implications for patients with endometriosis. Semin Reprod Med 2017; 35: 110–6. doi: 10.1055/s-0036-1597120 [DOI] [PubMed] [Google Scholar] - 4.Leiserowitz GS, Gumbs JL, Oi R, Dalrymple JL, Smith LH, Ryu J, et al. Endometriosis-related malignancies. Int J Gynecol Cancer 2003; 13: 466–71. doi: 10.1136/ijgc-00009577-200307000-00010 [DOI] [PubMed] [Google Scholar] - 5.Vergote I, Tropé CG, Amant F, Kristensen GB, Ehlen T, Johnson N, et al. Neoadjuvant chemotherapy or primary surgery in stage IIIC or IV ovarian cancer. N Engl J Med 2010; 363: 943–53. doi: 10.1056/NEJMoa0908806 [DOI] [PubMed] [Google Scholar] - 6.Su K-M, Wang P-H, Yu M-H, Chang C-M, Chang C-C. The recent progress and therapy in endometriosis-associated ovarian cancer. J Chin Med Assoc 2020; 83: 227–32. doi: 10.1097/JCMA.0000000000000262 [DOI] [PMC free article] [PubMed] [Google Scholar] - 7.Togashi K. Ovarian cancer: the clinical role of US, CT, and MRI. Eur Radiol 2003; 13 Suppl 4(Suppl 4): L87–104. doi: 10.1007/s00330-003-1964-y [DOI] [PubMed] [Google Scholar] - 8.Vargas HA, Barrett T, Sala E. Mri of ovarian masses. J Magn Reson Imaging 2013; 37: 265–81. doi: 10.1002/jmri.23721 [DOI] [PubMed] [Google Scholar] - 9.Tanaka YO, Okada S, Satoh T, Matsumoto K, Oki A, Saida T, et al. Differentiation of epithelial ovarian cancer subtypes by use of imaging and clinical data: a detailed analysis. Cancer Imaging 2016; 16: 3. doi: 10.1186/s40644-016-0061-9 [DOI] [PMC free article] [PubMed] [Google Scholar] - 10.Sohaib SAA, Reznek RH. Mr imaging in ovarian cancer. Cancer Imaging 2007; 7 Spec No A: S119–29. doi: 10.1102/1470-7330.2007.9046 [DOI] [PMC free article] [PubMed] [Google Scholar] - 11.Jung SE, Lee JM, Rha SE, Byun JY, Jung JI, Hahn ST. Ct and MR imaging of ovarian tumors with emphasis on differential diagnosis. Radiographics 2002; 22: 1305–25. doi: 10.1148/rg.226025033 [DOI] [PubMed] [Google Scholar] - 12.Tanase Y, Kawaguchi R, Takahama J, Kobayashi H. Factors that differentiate between endometriosis-associated ovarian cancer and benign ovarian endometriosis with mural nodules. Magn Reson Med Sci 2018; 17: 231–7. doi: 10.2463/mrms.mp.2016-0149 [DOI] [PMC free article] [PubMed] [Google Scholar] - 13.Nishio N, Kido A, Kataoka M, Kuwahara R, Nakao K, Kurata Y, et al. Longitudinal changes in magnetic resonance imaging of malignant and borderline tumors associated with ovarian endometriotic cyst comparing with endometriotic cysts without arising malignancy. Eur J Radiol 2018; 105: 175–81. doi: 10.1016/j.ejrad.2018.06.007 [DOI] [PubMed] [Google Scholar] - 14.Li M, Tan J, Zhang Y, Ai C, Wang H, Zhang H, et al. Assessing CT imaging features combined with CEA and CA125 levels to identify endometriosis-associated ovarian cancer. Abdom Radiol 2020;18 May 2020. doi: 10.1007/s00261-020-02571-x [DOI] [PubMed] [Google Scholar] - 15.Robinson KA, Menias CO, Chen L, Schiappacasse G, Shaaban AM, Caserta MP, et al. Understanding malignant transformation of endometriosis: imaging features with pathologic correlation. Abdom Radiol 2020; 45: 1762–75. doi: 10.1007/s00261-019-01914-7 [DOI] [PubMed] [Google Scholar] - 16.McDermott S, Oei TN, Iyer VR, Lee SI. MR imaging of malignancies arising in endometriomas and extraovarian endometriosis. Radiographics 2012; 32: 845–63. doi: 10.1148/rg.323115736 [DOI] [PubMed] [Google Scholar] - 17.Pejovic T, Thisted S, White M, Nezhat FR. Endometriosis and endometriosis-associated ovarian cancer (EAOC. Adv Exp Med Biol 2020; 1242: 73–87. doi: 10.1007/978-3-030-38474-6_5 [DOI] [PubMed] [Google Scholar] - 18.Takeuchi M, Matsuzaki K, Uehara H, Nishitani H. Malignant transformation of pelvic endometriosis: MR imaging findings and pathologic correlation. Radiographics 2006; 26: 407–17. doi: 10.1148/rg.262055041 [DOI] [PubMed] [Google Scholar] - 19.Li Q, Sun Y, Zhang X, Wang L, Wu W, Wu M, et al. Endometriosis-Associated ovarian cancer is a single entity with distinct clinicopathological characteristics. Cancer Biol Ther 2019; 20: 1029–34. doi: 10.1080/15384047.2019.1595278 [DOI] [PMC free article] [PubMed] [Google Scholar] - 20.Scarfone G, Bergamini A, Noli S, Villa A, Cipriani S, Taccagni G, et al. Characteristics of clear cell ovarian cancer arising from endometriosis: a two center cohort study. Gynecol Oncol 2014; 133: 480–4. doi: 10.1016/j.ygyno.2014.03.017 [DOI] [PubMed] [Google Scholar] - 21.Wang S, Qiu L, Lang JH, Shen K, Yang JX, Huang HF, et al. Clinical analysis of ovarian epithelial carcinoma with coexisting pelvic endometriosis. Am J Obstet Gynecol 2013; 208: 413.e1–413.e5. doi: 10.1016/j.ajog.2012.12.004 [DOI] [PubMed] [Google Scholar] - 22.Paik ES, Kim TJ, Choi CH, Kim BG, Bae DS, Lee JW. Clinical outcomes of patients with clear cell and endometrioid ovarian cancer arising from endometriosis. J Gynecol Oncol 2018; 29: e18. doi: 10.3802/jgo.2018.29.e18 [DOI] [PMC free article] [PubMed] [Google Scholar] - 23.Bast RC, Klug TL, St John E, Jenison E, Niloff JM, Lazarus H, et al. A radioimmunoassay using a monoclonal antibody to monitor the course of epithelial ovarian cancer. N Engl J Med 1983; 309: 883–7. doi: 10.1056/NEJM198310133091503 [DOI] [PubMed] [Google Scholar] - 24.Foster W. Diagnosing endometriosis: CA125 rules in, but not out. BJOG 2016; 123: 1769. doi: 10.1111/1471-0528.14224 [DOI] [PubMed] [Google Scholar] - 25.Zhang L, Chen Y, Wang K. Comparison of CA125, HE4, and Roma index for ovarian cancer diagnosis. Curr Probl Cancer 2019; 43: 135–44. doi: 10.1016/j.currproblcancer.2018.06.001 [DOI] [PubMed] [Google Scholar] - 26.Singh N. Synchronous tumours of the female genital tract. Histopathology 2010; 56: 277–85. doi: 10.1111/j.1365-2559.2009.03367.x [DOI] [PubMed] [Google Scholar] - 27.Tanaka YO, Yoshizako T, Nishida M, Yamaguchi M, Sugimura K, Itai Y. Ovarian carcinoma in patients with endometriosis: MR imaging findings. AJR Am J Roentgenol 2000; 175: 1423–30. doi: 10.2214/ajr.175.5.1751423 [DOI] [PubMed] [Google Scholar] - 28.Alizadeh M, Mahjoub S, Esmaelzadeh S, Hajian K, Basirat Z, Ghasemi M. Evaluation of oxidative stress in endometriosis: a case-control study. Caspian J Intern Med 2015; 6: 25–9. [PMC free article] [PubMed] [Google Scholar] - 29.Ma F-H, Qiang J-W, Zhang G-F, Li H-M, Cai S-Q, Rao Y-M. Magnetic resonance imaging for distinguishing ovarian clear cell carcinoma from high-grade serous carcinoma. J Ovarian Res 2016; 9: 40. doi: 10.1186/s13048-016-0251-x [DOI] [PMC free article] [PubMed] [Google Scholar] - 30.Choi HJ, Lee J-H, Seok Lee J, Choi J-I, Kang S, Lee S, et al. Ct findings of clear cell carcinoma of the ovary. J Comput Assist Tomogr 2006; 30: 875–9. doi: 10.1097/01.rct.0000220795.45782.1d [DOI] [PubMed] [Google Scholar] - 31.Tangjitgamol S, Manusirivithaya S, Sheanakul C, Leelahakorn S, Sripramote M, Thawaramara T, et al. Can we rely on the size of the lymph node in determining nodal metastasis in ovarian carcinoma? Int J Gynecol Cancer 2003; 13: 297–302. doi: 10.1046/j.1525-1438.2003.13192.x [DOI] [PubMed] [Google Scholar] - 32.Kobayashi H. Ovarian cancer in endometriosis: epidemiology, natural history, and clinical diagnosis. Int J Clin Oncol 2009; 14: 378–82. doi: 10.1007/s10147-009-0931-2 [DOI] [PubMed] [Google Scholar] - 33.Yoshimoto C, Takahama J, Iwabuchi T, Uchikoshi M, Shigetomi H, Kobayashi H. Transverse relaxation rate of cyst fluid can predict malignant transformation of ovarian endometriosis. Magn Reson Med Sci 2017; 16: 137–45. doi: 10.2463/mrms.mp.2016-0028 [DOI] [PMC free article] [PubMed] [Google Scholar] - 34.Forstner R, Thomassin-Naggara I, Cunha TM, Kinkel K, Masselli G, Kubik-Huch R, et al. ESUR recommendations for MR imaging of the sonographically indeterminate adnexal mass: an update. Eur Radiol 2017; 27: 2248–57. doi: 10.1007/s00330-016-4600-3 [DOI] [PMC free article] [PubMed] [Google Scholar] - 35.Sheikh-Sarraf M, Nougaret S, Forstner R, Kubik-Huch RA. Patient preparation and image quality in female pelvic MRI: recommendations revisited. Eur Radiol 2020; 30: 5374–83. doi: 10.1007/s00330-020-06869-8 [DOI] [PubMed] [Google Scholar]