Ovarian sex cord stromal tumors are rare tumors that account for approximately 7% of all ovarian tumors. According to the 2014 World Health Organization (WHO) ovarian tumor histological classification, these tumors are divided into pure stromal tumors, pure sex cord tumors, luteinized thecoma associated with sclerosing peritonitis, and mixed sex cord stromal tumors. Pure stromal tumors include three subtypes: fibroma, cellular fibroma, and thecoma; these are mainly distinguished based on whether they comprise theca cells, lutein cells, fibroblasts, and fibrocytes. This group of tumors has overlapping features in multidirectional differentiation through histology, which makes it difficult to obtain a pathological diagnosis in some cases. Therefore, these tumors are traditionally named ovarian thecoma-fibrothecoma (OTCA–FTCA) (1, 2).

OTCA–FTCA and ovarian granulosa cell tumors (OGCTs) are the most common sex cord stromal tumors and have a low incidence relative to other ovarian tumors. These tumors are usually discovered by chance during gynecological examinations or routine physical examinations as the symptoms are nonspecific. OTCA–FTCA accounts for 0.5–1.0% of ovarian tumors, is generally benign, and has an excellent prognosis after resection, but a small proportion of these tumors (1.0–5.0%) are malignant (3, 4). OGCTs are rare sex cord stromal tumors with a low malignant potential and account for only 5% of all malignant ovarian tumors, with adult and juvenile forms of subtypes. OGCTs have a low degree of malignancy, show growth patterns of benign tumors, and have potentially malignant behaviors, including local invasion, recurrence, and metastasis (5, 6). OTCA–FTCA is mainly found in menopausal women, and less than 10% occur before age 30 (7, 8); however, OGCTs are more common in postmenopausal women, and the juvenile type is rare and typically occurs before 30 years of age. Sometimes these tumors share similar clinical manifestations (such as elevated estrogen levels leading to endometrial hyperplasia and irregular vaginal bleeding). These tumors can have similar imaging findings, such as combined with cystic degeneration, edema, and hemorrhage, which may cause misdiagnosis in radiography and inappropriate choice of treatment of clinicians (9–11). Therefore, the preoperative diagnosis of OTCA–FTCA and OGCTs is particularly important.

Magnetic resonance imaging (MRI) has high resolution in soft tissues that clearly reveal the lesion characteristics, relationship between the tumor and surrounding tissues, and the status of lymph node disease (9). In particular, the semiquantitative parameters deprived from diffusion-weighted imaging (DWI) have gradually become one of the important tools for evaluating ovarian tumors (12). Texture analysis (TA) has been widely adopted in the differential diagnosis of tumors in recent years and is considered to be an effective means to assess tumor heterogeneity. Not only MRI-based texture analysis but also CT texture-based analysis of the whole tumor has demonstrated high sensitivity and specificity for the characterization of ovarian tumors and may assist in characterizing the differences in ovarian tumor patients. The application of MRI-based texture features combined with conventional MRI features may assist in improving the differentiation of ovarian tumors. These findings, in turn, may guide diagnostic protocols for future patients and can help radiologists make appropriate follow-up decisions (3, 4, 7).

Therefore, the purpose of this study was to identify the best features for distinguishing between OTCA–FTCA and OGCTs through conventional MRI, TA, and the combination of the two diagnostic methods to improve the accuracy of preoperative imaging-based diagnoses and help clinicians choose appropriate treatment methods.

OTCA–FTCA and OGCTs are the most common sex cord stromal tumors and have a low incidence relative to other ovarian tumors. The radiological knowledge of those rare ovarian tumors is still limited in the reported literature; furthermore, the imaging findings of the two entities are similar. Herein we performed a retrospective review of the MRI findings of 32 patients with OTCA–FTCA and 14 patients with OGCTs in this study at our single institution within 7 years. To the best of our knowledge, this is the first study to describe the detailed MRI sign and TA characteristic in this samples.

In our study, the clinical characteristics [age (years), size (maximum), size (average), menopausal status, presence of endometrial hyperplasia, and location] were compared, and there were no significant differences between the two tumors, indicating that they have similar clinical characteristics, as shown in Table 2 . Combined with literature reports, we found the following: (1) The incidence of intratumoral hemorrhage in this group of OGCTs was as high as 76%, which is higher than that reported in the literature (13). Intratumoral hemorrhage mainly manifested as high signal on T1WI and high signal or low signal on T2WI, and the fluid–fluid level due to hemorrhage could be seen in some lesions. In comparison, the incidence of intratumoral hemorrhage in OTCA–FTCA was only 16% (5/32). The multivariate logistic regression analysis found that the presence of intratumoral hemorrhage could help diagnose OGCTs (OR = 0.12, 95% CI: 0.001–0.284), which is consistent with previous reports that intratumoral hemorrhage is a typical feature of these tumors (14); (2) OTCA–FTCA is composed of theca cells, lutein cells, and fibroblasts. This group of tumors is prone to secondary degenerative changes, such as tumor stromal edema and mucinous degeneration, which may lead to high ADC values (1.50 ± 0.32 × 10³ mm²/s). In contrast, OGCTs are low-grade malignant tumors that histologically show diffuse, island, beam, follicular, and sarcoma-like growth patterns. These patterns often exist mixed, and the relatively tight arrangement results in more restricted water molecule diffusion with lower ADC values (1.27 ± 0.37 × 10³ mm²/s) than that of OTCA–FTCA. Therefore, the average ADC value was significantly different between the two tumors (Z = -1.982, P = 0.047) (15). When the ADC value was ≤1.34 × 10³ mm²/s, its sensitivity for diagnosing OGCTs was 71.34%, and the specificity was 65.62% (AUC = 0.685, 95% CI: 0.532 to 0.814, P = 0.048); (3) In our group, 94% (30/32) of the OTCA–FTCA tumors were mildly enhanced, 6% (2/32) were moderately enhanced, and none showed marked enhancement. In comparison, 57% (8/14) of the OGCTs were mildly enhanced, 43% were moderately enhanced, and none showed marked enhancement. There was a significant difference in the degree of enhancement between the two tumor types (OR = 0.89, 95% CI: 0.015–0.527). It is possible that OTCA–FTCA contains fibrous components, resulting in a lower blood supply and lower enhancement than OGCTs. This is also consistent with previous reports that OTCA–FTCA tumors have a low blood supply, resulting in mild enhancement on MRI (16–18); (4) OGCTs are mostly solid or cystic–solid, and it has been reported in the literature that a “honeycomb” or “sponge” cyst is the characteristic imaging manifestation (19). OTCA–FTCA is often prone to secondary cystic transformation when the tumor volume is large. Some scholars have reported that the cystic transformation rate is 76% (19/25) (20), so the tumor often appears as a cystic–solid or cystic mass, which may be preoperatively misdiagnosed as OGCTs or other ovarian tumor. Other scholars have divided these tumors into solid, cystic, and cystic–solid masses according to the degree of the cystic component. Cystic–solid masses are divided into intratumoral cysts and extratumoral cysts according to whether the cysts are located in the tumor. Intratumoral cysts are divided into peripheral, central, and diffuse types according to their location. The study showed that peritumoral cysts are a characteristic MRI sign (21). In our study, the types of cysts were divided into five degrees according to the degree of cystic degeneration (no cyst: 0°, 0–25%: 1°, 25–50%: 2°, 50–75%: 3°, and greater than 75%: 4°), and the forms of cystic transformation were divided into four forms (no cystic transformation, small cyst, large cyst, and mixed). Between the two tumor types, there was no significant difference in the degree of cystic transformation (Fisher = 0.149, P = 0.229), but there was a significant difference in the form of cystic transformation (Fisher = 0.006, P = 0.008), indicating that OGCTs mainly demonstrated mixed cystic changes, while OTCA–FTCA predominantly exhibited macrocystic changes. In this study, a weak correlation existed between tumor size and the degree of cystic transformation in the OGCTs group (Kendall’s tau-b = 0.618, P < 0.001), and no correlation was observed in the OTCA–FTCA group (Kendall’s tau-b = -0.025, P = 0.857). It is inconsistent with related reports (17) and may be caused by the small sample size.

In the multivariate logistic regression analysis, the IBD model established had an AUC of 0.935, and its sensitivity, specificity, and Youden index were 85.71%, 93.75%, and 0.794 (95% CI: 0.822 to 0.987, P < 0.0001), respectively, so the significant features, such as the mean ADC value, enhancement degree, and presence of intratumoral hemorrhage, were important predictors to distinguish between OGCTs and OTCA–FTCA.

TA is different from traditional empirical image analysis based on observations with the naked eye. TA can provide a large amount of imaging information that cannot be recognized by the naked eye by quantitatively analyzing the grayscale information of medical images, realizing the conversion from images to data, and constructing labels to describe the details of the lesion features. Thus, this information could be of value in helping clinicians develop reasonable treatment strategies (22). In recent years, TA has been regarded as an effective means to assess tumor heterogeneity. This method can be used to evaluate the gray-level intensity and position of the pixels within an image to derive texture features that provide a measure of intralesional heterogeneity. TA data are easy to obtain, and no additional imaging is required. In addition, TA plays a relatively important role in evaluating clinical curative effects and predicting prognosis. Many researchers have conducted excellent research, especially with radiomics, in predicting the development trends of tumor lesions (23, 24). The TA in the present study is based on the T2WI sequence because conventional T2WI can reveal the rich histopathological characteristics of tumors, for example, by determining the water content, degree of fibrotic change, necrosis, and hemorrhage (15).

As shown in Table 4 , the univariate analysis demonstrated that six texture features were significantly different between the OGCTs and OTCA–FTCA groups (P < 0.05). Among the six features, the log-sigma-2-0-mm-3D_glszm_SmallAreaEmphasis, log-sigma-2-0-mm-3D_glszm_SmallAreaHighGrayLevelEmphasis, and log-sigma-3-0-mm-3D_glcm_InverseVariance were derived from the image transform type of Laplacian of Gaussian. The wavelet-LLH_glcm_MCC, wavelet-HLH_glszm_SmallAreaHighGrayLevelEmphasis, and wavelet-HLL_glszm_LowGrayLevelZoneEmphasis were derived from the image transform type of wavelet. The features belong to the gray level co-occurrence matrix (GLCM), and the gray level size zone matrix can assess the second-order joint probability function and quantify gray level zones in the image (25). A gray level zone is defined as the number of connected voxels that share the same gray level intensity (26). In the multivariate logistic regression analysis with the forward step method, we found that two features from the image transform type of Laplacian of Gaussian—log-sigma-2-0-mm-3D_glszm_SmallAreaEmphasis and Log-sigma-20mm-3D_glszm_SmallAreaHighGrayLevelEmphasis—are independent risk predictors for distinguishing between OGCTs and OTCA–FTCA (P < 0.05). The Laplacian operator can highlight areas in the image where the intensity changes rapidly. The log-sigma-2-0-mm-3D_glszm_SmallAreaEmphasis and log-sigma-20mm-3D_glszm_SmallAreaHighGrayLevelEmphasis describe the distribution of small size zones and the proportion of the joint distribution of smaller size zones with higher gray level values, respectively (27). In our study, the log-sigma-2-0-mm-3D_glszm_SmallAreaEmphasis value of OTCA–FTCA was significantly lower than that of OGCTs, which means that smaller size zones and fine textures were observed in the solid lesions of OTCA–FTCA composed of similar theca cells, lutein cells, and fibroblasts (18). OTCA–FTCA also had significantly less intratumoral hemorrhage than OGCTs in the present study (P < 0.05). For the log-sigma-20mm-3D_glszm_SmallAreaHighGrayLevelEmphasis value, the OGCTs obviously contained a greater proportion of the joint distribution of smaller size zones with higher gray level values on T2WI scans than OTCA–FTCA (P < 0.05). The log-sigma-20mm-3D_glszm_SmallAreaHighGrayLevelEmphasis is a quantitative index used to compensate for the shortage of MRI findings on T2WI based on solid or cystic components that can be compared. Then, the TA-based predictive model was obtained and had a diagnostic performance/AUC, specificity, and sensitivity of 0.944, 92.86%, and 93.75%, respectively (P < 0.05).

The AUC of the IBD and TA combined prediction model to distinguish between OGCTs and OTCA–FTCA was 0.969. When compared with MRI features or TA parameters alone, the combined model showed no significant difference, even though the sensitivity and specificity of the combination were improved to some extent. Therefore, it is believed that the diagnostic performance of the combination model was similar to that of MRI-IBD or T2WI-TA features alone. Furthermore, the diagnostic performance of T2WI-TA parameters was similar to that of MRI-IBD features in helping to distinguish between OGCTs and OTCA–FTCA, which may be less strongly associated with the sample size. However, TA can provide another method to identify OGCTs and OTCA–FTCA.

The present study has several limitations. First of these is the limited study sample size (14 patients with OGCTs and 32 patients with OTCA–FTCA) due to the low incidence of these tumors relative to other ovarian tumors. It might have influence on the final results, such as the rigor of ROC curve analysis. Second, there was an inherent selection bias because the retrospective study was conducted in one institution. We urge the clarification of the imaging findings in larger population-based studies. Third, the ROIs of the ADC and TA were performed manually by radiologists based on individual habits, which may also have influence on the final results. In addition, we did not use MRI images other than T2-weighted images for TA in the present study.

In summary, compared with OTCA–FTCA, OGCTs more commonly exhibit intratumoral hemorrhage, mixed cystic degeneration, moderate enhancement, and low ADC values. Particularly, intratumoral hemorrhage may be a common and characteristic MR finding of OGCTs. When it is difficult to distinguish between OGCTs and OTCA–FTCA, TA described here may serve as a supplementary means, although this will require further large sample size validation before widespread implementation in clinical practice.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the ethics committee of the First Affiliated Hospital of USTC, Division of Life Sciences and Medicine, University of Science and Technology of China. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

N-YL, J-ND, and C-BW contributed to conception and design. N-YL organized the database, performed the statistical analysis, and wrote the first draft of the manuscript. BS, P-PW, Y-LC, C-BW, and YC contributed to the collection and arrangement of data. Y-LC, Y-QG and P-PW contributed to data analysis. BS, P-PW, and C-BW wrote sections of the manuscript. All authors contributed to manuscript revision and read and approved the submitted version.

This work was supported by the Anhui Provincial Natural Science Foundation for Youths, China (1908085QH364).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.