Exosome ocs combined with o-rads for appendiceal mass risk assessment

Abstract

The application discloses an adnexal mass benign and malignant risk assessment method based on exosome OCS and O-RADS combination. The biomarker combination provided by the application comprises ovarian cancer OCS score and imaging grading score, and the ovarian cancer OCS score is obtained based on the concentration of CA125 protein, HE4 protein and C5a protein in exosomes. The application first uses the OCS score in combination with the O-RADS imaging score, and by comprehensively combining the biological information and the imaging characteristics, the accurate diagnosis capability for the O-RADS 3 / 4 gray area patients is significantly improved, and unnecessary excessive medical treatment and missed diagnosis are reduced. The application also includes various clinical variables such as age, menopausal status and ultrasonic characteristics, and constructs a multi-dimensional combined diagnosis model, and further improves the diagnosis performance.

Description

Technical Field

[0001] This invention belongs to the field of biomedical detection technology, specifically relating to a method, system, and reagent kit for assessing the benign or malignant risk of adnexal masses by combining exosome biomarker scoring and imaging grading. Background Technology

[0002] Ovarian cancer is the leading cause of death among gynecological malignancies. Because early-stage symptoms are often subtle, many patients are diagnosed at an advanced stage, missing the optimal window for treatment. According to the Global Cancer Burden Report, ovarian cancer causes more than 200,000 women's deaths annually. The 5-year survival rate for ovarian cancer at early diagnosis can reach over 90%, but for patients with advanced-stage ovarian cancer, the 5-year survival rate is only around 30%. Therefore, early diagnosis is extremely important for improving the survival rate of ovarian cancer patients.

[0003] Currently, clinical diagnostic methods for ovarian cancer include imaging examinations (such as ultrasound, CT, and MRI), serum biomarker testing (such as serum CA125 and HE4), and pathological examination. However, these methods still face significant challenges in early screening and accurate diagnosis, especially in effectively differentiating ovarian cancer from numerous benign and malignant lesions.

[0004] Imaging examinations, especially ultrasound, are currently the primary means of ovarian cancer diagnosis. O-RADS (Ovarian-Adnexal Risk of Malignancy Algorithm), a scoring system in ultrasound imaging, is widely used in clinical practice to assess the malignancy risk of ovarian or adnexal masses. The O-RADS system scores masses based on imaging characteristics (such as cystic / solid nature, blood flow, mass size, and morphology), classifying ovarian masses into different risk levels (2-5) to provide a reference for clinical decision-making. However, this system has significant limitations in assessing some intermediate-risk cases (such as O-RADS 3 and O-RADS 4).

[0005] The O-RADS 3 / 4 Gray Zone Issue: For patients with O-RADS 3 and 4, the accuracy of the imaging scoring system is affected by operator experience, equipment quality, and image interpretation standardization issues, increasing the likelihood of misdiagnosis and missed diagnosis. The nature of the lesions in these gray zone patients often cannot be definitively diagnosed based on imaging characteristics, posing significant challenges to clinical decision-making, especially in deciding whether to proceed with surgery or continue observation. Furthermore, imaging examinations themselves do not provide sufficient biological information, while the malignancy of ovarian cancer is closely related to its molecular biological characteristics. Although imaging examinations can observe morphological changes in tumors, they have inherent limitations in identifying tumor molecular markers and detecting early lesions.

[0006] Serum biomarkers, especially CA125 and HE4, are commonly used methods for ovarian cancer diagnosis. Serum CA125 is a classic biomarker for ovarian cancer, with elevated levels in approximately 80% of ovarian cancer patients. However, the false-positive rate of serum CA125 is high, especially in benign gynecological diseases (such as endometriosis, ovarian cystadenoma, and uterine fibroids), where serum CA125 levels may also be elevated, leading to misdiagnosis. Furthermore, serum CA125 is not ideal in early-stage ovarian cancer; many patients with early-stage lesions maintain normal serum CA125 levels, thus missing the opportunity for early diagnosis.

[0007] To compensate for the inadequacy of serum CA125, serum HE4 (human epithelial protein 4) is often used clinically as an auxiliary diagnostic marker. Serum HE4 shows good sensitivity and specificity in certain types of ovarian cancer, especially for serous ovarian cancer. However, serum HE4 still has certain limitations, mainly manifested in its overall unstable sensitivity and false positives in benign diseases and other tumors.

[0008] In summary, the use of serum CA125 and serum HE4 as biomarkers alone is unlikely to effectively improve the early diagnosis rate of ovarian cancer, and may result in false positives in non-malignant diseases, leading to unnecessary surgery.

[0009] In recent years, exosomes have shown great potential as a novel biomarker in the early diagnosis and monitoring of cancer. Exosomes are nanoscale vesicles actively secreted by tumor cells, stably carrying tumor-related molecules (such as miRNAs, proteins, and lipids). These molecules can be detected in bodily fluids such as blood, thus providing a new approach for early cancer detection and monitoring. The Ovarian Cancer Score (OCS), an exosome-based scoring system for ovarian cancer, has shown promising clinical application prospects in the auxiliary diagnosis of ovarian cancer by detecting three biomarkers in exosomes: CA125, HE4, and C5a. However, current technologies still lack accurate and effective diagnostic methods to address the insufficient ability to differentiate between malignant and benign lesions in some individuals with unclear imaging findings (such as those with O-RADS 3 / 4), leading to missed diagnoses and overtreatment. Summary of the Invention

[0010] The purpose of this invention is to develop a new and more accurate combined diagnostic model using statistical and machine learning methods, combining exosome markers (OCS) and imaging grading (such as O-RADS), for assessing the benign or malignant risk of adnexal masses, solving the problem of accurate stratification of the gray area population on imaging, improving the early diagnosis capability of ovarian cancer, and reducing unnecessary surgery and missed diagnosis.

[0011] The technical solution adopted by the present invention to achieve the above objectives is as follows: This invention provides a combination of biomarkers for assessing the risk of benign or malignant adnexal masses, the combination of biomarkers including: ovarian cancer OCS score and imaging grade score.

[0012] As a preferred embodiment, the ovarian cancer OCS score is calculated based on the concentrations of CA125, HE4, and C5a proteins in exosomes.

[0013] As a preferred embodiment, the radiographic grading score includes O-RADS score, IOTA score, ADNEX model score, RMI score, or artificial intelligence-based radiographic analysis score.

[0014] The present invention also provides a system for assessing the benign or malignant risk of accessory blocks, the system comprising the following modules: The data acquisition module is used to acquire the imaging grading score data of the subjects and generate the OCS score based on the expression level data of exosomal CA125 protein, HE4 protein and C5a protein in the subjects' biological samples. The risk scoring module is used to input the acquired radiographic grading data and OCS score data into the constructed risk scoring model to calculate the risk probability value. The result determination module is used to compare the calculated risk probability value with a preset reference value and output the risk assessment result of the benign or malignant mass of the subject's appendage.

[0015] As a preferred embodiment, the formula for the risk scoring model is as follows: , X i This indicates that the model calculates the numerical result based on the biomarker expression of sample i, P(X). i ) represents the probability value of sample i to be cancer predicted by the scoring model; samples with a probability value greater than the reference value are considered to be ovarian cancer patients and are represented by 1; samples with a probability value less than the reference value are likely to be non-cancer individuals and are represented by 0, which is the final prediction result.

[0016] As a preferred embodiment, the risk scoring model is constructed using one or more machine learning algorithms selected from random forest, extreme random tree, neural network, gradient boosting machine, and linear model.

[0017] As a preferred implementation, an OCS score is generated using a validated calculation algorithm. An OCS score ≥ 0.506 is considered positive, indicating a malignant tumor; an OCS score < 0.506 is considered negative, indicating a benign lesion.

[0018] As a preferred embodiment, the risk scoring module further includes: taking one or more clinical features selected from age, menopausal status, blood flow signal, cystic-solid, ascites, papillary protrusions, morphology, and peritoneal nodules as input variables, and substituting OCS score data into the constructed risk scoring model to calculate the risk probability value.

[0019] The present invention also provides the application of the aforementioned biomarker combination or the aforementioned system in the preparation of products for assessing the benign or malignant risk of accessory masses.

[0020] The present invention also provides a method for assessing the benign or malignant risk of accessory masses, the method comprising the following steps: (1) Obtain the imaging grading data of the subjects; (2) Obtain blood samples from the subjects, isolate exosomes from the blood samples, detect the expression levels of CA125 protein, HE4 protein and C5a protein in the exosomes, and calculate the exosome ovarian cancer score according to the preset algorithm; (3) Substitute the imaging grading data obtained in step (1) and the exosome ovarian cancer OCS score calculated in step (2) into the constructed risk scoring model to calculate the risk value; (4) Compare the risk value calculated in step (3) with the preset reference value, and determine the benign or malignant risk of the subject's appendage mass based on the comparison results.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. Improve the diagnostic accuracy of patients in the gray zone: This invention is the first to combine the OCS score with the O-RADS imaging score. By integrating biological information and imaging features, it significantly improves the accuracy of diagnosis of patients in the O-RADS 3 / 4 gray zone, reducing unnecessary overtreatment and missed diagnosis.

[0022] 2. Multi-dimensional data fusion: This invention not only integrates exosome biomarkers and imaging scores, but also incorporates various clinical variables such as age, menopausal status, and ultrasound features to construct a multi-dimensional joint diagnostic model, further improving diagnostic performance.

[0023] 3. High sensitivity and high specificity: Experimental data show that the joint risk scoring model constructed in this invention exhibits excellent AUC (up to 0.99), sensitivity and specificity in both the training and validation queues, and the model has strong generalization ability.

[0024] 4. Non-invasive and highly reproducible: Based on liquid biopsy technology, only peripheral blood needs to be collected, which is minimally invasive and highly reproducible, making it suitable for large-scale clinical application. Attached Figure Description

[0025] Figure 1 This is a transmission electron microscopy (TEM) image of the serum exosomes detected in this invention.

[0026] Figure 2 This is a graph showing the particle size distribution of serum exosomes according to the present invention.

[0027] Figure 3 This is the ROC curve of the risk scoring model constructed by combining three biomarkers—O-RADS score, OCS score, and age—on the training queue in this embodiment of the invention.

[0028] Figure 4 This is the ROC curve of the risk scoring model constructed by combining two markers, O-RADS score and OCS score, in the training queue in this embodiment of the invention.

[0029] Figure 5 This is the ROC curve of the risk scoring model constructed by combining four biomarkers—O-RADS score, OCS score, age, and menopausal status—on the training queue in this embodiment of the invention.

[0030] Figure 6 This is the ROC curve of the risk scoring model constructed by combining seven biomarkers—OCS score, blood flow signal, cystic / solid condition, age, ascites, papillary protrusion, and morphology—on the training cohort in this embodiment of the invention.

[0031] Figure 7 This is the ROC curve of the risk scoring model constructed by combining eight biomarkers—OCS score, blood flow signal, cystic / solid condition, age, ascites, papillary protrusions, morphology, and peritoneal nodules—on the training cohort in this embodiment of the invention.

[0032] Figure 8 This is the ROC curve of the risk scoring model constructed by combining four markers—OCS score, cystic solidity, papillary protrusion, and morphology—on the training queue in this embodiment of the invention.

[0033] Figure 9 This is the ROC curve of the risk scoring model constructed by combining five biomarkers—OCS score, cystic solidity, papillary protrusion, morphology, and menopausal status—on the training cohort in this embodiment of the invention.

[0034] Figure 10 This is the ROC curve of the risk scoring model constructed by combining five biomarkers—OCS score, blood flow signal, cystic / solid condition, ascites, and papillary protrusion—in the training cohort in this embodiment of the invention.

[0035] Figure 11 This is the ROC curve of the risk scoring model constructed by combining four markers—OCS score, cystic-solid condition, ascites, and papillary protrusion—in this embodiment of the invention on the training cohort.

[0036] Figure 12 This is the ROC curve of the risk scoring model constructed by combining seven biomarkers—OCS score, cystic-solid condition, ascites, papillary protuberance, morphology, peritoneal nodules, and menopausal status—on the training cohort in this embodiment of the invention.

[0037] Figure 13 This is the ROC curve of the risk scoring model constructed by combining three biomarkers—O-RADS score, OCS score, and age—on the validation queue in this embodiment of the invention.

[0038] Figure 14 This is the ROC curve of the risk scoring model constructed by combining two markers, O-RADS score and OCS score, in the validation queue in this embodiment of the invention.

[0039] Figure 15 This is the ROC curve of the risk scoring model constructed by combining four biomarkers—O-RADS score, OCS score, age, and menopausal status—on the validation queue in this embodiment of the invention.

[0040] Figure 16 This is the ROC curve of the risk scoring model constructed by combining seven biomarkers—OCS score, blood flow signal, cystic / solid condition, age, ascites, papillary protrusion, and morphology—on the validation cohort in this embodiment of the invention.

[0041] Figure 17 This is the ROC curve of the risk scoring model constructed by combining eight biomarkers—OCS score, blood flow signal, cystic / solid condition, age, ascites, papillary protrusions, morphology, and peritoneal nodules—on the validation cohort in this embodiment of the invention.

[0042] Figure 18 This is the ROC curve of the risk scoring model constructed by combining four markers—OCS score, cystic solidity, papillary protrusion, and morphology—in the embodiment of the present invention on the validation queue.

[0043] Figure 19 This is the ROC curve of the risk scoring model constructed by combining five markers—OCS score, cystic solidity, papillary protrusion, morphology, and menopausal status—on the validation cohort in this embodiment of the invention.

[0044] Figure 20 This is the ROC curve of the risk scoring model constructed by combining five biomarkers—OCS score, blood flow signal, cystic-solid condition, ascites, and papillary protrusion—in the embodiment of this invention on the validation cohort.

[0045] Figure 21 This is the ROC curve of the risk scoring model constructed by combining four biomarkers—OCS score, cystic-solid condition, ascites, and papillary protrusion—in the embodiment of this invention on the validation cohort.

[0046] Figure 22 This is the ROC curve of the risk scoring model constructed by combining seven biomarkers—OCS score, cystic-solid condition, ascites, papillary protuberance, morphology, peritoneal nodules, and menopausal status—on the validation cohort in this embodiment of the invention. Detailed Implementation

[0047] The technical solution of the present invention will be described in detail below with reference to the embodiments. Unless otherwise specified, all reagents and biological materials used below are commercial products.

[0048] Example 1: Screening of biomarker combinations and construction of risk scoring models (1) Study cohort and clinical information To achieve the technical solution of this invention, a total of 240 patients with ovarian masses were included as the research sample, comprising 179 patients with benign diseases and 61 patients with malignant ovarian cancer. Standardized testing: All patients underwent standardized ultrasound imaging evaluation and serum tumor marker testing before surgery. All patients were diagnosed by postoperative pathological examination. Sample collection: All samples were blood samples collected from patients before surgery; serum was used for subsequent exosome extraction, marker detection, and data analysis. The patients enrolled in this study were divided into a training cohort and a validation cohort. The training cohort consisted of 165 patients, including 123 benign patients and 42 ovarian cancer samples. The validation cohort consisted of 75 patients, including 56 benign patients and 19 ovarian cancer samples. Table 1 shows the clinical information of the enrolled samples, indicating no significant difference in the proportion of samples from different categories.

[0049]

[0050] (2) Detection of exosomal tumor markers OCS in patients' blood 1) Blood collection and serum separation Venous blood samples (≥ 5 ml) from patients with ovarian cancer and benign masses were collected and stored in 10 ml BD blood collection tubes. After gentle inversion and mixing, serum separation was performed by centrifugation: First, the blood collection tubes were centrifuged at 2000 g and 4°C for 10 min, and the serum hemolysis grade was determined and recorded. Samples with a grade of 4 or lower were used for subsequent studies. Second, the supernatant was transferred to 1.5 mL EP tubes. Finally, the supernatant was aliquoted into 1.5 mL EP tubes and stored at -80°C for later use.

[0051] 2) Extraction by exosome precipitation method The exosome extraction procedure employed a human exosome CA125, HE4, C5a detection kit (chemiluminescence method, National Medical Device Registration No. 20243400138), using the exosome separation reagent (chemical precipitation method) to extract exosomes. The steps are as follows: Take an appropriate amount of serum sample: Transfer the filtered serum sample (e.g., 100 μL) to a 96-well deep-well receiving plate. Add exosome precipitation reagent: Add 25% of the serum volume of exosome precipitation reagent (Ex) to each well. For example, if taking 100 μL of filtered serum sample, add 25 μL of reagent Ex. Ensure the reagent and serum are thoroughly mixed. If it is difficult to mix, use a pipette to mix by blowing. Shaking and standing: Place the 96-well deep-well receiving plate on a horizontal shaker and shake (set the speed to a moderate level to ensure uniform mixing of the solution). If it is not completely mixed after shaking, use a pipette to mix by blowing. Let stand at room temperature for 30 minutes to ensure complete precipitation of exosomes.

[0052] Centrifugation and removal of supernatant: Place the 96-well deep-well receiving plate in a plate centrifuge and centrifuge at 4°C, 4700×g for 10 minutes (or at 4°C, 2200×g for 20 minutes). After centrifugation, gently remove the supernatant (be careful to preserve the precipitate at the bottom). The precipitate formed at the bottom of the 96-well deep-well plate is the extracted exosome vesicle, which can be used for subsequent exosome characterization or biomarker detection after lysis.

[0053] 3) Exosome lysis After obtaining exosome precipitation, the exosomes were further lysed using a human exosome CA125, HE4, C5a detection kit (chemiluminescence method, National Medical Device Registration No. 20243400138) with the lysis reagent (Ly) to extract proteins and other biomolecules. The specific steps are as follows: Add lysis reagent (Ly): Add an equal volume of lysis reagent (Ly) to the precipitate. For example, if taking 100 μL of filtered serum sample, add 100 μL of lysis reagent (Ly). Mix thoroughly and place on a horizontal shaker for at least 30 minutes. The shaking time can be adjusted according to experimental needs to ensure complete exosome lysis.

[0054] Post-lysis solution transfer and filtration: The lysed solution should be transparent. Transfer all the lysate to a 96-well filter plate. Filter the lysate to a 96-well deep-well receiving plate. The filtered liquid is the exosome protein extract, which can be used for subsequent protein detection (such as the detection of exosome markers CA125, HE4, and C5a).

[0055] 4) Biomarker detection and OCS score generation The filtered exosome protein extract was used for biomarker detection and OCS score generation. Biomarker detection: The concentrations of exosome CA125, HE4, and C5a were determined using a human exosome CA125, HE4, and C5a assay kit (chemiluminescence immunoassay, National Medical Device Registration Number 20243400138) via chemiluminescent immunoassay (CLIA). All procedures were performed according to the kit instructions to ensure the accuracy and consistency of the results. The concentrations of these three biomarkers in each sample were quantified to obtain the corresponding measurement data.

[0056] OCS Score Generation: Based on the test results, an OCS score is generated using a validated calculation algorithm (see the published literature Wang H, Zhu J, Zou D, et al. Multicenter study of ovarian cancer score for diagnosing ovarian cancer. Gynecol Oncol. 2025;193:58-64). The OCS score reflects the comprehensive performance of exosome markers in the sample and serves as a key indicator for ovarian cancer risk assessment. An OCS score ≥ 0.506 is considered positive, suggesting a possible malignant tumor; an OCS score < 0.506 is considered negative, suggesting a benign lesion.

[0057] (3) Identification of serum exosome characteristics The morphology of exosomes resuspended in PBS was examined using transmission electron microscopy (TEM). First, serum exosomes were fixed in 4% paraformaldehyde and then transferred to a carbon-coated copper mesh for an electron microscope. The copper mesh was washed twice with PBS, then once each with PBS containing glycine (50 mM) and 0.5% BSA. The mesh was then stained with 2% uranyl acetate. Finally, the morphology of the serum exosomes was characterized using a transmission electron microscope (H-7650, Hitachi High-Technologies, Japan). TEM analysis showed that the extracted exosomes exhibited a typical "horseshoe" morphology (see [link to TEM]). Figure 1 ).

[0058] The particle size distribution of serum exosomes was detected using a nanoparticle tracking analysis (NTA) system: Resuspended serum exosomes were diluted with PBS (1 × 10⁷ - 1 × 10⁹ / ml) and mixed thoroughly by pipetting. The sample was added to the sample chamber of the NTA instrument (NanoSight NS300, Malvern, UK). A 488 nm excitation module was used, and the camera lens parameters were set (shutter speed 890, gain 146, detection threshold 7). At least 200 complete tracks were analyzed for each video. The nanoparticle tracking data of serum exosomes were analyzed using NTA software (version 2.3). NTA detection showed that the main peak of the exosome particle size was 50 nm, consistent with the particle size distribution of exosomes (see [link to NTA software]). Figure 2 ).

[0059] (4) Acquisition of imaging data and O-RADS scoring Standardized ultrasound imaging was used, and the O-RADS scoring system was applied to classify the adnexal masses, as detailed below: 1) Steps of radiological assessment Imaging methods: All patients with benign and malignant ovarian cancer underwent standardized ultrasound examinations. Detailed scans of the adnexal mass were performed using high-resolution transvaginal ultrasound (TVS). Ultrasound examinations were performed by experienced clinicians or radiologists to ensure the accuracy and consistency of the imaging data. Imaging assessment focused primarily on the morphological characteristics, size, shape, and blood flow signal of the adnexal mass, and the data were interpreted in conjunction with the patient's clinical presentation. Color Doppler ultrasound was used to assess blood flow in the mass, and high-frequency ultrasound probes (e.g., 5-10 MHz) were typically used for detailed imaging evaluation.

[0060] Ultrasound image data acquisition: Location and size of the adnexal mass: First, accurately identify the location (left or right ovary), size, and shape of the adnexal mass, and measure its maximum diameter and vertical diameter. Morphological characteristics of the mass: Assess whether the mass is a simple cystic, mixed, or solid mass. Observe its boundaries and cystic wall structure to ensure they are smooth and regular. Septa and papillary structures: Examine for the presence of internal septa, papillary structures, and other imaging features. Papillary structures usually indicate a higher probability of malignancy. Blood flow characteristics: Use color Doppler ultrasound to assess the blood flow signal within and around the mass. High blood flow in the tumor is usually an important marker of malignancy. Other imaging features: Such as the presence of calcification, fluid accumulation (e.g., ascites), and lymphadenopathy within the mass.

[0061] 2) Application of the O-RADS scoring system: The O-RADS (Ovarian-Adnexal Risk of Malignancy Algorithm) scoring system is used to stratify the risk of adnexal masses based on their radiographic features. Based on morphological characteristics, blood flow signals, septal structures, and other features observed in ultrasound imaging, masses are classified into different levels from O-RADS 1 to O-RADS 5, as detailed below: O-RADS 1: Very low risk, usually a completely benign adnexal mass, such as a functional cyst. It presents as a simple, homogeneous cystic lesion, without septa, no blood flow signal, and with well-defined borders.

[0062] O-RADS 2: Low risk, commonly seen in benign cysts, with regular shape and clear borders.

[0063] O-RADS 3: Low to medium risk, with possible slight irregularities or minor septations, and weak blood flow signal.

[0064] O-RADS 4: Medium to high risk, with obvious septa, papillary structures, or enhanced blood flow signals.

[0065] O-RADS 5: High risk, with typical malignant features such as irregular mass, prominent papillary structure and abundant blood flow signal.

[0066] This scoring system helps clinicians make accurate risk assessments and diagnostic decisions when faced with adnexal masses.

[0067] (5) Discovery of biomarkers Samples were grouped according to pathological examination results. Based on OCS scores, O-RADS scores, and clinical characteristics in the training cohort, statistical and machine learning methods were used to discover biomarkers that could distinguish between benign controls and ovarian cancer patients. The process is as follows: 1) Training cohort grouping. Based on the pathological test results of the samples, the patient samples in the training cohort were divided into a benign control group and an ovarian cancer patient group, for a total of 2 groups.

[0068] 2) Statistical methods for biomarker screening. First, the Mann-Whitney U test, chi-square test, and Fisher's exact test were used to analyze the differences of all features between the two groups. Ten features with a statistically significant difference between the groups (p-value <= 0.05) were selected as candidate features (namely: O-RADS score, OCS score, age, menopausal status, blood flow signal, cystic-solid, ascites, papillary projections, morphology, and peritoneal nodules) for subsequent analysis.

[0069] 3) Machine learning methods for biomarker selection. Using the candidate features selected in step 2) as initial features, five machine learning algorithms—random forest, extremely randomized trees, neural network, gradient boosting machine (GBM), and linear model—were employed to calculate the feature importance of each biomarker in each algorithm. Biomarkers with importance > 0 were retained. This process was repeated until the importance of all features included in each algorithm was > 0. The resulting features are the biomarker combinations listed in Table 2, used for subsequent analysis.

[0070] (6) Construction of risk scoring model between benign control group and ovarian cancer patients Using benign control groups and ovarian cancer patients as classification prediction targets, the biomarker combination discovered in (5) was used, and five machine learning algorithms were employed: random forest, extremely randomized trees, neural network, gradient boosting machine (GBM), and linear model. Different hyperparameters were preset for each algorithm. Using training cohort sample data and pathological test results as the true values, multiple scoring models of different machine learning algorithms were trained. The model formulas are as follows: , X i This indicates that the model calculates the numerical result based on the biomarker expression of sample i, P(X). i The probability value (1) represents the predicted probability of sample i being cancer by the scoring model. Samples with a probability value greater than the reference value are considered to be ovarian cancer patients and are represented by 1; samples with a probability value less than the reference value are likely to be non-cancer individuals and are represented by 0. This is the final prediction result. The trained model is saved on the hard drive as a file. When calling the model, inputting the sample marker expression value will yield the model's prediction result.

[0071] In the training cohort, the classification performance of the model was evaluated using a 5-fold cross-validation method, and the generalization performance of the model was evaluated in the validation cohort. Model evaluation and validation metrics included the area under the receiver operating characteristic curve (AUC, ranging from 0 to 1), sensitivity (ranging from 0 to 1), and specificity (ranging from 0 to 1), with higher values ​​indicating better model classification performance. Table 2 shows the cross-validation evaluation results of various biomarker combinations in the training cohort. The OCS score was based on a combination of three biomarkers derived from exosomes: CA125, HE4, and C5a. The O-RADS score was a comprehensive score ranging from 0 to 5, calculated based on imaging features including blood flow signals, cystic / solid structures, ascites, papillary projections, morphology, and peritoneal nodules. The data in Table 2 show that the O-RADS total score can be directly used as an imaging indicator, or specific features can be extracted individually and combined with the OCS score to further improve diagnostic accuracy.

[0072]

[0073] The risk scoring model constructed using a combination of three biomarkers—O-RADS score, OCS score, and age—had an AUC, sensitivity, and specificity of 0.98, 0.98, and 0.95, respectively (Table 2). See also... Figure 3 The ROC curve of the risk scoring model constructed using a combination of three biomarkers—O-RADS score, OCS score, and age—is shown in the training cohort. The results indicate that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0074] The risk scoring model constructed using a combination of two biomarkers—O-RADS score and OCS score—had an AUC, sensitivity, and specificity of 0.97, 0.95, and 0.95, respectively (Table 2). See also... Figure 4 The ROC curve of the risk scoring model constructed using a combination of O-RADS and OCS scores is shown in the training cohort. The results indicate that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0075] The risk scoring model constructed using a combination of four biomarkers—O-RADS score, OCS score, age, and menopausal status—had an AUC, sensitivity, and specificity of 0.93, 0.98, and 0.88, respectively (Table 2). See also... Figure 5 The ROC curves of the risk scoring model constructed using a combination of four biomarkers—O-RADS score, OCS score, age, and menopausal status—are plotted on the training cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0076] The risk scoring model constructed from a combination of seven biomarkers—OCS score, blood flow signal, cystic / solid morphology, age, ascites, papillary projections, and morphology—had an AUC, sensitivity, and specificity of 0.98, 0.96, and 0.98, respectively (Table 2). See also... Figure 6 The ROC curve of the risk scoring model, constructed using a combination of seven biomarkers—OCS score, blood flow signal, cystic / solid condition, age, ascites, papillary projections, and morphology—is plotted on the training cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0077] The risk scoring model constructed from a combination of eight biomarkers—OCS score, blood flow signal, cystic / solid morphology, age, ascites, papillary projections, morphology, and peritoneal nodules—had an AUC, sensitivity, and specificity of 0.97, 0.95, and 0.98, respectively (Table 2). See also... Figure 7 The ROC curves of a risk scoring model constructed using a combination of eight biomarkers—OCS score, blood flow signal, cystic / solid condition, age, ascites, papillary projections, morphology, and peritoneal nodules—are plotted on the training cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0078] The risk scoring model constructed from a combination of four biomarkers—OCS score, cystic-solid structure, papillary projections, and morphology—had an AUC, sensitivity, and specificity of 0.97, 0.95, and 0.93, respectively (Table 2). See also... Figure 8 The ROC curve of the risk scoring model, constructed using a combination of four biomarkers—OCS score, cystic-solid structure, papillary projections, and morphology—is plotted on the training cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0079] The risk scoring model constructed from a combination of five biomarkers—OCS score, cystic-solid structure, papillary projections, morphology, and menopausal status—had an AUC, sensitivity, and specificity of 0.97, 0.96, and 0.93, respectively (Table 2). See also... Figure 9 The ROC curve of the risk scoring model, constructed using a combination of five biomarkers—OCS score, cystic / solid morphology, papillary projections, morphology, and menopausal status—is plotted on the training cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0080] The risk scoring model constructed using a combination of five biomarkers—OCS score, blood flow signal, cystic-solid morphology, ascites, and papillary projections—had an AUC, sensitivity, and specificity of 0.96, 0.96, and 0.98, respectively (Table 2). See also... Figure 10The ROC curve of the risk scoring model, constructed using a combination of five biomarkers—OCS score, blood flow signal, cystic / solid lesions, ascites, and papillary projections—is plotted on the training cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0081] The risk scoring model constructed using a combination of four biomarkers—OCS score, cystic-solid morphology, ascites, and papillary projections—had an AUC, sensitivity, and specificity of 0.96, 0.92, and 0.95, respectively (Table 2). See also... Figure 11 The ROC curve of the risk scoring model constructed using a combination of four biomarkers—OCS score, cystic-solid lesions, ascites, and papillary protuberances—is shown in the training cohort. The results indicate that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0082] The risk scoring model constructed from a combination of seven biomarkers—OCS score, cystic-solid morphology, ascites, papillary projections, morphology, peritoneal nodules, and menopausal status—had an AUC, sensitivity, and specificity of 0.96, 0.93, and 0.93, respectively (Table 2). See also... Figure 12 The ROC curves of a risk scoring model constructed using a combination of seven biomarkers—OCS score, cystic / solid lesions, ascites, papillary projections, morphology, peritoneal nodules, and menopausal status—are plotted on the training cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the training cohort, demonstrating superior predictive performance.

[0083] Example 2: Validation of the predictive efficacy of a risk score model for ovarian cancer patients and non-cancer controls To validate the predictive efficacy of the established risk scoring model for ovarian cancer, a separate validation cohort consisting of ovarian cancer patients and non-cancer controls was selected to assess the model's classification performance. Pathological examination results were used as the true values. Model evaluation metrics included the area under the receiver operating characteristic curve (AUC, ranging from 0 to 1), sensitivity (ranging from 0 to 1), and specificity (ranging from 0 to 1), with higher values ​​indicating better performance. Table 3 shows the evaluation results of various combinations of biomarkers in the validation cohort.

[0084]

[0085] The risk scoring model constructed using a combination of three biomarkers—O-RADS score, OCS score, and age—had an AUC, sensitivity, and specificity of 0.99, 0.96, and 0.89, respectively (Table 3). See also Figure 13The ROC curve of the risk scoring model constructed using a combination of three biomarkers—O-RADS score, OCS score, and age—is shown in the validation cohort. The results indicate that this risk prediction model exhibits high AUC, sensitivity, and specificity in the validation cohort, demonstrating superior predictive performance.

[0086] The risk scoring model constructed using a combination of two biomarkers—O-RADS score and OCS score—had an AUC, sensitivity, and specificity of 0.97, 0.95, and 0.89, respectively (Table 3). See also... Figure 14 The ROC curve of the risk scoring model constructed using a combination of O-RADS and OCS scores is shown in the validation cohort. The results indicate that this risk prediction model exhibits high AUC, sensitivity, and specificity in the validation cohort, demonstrating superior predictive performance.

[0087] The risk scoring model constructed using a combination of four biomarkers—O-RADS score, OCS score, age, and menopausal status—had an AUC, sensitivity, and specificity of 0.97, 0.95, and 0.89, respectively (Table 3). See also... Figure 15 The ROC curves of a risk scoring model constructed using a combination of four biomarkers—O-RADS score, OCS score, age, and menopausal status—are plotted on the validation cohort. The results show that, in the validation cohort, this risk prediction model exhibits high AUC, sensitivity, and specificity, demonstrating superior predictive power.

[0088] The risk scoring model constructed from a combination of seven biomarkers—OCS score, blood flow signal, cystic / solid morphology, age, ascites, papillary projections, and morphology—had an AUC, sensitivity, and specificity of 0.99, 0.98, and 0.89, respectively (Table 3). See also... Figure 16 The ROC curves of a risk scoring model constructed using a combination of seven biomarkers—OCS score, blood flow signal, cystic / solid condition, age, ascites, papillary projections, and morphology—are plotted on the validation cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the validation cohort, demonstrating superior predictive performance.

[0089] The risk scoring model constructed from a combination of eight biomarkers—OCS score, blood flow signal, cystic / solid morphology, age, ascites, papillary projections, morphology, and peritoneal nodules—had an AUC, sensitivity, and specificity of 0.97, 0.96, and 0.84, respectively (Table 3). See also... Figure 17 The ROC curves of a risk scoring model constructed using a combination of eight biomarkers—OCS score, blood flow signal, cystic / solid condition, age, ascites, papillary projections, morphology, and peritoneal nodules—are plotted on the validation cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the validation cohort, demonstrating superior predictive power.

[0090] The risk scoring model constructed from a combination of four biomarkers—OCS score, cystic-solid structure, papillary projections, and morphology—had an AUC, sensitivity, and specificity of 0.97, 0.95, and 0.89, respectively (Table 3). See also... Figure 18 The ROC curve of the risk scoring model constructed using a combination of four biomarkers—OCS score, cystic-solid structure, papillary projections, and morphology—is plotted on the validation cohort. The results show that, in the validation cohort, this risk prediction model exhibits high AUC, sensitivity, and specificity, demonstrating superior predictive performance.

[0091] The risk scoring model constructed from a combination of five biomarkers—OCS score, cystic-solid structure, papillary projections, morphology, and menopausal status—had an AUC, sensitivity, and specificity of 0.98, 0.98, and 0.89, respectively (Table 3). See also... Figure 19 The ROC curves of a risk scoring model constructed using a combination of five biomarkers—OCS score, cystic / solid morphology, papillary projections, morphology, and menopausal status—are plotted on the validation cohort. The results show that, in the validation cohort, this risk prediction model exhibits high AUC, sensitivity, and specificity, demonstrating superior predictive power.

[0092] The risk scoring model constructed using a combination of five biomarkers—OCS score, blood flow signal, cystic-solid morphology, ascites, and papillary projections—had an AUC, sensitivity, and specificity of 0.97, 0.93, and 0.74, respectively (Table 3). See also... Figure 20 The ROC curve of the risk scoring model constructed using a combination of five biomarkers—OCS score, blood flow signal, cystic / solid lesions, ascites, and papillary projections—is plotted on the validation cohort. The results show that in the validation cohort, this risk prediction model exhibits high AUC, sensitivity, and specificity, demonstrating superior predictive performance.

[0093] The risk scoring model constructed using a combination of four biomarkers—OCS score, cystic-solid morphology, ascites, and papillary projections—had an AUC, sensitivity, and specificity of 0.95, 0.98, and 0.89, respectively (Table 3). See also... Figure 21 The ROC curve of the risk scoring model constructed using a combination of four biomarkers—OCS score, cystic-solid abscess, ascites, and papillary protuberances—is plotted on the validation cohort. The results show that, in the validation cohort, this risk prediction model exhibits high AUC, sensitivity, and specificity, demonstrating superior predictive power.

[0094] The risk scoring model constructed from a combination of seven biomarkers—OCS score, cystic-solid morphology, ascites, papillary projections, morphology, peritoneal nodules, and menopausal status—had an AUC, sensitivity, and specificity of 0.99, 0.98, and 0.84, respectively (Table 3). See also... Figure 22The ROC curves of a risk scoring model constructed using a combination of seven biomarkers—OCS score, cystic / solid lesions, ascites, papillary projections, morphology, peritoneal nodules, and menopausal status—are plotted on the validation cohort. The results show that this risk prediction model exhibits high AUC, sensitivity, and specificity in the validation cohort, demonstrating superior predictive power.

[0095] Comparative Example 1: Predictive power of OCS score or O-RADS score used alone The evaluation results of using OCS score or O-RADS score alone as biomarkers in the risk scoring model of this application are shown in Table 4. The O-RADS score (cutoff of 4) in Table 4 means that O-RADS=4 is used as the benign or malignant threshold. A score less than 4 is predicted as benign, and a score greater than or equal to 4 is predicted as malignant. Then, the specificity and sensitivity are statistically analyzed by comparing it with the clinical true value.

[0096]

[0097] As shown in Table 4, the OCS score biomarker alone exhibits high sensitivity and good specificity in both the training and validation cohorts. The O-RADS score (cutoff of 4), a binary classification index, used as a single imaging biomarker, shows sensitivity close to that of the OCS score, but its specificity is significantly lower. Combining the OCS and O-RADS scores as biomarkers can further improve specificity while maintaining high sensitivity, achieving a synergistic effect between imaging and molecular biomarkers. The above are merely some preferred embodiments of the present invention, and the present invention is not limited to the contents of these embodiments. For those skilled in the art, various changes and modifications can be made within the scope of the present invention's technical solutions, and any changes and modifications made are within the protection scope of the present invention.

Claims

1. A combination of biomarkers for assessing the risk of benign or malignant malignancy in accessory masses, characterized in that, The biomarker combination includes: ovarian cancer OCS score and imaging grade score.

2. A combination of biomarkers for assessing the risk of benign or malignant malignancy in accessory masses, characterized in that, The biomarker combination includes: ovarian cancer OCS score and imaging grade score.

3. The biomarker combination according to claim 1, characterized in that, The imaging grading scores include O-RADS scores, IOTA scores, ADNEX model scores, RMI scores, or AI-based imaging analysis scores.

4. A system for assessing the benign or malignant risk of accessory blocks, characterized in that, The system includes: The data acquisition module is used to acquire the imaging grading score data of the subjects and generate the OCS score based on the expression level data of exosomal CA125 protein, HE4 protein and C5a protein in the subjects' biological samples. The risk scoring module is used to input the acquired radiographic grading data and OCS score data into the constructed risk scoring model to calculate the risk probability value. The result determination module is used to compare the calculated risk probability value with a preset reference value and output the risk assessment result of the benign or malignant mass of the subject's appendage.

5. The system for assessing the benign or malignant risk of accessory packages according to claim 4, characterized in that, The formula for the risk scoring model is as follows: ， X i This indicates that the model calculates the numerical result based on the biomarker expression of sample i, P(X). i ) represents the probability value of sample i to be cancer predicted by the scoring model; samples with a probability value greater than the reference value are considered to be ovarian cancer patients and are represented by 1; samples with a probability value less than the reference value are likely to be non-cancer individuals and are represented by 0, which is the final prediction result.

6. The system for assessing the benign or malignant risk of accessory packages according to claim 4 or 5, characterized in that: The risk scoring model is constructed using one or more machine learning algorithms selected from random forest, extreme random tree, neural network, gradient boosting machine, and linear model.

7. The system for assessing the benign or malignant risk of accessory blocks according to claim 4, characterized in that: The OCS score is generated using a validated calculation algorithm. An OCS score ≥ 0.506 is considered positive, indicating a malignant tumor; an OCS score < 0.506 is considered negative, indicating a benign lesion.

8. The system for assessing the benign or malignant risk of accessory packages according to claim 4, characterized in that, The risk scoring module further includes: taking one or more clinical features selected from age, menopausal status, blood flow signal, cystic-solid, ascites, papillary protrusions, morphology, and peritoneal nodules as input variables, and substituting OCS score data into the constructed risk scoring model to calculate the risk probability value.

9. The use of the combination of biomarkers according to any one of claims 1-3, or the system according to any one of claims 4-8, in the preparation of products for assessing the benign or malignant risk of accessory masses.

10. A method for assessing the benign or malignant risk of accessory masses, characterized in that, The method includes the following steps: (1) Obtain the imaging grading data of the subjects; (2) Obtain blood samples from the subjects, isolate exosomes from the blood samples, detect the expression levels of CA125 protein, HE4 protein and C5a protein in the exosomes, and calculate the exosome ovarian cancer OCS score according to the preset algorithm; (3) Substitute the imaging grading data obtained in step (1) and the exosome ovarian cancer score calculated in step (2) into the constructed risk scoring model to calculate the risk value; (4) Compare the risk value calculated in step (3) with the preset reference value, and determine the benign or malignant risk of the subject's appendage mass based on the comparison results.

Images