A novel diagnostic nomogram based on serological and ultrasound findings for preoperative prediction of malignancy in patients with ovarian masses

Nomogram development for predicting ovarian tumor malignancy using inflammatory biomarker and CA-125

Global challenges in ovarian cancer underscore the need for cost-effective screening. This study aims to assess the role of pretreatment Neutrophil-to-Lymphocyte Ratio (NLR), Lymphocyte-to-Monocyte-Ratio (LMR), Platelet-to-Lymphocyte Ratio (PLR), and CA-125 in distinguishing benign and malignant ovarian tumors, while also constructing nomogram models for distinguish benign and malignant ovarian tumor using inflammatory biomarkers and CA-125. This is a retrospective study of 206 ovarian tumor patients. We conducted bivariate analysis to compare mean values of CA-125, LMR, NLR, and PLR with histopathology results. Multiple regression logistic analysis was then employed to establish predictive models for malignancy. NLR, PLR, and CA-125 exhibited statistically higher levels in malignant ovarian tumors compared to benign ones (5.56 ± 4.8 vs. 2.9 ± 2.58, 278.12 ± 165.2 vs. 180.64 ± 89.95, 537.2 ± 1621.47 vs. 110.08 ± 393.05, respectively), while lower LMR was associated with malignant tumors compared to benign (3.2 ± 1.6 vs. 4.24 ± 1.78, p = 0.0001). Multiple logistic regression analysis revealed that both PLR and CA125 emerged as independent risk factors for malignancy in ovarian tumors (P(z) 0.03 and 0.01, respectively). Utilizing the outcomes of multiple regression logistic analysis, a nomogram was constructed to enhance malignancy prediction in ovarian tumors. In conclusion, our study emphasizes the significance of NLR, PLR, CA-125, and LMR in diagnosing ovarian tumors. PLR and CA-125 emerged as independent risk factors for distinguishing between benign and malignant tumors. The nomogram model offers a practical way to enhance diagnostic precision.

An explainable machine learning model to solid adnexal masses diagnosis based on clinical data and qualitative ultrasound indicators

BACKGROUND

Accurate characterization of newly diagnosed a solid adnexal lesion is a key step in defining the most appropriate therapeutic approach. Despite guidance from the International Ovarian Tumor Analyzes Panel, the evaluation of these lesions can be challenging. Recent studies have demonstrated how machine learning techniques can be applied to clinical data to solve this diagnostic problem. However, ML models can often consider as black-boxes due to the difficulty of understanding the decision-making process used by the algorithm to obtain a specific result.

AIMS

For this purpose, we propose an Explainable Artificial Intelligence model trained on clinical characteristics and qualitative ultrasound indicators to predict solid adnexal masses diagnosis.

MATERIALS & METHODS

Since the diagnostic task was a three-class problem (benign tumor, invasive cancer, or ovarian metastasis), we proposed a waterfall classification model: a first model was trained and validated to discriminate benign versus malignant, a second model was trained to distinguish nonmetastatic versus metastatic malignant lesion which occurs when a patient is predicted to be malignant by the first model. Firstly, a stepwise feature selection procedure was implemented. The classification performances were validated on Leave One Out scheme.

RESULTS

The accuracy of the three-class model reaches an overall accuracy of 86.36%, and the precision per-class of the benign, nonmetastatic malignant, and metastatic malignant classes were 86.96%, 87.27%, and 77.78%, respectively.

DISCUSSION

SHapley Additive exPlanations were performed to visually show how the machine learning model made a specific decision. For each patient, the SHAP values expressed how each characteristic contributed to the classification result. Such information represents an added value for the clinical usability of a diagnostic system.

CONCLUSIONS

This is the first work that attempts to design an explainable machine-learning tool for the histological diagnosis of solid masses of the ovary.

Artificial intelligence-based models enabling accurate diagnosis of ovarian cancer using laboratory tests in China: a multicentre, retrospective cohort study

BACKGROUND

Ovarian cancer is the most lethal gynecological malignancy. Timely diagnosis of ovarian cancer is difficult due to the lack of effective biomarkers. Laboratory tests are widely applied in clinical practice, and some have shown diagnostic and prognostic relevance to ovarian cancer. We aimed to systematically evaluate the value of routine laboratory tests on the prediction of ovarian cancer, and develop a robust and generalisable ensemble artificial intelligence (AI) model to assist in identifying patients with ovarian cancer.

METHODS

In this multicentre, retrospective cohort study, we collected 98 laboratory tests and clinical features of women with or without ovarian cancer admitted to three hospitals in China during Jan 1, 2012 and April 4, 2021. A multi-criteria decision making-based classification fusion (MCF) risk prediction framework was used to make a model that combined estimations from 20 AI classification models to reach an integrated prediction tool developed for ovarian cancer diagnosis. It was evaluated on an internal validation set (3007 individuals) and two external validation sets (5641 and 2344 individuals). The primary outcome was the prediction accuracy of the model in identifying ovarian cancer.

FINDINGS

Based on 52 features (51 laboratory tests and age), the MCF achieved an area under the receiver-operating characteristic curve (AUC) of 0·949 (95% CI 0·948-0·950) in the internal validation set, and AUCs of 0·882 (0·880-0·885) and 0·884 (0·882-0·887) in the two external validation sets. The model showed higher AUC and sensitivity compared with CA125 and HE4 in identifying ovarian cancer, especially in patients with early-stage ovarian cancer. The MCF also yielded acceptable prediction accuracy with the exclusion of highly ranked laboratory tests that indicate ovarian cancer, such as CA125 and other tumour markers, and outperformed state-of-the-art models in ovarian cancer prediction. The MCF was wrapped as an ovarian cancer prediction tool, and made publicly available to provide estimated probability of ovarian cancer with input laboratory test values.

INTERPRETATION

The MCF model consistently achieved satisfactory performance in ovarian cancer prediction when using laboratory tests from the three validation sets. This model offers a low-cost, easily accessible, and accurate diagnostic tool for ovarian cancer. The included laboratory tests, not only CA125 which was the highest ranked laboratory test in importance of diagnostic assistance, contributed to the characterisation of patients with ovarian cancer.

FUNDING

Ministry of Science and Technology of China; National Natural Science Foundation of China; Natural Science Foundation of Guangdong Province, China; and Science and Technology Project of Guangzhou, China.

TRANSLATION

For the Chinese translation of the abstract see Supplementary Materials section.

Predicting Survival of Patients with Nonmetastatic Breast Cancer Based on Fibrinogen-to-Albumin Ratio and Lymphocyte-to-Monocyte Ratio: A Nomogram-Based Assessment

Background

Parameters of systemic inflammation have received attention as prognostic surrogates in various malignant tumors. Fibrinogen-to-albumin ratio (FAR) and lymphocyte-to-monocyte ratio (LMR) correlate with tumor growth and dissemination. We aimed to bring the combination of FAR and LMR (FAR-LMR) together to establish novel nomograms for survival and recurrence in nonmetastatic breast cancer patients.

Methods

We retrospectively recruited 461 female patients with nonmetastatic breast cancer from January 2011 to December 2013 in our hospital and randomly assigned them into the training cohort (N = 318) and the validation cohort (N = 143). The potential predictive factors for overall survival (OS), locoregional recurrence-free survival (LRFS), and distant metastasis-free survival (DMFS) were assessed by Cox proportional hazards models and log-rank test.

Results

Elevated FAR was associated with poor OS (p < 0.001) and DMFS (p = 0.02), whereas increased LMR was associated with satisfactory OS (p = 0.01) and LRFS (p = 0.01). High FAR combined with low LMR was associated with less favorable OS (p = 0.001), LRFS (p = 0.005), and DMFS (p = 0.003) Based on multivariate analysis, FAR-LMR, tumor size, lymph node metastasis, age, and pathologic status contributed to prognostic nomograms of OS, DMFS, and LRFS. Nomograms presented exceptional performance for 3-, 5-, and 8-year OS, DMFS, and LRFS prediction compared with clinical TNM stage. The C-index was significantly higher than that of TNM stage, either of FAR or LMR (3-year: 0.709 vs. 0.621 vs. 0.544 vs. 0.641, 5-year: 0.761 vs. 0.597 vs. 0.605 vs. 0.677, 8-year: 0.84 vs. 0.62 vs. 0.539 vs. 0.623).

Conclusions

We developed and validated a convenient predictive model for the survival outcomes of patients with nonmetastatic breast cancer. The nomograms can be utilized as auxiliary tools to provide prognostic information.

Independent predictive value of blood inflammatory composite markers in ovarian cancer: recent clinical evidence and perspective focusing on NLR and PLR

Ovarian cancer (OC) is one of the deadliest malignant tumors affecting women worldwide. The predictive value of some blood inflammatory composite markers in OC has been extensively reported. They can be used for early detection and differential diagnosis of OC and can be used for predicting survival, treatment response, and recurrence in the affected patients. Here, we reviewed the predictive values of composite inflammatory markers based on complete blood count, namely neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), monocyte-to-lymphocyte ratio, and systemic inflammation index and markers based on blood protein, namely C-reactive protein-to-albumin ratio and prognostic nutritional index in OC, with a focus on NLR and PLR. We referred to the clinical studies on these six markers, reviewed the patient population, and summarized the marker cut-off values, significance, and limitations of these studies. All these studies were retrospective and most of them were single-center clinical studies with small sample sizes. We found that the cut-off values of these markers have not been unified, and methods used to determine these values varied among studies. The predictive value of these markers on survival was mainly reflected in the postoperative patients of multiple subtypes of ovarian cancer including epithelial OC, high-grade serous ovarian carcinoma, and ovarian clear cell carcinoma. We focused on NLR and PLR and calculated their pooled hazard ratios. NLR and PLR were reliable in predicting overall and progression-free survivals in patients with OC. Therefore, it is necessary to adjust important confounding factors and conduct a long-term follow-up prospective cohort study to further clarify the cut-off values of NLR and PLR and their clinical applications.

The prognostic role of systemic inflammatory markers in apparent early-stage ovarian cancer

BACKGROUND

Few studies analyzed the prognostic role of systemic inflammatory markers in early-stage ovarian cancer. The primary endpoint of the present study was to assess the prognostic impact of baseline inflammatory markers in early-stage ovarian cancer. The secondary endpoints were to compare the disease-free survival (DFS) of inflammatory markers with standard risk factors and to correlate these with BRCA mutational status.

METHODS

Retrospective, single-center, observational study. Patients with FIGO-stage I-II and IIIA1 epithelial ovarian cancer undergoing primary surgery between 10/2012 and 12/2019 were included. Inflammatory markers were evaluated on the results of the complete blood count and coagulation tests, performed before ovarian cancer surgery. The Receiver Operating Characteristic curve was used to determine the optimal cut-off value of different baseline inflammatory biomarkers for the DFS analysis.

RESULTS

Three hundred fifty-nine patients were included in the study period. Baseline neutrophil-lymphocyte ratio (NLR) ≥ 3 and systemic immune inflammation index (SII, defined as platelet x neutrophil-lymphocyte ratio) ≥ 1000 were associated with worse 3 year DFS and baseline SII ≥ 1000 was associated with worse 3 year OS. BRCA-mutated patients with SII ≥ 1000 and with NLR ≥ 3 had significantly worse DFS compared to SII < 1000 and with NLR < 3. FIGO stage > I was the only independent risk factor for higher risk of recurrence.

CONCLUSION

SII ≥ 1000 and NLR ≥ 3 were associated with worse 3 year DFS and SII ≥ 1000 was associated with worse 3 year OS. The subgroups of BRCA-mutated patients with higher inflammation markers (SII ≥ 1000 and NLR ≥ 3) were associated with worse DFS. These findings might be helpful to design personalized treatment and more intensive surveillance.

The Endometriotic Neoplasm Algorithm for Risk Assessment (e-NARA) Index Sheds Light on the Discrimination of Endometriosis-Associated Ovarian Cancer from Ovarian Endometrioma

BACKGROUND

Magnetic resonance (MR) relaxometry provides a noninvasive tool to discriminate endometriosis-associated ovarian cancer (EAOC) from ovarian endometrioma (OE) with high accuracy. However, this method has a limitation in discriminating malignancy in clinical use because the R2 value depends on the device manufacturer and repeated imaging is unrealistic. The current study aimed to reassess the diagnostic accuracy of MR relaxometry and investigate a more powerful tool to distinguish EAOC from OE.

METHODS

This retrospective study was conducted at our institution from December, 2012, to May, 2022. A total of 150 patients were included in this study. Patients with benign ovarian tumors (n = 108) mainly received laparoscopic surgery, and cases with suspected malignancy (n = 42) underwent laparotomy. Information from a chart review of the patients' medical records was collected.

RESULTS

A multiple regression analysis revealed that the age, the tumor diameter, and the R2 value were independent malignant predicting factors. The endometriotic neoplasm algorithm for risk assessment (e-NARA) index provided high accuracy (sensitivity, 85.7%; specificity, 87.0%) to discriminate EAOC from OE.

CONCLUSIONS

The e-NARA index is a reliable tool to assess the probability of malignant transformation of endometrioma.

Protein Panel of Serum-Derived Small Extracellular Vesicles for the Screening and Diagnosis of Epithelial Ovarian Cancer

Although ovarian cancer, a gynecological malignancy, has the highest fatality rate, it still lacks highly specific biomarkers, and the differential diagnosis of ovarian masses remains difficult to determine for gynecologists. Our study aimed to obtain ovarian cancer-specific protein candidates from the circulating small extracellular vesicles (sEVs) and develop a protein panel for ovarian cancer screening and differential diagnosis of ovarian masses. In our study, sEVs derived from the serum of healthy controls and patients with cystadenoma and ovarian cancer were investigated to obtain a cancer-specific proteomic profile. In a discovery cohort, 1119 proteins were identified, and significant differences in the protein profiles of EVs were observed among groups. Then, 23 differentially expressed proteins were assessed using the parallel reaction monitoring in a validation cohort. Through univariate and multivariate logistic regression analyses, a novel model comprising three proteins (fibrinogen gamma gene (FGG), mucin 16 (MUC16), and apolipoprotein (APOA4)) was established to screen patients with ovarian cancer. This model exhibited an area under the receiver operating characteristic curve (AUC) of 0.936 (95% CI, 0.888–0.984) with 92.0% sensitivity and 82.9% specificity. Another panel comprising serum CA125, sEV-APOA4, and sEV-CD5L showed excellent performance (AUC 0.945 (95% CI, 0.890–1.000), sensitivity of 88.0%, specificity of 93.3%, and accuracy of 89.2%) to distinguish malignancy from benign ovarian masses. Altogether, our study provided a proteomic signature of circulating sEVs in ovarian cancer. The diagnostic proteomic panel may complement current clinical diagnostic measures for screening ovarian cancer in the general population and the differential diagnosis of ovarian masses.

Serology-Based Model for Personalized Epithelial Ovarian Cancer Risk Evaluation

This study aimed to establish a prognosis-prediction model based on serological indicators in patients with epithelial ovarian cancer (EOC). Patients initially diagnosed as ovarian cancer and surgically treated in Fudan University Shanghai Cancer Center from 2014 to 2018 were consecutively enrolled. Serological indicators preoperatively were collected. A risk model score (RMS) was constructed based on the levels of serological indicators determined by receiver operating characteristic curves. We correlated this RMS with EOC patients’ overall survival (OS). Finally, 635 patients were identified. Pearson’s χ2 results showed that RMS was significantly related to clinical parameters. Kaplan−Meier analysis demonstrated that an RMS less than 3 correlated with a longer OS (p < 0.0001). Specifically, significant differences were perceived in the survival curves of different subgroups. Multivariate Cox analysis revealed that age (p = 0.015), FIGO stage (p = 0.006), ascites (p = 0.015) and RMS (p = 0.005) were independent risk factors for OS. Moreover, RMS combined with age, FIGO and ascites could better evaluate for patients’ prognosis in DCA analyses. Our novel RMS-guided classification preoperatively identified the prognostic subgroups of patients with EOC and showed higher accuracy than the conventional method, meaning that it could be a useful and economical tool for tailored monitoring and/or therapy.

Application Values of 2D and 3D Radiomics Models Based on CT Plain Scan in Differentiating Benign from Malignant Ovarian Tumors

Background

Accurate identification of ovarian tumors as benign or malignant is highly crucial. Radiomics is a new branch of imaging that has emerged in recent years to replace the traditional naked eye qualitative diagnosis.

Objective

This study is aimed at exploring the difference in the application potential of two- (2D) and three-dimensional (3D) radiomics models based on CT plain scan in differentiating benign from malignant ovarian tumors.

Method

A retrospective analysis was performed on 140 patients with ovarian tumors confirmed by surgery and pathology in our hospital from July 2017 to August 2020. These 140 patients were divided into benign group and malignant group according to the pathological results. The ITK-SNAP software was used to outline the regions-of-interest (ROI) of 2D or 3D tumors on the CT plain scan image of each patient; the texture features were extracted through analysis kit (AK), and the cases were randomly divided into training groups (n = 99) and validation group (n = 41) in a ratio of 7 : 3. The least absolute shrinkage and selection operator (LASSO) algorithm was used to perform dimensionality reduction, followed by the construction of the radiomics nomogram model using the logistic regression method. The receiver operating characteristic (ROC) curve was drawn, and the calibration curve and decision curve analysis (DCA) were used to evaluate and verify the results of the radiomics nomogram and compare the differences between 2D and 3D diagnostic performance.

Results

There were 396 quantitative radiomics feature parameters extracted from 2D group and the 3D group, respectively. The area under the curve (AUC) of the radiomics nomogram of the 2D training group and the validation group were 0.96 and 0.97, respectively. The accuracy, specificity, and sensitivity of the training set were 92.9%, 88.9%, and 96.3%, respectively, and those of the validation set were 90.2%, 82.6%, and 100.0%, respectively. The AUCs of the radiomics nomogram of the 3D training group and validation group were 0.96% and 0.99%, respectively. The accuracy, sensitivity, and specificity of the training set were 92.9%, 96.3%, and 88.9%, respectively, and those of the validation set were 97.6%, 95.7%, and 100.0%, respectively. DeLong's test indicated that there was no statistical significance between the two sets (P > 0.05).

Conclusions

For the differential diagnosis of benign and malignant ovarian tumors, the 2D and 3D radiomics nomogram models exhibited comparable diagnostic performance. Considering that the 2D model was cost-effective and time-efficient, it was more recommended to use 2D features in future research.