A prospective study to evaluate the risk malignancy index and its diagnostic implication in patients with suspected ovarian mass

Accuracy of machine learning in the preoperative identification of ovarian borderline tumors: a meta-analysis

AIM

The objective of this study is to explore the diagnostic value of machine learning (ML) in borderline ovarian tumors through meta-analysis.

METHODS

Pubmed, Embase, Web of Science, and Cochrane Library databases were comprehensively retrieved from database inception untill February 16, 2023. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was adopted to evaluate the risk of bias in the original studies. Sub-group analyses of ML were conducted according to clinical features and radiomics features. We separately discussed the discriminative value of ML for borderline vs benign and borderline vs malignant tumors.

RESULTS

Eighteen studies involving 12,778 subjects were included in our analysis. The modeling variables mainly consisted of radiomics features (n=13) and a small number of clinical features (n=5). When distinguishing between borderline and benign tumors, the ML model based on radiomic features achieved a c-index of 0.782 (95% CI: 0.732-0.831), sensitivity of 0.75 (95% CI: 0.67-0.82), and specificity of 0.75 (95% CI: 0.67-0.81) in the validation set. When distinguishing between borderline and malignant tumors, the ML model based on radiomic features achieved a c-index of 0.916 (95% CI: 0.891-0.940), sensitivity of 0.86 (95% CI: 0.78-0.91), and specificity of 0.88 (95% CI: 0.82-0.92) in the validation set. In addition, we analyzed the discriminatory ability of radiologists and found that their sensitivity was 0.26 (95% CI: 0.12-0.46) and specificity was 0.94 (95% CI: 0.90-0.97).

CONCLUSIONS

ML has tremendous potential in the preoperative diagnosis and differentiation of borderline ovarian tumors and may be more accurate than radiologists in diagnosing and differentiating borderline ovarian tumors.

Risk of Ovarian Malignancy Algorithm and Pelvic Mass Score for the prediction of malignant ovarian tumors: a prospective comparative study

Aim

Ovarian cancer is the seventh most common female cancer worldwide. Nevertheless, there is no available universal screening method for malignant ovarian masses. This study compares the value of the Risk of Ovarian Malignancy Algorithm (ROMA) and Pelvic Mass Score (PMS) scoring systems in the diagnosis of malignant ovarian masses.

Material and methods

This prospective comparative study was conducted from March 2021 until April 2022. A total of 258 women diagnosed with ovarian mass and eligible for surgical intervention according to institutional guidelines were enrolled in the study. Ultrasound was performed for the assessment of masses, ascites and metastases, also color flow Doppler was done to measure the resistance index of the mass vasculature. Preoperative venous blood samples were collected to measure CA 125 and HE4. PMS and ROMA scoring systems were calculated for each patient. All women were subjected to a surgical intervention (according to applicable institutional guidelines), using either open or laparoscopic techniques. Histopathological examination of the removed specimens was done, and in line with the recognized gold standard, the results were compared with the pre-operative diagnosis of both scoring systems.

Results

Both PMS and ROMA showed a high predictive probability for ovarian malignancies (AUC = 0.93, sensitivity = 83.3%, specificity = 90.37%; AUC = 0.91, sensitivity = 84.4%, specificity = 95.56%, respectively), yet no statistical significant difference was found between the two scoring systems (p = 0.353, 95% CI -0.025 to 0.070).

Conclusions

Both PMS and ROMA seem to be promising scoring systems for discriminating benign from malignant ovarian masses, but more research is needed to determine the optimum diagnostic pathway, especially one yielding the least false-negative results.

Comparison of Four Risk of Malignancy Indices for Preoperative Evaluation of Ovarian Masses: A Prospective Observational Study

BACKGROUND

Ovarian cancer imposes a significant health burden worldwide. Although various tumor markers are available to diagnose ovarian cancer, low-resource countries like India require a humble marker or index. The Risk of Malignancy Index (RMI) has been found to be a simple yet promising tool that can be used for this purpose. In this study, we attempted to validate various RMIs with the help of menopausal status, ultrasonogram score, cancer antigen (CA) 125 value and compare all four RMIs, which would be useful to differentiate benign and malignant ovarian masses. This could be an essential tool, especially in low-resource settings.

METHOD

This prospective study was conducted at Kalinga Institute of Medical Sciences in Odisha, India, from September 2020 to September 2022 involving 191 patients with ovarian mass with histopathology, which was deemed the "gold standard" diagnostic tool. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of RMI 1, 2, 3, and 4 were calculated and compared.  Results: Out of 191 patients, 32 (16%) had malignancy and 159 (83.2%) had benign pathology. It was apparent that RMI 4 was a better tool for the initial assessment of patients with ovarian masses with a sensitivity of 80.6%, specificity of 96.2%, PPV of 81%, NPV of 96% at a cutoff of 334, and an area under the curve value of 0.939.

CONCLUSION

RMI 4 followed by RMI 3 were relatively better indices than RMI 1 and RMI 2 for identifying benign and malignant ovarian masses. RMI 4 was a valuable and applicable method in diagnosing pelvic masses with a high risk of malignancy.

A comparison of the diagnostic performance of the O-RADS, RMI4, IOTA LR2, and IOTA SR systems by senior and junior doctors

Purpose

This study compared the diagnostic performance of the Ovarian-Adnexal Reporting and Data System (O-RADS), the Risk of Malignancy Index 4 (RMI4), the International Ovarian of Tumor Analysis Logistic Regression Model 2 (IOTA LR2), and the IOTA Simple Rules (IOTA SR) in predicting the malignancy of adnexal masses (AMs).

Methods

This retrospective study included 575 women with AMs between 2017 and 2020. All clinical messages, ultrasound images, and pathological findings were collected. Two senior doctors (group I) and two junior doctors (group II) used the four systems to classify AMs. The postoperative pathological diagnosis was used as the gold standard to evaluate the diagnostic efficiency. A receiver operating characteristic curve was used to test the diagnostic performance. The interrater agreement between the two groups was tested using kappa values.

Results

Of all 592 AMs, 447 (75.5%) were benign, 123 (20.8%) were malignant, and 22 (3.7%) were borderline. The intergroup consistency test yielded kappa values of 0.71, 0.92, 0.68, and 0.77 for the O-RADS, RMI4, IOTA LR2, and IOTA SR, respectively. To predict malignant lesions, the areas under the curve of the O-RADS, RMI4, IOTA LR2, and IOTA SR systems were 0.90, 0.89, 0.90, and 0.86 for group I and 0.89, 0.87, 0.88, and 0.84 for group II, respectively. The O-RADS had the highest sensitivity (91.0% in group I and 84.8% in group II).

Conclusion

The four diagnostic systems could compensate for junior doctors’ inexperience in predicting malignant adnexal lesions. The O-RADS performed best and showed the highest sensitivity.

MRI-Based Multiple Instance Convolutional Neural Network for Increased Accuracy in the Differentiation of Borderline and Malignant Epithelial Ovarian Tumors

BACKGROUND

Preoperative differentiation of borderline from malignant epithelial ovarian tumors (BEOT vs. MEOT) is challenging and can significantly impact surgical management.

PURPOSE

To develop a multiple instance convolutional neural network (MICNN) that can differentiate BEOT from MEOT, and to compare its diagnostic performance with that of radiologists.

STUDY TYPE

Retrospective study of eight clinical centers.

SUBJECTS

Between January 2010 and June 2018, a total of 501 women (mean age, 48.93 ± 14.05 years) with histopathologically confirmed BEOT (N = 165) or MEOT (N = 336) were divided into the training (N = 342) and validation cohorts (N = 159).

FIELD STRENGTH/SEQUENCE

Three axial sequences from 1.5 or 3 T scanner were used: fast spin echo T2-weighted imaging with fat saturation (T2WI FS), echo planar diffusion-weighted imaging, and 2D volumetric interpolated breath-hold examination of contrast-enhanced T1-weighted imaging (CE-T1WI) with FS.

ASSESSMENT

Three monoparametric MICNN models were built based on T2WI FS, apparent diffusion coefficient map, and CE-T1WI. Based on these monoparametric models, we constructed an early multiparametric (EMP) model and a late multiparametric (LMP) model using early and late information fusion methods, respectively. The diagnostic performance of the models was evaluated using the receiver operating characteristic (ROC) curve and compared to the performance of six radiologists with varying levels of experience.

STATISTICAL TESTS

We used DeLong test, chi-square test, Mann-Whitney U-test, and t-test, with significance level of 0.05.

RESULTS

Both EMP and LMP models differentiated BEOT from MEOT, with an area under the ROC curve (AUC) of 0.855 (95% CI, 0.795-0.915) and 0.884 (95% CI, 0.831-0.938), respectively. The AUC of the LMP model was significantly higher than the radiologists' pooled AUC (0.884 vs. 0.797).

DATA CONCLUSION

The developed MICNN models can effectively differentiate BEOT from MEOT and the diagnostic performances (AUCs) were more superior than that of the radiologists' assessments.

LEVEL OF EVIDENCE

3 TECHNICAL EFFICACY STAGE: 2.

[Huge ovarian tumor in a perimenopausal woman: a case report]

Huge ovarian tumor is usually diagnosed at an advanced stage, in particular in developing countries. It can cause surgical complications. Anatomopathological examination is essential to establish the diagnosis. The role of tumor marker CA125 in a perimenopausal woman with pelvic mass has been widely debated.

Comparison of the four malignancy risk indices in the discrimination of malignant ovarian masses: A cross-sectional study

OBJECTIVE

To evaluate the accuracy of the four malignancy risk indices to distinguish benign from malignant ovarian masses MATERIALS AND METHODS: This was an observational cross-sectional study conducted on 155 patients between January 2016 and January 2019. Women with ovarian masses planned for surgical management were recruited from the outpatient Gynecology clinic of the hospital. The risk of malignancy index (RMI 1-4) was calculated for all women with ovarian masses. Biopsies obtained from the ovarian masses after the surgical intervention was sent to the pathology lab for histopathological examination. The histopathologic diagnosis of the ovarian masses was considered the gold standard for diagnosis.

RESULTS

The participants' mean age in the group of patients with benign masses was 33.50 ± 14.53 years versus 45.09 ± 13.67 years in the malignant group. The two most prominent features in the malignant group were solid areas in 85.3 % of malignant masses and about 91.2 % of malignant masses showing size <7 cm in their largest diameter. The RMI's most sensitive individual parameter was the CA-125 level, while the lowest sensitivity was for the menopausal status. RMI 2 had the highest sensitivity of 76.47 %, while RMI 1 and 3 had the highest specificity, 92.56 %. RMI 2 had the highest AUC, 0.83.

CONCLUSIONS

RMI 2 is a simple and reliable tool and had the best performance among all RMIs in benign discrimination from malignant ovarian masses with high sensitivity and accuracy.

Scoring systems for the evaluation of adnexal masses nature: current knowledge and clinical applications

Adnexal masses are a common finding in women, with 20% of them developing at least one pelvic mass during their lifetime. There are more than 30 different subtypes of adnexal tumours, with multiple different subcategories, and the correct characterisation of the pelvic masses is of paramount importance to guide the correct management. On that basis, different algorithms and scoring systems have been developed to guide the clinical assessment. The first scoring system implemented into the clinical practice was the Risk of Malignancy Index, which combines ultrasound evaluation, menopausal status, and serum CA-125 levels. Today, current guidelines regarding female patients with adnexal masses include the application of International Ovarian Tumours Analysis simple rules, logistic regression model 1 (LR1) and LR2, OVERA, cancer ovarii non-invasive assessment of treating strategy, and assessment of Different Neoplasias in the adnexa. In this scenario, the choice of the scoring system for the discrimination between benign and malignant ovarian tumours can be complex when approaching patients with adnexal masses. This review aims to summarise the available evidence regarding the different scoring systems to provide a complete overview of the topic.

Comparison of HE 4, CA 125, ROMA score and ultrasound score in the differential diagnosis of ovarian masses

OBJECTIVE

To compare the value of ultrasonography (USG) score, cancer antigen 125 (CA 125), human epididymis protein 4 (HE 4) and risk of malignancy algorithm (ROMA) in differential diagnosis ovarian masses.

METHODS

This prospective study was conducted between May 2012 and September 2013 in a single center. Eighty-four women who had an ovarian mass on imaging and underwent surgery were included. The diagnostic performances of CA 125, HE 4, ROMA score and USG score for ovarian cancer were analyzed.

RESULTS

There were 65 (77.3 %) women with benign ovarian tumors and 19 (22.7 %) women with malignant ovarian tumors. According to receiver operating characteristic (ROC) analysis; area under curve (AUC) was 0.874 for the USG score (p < 0.001), 0.794 for the CA 125 (p < 0.001), 0.9 for the HE 4 (p < 0.001), and 0.893 for the ROMA (p < 0.001). The USG score ≥ 3 had a sensitivity of 68.4 % and specificity of 90.7 %.The CA 125 ≥ 35 IU/l, had a sensitivity of 84.2 %, specificity of 49.2 %, the HE 4 ≥ 150 pM, had a sensitivity of 84.2 %, specificity of 98.4 % and the ROMA score had a sensitivity of 84.2 %, specificity of 75.3 % CONCLUSION: The HE 4 had higher accuracy than ROMA score, USG score and CA 125, in predicting ovarian cancer. Besides, the USG score was a simple and achievable method with acceptable performance.

Is There a Role of Elevated CA 19-9 Levels in the Evaluation of Clinical Characteristics of Mature Cystic Ovarian Teratomas? A Systematic Review and Meta-analysis

The role of preoperative CA 19-9 levels in patients with ovarian mature cystic teratoma (MCT) and the association of elevated levels of the biomarker with patients’ and tumor characteristics were evaluated. Four electronic databases were searched for articles published up to September 2019. Trials that evaluated the significance of elevated CA 19-9 in patients with ovarian MCTs and publications with > 20 patients were considered eligible for inclusion. Seven studies that included 995 patients with an ovarian MCT who were evaluated with elevated (n = 364) or normal (n = 631) CA 19-9 levels were included. Mean tumor size was significantly increased in patients with elevated CA 19-9 levels (p = 0.038). The rate of ovarian torsion was significantly increased in the elevated CA 19-9 group (p = 0.04). The present study highlights the importance of CA 19-9 as a marker in the diagnosis of MCT, and a meta-analysis supports that it could raise a high degree of clinical suspicion of early recognition of torsion and early surgical management due to complications related to increased size. Nonetheless, the diagnostic value of CA 19-9 is still limited and CA 19-9 can still serve only as a supplementary diagnostic tool in patients with MCTs.

Clinical and pathological features of women with adnexal masses admitted as emergency cases to the Gynaecology Department of West Kazakhstan University

Aim of the study

To detect the clinical and pathological features of women with adnexal masses (AMs) admitted as emergency cases to the Gynaecology Department of West Kazakhstan University.

Material and methods

A retrospective analysis of the data of women with AMs admitted as an emergency cases to the Gynaecology Department of West Kazakhstan University. The collected data include: age, age of menarche and age of menopause, presenting symptoms, admission criteria – either self-referral or refereed from another department, ultrasound findings, associated pregnancy, associated pathology of the female genital tract, and post-operative histological results of surgically excised AMs (gold standard).

Results

77.04% (245/318) of the studied AMs were found in the reproductive age group. The main causes for surgical intervention for the studied AMs was ruptured ovarian cyst in 27.1% or adnexal torsion in 9.7%. The available histological results of the surgically managed AMs showed the following: functional ovarian cyst in 36.2% (115/318), benign ovarian neoplasms (BONs) in 18.55% (59/318), and borderline malignant ovarian tumours in 0.63% (2/318). 44.34% of the studied AMs were associated with pregnancy, 49.3% with chronic tubo-ovarain diseases such as salpingo-oophoritis, 14.8% with cervical pathology and pelvic inflammatory diseases, 11.3% with uterine leiomyomas, and 4.4% with endometrial hyperplasia.

Conclusions

AMs were more common in the reproductive age group (77.04%), and 44.34% of the studied AMs were associated with pregnancy. The main causes of surgical intervention for the studied AMs were ruptured ovarian cyst in 27.1% or adnexal torsion in 9.7%.

Diagnostic extended usefulness of RMI: comparison of four risk of malignancy index in preoperative differentiation of borderline ovarian tumors and benign ovarian tumors

Background

This study aimed to examine the performance of the four risk of malignancy index (RMI) in discriminating borderline ovarian tumors (BOTs) and benign ovarian masses in daily clinical practice.

Methods

A total of 162 women with BOTs and 379 women with benign ovarian tumors diagnosed at the Second Affiliated Hospital of Harbin Medical University from January 2012 to December 2016 were enrolled in this retrospective study. Also, we classified these patients into serous borderline ovarian tumor (SBOT) and mucinous borderline ovarian tumor (MBOT) subgroup. Preoperative ultrasound findings, cancer antigen 125 (CA125) and menopausal status were reviewed. The area under the curve (AUC) of receiver operator characteristic curves (ROC) and performance indices of RMI I, RMI II, RMI III and RMI IV were calculated and compared for discrimination between benign ovarian tumors and BOTs.

Results

RMI I had the highest AUC (0.825, 95% CI: 0.790–0.856) among the four RMIs in BOTs group. Similar results were found in SBOT (0.839, 95% CI: 0.804–0.871) and MBOT (0.791, 95% CI: 0.749–0.829) subgroups. RMI I had the highest specificity among the BOTs group (87.6, 95% CI: 83.9–90.7%), SBOT (87.6, 95% CI: 83.9–90.7%) and MBOT group (87.6, 95% CI: 83.9–90.7%). RMI II scored the highest overall in terms of sensitivity among the BOTs group (69.75, 95% CI: 62.1–76.7%), SBOT (74.34, 95% CI: 65.3–82.1%) and MBOT (59.18, 95% CI: 44.2–73.0%) group.

Conclusion

Compared to other RMIs, RMI I was the best-performed method for differentiation of BOTs from benign ovarian tumors. At the same time, RMI I also performed best in the discrimination SBOT from benign ovarian tumors.

Surgical treatment of large adnexal masses: a retrospective analysis of 330 consecutive cases

Introduction: In the present study, perioperative outcomes of laparoscopy (LPS) were compared to open surgery (OS) for the treatment of large adnexal masses (AM).Material and methods: Retrospective observational cohort study. Data of consecutive patients who underwent ovarian cystectomy or salpingo-oophorectomy for large AM (diameter ≥10 cm) at a referral minimally invasive gynecologic center were analyzed. Propensity score match (PSM) analysis was used to minimize covariate imbalances between the two groups.Results: Overall 330 patients, 285 (86.4%) LPSs and 45 (13.6%) OSs were included. PSM showed LPS (vs. OS) to be associated with less intraoperative blood loss (mL: 131.1 ± 52.6 vs. 545.5 ± 101.2; p = .007), shorter operative time (min: 84.8 ± 77.9 vs. 123.7 ± 70.1; p < .001), but higher rate of spillage (54.5% vs. 12.1%; p < .001). Among the LPS group, a positive correlation between AM size and both conversion to open surgery and need for mini-laparotomy was found (p < .05).Conclusions: An accurate patient selection, a dedicated workup, and an appropriate counselling are mandatory before LPS for large AM. The increased risks of intraoperative spillage associated with the minimally invasive approach should be acknowledged.

A preliminary study: The sequential use of the risk malignancy index and contrast-enhanced ultrasonography in differential diagnosis of adnexal masses

The aim of this study was to explore the sequential use of risk malignancy index (RMI) combined with contrast-enhanced ultrasonography (CEUS) in identification diagnosis of adnexal masses.This study contained 2 steps: first, 151 patients were analyzed retrospectively with RMI 1, RMI 2, and RMI 3 indices; receiver operating characteristic (ROC) curves were plotted to analyze area under the curves (AUC), and then RMI cut-off value was obtained according to maximum Youden index (YI, Sensitivity + Specificity - 1) and calculating diagnostic sensitivity, specificity, positive/negative predictive value, and accuracy. Second, 151 cases were divided into 2 groups randomly (105 in study group and 46 in test group); in the study group, the lower cut-off value (LC), upper cut-off value (UC), CEUS cut-off value according to maximum YI, and then these cut-offs were validated in test group.There was no statistical significance in 3 RMI models (P = .35), and RMI1 model was established randomly for following study. When the RMI1 cut-off value was 149, the YI was maximal (0.53), and the sensitivity, specificity, positive/negative predictive value, and accuracy were 71.0%, 81.7%, 77.1%, 75.6%, and 76.2%, respectively. The LC was 15 (sensitivity was 98.0%), the UC was 3000 (specificity was 98.0%), and the CEUS cut-off value was 7 (maximal YI was 0.81). In the test group (46 cases), combining RMI1 LC (15) and UC (3000) with CEUS cut-off value (7), the sensitivity, specificity, positive/negative predictive value, and accuracy were up to 85.7%, 92.0%, 90.0%, 88.5%, and 89.1%, respectively.CEUS can help RMI to make a more effective differential diagnosis of the adnexal mass. Further validation by additional multicenter prospective trials is required.

Intra-Operative Frozen Sections for Ovarian Tumors – A Tertiary Center Experience

Background: Accuracy of diagnosis with intra-operative frozen sections is extremely important in the evaluation of ovarian tumors so that appropriate surgical procedures can be selected. Study design: All patients who with intra-operative frozen sections for ovarian masses in a tertiary center over nine years from June 2008 until April 2017 were reviewed. Frozen section diagnosis and final histopathological reports were compared. Main outcome measures: Sensitivity, specificity, positive and negative predictive values of intra-operative frozen section as compared to final histopathological results for ovarian tumors. Results: A total of 92 cases were recruited for final evaluation. The frozen section diagnoses were comparable with the final histopathological reports in 83.7% of cases. The sensitivity, specificity, positive predictive value and negative predictive value for benign and malignant ovarian tumors were 95.6%, 85.1%, 86.0% and 95.2% and 69.2%, 100%, 100% and 89.2% respectively. For borderline ovarian tumors, the sensitivity and specificity were 76.2% and 88.7%, respectively; the positive predictive value was 66.7% and the negative predictive value was 92.7%. Conclusion: The accuracy of intra-operative frozen section diagnoses for ovarian tumors is high and this approach remains a reliable option in assessing ovarian masses intra-operatively.