Application of Machine Learning Algorithms for Risk Stratification and Efficacy Evaluation in Cervical Cancer Screening Among the ASCUS/LSIL Population: Evidence from the Korean HPV Cohort Study.

Purpose: We assessed human papillomavirus (HPV) genotype-based risk stratification and the efficacy of cytology testing for cervical cancer screening in patients with atypical squamous cells of undetermined significance (ASCUS)/low-grade squamous intraepithelial lesion (LSIL). Materials and Methods: Between 2010 and 2021, we monitored 1,237 HPV-positive women with ASCUS/LSIL every 6 months for up to 60 months. HPV infections were categorized as persistent (HPV positivity consistently observed post-enrollment), negative (HPV negativity consistently observed post-enrollment), or non-persistent (neither consistently positive nor negative). HPV genotypes were grouped into high-risk (Hr) groups 1 (types 16, 18, 31, 33, 45, 52, and 58) and 2 (types 35, 39, 51, 56, 59, 66, and 68) and a low-risk group. Hr1 was subdivided into types a) 16 and 18; b) 31, 33, and 45; and c) 52 and 58. Cox regression and machine learning (ML) algorithms were used to analyze progression rates. Results: Among 1,273 participants, 17.6% with persistent HPV infections experienced disease progression versus no progression in the HPV-negative group (p<0.001). Cox analysis revealed the highest hazard ratios (HRs) for Hr1-a (11.6, p<0.001), followed by Hr1-b (9.26, p<0.001) and Hr1-c (7.21, p<0.001). HRs peaked at 12-24 months, with Hr1-a maintaining significance at 24-36 months (10.7, p=0.034). ML analysis identified the final cytology change pattern as the most significant factor, with 14-15 months the optimal time for detecting progression from the first examination. Conclusion: In ASCUS/LSIL cases, follow-up strategies should be based on HPV risk types. Annual follow-up was the most effective monitoring for detecting progression/regression.

Integrated analysis of spatial transcriptomics and CT phenotypes for unveiling the novel molecular characteristics of recurrent and non-recurrent high-grade serous ovarian cancer.

BACKGROUND: High-grade serous ovarian cancer (HGSOC), which is known for its heterogeneity, high recurrence rate, and metastasis, is often diagnosed after being dispersed in several sites, with about 80% of patients experiencing recurrence. Despite a better understanding of its metastatic nature, the survival rates of patients with HGSOC remain poor. METHODS: Our study utilized spatial transcriptomics (ST) to interpret the tumor microenvironment and computed tomography (CT) to examine spatial characteristics in eight patients with HGSOC divided into recurrent (R) and challenging-to-collect non-recurrent (NR) groups. RESULTS: By integrating ST data with public single-cell RNA sequencing data, bulk RNA sequencing data, and CT data, we identified specific cell population enrichments and differentially expressed genes that correlate with CT phenotypes. Importantly, we elucidated that tumor necrosis factor-α signaling via NF-κB, oxidative phosphorylation, G2/M checkpoint, E2F targets, and MYC targets served as an indicator of recurrence (poor prognostic markers), and these pathways were significantly enriched in both the R group and certain CT phenotypes. In addition, we identified numerous prognostic markers indicative of nonrecurrence (good prognostic markers). Downregulated expression of PTGDS was linked to a higher number of seeding sites (≥ 3) in both internal HGSOC samples and public HGSOC TCIA and TCGA samples. Additionally, lower PTGDS expression in the tumor and stromal regions was observed in the R group than in the NR group based on our ST data. Chemotaxis-related markers (CXCL14 and NTN4) and markers associated with immune modulation (DAPL1 and RNASE1) were also found to be good prognostic markers in our ST and radiogenomics analyses. CONCLUSIONS: This study demonstrates the potential of radiogenomics, combining CT and ST, for identifying diagnostic and therapeutic targets for HGSOC, marking a step towards personalized medicine.

Impact of Preoperative Gum Chewing on Postoperative Anti-Emetic Use in Robot-Assisted Laparoscopic Surgery for Benign Ovarian Masses: A Prospective, Single-Blinded Randomized Controlled Trial.

Background and Objectives: Postoperative nausea and vomiting (PONV) is a common issue for females undergoing gynecological surgeries, including those assisted by robotic systems. Despite available prophylactic measures, the incidence of PONV remains high, negatively impacting recovery and increasing healthcare costs. This study evaluates whether preoperative gum chewing reduces the need for anti-emetic drugs in females undergoing robot-assisted laparoscopic surgery for benign ovarian mass. Materials and Methods: This prospective, single-blinded, randomized controlled trial enrolled 92 adult females scheduled for robot-assisted laparoscopic surgery to treat benign ovarian mass. Following exclusions, the remaining participants were randomly assigned to either a gum-chewing group or a no-gum-chewing group. The gum-chewing group chewed sugar-free gum for 15 min in the holding area before surgery. The primary outcome measured was the need for anti-emetics to control PONV during the first hour in the post-anesthesia care unit (PACU). Secondary outcomes included the number of anti-emetic requests. No preemptive anti-emetics were administered during surgery. Results: Out of the initial 92 patients, 88 were included in the final analysis, with 44 in each group. The incidence of PONV requiring anti-emetics in the PACU was significantly lower in the gum-chewing group (79.5%) compared to the no-gum-chewing group (95.5%). Additionally, the number of anti-emetic requests was higher in the no-gum-chewing group. No postoperative complications such as tooth or jaw pain/injury or gastric content regurgitation were reported. Conclusions: Preoperative gum chewing for 15 min immediately before surgery significantly reduced the incidence of PONV in females undergoing robot-assisted laparoscopic surgery for benign ovarian mass. This simple, non-pharmacological intervention improved patient comfort and reduced the need for anti-emetic medications without any adverse effects. Further studies are needed to confirm these findings and to develop guidelines for incorporating preoperative gum chewing into clinical practice.

Major clinical research advances in gynecologic cancer in 2023: a tumultuous year for endometrial cancer.

In the 2023 series, we summarized the major clinical research advances in gynecologic oncology based on communications at the conference of Asian Society of Gynecologic Oncology Review Course. The review consisted of 1) Endometrial cancer: immune checkpoint inhibitor, antibody drug conjugates (ADCs), selective inhibitor of nuclear export, CDK4/6 inhibitors WEE1 inhibitor, poly (ADP-ribose) polymerase (PARP) inhibitors. 2) Cervical cancer: surgery in low-risk early-stage cervical cancer, therapy for locally advanced stage and advanced, metastatic, or recurrent setting; and 3) Ovarian cancer: immunotherapy, triplet therapies using immune checkpoint inhibitors along with antiangiogenic agents and PARP inhibitors, and ADCs. In 2023, the field of endometrial cancer treatment witnessed a landmark year, marked by several practice-changing outcomes with immune checkpoint inhibitors and the reliable efficacy of PARP inhibitors and ADCs.

MGMT Methylation Is Associated with Human Papillomavirus Infection in Cervical Dysplasia: A Longitudinal Study.

Cervical premalignancy/malignancy, as detected by cervical cytology or biopsy, can develop as a result of human papillomavirus (HPV) infection. Meanwhile, DNA methylation is known to be associated with carcinogenesis. In this study, we thus attempted to identify the association between MGMT methylation and persistent HPV infection using an Epi-TOP MPP assay. Integrative analysis of DNA methylation was carried out here using longitudinal cervical cytology samples of seven patients with atypical squamous cells of undetermined significance/low-grade squamous intraepithelial lesion (ASC-US/LSIL). Then, a gene expression analysis using the longitudinal cervical cytology samples and a public database (The Cancer Genome Atlas (TCGA)) was performed. Upon comparing the ASC-US or LSIL samples at the 1st collection and the paired samples at the 2nd collection more than 6 months later, we found that they became hypermethylated over time. Then, using the longitudinal data, we found that the MGMT methylation was associated with HPV infection. Moreover, TCGA dataset revealed an association between downregulated MGMT mRNA expression and poor overall survival. This decreased MGMT mRNA expression was observed to have an inverse relationship with MGMT methylation levels. In this study, we found that the MGMT methylation level could potentially serve as a valuable prognostic indicator for the transition from ASC-US/LSIL to cervical cancer.

Shallow Whole-Genome Sequencing of Cell-Free DNA (cfDNA) Detects Epithelial Ovarian Cancer and Predicts Patient Prognosis.

Despite the progress in diagnostics and therapeutics, epithelial ovarian cancer (EOC) remains a fatal disease. Using shallow whole-genome sequencing of plasma cell-free DNA (cfDNA), we investigated biomarkers that could detect EOC and predict survival. Plasma cfDNA from 40 EOC patients and 20 healthy subjects were analyzed by shallow whole-genome sequencing (WGS) to identify copy number variations (CNVs) and determine the Z-scores of genes. In addition, we also calculated the genome-wide scores (Gi scores) to quantify chromosomal instability. We found that the Gi scores could distinguish EOC patients from healthy subjects and identify various EOC histological subtypes (e.g., high-grade serous carcinoma). In addition, we characterized EOC CNVs and demonstrated a relationship between RAB25 amplification (alone or with CA125), and disease-free survival and overall survival. This study identified RAB25 amplification as a predictor of EOC patient survival. Moreover, we showed that Gi scores could detect EOC. These data demonstrated that cfDNA, detected by shallow WGS, represented a potential tool for diagnosing EOC and predicting its prognosis.

Ovarian tumor diagnosis using deep convolutional neural networks and a denoising convolutional autoencoder.

Discrimination of ovarian tumors is necessary for proper treatment. In this study, we developed a convolutional neural network model with a convolutional autoencoder (CNN-CAE) to classify ovarian tumors. A total of 1613 ultrasound images of ovaries with known pathological diagnoses were pre-processed and augmented for deep learning analysis. We designed a CNN-CAE model that removes the unnecessary information (e.g., calipers and annotations) from ultrasound images and classifies ovaries into five classes. We used fivefold cross-validation to evaluate the performance of the CNN-CAE model in terms of accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC). Gradient-weighted class activation mapping (Grad-CAM) was applied to visualize and verify the CNN-CAE model results qualitatively. In classifying normal versus ovarian tumors, the CNN-CAE model showed 97.2% accuracy, 97.2% sensitivity, and 0.9936 AUC with DenseNet121 CNN architecture. In distinguishing malignant ovarian tumors, the CNN-CAE model showed 90.12% accuracy, 86.67% sensitivity, and 0.9406 AUC with DenseNet161 CNN architecture. Grad-CAM showed that the CNN-CAE model recognizes valid texture and morphology features from the ultrasound images and classifies ovarian tumors from these features. CNN-CAE is a feasible diagnostic tool that is capable of robustly classifying ovarian tumors by eliminating marks on ultrasound images. CNN-CAE demonstrates an important application value in clinical conditions.

Whole-Exome Sequencing Reveals Clinical Potential of Circulating Tumor DNA from Peritoneal Fluid and Plasma in Endometrial Cancer.

Endometrial cancer (EC) is the most common type of gynecological cancer. Studies comparing tumor gDNA and ctDNA isolated from the plasma and peritoneal fluid of EC patients are limited. Whole-exome sequencing and P53 immunohistochemistry of 24 paired tissue, plasma, and peritoneal fluid samples from 10 EC patients were performed to analyze somatic mutations, copy number alterations, microsatellite instability, and mutational signatures. Mutations in cancer-related genes (KMT2C, NOTCH2, PRKAR1A, SDHA, and USP6) and genes related to EC (ARID1A, CTNNB1, PIK3CA, and PTEN) were identified with high frequencies among the three samples. TP53 and POLE mutations, which are highly related to the molecular classification of EC, were identified based on several key observations. The ctDNA of two patients with negative peritoneal fluid presented TP53 mutations concordant with those in tissues. ctDNA from the plasma and peritoneal fluid of a patient with positive cytology harbored both TP53 and POLE mutations, although none were detected in tissues. Additionally, the patient presented with wild type P53 immunohistochemistry, with a focal "high" expression in a "low" wild type background. The tissues and peritoneal fluid of 75% EC patients showed concordant microsatellite instability. Furthermore, we observed strong mutational concordance between the peritoneal fluid and tumors. Our data suggest that the ctDNA from peritoneal fluid might be a suitable biomarker for identifying the mutational landscape of EC and could complement tumor heterogeneity.

Molecular genetic evidence supporting diverse histogenic origins of germ cell tumors.

Germ cell tumors (GCTs) originate during the histogenesis of primordial germ cells to mature gametes. Previous studies identified five histogenic mechanisms in ovarian mature teratomas (type I: failure of meiosis I; type II: failure of meiosis II; type III: duplication of the genome of a mature gamete; type IV: no meiosis; and type V: fusion of two different ova), but those of other GCTs remain elusive. In this study, we analyzed 84 GCTs of various pathologic types to identify the histogenesis using single-nucleotide polymorphism array by analyzing copy-neutral loss of heterozygosity (CN-LOH) and copy number alterations (CNAs). We detected types I and II in ovarian teratomas, type III in ovarian teratomas and yolk sac tumors (YSTs), and type IV in all GCT types. The GCTs with multiple-type histogenesis (I-IV) (ovarian mature/immature teratomas and YST) show meiotic CN-LOH with scant CNAs. Type IV-only GCTs are either with mitotic CN-LOH and abundant CNAs (seminoma, dysgerminoma, testicular mixed GCTs) or with scant CNAs and no CN-LOH (pediatric testicular and mediastinal teratomas). The development sequences of CN-LOH and CNA are different between the multiple type (I-IV) GCTs and type IV-only GCTs. We analyzed two different histologic areas in eight GCTs (one mature teratoma with a mucin-secreting adenoma, two immature teratomas, and five mixed GCTs). We found that GCTs (mature teratoma, immature teratoma, and mixed GCT) showed different genomic alterations between histologic areas, suggesting that genomic differences within a GCT could accompany histologic differentiation. Of note, we found evidence for collision tumors in a mixed GCT. Our data indicate that GCTs may have various histogenesis and intratumoral genomic differences, which might provide important information for the identification of GCTs, especially for those with different histologic areas. © 2021 The Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Single Cell Gene Transcriptome Analysis of Ovarian Mature Teratomas.

Teratoma is a type of germ cell tumor that originates from totipotential germ cells that are present in gonads, which can differentiate into any of the cell types found in adult tissues. Ovarian teratomas are usually mature cystic teratomas (OMCTs, also known as dermoid cysts). Chromosome studies in OMCTs show that the chromosomes are uniformly homozygous with karyotype of 46, XX, indicating that they may be parthenogenic tumors that arise from a single ovum after thefirst meiotic division. However, the tissues in OMCTs have been known to be morphologically and immunophenotypically identical to the orthotopic tissues. Currently, expression profiles of tissue components in OMCTs are not known. To identify whether OMCT tissues are expressionally similar to or different from the orthotopic tissues, we adopted single-cell RNA-sequencing (scRNA-seq), and analyzed transcriptomes of individual cells in heterogenous tissues of two OMCTs. We found that transcriptome profiles of the OMCTs at single cell level were not significantly different from those of normal cells in orthotopic locations. The present data suggest that parthenogeneticlly altered OMCTs may not alter expression profiles of inrivirual tissue components in OMCTs.

Plasma Protein Biomarkers Associated with Higher Ovarian Cancer Risk in BRCA1/2 Carriers.

Ovarian cancer (OC) is the most lethal gynecologic malignancy and in-time diagnosis is limited because of the absence of effective biomarkers. Germline BRCA1/2 genetic alterations are risk factors for hereditary OC; risk-reducing salpingo-oophorectomy (RRSO) is pursued for disease prevention. However, not all healthy carriers develop the disease. Therefore, identifying predictive markers in the BRCA1/2 carrier population could help improve the identification of candidates for preventive RRSO. In this study, plasma samples from 20 OC patients (10 patients with BRCA1/2 wild type (wt) and 10 with the BRCA1/2 variant (var)) and 20 normal subjects (10 subjects with BRCA1/2wt and 10 with BRCA1/2var) were analyzed for potential biomarkers of hereditary OC. We applied a bottom-up proteomics approach, using nano-flow LC-MS to analyze depleted plasma proteome quantitatively, and potential plasma protein markers specific to the BRCA1/2 variant were identified from a comparative statistical analysis of the four groups. We obtained 1505 protein candidates from the 40 subjects, and SPARC and THBS1 were verified by enzyme-linked immunosorbent assay. Plasma SPARC and THBS1 concentrations in healthy BRCA1/2 carriers were found to be lower than in OC patients with BRCA1/2var. If plasma SPARC concentrations increase over 337.35 ng/mL or plasma THBS1 concentrations increase over 65.28 µg/mL in a healthy BRCA1/2 carrier, oophorectomy may be suggested.

Plasma-derived exosomal miR-4732-5p is a promising noninvasive diagnostic biomarker for epithelial ovarian cancer.

BACKGROUND: Exosomal miRNAs regulate gene expression and play important roles in several diseases. We used exosomal miRNA profiling to investigate diagnostic biomarkers of epithelial ovarian cancer (EOC). METHODS: In total, 55 individuals were enrolled, comprising healthy (n = 21) and EOC subjects (n = 34). Small mRNA (smRNA) sequencing and real-time PCR (RT-PCR) were performed to identify potential biomarkers. Receiver operating characteristic (ROC) curves were conducted to determine biomarker sensitivity and specificity. RESULTS: Using smRNA sequencing, we identified seven up-regulated (miR-4732-5p, miR-877-5p, miR-574-3p, let-7a-5p, let-7b-5p, let-7c-5p, and let-7f-5p) and two down-regulated miRNAs (miR-1273f and miR-342-3p) in EOC patients when compared with healthy subjects. Of these, miR-4732-5p and miR-1273f were the most up-regulated and down-regulated respectively, therefore they were selected for RT-PCR analysis. Plasma derived exosomal miR-4732-5p had an area under the ROC curve of 0.889, with 85.7% sensitivity and 82.4% specificity in distinguishing EOC patients from healthy subjects (p<0.0001) and could be a potential biomarker for monitoring the EOC progression from early stage to late stage (p = 0.018). CONCLUSIONS: Plasma derived exosomal miR-4732-5p may be a promising candidate biomarker for diagnosing EOC.

Peripheral blood BRCA1 methylation profiling to predict familial ovarian cancer.

OBJECTIVE: Familial cancer appears at a young age and its incidence is increasing. About 12% of familial ovarian cancer cases are associated with BRCA1/2 mutations (BRCAm). In this study, we investigated BRCA1 methylation may predict ovarian cancer in those with a family history of cancer (FHC) but without BRCA1/2 mutations (BRCAwt). METHODS: Using peripheral blood DNA from 55 subjects without a history of cancer [cancer(-)] and 52 ovarian cancer patients, we examined BRCA1 promoter methylation through bisulfite sequencing of the promoter and expressed the results as the cumulative methylation index. Then, we evaluated the BRCA1 promoter methylation according to BRCA1/2 germline mutations. RESULTS: BRCA1 methylation was more prevalent in the BRCAm cancer(-) group than in the BRCAwt cancer(-) group and ovarian cancer patients (p=0.031 and p=0.019, respectively). In the BRCAwt cancer(-) group, BRCA1 methylation was more prevalent in those with an FHC than in those without one and in the BRCAm cancer(-) group with an FHC (p=0.001 and p<0.001, respectively). CONCLUSION: Our data suggest a predictive role of BRCA1 methylation profile for ovarian cancer in those without a history of cancer but with an FHC. BRCA1 methylation has important implications for diagnostic and predictive testing of those with BRCAwt cancer(-) status with FHC.

An Application of Machine Learning That Uses the Magnetic Resonance Imaging Metric, Mean Apparent Diffusion Coefficient, to Differentiate between the Histological Types of Ovarian Cancer.

This retrospective single-center study included patients diagnosed with epithelial ovarian cancer (EOC) using preoperative pelvic magnetic resonance imaging (MRI). The apparent diffusion coefficient (ADC) of the axial MRI maps that included the largest solid portion of the ovarian mass was analysed. The mean ADC values (ADCmean) were derived from the regions of interest (ROIs) of each largest solid portion. Logistic regression and three types of machine learning (ML) applications were used to analyse the ADCs and clinical factors. Of the 200 patients, 103 had high-grade serous ovarian cancer (HGSOC), and 97 had non-HGSOC (endometrioid carcinoma, clear cell carcinoma, mucinous carcinoma, and low-grade serous ovarian cancer). The median ADCmean of patients with HGSOC was significantly lower than that of patients without HGSOCs. Low ADCmean and CA 19-9 levels were independent predictors for HGSOC over non-HGSOC. Compared to stage I disease, stage III disease was associated with HGSOC. Gradient boosting machine and extreme gradient boosting machine showed the highest accuracy in distinguishing between the histological findings of HGSOC versus non-HGSOC and between the five histological types of EOC. In conclusion, ADCmean, disease stage at diagnosis, and CA 19-9 level were significant factors for differentiating between EOC histological types.

Distinct genomic profiles of gestational choriocarcinoma, a unique cancer of pregnant tissues.

Little is known about genomic alterations of gestational choriocarcinoma (GC), unique cancer that originates in pregnant tissues, and the progression mechanisms from the nonmalignant complete hydatidiform mole (CHM) to GC. Whole-exome sequencing (20 GCs) and/or single-nucleotide polymorphism microarray (29 GCs) were performed. We analyzed copy-neutral loss-of-heterozygosity (CN-LOH) in 29 GCs that exhibited androgenetic CN-LOHs (20 monospermic, 8 dispermic) and no CN-LOH (one with NLRP7 mutation). Most GCs (25/29) harboring recurrent copy number alterations (CNAs) and gains on 1q21.1-q44 were significantly associated with poor prognosis. We detected five driver mutations in the GCs, most of which were chromatin remodeling gene (ARID1A, SMARCD1, and EP300) mutations but not in common cancer genes such as TP53 and KRAS. One patient's serial CHM/invasive mole/GC showed consistent CN-LOHs, but only the GC harbored CNAs, indicating that CN-LOH is an early pivotal event in HM-IM-GC development, and CNAs may be a late event that promotes CHM progression to GC. Our data indicate that GCs have unique profiles of CN-LOHs, mutations and CNAs that together differentiate GCs from non-GCs. Practically, CN-LOH and CNA profiles are useful for the molecular diagnosis of GC and the selection of GC patients with poor prognosis for more intensive treatments, respectively.

Genomic, transcriptomic, and viral integration profiles associated with recurrent/metastatic progression in high-risk human papillomavirus cervical carcinomas.

Acquisition of recurrent/metastatic potential by a tumor cell defines a critical step in malignant progression. However, understanding of metastatic progression at the molecular level is scarce for cervical carcinomas (CES). In this study, we performed genomic, transcriptomic, and viral profiling of five pairs of primary (CES-P) and matched recurrent/metastatic tumors (CES-R/M) with high risk human papillomavirus. Whole exome sequencing revealed mutation features of CES-R/M including elevated mutation burdens and prevalent copy number alterations compared to their matched CES-P. A relative deficit of APOBEC-related mutation signatures accompanying the transcriptional downregulation of APOBEC3A was observed for CES-R/M. Mutations in genes encoding epigenetic regulators were commonly observed as CES-R/M-specific alterations. Immunoprofiling and gene set analysis revealed CES-Ps were enriched with transcripts representing activated anticancer immunity such as interferon-gamma pathway, while CES-R/M exhibited upregulation of genes involved in epithelial-mesenchymal transition and angiogenesis. Viral capture sequencing revealed that integration sites remained enriched in viral E1 protein domain during malignant progression. Moreover, we found transcriptional upregulation of POSTN and downregulation of APOBEC3A were associated with unfavorable clinical outcomes in CES. Comprehensive genomic and transcriptomic profiling of a rare cohort including CES-R/M identified metastases-specific features to advance the molecular understanding into CES metastatic progression with potential clinical implications.

Incidence and clinical course of septic shock in neutropenic patients during chemotherapy for gynecological cancers.

OBJECTIVE: To identify the incidence and clinical course of septic shock combined with neutropenia during chemotherapy in gynecological cancer patients. METHODS: We retrospectively reviewed the medical records of all gynecological cancer patients who received intravenous chemotherapy between March 2009 and March 2018. Patients diagnosed with neutropenic septic shock (NSS) during the course of chemotherapy were identified. We calculated the overall incidence and mortality rate of NSS, and analyzed risk factors and clinical course. RESULTS: A total of 1,009 patients received 10,239 cycles of chemotherapy during the study period. Among these, 30 (3.0%) patients had 32 NSS events, of which 12 (1.2%) died. With respect to patient age during the first course of chemotherapy, the incidence of NSS after the age of 50 was significantly higher than that in patients under 50 (3.9% vs. 1.4%, p=0.034). As the number of chemotherapy courses increased, the incidence of NSS increased, and linear-by-linear association analysis showed a positive correlation (p=0.004). NSS events occurred on average 7.8 days after the last cycle of chemotherapy, and the median duration of vasopressor administration was 23.3 hours. The median age (64.0 vs. 56.5, p=0.017) and peak heart rate (149.5 min-1 vs. 123.5 min-1, p=0.015) were significantly higher in the group of patients who subsequently died of NSS than in those who survived. CONCLUSION: The overall incidence of NSS in gynecological cancer patients receiving chemotherapy was 3.0%, which is higher than previously estimated. Peak heart rate during NSS events may be an indicator for predicting survival.

Classification of cervical neoplasms on colposcopic photography using deep learning.

Colposcopy is widely used to detect cervical cancers, but experienced physicians who are needed for an accurate diagnosis are lacking in developing countries. Artificial intelligence (AI) has been recently used in computer-aided diagnosis showing remarkable promise. In this study, we developed and validated deep learning models to automatically classify cervical neoplasms on colposcopic photographs. Pre-trained convolutional neural networks were fine-tuned for two grading systems: the cervical intraepithelial neoplasia (CIN) system and the lower anogenital squamous terminology (LAST) system. The multi-class classification accuracies of the networks for the CIN system in the test dataset were 48.6 ± 1.3% by Inception-Resnet-v2 and 51.7 ± 5.2% by Resnet-152. The accuracies for the LAST system were 71.8 ± 1.8% and 74.7 ± 1.8%, respectively. The area under the curve (AUC) for discriminating high-risk lesions from low-risk lesions by Resnet-152 was 0.781 ± 0.020 for the CIN system and 0.708 ± 0.024 for the LAST system. The lesions requiring biopsy were also detected efficiently (AUC, 0.947 ± 0.030 by Resnet-152), and presented meaningfully on attention maps. These results may indicate the potential of the application of AI for automated reading of colposcopic photographs.

Clinical Implications of Circulating Tumor DNA from Ascites and Serial Plasma in Ovarian Cancer.

PURPOSE: The purpose of this study was to identify the clinical utility of circulating tumor DNA (ctDNA) from ascites and serial plasma samples from epithelial ovarian cancer (EOC) patients. MATERIALS AND METHODS: Using targeted next-generation sequencing, we analyzed a total of 55 EOC samples including ctDNA from ascites and serial plasma and gDNA from tumor tissues. Tumor tissues and ascites were collected during debulking surgeries and plasma samples were collected before and after the surgeries. Because one EOC patient underwent secondary debulking surgery, a total of 11 tumor tissues, 33 plasma samples, and 11 ascites samples were obtained from the 10 patients. RESULTS: Of the 10 patients, nine (90%) contained somatic mutations in both tumor tissues and ascites ctDNA. This mutational concordance was confirmed through correlation analysis. The mutational concordance between ascites and tumor tissues was valid in recurrent/progressive ovarian cancer. TP53 was the most frequently detected gene with mutations. ctDNA from serial plasma samples identified EOC progression/recurrence at a similar time or even more rapidly than cancer antigen 125, an established serum protein tumor marker for EOC. CONCLUSION: Our data suggest that ascites ctDNA can be used to identify the mutational landscape of ovarian cancer for therapeutic strategy planning.