Integrated multiomic predictors for ovarian cancer survival

Predicting Survival Among Colorectal Cancer Patients: Development and Validation of Polygenic Survival Score

Purpose

Colorectal cancer is the second leading cause of cancer-related death in the United States. A multi-omics approach has contributed in identifying various cancer-specific mutations, epigenetic alterations, and cells response to chemotherapy. This study aimed to determine the factors associated with colorectal cancer survival and develop and validate a polygenic survival scoring system (PSS) using a multi-omics approach.

Patients and Methods

Data were obtained from the Cancer Genome Atlas (TCGA). Colon Adenocarcinoma (TCGA-COAD) data were used to develop a survival prediction model and PSS, whereas rectal adenocarcinoma (TCGA-READ) data were used to validate the PSS. Cox proportional hazards regression analysis was conducted to examine the association between the demographic characteristics, clinical variables, and mRNA gene expression.

Results

Overall accuracy of PSS was also evaluated. The median overall survival for TCGA-COAD patients was 7 years and for TCGA-READ patients was 5 years. The multivariate Cox proportional hazards model identified age, cancer stage, and expression of nine genes as predictors of colon cancer survival. Based on the median PSS of 0.38, 48% of TCGA-COAD patients had high mortality risk. Patients in the low risk group had significantly higher 5-year survival rates than those in the high group (p <0.0001). The PSS demonstrated a high overall accuracy in predicting colorectal cancer survival.

Conclusion

This study integrated clinical and transcriptome data to identify survival predictors in patients with colorectal cancer. PSS is an accurate and valid measure for estimating colorectal cancer survival. Thus, it can serve as an important tool for future colorectal cancer research.

A chemotherapy response prediction model derived from tumor-promoting B and Tregs and proinflammatory macrophages in HGSOC

Background

Most patients with high-grade serous ovarian cancer (HGSOC) experienced disease recurrence with cumulative chemoresistance, leading to treatment failure. However, few biomarkers are currently available in clinical practice that can accurately predict chemotherapy response. The tumor immune microenvironment is critical for cancer development, and its transcriptomic profile may be associated with treatment response and differential outcomes. The aim of this study was to develop a new predictive signature for chemotherapy in patients with HGSOC.

Methods

Two HGSOC single-cell RNA sequencing datasets from patients receiving chemotherapy were reinvestigated. The subtypes of endoplasmic reticulum stress-related XBP1+ B cells, invasive metastasis-related ACTB+ Tregs, and proinflammatory-related macrophage subtypes with good predictive power and associated with chemotherapy response were identified. These results were verified in an independent HGSOC bulk RNA-seq dataset for chemotherapy. Further validation in clinical cohorts used quantitative real-time PCR (qRT-PCR).

Results

By combining cluster-specific genes for the aforementioned cell subtypes, we constructed a chemotherapy response prediction model containing 43 signature genes that achieved an area under the receiver operator curve (AUC) of 0.97 (p = 2.1e-07) for the GSE156699 cohort (88 samples). A huge improvement was achieved compared to existing prediction models with a maximum AUC of 0.74. In addition, its predictive capability was validated in multiple independent bulk RNA-seq datasets. The qRT-PCR results demonstrate that the expression of the six genes has the highest diagnostic value, consistent with the trend observed in the analysis of public data.

Conclusions

The developed chemotherapy response prediction model can be used as a valuable clinical decision tool to guide chemotherapy in HGSOC patients.

Toward More Comprehensive Homologous Recombination Deficiency Assays in Ovarian Cancer, Part 1: Technical Considerations

Simple Summary

High-grade serous ovarian cancer (HGSOC) is the most frequent and lethal form of ovarian cancer and is associated with homologous recombination deficiency (HRD) in 50% of cases. This specific alteration is associated with sensitivity to PARP inhibitors (PARPis). Despite vast prognostic improvements due to PARPis, current molecular assays assessing HRD status suffer from several limitations, and there is an urgent need for a more accurate evaluation. In these companion reviews (Part 1: Technical considerations; Part 2: Medical perspectives), we develop an integrative review to provide physicians and researchers involved in HGSOC management with a holistic perspective, from translational research to clinical applications.

Abstract

High-grade serous ovarian cancer (HGSOC), the most frequent and lethal form of ovarian cancer, exhibits homologous recombination deficiency (HRD) in 50% of cases. In addition to mutations in BRCA1 and BRCA2, which are the best known thus far, defects can also be caused by diverse alterations to homologous recombination-related genes or epigenetic patterns. HRD leads to genomic instability (genomic scars) and is associated with PARP inhibitor (PARPi) sensitivity. HRD is currently assessed through BRCA1/2 analysis, which produces a genomic instability score (GIS). However, despite substantial clinical achievements, FDA-approved companion diagnostics (CDx) based on GISs have important limitations. Indeed, despite the use of GIS in clinical practice, the relevance of such assays remains controversial. Although international guidelines include companion diagnostics as part of HGSOC frontline management, they also underscore the need for more powerful and alternative approaches for assessing patient eligibility to PARP inhibitors. In these companion reviews, we review and present evidence to date regarding HRD definitions, achievements and limitations in HGSOC. Part 1 is dedicated to technical considerations and proposed perspectives that could lead to a more comprehensive and dynamic assessment of HR, while Part 2 provides a more integrated approach for clinicians.

Prognostic value of E-Cadherin and its tumor suppressor role in Saudi women with advanced epithelial ovarian cancer

The extracellular matrix (ECM) disruption and cytoskeleton reorganization are crucial events in tumor proliferation and invasion. E-Cadherin (E-CAD) is a member of cell adhesion molecules involved in cell-cell junctions and ECM stability. The loss of E-CAD expression is associated with cancer progression and metastasis. This retrospective study aimed to assess E-CAD protein expression in ovarian cancer (OC) tissues and to evaluate its prognostic value. Patients and Methods: 143 formalin-fixed and paraffin-embedded (FFPE) blocks of primary advanced stages OC were retrieved and used to construct Tissue microarrays. Automated immunohistochemistry technique was performed to evaluate E-CAD protein expression patterns in OC. Results: E-CAD protein expression was significantly correlated with OC histological subtype (p < 0.0001), while borderline significant correlations were observed with both tumor grade (p = 0.06) and stage (p = 0.07). Interestingly, Kaplan-Meier survival analysis showed that OC patients with membranous E-CAD expression survived longer than those with no E-CAD expression mainly those at advanced stages (p < 0.009). Further in silico analysis confirms the key roles of E-CAD in OC molecular functions. Conclusion: we reported a prognosis value of membranous E-CAD in advanced stage OC patients. Further validation using larger cohorts is recommended to extract clinically relevant outcomes towards better OC management and individualized oncology.

Integrated multi-omics analysis of ovarian cancer using variational autoencoders

Cancer is a complex disease that deregulates cellular functions at various molecular levels (e.g., DNA, RNA, and proteins). Integrated multi-omics analysis of data from these levels is necessary to understand the aberrant cellular functions accountable for cancer and its development. In recent years, Deep Learning (DL) approaches have become a useful tool in integrated multi-omics analysis of cancer data. However, high dimensional multi-omics data are generally imbalanced with too many molecular features and relatively few patient samples. This imbalance makes a DL based integrated multi-omics analysis difficult. DL-based dimensionality reduction technique, including variational autoencoder (VAE), is a potential solution to balance high dimensional multi-omics data. However, there are few VAE-based integrated multi-omics analyses, and they are limited to pancancer. In this work, we did an integrated multi-omics analysis of ovarian cancer using the compressed features learned through VAE and an improved version of VAE, namely Maximum Mean Discrepancy VAE (MMD-VAE). First, we designed and developed a DL architecture for VAE and MMD-VAE. Then we used the architecture for mono-omics, integrated di-omics and tri-omics data analysis of ovarian cancer through cancer samples identification, molecular subtypes clustering and classification, and survival analysis. The results show that MMD-VAE and VAE-based compressed features can respectively classify the transcriptional subtypes of the TCGA datasets with an accuracy in the range of 93.2-95.5% and 87.1-95.7%. Also, survival analysis results show that VAE and MMD-VAE based compressed representation of omics data can be used in cancer prognosis. Based on the results, we can conclude that (i) VAE and MMD-VAE outperform existing dimensionality reduction techniques, (ii) integrated multi-omics analyses perform better or similar compared to their mono-omics counterparts, and (iii) MMD-VAE performs better than VAE in most omics dataset.

Creation and validation of models to predict response to primary treatment in serous ovarian cancer

Nearly a third of patients with high-grade serous ovarian cancer (HGSC) do not respond to initial therapy and have an overall poor prognosis. However, there are no validated tools that accurately predict which patients will not respond. Our objective is to create and validate accurate models of prediction for treatment response in HGSC. This is a retrospective case–control study that integrates comprehensive clinical and genomic data from 88 patients with HGSC from a single institution. Responders were those patients with a progression-free survival of at least 6 months after treatment. Only patients with complete clinical information and frozen specimen at surgery were included. Gene, miRNA, exon, and long non-coding RNA (lncRNA) expression, gene copy number, genomic variation, and fusion-gene determination were extracted from RNA-sequencing data. DNA methylation analysis was performed. Initial selection of informative variables was performed with univariate ANOVA with cross-validation. Significant variables (p < 0.05) were included in multivariate lasso regression prediction models. Initial models included only one variable. Variables were then combined to create complex models. Model performance was measured with area under the curve (AUC). Validation of all models was performed using TCGA HGSC database. By integrating clinical and genomic variables, we achieved prediction performances of over 95% in AUC. Most performances in the validation set did not differ from the training set. Models with DNA methylation or lncRNA underperformed in the validation set. Integrating comprehensive clinical and genomic data from patients with HGSC results in accurate and robust prediction models of treatment response.