Multivariant Transcriptome Analysis Identifies Modules and Hub Genes Associated with Poor Outcomes in Newly Diagnosed Multiple Myeloma Patients.

The molecular mechanisms underlying chemoresistance in some newly diagnosed multiple myeloma (MM) patients receiving standard therapies (lenalidomide, bortezomib, and dexamethasone) are poorly understood. Identifying clinically relevant gene networks associated with death due to MM may uncover novel mechanisms, drug targets, and prognostic biomarkers to improve the treatment of the disease. This study used data from the MMRF CoMMpass RNA-seq dataset (N = 270) for weighted gene co-expression network analysis (WGCNA), which identified 21 modules of co-expressed genes. Genes differentially expressed in patients with poor outcomes were assessed using two independent sample t-tests (dead and alive MM patients). The clinical performance of biomarker candidates was evaluated using overall survival via a log-rank Kaplan-Meier and ROC test. Four distinct modules (M10, M13, M15, and M20) were significantly correlated with MM vital status and differentially expressed between the dead (poor outcomes) and the alive MM patients within two years. The biological functions of modules positively correlated with death (M10, M13, and M20) were G-protein coupled receptor protein, cell-cell adhesion, cell cycle regulation genes, and cellular membrane fusion genes. In contrast, a negatively correlated module to MM mortality (M15) was the regulation of B-cell activation and lymphocyte differentiation. MM biomarkers CTAG2, MAGEA6, CCND2, NEK2, and E2F2 were co-expressed in positively correlated modules to MM vital status, which was associated with MM's lower overall survival.

Identification of Gene Co-Expression Networks Associated with Consensus Molecular Subtype-1 of Colorectal Cancer.

Colorectal cancer (CRC) is driven in part by dysregulated Wnt, Ras-Raf-MAPK, TGF-ß, and PI3K-Akt signaling. The progression of CRC is also promoted by molecular alterations and heterogeneous-yet interconnected-gene mutations, chromosomal instability, transcriptomic subtypes, and immune signatures. Genomic alterations of CRC progression lead to changes in RNA expression, which support CRC metastasis. An RNA-based classification system used for CRC, known as consensus molecular subtyping (CMS), has four classes. CMS1 has the lowest survival after relapse of the four CRC CMS phenotypes. Here, we identify gene signatures and associated coding mRNAs that are co-expressed during CMS1 CRC progression. Using RNA-seq data from CRC primary tumor samples, acquired from The Cancer Genome Atlas (TCGA), we identified co-expression gene networks significantly correlated with CMS1 CRC progression. CXCL13, CXCR5, IL10, PIK3R5, PIK3AP1, CCL19, and other co-expressed genes were identified to be positively correlated with CMS1. The co-expressed eigengene networks for CMS1 were significantly and positively correlated with the TNF, WNT, and ERK1 and ERK2 signaling pathways, which together promote cell proliferation and survival. This network was also aligned with biological characteristics of CMS1 CRC, being positively correlated to right-sided tumors, microsatellite instability, chemokine-mediated signaling pathways, and immune responses. CMS1 also differentially expressed genes involved in PI3K-Akt signaling. Our findings reveal CRC gene networks related to oncogenic signaling cascades, cell activation, and positive regulation of immune responses distinguishing CMS1 from other CRC subtypes.

Multivariate transcriptome analysis identifies networks and key drivers of chronic lymphocytic leukemia relapse risk and patient survival.

BACKGROUND: Chronic lymphocytic leukemia (CLL) is an indolent heme malignancy characterized by the accumulation of CD5+ CD19+ B cells and episodes of relapse. The biological signaling that influence episodes of relapse in CLL are not fully described. Here, we identify gene networks associated with CLL relapse and survival risk. METHODS: Networks were investigated by using a novel weighted gene network co-expression analysis method and examining overrepresentation of upstream regulators and signaling pathways within co-expressed transcriptome modules across clinically annotated transcriptomes from CLL patients (N = 203). Gene Ontology analysis was used to identify biological functions overrepresented in each module. Differential Expression of modules and individual genes was assessed using an ANOVA (Binet Stage A and B relapsed patients) or T-test (SF3B1 mutations). The clinical relevance of biomarker candidates was evaluated using log-rank Kaplan Meier (survival and relapse interval) and ROC tests. RESULTS: Eight distinct modules (M2, M3, M4, M7, M9, M10, M11, M13) were significantly correlated with relapse and differentially expressed between relapsed and non-relapsed Binet Stage A CLL patients. The biological functions of modules positively correlated with relapse were carbohydrate and mRNA metabolism, whereas negatively correlated modules to relapse were protein translation associated. Additionally, M1, M3, M7, and M13 modules negatively correlated with overall survival. CLL biomarkers BTK, BCL2, and TP53 were co-expressed, while unmutated IGHV biomarker ZAP70 and cell survival-associated NOTCH1 were co-expressed in modules positively correlated with relapse and negatively correlated with survival days. CONCLUSIONS: This study provides novel insights into CLL relapse biology and pathways associated with known and novel biomarkers for relapse and overall survival. The modules associated with relapse and overall survival represented both known and novel pathways associated with CLL pathogenesis and can be a resource for the CLL research community. The hub genes of these modules, e.g., ARHGAP27P2, C1S, CASC2, CLEC3B, CRY1, CXCR5, FUT5, MID1IP1, and URAHP, can be studied further as new therapeutic targets or clinical markers to predict CLL patient outcomes.

Transcriptome Network Analysis Identifies CXCL13-CXCR5 Signaling Modules in the Prostate Tumor Immune Microenvironment.

The tumor immune microenvironment (TIME) consists of multiple cell types that contribute to the heterogeneity and complexity of prostate cancer (PCa). In this study, we sought to understand the gene-expression signature of patients with primary prostate tumors by investigating the co-expression profiles of patient samples and their corresponding clinical outcomes, in particular "disease-free months" and "disease reoccurrence". We tested the hypothesis that the CXCL13-CXCR5 axis is co-expressed with factors supporting TIME and PCa progression. Gene expression counts, with clinical attributes from PCa patients, were acquired from TCGA. Profiles of PCa patients were used to identify key drivers that influence or regulate CXCL13-CXCR5 signaling. Weighted gene co-expression network analysis (WGCNA) was applied to identify co-expression patterns among CXCL13-CXCR5, associated genes, and key genetic drivers within the CXCL13-CXCR5 signaling pathway. The processing of downloaded data files began with quality checks using NOISeq, followed by WGCNA. Our results confirmed the quality of the TCGA transcriptome data, identified 12 co-expression networks, and demonstrated that CXCL13, CXCR5 and associated genes are members of signaling networks (modules) associated with G protein coupled receptor (GPCR) responsiveness, invasion/migration, immune checkpoint, and innate immunity. We also identified top canonical pathways and upstream regulators associated with CXCL13-CXCR5 expression and function.