ABSTRACT
Background: The tumor immune microenvironment (TIME) plays an important role in the development and prognosis of bladder cancer. It is essential to conduct a risk model to explore the prognostic value of the immunologic genes and establish an individualized prognostic signature for predicting the survival of patients with bladder cancer. Method: The differentially expressed immunologic genes (DEGs) are identified in The Cancer Genome Atlas (TCGA). The nonnegative matrix factorization (NMF) was used to stratify the DEGs in TCGA. We used the least absolute shrinkage and selection operator (LASSO) Cox regression and univariate Cox analysis to establish a prognostic risk model. A nomogram was used to establish an individualized prognostic signature for predicting survival. The potential pathways underlying the model were explored. Results: A total of 1,018 DEGs were screened. All samples were divided into two clusters (C1 and C2) by NMF with different immune cell infiltration, and the C2 subtype had poor prognosis. We constructed a 15-gene prognostic risk model from TCGA cohort. The patients from the high-risk group had a poor overall survival rate compared with the low-risk group. Time-dependent ROC curves demonstrated good predictive ability of the signature (0.827, 0.802, and 0.812 for 1-, 3-, and 5-year survival, respectively). Univariate and multivariate Cox regression analyses showed that the immunologic prognostic risk model was an independent factor. The decision curve demonstrated a relatively good performance of the risk model and individualized prognostic signature, showing the best net benefit for 1-, 3-, and 5-year OS. Gene aggregation analysis showed that the high-risk group was mainly concentrated in tumorigenesis and migration and immune signaling pathways. Conclusion: We established a risk model and an individualized prognostic signature, and these may be useful biomarkers for prognostic prediction of patients with bladder cancer.
ABSTRACT
The aim of the present study was to develop a novel method for identifying pathways associated with renal cell carcinoma (RCC) based on a gene co-expression network. A framework was established where a co-expression network was derived from the database as well as various co-expression approaches. First, the backbone of the network based on differentially expressed (DE) genes between RCC patients and normal controls was constructed by the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database. The differentially co-expressed links were detected by Pearson's correlation, the empirical Bayesian (EB) approach and Weighted Gene Co-expression Network Analysis (WGCNA). The co-expressed gene pairs were merged by a rank-based algorithm. We obtained 842; 371; 2,883 and 1,595 co-expressed gene pairs from the co-expression networks of the STRING database, Pearson's correlation EB method and WGCNA, respectively. Two hundred and eighty-one differentially co-expressed (DC) gene pairs were obtained from the merged network using this novel method. Pathway enrichment analysis based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database and the network enrichment analysis (NEA) method were performed to verify feasibility of the merged method. Results of the KEGG and NEA pathway analyses showed that the network was associated with RCC. The suggested method was computationally efficient to identify pathways associated with RCC and has been identified as a useful complement to traditional co-expression analysis.
