Abstract
Gene expression profiling has been vastly used to extract the genes that can predict the clinical outcome in patients with diverse cancers, including diffuse large B-cell lymphoma (DLBCL). With the aid of bioinformatics and computational analysis on gene expression data, various prognostic gene signatures for DLBCL have been recently developed. The major drawback of the previous signatures is their inability to correctly predict survival in external data sets. In other words, they are not reproducible in other datasets. Hence, in this study, we sought to determine the gene(s) that can reproducibly and robustly predict survival in patients with DLBCL. Gene expression data were extracted from 7 datasets containing 1636 patients (GSE10846 [n=420], GSE31312 [n=470], GSE11318 [n=203], GSE32918 [n=172], GSE4475 [n=123], GSE69051 [n=157], and GSE34171 [n=91]). Genes significantly associated with overall survival were detected using the univariate Cox proportional hazards analysis with a P value <0.001 and a false discovery rate (FDR) <5%. Thereafter, significant genes common between all the datasets were extracted. Additionally, chromosomal aberrations in the corresponding region of the final common gene(s) were evaluated as copy number alterations using the single nucleotide polymorphism (SNP) data of 570 patients with DLBCL (GSE58718 [n=242], GSE57277 [n=148], and GSE34171 [n=180]). Our results indicated that reticulon family gene 1 (RTN1) was the only gene that met our rigorous pipeline criteria and associated with a favorable clinical outcome in all the datasets (P<0.001, FDR<5%). In the multivariate Cox proportional hazards analysis, this gene remained independent of the routine international prognostic index components (i.e., age, stage, lactate dehydrogenase level, Eastern Cooperative Oncology Group [ECOG] performance status, and number of extranodal sites) (P<0.0001). Furthermore, no significant chromosomal aberration was found in the RTN1 genomic region (14q23.1: Start 59,595,976/ End 59,870,966).
Data Mining Approaches to Diffuse Large B–Cell Lymphoma Gene Expression Data Interpretation