Identification of distinctive long noncoding RNA competitive interactions and a six-methylated-gene prognostic signature in acute myeloid leukemia with -5/del(5q) or -7/del(7q).

Abstract

Acute myeloid leukemia (AML) with -5/del(5q) or -7/del(7q) has special clinical and biological characteristics, but its molecular mechanisms and risk stratification remain unknown. The RNA sequencing and DNA methylation of 23 patients with -5/del(5q) or -7/del(7q) and 128 patients with other subtypes of acute myeloid leukemia were obtained from The Cancer Genome Atlas (TCGA). The regulatory mechanisms of competitive endogenous RNA (ceRNA) network and DNA methylation on gene expression were explored. To find robust and specific risk stratification for this AML subtype, a prognostic model was established and evaluated through four independent data sets. We identified 966 differentially expressed long noncoding RNA, 2274 differentially expressed genes, and 47 differentially expressed microRNAs, and constructed a ceRNA network. After the integrated analysis of differentially methylated and expressed genes, 19 genes showed the opposite trend between the methylation variation and gene expression. An six-methylated-gene prognostic signature which highly correlated with overall survival was established, and the performance was validated by leave-one-out cross validation method and permutation test. Furthermore, the excellent prognostic value of this model was supported by an independent cohort, while specificity of this model was validated by three independent data sets, suggesting it as a predictive classifier with high efficiency for distinguishing those with -5/del(5q) or -7/del(7q) from other AML subtypes. The ceRNA network may provide new ideas for the diagnosis and treatment for patients with -5/del(5q) or -7/del(7q).The six-methylated-gene prognostic signature was a robust, specific, and clinically practical risk stratification for the outcome of patients with AML having -5/del(5q) or -7/del(7q).