Prediction of competing endogenous RNA coexpression network as prognostic markers in AML

Recently, competing endogenous RNAs (ceRNAs) hypothesis has gained a great interest in the study of molecular biological mechanisms of cancer occurrence and progression. However, studies on leukemia are limited, and there is still a lack of comprehensive analysis of lncRNA-miRNA-mRNA ceRNA regulatory network of AML based on high-throughput sequencing and large-scale sample size. We obtained RNA-Seq data and compared the expression profiles between 407 normal whole blood (GTEx) and 151 bone marrows of AML (TCGA). The similarity between two sets of genes with trait in the network was analyzed by weighted correlation network analysis (WGCNA). MiRcode, starBase, miRTarBase, miRDB and TargetScan was used to predict interactions between lncRNAs, miRNAs and target mRNAs. At last, we identified 108 lncRNAs, 10 miRNAs and 8 mRNAs to construct a lncRNA-miRNA-mRNA ceRNA network, which might act as prognostic biomarkers of AML. Among the network, a survival model with 8 target mRNAs (HOXA9+INSR+KRIT1+MYB+SPRY2+UBE2V1+WEE1+ZNF711) was set up by univariate and multivariate cox proportional hazard regression analysis, of which the AUC was 0.831, indicating its sensitivity and specificity in AML prognostic prediction. CeRNA networks could provide further insight into the study on gene regulation and AML prognosis.