DesA Prognostic Risk Model of LncRNAs in Patients With Acute Myeloid Leukaemia Based on TCGA Data

Purpose: This study aimed to combine the clinical data of acute myeloid leukaemia (AML) from The Cancer Genome Atlas (TCGA) database to obtain prognosis-related biomarkers, construct a prognostic risk model using long non-coding RNAs (lncRNAs) in AML and help patients with AML make clinical treatment decisions. Methods: We analysed the transcriptional group information of 151 patients with AML obtained from TCGA and extracted the expressions of lncRNAs. According to the mutation frequency, the patients were divided into the high mutation group (genomic unstable group, top 25% of mutation frequency) and low mutation group (genomic stable group, 25% after mutation frequency). The 'limma' R package was used to analyse the difference in lncRNA expressions between the two groups, and the "survival," "caret," and "glmnet" R packages were used to screen lncRNAs that are related to clinical prognosis. Subsequently, a prognosis-related risk model was constructed and verified through different methods. Results: According to the lncRNA expression data in TCGA, we found that seven lncRNAs (i.e. AL645608.6, LINC01436, AL645608.2, AC073534.2, LINC02593, AL512413.1, and AL645608.4) were highly correlated with the clinical prognosis of patients with AML, so we constructed a prognostic risk model of lncRNAs based on LINC01436, AC073534.2, and LINC02593. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes pathway analyses of differentially expressed lncRNA-related target genes were performed, receiver operating characteristic (ROC) curves were created, the applicability of the model in children was assessed using the TARGET database and the model was externally verified using the GEO database. Furthermore, different expression patterns of lncRNAs were validated in various AML cell lines derived from Homo sapiens. Conclusions: We have established a lncRNA prognostic model that can predict the survival of patients with AML. The Kaplan-Meier analysis showed that this model distinguished survival differences between patients with high- and low-risk status. The ROC analysis confirmed this finding and showed that the model had high prediction accuracy. The Kaplan-Meier analysis of the clinical subgroups showed that this model can predict prognosis independent of clinicopathological factors. Therefore, the proposed prognostic lncRNA risk model can be used as an independent biomarker of AML.