Bioinformatics Study On Prognostic Markers For Cytogenetically Normal Acute Myeloid Leukemia

Acute myeloid leukemia?AML?is a highly heterogeneous hematopoietic malignancy,accounting for about 60%of adult acute leukemia.At present,based on traditional cytogenetic detection technology,abnormal karyotypes can be found in more than half of AML patients,and AML can be classified into low-risk group,intermediate-risk group and high-risk group based on the cytogenetic risk classification.Among them,nearly half of AML patients are cytogenetically normal?CN-AML?with an intermediate prognosis and markedly diverse outcomes.Therefore,it is extremely important to perform molecular screening of CN-AML to identify abnormally expressed genes with potential prognostic value,which further promoting CN-AML refinement and improvement of clinical treatment options.With genome-wide sequencing for the pathogenesis of AML,more and more leukemia genes have been identified and defined.The Cancer Genome Atlas?TCGA?and Gene Expression Omnibus?GEO?databases that serve as good research platforms to provide big data.Based on those platforms,this study integrates the AML expression data and has achieved the following three research results:1.Screening the prognosis gene STAB1 in CN-AML and exploring its mechanism and drug sensitivity.To identify genes with potential prognostic value in CN-AML,this study based on 2-year Overall Survival?OS?criteria,combining differential expression analysis with Cox multivariate analysis,and finally obtained six differentially expressed genes that independent of several clinical factors.Further,survival analysis was performed using four independent GEO data sets,and STAB1 was found to be the only gene whose expression related with OS in three GEO data sets.In TCGA dataset and GSE6891 dataset,the STAB1 high expression(STAB1^high)group had a shorter event-free survival?EFS?.Meanwhile,differential expression analysis between the STAB1^high and STAB1^low groups showed that up-regulated genes in STAB1^high group were enriched in pathways associated with cell adhesion and migration and immune responses.Further,it was confirmed by in vitro experiments that the interference of the STAB1 gene would inhibit the proliferation of KG1a and NB4 leukemia cells.Analysis of the correlation between STAB1 and cancer drug targets indicated that the STAB1^low group was more sensitive to the BCL2 inhibitor Venetoclax,and this study also confirmed this in cell lines.In summary,this work combines bioinformatics and experiments to validate STAB1 as a prognostic factor for CN-AML in multiple data sets,and explore the underlying mechanisms and therapeutic strategies.2.Based on six membrane protein genes,an MPG6 model for predicting the prognosis of CN-AML was constrcuted.Further to construct a more stable multi-gene prognostic prediction model for CN-AML patients,we combined univariate analysis and Cox multivariate analysis to analyze the TCGA CN-AML dataset.We obtained 23 membrane protein genes independent of several clinical factors.By correlation analysis and linear regression analysis,this research constructed a regression model for CN-AML prognosis:?6-Membrane Protein Genes,MPG6?Score=?0.0492×CD52?–?0.0018×CD96?+?0.0131×EMP1?+?0.2058×TSPAN2?+?0.0234×STAB1?–?0.3658×MBTPS1?.After validated in four CN-AML GEO datasets,consistent results showed that high MPG6 scores had lower survival rates than those in CN-AML patients with lower MPG6 scores.Meanwhile,after combining multiple clinical factors with MPG6 score in the logistic regression model,it can be observed that the MPG6 scoring model greatly improves the predictive performance?TCGA CN-AML dataset:AUC=0.762 vs.0.624;GSE6891 dataset,AUC=0.912 vs.0.614?,which indicates that MPG6 has good clinical application potential.3.Gene differential expression and regulatory network analysis for AML with different cytigentics.To investigate the difference between CN-AML and other two prognostic cytogenetics,this study performed a differential analysis for those three prognostic cytogenetics:favorable cytogenetics,CN-AML and poor cytogenetics.In this study,we found that 11 genes and two miRNAs were differentially expressed among AMLs of favorable,CN-AML and poor risk cytogenetics groups.DEGs were enriched in membrane related processes.In cytogenetically normal AML?CN-AML?,expression levels of most HOX genes in FLT3 or NPM1 mutated samples were higher than those non-mutated samples?p<0.05?.In survival analysis,membrane related genes and several miRNAs were significant on prognostic outcome in the comparison of favorable cytogenetics and CN-AML group.Notably,six HOXA?HOXA3/5/6/7/9/10?and three HOXB?HOXB2/3/4?genes were significantly in lower expression and higher methylation in AML with favorable cytogenetics compared with other two groups?p<0.05?.Meanwhile,the miRNA-HOX gene co-regulatory networks revealed that HOXA5 was a hub node and regulated an AML oncogene SPARC,which was a DEG in the comparisons of favorable cytogenetics vs CN-AML and poor cytogenetics vs CN-AML.In summary,based on the existing CN-AML data,this paper identified STAB1 as a prognostic marker of CN-AML and construct a CN-AML prognostic model with six membrane proteins,and verified it in multiple independent data sets.