Construction of prognostic risk prediction model based on high-throughput sequencing expression profile data in childhood acute myeloid leukemia

Abstract

This study aimed to identify critical prognostic molecular markers in Childhood acute myeloid leukemia (AML) and construct nomogram-based model for prognostic prediction. The RNA-sequencing profiles and corresponding clinical information were downloaded from TCGA database. Differential expressed genes (DEG) were screened using limma package, subsequently following by GO and KEGG pathway analysis. Univariate and multivariate cox regression analysis were performed to screen critical DEGs. Nomogram-based prediction model were constructed to identify clinical factors with independent prognostic values, and the accuracy of this model was validated. A total of 214 DEGs were identified from relapse AML samples compared with non-relapse samples. These DEGs were mainly involved in twenty GO terms and three signaling pathways, such as chromatin assembly or disassembly, cytokine-cytokine receptor interaction, and JAK-STAT signaling pathway. Among these genes, Univariate and multivariate cox regression analysis results showed that relapse and risk score were significantly correlated with survival outcomes. Finally, the accuracy ability of nomogram-based prediction model was validated. These six DEGs (ABCA5, CYP7A1, HERC5, etc.) play major roles in AMLs progression. Our nomogram-based prognostic predictive model might be an effective method to estimate survival probability of AML patients with different risk status.