Abstract
Acute myeloid leukemia (AML) is a heterogeneous hematological malignancy with dismal prognosis. Identification of better biomarkers remained a priority to improve established stratification and guide therapeutic decisions. Therefore, we extracted the RNA sequence data and clinical characteristics of AML from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression database (GTEx) to identify the key factors for prognosis. We found UNC93B1 was highly expressed in AML patients and significantly linked to poor clinical features (p < 0.05). We further validated the high expression of UNC93B1 in another independent AML cohort from GEO datasets (p < 0.001) and performed quantitative PCR of patient samples to confirm the overexpression of UNC93B1 in AML (p < 0.005). Moreover, we discovered high level of UNC93B1 was an independent prognostic factor for poorer outcome both in univariate analysis and multivariate regression (p < 0.001). Then we built a nomogram model based on UNC93B1 expression, age, FAB subtype and cytogenetic risk, the concordance index of which for predicting overall survival was 0.729 (p < 0.001). Time-dependent ROC analysis for predicting survival outcome at different time points by UNC93B1 showed the cumulative 2-year survival rate was 43.7%, and 5-year survival rate was 21.9%. The differentially expressed genes (DEGs) between two groups divided by UNC93B1 expression level were enriched in innate immune signaling and metabolic process pathway. Protein-protein interaction (PPI) network indicated four hub genes (S100A9, CCR1, MRC1 and CD1C) interacted with UNC93B1, three of which were also significantly linked to inferior outcome. Furthermore, we discovered high UNC93B1 tended to be infiltrated by innate immune cells, including Macrophages, Dendritic cells, Neutrophils, Eosinophils, and NK CD56dim cells. We also found UNC93B1 had a significantly positive correlation with CD14, CD68 and almost all Toll-like receptors. Finally, we revealed negatively correlated expression of UNC93B1 and BCL2 in AML and conjectured that high-UNC93B1 monocytic AML is more resistant to venetoclax. And we found high MCL-1 expression compensated for BCL-2 loss, thus, we proposed MCL-1 inhibitor might overcome the resistance of venetoclax in AML. Altogether, our findings demonstrated the utility of UNC93B1 as a powerful poor prognostic predictor and alternative therapeutic target.