HOX Gene Signatures Predict Survival Outcomes of Glioblastoma Patients

Survival Prediction in Glioblastoma

Abstract

Diffuse gliomas represent over 80\% of malignant brain tumors ranging from low-grade to aggressive high-grade lesions. Molecular characterization of gliomas has led to the development of a more accurate World Health Organization (WHO) classification system comprising specific glioma subtypes. The adult-type diffuse glioma was classified into three categories: astrocytomas (IDH-mutated), oligodendrogliomas (IDH-mutated and 1p/19q-deleted), and glioblastomas (IDH-wildtype (IDH-wt)) in 2021. The HOX (homologous box) gene is known as the leading gene of cell differentiation and vertebrate growth. HOX genes display important roles by regulating several hallmarks of cancer. HOX genes are virtually absent in healthy adult brains when they are detected in malignant brain tumors, namely gliomas. There is a need for new molecular biomarkers that can accurately predict patient outcomes. Our goal is to characterize glioblastoma and identify the HOX gene signatures of the outcome to understand which HOX-gene biomarkers predict bad survival in glioblastoma, IDH-wt. We used 237 and 310 IDH-wt glioblastoma gene expression (RNAseq) from The Cancer Genome Atlas (TCGA) and Chinese Glioma Genome Atlas (CGGA) respectively. To identify prognostic HOX genes, supervised analysis and COX analysis on grade 3 vs 2 and grade 4 vs 3 were performed separately, and the common significant HOX genes between supervised analysis and COX analysis. The selected 8-HOX genes (\textit{HOXA2, HOXA3, HOXC11, HOXC6, HOXC9, HOXD11, HOXD12, HOXD4}) were used for gene expression and survival analysis. Notably, 8-HOX gene signatures in the TCGA dataset split the glioblastoma cohort into two prognostic groups that strongly predict the survival probability of glioblastoma patients ($p<0.01$). Furthermore, the identified 8 HOX-gene signatures associated with patient survival were validated by using an independent CGGA dataset ($p<0.01$). Thus, glioblastoma can further be stratified into clinically relevant categories based on HOX gene expression which shows the importance of HOX genes in the outcome of IDHwt gliomas to identify relevant molecular subtypes for practical tumor classification.