Machine learning-based screening of an epithelial-mesenchymal transition-related long non-coding RNA signature reveals lower-grade glioma prognosis and the tumor microenvironment and predicts antitumor therapy response

Epithelial-mesenchymal transition (EMT) confers high invasive and migratory capacity to cancer cells, which limits the effectiveness of tumor therapy. Long non-coding RNAs (lncRNAs) can regulate the dynamic process of EMT at different levels through various complex regulatory networks. We aimed to comprehensively analyze and screen EMT-related lncRNAs to characterize lower-grade glioma (LGG) tumor biology and provide new ideas for current therapeutic approaches. We retrieved 1065 LGG samples from the Cancer Genome Atlas and Chinese Glioma Genome Atlas by machine learning algorithms, identified three hub lncRNAs including CRNDE, LINC00665, and NEAT1, and established an EMT-related lncRNA signature (EMTrLS). This novel signature had strong prognostic value and potential clinical significance. EMTrLS described LGG genomic alterations and clinical features including gene mutations, tumor mutational burden, World Health Organization (WHO) grade, IDH status, and 1p/19q status. Notably, stratified analysis revealed activation of malignancy-related and metabolic pathways in the EMTrLS-high cohort. Moreover, the population with increased EMTrLS scores had increased cells with immune killing function. However, this antitumor immune function may be suppressed by increased Tregs and macrophages. Meanwhile, the relatively high expression of immune checkpoints explained the immunosuppressive state of patients with high EMTrLS scores. Importantly, we validated this result by quantifying the course of antitumor immunity. In particular, EMTrLS stratification enabled assessment of the responsiveness of LGG to chemotherapeutic drug efficacy and PD1 blockade. In conclusion, our findings complement the foundation of molecular studies of LGG, provide valuable insight into our understanding of EMT-related lncRNAs, and offer new strategies for LGG therapy.