Background: Individuals with low-grade glioma (LGG) have a dismal prognosis, and most patients will eventually progress to high-grade disease. Therefore, it is crucial to accurately determine their prognoses.
Methods: Seventy-nine NK cell genes were downloaded from the LM22 database and univariate Cox regression analysis was utilized to detect NK cell-related genes affecting prognosis. Molecular types were established for LGG using the "ConsensusClusterPlus" R package. The results from a functional enrichment analysis and the immune microenvironment were intensively explored to determine molecular heterogeneity and immune characteristics across distinct subtypes. Furthermore, a RiskScore model was developed and verified using expression profiles of NK cells, and a nomogram consisting of the RiskScore model and clinical traits was constructed. Moreover, pan-cancer traits of NK cells were also investigated.
Results: The C1 subtype included the greatest amount of immune infiltration and the poorest prognosis among well-established subtypes. The majority of enriched pathways were those involved in tumor progression, including epithelial-mesenchymal transition and cell cycle pathways. Differentially expressed genes among distinct subtypes were determined and used to develop a novel RiskScore model. This model was able to distinguish low-risk patients with LGG from those with high-risk disease. An accurate nomogram including the RiskScore, disease grade and patient's age was constructed to predict clinical outcomes of LGG patients. Finally, a pan-cancer analysis further highlighted the crucial roles of NK cell-related genes in the tumor microenvironment.
Conclusions: An NK cell-related RiskScore model can accurately predict the prognoses of patients with LGG and provide valuable insights into personalized medicine.
Keywords: RiskScore model; bioinformatics; low-grade glioma (LGG); natural killer (NK) cells; tumor immune microenvironment.