Prognostic model construction of disulfidptosis-associated genes in brain glioma and their relevance in immune checkpoints

Abstract

Abstract Background: Disulfidptosis is a novel form of programmed cell death (PCD) triggered by disulfide stress and is implicated to have a potential role in various cancers. Although the role of disulfidptosis has been assessed in various cancers, its role in Glioma (GBMLGG) prognosis and immunotherapy response remains unexplored. Results: A prognostic analysis of 15 disulfidptosis-associated genes in a pan-cancer dataset revealed significant differences in their prognostic performance in several cancer types. An analysis of GBMLGG revealed that the prognosis of GBMLGG predicted by the 15 disulfidptosis-associated genes had grouped without any clusters, and the difference in the prognosis among the subgroups was highly significant. Subsequently, seven target genes were identified using 10 machine learning algorithms including 101 algorithm combinations, and their prognostic prediction was validated by constructing receiver operating characteristic curves and conducting Kaplan-Meier (KM) analyses, which showed very satisfactory results. The complete predictive nomogram was then constructed. Furthermore, we analyzed the correlation between the target genes and immune cells, tumor mutational burden, and immune checkpoints, and the results showed that 13 cell lines and 19 immune checkpoint–associated genes were significantly correlated with the target genes. These results were also verified through single-cell analysis. Conclusions: An in-depth analysis of GBMLGG confirmed the strong correlation between the target genes of disulfidptosis-associated cell death and GBMLGG prognosis in multiple ways. The prognostic predictive nomogram was constructed using powerful machine learning algorithms to screen for the final target genes, which demonstrated excellent results when tested using several datasets. Trial registration: Not applicable