Integrated Multiomic Analysis Reveals disulfidptosis Subtypes in Glioblastoma

Glioblastoma (GBM) stands out as the most aggressive and deadly brain tumor, presenting an alarming scarcity of viable treatment alternatives. Recognizing the pronounced heterogeneity in the prognosis of GBM patients and subtyping them is crucial for precision treatment. disulfidptosis, recently identified as a form of cell death, is closely associated with tumor glucose metabolism abnormalities and disulfide stress, primarily occurring in tumors with elevated expression of SLC7A11. Previous studies have shown that gliomas exhibit high expression of SLC7A11, high metabolism, and glucose deficiency. Therefore, we conducted an in-depth investigation into the role of disulfidptosis in glioblastoma. Samples from glioblastoma patients who underwent surgical treatment at Tiantan Hospital from January 2022 to December 2023 were collected for transcriptome sequencing. Simultaneously, patient data from the CGGA and TCGA databases were gathered. Based on the disulfidptosis features, GBM patients were classified into two subtypes using consensus clustering, with patients in disulfidptosis related group cluster(DRGcluster) A exhibiting significantly improved overall survival (OS), whereas those in DRGcluster B demonstrated the opposite trend. Additionally, the two DRGcluster subtypes exhibited distinct patterns in the tumor immune microenvironment, including differences in immune cell infiltration or cytokine expression. Additional scrutiny of immune dysfunction, exclusion, and subclass mapping in tumor analysis revealed a heightened likelihood of positive responses to immunotherapy, notably anti-PD1 treatment, among patients categorized in DRGcluster A. The pRRophetic algorithm further uncovered noteworthy distinctions in IC50 values for prevalent chemotherapeutic and targeted treatments across diverse DRGclusters. Lastly, utilizing gene sets closely associated with glioblastoma occurrence from the  Weighted correlation network analysis (WGCNA) model and significantly differential gene sets between DRGcluster A and B patients, we developed an 8-gene disulfidptosis high-low risk subtype predictor using the Least absolute shrinkage and selection operator (LASSO) machine learning algorithm. The predictor was validated for survival in an external independent GBM cohort from CGGA. This novel disulfidptosis-based classification holds promise as a prognostic predictor for GBM. Furthermore, the pRRophetic algorithm indicated significant differences in half maximal inhibitory concentration 50 (IC50) values for common chemotherapy and targeted therapy among patients in different risk groups, suggesting potential guidance for physicians in selecting patients with advantages in chemotherapy and targeted therapy.