Feed

Gliomas exhibit considerable molecular heterogeneity and immunological complexity, emphasizing the need for effective biomarkers and therapeutic targets. In this study, the Chinese Glioma Genome Atlas (CGGA-325/693) and The Cancer Genome Atlas (TCGA-LGG/GBM) cohorts were used to explore the pathological role of PHD finger protein 23 (PHF23) in gliomas. Machine learning algorithms were performed to construct a PHF23-related prognosis signature (PHF23-RPS). Our analysis revealed significant upregulation of PHF23 in high-grade gliomas, while the PHF23-RPS exhibited strong predictive performance (AUC = 0.853). Two molecular subtypes were identified; Cluster 2 was characterized as "inflamed yet immunosuppressive". This subtype displayed a tumor mutational burden (TMB) paradox, where elevated TMB failed to translate into survival benefits due to extensive M2 macrophage infiltration and checkpoint-mediated immune exhaustion. Pharmacogenomic screening and molecular dynamics simulations identified Entospletinib as a potential candidate targeting this immunosuppressive barrier, showing a stable binding affinity (-7.7 kcal/mol). Functional assays, including in vitro experiments and in vivo experiments via a male BALB/c nude mouse orthotopic glioma model (n = 6/group), confirmed that PHF23 silencing inhibited glioma malignancy. Our results identify PHF23 as a critical oncogenic driver in glioma and support the PHF23-RPS for risk stratification. Entospletinib may offer a potential immunomodulatory option for high-risk gliomas.