グリア腫における新規予後予測因子および免疫療法応答予測因子としてのCLDN22：包括的なマルチオミクス解析
Hui Zheng1, Jingsong Cheng1, Jialin Liu1
1Department of Neurosurgery, The Second Affiliated Hospital of Chongqing Medical University, Chongqing, China, cqmu.edu.cn.
Background:
The claudin gene family plays crucial roles in cancer biology, yet their comprehensive molecular characteristics and clinical implications in gliomas remain unclear.
Methods:
Multiomics data from The Cancer Genome Atlas (TCGA) were analyzed, and differential expression analysis was performed between glioma and normal samples. Consensus clustering was applied to identify molecular subtypes. Multiple machine learning algorithms, including least absolute shrinkage and selection operator (LASSO), extreme gradient boosting (XGBoost), Boruta, prediction analysis of microarrays (PAMR), and random forest, were employed for feature selection. Immune characteristics were evaluated using Estimation of STromal and Immune cells in MAlignant Tumors using Expression data (ESTIMATE), cell-type enrichment analysis by gene expression signatures (xCell), and Cell-type Identification By Estimating Relative Subsets Of RNA Transcripts (CIBERSORT) algorithms. Drug sensitivity analysis was conducted using the Genomics of Drug Sensitivity in Cancer (GDSC) database. Functional enrichment analysis was performed based on Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways.
Results:
We identified distinct regulatory patterns of claudin family genes involving CNV and DNA methylation. Consensus clustering revealed two molecular subtypes with significant differences in survival (p < 0.001) and immune profiles. CLDN22 emerged as the most robust biomarker through machine learning integration. High CLDN22 expression correlated with poor prognosis, higher tumor grade, mesenchymal subtype, and IDH wild-type status. CLDN22 showed superior predictive power for immunotherapy response compared to traditional biomarkers in multiple cohorts, particularly for anti-MAGE-A3 (AUC = 0.646), CAR-T (AUC = 0.644), and anti-PD-1 (AUC = 0.646) therapies. Functional analysis revealed CLDN22's involvement in cell adhesion, tight junction signaling, and immune cell migration. Drug sensitivity analysis identified distinct therapeutic vulnerabilities based on CLDN22 expression levels.
Conclusion:
Our comprehensive analysis establishes CLDN22 as a novel prognostic and predictive biomarker in gliomas with significant implications for patient stratification and therapeutic decision-making. These findings provide new insights into glioma biology and potential therapeutic strategies, though further experimental validation is warranted.
