Optimized risk stratification strategy for glioma patients based on the feature genes of poor immune cell infiltration patterns
Heng-Tong Wan,Zhen-Jin Su,Ze-Shang Guo,Peizhen Wen,Xin-Yu Hong
DOI: https://doi.org/10.1007/s00432-023-05209-9
2023-08-03
Journal of Cancer Research and Clinical Oncology
Abstract:BackgroundGliomas, originating from glial cells within the brain or spinal cord, are common central nervous system tumors with varying degrees of malignancy that influence the complexity and difficulty of treatment. The current strategies, including traditional surgery, radiotherapy, chemotherapy, and emerging immunotherapies, have yielded limited results. As such, our study aims to optimize risk stratification for a more precise treatment approach. We primarily identify feature genes associated with poor immune cell infiltration patterns through various omics algorithms and categorize glioma patients based on these genes to enhance the accuracy of patient prognosis assessment. This approach can underpin individualized treatment strategies and facilitate the discovery of new therapeutic targets.MethodsWe procured datasets of gliomas and normal brain tissues from TCGA, CGGA, and GTEx databases. Clustering was conducted using the input of 287 immune cell feature genes. Hub genes linked with the poor prognosis subtype (C1) were filtered through WGCNA. The TCGA dataset served as the discovery cohort and the CGGA dataset as the external validation cohort. We constructed a prognostic model related to feature genes from poor immune cell infiltration patterns utilizing LASSO-Cox regression. Comprehensive analyses of genomic heterogeneity, tumor stemness, pathway relevance, immune infiltration patterns, treatment response, and potential drugs were conducted for different risk groups. Gene expression validation was performed using immunohistochemistry (IHC) on 98 glioma samples and 11 normal brain tissue samples.ResultsUsing the filtered immune cell-related genes, glioma patients were stratified into C1 and C2 subtypes through clustering. The C1 subtype exhibited a worse prognosis, with upregulated genes primarily enriched in immune response, extracellular matrix, etc., and downregulated genes predominantly enriched in neural signal transduction and neural pathway-related aspects. Seven advanced algorithms were used to elucidate immune cell infiltration patterns of different subtypes. In addition, WGCNA identified hub genes from poor immune infiltration patterns, and a prognostic model was constructed accordingly. High-risk patients demonstrated shorter survival times and higher risk scores as compared to low-risk patients. Multivariate Cox regression analysis revealed that, after adjusting for confounding clinical factors, risk score was a vital independent predictor of overall survival (OS) (P < 0.001). The established nomogram, which combined risk scores with WHO grade and age, accurately predicted glioma patient survival rates at 1, 3, and 5 years, with AUCs of 0.908, 0.890, and 0.812, respectively. This risk score enhanced the nomogram's reliability and informed clinical decision-making. We also comprehensively analyzed genomic heterogeneity, tumor stemness, pathway relevance, immune infiltration patterns, treatment response, and potential drugs for different risk groups. In addition, we conducted preliminary validation of the potential PLSCR1 gene using IHC with a large sample of gliomas and normal brain tissues.ConclusionOur optimized risk stratification strategy for glioma patients has the potential to improve the accuracy of prognosis assessment. The findings from our omics research not only enhance the understanding of the functions of feature genes related to poor immune cell infiltration patterns but also offer valuable insights for the study of glioma prognostic biomarkers and the development of individualized treatment strategies.Graphical abstract
oncology