胶质瘤起源于脑或脊髓的胶质细胞，是一种常见的中枢神经系统肿瘤，其恶性程度不同，影响了治疗的复杂性和难度。目前的策略包括传统手术、放射治疗、化疗和新兴的免疫治疗等，但结果有限。因此，本研究旨在优化风险分层以实现更精确的治疗方法。我们主要通过各种组学算法识别与免疫细胞浸润模式差的特征基因，并根据这些基因对胶质瘤患者进行分类，以增强患者预后评估的准确性。这种方法可以支持个体化的治疗策略，并促进新的治疗靶点的发现。我们从TCGA、CGGA和GTEx数据库获得了胶质瘤和正常脑组织的数据集。我们使用287个免疫细胞特征基因进行聚类分析。通过WGCNA筛选出与不良预后亚型（C1）相关的核心基因。TCGA数据集用作发现队列，CGGA数据集用作外部验证队列。我们利用LASSO-Cox回归建立了与免疫细胞浸润模式差相关特征基因的预后模型。对不同风险组进行了基因组异质性、肿瘤干性、通路相关性、免疫浸润模式、治疗反应和潜在药物的综合分析。我们还对98个胶质瘤样本和11个正常脑组织样本进行了基因表达验证，使用免疫组化（IHC）方法进行。使用筛选出的与免疫细胞相关的基因，将胶质瘤患者分为C1和C2亚型。C1亚型表现出更差的预后，上调基因主要富集在免疫响应、细胞外基质等方面，下调基因主要富集在神经信号转导和神经通路相关方面。我们使用了七种高级算法来阐明不同亚型的免疫细胞浸润模式。此外，WGCNA识别出与免疫浸润模式差相关的核心基因，并相应地构建了预后模型。高风险患者的生存时间较短，风险评分较高与低风险患者相比。多变量Cox回归分析显示，调整了混杂的临床因素后，风险评分是总生存期（OS）的重要独立预测因子（P < 0.001）。建立的标示图表将风险评分与WHO分级和年龄结合起来，准确预测了胶质瘤患者1、3和5年的生存率，AUC分别为0.908、0.890和0.812。这种风险评分增强了标示图表的可靠性，并指导临床决策。我们还对不同风险组进行了基因组异质性、肿瘤干性、通路相关性、免疫浸润模式、治疗反应和潜在药物的综合分析。此外，我们还对大量胶质瘤和正常脑组织样本进行了潜在PLSCR1基因的初步验证。我们对胶质瘤患者进行了优化的风险分层策略，可以提高预后评估的准确性。我们的组学研究结果不仅增进了理解与免疫细胞浸润模式差相关的特征基因功能，还为胶质瘤预后生物标志物研究和个体化治疗策略的发展提供了有价值的见解。© 2023作者，独家许可给施普林格-弗朗厄集团德国公司，属于施普林格-自然出版集团。

Gliomas, 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.We 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.Using 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.Our 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.© 2023. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.