Haghshenas Zahra, Nazari Elham, Khalili-Tanha Ghazaleh, Razzaghi Zahra
Proteomics Research Center, System Biology Institute, Faculty of Paramedical Sciences, Shahid Beheshti University of Medical Sciences, Tehran, Iran.
Department of Medical Genetics and Molecular Medicine, School of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran.
Med J Islam Repub Iran. 2025 Apr 1;39:49. doi: 10.47176/mjiri.39.49. eCollection 2025.
Approximately 80% of all malignant brain tumors and the most common cause of death that occur as a result of primary brain tumors belong to glioma. Hence, identifying effective biomarkers for early diagnosis and prognosis can have a significant impact on patient treatment. Recent years have witnessed a significant increase in the use of machine learning (ML) to analyze RNAseq data to identify new cancer biomarkers. In this study, diagnostic and prognostic biomarkers for Glioma were identified through the collection of patient data from the TCGA database and analysis using ML algorithms and bioinformatics.
The study used ML to analyze ribonucleic acid (RNA) expression profiles from Glioma patients (GBMLGG) to identify differentially expressed genes (DEGs). In general, the sample of 1012 patients and 35 controls, which included 613 men and 434 women, was used in this study. Biomarkers of prognosis have been identified using the Kaplan-Meier analysis of survival curves. The coexpression of DEGs, protein-protein interactions (PPIs), and the correlation between DEGs and clinical data were also examined. The receiver operating characteristic (ROC) curve analysis was used to determine diagnostic markers.
After normalization and filtering, we identified 3172 DEGs with a log fold change |FC| ≥ 1 and < 0.0.05. According to a survival analysis, 15 upregulated genes (GRAPL, LOC339240, LOC723809, NODAL, SILV, SPINK8, TAC4, ANG, CD209, F2RL2, LYZ, SLAMF7, psiTPTE22, SFRP4 and DKFZP) and 9 downregulated genes (PCDHGC5, CES8, CHD5, DNAJC6, DNM1, KIRREL3, NCOA7, RASAL1, SNCA) were associated with overall survival (OS). In addition, the ML algorithm identified 20 genes, among which PSD, TUBA4A, and PCDHGC5 were identified as candidates with high correlation coefficients.
Generally, our results showed that immune-related genes play a crucial role in the development, progression, and pathogenesis of gliomas. Five immune-related genes-including SLAMF7, CD209, TAC4, HLA-DRB68, and LYZ-were found to be diagnostic and prognostic biomarkers of the disease.
所有恶性脑肿瘤中约80%以及原发性脑肿瘤导致的最常见死亡原因都属于胶质瘤。因此，识别早期诊断和预后的有效生物标志物对患者治疗会产生重大影响。近年来，利用机器学习（ML）分析RNA测序数据以识别新的癌症生物标志物的应用显著增加。在本研究中，通过从TCGA数据库收集患者数据并使用ML算法和生物信息学进行分析，确定了胶质瘤的诊断和预后生物标志物。
该研究使用ML分析胶质瘤患者（GBMLGG）的核糖核酸（RNA）表达谱，以识别差异表达基因（DEG）。总体而言，本研究使用了1012例患者和35例对照的样本，其中包括613名男性和434名女性。使用生存曲线的Kaplan-Meier分析确定预后生物标志物。还检查了DEG的共表达、蛋白质-蛋白质相互作用（PPI）以及DEG与临床数据之间的相关性。使用受试者工作特征（ROC）曲线分析来确定诊断标志物。
经过标准化和筛选后，我们鉴定出3172个差异表达基因，其对数变化倍数|FC|≥1且<0.05。根据生存分析，15个上调基因（GRAPL、LOC339240、LOC723809、NODAL、SILV、SPINK8、TAC4、ANG、CD209、F2RL2、LYZ、SLAMF7、psiTPTE22、SFRP4和DKFZP）和9个下调基因（PCDHGC5、CES8、CHD5、DNAJC6、DNM1、KIRREL3、NCOA7、RASAL1、SNCA）与总生存期（OS）相关。此外，ML算法鉴定出20个基因，其中PSD、TUBA4A和PCDHGC5被鉴定为具有高相关系数的候选基因。
总体而言，我们的结果表明免疫相关基因在胶质瘤的发生、发展和发病机制中起关键作用。发现包括SLAMF7、CD209、TAC4、HLA-DRB68和LYZ在内的五个免疫相关基因是该疾病的诊断和预后生物标志物。
