T细胞基因筛选扩张型心肌病生物标志物并识别治疗药物

Exploring Key Genes and Potential Drugs of Dilated Cardiomyopathy Based on Bioinformatics Data

摘要: 目的：通过T细胞基因筛选寻找扩张型心肌病(DCM)的潜在病理生物标志物并筛选相应的治疗药物。方法：1) 收集3个DCM微阵列数据集并去除批次效应，利用CIBERSORT算法评估正常组织与DCM病变组织中的免疫浸润差异。2) 采用LASSO逻辑回归分析和多因素逻辑回归分析筛选与DCM有关的gamma delta T cells (γδT)相关基因并建立gamma delta T cells相关基因评分(DTAGS)。3) 使用ROC曲线下面积(Area under Curve, AUC)，校准曲线以及临床决策曲线(Decision Curve Analysis, DCA)对DCM诊断模型的区分度、校准度以及临床获益程度进行评估，进行GO和KEGG通路分析。4) 使用DGIdb数据库分析DCM的潜在药物。结果：1) 共纳入108个DCM心内膜样本和24个正常对照心内膜样本，免疫浸润分析显示与正常组织相比，DCM病变组织中的γδT细胞的浸润程度显著升高。2) LASSO分析和多因素分析确定了8个γδT相关基因用于构建DCM的诊断模型DTAGS，模型性能良好(AUC = 0.857)，DTAGS诊断DCM与实际诊断率有良好的一致性，决策曲线表明该诊断模型有较好的诊断效能。3) DTAGS与细胞间的信号转导、通道活性有显著相关。4) 基于DGIdb数据库筛选到Dasatinib，Pazopanib hydrochloride等药物作为DCM的潜在治疗药物。结论：我们建立了8个γδT相关基因的DTAGS诊断模型可以作为临床实践中预测DCM的有力工具。并基于这些基因筛选了DCM的潜在的治疗药物。

Abstract: Objectives: To search for potential pathological biomarkers of dilated cardiomyopathy (DCM) and screen corresponding therapeutic drugs. Methods: 1) Three DCM microarray datasets were collected and batch effects were removed, and differences in immune infiltration in normal versus DCM lesions tissues were assessed using the CIBERSORT algorithm. 2) LASSO logistic regression analysis and multi-factor logistic regression analysis were used to screen gamma delta T cells (gamma delta T) related genes associated with DCM and to establish gamma delta T cells related gene score (DTAGS). 3) Differentiation, calibration, and clinical benefit of the DCM diagnostic model was assessed using the Area under Curve (AUC), calibration curve, and Decision Curve Analysis (DCA). GO and KEGG pathway analysis was performed. 4) The DGIdb database was used to analyze potential drugs for DCM. Results: 1) A total of 108 DCM endocardial samples and 24 normal controls endocardial samples were included. Immuno-infiltration analysis showed significantly higher infiltration of γδT cells in DCM lesions tissues compared to normal tissues. 2) LASSO analysis and multifactorial analysis identified eight γδT-related genes for the construction of DTAGS, a diagnostic model for DCM. The model performance was good (AUC = 0.857). The DTAGS diagnostic DCM was in good agreement with the actual diagnostic rate, and the decision curve indicated that the diagnostic model had good diagnostic efficacy. 3) DTAGS is significantly related to signal transduction and channel activity between cells. 4) Drugs such as Dasatinib and Pazopanib hydrochloride were screened as potential therapeutic agents for DCM based on the DGIdb database. Conclusion: We developed a DTAGS diagnostic model of eight γδT-associated genes that can be used as a powerful tool for predicting DCM in clinical practice. Potential therapeutic agents for DCM were screened based on these genes.