基于单细胞数据研究泛癌相关坏死性凋亡的生物标志物

Biomarkers of Pan-Cancer-Associated Necroptosis Studied Based on Single-Cell Data

摘要: 目的：基于单细胞测序数据筛选泛癌相关坏死性凋亡的生物标志物，阐明生物标志物对肿瘤的预后和免疫治疗的价值。方法：采用高通量基因表达数据库中下载的23种癌症的单细胞转录组测序数据，基于基因名片数据库中下载的坏死性凋亡基因集对单细胞数据打分，筛选差异表达基因。通过机器学习模型筛选差异基因得到泛癌相关坏死性凋亡的基因集。采用TCGA数据库下载的33种癌症的转录组数据和临床数据，利用基因集变异分析对基因集打分。基于GO、KEGG和GSEA分析基因集的功能和通路、使用COX分析、Kaplan-Meier分析和深度生存模型分析生物标志物的预后价值，并利用深度学习生存分析模型验证基因集预后效果。使用cBioPortal数据库分析免疫浸润相关性。结果：通过单细胞数据筛选出包含34个基因的生物标志物。基因集在大多数肿瘤中表达升高，在坏死性凋亡通路上高表达，并且在葡萄膜黑色素瘤、胰腺腺癌、间皮瘤等多种癌症中与较差的预后相关。基因集的评分与多种免疫细胞浸润相关，包括巨噬细胞、T细胞。结论：基因集在多种肿瘤中高表达，并于预后不良相关，且与免疫细胞密切相关。基因集可能作为一种生物标志物和免疫治疗效果的预测因子，成为有潜力的新的治疗靶点。

Abstract: Objective: To screen pan-cancer-associated necroptosis biomarkers based on single-cell sequencing data, and to elucidate the value of biomarkers for tumor prognosis and immunotherapy. Methods: Single-cell transcriptome sequencing data of 23 cancers downloaded from a high-throughput gene expression database were used to screen differentially expressed genes by scoring the single-cell data based on the necroptosis gene set downloaded from the gene business card database. The gene set of pan-cancer-associated necroptosis was obtained by screening differential genes through machine learning model. Transcriptomic data and clinical data of 33 cancers downloaded from TCGA database were used to score the gene set using gene set variant analysis. The prognostic value of the biomarkers was analyzed based on GO, KEGG, and GSEA analysis of the gene set’s functions and pathways, using COX analysis, Kaplan-Meier analysis, and deep survival modeling, and the prognostic effect of the gene set was validated using a deep learning survival analysis model. Immune infiltration correlations were analyzed using the cBioPortal database. Results: Biomarkers containing 34 genes were screened by single-cell data. Gene sets had elevated expression in most tumors, were highly expressed on the necroptosis pathway, and were associated with poorer prognosis in several cancers, including uveal melanoma, pancreatic adenocarcinoma, and mesothelioma. Gene set scores were associated with infiltration of a variety of immune cells, including macrophages and T cells. Conclusion: Gene sets are highly expressed in a variety of tumors and are associated with poor prognosis and are strongly associated with immune cells. Gene sets may be potential new therapeutic targets as a biomarker and predictor of immunotherapy efficacy.