基于GEO数据库与R语言的喉鳞状细胞癌分子特征分析

Analysis of Molecular Characterization of Laryngeal Squamous Cell Carcinoma Based on GEO Datasets and R Language

摘要: 目的：基于GEO数据库与R语言生物信息学方法，对公开LSCC表达谱数据集GSE59102进行再分析，系统刻画其差异表达基因、GO功能和KEGG通路富集特征，并从细胞周期、细胞外基质重塑和炎症微环境等角度归纳该数据集中LSCC的主要分子改变，为后续可验证的机制研究提供线索。方法：从GEO数据库下载LSCC表达谱数据集GSE59102作为主要分析队列，GSE10288作为辅助验证队列。采用R语言对表达矩阵进行预处理、标准化和差异表达分析，以adjusted P value < 0.05且|log₂FC| > 1为筛选标准获得差异表达基因(DEGs)，并绘制火山图和热图。随后利用cluster Profiler包进行Gene Ontology (GO)和Kyoto Encyclopedia of Genes and Genomes (KEGG)富集分析，并结合既往文献对主要差异基因及其可能涉及的生物学过程进行讨论。结果：GSE59102共纳入29例肿瘤样本和13例正常/安全边界样本。按照上述阈值共筛选出3439个显著差异表达基因，包括1526个上调基因和1913个下调基因。上调基因中MMP12、HOXC13、LINC01980、CDC6、CASC9、HOXC13-AS、FGD6和MCM2等变化较为明显；下调基因主要包括PLP1、CRISP2、NUCB2、VSX1、CAB39L、CRISP3和EHF等。前50个差异基因的热图能够较好地区分肿瘤组和正常组。GO富集结果主要涉及nuclear division、chromosome segregation、extracellular matrix、basement membrane、integrin binding及cytokine activity等条目；KEGG富集主要涉及Cell cycle、DNA replication、Cytokine-cytokine receptor interaction和Complement and coagulation cascades等通路。由于GSE10288平台覆盖基因较少，两个队列共同且同向变化的基因数量有限，未进一步开展稳定性不足的联合富集分析。结论：基于GSE59102数据集的分析结果提示，LSCC组织的转录组异常主要表现为细胞周期激活、细胞外基质重塑及炎症/细胞因子信号改变。MMP12、HOXC13、CASC9、CDC6和MCM2等基因可作为后续实验验证的候选分子。由于外部验证受平台覆盖范围限制，本文结果仍主要反映单一公开数据集中的分子特征，其稳定性和普适性有待更多独立队列及实验研究进一步确认。

Abstract: Objective: To reanalyze the public LSCC expression dataset GSE59102 using GEO data and R-based bioinformatics methods, to systematically characterize differentially expressed genes and their GO/KEGG enrichment patterns, and to summarize the major molecular features of LSCC in this dataset from the perspectives of cell-cycle regulation, extracellular matrix remodeling and inflammatory microenvironment, providing clues for subsequent verifiable mechanistic studies. Methods: The LSCC expression profile dataset GSE59102 was downloaded from the GEO database as the main discovery cohort, and GSE10288 was used as an auxiliary validation cohort. R was employed to perform preprocessing, normalization, and differential expression analysis of the expression matrix. Differentially Expressed Genes (DEGs) were identified using adjusted P value < 0.05 and |log₂FC| > 1 as the threshold, and volcano plots and heatmaps were generated. Subsequently, the cluster Profiler package was used to perform Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. The representative differentially expressed genes and their potential biological processes were then discussed in the context of published studies. Results: In GSE59102, 29 tumor samples and 13 normal/safety-margin samples were included. A total of 3439 significant DEGs were identified, including 1526 upregulated genes and 1913 downregulated genes. MMP12, HOXC13, LINC01980, CDC6, CASC9, HOXC13-AS, FGD6 and MCM2 were among the representative upregulated genes, whereas PLP1, CRISP2, NUCB2, VSX1, CAB39L, CRISP3 and EHF were among the downregulated genes. The heatmap of the top 50 differentially expressed genes effectively distinguished the tumor group from the normal group. Go enrichment results primarily involved terms such as nuclear division, chromosome segregation, extracellular matrix, basement membrane, integrin binding and cytokine activity; KEGG enrichment primarily involved pathways such as Cell cycle, DNA replication, Cytokine-cytokine receptor interaction and Complement and coagulation cascades. Because of the limited gene coverage of GSE10288, the number of genes that were commonly and unidirectionally altered across the two cohorts was limited, no further joint enrichment analysis was conducted due to insufficient stability. Conclusion: The analysis based on GSE59102 suggests that LSCC tissues are characterized by abnormal activation of cell-cycle programs, extracellular matrix remodeling and inflammatory/cytokine signaling. MMP12, HOXC13, CASC9, CDC6 and MCM2 may be considered candidate molecules for further experimental validation. Because external validation was limited by platform gene coverage, the present findings should be interpreted as molecular features observed in a specific public dataset, and their stability and generalizability require confirmation in additional independent cohorts and experimental models.