Yang Dunpeng, Zhang Wentian, Wang Qibin
Department of Thoracic Surgery, Xuzhou Central Hospital, Xuzhou, Jiangsu, China.
Department of Thoracic Surgery, Shanghai Pulmonary Hospital, Shanghai, China.
Discov Oncol. 2025 Jul 28;16(1):1430. doi: 10.1007/s12672-025-03224-5.
Lung cancer pathogenesis involves complex interactions between immune system components and metabolic pathways. However, the causal relationships between these factors remain unclear. This study aimed to employ Mendelian randomization (MR) analysis to establish causal links between immunological markers, metabolic factors, and lung cancer development, while integrating multi-omics data for comprehensive molecular characterization.
We conducted a two-sample MR analysis of 731 immunological features and 1,400 metabolites using inverse-variance weighted methodology. The analysis was supplemented with single-cell expression profiling, functional enrichment analysis, and machine learning approaches. Multiple MR methodologies, including MR-Egger and heterogeneity testing via Cochran's Q statistic, were employed to validate findings. We specifically investigated the mediating role of sphingomyelin in the relationship between T cell %lymphocyte levels and lung cancer risk.
MR analysis identified 25 blood cell phenotypes significantly linked to lung cancer susceptibility, with memory antigen-presenting cells showing notable risk association (OR = 1.0763, 95%CI: 1.0147-1.1417, P = 0.0145). Seventy-six metabolites demonstrated causal influences on lung cancer pathogenesis, particularly within sphingolipid metabolism pathways. Single-cell profiling revealed significant differential expression of six genes (APCDD1L, CNTNAP4, GNG13, KIRREL2, LINC00628, and LIPK) between normal and tumor tissues. Machine learning models constructed from these findings demonstrated robust predictive performance in both TCGA and GEO datasets.
Through rigorous MR analysis, this study establishes causal relationships between specific immune markers, metabolic pathways, and lung cancer development.
肺癌发病机制涉及免疫系统成分与代谢途径之间的复杂相互作用。然而，这些因素之间的因果关系仍不清楚。本研究旨在采用孟德尔随机化（MR）分析来建立免疫标志物、代谢因素与肺癌发生之间的因果联系，同时整合多组学数据进行全面的分子特征分析。
我们使用逆方差加权方法对731个免疫特征和1400种代谢物进行了两样本MR分析。该分析辅以单细胞表达谱分析、功能富集分析和机器学习方法。采用多种MR方法，包括MR-Egger和通过 Cochr an Q统计量进行的异质性检验，以验证研究结果。我们特别研究了鞘磷脂在T细胞%淋巴细胞水平与肺癌风险关系中的中介作用。
MR分析确定了25种血细胞表型与肺癌易感性显著相关，记忆抗原呈递细胞显示出显著的风险关联（OR = 1.0763，95%CI：1.0147 - 1.1417，P = 0.0145）。76种代谢物对肺癌发病机制有因果影响，特别是在鞘脂代谢途径中。单细胞分析揭示了正常组织和肿瘤组织之间六个基因（APCDD1L、CNTNAP4、GNG13、KIRREL2、LINC00628和LIPK）的显著差异表达。基于这些发现构建的机器学习模型在TCGA和GEO数据集中均表现出强大的预测性能。
通过严格的MR分析，本研究建立了特定免疫标志物、代谢途径与肺癌发生之间的因果关系。
