Institute of Biomedical Informatics, National Yang Ming Chiao Tung University, Taipei, Taiwan.
Division of General Surgery, Department of Surgery, Kaohsiung Chang Gung Memorial Hospital, Kaohsiung, Taiwan.
J Transl Med. 2023 Jul 20;21(1):485. doi: 10.1186/s12967-023-04355-5.
The nuclear factor kappa B (NFκB) regulatory pathways downstream of tumor necrosis factor (TNF) play a critical role in carcinogenesis. However, the widespread influence of NFκB in cells can result in off-target effects, making it a challenging therapeutic target. Ensemble learning is a machine learning technique where multiple models are combined to improve the performance and robustness of the prediction. Accordingly, an ensemble learning model could uncover more precise targets within the NFκB/TNF signaling pathway for cancer therapy.
In this study, we trained an ensemble learning model on the transcriptome profiles from 16 cancer types in the TCGA database to identify a robust set of genes that are consistently associated with the NFκB/TNF pathway in cancer. Our model uses cancer patients as features to predict the genes involved in the NFκB/TNF signaling pathway and can be adapted to predict the genes for different cancer types by switching the cancer type of patients. We also performed functional analysis, survival analysis, and a case study of triple-negative breast cancer to demonstrate our model's potential in translational cancer medicine.
Our model accurately identified genes regulated by NFκB in response to TNF in cancer patients. The downstream analysis showed that the identified genes are typically involved in the canonical NFκB-regulated pathways, particularly in adaptive immunity, anti-apoptosis, and cellular response to cytokine stimuli. These genes were found to have oncogenic properties and detrimental effects on patient survival. Our model also could distinguish patients with a specific cancer subtype, triple-negative breast cancer (TNBC), which is known to be influenced by NFκB-regulated pathways downstream of TNF. Furthermore, a functional module known as mononuclear cell differentiation was identified that accurately predicts TNBC patients and poor short-term survival in non-TNBC patients, providing a potential avenue for developing precision medicine for cancer subtypes.
In conclusion, our approach enables the discovery of genes in NFκB-regulated pathways in response to TNF and their relevance to carcinogenesis. We successfully categorized these genes into functional groups, providing valuable insights for discovering more precise and targeted cancer therapeutics.
肿瘤坏死因子（TNF）下游的核因子 kappa B（NFκB）调控途径在癌变过程中起着关键作用。然而，NFκB 在细胞中的广泛影响可能导致脱靶效应，使其成为一个具有挑战性的治疗靶点。集成学习是一种机器学习技术，其中多个模型被组合以提高预测的性能和稳健性。因此，集成学习模型可以在 NFκB/TNF 信号通路中发现更精确的癌症治疗靶点。
在这项研究中，我们在 TCGA 数据库中 16 种癌症类型的转录组谱上训练了一个集成学习模型，以鉴定一组与癌症中 NFκB/TNF 通路一致相关的稳健基因。我们的模型使用癌症患者作为特征来预测参与 NFκB/TNF 信号通路的基因，并可以通过切换患者的癌症类型来适应预测不同癌症类型的基因。我们还进行了功能分析、生存分析和三阴性乳腺癌的案例研究，以证明我们的模型在转化癌症医学中的潜力。
我们的模型准确地识别了 NFκB 响应 TNF 在癌症患者中调节的基因。下游分析表明，所鉴定的基因通常参与经典的 NFκB 调控途径，特别是适应性免疫、抗细胞凋亡和细胞对细胞因子刺激的反应。这些基因被发现具有致癌特性，并对患者的生存产生不利影响。我们的模型还可以区分特定的癌症亚型，即三阴性乳腺癌（TNBC），已知其受到 TNF 下游 NFκB 调控途径的影响。此外，还鉴定了一个称为单核细胞分化的功能模块，该模块可以准确预测 TNBC 患者和非 TNBC 患者的短期生存不良，为开发癌症亚型的精准医学提供了潜在途径。
总之，我们的方法能够发现 NFκB 调节的通路中响应 TNF 的基因及其与癌变的相关性。我们成功地将这些基因分类到功能组中，为发现更精确和靶向的癌症治疗方法提供了有价值的见解。
