Fu Ziqiao, Zeng Jia, Xiong Xiaomin, Zhong Weimin
Respiratory and Critical Care Medicine, Guangyuan Central Hospital, Guangyuan, 628000, Sichuan, China.
The Fifth Hospital of Xiamen, Xiamen, 361101, Fujian, P.R. China.
Discov Oncol. 2025 Aug 24;16(1):1605. doi: 10.1007/s12672-025-03456-5.
PANoptosis is a new form of inflammatory programmed cell death, that emphasizes the interaction between pyroptosis, apoptosis and necroptosis. This study aimed to investigate clinical implications of PANoptosis in lung adenocarcinoma.
The ConsensusClusterPlus software was firstly utilized to identify molecular subtypes in lung adenocarcinoma (LUAD) based upon expression of PANoptosis-related regulators. Then, subtype-associated modules were further screened by using the weight gene correlation network analysis (WGCNA). A PANoptosis-related signature (PRS) was developed using a 10-fold cross-validation framework and 101 combinations of 10 machine-learning algorithms based on module genes. The predictive value of PRS was thoroughly assessed in relation to prognosis and immunotherapy. In addition, chemotherapy drugs sensitivity and candidate drug targets were further screened through cell line analysis.
Two molecular subtypes related to PANoptosis were distinguished by analyzing the expression of PANoptosis regulators. A total of 789 genes associated with subtype were identified in the yellow module using the WGCNA algorithm. Based on subtype-related genes, the optimal PRS was constructed using the ridge algorithm out of 101 algorithm combinations, and displayed a robust and reliable performance in predicting the survival of LUAD patients across multiple cohorts. Multivariate cox regression analysis result demonstrated that PRS can be serve as an independent prognostic factor. A nomogram, constructed by PRS and independent clinical factors demonstrated outstanding predictive accuracy for overall survival (OS) of LUAD patients. Additionally, the patients in low PRS group exhibited a favorable survival outcome, increased immune cell infiltration, and lower tumor immune dysfunction and exclusion (TIDE) score. Conversely, the high PANoptosis group showed a correlation with higher rates of somatic single nucleotide polymorphisms (SNP) mutation and copy number variation (CNV). The individuals with a high PANoptosis score displayed higher sensitivity to docetaxel and gemcitabine. Ultimately, seven drugs (SB-743921, GSK461364, BI-2536, deferasirox, VLX600, VE-822, epothilone-b) and a therapeutic target (TRPA1) were predicted to the high PANoptosis group patients.
The present study developed a PRS using 101 machine learning combination algorithms, which could aid in risk stratification and prognosis for LUAD patients. The candidate drugs and target may provide new insights in the treatment of high PRS group patients.
PAN细胞焦亡是一种新的炎症程序性细胞死亡形式，强调细胞焦亡、凋亡和坏死性凋亡之间的相互作用。本研究旨在探讨PAN细胞焦亡在肺腺癌中的临床意义。
首先利用ConsensusClusterPlus软件，根据PAN细胞焦亡相关调节因子的表达情况，识别肺腺癌（LUAD）的分子亚型。然后，通过加权基因共表达网络分析（WGCNA）进一步筛选亚型相关模块。基于模块基因，使用10倍交叉验证框架和10种机器学习算法的101种组合，开发了一种PAN细胞焦亡相关特征（PRS）。全面评估了PRS在预后和免疫治疗方面的预测价值。此外，通过细胞系分析进一步筛选化疗药物敏感性和候选药物靶点。
通过分析PAN细胞焦亡调节因子的表达，区分出两种与PAN细胞焦亡相关的分子亚型。使用WGCNA算法在黄色模块中鉴定出789个与亚型相关的基因。基于亚型相关基因，从101种算法组合中使用岭算法构建了最佳PRS，并在预测多个队列中LUAD患者的生存情况时表现出稳健可靠的性能。多变量cox回归分析结果表明，PRS可作为独立的预后因素。由PRS和独立临床因素构建的列线图对LUAD患者的总生存期（OS）显示出出色的预测准确性。此外，低PRS组患者表现出良好的生存结果、免疫细胞浸润增加以及较低的肿瘤免疫功能障碍和排除（TIDE）评分。相反，高PAN细胞焦亡组与较高的体细胞单核苷酸多态性（SNP）突变率和拷贝数变异（CNV）相关。PAN细胞焦亡评分高的个体对多西他赛和吉西他滨表现出更高的敏感性。最终，预测高PAN细胞焦亡组患者有7种药物（SB-743921、GSK461364、BI-2536、地拉罗司、VLX600、VE-822、埃坡霉素-b）和一个治疗靶点（TRPA1）。
本研究使用101种机器学习组合算法开发了一种PRS，可帮助对LUAD患者进行风险分层和预后评估。候选药物和靶点可能为高PRS组患者的治疗提供新的见解。
