Machine Learning Screening and Validation of PANoptosis-Related Gene Signatures in Sepsis

Abstract

Jingjing Xu,1 Mingyu Zhu,1 Pengxiang Luo,2 Yuanqi Gong1 1Department of Intensive Care Unit, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, Nanchang, 330006, People’s Republic of China; 2Tongji Medical College, Huazhong University of Science and Technology, Wuhan, 430030, People’s Republic of ChinaCorrespondence: Yuanqi Gong, Department of Intensive Care Unit, The Second Affiliated Hospital, Jiangxi Medical College, Nanchang University, No. 1 Minde Road, Donghu District, Nanchang, 330006, People’s Republic of China, Email [email protected]: Sepsis is a syndrome marked by life-threatening organ dysfunction and a disrupted host immune response to infection. PANoptosis is a recent conceptual development, which emphasises the interconnectedness among multiple programmed cell deaths in various diseases. Nevertheless, the role of PANoptosis in sepsis is still unclear.Methods: We utilized the GSE65682 dataset to identify PANoptosis-related genes (PRGs) and associated immune characteristics in sepsis, classified sepsis samples based on PRGs using the ConsensusClusterPlus method and applied the Weighted Gene Co-Expression Network Analysis (WGCNA) algorithm to identify cluster-specific hub genes. Based on PANoptosis -specific DEGs, we compared results from machine learning models and the best-performing model was selected. Predictive efficiency was validated through external dataset, nomogram, survival analysis, quantitative real-time PCR, and western blot.Results: The expression levels of PRGs were generally dysregulated in sepsis patients compared with normal samples, and higher PRGs expression correlated with increased immune cell infiltration. In addition, two distinct PANoptosis-related clusters were defined, and functional analysis indicated that DEGs associated with these clusters were primarily linked to immune-related pathways. The SVM model was selected as best-performing model, with lower residuals and the highest area under the curve (AUC = 0.967), which was then validated in an external dataset (AUC = 0.989) and through in vivo experiments. Additional validation through nomogram and survival analysis further confirmed its substantial predictive efficacy.Conclusion: Our findings exposed the intricate association between PANoptosis and sepsis, offering important insights on sepsis diagnosis and potential therapeutic targets.Keywords: sepsis, PANoptosis, immune infiltration, machine learning, prediction model

Keywords

• sepsis
• panoptosis
• immune infiltration
• machine learning
• prediction model