The significance of long chain non-coding RNA signature genes in the diagnosis and management of sepsis patients, and the development of a prediction model
Yong Bai,Jing Gao,Yuwen Yan,Xu Zhao
DOI: https://doi.org/10.3389/fimmu.2024.1450014
IF: 7.3
2024-12-13
Frontiers in Immunology
Abstract:Background: Sepsis is a life-threatening organ dysfunction condition produced by dysregulation of the host response to infection. It is now characterized by a high clinical morbidity and mortality rate, endangering patients' lives and health. The purpose of this study was to determine the value of Long chain non-coding RNA (LncRNA) RP3_508I15.21, RP11_295G20.2, and LDLRAD4_AS1 in the diagnosis of adult sepsis patients and to develop a Nomogram prediction model. Methods: We screened adult sepsis microarray datasets GSE57065 and GSE95233 from the GEO database and performed differentially expressed genes (DEGs), weighted gene co-expression network analysis (WGCNA), and machine learning methods to find the genes by random forest (Random Forest), least absolute shrinkage and selection operator (LASSO), and support vector machine (SVM), respectively, with GSE95233 as the training set and GSE57065 as the validation set. Differentially expressed genes (DEGs), weighted gene co-expression network analysis (WGCNA), boxplot statistical analysis, and ROC analysis by Random Forest, Least Absolute Shrinkage and Selection Operator (LASSO), and Support Vector Machine (SVM) machine learning methods were used to identify characteristic genes and build the Nomogram Prediction model. Results: GSE95233 yielded a total of 1069 genes, 102 of which were sepsis-related and 22 of which were non-sepsis controls. GSE57065 yielded a total of 899 genes, with 467 up-regulated and 432 down-regulated, including 82 sepsis-related genes and 25 non-sepsis control genes. WGCNA analysis excluded outlier samples, leaving 2,029 genes for relationship analysis between sepsis- and non-sepsis patient-associated LncRNA network representation modules, as well as Wein plots of differential genes versus genes in key modules of weighted co-expression network analysis to analyze gene intersections. Machine Learning found the sepsis-related characteristic LncRNAs RP3-508I15.21, RP11-295G20.2, LDLRAD4-AS1, and CTD-2542L18.1. The datasets GSE95233 and GSE57065 were analyzed using Boxplot against the screened genes listed above, respectively. The p-value between the sepsis and non-sepsis groups was less than 0.05, indicating that anomalies were statistically significant. CTD-2542L18.1 in dataset GSE57065 had an AUC value of 0.638, which was less than 0.7 and did not indicate diagnostic significance, but RP3-508I15.21, RP11-295G20.2, and LDLRAD4-AS1 had AUC values more than 0.7 after ROC analysis. All four sepsis-associated LncRNA ROC analyses in dataset GSE95233 exhibited AUC values more than 0.7, indicating diagnostic significance. Conclusion: LncRNAs RP3_508I15.21, RP11_295G20.2, and LDLRAD4_AS1 have some utility in the diagnosis and treatment of adult sepsis patients, as well as some reference importance in guiding the diagnosis and treatment of clinical sepsis.
immunology