Identifying novel aging-related diagnostic and prognostic models and aging-targeted drugs for sepsis patients

Abstract

Sepsis is defined as a dysfunctional, life-threatening response to infection leading to multiorgan dysfunction and failure. During the past decade, studies have highlighted the relationship between sepsis and aging. However, the role of aging-related mechanisms in the progression and prognosis of sepsis remains unclear. In the present study, we divided sepsis patients into High- and Low-aging groups based on the gene set variation analysis (GSVA) scores of GOBP-AGING gene set. Sepsis patients in the high-aging group exhibited higher levels of infiltration of innate immune cells, lower levels of infiltration of adaptive immune cells, and a worse prognosis than those in the Low-aging group. Additionally, the MPO to MME ratio (MPO/MME) appears to be an effective biomarker for predicting the prognosis of sepsis patients. Moreover, ARG1/SEC63 and ARG1/CDKN1C appear to be effective and robust biomarkers for the early diagnosis of sepsis patients. Finally, we found that thalidomide (TAL) significantly ameliorated LPS induced inflammation and organ injury and attenuated LPS induced cellular senescence in lung and kidney. Overall, this study provides new insights into the heterogeneity of sepsis, reveals the vital role of aging-related markers in the prognosis and diagnosis of sepsis and demonstrates that TAL is a novel aging-targeted drug for sepsis patients by attenuating LPS induced cellular senescence.

Keywords: Aging; Classification; Diagnosis; Prognosis; Sepsis; Thalidomide.

Fig. 1

Two sepsis groups were identified according to the GSVA scores of the GOBP_AGING gene set. (A,B) GSVA scores of the GOBP_AGING gene set were significantly higher in sepsis patients than in normal controls from GSE65682, GSE26440-26378, GSE185263, and GSE95233. (C) UMAP visualization of different cell types from human PBMCs. Comparison of GOBP_AGING scores between sepsis patents and healthy controls in CD4 + T cells (D), monocytes (E), and B cells (F). (G) Sepsis patients from GSE68682 were divided into High- and Low-aging groups according to the median score of the GOBP_AGING. (H) The outcome of sepsis patients in the two sepsis groups. (I) GSVA scores of aging-related gene sets were significantly higher in the High-aging group than in the Low-aging group from GSE65682. GSVA gene set variation analysis.

Fig. 2

Clinical features of the aging-related sepsis groups. (A) Kaplan‒Meier survival analysis of the overall survival of the two sepsis groups from GSE65682. (B,C) The outcome of sepsis patients in two groups from GSE16440-26378 and E-MTAB-4421. (D,E) Correlation of the two groups with sepsis endotypes published previously by Baghela and Scicluna.

Fig. 3

Transcriptome features of the aging-related sepsis groups. (A) The heatmap plot shows the GSVA score of 50 hallmark gene sets of the two sepsis groups and normal controls. The up-regulated (B) and down-regulated (C) processes or pathways in the High-aging group.

Fig. 4

Immune cell infiltration and immune function levels of the aging-related sepsis groups. The up-regulated (A) and down-regulated (B) immune cell infiltration or immune function levels in the High-aging group. (C) Correlation between immune cell infiltration or immune function levels and GSVA scores of GOBP_AGING.

Fig. 5

Construction of an aging-related sepsis DEG-based prognostic model. (A) Flow chart of the process of the identification of aging-related prognostic DEGs. (B) Results of LASSO Cox regression screening of four genes that are essential for the prognosis of sepsis. The upper panel shows the shrinkage profile of LASSO coefficients. The lower panel is the cross-validation plot of LASSO. (C) Gene expression changes and HR of all five genes from GSE65682, GSE185263, GSE95233, and GSE26440-26378. (D) The AUCs of different models in predicting the survival of sepsis patients within 28 days. (E) The time-dependent ROC for 7, 14, and 28 day overall survival predictions for the MPO/MME. (F) Kaplan‒Meier survival analysis of overall survival of the high and low groups divided by the value of MPO/MME (1.050) from GSE65682. (G) The outcome of sepsis patients in the high and low groups from MTAB-5273, E-MTAB-5274, and GSE95233. LASSO least absolute shrinkage and selection operation, FC fold change, HR hazard ratio, DEGs differentially expressed genes, AUC area under the curve, ROC receiver operating characteristic curve.

Fig. 6

Construction of an aging-related sepsis-DEG-based diagnostic model. (A) Flow chart of the process of the identification of aging-related DEGs. (B) The performance of the models in the diagnosis of sepsis patients in the GSE65682 cohort. (C) The optimal cutoff values of ARG1/SEC63 and ARG1/CDNK1C. (D) The diagnostic value of ARG1/SEC63 (1.036) and ARG1/CDNK1C (1.026) in the external cohorts GSE95233, GSE26440-26378, and GSE185263.

Fig. 7

TAL attenuated lung and kidney injury in LPS induced sepsis mice model (A) Flow chart of the process of candidate small-molecule drug identification. (B) Experimental design for establishing LPS-induced sepsis mice model and assessing the effect of TAL treatment. Serum levels of TNF-α(C) and IL-1β (D) from negative control (NC) and LPS induced sepsis mice with TAL or vehicle administration. ***P < 0.001 vs NC group; ^###P < 0.001 vs LPS group. (E) Representative lung and kidney sections stained with HE from NC and LPS induced sepsis mice with TAL or vehicle administration. Scale: 100 μm. Serum levels of creatinine (F) and BUN (G) from NC and LPS induced sepsis mice with TAL or vehicle administration. ***P < 0.001 vs NC group; ^###P < 0.001 vs LPS group. (H) Expression level of KIM-1 and NAGL in kidney tissues from NC and LPS induced sepsis mice with TAL or vehicle administration.

Fig. 8

TAL ameliorated cellular senescence of lung and kidney in LPS induced sepsis mice mode. Representative IHC staining of p16 and p21 f. from NC and LPS induced sepsis mice with TAL or vehicle administration. Scale: 40 μm.