Table_7_Construction and validation of a robust prognostic model based on immune features in sepsis.docx

PurposeSepsis, with life-threatening organ failure, is caused by the uncontrolled host response to infection. Immune response plays an important role in the pathophysiology of sepsis. Immune-related genes (IRGs) are promising novel biomarkers that have been used to construct the diagnostic and prognostic model. However, an IRG prognostic model used to predict the 28-day mortality in sepsis was still limited. Therefore, the study aimed to develop a prognostic model based on IRGs to identify patients with high risk and predict the 28-day mortality in sepsis. Then, we further explore the circulating immune cell and immunosuppression state in sepsis.
Materials and methodsThe differentially expressed genes (DEGs), differentially expressed immune-related genes (DEIRGs), and differentially expressed transcription factors (DETFs) were obtained from the GEO, ImmPort, and Cistrome databases. Then, the TFs-DEIRGs regulatory network and prognostic prediction model were constructed by Cox regression analysis and Pearson correlation analysis. The external datasets also validated the reliability of the prognostic model. Based on the prognostic DEIRGs, we developed a nomogram and conducted an independent prognosis analysis to explore the relationship between DEIRGs in the prognostic model and clinical features in sepsis. Besides, we further evaluate the circulating immune cells state in sepsis.
ResultsA total of seven datasets were included in our study. Among them, GSE65682 was identified as a discovery cohort. The results of GSEA showed that there is a significant correlation between sepsis and immune response. Then, based on a P value <0.01, 69 prognostic DEIRGs were obtained and the potential molecular mechanisms of DEIRGs were also clarified. According to multivariate Cox regression analysis, 22 DEIRGs were further identified to construct the prognostic model and identify patients with high risk. The Kaplan–Meier survival analysis showed that high-risk groups have higher 28-day mortality than low-risk groups (P=1.105e-13). The AUC value was 0.879 which symbolized that the prognostic model had a better accuracy to predict the 28-day mortality. The external datasets also prove that the prognostic model had an excellent prediction value. Furthermore, the results of correlation analysis showed that patients with Mars1 might have higher risk scores than Mars2-4 (P=0.002). According to the previous study, Mars1 endotype was characterized by immunoparalysis. Thus, the sepsis patients in high-risk groups might exist the immunosuppression. Between the high-risk and low-risk groups, circulating immune cells types were significantly different, and risk score was significantly negatively correlated with naive CD4+ T cells (P=0.019), activated NK cells (P=0.0045), monocytes (P=0.0134), and M1 macrophages (P=0.0002).
ConclusionsOur study provides a robust prognostic model based on 22 DEIRGs which can predict 28-day mortality and immunosuppression status in sepsis. The higher risk score was positively associated with 28-day mortality and the development of immunosuppression. IRGs are a promising biomarker that might facilitate personalized treatments for sepsis.

## 研究目的
脓毒症（Sepsis）是指宿主对感染产生失控性宿主反应后引发的危及生命的器官功能衰竭。免疫反应在脓毒症的病理生理过程中发挥重要作用。免疫相关基因（immune-related genes, IRGs）作为极具潜力的新型生物标志物，已被用于构建脓毒症的诊断及预后模型。然而，目前用于预测脓毒症28天死亡率的免疫相关基因预后模型仍较为有限。因此，本研究旨在构建基于免疫相关基因的预后模型，以识别高危脓毒症患者并预测其28天死亡率。此外，本研究还进一步探讨了脓毒症患者的循环免疫细胞状态及免疫抑制状态。

## 材料与方法
本研究从GEO、ImmPort及Cistrome数据库中获取差异表达基因（differentially expressed genes, DEGs）、差异表达免疫相关基因（differentially expressed immune-related genes, DEIRGs）以及差异表达转录因子（differentially expressed transcription factors, DETFs）。随后，通过Cox回归分析与Pearson相关分析构建转录因子-差异表达免疫相关基因调控网络及预后预测模型。利用外部数据集验证该预后模型的可靠性。基于预后相关差异表达免疫相关基因，本研究构建列线图（nomogram）并开展独立预后分析，以探究预后模型中差异表达免疫相关基因与脓毒症患者临床特征的关联。此外，本研究还进一步评估了脓毒症患者的循环免疫细胞状态。

## 研究结果
本研究共纳入7个数据集，其中GSE65682被选为发现队列。基因集富集分析（Gene Set Enrichment Analysis, GSEA）结果显示，脓毒症与免疫反应存在显著相关性。以P<0.01为筛选标准，本研究共获得69个预后相关差异表达免疫相关基因，并阐明了这些基因的潜在分子机制。通过多因素Cox回归分析，进一步筛选出22个差异表达免疫相关基因以构建预后模型并识别高危患者。Kaplan-Meier生存分析结果显示，高危组患者的28天死亡率显著高于低危组（P=1.105e-13）。该预后模型的曲线下面积（area under the curve, AUC）为0.879，表明其预测脓毒症28天死亡率的准确性较好。外部数据集验证结果同样证实该预后模型具有优异的预测效能。相关性分析结果显示，Mars1型患者的风险评分可能高于Mars2-4型患者（P=0.002）。既往研究表明，Mars1内表型以免疫麻痹为特征，因此高危组脓毒症患者可能存在免疫抑制状态。高危组与低危组患者的循环免疫细胞亚型存在显著差异，且风险评分与初始CD4+T细胞（naive CD4+ T cells）、活化自然杀伤（NK）细胞、单核细胞及M1型巨噬细胞均呈显著负相关（分别为P=0.019、P=0.0045、P=0.0134、P=0.0002）。

## 研究结论
本研究构建了一个基于22个差异表达免疫相关基因的稳健预后模型，该模型可预测脓毒症患者的28天死亡率及免疫抑制状态。较高的风险评分与28天死亡率及免疫抑制的发生呈正相关。免疫相关基因是极具潜力的生物标志物，有望推动脓毒症的个体化治疗。