Original code.

This study aims to explore the molecular subtypes of sepsis and the correlation between immune-related genes and the prognosis of patients with sepsis. Utilizing the Gene Expression Omnibus dataset (GSE65682) with 479 patients with sepsis as the training set and 164 patients treated at our hospital as the independent validation cohort. An unsupervised cluster analysis was used to identify potential molecular subtypes of sepsis, and a weighted gene co-expression network analysis was performed to identify gene modules. Gene Ontology, Kyoto Encyclopedia of Genes, and Genomes enrichment analyses were performed, and the immune status was also evaluated. Using LASSO regression and multivariate Cox regression, an immune-related gene prognostic model was developed, validated, and evaluated, followed by an individual risk scoring system. We identified two molecular subtypes of sepsis that are associated with distinct immune response patterns and clinical outcomes. Patients in Cluster A exhibited poorer survival and enrichment of pro-inflammatory pathways, while those in Cluster B had better outcomes and enrichment of immune regulatory pathways. A 10-gene prognostic model was constructed, stratifying patients into high- and low-risk groups using the estimated risk score that was confirmed to be an independent prognostic factor in both the training (hazard ratio [HR]: 1.126, 95% confidence interval [CI]: 1.096–1.156, P < 0.001) and validation datasets (HR: 1.149, 95% CI: 1.085–1.216, P < 0.001). A risk scoring system was developed based on the risk score and clinical parameters, with estimated mortality probabilities of 0.132 (7-day), 0.211 (14-day), and 0.258 (21-day). High-risk patients had significantly worse prognoses, and this was validated in the independent cohort. Distinct immune cell profiles were found between the two subtypes and risk groups, with B cells, CD8 + T cells, and NK cells elevated in Cluster B. This study identified immune-related molecular subtypes of sepsis and developed a prognostic model that accurately predicts sepsis mortality. These findings provide insights into the immune dysregulation in sepsis and can potentially be used for developing personalized treatment strategies and improving clinical decision-making in sepsis management.

本研究旨在探究脓毒症的分子亚型以及免疫相关基因与脓毒症患者预后的相关性。本研究以包含479例脓毒症患者的基因表达综合数据库（Gene Expression Omnibus，GEO）GSE65682数据集作为训练集，以本院收治的164例患者作为独立验证队列。采用无监督聚类分析识别脓毒症潜在分子亚型，并通过加权基因共表达网络分析筛选基因模块。随后开展基因本体（Gene Ontology，GO）、京都基因与基因组百科全书（Kyoto Encyclopedia of Genes and Genomes，KEGG）富集分析，并评估患者免疫状态。利用LASSO回归与多因素Cox回归构建免疫相关基因预后模型，完成模型验证与效能评估，并建立个体化风险评分系统。本研究筛选出两种脓毒症分子亚型，二者具有截然不同的免疫应答模式与临床结局：A簇患者生存率更低，且促炎通路富集显著；B簇患者预后更佳，免疫调控通路富集更为明显。本研究构建了包含10个基因的预后模型，通过估算风险评分将患者划分为高风险组与低风险组，该风险评分在训练集（风险比[HR]：1.126，95%置信区间[CI]：1.096~1.156，P<0.001）与验证集（HR：1.149，95%CI：1.085~1.216，P<0.001）中均被证实为独立预后因素。基于风险评分与临床参数建立了风险评分系统，其估算的7天、14天、21天死亡率分别为0.132、0.211与0.258。高风险组患者预后显著更差，该结论在独立队列中得到验证。两种亚型与风险分组间的免疫细胞浸润特征存在显著差异：B簇中B细胞、CD8+T细胞与自然杀伤（NK）细胞水平显著升高。本研究明确了脓毒症的免疫相关分子亚型，并构建了可精准预测脓毒症死亡率的预后模型。上述发现为脓毒症免疫失调机制提供了新见解，有望为开发脓毒症个性化治疗策略、优化临床诊疗决策提供理论依据。