目的·基于基因表达综合数据库（Gene Expression Omnibus，GEO）筛选儿童脓毒症预后相关长链非编码RNA（long non-coding RNA，lncRNA），并构建竞争性内源RNA（competing endogenous RNA，ceRNA）调控网络，探索其在儿童脓毒症预后评估中的潜在应用价值。方法·基于GEO转录组学数据集（GSE4607、GSE26440、GSE26378和GSE9692），对比儿童脓毒症休克患儿存活组与非存活组中lncRNA的表达差异，利用多元线性回归和LASSO分析筛选潜在特征lncRNA，采用受试者工作特征（receiver operating characteristic，ROC）曲线分析其预后评估能力。预测lncRNA、微小RNA（microRNA，miRNA）和mRNA的互作，构建蛋白互作（protein-protein interaction，PPI）网络，进行通路富集分析。结果·筛选出55个预后相关的差异表达lncRNA。LASSO回归分析明确6个潜在lncRNA，包括miR503宿主基因（miR503 host gene，MIR503HG）、TAPT1反义RNA1（TAPT1 antisense RNA 1，TAPT1-AS1）、膀胱癌细胞凋亡相关转录物（apoptosis associated transcript in bladder cancer，AATBC）、SBF2反义RNA1（SBF2 antisense RNA 1，SBF2-AS1）、MGC16275及宿主小核仁RNA基因15（small nucleolar RNA host gene 15，SNHG15），组成6-lncRNA特征（6-lncRNA signature，lncSig6）；lncSig6预测儿童脓毒症休克预后内部验证及外部验证的ROC曲线下面积（area under the curve，AUC）分别为0.859（95% CI 0.722~0.996）和0.854（95% CI 0.687~1.000）。进一步构建基于差异表达MIR503HG，SNHG15和SBF2-AS1的lncRNA-miRNA-mRNA网络，根据核心蛋白PPI网络和GO/KEGG信号通路分析结果提示：lncRNA经海绵化miRNA调控靶基因，涉及叉头盒蛋白O（forkhead box O，FoxO）信号通路、磷脂酰肌醇3激酶-蛋白激酶B（phosphatidylinositol 3 kinase-protein kinase B，PI3K-AKT）信号通路、细胞衰老信号通路、胰岛素信号通路、缺氧诱导蛋白-1（hypoxia-inducible protein-1，HIF-1）信号通路，和晚期糖化终产物（advanced glycation end products，AGEs）及其受体（receptor of AGEs，RAGE）信号通路。结论·lncSig6可以作为预测儿童脓毒症休克预后的评估方法，构建的ceRNA分子网络可为信号通路研究提供依据。

Objective ·To screen a long non-coding RNA (lncRNA) signature and construct a competing endogenous RNA (ceRNA) network associated with the prognosis of pediatric sepsis based on the Gene Expression Omnibus (GEO) database, and explore their potential application value in the prognosis assessment of children with sepsis. Methods ·Microarray data in GSE4607, GSE26440, GSE26378, and GSE9692 in the GEO database were used to compare the differences in lncRNA profiles between the survival and non-survival groups of children with septic shock. Then, multivariate linear regression, LASSO analysis, and receiver operating characteristic (ROC) curves were used to evaluate the capacity of the lncRNA signature for predicting the outcome of pediatric sepsis. The potential targeted microRNAs (miRNAs) and their downstream mRNAs, targeted by the screened lncRNAs, were used to construct a protein-protein interaction (PPI) network and perform pathway enrichment analysis. Results ·Transcriptomic data from GSE4607, GSE26440, GSE26378 and GSE9692 revealed 55 differentially expressed lncRNAs associated with prognosis, and miR503 host gene (MIR503HG), TAPT1 antisense RNA 1 (TAPT1-AS1), apoptosis-associated transcript in bladder cancer (AATBC), SBF2 antisense RNA 1 (SBF2-AS1), MGC16275, and small nucleolar RNA host gene 15 (SNHG15) were identified as a 6-lncRNA signature (lncSig6) associated with the prognosis of pediatric septic shock by LASSO regression analysis. The area under the ROC curve (AUC) of lncSig6 was 0.859 (95% CI 0.722‒0.996) and 0.854 (95% CI 0.687‒1.000) in internal and external validation, respectively. As lncRNA act as miRNA sponge, a lncRNA-miRNA-mRNA network based on 3 lncRNAs (MIR503HG, SNHG15, and SBF2-AS1) was constructed and involved in the regulation of signaling pathways, including forkhead box O (FoxO) signaling pathway, phosphatidylinositol 3 kinase-protein kinase B (PI3K-AKT) signaling pathway, cell senescence, insulin signaling pathway, hypoxia-inducible protein-1 (HIF-1) signaling pathway and advanced glycation end products (AGEs) and receptor of AGEs (RAGE) signaling pathway. Conclusion ·The lncSig6 can be used as an evaluation method to predict the prognosis of septic shock in children, and the constructed ceRNA molecular networks can provide an experimental basis for the study of signaling pathways.