Identification of a ferroptosis-related gene signature for predicting the prognosis of cholangiocarcinoma

Cholangiocarcinoma (CCA) is a highly intractable malignancy with poor prognosis. Ferroptosis, a newly explored type of programmed cell death, plays a critical role in the initiation and progression of a tumor. Herein, we aimed to identify a ferroptosis-related risk model to evaluate the prognosis of CCA. Differentially expressed genes (DEGs) were retrieved from three GEO cohorts. Univariate and LASSO analysis were employed to build a ferroptosis-related gene signature. Next, the predictive value was assessed in a training and a validation cohort. Metascape Online analysis, ESTIMATE and CIBERSORT algorithms, and ssGSEA were employed to perform the functional analysis between different risk groups. Finally, the expression of prognostic genes was validated with RT-qPCR. We identified 51 differentially expressed ferroptosis genes and established the prognostic signature containing five ferroptosis-related genes. The K-M curves and the ROC curves revealed a favorable predictive efficacy of the prognostic signature. Functional enrichment analysis indicated that immune-related responses were greatly enriched between different risk groups. Five prognostic genes were also differentially expressed in CCA cell lines. We developed a novel ferroptosis-related gene signature for CCA with high predictive accuracy. The analysis of the immune infiltration status may provide a potential therapeutic alternative to CCA.

胆管癌（Cholangiocarcinoma, CCA）是一种预后不良的高度难治性恶性肿瘤。铁死亡（Ferroptosis）作为一种新近被探索的程序性细胞死亡类型，在肿瘤的发生与进展中发挥关键作用。本研究旨在构建一种铁死亡相关风险模型，以评估胆管癌患者的预后情况。我们从3个GEO队列中检索得到差异表达基因（differentially expressed genes, DEGs），采用单因素分析及最小绝对收缩和选择算子（least absolute shrinkage and selection operator, LASSO）分析构建铁死亡相关基因特征。随后，在训练队列与验证队列中评估该模型的预测价值。本研究采用Metascape在线分析工具、ESTIMATE算法、CIBERSORT算法以及单样本基因集富集分析（single-sample gene set enrichment analysis, ssGSEA）对不同风险组间进行功能分析。最后，通过实时定量聚合酶链反应（reverse transcription quantitative polymerase chain reaction, RT-qPCR）验证预后基因的表达水平。本研究共筛选得到51个差异表达铁死亡相关基因，并构建了包含5个铁死亡相关基因的预后特征模型。Kaplan-Meier曲线（K-M曲线）与受试者工作特征曲线（receiver operating characteristic curve, ROC）分析结果显示，该预后特征模型具有良好的预测效能。功能富集分析结果表明，不同风险组间免疫相关反应显著富集。5个预后基因在胆管癌细胞系中同样存在差异表达。本研究构建了一种全新的胆管癌铁死亡相关基因特征模型，其预测准确性较高。免疫浸润状态分析可为胆管癌的潜在治疗方案提供新思路。