Department of Clinical Laboratory Sciences and Medical Biotechnology, College of Medicine, National Taiwan University, Taipei, Taiwan.
Institute of Chemistry, Academia Sinica, Taipei, Taiwan.
Clin Epigenetics. 2024 Oct 3;16(1):137. doi: 10.1186/s13148-024-01751-6.
As the global use of extracorporeal membrane oxygenation (ECMO) treatment increases, survival rates have not correspondingly improved, emphasizing the need for refined patient selection to optimize resource allocation. Currently, prognostic markers at the molecular level are limited.
Thirty-four cardiogenic shock (CS) patients were prospectively enrolled, and peripheral blood mononuclear cells (PBMCs) were collected at the initiation of ECMO (t0), two-hour post-installation (t2), and upon removal of ECMO (tr). The PBMCs were analyzed by comprehensive epigenomic assays. Using the Wilcoxon signed-rank test and least absolute shrinkage and selection operator (LASSO) regression, 485,577 DNA methylation features were analyzed and selected from the t0 and tr datasets. A random forest classifier was developed using the t0 dataset and evaluated on the t2 dataset. Two models based on DNA methylation features were constructed and assessed using receiver operating characteristic (ROC) curves and Kaplan-Meier survival analyses.
The ten-feature and four-feature models for predicting in-hospital mortality attained area under the curve (AUC) values of 0.78 and 0.72, respectively, with LASSO alpha values of 0.2 and 0.25. In contrast, clinical evaluation systems, including ICU scoring systems and the survival after venoarterial ECMO (SAVE) score, did not achieve statistical significance. Moreover, our models showed significant associations with in-hospital survival (p < 0.05, log-rank test).
This study identifies DNA methylation features in PBMCs as potent prognostic markers for ECMO-treated CS patients. Demonstrating significant predictive accuracy for in-hospital mortality, these markers offer a substantial advancement in patient stratification and might improve treatment outcomes.
随着全球体外膜肺氧合（ECMO）治疗的应用增加，生存率并未相应提高，这强调了需要精细化患者选择以优化资源分配。目前，分子水平的预后标志物有限。
前瞻性纳入 34 例心源性休克（CS）患者，在 ECMO 启动时（t0）、安装后两小时（t2）和 ECMO 移除时（tr）采集外周血单核细胞（PBMC）。通过综合表观基因组分析对 PBMC 进行分析。使用 Wilcoxon 符号秩检验和最小绝对收缩和选择算子（LASSO）回归，从 t0 和 tr 数据集分析和选择 485,577 个 DNA 甲基化特征。使用 t0 数据集开发随机森林分类器，并在 t2 数据集上进行评估。基于 DNA 甲基化特征构建并评估了两个模型，通过接收者操作特征（ROC）曲线和 Kaplan-Meier 生存分析进行评估。
用于预测住院死亡率的十特征和四特征模型的 AUC 值分别为 0.78 和 0.72，LASSO alpha 值分别为 0.2 和 0.25。相比之下，包括 ICU 评分系统和静脉动脉 ECMO 后生存（SAVE）评分在内的临床评估系统没有达到统计学意义。此外，我们的模型与住院生存率显著相关（p<0.05，对数秩检验）。
本研究确定了 PBMC 中的 DNA 甲基化特征作为 ECMO 治疗 CS 患者的潜在预后标志物。这些标志物对住院死亡率具有显著的预测准确性，为患者分层提供了重大进展，并可能改善治疗结果。
