Qian Shirui, Li Ping, Hou Xiaotong, Dong Nianguo
Department of Cardiovascular Surgery, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China.
Centre for Cardiac Intensive Care, Beijing Anzhen Hospital, Beijing Institute of Heart, Lung and Blood Vessel Diseases, Capital Medical University, Beijing, China.
Sci Rep. 2025 Jul 1;15(1):20573. doi: 10.1038/s41598-025-06607-4.
This study was aimed at developing and internally validating nomograms for predicting mortality during venoarterial-extracorporeal membrane oxygenation (VA-ECMO) and in-hospital mortality risk in patients treated with VA-ECMO. A total of 7260 patients treated with VA-ECMO from January 2017 to December 2023 were extracted from the Chinese society of extra corporeal life support registry database. The entire cohort was randomly assigned to derivation and validation cohorts at a ratio of 2:1. Multivariable Cox proportional hazards regression was conducted using bootstrapping with the likelihood ratio test and Akaike information criterion. Approximately 24% of patients died during VA-ECMO assistance, and 51% died in the hospital. The nomogram PROFIT was constructed with ten pre- and immediately post-ECMO parameters: age, body mass index (BMI), intra-aortic balloon pump before VA-ECMO, history of cardiac arrest, worst mean arterial pressure (MAP), potential of hydrogen (pH) and serum lactate levels before VA-ECMO, site of ECMO installation, peripheral cannulation and distal perfusion. Additionally, nomogram POSITIVE was also established with ten parameters: age, sex, BMI, history of cardiac arrest, MAP, pH, and serum lactate levels before VA-ECMO support, the occurrence of cardiac arrest before VA-ECMO, type of sedation and prior continuous renal replacement therapy. The area under the receiver operating characteristics (AUC) of the nomogram PROFIT (0.72 [95% CI 0.70-0.74]) and POSITIVE (0.71 [95% CI 0.68-0.73]) outperformed the SAVE score, which indicated that the nomograms were capable of effectively identifying patients with a high risk of mortality. Both nomograms demonstrated outstanding discrimination and calibration in derivation and validation cohorts. In patients treated with VA-ECMO, the nomogram PROFIT may serve as a valuable tool for predicting mortality during VA-ECMO assistance, and the nomogram POSITIVE can predict in-hospital mortality with high reliability. However, these tools still require external validation in other patient populations requiring VA-ECMO support.
本研究旨在开发并内部验证用于预测静脉-动脉体外膜肺氧合（VA-ECMO）期间死亡率以及接受VA-ECMO治疗患者的院内死亡风险的列线图。从中国体外生命支持注册数据库中提取了2017年1月至2023年12月期间接受VA-ECMO治疗的7260例患者。整个队列以2:1的比例随机分为推导队列和验证队列。使用似然比检验和赤池信息准则的自抽样法进行多变量Cox比例风险回归。约24%的患者在VA-ECMO支持期间死亡，51%的患者在院内死亡。列线图PROFIT由10个ECMO前和ECMO后即刻的参数构建而成：年龄、体重指数（BMI）、VA-ECMO前的主动脉内球囊泵、心脏骤停病史、最差平均动脉压（MAP）、VA-ECMO前的酸碱度（pH）和血清乳酸水平、ECMO安装部位、外周插管和远端灌注。此外，列线图POSITIVE也由10个参数建立：年龄、性别、BMI、心脏骤停病史、MAP、pH、VA-ECMO支持前的血清乳酸水平、VA-ECMO前心脏骤停的发生情况、镇静类型和既往持续肾脏替代治疗。列线图PROFIT（0.72 [95%可信区间0.70 - 0.74]）和POSITIVE（0.71 [95%可信区间0.68 - 0.73]）的受试者操作特征曲线下面积（AUC）优于SAVE评分，这表明列线图能够有效识别高死亡风险患者。两个列线图在推导队列和验证队列中均表现出出色的区分度和校准度。在接受VA-ECMO治疗的患者中，列线图PROFIT可作为预测VA-ECMO支持期间死亡率的有价值工具，列线图POSITIVE能够高度可靠地预测院内死亡率。然而，这些工具仍需要在其他需要VA-ECMO支持的患者群体中进行外部验证。
