体外心肺复苏患者急性肾损伤列线图预测模型的建立与验证:一项跨越连续九年的多中心回顾性队列研究
该多中心回顾性队列研究纳入108例ECPR患者建立AKI预测模型,AKI发生率为72.2%。多因素Logistic回归显示SOFA评分升高(OR=1.288)、SCr升高(OR=1.010)、血乳酸升高(OR=1.151)和低灌注时间延长(OR=1.059)为AKI独立危险因素。基于上述因素构建的列线图模型AUC为0.858(95%CI 0.782-0.934),敏感度71.8%、特异度83.3%,内部验证C-index为0.822。外部验证队列31例,AUC为0.833(95%CI 0.654-1.000),敏感度70.0%、特异度85.7%。
要点
- 建模队列108例ECPR患者中78例发生AKI,AKI发生率为72.2%。
- SOFA评分(OR=1.288,95%CI 1.055-1.571)、SCr(OR=1.010,95%CI 1.002-1.018)、血乳酸(OR=1.151,95%CI 1.036-1.279)和低灌注时间(OR=1.059,95%CI 1.007-1.114)为AKI独立危险因素。
- 列线图模型预测AKI的AUC为0.858(95%CI 0.782-0.934),敏感度71.8%、特异度83.3%,内部验证C-index为0.822。
- 外部验证队列31例,模型AUC为0.833(95%CI 0.654-1.000),敏感度70.0%、特异度85.7%。
- Hosmer-Lemeshow检验(χ2=6.402,P=0.602)和校准曲线显示模型拟合良好,DCA提示多数患者净获益较高。
适用人群与场景
- 建模队列:2016年1月至2024年8月入住嘉兴市第一医院、在心肺复苏期间接受ECPR的患者。
- 外部验证队列:2024年5月至2025年6月入住桐乡市第一人民医院和平湖市第一人民医院、在心肺复苏期间接受ECPR的患者。
诊断标准与关键阈值
- 根据ECMO支持期间是否发生AKI分为AKI组和非AKI组。
研究设计与关键结果
- 多中心回顾性队列研究,建模队列108例ECPR患者,其中78例发生AKI、30例未发生,AKI发生率72.2%。
- 与非AKI组相比,AKI组SOFA评分、SCr、血乳酸更高,PCT更低,低灌注时间更长(均P<0.05)。
- 多因素Logistic回归显示SOFA评分(OR=1.288,95%CI 1.055-1.571,P=0.013)、SCr(OR=1.010,95%CI 1.002-1.018,P=0.015)、血乳酸(OR=1.151,95%CI 1.036-1.279,P=0.009)和低灌注时间(OR=1.059,95%CI 1.007-1.114,P=0.026)为AKI独立危险因素。
- 列线图模型预测AKI的AUC为0.858(95%CI 0.782-0.934,P<0.001),敏感度71.8%、特异度83.3%;1 000次Bootstrap重采样后C-index为0.822。
- Hosmer-Lemeshow检验(χ2=6.402,P=0.602)和校准曲线显示预测概率与理想概率拟合良好;DCA提示使用列线图模型对多数患者可获得更高净获益。
- 外部验证队列31例,与建模队列在性别、年龄、基础疾病及四个核心独立危险因素等基线数据上无显著差异;模型预测AKI的AUC为0.833(95%CI 0.654-1.000,P<0.001),敏感度70.0%、特异度85.7%。
证据类型、等级与局限
- 证据类型:多中心回顾性队列研究(建模队列+外部验证队列)。
- 局限:原文未报告局限性。
对床旁实践的改变
- 作者认为,基于低灌注时间联合SOFA评分、SCr和血乳酸建立的列线图模型经内部和外部验证均具有良好的预测价值,可用于预测ECPR患者AKI。
提炼说明
- 本条依据论文摘要整理;摘要未涉及「处置流程与药物剂量」「红旗征与不能漏的情况」「指南或共识出处」,故未列出这些栏目;全文请点「阅读原文」。
英文原文(摘要)· 点击展开
OBJECTIVE: To establish a predictive model for acute kidney injury (AKI) in patients undergoing extracorporeal cardiopulmonary resuscitation (ECPR), and to evaluate and validate its predictive value.
METHODS: A multicenter retrospective cohort study was conducted. 1) The clinical data of the patients undergoing ECPR during cardiopulmonary resuscitation (CPR) admitted to the First Hospital of Jiaxing between January 2016 and August 2024 were collected as the modeling cohort. The clinical data included patient characteristic information, relevant treatment information during ECPR, post-extracorporeal membrane oxygenation (ECMO) operation parameters, and ECMO-related variable information. The patients were divided into an AKI group and a non-AKI group according to the occurrence of AKI during ECMO support. The differences in clinical characteristics between the two groups were compared. Multivariate Logistic regression analysis was used to screen independent risk factors for AKI during ECMO therapy, and a nomogram model was established. The predictive value of the model was evaluated by receiver operator characteristic curve (ROC curve). Internal validation of the model was performed using the Bootstrap method with 1 000 resamplings. The predictive performance of the nomogram model was verified using calibration curves and the Hosmer-Lemeshow test, and the clinical utility of the model was assessed by decision curve analysis (DCA). 2) The clinical data of the patients who received ECPR during CPR admitted to the First People's Hospital of Tongxiang and the First People's Hospital of Pinghu from May 2024 to June 2025 were selected for external validation. The predictive efficacy of the model was evaluated by ROC curve analysis.
RESULTS: 1) A total of 108 ECPR patients were finally enrolled in the modeling cohort, among whom 78 developed AKI during ECMO treatment and 30 did not, with an AKI incidence of 72.2%. Compared with the non-AKI group, the patients in the AKI group had higher Sequential Organ Failure Assessment (SOFA) score, serum creatinine (SCr), blood lactic acid, lower procalcitonin (PCT), and longer hypoperfusion time (all P<0.05). There were no significant differences in other clinical data between the two groups. Multivariate Logistic regression analysis showed that increased SOFA score [odds ratio (OR)=1.288, 95% confidence interval (95%CI) was 1.055-1.571, P=0.013], SCr (OR=1.010, 95%CI was 1.002-1.018, P=0.015) and blood lactic acid (OR=1.151, 95%CI was 1.036-1.279, P=0.009), and prolonged hypoperfusion time (OR=1.059, 95%CI was 1.007-1.114, P=0.026) were independent risk factors for AKI during ECMO in ECPR patients. A nomogram prediction model was constructed based on the above independent risk factors. ROC curve analysis showed that the area under the ROC curve (AUC) of the nomogram model for predicting AKI in ECPR patients was 0.858 (95%CI was 0.782-0.934, P<0.001), with a sensitivity of 71.8% and a specificity of 83.3%. After 1 000 Bootstrap resamplings, the C-index was 0.822. The Hosmer-Lemeshow test and calibration curve showed good fitness between the predicted and ideal probabilities (χ2=6.402, P=0.602), indicating favorable model performance. DCA results suggested that using the nomogram model achieved higher net benefit for most patients. 2) A total of 31 patients who received ECPR were enrolled for external validation. There were no significant differences in baseline data such as gender, age, or underlying diseases as well as four core independent risk factors for constructing a nomogram predictive model between the external validation cohort and the primary cohort, meeting the requirements of external validation design. ROC curve analysis showed that the AUC of the nomogram model for predicting AKI in ECPR patients was 0.833 (95%CI was 0.654-1.000, P<0.001), with a sensitivity of 70.0% and a specificity of 85.7%.
CONCLUSIONS: A nomogram model for predicting AKI in ECPR patients is established based on hypoperfusion time combined with SOFA score, SCr, and blood lactic acid. The model has been confirmed to possess good predictive value through both internal and external validation.
参考来源
- Shen Y, Jin T, Li C, et al. [Establishment and validation of a nomogram model for predicting acute kidney injury in patients undergoing extracorporeal cardiopulmonary resuscitation: a multicenter retrospective cohort study spanning nine consecutive years]. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2026. PMID: 42200244. doi:10.3760/cma.j.cn121430-20250801-00418
供执业医师抢救与诊疗参考,不替代临床判断;剂量请结合体重、肝肾功能和本院药品规格复核,以指南原文为准。