Fan Zeyuan, Li Yang, Ji Hanhua, Jian Xinwen
Department of Cardiovascular Diseases, Civil Aviation General Hospital, Civil Aviation Clinical Medical College of Peking University, Beijing, China,
Department of Cardiovascular Diseases, Civil Aviation General Hospital, Civil Aviation Clinical Medical College of Peking University, Beijing, China.
Kidney Blood Press Res. 2018;43(6):1832-1841. doi: 10.1159/000495815. Epub 2018 Dec 7.
BACKGROUND/AIMS: Cardiorenal syndrome type 1(CRS1) is a serious clinical condition in patients with acute heart failure (AHF) associated with adverse clinical outcomes. Although several biomarkers for identifying CRS1 have been reported, early and accurate predicting CRS1 still remains a challenge. This study was aimed to develop and validate an individualized predictive nomogram for the risk of CRS1 in patients with AHF.
A total of 1235 AHF patients between 2013 and 2018 were included in this study. The patients were randomly classified into training set (n=823) and validation set (n=412). All data of the training set were used to screen the predictors of CRS1 via univariate and multivariate analyses. A nomogram was developed based on these predictors and validated by internal and external validation. The nomogram validation comprised discriminative ability determined by the area under the curve (AUC) of receiver-operating characteristic (ROC) curve and the predictive accuracy by calibration plots.
The overall incidence of CRS1 was 31.7%. Multivariate logistic regression revealed that age, diabetes, NYHA class, eGFR, hs-CRP and uAGT were independently associated with CRS1. A nomogram developed based on the six variables was with the AUC 0.885 and 0.823 on internal and external validation, respectively. Calibration plots showed that the predicted and actual CRS1 probabilities were fitted well on both internal and external validation.
The proposed nomogram could predict the individualized risk of CRS1 with good accuracy, high discrimination, and potential clinical applicability in patients with AHF.
背景/目的：1型心肾综合征（CRS1）是急性心力衰竭（AHF）患者中一种严重的临床情况，与不良临床结局相关。尽管已有多种用于识别CRS1的生物标志物的报道，但早期准确预测CRS1仍然是一项挑战。本研究旨在开发并验证一种针对AHF患者CRS1风险的个体化预测列线图。
本研究纳入了2013年至2018年间共1235例AHF患者。患者被随机分为训练集（n = 823）和验证集（n = 412）。训练集的所有数据用于通过单因素和多因素分析筛选CRS1的预测因素。基于这些预测因素开发列线图，并通过内部和外部验证进行验证。列线图验证包括通过受试者工作特征（ROC）曲线下面积（AUC）确定的判别能力以及校准图的预测准确性。
CRS1的总体发生率为31.7%。多因素逻辑回归显示，年龄、糖尿病、纽约心脏协会（NYHA）心功能分级、估算肾小球滤过率（eGFR）、高敏C反应蛋白（hs-CRP）和尿血管紧张素原（uAGT）与CRS1独立相关。基于这六个变量开发的列线图在内部验证和外部验证中的AUC分别为0.885和0.823。校准图显示，在内部和外部验证中，预测的和实际的CRS1概率拟合良好。
所提出的列线图能够准确预测AHF患者CRS1的个体化风险，具有高判别性和潜在的临床适用性。
