Combined Risk Stratification With Patient Characteristics and Biomarkers in Patients Treated With the Impella for Cardiogenic Shock

Abstract

Background: In patients with cardiogenic shock (CS), a percutaneous microaxial ventricular assist device (Impella, Abiomed, Danvers, MA) is a choice for temporary mechanical circulatory support. Given the high morbidity and mortality in this patient population, early risk stratification is relevant when making treatment decisions.

Methods: Using nationwide registry data between February 2020 and December 2022 in Japan, we included a total of 4122 patients with cardiogenic shock treated with the Impella devices. Using logistic regression analysis, we incorporated patient characteristics and biomarkers to develop a risk-stratifying model for in-hospital mortality. The model was also tested if applicable to composite outcomes of in-hospital death and major complications.

Results: Of the 4122 patients with cardiogenic shock, the Impella was indicated for acute myocardial infarction in 2575 (62.5%). Multivariable analysis identified 4 patient characteristics (age, body mass index, out-of-hospital cardiac arrest, and blood pressure) and 6 biomarkers (lactate, lactate dehydrogenase, creatinine, total bilirubin, albumin, and creatinine kinase) with cutoff values as factors significantly associated with in-hospital mortality. We developed a risk-stratifying model using the 10 variables, which was predictive of in-hospital death (area under the curve, 0.711; P<0.001). Adding biomarkers to patient characteristics significantly improved the diagnostic accuracy (area under the curve, from 0.649 to 0.711; P<0.001). This risk score was also predictive of death and major complications (area under the curve, 0.680; P<0.001).

Conclusions: In patients with cardiogenic shock treated with the Impella devices, our risk-stratifying system, consisting of 4 patient characteristics and 6 biomarkers, strongly correlated with in-hospital mortality, potentially facilitating clinical decision-making.

Keywords: Impella; cardiogenic shock; mechanical circulatory support; risk stratification.

Figure 1. Study flow.

Impella (Abiomed, Danvers, MA) is an intravascular microaxial left ventricular assist device. ALT indicates alanine aminotransferase; AMI, acute myocardial infarction; AST, aspartate aminotransferase; BMI, body mass index; CAD, coronary artery disease; CK, creatinine kinase; CRP, C‐reactive protein; CS, cardiogenic shock; DLP, dyslipidemia; HD, hemodialysis; HF, heart failure; HR, heart rate; LDH, lactate dehydrogenase; OHCA, out‐of‐hospital cardiac arrest; ROC, receiver operating characteristic; SBP, systolic blood pressure; T‐Bil, total bilirubin; and TIA, transient ischemic attack.

Figure 2. Receiver operating characteristic curve analysis of either patient characteristics or biomarkers or both for in‐hospital mortality.

All 23 variables as shown in Figure 1 (14 patient characteristics and 9 biomarkers) are incorporated into the model. The optimism‐corrected AUCs by a bootstrap method are 0.704, 0.684, and 0.745 for patient characteristics alone, biomarkers alone, and patient characteristics plus biomarkers, respectively. AUC indicates area under the curve.

Figure 3. Components of the J‐PVAD Mortality Risk Score.

BP indicates blood pressure; CK, creatinine kinase; J‐PVAD, Japan Registry for Percutaneous Ventricular Assist Device; LDH, lactate dehydrogenase; and OHCA, out‐of‐hospital cardiac arrest.

Figure 4. Receiver operating characteristic curve analysis of either patient characteristics or biomarkers or both for in‐hospital mortality.

A total of 10 variables shown in Figure 3 (4 patient characteristics and 6 biomarkers) are incorporated into the model. The optimism‐corrected AUCs by a bootstrap method are 0.648, 0.662, and 0.710 for patient characteristics alone, biomarkers alone, and patient characteristics plus biomarkers, respectively. AUC indicates area under the curve.

Figure 5. Distribution of the J‐PVAD Mortality Risk Score and rates of in‐hospital death and complications.

J‐PVAD indicates Japan Registry for Percutaneous Ventricular Assist Device.

Figure 6. Receiver operating characteristic curve analysis of either patient characteristics or biomarkers or both for in‐hospital mortality and major complications.

A total of 10 variables shown in Figure 3 (4 patient characteristics and 6 biomarkers) are incorporated into the model. The optimism‐corrected AUCs by a bootstrap method are 0.622, 0.643, and 0.680 for patient characteristics alone, biomarkers alone, and patient characteristics plus biomarkers, respectively. AUC indicates area under the curve.