Modeling creatine-kinase MB concentrations following coronary artery bypass grafting

Jamie L Romeo,Pieter A Vriesendorp,Kim Gerritsen,Musafir Nader,Edris Mahtab,Jos G Maessen,Arnoud W J Van't Hof,Can Gollmann-Tepeköylü,Frank van Rosmalen,Iwan C C van der Horst,Alma M A Mingels,Samuel Heuts

DOI: https://doi.org/10.1016/j.jtcvs.2024.08.019

2024-08-21

Abstract:Background: An increase in cardiac biomarkers is a prerequisite for diagnosing periprocedural myocardial infarction (PMI) after coronary artery bypass grafting (CABG). Early-phase risk detection may be aided by modeling time-dependent serum creatine kinase-MB (CK-MB) concentrations. This study aimed to model the kinetics of CK-MB while identifying its influencing factors. Methods: Patients who underwent elective CABG and had CK-MB measurements within 72 hours postoperatively were included. The primary outcome was the modeled post hoc kinetics of CK-MB in patients without potential PMI. These patients were defined as having no potential PMI based on the absence of ischemic electrocardiographic abnormalities, imaging abnormalities, in-hospital cardiac arrest, mortality, or postoperative unplanned catheterization. A web-based application was created using mixed-effect modeling to provide an interactive and individualized result. Results: A total of 1589 CK-MB measurements from 635 patients who underwent elective isolated CABG were available for analysis. Of these, 609 patients (96%) had no potential PMI and 26 (4%) had potential PMI. Male sex, aortic cross-clamp time, and cardioplegia type significantly impacted CK-MB concentrations. The diagnostic accuracy of the model had an area under the receiver operating characteristic curve of 82.8% (95% confidence interval, 72.6%-90.2%). A threshold of 7 μg/L yielded a sensitivity of 94% and a specificity of 80% (positive predictive value, 17%; negative predictive value, 99%) for excluding potential PMI in our study population. Conclusions: CK-MB release after CABG depends on the timing of measurement, patient sex, aortic cross-clamp time, and cardioplegia type. The model (available at https://www.cardiomarker.com/ckmb) can be validated, reproduced, refined, and applied to other biomarkers.