Abstract
Myeloperoxidase (MPO) is associated with conduction inhomogeneity. We hypothesized that MPO in selectively drained pericardial fluid (PCF), a direct window into the cardiac microenvironment, may improve postoperative atrial fibrillation (POAF) risk prediction beyond existing risk scores. A total of 469 consecutive patients undergoing coronary artery bypass grafting (CABG) were enrolled in this study (n = 201 in derivation cohort and n = 268 in validation cohort). The primary outcome was new-onset POAF within the first 7 days postoperatively. MPO concentrations were measured in selectively drained PCF and peripheral blood at baseline, 0 and 6 h postoperatively. We further compared MPO levels between pure pericardial fluid and mixed drainage fluid (via Y-connector). A new prediction model, the pcMPO-AF rule, was developed using multivariable logistic regression and validated internally using bootstrapping and externally in an independent cohort. Approximately 98.0% of patients underwent off-pump CABG. POAF occurred in 31.8% and 35.1% of the derivation and validation cohorts, respectively. Pericardial MPO at 6 h postoperatively emerged as the strongest independent predictor of POAF. MPO concentrations in selectively drained PCF were 25-fold higher than mixed drainage samples and 1,648-fold higher than serum levels (both P < 0.001). The pcMPO-AF rule demonstrated good discrimination, with AUCs of 0.908 in derivation and 0.865 in validation cohorts, outperforming POAF, CHA₂DS₂-VASc, and HATCH scores. MPO in selectively drained PCF is a potent biomarker of POAF. The pcMPO-AF rule integrated PCF biomarkers with clinical factors, providing superior predictive performance by capturing both the vulnerable atrial substrate and acute inflammatory trigger.