The pcMPO-AF rule for predicting postoperative atrial fibrillation after coronary artery bypass grafting

Figures

• 

  Fig. 1. Pericardial and serum MPO levels are elevated in patients with POAF. A. Patients who developed POAF exhibited higher pericardial and serum MPO across the 6-hour time course (Pgroup < 0.001 for both), especially at 6 h postoperatively in pericardial fluid and baseline in blood (Ppost−op 6 h < 0.001, Pbaseline < 0.001, respectively). ***P < 0.001 in comparisons between groups after Bonferroni correction for multiple comparisons at each time point. Pbaseline and Ppost−op 6 h values were calculated by Mann-Whitney U test. B. ROC curves for the prediction of POAF. Among the five parameters, the strongest predictors were pericardial MPO at post-op 6 h (AUC = 0.831) and serum MPO at baseline (AUC = 0.795), with optimal cutoff values of 389,054.9 ng/mL (sensitivity 90.6%, specificity 62.0%) and 163.7 ng/mL (sensitivity 59.4%, specificity 79.6%), respectively. Abbreviations: MPO, myeloperoxidase; POAF, postoperative atrial fibrillation; ROC, receiver operating characteristic; AUC, area under the curve.

• 

  Fig. 2. Incremental Predictive Performance of Models for POAF. The first model includes clinical variables (age and left atrial diameter), with an AUC of 0.701. The second model combines a surgical variable (intra-aortic balloon pump) and clinical variables, increasing the AUC to 0.734. The highest AUC was observed when both serum and pericardial MPO were added to the model, which improved the AUC from 0.734 to 0.904. Abbreviations: AUC, area under the curve; MPO, myeloperoxidase; POAF, postoperative atrial fibrillation.

• 

  Fig. 3. Nomogram of the pcMPO-AF rule for POAF risk prediction. The prediction is based on the following logistic model: logit(POAF) = − 4.521 + 3.086× (pericardial MPO at post-op 6 h ≥ 389,054.9) + 2.775 × (IABP use) + 2.022 × (serum MPO at baseline ≥ 163.7) + 1.481 × (left atrial diameter ≥ 56.5) + 0.928 × (age ≥ 66.5), where each predictor is coded as 1 if the criterion is met or 0 otherwise. The nomogram can be used by determining the position of each variable on the corresponding axis, drawing a line to the point axis, and adding points from all variables. Next, a line was drawn from the total point axis to determine POAF probabilities using the bottom line of the nomogram. Abbreviations: MPO, myeloperoxidase; POAF, postoperative atrial fibrillation; IABP, intra-aortic balloon pump.

• 

  Fig. 4. Validation of the model’s predictive performance for estimating the risk of POAF. The AUCs were 0.908 (A) for the derivation cohort and 0.865 (B) for the external validation cohort. Calibration curves showed good agreement between predicted and observed outcomes in both the derivation (C) and external validation (D) cohorts. Decision curve analysis indicated that the nomogram conferred greater net benefit than the “All or None” strategy across a range of threshold probabilities in both the derivation (E) and external validation (F) cohorts. Abbreviations: POAF, postoperative atrial fibrillation; AUC, area under the curve.