Haitham Abu Khadija, Mohammad Alnees, Nizar Abu Hamdeh, Nicholay Teodorovich, Mohammad Masu'd, Ramon Cohen, Duha Najajra, Alexander Kogan, Adam Mahamid, Jebrin Alkrinawi, Eyal Nachum, Omar Ayyad, Alena Kirzhner, Kareem Ibraheem, Osama Hroub, Ehud Raanani, Ehud Karni, Yahya Z Fraitekh, Abdalaziz Darwish, Leonid Sternik
In a large multicenter CABG cohort, the AMP score was independently associated with 30-day stroke and provided incremental risk stratification beyond conventional clinical factors. As a pragmatic surrogate derived from routinely available data, the AMP score may support early postoperative risk assessment, although external validation is warranted.
BACKGROUND: Stroke remains a devastating complication after coronary artery bypass grafting (CABG), often related to intraoperative aortic manipulation and embolization. However, no simple and scalable surrogate exists to approximate procedural complexity and embolic risk using routinely collected variables. We developed and evaluated the Aortic Manipulation Proxy (AMP) score for predicting 30-day stroke after isolated CABG.
METHODS: We conducted a retrospective international multicenter cohort study including 6,818 adults undergoing isolated CABG. The primary outcome was 30-day postoperative stroke (n = 133; 1.95%). The AMP score was constructed from routinely documented operative variables. Associations were assessed using Cox proportional hazards models stratified by center. Discrimination was evaluated using the area under the receiver operating characteristic curve (AUC), and incremental predictive value was examined using ΔAUC and decision curve analysis. Sensitivity analysis, excluding postoperative atrial fibrillation and internal validation using bootstrap resampling, was performed.
RESULTS: Higher AMP scores were associated with a graded increase in 30-day stroke risk. In center-stratified models, AMP independently predicted stroke per 1-SD increase (HR 1.46; 95% CI, 1.23-1.73; p < 0.001) and per 2-unit increase (HR 1.62; 95% CI, 1.30-2.01; p < 0.001). Kaplan-Meier analysis demonstrated stepwise separation across AMP tertiles (log-rank p = 0.005). Addition of AMP to a clinical model modestly improved discrimination (AUC 0.731 vs. 0.707; ΔAUC = 0.024; p = 0.021) and demonstrated improved net benefit across clinically relevant thresholds. Findings were consistent in sensitivity analyses and bootstrap validation.
CONCLUSIONS: In a large multicenter CABG cohort, the AMP score was independently associated with 30-day stroke and provided incremental risk stratification beyond conventional clinical factors. As a pragmatic surrogate derived from routinely available data, the AMP score may support early postoperative risk assessment, although external validation is warranted.