Abstract 14555: Early Predictors That Identify Patients With Significant Left Main and Three-Vessel Coronary Artery Disease in Non-ST Elevation Acute Coronary Syndrome: Validating the MADISON Score

Abstract

Introduction: In patients with non-ST elevation acute coronary syndrome (NSTE-ACS), identifying those with a high probability of having high-risk disease such as left main (LM) or three-vessel disease (3VD) could influence decisions to preload with P2Y12 inhibitors and timing of coronary angiography. The MADISON score was developed to predict LM/3VD on coronary angiography in patients with NSTE-ACS. In this study, we validated the MADISON score using data from an external cohort. Methods: The MADISON score was first developed using a retrospective cohort of 1700 patients presenting with NSTE-ACS at the University of Wisconsin-Madison (UW) (2011-2018). Those with prior CABG, cardiac arrest/shock, and bleeding on initial presentation were excluded from the analysis. Multivariable estimates were employed and model performance was assessed using receiver operating characteristic curve analysis. The score was then validated using a retrospective Houston Methodist Hospital (HMH) cohort of 622 patients (2019-2022). Results: The MADISON score variables and their adjusted odds ratios (95% CI) are presented in Figure 1A. Risk categories and corresponding LM/3VD probabilities are presented in Figure 1B. In the UW cohort, the area under the curve (AUC) was 0.68 (0.65-0.71) (Figure 2A). Interestingly, the AUC was also 0.68 (0.64-0.72) in the HMH validation cohort (Figure 2B). Conclusions: The MADISON score performed well and similarly between 2 large, independent cohorts in predicting patients most likely to have LM/3VD. The MADISON score is a simple and practical tool to guide decisions. Prospective validation of the score is planned to evaluate its impact on clinical outcomes.