Résumé
Coronary artery disease (CAD) severity assessed by angiographic features is a strong predictor of outcomes. However, recommended risk stratification tools, such as GRACE risk score, do not incorporate these angiographic features.
To assess the additional prognostic value of angiographic features after ACS for in-hospital outcomes prediction beyond traditional stratification tool.
From 7th to 21st April 2021, the Addiction in Intensive Care Unit (ADDICT-ICCU) study recruited all consecutive patients admitted to ICCUs across 39 French hospitals. This prespecified study enrolled patients presenting with ACS, with available invasive coronary angiography. Patients with Killip class≥2 at admission, thrombolysis, or history of coronary-artery-bypass graft were excluded. An independent and centralized angiographic core laboratory analyzed all angiograms. The outcome was in-hospital major adverse event (MAE) defined as a composite of all-cause death, cardiac arrest, cardiogenic shock requiring inotropic or mechanical support, and congestive heart failure. An angiographic model was built by incorporating angiographic features, selected through a stepwise process, into the initial model based on the GRACE risk score.
This study included 446 patients (mean age 62±13 years; 77% male; 47% ST-elevation myocardial infarction). In-hospital MAE occurred in 39 (9%) patients. Patients experiencing MAE were older (67 vs. 61 years; P<0.01), had a higher prevalence of diabetes (33 vs. 19%; P=0.04), and lower left ventricular ejection fraction (LVEF) (43 vs. 54%; P<0.01) compared to those without events. Concerning coronary angiography features, patients with in-hospital MAE exhibited a higher mean syntax score (16 vs. 10; P<0.01), and higher rates of left main stenosis (18 vs. 7%; P=0.02), massive calcifications (23 vs. 6%; P<0.01), and chronic total occlusion (26 vs. 7%; P<0.01) as compared to those without events. For in-hospital MAE prediction, the GRACE risk score demonstrated a ROC-AUC of 0.69. The addition of selected angiographic features to the GRACE risk score improved model discrimination with a ROC-AUC of 0.75 (P=0.05) and IDI of 7% (P<0.01), along with a reclassification improvement (continuous NRI=64%; P<0.01) as illustrated in Fig. 1.
In patients presenting ACS, angiographic features of CAD severity improved performance of GRACE risk score for in-hospital adverse outcomes prediction.