Abstract
BACKGROUND: There is emerging evidence that plaque features may play a critical role in future acute coronary syndrome. In this study, we analyzed plaque features using an artificial intelligence-enabled algorithm in a clinical cohort who developed non-ST-elevation myocardial infarction (NSTEMI) following coronary CT angiogram (CCTA). METHODS: We performed a case-control study selected from 13 751 consecutive cases in a single center referred for outpatient CCTA. After a follow-up of 4.3 ± 4 years, 48 patients without preexisting coronary disease developed NSTEMI. Controls (N = 187) were matched to the cases on age, gender, BMI, and kilovoltage for CTA acquisition. Quantitative plaque analysis was performed using artificial intelligence-enabled Autoplaque software (Autoplaque version 3.0; Cedars-Sinai Medical Center, Los Angeles, California, USA). Multivariable Cox proportional hazards models were performed to identify the predictors of NSTEMI. RESULTS: The mean age was 64 ± 11 years. Both case and control groups had mild stenosis at baseline (26 vs 17%, P = 0.01). The total calcified plaque and fibrous plaque volume were not different (P = 0.10 and P = 0.13, respectively). Necrotic core plaque volume and fibrous fatty plaque volume were higher in the NSTEMI group (28 ± 29 vs 9 ± 13 mm3, 169 ± 157 vs 84 ± 105 mm3, respectively, both P < 0.01). In multivariable Cox regression, necrotic core volume portended the greatest risk of NSTEMI, a seven-fold higher than that of total plaque volume. CONCLUSION: Using artificial intelligence-enabled plaque analysis, noncalcified plaque volume, especially necrotic core and fibrous fatty plaque volume are important precursors for future NSTEMI events.