Additive Value of Semi-Automated Quantification of Coronary Artery Disease Using Cardiac CT-Angiography to Predict for Future Acute Coronary Syndrome

Study Questions:

Can a semi-automated coronary plaque algorithm identify plaque features on coronary computed tomographic angiography (CCTA) that are associated with risk of future acute coronary syndrome (ACS)?


From a population of 1,650 individuals undergoing CCTA for stable chest pain at two institutions and with follow-up for ACS (mean follow-up 26 ± 10 months), this retrospective study compared CCTA findings between 25 individuals with subsequent ACS to 101 randomly selected controls with coronary artery disease (CAD), but no ACS. Semi-automated quantification of plaque features (plaque volume, burden, area, % noncalcified, mean attenuation, and remodeling index) and standard CCTA findings (calcium score, luminal stenosis severity, and qualitative plaque composition) were compared between groups.


Mean age was 63 ± 10 years and 75 (60%) were male. Between individuals with ACS and controls, there were no significant differences in mean age, gender, or risk factors other than diabetes (24% vs. 8%, p = 0.03). Using standard CCTA findings, there were no differences between those with ACS and controls with regard to calcium score, stenosis severity, or the number of segments with noncalcified plaque (p = not significant for each); using semi-automated quantification, there were significant differences in the median number of plaques (4 vs. 2, p = 0.02), total plaque volume (94 vs. 29 mm3, p = 0.001), total noncalcified plaque volume (28 vs. 4 mm3, p < 0.001), highest plaque volume (56 vs. 24 mm3, p < 0.001), highest plaque burden (57 vs. 36 mm3, p < 0.001), plaque with largest area (9 vs. 5 mm2, p = 0.006), plaque with highest noncalcified proportion (62 vs. 26%, p = 0.006), plaque with lowest mean attenuation (206 vs. 282 Hounsfield units, p = 0.005), and plaque with highest remodeling index (1.5 vs. 1.3, p = 0.01). In ACS patients, there were significant differences between culprit (n = 24) and nonculprit (n = 74) lesions in regard to semi-quantitative plaque volume, noncalcified volume, burden, area, and mean attenuation (p < 0.05 for each). Finally, for prediction of ACS, the area under the receiver-operator curve improved from 0.64 using Framingham risk score and standard CCTA findings to 0.79 for a model additionally incorporating semi-automated plaque quantification.


Coronary plaque characteristics on CCTA using a semi-automated plaque quantification algorithm may identify individuals at risk of future ACS, and appear incremental to standard CCTA findings.


There is increasing data that luminal stenosis alone is a limited predictor of future ACS. This study builds on prior literature by Motoyama et al., who identified potential CCTA predictors of future ACS, such as positive plaque remodeling and low-attenuation plaque. The present findings identify multiple measures of plaque volume, burden, composition, and remodeling that are significantly higher in patients with subsequent ACS as compared to controls with CAD, but no ACS. Importantly, significant differences were also observed in plaque measures between future culprit lesions versus nonculprit lesions in patients with subsequent ACS. These findings are intriguing and prompt a need for future prospective studies to confirm these results.

Clinical Topics: Acute Coronary Syndromes, Invasive Cardiovascular Angiography and Intervention, Noninvasive Imaging, Interventions and ACS, Interventions and Coronary Artery Disease, Interventions and Imaging, Angiography, Nuclear Imaging

Keywords: Coronary Artery Disease, Acute Coronary Syndrome, Follow-Up Studies, Plaque, Atherosclerotic, Coronary Angiography, Chest Pain, Forecasting, Risk Factors, Diabetes Mellitus

< Back to Listings