Coronary artery calcium score to predict coronary CT angiography interpretability: an old problem revisited

Abstract Funding Acknowledgements Type of funding sources: None. Introduction Clinical guidelines recommend against the use of coronary computed tomography angiography (CCTA) in patients with heavy calcification due to interpretability concerns, but no specific approach or threshold is provided. Recently, alternative methods have been proposed as more reliable predictors of CCTA interpretability than the classic coronary artery calcium score (CACS). The purpose this study was to compare the performance of different measures of coronary calcification as predictors of CCTA interpretability. Methods We conducted a retrospective analysis of consecutive patients undergoing CACS and CCTA between 2018 and 2020. The key exclusion criteria were known coronary artery disease, CACS of zero, and presence of non-assessable coronary lesions for reasons other than calcification (movement/gating artifacts or vessel diameter < 2mm). CCTA studies were considered non-interpretable if the main reader considered one or more coronary lesions non-assessable due to calcification. Three different measures of coronary calcification were compared using ROC curve analysis: 1) total CACS; 2) CACS-to-lesion ratio (total CACS divided by the number of calcified plaques); and 3) calcium score of the most calcified plaque. Decision-tree analysis was performed to identify the algorithm that best predicts CCTA interpretability. Results A total of 432 patients (191 women, mean age 64 ± 11 years) were included. Overall, 31 patients (7.2%) had a non-interpretable CCTA due to calcification. Patients with non-interpretable CCTA had higher CACS (median 589 vs. 50 AU, p < 0.001), higher CACS-to-lesion ratio (median 43 vs. 14 AU/lesion, p < 0.001), and higher score of the most calcified plaque (median 445 vs. 43 AU, p < 0.001). Among the 3 methods, CACS showed the highest discriminative power to predict a non-interpretable CCTA (C-statistic 0.93, 95%CI 0.89-0.95, p < 0.001) - Figure. Decision-tree analysis identified a single-variable algorithm (CACS value ≤ 515 AU) as the best discriminator of CCTA interpretability: 396 of the 409 patients (97%) with CACS ≤ 515 AU had an interpretable CCTA, whereas only 5 of the 23 patients (22%) with CACS > 515 AU had an interpretable test, yielding a total of 96% correct predictions. Conclusions The recently proposed and more complex measures of coronary calcification seem unable to outperform total CACS as a predictor of CCTA interpretability. A simple CACS cutoff-value around 500 AU remains the best discriminator for this purpose.