Improving the diagnostic performance of the real world coronary computed tomography angiography including uninterpretable segments.

The purpose was to investigate the diagnostic performance of coronary computed tomography angiography (CTA) when non-calcified uninterpretable segments were determined as either obstructive or patent. We also investigated the factors that could improve the diagnosis of CTA. A total of 268 patients without known coronary artery disease who were clinically indicated for coronary angiogram (CAG) within 50days of coronary CTA were retrospectively included. The diagnostic performance of CTA was assessed with CAG as a reference, whereas stenosis of ≥50% was considered obstructive. We compared the results when non-calcified uninterpretable segments were determined as obstructive or patent. Coronary risk factors as well as contrast medium arrival time adjusted by heart rate (CATHR) were investigated for improvement of CTA diagnosis. Area under the receiver operating characteristic curve (AUC) improved when uninterpretable segments were determined as patent rather than obstructive (0.79 vs 0.73, p=0.02). Multivariate analysis showed that CATHR was a predictor of CAG stenosis (odds ratio 1.13, p=0.046) while other risk factors were not. Adding CATHR further improved the AUC to 0.82 (p=0.003). The accuracy, sensitivity, specificity, positive predictive value and negative predictive value of CTA stenosis (uninterpretable segments as obstructive) were 72%, 99%, 32%, 68% and 95%. The values were 78%, 89%, 61%, 77% and 80% when CATHR was added and uninterpretable segments determined as patent. The diagnostic performance of coronary CTA improved when non-calcified uninterpretable segments were determined as patent rather than obstructive. Adding CATHR could further improve the specificity.