Article Text
Abstract
Objective Fractional flow reserve derived from coronary CT angiography (FFR-CT) is a novel tool for assessing the significance of coronary artery stenosis. The primary aim of this prospective study was to evaluate the diagnostic performance of a novel on-site rapid FFR-CT algorithm as compared to invasive FFR as the gold standard in a real world workflow. Our secondary aim was to determine whether the FFR-CT diagnostic performance was affected by inter-observer variations in lumen segmentation.
Methods We enrolled 44 consecutive patients (64.6 ±8.9 years, 34% female) with 60 coronary atherosclerotic lesions who underwent coronary computed tomography angiography (CTA) and were referred for invasive coronary angiography (ICA) in two European centres. ICA with FFR measurements were performed within 60 days after coronary CTA in all lesions. An FFR value of ≤0.8 was considered significant. Coronary CTA scans were evaluated by two expert readers, who determined the effective diameter stenosis (EDS) and manually adjusted the semi-automated coronary lumen segmentations. All extracted vessels were evaluated by an on-site FFR simulator to calculate the FFR-CT values.
Results The mean FFR-CT value was 0.77 ±0.15 and the average coronary CTA based EDS was 43.6±16.9%. On-site lumen segmentation, manual adjustment and FFR-CT simulations were performed in an average of 9 minutes, (range: 3–25 min). The sensitivity, specificity, positive predictive value and negative predictive value of FFR-CT vs. EDS with a cut-off of 50% were the following: 90.5%, 71.8%, 63.3% and 93.3% vs. 52.4%, 87.2%, 68.8% and 77.3%. FFR-CT demonstrated significantly better diagnostic performance as compared to EDS (AUC: 0.89 vs. 0.74 respectively; p<0.001). The FFR-CT AUCs of the two readers did not show any significant difference (0.89 vs. 0.88; p=0.74).
Conclusion On-site FFR-CT simulation is feasible and has better diagnostic performance than anatomical stenosis assessment. The diagnostic performance of the FFR-CT simulation algorithm does not depend on the readers who adjust the semi-automated lumen segmentation adjustments.