Diagnostic accuracy of coronary computed tomography angiography-derived fractional flow reserve

Abstract

Background: Fractional flow reserve (FFR) is a widely used gold standard to evaluate ischemia-causing lesions. A new method of non-invasive approach, termed as AccuFFRct, for calculating FFR based on coronary computed tomography angiography (CCTA) and computational fluid dynamics (CFD) has been proposed. However, its diagnostic accuracy has not been validated.

Objectives: This study sought to present a novel approach for non-invasive computation of FFR and evaluate its diagnostic performance in patients with coronary stenosis.

Methods: A total of 54 consecutive patients with 78 vessels from a single center who underwent CCTA and invasive FFR measurement were retrospectively analyzed. The CT-derived FFR values were computed using a novel CFD-based model (AccuFFRct, ArteryFlow Technology Co., Ltd., Hangzhou, China). Diagnostic performance of AccuFFRct and CCTA in detecting hemodynamically significant coronary artery disease (CAD) was evaluated using the invasive FFR as a reference standard.

Results: Diagnostic accuracy, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for AccuFFRct in detecting FFR ≤ 0.8 on per-patient basis were 90.7, 89.5, 91.4, 85.0 and 94.1%, respectively, while those of CCTA were 38.9, 100.0, 5.71, 36.5 and 100.0%, respectively. The correlation between AccuFFRct and FFR was good (r = 0.76 and r = 0.65 on per-patient and per-vessel basis, respectively, both p < 0.0001). Area under the curve (AUC) values of AccuFFRct for identifying ischemia per-patient and per-vessel basis were 0.945 and 0.925, respectively. There was much higher accuracy, specificity and AUC for AccuFFRct compared with CCTA.

Conclusions: AccuFFRct computed from CCTA images alone demonstrated high diagnostic performance for detecting lesion-specific ischemia, it showed superior diagnostic power than CCTA and eliminated the risk of invasive tests, which could be an accurate and time-efficient computational tool for diagnosing ischemia and assisting clinical decision-making.

Keywords: CT-derived FFR; Computational fluid dynamics; Coronary computed tomography angiography; Fractional flow reserve.

Fig. 1

Per-patient correlations (r = 0.76, p < 0.0001) and per-vessel correlations (r = 0.65, p < 0.0001) between wire-based FFR and AccuFFRct. FFR fractional flow reserve

Fig. 2

Bland–Altman plot of FFR and AccuFFRct on the per-patient and per-vessel basis, respectively. FFR = fractional flow reserve

Fig. 3

AccuFFRct results with invasive FFR measurement. a CCTA demonstrating 80% stenosis at the proximal to middle portion of LAD (green arrow); b a computed AccuFFRct value of 0.79 (red arrow); c the corresponding measured FFR value of 0.75, demonstrating stenosis ischemia. FFR fractional flow reserve; CCTA coronary computed tomography angiography, LAD left anterior descending artery

Fig. 4

Areas under the curve (AUC) for receiver-operating characteristics (ROC) of AccuFFRct and CCTA, for per-patient and per-vessel basis. CCTA coronary computed tomography angiography

Fig. 5

Flowchart for computing AccuFFRct: a CCTA image data; b segmented 3D coronary artery model; c segmented 3D ventricle model; d mesh generation; e coronary flow computational algorithm for computing coronary flow, pressure from Navier–Stokes flow governing equations; f AccuFFRct distribution over the coronary artery tree. g CCTA coronary computed tomography angiography, 3D three-dimensional