Diagnostic accuracy of noninvasive fractional flow reserve derived from computed tomography angiography in ischemia-specific coronary artery stenosis and indeterminate lesions: results from a multicenter study in China

Abstract

Background: To determine the diagnostic performance of a novel computational fluid dynamics (CFD)-based algorithm for in situ CT-FFR in patients with ischemia-induced coronary artery stenosis. Additionally, we investigated whether the diagnostic accuracy of CT-FFR differs significantly across the spectrum of disease and analyzed the influencing factors that contribute to misdiagnosis.

Methods: Coronary computed tomography angiography (CCTA), invasive coronary angiography (ICA), and FFR were performed on 324 vessels from 301 patients from six clinical medical centers. Local investigators used CCTA and ICA to conduct assessments of stenosis, and CT-FFR calculations were performed in the core laboratory. For CCTA and ICA, CT-FFR ≤ 0.8 and a stenosis diameter ≥ 50% were identified as lesion-specific ischemia. Univariate logistic regression models were used to assess the effect of features on discordant lesions (false negative and false positive) in different CT-FFR categories. The diagnostic performance of CT-FFR was analyzed using an invasive FFR ≤ 0.8 as the gold standard.

Results: The Youden index indicated an optimal threshold of 0.80 for CT-FFR to identify functionally ischemic lesions. On a per-patient basis, the diagnostic sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV) for CT-FFR were 96% (91%-98%), 92% (87%-96%), 94% (90%-96%), 91% (85%-95%), and 96% (92%-99%), respectively. The diagnostic efficacy of CT-FFR was higher than that of CCTA without the influence of calcification. Closer to the cut point, there was less certainty, with the agreement between the invasive FFR and the CT-FFR being at its lowest in the CT-FFR range of 0.7-0.8. In all lesions, luminal stenosis ≥ 50% significantly affected the risk of reduced false negatives (FN) and false positives (FP) results by CT-FFR, irrespective of the association with calcified plaque.

Conclusions: In summary, CT-FFR based on the new parameter-optimized CFD model has a better diagnostic performance than CTA for lesion-specific ischemia. The presence of calcified plaque has no significant effect on the diagnostic performance of CT-FFR and is independent of the degree of calcification. Given the range of applicability of our software, its use at a CT-FFR of 0.7-0.8 requires caution and must be considered in the context of multiple factors.

Keywords: computational fluid dynamics (CFD); computed tomography (CTA); coronary artery disease (CAD); coronary computed tomography-derived fractional flow reserve (CT-FFR); indeterminate lesions.

Figure 1

Study enrolment. CCTA, coronary computed tomography angiography; CT-FFR, coronary computed tomography angiography derived fractional flow reserve; FFR, fractional flow reserve.

Figure 2

(A) multiplanar reformatting of (a) coronary CT angiography, (b) CT-FFR of the right coronary artery system, and CT-FFR value (0.886) was measured (c) invasive coronary angiography, (d) invasive FFR measurements, the FFR measured at the corresponding position was 0.88. (B) (a) Routine coronary computed tomography angiography are received and reconstruction of anatomical model, (b) Coronary artery physiological models were measured at corresponding locations, (c) The physical laws of fluid dynamics are used to calculate coronary blood flow, and calculation of fractional flow reserve from a standard acquired coronary computed tomography datasets.

Figure 3

Receiver operating characteristic curve for diagnostic performance of CT-FFR and CCTA. The area under receiver operating characteristic curve (AUC) for the detection of ischemia by CT-FFR, and CCTA in lesions using invasive FFR as the reference standard; CCTA, coronary computed tomography angiography; CT-FFR, fractional flow reserve derived from coronary computed tomography angiography. (A) Per-patient; (B) Per-vessel; (C) Per patient in the high calcification group; (D) Per patient in the mild-to-moderate calcification.

Figure 4

Bland-Altman and scatter plots for the association of CT-FFR and FFR. For all patients (n = 303) (panels A,B); for all vessels (n = 324) (panels C,D); for vessels with a CTFFR of 0.7–0.8 (n = 91) (panels E,F). CT-FFR, fractional flow reserve derived from coronary computed tomography angiography; FFR, fractional flow reserve.

Figure 5

(A) diagnostic performance of CT-FFR in different categories. The diagnostic accuracy of lesions in CT-FFR (0.70–0.80) was the lowest, and the further from the zone, the higher the accuracy. (B) The area under the receiver operating characteristic curve (AUC) for detecting CT-FFR (0.7–0.8), the AUC was significantly lower in this interval compared to the overall.