Diagnostic performance of on-site computation of quantitative flow ratio by a coronary computed tomography angiography based algorithm: comparison of distal and lesion-specific measurements

Abstract

Introduction Guidelines recommend secondary ischemia assessment following a coronary computed tomography angiography (CTA) with suspected obstructive coronary artery disease (CAD). Coronary CTA-derived quantitative flow ratio (CT-QFR) is an on-site technique performed on acquired CTA images that estimates the functional severity of a coronary stenosis. However, CT-QFR measurements are available throughout the coronary vessel with no clear recommendations as to which specific values should be used for identifying obstructive CAD, e.g. most distal or lesion-specific values. Purpose First, to investigate the feasibility of CT-QFR and the correlation and agreement with invasive fractional flow reserve (FFR). Secondly, to compare the diagnostic performance of distal versus lesion-specific CT-QFR for identifying obstructive CAD defined by invasive coronary angiography (ICA) with FFR. Methods A total of 1732 prospectively included patients with symptoms suggestive of CAD referred for CTA were included. All patients with ≥50% diameter stenosis (DS) on CTA were subsequently referred for ICA with conditional FFR in lesions with 30–89%DS. Obstructive CAD was defined by ICA as FFR ≤0.80 or high-grade stenosis by visual assessment (≥90%DS). A blinded analysis of CT-QFR was performed in patients referred to ICA with measurements at the distal end of a vessel (distal CT-QFR) and 1 cm distal to stenotic lesions on CTA (lesion-specific). CT-QFR ≤0.80 was defined as abnormal. For correlation analyses to invasive FFR, CT-QFR was assessed corresponding to the position of the invasive pressure sensor. Results In total, 445/1732 (25%) patients had suspected obstructive CAD at CTA and underwent subsequent ICA. CT-QFR analysis was feasible in 423/445 (95%) patients. CT-QFR correlated (Pearson's rho 0.54, p<0.001) and agreed (mean difference –0.02±0.09) to FFR with CT-QFR overestimating FFR (Fig. 1). Obstructive CAD was identified in 190/423 (44%) patients by ICA. Distal and lesion-specific CT-QFR classified 196 (46%) and 171 (40%) patients as abnormal, respectively. Areas under the receiver-operating characteristic curves for distal versus lesion-specific CT-QFR were similar (0.86 (95% CI: 0.82–0.89) vs. 0.86 (0.82–0.90), p=0.80). Sensitivities for distal and lesion-specific CT-QFR were 78% (95% CI: 71–84) vs. 74% (67–80), p=0.01, respectively, and specificities 79% (95% CI: 74–84) vs. 87% (82–91), p<0.01, respectively. Distal and lesion-specific CT-QFR had similar diagnostic accuracy (79 (95% CI: 75–83), vs. 81 (77–85), p=0.07) (Fig. 2). Conclusion In patients with suspected obstructive CAD on CTA, non-invasive estimation of FFR using CT-QFR is feasible with moderate correlation and good agreement with invasive FFR. Overall diagnostic performance of distal and lesion-specific values for discriminating obstructive CAD by invasive FFR are similar. The use of CT-QFR could therefore potentially reduce the need for referral to invasive angiography after CTA. Funding Acknowledgement Type of funding sources: Public Institution(s). Main funding source(s): Aarhus UniversityRegion Mid Jutland