Diagnostic performance of a novel automated CT-derived FFR technology in detecting hemodynamically significant coronary artery stenoses: A multicenter trial in China

Background and aims: Computed tomography-derived fractional flow reserve (CT-derived FFR) algorithms have emerged as promising noninvasive methods for identifying hemodynamically significant coronary artery disease (CAD). However, its broad adaption is limited by the complex workflow, slow processing, and supercomputer requirement. Therefore, CT-derived FFR solutions capable of producing fast and accurate results could help deliver time-sensitive results rapidly and potentially alter patient management. The current study aimed to determine the diagnostic performance of a novel CT-derived FFR algorithm, esFFR, on patients with CAD was evaluated. Methods: 329 patients from 6 medical centers in China were included in this prospective study. CT-derived FFR calculations were performed on 350 vessels using the esFFR algorithm using patients' presenting coronary computed tomography angiography (CCTA) images, and results and processing speed were recorded. Using invasive FFR measurements from direct coronary angiography as the reference standard, the diagnostic performance of esFFR and CCTA in detecting hemodynamically significant lesions were compared. Post-hoc analyses were performed for patients with calcified lesions or stenoses within the CT-derived FFR diagnostic "gray zone." Results: The esFFR values correlated well with invasive FFR. The sensitivity, specificity, accuracy, positive and negative predictive value for esFFR were all above 90%. The overall performance of esFFR was superior to CCTA. Coronary calcification had minimal effects on esFFR's diagnostic performance. It also maintained 85% of diagnostic accuracy for "gray zone" lesions, which historically was <50%. The average esFFR processing speed was 4.6 +/- 1.3 minutes. Conclusions: The current study demonstrated esFFR had high diagnostic efficacy and fast processing speed in identifying hemodynamically significant CAD.