Optimized interpretation of fractional flow reserve derived from computed tomography: Comparison of three interpretation methods

An optimal system for interpreting fractional flow reserve (FFR) values derived from CT (FFRCT) is lacking. We sought to evaluate performance of three FFRCT measurements in detecting ischemia by comparing them with invasive FFR.For 73 vessels in 50 patients who underwent coronary CT angiography (CCTA) and FFRCT analysis followed by invasive FFR, the greatest diameter stenosis on CCTA, FFRCT difference between distal and proximal to the stenosis (ΔFFRCT), FFRCT 2 cm distal to the stenosis (lesion-specific FFRCT), and the lowest FFRCT in distal vessel tip were calculated. Significant obstruction (≥50% diameter stenosis) and ischemia (lesion-specific FFRCT ≤0.80, the lowest FFRCT ≤0.80, or ΔFFRCT ≥0.12 based on the greatest Youden index) were compared with invasive FFR (≤0.80).Forty (55%) vessels demonstrated ischemia during invasive FFR. On multivariable generalized estimating equations, ΔFFRCT (odds ratio [OR] 10.2, p < 0.01) remained a predictor of ischemia over CCTA (OR 2.9), lesion-specific FFRCT (OR 3.1), and the lowest FFRCT (OR 0.9) (p > 0.05 for all). Area under the curve (AUC) of ΔFFRCT (0.86) was higher than CCTA (0.66), lesion-specific FFRCT (0.71), and the lowest FFRCT (0.65) (p < 0.01 for all). Addition of each FFRCT measure to CCTA showed improvement of AUC and significant net reclassification improvement (NRI): ΔFFRCT (AUC 0.84, NRI 1.24); lesion-specific FFRCT (AUC 0.77, NRI 0.83); and the lowest FFRCT (AUC 0.76, NRI 0.59) (p < 0.01 for all).Compared with diameter stenosis, ΔFFRCT, lesion-specific FFRCT, and the lowest FFRCT improved ischemia discrimination and reclassification, with ΔFFRCT being superior in identifying and discriminating ischemia.