Letters Section Editor: Jody W. Zylke, MD, Senior Editor.
Author Affiliations: Weatherhead PET Center for Preventing and Reversing Atherosclerosis, University of Texas Medical School, Houston (email@example.com).
To the Editor: Modeling FFR using CT angiograms represents the latest of many attempts to predict physiology from anatomy alone. All such techniques over the last 40 years (eg, percentage diameter stenosis, minimum lumen area or diameter, and stenosis flow reserve) have shown reasonable group agreement but unacceptable imprecision for individual application.
A large scatter between FFR and FFRCT most likely existed in the DeFACTO study,1 although it was not explicit in the article, leaving readers to infer its existence from the tables comparing dichotomized FFR and FFRCT. At best, the paired plot of FFR and FFRCT would mirror the results of the Diagnosis of Ischemia-Causing Stenoses Obtained Via Noninvasive Fractional Flow Reserve (DISCOVER-FLOW) trial,2 which showed wide limits of agreement between FFRCT and FFR of ±0.13 in addition to a slight bias of −0.02 (underestimation by FFRCT) and correlation of 0.72. The discordance may be even larger in the DeFACTO study because per-patient accuracy (73% vs 87%) and specificity (54% vs 82%) were lower than in the DISCOVER-FLOW study.
Johnson NP, Kirkeeide RL, Gould KL. Noninvasive Approach to Assess Coronary Artery Stenoses and Ischemia. JAMA. 2013;309(3):233-236. doi:10.1001/jama.2012.157189