Author + information
- James K. Min, MD⁎ ()
- ↵⁎Cedars-Sinai Heart Institute, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, S. Mark Taper Building, Room 1253, Los Angeles, California 90048
We thank Dr. Martus and colleagues for their interest in DISCOVER-FLOW, a prospective multicenter study that demonstrates the high diagnostic performance of fractional flow reserve derived from typically acquired coronary computed tomography angiograms (FFRCT) (1) . We share their enthusiasm for this “game-changing” technology that enables noninvasive computation of coronary flow and pressure for the determination of lesion-specific ischemia.
Dr. Martus and colleagues suggest that in the DISCOVER-FLOW study, the performance of CT stenosis severity and FFRCT against an invasive FFR reference standard may have been affected by workup bias because the decision to perform FFR was based on clinical indications after identification of ≥50% stenosis by CT. Dr. Martus and colleagues propose a statistical correction to account for individuals with no or mild stenoses by CT who were not subjected to FFR. This approach is certainly suitable to reduce the relative contribution of referral bias when performance measures of CT stenosis are judged against invasive angiography provoked by CT findings because the ability to identify or exclude a stenosis applies equally and universally to all patients undergoing CT. However, this approach is inappropriate for assessing the performance of FFRCT because the population for whom FFRCT would be expected to be applied are those with CT-identified stenoses that could both cause ischemia and be eligible for revascularization. In this regard, a >50% stenosis threshold in vessels ≥2 mm in diameter was chosen as an appropriate cutoff, given the low rates of ischemia for lesions with <50% stenosis and, more importantly, the widely accepted reluctance to revascularize nonobstructive coronary lesions. Even more fundamental, subjecting individuals who lacked any CT or invasive evidence of significant disease to FFR would have been both logistically difficult as well as questionably ethical.
Dr. Martus and colleagues note that the limits of agreement between FFRCT and invasive FFR increase in a manner that is inversely proportional to the FFR values. They provide a figure that illustrates a very shallow negative slope of a regression line superimposed on a Bland-Altman plot that visually begins to diverge from the average of FFR and FFRCT at values <0.75. There are numerous technological explanations for this, which are the subject of a review that we are preparing, but this negligible divergence has limited bearing on the clinical application of FFRCT. Robust standards for ischemia based on FFR have been firmly established at values ≤0.80 in randomized trials, and all values ≤0.80 should thus be considered ischemia causing (2). To date, the relative impact of lesions with different values of ≤0.80 remains unexplored. Further, values >0.80, where the limits of agreement of FFRCT and FFR are very close, are widely accepted as pathognomonic for those that are unquestionably not ischemia causing and for which revascularization can and should be safely avoided. The DISCOVER-FLOW study results support the high diagnostic performance of FFRCT for both of these groups.
- American College of Cardiology Foundation
- Koo B.K.,
- Erglis A.,
- Doh J.H.,
- et al.