Semi-Quantitative Versus Visual Analysis of Adenosine Perfusion Magnetic Resonance Imaging in Intermediate-Grade Coronary Artery Stenosis Using Fractional Flow Reserve as the Reference: A Pilot Study

Abstract

Background: To evaluate the diagnostic accuracy of semi-quantitative adenosine perfusion magnetic resonance imaging (MRI) to determine fractional flow reserve (FFR) <= 0.80 intermediate-grade coronary stenoses as compared to visual analysis. Methods: Forty-six patients (mean age 61 +/- 9 years; 33 males) with 49 intermediate-grade stenoses (59 +/- 7.6%; range, 42-70% minimal diameter reduction) underwent adenosine perfusion MRI and FFR measurement within four months in this retrospective study. MRI was visually assessed by two experienced readers twice with one-year interval, the second time with the knowledge of the diseased artery. The stress subendocardial myocardial enhancement maximal upslope was evaluated distal to the coronary stenosis (=RISK) and divided by the same value in remote myocardium supplied by normal arteries (=REMOTE) to obtain the relative myocardial perfusion index (RMPI). Results: The average FFR value was 0.84 +/- 0.09 and 15/49(31%) intermediate-grade stenoses were FFR <= 0.80. The kappa-values for interobserver agreement assessing inducible perfusion defects on visual readings was 0.20 on the first reading and increased to 0.62 with the knowledge of the stenosis location. Consensus readings had a diagnostic accuracy of 82%(40/49) in identifying FFR <= 0.80 stenoses on both blinded and unblinded readings with regards to the knowledge of the stenosis location. Meanwhile, stress subendocardial RMPI had higher accuracy (43/49[88%]) than visual readings to predict FFR <= 0.80 stenoses, using a cutoff value of 0.84. Conclusion: By assessing perfusion changes in remote myocardium, semi-quantitative MRI analysis using stress subendocardial RMPI can provide an equal or more accurate alternative to visual analysis in identifying FFR <= 0.80 intermediate-grade stenoses. Larger cohorts of patients are required to validate this approach.