Stress perfusion CMR: improved correlation with invasive fractional flow reserve after correction for perfusion changes in remote myocardium.

PURPOSE This study was undertaken, considering fractional flow reserve (FFR) as the reference: (i) to evaluate a predictive model of flow reserve using adenosine cardiac magnetic resonance (CMR) time-signal intensity measurements collected in the area distal to a focal coronary artery stenosis (CAS) and (ii) to assess the incremental value of correcting this model by including similar measurements in remote areas. METHOD AND MATERIALS This retrospective study was approved by the hospital ethics committee. Forty-six patients (mean age 61±9 years; 33 males) who underwent both adenosine first-pass CMR and FFR in the work-up for a focal CAS (n=49) were included after written informed consent. Areas-at-risk (RISK) and remote adenosine/rest time-signal intensity parameters were evaluated. Boosting models were elaborated to predict the FFR value from (i)the whole (extended) and (ii)RISK-only parameters. The relationship between the predictions and FFR value was described with Bland-Altman and summarized with intra-class correlation(ICC). Diagnostic accuracies of the models predicting FFR<=0.80 were calculated. RESULTS The average FFR value was 0.84±0.09 (0.60-0.98 range), 15(31%) were <=0.80. Decreasing FFR was associated with opposite effects on myocardial time-signal intensity responses downstream of the CAS or remotely. Compared to the RISK-only models, the extended models exhibited higher correlations with the FFR value (0.73;95%CI,0.57-0.84 versus 0.25;95%CI,0.03-0.50) and diagnostic accuracy to predict FFR<=0.80 CAS [44/49 (90%; 95%CI,78-98) vs 36/49 (73%;95%CI,55-88)]. CONCLUSION When evaluating the functional significance of a CAS using adenosine first-pass CMR, considering time-signal intensity measurements in remote areas allows a better correlation with invasive FFR and improved diagnostic accuracy for FFR<=0.80. CLINICAL RELEVANCE/APPLICATION To improve the clinical relevance of evaluating coronary stenoses by using adenosine first-pass CMR, each perfusion-related parameter should be corrected by its value in remote segments. Clos