Inter-observer agreement and diagnostic accuracy of myocardial perfusion reserve quantification by cardiovascular magnetic resonance at 3 Tesla in comparison to quantitative coronary angiography

BACKGROUND Quantification of cardiovascular magnetic resonance (CMR) myocardial perfusion reserve (MPR) at 1.5 Tesla has been shown to correlate to invasive evaluation of coronary artery disease (CAD) and to yield good inter-observer agreement. However, little is known about quantitative adenosine-perfusion CMR at 3 Tesla and no data about inter-observer agreement is available. Aim of our study was to evaluate inter-observer agreement and to assess the diagnostic accuracy in comparison to quantitative coronary angiography (QCA). METHODS Fifty-three patients referred for coronary x-ray angiography were previously examined in a 3 Tesla whole-body scanner. Adenosine and rest perfusion CMR were acquired for the quantification of MPR in all segments. Two blinded and independent readers analyzed all images. QCA was performed in case of coronary stenosis. QCA data was used to assess diagnostic accuracy of the MPR measurements. RESULTS Inter-observer agreement was high for all myocardial perfusion territories (ρ = 0.92 for LAD, ρ = 0.93 for CX and RCA perfused segments). Compared to QCA receiver-operating characteristics yielded an area under the curve of 0.78 and 0.73 for RCA, 0.66 and 0.69 for LAD, and 0.52 and 0.53 for LCX perfused territories. CONCLUSIONS Inter-observer agreement of MPR quantification at 3 Tesla CMR is very high for all myocardial segments. Diagnostic accuracy in comparison to QCA yields good values for the RCA and LAD perfused territories, but moderate values for the posterior LCX perfused myocardial segments.