Myocardial blood flow determination from contrast-free magnetic resonance imaging quantification of coronary sinus flow

Background: Determination of myocardial blood flow (MBF) with MRI is usually performed with dynamic contrast enhanced imaging (MBFDCE). MBF can also be determined from coronary sinus blood flow (MBFCS), which has the advantage of being a noncontrast technique. However, comparative studies of MBFDCE and MBFCS in large cohorts are lacking. Purpose: To compare MBFCS and MBFDCE in a large cohort. Study Type: Prospective, sequence-comparison study. Population: 147 patients with type 2 diabetes mellitus (age: 56+/-12 years; 106 male; diabetes duration: 12.9+/-8.1 years), and 25 age-matched controls. Field Strength/Sequences: 1.5 Tesla scanner. Saturation recovery sequence for MBFDCE vs. phase-contrast gradient-echo pulse sequence (free-breathing) for MBFCS. Assessment: MBFDCE and MBFCS were determined at rest and during coronary dilatation achieved by administration of adenosine at 140 mu g/kg/min. Myocardial perfusion reserve (MPR) was calculated as the stress/rest ratio of MBF values. Coronary sinus flow was determined twice in the same imaging session for repeatability assessment. Statistical Tests: Agreement between MBFDCE and MBFCS was assessed with Bland and Altman's technique. Repeatability was determined from single-rater random intraclass and repeatability coefficients. Results: Rest and stress flows, including both MBFDCE and MBFCS values, ranged from 33 to 146 mL/min/100 g and 92 to 501 mL/min/100 g, respectively. Intraclass and repeatability coefficients for MBFCS were 0.95 (CI 0.90; 0.95) and 5 mL/ min/100 g. In Bland-Altman analysis, mean bias at rest was -1.1 mL/min/100 g (CI -3.1; 0.9) with limits of agreement of -27 and 24.8 mL/min/100 g. Mean bias at stress was 6.3 mL/min/100 g (CI -1.1; 14.1) with limits of agreement of -86.9 and 99.9. Mean bias of MPR was 0.11 (CI: -0.02; 0.23) with limits of agreement of - 1.43 and 1.64. Conclusion: MBF may be determined from coronary sinus blood flow, with acceptable bias, but relatively large limits of agreement, against the reference of MBFDCE.