Magnetic resonance (MR) first-pass myocardial perfusion imaging during hyperaemia detects coronary artery stenoses in humans with test sensitivity depending on contrast medium (CM)-induced signal change in myocardium. In this prospective multi-centre study, the effect of CM dose on myocardial signal change and on diagnostic performance was evaluated using a stress-only approach. Ninety-four patients with known or suspected coronary artery disease (CAD) were randomised to 0.05,0.10, or 0.15 mmol/kg body weight of an extravascular CM (Gd-DTPA) and X-ray coronary angiography was performed within 30 days prior/after the MR examination. A multi-slice MR technique with identical hardware and software in all centres was used during hyperaemia (adenosine 0.14 mg/kg/min) to monitor myocardial CM wash-in kinetics and data were analysed semi-automatically in a core laboratory. Protocol violations resulted in 80 complete studies with CAD (defined as > or =1 vessel with diameter stenosis > or =50% on quantitative coronary angiography) present in 19/29, 13/24, and 20/27 patients for doses 1, 2, and 3, respectively. In normal myocardium, the upslope increased with CM dose (overall-p<0.0001, ANOVA). For CAD detection the area under the receiver operator characteristics curve for subendocardial data (3 slices with quality score<4 representing 86% of cases) was 0.91+/-0.07 and 0.86+/-0.08 for doses 2 and 3, respectively, and was lower for dose 1 (0.53+/-0.13, p<0.01 and p<0.02 vs. doses 2 and 3, respectively). Corresponding sensitivities/specificities (95% confidence intervals) for pooled doses 2/3 were 93% (77-99%; ns vs. dose 1) and 75% (48-92%;p<0.05 vs. dose 1), respectively. With increasing doses of CM, a higher signal response in the myocardium was achieved and consequently this stress-only protocol, with CM doses of 0.10-0.15 mmol/kg combined with a semi-automatic analysis, yielded a high diagnostic performance for the detection of CAD.
