For this study 360 men and women, aged 18–60 years (mean age 31 ± 11 years, 37% women), were selected from a larger study of 480 healthy volunteers who were prospectively recruited over 1.5 years with the aim of acquiring normal cardiac MRI reference values in healthy athletes and non-athletes. The only selection criteria for the 360 subjects was that there was sufficient remaining scan time to perform the free-breathing 3D-MRCA additional to the standard cardiac anatomic and functional MRI performed in all 480 volunteers [9, 10].

Before participating in the larger study, all persons were interviewed, completed a questionnaire regarding training history, past medical and family history, and underwent resting electrocardiography and blood pressure measurement before cardiac MRI. Persons could only participate in the larger study if there was no history of illness, hypertension or medication. Persons with hypertension, unexpected findings on the electrocardiogram or cardiac MRI during the study, or contraindications for MRI, were excluded. Electrocardiograms were read by a cardiologist according to the European Society of Cardiology pre-participation guidelines [11]. No one who underwent the additional 3D-MRCA was excluded from this study.

There were 207 athletes (exercising at least 9 h per week), and 153 healthy non-athletes (exercising ≤3 h per week). The Institutional Ethics Committee of the University Medical Center Utrecht approved the study and all persons gave written informed consent.

All scans were performed on a 1.5-T MRI system (Achieva release 1.5, Philips Medical Systems, Best, The Netherlands). All images were acquired by the same operator (N. H. P) using vector-electrocardiogram triggering with a five-element phased-array cardiac coil for signal reception. The standard protocol, of which the results are described separately, included breathhold steady-state-free-precession cines of different views of both ventricles and flow measurements of the cardiac valves [12].

In addition, a free-breathing steady-state-free-precession 3D-MRCA balanced turbo-field-echo sequence containing the proximal course of the coronary arteries was planned on a coronal view, starting from around 2 cm above the aortic root to about 5 cm below the aortic root [13]. The 3D-MRCA was performed in cartesian acquisition mode in the transversal plane including a T2 preparation pulse and a fat suppression preparation pulse, repetition/echo time 5.4/2.6, flip angle 90°, rectangular field-of-view (FOV) 100%, phase resolution 100% and sensitivity encoding reduction of two in the anterior-posterior, and one in the feet-head direction. Turbo-field-echo factors were maximized individually to match the end-diastolic resting period obtained by visually identifying the time window of least RCA movement during diastole (resting period) on a four-chamber cine [8]. A respiratory navigator was used by placing the 5-mm 2D selective excitation pulse on the highest point of the liver on the right side of the diaphragm. To optimize scan-time while maintaining a reconstructed isotropic in-plane resolution of 0.55 mm², the FOV was minimized and matrix size was adjusted without introducing fold-over artifacts per scan. Typically the matrix was reconstructed to obtain a twofold higher resolution in two directions after reconstruction; a matrix of 256 × 256 was reconstructed to 512 × 512, so that the scanned in-plane spatial resolution of 1.1 mm² was reconstructed to an in-plane resolution of 0.55 mm². We planned the nominal scan duration under 4 min so that the effective scan duration would not exceed 10 min with a minimum navigator efficiency of 40%. Exact nominal and effective scan duration and navigator efficiency were not recorded. In this pragmatic approach, if the heart rate was less than 60 beats per minute (bpm), we chose the 139 slice package with 0.55 mm slice thickness. If the scan duration prognosis exceeded the 10 min maximum (including navigator efficiency below 40% or heart rate above 60 bpm), the slice thickness and the number of slices were adjusted from 139 slices of 0.55 mm (isotropic voxels of 0.55 mm³ after reconstruction) to 70 slices of 1 mm (anisotropic voxels), or if necessary to 35 slices of 2 mm (anisotropic voxels). No medication was given to lower the heart rate.

The 3D-MRCA dataset was transferred to a Vitrea workstation version 4.0 (Vital Images) for source image analysis, slab maximum-intensity-projection, and 3-dimensional reconstructions.

A radiologist experienced in cardiac MRI (B. K. V.), and blinded for participant characteristics, scored all 3D datasets on the source images. The overall image quality of the 3D-MRCA was visually graded for CAA; 0—indicating uninterpretable poor quality (technically unsatisfactory scan/ non diagnostic); 1—moderate (suboptimal visualization of all the major coronary arteries but assessment is still possible); 2—good visibility of all the proximal coronary arteries. The three coronary arteries were also assessed separately using above-mentioned grading system. The anatomic position of the three major coronary arteries and their proximal course in relation to the aortic root and the pulmonary trunk, the presence of proximal lumen reduction of more than 50%, and the presence of myocardial bridging were evaluated.

Final diagnosis was determined for positive findings in a consensus meeting with a cardiologist experienced in cardiac MRI (M. J. C.). Subjects with a CAA or proximal stenosis on 3D-MRCA were offered a low-dose (1–3 mSv) prospectively triggered multidetector CT angiography (MDCTA), and if necessary additional testing, for confirmation. Participants with myocardial bridging were not offered MDCTA as it is a common incidental finding on CT and only very rarely associated with sudden cardiac death in asymptomatic people [4, 6, 14]. The 3D-MRCA diagnosis was not systematically confirmed in all subjects as MDCTA was not mandatory because of the required radiation exposure in a perceived healthy population.