Synopsis

Keywords: Heart, Cardiovascular

Contrast enhanced MRA of the aorta is typically performed without cardiac gating, greatly limiting its value in the aortic root and its more widespread utilization. We implemented a cardiac gated sequence for breath-held MRA of the thoracic aorta with flexible contrast during the steady-state distribution of ferumoxytol and compared it to ungated MRA with similar resolution. The resulting images were compared for definition of the aortic annulus, leaflets and ascending aorta. The gated sequence performed significantly better than the ungated sequence for all measured parameters and holds promise as a high quality alternative to gated CTA in appropriate clinical circumstances.

Introduction

First pass, contrast enhanced MR angiography (CEMRA) of the thorax is normally performed without cardiac gating (1). Commonly available techniques for breath-held, gated MRA are inefficient and inflexible, resulting in a long breath hold-relative to an ungated acquisition. Moreover, the complexity involved in matching the acquisition window both to the first pass of a contrast bolus and a specific phase of the cardiac cycle increases the risk of technical failure. However, with ungated MRA, the aortic root is usually blurred due to pulsation artifact, undermining diagnostic utility. For this reason, imaging of the thoracic aorta is most frequently performed with gated CT angiography or, less commonly, non-contrast MRA (2). During the steady state distribution of ferumoxytol, the T1 of the blood remains reliably short for several hours, obviating the need to time a contrast bolus (3). The procedural complexity of acquiring a gated study is thereby greatly diminished, as is the risk of technical failure, since the acquisition can be repeated. We hypothesized that, relative to ungated MRA, a flexible, 3D gated acquisition during the steady state distribution of ferumoxytol will reliably improve the quality of MRA in the aortic root and ascending aorta.

Methods

Twenty consecutive adult patients underwent MRA of the thorax following infusion of ferumoxytol (Feraheme, Covis Pharma), 4 mg/kg. Patients were studied as part of routine cardiac MR workup for suspected aortic pathology on a background of renal failure or congenital heart disease, under an institutional IRB. Ungated breath-held 3D MRA was acquired in the coronal plane, followed by gated breath-held 3D MRA. 15 patients were scanned at 1.5T (Siemens Avanto Fit) and 5 patients at 3.0T (Siemens Prisma Fit). Acquisition time for the ungated MRA was 17-19 seconds with voxel dimensions 1 x 1.2 x 1.3 mm and for the gated MRA 19-22 seconds with voxel dimensions 1 x 1.3 x 1.4 mm. For both sequences the k-space trajectory was Cartesian with TR/TE = 2.9/1.0 ms per line (bandwidth 650 Hz /pixel) and an asymmetric echo with partial Fourier in frequency. The k-space trajectory for the gated acquisition was flexible and segmented, so that within the same cardiac cycle a segment contained lines from both ky and kz (Figure 1) and the segments were offset so that the center of k-space was positioned in diastole. Elliptical filtering in ky-kz was added to the gated sequence to minimize the difference in breath hold duration between it and the ungated acquisition. The gating efficiently was approximately 70%, relative to ungated MRA, avoiding the most dynamic 30% of the cardiac cycle and weighting the central lines towards late diastole. Two experienced cardiovascular radiologists scored the images blindly for definition of the aortic annulus and valve leaflets (3 point scale; 3 = sharply defined) and for definition of the aortic root and walls of the ascending aorta (4 point scale; 4 = homogeneous high signal with no pulsation artifact). Differences between gated and ungated images were determined by using two-tailed t-test. Interobserver agreement was determined by using Gwet’s AC1 statistic, with the following grading: 0.00-0.20, poor; 0.21-0.40, fair; 0.41-0.60, moderate; 0.61-0.80, good; 0.81-1.00, very good.

Results

All patients underwent FE-MRA without any adverse events and there were no technical failures in the study group. There was good interobserver agreement (AC1=0.76). Gated images were consistently scored higher than ungated images for annulus (p<0.0001, 2.57 ± 0.50 vs 1.57 ± 0.50) leaflet (p<0.0001, 2.40 ± 0.59 vs 1.32 ± 0.52) definition and proximal ascending aortic quality (p<0.0001, 3.57 ± 0.50 vs 2.32 ± 0.76) (Figure 2 b,c,d). Ungated images routinely suffered from some degree of motion degradation of the aortic root (Figure 2a). Pulsation artifact in the ascending aorta was most marked on ungated images in patients with good cardiac contractility.

Conclusion

Breath-held, cardiac gated MRA of the thoracic aorta is highly successful and practical when performed during the steady state distribution of ferumoxytol, such that it may serve as an alternative to CTA in appropriate patient groups. Flexible k-space ordering allows for gating using either ECG or pulse triggering. Further advances in image acceleration techniques hold promise for even higher performance and resolution.

Acknowledgements

No acknowledgement found.