Ferumoxytol dose optimization for three-dimensional whole-heart magnetic resonance imaging in patients with congenital heart disease.
BACKGROUND Ferumoxytol is used off-label as a contrast agent for cardiovascular magnetic resonance (CMR). However, the dose-response relationship between ferumoxytol, longitudinal relaxation, and image quality in pediatric and young patients remains unclear. OBJECTIVE To evaluate the effects of sequential ferumoxytol dose escalation on three-dimensional (3D) whole-heart (WH) image quality and blood-pool nulling inversion time (TI), a physiologic surrogate of post-contrast T1 shortening. MATERIALS AND METHODS In this prospective study, 45 patients with congenital heart disease underwent ferumoxytol-enhanced 3D WH CMR at 1.5 T using sequential dose regimens (1 mg/kg→2 mg/kg or 2 mg/kg→3 mg/kg). Right coronary artery (RCA) image quality and visible length, contrast-to-noise ratio (CNR), and image quality of the cardiac chambers and great vessels, as well as diagnostic completeness, were assessed. Blood-pool and myocardial nulling TI were measured using Look-Locker sequences. Baseline-adjusted change-score regression was the primary analysis, with mixed-effects ANCOVA performed to model TI behavior while accounting for repeated within-patient measurements. RESULTS No significant differences were observed in CNR, RCA length, image quality, or diagnostic completeness between all doses (all P>0.05). Myocardial and blood-pool nulling TI shortened significantly with dose escalation (P≤0.02). Change in blood-pool TI after the second dose was strongly predicted by TI after the first dose (β=-0.63, P<0.001), whereas starting dose (1 mg/kg or 2 mg/kg) was not associated with TI change (P=0.70). At 1 mg/kg, prolonged myocardial nulling TI may encroach upon the systolic rest period. CONCLUSIONS Ferumoxytol-enhanced 3D WH imaging provides diagnostic-quality assessment across 1-3 mg/kg dosing. Dose-dependent TI shortening follows a nonlinear, saturating pattern, supporting optimized low-dose protocols.