Dynamic Fractional Myocardial Blood Volume Mapping Using Ferumoxytol-Enhanced MR Multitasking

Background: Fractional myocardial blood volume (fMBV) derived from multi-dose ferumoxytol-enhanced imaging has emerged as a promising biomarker of myocardial perfusion. The current approach uses multi-dose, breath-held T1 MOLLI, which requires image registration across multiple inversion times and ferumoxytol doses, and cannot estimate fMBV dynamically^1,2. Cardiac MR Multitasking³ enables motion-resolved simultaneous acquisition of cardiac, respiratory, and T1-dynamic information without ECG triggering or breath-holds, providing dynamic cardiac phase-resolved fMBV and potentially improving the spatial reliability of the fMBV quantification for perfusion assessment without needing stress pharmacologic agents. We aim to show the potential of ferumoxytol-enhanced MR Multitasking for dynamic multiphase fMBV estimates.

Methods: Eighteen healthy volunteers (15 females, 3 males, 28.2±8.4 years) and 8 patients with ischemic heart disease (IHD) (males, 67.4±8.1 years) underwent multi-dose ferumoxytol imaging (0, 0.125, 2.0, 4.0mg/kg) using both 5(3)3(3)3 MOLLI and a free breathing 2D Multitasking T1 mapping sequence. For MOLLI, non-rigid registration was applied across inversion times and doses prior to the T1 fitting using the InSil algorithm⁴. Multitasking T1 maps were fit for 30 cardiac bins at each of the 4 doses, followed by non-rigid registration of the 4 derived T1 maps. fMBV maps were computed using the model from Colbert et al.1, fitting voxel-wise changes in R1 (1/T1) versus blood pool R1 to estimate fMBV. For comparison, the MOLLI and Multitasking fMBV maps for each subject were co-registered; endocardial and epicardial contours were drawn on diastolic mid-ventricular short-axis slices. Multitasking dynamic fMBV maps were computed across 30 cardiac bins and the mean fMBV values were assessed. Mean end-diastolic fMBV values were compared using Bland-Altman plots and paired t-tests.

Results: Multitasking produced motion-resolved fMBV maps. Mean diastolic fMBV values were 10.2% (MOLLI) and 10.3% (Multitasking). The mean difference (bias) of -0.15% was not significant (p=0.77), and 95% limits of agreement ranged from -4.48% to 4.17% (Fig 1).In healthy volunteers, changes in fMBV across different cardiac phases were consistent with healthy physiologic perfusion values^5-7 (Fig 2). Fig 3 illustrates fMBV values in a patient with IHD and end-stage renal disease. Our proposed method identified scar tissue in the inferior myocardial segment, where reduced and blunted fMBV curves corresponded with the fixed perfusion defect on stress cardiac PET and with reduced circumferential strain.

Conclusion: Ferumoxytol-enhanced Multitasking enables free-breathing, motion-resolved T1 mapping for fMBV estimation. In the diastolic phase, results agree with the current MOLLI-derived fMBV approach. By intrinsically resolving cardiac and respiratory motion, Multitasking simplifies the complex inter-dose registration and provides dynamic fMBV mapping as a quantitative biomarker of myocardial perfusion.

Bland-Altman plot comparing fMBV (%) from MOLLI and Multitasking T1 mapping. The solid blue line shows the mean difference (bias = −0.154 %), and dashed blue lines indicate the 95% limits of agreement (−4.478 % to 4.170 %). Mean fMBV values were similar between MOLLI (10.167 %) and Multitasking (10.321 %), with no significant difference (paired t-test, p = 0.771). A comparison of MOLLI and Multitasking fMBV maps for one healthy subject is also shown.
Dynamic fMBV values for one healthy volunteer. The curve shows an example of fMBV values across the cardiac cycle for one healthy volunteer. Cardiac bins are shown as 2x concatenation along cardiac phases to improve user-readability. The lowest fMBV value is measured in systole, whereas the highest value occurred in diastole. fMBV maps are also shown for representative cardiac phases.
Dynamic fMBV for one patient with ischemic heart disease and end-stage renal disease where administration of gadolinium contrast agents for scar imaging was prohibitive. (a) Dynamic fMBV values across the cardiac cycle and (b) cardiac phase-specific fMBV maps for one IHD patient with likely scar tissue (white arrows) in the inferior segment confirmed by Rb-82 stress cardiac PET (c). The dynamic behavior of fMBV values in the inferior segment (purple curve) is blunted and mean fMBV values are reduced relative to remote segments. Cardiac phase-specific fMBV maps for the diastolic and systolic phases are shown. Cardiac bins are shown 2x concatenated to improve user-readability. (d) Circumferential strain is severely reduced in the basal and mid-inferior segments, consistent with regional wall motion abnormalities.