Fractional myocardial blood volume by ferumoxytol-enhanced MRI: Estimation of ischemic burden

Purpose To investigate model-fitted fractional myocardial blood volume (fMBV) derived from ferumoxytol-enhanced MRI as a measure of myocardial tissue hypoperfusion at rest. Methods We artificially induced moderate to severe focal coronary stenosis in the left anterior descending artery of 19 swine by percutaneous delivery of a 3D-printed coronary implant. Using the MOLLI pulse sequence, we acquired T-1 maps at 3 T after multiple incremental ferumoxytol doses (0.0-4.0 mg/kg). We computed pixel-wise fMBV using a multi-compartmental modeling approach in 19 ischemic swine and 4 healthy swine. Results Ischemic myocardial segments showed a mean MRI-fMBV of 11.72 +/- 3.00%, compared with 8.23 +/- 2.12% in remote segments and 8.38 +/- 2.23% in normal segments. Ischemic segments showed a restricted transvascular water-exchange rate (k(i) = 15.32 +/- 8.69 s(-1)) relative to remote segments (k(i) = 17.78 [11.60, 26.36] s(-1)). A mixed-effects model found significant difference in fMBV (p = 0.002) and water-exchange rate (p < 0.001) between ischemic and remote myocardial regions after adjusting for biological sex and slice location. Analysis of fMBV as a predictor of impaired myocardial contractility using receiver operating characteristics showed an area under the curve of 0.89 (95% confidence interval [CI] 0.80, 0.95). An MRI-fMBV threshold of 9.60% has a specificity of 90.0% (95% CI 76.3, 97.2) and a sensitivity of 72.5% (95% CI 56.1, 83.4) for prediction of impaired myocardial contractility. Conclusions Model-fitted fMBV derived from ferumoxytol-enhanced MRI can distinguish regions of ischemia from remote myocardium in a swine model of myocardial hypoperfusion.