Using a recently developed in vivo dwCMR technique, we have demonstrated in a large animal model of chronic MI that ADC was significantly increased in infarcted regions compared to remote myocardium. A chronic MI porcine model was chosen because previous ex vivo dwCMR studies established that the ADC increase found in chronic infarcts was due to the underlying tissue microstructural presence of myocardial fibrosis. Additionally, the presence of edema would be absent leaving only bulk motion as the major confounding source of an observed in vivo ADC increase. Infarct quantification utilized two common semi-automatic approaches, threshold and FWHM, to avoid potential inter-observer bias and provide a more quantitative infarct characterization. RWM analysis was performed and compared to ADC maps to determine if any observed ADC change was significantly influenced by bulk motion.
ADC estimation of infarct volume and location demonstrated substantial agreement and correspondence with LGE. These findings were highly reproducible between two blinded reviewers and repeated analysis of a single reviewer. In yielding the best agreement between ADC and LGE, FWHM outperformed the threshold method and yielded better of inter- and intra-observer reproducibility. In the infarcted regions identified by LGE, ADC was significantly increased while RWM was significantly decreased when compared with remote myocardium. In the regions with elevated ADC, the tissue’s bulk motion assessed by cine imaging was significantly reduced compared with remote regions. Consequently, if residual bulk-motion weighting were to exist and dominated the overall change in ADC, then the expected ADC would have instead been decreased relative to remote regions as opposed to the observed increase. Furthermore, high intraclass correlation and agreement in estimating infarct volume when compared to LGE suggests that bulk motion has little effect on the observed increase in ADC. A more rigorous follow-up experiment that includes applying dwCMR immediately before and after the myocardium is arrested will be needed to conclusively demonstrate that an in vivo dwCMR technique is completely free of bulk motion. Such an experiment at the time of our study was logistically infeasible. Additionally, T2 weighting visualized by the least diffusion-weighted measurement (b0) depicted isointensity between the infarct and remote regions. Therefore, the results strongly suggest that the observed increase in ADC found in infarct regions was a reflection of underlying tissue microstructure change as opposed to the presence of bulk motion or edema .
The significant increase in ADC (~70%) found in the study is consistent with the significant increases found in previous ex vivo studies that used similar large animal models (50-80%) ,. However, the absolute infarct (2.3 μm2/ms) and remote (1.4 μm2/ms) myocardial ADC values are larger than what was presented ex vivo (Pop, et al.: 1.1 μm2/ms infarct vs 0.62 μm2/ms remote, Wu, et al.: 1.0 μm2/ms infarct vs 0.67 μm2/ms remote). This disparity could be attributed to the use of formalin, dehydration, and differences in temperature, which are known to perturb absolute ADC values in ex vivo conditions . In a recent in vivo chronic MI patient study , infarct ADC was also significantly increased, but did not exhibit the same large relative increase (~10%) as the work presented. This is most likely due to the variety in presentation of chronic infarction in the patients recruited, in which some did not receive any percutaneous intervention while others did. This is in comparison to the 150 min LAD occlusion and reperfusion of the animal model used in this study. A follow-up study using the diffusion-prepared TSE technique in a similar patient demographic needs to be conducted to confirm that the large relative increase in ADC found in this study is present in chronic MI patients as well.
Despite the potential for ADC to quantify the degree of fibrosis shown by Pop, et al., the in vivo quantification of ADC in this study cannot precisely predict the severity of fibrosis because the degree of bulk motion weighting was not fully quantified. Cine imaging used in this study can only qualitatively yield the radial displacement of the myocardium as opposed to more sophisticated myocardial strain or phase contrast tissue mapping , which can quantitatively measure bulk motion entirely. Therefore, the observed elevated ADC can currently only detect the presence of fibrosis similar to LGE imaging. A future study using myocardial strain or phase contrast tissue mapping to discern the exact amount of bulk motion weighting affecting the proposed in vivo dwCMR technique should be conducted in combination with the aforementioned arrested myocardium experiment.
Limitations of the diffusion-prepared TSE technique used in this study include potentially long scan times and low spatial coverage. Although the TSE readout generally yields less image artifacts than balanced steady-state free precession at 3 T, it unfortunately has longer echo spacing leading to an overall two to three times increase in scan time. Therefore to keep the imaging time to a manageable 10 minutes, the spatial coverage was reduced to three slices. Consequently, the infarct volume estimated in this study was not calculated from the whole LV potentially yielding a deficient estimate of the true infarct volume. However, the focus of this study was on the comparison of ADC with LGE and despite the limited spatial coverage, the two yielded comparable estimates in infarct volume. Furthermore since the diffusion-preparation approach is inherently multi-shot, both scan time and spatial coverage limitations could be overcome with a 3D whole-heart radial acquisition with parallel imaging and iterative reconstruction . Potentially, this technique could markedly reduce the scan time while providing whole-heart coverage.
Another limitation of this study was that the animal model used only yielded transmural infarcts at the LAD territory. Consequently, the ability to detect subendocardial infarction with dwCMR was not tested, which is often of clinical importance in viability imaging of chronic MI. Non-transmural infarcts are generally much smaller in size, which could be more difficult to detect with the lower spatial resolution of dwCMR. Additionally, infarction was only induced in the LAD territory, which limits the study in testing the performance of the new technique to locate infarction in other arterial territories. Further studies will be needed to specifically investigate the ability of dwCMR ability to detect non-transmural infarction at different arterial territories.