The focus of this study was to apply 4D flow MRI to compare regional and global LV flow between anterior AMI patients and controls. To do so, through-plane flows in the LV base, mid-ventricle, and apex, as well as global KEi
EDV and the fraction of LV flow in each of the 4 flow compartments, were measured in 12 patients and 19 controls. AMI patients had reduced through-plane flow in the apex at peak systole and diastole, and in the mid-ventricle at peak systole compared to controls, even after correcting for age, sex, and heart rate differences between the groups. While the other flow measurements were also lower in patients than in controls, the differences were not significant after correcting for age, sex, and heart rate differences. This finding supports the notion of stasis in the LV apex of anterior AMI subjects contributing to the elevated rates of LVT experienced by this group. The implication of this finding is that 4D flow MRI of the LV apex may be a valuable tool in larger studies on post-AMI LVT risk assessment. While the more traditional hemodynamic metric, LVEF, was also lower in patients than in controls, this measure does not offer insights into the location of flow impairment. This study did not find the reduction in KEi
EDV that Garg et al. found in a cohort of 48 MI subjects [
10]. This is likely caused by the smaller sample size of this study and the use of different scanners for different subjects, which we elaborate on in the limitations paragraph below. Flow compartment analysis revealed a marked shift away from blood entering and leaving the LV in one heartbeat (direct flow) towards blood starting and residing in the LV for > = 2 heartbeats (residual volume) in anterior AMI patients. This increase in residual volume is consistent with the decreased through-plane flow in the apex (since residual flow pathlines are typically located in or near the apex) and with the notion of increased stasis post MI. Our control data have similar flow rates to those reported in 1995 [
21] and similar KE values to those reported in 2015 [
22] and 2016 [
23] in healthy controls. Our control data however displays a greater fraction of flow in the direct flow compartment and less in the residual volume compartment than Eriksson et al. found in 2010 [
19] and 2013 [
24]. We attribute this to differences in acquisition such as k-space trajectory (radial vs. cartesian), spatial resolution (2.5 × 2.5 × 2.5mm
3 vs. 3x3x3mm
3), number of cardiac frames (20 vs. 40), and scan duration (9–14 min. vs. 16–57 min.), although it is unclear how this difference led to the observed differences in flow compartment distribution. Despite the fact that our baseline LV flow compartment distribution in controls differs from that in the literature, our finding of flow compartment shifts in AMI patients relative to controls is valid since all of the 4D flow data presented herein was acquired with the same sequence.
Peak systolic flow in the LV base was negatively correlated with heart rate and circumferential strain and positively correlated with stroke volume and ejection fraction. The negative correlation of peak flow with heart rate may result from compensatory heart rate increases in order to maintain cardiac output in the infarcted individuals with reduced intraventricular flow. The negative correlation with circumferential strain means that individuals with the largest magnitude of strain had the highest flow in the LV base (because circumferential strain is negative). Ventricles that deformed more where able to push more blood through at peak systole. This correlation may not have been significant in other LV regions due to sample size limitations. The positive correlations of peak systolic intraventricular flows with SV and EF are expected since SV is the total flow over one cardiac cycle at the aortic valve, and EF is directly related to SV. The lack of significant correlation between the LV flow metrics and infarct size could be attributed to the small sample size of the MI group.
Limiations
One limitation of this pilot study is the relatively low number of subjects in each cohort. Additionally, 4D flow data in patients was collected as an adjunct to clinical CMR exams. Accordingly, multiple scanners (1.5 T and 3 T) were used depending on clinical availability. As a result, imaging parameters and image quality slightly varied between patients, and quality might have been compromised in comparison to controls (all imaged at 3 T). The use of 1.5 T scanners for some AMI patients limited sensitivity but did not create a systematic bias in AMI patient flow versus controls, as data from Lotz et al. show that the use of 1.5 T versus 3 T scanners reduces flow measurement precision but not accuracy [
25]. Likewise, the difference in contrast agent used between the two groups may have created different blood T1 constants and therefore velocity-to-noise ratios in the two groups but would not create a bias between the through-plane flow measurements in each group. The use of different scanners and contrast agents may have affected our KE measurements, however, as erroneously measured high-velocity voxels can dominate the KE sum since it is derived from velocity squared. Moreover, the velocity encoding parameter selection at 100–150 cm/s may have been higher than optimal for detecting differences in low velocities (stasis), such as those present in the apex. However, a lower velocity encoding parameter would have resulted in aliased and incorrect flow measurements in the LV base and mid-ventricle. Further studies investigating and validating the optimal 4D flow acquisition settings in the context of AMI are warranted. While the relatively low sample size, increased noise in patient data, and perhaps suboptimal VENC setting may have reduced the power of this study, this study did shed light on which regions of the intra-LV flow field are most affected after anterior AMI. Another limitation of this study is the comparison of the prospectively acquired AMI cohort with the control cohort from a previous study. As such, there were differences between the AMI and control groups such as age, sex and heart rate. While our model corrected for differences in age, sex, and heart rate between the groups, we acknowledge that a prospectively matched study would provide stronger evidence of post-AMI flow reductions.