Acute intra-cavity 4D flow cardiovascular magnetic resonance predicts long-term adverse remodelling following ST-elevation myocardial infarction

Fifty prospectively recruited STEMI patients reperfused by primary percutaneous coronary intervention (PCI) within 12 h of symptoms onset, underwent serial CMR scans at 5 ± 2 days and 378 ± 23 days following their index presentation. Study inclusion criteria were (a) MI as defined by current international guidelines [5], (b) revascularisation via primary PCI within 12 h after onset of symptoms and (c) no contraindications to CMR. Exclusion criteria were (a) previous revascularisation procedure (coronary artery bypass grafts or PCI), (b) known cardiomyopathy, (c) severe valvular heart disease, (d) atrial fibrillation and (e) haemodynamic instability lasting longer than 24 h following PCI and contraindication. The study protocol was approved by the institutional research ethics committee and complied with the Declaration of Helsinki. (NIHR 33963, REC 17/YH/0062).

The study protocol included a CMR scan within 3–7 days of index presentation (acute scan), a second scan at 12 months. CMR examinations were performed on a 3T CMR system (Achieva TX, Philips Healthcare, Best, The Netherlands) equipped with a 32-channel cardiac phased array receiver coil, MultiTransmit technology and high-performance gradients with Gmax = 80mT/m and slew rate = 100 mT/m/ms. Survey images were used to plan vertical long-axis, horizontal long-axis, 3-chamber (LV outflow tract) views and the LV volume contiguous short axis stack. Cine imaging used a balanced steady-state free precession (bSSFP) pulse sequence (echo time (TE)/repetition time (TR)/flip angle 1.3 ms/2.6 ms/40°, spatial resolution 1.6 × 2.0 × 10 mm, typical temporal resolution 25 ms, slice thickness 8 mm, and 30 phases per cardiac cycle). Modified Look-Locker inversion recovery (MOLLI) to determine the T1-inversion time. LGE imaging was done at 15-min from gadolinium-based contrast injection, using phase sensitive inversion recovery (PSIR) spoiled gradient echo (GE) sequence (SENSE factor 1.7, typical TE/TR of 3.0/6.1 ms, flip angle of 25°, slice thickness of 10 mm and with Look-Locker scout determined T1-inversion time).

An unnavigated free-breathing 4D flow data acquisition was planned in the trans-axial plane while ensuring complete ventricle coverage. A 3D echo planar imaging (EPI)-based, fast field echo (FFE) sequence was used with retrospective cardiac gating; 30 phases were reconstructed across the cardiac cycle. Sequence parameters were as follows: acquired voxel size = 3 × 3 × 3 mm³, reconstructed voxel size = 2.23 × 2.23 × 3 mm³, field of view (FOV) = 400 × 300 mm², TR = 8.1 ms, TE = 3.5 ms, flip angle = 10°, number of signal averages = 1, VENC = 150 cm/s, EPI factor = 5. 4D flow data reconstruction, error and quality check methods were done as from previously published literature [3].

Cine and LGE data were analyzed using cvi⁴² software (Circle Cardiovascular Imaging Inc, Calgary, Canada). Cine-images were used to derive LV volumes and LVEF, while LGE images were used to derive infarct size and identify microvascular obstruction (MVO). On LGE images, the threshold used for identifying infarcted tissue was set to 5 standard deviations above remote myocardial tissue signal intensity. MVO was defined as dark zones within an area of LGE at 15 min. Adverse remodelling was defined as an increase in LV end-diastolic volume (LVEDV) indexed for body surface area (LVEDVI) > 20% at 12 months from baseline [6]. 4D flow data was analyzed using the research software tool MASS (Leiden University Medical Center, Leiden, The Netherlands). Cine short-axis segmentation was used to define the boundaries of the region for LV blood flow parameter estimation. Prior to these calculations, spatial misalignment between the cine short-axis stack and 4D flow CMR data were corrected by rigid registration as previously described [3]. 4D flow measurements are described in Table 1.

Statistical analyses were performed in SPSS (version 21.0; Statistical Package for the Social Sciences, International Business Machines, Inc., Armonk, New York, USA). Normality was checked using the Shapiro–Wilk test. Continuous variables are reported as mean ± SD. Comparison between quantitative variables was performed by independent-sample parametric (unpaired Student’s t-test) or non-parametric (Mann–Whitney) statistical test as appropriate. For comparing results from initial and repeated measurements, paired t-tests and ANOVA with Bonferroni post-hoc comparisons were used. Pearson correlation analysis was used to calculate the correlation coefficient between LVEF, infarct size and 4D flow parameters. All tests were assumed to be statistically significant when p < 0.05.

The study flowchart is shown in Fig. 1. Of the 54 patients prospectively recruited for the study, 3 patients experienced claustrophobia during the acute CMR scan, and 1 patient did not attend the follow-up CMR scan. The acute and 12-month CMR scans from the remaining 50 patients were used for statistical analysis. Patient demographics are displayed in Table 2. All patients received 12 months of dual antiplatelet and angiotensin converting enzyme inhibitor therapy, and all but 1 were on beta-blockers at the time of their 12-month scan. There was no significant difference in pre-scan systolic blood pressure between adverse and non-adverse remodellers (140 ± 34 vs 138 ± 28 mmHg, p = 0.52).

The CMR characteristics are summarised in Table 3. In the acute scan, the mean LVEDV was 152 ± 41mls, with a mean LVEF was 43 ± 9% and mean infarct size of 14 ± 11 g. By 12 months, across the entire cohort, the mean LVEDV increased to 164 ± 52mls and 12 out of the 50 patients fulfilled the criteria for adverse remodelling.

On the acute scan, average systolic KE indexed for end-diastolic volume (KE_iEDV) across the cohort was 10.8 ± 3.9 μJ/ml which is higher than previously reported values in acute MI patients by our group (9.2 ± 3.8 µJ/ml [3], but average systolic KE_iEDV was noted to decrease with worsening LV systolic function (ANOVA p = 0.01, Table 4).

Across the cohort, systolic in-plane KE % decreased with worse LV systolic function (ANOVA p < 0.01, Table 4 and Fig. 2a). Systolic in-plane KE also correlated with infarct size (p < 0.01, Fig. 2b). There was a significant correlation between in-plane KE in the acute scan and change in LVEDV over 12 months (p < 0.01, Fig. 2c). Patients who remodelled by 12 months (Table 5) had significantly higher systolic in-plane KE (54.0 ± 12.2% vs 42.5 ± 10.0%, p = 0.01 as shown on Fig. 2d), and lower average systolic KE (8.5 ± 3.1 μJ/ml vs 11.6 ± 3.7 μJ/ml, p = 0.01) on their acute scan than patients who did not undergo adverse remodelling. Figure 3 demonstrates time-curves of 2 separate subjects who both suffered an inferior STEMI with similar initial LVEF; patient A underwent adverse remodelling at 12 months while patient B does not. In comparison to patient B, on the acute scan, patient A has less through-plane and more in-plane KE during systole. There were no significant differences in acute diastolic in-plane KE between adverse and non-adverse remodellers (10 ± 5 vs 8 ± 4 μJ/ml, p = 0.22).

When comparing the proportion of blood volume entering and leaving the LV cavity across 2 cardiac cycles, patients with more severe LV dysfunction tended to have less direct flow (p = 0.86), significantly higher proportion of delayed ejection flow (p = 0.003) and residual volume (p = 0.012), as shown in Table 4. Patients who had adverse remodelling at 12-months had significantly reduced direct flow (11 ± 4% vs 27 ± 9%, p < 0.01), increased delayed ejection flow (22 ± 9% vs 12 ± 2%, p < 0.01) and increased residual volume (64 ± 14% vs 34 ± 12%, p < 0.01) across 2 cardiac cycles in their acute 4D flow CMR scan, as shown in Table 5 and Fig. 4.

As contractility is lost in infarct segments, LV contraction becomes asymmetrical, meaning that wall tension is no longer homogeneously distributed in the cavity. Previous authors have attributed this to be the cause of greater ‘in-plane’ flow KE across the cavity [3]. This asymmetrical contraction is thought to exert heterogenous haemodynamic forces on the LV wall, which can cause stretching of the LV wall and lead to cavity dilatations over time [3, 4, 7, 8]. Garg et al. highlighted that larger infarct size correlated with greater in-plane KE following MI [3]. Results from our study matched this pattern, and in addition, demonstrates a direct association between in-plane KE and increase in LVEDV over 12-months, providing a link between infarct size, interventricular flow and long-term adverse remodelling. In-plane KE therefore provides an additional measurement of mechanistic function during systole beyond LVEF alone, which can be useful in predicting adverse remodelling. It is worth highlighting however that in-plane flow is likely one of several factors exerting pressure on the LV wall, and other contributing factors such as systemic blood pressure and LV end-diastolic pressures were not formally assessed in this study. In addition, distension of the LV wall during diastole is also likely to impact on cavity stretching and subsequent remodelling, however our results did not detect a significant difference in diastolic in-plane KE between adverse and non-adverse remodellers.

In a previous study, Stoll et al. performed 4D-flow CMR in heart failure patients (dilated cardiomyopathy and ischaemic heart disease) and demonstrated them to have decreased direct flow and increased residual volume following across 2 cardiac cycles than controls [2]. The degree of derangement in KE parameters correlated with myocardial dilatations and brain natriuretic peptide levels—neurohormones which are released in response to stretching of the cavity walls. Garg et al. also demonstrated how increased residual volume was predictive of the formation of LV thrombus, as a consequence of reduced diastolic LV wash-out [4]. Our results add to this finding by showing that decreased direct flow, and consequently increased residual volume was associated with long-term adverse remodelling. Like previous authors, we hypothesise that the increased stress placed on the LV cavity from the increased residual volume over time leads to stretching of the LV wall, with subsequent increase in cavity size [2, 4].

The impact of ischaemic injury on myocardial strain has been explored previously. Echocardiography based studies using echo-particle image velocimetry analysis and speckle-tracking, have shown that alterations in energy dissipation index and KE fluctuation index can be used to explain impairments in both LVEF and global wall motion indices following STEMI [9]. CMR strain imaging, which provides superior spatial resolution to speckle-tracking, have found that circumferential strain can be used to predict the recovery of long-term LV function, however associations between strain parameters and adverse LV remodelling remain unclear [10]. The impact of intraventricular flow and reduced LV-wash out on global and regional strain parameters has not yet been explored and may provide further mechanistic insights into the pathophysiology of adverse remodelling following MI.

Recruiting participants after STEMI for complex acute and longitudinal imaging was challenging, and the study sample size was therefore relatively small but aligned with similar studies [3]. The temporal resolution of the 4D flow CMR was 40 ms, which may affect the quality of KE and TD assessment. The LV geometry was defined by LV cine stack which was done using breath-hold technique while the 4D flow was done using free breathing. Hence, although spatial miss-registration was corrected for, other issues still remain including difference in heart rate and physiological conditions. This may have impact on the time-varying flow characteristics which could not be corrected for. Results from this study cannot be applied to patients with significant valvulopathy, cardiomyopathies and congenital heart disease.