Residual Myocardial Iron Following Intramyocardial Hemorrhage During the Convalescent Phase of Reperfused ST-Segment–Elevation Myocardial Infarction and Adverse Left Ventricular Remodeling

Study Population

Fifty STEMI patients reperfused by PPCI were prospectively recruited between August 2013 and July 2014. The main exclusion criteria were previous MI and standard recognized contraindications to CMR, such as estimated glomerular filtration rate <30 mL/min, ferromagnetic implants, and claustrophobia. The acute CMR scan was performed at 4±2 days post STEMI and a follow-up CMR scan at 5±2 months. The UK National Research Ethics Service approved this study, and all patients provided written informed consent.

CMR Acquisition

CMR was performed on a 1.5 Tesla scanner (Magnetom Avanto; Siemens Medical Solutions) using a 32-channel phased-array cardiac coil. The imaging protocol included balanced steady-state free precession cines, native T1, T2, and T2* mapping (Work In Progress 448b; Siemens Healthcare), and late gadolinium enhancement (LGE) LV short-axis images. For the acute scan, whole LV short-axis coverage for T1 and T2 maps and 3 (basal, mid, and apical) LV short-axis slices T2* maps were acquired. On the follow-up scan, only basal, mid, and apical short-axis slices were acquired for native T1, T2, and T2* maps.

Native T1 Mapping

Native T1 maps were acquired with a steady-state free precession–based Modified Look-Locker Inversion Recovery (MOLLI) sequence using a 5s(3s)3s modified sampling protocol.¹⁹ The acquisition parameters were flip angle =35°; pixel bandwidth =977 Hz/pixel; matrix =256×144; echo time =1.1 ms; and slice thickness =6 mm. A pixel-wise colored T1 map was generated after performing motion correction and a nonlinear least-square curve fitting of the set of images acquired at different inversion times.

T2 Mapping

Colored T2 maps consisting of pixel-wise T2 values were generated inline after motion correction and fitting to estimate T2 relaxation times²⁰ after acquiring 3 single-shot images at different T2 preparation times (0, 24, and 55 ms, respectively). The acquisition parameters were flip angle =70°; pixel bandwidth =930 Hz/pixel; matrix =116×192; echo time =1.1 ms; repetition time =3×R-R interval; and slice thickness =6 mm.

T2* Mapping

T2* maps were obtained using the following imaging parameters: bandwidth 814(×8) Hz/pixel; echo times×8: 2.7, 5, 7.3, 9.6, 11.9, 14.2, 16.5, and 18.8 ms; flip angle =18°; acquisition matrix =256×115; and slice thickness =8 mm. A colored pixel-wise T2* map was generated by the scanner.

Late Gadolinium Enhancement

LGE images were acquired 10 to 15 minutes after the injection of 0.1 mmol/kg of gadoterate meglumine (Gd-DOTA marketed as Dotarem; Guerbet S.A., Paris, France), using either a standard segmented fast low-angle shot 2-dimensional inversion-recovery gradient echo sequence (n=16) or a respiratory motion-corrected, free-breathing single-shot steady-state free precession averaged phase-sensitive inversion recovery sequence²¹ (n=24).

Figure 1 shows an example of the basal short axis images acquired during the acute scan and at follow-up in a patient with an inferior STEMI.

CMR Analysis

Imaging analysis was performed using CVI42 software (Version 5.1.2[303], Calgary, Canada).

Adverse LV remodeling was defined as a ≥20% increase in LV end-diastolic volume (EDV) between the acute and follow-up scans.²²

MI size was quantified in grams and as a percentage of the LV (%LV) using a signal-intensity threshold of 5 standard deviations (SD)²³ above the mean remote myocardium. The area at risk (AAR) was assessed from the T2 maps using a signal intensity threshold of 2SD above the mean remote myocardium and expressed as %LV. The presence of MVO (late MVO) was defined as areas of hypoenhancement on the LGE images and was quantified and expressed as %LV. Areas of MVO were included as part of the MI zone and AAR.

A hypointense core on the T2* maps with a T2* value of <20 ms was used to identify the presence of IMH on the acute scans and residual myocardial iron on the follow-up scans.^24–28 The extent of IMH and residual iron was quantified using manual region of interest (ROI) delineation around the hypointense core on the basal, mid, and apical slices and expressed as %LV.

For the follow-up T1 and T2 maps, a ROI was manually drawn in hyperenhanced area using a signal intensity threshold of 2SD from the remote myocardium (using the LGE images as reference) and was denoted as T1_Infarct and T2_Infarct, respectively.

A second ROI was copied from the hypointense core of the T2* map on to the T1 and T2 maps and denoted as T1_Core and T2_Core as illustrated in Figure 2.

In cases when there were no areas of hyperenhancement on the follow-up maps using a 2SD threshold, an ROI from the areas of LGE was copied on to the maps for representative values. An ROI was also drawn in the remote myocardium and expressed as T1_Remote and T2_Remote.

Statistical Analysis

Statistical analysis was performed using SPSS version 22 (IBM Corporation, IL). Continuous data were expressed as mean±SD or median (interquartile range), and categorical data were reported as frequencies and percentages. Normality was assessed using Shapiro–Wilk test. Independent groups (those with and without IMH; with and without paired CMR scans; with and without residual iron at follow-up; with and without adverse LV remodeling; without adverse LV remodeling but with and without residual iron) were compared with unpaired Student’s t test for normally distributed data and with Mann–Whitney U test for non-normally distributed data. Comparison between paired acute and follow-up scan was performed using paired Student’s t test for normally distributed data and Wilcoxon signed-rank test for non-normally distributed data. To compare T1 and T2 values in the remote, core, and infarct territories between patients and divided between those with and without residual iron, a linear mixed model was used, with the patients as a random factor and the territory within the patients (remote, core, and infarct) being a fixed factor. Comparison of 3 independent groups (no residual iron, residual iron only, and residual iron and adverse LV remodeling) was performed using Kruskal–Wallis Test and post hoc pairwise comparisons. Categorically variables such as incidence of MVO or IMH were compared using chi-squared test or Fisher exact test. Pearson rho correlation coefficient was assessed between MVO, acute MI size, IMH, and adverse LV remodeling as a continuous variable, respectively, and their correlations were compared. Receiver operating characteristic curve was also used to compare the diagnostic performance of MVO, acute MI size, and IMH to predict adverse LV remodeling. All statistical tests were 2-tailed, and P<0.05 was considered statistically significant.

IMH and Edema-Based AAR

Patients with IMH had lower T2_Core (50 [46–53] ms versus 55 [52-59] ms; P=0.001) but similar T2_Infarct (64 [62–64] ms versus 64 [62–69] ms; P=0.29) and T2 in the salvaged myocardium (65 [61–68] ms versus 63 [62–67] ms; P=0.43). However, in patients with IMH, edema-based AAR was larger (46 [40–55] %LV versus 31 [24–43] %LV; P=0.009) and myocardial salvage index was smaller (0.24 [0.16–0.43] versus 0.61 [0.36–0.88]; P=0.009), when compared with those without IMH.

Residual Myocardial Iron on the Follow-Up Scan in a Subset of Patients

A subset of 28 patients who completed the follow-up scans had matching and interpretable T2* maps. There was no difference in the CMR parameters between those with paired T2* maps (n=28) and those without (n=12; see Table in the Data Supplement). On the acute scan, 15/28 (54%) patients had IMH and 18/28 (64%) had MVO. Fifteen of 18 (83%) patients had MVO with IMH. On the follow-up scan, 13/15 (87%) patients had evidence of residual myocardial iron.

T1 and T2 of the Core, Infarct Zone, and Remote Myocardium at Follow-Up

Using a linear mixed model to adjust for within-patient interaction, in patients without residual myocardial iron, there was no difference between T1_Core and T1_Infarct (1042 [1015–1140] ms versus 1113 [1092–1140] ms; P=0.20), and T1_Infarct was significantly higher than T1_Remote (1113 [1092–1140] ms versus 1007 [966–1039] ms; P<0.001). However, T2_Core was lower than T2_Infarct (47 [45–49] ms versus 53 [51–55] ms; P=0.001) and similar to T2_Remote (47 [45–49] ms versus 47 [45–48] ms; P=1.0), and T2_Infarct was significantly higher than T2_Remote (53 [51–55] ms versus 47 [45–48] ms; P<0.001; Figure 3).

In patients with residual myocardial iron, both T1_Core and T2_Core were similar to T1_Remote and T2_Remote (T1: 1004 [90–1028] ms versus 1015 [989–1020] ms; P=1.0; T2: 47 [45–50] ms versus 49 [46–51] ms; P=1.0) but significantly lower than T1_Infarct and T2_Infarct (T1: 1004 [90–1028] ms versus 1162 [1132–1216] ms; P<0.001; T2: 47 [45–50] ms versus 60 [58–64] ms; P<0.001) respectively (Figure 3).

Patients with residual myocardial iron had lower T1_Core, similar T2_Core, and higher T1_Infarct and T2_Infarct when compared with patients without. There was no difference in the T1_Remote and T2_Remote between these 2 groups as illustrated in Figure 3.

Figure 4 shows an example of 3 patients with acute and follow-up CMR images showing MVO, IMH, residual myocardial iron, and persistently elevated T2 in the surrounding myocardium within the infarct zone at follow-up.

IMH and Adverse LV Remodeling

Eight out of 40 (20%) patients developed adverse LV remodeling. Patients with adverse LV remodeling had larger MI size (30.1±7.3 %LV versus 16.9±9.8 %LV; P=0.01), had an increased incidence (100% versus 50%; P=0.10) and extent of MVO (8.0±3.2 %LV versus 3.6±2.6 %LV; P=0.001), and were more likely to have IMH (100% versus 60%; P=0.04) on the acute scans when compared with those without adverse LV remodeling.

Residual Myocardial Iron and Adverse LV Remodeling

In patients with adverse LV remodeling, T2_Infarct was significantly higher than in those without adverse LV remodeling and without residual myocardial iron (60 [54–64] ms versus 53 [51–56] ms; P=0.025), but similar to those with residual myocardial iron but no adverse LV remodeling (60 [54–64] ms versus 60 [58–63] ms; P=1.0) as shown in Figure 5A.

However, when looking at those patients who did not develop adverse LV remodeling, those with residual iron had a significantly higher change in EDV (8% [−2% to 14%] versus −4% [−7% to 5%]; P=0.043) when compared with those patients without residual myocardial iron (Figure 5B).

Relationship Between Acute MI Size, MVO, and IMH and Adverse LV Remodeling

Acute MI size, extent of MVO, and IMH correlated with the change in LV EDV (Pearson’s rho of 0.64, 0.59, and 0.66, respectively; P values of 0.18 and 0.62, respectively, for correlation coefficient comparison) and performed equally well on receiver operating characteristic curve for predicting adverse LV remodeling (area under the curve: 0.99, 0.94, and 0.95, respectively, P=0.19 for receiver operating characteristic curve comparison).

Limitations

A significant proportion of the T2* maps were not interpretable in our study, and this partly highlights the challenge of performing a comprehensive CMR scan with multiparametric mapping in patients with an acute STEMI and may be partly because of the suboptimal per-slice rather than whole-heart shimming protocol we used. Of note, 18% of T2* data were not available in the large study by Carrick et al³⁴ either because of patients' intolerance of the scan or noninterpretable T2* maps. Although we quantified the extent of IMH in our study, we only used basal, mid, and apical short-axis T2* maps, and the quantification may have been more accurate if whole LV coverage T2* maps had been performed. Furthermore, only basal, mid, and apical short-axis T1 and T2 maps were acquired at follow-up. We used 2 different LGE sequences, and the difference in signal-to-noise ratio between the 2 may have affected the MVO size quantification. LGE images were acquired between 10 and 15 minutes, and this may have overestimated the MI and MVO size, and whether acquiring LGE images at 25 minutes³⁶ would have yielded different diagnostic performances for LGE, MVO, and IMH to predict adverse LV remodeling warrants further investigation. Receiver operating characteristic curve analysis was performed in 28 patients only, and therefore, we did not provide data on sensitivity/specificity and cutoff values because of small sample size. We did not measure blood inflammation markers to support the proposal of persistent inflammation because of residual myocardial iron in these patients. Histological validation is required to determine whether the elevated T2 values in the myocardium surrounding residual iron in the infarct zone was because of persistent inflammation. Our study was small, and we could not establish a causal relationship between IMH and adverse LV remodeling.

Conclusion

The majority of reperfused STEMI patients with IMH had residual myocardial iron at follow-up. This was associated with adverse LV remodeling and persistently elevated T2 values in the surrounding infarct tissue and may be suggestive of ongoing inflammation. Furthermore, IMH was as strongly associated with adverse LV remodeling as acute MI size and MVO. Therefore, patients with IMH or residual myocardial iron could be future potential targets for therapeutic interventions with anti-inflammatory agents or chelation therapy aiming to prevent adverse LV remodeling and improve clinical outcomes in these patients.