Myocardial Extracellular Volume Estimation by CMR Predicts Functional Recovery Following Acute MI

Objectives In the setting of reperfused acute myocardial infarction (AMI), the authors sought to compare prediction of contractile recovery by infarct extracellular volume (ECV), as measured by T1-mapping cardiac magnetic resonance (CMR), with late gadolinium enhancement (LGE) transmural extent. Background The transmural extent of myocardial infarction as assessed by LGE CMR is a strong predictor of functional recovery, but accuracy of the technique may be reduced in AMI. ECV mapping by CMR can provide a continuous measure associated with the severity of tissue damage within infarcted myocardium. Methods Thirty-nine patients underwent acute (day 2) and convalescent (3 months) CMR scans following AMI. Cine imaging, tissue tagging, T2-weighted imaging, modified Look-Locker inversion T1 mapping natively and 15 min post–gadolinium-contrast administration, and LGE imaging were performed. The ability of acute infarct ECV and acute transmural extent of LGE to predict convalescent wall motion, ejection fraction (EF), and strain were compared per-segment and per-patient. Results Per-segment, acute ECV and LGE transmural extent were associated with convalescent wall motion score (p < 0.01; p < 0.01, respectively). ECV had higher accuracy than LGE extent to predict improved wall motion (area under receiver-operating characteristics curve 0.77 vs. 0.66; p = 0.02). Infarct ECV ≤0.5 had sensitivity 81% and specificity 65% for prediction of improvement in segmental function; LGE transmural extent ≤0.5 had sensitivity 61% and specificity 71%. Per-patient, ECV and LGE correlated with convalescent wall motion score (r = 0.45; p < 0.01; r = 0.41; p = 0.02, respectively) and convalescent EF (p < 0.01; p = 0.04). ECV and LGE extent were not significantly correlated (r = 0.34; p = 0.07). In multivariable linear regression analysis, acute infarct ECV was independently associated with convalescent infarct strain and EF (p = 0.03; p = 0.04), whereas LGE was not (p = 0.29; p = 0.24). Conclusions Acute infarct ECV in reperfused AMI can complement LGE assessment as an additional predictor of regional and global LV functional recovery that is independent of transmural extent of infarction.

infarct extent in vivo (2). The convention of "bright is dead" appears robust in chronic infarction (3), but in AMI, factors such as myocardial edema and the effects of reperfusion therapy add complexity to infarct imaging, reducing the accuracy of LGE to predict recovery of regional wall motion (4)(5)(6). Processes within the infarct zone, such as microvascular obstruction (MO) or intramyocardial hemorrhage, impair functional recovery independent of infarct size (7,8), demonstrating that differing degrees of infarct suggesting that it may be sensitive to severity of tissue damage in myocardial infarction (9), but the method has not been evaluated in patients with AMI.
We hypothesized that ECV estimation by CMR in reperfused ST-segment elevation AMI offers additional predictive value for functional contractile recovery as compared to transmural extent of LGE hyperenhancement.

METHODS
Patients with first ST-segment elevation AMI, revascularized by primary percutaneous coronary intervention within 12 h of onset of pain, were prospectively recruited from a single tertiary center.
ST-segment elevation AMI was defined as per current guidelines (10). Exclusion criteria were previous AMI or coronary artery bypass grafting, estimated glomerular filtration rate <30 ml/min/1.73 m 2 , cardiomyopathy, or contraindications to CMR. The study protocol was approved by the institutional research ethics committee and complied with the Declaration of Helsinki; all patients gave written informed consent. Clinical management (including anticoagulation and use of aspiration catheters during primary percutaneous coronary intervention) was at the discretion of the responsible clinician, reflecting contemporary practice and guidelines, and performed blind to CMR results. All patients were considered for beta-blockade, angiotensin-converting enzyme inhibitors, statins, dual antiplatelet therapy, and cardiac rehabilitation. A venous blood sample for hematocrit was obtained at the start of each scan.  LGE signal intensity was recorded in arbitrary units as generated by cvi42 analysis software (Online Appendix).
Regional wall motion anomaly was graded persegment and per-patient from cine imaging by an experienced cardiologist (A.K., 4 years' CMR experience), blinded to the results of strain and LGE and scored as: 0 ¼ normal; 1 ¼ mild or moderate hypo- Myocardial ECV was calculated from native and post-contrast MOLLI images (14). For the per-segment analysis, each of 16 segments had mean ECV evaluated, blinded to the results of LGE and taking care to avoid partial-volume interaction with blood pool or pericardium. Regions of interest were manually motion corrected as required. For the per-patient analysis, T1 was calculated for infarcted and remote myocardium using a region of interest within the infarct and remote zone, with a conservative region of interest, and avoiding partial-volume effects from neighboring tissue, MO, or blood pool ( Figure 1).
The fit of the T1 curve was assessed; for both methods, regions with R 2 < 0.95 were rejected.
Mid-myocardial end-systolic circumferential strain was measured through the infarct and remote zones using tissue tagged imaging ( Figure 2).

RESULTS
PATIENT CHARACTERISTICS. Recruitment details are given in the Online Appendix. Thirty-nine   Figure 1). Transmural extent of LGE decreased  significantly between acute and convalescent visits ( Table 2). Adding acute ECV analysis to a 50% LGE transmural extent cutoff for prediction of wall motion improvement in dysfunctional segments increased sensitivity from 61% to 75% and specificity from 71% to 85%, and showed a trend toward improved prediction of convalescent wall motion score ( Figure 6) and functional recovery (Online Figure 2) across the range of LGE transmural extent.
PER PATIENT. Performance of acute ECV and acute LGE to predict markers of LV function per patient are shown in Table 3. ECV and LGE correlated with convalescent wall motion score (r ¼ 0.45; p < 0.01; r ¼ 0.41; p ¼ 0.02, respectively) and convalescent ejection fraction (EF) (r ¼ À0.56; p < 0.01; r ¼ À0.34;     other abbreviations as in Figure 1.  ECV-derived measurement of infarct severity also complemented LGE measurements of infarct extent, and increased the sensitivity and specificity to predict Groups are split by a cutoff value of 0.5 for acute ECV. A higher score indicates more severe segmental dysfunction. The left bar in each quartile represents patients with ECV <0.5; the right bar represents ECV >0.5. Abbreviations as in Figure 1. Abbreviations as in Table 2.
functional improvement in patients with >50% transmural LGE that was independent of LGE analysis. ECV had higher accuracy than LGE to predict improvement in regional wall motion score ( Figure 5), higher degree of correlation with infarct zone strain ( Figure 6), and reduced spread of values across the range of wall motion abnormalities (Figure 3).
LGE   Results of the present study were consistent with these previous reports and showed that higher transmural extent of scar was associated with impaired wall motion both acutely and at 90 days ( Figure 3) and with lower improvement in wall motion score over time ( Figure 4). However, these previous reports and our data also indicate that, whereas LGE can accurately predict functional outcome in AMI in areas of no LGE or with full transmural infarction, its accuracy is reduced in intermediate (25% to 75%) transmural infarct extent.
Peri-infarct edema and remodeling of the infarct zone over time may lead to comparatively high transmural extent acutely (4,16,19,20). Additionally, LGE cannot differentiate degrees of severity of tissue damage within the hyperenhanced infarct zone. By contrast, this study uses ECV, not to delineate spatial extent of infarction, but as a measure of infarct severity. Our data suggest that ECV can provide characterization in the diagnostic quandary of intermediate LGE extent, and adds an additional dimension to assessment of infarct transmurality by LGE. In histological studies, myocardial infarcts maintain foci of preserved myocytes within areas of necrosis (20), raising potential for functional recovery. Interstitial expansion in infarction is also variable and may depend on the extent of local reperfusion (19). These observations may, in part, explain contractile recovery found within the infarct zone (7,8), and the range of infarct ECV values observed in the present study.
ECV estimation by CMR is validated in models of chronic fibrosis rather than acute infarction (14).
Studies to date have mainly focused on native T1 in acute infarction, with similar findings to this study.
Messroghli et al. (21) found that native T1 maps were sensitive to acute infarction, but did not correlate this with functional recovery. Dall'Armellina et al. (22) found that native T1 was correlated with functional recovery on a segmental basis, and accuracy of T1 was Equilibrium-contrast ECV estimation may be more accurate for high ECV values than a bolus method (14); however, the bolus method used in this study can be more easily integrated into existing clinical protocols. The optimal threshold for hyperenhancement is debated (27,28); in the present study, we therefore deliberately evaluated the optimal threshold given our setup and pulse sequence.

CONCLUSIONS
This study demonstrates that addition of CMR- Kidambi et al.