Strain Measures Predict Outcome after ST-Segment–Elevation Myocardial Infarction: Now What?

See also the article by Mangion et al in this issue.

Introduction

Studies of myocardial strain after ST-segment–elevation myocardial infarction (STEMI) measured by using cardiac MRI were initially performed in the 1990s, first in animal models (1) and then in patients (2). These were descriptive studies using myocardial tissue tagging, demonstrating dysfunction in adjacent noninfarcted regions that slowly recovered over several weeks (1) and mild acute dysfunction in remote regions (2) that recovered more quickly. Recent T1 mapping measures in patients after myocardial infarction (MI) demonstrate increased T1 in these noninfarcted areas, likely representing inflammation and/or edema as the etiology of dysfunction outside of the infarct zone (3). The application of contrast material–enhanced cardiac MRI and the development of late gadolinium enhancement (4) enabled identification of infarct size as well as microvascular obstruction as important markers of adverse outcome after MI (5). Because many of these markers (strain, infarct size, microvascular obstruction) track together in patients after MI, it has been difficult to tease out which is the most predictive of adverse outcomes.

Larger outcome studies in STEMI are more recent, enabled by the increasing number of patients undergoing MRI after STEMI. In fact, elevated T1 in noninfarcted regions is associated with increased major adverse cardiac events after MI as previously demonstrated by the authors of the present study in the same patient population (3). In addition, studies of strain have been recently performed in larger patient groups. One study of 323 patients after STEMI demonstrated that feature-tracking measures of global longitudinal strain was an independent predictor of outcomes including death, hospitalization due to heart failure, and reinfarction over 2 years (6). However, it was not additive to standard cardiac MRI risk markers such as those mentioned above. A larger multicenter study of 1235 patients with STEMI and 440 patients with non-STEMI demonstrated additive power of global longitudinal strain with feature tracking to infarct size and left ventricular ejection fraction (7). Differences between the findings of these two studies are likely due to differences in statistical power. Global longitudinal strain is predictive of outcome in many clinical conditions whether it is measured by using cardiac MRI or echocardiography.

With this as background, the article by Mangion et al in the present issue of Radiology (8) examines the predictive value of strain measured in two ways on post-STEMI outcome. In addition to feature tracking, these investigators also used cine displacement encoding with stimulated echoes (DENSE) for strain measurement (9). The latter is a high-spatial-resolution method for measuring strain and is especially robust for measurement of circumferential strain, which may be the most sensitive marker of intramyocardial dysfunction after MI. As discussed by the authors, midwall fibers are primarily circumferential, so if the subendocardium is lost due to infarction, it is the midwall that must pick up the slack. In the present study, 300 participants were recruited to undergo cardiac MRI after primary percutaneous intervention for STEMI. Only 261 (87%) had adequate cine DENSE images for analysis, which is one of the limitations of the technique, because it is technically challenging to acquire in sicker patients. Another 1.8% of the DENSE segments were inadequate for analysis.

The outcomes in this study were remarkably good. After 4 years of follow-up, only 21 (8%) of participants developed a major adverse cardiac event, pointing out the impressive advances in post-STEMI care that have occurred in the past 2 decades. That being said, patients with STEMI recruited into cardiac MRI studies may tend to be more stable patients. DENSE-derived strain had a higher area under the receiver operating curve of 0.76 compared with that of feature tracking at 0.62. It did not perform better than did infarct size or microvascular obstruction. Interestingly, neither left ventricular ejection fraction nor myocardial salvage index (T2-weighted imaging measures of area at risk−infarct size)/area at risk) were predictive.

The authors compared the predictive abilities of these various markers in a multivariable model. Herein lies the major flaw of the study. With only 21 patients with events, the authors only had statistical power to put two variables into a multivariate model. Thus, they were forced to compare DENSE-measured strain and other markers to infarct size alone, one by one. This is clearly not ideal. What is sorely needed is a multicenter study using DENSE to have adequate power to be able to put strain, infarct size, microvascular obstruction, left ventricular ejection fraction, et cetera, into a model simultaneously to enable identification of the most potent marker of outcome. This will likely require a sample size of well over 1000 patients, given the favorable outcome of patients following STEMI in the modern era.

By using the methods and sample size at hand, the authors were able to show that DENSE did perform better than infarct size, whereas the other markers did not. This is welcome news as we enter a future era of working toward avoidance of gadolinium-based contrast agents when possible. Should T1 mapping become standardized in the measures of acute and chronic infarct size (10), along with strain, these could become standard methods of assessing patients after STEMI. Another take-home message of the study is that DENSE performed better than did feature tracking. The latter has limitations related to through-plane motion and challenges regarding reproducibility. Should advances in DENSE pulse sequences and automated analysis come to fruition, it is likely that it could become the reference standard approach for measuring strain. A contrast-free imaging approach with T1 mapping and DENSE imaging could check all the boxes for prognostic assessment of the patient after MI.

Nearly a decade ago, I proposed a multicenter registry of patients after STEMI to the National Heart, Lung, and Blood Institute (NHLBI), hypothesizing an improved prognostic power of cardiac MRI over echocardiography. However, the proposal was turned down early in the process, primarily because the question was raised, how would it change management? At present, all patients after STEMI receive aspirin, beta-blockade, angiotensin-converting enzyme inhibition, statin, and adenosine diphosphate receptor antagonists, assuming they received a stent and as long as all can be tolerated. No difference in management is indicated at present if a large infarct, microvascular obstruction, or reduced strain is identified. This limits the utility of cardiac MRI in the patient after STEMI. Until such time as a management change (eg, implantable cardioverter defibrillator or neprilysin inhibition) is indicated in such patients, this will not change. Elegant MRI studies such as the present remain proof-of-principle demonstration of important post-MI pathophysiology, as well as examples of the power of MRI to demonstrate prognosis in cardiac patients. The next steps will require understanding how best to use this powerful information to improve patient outcome. Post-STEMI care has improved so dramatically that it has become difficult to demonstrate additive values of new therapies, because it requires tens of thousands of patients and large sums of money to do this. Cardiac MRI likely holds the key to identifying the value of future novel therapeutics in the patient following STEMI.