The Imaging of Myocardial Scars on MRI: The Role of a Radiologist

Introduction

Myocardial infarction (MI) is defined as necrosis of myocytes caused by ischemia and is divided into five types according to the new definition of MI by the Joint ESC/ACCF/ AHA/WHF Task Force [1]. Clinically, MI is classified as type I if related to a primary coronary event, as type II if due to increased oxygen demand or decreased supply, type III if there is sudden unexpected cardiac death, and type IV and V are related to coronary interventional procedures. Delayed- enhanced magnetic resonance imaging (DE-MRI) has proven, since the studies from KIM [2], to be a valuable and accurate diagnostic tool to detect MI [3].

Although the relationship of DE-MRI to the underlying pathophysiology is not completely elucidated, it is thought that delayed-enhancement of MI scars is due to an increased extracellular space, allowing a larger distribution volume for the extracellular contrast agent [4]. The contrast distribution volume within a tissue has a direct correlation with the longitudinal relaxation rate [5]. The higher the contrast concentration, the shorter the longitudinal relaxation time of the tissue will be. As a result, different tissues with different contrast distribution volumes will also have different T1.

After contrast injection, the T1 values in normal and infarcted myocardium tissues are distinct. This divergence allows, with a non-selective 1808 inversion pulse and an adequate inversion time, nulling the signal of normal myocardium before image acquisition [6].

Different prevalence of unrecognized myocardial infarctions (UMIs) have been published in epidemiologic studies [7]. Traditionally, UMIs are defined as the appearance, in a non-acute setting, of a new diagnostic Q wave. In a meta- analysis done by AMMAR et al., the occurrence of UMI as a proportion of all infarctions varied from 4 to 44% [8]. The main reasons found for this high variation were the different ECG criteria used in the different studies and the lack of a gold standard for the diagnosis of UMI. In the 70-year-old population in the Prospective Investigation of the Vasculature of Uppsala Seniors (PIVUS) study, there was a prevalence of MI scars detected by DE-MRI of 24.2%, in which UMI represented 80% of those [9].

The aim of this study was to explore differences in tissue characteristics between the previously found UMIs and recognized myocardial infarctions (RMIs) [9], by further assessing the signal intensity (SI) at DE-MRI.

Results

In both groups, the mean SI of the scars was different from the mean SI of the normal myocardium (UMI group, P-valueB0.0001; RMI group, P-value 0.0003). The inversion time chosen to null viable myocardium did not differ between the UMI, RMI, and normal subjects groups (P-value 0.3852). The mean SI ratio in the UMI group (4.593.0, mean 9SD) was lower than in the RMI group (8.995.1) (P-value 0.001) (Figure 3).

Click to enlarge

This difference was still significant (PB0.0001) after adjustment for gender, BMI, time of image acquisition after gadolinium injection, scar transmurality or total MI scar mass (Table 2). The result was the same if weight was used instead of BMI. An example of an UMI scar and a RMI scar are shown in Figsures 1 and 2, respectively.

Discussion

The present study demonstrates that the UMIs found in the PIVUS population have a different behavior from RMIs in DE-MRI, as was elucidated by the difference in SI ratio between the two groups. The SI ratio represents the mean scar tissue SI normalized to the mean SI of normal myocardium. One possible explanation for the difference in SI ratio could be based on potential partial volumes effects in the UMIs, since some of them were small. The 3D acquisition chosen to cover the whole left ventricle was comparable to the sequence used on similar machines resulting in a slice thickness of 10 mm [12]. The used slice thickness could possibly introduce partial volume effects that could be responsible for the decreased SI in the scar area in the UMI group.

In this study, scar tissue volume was not found to have any influence on the SI ratio. The result was the same when relating the SI ratio to gender, BMI, time after gadolinium injection, and scar transmurality. The only parameter that was significant in predicting the SI ratio was whether it was an UMI scar or a RMI scar. A second explanation for the disparity in SI ratio could be that the two groups would primarily diverge in the SI of the normal myocardium, which was used as a reference to normalize the SI of the infarction area. Due to the study protocol, it was not possible to scan the myocardium prior to contrast injection. Therefore, it was not possible to test the underlying assumption of comparable relaxation rates, of normal and infarcted myocardium, prior to contrast injection. Nevertheless, the SI of the normal myocardium was normalized by different inversion times after the inversion pulse. Differences in normal myocardium could be detected by variation in the inversion times used in the different groups.

For this reason, the inversion times were compared and no difference was found between the groups. Having excluded these possible confounding factors, the reason for the result is likely to originate in the scar. The variable likely to explain the differences in SI is different amount of contrast agent in the scar tissues and, consequently, different contrast distribution volumes in the scar tissues. This study aimed to do an ad hoc analysis to investigate whether the UMI scars previously defined had the same SI as the RMI scars in DE-MRI, in order to further explore differences between these groups. Previous epidemiological studies reveal that subjects with UMI have a similar mortality rate as the ones with RMI [13, 14]. In these studies, UMI diagnosis is based on ECG criteria and there is a lack of consensus in the definition of ECG-detected UMI. Nowadays, high resolution DE-MRI provides a more sensitive method to detect MI scars [15], some of which are small, such as the UMIs detected in our PIVUS study, with a mean volume of 1.9% of the total left ventricle mass [9]. Some of these scars might not have the sufficient extent of necrosis to produce a significant Q wave on ECG [16] and, therefore, their long-term cardiovascular risk has never been evaluated in a population-based study. However, some studies indicate that small myocardial scars detected by DE-MRI have an important prognostic value in the patients’ outcome. Kwong et al. report a seven- fold increased risk of major adverse cardiac events in patients with clinical suspicion of coronary artery disease and without a previously known MI, who had a small (mean 1.4% of left ventricle mass) MI scars detected by DE-MRI [17]. The present study had some drawbacks. The contrast dose was not individually adapted, since cardiac-MRI was performed after whole-body MRA, leading to a higher dose of contrast than it is recommended for DE-MRI [6, 12].

This implies that imaging acquisition should be even more delayed to allow contrast wash-out from the left ventricle cavity and to allow a better identification and delineation of small subendocardial scars. A second drawback was that the timing of imaging acquisition after the contrast injection was not ideal. It has been established that the best timing for DE MRI is between 25 and 30 min [12], while normally a scan delay of 10 min is clinically used [6]. The interval time in our study was 25 to 64 min. There was only one subject scanned at 64 min and, in that subject, there was an obvious enhancement. Excluding this subject, the interval time of imaging acquisition was between 25 and 45 min. To some extent, the longer waiting time might have compensated for the higher used dose. Additionally, these problems were compensated by individually adjusting the inversion time [12].

It is also known that the MI scars evolve over time, with infarction resorption occurring mainly in the first 8 weeks after coronary artery occlusion [18]. In this study, the age and evolution of the MI scars are not known. Nevertheless, most of them are likely to be chronic, since if several were acute or sub acute they would have occurred in a short time span, which is unlikely. This means that the scars are most probably old and that at the time of the study they were likely to be stable in terms of evolvement.

Conclusion

In conclusion, this is the first study to compare SI characteristics of UMIs and RMIs detected by DEMRI. The difference found in SI ratio between these two groups most likely reflects different contrast distribution volume of the tissues, which might indicate that UMI and RMI tissues diverge in composition.