Background
Cardiovascular magnetic resonance (CMR) has emerged as a popular and useful tool for noninvasive myocardial tissue characterization [
1]. CMR provides valuable anatomical and functional information through high spatial resolution images and soft tissue contrast, without exposing patients to ionizing radiation. There is particular interest in using CMR to detect and track myocardial water content, because edema is a feature of many cardiovascular conditions [
2‐
4]. T2-weighted (T2W) CMR sequences have been used for this task [
5], but several problems inherent to these sequences have limited the widespread acceptance of this sequence to detect edema [
6]. New T2-mapping sequences have recently been proposed to overcome some of these limitations [
7‐
10] and provide absolute quantification of myocardial T2 relaxation times that can be compared among studies, the reference standard being T2 turbo-spin-echo (T2-TSE) [
11,
12]. However, these methods are either time-consuming or require specialized software for data acquisition and/or post-processing, factors which limit their routine clinical use.
It is noteworthy that no T2-mapping sequence has ever been validated for quantification of myocardial water content against direct measurement by a gold standard technique. The validity of T2-mapping sequences for this task has been assumed based on their ability to retrospectively delineate the hypoperfused myocardial territory supplied by the occluded coronary artery (the area at risk). However, regional T2 relaxation time in the post-ischemia/reperfusion area is affected by tissue characteristics [
13,
14], the application of cardioprotective therapies [
15,
16], and the timing of image acquisition [
17], and therefore the evidence supporting the validity of T2-mapping for myocardial edema quantification is weak.
In this study, we sought to provide in vivo validation of a sequence for fast and accurate T2-mapping of the myocardium using the gradient-spin-echo (GraSE) technique [
18], which could be rapidly and easily integrated in daily protocols as it is commercially available from many vendors. To achieve this goal, we used a closed-chest pig model of ischemia/reperfusion in which animals were serially scanned and sacrificed at different time-points after reperfusion for direct quantification of myocardial water content.
Discussion
In this study we present an in vivo validation of a CMR sequence for fast and accurate T2-mapping of the heart using the gradient-spin-echo (GraSE) technique that can be easily integrated in routine protocols, overcoming some of the limitations of current CMR mapping sequences for myocardial T2 quantification. For this endeavor we developed a closed-chest large animal model of ischemia/reperfusion in which animals had serial CMR scans and were sacrificed at serial time-points after reperfusion for direct quantification of myocardial water content. Two mapping sequences were used to quantify myocardial T2 relaxation time: the well-established reference T2-TSE technique and the newer T2-GraSE technique, which further speeds up the TSE sequence [
18]. The translational pig model of myocardial infarction used in this study allows examination of a wide range of myocardial T2 relaxation times and myocardial water content [
17], and produced values that closely mimic those clinically observed in several pathological conditions, strengthening the present validation.
Accurate noninvasive detection and quantification of myocardial edema is of great scientific and clinical interest given the occurrence of edema in several cardiovascular diseases and its usefulness for diagnosis and its correlation with ventricular remodeling and prognosis [
20‐
22]. Many studies over the past decades have investigated the use of CMR with T2-weighted (T2W) sequences to monitor in the post-ischemic myocardium, since this approach is considered especially suited to the detection of high water content in this setting [
23]. However, several problems inherent to T2W-CMR have limited the widespread uptake of this sequence for the detection of edema [
6]. These problems include variations in surface coil sensitivity, motion artifacts, incomplete blood suppression, and the subjectivity of image interpretation [
24].
A number of T2-mapping sequences have been recently proposed as a route to overcoming some of these limitations [
7‐
9] and providing absolute quantification of regional T2 relaxation times that can be compared across studies. However, these methods are either time-consuming or require specialized software for data acquisition and/or post-processing, factors that impede their clinical routine use. Compared with these other approaches, T2-GraSE mapping has many advantages, including an acceptable acquisition time for integration into daily clinical CMR protocols, reduced energy requirements, and the use of standard post-processing methods. In this study, T2-GraSE mapping was 3-times faster than conventional reference standard T2-TSE mapping due to the interleaving of the EPI readout between two consecutive 180° pulses. The applicability of myocardial T2-mapping using the GraSE technique in humans has been reported recently [
12,
25]; however, these studies mostly examined healthy hearts and therefore a narrow range of myocardial T2 relaxation times, and did not validate the sequence against directly determined tissue water content. Our study provides robust validation of T2-GraSE over a wide spectrum of myocardial T2 relaxation times and water contents that reflect a range of potential clinical scenarios.
Descriptions of previous T2-mapping sequences have relied on their ability to retrospectively identify the hypoperfused myocardial territory supplied by the occluded coronary artery—the area at risk—and there are no published data validating these techniques against true myocardial water content. Regional T2 relaxation time in the ischemic area can be altered depending on tissue characteristics [
13,
14], the application of cardioprotective therapies [
15,
16] or the timing of imaging acquisition [
17], highlighting the need of establishing the relationship between T2-mapping and true myocardial water content directly quantified in the tissue. This question has been explored in only a few studies conducted over 20 years ago [
26‐
30], and these studies were performed in low magnetic fields or with excised hearts, factors well known to affect T2 relaxation time [
31]. The present study is thus the first to provide in vivo validation of T2-mapping against actual tissue myocardial water content in magnetic fields used in current clinical practice. We believe it is important to assess the association between actual water content and T2 relaxation time at 3 Tesla since no clear relationship has been established between these parameters at different field strength. In this regard, in vitro analysis have demonstrated an increase on myocardial T2 at 3 Tesla systems compared to 1.5Tesla [
32], while in-vivo studies suggest equivalent T2 values at 3-Tesla with those previously reported at 1.5 Tesla [
33].
Our data demonstrate similarly good correlation between myocardial water content and both T2-mapping techniques examined. In our study ≈ 25-30 % of T2 relaxation time variance was not completely explained by water content changes, highlighting the influence on T2 values of other tissue characteristics and components [
6,
17] including the proportion of free/bound water as well as its location (intracellular/extracellular) in the tissue [
34]. These data should be taken into account when interpreting clinical studies using these sequences.
In summary, we provide in vivo validation of a CMR sequence for fast and accurate T2-mapping of the heart using the gradient-spin-echo (GraSE) technique. This approach can be easily integrated in routine protocols since it is available for all equipment, and overcomes some of the limitations of current CMR mapping sequences for T2 quantification.
Limitations
Although tissue changes in the post-I/R myocardium in pigs are similar to those in humans, we cannot rule out the existence of subtle histological differences between human and pig after infarction. As a consequence, the ≈ 30 % of T2 relaxation time variance that was not completely explained by water content changes in the present study might differ slightly from the value in humans. However, experimental studies allow validation, as shown here with the direct quantification of myocardial water content.
The pig is one of the most clinically translatable large animal models for the study of myocardial infarction and related issues due to its anatomical and functional similarities to humans [
35], and also shows similar T2 values in the ischemic and remote myocardium to those seen in humans [
8,
12].
In this study the ROIs for T2 relaxation time quantification covered the entire wall thickness and were individually adjusted by hand to carefully avoid the right and left ventricular cavities. The ROIs therefore might include different myocardial states (e.g., hemorrhage, microvascular obstruction, collagen) given that reperfused myocardium is a very heterogeneous condition [
36]. However, we adopted this approach to match the analysis of water content, which was evaluated in the entire wall thickness. Both T2 mapping sequences did not apply any correction for respiratory motion. However, all scans were performed during free breathing; therefore animals had an abdominal breathing pattern with minimal chest movement in antero-posterior direction minimizing the ghosting artefacts in short axis view. Three-dimensional (3D) T2-mapping sequences have been very recently described and might benefit from higher spatial resolution, therefore reducing partial-volume averaging effects and misregistration between images [
37,
38]. However, the proposed implementation of the T2-GraSE mapping can be performed in many modern scanners in a reasonable acquisition time, while 3D T2-mapping normally take longer acquisition times.
Acknowledgements
We are greatly indebted to Gonzalo J Lopez-Martin who carried out images acquisition. We thank Tamara Córdoba, Oscar Sanz, Eugenio Fernández and other members of the CNIC animal facility and farm for outstanding animal care and support. Simon Bartlett (CNIC) provided English editing.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
RFJ carried out the experimental procedures, participated in the data analysis and design of the study, performed the statistical analysis and drafted the manuscript. JSG conceived and carried out the application of the sequence, participated in images acquisition and design of the study, and helped to draft the manuscript. JA helped to perform experimental procedures and to draft the manuscript. MT carried out the data analysis, and helped to perform experimental procedures and to draft the manuscript. CGA helped to perform experimental procedures and to draft the manuscript. BI participated in the design of the study. BI and VF coordinated and helped to draft the manuscript. All authors read and approved the final manuscript; and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.