Cardiovascular magnetic resonance of myocardial edema using a short inversion time inversion recovery (STIR) black-blood technique: Diagnostic accuracy of visual and semi-quantitative assessment

The ability of magnetic resonance (MR) techniques to assess myocardial infarction has attracted the interest of cardiovascular researchers from the very beginning of cardiovascular MR (CMR). The behavior of acute myocardial infarction (MI) in dogs was described by Higgins as early as 1983 and both T1 and T2 were found to increase after coronary occlusion [1]. Wisenberg et al. observed that this increase starts in the first hour after myocardial reperfusion [2]. In 1988, Been et al. published T1 maps of patients with acute myocardial infarction demonstrating elevated T1 values in areas of acute MI [3]. It could be shown that these increases of relaxation times are due to myocardial edema [1, 4], which is regarded as a phenomenon that is invariably linked to acute myocardial injury.

A new approach for the visualization of myocardial edema in humans was introduced in 1996. In their work, Simonetti adapted the short time-from-inversion inversion recovery (STIR) black-blood technique for cardiac applications [5] and showed examples of clinical patients, where the areas of acute infarction could be clearly visualized with this technique as regions with hyper-intense signal. Only a limited number of CMR centers used this approach, until Abdel-Aty published in 2004 the results of a study comparing STIR images in acute and chronic myocardial infarction [6]. In this study, STIR images had a specificity of 96% to differentiate patients with acute from those with chronic MI. The authors concluded that an imaging approach combining late gadolinium enhancement (LGE) and T2-weighted CMR accurately differentiates acute from chronic MI.

Since STIR was perceived as a T2-weighted technique, several other publications subsequently investigated alternative T2 techniques to assess myocardial edema. Some authors speculated that the hyper-intense area on T2-weighted techniques represents the area at risk, and several animal studies support this concept [7, 8].

Edema imaging has been shown to be also useful in other acute myocardial diseases, such as transplant rejection [9], myocarditis [10], and stress-induced (Takotsubo) cardiomyopathy [11].

The centers involved in these studies have translated the results from their research into their clinical practice, whereas some other centers aiming to do the same have experienced major discrepancies between their findings with T2-weighted techniques and clinical status of their patients. Specifically, a large prevalence of image artifacts and a significant variation in results depending on the choice of pulse sequence parameters has been reported [12], leading to a considerable amount of uncertainty and confusion among CMR centers and physicians new to edema imaging.

With this discussion in mind, we set to investigate if the practice to use STIR for the detection of myocardial edema in both clinical and research settings was justified. The aim of this study was therefore to review the technical principles behind STIR, and to assess the diagnostic accuracy of STIR in patients who underwent CMR in both the acute and chronic stages after MI. We hypothesized that STIR imaging is an accurate and robust way of detecting myocardial edema in non-selected patients with acute myocardial infarction.

Over a period of 14 months, 46 consecutive patients with acute first MI were prospectively enrolled to undergo CMR within the first week and at 6 months after MI at the Leeds General Infirmary, Leeds, UK. Exclusion criteria included clinical instability and the normal contraindications to CMR or the administration of gadolinium (Gd) contrast; no other exclusion criteria were applied. The local ethics committee approved the study, and written informed consent was obtained from all patients.

Our study shows that STIR imaging with appropriate pulse sequence parameters allows for visualization and semi-quantitative assessment of myocardial edema due to acute myocardial infarction with high diagnostic accuracy.

The results of our study demonstrate the high image quality that can be achieved if a STIR sequence with appropriate pulse sequence properties is available. In our patients, we could not only assess transmurality of myocardial edema, but also clearly visualize hypointense cores within the edematous zone, a phenomenon that is related to hypoenhanced cores on LGE images [21, 22], probably represents intramyocardial hemorrhage [23] and recently has been shown to be linked to adverse clinical outcome [24]. Image quality was also sufficient to assess the presence of edema on the segmental level rather than case-by-case. As a limitation, it has to be kept in mind that most patients in our study suffered from ST-elevation MI and therefore presented with severe myocardial lesions. Results might differ in patients with less severe ischemic or non-ischemic events.

This study confirms the high accuracy of STIR for differentiating acute from chronic cases of myocardial infarction that was first reported in the study by Abdel-Aty et al. [6] In contrast to that study we found that hyperintensity on STIR was not always transmural, indicating superior spatial resolution of our image data. As MSI was larger in reperfused than in non-reperfused infarcts, the hyper-intense areas in our STIR images can be attributed to the area at risk. The finding that the area at risk does not necessarily have fully transmural extent is supported by the observation that not all non-reperfused infarcts lead to transmural necrosis on LGE, and might be explained by the development of significant collateral blood supply stimulated by repeated transitory ischemia in some individuals prior to the final occlusion of an epicardial coronary vessel.

The increase in T2 associated with edema in acute MI is relatively small (25), and estimated to be on the order of 25-50% for T2 in range 45-50 ms for normal myocardium and 60-65 ms for acute MI [25]. Thus signal uniformity is critical for determination of the size of elevated T2 region. In this case we were able to benefit from the coil sensitivity-based homogeneity correction that comes with CLEAR.

The STIR technique is based on a double-inversion recovery black-blood turbo spin echo (TSE; equivalent: fast spin echo) pulse sequence, into which a third 180° inversion pulse is integrated. The specific feature of this third inversion pulse is its relatively short TI (140-180 ms), which leads to a suppression of the signal from fatty tissue and from anything else with a T1 relaxation time of approximately 200-250 ms. As illustrated by Dwyer et al in 1988 [26], the second effect of this inversion pulse is far more important in edema imaging: the early inversion of the relaxation curve before the TSE 90° pulse leads to an inversion of the effects of long T1 on signal intensity. The presence of increased tissue water with edema causes an increase in both T1 and T2 relaxation times. In comparison to a standard TSE sequence (without an inversion pulse), where an increase in T1 value causes a reduction in signal intensity, the inversion of the net magnetization prior to the TSE 90° pulse in STIR sequences results in an increased signal intensity for longer T1 relaxation values, provided the relaxation curves do not cross zero. If T2 weighting is also introduced by increasing the echo time, TE, the relative signal intensity is increased for increased T2 relaxation values. So in the case of a standard TSE pulse sequence, the contrast arising from T1-weighting opposes the contrast arising from T2 weighting, while in the case of a T2-weighted STIR pulse sequence, the T1 and T2 effects of water are combined which boosts the contrast between edematous and non-edematous tissue beyond the limits of pure T2 weighting (Figure 5). The above signal dependencies translate to image pixel intensities provided that a magnitude (or modulus) image display is chosen that ignores the sign of the magnetization prior to readout by the 90° pulse.

While these theoretical considerations clearly show that the STIR pulse sequence is favorable for the visualization of myocardial edema, at the moment the theoretical advantages are not realized in all CMR centers. A possible reason for this might be the lack of standardization of STIR pulse sequence parameters. From our experience there are numerous pulse sequence parameters in STIR that need to be set appropriately in order to optimize image quality and maximize the edema contrast. The list of these parameters includes, among others, timing and design of STIR and black-blood inversion pulses; quality of TSE read-out pulses; bandwidth; and slice thickness. Dwyer et al. also addressed the importance of allowing a sufficient repetition time (TR), when they systematically investigated signal evolution in STIR and predicted peak T1 values for additive T1 and T2 behavior at a given TR [26]. In practice, TR for myocardial imaging at 1.5 Tesla should be set to two RR intervals for heart rates ≤ 80/min, and to > 2 RR intervals for heart rates above that value. As most TSE implementations do not allow for such values, the same effect can be achieved by prescribing half of the given heart rate, which will lead the MR system to omit every other heart beat and thus effectively double the TR, for example, if the patient's heart rate is 90, prescribing a scanner heart rate of 45 would double the TR. While the choice of a long (> 80 ms) echo time (TE) would seem obvious to achieve high T2 weighting, considerably shorter TE has been used in some studies on STIR [27]. Apart from the fact that on some MR systems, the signal-to-noise ratios achievable with long TE are not sufficient to generate diagnostic images when the standard configurations of black-blood pulse sequences provided by the manufacturer are used, the choice of shorter TE might be justified as the contrast between typical remote and edematous myocardium should not change significantly for TE beyond 50 ms (Figure 5).

One concern in relation to STIR imaging that is frequently raised is the fact that slow-flowing blood in the ventricular cavity can give bright signal at the endocardial border. This can be mistaken for myocardial hyperintensity. In our study, we used cine images at the same spatial and temporal location as a reference for endocardial contour detection. This method effectively helps avoid confusion with intracavity signals.

A number of other methods of T2 weighted imaging have recently been described including bright [28, 29] and dark blood [30] methods. These have recently been compared and showed significant variation in the size of the area of edema [31]. STIR imaging seems to provide a reasonable balance between sensitivity and specificity when all available clinical and angiographic data are used. In addition, STIR has the advantage of being the most widely available sequence. It has also available for more than 10 years, which results in better knowledge of its use and potential artifacts.

While our study shows the high value of STIR for the qualitative ("yes/no") detection of focal myocardial edema, this technique is limited as any other conventional MR technique in the way that it does not allow for absolute quantification of signal intensity. In other words, the identification of myocardial edema with STIR is based on its perceptible contrast to non-edematous myocardium rather than on absolute thresholds. Therefore, applications involving diffuse processes or serial follow-up (e.g. for the assessment of inflammatory activity) should benefit from the use of alternative MR strategies enabling absolute signal quantification such as T1 mapping [20] or T2 mapping [32, 33].

CMR using STIR enables the detection of infarct-related myocardial edema with high spatial resolution and high diagnostic accuracy. STIR imaging possesses unique properties beyond T2-weighting and fat suppression, and should be regarded as an optimized edema-weighted technique rather than a purely T2-weighted method. CMR centers and individuals interested in the use of STIR should compare their results to those of other centers in order to optimize their protocols, and should not settle for sub-optimal image quality.