Cardiovascular magnetic resonance in heart transplant patients: diagnostic value of quantitative tissue markers: T2 mapping and extracellular volume fraction, for acute rejection diagnosis

Heart transplantation is the therapy of choice for end-stage heart failure to improve survival and quality of life. Survival after cardiac transplantation is linked to the occurrence of complications, especially the risk of acute rejection during the first year [1, 2]. Approximatively 20 to 40% of patients have at least one episode of acute rejection during the first year [3]. Clinical monitoring and echocardiography have failed to detect early rejection [4]. So far, endomyocardial biopsy (EMB) is the gold standard procedure to diagnose acute rejection. Although there is no international consensus on the optimal frequency of EMB, an average of 10 to 15 biopsies are performed during the first year [5]. The procedure requires radioscopy (irradiation) and a large introduction system. This is an invasive procedure with 0.5 to 3% of complications [6, 7]. Most serious complications are tamponade by cardiac perforation, ventricular arrhythmia and tricuspid valve injury leading to regurgitation [8]. Moreover, the accuracy of EMB to diagnose rejection is imperfect due to a non-uniform pathologic process [9]. Indeed, biopsy negative rejection (true rejection with negative biopsy) is a well-known recognized entity and represents 10 to 20% of rejection [10]. There is also significant interobserver variability in the histologic interpretation of myocardial samples [11]. Clinicians need a reliable and non-invasive process for diagnosing rejection and for limiting the number of biopsies performed.

Rejection in heart transplantation is mediated either by T lymphocytes that have been activated against donor antigens (cellular rejection) or by activated B cells secreting antibody molecules (antibody-mediated rejection: AMR). These processes lead to cellular infiltration, myocardial oedema and inflammation with eventual myocytes damage and haemorrhage [12].

Cardiovascular magnetic resonance (CMR) is the gold standard imaging modality for assessing cardiac morphology, ventricular volumes, systolic function and myocardial mass [13]. In addition, CMR allows for assessing the activity of inflammatory changes using markers for myocardial oedema, hyperaemia, capillary leak and irreversible injury applying a combination of non-contrast (T2 weighted imaging and more recently parametric mapping techniques (T1 and T2 mapping) and gadolinium enhanced technique [14‐16].

Native T1 values are higher with increased extracellular compartment by fibrosis [17] oedema [18] and amyloid [19]. From native and post-contrast T1, we can calculate extracellular volume fraction (ECV) which represents the interstitial volume [20]. Expansion of interstitial volume occurs with diffuse fibrosis, oedema and infiltrative diseases [21]. Recent studies have shown the improvement of these mapping techniques in the diagnostic accuracy of CMR in patients with suspected myocarditis [22‐24]. T1 and T2 mapping sequences accurately diagnoses interstitial edema and extracellular space expansion and can potentially detect acute allograft rejection. In heart transplantation, few studies have assessed these mapping techniques [25‐30]. Marie et al., more than 10 years ago, analysed T2 mapping in acute heart transplant rejection using a black-blood sequence on a low 0.5 Tesla magnet [25, 26]. They found that a T2 value ≥56 ms had an accurate detection of acute cardiac rejection defined by grade ≥ 2 (sensitivity 89%, specificity 70%). Bonnemains et al. confirmed these results on a higher field magnet (1.5 T) on 60 patients with a T2 value > 60 ms associated with risk of rejection higher than grade 2 [30].

To our knowledge, no studies have evaluated multisequential approach using T2 mapping, T1 mapping and ECV in heart transplant patients. The purpose of the study was to evaluate T2, T1 and ECV quantification as novel tissue markers to detect acute rejection in comparison to the gold standard strategy based on EMB.

To our knowledge, this is the first study combining CMR T2 and T1 mapping with segmental and global analysis in a prospective cohort of heart transplant patients. The major findings were:

Basal T2 and basal ECV had a high diagnostic accuracy to identify patient with acute rejection. Combining T2 and ECV increased the sensitivity leading to the detection of all cardiac rejection episodes and potentially decreasing by 63% the number of EMBs. Segmental analysis of T2 demonstrated that acute rejection is a diffuse process involving all segments of the myocardium.

Because of interstitial edema and lymphocyte infiltrate, free water content is increased in acute rejection. As a result, tissue T2 relaxation time is increased and causes hyperintense areas on T2 weighted image in comparison with normal remote myocardium. Previous studies did not show significant increase in overall myocardial T2 signal ratios in patients with cellular rejection [33, 34]. More recently, Krieghoff et al. [35] in 73 heart transplant patients found that edema ratio had a 63% sensitivity, 78% specificity, 30% positive predictive value and 93% negative predictive value for cellular acute rejection grade ≥ 1b (ISHLT 1990 classification [36]).

Nevertheless, these sequences are limited by a low signal-to-noise ratio, motion artifacts, variable signal intensity and the possibility of inflammation of the skeletal muscle distorting the T2 ratio results. One of the strengths of T2 mapping is the possibility to measure T2 relaxation times in the absence of a ROI in the skeletal muscle and without comparison with normal myocardium. Current T2 mapping sequences appear robust and have led to improve image quality and reproducibility [37]. Our study confirms the high image quality of T2 mapping sequence with 98.6% of the analyzable segments (489/496) in accordance to Usman study (6 of 74 scans excluded) [29] and Butler study (6 of 77 scans excluded) [28].

Early studies in animal trials have showed, with histologic rejection, an increased T2 relaxation time with T2 values in rejecting animal from 46 to 68 ms [38, 39]. However, these results were limited by older sequence techniques on variable magnet (from 0.25 to 1.5 Tesla).

In humans, several studies have compared T2 relaxation times to transplant rejection [25‐30]. Marie et al. [26], on 123 comparisons CMR-EMB using inversion-recovery/spin echo sequences on a low 0.5 T magnet, showed that a T2 ≥ 56 ms had sensitivity of 89% and specificity of 70% to detect moderate acute rejection using the previous ISHLT 1990 criteria [36]. More recently, Usman et al. [29] using updated imaging sequence (single shot T2 prepared bSSFP) and updated grading ISHLT classification, found that the optimal T2 time cut off was 56.4 ms with sensitivity of 86.5% and specificity of 94.6% to diagnose acute rejection in 8 rejection cases. Butler et al. [28], on 73 comparisons EMB-CMR using a modified half Fourier acquisition single shot turbo spin echo sequence, found that T2 relaxation time was 63 ± 6 ms in 15 patients with positive biopsy (defined by grade ≥ 2R or antibody mediated rejection) versus 57 ± 6 ms in 58 patients with negative biopsy. However, the authors did not use most recent T2 pre-bSSFP less sensitive to motion artifacts and incomplete blood suppression. A recent study [40] is not in agreement with the current study and the 3 other previously mentioned [26, 28, 29]. Miller et al. [40], on 88 comparisons CMR-EMB using a similar T2 prepared bSSFP sequence to ours, did not find significantly difference in T2 values between rejectors and non rejectors (58.8 ± 3.5 vs 57.0 ± 3.2, p = 0.242). However, the study design was different with statistical adjustment for serial measurements within the same patient. As discussed by the authors, if the statistical methods used by Usman [29] and Marie [26] were applied, the differences in T2 (and T1) between cardiac rejection group would be significant. In accordance with Usman et al. [29], we found that a T2 time cut-off 57.7 ms had sensitivity of 70%, specificity of 96% and an excellent diagnostic accuracy of 90% to diagnose acute rejection. However, our definition of significant acute cellular rejection was different from the 2 aforementioned studies [28, 29]. In our study, histological grade ≥ 1R was considered relevant rejection and treated (either changes in long term medication or IV steroids). In the other studies, only cellular rejection episodes ≥2R were relevant. A revision of the ISHLT-1990 criteria has been proposed merging the previous grade 1A, 1B and 2 within new 1R [12, 36]. The main difference from the previous set of criteria includes the addition of the ISHLT-1990 grade 2 (single focus of myocyte injury) with the mild 1R category. It has been demonstrated that grade 2 biopsies had significantly more edema (higher T2) than grade 1A and 1B [26]. Grade 1R has now become unspecific and 1R biopsies correspond to very different situations. A similar T2 cut off between our study and Usman’s can be explained by a majority of ISHLT-1990 grade 2 categories in our group of rejection. A large CMR multicenter study in heart transplant patients is ongoing to confirm diagnostic value of T2 mapping and to study the prediction value of T2 when biopsies are negative [41].

Until now, myocardial signal intensity was measured as ROI in the septum most of the time. Interestingly, we performed a segmental T2 analysis by measuring T2 myocardial signal intensity as ROI in the 17 American Heart Association segment models [31]. Although rejection’s distribution is often heterogeneous on biopsies, we have demonstrated that acute graft rejection is a diffuse process involving all segments of the myocardium. However, measuring T2 relaxation time in each segment is time consuming and does not provide a better diagnostic accuracy than the global T2 measurement. In daily life, we do not recommend performing a segmental T2 analysis. A global T2 measurement or a ROI in the septum corresponding to the anteroseptal and inferoseptal segments seem to be accurate.

In contrast to T2 studies, limited data are available on T1 mapping. Increased native T1 values have been shown in patients with myocarditis compared with controls [23, 24]. Although the mechanism is unclear, it could reflect an increase in intracellular and extracellular water content, possibly related to potential T2 effects in the T1 mapping sequence [42]. In heart transplant patients, data on T1 quantification are scare. Early studies in animal trials have found a prolonged T1 relaxation time correlated to the severity of rejection [39]. In humans, previous studies did not confirm these results [43].

Although our results cannot be compared to previous studies with different employed CMR sequences or different timing post transplantation [40], we confirmed that T1 mapping values before and after contrast injection were not significantly increased in the presence of rejection. Measuring absolute myocardial T1 values is limited due to several factors that can alter T1 measurement such as the acquisition scheme, magnetization transfer, flow and T2 effect [44, 45]. Moreover, T1 values vary with the CMR sequence and the algorithm of T1 calculation leading to discrepancies when comparing results between different studies. In contrast, ECV quantification should not be affected by confounding variables mentioned before. Values of myocardial ECV are very similar in different studies in the range of 24–28% [46, 47]. ECVs values also show better agreement with histological measure of the collagen volume fraction than isolated post contrast T1 [48].

Our data demonstrated that basal ECV had an excellent sensitivity of 89% and a good specificity of 77% to diagnose acute rejection. ECV at apical level was not significantly different in patient with acute rejection probably due to the poor image quality in apical segments and a thin wall.

When we use a multi-parametric sequential approach combining the excellent specificity of T2 (96%) with the good specificity of ECV (85%), we identified all acute rejection episodes potentially decreasing by 63% the need of EMB (Fig. 4). There has been little previous investigation on multi-sequential approach in heart transplant patients. Using T2 weighted imaging early and late gadolinium Enhancement (LGE), Taylor et al. [33], on 50 heart transplant recipients, found that the presence of either elevated early relative myocardial contrast enhancement (> 3.5) or increased relative STIR intensity (> 2) had sensitivity of 100% and specificity of 73% for confirmed acute rejection. More recently, using myocardial tissue characterization (T1 and T2 mapping and LGE) and regional ventricular function (mid ventricular short axis tagged images), Miller et al. [40], demonstrated significant reduction in circumferential strain without difference in T1 and T2 values.

Our study suggests evidence for using a multi-parametric sequential approach to diagnose acute rejection in heart transplant recipients. Basal T2 mapping combining with basal ECV provided the best diagnostic accuracy leading to a potential reduction in the need for invasive EMBs by more than 50%. This is a preliminary study with a small sample size. Larger studies are required to confirm these data.