Myocardial T2 signal intensity (SI) is influenced by myocardial water content and can clinically detect myocardial inflammation associated with myocarditis 26, Tako-Tsubo cardiomyopathy 27, and acute myocardial infarction39. The ability of T2 SI to detect heart transplant rejection has been inconsistent in the literature2935. Aherne et al. showed in a dog model that T2 SI was initially similar between untreated allografts and non-transplant controls, but by day seven, T2 SI was 66% higher in the untreated allograft group compared to controls29. Smart et al., and Revel et al. found no difference in T2 weighted SI in patients with biopsy proven rejection compared to those without rejection3438. Notably, Smart et al. did show that serial signal intensities for a given patient increased with biopsy proven rejection and decreased with anti-rejection therapy, however the specificity was only 33%34. Alemnar et al. found no association between T2 STIR values and transplant rejection in a group of 40 transplant patients 35. Despite its utility in other myocardial disease states, T2 signal intensity has shown mixed results in diagnosing heart transplant rejection.

T2 relaxation time is the decay time constant of magnetic signal after an excitatory pulse. T2 relaxation time is calculated by plotting the spin echo signal intensity against varying echo times and is believed to lengthen in proportion to the degree of myocardial edema (Figure 1). Long T2 relaxation times are associated with high tissue water content in models of myocardial infarction25, myocarditis26, and animal models of acute rejection 29303140 and is therefore a biologically plausible variable to detect human heart transplant rejection.

Normal myocardial T2 relaxation times vary as a function of magnetic field strength and measured values will depend on whether or not an appropriate pulse sequence for quantifying T2 has been used. The last point cannot be stressed enough given that not all T2-weighted pulse sequences are appropriate for obtaining accurate T2 measurements. For this reason, CMR studies generally define their own normal T2 relaxation times from a group of controls and describe abnormal T2 relaxation as more than two standard deviations above the mean 162033.

Since 1985, there have been eleven animal trials evaluating T2 relaxation time and transplant rejection. These trials have used predominantly rat and dog models and applied a variety of imaging platforms, transplant techniques and anti-rejection regimens. Nevertheless, they demonstrated that T2 relaxation times increased with histologic rejection 2930404142434445 and ex-vivo myocardial water content 29304041 (Table 1). Furthermore, they also demonstrated that the prolongation of T2 relaxation times observed in transplant rejection could be prevented by the addition of immunosuppressive agents such as cyclosporine 2942.

There have been eight human trials totaling 302 patients (521 CMR scans) comparing T2 relaxation times to transplant rejection as determined by endomyocardial biopsy (Table 2) 464748. Four trials showed significant correlation between T2 relaxation times and transplant rejection (Table 3) 16203233. The two trials that did not find an association between T2 and rejection both gated their image acquisition to ventricular systole 3637 which often leads to signal loss and poor image quality 4950.

In an early investigation, Wisenberg et al performed CMR (0.15 T) on ten healthy volunteers to establish normal values for T2 relaxation times and compared them with those obtained from 25 transplant patients scanned within 24 hours of endomyocardial biopsy 33. Transplant patients were scanned immediately following heart transplantation out to a maximum of 107 days. Patients who were scanned more than 24 days after heart transplantation showed significant correlation between T2 relaxation times and endomyocardial biopsy. All patients who were scanned in the first 24 days post transplantation had elevated T2 relaxation times irrespective of their biopsy results. After 24 days, a T2 relaxation time of > 46 ms (i.e., 2 standard deviations above control T2 relaxation times) achieved a sensitivity of 93% and specificity of 96% for detecting rejection. A second control group of patients undergoing non-transplant thoracotomy displayed T2 relaxation times that were not different from healthy volunteers. This suggests that T2 relaxation times are not able to discriminate rejecting and non-rejecting allografts in the peri-operative period. It also suggests that there are early causes of inflammation in heart transplantation not related to rejection. Pereira et al. have shown that in the first week post heart transplantation, there are transient increases in myocardial wall thickness that presumably reflect myocardial edema and are correlated to the length of cold ischemic time, but not to the presence of rejection 51. Previous work in pig models have shown significant myocardial edema resulting from the administration of cardio-protective solutions used during organ harvesting 5253. These findings suggest that T2 relaxation times are not able to discriminate rejecting and non-rejecting allografts in the peri-operative period due to normal inflammation and edema that occurs early after heart transplantation.

In a series of studies by Marie et al., it was found that T2 relaxation times were significantly higher in patients with International Society of Heart and Lung Transplant (ISHLT 1990) grade 2 or 3 rejection compared to those with less than grade 2 rejection (60 ms vs. 51 ms, p = 0.0001) 1620. The sensitivity in these studies was 89% and specificity was 75% and 91%, respectively 1620. Marie et al. also demonstrated that T2 relaxation times normalized following treatment of the rejection episode (60 ms vs. 49 ms) 16. In a subgroup analysis16, patients categorized as 'false positive' (i.e. elevated T2 relaxation time with no biopsy evidence of rejection) were significantly more likely to develop rejection in the subsequent three months than patients with both normal CMR and normal biopsy results (Sensitivity 63%, Specificity 78%).

Any assessment of diagnostic accuracy is predicated on the accuracy of the 'Gold Standard'. The utility of endomyocardial biopsy is clear, however, its diagnostic accuracy is not well characterized and is essentially impossible to study in a human population. The well recognized, and not uncommon, entity of biopsy negative rejection suggests that the sensitivity of endomyocardial biopsy is less than 100%. Therefore disagreements between CMR and biopsy may be a deficiency in CMR, endomyocardial biopsy or both. The subgroup analysis by Marie et al. that suggests that CMR might predict rejection in patients who are CMR 'positive' and biopsy negative raises the exciting possibility that CMR may be more sensitive than biopsy at detecting rejection.

Signal dropout and motion artefacts are well recognized problems affecting image quality in T2 weighted spin-echo techniques. These pulse sequence deficiencies are accentuated by low magnetic field strength, arrhythmia, and long scan times. Most of the papers in the field of T2 imaging and transplant rejection predate the newer and shorter Turbo Spin Echo (TSE) sequences and relied on the much slower spin echo techniques with long scan times. In addition many studies presented in this review obtained their images using magnetic field strength well below the current standard of 1.5 T, which yields lower signal to noise ratios. Long scan times and low field strength could both have had significant negative impact on image quality. TSE sequences reduce scan times, but T2 quantification can be adversely affected by stimulated echoes. TSE sequences are also sensitive to RF and static field inhomogeneities 54. T2 weighted images are prone to signal gradation related to the proximity of the sampled area to the acquisition coil. This has been largely corrected with newer signal intensity correction algorithms, however some of the studies presented in this review used T2 weighted images that were likely subject to this potential error. Biological factors may also affect T2 quantification as many tissues types including the heart have been shown to have multi-exponential T2 behaviour 55. Under sampling the T2 decay curve can lead to a two fold error in T2 estimation when multiexponential T2 behaviour is present54.

He et al. have recently developed a T2 quantification sequence for the purpose of measuring myocardial iron in patients with thalassemia 56. This novel spin-echo technique has since been installed and validated at four other international centres 57. The coefficient of variation for local inter-study and inter-site variability was reported as 4.4% and 5.2%, respectively.

Despite the limitations of T2 weighted imaging mentioned above, previous work in transplant rejection studies has shown a consistent linear association between T2 relaxation times and the degree of acute transplant rejection. Furthermore, the relatively mild degree of rejection detectable with this technique (i.e. grade 2 ISHLT 1990) would not even require therapy in many contemporary transplant settings. This suggests that the relationship between T2 relaxation and rejection is highly sensitive and very unlikely to miss any cases of advanced rejection. Improvements in T2 imaging in the current era such as higher field strengths, fast TSE sequences, and improved blood and fat suppression techniques will likely strengthen the association between T2 relaxation times and transplant rejection49.

Gadolinium based contrast agents are by far the most common contrast agents used in clinical CMR imaging. Intravenous gadolinium increases signal intensity on T1 weighted images acquired early after contrast administration, in proportion to the degree of tissue perfusion and is thought to reflect the hyperemia seen in inflamed tissue. Increase in signal intensity early after contrast injection (early enhancement) has shown utility in the diagnosis of other disorders of myocardial inflammation such as myocarditis 26. Abdel-Aty et al. found that early enhancement alone was too insensitive for diagnosing myocarditis, but was useful when used in combination with T2 values and late gadolinium enhancement in a scoring system for the diagnosis of myocarditis 26. Yoshida et al. used a non-working allograft heart transplant model in dogs to demonstrate a 25–42% increase in signal intensity post gadolinium in rejecting allografts compared to native hearts 58. Konstam et al. showed that T1 weighted maximal myocardial signal intensity post gadolinium infusion could identify three distinct grades of rejection in a rat model 59.

In two human trials of transplant rejection, post contrast signal intensity tended to increase with degree of rejection although it could not consistently identify the full spectrum of abnormal endomyocardial biopsies diagnostic of rejection 3536. Alemnar et al. tested several variables of contrast enhanced myocardial signal intensity in 40 heart transplant patients and found no association with rejection35. Mousseaux et al. examined 39 heart transplant patients for an association between biopsy proven rejection and myocardial enhancement (post contrast myocardial SI – pre contrast myocardial SI/pre contrast myocardial SI) within ten minutes post gadolinium injection 36. They found an increase in myocardial enhancement in rejectors compared with non-rejectors (mean enhancement: grade 1 rejection = 70 +/- 14%, grade 2 or 3 rejection = 81 +/- 27%, non-rejectors 53% +/- 24, p < 0.05). However, myocardial enhancement was not able to discriminate rejection severity 36.

The correlation between biopsy proven rejection and echocardiographically determined ventricular morphology is specific in severe cases of acute cellular rejection but is too insensitive to be used as a screening tool 60. It has been postulated that the superior spatial resolution of CMR may lead to improved sensitivity in diagnosing rejection on the basis of changes to ventricular morphology 23.

Myocardial wall thickness has been shown to increase in both animal 4161 and human 3338 CMR trials of transplant rejection. Several animal studies showed that increased wall thickness during acute rejection was correlated to ex-vivo total myocardial water content 29304062. Wall thickness was not capable of accurately identifying the severity of a rejection episode.

Revel et al. studied 29 heart transplant patients using CMR and found that wall thickness increased during acute rejection and decreased as the rejection episode resolved 38. Wisenberg also showed that left ventricular wall thickness was increased in patients with rejection (ISHLT ≥ grade 2) compared to those without rejection (21 mm vs. 13 mm) 33. Alemnar et al. performed CMR on 40 transplant patients receiving contemporary anti-rejection therapies and found no significant differences in ventricular volume, wall thickness, and ejection fraction between those with and without histologic evidence of rejection 35. In the late 1990s, animal trials by Yoshida and Walpoth found that hearts undergoing rejection had reduced ejection fraction and stroke volume, although these changes were only significant when rejection was moderate or severe 6162.

Changes in ventricular morphology and systolic function as measured by CMR are associated with rejection 606364. Despite the excellent spatial resolution of CMR, these variables are probably of insufficient sensitivity to detect the early and milder forms of rejection that are of clinical interest.