Multi-parametric cardiovascular magnetic resonance with regadenoson stress perfusion is safe following pediatric heart transplantation and identifies history of rejection and cardiac allograft vasculopathy

We conducted a single-center retrospective cross-sectional case–control study that was IRB approved and HIPAA compliant. Informed consent was waived. PHT recipients who underwent a clinically ordered comprehensive stress perfusion structure–function CMR between May 2016 and January 2019 were identified. Eighteen historical institutional controls who had a normal structure–function CMR (without stress perfusion) for various indications including abnormal electrocardiogram (ECG) (n = 9), chest pain (n = 2), syncope (n = 2), family history of cardiomyopathy (n = 1), obesity (n = 1), suspected atrial mass (n = 1) and other systemic diseases (n = 2) were age and sex-matched to the PHT. Since there were no institutional healthy control patients who had undergone stress perfusion imaging, we identified a separate cohort of 16 non-transplant patients who underwent stress perfusion CMR with negative results performed for various indications where suspicion for microvascular dysfunction was minimal. Indications for CMR in this cohort were: post-coronary artery unroofing with chest pain (n = 4), single coronary artery (n = 2), history of arterial switch procedure (n = 1), failing Fontan (n = 1), remote history of Kawasaki disease with normal coronaries (n = 6), congenital coronary artery aneurysm (n = 1) and aortic stenosis with syncope (n = 1).

Demographic and clinical data collection included transplant surgical history, number of rejection episodes and history of CAV per clinical history or cardiac catheterization report. A Rejection Score (RS) was devised to quantify the clinical rejection history by adding the total number of rejection episodes (acute cellular rejection ≥ 1B, any episode of antibody mediated rejection (AMR) based on endomyocardial biopsy (EMB) as per ISHLT criteria [44] or any episode of biopsy negative rejection that was treated clinically) in order to generate a score. In addition, CAV grading was performed by independent expert blinded review of coronary angiograms from the cardiac catheterization done closest to the time of the CMR by an adult interventional cardiologist with > 40 years of clinical experience. Angiograms were graded based on ISHLT CAV scale [45].

CMR was performed on a 1.5 T scanner (Aera, Siemens Healthineers, Erlangen, Germany). A comprehensive protocol including structure and function assessment and stress perfusion testing was followed (Fig. 1) with the use of gadobutrol (Gadavist, Bayer HealthCare, Berlin, Germany), as the gadolinium-based contrast agent and regadenoson (Lexiscan; Astellas Pharm; Northbrook, Illinois, USA) as the pharmacologic stressor. Regadenoson is a selective cardiac A2A adenosine receptor agonist that vasodilates the coronary arteries thereby increasing myocardial blood flow.

Following image localizer scans, 2D cine balanced steady-state free-precession (bSSFP) images in the two-chamber, three-chamber and four-chamber planes were obtained in standard clinical fashion (TR = 3.0 ms TE = 1.26–1.3 ms; flip angle = 90°, slice thickness = 6 mm, in plane resolution = 1.0 × 1.0 mm²). T2 mapping was performed prior to contrast injection. Three T2-prepared bSSFP images with varying T2-prep times (0, 24, and 55 ms) were acquired in a breath-held fashion in three short axis orientations (base, mid, apex) through the LV (TR = 2.5 ms, TE = 1.1 ms, slice thickness = 8 mm, in-plane resolution 1.9 × 1.9 mm²). These levels were pre-selected by the technologist at the time of the scan: basal level defined at the mitral valve leaflet tips, apical level near the apex where the ventricular cavity was still visible at end-systole and mid ventricular level mid-way between those. Next, T1 mapping with a modified Look-Locker inversion recovery (MOLLI) sequence was used to measure native (pre-contrast) longitudinal relaxation T1 times of myocardium at the same three levels as well as within the blood pool. Pre-contrast T1 maps were obtained using a single-breath-hold, ECG-triggered, MOLLI sequence (TR = 2.6 ms, TE = 1.0 ms, slice thickness = 6 mm, in-plane resolution 0.7 × 0.7 mm²). Myocardial stress perfusion was then performed using regadenoson at a dose of 6–10 mcg/kg (up to a maximum dose of 400 mcg) [46] with close patient monitoring of symptoms and vital signs. First-pass contrast-enhanced images were acquired at 60–90 s at the same three short axis levels. Single-shot fast gradient echo, advanced motion corrected fast gradient echo sequences were acquired during stress after the intravenous administration of 0.05 mmol/kg of gadobutrol (TR = 2.5 ms TE = 1.1 ms, inversion time (TI) 150 ms, flip angle = 12°, slice thickness = 8 mm, inplane resolution = 2.8 × 2.8 mm²). Following image acquisition, a single dose of 75 mg of intravenous aminophylline was given to reverse the effects of regadenoson. Between stress and rest perfusion, ECG-gated 2-D Cine bSSFP imaging of the ventricles in short-axis from base to apex was obtained. Contrast-enhanced first pass perfusion imaging at rest was obtained following a second 0.05 mmol/kg dose of gadobutrol. Finally, a third dose of 0.05 mmol/kg dose of gadobutrol was injected with post-contrast T1 mapping. LGE was then obtained using segmented inversion-recovery sequences in three orientations: 4-chamber, 2-chamber and short axis from base to apex (TR = 2.8 ms, TE = 1.2 ms, slice thickness = 8 mm, flip angle 50°, inplane resolution = 1.4 × 1.4 mm²) around 20–30 min from the time of the initial contrast injection. For LGE sequences, the inversion time was selected using an inversion time scout scan to optimally null the normal myocardium.

Patients were advised to refrain from the use of caffeine containing products for 24 h prior to the study. A baseline ECG was obtained prior to start of the exam. Heart rate and blood pressure were recorded at the start of the CMR and monitored during the test. A pediatric cardiologist, pediatric radiologist, a technologist and a nurse were present during induction of stress. Following administration of regadenoson, vitals were recorded every minute for 5 min, every 5 min for the next 20 min and then every 10–15 min until an hour from the time of stress.

Adverse events were documented and classified as major or minor. Major adverse events would result in need for termination of the study, and included arrhythmia, hypotension, atrioventricular block and bronchoconstriction. Minor adverse events would not typically result in termination of the test and included anxiety, chest pain, tingling in the limbs and gastrointestinal effects such as nausea and abdominal pain. Patients were monitored for an hour after the completion of the study prior to discharge.

Data are reported as means ± standard deviations or median (interquartile range, IQR) as appropriate. Intraclass correlation coefficient was calculated to assess inter-observer variability in MPRI and FT strain in a subset of patients. To assess significant group differences, an unpaired t-test (for normally distributed data) or a Wilcoxon rank-sum test (for non-normally distributed data) was performed between cases and controls. Segmental differences in the CMR parametric data were displayed as AHA 16-segment maps. Additionally, Pearson’s correlation coefficients (r) were obtained to determine the relationships between clinical variables and CMR structure/function variables. Univariate and multivariable logistic regression model was used to determine association of CMR parameters with the rejection score. Significance was determined by p < 0.05.

Ventricular volumes were smaller and cardiac output was higher in PHT compared to controls (Table 2). Baseline heart rate (prior to administration of regadenoson) was also higher in PHT. Global pre-contrast T1 and ECV were significantly higher and pre-contrast T1 was higher in all 16 AHA segments (Fig. 2) compared to controls. Nine of the 16 (56%) AHA segments (basal and mid-ventricular septum, basal and mid-ventricular anterolateral wall and the mid-ventricular and apical anterior walls of the LV) showed significantly elevated ECV as well. These segments corresponded mostly with left anterior descending coronary artery (LAD) territories with some overlap with right coronary artery (RCA) distribution. Global T2 values were not significantly different in PHT compared to controls, however 4/16 (25%) of the AHA segments (mid-ventricular septum, mid-anterior and mid-inferior walls) did show significantly elevated T2 in PHT. These segments corresponded to both LAD and RCA territories, similar to ECV. LGE was present in three (11.5%) PHT and none of the controls. There was no particular coronary distribution pattern, with LGE demonstrated in the mid-myocardial apical inferior wall (n = 2), inferior septal hinge point (n = 1) and subepicardial to mid-myocardial circumferential distribution from base to apex (n = 1).

GLS rate and GCS rate were higher in this cohort of PHT (Table 3). All mid-ventricular and apical segments and most of the basal segments had significantly higher systolic circumferential strain rate (Fig. 3). A Bland–Altman analysis between two blinded reviewers showed agreement with no significant bias in GLS (bias: 0.8 ± 2.1, 95% LOA − 3.3 to 4.8%; p = 0.45) and GCS measurement (Bias: − 0.6 ± 3.4, 95% LOA − 7.3 to 6.1%; p = 0.71) (Fig. 4a, b).

Stress perfusion was reliably performed in 24/26 (92.3%) CMR exams. Deviation from standard protocol with exclusion of stress perfusion testing occurred because of the clinical situation (1 with active rejection) or patient factors (1 for limited scan due to difficulty lying in scanner). There were no major adverse reactions and 1/24 (4.1%) PHT reported a minor adverse reaction with nausea and gastrointestinal discomfort. Two of 16 (12.5%) perfusion controls had minor adverse reactions including limb numbness and chest discomfort that self-resolved. Qualitative inducible subendocardial defects were noted in 3/12 (25%) of PHT with known CAV: 1/1 (100%) with CAV2 and 2/11 (18%) with CAV1. These qualitative perfusion defects overlapped the CAV coronary territory in 3/3 (100%) PHT with positive findings. Compared to the control group, PHT were found to have a significantly lower global MPRI (PHT mean: 0.69 ± − 0.21; PHT control mean: 0.94 ± 0.22; p < 0.001) with no significant differences between CAV 0 versus CAV 1. MPRI did not correlate with ischemic time during heart transplant surgery. A Bland–Altman analysis between two blinded reviewers showed agreement with no significant bias in MPRI measurement (Bias: − 0.04 ± 0.15, 95% LOA − 0.33 to 0.25%; p = 0.44) (Fig. 4c).

There were significant correlations between tissue structure and cardiac function noted in PHT. Increased global myocardial T2 value was significantly associated with lower LVEF (r = − 0.57, p = 0.005), reduced GCS (r = − 0.73, p < 0.001) and lower GLS (r = − 0.49, p = 0.03) (Fig. 5). In addition, there were significant relationships between tissue parameters: T2 with ECV (r = 0.68, p < 0.001), and T2 with T1 (r = 0.68, p < 0.001).

With increasing years since transplant, there was a significant increase in the global pre-contrast T1 values (r = 0.45, p = 0.03) and ECV values (r = 0.46, p = 0.02) in PHT recipients. There was higher GCS noted in those who were further out in years post-transplant (r = 0.45, p = 0.025). No correlations were noted between age and T2, T1 and ECV or MPRI.

There were modest correlations between higher rejection scores and elevated global T1 (r = 0.38, p = 0.05), T2 (r = 0.39, p = 0.058) and ECV (r = 0.68, p < 0.001). Significant structural differences were also noted between CAV 0 versus CAV 1: CAV 1 showed higher global T1: (1037 ± 28 vs 1072 ± 53 ms; p = 0.04), higher global ECV (25.4 ± 1.9 vs 28.9 ± 4.6% p = 0.02) and higher maximum segmental T2 (52.3 ± 4.0 vs 56.5 ± 4.8; p = 0.04) (Fig. 6).

As has been shown in previous adult studies [51], we found that myocardial native T1 and ECV were higher even in our relatively healthy PHT cohort. These differences were noted at the segmental level with elevated native T1 values in PHT in all AHA segments and elevated ECV in greater than 50% of the segments, in no specific coronary distribution pattern. Additionally, in this cross-sectional cohort we noted that with increasing duration since transplant, the T1 and ECV values rose, suggesting ongoing myocardial remodeling with time that may reflect progressive chronic graft failure. While global T2 values were not different between PHT and healthy controls, segmental differences were noted.

Many adult studies have shown that elevated native T1, T2 and ECV may be markers of acute rejection and CAV [21, 23‐26, 28, 49]. One recent pediatric study showed that there may be trends in T2 that indicate allograft rejection [43]. In our cohort, only one subject had acute rejection at the time of CMR and that patient showed significant abnormalities in all three parametric variables (T1 = 1193 ms, ECV = 32.1% and T2 = 60 ms), which corroborates studies in which T2 signal has been indicated as a potential marker of acute rejection [23, 24, 26, 43]. In addition, T2 and ECV have shown to be independently associated with cardiac events (cardiac death, myocardial infarction, coronary revascularization, and heart failure hospitalization) and noncardiac death or hospitalization in adult patients [32]. We did not have longitudinal follow up of our PHT recipients yet and suggest that more studies with CMR performed at the time of acute rejection and significant CAV diagnosis are needed.

A small subset of our PHT cohort (11.5%) showed positive LGE. Studies in adult transplant recipients demonstrate that presence and severity of LGE predict long-term outcomes such as mortality and major adverse cardiac events [56, 57], and are more common with higher CAV grades [58], however, the role of LGE imaging in PHT surveillance is not established. In a small pediatric study [39], prognostic value of LGE in determining cellular rejection was not seen, although the study was likely underpowered to detect the outcome. Similar to our cohort, LGE is known to be fairly common in adult heart transplant recipients at the time of their first CMR [31] and appears to be relatively stable over time.

CMR stress perfusion testing was reliably and safely performed in our study. Only one PHT in our cohort had a minor adverse event and no serious adverse events were noted. This low incidence is similar to that reported in other pediatric cohorts [46]. Three of 12 PHT with known CAV showed abnormalities on stress perfusion testing, suggesting that qualitative CMR perfusion assessment has a high positive predictive value (100%) but a low negative predictive value (61%) for detection of early CAV. In other words, qualitative perfusion assessment with its associated technical challenges in the pediatric population may not be an ideal rule in or rule out test for CAV detection. Given the diffuse and heterogenous nature of CAV, semi-quantitative MPRI and GLS have been described as promising potential markers of CAV in late-stage heart transplantation [34]. In our cohort, compared to controls, MPRI was decreased in PHT even without CAV suggesting that perfusion reserve is likely abnormal in most PHT recipients. These differences were noted despite the lack of a normal control cohort as described above. MPRI was not different between angiographically determined CAV 0 and CAV 1, however, this could be due to the relatively subjective nature of angiographic interpretation of early CAV and/or the absence of higher-grade CAV subjects in our cohort.

The safety profile, clinical findings and good reproducibility of MPRI measurements shown across observers, should motivate further studies to better quantify myocardial perfusion as an indicator of CAV in PHT.

Despite similar body sizes between PHT and age-matched controls, ventricular volumes in PHT were smaller even though cardiac output was overall higher. This has been previously described in pediatric cohorts [59, 60] in which a wide range of size mismatch between donors and recipients is necessary and acceptable [61]. These observed differences may be a consequence of initial size mismatching and chronic adaptive or maladaptive remodeling that occurs over time.

Echocardiography-derived strain and CMR derived myocardial velocities are often abnormal during an acute rejection episode or with cumulative history of repetitive rejections [11‐13, 28, 37, 37, 40, 41, 62, 63]. Miller et al. [30] have reported an association of reduced GCS by CMR tagging with clinically significant ACR (grade 2 R or higher) in adult heart transplant recipients. Grotenhuis et al. [40] showed similar results in a pediatric cohort. Controversy exists, however, regarding which strain dimension is affected with rejection, with several echocardiography-based studies showing a reduction in GLS with preservation of GCS. Several studies have shown abnormal contractility following heart transplantation even in the absence of significant rejection [64, 65] and altered systolic deformation patterns in healthy transplanted hearts compared to native hearts [52]. LV echocardiography-derived strain parameters have also been shown to change over time in healthy PHT recipients [55]. Overall, patterns of strain are not well understood in those with acute or chronic rejection or in those with healthy transplanted hearts.

In our cohort, we noted that while GLS and GCS were not significantly different between PHT and healthy controls, there were significant differences in strain rate. We also observed higher GCS in PHT further from transplant. These differences could be due to a variety of factors including differences in age, heart rate, cardiac size, donor-recipient size mismatch or loading conditions. Importantly, the mechanistic and clinical implications are unclear given the cross-sectional nature of our study design and prospective, longitudinal studies could answer some of these questions and clarify significance.