Cardiac Magnetic Resonance Imaging Non-Invasively Detects Rejection in Pediatric Heart Transplant Recipients

Endomyocardial biopsy (EMB) is commonly used as surveillance for acute rejection (AR) in pediatric heart transplant recipients (PHTx). Given the potential morbidity associated with cardiac catheterization, a non-invasive tool to either diagnose or screen for AR in PHTx would be beneficial. Non-invasive imaging may also allow for more frequent screening at a lower cost.

Cardiac magnetic resonance (CMR) parametric mapping has potential to serve as this tool. Native T1 mapping detects myocardial edema and fibrosis and T2 mapping is sensitive for myocardial edema. Extracellular volume (ECV) mapping, calculated from native and post-contrast T1 maps and a hematocrit, is sensitive for extracellular matrix expansion, including edema. Adult studies have demonstrated the potential of these techniques for diagnosing AR.^{1, 2} However, there have been only limited pediatric data.³ We hypothesized that native T1, T2, and ECV mapping would associate with AR or no AR in PHTx.

The institutional review board approved this study and participants, or their parents/guardians, signed appropriate consents/assents. PHTx were prospectively enrolled at two sites at the time of surveillance EMB or EMB performed for clinical concerns for AR. AR was defined as a clinical change or positive EMB requiring intensification of immunosuppression. Diagnosis of AR was made by the treating cardiologist blinded to CMR results; all AR cases were reviewed by a PHTx cardiologist with >30 years of experience blinded to CMR results. Given the relatively high incidence of EMB negative AR in PHTx, and the importance of better understanding these patients, these subjects were included in the AR group. This improved generalizability, though the potential for false positive subjects would decrease the difference between groups.

Subjects were excluded if they were within 6 months from PHTx, were unable to undergo CMR without sedation/anesthesia, had known CAV, or CMR with contrast was contraindicated. Subjects were only included once in the study; subjects enrolled initially without AR (NoAR) and subsequently with AR were removed from the NoAR group.

CMR included standard volumetrics, late gadolinium enhancement (LGE), native T1, T2, and extracellular volume (ECV) mapping. Parametric mapping was performed in 4 short axis slices at the base and mid-left ventricle (LV). LGE quantification used a 5-standard deviation technique. Qmaps (Medis, Leiden, The Netherlands) was used to analyze parametric mapping and Qstrain (Medis) to calculate global longitudinal and circumferential strain (GLS and GCS). Native T1 and T2 Z-scores for each magnet were calculated based on normal values for the base, mid, and global LV (average of base and mid).

Wilcoxon rank-sum (continuous variables) and Chi-square (categorical variables) tests were used to determine differences between AR and NoAR. Area under the curve (AUC) was estimated using separate univariate logistic regression models. Data are available from the corresponding author upon reasonable request.

Thirty subjects were enrolled with a median age of 16 years old (IQR13,18). There were 12 AR and 18 NoAR subjects. All AR episodes were unique. The median absolute time between EMB and CMR was 1 day (IQR1,4). There was no evidence of demographic differences between AR and NoAR (Table 1).

Table 1.

Comparison of demographic and cardiac magnetic resonance data between patients with and without acute rejection.

	Acute Rejection (N=12) Median (IQR) or % (N)	No Rejection (N=21) Median (IQR) or % (N)	P-value
Age (years)	16 (11,19)	16 (13,17)	0.841*
	range (5-20)	range (8-19)
Height (cm)	157 (144,166)	164 (152,173)	0.340*
Weight (kg)	65 (44,84)	63 (47,75)	0.525*
Male gender	42% (N=5)	56% (N=10)	0.362†
Heart rate (bpm)	100 (85,117)	95 (80,105)	0.418*
Time from transplant (years)	5.7 (2.6,12.5)	2.1 (1.0, 4.0)	0.072*
Prior Rejection	42% (N=5)	17% (N=3)	0.092†
LVEF (%)	54 (43,56)	60 (59,64)	0.004 *
LVEDVi (ml/m2)	75 (65,81)	67 (65,80)	0.719*
LVESVi (ml/m2)	37 (25,43)	27 (22,33)	0.107*
Cardiac index (L/min/m2)	3.7 (3.2,4.1)	3.8 (3.4,4.2)	0.807*
RVEF (%)	51 (44,57)	58 (53,63)	0.041 *
LV mass indexed (g/m2)	38 (34,50)	36 (30,41)	0.271*
LGE present	67% (N=8)	56% (N=10)	0.858†
LGE 5SD (%)	12.3 (10,18)	13.7 (6.4,20.2)	0.887*
GLS (%)	−14.6 (−20.4,−11.5)	−22.1 (−24.3,−19.4)	0.011 *
GCS (%)	−26.3 (−31.1,−22.6)	−31.2 (−34.4,−28.8)	0.016 *
Base ECV (%)	32.5 (31.6,36.8)	26.1 (23.6,32.0)	0.007 *
Mid ECV (%)	35.3 (32.0,39.7)	29.2 (26.7,32.8)	0.009 *
Global ECV (%)	34.9 (31.7,38.4)	27.2 (24.5,32.9)	0.004 *
Base Native T1 Z-score	3.6 (2.0,6.4)	2.5 (1.3,3.9)	0.14*
Mid Native T1 Z-score	2.8 (1.8,5.0)	1.8 (0.9,3.1)	0.13*
Global Native T1 Z-score	3.3 (2.1,5.9)	2.3 (1.1,3.8)	0.08*
Base T2 Z-score	4.2 (1.9,5.9)	0.2 (−0.5,2.3)	0.005 *
Mid T2 Z-score	3.1 (1.2,6.0)	0.9 (−0.3,2.0)	0.01 *
Global T2 Z-score	3.7 (1.8,6.8)	0.6 (−0.3,2.9)	0.009 *

IQR (interquartile range); LVEF (left ventricular ejection fraction); LVEDVi (indexed left ventricular end diastolic volume); LVESVi (indexed left ventricular end systolic volume); RVEF (right ventricular ejection fraction); LGE (late gadolinium enhancement); LGE 5SD (percent LGE using the 5-standard deviation method); GLS (global longitudinal strain); GCS (global circumferential strain); ECV (extracellular volume).

^*Statistical analysis for continuous variables performed using a Wilcoxon rank-sum.

^†Statistical analysis for categorical variables performed using a Chi-square.

LV ejection fraction (LVEF) and right ventricular ejection fraction (RVEF) were lower in AR (Table 1). Subjects with and without AR had a high prevalence of LGE, but there was no difference in presence or percentage of LGE between groups. T2 Z-scores and ECV were increased in AR vs NoAR; the difference in native T1 Z-scores did not reach statistical significance (Table 1). GLS was worse in AR. The AUC for variables of interest were: base ECV=0.80, mid ECV=0.79, average ECV=0.82, T2 Z-base=0.81, T2 Z-mid=0.78, T2 Z-average=0.79, GLS=0.79, GCS=0.76.

This manuscript demonstrates that T2, ECV, and myocardial strain are increased in PHTx recipients with AR. This is the first manuscript of which we are aware to evaluate a combination of parametric mapping and strain differences in pediatric subjects with AR. It is also the first pediatric study to use Z-scores for T1 and T2-mapping, which are necessary for generalizability between magnets.

Non-invasive myocardial assessment using CMR has promise to decrease surveillance risk. In older children not requiring anesthesia, CMR can be performed more frequently, increasing the chances of early detection and therapy for AR. Although there was overlap of values between patients with and without AR, modeling that includes multiple variables has potential to help diagnose AR.

As some subjects with AR underwent CMR after EMB, areas of inflammation may have already begun to normalize, thus decreasing the detectable difference between groups. Children under 8 years old were excluded due to the risks of sedation/anesthesia; therefore, the utility of these techniques in smaller children is unclear.

ECV, T2, and GLS can non-invasively distinguish between PHTx with and without AR. As this is a small study, this work is hypothesis generating and future, multicenter studies are necessary to confirm these results and explore whether CMR can decrease the number of surveillance EMBs in PHTx.