Cardiac magnetic resonance feature tracking of the right ventricle in convalescent Kawasaki disease in a large single center

Abstract Background The changes in right ventricular (RV) contractility of Kawasaki disease (KD) still remain unclear. Hypothesis We aimed to determine whether RV systolic dysfunction can be detected by cardiac magnetic resonance (CMR) feature tracking and to find its association with coronary artery lesions (aneurysm, thrombosis and stenosis). Methods Peak systolic myocardial longitudinal, radial and circumferential strain and the strain rate (RVSL, RVSR, RVSC, RVSRL, RVSRR and RVSRC) in the global RV and three levels (basal, middle and apical) were measured in 66 patients with convalescent KD. A total of 20 controls were included. Comparisons were made with controls and among KD subgroups divided with coronary artery lesions. Results RVSC (−10.575% vs. −10.760%), RVSL (−18.150% vs. −18.712%) and RVSRC (−0.815/s vs. −0.924/s) were slightly lower in KD group without significant difference. All the strain and strain rate presented lowest in the basal level. In subgroup comparison, lower RVSL and RVSRL were observed in the giant coronary artery aneurysm (CAA) group; RVSR (15.844% vs. 16.897%), RVSRR (1.245/s vs. 1.322/s) and RVSRC (−0.715/s vs. −0.895/s) were lower in thrombosed group; RVSRL (−1.27/s vs. −1.503/s) were lower in stenosis group. All the comparison in subgroups did not reach significant difference. From the analysis of receiver operating characteristic curve, RVSRL had a better ability to identify KD with giant CAA and stenosis. For the identification of thrombosis, RVSRC had a better ability. Conclusions Lower strain and strain rates of RV were detected in convalescent KD. More pronounced in those with persisting coronary artery lesions.


| INTRODUCTION
As a systemic vasculitis of medium sized vessels, Kawasaki disease (KD) is the most common cause of acquired heart disease among children in the developed world. It was reported that coronary artery lesions (CALs) is the dominant common complication, mainly including coronary artery aneurysm (CAA), thrombosis and stenosis. The incidence of CAA is about 5% and huge CAA can usually form thrombosis and ischemic cardiac disease. In the chronic stage, the reconstruction of coronary artery continues and induces stenosis and micro-flow impairment.
Myocarditis, the most common non-CAL, accounts for 50%-70% of patients in the acute phase, followed by myocardial fibrosis. For KD, the routine cardiac function often displays normal with routine measurements, 1 except for patients with ischemic cardiomyopathy and severe CALs. 2-4 However, lower strain has both been detected in the acute and convalescent phase of KD. [5][6][7] The strain and strain rate are the main indices in the paper. This methodology has the potential to elucidate subtle impairment in myocardial mechanics that cannot be demonstrated by routine cardiac imaging modality.
Compared with echocardiography, cardiac MR feature tracking (CMR-FT) allows for the quantitative assessment of regional and global myocardial mechanic deformation that is related to myocarditis and myocardial fibrosis. Myocarditis has been found in the entire heart, and the fibrosis continues to the convalescent stage. Previous studies have summarized the left ventricular (LV) myocardial functional impairments in the acute and chronic phase in detail. However, the abnormal mechanics in right ventricle (RV) of KD patients still remains undefined. This is the first study to apply CMR-FT in KD to examine mechanic deformation in RV and to discover its association with CALs. Our goal was to define the variation of RV myocardial deformation within KD cohorts and to assess if variation in myocardial deformation may help identify subgroups with higher risk. A combination of 16-elements phased array surface coil and spine coil were used. Initially, a survey examination in three orthogoal planes was performed to localization of the heart, and then an interactive sequence was performed to define patient-specific relevant orientations (including axial, coronal, sagittal, two-chamber, four-chamber and short axial views). Electrocardiography (ECG)-gated two-dimensional steady-

| CMR analysis
The analysis of CMR-FT was performed by axial short and longitude cine images with commercial software Circle (CMR42, version 5.6.5, Cardiovascular Imaging Inc., Calgary, Alberta, Canada), as previously described. 7 Global and regional strain analyses of the RV free wall in three directions (longitudinal, circumferential and radial) were mea- All the CMR studies were analyzed by 1 radiologist who was blinded to the patient clinical data. In addition, random samples of 20 cases were re-evaluated by the same radiologist to assess intra-observer and by another investigator for inter-observer agreement.
There exist no consensus normal values for myocardial deformation in children, so we compared the KD group with the control group. 9

| Statistical analysis
All data were reported as mean ± SD. Unpaired Student's t-test and oneway ANOVA were adopted to compare the strain and strain rate between subgroups. Diagnostic accuracy of CMR parameters was evaluated through receiver operating characteristic (ROC) curve analysis. For area, under the curve (AUC), a value of 0.9-1.0 was considered excellent, 0.75-0.9 marked good, 0.6-0.75 stood for moderate and 0.5-0.6 was poor. Best cut-off values, sensitivity and specificity were derived from ROC curves. Intra-and inter-observer variability was assessed using Cronbach's α. Statistical analysis was performed with SPSS software (SPSS Inc., Chicago, IL, version 26.0).
A p-value of ≤.05 was regarded statistically significant.

| Demographic information of patients and controls
In Table 1

| CMR-FT of RV in KD group versus controls
Compared with controls, RVSC, RVSL and RVSRC were slightly lower in KD group ( Table 2). The above differences did not reach statistical significance. Among the three levels (apical, middle and basal), all the lowest strain and strain rates were observed in the basal segment, which reached statistical significance (Table 3).  (Table 2).

| CMR-FT comparisons in KD subgroups
When we compared patients suffering from thrombi (n = 13) with those without thrombi (n = 53), RVSR, RVSRR and RVSRC were lower in thrombosed group, but did not reach statistical difference.
When patients catching with stenosis (n = 4) were compared with those without stenosis (n = 62), RVSRL were lower in stenosis group, but did not reach statistical significant difference.

| ROC analysis
ROC curve revealed that RVSRL had a better ability to identify KD from giant CAA and stenosis. The cut-off value for RVSRL was  Table S1.

| Inter-and intra-observer agreements
Twenty subjects were randomly selected from KD group to assess interand intra-observer agreement of CMR-FT analysis. We evaluated the intra-observer and inter-observer agreement by Cronbach's α. If the value is higher than 0.8, the reliability is high. If the value is between 0.7 and 0.8, the reliability is good. If the value is between 0.6 and 0.7, the reliability is acceptable. If the value is less than 0.6, the reliability is poor. According to CMR-based strain parameters, we identified the good reliability in the inter-and intra-observer agreement, except for the inter-observer agreement of RVSRR (Cronbach's α = 0.651).

| DISCUSSION
As a systemic vasculitis of medium sized vessels, perhaps the pericardium, myocardium, valves and the coronary arteries were inflamed in F I G U R E 2 A, A 2-year-old girl with a giant CAA (arrow) in the proximal segment of LAD. B, A giant aneurysm in the proximal segment of RCA associated with mural thrombus (arrow), presenting higher signal on 3D-SSPF images. C, Multiple CAAs were found along the whole RCA and the stenosis could be found between the proximal giant CAAs (arrow). CAA: coronary artery aneurysm; LAD: left anterior descending artery; RCA: right coronary artery; SSPF: steady-state free precession KD. 10 Myocardial inflammation is the most common noncoronary cardiac abnormalities and has been found in about 50%-70% patients in acute phase. Biopsy studies suggested that the myocarditis ranged from the regional segments to the entire heart. 11 It has been proved that the strain analysis was superior to routine cardiac function for evaluation of subclinical LV systolic dysfunction in many diseases, including KD. 13,14 It was obvious that in KD, LV myocardial strain and strain rate were reduced despite the normal LV systolic function. 1,15,16 RV myocardial strain, and both global and segmental strains, have also been found impaired in many clinical scenarios, such as congenital heart disease, pulmonary hypertension and heart failure. 17,18 In a research of 364 patients, the strain analysis by CMR-FT was significantly associated with future cardiac events, especially the RV global radial strain and LV global transverse strain 19 .
In researches of repaired TOF patients, reduced strain values were related with cardiac function and functional capacity of the cardiopulmonary exercise test. 20 RCA is often affected in KD, but the researches of RV contractility are limited. Based on a recent data, the global and regional RV strain values of KD patients were decreased compared with control subjects. 7 However, in another research, including 15 KD patients, there was no significant difference in the RV global longitudinal strain and strain rate. 16 Overall, the existing evidence about myocardial deformation studies of RV in KD is poor.
In our study, a considerable proportion of patients suffer from normal EF and abnormal RV strain indices, indicating that myocardial mechanic deformation impairment starts before RV cardiac dysfunction. KD patients were revealed to have lower RVSC, RVSL and RVSRC compared with age-matched normal controls. The results did not reach statistic difference, which may be caused by the preserved EF in this group. Meanwhile, we also found that the proximal segments presented more obvious strain abnormality, which was related with the CAAs that mostly existed in the proximal segments of RCA.
The paper had unraveled the non-uniform mechanics of RV, characterized by increased strain values at the apex. 21 That would also contribute to the increase of strain and strain value of the middle and apex level.
In subgroup analysis, a more pronounced decrease was shown in those with persisting CALs versus those without persisting CALs. 7 T A B L E 2 Comparison of RV strain and strain rate between healthy volunteers, KD patients and subgroups  So the impairment of longitude strain was the most obvious one in KD. Meanwhile, the systolic strain rate was regarded as the a more robust parameter, reflecting myocardial function, rather than strain, 31-33 because strain may be affected by interstitial inflammation/edema, while strain rate may not be affected unless that actual myocyte injury occurs. The results of this study offer additional insights into the value of the early detection of RV contractility deterioration as an independent predictor of more severe CALs and myocardial fibrosis.
For echocardiography, currently, the qualitative evaluation of RV cardiac function is a diagnostic challenge due to the unusual shape and uneven contractility pattern. 34 CMR has been the gold standard method in RV function evaluation, and FT-CMR can add value for RV contractility abnormalities. [34][35][36][37][38] Reproducible and repeatable quantification of RV strain indices are also vital to monitor patients. 39 The analysis of RV function by FT-CMR was feasible and reproducible. 31,40 In our research, the reproducibility was good, as reported in other papers. 41 Moreover, the CVI 42 software also proved excellent intraand inter-observer reproducibility for RV regional strain values. 31,41,42 The good reproducibility, inter-reader and intra-reader variability, make FT-CMR a robust tool to support decision-making and followup. 43 Meanwhile, CMR can afford the assessment of myocardial perfusion, scarring in a single examination, as well as the evaluation of regional wall motion. 33,44 Although the number of patients in this study was large enough, it still existed several limitations. (1) We did not include KD children without CALs, so we missed the data of this subgroup.
(2) Because of the difficulty of assessment to the thin-walled RV, the evaluation of RV myocardial perfusion and scarring was not conducted. (3) The accuracy of CMR and echocardiography of strain analysis was not compared. Therefore, future studies should be conducted to solve these problems.

| CONCLUSION
In KD, a reduction in strain of RV was observed in the convalescent stage, detecting subclinical functional abnormalities. A more pronounced decrease was displayed in those with persisting CALs.
These results provide new insight into the effects of strain analysis on ventricular mechanics and may prompt modifications in management.

ACKNOWLEDGMENTS
The National Key Research and Development Program of China (2016YFC1000500).

CONFLICT OF INTEREST
The authors declare that they have no competing interests.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request

ETHICS APPROVAL
Informed consent forms were signed by the parents and ethics approval was approved by the institutional review boards of the Children's Hospital of Fudan University.