Multiparametric cardiac magnetic resonance imaging in pediatric and adolescent patients with acute myocarditis

Background The diagnostic value of cardiac magnetic resonance imaging (MRI) employing the 2018 Lake Louise criteria in pediatric and adolescent patients with acute myocarditis is undefined. Objective To evaluate the diagnostic value of the Lake Louise criteria in pediatric and adolescent patients with suspected acute myocarditis and to show the utility of cardiac MRI for follow-up in this patient cohort. Materials and methods Forty-three patients (age range: 8–21 years) with suspected acute myocarditis and 13 control patients who underwent cardiac MRI were retrospectively analyzed. T2-weighted and late gadolinium enhancement imaging were performed in all patients. T1 and T2 mapping were available in 26/43 patients (60%). The Lake Louise criteria were assessed. In 27/43 patients (63%), cardiac MRI follow-up was available. Receiver operating characteristic analysis, Pearson’s correlation coefficient and paired Student’s t-test were used for statistical analysis. Results In the total cohort, the Lake Louise criteria achieved a sensitivity of 86% (95% confidence interval [CI]: 72–95%) and a specificity of 100% (95% CI: 79–100%) for the diagnosis of acute myocarditis. In the subgroup of patients with available mapping parameters, the diagnostic performance of the Lake Louise criteria was higher when mapping parameters were implemented into the score (area under the receiver operating characteristic curve: 0.944 vs. 0.870; P=0.033). T2 relaxation times were higher in patients with admission to the intermediate care unit and were associated with the length of intermediate care unit stay (r=0.879, P=0.049). Cardiac MRI markers of active inflammation decreased on follow-up examinations (e.g., T1 relaxation times: 1,032±39 ms vs. 975±33 ms, P<0.001; T2 relaxation times: 58±5 ms vs. 54±5 ms, P=0.003). Conclusion The Lake Louise criteria have a high diagnostic performance for the diagnosis of acute myocarditis and are a valuable tool for follow-up in pediatric and adolescent patients. The mapping techniques enhance the diagnostic performance of the 2018 Lake Louise criteria. Supplementary Information The online version contains supplementary material available at 10.1007/s00247-021-05169-7.


Introduction
Acute myocarditis is an inflammatory disorder of the myocardium primarily caused by infections, autoimmune processes, systemic diseases, drugs or toxins and can affect both adults and children [1]. Because subclinical manifestation is common, the true incidence of acute myocarditis in children is unknown. Estimated incidence rates range between 1 and 2 per 100,000 children [2,3]. Males seem to be affected more often [2,4]. Although incidence of acute myocarditis is relatively low in children, it accounts for about one-third of childhood dilated cardiomyopathies [5] and is related to sudden cardiac death in young patients [6,7]. Diagnosis of myocarditis is challenging because of the wide spectrum of clinical presentation. Furthermore, children particularly may show atypical manifestations [8]. The clinical course is heterogeneous, varying from fast convalescence to delayed recovery or chronic heart failure and, rarely, serious longterm morbidity [8,9]. Due to its invasiveness and possible sampling error, endomyocardial biopsythe diagnostic reference standardis used infrequently in clinical practice. Nowadays, cardiac magnetic resonance imaging (MRI) plays an important role for diagnostic work-up in patients with suggested myocarditis. Cardiac MRI is a noninvasive and radiation-free technique that is suitable for pediatric and adolescent patients and can be of particular value to differentiate acute myocarditis from myocardial infarction, primary cardiomyopathy or congenital heart disease to spare young patients from coronary angiography [10,11]. The Lake Louise criteria were established in 2009 for the diagnosis of myocarditis by cardiac MRI. They include three aspects of myocardial inflammation (edema, hyperemia and necrosis) in a two-out-of-three approach (high signal intensities on T2-weighted images, early gadolinium enhancement and late gadolinium enhancement) [12]. In 2018, the implementation of quantitative mapping techniques led to a revision of the original Lake Louise criteria [13][14][15]. According to the 2018 Lake Louise criteria, diagnosis of myocarditis can be made when two main criteria are fulfilled, at least one T1-based criterion (increased myocardial T1 relaxation time, increased extracellular volume fraction or positive late gadolinium enhancement) and one T2-based criterion (increased myocardial T2 relaxation time or visual myocardial edema/increased T2 signal intensity ratio) [16]. Previous studies evaluated the performance of cardiac MRI diagnostic parameters primary in adults, but the diagnostic value of quantitative parameters remains undetermined in children and adolescents.
The purpose of this cardiac MRI study was to evaluate (1) the diagnostic performance of the 2018 Lake Louise criteria for acute myocarditis in pediatric and adolescent patients and (2) to assess the value of cardiac MRI for follow-up of the disease.

Materials and methods
This retrospective, case-control study was approved by the institutional ethics committee and the requirement for written informed consent was waived. Patients with clinically defined acute myocarditis and control patients were included in this study. The department's registry was searched for children and adolescents (ages 2 to 21 years old, according to the American Academy of Pediatrics [17]) who had undergone cardiac MRI for the evaluation of acute myocarditis on the same clinical whole-body MRI system between January 2014 and April 2020. Patients with preexisting cardiovascular disease were excluded. The diagnosis of myocarditis was made according to current criteria for clinically suspected myocarditis as recommended by the European Society of Cardiology Working Group on Myocardial and Pericardial Diseases [8]. The main criteria for clinical diagnosis were acute chest pain, signs of acute myocardial injury (electrocardiogram [ECG] changes and/or elevated troponin), and a constellation of signs associated with infection (elevated leucocytes and/or C-reactive protein and/or confirmed infectious disease). This clinical validation approach was used for the diagnosis of acute myocarditis in this study as previously reported [14][15][16]18]. Clinical evidence was the reference standard against which the diagnostic performance of cardiac MRI parameters was tested. As a control group, outpatients were retrospectively reviewed for those who had been referred for nonspecific cardiac symptoms or thoracic discomfort to rule out structural heart disease. All control patients had an unremarkable past medical history of cardiovascular disease. Electrocardiographic and echocardiographic results were unremarkable and no cardiac risk factors were present. All control participants had normal cardiac MRI results without structural abnormalities.

Cardiac magnetic resonance
All investigations were performed on a clinical whole-body MRI system (Ingenia 1.5 tesla; Philips Healthcare, Best, the Netherlands). For signal reception, a 32-channel torso coil with digital interface was used. ECG-gated steady state free precession cine images were acquired in short-axis, fourchamber and two-chamber views. T2-weighted short tau inversion recovery (STIR) sequences were acquired in shortaxis, two-chamber and transversal views to visualize myocardial edema and calculate the T2 signal intensity ratio. Late gadolinium enhancement imaging was based on segmented inversion recovery gradient echo sequences in short-axis, two-chamber and four-chamber orientation. Cardiac MRI protocol consisted of myocardial T1 and T2 mapping acquired in end-diastole in apical, midventricular and basal short-axis views. Post-contrast myocardial T1 maps were performed 10 min after contrast injection in the same orientation. T1 mapping was based on a standard 3(3)3(3)5 modified Look-Locker inversion recovery (MOLLI) acquisition scheme [19]. A 6-echo gradient spin echo (GraSE) sequence allowed acquisition of T2 mapping [20]. For contrast enhancement, a single bolus of 0.2 mmol/kg body weight of gadobutrol (Gadovist; Bayer Healthcare, Berlin, Germany) was given. For extracellular volume fraction estimation, the hematocrit blood levels on the cardiac MRI day were available. A detailed description of the cardiac MRI sequence parameters is provided in Online Supplementary Material 1.

Cardiac image analysis
Images were analyzed by two radiologists (J.A.L. and A.I., with 8 and 3 years of experience in cardiac MRI, respectively) using dedicated software (IntelliSpace Portal version 10.1; Philips Medical System, Best, the Netherlands). Readers were blinded to the clinical information. All volumes and masses were indexed to body surface area using the Mosteller method. Visual analysis of T2 STIR and late gadolinium enhancement images was evaluated separately by consensus agreement of the two readers for focal myocardial hyperintensity as markers for myocardial edema and necrosis, respectively. Semiquantitative analysis allowed for determination of quantitative T2 signal intensity ratio (global myocardial edema) [12,14]. Myocardial relaxation maps were motion corrected using FEIR (fast elastic image registration) by dedicated software (IntelliSpace Portal version 10.1, Philips Medical System). Myocardial T1 and T2 relaxation times and hematocrit corrected extracellular volume fraction values (using preand post-contrast T1 values) were calculated using a segmental approach as previously described [14][15][16]. The 2018 Lake Louise criteria were applied as recommended [13].

Statistical analysis
Prism (version 8.4.3; GraphPad Software, San Diego, CA), SPSS Statistics (version 26; IBM Corp., Armonk, NY) and MedCalc (version 18.11.3; MedCalc Software bvba, Ostend, Belgium) were used for statistical analysis. The Kolmogorov-Smirnov test was applied to assess normality. Continuous patient characteristics are presented as mean±standard deviation or as absolute frequency. Continuous variables between the two groups were compared using the Student's t-test. Dichotomous variables were compared using the Fisher exact test. For intraindividual comparisons, the paired Student's t-test and McNemar's test were used. Correlation analysis was performed using Pearson's correlation coefficient. A receiver operating characteristic (ROC) analysis was performed to calculate area under the curve (AUC). Optimal cutoff values were determined using the Youden index and sensitivities and specificities were calculated. Differences between ROC curves were tested using the DeLong method. The level of statistical significance was set to P<0.05.
T1 and T2 mapping was available in a subgroup of 26/43 patients and in all control patients. Myocardial native T1 relaxation time as a marker of acute myocardial injury was higher in patients with acute myocarditis than in healthy controls (1,031±46 ms vs. 962±17 ms; P<0.001) (Fig. 1).

Diagnostic performance
In the total cohort (n=43), the 2018 Lake Louise criteria could only be applied in a limited way because mapping parameters were not available in 17/43 patients. The application of the 2018 Lake Louise criteria in the total cohort reached a sensitivity of 37 . In this subgroup, the 2018 Lake Louise criteria yielded a significantly higher diagnostic performance (AUC: 0.944, sensitivity: 89%, specificity: 100%) compared to the modified 2018 Lake Louise criteria with exclusion of mapping parameters (AUC: 0.870, sensitivity: 73%, specificity: 100%; P=0.034). Mapping parameters allowed for the diagnosis of acute myocarditis in four additional patients. Furthermore, a combined non-contrast score of native T1  Table 2 summarizes all cutoff values, sensitivities and specificities with confidence intervals for all evaluated parameters. Figure 2 visualizes the AUC values for single variables and the 2018 Lake Louise criteria of the total cohort, and a combined score of native quantitative myocardial parameters.  Table 3 and Fig. 3. Representative clinical cardiac MRI examples of patients with acute myocarditis are presented in Fig. 4 (case with a typical pattern of acute myocarditis) and Fig. 5 (case with a diffuse pattern of acute myocarditis, which was diagnosed by mapping parameters according to the 2018 Lake Louise criteria).

Discussion
In this study, we evaluated the diagnostic performance of current cardiac MRI criteria for acute myocarditis in pediatric and adolescent patients. The main finding is that the 2018 Lake Louise criteria have a high performance for the diagnosis of acute myocarditis in pediatric and adolescent patients, especially when mapping parameters are implemented. Monitoring of disease/follow-up is feasible with cardiac MRI, especially with mapping parameters. Extensive increase of T2 relaxation times is associated with a severe course of disease. Furthermore, a combined non-contrast score solely based on native mapping parameters yielded a diagnostic accuracy comparable to the 2018 Lake Louise criteria. Therefore, the use of a non-contrast cardiac MRI protocol in children may be a viable option in cases where intravenous contrast is sought to be avoided.
Several studies have investigated the diagnostic performance of myocardial T1 and T2 mapping and extracellular volume fraction in adults [14,[21][22][23][24]. However, cardiac MRI studies evaluating pediatric myocarditis are underrepresented. The published cutoff values for mapping parameters in acute myocarditis show a wide heterogeneity because myocardial T1 and T2 relaxation times depend on different centerspecific technical parameters as well as on the employed scanning sequence, field strength, manufacturer, software and type of analysis [25]. Furthermore, physiological factors may influence T1 and T2 relaxation times, e.g., sex, age, heart rate or even hydration status [26][27][28][29]. Previous studies have shown an age-related impact on myocardial T1 relaxation times. Rosmini et al. [30] found that native T1 relaxation times were slightly lower with increasing age. Plausible reasons might be myocardial lipofuscin or hemosiderin accumulation. Roy et al. [28] described higher T1 and extracellular volume fraction values in male participants of increased age, probably related to diffuse myocardial fibrosis. The determined cutoff values for the diagnosis of acute myocarditis in children and adolescents in our study differ from cutoff values for adults, which were determined in a previous study by applying the same mapping sequences and overall technical parameters [15], which might imply that appropriate cutoff values are needed for specific patient cohorts.

Lake Louise criteria
In a previous adult study, the 2018 Lake Louise criteria (sensitivity: 87.5%, specificity: 96.2%) outperformed the original 2009 Lake Louise criteria (sensitivity: 72.5%, specificity: 96.2%) for diagnosis of acute myocarditis [16]. In a pediatric cardiac MRI study by Cornicelli et al. [31], only 57% of patients met the original 2009 Lake Louise criteria, but the additional use of mapping parameters yielded a higher diagnostic performance. In particular, a combined score of global native T1 and T2 relaxation times yielded a sensitivity of 83% and a specificity of 96% [31]. However, the appropriate application of the 2018 Lake Louise criteria could not be done  due to its later release. In our study, we also could confirm high diagnostic accuracy for 2018 Lake Louise criteria for a pediatric and adolescent cohort, which was significantly higher when mapping parameters were implemented into the score (sensitivity: 89% vs. 73%, specificity: 100% vs. 100%).

T1 mapping
T1 mapping as a single parameter achieved high diagnostic values according to a recent meta-analysis in adults with acute myocarditis (pooled sensitivity: 85%, pooled specificity: 86%) [32]. High sensitivity (91%) and specificity (86%) of T1 mapping were also observed in the reported pediatric cohort by Cornicelli et al. [31]. We could also show a comparable diagnostic performance of T1 mapping with an AUC of 0.873 and a slightly lower sensitivity (sensitivity: 82%, specificity: 100%) in our cohort. These results show that myocardial T1 mapping is a valuable tool for detecting acute myocarditis in a pediatric cohort. An increase of myocardial T1 values is mainly driven by intracellular or extracellular edema, hyperemia and necrosis in acute myocarditis. However, it is also increased in systemic or chronic diseases related to protein deposition, e.g., in diffuse fibrosis causes like chronic myocarditis or ischemic and nonischemic cardiomyopathies [33,34]. Therefore, myocardial T1 mapping as a single parameter is not specific for acute myocarditis and should always be interpreted in clinical context.

Extracellular volume fraction
Extracellular volume fraction was implemented as another T1-based criterion in the 2018 Lake Louise criteria [13]. It reflects the volume of cell-free heart tissue, including the intracapillary plasma volume, which is increased in acute myocarditis [33]. However, it also includes the space occupied  by the extracellular matrix, and is therefore a surrogate for myocardial fibrosis [32,33]. In our study, the diagnostic performance of extracellular volume fraction (AUC: 0.687, sensitivity: 58%, specificity: 88%) was lower compared to T1 or T2 mapping, which differs from the pediatric study of Cornicelli et al. [31]. However, a recent meta-analysis also showed lower overall performance of extracellular volume fraction compared to native T1 and T2 mapping [18]. Furthermore, previous studies could demonstrate a strong correlation between extracellular volume fraction values and histological myocardial fibrosis [33,35]. Thus, extracellular volume fraction could still be normal in the early stage of disease and probably increase in severe or chronic courses of myocarditis.

T2 mapping
While neither extracellular volume fraction nor native T1 mapping strictly represent acute myocardial injury and are also biomarkers of myocardial fibrosis, T2 mapping has shown high sensitivity for detecting myocardial edema (e.g., inflammatory or ischemic causes) and is considered specific for acute myocardial diseases [20,22]. Therefore, applying T2 mapping as a parameter of myocardial edema can help to discriminate between acute and convalescent myocarditis and also between chronic myocarditis and noninflammatory dilated cardiomyopathy [22,36,37]. For this reason, it was implemented in the 2018 Lake Louise criteria as a T2-based criterion [13]. In our study, T2 relaxation times reached the highest sensitivity (88%) of all single parameters and showed the highest diagnostic performance of all quantitative markers (AUC: 0.899), which is in line with the results of previous adult and pediatric studies [15,16,31,32]. Although T2 mapping is less dependent on field strength compared to T1 mapping, most of the above-mentioned technical and physiological limitations also apply to T2 mapping.

Non-contrast score
According to the 2018 Lake Louise criteria, the diagnosis of acute myocarditis can be principally based on native parameters only (increased native T1 and T2 relaxation times). In our cohort, a combined native score of T1 and T2 mapping showed a high diagnostic performance (AUC: 0.920, sensitivity: 85%, specificity: 100%), which was comparable to the original 2018 Lake Louise criteria (AUC: 0.944, sensitivity: 89%, specificity: 100%). Although complications regarding gadolinium-based contrast agents are rare, especially since the introduction of new-generation agents, the application has several drawbacks. Peripheral intravenous insertion in children is often a traumatic procedure and extravasation may occur. Furthermore, anaphylactic reaction and nephrogenic systemic fibrosis in end-stage renal disease are extremely rare but serious complications [38]. Given the recent literature on gadolinium retention within the brain and body of adults and children after repetitive contrast-enhanced MRI scans, longterm effects and clinical implications remain unclear [38]. Therefore, guidelines recommend cautious use of gadolinium contrast agents in children. Our results indicate that a non-contrast cardiac MRI protocol is a viable option for the diagnosis of acute myocarditis in cases where intravenous contrast is sought to be avoided. Furthermore, our follow-up results indicate that a noncontrast score can sufficiently track disease activity or therapeutic response. But late gadolinium enhancement imaging remains an essential part of diagnostics in acute myocarditis, especially for detecting focal myocarditis (which may be missed by native quantitative mapping parameters), for discriminating between ischemic and nonischemic cardiomyopathies as well as risk stratification based on visualization and quantification of myocardial scar formation [39][40][41].

Clinical correlations
Because the clinical manifestation of myocarditis varies with broad spectrums of symptoms ranging from asymptomatic courses to fulminant presentations with cardiogenic shock, the diagnosis of myocarditis might be difficult. Therefore, in clinical routine, myocarditis is often diagnosed using a multimodal approach with clinical, laboratory, imaging-based and pathological parameters. Our results show correlations between quantitative parameters and disease severity. T1 and T2 values were higher in patients with reduced LVEF. Patients requiring intermediate care unit treatment presented with higher T1 and T2 relaxation times, but without significant differences in extracellular volume fraction, LVEF or troponin level, when compared to patients not requiring intermediate care unit treatment. Notably, T2 relaxation times were directly associated with the duration of intermediate care unit treatment and the extent of systemic inflammation (C-reactive protein, white blood cell count). These findings indicate the diagnostic value of mapping parameters regarding the detection of severe courses of myocarditis, which are associated with increased mortality [42]. Furthermore, quantitative parameters have been shown to be appropriate markers to discriminate between active inflammation and the convalescent stage of the disease [22]. Our results also show that markers of active inflammation (T1 and T2 relaxation times, T2 signal intensity ratio) decreased at follow-up. However, visual myocardial edema and late gadolinium enhancement were still present in most patients. Therefore, mapping parameters seem to be a valuable tool to track disease activity or therapeutic response.

Limitations
Our study has some limitations. First, endomyocardial biopsy was not available as a reference standard because it is not routinely used in the diagnostic work-up of acute myocarditis at our institution due to relevant disadvantages, e.g., sampling error leading to false-negative results in focal myocarditis or its periprocedural risks, especially in sick children [43,44]. Therefore, clinical evidence for myocarditis was the reference standard used in this study, which has some drawbacks because this might have influenced the patient composition. Second, most included patients showed an infarct-like pattern of myocarditis. However, cutoff values might be different in patients with more chronic symptoms. Third, clinical data are partially incomplete because of the retrospective study design. Fourth, infants and young children are underrepresented in our cohort; however, the overall incidence rate of acute myocarditis is very low in this age group. Fifth, our single-center cohort is small, therefore additional prospective studies are needed before our results can be generalized.

Conclusion
The 2018 Lake Louise criteria including mapping parameters have a high performance for the diagnosis of acute myocarditis in children and adolescents and should be implemented into routine diagnostic work-up. Mapping parameters might be especially useful in tracking inflammatory disease activity.
Funding Open Access funding enabled and organized by Projekt DEAL.

Declarations
Conflicts of interest Claus C. Pieper declares only activities not related to the present article: He is a consultant for and has grants/grants pending with Guerbet and has received payment for lectures including service on speakers' bureaus from Philips Healthcare, Bayer Vital, and Guerbet. Ulrike Attenberger declares only activities not related to the present article: She has received payment for lectures including service on speakers bureaus from Siemens Healthineers. The remaining authors declare no conflicts of interest. Julian A. Luetkens declares only activities not related to the present article: He has received payment for lectures including service on speakers bureaus from Philips Healthcare.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.