Inclusion

Of the 68 patients that presented with multisystem inflammatory syndrome in children, 20 patients were excluded due to the lack of echocardiographic imaging in the acute phase to enable retrospective speckle-tracking measurements. Therefore, 48, 47, and 46 patients were included in our acute, early convalescence (6 weeks), and late convalescence (6 months) analyses, respectively. Of the 15 patients with persistent global longitudinal strain abnormalities at 6 months, echocardiographic imaging was available for 9/15 at >12−≤18 months post-onset (Fig. 1). Of the five patients with persisting global longitudinal strain abnormalities at >12−≤18 months, echocardiographic imaging was available for 3/5 at >18−≤24 months post-onset.

Figure 1. Patient inclusion in cohort study. Abbreviations: GLS = global longitudinal strain.

Clinical outcomes

Our cohort consisted of 48 patients (median age 11.9 [interquartile range 6.9−14.5] years old, 58.3% male). A total of 66.7% of the cases presented with symptoms that fit the criteria of a complete Kawasaki disease presentation as well. Nearly all (91.7%) children presented with abdominal symptoms (vomiting, diarrhoea, and abdominal pain). Neurological symptoms were seen in 56.3%, mostly consisting of headaches (29.2%) and an altered mental state (29.2%), which also included hallucinations (8.3%), dizziness (8.3%), and fainting (6.3%). Circulatory shock defined as the need for inotropic support or fluid resuscitation >20 mL/kg was the most frequent complication (58.3%). Overall, inflammatory markers (C-reactive protein, leukocytes, sedimentation), cardiac-specific markers (N-terminal pro b-type natriuretic peptide, troponin T), haemostatic components (D-dimer, fibrinogen), and ferritin were elevated in the acute phase of multisystem inflammatory syndrome in children. In addition, hypoalbuminemia and hyponatremia were seen in the acute phase of our patients. All patients recovered quickly after treatment and were discharged without heart failure treatment. A detailed overview of the patient characteristics and clinical presentation can be seen in Table S1.

Echocardiography

Acute echocardiography findings

There were 48 patients with a four-chamber view available in the acute phase for speckle-tracking analysis. In the acute phase, 81.3% (n = 39) of patients presented with a decreased left ventricle ejection fraction and/or fractional shortening (Table 1, Fig. 2). Four (8.5%) patients had an abnormal global longitudinal strain (>[−17]%), despite a preserved left ventricle ejection fraction and fractional shortening (example shown in Figure S1). Two of these had hypotension, of which one required inotropic support and also had acute kidney injury; in the other IVIG infusion was halted prematurely. The third patient had respiratory insufficiency requiring nasal high-flow therapy (consolidations on chest radiograph) and the last did not have any complications.

Table 1. Over-time proportion of patients with left ventricle ejection fraction, fractional shortening, and GLS below cut-off

*Based on left ventricle ejection fraction <50% and/or fractional shortening <28%(4).

Abbreviations: GLS = global longitudinal strain.

Figure 2. Left ventricle ejection fraction (A), fractional shortening (B), and GLS (C) over in first six months post-onset of disease with the pink range representing abnormal values; available GLS values in patients with abnormal GLS level at 6 months (D). Abbreviations: LVEF = left ventricle ejection fraction, FS = fractional shortening, GLS = global longitudinal strain.

Associations of laboratory measurements with echocardiography and cardiac MRI in the acute phase

Table 2 shows a detailed description of the acute blood parameter values that were significantly worse in patients with a decreased left ventricle ejection fraction, fractional shortening, and global longitudinal strain than those with a normal cardiac function during follow-up. Patients with a decreased left ventricle ejection fraction or fractional shortening had higher acute N-terminal pro b-type natriuretic peptide values than patients with a normal left ventricle ejection fraction or fractional shortening in the acute phase, while higher cardiac troponin T values were not associated with decreased left ventricle ejection fraction and fractional shortening in the acute phase.

Table 2. Associated blood parameters in patients with decreased LVEF, FS, or GLS per timeframe

^aSignificance (p value) based on Mann–Whitney U test.

Abbreviations: LVEF = left ventricle ejection fraction, FS = fractional shortening, GLS = global longitudinal strain.

Due to the limited number of cardiac MRIs performed, we were not able to assess the association of clinical outcomes and laboratory measurements with the cardiac MRI data. However, all of the patients with T1/T2 abnormalities as per cardiac MRI also had cardiac dysfunction based on left ventricle ejection fraction, fractional shortening, and global longitudinal strain.

Associations of laboratory measurements with prolonged global longitudinal strain abnormalities

Noticeably, different peak values of soluble blood markers in the acute phase were associated with the acute cardiac outcomes than with the late cardiac outcomes. Troponin T (P < 0.001) and N-terminal pro b-type natriuretic peptide (P = 0.02) were both significantly associated with a global longitudinal strain >[−17]% at 6 weeks (Table 2). C-reactive protein (P = 0.01), troponin T (P = 0.02), and creatinine (P = 0.003) values were all significantly associated with a global longitudinal strain >[−17]% at 6 months (Table 2, Fig. 3). In a logistic regression analysis, patients with elevated troponin T (above median, >62.0 ng/L [normal range 0–14 ng/L]) were 9.3 (95% CI 1.3–67.6) times more likely to exhibit abnormal global longitudinal strain during late convalescence disease than patients with a troponin level lower than 62.0 ng/L. This effect was independent of age at onset (continuous, in months), elevated C-reactive protein (above median, >187.9 mg/L [normal range 0–5 mg/L]), and decreased fractional shortening (<28%), which were not significantly associated with the prediction of a prolonged abnormal global longitudinal strain. The logistic regression model was statistically significant (χ²(4) = 12.3, P = 0.02) and explained 41.0% (Nagelkerke R²) of the variance in abnormal global longitudinal strain and correctly classified 76.5% of cases. We validated these findings in a Cox Regression analysis and found that an elevated troponin T value was also significantly associated with the time to recovery of the global longitudinal strain with a hazard ratio of 0.4 (P = 0.05, confidence interval 0.1–1.0), while age at onset, elevated C-reactive protein, and decreased fractional shortening were not. Furthermore, the peak acute values of troponin significantly correlated with the global longitudinal strain during late convalescence (r = 0.5, P = 0.002).

Figure 3. Acute levels of C-reactive protein (A), creatinine (B), N-terminal pro b-type natriuretic peptide (C), and troponin T (D) in patients with a normal (≤[−17]%) and abnormal (>[−17]%) GLS after 6 months. Abbreviations: GLS = global longitudinal strain, NT-pro-BNP = N-terminal pro b-type natriuretic peptide.

Conventional echocardiography in multisystem inflammatory syndrome in children

Regarding acute cardiac outcomes in our multisystem inflammatory syndrome in children population, cardiac dysfunction based on a decreased left ventricle ejection fraction or fractional shortening was seen in 81.3% with vasoactive support needed in 50% of our patients. This indicates a skewed presentation of severe multisystem inflammatory syndrome in children cases in the Amsterdam UMC, a tertiary hospital and expertise centre for Kawasaki disease. Previous studies have reported left ventricle ejection fraction dysfunction in up to 67%^{Reference Gaitonde, Ziebell and Kelleman9} and cardiac shock in about half of multisystem inflammatory syndrome in children cases with rapid left ventricle ejection fraction recovery within several days post-treatment in most patients.^{Reference Feldstein, Rose and Horwitz1,Reference Whittaker, Bamford and Kenny2} Similarly, in our study, left ventricle ejection fraction and fractional shortening had normalised in the majority of the patients during early and late convalescence.

Global longitudinal strain in multisystem inflammatory syndrome in children

Global longitudinal strain was decreased in 85.1% of the multisystem inflammatory syndrome in children patients in the acute phase. This is consistent with findings in other studies identified in our systematic search.^{Reference Basar, Usta, Akgun, Gungor, Sonmez and Babaoglu5–Reference Theocharis, Wong and Pushparajah19} In these studies, up to 90% of patients had an abnormal global longitudinal strain during admission.^{Reference Theocharis, Wong and Pushparajah19}

Similar to the subset of studies that included speckle-tracking measurements during follow-up,^{Reference Basar, Usta, Akgun, Gungor, Sonmez and Babaoglu5,Reference Das, Hill and Trivedi8,Reference Garbin, Raso and Piersanti10–Reference Matsubara, Kauffman and Wang13,Reference Sanil, Misra and Safa16,Reference Sirico, Basso and Sabatino18} global longitudinal strain values improved significantly during follow-up, but showed delayed recovery when compared to conventional parameters (left ventricle ejection fraction, fractional shortening). Remarkably, 35.7% of the multisystem inflammatory syndrome in children patients had persistently abnormal global longitudinal strain after 6 months, a large proportion when compared to a previous study with a follow-up of six months, which reported a persistently impaired global longitudinal strain in 13% of the multisystem inflammatory syndrome in children patients (n = 23)^{Reference Sirico, Basso and Sabatino18} and another study that reported normal cardiac function in all multisystem inflammatory syndrome in children patients after six months (n = 32).^{Reference Garbin, Raso and Piersanti10} While global longitudinal strain is known to have a low inter- and intra-operator variability and a low inter-vendor variation,^{Reference Plana, Galderisi and Barac22} these discrepancies may be due to the skewed presentation of severe multisystem inflammatory syndrome in children cases in the Amsterdam UMC, which was mentioned previously. Furthermore, as mentioned in the methods, we measured the global longitudinal strain in 4-chamber view, instead of deriving it from the averaged three standard apical views.^{Reference Mor-Avi, Lang and Badano30} Although the two are strongly correlated^{Reference Farsalinos, Daraban, Unlu, Thomas, Badano and Voigt31,Reference Gayeski, Connett and Honig32} and we chose this approach to be able to include a larger patient population, it could have led to a slight overestimation of strain abnormalities, since an overall global longitudinal strain can be normal even if a single plane is abnormal. Contrastingly, our proportion of global longitudinal strain abnormalities could also have been underestimated, since our global longitudinal strain cut-off was based on the lower limit reported in children between 0 and 1 year old. It should be noted that this cut-off was quite lenient since the mean global longitudinal strain in children of all age groups is −20.2% (95% confidence interval between −20.8% and −19.6%).^{Reference Levy, Machefsky and Sanchez29} The clinical implications of an abnormal global longitudinal strain in patients with a recovered left ventricle ejection fraction and fractional shortening are unclear. Only two of our patients had clinical complaints during late convalescence that may have been linked to the prior multisystem inflammatory syndrome in children episode. The latter patients had a persistently abnormal global longitudinal strain >1-year post-admission and had premature ventricular contractions that were first noticed during the acute disease episode. While these premature ventricular contractions could have been pre-existent, the female patient also had clinical complaints of exercise intolerance. Previous studies in adult heart failure patients with a recovered left ventricle ejection fraction suggest that an abnormal global longitudinal strain may predict future deterioration of left ventricular function.^{Reference Adamo, Perry, Novak, Makan, Lindman and Mann35} In this population, global longitudinal strain >[−14]% is also associated with major adverse cardiac events (cardiovascular death, emergency heart transplantation, left ventricular assist device, or intra-aortic balloon pump implantation),^{Reference Cameli, Pastore and Mandoli36} although extrapolation of data from adult heart failure patients to previously healthy children with acute ventricular dysfunction secondary to a hyperinflammatory disease is difficult.

Noticeably, the global longitudinal strain in three patients with an abnormal global longitudinal strain between >12 and ≤18 months had worsened at their consecutive check-ups between >18 and ≤24 months. Although no hard conclusions can be drawn, the lack of normalisation of global longitudinal strain is concerning. Even though the long-term implications of global longitudinal strain abnormalities are still unclear, we advocate that careful cardiac follow-up is warranted in patients with a persistent abnormal global longitudinal strain, since they may be at risk for cardiovascular events in the future.

Cardiac MRI in multisystem inflammatory syndrome in children

Our cardiac MRI findings indicated acute myocardial involvement in a majority of the patients in the acute phase (77.8% [7/9]). This is high compared to previous reports describing myocarditis in up to 73%,^{Reference McMurray, May, Cunningham and Jones33,Reference Dufort, Koumans and Chow37} possibly due to the limited subgroup investigated in our study (only 9/48 patients with a median left ventricle ejection fraction of 33.0% vs. 39.0% of the complete study population).

Thankfully, later cardiac MRIs did not show signs of myocardial involvement or fibrosis. Similar to previous reports^{Reference Webster, Patel and Carr38,Reference Bartoszek, Malek, Barczuk-Falecka and Brzewski39} and in contrast to typical myocarditis,^{Reference Patel, Kelleman and West40,Reference Vukomanovic, Krasic and Prijic41} none of the patients in our study required heart failure treatment after discharge and systolic function and troponin levels normalised promptly post-treatment. Other studies found late gadolinium enhancement in 20% (4/20)^{Reference Theocharis, Wong and Pushparajah19} and 35% (8/23)^{Reference Sirico, Basso and Sabatino18} of multisystem inflammatory syndrome in children patients during the acute phase of disease. One of the studies we identified in our systematic search found residual oedema in one out of nine patients after 204 days of disease onset.^{Reference McAree, Hauck and Arzu14} The patient had presented with left ventricle dysfunction and elevated troponin in the acute phase, which had completely normalised during follow-up. However, no fibrosis was seen on the cardiac MRI of the patient. Other studies focused on cardiac MRI abnormalities two^{Reference Webster, Patel and Carr38} to three^{Reference Bartoszek, Malek, Barczuk-Falecka and Brzewski39} months post-diagnosis did not report any abnormalities either, concurrent with our own findings.

Prediction of prolonged global longitudinal strain abnormalities

In our multivariable analyses, for the first time, the highest acute value of cardiac troponin significantly predicted delayed global longitudinal strain recovery, suggesting subclinical myocardial involvement in part of the multisystem inflammatory syndrome in children patients. Cardiac biomarkers (N-terminal pro b-type natriuretic peptide, troponin) have previously been associated with disease severity (ICU admission^{Reference Whittaker, Bamford and Kenny2,Reference Lee, Day-Lewis and Henderson42} ) and decreased cardiac function (based on left ventricle ejection fraction, fractional shortening, and/or global longitudinal strain) in multisystem inflammatory syndrome in children.^{Reference Basar, Usta, Akgun, Gungor, Sonmez and Babaoglu5,Reference Basu, Kim and Sharron6,Reference Gaitonde, Ziebell and Kelleman9–Reference He, Leone and Frye11,Reference Matsubara, Kauffman and Wang13,Reference Sirico, Basso and Sabatino18,Reference Kobayashi, Dionne and Ferraro20} Two other studies investigated the potential of N-terminal pro b-type natriuretic peptide and troponin I to predict an abnormal global longitudinal strain.^{Reference McAree, Hauck and Arzu14,Reference Sanil, Misra and Safa16} One did not find an association between the cardiac biomarkers and global longitudinal strain abnormalities at 3−10 weeks post-onset, although 3/9 (33%) patients with an abnormal troponin I during acute illness had an abnormal global longitudinal strain and circumferential strain at subacute follow-up.^{Reference McAree, Hauck and Arzu14} The other study found an association between the initial troponin I value and longitudinal strain value 10 weeks post-onset.^{Reference Sanil, Misra and Safa16} It should, however, be noted that the outcomes associated with troponin I may differ from those associated with troponin T, although both are associated with cardiovascular events.^{Reference Welsh, Preiss and Hayward43}

Elevated levels of troponin in patients receiving cardiotoxic chemotherapy have also been described to be a sensitive measurement for early detection of cardiotoxicity.^{Reference Sawaya, Sebag and Plana25} In line with our findings, N-terminal pro b-type natriuretic peptide levels have not been consistently described as predictor for cardiotoxicity.^{Reference Plana, Galderisi and Barac22}

Future studies

In the future, we recommend in-depth analyses of the prognostic value of an integrated approach of imaging techniques and biomarkers in severe inflammatory disorders, such as multisystem inflammatory syndrome in children, similar to the approach used to predict cardiotoxicity related to cancer treatment.^{Reference Heidenreich, Bozkurt and Aguilar4,Reference Plana, Galderisi and Barac22} Since the incidence of multisystem inflammatory syndrome in children seems to be diminishing, concurrent with the rise in paediatric SARS-CoV-2 vaccination status and possibly decreased abilities to induce a hyperinflammatory response of the currently ruling SARS-CoV-2 variant omicron,^{Reference Holm, Espenhain and Glenthoj44} such studies should also focus on other similar hyperinflammatory disorders, such as Kawasaki disease and toxic shock syndrome. Furthermore, mechanisms underlying persisting global longitudinal strain abnormalities should be further elucidated, for instance, by investigating if these findings may be linked to the mitochondrial dysfunction recently implicated in long coronavirus disease 19.^{Reference Guarnieri, Dybas and Fazelinia45} It would be interesting to evaluate if patients with mitochondrial dysfunction and long coronavirus disease 19 have persisting global longitudinal strain abnormalities, as has been suggested in studies with patients that have fully recovered from coronavirus disease 19.^{Reference De and Bansal46,Reference Gherbesi, Bergamaschi and Cusmano47}

Study limitations and strengths

Limitations of the current study include the limited number of subjects, due to the rarity of multisystem inflammatory syndrome in children, limiting the (multivariable) analyses that we could perform within the current patient population. Also, a single observer (experienced paediatric cardiologist specialised in Kawasaki disease) recalculated the echocardiographic (left ventricle ejection fraction, fractional shortening, and global longitudinal strain) values retrospectively in this study, which could introduce bias, but minimises the risk of inter- and intra-observer variability. Moreover, there may have been selection bias, since we could only include patients of whom echocardiographic data in the acute phase were available, likely representing a more severely affected subpopulation of multisystem inflammatory syndrome in children patients. Lastly, only 9 of the 15 patients with global longitudinal strain abnormalities at 6 months had echocardiography available between >12 and ≤18 months.

However, the maximum follow-up of previous studies did not exceed 4.4 months (due to the novelty of the disease),^{Reference Basar, Usta, Akgun, Gungor, Sonmez and Babaoglu5,Reference Das, Hill and Trivedi8,Reference Garbin, Raso and Piersanti10–Reference Matsubara, Kauffman and Wang13,Reference Sanil, Misra and Safa16,Reference Sirico, Basso and Sabatino18} with the exception of two studies that investigated cardiac outcomes in multisystem inflammatory syndrome in children up to 6 months.^{Reference Garbin, Raso and Piersanti10,Reference Sirico, Basso and Sabatino18} This stresses another major strength of our current study, which adds insight into the dynamics during late convalescence, up to two years after the disease episode. Also, to our knowledge, this is the first study to perform a multivariable regression analysis to investigate the correlation of delayed cardiac recovery with cardiac biomarkers in multisystem inflammatory syndrome in children.