Longitudinal Outcomes for Multisystem Inflammatory Syndrome in Children

All consecutive pediatric patients hospitalized for MIS-C at Columbia University Irving Medical Center from April 17 to December 31, 2020, were included. We included any individual <21 years of age, meeting Centers for Disease Control and Prevention definition for MIS-C, with documented current or recent SARS-CoV-2 reverse transcriptase polymerase chain reaction (RT-PCR) or serology or documented recent household contacts positive for SARS-CoV-2 if testing was unavailable at presentation.^1,2  Patients were excluded if they had classic Kawasaki disease or if an alternate, plausible diagnosis was made.^3–5  All cases were reviewed and approved by an interdisciplinary team (R.W., A.F., U.S.K., M.G., J.D.M.). Questionable cases were independently evaluated (P.Z.).

Patients were admitted to either general pediatric units or ICUs and managed by a dedicated interdisciplinary MIS-C team of pediatricians, pediatric cardiologists, rheumatologists, immunologists, and infectious disease specialists. Patients were treated in accordance with our institutional clinical protocol and discharged on a steroid taper and aspirin (Fig 1).

Patients were managed in the Columbia University Interdisciplinary MIS-C Follow-up Program. Visits were scheduled ∼2 weeks (1–4 weeks), 2 months (1–4 months), and 6 months (4–9 months) after discharge (Fig 2). Inflammatory and other laboratory markers were tested at baseline and reassessed at follow-up visits until they normalized. Comprehensive cytokine profiling and immunoglobulin electrophoresis were performed on a subset of patients. Echocardiograms were deidentified and read by our MIS-C follow-up program echocardiogram core laboratory, following a standardized protocol (K.M.F, N.P., and A.M.S.). Coronary arteries were measured and z scores were derived by using previously established normative values (Boston Children’s z score system).⁶  A subset of 10 echocardiograms were read by all 3 readers (K.M.F., N.P., A.M.S.) and intraclass correlation coefficient used to assess interrater reliability of echocardiographic measurements (interrater reliability 0.72–0.89). Coronary artery abnormalities were classified by z scores as follows: normal <2, dilation 2 to <2.5, aneurysm ≥2.5. Left ventricular dilation and function were measured qualitatively and quantitatively; ejection fraction was calculated by using the 5/6 area × length method.⁷ 

Patient demographics and clinical characteristics were described by using standard summary statistics and univariable analyses. For laboratory values, the percentages of patients with values out of range were calculated on admission and at each follow-up time point. Major cardiac sequelae were examined as a composite outcome, defined as moderate or severe dysfunction, ventricular dilation, or valve regurgitation, or coronary artery z score >2.0. Our intention had been to use survival analysis to assess associations between biomarkers and time to recovery from cardiac sequelae. Given the paucity of long-term adverse outcomes and high level of predictor collinearity, we assessed only the effects of patient characteristics on in-hospital cardiac sequelae. To assess loss to follow-up bias, we assessed differences in follow-up between patients with and without major cardiac sequelae on admission and at each follow-up interval. Analyses were performed by using Stata software version 14.2 (Stata Corp, College Station, TX) and R version 4.0 (R Foundation for Statistical Computing, Vienna, Austria). This study was approved by the Columbia University Irving Medical Center Institutional Review Board.

In total, 58 patients were evaluated for MIS-C during the 7 weeks after the 2020 New York City spring SARS-CoV-2 surge (April 17 to June 6, 2020). Five additional patients were evaluated between June 7 and December 31, 2020. Of these, 9 patients were excluded as classic Kawasaki disease and 9 for other alternate, plausible medical diagnoses, for a total of 45 cases of MIS-C. Of the patients with Kawasaki disease, 3 did have positive coronavirus disease 2019 (COVID-19) antibody results but had lymphocytosis, thrombocytosis, and anemia.^4,5 

The median length of in-hospital stay was 5 days (interquartile range [IQR] 4–8; range 2–32 days). Many patients returned to their home institutions for follow-up as the pandemic evolved. In total, 96% (n = 43 of 45) of patients returned for at least 1 follow-up assessment with the Columbia University Interdisciplinary MIS-C Follow-up Program; 39 (87%) were seen at or around 2 weeks postdischarge (mean 14 days; SD 4 days), including all with moderate to severe dysfunction, dilation, atrioventricular valve regurgitation, or coronary artery dilation. Thirty-one patients (69%) were seen at or around 2 months (mean 1.9 ± 0.6 months) and 24 (53%) at or around 6 to 9 months (mean 6.7 ± 1.0 months). The median time to last follow-up was at 5.8 months postdischarge (IQR 1.3–6.8), with 37 (84%) of children completing at least 1 visit beyond 2 months postdischarge.

Admission characteristics are described in Table 1. SARS-CoV-2 serology was tested in 96% of patients; all 43 test results were positive. Two patients presented when serology testing was not available, had documented previous positive PCR results, and had household contacts who had received previous positive SARS-CoV-2 PCR results. PCR results were acutely positive in 27% (n = 12 of 45 tested) of children. The majority had no underlying medical condition (84%, [n = 38]); 6 children (14%) had asthma and 1 had von Willebrand disease. One of the 6 children with asthma was also of late-preterm gestation and one also had chronic renal disease. Twenty-four percent (n = 11) of children presented with at least mild respiratory distress or oxygen requirement and 64% (n = 29) had vasodilatory shock. All patients survived to hospital discharge.

The majority of patients required intensive care (76% [n = 34]), with 64% (n = 29) requiring inotropes or pressors and 31% (n = 14) requiring some form of respiratory support. Immunomodulatory medications were administered to most patients; 100% (n = 45) received methylprednisolone and 91.1% (n = 41) received intravenous immunoglobulin. Biologics were administered in 11% of patients: 9% (n = 4) received anakinra (recombinant anti-interleukin [IL] IL-1RI antibody) as monotherapy (with intravenous immunoglobulin and steroids) and 1 patient with a mutation in complement regulatory gene CFHR1 received anakinra in combination with eculizumab (recombinant anti-C5 monoclonal antibody). One patient early in the pandemic received remdesivir. No child required extracorporeal membrane oxygenation (ECMO). No child developed secondary infections.

Inflammatory and immune profiles for children over time are described in Table 2. On admission, laboratory evaluation was consistent with marked, nonspecific inflammation, with C-reactive protein (CRP), ferritin, and D-dimer all markedly elevated in the majority of patients (87%–98%). Troponin and N-terminal pro–brain natriuretic peptide (NT-proBNP) were also elevated in 64% and 91% of patients, respectively, consistent with cardiac involvement. These parameters peaked at or shortly after admission and then gradually normalized. By the first follow-up, CRP, troponin, and NT-proBNP had normalized in nearly all tested patients (97%–100%); ferritin and D-dimer were not reassessed (Fig 3).

Forty-four percent of patients (n = 20) were mildly thrombocytopenic on admission. Platelet counts peaked on hospital day 4, with 34% (n = 15) developing thrombophilia. Twenty-six percent (n = 11) had persistent thrombophilia at 1 to 4 months.

Twenty patients underwent comprehensive cytokine profiling during hospitalization. Soluble IL-2R, IL-6, IL-10, IL-17, IL-18, and C-X-C Motif Chemokine Ligand 9 were elevated in the majority of patients. Four patients had elevated interferon-γ (IFN-γ). All other measured cytokines were undetectable (IL-2, IL-12, IL-4, IL-5, IL-8, IL-13, IL-1β, and tumor necrosis factor α).

On admission, significant lymphopenia was evident in 66% of patients, with a median absolute lymphocyte count of 0.9 × 10³/μL (0.5–2.1). Quantitative immunoglobulins were normal. Lymphocyte subsets were assessed in 12 patients on admission, among whom 83% (n = 10) showed evidence of lymphopenia. CD3⁺CD8⁺ cells were nearly uniformly low, whereas CD3⁺CD4⁺ and CD56⁺ (natural killer) cells were low in two-thirds of patients (n = 8). CD19⁺ (B cells) were low in 25% of patients (n = 3). At 1 to 4 weeks postdischarge, lymphocyte subsets were assessed in 24 patients. Despite ongoing therapy with glucocorticoids, the lymphocyte population had shifted, with 42% of patients (n = 10) demonstrating marked lymphocytosis (ie, markedly elevated CD3⁺ and CD19⁺ counts). In addition, the subsets within this CD3⁺ population were altered, with nearly all patients demonstrating an increased double-negative T cell (DNT) (CD3⁺, CD4⁻CD8⁻) population (96% [n = 23 of 24]), irrespective of lymphocytosis. DNTs remained elevated in 92% (n = 12/13) patients throughout follow-up. Study of these cells in a patient at >6 months revealed the majority to be γδ T cells (Supplemental Fig 4).

Table 3 describes echocardiographic findings in children on admission and over time. During initial hospitalization, 22% of patients (n = 10) had moderate to severe left or right ventricular dysfunction, and 53% (n = 24) had at least mild dysfunction. In total, 36% (n = 16) had at least mild right or left ventricular dilation. Twenty-seven percent (n = 12) had moderate to severe atrioventricular valve regurgitation, divided evenly between those with and without ventricular dysfunction. All patients with moderate or severe mitral regurgitation also had moderate or severe tricuspid regurgitation. One patient presented with a large pericardial effusion requiring pericardiocentesis. The effusion resolved after drainage.

Although clinical echocardiograms recorded “prominent” coronaries in 40% of patients on initial hospitalization (n = 18), objective coronary artery dilation and aneurysms, as assessed by our MIS-C follow-up program echocardiogram core laboratory, were present in only 9% (n = 4) and 7% (n = 3) of patients, respectively. Coronary artery abnormalities were significantly associated with myocardial dysfunction, with 4 of the 7 patients with enlarged coronaries also exhibiting severe dysfunction (P = .035).

Only higher CRP (P = .030) and higher NT-proBNP (P = .028) were significantly associated with severe cardiac presentation (moderate to severe dysfunction, ventricular dilation, atrioventricular valve regurgitation, or coronary artery dilation). Neither cytokine, lymphocyte subtype, nor immunoglobulin levels were associated with severe presentation.

Thirty-nine children returned for follow-up 1 to 4 weeks postdischarge (87%). All children with moderate or severe dysfunction, ventricular dilation, valve regurgitation, or coronary artery dilation during admission were reevaluated. Eleven percent of patients (n = 5) had mild ventricular dysfunction. Of these 5 patients, 3 had mild ventricular dilation (P = .010); another 2 children had mild ventricular dilation with normal function. There was no association between persistent pericardial effusions and either ventricular dysfunction (P = .436) or dilation (P = .072). All coronary measurements had normalized.

Thirty-one patients were reevaluated between 1 to 4 months postdischarge (69%). Two of 31 children had mild (biventricular) dysfunction (6%). One additional child had mild right ventricular dilation (3%) and one had mild left ventricular dilation (3%). Eight patients (26%) had mild atrioventricular valve regurgitation. Only 1 child with regurgitation was among those with dysfunction. No patient with ventricular dysfunction had dilation. In total, 11 children had persistent mild abnormalities on echocardiogram (24%), but none had more severe findings.

At 4 to 9 months postdischarge, 24 children had returned for follow-up (55%); 84% of children (n = 38) completed at least 1 visit >2 months postdischarge. No child was readmitted to the hospital or reported secondary infections during or after hospital discharge. Only 1 child had persistent mild biventricular dysfunction. A second child had persistent mild mitral and tricuspid valve regurgitation. All had normal ventricular size and normal coronary artery measurements. One additional child, last seen at 5 weeks postdischarge, returned late for the 4- to 9-month follow-up (11.5 months), and so is excluded in our totals. This child also had a normal echocardiogram.

We report the midterm inflammatory, immunologic, and cardiac outcomes for the largest single–hospital system cohort from an interdisciplinary MIS-C follow-up program. Two-thirds of patients admitted with MIS-C presented with or developed signs of vasodilatory shock and three-quarters required admission to an ICU. Consistent with other reports, patients presented with a marked, diffuse, systemic inflammatory reaction, evidenced by severe elevations in all inflammatory markers and cytokines. Two-thirds of patients had lymphopenia, nearly half were thrombocytopenic, and almost all had moderate ferritin elevations. Cardiac involvement was evidenced by moderate to severe ventricular dysfunction, dilation, valve regurgitation, or coronary artery dilation in nearly half of the children, and at least mild echocardiographic abnormalities were present in >80% of children. Steroids and immunoglobulins were initiated soon after clinical presentation. There was rapid clinical improvement of inflammatory and immunologic markers and resolution of all moderate to severe cardiac findings by first follow-up. Two children had persistent mild abnormalities on echocardiogram at 4.3 and 7.8 months postdischarge.

Laboratory markers of inflammation and immunologic derangement on admission in our cohort were similar to those described by others, including diffuse elevations in acute phase reactants, lymphopenia, thrombocytopenia, and sustained and progressive NT-proBNP and troponin concentrations.^5,8–10  In our cohort, however, only 1 child was intubated, no children died, and no children required ECMO. This is in contrast to other reports revealing rates of intubation of 20% to 40%, 5% ECMO, and 1% to 2% mortality.^5,11–14  Early use of high-dose glucocorticoids may have played a role in the lower morbidity in our cohort.^13,15–18  It is also possible that there might have been differences in the COVID-19 strains impacting New York in spring 2020 or in the timing of patient presentations at that time. Ferritin levels in our cohort were lower than in other reports.^10,13,17 

Aggressive use of steroids may also explain the lower incidence of coronary artery abnormalities in our cohort. Ramcharan et al,¹⁹  for example, described “prominence” of coronaries as positive findings in 14 of their 15 patients. Theocharis et al,²⁰  using cardiac computed tomography, described uniform coronary dilation in 9 of their 12 patients. That said, differences coronary artery dilation may also lie in varying definitions used across reports. Our clinical team initially described coronary artery prominence in nearly half our cohort. Because the pandemic was evolving, anxiety was high, and clinical care teams preferred to err toward high sensitivity rather than miss any abnormalities. Subsequent, blinded objective measures by our MIS-C follow-up program echocardiogram core laboratory revealed a more modest incidence of coronary dilation. Again, it is also possible that the COVID-19 strains impacting New York in spring 2020 or the timing of patient presentations could be different from those impacting the rest of the country or world.

In our data, during the acute phase, robust innate immune response with associated IFN-γ upregulation was a hallmark, as universally described in previous studies.^5,14,21,22  Observed elevations in IL-6, IL-2R, IL-10, CXCL-9, and IL-18^5,21  support the concept of interplay between T-cell activation and IFN-γ activation in the pathogenesis of MIS-C. CXCL-9, in particular, was highly elevated in our cohort, reaffirming IFN-γ activation.²³  In contrast to a previously published report, our patients had low, nearly undetectable levels of tumor necrosis factor α on admission. Although this may be the result of varying assay sensitivity between institutions, this warrants further evaluation.²⁴ 

We observed differential T- and B-cell subset lymphopenia during the acute phase with more significant reduction in T-cell numbers. Although this has been previously described as a hallmark of MIS-C,^5,21,25–27  a subset of our patients also demonstrated significant lymphocytosis during recovery and convalescence. Lymphocyte counts have been noted to constitute a relatively high percentage of total cells in patients during the convalescent phase of Kawasaki disease²⁸ ; however, absolute lymphocytosis has not been reported at any stage in either MIS-C or Kawasaki disease. Lymphopenia is typically seen with steroid use rather than lymphocytosis, and supraphysiologic steroid dosing, as we used in our patients, is directly lymphotoxic.^29,30  Thus, the lymphocytosis seen in these patients while on corticosteroid therapy is paradoxical and may be a rebound from the processes that caused lymphopenia during acute disease. Interrogation of lymphocyte subpopulations revealed that by 1 to 4 weeks postrecovery, the majority of patients displayed marked increases in the proportion of DNTs.³¹  A specific subtype of DNTs known as γδ T cells (as opposed to αβ DNTs, which are seen in congenital lymphoproliferative syndromes³¹ ) are known to proliferate in the setting of acute infection³²  and potentially persist for short periods during recovery but may also play a more complex role in response to acute viral infection.^33,34  However, in our patient sample, these cells persisted long after acute infection and well beyond any state of active inflammation. They likely represent a prolonged postinflammatory recovery cell population, but further study is ongoing to better define this observation.

Our cohort of patients demonstrated excellent recovery and response to therapy. The majority of patients had short hospital stays and rapid return to baseline health. Although these findings are encouraging, persistent mild abnormalities in a small minority of patients raise questions about long-term follow-up and physical activity. The American Academy of Pediatrics and the American College of Cardiology currently recommend restriction from exercise for a minimum of 3 to 6 months for children who had severe symptoms associated with a COVID-19 infection, including MIS-C, with cardiac clearance before resumption of training or competition.^35,36  Case series have revealed diffuse myocardial edema without fibrosis in the acute phase of illness, suggestive of a postviral myocarditis.³⁷  Data regarding changes on cardiac MRI in these patients are limited to case series with findings consistent with edema and hyperemia in the acute phase of MIS-C.³⁷  Because it is possible that more subtle findings that might put patients at risk for sudden cardiac arrest or future cardiomyopathies would only be observed on advanced imaging, we recommended cardiac MRI for all patients with mild or more systolic dysfunction (left ventricular ejection fraction <50%) any time during initial hospitalization. The National Institutes of Health, Centers for Disease Control and Prevention, and others are currently funding multicenter investigations of children with MIS-C. These will be critical for providing long-term answers for patients and families. In the meantime, we recommend a relatively conservative, individualized approach, with cardiac clearance recommended before sports participation for children hospitalized with MIS-C.

Because the understanding of MIS-C evolved over the course of the 2020 SARS-CoV-2 pandemic, case definitions changed. In an earlier report from our institution, for example, researchers used a broader New York State Department of Health interim case definition for MIS-C, current during spring 2020.¹⁸  That case definition resulted in the inclusion of 7 patients for whom alternate plausible diagnoses were eventually established and an additional 5 patients (with lymphocytosis, thrombocytosis, and anemia) now defined as classic Kawasaki disease. Inclusion of alternate diagnoses might have unpredictable effects on morbidity and mortality and might explain some variation in inpatient outcomes between our studies and reports from other institutions. Inclusion of patients with Kawasaki disease in our cohort would have decreased the median age, reduced the incidence of myocardial dysfunction and vasodilatory shock, and reduced our ability to detect T-cell and other inflammatory abnormalities. Other systemwide effects of the pandemic may bias the severity of our clinical cohort. During the spring of 2020, all children admitted within the NewYork-Presbyterian hospital system were transferred to Columbia University Irving Medical Center/NewYork-Presbyterian Morgan Stanley Children’s Hospital for care to make room for adults at other institutions. In addition, echocardiogram availability was limited during the pandemic to reduce person-to-person transmission of COVID-19 in clinical care spaces. Patients with more severe disease were prioritized for earlier follow-up appointments. It is therefore possible that patients with inconsistent follow-up might have recovered even earlier than described here. Because of the observational design of this report, we are unable to make definitive conclusions regarding patient outcomes and treatment regimen.

Our study reveals generally encouraging medium-term outcomes, including rapid normalization of inflammatory markers and significant cardiac abnormalities in the majority of patients with MIS-C. Immunologic data suggest that immune derangements may extend for months after initial hospitalization. The exact nature and potential for long-term cardiac fibrosis, exercise intolerance, or other changes remain unknown. Given the unexpected, persistent immunologic findings, both optimism and continued caution and follow-up are recommended to ensure the health of children with MIS-C and to further inform our understanding of this disease.

We acknowledge the contributions of the following members of the Columbia University Irving Medical Center Pediatric-Adult Congenital Heart Research Collaborative and the Columbia University Interdisciplinary MIS-C Follow-up Program: Alexis Boneparth, MD; Eva Cheung, MD; Tarif Choudhary, MD; Elizabeth Duvergeur, BA; Yannett Franklin; Michael Fremed, MD; Katrina Golub, MPH; Leo Liberman, MD, MS; and Eric Silver, MD.

We also acknowledge George Vlad for performing flow cytometry to phenotype DNTs.

COVID-19

coronavirus disease 2019

CRP

C-reactive protein

DNT

double-negative T cell

ECMO

extracorporeal membrane oxygenation

IFN-γ

interferon-γ

IL

interleukin

IQR

interquartile range

MIS-C

multisystem inflammatory syndrome in children

NT-proBNP

N-terminal pro–brain natriuretic peptide

PCR

reverse transcriptase polymerase chain reaction

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2