Main

At the start of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic, most children were initially spared, with very few developing moderate to severe coronavirus disease 2019 (COVID-19)1,2,3,4,5,6,7. As the virus spread globally, the majority of children <18 years of age had asymptomatic infection or mild COVID-19, and hospitalization was rare8,9,10,11,12. Those children who did develop severe COVID-19 generally had risk factors, including underlying respiratory, neurological, or immune disorders13. This pattern contrasts with other respiratory viruses, including respiratory syncytial virus (RSV), influenza virus, and parainfluenza virus, known to cause severe disease in young children14,15. In April 2020, a post-infectious syndrome known as multisystem inflammatory syndrome in children (MIS-C) emerged in individuals <21 years of age, with the majority of patients requiring intensive care for life-threatening complications. Investigating age-associated determinants of the spectrum of clinical outcomes in children and adults related to SARS-CoV-2 infection is paramount for understanding host susceptibility and outcome and could help optimize disease prevention and treatment.

This Review synthesizes important principles of developmental immunology with evidence linking immunobiology and clinical outcomes of SARS-CoV-2 infections in children and adults. Because severe COVID-19 and MIS-C are uncommon in children, sample sizes for published studies are limited. To address this limitation, we highlight findings that are common as well as disparate across published studies, with the goal of identifying fundamental distinctions between the pediatric and adult response to SARS-CoV-2 infections. Owing to the spectrum of disease associated with SARS-CoV-2 infections, we compare the immunological response of children and adults within distinct clinical phenotypes. These phenotypes include asymptomatic SARS-CoV-2 infections, mild COVID-19 (characterized by the presence of upper respiratory symptoms not requiring supportive care), and symptomatic COVID-19 prompting medical attention. Within each section, we discuss hypotheses that have been investigated to explain why children are more protected from moderate to severe COVID-19 than adults.

Asymptomatic SARS-CoV-2 infection to mild COVID-19

Overall, studies thus far have shown that responses to asymptomatic or mild COVID-19 are similar between adults and children. Three studies found that the viral loads of children with asymptomatic SARS-CoV-2 infection or mild disease are comparable16 to, or slightly lower than, viral loads in adults with similarly mild SARS-CoV-2 infections17,18. Jia et al. showed that the cytokine profiles of children with mild COVID-19 resemble profiles from healthy children, reflecting a low level of inflammation19. Similarly, Cohen et al. found no differences in the percentage of interferon-γ (IFN-γ)-producing CD4+ or CD8+ T cells with T cell receptors (TCR) specific for structural, accessory, and non-structural proteins in nine asymptomatic children compared with that in nine adults, with comparable capacities for generating multiple cytokines20. This similarity occurred despite a reduced level of cross-reactive antibodies to β-coronaviruses other than SARS-CoV-2 in children compared with adults with similarly mild infections, suggesting that differences in prior infections do not substantially affect the T cell response in individuals with mild SARS-CoV-2 infections20.

Parallel to a robust cellular response to SARS-CoV-2, children are capable of mounting a robust serological response to mild SARS-CoV-2 infections. A study showed that children, adolescents, and adults with mild COVID-19 had similar levels of anti-spike IgG with comparable functionality, including the capacity for activating complement and phagocytic cells21. Two studies showed even higher neutralizing activity in anti-SARS-CoV-2 antibodies from children <10 years of age than in adolescents and young adults22,23. The effect of disease severity on the antibody response is indicated by lower levels of convalescent antibodies against the spike and nucleocapsid proteins in children with asymptomatic or mild infections than in those with moderate COVID-19 requiring hospitalization24,25. Collectively, these studies do not show major differences in the viral load or immune response to mild COVID-19 between children and adults.

Moderate to severe COVID-19

In contrast with the comparable pediatric and adult immune responses during mild COVID-19, notable differences occur in patients hospitalized with COVID-19. Even though the number of pediatric individuals with COVID-19 increased after the spread of the Delta variant26, compared with hospitalized adults, most children and adolescents hospitalized for COVID-19 required shorter hospitalizations and less respiratory support, and mortality was less common27,28. In the USA, individuals >50 years of age account for at least 90% of all deaths29. Differences in immune responses could be either predictive or reflective of COVID-19 disease severity. Additionally, comorbid conditions differ between adults and children, as do alcohol and tobacco use and obesity, which can all affect immune outcomes. In this section, we discuss the differences in the immune systems of children and adults that have been investigated as potential drivers of age-associated clinical outcomes from COVID-19 (Fig. 1).

Fig. 1: Factors that reduce the severity of SARS-CoV-2 infection in children.
figure 1

Children tend to have fewer comorbidities than adults. Protective innate immune factors include reduced expression of ACE2 and a more robust IFN response. Compared with adults, children also have more T cells owing to greater thymic output, more T cells that recognize non-structural viral proteins, and less T cell exhaustion.

Full size image

Potential contributors to differences in adult versus pediatric COVID-19

In this section, we review studies pointing to developmental differences in innate and adaptive immunity (Fig. 2). Studies have focused particularly on differences in IFN signaling and T cell function as contributors to the increased susceptibility to severe COVID-19 in adults. In support of this idea, the prevalence of severe COVID-19 was higher in immunodeficient cohorts enriched for disorders of T cell development and/or function30,31,32, compared with that of the general pediatric population or cohorts with a broader diversity of primary or secondary immunodeficiencies, such as antibody deficiencies, autoimmune disorders, or those being treated with immunosuppressive medications6,33,34. The increased frequency of severe COVID-19 in individuals with reduced T cell numbers parallels the increased mortality from COVID-19 in older people, another population with reduced T cell number and function35,36. Notably, comorbidities associated with severe COVID-19 (ref. 37), including cardiovascular disease, diabetes, malignancies, renal failure, and neurodegenerative disease, occur more frequently in adults than in children. Some chronic diseases, such as cerebrovascular disease, obesity, and diabetes, can result in a dysfunctional immune response to infection through epigenetic and metabolic reprogramming of immune cells38,39,40. Other comorbidities, such as chronic lung disease, can result in anatomical susceptibility to more-severe COVID-19. When present in children and young adults, these comorbidities also increase the risk of severe COVID-19 (refs. 6,9,10,13,41).

Fig. 2: Developmental differences in pediatric and adult immune response to viral infection.
figure 2

Innate cell function reaches adult levels in early childhood. B cell and T cell function continues to develop throughout childhood. By contrast, regulatory T cell function wanes with aging.

Full size image

SARS-CoV-2 entry factors

Upon encountering respiratory mucosa, the receptor-binding domain (RBD) of the SARS-CoV-2 spike protein binds to the widely expressed angiotensin-converting enzyme 2 (ACE2) receptor42,43,44. The subsequent fusion of viral and cellular membranes is facilitated by cleavage of the viral spike protein by host proteases, such as transmembrane serine protease 2 (TMPRSS2) and cathepsin L, thereby enabling cellular entry45,46. TMPRSS2 expression in the airway epithelium increases with age47,48,49,50. A meta-analysis of 31 single-cell RNA-sequencing studies from 228 individuals without a history of lung disease also showed a positive correlation between age and the expression of ACE2 in alveolar cells51. However, distinctions in the expression of viral entry factors between children and adults are less clear, as only some studies support the hypothesis that fewer ACE2 receptors in children can account for reduced viral entry into the lungs47,52,53.

In adults with COVID-19, higher viral loads measured through nasopharyngeal or oropharyngeal swabs have been associated with more-severe disease in some, but not all, studies, particularly as asymptomatic individuals might have viral loads comparable to those found in hospitalized individuals16,54,55,56,57. Although studies have detected higher viral loads in symptomatic children than in asymptomatic children, viral loads have not been predictive of disease severity in children58,59. Furthermore, children and adults with similar disease severity have comparable viral loads16,17. The variability in these studies might arise from differences in collection techniques or the inability to sample the lower respiratory tract in non-intubated individuals. Nonetheless, studies thus far have not conclusively shown that differences in viral load account for improved clinical outcomes that occur in the majority of children with COVID-19.

Interferon signaling

Although the ACE2 receptor is critical for SARS-CoV-2 cellular entry, other viral components activate widely expressed pattern-recognition receptors to initiate the innate immune response60. SARS-CoV-2 envelope protein and spike protein are sensed by Toll-like receptor 2 (TLR2) and TLR4, respectively60,61. Single-stranded viral genomic RNA activates TLR7/8, while the virus’s double-stranded RNA intermediates bind TLR3, and cytosolic RNA is sensed by the RIG-I-like receptors62. Activation of TLRs and RIG-I induces a robust IFN response. As a counter-regulatory mechanism, SARS-CoV-2 proteins, such as ORF9b, non-structural proteins 1 and 13, and the nucleocapsid protein, can inhibit the host IFN response62,63. Differences in IFN signaling have been investigated in children and adults with COVID-19. Adults with severe COVID-19 have lower circulating levels of type I IFNs, which has been attributed to neutralizing antibodies to type I IFNs64,65,66, as well as to reduced type I IFN secretion from plasmacytoid dendritic cells67,68. Pierce et al. compared the immune response in nasal secretions from 12 children and 27 adults hospitalized for moderate to severe COVID-19; children had much milder disease than the adults, requiring no respiratory support and resulting in no mortality52. Compared with adults, the children in this study had increased expression of genes downstream of type I and II IFNs and the NLR family pyrin domain-containing 3 (NLRP3) inflammasome, but both groups had similar viral loads and anti-SARS-CoV-2 IgG and IgA titers in nasal fluid52. In correlating the increased IFN and inflammasome signaling in the nasal mucosa with better clinical outcomes, this study further supports the hypothesis that children are protected from severe disease owing to more robust innate immune responses that enable them to overcome SARS-CoV-2. One caveat is that upper respiratory samples might not represent the cellular composition and gene expression of cells and fluid from lower airways or peripheral blood69,70.

The heightened IFN response in children does not arise exclusively from intrinsic age-associated differences in this pathway. IFN responses are reduced from the time of birth until approximately 2 years of age. Then, between 2 and 5 years of age, the serum cytokine profiles shift toward that of an adult, with increased expression of type I IFNs, reduced IL-6 and IL-23, and increased IL-12 and IL-10 (refs. 71,72). A reduction in the IFN response occurs again in adults >65 years of age, in part due to age-associated reductions in interferon-regulatory factor 8 (IRF8), which is a transcription factor downstream of RIG-I signaling, and proteolysis of tumor receptor-associated factor 3 (TRAF3), which is a signaling protein utilized by multiple TLRs73. Additionally, as individuals age, type I IFN receptors in CD4+ T cells are more frequently in complex with Src homology region 2 domain–containing phosphatase 1 (SHP-1), an inhibitory protein tyrosine phosphatase that restrains type I IFN signaling74,75. Exogenous factors that reduce the IFN response also accumulate with age. These factors include chronic viral infections, such as human cytomegalovirus, known to cause exhaustion of plasmacytoid dendritic cells and reduced IFN secretion, as well as anti-IFN autoantibodies64,76,77,78. Although the effects of reduced IFN signaling are most pronounced in individuals >65 years of age, these factors progressively impair the IFN response and might contribute to the increased frequency of severe COVID-19 in middle-aged and older adults.

T cell homeostasis and activation

T cell lymphopenia in the blood and airway fluid is a marker of COVID-19 severity in children and adults52,70,79. In a study comparing the immune response of 65 children and young adults under 24 years of age with that of 60 adults, all of whom were hospitalized for COVID-19, children had higher absolute lymphocyte counts27. However, compared with children and infants, adults generated increased percentages of spike-specific IFN-γ+CD4+ T cells27. In another study analyzing T cell responses (with 34 children and 36 adults), the overall magnitude of SARS-CoV-2-specific CD4+ and CD8+ T cell responses was lower in the blood of children than in that of adults20. The TCR specificities of CD4+ T cells varied widely between children and adults, with adult T cells being specific mainly for structural proteins. By contrast, CD4+ T cells from children consisted of a substantial proportion that were specific for open reading frame 1ab (Orf1ab), which encodes 16 non-structural proteins20. As some of these non-structural proteins are known to block the host IFN response63, these studies raise the possibility that differences in T cell specificity might contribute to the age-associated decrease in the IFN response to SARS-CoV-2 infection.

Given the dynamic changes affecting T cell development throughout life, multiple age-related differences in T cell number and function probably contribute to the increased frequency of severe COVID-19 in adults relative to that in children. Thymic output is maximal at 1 year of age, and decreases gradually thereafter, with negligible output occurring by ~85 years of age80. The TCR repertoire is another factor that evolves throughout a lifetime. Pathogen-specific memory responses expand as infants and young children encounter new infections, including the formation of tissue-resident memory T cells81. Although some studies have identified T cell cross-reactivity to epitopes of SARS-CoV-2 and those of other human coronaviruses, a definitive correlation between clinical outcomes and T cell epitope cross-reactivity has not been made82. Aging is also associated with shifts in the transcriptional profiles of T cells. Aging mice housed in specific-pathogen-free environments accumulate cytotoxic and effector CD4+ T cells, with increased expression of genes associated with chronic inflammation (such as S100 protein family members), the cytotoxic T cell response to viral infections (such as granzymes and the transcription factor eomesodermin), and T cell exhaustion83. The similarities between T cell dysfunction in aging adults and in those with severe COVID-19 suggest that the gradual onset of T cell dysfunction in adulthood might set the stage for further dysfunctional T cell responses that are associated with severe COVID-19. Although some studies have delved deeply into mechanisms of T cell dysfunction in older people84,85, additional investigations are needed to delineate the threshold of T cell dysfunction associated with susceptibility to severe COVID-19.

Serological responses

SARS-CoV-2 drives the expansion of plasmablasts and plasma cells during both mild and severe COVID-19, even more so than influenza vaccination and infection86,87,88,89. Children and adults hospitalized with COVID-19 have comparable levels of plasmablasts, regardless of need for ventilatory support90. Increased plasmablast frequencies correlate with percentages of proliferating CD4+ T cells, rather than with levels of anti-spike antibodies90,91. Adults with acute severe COVID-19 exhibited increased levels of neutralizing antibodies compared with titers from convalescent adults or pediatric individuals with MIS-C, the latter two of which were comparable in one study70.

The differences in the serological response to COVID-19 in children and adults are concordant with known age-associated differences in B cell development (Fig. 1). The response to T cell–independent antigens takes ~2 years to mature, as infants produce more IgG1 than IgG2, the latter of which preferentially recognizes polysaccharide antigens92,93. Environmental factors also shape B cell development. Clonal expansion and somatic hypermutation begin during infancy and increase with environmental and infectious exposures93,94,95. The rate of somatic hypermutation can be increased, even during infancy, through exposure to infectious pathogens94. However, also important to note is that the severity of COVID-19 in adults also affects the serological response8,9,10,11,12. Indeed, Bartsch et al. showed that adults with severe COVID-19 have increased titers of anti-SARS-CoV-2 IgG and IgA with more functionality in activating complement and phagocytic cells, compared with either adults or children with mild SARS-CoV-2 infections21.

Antibody repertoires also evolve with age and infectious exposures. With respect to other common coronaviruses, including 229E, NL63, OC43, and KHU1, a study of 231 children and 1,168 adults found that children had increased levels of antibodies against the spike, nucleocapsid, and matrix structural proteins, whereas a non-structural protein of unknown function was the most common target of antibodies in adults96. Age-associated differences in epitope reactivity have also been identified in antibodies against other respiratory viruses, including RSV and influenza virus, suggesting that serological immunity is achieved through targeting different viral epitopes in children than in adults97,98,99,100. These differences have been attributed, at least in part, to original antigenic sin or immunological imprinting, in which exposures to specific pathogen epitopes skew subsequent antibody repertoires as a result of the reliance of adaptive immunity on existing memory cells100. Antigenic sin has been shown to contribute to differences in age-associated mortality with influenza virus subtypes101,102. For the response to COVID-19, no definitive associations have been made between prior coronavirus infections and the serological response to SARS-CoV-2. A study of 37 adults with COVID-19 found that serological responses to the spike protein of other betacoronaviruses were inversely proportional to IgM and IgG antibodies against the spike and nucleocapsid proteins of SARS-CoV-2 (ref. 103). Similarly, a study of 232 people with COVID-19 identified a negative correlation between antibody responses to SARS-CoV-2 and other human coronaviruses104. By contrast, a study of individuals hospitalized for COVID-19 (10 young adults 27–39 years of age and 20 adults 69–82 years of age, of whom 9 required ventilation) found positive correlations between antibodies to the spike protein of other human betacoronaviruses and SARS-CoV-2; serological memory to earlier coronaviruses did not correlate with clinical outcomes105. Another study of 65 children and 60 adults did not find significant correlations between serological memory to other human coronaviruses and either age or outcome27.

The heterogeneous outcomes of these studies indicate the complexity of investigating immunological imprinting in diverse human populations with a broad range of environmental and infectious exposures. Future studies with animal models of serial viral infections might provide more-definitive conclusions.

MIS-C

In April 2020, in the wake of viral surges occurring in Italy and the UK106,107,108,109, a number of children presented with mucocutaneous features resembling Kawasaki disease or toxic shock syndrome. Soon after, descriptive reports by Feldstein et al. and Dufort et al., totaling almost 300 cases in the USA, confirmed that the timing was ~4 weeks after the peak of SARS-CoV-2 activity in these regions110,111. This disease entity is now referred to as pediatric inflammatory multisystem syndrome temporally associated with SARS-CoV-2 (PIMS-TS)108 or MIS-C112.

Symptoms and diagnosis

The diagnostic criteria from the US Centers for Disease Control for MIS-C include at least 24 hours of fever and inflammation severe enough to affect at least two organ systems and require hospitalization in an individual <21 years of age who has no alternative diagnosis explaining the illness112. These symptoms must occur during a current or recent SARS-CoV-2 infection or within 4 weeks of exposure to a confirmed COVID-19 case112. At least 75% of children in published cohorts have positive SARS-CoV-2 antibody testing, but fewer (up to 52%) have positive respiratory testing, suggesting the post-infectious timing of MIS-C109,110,111,113. Milder symptoms of MIS-C include fever, rash, and gastrointestinal symptoms, but it can progress to life-threatening cytopenias, coagulopathy, myocardial dysfunction, coronary aneurysms, and shock1,2,3,4,5. The majority of children diagnosed with MIS-C were previously healthy and most were critically ill, mostly from shock and/or left ventricular dysfunction, with less-severe or no respiratory involvement110. Approximately 40% of the children met criteria for Kawasaki disease, a vasculitic syndrome associated with coronary artery aneurysms110. In a study of >1,000 children with MIS-C or severe COVID-19, those with MIS-C were more likely to have cardiovascular and/or mucocutaneous involvement, with higher CRP levels and a higher ratio of neutrophils to lymphocytes113. MIS-C predominantly affects those between 6 and 12 years of age, rather than younger or older individuals113,114. MIS-C has been reported in Europe, North and South America, the Middle East, South Africa, and South Asia115. Relatively fewer cases have been reported from East Asia, which has a high frequency of Kawasaki disease, and it is unclear whether this is due to differences in the genetic background of local populations or environmental conditions115,116. Compared with non-Hispanic white children in the USA, MIS-C occurs more frequently than COVID-19 in non-Hispanic Black children113,114.

Immune cell activation in MIS-C

Cellular activation affecting multiple hematopoietic lineages is a consistent feature of MIS-C. High levels of inflammatory cytokines, including IFN-α, IFN-γ, IL-1β, IL-6, IL-8, IL-10, and IL-17, have been identified in all studies of MIS-C117,118,119,120,121,122,123, but no single archetypal cytokine signature of MIS-C has emerged. Evidence of myeloid cell activation in children with MIS-C includes increased expression of CD64 on monocytes and neutrophils, as well as of the alarmin-related S100A genes, both of which are also elevated in adults with severe COVID-19 (refs. 91,117). Natural killer cells from children with MIS-C have also been shown to be highly active, with higher expression of perforin and granzymes than in those from healthy adults91. Single-cell RNA-seq of peripheral blood mononuclear cells from eight children with MIS-C, compared with four children with COVID-19 pneumonia and nine healthy adults, showed that those with MIS-C and COVID-19 had similarly increased expression of genes downstream of inflammatory cytokines (type I/II IFNs, IL-1, IL-6, and IL-17) and NF-κB activation, even though all children with MIS-C had been treated with immunomodulatory therapies prior to study enrollment123. The inflammatory signature in the MIS-C cohort was enriched for monocytes and dendritic cells, highlighting the contribution of these cells to the systemic inflammation that is characteristic of MIS-C123.

Commensurate with multilineage immune activation, children with MIS-C have tissue inflammation, including increased gastrointestinal permeability and schistocytes; elevated circulating levels of complement C5b9, which is indicative of microangiopathy; and release of troponin and natriuretic peptides, indicative of cardiac inflammation109,110,111,113,118,121,122,124. Children with MIS-C might also have mucocutaneous inflammation and coronary artery aneurysms, features characteristic of Kawasaki disease. However, studies have identified distinctions between the immunological abnormalities associated with these diseases. Compared with individuals with Kawasaki disease, those with MIS-C have more severe lymphopenia, neutropenia, and thrombocytopenia, with increased expression of HLA-DR, a T cell activation marker, and CD57, a marker of CD4+ T cell senescence118,119. Circulating levels of proteins associated with endothelial damage are higher in people with Kawasaki disease than in children with MIS-C119, reflecting the more frequent coronary artery involvement in Kawasaki disease118. These differences might explain why the great majority of coronary artery aneurysms in MIS-C resolve within 30 days, which contrasts with the persistent aneurysms associated with Kawasaki disease.

T cells

The function of T cells in MIS-C pathogenesis has also been investigated. One study found a strong enrichment for patrolling CX3CR1+CD8+ T cells, potentially contributing to the vascular manifestations of MIS-C90. PD1+CD39+CD8+ T cells were also increased in number, nominally higher than levels in adults with severe disease, indicative of prolonged activation and TCR signaling. Three separate studies, each with unique cohorts of children with MIS-C, identified pronounced skewing of the TCR Vβ repertoire91,125,126, with up to 24% of the sequenced clones expressing the TRBV11-2 V-region (Vβ21.3) in one study126. The polyclonal nature of this expansion, highlighted in two studies125,126, is suggestive of superantigen-driven activation. Concordant with these findings, a superantigen-like motif has been identified in the SARS-CoV-2 spike protein sequence127. TCR skewing has been similarly identified in adults with severe COVID-19, although the Vβ signatures identified in adults with severe COVID-19 seem to differ, thus far, from those identified in children with MIS-C127,128. By contrast, TRBV11-2 enrichment was not detected in individuals with mild SARS-CoV-2, Kawasaki disease, or toxic shock syndrome, or in age-matched febrile children91,125,126. After resolution of the disease, the repertoire of T cells from children with MIS-C re-normalized in distribution91. In light of the skewed TCR repertoires evident in some who have MIS-C, studies have investigated post-infectious sources of antigens122,129. Given the frequency of gastrointestinal symptoms in children with MIS-C6,108,109,110,113, the gastrointestinal tract has been hypothesized to function as a reservoir for SARS-CoV-2 (refs. 122,129,130), as the virus is known to infect enterocytes131. Compared with healthy children and those with COVID-19, a cohort of 19 children with MIS-C investigated by Yonker et al. shows evidence of increased gastrointestinal permeability and circulating levels of SARS-CoV-2 spike protein and the S1 subunit, the latter of which correlated with TRBV11-2 expression122. Additional studies with larger, multicenter MIS-C cohorts are needed to understand the generalizability of these findings.

Antibodies

The serological features of MIS-C reflect the post-infectious timing of this disease. Children with MIS-C have IgG antibodies to SARS-CoV-2 with levels and neutralizing capacity similar to those of convalescent adults, irrespective of concomitant RT–PCR positivity21,120. Additionally, higher levels of total and neutralizing antibodies against the spike and RBD proteins have been found in children with MIS-C compared with those with acute COVID-19 (ref. 132). These findings suggest that the nasopharyngeal load of virus reflects an antecedent, rather than acute, infection120,124. In addition to robust antibody responses to SARS-CoV-2, children with MIS-C have a concomitant expansion of antibodies against other common viruses, including common coronaviruses, influenza virus, RSV, and Epstein–Barr virus21. This diversity of antibodies is similar to the antibody profiles of non-hospitalized adults with mild to moderate COVID-19; by contrast, adults with severe COVID-19 requiring hospitalization have more limited expansion of antibodies to pathogens other than SARS-CoV-2 (refs. 21,104). The cause of differences in the breadth of the serological response is not known, but investigators have hypothesized that attenuated immunity with aging leads to a narrower humoral response during SARS-CoV-2 infection104. Children with MIS-C have increased frequencies of proliferating CD27+CD38+ plasmablasts, which correlates with increased percentages of proliferating CD4+ T cells—a feature that diminishes with age90,91.

Differences in age-associated morbidity and mortality to subtypes of respiratory viruses have prompted investigations of how exposure to other coronavirus infections affects susceptibility to MIS-C. A small study of three children with MIS-C identified a lack of antibodies against the common human coronaviruses HKU1 and betacoronavirus 1 (ref. 119). However, subsequent studies have identified comparable or increased levels of antibodies to HKU1 and other seasonal coronaviruses in children hospitalized with MIS-C, compared with those with COVID-19 and healthy children, with no consistent correlation across studies in the relationship between antibodies against seasonal coronaviruses and neutralizing antibodies to SARS-CoV-2 (refs. 133,134).

The end-organ inflammation characteristic of MIS-C has prompted studies investigating the contributions of autoantibodies to the pathogenesis of MIS-C. Increased levels of circulating autoantibodies have been identified in children with MIS-C91,119,120, including in samples taken prior to administration of intravenous immunoglobulins. These autoantibodies recognize antigens in endothelial cells and the gastrointestinal tract, those associated with rheumatological disorders (systemic lupus erythematosus, Sjögren’s disease, and idiopathic inflammatory myopathies), and those against widely expressed tissue antigens (MAPK2 and members of the casein kinase family)91,119,120. Additionally, two studies have found evidence of B cell clonal expansion in children with MIS-C91,135, with the study by Porritt et al. showing that a higher frequency of immunoglobulin heavy-chain genes is associated with autoimmune disorders in individuals with TRBV11-2 expression135. Notably, studies of children with MIS-C have not identified autoantibodies against IFN or IFN-stimulated genes123,136, thus distinguishing autoantibodies in children with MIS-C from those detected in adults with severe COVID-19 (refs. 64,78).

Genetic risk factors

Genetic risk factors associated with susceptibility to MIS-C are also distinct from those associated with severe COVID-19 in adults. Multiple studies of adults with severe COVID-19 have identified an association between severe COVID-19 and loss-of-function variants of TLR7, a sensor for SARS-CoV-2 single-stranded RNA137,138,139,140. TLR7 signaling promotes IFN signaling, the secretion of IL-6, IL-1β. and IL-23, and generation of the TH17 subset of helper T cells in response to viral infections141,142. Thus, reduced function of the protein encoded by TLR7, which is located on the X chromosome, is one mechanism contributing to the increased prevalence of severe COVID-19 in men compared with women143,144,145. In contrast to these risk variants for COVID-19, variants impairing negative regulators of inflammation have been detected in children with MIS-C146,147. These include SOCS1 (encoding suppressor of cytokine signaling 1), XIAP (encoding X-linked inhibitor of apoptosis), and CYBB (encoding cytochrome b-245, beta subunit)147. Even 7 months after recovery, transcriptional evidence of increased IFN and inflammatory signaling can be detected in individuals with these genetic risk variants, thus constituting a counterpoint to the reduced IFN signature associated with severe COVID-19 in adults. Extending beyond monogenic risk factors, studies have identified HLA class I alleles as factors potentially increasing a child’s risk of MIS-C. In the study by Porritt et al., which had a predominance of Hispanic patients from the USA, four of seven patients with MIS-C shared the HLA class I alleles A02, B35, and C04125125. By contrast, this enrichment for specific HLA class I alleles was not found in the sub-cohort of HLA-genotyped patients presented by Moreews et al., which was composed predominantly of children from Afro-Caribbean backgrounds126. Thus, larger cohort studies are necessary to quantify the association between HLA genotype on the susceptibility to MIS-C.

Conclusion

Many limitations of the studies presented here arise from challenges inherent to a pandemic. Children have smaller blood volumes than adults, which limits sample volumes, and the lower prevalence of SARS-CoV-2-associated complications in pediatric populations than in adult cohorts reduced study sample sizes148. The relative rarity of intubation required for children with COVID-19 prompted the use of nasopharyngeal samples as surrogates. The viral load, immune cell composition, and gene expression profiles of nasal samples and airway samples are distinct and not interchangeable69. The immunomodulatory treatments given to patients with MIS-C affects the transcriptional signature of immune cells, as seen in longitudinal studies of MIS-C90,117,119. Historically, these types of limitations have been overcome by the use of animal models of disease to confirm the findings from patient studies. However, owing to interspecies differences in the susceptibility and response to SARS-CoV-2 infection, there are no animal models that recapitulate the two most severe manifestations of SARS-CoV-2 infection in humans: acute respiratory distress syndrome or MIS-C149.

Nevertheless, collectively, these studies show that the immune response of children and adults to mild SARS-CoV-2 infection are similar, but diverge after the development of severe disease (Fig. 3). Adults with severe COVID-19 might have loss of immune synchrony: T cell lymphopenia, reduced circulating follicular helper T cells, and impaired IFN-mediated signaling juxtaposed with high frequencies of activated SARS-CoV-2-specific T cells, robust humoral responses, and myeloid cell activation. By contrast, children with MIS-C exhibit multilineage immune cell activation beyond that which occurs in pediatric or adult COVID-19, evidenced by greater IFN and inflammasome signaling, widespread T cell activation, and activation of a broad humoral response reactive to common viruses as well as SARS-CoV-2. The association of severe COVID-19 in children and adults with pre-existing medical conditions underscores the contributions of these comorbidities to disease severity. Although studies have started to identify genetic risk factors for MIS-C, the constellation of factors that lead to MIS-C are mostly undefined. The spread of viral variants among unvaccinated pediatric populations may also change the spectrum of disease in children. The solutions to these questions will emerge as investigators continue to develop multicenter studies with larger cohorts alongside new tools for studying SARS-CoV-2 pathogenesis.

Fig. 3: Comparison of the immune response to SARS-CoV-2 in children and adults.
figure 3

Age-related differences in the immune response to SARS-CoV-2 infection are most apparent in moderate to severe COVID-19 and MIS-C.

Full size image