Articles Epigenetic profiling linked to multisystem inflammatory syndrome in children (MIS-C): A multicenter, retrospective study Veronica Davalos,a,1 Carlos A. García-Prieto,a,b,1 Gerardo Ferrer,a,c Sergio Aguilera-Albesa,d Juan Valencia-Ramos,e Agustí Rodríguez-Palmero,f,g,h Montserrat Ruiz,f,g Laura Planas-Serra,f,g Iolanda Jordan,i Iosune Alegría,d Patricia Flores-Perez,j onica Cantarín,j Victoria Fumad Ver o,k Maria Teresa Viadero,l Carlos Rodrigo,h Maria Mendez-Hern andez,h Eduardo Lopez-Granados, Roger Colobran, Jacques G. Riviere, Pere Soler-Palacín, Aurora Pujol,f,g,p* and g,m n o o Manel Esteller a,c,p,q** a Josep Carreras Leukaemia Research Institute (IJC), Badalona, Barcelona, Catalonia, Spain b Life Sciences Department, Barcelona Supercomputing Center (BSC), Barcelona, Catalonia, Spain c Centro de Investigacio n Biom edica en Red de Cancer (CIBERONC), Spain d Navarra Health Service Hospital, Pamplona, Spain e University Hospital of Burgos, Burgos, Spain f Neurometabolic Diseases Laboratory, Bellvitge Biomedical Research Institute (IDIBELL), L'Hospitalet de Llobregat, Barcelona, Catalonia, Spain g Centro de Investigaci on Biom edica en Red de Enfermedades Raras (CIBERER), Spain h Germans Trias i Pujol Research Institute (IGTP), Universitat Auto noma de Barcelona (UAB), Badalona, Barcelona, Spain i Pediatric Critical Care Unit, Hospital Universitari Sant Joan de Deu, Barcelona, Catalonia, Spain j Pediatrics Department, Hospital Universitario Nin ~o Jes us, Madrid, Spain k Unitat de Malalties Infeccioses i Importades, Servei de Pediatría, Infectious and Imported Diseases, Pediatric Unit, Hospital Universitari Sant Joan de De u, Barcelona, Catalonia, Spain l Servicio de Pediatría del Hospital Universitario Marqu es de Valdecilla, Santander, Spain m Department of Immunology, La Paz University Hospital, Madrid, Spain; La Paz Institute of Biomedical Research, Madrid, Spain n Immunology Division, Department of Clinical and Molecular Genetics, Hospital Universitari Vall d'Hebron (HUVH), Vall d'Hebron Research Institute (VHIR), Universitat Auto noma de Barcelona, Barcelona, Catalonia, Spain o Pediatric Infectious Diseases and Immunodeficiencies Unit, Hospital Universitari Vall d'Hebron (HUVH), Vall d'Hebron Research Institute (VHIR), Universitat Auto noma de Barcelona, Barcelona, Catalonia, Spain p Instituci o Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Catalonia, Spain q Physiological Sciences Department, School of Medicine and Health Sciences, University of Barcelona (UB), Catalonia, Spain Summary Background Most children and adolescents infected with the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) remain asymptomatic or develop a mild coronavirus disease 2019 (COVID-19) that usually does not eClinicalMedicine 2022;50: 101515 require medical intervention. However, a small proportion of pediatric patients develop a severe clinical condition, Published online 25 June multisystem inflammatory syndrome in children (MIS-C). The involvement of epigenetics in the control of the 2022 immune response and viral activity prompted us to carry out an epigenomic study to uncover target loci regulated by https://doi.org/10.1016/j. DNA methylation that could be altered upon the appearance of MIS-C. eclinm.2022.101515 Methods Peripheral blood samples were recruited from 43 confirmed MIS-C patients. 69 non-COVID-19 pediatric samples and 15 COVID-19 pediatric samples without MIS-C were used as controls. The cases in the two groups were mixed and divided into discovery (MIS-C = 29 and non-MIS-C = 56) and validation (MIS-C = 14 and non-MIS- C = 28) cohorts, and balanced for age, gender and ethnic background. We interrogated 850,000 CpG sites of the human genome for DNA methylation variants. Findings The DNA methylation content of 33 CpG loci was linked with the presence of MIS-C. Of these sites, 18 (54.5%) were located in described genes. The top candidate gene was the immune T-cell mediator ZEB2; and others *Corresponding author at: Neurometabolic Diseases Laboratory, Bellvitge Biomedical Research Institute (IDIBELL), L'Hospitalet de Llobregat, 08908 Barcelona, Catalonia, Spain. **Corresponding author at: Josep Carreras Leukaemia Research Institute (IJC), Carretera de Can Ruti, Camı de les Escoles s/n, 08916 Badalona, Barcelona, Catalonia, Spain. E-mail addresses: apujol@idibell.cat (A. Pujol), mesteller@carrerasresearch.org (M. Esteller). 1 These authors contributed equally. www.thelancet.com Vol 50 Month August, 2022 1 Articles highly ranked candidates included the regulator of natural killer cell functional competence SH2D1B; VWA8, which contains a domain of the Von Willebrand factor A involved in the pediatric hemostasis disease; and human leukocyte antigen complex member HLA-DRB1; in addition to pro-inflammatory genes such as CUL2 and AIM2. The identi- fied loci were used to construct a DNA methylation profile (EPIMISC) that was associated with MIS-C in both cohorts. The EPIMISC signature was also overrepresented in Kawasaki disease patients, a childhood pathology with a possible viral trigger, that shares many of the clinical features of MIS-C. Interpretation We have characterized DNA methylation loci that are associated with MIS-C diagnosis. The identified genes are likely contributors to the characteristic exaggerated host inflammatory response observed in these patients. The described epigenetic signature could also provide new targets for more specific therapies for the disorder. Funding Unstoppable campaign of Josep Carreras Leukaemia Foundation, Fundaci o La Marat o de TV3, Cellex Foundation and CERCA Programme/Generalitat de Catalunya. Copyright Ó 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) Keywords: Multisystem inflammatory syndrome in children; COVID-19; Kawasaki disease; Epigenetics; DNA methylation pathophysiological mechanisms of the disorder, and Research in context highlight its close similarity to Kawasaki disease. The genes identified could also be candidate targets for more precise treatments of the disease. Most impor- Evidence before this study tantly, the assessment of the DNA methylation levels of Most members of the pediatric population infected with these loci can be swiftly added to the measurement of the SARS-CoV-2 virus, which is responsible for the other biochemical and clinical parameters to improve COVID-19 pandemic, escape severe disease. However, early MIS-C diagnosis. in a few cases, a rare and serious health condition, known as multisystem inflammatory syndrome in chil- dren (MIS-C), may occur. The clinical spectrum of MIS-C can affect multiple organ systems, often requiring admission to intensive care unit. Risk factors for the dis- ease are not well defined, and the hyperinflammatory Introduction condition resembles another rare disorder known as In late 2019, an unexpected increase in the number of Kawasaki disease. To our knowledge, this is the first epi- pneumonia cases in China led to the identification of genomic study of MIS-C after acute SARS-CoV-2 infec- the severe acute respiratory syndrome coronavirus 2 tion. Our search of PubMed on January 20th, 2022, (SARS-CoV-2),1 and the subsequent worldwide spread limited to articles in English, but not by date, using the of the derived disease, termed COVID-19. At the time of terms “MIS-C”, “epigenomics”, “DNA methylation”, and “marker”, identified no studies addressing this topic. the writing (January 24th, 2022), more than 350 million confirmed cases and more than 5,6 million deaths have Added value of this study been reported worldwide (https://coronavirus.jhu.edu/ map.html). High mortality of COVID-19 patients with Our results indicate the existence of distinct DNA meth- serious respiratory failure linked to acute respiratory ylation loci that distinguish MIS-C patients from COVID- distress syndrome (ARDS) and interstitial pneumonia 19 pediatric patients without MIS-C, and from healthy children and adolescents without SARS-CoV-2 infection. have been associated with male sex, old age and con- The epigenetic sites found were mostly located within comitant medical conditions, such as diabetes, obesity, genes associated with immune response and pro- hypertension, and cardiovascular pathology.2 In com- inflammatory pathways. Taking advantage of these parison to the presentation in adults, most children and DNA methylation markers, we produced an epigenomic adolescents with SARS-CoV-2 infection are fully asymp- profile that exhibited great accuracy in predicting MIS-C tomatic or have very mild clinical manifestations.3 The diagnosis. We have named this profile the EPIMISC. severity of the disease in the pediatric population also depends on their underlying conditions, and children Implications of all the available evidence may manifest ARDS and pneumonia as do adults.3 Our research has revealed new biomarkers linked to However, a more specific complication appeared in MIS-C onset that provide new information about the April 2020,4 a new and rare syndrome, termed 2 www.thelancet.com Vol 50 Month August, 2022 Articles multisystem inflammatory syndrome in children (MIS- from 15 pediatric COVID-19 cases with no evidence of C). This is also known as pediatric inflammatory multi- MIS-C, and from 69 healthy children and adolescents system syndrome temporally associated with SARS- collected during the pre-COVID-19 era (before Decem- CoV-2 infection (PIMS-TS). MIS-C arises days to weeks ber 2019), in the setting of routine surgical procedures after the initial infection.4−10 Unlike severe adult such as circumcision, orchiopexy, inguinal or umbilical COVID-19 patients, who are characterized by respira- hernia repair, adenoidectomy, tonsillectomy or tym- tory failure, MIS-C patients show a broad spectrum of panic membrane incision; or from unaffected sibling additional clinical features (e.g., rash, fever, abdominal controls collected in previous studies. The data from and/or chest pain, conjunctival hyperemia, etc.) as a these samples are summarized in Tables S1 and S2, result of multiple organ involvement (e.g., the cardio- respectively. The protocol of this retrospective study was vascular, gastrointestinal, mucocutaneous, or hemato- approved by the ethics review boards of the participating logical systems, amongst others). Although it is a rare institutions. Written informed consent was obtained disease, MIS-C is a serious health condition that require from all participants. The study protocol is described in admission to intensive care unit in around 60% of the Supplementary Methods. cases, and ultimately lead to death to a not negligible 2% of cases.5 The exact pathways that give rise to the clinical manifestations of MIS-C, and the factors predis- DNA methylation data and computational analyses posing to development of the disease are largely The DNA methylation status of the studied samples was unknown.410 obtained using the Infinium MethylationEPIC Array In the context of adult COVID-19, in addition to the (»850,000 CpG sites) (Supplementary Methods). The aforementioned concomitant medical conditions,2 MIS-C epigenetic signature, referred to hereafter as EPI- genetic studies suggest that several genetic loci are asso- MISC, was obtained by first identifying the probes differ- ciated with the severity of the disease (summarized in entially methylated between MIS-C cases and healthy Supplementary Materials). We have also recently shown control donors, filtering out in a second step those probes that epigenetic variation, particularly DNA methylation, found to be differentially methylated between pediatric which is altered in many human diseases,11 is also asso- COVID-19 cases and healthy controls (Figure S1). This ciated with adult COVID-19 severity.12 DNA and RNA approach enabled us to effectively discover the differen- viral activity are controlled by DNA methylation tially methylated probes between MIS-C and non-MIS-C changes,11 but more importantly, this epigenetic mark cases. This involved deriving a linear model adjusted by is key to proper immune system activity and could pre- the age covariate with the limma R package (v3.46.0), dict the efficacy of immune-related therapies.13 To inves- using the methylation values of the discovery dataset. A tigate if epigenetic changes are involved in MIS-C, we significance threshold for CpGs with a False Discovery undertook a comprehensive epigenomic study to iden- Rate (FDR) adjusted P-value <0.05 and an absolute mean tify candidate DNA methylation loci linked to the dis- methylation beta difference between groups of >0.15 was ease that distinguish these patients from standard established. The linear model was adjusted by the age COVID-19 pediatric patients and from SARS-CoV-2- covariate after performing a principal component analysis uninfected children and adolescent subjects of the pre- (PCA) that identified disease status and age as the greatest COVID-19 era. sources of variation in our dataset (Supplementary Meth- ods). The significantly differential DNA methylation sites (Table S3) were used to train a supervised classification model based on a ridge-regularized logistic regression to Methods predict MIS-C diagnosis using the glmnet R package Study design and participants (v4.1-1). The classification model was optimized by tuning Whole blood samples and clinical data from 43 patients parameters (best performance with alpha = 0 from ridge with MIS-C were previously collected between April regression, and regularization parameter lambda = 0.1) 16th, 2020 and August 17th, 2021 from seven Hospitals after resampling with 10-fold cross-validation carried out in Spain. MIS-C was diagnosed based on the case defini- three times using the caret package in R (v6.0-86). Once tion provided by the World Health Organization. the model and tuning parameters values have been Briefly, children with clinical and biochemical evidence defined after resampling, our model performance was of inflammation in at least two systems, without other assessed using the receiver operating characteristic (ROC) cause, and with evidence of SARS-CoV2 infection or and calibration curves. Further details are provided in Sup- close contact. The complete description is available in plementary Methods. https://www.who.int/news-room/commentaries/detail/ multisystem-inflammatory-syndrome-in-children-and- adolescents-with-covid-19. Clinicopathological charac- Role of funders teristics of the MIS-C patients studied are summarized Funded by the Josep Carreras Leukaemia Foundation, in Table 1. Whole blood samples were also obtained the Cellex Foundation and the CERCA Programme of www.thelancet.com Vol 50 Month August, 2022 3 Articles Characteristics MIS-C cohorts Discovery cohort (N = 29) Validation cohort (N = 14) Entire cohort (N = 43) Gender - Frequency (%) Female 9 (31.0%) 8 (57.1%) 17 (39.5%) Male 20 (69.0%) 6 (42.9%) 26 (60.5%) Age (years) - Median [range] 8.0 [0.5−17] 6.5 [1−11] 7.0 [0.5−17] Age group- Frequency (%) ≤2 yr 5 (17.2%) 1 (7.1%) 6 (14.0%) 3−5 yr 4 (13.8%) 5 (35.7%) 9 (20.9%) 6−9 yr 7 (24.1%) 5 (35.7%) 12 (27.9%) 10−13 yr 8 (27.6%) 3 (21.4%) 11 (25.6%) 14−18 yr 5 (17.2%) 0 (0.0%) 5 (11.6%) Ethnicity - Frequency (%) West-Eurasia 20 (69.0%) 11 (78.6%) 31 (72.1%) Central-South America 6 (20.7%) 2 (14.3%) 8 (18.6%) African 3 (10.3%) 1 (7.1%) 4 (9.3%) Underlying conditions - Frequency (%) Previously healthy 26 (89.7%) 14 (100%) 40 (93.0%) Asthma 3 (10.3%) 0 (0%) 3 (7.0%) SARS-CoV-2 status- Frequency (%) IgG and/or PCR positive 26 (89.7%) 13 (92.9%) 39 (90.7%) Near contact with SARS-CoV-2 positive* 3 (10.3%) 1 (7.1%) 4 (9.3%) Detection of additional virus - Frequency (%)y 5 (17.2%) 0 (0.0%) 5 (11.6%) Organ system involvement - Frequency (%) Two systems 4 (13.8%) 2 (14.3%) 6 (14.0%) Three systems 5 (17.2%) 6 (42.9%) 11 (25.6%) Four or more systems 20 (69.0%) 6 (42.9%) 26 (60.5%) Gastrointestinal involvement # - Frequency (%) 26 (89.7%) 11 (78.6%) 37 (86.0%) Respiratory involvement # - Frequency (%) 21 (72.4%) 6 (42.9%) 27 (62.8%) Cardiovascular involvement # - Frequency (%) 22 (75.9%) 11 (78.6%) 33 (76.7%) Mucocutaneous involvement # - Frequency (%) 19 (65.5%) 10 (71.4%) 29 (67.4%) Hematologic involvement # - Frequency (%) 21 (72.4%) 8 (57.1%) 29 (67.4%) Neurologic involvement # - Frequency (%) 6 (20.7%) 2 (14.3%) 8 (18.6%) Renal involvement # - Frequency (%) 3 (10.3%) 0 (0.0%) 3 (7.0%) Musculoskeletal involvement # - Frequency (%) 1 (3.4%) 1 (7.1%) 2 (4.7%) Highest level of care - Frequency (%) Home 1 (3.4%) 0 (0.0%) 1 (2.3%) Ward 8 (27.6%) 9 (64.3%) 17 (39.5%) Intensive care unit 20 (69.0%) 5 (35.7%) 25 (58.1%) Oxygen supplementation - Frequency (%) None 9 (31.0%) 10 (71.4%) 19 (44.2%) Nasal cannula 7 (24.1%) 1 (7.1%) 8 (18.6%) Non-Invasive Ventilation or High Flow Oxygen 10 (34.5%) 2 (14.3%) 12 (27.9%) Mechanical Ventilation 2 (6.9%) 1 (7.1%) 3 (7.0%) Extracorporeal membrane oxygenation 1 (3.4%) 0 (0%) 1 (2.3%) Table 1: Clinicopathological characteristics of the studied MIS-C patients. * For cases of unknown SARS-CoV-2 status. y Additional viruses: Parainfluenza virus type 4 (HPIV-4), Rhinovirus/Enterovirus (HRV/ENT) and Adenovirus. # Following the definitions used for organ involvement in Feldstein et al., N Engl J Med, 2020. the Generalitat de Catalunya. Additional support was analysis, data interpretation, or the writing of the manu- provided by the Fundacio La Marat o de TV3 (202131-32- script. The authors collected the data, and had full 33), MCIU/AEI/FEDER (RTI2018-094049-B-I00) and access to all of the data in the study. They also took the AGAUR (2017SGR1080). The sponsors of the study final decision and had responsibility for submitting the had no role in the study design, data collection, data study results for publication. 4 www.thelancet.com Vol 50 Month August, 2022 Articles Characteristics Cohorts Discovery cohort Validation cohort Entire cohort (N = 85) (N = 42) (N = 127) Cases - Frequency (%) MIS-C 29 (34.1%) 14 (33.3%) 43 (33.9%) Non-MIS-C 56 (65.9%) 28 (66.7%) 84 (66.1%) Gender - Frequency (%) Female 36 (42.4%) 22 (52.4%) 58 (45.7%) Male 49 (57.6%) 20 (47.6%) 69 (54.3%) Age (years) - Median [range] 9.0 [0−17] 7.5 [0−16] 8.0 [0 - 17] Age group- Frequency (%) ≤2 yr 11 (12.9%) 3 (7.1%) 14 (11.0%) 3−5 yr 11 (12.9%) 13 (31.0%) 24 (18.9%) 6−9 yr 22 (25.9%) 12 (28.6%) 34 (26.8%) 10−13 yr 23 (27.1%) 7 (16.7%) 30 (23.6%) 14−18 yr 18 (21.2%) 7 (16.7%) 25 (19.7%) Ethnicity - Frequency (%) West-Eurasia 61 (71.8%) 25 (59.5%) 86 (67.7%) Central-South America 7 (8.2%) 4 (9.5%) 11 (8.7%) African 4 (4.7%) 2 (4.8%) 6 (4.7%) Unknown 13 (15.3%) 11 (26.2%) 24 (18.9%) Table 2: Clinicopathological characteristics of the discovery and validation cohorts. Results To optimize our analyses, we compared the MIS-C group (n = 43) with the non-MIS-C group (n = 84). The Patients and epigenomic study latter group comprised the pediatric COVID-19 cases Between April 16th, 2020 and August 17th, 2021, we without MIS-C (n = 15) and the pediatric controls obtained whole blood samples from 43 patients diag- obtained before the COVID-19 pandemic (n = 69). The nosed with MIS-C, using the case definition provided by 127 samples collected were divided into discovery and the World Health Organization and summarized in validation cohorts (85 and 42 cases, respectively) Methods. Table 1 lists the clinicopathological features of (Table 2). There were no significant differences between the MIS-C patients studied. MIS-C-associated laboratory the two cohorts with respect to the frequencies of MIS- findings for these cases are summarized in Figure S2. C and non-MIS-C cases (Fisher’s exact test, P = 1), gen- Overall, the median age was 7.0 years old (Interquartile der (Fisher’s exact test, P = 0.345), age (Mann−Whitney range, IQR = 7), and the majority of the children were −Wilcoxon test, P = 0.282) and ethnicity (Fisher’s exact male (26 cases; 60.5%) and from a West-Eurasian eth- test, P = 0.579) (Table 2). DNA from the whole blood nic background (31 cases, 72.1%). Most patients exhib- samples was purified for all cases and analyzed to deter- ited a previous healthy status (40 cases, 93%) and IgG mine DNA methylation status. The study aimed to char- and/or PCR positivity for SARS-CoV-2 (39 cases, acterize those genomic sites with a distinct DNA 90.7%). Most cases had affectation of four or more of methylation status in MIS-C patients compared with their organ systems (26 cases, 60.5%), and were admit- the non-MIS-C population. The overall study design is ted to an intensive care unit (25 cases, 58.1%). As previ- illustrated in Figure S1. ously described in other MIS-C series, only a few cases presented prominent respiratory symptoms that required mechanical ventilation (3 cases, 7%), in con- Epigenomic analysis of MIS-C in the discovery cohort trast to the classic severe COVID-19 illness, which Using the experimental and bioinformatic pipeline often requires active and interventional oxygen sup- shown in Figure S1 and described in Supplementary plementation. We also collected whole blood samples Methods, the DNA methylation analysis of 85 pediatric from 15 pediatric COVID-19 patients with IgG-posi- individuals in the discovery cohort identified 33 CpG tive and/or PCR-positive status for SARS-CoV-2, but sites with a distinct methylation status between MIS-C without MIS-C (Table S1). Finally, we obtained whole (n = 29) and non-MIS-C (n = 56) cases (Table S3). The blood samples from 69 children collected before Volcano plot of the fully adjusted P-values from the December 2019, when the COVID-19 disease first DNA methylation loci linked to MIS-C diagnosis in the appeared (Table S2). discovery cohort is shown in Figure 1. The genomic www.thelancet.com Vol 50 Month August, 2022 5 Articles Figure 1. The volcano plot shows significant differences in the DNA methylation status of 850K CpG sites between MIS-C and non-MISC using the described experimental and bioinformatic pipeline. Y-axis shows the -log10 P-value and X-axis shows the mean methylation difference according to beta value. A total of 33 CpGs with a delta beta >0.15 and FDR adjusted P-value <0.05 are shown in red. For those with an associated coding sequence, the gene name is also indicated. CpG-sites that exhibited a methylation beta value difference <0.15 and/or FDR adjusted P-value >0.05 are shown in grey. Dashed lines indicate cut-offs for sig- nificance. annotation of these differentially methylated 33 CpG coupled receptor 111 (GPR111) gene, the SH2 domain sites is described in Table S3. Fifteen (45.45%) of the containing 1B (SH2D1B) gene and the ubiquitin-protein identified sites were located in regions of the genome ligase component Cullin-2 (CUL2) exhibit activities that with no currently annotated gene sequences; three could directly relate to MIS-C (Table 3). ZEB2 promotes (9.1%) were associated with three long non-coding terminal differentiation of effector and memory T cell RNAs (LINC00880, LOC645434, LOC100996286); populations during infections and the development of and the other 15 (45.45%) CpG loci were located within plasmacytoid dendritic cells, monocytes, B-cells, natural 15 known protein-coding genes (Table 3). killer cells, and macrophages.14 GPR111 is involved in To investigate further the activities of the 15 candi- tolerance induction, granulopoiesis and the control of date coding genes identified by the MIS-C DNA methyl- cytotoxicity.15 SH2D1B is a unique adaptor protein that ation screening, we performed an enrichment analysis enhances innate and adaptive immune responses to (Supplementary Methods). Significantly enriched Gene antigens.16,17 In this regard, the SH2D1B signaling Ontology (GO) biological processes (hypergeometric pathway has the potential to be co-opted to produce test, FDR adjusted P-value < 0.05) included “regulation enhanced vaccination responses.16 CUL2 is a mediator of inflammatory response to antigenic stimulus” and of inflammation and, in this regard, its pharmacological “regulation of immune response”. All these enriched inhibition protects against hyperinflammatory processes and pathways indicate that a broad exag- responses,18 a finding that could be relevant for those gerated engagement of the immune response to the MIS-C patients that do not respond to the standard SARS-CoV-2 infection contributes to the characteris- treatment. tic hyperinflammatory clinical picture observed in Of the other genes with a distinct DNA methylation these children. profile in MIS-C patients (Table 3), the cases of AIM2 Of the 15 candidate coding genes derived from the (absent in melanoma 2) and PM20D1 (peptidase M20 MIS-C epigenomic analysis, among the highest ranked domain-containing 1) are particularly interesting coding genes according to the DNA methylation differ- because methylation events at these loci are also charac- ence and adjusted P-value derived from the MIS-C teristic of adults who develop severe COVID-19 epigenomic analysis (Table S3), the zinc finger E-box disease.12 The AIM2 gene is related to the hyperinflam- binding homeobox 2 (ZEB2) gene, the G protein- matory manifestation of MIS-C patients since triggers 6 www.thelancet.com Vol 50 Month August, 2022 Articles Gene symbol Gene name Gene function Adjusted P-value ADCY3 Adenylate cyclase 3 Catalyzes the synthesis of cyclic AMP (cAMP) from <0.001 ATP. ADCY3 variants have been associated to risk/susceptibility to obesity, diabetes and chronic inflammatory diseases. AIM2 Absent in melanoma 2 Assembles the macromolecular inflammasome <0.001 complex. CUL2 Cullin 2 Mediator of inflammation. 0.0096 CYREN Cell cycle regulator of NHEJ Cell-cycle-specific regulator of classical non- 0.0136 homologous end joining (NHEJ) of DNA double- strand break (DSB) repair. GPR111 G Protein-Coupled Receptor 111 Member of the adhesion G protein-coupled recep- <0.001 tors (aGPCRs). HLA-DRB1 Major histocompatibility com- Encodes a beta chain of antigen-presenting major 0.0421 plex, class II, DR beta 1 histocompatibility complex class II (MHCII) molecule. KIF13A Kinesin family member 13A Motor protein that also mediates the trafficking of <0.001 influenza A virus ribonucleoproteins, and trans- port of an arenavirus protein. NDST2 N-deacetylase and N-sulfotrans- Enzyme with dual functions in processing glucos- 0.0010 ferase 2 amine and heparin polymers. PM20D1 Peptidase M20 domain contain- Enzyme that regulates the production of N-fatty- 0.0328 ing 1 acyl amino acids. Considered a metabolic dis- ease-associated gene also linked to neurode- generative disorders. RARG Retinoic acid receptor gamma Receptor for retinoic acid. Act as transcriptional <0.001 regulator. SH2D1B SH2 domain containing 1B Adaptor protein for the signaling lymphocytic acti- <0.001 vation molecule family of receptors that enhan- ces immune responses to antigens, including viral proteins such as HIV-Gag. SSUH2 Ssu-2 homolog A putative chaperone protein. 0.0092 VWA8 Von Willebrand factor A domain- Mitochondrial ATPase protein. <0.001 containing protein 8 ZAK ZAK1 Homolog, Leucine Zipper Mitogen-activated protein kinase, also known as <0.001 And Sterile-Alpha Motif Kinase MAP3K20. ZEB2 Zinc finger E-box binding ZEB2 is a DNA-binding transcriptional repressor. <0.001 homeobox 2 Table 3: Epigenetic changes in coding genes associated with MIS-C diagnosis. caspase-1 and unleashes pro-inflammatory cytokines clinical severity of adult COVID-19 cases,22−24 and such as IL-1b and IL-18,19 which are also involved in the CD8+ T-cells from critically ill adult COVID-19 patients innate immune response to viral infections. Regarding show upregulation of the HLA-DRB1 gene.25 PM20D1, recent data suggest that it contributes to auto- We also investigated whether the DNA methylation immune disorders and allergies,20,21 all of which are status of MIS-C was distinct from that of non-MIS-C pathologies with an important hyperinflammatory com- groups for genes that, according to the literature, are ponent. In this study, we have identified that a DNA likely candidates for adult COVID-19. The 47 genes ana- methylation site of the HLA-DRB1 (major histocompati- lyzed were the ACE2 receptor and TMPRSS2 protease, bility complex, class II, DR beta 1) gene is linked to GWAS-derived genes, genes associated with inborn MIS-C. Interestingly, our study of adult COVID-19 errors of type I IFN immunity in cases with life-threat- cases identified that an epigenetic mark in HLA-C ening COVID-19, and other genes involved in immune (major histocompatibility complex, class I, C) was asso- host-cell pathways (Table S4). Only one gene, VWA8 ciated with the severe disease.12 Importantly, allelic gen- (Von Willebrand factor A domain-containing protein 8), otypes of HLA-DRB1 have been associated with the was shared in the list of COVID-19 associated loci www.thelancet.com Vol 50 Month August, 2022 7 Articles (Table S4) and in our MIS-C associated DNA methyla- clustering using the EPIMISC signature differentiated tion sites (Table 3). A single nucleotide polymorphism two branches that were significantly enriched with in VWA8 has been linked to hospitalized cases in respect to each condition, MIS-C vs. non-MIS-C (Fish- COVID-19 cases.26 For MIS-C genetic susceptibility er’s exact test, P = 4.3e-09) (Figure S4). Most important, very little is known. Three genes with reported sequence we found that the EPIMISC signature kept its value in variants for MIS-C (SOCS1, XIAP and CYBB)27 were not our validation cohort, being associated with the disease differentially methylated in our cohorts (Table S5), in with a specificity of 92.86% (95% CI = 76.50% to line with the idea that, for the same candidate target, 99.12%), a sensitivity of 85.71% (95% CI = 57.19% to genetic and epigenetic alteration are usually mutually 98.22%), and PPV and NPV of 85.71% (95% exclusive. CI = 57.19% to 98.22%) and 92.86% (95% CI = 76.50% to 99.12%), respectively. The accuracy was 90.48% (95% CI = 77.38% to 97.34%) and the Kappa value was Testing MIS-C-associated DNA methylation markers in 0.7857 (95% CI = 0.5865 to 0.9849). The ROC curve the validation cohort, and development of the EPIMISC and AUC [89.29% (95% CI = 78.61% to 99.97%)] signature alongside the calibration curve were also determined The DNA methylation status of single CpG sites linked (Figure S5). The EPIMISC signature in the validation to the presence of MIS-C in the discovery cohort cohort also distinguished two branches with respect to (n = 85) was confirmed in the validation cohort (n = 42). MIS-C and non-MIS-C samples (Fisher’s exact test, Overall, when we individually analyzed the 33 CpGs P = 3.1e-06) (Figure S6). Finally, for the entire cohort, whose DNA methylation levels differed significantly EPIMISC was associated with MIS-C diagnosis with a between the MIS-C and non-MIS-C cases, 20 (60.6%) specificity of 96.43% (95% CI = 89.92% to 99.26%), a were also significantly associated with the severe pediat- sensitivity of 90.70% (95% CI = 77.86% to 97.41%), ric disorder in the validation cohort (Table S6). Of these and PPV and NPV of 92.86% (95% CI = 80.52% to 20 CpG sites, seven loci were located in the aforemen- 98.50%) and 95.29% (95% CI = 88.39% to 98.70%), tioned gene coding-containing sequences (35%). Impor- respectively. Its accuracy was 94.49% (95% tantly, when we interrogated all the samples as an CI = 88.97% to 97.76%) and the Kappa value was entire set, comprising the discovery and validation 0.8762 (95% CI = 0.7872 to 0.9653). The ROC curve cohorts (n = 127), 24 of 33 (72.7%) individual CpG and AUC (93.56% [95% CI = 88.74% to 98.39%)] along- sites remained associated with MIS-C development side the calibration curve were also determined (Figure (Table S6). S7). The application of the EPIMISC signature for the The discovery of single DNA methylation sites linked entire cohort also classified samples as MIS-C or non- to the presence of MIS-C could be very helpful, but the MIS-C (Fisher’s exact test, P = 6.5e-14) (Figure 2). The establishment of an overall epigenomic signature could five cases with concomitant viral infections (Table 1) also be of great value to our understanding of the patho- were all of them EPIMISC positive, whereas the epige- physiological basis of the diseases and its clinical man- nomic signature was observed in 2 of 4 (50%) cases clas- agement. To achieve this, we selected the 33 sified clinically as MIS-C but without any biological significantly differential DNA methylation sites that probe of SARS-CoV-2 infection (Table 1). were associated with the occurrence of MIS-C (Table S3) To further assess the specificity of the EPIMISC sig- to train our discovery set, using a supervised classifica- nature for the disease, we run our classification model tion model based on ridge-regularized logistic regres- to establish whether it was also overrepresented in avail- sion (see Supplementary Methods). By this method, we able public DNA methylation datasets (GEO data reposi- obtained a DNA methylation signature, hereafter tory) for other distinct pediatric inflammatory disorders. referred to as EPIMISC, that was associated with MIS-C We found that the EPIMISC signature was not present diagnosis (EPIMISC positive). It had a specificity of in juvenile localized scleroderma (GEO GSE175379), 98.21% (95% confidence interval CI = 90.45% to juvenile systemic sclerosis (GEO GSE175379), or atopic 99.95%), a sensitivity of 93.10% (95% CI = 77.23% to dermatitis (GEO GSE152084). Similarly, the EPIMISC 99.15%), and positive and negative predictive values signature was almost non-existent in the general popu- (PPV and NPV) of 96.43% (95% CI = 81.65% to lation (0.4%, 1 of 241 donors) (GEO GSE142512; GEO 99.91%) and 96.49% (95% CI = 87.89% to 99.57%), GSE132181). These samples were collected before the respectively. Its accuracy was 96.47% (95% emergence of COVID-19, so the donors had never been CI = 90.03% to 99.27%) and the Kappa value was exposed to the SARS-CoV-2 virus. Our observation that 0.9208 (95% CI = 0.8329 to 1). We also plotted the two of the targeted methylated genes within EPIMISC Receiver Operating Characteristic (ROC) curve and cal- were shared with severe adult COVID-19 cases (AIM2 culated the Area Under the Curve [AUC = 95.66% and PM20D1) prompted us to investigate whether the (95% CI = 90.65% to 100%)] together with the calibra- EPIMISC signature was also present in non-pediatric tion curve to further assess and visualize the model’s COVID-19 cases.12 We found that although EPIMISC performance (Figure S3). Supervised hierarchical was almost completely absent from asymptomatic and 8 www.thelancet.com Vol 50 Month August, 2022 Articles Figure 2. Heatmap representing the entire cohort of MIS-C and non-MIS-C cases clustered by methylation beta values of the 33 CpGs defining the EPIMISC signature. Cluster analysis was performed using the Ward.D clustering method and assuming Manhattan distances. mild adult COVID-19 patients (1%, 2 of 194), it was rhinovirus/enterovirus (33%), influenza A (17%), meta- present in 24.9% (53 of 213) of adult COVID-19 patients pneumovirus (13%), influenza B (11%), other coronavi- with clinical severity. MIS-C and critically-ill COVID-19 rus (11%), respiratory syncytial virus (9%), patients show some distinct clinicopathological charac- parainfluenza (5%), and adenovirus (2%) cases. Most teristics, but also some commonalities. This last obser- important, the EPIMISC signature was absent in all vation can relate to the targeting of similar cellular HIV cases (n = 70) analyzed in a recently published networks. For example, the activation of the inflamma- cohort (GSE140800; Ref.30). Thus, overall, these results some pathobiological pathway represented by the AIM2 further support the specificity of the EPIMISC gene occurs in COVID-19 adult patients28 and it was signature. also associated with the severity of the disease in adult MIS-C is considered a new pediatric inflammatory cases12 and, at the same time, the AIM2 gene is a key entity associated with SARS-CoV-2 infection, but there component of the EPIMISC signature identified herein. is clinical overlap with another disorder, Kawasaki But targeting of distinct pathobiological pathways disease,3134 a childhood febrile and systemic vasculitis between both disorders also occur. For example the thought to be triggered by exposure to a novel ribonu- genes in our EPIMISC signature showed a significant cleic acid, as occurs in viral infections. This is a similar enriched Gene Ontology (GO) biological processes scenario to that presented by SARS-CoV-2 and MIS-C. (hypergeometric test, adjusted P < 0.05) of “regulation Remarkably, when we run our classification model to of natural killer cell mediated immunity (GO:0002715)” assess the presence of the EPIMISC signature in DNA and “peptide antigen assembly with MHC protein com- methylation profiles of Kawasaki disease patients avail- plex (GO:0002501)” that were not observed in critically- able in the GEO database (GEO GSE84624),35 we found ill COVID-19 adults.12 Thus, epigenetic and clinical the EPIMISC signature in 95.8% (23 of 24) of the cases. commonalities and singularities between both disorders There are other similarities between the two disorders. occur. Finally, we also wondered if the identified epige- For example, beta-catenin contributed to the pathogene- nomic profile was overrepresented in diseases involving sis of Kawasaki disease,35 and a highly ranked gene of other viral infections. We observed that the EPIMISC our EPIMISC signature was ZEB2. This gene is signature was present only in 7.8% (5 out 64) of patients involved in the epithelial-mesenchymal transition with other viral respiratory infections (GSE167202; (EMT), as it also occurs with the Wnt/beta-catenin sig- Ref.29). The interrogated GSE167202 cohort included naling, but it is also essential for regulating www.thelancet.com Vol 50 Month August, 2022 9 Articles hematopoiesis.36 Another example of features common frequently affected in MIS-C, respiratory function is to the two clinical entities was the suggested activation also commonly compromised, with a wide spectrum of of neutrophils in Kawasaki disease.37 Using a deconvo- consequences, from simple cough and shortness of lution approach to calculate hematological cell popula- breath to a requirement for mechanical ventilation.5,7 tions (Supplementary Methods), we found that our The overlap between the epigenomic landscapes that MIS-C cases were also enriched in neutrophils relative we found to be associated with MIS-C and Kawasaki dis- to the non-MIS-C cases (Mann−Whitney−Wilcoxon ease might have also consequences for understanding test, P = 1.7e-05). Finally, although the EPIMISC signa- the mechanisms involved in the onset of both condi- ture was overrepresented in the Kawasaki disease, four tions.33 Our findings are in line with the reported CpGs were distinctly methylated between both disor- appearance of Kawasaki’s disease-like features in at least ders. One site was not associated with any known gene 40% of MIS-C patients.5 MIS-C patients with Kawasaki (cg16729631), and another was located in the GPR111 disease-like features are frequently under 5 years of gene, which was a highly ranked candidate for the MIS- age,5 similar to the age of Kawasaki disease patients. In C cases. The other two sites were located in SSUH2 fact, 35% of our patients were younger than 5 years old. (ssu-2 homolog), a protein chaperone,38 and RARG (ret- The high degree of enrichment of the EPIMISC signa- inoic acid receptor gamma) which is associated with ture in Kawasaki disease reinforces the invoked role of rubella virus-induced cytokine immune responses.39 viral infection in this disorder, as it is also suggested by These data are germane to similar findings showing the peak in cases following the 2009 influenza A H1N1 that MIS-C and Kawasaki disease share many inflam- pandemic.32 Since MIS-C and Kawasaki disease have matory biomarkers, but others are unique such as the similar underlying DNA methylation defects, epigenetic high concentration of IFN-gamma-induced CXCL9 in drugs combined with immunomodulatory agents, tar- MIS-C cases.34 The DNA methylation analysis of 33 geting viral mimicry and inflammation, could be reported GWAS-derived candidate genes for Kawasaki assessed. disease did not show any significant CpG methylation Limitations of the study are mainly associated with difference in our cohorts (Table S7). Overall, our results sample availability, since MIS-C is a rare and novel dis- highlight the close epigenetic resemblance of MIS-C ease. First, the number of cases is relatively low, and Kawasaki disease, further suggesting that a viral although it is in line with previous studies defining infection could unleash the plethora of clinical manifes- molecular profiles in MIS-C.27,40−47 It should be tations that they share. highlighted, to the best of our knowledge, this is the first epigenomic profiling of MIS-C cases. A second lim- itation is the lack of ethnic heterogeneity, directly Discussion related to the ethnic distribution in the studied popula- To the best of our knowledge, this is the first study to tion, enriched in West-Eurasia origin. This fact could establish the epigenomic profile of MIS-C patients upon underestimate key intrinsic features of other popula- diagnosis. Gene Ontology analyses showed enrichment tions showing potential enhanced risk of MIS-C in pre- of the differentially methylated sites in genes associated vious studies, such as black children.7,48 A third with an immune response triggered by the SARS-CoV-2 limitation to consider is the possible existence of addi- infection. This immune overreaction may well explain tional unmeasured confounding factor, other than age the hyperinflammatory phenotype manifested in these or those that were not statistically significant in our children and why multiple body organs and tissues are analysis. affected. ZEB2, GPR111, SH2D1B, and HLA-DRB1 are In conclusion, we report that MIS-C patients exhibit examples of targeted genes, all of which are involved in a well-defined set of epigenetic loci that are associated the generation of immune and inflammatory responses with the diagnosis of the disorder and support a direct to virus. Interestingly, the EPIMISC signature that we role of a hyperactivated immune response in the charac- found to be associated with the presence of MIS-C in teristic features of overinflammation and multisystem the discovery and validation cohorts was not linked to organ involvement. These DNA methylation sites were other pediatric inflammatory disorders that occur with- used to construct an epigenomic signature, EPIMISC, out involvement of a viral agent. that is associated with the disease. This profile was Some of the identified MIS-C epigenetic targets, absent in non-viral inflammatory processes in children, such as AIM2 and PM20D1, and the EPIMISC signa- but present to a certain degree in severe adult COVID- ture overall, are also present in some severe adult 19 cases. The profile overlaps with that of another COVID-19 cases, confirming that both processes (MIS- inflammatory syndrome, Kawasaki disease, where a C in pediatrics and severe acute respiratory distress syn- trigger by viral infection can now also be further drome in adults) are inflammatory post-infectious com- strengthened. These findings provide essential clues plications and probably could be differently treated than that will help us to understand the immune mecha- the initial phase of the viral infection. Although the gas- nisms that go awry in MIS-C cases, to identify patients trointestinal and cardiovascular systems are the most likely to have worse outcomes, and to suggest actionable 10 www.thelancet.com Vol 50 Month August, 2022 Articles candidate genes for more specific treatments. Together Supplementary materials with genetic, serological and clinical parameters,49 the Supplementary material associated with this article can EPIMISC signature could help in patient stratification be found in the online version at doi:10.1016/j. and to identify highly susceptible patients who require eclinm.2022.101515. close attention and early active treatments to prevent the progression of the disease. Most importantly, we have identified new biomarkers for diagnosing MIS-C References 1 Zhu N, Zhang D, Wang W, et al. A novel coronavirus from patients that could be useful as the COVID-19 pandemic patients with pneumonia in China, 2019. N Engl J Med. progresses and seroconversion increases, reducing the 2020;382:727–733. 2 Zhou F, Yu T, Du R, et al. Clinical course and risk factors for mor- value of knowing the history of exposure and serology tality of adult inpatients with COVID-19 in Wuhan, China: a retro- for defining the MIS-C clinical entity, a key point once spective cohort study. Lancet. 2020;395:1054–1062. COVID-19 turns into an endemic disease worldwide. 3 G€otzinger F, Santiago-Garcıa B, Noguera-Julian A, et al. COVID-19 in children and adolescents in Europe: a multinational, multicentre Finally, the identified epigenetic sites could be useful cohort study. Lancet Child Adolesc Health. 2020;4:653–661. for following up these patients, including how we moni- 4 Whittaker E, Bamford A, Kenny J, et al. Clinical characteristics of tor the efficacy of immunomodulation therapies and 58 children with a pediatric inflammatory multisystem syndrome temporally associated with SARS-CoV-2. JAMA. 2020;324:259– how we can detect at an early stage the MIS-C cases 269. whose disease will rapidly worsen. 5 Feldstein LR, Rose EB, Horwitz SM, et al. Multisystem inflamma- tory syndrome in U.S. children and adolescents. N Engl J Med. 2020;383:334–346. 6 Jiang L, Tang K, Levin M, et al. COVID-19 and multisystem inflam- matory syndrome in children and adolescents. Lancet Infect Dis. Contributors 2020;20:e276–e288. VD, CAGP, AP, and ME designed the study, contributed 7 Abrams JY, Oster ME, Godfred-Cato SE, et al. Factors linked to to the analysis, and wrote the first draft of the manu- severe outcomes in multisystem inflammatory syndrome in chil- dren (MIS-C) in the USA: a retrospective surveillance study. Lancet script. In-depth clinical and pathological characteriza- Child Adolesc Health. 2021;5:323–331. tion and recruitment of patients were carried out by GF, 8 Penner J, Abdel-Mannan O, Grant K, et al. 6-month multidisciplin- SAA, JVR, ARP, MR, LPS, IJ, IA, PFP, VC, VF, ary follow-up and outcomes of patients with paediatric inflamma- tory multisystem syndrome (PIMS-TS) at a UK tertiary paediatric MTV, CR, MMH, ELG, RC, JGR, PSP and AP. All hospital: a retrospective cohort study. Lancet Child Adolesc Health. authors helped draft the manuscript or revised it crit- 2021;5:473–482. 9 Harwood R, Allin B, Jones CE, et al. A national consensus manage- ically for significant intellectual content, and made ment pathway for paediatric inflammatory multisystem syndrome substantial contributions to the concept and design temporally associated with COVID-19 (PIMS-TS): results of a of the study, and to the acquisition, analysis and national Delphi process. Lancet Child Adolesc Health. 2021;5:133– 141. interpretation of data. 10 Ahmed M, Advani S, Moreira A, et al. Multisystem inflammatory syndrome in children: a systematic review. EClinicalMedicine. 2020;26:100527. 11 Berdasco M, Esteller M. Clinical epigenetics: seizing opportunities Data sharing statement for translation. Nat Rev Genet. 2019;20:109–127. 12 Castro de Moura M, Davalos V, Planas-Serra L, et al. Epigenome- The complete DNA methylation raw data of the all the wide association study of COVID-19 severity with respiratory fail- studied MIS-C and non-MIS-C samples cases are avail- ure. EBioMedicine. 2021;66:103339.  13 Villanueva L, Alvarez-Errico D, Esteller M. The contribution of epi- able on the GEO repository under accession number genetics to cancer immunotherapy. Trends Immunol. 2020;41:676– GSE193879. 691. 14 Scott CL, Omilusik KD. ZEBs: novel players in immune cell devel- opment and function. Trends Immunol. 2019;40:431–446. 15 Hamann J, Hsiao CC, Lee CS, Ravichandran KS, Lin HH. Adhe- Declaration of interests sion GPCRs as modulators of immune cell function. Handb Exp Pharmacol. 2016;234:329–350. Dr. Esteller declares grants from Ferrer International, 16 O'Connell P, Amalfitano A, Aldhamen YA. SLAM family receptor personal fees from Quimatryx, outside the submitted signaling in viral infections: HIV and beyond. Vaccines (Basel). work. Dr. Riviere reports personal fees from Grifols, 2019;7:184. 17 Aldhamen YA, Seregin SS, Schuldt NJ, et al. Vaccines expressing CSL behring and Takeda, outside the submitted work. the innate immune modulator EAT-2 elicit potent effector memory The other authors declare no conflicts of interest. T lymphocyte responses despite pre-existing vaccine immunity. J Immunol. 2012;189:1349–1359. 18 Curtis VF, Ehrentraut SF, Campbell EL, et al. Stabilization of HIF through inhibition of Cullin-2 neddylation is protective in mucosal inflammatory responses. FASEB J. 2015;29:208–215. Acknowledgements 19 Kumari P, Russo AJ, Shivcharan S, Rathinam VA. AIM2 in health We thank the Health Department and the Centres de and disease: Inflammasome and beyond. Immunol Rev. Recerca de Catalunya (CERCA) Programme of the Gen- 2020;297:83–95. 20 Li X, Zhao X, Xing J, Li J, et al. Different epigenome regulation and eralitat de Catalunya, the Josep Carreras Leukaemia transcriptome expression of CD4+ and CD8+ T cells from monozy- Foundation, Fundacio La Marato de TV3 and the Cellex gotic twins discordant for psoriasis. Australas J Dermatol. 2020;61: Foundation for institutional support. We also wish to e388–e394. 21 Imran S, Neeland MR, Koplin J, et al. Epigenetic programming thank all the patients, family members and staff from underpins B-cell dysfunction in peanut and multi-food allergy. all the units that participated in the study. Clin Transl Immunol. 2021;10:e1324. www.thelancet.com Vol 50 Month August, 2022 11 Articles 22 Amoroso A, Magistroni P, Vespasiano F, et al. HLA and AB0 poly- and progenitor cell differentiation and hematopoietic lineage fidel- morphisms may influence SARS-CoV-2 infection and COVID-19 ity. PLoS Biol. 2021;19:e3001394. severity. Transplantation. 2021;105:193–200. 37 Huang LH, Kuo HC, Pan CT, Lin YS, Huang YH, Li SC. Multio- 23 Anzurez A, Naka I, Miki S, et al. Association of HLA-DRB1*09:01 mics analyses identified epigenetic modulation of the S100A gene with severe COVID-19. HLA. 2021;98:37–42. family in Kawasaki disease and their significant involvement in 24 Castelli EC, de Castro MV, Naslavsky MS, et al. MHC Variants neutrophil transendothelial migration. Clin Epigenet. 2018;10:135. associated with symptomatic versus asymptomatic SARS-CoV-2 38 Reinartz A, Ehling J, Franz S, et al. Small intestinal mucosa expres- infection in highly exposed individuals. Front Immunol. 2021;12: sion of putative chaperone fls485. BMC Gastroenterol. 2010;10:27. 742881. 39 Ovsyannikova IG, Dhiman N, Haralambieva IH, et al. Rubella vac- 25 Li S, Wu B, Ling Y, et al. Epigenetic landscapes of single-cell chro- cine-induced cellular immunity: evidence of associations with poly- matin accessibility and transcriptomic immune profiles of T Cells morphisms in the Toll-like, vitamin A and D receptors, and innate in COVID-19 patients. Front Immunol. 2021;12:625881. immune response genes. Hum Genet. 2010;127:207–221. 26 Mousa M, Vurivi H, Kannout H, et al. Genome-wide association 40 Ramaswamy A, Brodsky NN, Sumida TS, et al. Immune dysregula- study of hospitalized COVID-19 patients in the United Arab Emi- tion and autoreactivity correlate with disease severity in SARS-CoV- rates. EBioMedicine. 2021;74:103695. 2-associated multisystem inflammatory syndrome in children. 27 Chou J, Platt CD, Habiballah S, et al. Mechanisms underlying Immunity. 2021;54(5):1083–1095. genetic susceptibility to multisystem inflammatory syndrome in 41 Porritt RA, Binek A, Paschold L, et al. The autoimmune signature children (MIS-C). J Allergy Clin Immunol. 2021;148:732–738. e1. of hyperinflammatory multisystem inflammatory syndrome in 28 Junqueira C, Crespo A, ^ Ranjbar S, et al. FcgR-mediated SARS-CoV- children. J Clin Invest. 2021;131(20):e151520. 2 infection of monocytes activates inflammation. Nature. 2022;606 42 Beckmann ND, Comella PH, Cheng E, et al. Downregulation of (7914):576–584. exhausted cytotoxic T cells in gene expression networks of multi- 29 Konigsberg IR, Barnes B, Campbell M, et al. Host methylation pre- system inflammatory syndrome in children. Nat Commun. 2021;12 dicts SARS-CoV-2 infection and clinical outcome. Commun Med (1):4854. (Lond). 2021;1(1):42. 43 Pfeifer J, Thurner B, Kessel C, et al. Autoantibodies against inter- 30 Oriol-Tordera B, Berdasco M, Llano A, et al. Methylation regulation leukin-1 receptor antagonist in multisystem inflammatory syn- of antiviral host factors, interferon stimulated genes (ISGs) and T- drome in children: a multicentre, retrospective, cohort study. cell responses associated with natural HIV control. PLoS Pathog. Lancet Rheumatol. 2022;4(5):e329–e337. 2020;16(8):e1008678. 44 Gruber CN, Patel RS, Trachtman R, et al. Mapping systemic 31 Verdoni L, Mazza A, Gervasoni A, et al. An outbreak of severe inflammation and antibody responses in multisystem inflamma- Kawasaki-like disease at the Italian epicentre of the SARS-CoV-2 tory syndrome in children (MIS-C). Cell. 2020;183(4):982–995. epidemic: an observational cohort study. Lancet. 2020;395:1771– e14. 1778. 45 Consiglio CR, Cotugno N, Sardh F, et al. The immunology of mul- 32 Ouldali N, Pouletty M, Mariani P, et al. Emergence of Kawasaki dis- tisystem inflammatory syndrome in children with COVID-19. Cell. ease related to SARS-CoV-2 infection in an epicentre of the French 2020;183(4):968–981.e7. COVID-19 epidemic: a time-series analysis. Lancet Child Adolesc 46 Vella LA, Giles JR, Baxter AE, et al. Deep immune profiling of MIS- Health. 2020;4:662–668. C demonstrates marked but transient immune activation com- 33 Sancho-Shimizu V, Brodin P, Cobat A, et al. SARS-CoV-2-related pared to adult and pediatric COVID-19. Sci Immunol. 2021;6(57): MIS-C: A key to the viral and genetic causes of Kawasaki disease? eabf7570. J Exp Med. 2021;218:e20210446. 47 Esteve-Sole A, Anton J, Pino-Ramirez RM, et al. Similarities and 34 Rodriguez-Smith JJ, Verweyen EL, Clay GM, et al. Inflammatory differences between the immunopathogenesis of COVID-19- biomarkers in COVID-19 associated multisystem inflammatory related pediatric multisystem inflammatory syndrome and Kawa- syndrome in children, Kawasaki disease, and macrophage activa- saki disease. J Clin Invest. 2021;131(6):e144554. tion syndrome: a cohort study. Lancet Rheumatol. 2021;3:e574– 48 Middelburg JG, Crijnen TEM, D'Antiga L, et al. Association of eth- e584. nicity with multisystem inflammatory syndrome in children 35 Chen KD, Huang YH, Ming-Huey Guo M, et al. The human related to SARS-CoV-2 infection: an international case-referent blood DNA methylome identifies crucial role of b-catenin in study. Front Pediatr. 2021;9:707650. the pathogenesis of Kawasaki disease. Oncotarget. 2018;9: 49 Geva A, Patel MM, Newhams MM, et al. Data-driven clustering 28337–28350. identifies features distinguishing multisystem inflammatory syn- 36 Wang J, Farkas C, Benyoucef A, et al. Interplay between the EMT drome from acute COVID-19 in children and adolescents. EClinical transcription factors ZEB1 and ZEB2 regulates hematopoietic stem Medicine. 2021;40:101112. 12 www.thelancet.com Vol 50 Month August, 2022