Introduction

To date, more than 50 loci associated to COVID-19 susceptibility, hospitalization, and severity have been identified using genome-wide association studies (GWAS)^1,2. The COVID-19 Host Genetics Initiative (HGI) has made significant efforts³ to augment the power to identify disease loci by recruiting individuals from diverse populations and conducting a trans-ancestry meta-analysis. Despite this, the lack of genetic diversity and a focus on cases of European ancestries still predominate in the studies^4,5. Besides, while trans-ancestry meta-analyses are a powerful approach for discovering shared genetic risk variants with similar effects across populations⁶, they may fail to identify risk variants that have larger effects on particular underrepresented populations. Genetic disease risk has been shaped by the particular evolutionary history of populations and the environmental exposures⁷. Their action is particularly important for infectious diseases due to the selective constrains that are imposed by the host-pathogen interactions^8,9. Literature examples of this in COVID-19 severity includes a DOCK2 gene variant in East Asians¹⁰, and frequent loss of function variants in IFNAR1 and IFNAR2 genes in Polynesian and Inuit populations, respectively^11,12.

Including diverse populations in case-control GWAS studies with unrelated participants usually require a prior classification of individuals in genetically homogeneous groups, which are typically analysed separately to control the population stratification effects¹³. Populations with recent admixture impose an additional challenge to the GWAS due to their complex genetic diversity and linkage disequilibrium (LD) patterns, requiring the development of alternative approaches and a careful inspection of results to reduce the false positives due to population structure⁷. In fact, there are benefits in study power from modelling the admixed ancestries either locally, at regional scale in the chromosomes, or globally, across the genome, depending on factors such as the heterogeneity of the risk variant in frequencies or the effects among the ancestry strata¹⁴. Despite the development of novel methods specifically tailored for the analysis of admixed populations¹⁵, the lack of a standardized analysis framework and the difficulties to confidently cluster the admixed individuals into particular genetic groups often leads to their exclusion from GWAS.

The Spanish Coalition to Unlock Research on Host Genetics on COVID-19 (SCOURGE) recruited COVID-19 patients between March and December 2020 from hospitals across Spain and from March 2020 to July 2021 in Latin-America (https://www.scourge-covid.org). A first GWAS of COVID-19 severity among Spanish patients of European descent revealed novel disease loci and explored age and sex varying effects of the genetic factors¹⁶. Here we present the findings of a GWAS meta-analysis in admixed American (AMR) populations, comprising individuals from the SCOURGE Latin-American cohort and the HGI studies, which allowed to identify two novel severe COVID-19 loci, BAZ2B and DDIAS. Further analyses modelling the admixture from three genetic ancestral components and performing a trans-ethnic meta-analysis led to the identification of an additional risk locus near CREBBP. We finally assessed a cross-ancestry polygenic risk score model with variants associated with critical COVID-19.

Results

Meta-analysis of COVID-19 hospitalization in admixed Americans

Study cohorts

Within the SCOURGE consortium, we included 1,608 hospitalized cases and 1,887 controls (not hospitalized COVID-19 patients) from Latin-American countries and from recruitments of individuals of Latin-American descend conducted in Spain (Supplementary Table 1). Quality control details and estimation of global genetic inferred ancestry (GIA) (supplementary Figure 1) are described in Methods, whereas clinical and demographic characteristics of patients included in the analysis are shown in Table 1. Summary statistics from the SCOURGE cohort were obtained under a logistic mixed model with the SAIGE model (Methods). Another seven studies participating in the COVID-19 HGI consortium were included in the meta-analysis of COVID-19 hospitalization in admixed Americans (Figure 1).

Figure 1.

Table 1.

GWAS meta-analysis

We performed a fixed-effects GWAS meta-analysis using the inverse of the variance as weights for the overlapping markers. The combined GWAS sample size consisted of 4,702 admixed AMR hospitalized cases and 68,573 controls.

This GWAS meta-analysis revealed genome-wide significant associations at four risk loci (Table 2, Figure 2), two of which (BAZ2B and DDIAS) were novel discoveries. Variants of these loci were prioritized by positional and expression quantitative trait loci (eQTL) mapping with FUMA, identifying four lead variants linked to other 310 variants and 31 genes (Supplementary Tables 2-4). A gene-based association test revealed a significant association in BAZ2B and in previously known COVID-19 risk loci: LZTFL1, XCR1, FYCO1, CCR9, and IFNAR2 (Supplementary Table 5).

Figure 2.

Table 2.

Table 3.

Located within the gene BAZ2B, the sentinel variant rs13003835 is an intronic variant associated with an increased risk of COVID-19 hospitalization (Odds Ratio [OR]=1.20, 95% Confidence Interval [CI]=1.12-1.27, p=3.66×10^-8). This association was not previously reported in any GWAS of COVID-19 published to date. Interestingly, rs13003835 did not reach significance (p=0.972) in the COVID-19 HGI trans-ancestry meta-analysis including the five population groups¹. Based on our mapping strategy (see Methods), we also prioritized PLA2R1, LY75, WDSUB1, and CD302 in this locus.

The other novel risk locus is led by the sentinel variant rs77599934, a rare intronic variant located in chromosome 11 within DDIAS and associated with risk of COVID-19 hospitalization (OR=2.27, 95%CI=1.70-3.04, p=2.26×10^-8). The PRCP gene was an additional prioritized gene at this locus.

We also observed a suggestive association with rs2601183 in chromosome 15, which is located between ZNF774 and IQGAP1 (allele-G OR=1.20, 95%CI=1.12-1.29, p=6.11×10^-8, see Supplementary Table 2), which has not yet been reported in any other GWAS of COVID-19 to date. This sentinel variant is in perfect LD (r²=1) with rs601183, an eQTL of ZNF774 in the lung.

The GWAS meta-analysis also pinpointed two significant variants at known loci, LZTFL1 and FOXP4. The SNP rs35731912 was previously associated with COVID-19 severity in EUR populations¹⁷, and it was mapped to LZTFL1. As for rs2477820, while it is a novel risk variant within gene FOXP4, it has a moderate LD (r²=0.295) with rs2496644, which has been linked to COVID-19 hospitalization¹⁸. This is consistent with the effects of LD in tag-SNPs when conducting GWAS in diverse populations.

Functional mapping of novel risk variants

Bayesian fine mapping

We performed different approaches to narrow down the prioritized loci to a set of most probable genes driving the associations. First, we computed credible sets at the 95% confidence for causal variants and annotated them with VEP and the V2G aggregate scoring (Supplementary Table 6, Supplementary Figure 3). The 95% confidence credible set from the region of chromosome 2 around rs13003835 included 76 variants. However, the approach was unable to converge allocating variants in a 95% confidence credible set for the region in chromosome 11.

Colocalization of eQTLs

To determine if the novel genetic risk loci were associated with gene expression in relevant tissues (whole blood, lung, lymphocytes, and oesophagus mucosa), we computed the posterior probabilities (PP) of colocalization for overlapping variants allocated to the 95% confidence credible set. We used the GTEx v8 tissues as the main expression dataset, although it is important to consider that the eQTL associations were carried out mainly on individuals of EUR ancestries. To confirm the colocalization in other ancestries, we also performed analyses on three expression datasets computed on admixed AMR, leveraging data from individuals with high African GIA, high Native-American ancestry, and from a pooled cohort (Methods). Results are shown in the supplementary Table 7.

Five genes (LY75, BAZ2B, CD302, WDSUB1, and PLA2R1) were the candidates for eQTL colocalization in the associated region in chromosome 2. However, LY75 emerged as the most likely causal gene for this locus since the colocalization in whole blood was supported with a PP for H4 (PPH4) of 0.941 and with robust results (supplementary Figure 4). Moreover, this also allowed to prioritize rs12692550 as the most probable causal variant for both traits at this locus with a PP_SNP_H4 of 0.74. Colocalization with gene expression data from admixed AMR validated this finding. LY75 also had evidence of colocalization in lungs (PPH4=0.887) and the esophagus mucosa (PPH4=0.758). However, we could not prioritize a single causal variant in these two other tissues and sensitivity analyses revealed a weak support.

CD302 and BAZ2B were the second and third most likely genes that could drive the association, respectively, according to the colocalization evidence. CD302 was the most probable according to the high AFR genetic ancestries dataset (supplementary Figure 5).

Despite the chromosome 11 region failing to colocalize with gene expression associations for any of the tissues, the lead variant rs77599934 is in moderate-to-strong LD (r²=0.776) with rs60606421, which is an eQTL associated to a reduced expression of DDIAS in the lungs (supplementary Figure 6). The highest PPH4 for DDIAS was in the high AFR genetic ancestry expression dataset (0.71).

Transcriptome-wide association study (TWAS)

Five novel genes, namely SLC25A37, SMARCC1, CAMP, TYW3, and S100A12 (supplementary Table 8) were found significantly associated in the cross-tissue TWAS. To our knowledge, these genes have not been reported previously in any COVID-19 TWAS or GWAS analyses published to date. In the single tissue analyses, ATP5O and CXCR6 were significantly associated in lungs, CCR9 was significantly associated in whole blood, and IFNAR2 and SLC25A37 were associated in lymphocytes.

Likewise, we carried out the TWAS analyses using the models trained in the admixed populations. However, no significant gene-pairs were detected in this case. The 50 genes with the lowest p-values are shown in the supplementary Table 9.

Genetic architecture of COVID-19 hospitalization in AMR populations

Allele frequencies of rs13003835 and rs77599934 across ancestries

Neither rs13003835 (BAZ2B) or rs77599934 (DDIAS) were significantly associated in the COVID-19 HGI B2 cross-population or population-specific meta-analyses. Thus, we investigated their allele frequencies (AF) across populations and compared their effect sizes.

According to gnomAD v3.1.2, the T allele at rs13003835 (BAZ2B) has an AF of 43% in admixed AMR groups while AF is lower in the EUR populations (16%) and in the global sample (29%). Local ancestry inference (LAI) reported by gnomAD shows that within the Native-American component, the risk allele T is the major allele, whereas it is the minor allele within the African and European LAI components. These large differences in AF might be the reason underlying the association found in AMR populations. However, when comparing effect sizes between populations, we found that they were in opposite direction between SAS-AMR and EUR-AFR-EAS and that there was a large heterogeneity among them (Figure 3).

Figure 3.

rs77599934 (DDIAS) had an AF of 1.1% for the G allele in the non-hospitalized controls (Table 2), in line with the recorded gnomAD AF of 1% in admixed AMR groups. This variant has potential to be population-specific variant, given the allele frequencies in other population groups such as EUR (0% in Finnish, 0.025% in non-Finnish), EAS (0%) and SAS (0.042%) and its greater effect size over AFR populations (Figure 3). Examining the LAI, the G allele occurs at 1.1% frequency in the African component while it is almost absent in the Native-American and European. Due to its low MAF, rs77599934 was not analyzed in the COVID-19 HGI B2 cross-population meta-analysis and was only present in the HGI B2 AFR population-specific meta-analysis, precluding the comparison (Figure 3). For this reason, we retrieved the variant with the lowest p-value within a 50 kb region around rs77599934 in the COVID-19 HGI cross-population analysis to investigate if it was in moderate-to-strong LD with our sentinel variant. The variant with the smallest p-value was rs75684040 (OR=1.07, 95%CI=1.03-1.12, p=1.84×10^-3). Yet, LD calculations using the 1KGP phase 3 dataset indicated that rs77599934 and rs75684040 were poorly correlated (r²=0.11).

Cross-population meta-analyses

We carried out two cross-ancestry inverse variance-weighted fixed-effects meta-analyses with the admixed AMR GWAS meta-analysis results to evaluate whether the discovered risk loci replicated when considering other population groups. In doing so, we also identified novel cross-population COVID-19 hospitalization risk loci.

First, we combined the SCOURGE Latin American GWAS results with the HGI B2 ALL analysis (supplementary Table 10). We refer to this analysis as the SC-HGI_ALL meta-analysis. Out of the 40 genome-wide significant loci associated with COVID-19 hospitalization in the last HGI release¹, this study replicated 39 and the association was stronger than in the original study in 29 of those (supplementary Table 11). However, the variant rs13003835 located in BAZ2B did not replicate (OR=1.00, 95%CI=0.98-1.03, p=0.644).

In this cross-ancestry meta-analysis, we replicated two associations that were not found in HGIv7 albeit they were sentinel variants in the latest GenOMICC meta-analysis². We found an association at the CASC20 locus led by the variant rs2876034 (OR=0.95, 95%CI=0.93-0.97, p=2.83×10^-8). This variant is in strong LD with the sentinel variant of that study (rs2326788, r²=0.92), which was associated with critical COVID-19². Besides, this meta-analysis identified the variant rs66833742 near ZBTB7A associated with COVID-19 hospitalization (OR=0.94, 95%CI=0.92-0.96, p=2.50×10^-8). Notably, rs66833742 or its perfect proxy rs67602344 (r²=1) are also associated with upregulation of ZBTB7A in whole blood and in esophagus mucosa. This variant was previously associated with COVID-19 hospitalization².

In a second analysis, we also explored the associations across the defined admixed AMR, EUR, and AFR ancestral sources by combining through meta-analysis the SCOURGE Latin American GWAS results with the HGI studies in EUR, AFR, and admixed AMR, and excluding those from EAS and SAS (Supplementary Table 12). We refer to this as the SC-HGI_3POP meta-analysis. The association at rs13003835 (BAZ2B, OR=1.01, 95%CI=0.98-1.03, p=0.605) was not replicated and rs77599934 near DDIAS could not be assessed, although the association at the ZBTB7A locus was confirmed (rs66833742, OR=0.94, 95%CI=0.92-0.96, p=1.89×10^-8). The variant rs76564172 located near CREBBP also reached statistical significance (OR=1.31, 95% CI=1.25-1.38, p=9.64×10^-9). The sentinel variant of the region linked to CREBBP (in the trans-ancestry meta-analysis) was also subjected a Bayesian fine mapping (supplementary Table 6) and colocalization with eQTLs under the GTEx v8 MASHR models in lungs, esophagus mucosa, whole blood, and transformed lymphocytes. Eight variants were included in the credible set for the region in chromosome 16 (meta-analysis SC-HGI_3POP), although CREBBP did not colocalize in any of the tissues.

Polygenic risk score models

Using the 49 variants associated with disease severity that are shared across populations according to the HGIv7, we constructed a polygenic risk score (PGS) model to assess its generalizability in the admixed AMR (Supplementary Table 13). First, we calculated the PGS for the SCOURGE Latin Americans and explored the association with COVID-19 hospitalization under a logistic regression model. The PGS model was associated with a 1.48-fold increase in COVID-19 hospitalization risk per every PGS standard deviation. It also contributed to explain a slightly larger variance (R2=1.07%) than the baseline model.

Subsequently, we divided the individuals into PGS deciles and percentiles to assess their risk stratification. The median percentile among controls was 40, while in cases it was 63. Those in the top PGS decile exhibited a 5.90-fold (95% CI=3.29-10.60, p=2.79×10^-9) greater risk compared to individuals in the lowest decile, whereas the effects for the rest of the comparisons were much milder.

We also examined the distribution of PGS scores across a 5-level severity scale to further determine if there was any correspondence between clinical severity and genetic risk. Median PGS scores were lower in the asymptomatic and mild groups, whereas higher median scores were observed in the moderate, severe, and critical patients (Figure 4). We fitted a multinomial model using the asymptomatic class as reference and calculated the OR for each category (Supplementary Table 13), observing that the disease genetic risk was similar among asymptomatic, mild, and moderate patients. Given that the PGS was built with variants associated with critical disease and/or hospitalization and that the categories severe and critical correspond to hospitalized patients, these results underscore the ability of cross-ancestry PGS for risk stratification even in an admixed population.

Figure 4.

Finally, we incorporated the novel lead SNPs from our AMR meta-analysis (rs13003835, rs2477820, and rs77599934) into the PGS model. Their inclusion in the model contributed to explain a larger variance (R2=1.74%) than the model without them. This result, however, should be taken with caution given the risk of overfitting due to the use of the same subjects both for the derivation and testing of the variants.

Discussion

We have conducted the largest GWAS meta-analysis of COVID-19 hospitalization in admixed AMR to date. While the genetic risk basis discovered for COVID-19 is largely shared among populations, trans-ancestry meta-analyses on this disease have primarily included EUR samples. This dominance of GWAS in Europeans, and the subsequent bias in sample sizes, can mask population-specific genetic risks (i.e., variants that are monomorphic in some populations) or be less powered to detect risk variants having higher allele frequencies in population groups other than Europeans. In this sense, after combining data from admixed AMR patients, we found two risk loci which are first discovered in a GWAS of Latin-American populations. Interestingly, the sentinel variant rs77599934 in the DDIAS gene is a rare coding variant (∼1% for allele G) with a large effect on COVID-19 hospitalization that is nearly monomorphic in most of the other populations. This has likely led to its exclusion from the cross-populations meta-analyses conducted to date, remaining undetectable.

Fine mapping of the region harbouring DDIAS did not reveal further information about which gene could be the more prone to be causal, or about the functional consequences of the risk variant. However, DDIAS, known as damage-induced apoptosis suppressor gene, is itself a plausible candidate gene. It has been linked to DNA damage repair mechanisms: research showed that depletion of DDIAS led to an increase of ATM phosphorylation and the formation of p53-binding protein (53BP1) foci, a known biomarker of DNA double-strand breaks, suggesting a potential role in double-strand break repair¹⁹. Similarly, elevated levels of phosphorylated nuclear histone 2AXγ were detected after knocking down DDIAS, further emphasizing its role in DNA damage²⁰. Interestingly, a study found that the infection by SARS-CoV-2 also triggered the phosphorylation of the ATM kinase and inhibited repair mechanisms, causing the accumulation of DNA damage²¹. This same study reported the activation of the pro-inflammatory pathway p38/MAPK by the virus, which was as well prompted after knocking-down DDIAS²⁰.

Regarding lung function, the role of DDIAS in lung cancer has been widely studied. It has been proposed as a potential biomarker for lung cancer after finding that it interacts with STAT3 in lung cancer cells, regulating IL-6^22,23 and thus mediating inflammatory processes. Furthermore, another study determined that its blockade inhibited lung cancer cell growth²⁰. The sentinel variant was in strong LD with an eQTL that reduced gene expression of DDIAS in lung, and our findings suggest that DDIAS gene may be indeed involved in viral response. Hence, one reasonable hypothesis is that reduced expression of DDIAS could potentially facilitate SARS-CoV-2 infection through the downregulation of pathways involved in DNA repairment and inflammation. Another prioritized gene from this region was PRCP, an angiotensinase that has been linked to hypertension and for which a hypothesis on its role on COVID-19 progression has been raised^24,25.

The risk region found in chromosome 2 prioritized more than one gene. The lead variant rs13003835 is located within BAZ2B. BAZ2B encodes one of the regulatory subunits of the Imitation switch (ISWI) chromatin remodelers²⁶ constituting the BRF-1/BRF-5 complexes with SMARCA1 and SMARCA5, respectively, and the association signal colocalized with eQTLs in whole blood. The gene LY75 (encoding the lymphocyte antigen 75) also colocalized with eQTLs in whole blood, esophagus mucosa, and lung tissues. Lymphocyte antigen 75 is involved in immune processes through antigen presentation in dendritic cells and endocytosis²⁷, and has been associated with inflammatory diseases, representing also a compelling candidate for the region. Increased expression of LY75 has been detected within hours after the infection by SARS-CoV-2^28,29. Lastly, the signal of CD302 colocalized in individuals with high AFR ancestral admixture in whole blood. This gene is located in the vicinity of LY75 and both conform the readthrough LY75-CD302. It is worth noting that differences in AF for this variant suggest that analyses in AMR populations might be more powered to detect the association, supporting the necessity of population-specific studies.

A third novel risk region was observed in chromosome 15, between the genes IQGAP1 and ZNF774, although not reaching genome-wide significance.

Secondary analyses revealed five TWAS-associated genes, some of which have been already linked to severe COVID-19. In a comprehensive multi-tissue gene expression profiling study³⁰, decreased expression of CAMP and S100A8/S100A9 genes in COVID-19 severe patients was observed, while another study detected the upregulation of SCL25A37 among severe COVID-19 patients³¹. SMARCC1 is a subunit of the SWI/SNF chromatin remodelling complex that has been identified as pro-viral for SARS-CoV-2 and other coronavirus strains through a genome-wide screen³². This complex is crucial for ACE2 expression and the viral entry in the cell³³.

To explore the genetic architecture of the trait among admixed AMR populations, we performed two cross-ancestry meta-analyses including the SCOURGE Latin-American cohort GWAS findings. We found that the two novel risk variants did not associate with COVID-19 hospitalization outside the population-specific meta-analysis, highlighting the importance of complementing trans-ancestry meta-analyses with group-specific analyses. Notably, this analysis did not replicate the association at the DSTYK locus, which was associated with severe COVID-19 in Brazilian individuals with higher European admixture³⁴. This lack of replication supports the initial hypothesis of that study suggesting that the risk haplotype derived from European populations, as we have reduced the weight of this ancestral contribution in our study by excluding those individuals.

Moreover, these cross-ancestry meta-analyses pointed to three loci that were not genome-wide significant in the HGIv7 ALL meta-analysis: a novel locus at CREBBP, and two loci at ZBTB7A and CASC20 that were reported in another meta-analysis. CREBBP and ZBTB7A achieved a stronger significance when considering only EUR, AFR, and admixed AMR GIA groups. According to a recent study, elevated levels of the ZBTB7A gene promote a quasi-homeostatic state between coronaviruses and host cells, preventing cell death by regulating oxidative stress pathways³⁵. This gene is involved in several signalling pathways, such as B and T cell differentiation³⁶. On a separate note, CREBBP encodes the CREB binding protein (CBP), involved in transcription activation, that is known to positively regulate the type I interferon response through virus-induced phosphorylation of IRF-3³⁷. Besides, the CREBP/CBP interaction has been implicated in SARS-CoV-2 infection³⁸ via the cAMP/PKA pathway. In fact, cells with suppressed CREBBP gene expression exhibit reduced replication of the so called Delta and Omicron SARS-CoV-2 variants³⁸.

The cross-population PGS model effectively stratified individuals based on their genetic risk and demonstrated consistency with the clinical severity classification of the patients. The inclusion of the new variants in the PGS model slightly improved the predictive value of the PGS. However, it is important to confirm this last finding in an external admixed AMR cohort to address potential overfitting arising from using the same individuals both for the discovery of the associations and for testing the model.

This study is subject to limitations, mostly concerning the sample recruitment and composition. The SCOURGE Latino-American sample size is small and the GWAS is underpowered. Another limitation is the difference in case-control recruitment across sampling regions that, yet controlled for, may reduce the ability to observe significant associations driven by different compositions of the populations. In this sense, the identified risk loci might not replicate in a cohort lacking any of the parental population sources from the three-way admixture. Likewise, we could not explicitly control for socio-environmental factors that could have affected COVID-19 spread and hospitalization rates, although genetic principal components are known to capture non-genetic factors. Finally, we must acknowledge the lack of a replication cohort. We have used all the available GWAS data for COVID-19 hospitalization in admixed AMR in this meta-analysis due to the low number of studies conducted. Therefore, we had no studies to replicate or validate the results. These concerns may be addressed in the future by including more AMR GWAS studies in the meta-analysis, both by involving diverse populations in study designs and by supporting research from countries in Latin-America.

This study provides novel insights into the genetic basis of COVID-19 severity, emphasizing the importance of considering host genetic factors through using non-European populations, especially of admixed sources. Such complementary efforts can pin down new variants and increase our knowledge on the host genetic factors of severe COVID-19.

Materials and methods

GWAS in Latin Americans from SCOURGE

The SCOURGE Latin American cohort

A total of 3,729 of COVID-19 positive cases were recruited across five countries from Latin America (Mexico, Brazil, Colombia, Paraguay, and Ecuador) by 13 participating centres (supplementary Table 1) from March 2020 to July 2021. In addition, we included 1,082 COVID-19 positive individuals recruited between March and December 2020 in Spain who either had evidence of origin from a Latin American country or showed inferred genetic admixture between AMR, EUR, and AFR (with < 0.05% SAS/EAS). These individuals were excluded from a previous SCOURGE study that focused on participants with European genetic ancestries¹⁶. We used hospitalization as a proxy for disease severity and defined as cases those COVID-19 positive patients that underwent hospitalization as a consequence of the infection and used as controls those that did not need hospitalization due to COVID-19.

Samples and data were collected with informed consent after the approval of the Ethics and Scientific Committees from the participating centres and by the Galician Ethics Committee Ref 2020/197. Recruitment of patients from IMSS (in Mexico, City), was approved by of the National Comitte of Clinical Research, from Instituto Mexicano del Seguro Social, Mexico (protocol R-2020-785-082).

Samples and data were processed following normalized procedures. The REDCap electronic data capture tool^39,40, hosted at Centro de Investigación Biomédica en Red (CIBER) from the Instituto de Salud Carlos III (ISCIII), was used to collect and manage demographic, epidemiological, and clinical variables. Subjects were diagnosed for COVID-19 based on quantitative PCR tests (79.3%), or according to clinical (2.2%) or laboratory procedures (antibody tests: 16.3%; other microbiological tests: 2.2%).

SNP array genotyping

Genomic DNA was obtained from peripheral blood and isolated using the Chemagic DNA Blood 100 kit (PerkinElmer Chemagen Technologies GmbH), following the manufacturer’s recommendations.

Samples were genotyped with the Axiom Spain Biobank Array (Thermo Fisher Scientific) following the manufacturer’s instructions in the Santiago de Compostela Node of the National Genotyping Center (CeGen-ISCIII; http://www.usc.es/cegen). This array contains probes for genotyping a total of 757,836 SNPs. Clustering and genotype calling were performed using the Axiom Analysis Suite v4.0.3.3 software.

Quality control steps and variant imputation

A quality control (QC) procedure using PLINK 1.9⁴¹ was applied to both samples and the genotyped SNPs. We excluded variants with a minor allele frequency (MAF) <1%, a call rate <98%, and markers strongly deviating from Hardy-Weinberg equilibrium expectations (p<1×10^-6) with mid-p adjustment. We also explored the excess of heterozygosity to discard potential cross-sample contaminations. Samples missing >2% of the variants were filtered out. Subsequently, we kept the autosomal SNPs and removed high LD regions and conducted LD-pruning (windows of 1,000 SNPs, with step size of 80 and r² threshold of 0.1) to assess kinship and estimate the global ancestral proportions. Kinship was evaluated based on IBD values, removing one individual from each pair with PI_HAT>0.25 that showed a Z0, Z1, and Z2 coherent pattern (according to the theoretical expected values for each relatedness level). Genetic principal components (PCs) were calculated with PLINK with the subset of LD pruned variants.

Genotypes were imputed with the TOPMed version r2 reference panel (GRCh38) using the TOPMed Imputation Server and variants with Rsq<0.3 or with MAF<1% were filtered out. A total of 4,348 individuals and 10,671,028 genetic variants were included in the analyses.

Genetic admixture estimation

Global genetic inferred ancestry (GIA), referred to the genetic similarity to the used reference individuals, was estimated with the ADMIXTURE⁴² v1.3 software following a two-step procedure. First, we randomly sampled 79 European (EUR) and 79 African (AFR) samples from The 1000 Genomes Project (1KGP)⁴³ and merged them with the 79 Native American (AMR) samples from Mao et al.⁴⁴ keeping the biallelic SNPs. LD-pruned variants were selected from this merge using the same parameters as in the QC. We then run an unsupervised analysis with K=3 to redefine and homogenize the clusters and to compose a refined reference for the analyses, by applying a threshold of ≥95% of belonging to a particular cluster. As a result of this, 20 AFR, 18 EUR, and 38 AMR individuals were removed. The same LD-pruned variants data from the remaining individuals were merged with the SCOURGE Latin American cohort to perform a supervised clustering and estimated admixture proportions. A total of 471 samples from the SCOURGE cohort with >80% estimated European GIA were removed to reduce the weight of the European ancestral component, leaving a total of 3,512 admixed American (AMR) subjects for downstream analyses.

Association analysis

Results for the SCOURGE Latin Americans GWAS were obtained testing for COVID-19 hospitalization as a surrogate of severity. To accommodate the continuum of GIA in the cohort, we opted for a joint testing of all the individuals as a single study using a mixed regression model, as this approach has demonstrated a greater power and to sufficiently control population structure⁴⁵. The SCOURGE cohort consisted of 3,512 COVID-19 positive patients: cases (n=1,625) were defined as hospitalized COVID-19 patients and controls (n=1,887) as non-hospitalized COVID-19 positive patients.

Logistic mixed regression models were fitted using the SAIGEgds⁴⁶ package in R, which implements the two-step mixed SAIGE⁴⁷ model methodology and the SPA test. Baseline covariables included sex, age, and the first 10 PCs. To account for a potential heterogeneity in the recruitment and hospitalization criteria across the participating countries, we adjusted the models by groups of the recruitment areas classified in six categories: Brazil, Colombia, Ecuador, Mexico, Paraguay, and Spain. This dataset has not been used in any previously GWAS of COVID-19 published to date.

Meta-analysis of Latin-American populations

The results of the SCOURGE Latin American cohort were meta-analyzed with the AMR HGI-B2 data, conforming our primary analysis. Summary results from the HGI freeze 7 B2 analysis corresponding to the admixed AMR population were obtained from the public repository (April 8, 2022: https://www.covid19hg.org/results/r7/), summing up 3,077 cases and 66,686 controls from seven contributing studies. We selected the B2 phenotype definition because it offered more power and the presence of population controls not ascertained for COVID-19 does not have a drastic impact in the association results.

The meta-analysis was performed using an inverse-variance weighting method in METAL⁴⁸. Average allele frequency was calculated and variants with low imputation quality (Rsq<0.3) were filtered out, leaving 10,121,172 variants for meta-analysis. Heterogeneity between studies was evaluated with the Cochran’s-Q test. The inflation of results was assessed based on a genomic control (lambda).

Definition of the genetic risk loci and putative functional impact

Definition of lead variant and novel loci

To define the lead variants in the loci that were genome-wide significant, an LD-clumping was performed on the meta-analysis data using a threshold p-value<5×10^-8, clump distance=1500 kb, independence set at a threshold r²=0.1 and used the SCOURGE cohort genotype data as LD reference panel. Independent loci were deemed as a novel finding if they met the following criteria: 1) p-value<5×10^-8 in the meta-analysis and p-value>5×10^-8 in the HGI B2 ALL meta-analysis or in the HGI B2 AMR and AFR and EUR analyses when considered by separate; 2) Cochran’s Q-test for heterogeneity of effects is <0.05/N_loci, where N_loci is the number of independent variants with p<5×10^-8; and 3) the nearest gene has not been previously described in the latest HGIv7 update.

Annotation and initial mapping

Functional annotation was done with FUMA⁴⁹ for those variants with a p-value<5×10^-8 or in moderate-to-strong LD (r²>0.6) with the lead variants, where the LD was calculated from the 1KGP AMR panel. Genetic risk loci were defined by collapsing LD-blocks within 250 kb. Then, genes, scaled CADD v1.4 scores, and RegulomeDB v1.1 scores were annotated for the resulting variants with ANNOVAR in FUMA⁴⁹. Gene-based analysis was also performed using MAGMA⁵⁰ as implemented in FUMA, under the SNP-wide mean model using the 1KGP AMR reference panel. Significance was set at a threshold p<2.66×10^-6 (which assumes that variants can be mapped to a total of 18,817 genes).

FUMA allowed us to perform an initial gene mapping by two approaches: (1) positional mapping, which assigns variants to genes by physical distance using 10-kb windows; and (2) eQTL mapping based on GTEx v.8 data from whole blood, lungs, lymphocytes, and oesophagus mucosa tissues, establishing a False Discovery Rate (FDR) of 0.05 to declare significance for variant-gene pairs.

Subsequently, to assign the variants to the most likely gene driving the association, we refined the candidate genes by fine mapping the discovered regions and implementing functional mapping.

To conduct a Bayesian fine mapping, credible sets for the genetic loci considered novel findings were calculated on the results from each of the three meta-analyses to identify a subset of variants most likely containing the causal variant at 95% confidence level, assuming that there is a single causal variant and that it has been tested. We used corrcoverage (https://cran.rstudio.com/web/packages/corrcoverage/index.html) for R to calculate the posterior probabilities of the variant being causal for all variants with an r²>0.1 with the leading SNP and within 1 Mb except for the novel variant in chromosome 19, for which we used a window of 0.5 Mb. Variants were added to the credible set until the sum of the posterior probabilities was ≥0.95. VEP (https://www.ensembl.org/info/docs/tools/vep/index.html) and the V2G aggregate scoring from Open Targets Genetics (https://genetics.opentargets.org) were used to annotate the biological function of the variants contained in the fine-mapped credible sets.

Colocalization analysis

We also conducted colocalization analyses to identify the putative causal genes that could act through the regulation of gene expression. FUMA’s eQTL mapping enabled the identification of genes whose expression was associated with the variants in whole blood, lungs, lymphocytes, and oesophagus mucosa tissues. We combined this information with the VEP and V2G aggregate scoring to prioritize genes. For the fine-mapping regions, we included the variants within the calculated credible sets. In the cases where the fine mapping was unsuccessful, we considered variants within a 0.2 Mb window of the lead variant.

For each prioritized gene, we then run COLOC⁵¹ to assess the evidence of colocalization between association signals and the eQTLs in each tissue, when at least one variant overlapped between them. COLOC estimates the posterior probability of two traits sharing the same causal variant in a locus. Prior probabilities of a variant being associated to COVID-19 phenotype (p1) and gene expression (p2) were set at 1×10^-4, while pp2 was set at 1×10^-6 as they are robust thresholds⁵². The posterior probability of colocalization (PP4) > 0.75 and a ratio PP4/PP3>3 were used as the criteria to support evidence of colocalization. Additionally, a threshold of PP4.SNP >0.5 was chosen for causal variant prioritization. In cases were colocalization of a single variant failed, we computed the 95% credible sets. The eQTL data was retrieved from GTEx v8 and only significant variant-gene pairs were considered in the analyses.

Colocalization in whole-blood was also performed using the recent published gene expression datasets derived from a cohort of African Americans, Puerto Ricans, and Mexican Americans (GALA II-SAGE)⁵³. We used the results from the pooled cohort for the three discovered loci, and from the AFRHp5 (African genetic ancestry>50%) and IAMHp5 (Native American genetic ancestry>50%) cohorts for the risk loci in chromosomes 2 and 11. Results are shown in the Supplementary Table 10.

Sensitivity plots are shown in supplementary Figures 4 and 5.

Transcription-wide association studies

Transcriptome-wide association studies (TWAS) were conducted using the pretrained prediction models with MASHR-computed effect sizes on GTEx v8 datasets^54,55. Results from the Latin-American meta-analysis were harmonized and integrated with the prediction models through S-PrediXcan⁵⁶ for lungs, whole blood, lymphocytes and oesophagus mucosa tissues. Statistical significance was set at p-value<0.05 divided by the number of genes that were tested for each tissue. Subsequently, we leveraged results for all 49 tissues and run a multi-tissue TWAS to improve power for association, as demonstrated recently⁵⁷. TWAS was also conducted with the MASHR models for whole-blood in the pooled admixed AMR from the GALA and SAGE studies⁵³.

Cross-population meta-analyses

We conducted two additional meta-analyses to investigate the ability of combining populations to replicate our discovered risk loci. This methodology enabled the comparison of effects and the significance of associations in the novel risk loci between the results from analyses that included or excluded other population groups.

The first meta-analysis comprised the five populations analysed within HGI (B2-ALL). Additionally, to evaluate the three GIA components within the SCOURGE Latin-American cohort⁵⁸, we conducted a meta-analysis of the admixed AMR, EUR, and AFR cohorts (B2). All summary statistics were retrieved from the HGI repository. We applied the same meta-analysis methodology and filters as in the admixed AMR meta-analysis. Novel variants from these meta-analyses were fine-mapped and colocalized with gene expression.

Cross-population Polygenic Risk Score

A polygenic risk score (PGS) for critical COVID-19 was derived combining the variants associated with hospitalization or disease severity that have been discovered to date. We curated a list of lead variants that were: 1) associated to either severe disease or hospitalization in the latest HGIv7 release¹ (using the hospitalization weights); or 2) associated to severe disease in the latest GenOMICC meta-analysis² that were not reported in the latest HGI release. A total of 49 markers were used in the PGS model (see supplementary Table 13) since two variants were absent from our study.

Scores were calculated and normalized for the SCOURGE Latin-American cohort with PLINK 1.9. This cross-ancestry PGS was used as a predictor for hospitalization (COVID-19 positive that were hospitalized vs. COVID-19 positive that did not necessitate hospital admission) by fitting a logistic regression model. Prediction accuracy for the PGS was assessed by performing 500 bootstrap resamples of the increase in the pseudo-R-squared. We also divided the sample in deciles and percentiles to assess risk stratification. The models were fit for the dependent variable adjusting for sex, age, the first 10 PCs, and the sampling region (in the Admixed AMR cohort) with and without the PGS, and the partial pseudo-R2 was computed and averaged among the resamples.

A clinical severity scale was used in a multinomial regression model to further evaluate the power of this cross-ancestry PGS for risk stratification. This severity strata were defined as follows: 0) asymptomatic; 1) mild, that is, with symptoms, but without pulmonary infiltrates or need of oxygen therapy; 2) moderate, that is, with pulmonary infiltrates affecting <50% of the lungs or need of supplemental oxygen therapy; 3) severe disease, that is with hospital admission and PaO₂<65 mmHg or SaO₂<90%, PaO₂/FiO₂<300, SaO₂/FiO₂<440, dyspnea, respiratory frequency≥22 bpm, and infiltrates affecting >50% of the lungs; and 4) critical disease, that is with an admission to the ICU or need of mechanical ventilation (invasive or non-invasive). We also included the novel risk variants as predictors alongside the PRS to determine if they provided increased prediction ability.

Data availability

Summary statistics from the SCOURGE Latin-American GWAS will be available at https://github.com/CIBERER/Scourge-COVID19.

Funding

Instituto de Salud Carlos III (COV20_00622 to A.C., COV20/00792 to M.B., COV20_00181 to C.A., COV20_1144 to M.A.J.S. and A.F.R., PI20/00876 to C.F.); European Union (ERDF) ‘A way of making Europe’. Fundación Amancio Ortega, Banco de Santander (to A.C.), Estrella de Levante S.A. and Colabora Mujer Association (to E.G.-N.) and Obra Social La Caixa (to R.B.); Agencia Estatal de Investigación (RTC-2017-6471-1 to C.F.), Cabildo Insular de Tenerife (CGIEU0000219140 ‘Apuestas científicas del ITER para colaborar en la lucha contra la COVID-19’ to C.F.) and Fundación Canaria Instituto de Investigación Sanitaria de Canarias (PIFIISC20/57 to C.F.).

SD-DA was supported by a Xunta de Galicia predoctoral fellowship.

Author contributions

Study design: RC, AC, CF. Data collection: SCOURGE cohort group. Data analysis: SD-DA, RC, ADL, CF, JML-S. Interpretation: SD-DA, RC, ADL. Drafting of the manuscript: SD-DA, RC, ADL, CF, AR-M, AC. Critical revision of the manuscript: SD-DA, RC, ADL, AC, CF, JAR, AR-M, PL. Approval of the final version of the publication: all co-authors.

Acknowledgements

The contribution of the Centro National de Genotipado (CEGEN), and Centro de Supercomputación de Galicia (CESGA) for funding this project by providing supercomputing infrastructures, is also acknowledged. Authors are also particularly grateful for the supply of material and the collaboration of patients, health professionals from participating centers and biobanks. Namely Biobanc-Mur, and biobancs of the Complexo Hospitalario Universitario de A Coruña, Complexo Hospitalario Universitario de Santiago, Hospital Clínico San Carlos, Hospital La Fe, Hospital Universitario Puerta de Hierro Majadahonda—Instituto de Investigación Sanitaria Puerta de Hierro—Segovia de Arana, Hospital Ramón y Cajal, IDIBGI, IdISBa, IIS Biocruces Bizkaia, IIS Galicia Sur. Also biobanks of the Sistema de Salud de Aragón, Sistema Sanitario Público de Andalucía, and Banco Nacional de ADN.

Supplementary Material for: Novel risk loci for COVID-19 hospitalization among admixed American populations

Supplementary Tables are provided in a separate excel file

Supplementary figures

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.

Supplementary Figure 4.

Supplementary Figure 5.

Supplementary Figure 6.

Scourge Cohort Group

Full list of cohort members and affiliations

Javier Abellan^1,2; René Acosta-Isaac³; Jose María Aguado^4,5,6,7; Carlos Aguilar⁸; Sergio Aguilera-Albesa^9,10; Abdolah Ahmadi Sabbagh¹¹; Jorge Alba¹²; Sergiu Albu^13,14,15; Karla A.M. Alcalá-Gallardo¹⁶; Julia Alcoba-Florez¹⁷; Sergio Alcolea Batres¹⁸; Holmes Rafael Algarin-Lara^19,20; Virginia Almadana²¹; Julia Almeida^22,23,24,25; Berta Almoguera^26,27; María R. Alonso²⁸; Nuria Alvarez²⁸; Rodolfo Alvarez-Sala Walther¹⁸; Mónica T. Andrade ^29,30; Álvaro Andreu-Bernabeu^31,6; Maria Rosa Antonijoan³²; Eunate Arana-Arri^33,34; Carlos Aranda^35,36; Celso Arango^31,37,6; Carolina Araque^38,39; Nathalia K. Araujo⁴⁰; Izabel M.T. Araujo⁴¹; Ana C. Arcanjo^42,43,44; Ana Arnaiz^45,46,47; Francisco Arnalich Fernández⁴⁸; María J. Arranz⁴⁹; José Ramon Arribas Lopez⁴⁸; Maria-Jesus Artiga⁵⁰; Yubelly Avello-Malaver⁵¹; Carmen Ayuso^26,27; Ana Margarita Baldión⁵¹; Belén Ballina Martín¹¹; Raúl C. Baptista-Rosas^52,53,54; Andrea Barranco-Díaz²⁰; María Barreda-Sánchez^55,56; Viviana Barrera-Penagos⁵¹; Moncef Belhassen-Garcia^57,58; Enrique Bernal⁵⁵; David Bernal-Bello⁵⁹; Joao F. Bezerra⁶⁰; Marcos A.C. Bezerra⁶¹; Natalia Blanca-López⁶²; Rafael Blancas⁶³; Lucía Boix-Palop⁶⁴; Alberto Borobia⁶⁵; Elsa Bravo⁶⁶; María Brion^67,68; Óscar Brochado-Kith^69,7; Ramón Brugada^70,71,68,72; Matilde Bustos⁷³; Alfonso Cabello⁷⁴; Juan J. Caceres-Agra⁷⁵; Esther Calbo⁷⁶; Enrique J. Calderón^77,78,79; Shirley Camacho⁸⁰; Cristina Carbonell^81,58; Servando Cardona-Huerta⁸²; Antonio Augusto F. Carioca⁸³; Maria Sanchez Carpintero^35,36; Carlos Carpio Segura¹⁸; Thássia M.T. Carratto⁸⁴; José Antonio Carrillo-Avila⁸⁵; Maria C.C. Carvalho⁸⁶; Carlos Casasnovas^87,88,27; Luis Castano^{33,34,27,89,90}; Carlos F. Castaño^35,36; Jose E. Castelao⁹¹; Aranzazu Castellano Candalija⁹²; María A. Castillo⁸⁰; Yolanda Cañadas³⁶; Francisco C. Ceballos²⁷; Jessica G. Chaux³⁹; Walter G. Chaves-Santiago^93,39; Sylena Chiquillo-Gómez^19,20; Marco A. Cid-Lopez¹⁶; Oscar Cienfuegos-Jimenez⁸²; Rosa Conde-Vicente⁹⁴; M. Lourdes Cordero-Lorenzana⁹⁵; Dolores Corella^96,97; Almudena Corrales^98,99; Jose L. Cortes-Sanchez^82,100; Marta Corton^26,27; Tatiana X. Costa¹⁰¹; Raquel Cruz^27,102; Marina S. Cruz⁴⁰; Luisa Cuesta¹⁰³; Gabriela C.R. Cunha¹⁰⁴; David Dalmau^105,76; Raquel C.S. Dantas-Komatsu⁴⁰; M. Teresa Darnaude¹⁰⁶; Alba De Martino-Rodríguez^107,108; Juan De la Cruz Troca^109,110,78; Juan Delgado-Cuesta¹¹¹; Aranzazu Diaz de Bustamante¹⁰⁶; Covadonga M. Diaz-Caneja^31,37,6; Beatriz Dietl⁷⁶; Silvia Diz-de Almeida^27,102; Elena Domínguez-Garrido¹¹²; Alice M. Duarte⁴¹; Anderson Díaz-Pérez²⁰; Jose Echave-Sustaeta¹¹³; Rocío Eiros¹¹⁴; César O. Enciso-Olivera^38,39; Gabriela Escudero¹¹⁵; Pedro Pablo España¹¹⁶; Gladys Mercedes Estigarribia Sanabria¹¹⁷; María Carmen Fariñas^45,46,47; Marianne R. Fernandes^118,119; Lidia Fernandez-Caballero^26,27; María J. Fernandez-Nestosa¹²⁰; Ramón Fernández^45,121; Silvia Fernández Ferrero¹¹; Yolanda Fernández Martínez¹¹; Ana Fernández-Cruz¹²²; Uxía Fernández-Robelo¹²³; Amanda Fernández-Rodríguez^69,7; Marta Fernández-Sampedro^45,47,46; Ruth Fernández-Sánchez^26,27; Tania Fernández-Villa^124,78; Carmen Fernéndez Capitán⁹²; Patricia Flores-Pérez¹²⁵; Vicente Friaza^78,79; Lácides Fuenmayor-Hernández²⁰; Marta Fuertes Núñez¹¹; Victoria Fumadó¹²⁶; Ignacio Gadea¹²⁷; Lidia Gagliardi^35,36; Manuela Gago-Domínguez^128,129; Natalia Gallego¹³⁰; Cristina Galoppo¹³¹; Carlos Garcia-Cerrada^1,2,132; Josefina Garcia-García⁵⁵; Inés García^26,27; Mercedes García^35,36; Leticia García^35,36; María Carmen García Torrejón^133,2; Irene García-García⁶⁵; Carmen García-Ibarbia^45,47,46; Andrés C. García-Montero¹³⁴; Ana García-Soidán¹³⁵; Elisa García-Vázquez⁵⁵; Aitor García-de-Vicuña^33,136; Emiliano Garza-Frias⁸²; Jesus Gaytán-Martínez¹³⁷, Angela Gentile¹³¹; Belén Gil-Fournier¹³⁸; Fernan Gonzalez Bernaldo de Quirós¹³⁹; Manuel Gonzalez-Sagrado⁹⁴; Hugo Gonzalo Benito¹⁴⁰; Beatriz González Álvarez^107,108; Anna González-Neira²⁸; Javier González-Peñas^31,6,37; Oscar Gorgojo-Galindo¹⁴¹; Florencia Guaragna¹³¹; Genilson P. Guegel¹⁴²; Beatriz Guillen-Guio⁹⁸; Encarna Guillen-Navarro^{55,143,144,27}; Pablo Guisado-Vasco¹¹³; Luz D. Gutierrez-Castañeda^145,39; Juan F. Gutiérrez-Bautista¹⁴⁶; Luis Gómez Carrera¹⁸; María Gómez García¹²⁸; Ángela Gómez Sacristán¹⁴⁷; Javier Gómez-Arrue^107,108; Mario Gómez-Duque^93,39; Miguel Górgolas⁷⁴; Sarah Heili-Frades¹⁴⁸; Estefania Hernandez¹⁴⁹; Luis D. Hernandez-Ortega^150,151; Cristina Hernández Moro¹¹; Guillermo Hernández-Pérez⁸¹; Rebeca Hernández-Vaquero¹⁵²; Belen Herraez²⁸; M. Teresa Herranz⁵⁵; María Herrera^35,36; María José Herrero^153,154; Antonio Herrero-Gonzalez¹⁵⁵; Juan P. Horcajada^{156,157,14,158,7}; Natale Imaz-Ayo³³; Maider Intxausti-Urrutibeaskoa¹⁵⁹; Rafael H. Jacomo¹⁶⁰; Rubén Jara⁵⁵; Perez Maria Jazmin¹³¹; María A. Jimenez-Sousa^69,7; Ángel Jiménez^35,36; Pilar Jiménez¹⁴⁶; Ignacio Jiménez-Alfaro¹⁶¹; Iolanda Jordan^162,163,78; Rocío Laguna-Goya^164,165; Daniel Laorden¹⁸; María Lasa-Lazaro^164,165; María Claudia Lattig^80,166; Ailen Lauriente¹³¹; Anabel Liger Borja¹⁶⁷; Lucía Llanos¹⁶⁹; Esther Lopez-Garcia^{109,110,78,170}; Rosario Lopez-Rodriguez^26,27; Leonardo Lorente¹⁷¹; José E. Lozano¹⁷²; María Lozano-Espinosa¹⁶⁷; Andre D. Luchessi¹⁷³; Eduardo López Granados^174,175,27; Amparo López-Bernús^81,58; Miguel A. López-Ruz^176,177,178; Aluísio X. Magalhães-Brasil ¹⁷⁹;Ignacio Mahillo^180,181,99; Esther Mancebo^164,165; Carmen Mar¹¹⁶; Cristina Marcelo Calvo⁹²; Miguel Marcos^81,58; Alba Marcos-Delgado¹²⁴; Pablo Mariscal Aguilar¹⁸; Marta Martin-Fernandez¹⁸²; Laura Martin-Pedraza⁶²; Amalia Martinez¹⁸³; Iciar Martinez-Lopez^184,185; Oscar Martinez-Nieto^51,166; Pedro Martinez-Paz¹⁴⁰; Angel Martinez-Perez¹⁸⁶; Michel F. Martinez-Resendez⁸²; María M. Martín¹⁸⁷; María Dolores Martín¹⁸⁸; Vicente Martín^124,78; Caridad Martín-López¹⁶⁷; José-Ángel Martín-Oterino^81,58; María Martín-Vicente⁶⁹; Ricardo Martínez¹⁴⁹; Juan José Martínez^88,27; Silvia Martínez^45,47; Violeta Martínez Robles¹¹; Eleno Martínez-Aquino¹⁸⁹; Óscar Martínez-González¹⁹⁰; Andrea Martínez-Ramas^26,27; Laura Marzal^26,27; Alicia Marín Candon⁶⁵; Jose Antonio Mata-Marin,¹³⁷ Juliana F. Mazzeu^179,191,192; Jeane F.P. Medeiros⁴⁰; Francisco J. Medrano^77,78,79; Xose M. Meijome^193,194; Natalia Mejuto-Montero¹⁹⁵; Celso T. Mendes-Junior ^84,196,197; Humberto Mendoza Charris^66,20; Eleuterio Merayo Macías¹⁹⁸; Fátima Mercadillo¹⁹⁹; Arieh R. Mercado-Sesma^150,151; Pablo Minguez^26,27; Antonio J J. Molina^124,78; Elena Molina-Roldán²⁰⁰; Juan José Montoya¹⁴⁹; Patricia Moreira-Escriche²⁰¹; Xenia Morelos-Arnedo^66,20; Victor Moreno Cuerda^1,2; Alberto Moreno Fernández⁹²; Antonio Moreno-Docón⁵⁵; Junior Moreno-Escalante²⁰; Rubén Morilla^79,202; Patricia Muñoz García^203,99,6; Ana Méndez-Echevarria²⁰⁴; Pablo Neira¹³¹; Julian Nevado^27,131,205; Israel Nieto-Gañán¹³⁵; Joana F.R. Nunes⁴²; Rocio Nuñez-Torres²⁸; Antònia Obrador-Hevia^206,207; J. Gonzalo Ocejo-Vinyals^45,47; Virginia Olivar¹³¹; Silviene F. Oliveira^{179,208,208,210,211}; Lorena Ondo^26,27; Alberto Orfao^22,23,24,25; Luis Ortega²¹²; Eva Ortega-Paino⁵⁰; Fernando Ortiz-Flores^45,47; Rocio Ortiz-Lopez^213,82; José A. Oteo^12,214; Harry Pachajoa^215,216; Manuel Pacheco¹⁴⁹; Fredy Javier Pacheco-Miranda²⁰; Irene Padilla Conejo¹¹; Sonia Panadero-Fajardo⁸⁵; Mara Parellada^31,37,6; Roberto Pariente-Rodríguez¹³¹; Estela Paz-Artal^164,165,217; Germán Peces-Barba^218,99; Miguel S. Pedromingo Kus²¹⁹; Celia Perales¹²⁷; Patricia Perez²²⁰; Gustavo Perez-de-Nanclares^33,221; Teresa Perucho²²²; Aline Pic-Taylor ^{42,208,209,211}, Lisbeth A. Pichardo¹¹; Mel·lina Pinsach-Abuin^70,68; Luz Adriana Pinzón^93,39; Guillermo Pita³⁰; Francesc Pla-Junca^223,27; Laura Planas-Serra^88,27; Ericka N. Pompa-Mera^224,137; Gloria L. Porras-Hurtado¹⁴⁹; Aurora Pujol^88,27,225; César Pérez²²⁶; Felipe Pérez-García^227,228; Patricia Pérez-Matute²¹⁴; Alexandra Pérez-Serra^70,68; M. Elena Pérez-Tomás⁵⁵; María Eugenia Quevedo Chávez^19,20; Maria Angeles Quijada^30,229; Inés Quintela¹²⁸; Diana Ramirez-Montaño²³⁰; Soraya Ramiro León¹³¹; Pedro Rascado Sedes²³¹; Delia Recalde^107,108; Emma Recio-Fernández²¹⁴; Salvador Resino^69,7; Adriana P. Ribeiro^29,30,232; Carlos S. Rivadeneira-Chamorro³⁹; Diana Roa-Agudelo⁵¹; Montserrat Robelo Pardo²³¹; Marilyn Johanna Rodriguez³⁹; German Ezequiel Rodriguez Novoa¹³¹; Fernando Rodriguez-Artalejo^{109,110,78,170}; Carlos Rodriguez-Gallego^233,234; José A. Rodriguez-Garcia¹¹; María A. Rodriguez-Hernandez⁷³; Antonio Rodriguez-Nicolas¹⁴⁶; Agustí Rodriguez-Palmero^235,88; Paula A. Rodriguez-Urrego⁵¹; Belén Rodríguez Maya¹; Marena Rodríguez-Ferrer²⁰; Emilio Rodríguez-Ruiz^231,129; Federico Rojo^236,25; Andrea Romero-Coronado²⁰; Filomeno Rondón García¹¹; Lidia S. Rosa²³⁶; Antonio Rosales-Castillo²³⁷; Cladelis Rubio^238,239; María Rubio Olivera^35,36; Montserrat Ruiz^88,27; Francisco Ruiz-Cabello^146,177,240; Eva Ruiz-Casares²²²; Juan J. Ruiz-Cubillan^45,47; Javier Ruiz-Hornillos^241,36,242; Pablo Ryan^{243,244,245,7}; Hector D. Salamanca^38,39; Lorena Salazar-García⁸⁰; Giorgina Gabriela Salgueiro Origlia ⁹²; Cristina Sancho-Sainz¹⁵⁹; Jorge Luis Sandoval-Ramírez,¹³⁷ Anna Sangil⁶⁴; Arnoldo Santos²²⁶; Ney P.C. Santos¹¹⁸; Amanda C.M. Saúde ^30,246 Agatha Schlüter^88,27; Sonia Segovia^223,247,248; Alex Serra-Llovich²⁴⁹; Fernando Sevil Puras⁸; Marta Sevilla Porras^27,130; Miguel A. Sicolo^250,251; Vivian N. Silbiger¹⁷³; Nayara S. Silva²⁵²; Fabiola T.C. Silva⁴⁰; Cristina Silván Fuentes²⁷; Jordi Solé-Violán^253,99,254; José Manuel Soria¹⁸⁶; Jose V. Sorlí^96,97; Renata R. Sousa¹⁷⁹; Juan Carlos Souto³; Karla S.C. Souza⁸⁶; Vanessa S. Souza¹⁰⁴; John J. Sprockel^93,39; David A. Suarez-Zamora⁵¹; José Javier Suárez-Rama¹²⁸; Pedro-Luis Sánchez^114,58; Antonio J. Sánchez López²⁵⁵; María Concepción Sánchez Prados¹⁸; Javier Sánchez Real¹¹; Jorge Sánchez Redondo^1,256; Clara Sánchez-Pablo¹¹⁴; Olga Sánchez-Pernaute²⁵⁷; Xiana Taboada-Fraga¹⁹⁵; Eduardo Tamayo^258,140,7; Alvaro Tamayo-Velasco²⁵⁹; Juan Carlos Taracido-Fernandez¹⁵⁵; Nathali A.C. Tavares²⁶⁰; Carlos Tellería^107,108; Jair Antonio Tenorio Castaño^27,130,205; Alejandro Teper¹³¹; Ronald P. Torres Gutiérrez²²¹; Juan Torres-Macho²⁶¹; Lilian Torres-Tobar³⁹; Jesús Troya²⁴³; Miguel Urioste¹⁹⁹; Juan Valencia-Ramos²⁶²; Agustín Valido^21,263; Juan Pablo Vargas Gallo^264,265; Belén Varón²⁶⁶; Romero H.T. Vasconcelos²⁶⁰; Tomas Vega²⁶⁷; Santiago Velasco-Quirce²⁶⁸; Julia Vidán Estévez¹¹; Miriam Vieitez-Santiago^45,47; Carlos Vilches²⁶⁹; Lavinia Villalobos¹¹; Felipe Villar²¹⁸; Judit Villar-Garcia^270,271,272; Cristina Villaverde^26,27; Pablo Villoslada-Blanco²¹⁴; Ana Virseda-Berdices⁶⁹; Valentina Vélez-Santamaría^87,88; Virginia Víctor^35,36; Zuleima Yáñez²⁰; Antonio Zapatero-Gaviria²⁷³; Ruth Zarate²⁷⁴; Sandra Zazo²³⁶; Gabriela V. da Silva⁴¹; Raimundo de Andrés²⁷⁵; Jéssica N.G. de Araújo²⁵²; Carmen de Juan²⁰¹; Julianna Lys de Sousa Alves Neri²⁷⁶; Carmen de la Horra⁷⁹; Ana B. de la Hoz³³; Victor del Campo-Pérez²⁷⁷; Manoella do Monte Alves^278,279; Katiusse A. dos Santos⁸⁶; Yady Álvarez-Benítez^19,20; Felipe Álvarez-Navia^81,58; María Íñiguez²¹⁴; Miguel López de Heredia²⁷; Ingrid Mendes²⁷; Rocío Moreno²⁷; Esther Sande^27,129,102; Carlos Flores^{280,98,99,234}; José A. Riancho^45,46,47,27; Augusto Rojas-Martinez⁸²; Pablo Lapunzina^27,130,205; Angel Carracedo^{27,129,102,128}

Scourge Cohort Group’s filiations

¹ Hospital Universitario Mostoles, Medicina Interna, Madrid, Spain

² Universidad Francisco de Vitoria, Madrid, Spain

³ Haemostasis and Thrombosis Unit, Hospital de la Santa Creu i Sant Pau, IIB Sant Pau, Barcelona, Spain

⁴ Unit of Infectious Diseases, Hospital Universitario 12 de Octubre, Instituto de Investigación Sanitaria Hospital 12 de Octubre (imas12), Madrid, Spain

⁵ Spanish Network for Research in Infectious Diseases (REIPI RD16/0016/0002), Instituto de Salud Carlos III, Madrid, Spain

⁶ School of Medicine, Universidad Complutense, Madrid, Spain

⁷ Centro de Investigación Biomédica en Red de Enfermedades Infecciosas (CIBERINFEC), Instituto de Salud Carlos III, Madrid, Spain

⁸ Hospital General Santa Bárbara de Soria, Soria, Spain

⁹ Pediatric Neurology Unit, Department of Pediatrics, Navarra Health Service Hospital, Pamplona, Spain

¹⁰ Navarra Health Service, NavarraBioMed Research Group, Pamplona, Spain

¹¹ Complejo Asistencial Universitario de León, León, Spain

¹² Hospital Universitario San Pedro, Infectious Diseases Department, Logroño, Spain

¹³ Fundación Institut Guttmann, Institut Universitari de Neurorehabilitació adscrit a la UAB, Hospital de Neurorehabilitació, Barcelona, Spain

¹⁴ Universitat Autònoma de Barcelona (UAB), Barcelona, Spain

¹⁵ Fundació Institut d’Investigació en Ciències de la Salut Germans Trias i Pujol, Barcelona, Spain

¹⁶ Hospital General de Occidente, Guadalajara, Mexico

¹⁷ Microbiology Unit, Hospital Universitario N. S. de Candelaria, Santa Cruz de Tenerife, Spain

¹⁸ Hospital Universitario La Paz-IDIPAZ, Servicio de Neumología, Madrid, Spain

¹⁹ Camino Universitario Adelita de Char, Mired IPS, Barranquilla, Colombia

²⁰ Universidad Simón Bolívar, Facultad de Ciencias de la Salud, Barranquilla, Colombia

²¹ Hospital Universitario Virgen Macarena, Neumología, Seville, Spain

²² Departamento de Medicina, Universidad de Salamanca, Salamanca, Spain

²³ Centro de Investigación del Cáncer (IBMCC) Universidad de Salamanca - CSIC, Salamanca, Spain

²⁴ Biomedical Research Institute of Salamanca (IBSAL) Salamanca, Spain

²⁵ Centre for Biomedical Network Research on Cancer (CIBERONC), Instituto de Salud Carlos III, Madrid, Spain

²⁶ Department of Genetics & Genomics, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

²⁷ Centre for Biomedical Network Research on Rare Diseases (CIBERER), Instituto de Salud Carlos III, Madrid, Spain

²⁸ Spanish National Cancer Research Centre, Human Genotyping-CEGEN Unit, Madrid, Spain

²⁹ Hospital das Forças Armadas, Brazil

³⁰ Exército Brasileiro, Brazil

³¹ Department of Child and Adolescent Psychiatry, Institute of Psychiatry and Mental Health, Hospital General Universitario Gregorio Marañón (IiSGM), Madrid, Spain

³² Clinical Pharmacology Service, Hospital de la Santa Creu i Sant Pau, IIB Sant Pau, Barcelona, Spain

³³ Biocruces Bizkai HRI, Barakaldo, Bizkaia, Spain

³⁴ Cruces University Hospital, Osakidetza, Barakaldo, Bizkaia, Spain

³⁵ Hospital Infanta Elena, Valdemoro, Madrid, Spain

³⁶ Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

³⁷ Centre for Biomedical Network Research on Mental Health (CIBERSAM), Instituto de Salud Carlos III, Madrid, Spain

³⁸ Fundación Hospital Infantil Universitario de San José, Bogotá, Colombia

³⁹ Fundación Universitaria de Ciencias de la Salud, Bogotá, Colombia

⁴⁰ Universidade Federal do Rio Grande do Norte, Programa de Pós-graduação em Ciências da Saúde, Natal, Brazil

⁴¹ Universidade Federal do Rio Grande do Norte, Departamento de Medicina Clínica, Natal, Brazil

⁴² Departamento de Genética e Morfologia, Instituto de Ciências Biológicas, Universidade de Brasília, Brasilia, Brazil

⁴³ Colégio Marista de Brasilia, Brazil

⁴⁴ Associação Brasileira de Educação e Cultura, Brazil

⁴⁵ IDIVAL, Santander, Spain

⁴⁶ Universidad de Cantabria, Santander, Spain

⁴⁷ Hospital U M Valdecilla, Santander, Spain

⁴⁸ Hospital Universitario La Paz-IDIPAZ, Servicio de Medicina Interna, Madrid, Spain

⁴⁹ Fundació Docència I Recerca Mutua Terrassa, Barcelona, Spain

⁵⁰ Spanish National Cancer Research Center, CNIO Biobank, Madrid, Spain

⁵¹ Fundación Santa Fe de Bogota, Departamento Patologia y Laboratorios, Bogotá, Colombia

⁵² Hospital General de Occidente, Zapopan, Jalisco, Mexico

⁵³ Centro Universitario de Tonalá, Universidad de Guadalajara, Tonalá, Jalisco, Mexico

⁵⁴ Centro de Investigación Multidisciplinario en Salud, Universidad de Guadalajara, Tonalá, Jalisco, Mexico

⁵⁵ Instituto Murciano de Investigación Biosanitaria (IMIB-Arrixaca), Murcia, Spain

⁵⁶ Universidad Católica San Antonio de Murcia (UCAM), Murcia, Spain

⁵⁷ Hospital Universitario de Salamanca-IBSAL, Servicio de Medicina Interna-Unidad de Enfermedades Infecciosas, Salamanca, Spain

⁵⁸ Universidad de Salamanca, Salamanca, Spain

⁵⁹ Hospital Universitario de Fuenlabrada, Department of Internal Medicine, Madrid, Spain

⁶⁰ Escola Tecnica de Saúde, Laboratorio de Vigilancia Molecular Aplicada, Pará, Brazil

⁶¹ Federal University of Pernambuco, Genetics Postgraduate Program, Recife, PE, Brazil

⁶² Hospital Universitario Infanta Leonor, Servicio de Alergia, Madrid, Spain

⁶³ Hospital Universitario del Tajo, Servicio de Medicina Intensiva, Aranjuez, Spain

⁶⁴ Hospital Universitario Mutua Terrassa, Barcelona, Spain

⁶⁵ Hospital Universitario La Paz-IDIPAZ, Servicio de Farmacología, Madrid, Spain

⁶⁶ Alcaldía de Barranquilla, Secretaría de Salud, Barranquilla, Colombia

⁶⁷ Instituto de Investigación Sanitaria de Santiago (IDIS), Xenética Cardiovascular, Santiago de Compostela, Spain

⁶⁸ Centre for Biomedical Network Research on Cardiovascular Diseases (CIBERCV), Instituto de Salud Carlos III, Madrid, Spain

⁶⁹ Unidad de Infección Viral e Inmunidad, Centro Nacional de Microbiología (CNM), Instituto de Salud Carlos III (ISCIII), Madrid, Spain

⁷⁰ Cardiovascular Genetics Center, Institut d’Investigació Biomèdica Girona (IDIBGI), Girona, Spain

⁷¹ Medical Science Department, School of Medicine, University of Girona, Girona, Spain

⁷² Hospital Josep Trueta, Cardiology Service, Girona, Spain

⁷³ Institute of Biomedicine of Seville (IBiS), Consejo Superior de Investigaciones Científicas (CSIC)-University of Seville-Virgen del Rocio University Hospital, Seville, Spain

⁷⁴ Division of Infectious Diseases, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

⁷⁵ Intensive Care Unit, Hospital Universitario Insular de Gran Canaria, Las Palmas de Gran Canaria, Spain

⁷⁶ Hospital Universitario Mutua Terrassa, Terrassa, Spain

⁷⁷ Departamento de Medicina, Hospital Universitario Virgen del Rocío,Universidad de Sevilla, Seville, Spain

⁷⁸ Centre for Biomedical Network Research on Epidemiology and Public Health (CIBERESP), Instituto de Salud Carlos III, Madrid, Spain

⁷⁹ Institute of Biomedicine of Seville (IBiS), Consejo Superior de Investigaciones Científicas (CSIC)-University of Seville-Virgen del Rocio University Hospital, Seville, Spain

⁸⁰ Universidad de los Andes, Facultad de Ciencias, Bogotá, Colombia

⁸¹ Hospital Universitario de Salamanca-IBSAL, Servicio de Medicina Interna, Salamanca, Spain

⁸² Tecnologico de Monterrey, Escuela de Medicina y Ciencias de la Salud and Hospital San Jose TecSalud, Monterrey, Mexico

⁸³ University of Fortaleza (UNIFOR), Department of Nutrition. Fortaleza, Brazil

⁸⁴ Departamento de Química, Faculdade de Filosofia, Ciências e Letras de Ribeirão Preto, Universidade de São Paulo, Brazil

⁸⁵ Andalusian Public Health System Biobank, Granada, Spain

⁸⁶ Universidade Federal do Rio Grande do Norte, Programa de Pós-Graduação em Ciências Farmacêuticas, Natal, Brazil

⁸⁷ Neuromuscular Unit, Neurology Department, Hospital Universitari de Bellvitge, L’Hospitalet de Llobregat (Barcelona), Spain

⁸⁸ Bellvitge Biomedical Research Institute (IDIBELL), Neurometabolic Diseases Laboratory, L’Hospitalet de Llobregat, Spain

⁸⁹ Centre for Biomedical Network Research on Diabetes and Metabolic Associated Diseases (CIBERDEM), Instituto de Salud Carlos III, Madrid, Spain

⁹⁰ University of Pais Vasco, UPV/EHU, Bizkaia, Spain

⁹¹ Oncology and Genetics Unit, Instituto de Investigacion Sanitaria Galicia Sur, Xerencia de Xestion Integrada de Vigo-Servizo Galego de Saúde, Vigo, Spain

⁹² Hospital Universitario La Paz, Hospital Carlos III, Madrid, Spain

⁹³ Hospital de San José, Sociedad de Cirugía de Bogota, Bogotá, Colombia

⁹⁴ Hospital Universitario Río Hortega, Valladolid, Spain

⁹⁵ Servicio de Medicina intensiva, Complejo Hospitalario Universitario de A Coruña (CHUAC), Sistema Galego de Saúde (SERGAS), A Coruña, Spain

⁹⁶ Valencia University, Preventive Medicine Department, Valencia, Spain

⁹⁷ Centre for Biomedical Network Research on Physiopatology of Obesity and Nutrition (CIBEROBN), Instituto de Salud Carlos III, Madrid, Spain

⁹⁸ Research Unit, Hospital Universitario N.S. de Candelaria, Santa Cruz de Tenerife, Spain

⁹⁹ Centre for Biomedical Network Research on Respiratory Diseases (CIBERES), Instituto de Salud Carlos III, Madrid, Spain

¹⁰⁰ Otto von Guericke University, Departament of Microgravity and Translational Regenerative Medicine, Magdeburg, Germany

¹⁰¹ Maternidade Escola Janário Cicco, Natal, Brazil

¹⁰² Centro Singular de Investigación en Medicina Molecular y Enfermedades Crónicas (CIMUS), Universidade de Santiago de Compostela, Santiago de Compostela, Spain

¹⁰³ Institute of Psychiatry and Mental Health, Hospital General Universitario Gregorio Marañón (IiSGM), Madrid, Spain

¹⁰⁴ Programa de Pós Graduação em Ciências da Saúde, Faculdade de Medicina, Universidade de Brasília, Brasilia, Brazil

¹⁰⁵ Fundació Docència I Recerca Mutua Terrassa, Terrassa, Spain

¹⁰⁶ Hospital Universitario Mostoles, Unidad de Genética, Madrid, Spain

¹⁰⁷ Instituto Aragonés de Ciencias de la Salud (IACS), Zaragoza, Spain

¹⁰⁸ Instituto Investigación Sanitaria Aragón (IIS-Aragon), Zaragoza, Spain

¹⁰⁹ Department of Preventive Medicine and Public Health, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain

¹¹⁰ IdiPaz (Instituto de Investigación Sanitaria Hospital Universitario La Paz), Madrid, Spain

¹¹¹ Hospital Universitario Virgen del Rocío, Servicio de Medicina Interna, Seville, Spain

¹¹² Unidad Diagnóstico Molecular. Fundación Rioja Salud, La Rioja, Spain

¹¹³ Hospital Universitario Quironsalud Madrid, Madrid, Spain

¹¹⁴ Hospital Universitario de Salamanca-IBSAL, Servicio de Cardiología, Salamanca, Spain

¹¹⁵ Hospital Universitario Puerta de Hierro, Servicio de Medicina Interna, Majadahonda, Spain

¹¹⁶ Biocruces Bizkaia Health Research Institute, Galdakao University Hospital, Osakidetza, Bizkaia, Spain

¹¹⁷ Instituto Regional de Investigación en Salud-Universidad Nacional de Caaguazú, Caaguazú, Paraguay

¹¹⁸ Universidade Federal do Pará, Núcleo de Pesquisas em Oncologia, Belém, Pará, Brazil

¹¹⁹ Hospital Ophir Loyola, Departamento de Ensino e Pesquisa, Belém, Pará, Brazil

¹²⁰ Universidad Nacional de Asunción, Facultad de Politécnica, Paraguay

¹²¹ Fundación Asilo San Jose, Santander, Spain

¹²² Unidad de Enfermedades Infecciosas, Servicio de Medicina Interna, Hospital Universitario Puerta de Hierro, Instituto de Investigación Sanitaria Puerta de Hierro - Segovia de Arana, Madrid, Spain

¹²³ Urgencias Hospitalarias, Complejo Hospitalario Universitario de A Coruña (CHUAC), Sistema Galego de Saúde (SERGAS), A Coruña, Spain

¹²⁴ Grupo de Investigación en Interacciones Gen-Ambiente y Salud (GIIGAS) - Instituto de Biomedicina (IBIOMED), Universidad de León, León, Spain

¹²⁵ Hospital Universitario Niño Jesús, Pediatrics Department, Madrid, Spain

¹²⁶ Unitat de Malalties Infeccioses i Importades, Servei de Pediatría, Infectious and Imported Diseases, Pediatric Unit, Hospital Universitari Sant Joan de Deú, Barcelona, Spain

¹²⁷ Microbiology Department, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

¹²⁸ Fundación Pública Galega de Medicina Xenómica, Sistema Galego de Saúde (SERGAS) Santiago de Compostela, Spain

¹²⁹ Instituto de Investigación Sanitaria de Santiago (IDIS), Santiago de Compostela, Spain

¹³⁰ Instituto de Genética Médica y Molecular (INGEMM), Hospital Universitario La Paz-IDIPAZ, Madrid, Spain

¹³¹ Hospital de Niños Ricardo Gutierrez, Buenos Aires, Argentina

¹³² Centre for Biomedical Network Research on Rare Diseases (CIBERER), Instituto de Salud Carlos III, Madrid, Spain Universidad Francisco de Vitoria, Madrid,Spain

¹³³ Hospital Infanta Elena, Servicio de Medicina Intensiva, Valdemoro, Madrid, Spain

¹³⁴ University of Salamanca, Biomedical Research Institute of Salamanca (IBSAL), Salamanca, Spain

¹³⁵ Department of Immunology, IRYCIS, Hospital Universitario Ramón y Cajal, Madrid, Spain

¹³⁶ Osakidetza, Cruces University Hospital, Bizkaia, Spain

¹³⁷ Instituto Mexicano del Seguro Social, IMSS. Centro Médico Nacional La Raza. Hospital de Infectología. Mexico City, Mexico.

¹³⁸ Hospital Universitario de Getafe, Servicio de Genética, Madrid, Spain

¹³⁹ Ministerio de Salud Ciudad de Buenos Aires, Buenos Aires, Argentina

¹⁴⁰ Hospital Clinico Universitario de Valladolid, Unidad de Apoyo a la Investigación, Valladolid, Spain

¹⁴¹ Universidad de Valladolid, Departamento de Cirugía, Valladolid, Spain

¹⁴² Secretaria Municipal de Saude de Apodi, Natal, Brazil

¹⁴³ Sección Genética Médica - Servicio de Pediatría, Hospital Clínico Universitario Virgen de la Arrixaca, Servicio Murciano de Salud, Murcia, Spain

¹⁴⁴ Departamento Cirugía, Pediatría, Obstetricia y Ginecología, Facultad de Medicina, Universidad de Murcia (UMU), Murcia, Spain

¹⁴⁵ Hospital Universitario Centro Dermatológico Federico Lleras Acosta, Bogotá, Colombia

¹⁴⁶ Hospital Universitario Virgen de las Nieves, Servicio de Análisis Clínicos e Inmunología, Granada, Spain

¹⁴⁷ Pneumology Department, Hospital General Universitario Gregorio Marañón (iiSGM), Madrid, Spain

¹⁴⁸ Intermediate Respiratory Care Unit, Department of Pneumology, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

¹⁴⁹ Clinica Comfamiliar Risaralda, Pereira, Colombia

¹⁵⁰ Centro Universitario de Tonalá, Universidad de Guadalajara, Guadalajara, Mexico

¹⁵¹ Centro de Investigación Multidisciplinario en Salud, Universidad de Guadalajara, Guadalajara, Mexico

¹⁵² Unidad de Cuidados, Intensivos Hospital Clínico Universitario de Santiago (CHUS), Sistema Galego de Saúde (SERGAS), Santiago de Compostela, Spain

¹⁵³ IIS La Fe, Plataforma de Farmacogenética, Valencia, Spain

¹⁵⁴ Universidad de Valencia, Departamento de Farmacología, Valencia, Spain

¹⁵⁵ Data Analysis Department, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

¹⁵⁶ Hospital del Mar, Infectious Diseases Service, Barcelona, Spain

¹⁵⁷ Institut Hospital del Mar d’Investigacions Mèdiques (IMIM), Barcelona, Spain

¹⁵⁸ CEXS-Universitat Pompeu Fabra, Spanish Network for Research in Infectious Diseases (REIPI), Barcelona, Spain

¹⁵⁹ Biocruces Bizkaia Health Research Institute, Basurto University Hospital, Osakidetza, Bizkaia, Spain

¹⁶⁰ Sabin Medicina Diagnóstica, Brazil

¹⁶¹ Opthalmology Department, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

¹⁶² Hospital Sant Joan de Deu,Pediatric Critical Care Unit, Barcelona, Spain

¹⁶³ Paediatric Intensive Care Unit, Agrupación Hospitalaria Clínic-Sant Joan de Déu, Esplugues de Llobregat, Barcelona, Spain

¹⁶⁴ Hospital Universitario 12 de Octubre, Department of Immunology, Madrid, Spain

¹⁶⁵ Instituto de Investigación Sanitaria Hospital 12 de Octubre (imas12), Transplant Immunology and Immunodeficiencies Group, Madrid, Spain

¹⁶⁶ SIGEN Alianza Universidad de los Andes - Fundación Santa Fe de Bogotá, Bogotá, Colombia

¹⁶⁷ Hospital General de Segovia, Medicina Intensiva, Segovia, Spain

¹⁶⁸ Programa de Pós-Graduação em Biologia Animal, Universidade de Brasília, Brasília, Brazil

¹⁶⁹ Clinical Trials Unit, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

¹⁷⁰ IMDEA-Food Institute, CEI UAM+CSIC, Madrid, Spain

¹⁷¹ Intensive Care Unit, Hospital Universitario de Canarias, La Laguna, Spain

¹⁷² Dirección General de Salud Pública, Consejería de Sanidad, Junta de Castilla y León, Valladolid, Spain

¹⁷³ Universidade Federal do Rio Grande do Norte, Departamento de Analises Clinicas e Toxicologicas, Natal, Brazil

¹⁷⁴ Hospital Universitario La Paz-IDIPAZ, Servicio de Inmunología, Madrid, Spain

¹⁷⁵ La Paz Institute for Health Research (IdiPAZ), Lymphocyte Pathophysiology in Immunodeficiencies Group, Madrid, Spain

¹⁷⁶ Hospital Universitario Virgen de las Nieves, Servicio de Enfermedades Infecciosas, Granada, Spain

¹⁷⁷ Instituto de Investigación Biosanitaria de Granada (ibs GRANADA), Granada, Spain

¹⁷⁸ Universidad de Granada, Departamento de Medicina, Granada, Spain

¹⁷⁹ Faculdade de Medicina, Universidade de Brasília, Brasilia, Brazil

¹⁸⁰ Fundación Jiménez Díaz, Epidemiology, Madrid, Spain

¹⁸¹ Universidad Autónoma de Madrid, Department of Medicine, Madrid, Spain

¹⁸² Universidad de Valladolid, Departamento de Medicina, Valladolid, Spain

¹⁸³ Hospital Universitario Infanta Leonor, Servicio de Medicina Intensiva, Madrid, Spain

¹⁸⁴ Unidad de Genética y Genómica Islas Baleares, Islas Baleares, Spain

¹⁸⁵ Hospital Universitario Son Espases, Unidad de Diagnóstico Molecular y Genética Clínica, Islas Baleares, Spain

¹⁸⁶ Genomics of Complex Diseases Unit, Research Institute of Hospital de la Santa Creu i Sant Pau, IIB Sant Pau, Barcelona, Spain

¹⁸⁷ Intensive Care Unit, Hospital Universitario N. S. de Candelaria, Santa Cruz de Tenerife, Spain

¹⁸⁸ Preventive Medicine Department, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

¹⁸⁹ Servicio de Medicina Interna, Sanatorio Franchin, Buenos Aires, Argentina

¹⁹⁰ Hospital Universitario del Tajo, Servicio de Medicina Intensiva, Toledo, Spain

¹⁹¹ Programa de Pós-Graduação em Ciências Médicas, Universidade de Brasília, Brasilia, Brazil

¹⁹² Programa de Pós-Graduação em Ciências da Saúde, Universidade de Brasília, Brasilia, Brazil

¹⁹³ Hospital El Bierzo, Gerencia de Asistencia Sanitaria del Bierzo (GASBI), Gerencia Regional de Salud (SACYL), Ponferrada, Spain

¹⁹⁴ Grupo INVESTEN, Instituto de Salud Carlos III, Madrid, Spain

¹⁹⁵ Unidad de Cuidados Intensivos, Complejo Universitario de A Coruña (CHUAC), Sistema Galego de Saúde (SERGAS), A Coruña, Spain

¹⁹⁶ Programa de Pós-Graduação em Genética da Faculdade de Medicina de Ribeirão Preto

¹⁹⁷ Programa de Pós-Graduação em Química da Faculdade de Filosofia, Ciências e Letras de Ribeirão Preto

¹⁹⁸ Hospital El Bierzo, Unidad Cuidados Intensivos, León, Spain

¹⁹⁹ Spanish National Cancer Research Centre, Familial Cancer Clinical Unit, Madrid, Spain

²⁰⁰ Instituto de Investigación Sanitaria San Carlos (IdISSC), Hospital Clínico San Carlos (HCSC), Madrid, Spain

²⁰¹ Hospital Universitario Severo Ochoa, Servicio de Medicina Interna, Madrid, Spain

²⁰² Universidad de Sevilla, Departamento de Enfermería, Seville, Spain

²⁰³ Hospital General Universitario Gregorio Marañón (IiSGM), Madrid, Spain ²⁰⁴ Hospital Universitario La Paz-IDIPAZ, Servicio de Pediatría, Madrid, Spain ²⁰⁵ ERN-ITHACA-European Reference Network

²⁰⁶ Unidad de Genética y Genómica Islas Baleares, Unidad de Diagnóstico Molecular y Genética Clínica, Hospital Universitario Son Espases, Islas Baleares, Spain

²⁰⁷ Instituto de Investigación Sanitaria Islas Baleares (IdISBa), Islas Baleares, Spain

²⁰⁸ Programa de Pós-Graduação em Biologia Animal, Universidade de Brasília, Brasília, Brazil

²⁰⁹ Programa de Pós-Graduação em Ciências da Saúde, Universidade de Brasília, Brasília, Brazil

²¹⁰ Programa de Pós-Graduação Profissional em Ensino de Biologia, Universidade de Brasília, Brasília, Brazil

²¹¹ Programa de Pós-Graduação em Ciências Médicas, Universidade de Brasília, Brasília, Brazil

²¹² Anatomía Patológica, Instituto de Investigación Sanitaria San Carlos (IdISSC), Hospital Clínico San Carlos (HCSC), Madrid, Spain

²¹³ Tecnológico de Monterrey, Monterrey, Mexico

²¹⁴ Infectious Diseases, Microbiota and Metabolism Unit, CSIC Associated Unit, Center for Biomedical Research of La Rioja (CIBIR), Logroño, Spain

²¹⁵ Centro de Investigación en Anomalías Congénitas y Enfermedades Raras (CIACER), Universidad Icesi

²¹⁶ Departamento de Genetica, Fundación Valle del Lili

²¹⁷ Universidad Complutense de Madrid, Department of Immunology, Ophthalmology and ENT, Madrid, Spain

²¹⁸ Department of Neumology, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

²¹⁹ Hospital Nuestra Señora de Sonsoles, Ávila, Spain

²²⁰ Inditex, A Coruña, Spain

²²¹ Osakidetza, Cruces University Hospital, Barakaldo, Bizkaia, Spain

²²² GENYCA, Madrid, Spain

²²³ Neuromuscular Diseases Unit, Department of Neurology, Hospital de la Santa Creu i Sant Pau, Universitat Autònoma de Barcelona, Barcelona, Spain

²²⁴ Instituto Mexicano del Seguro Social (IMSS), Centro Médico Nacional Siglo XXI, Unidad de Investigación Médica en Enfermedades Infecciosas y Parasitarias, Mexico City, Mexico

²²⁵ Catalan Institution of Research and Advanced Studies (ICREA), Barcelona, Spain

²²⁶ Intensive Care Department, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

²²⁷ Hospital Universitario Príncipe de Asturias, Servicio de Microbiología Clínica, Madrid, Spain

²²⁸ Universidad de Alcalá de Henares, Departamento de Biomedicina y Biotecnología, Facultad de Medicina y Ciencias de la Salud, Madrid, Spain

²²⁹ Drug Research Centre, Institut d’Investigació Biomèdica Sant Pau, IIB-Sant Pau, Barcelona, Spain

²³⁰ Departamento de Genetica, Clinica imbanaco

²³¹ Unidad de Cuidados Intensivos, Hospital Clínico Universitario de Santiago (CHUS), Sistema Galego de Saúde (SERGAS), Santiago de Compostela, Spain

²³² Universidade de Brasília, Brasilia, Brazil

²³³ Department of Immunology, Hospital Universitario de Gran Canaria Dr. Negrín, Las Palmas de Gran Canaria, Spain

²³⁴ Department of Clinical Sciences, University Fernando Pessoa Canarias, Las Palmas de Gran Canaria, Spain

²³⁵ University Hospital Germans Trias i Pujol, Pediatrics Department, Badalona, Spain

²³⁶ Department of Pathology, Biobank, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

²³⁷ Hospital Universitario Virgen de las Nieves, Servicio de Medicina Interna, Granada, Spain

²³⁸ Fundación Universitaria de Ciencias de la Salud, Grupo de Ciencias Básicas en Salud (CBS), Bogotá, Colombia

²³⁹ Sociedad de Cirugía de Bogotá, Hospital de San José, Bogotá, Colombia

²⁴⁰ Universidad de Granada, Departamento Bioquímica, Biología Molecular e Inmunología III, Granada, Spain

²⁴¹ Hospital Infanta Elena, Allergy Unit, Valdemoro, Madrid, Spain

²⁴² Faculty of Medicine, Universidad Francisco de Vitoria, Madrid, Spain

²⁴³ Hospital Universitario Infanta Leonor, Madrid, Spain

²⁴⁴ Complutense University of Madrid, Madrid, Spain

²⁴⁵ Gregorio Marañón Health Research Institute (IiSGM), Madrid, Spain

²⁴⁶ Colégio Militar de Brasília

²⁴⁷ The John Walton Muscular Dystrophy Research Centre, Newcastle University and Newcastle Hospitals NHS Foundation Trust, Newcastle upon Tyne, UK.

²⁴⁸ Neuromuscular Unit, Neuropediatrics Department, Institut de Recerca Sant Joan de Déu, Hospital Sant Joan de Déu, Spain

²⁴⁹ Fundació Docència i Recerca Mutua Terrassa, Terrassa, Spain

²⁵⁰ Casa de Saúde São Lucas, Natal, Brazil

²⁵¹ Hospital Rio Grande, Rio Grande do Norte, Natal, Brazil

²⁵² Universidade Federal do Rio Grande do Norte, Pós-graduação em Biotecnologia - Rede de Biotecnologia do Nordeste (Renorbio), Natal, Brazil

²⁵³ Intensive Care Unit, Hospital Universitario de Gran Canaria Dr. Negrín, Las Palmas de Gran Canaria, Spain

²⁵⁴ Universidad Fernando Pessoa Canarias, Las Palmas de Gran Canaria, Spain

²⁵⁵ Biobank, Puerta de Hierro-Segovia de Arana Health Research Institute, Madrid, Spain

²⁵⁶ Universidad Rey Juan Carlos, Madrid, Spain

²⁵⁷ Reumathology Service, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

²⁵⁸ Hospital Clinico Universitario de Valladolid, Servicio de Anestesiologia y Reanimación, Valladolid, Spain

²⁵⁹ Hospital Clinico Universitario de Valladolid, Servicio de Hematologia y Hemoterapia, Valladolid, Spain

²⁶⁰ Hospital Universitario Lauro Wanderley, Brazil

²⁶¹ Hospital Universitario Infanta Leonor, Servicio de Medicina Interna, Madrid, Spain

²⁶² University Hospital of Burgos, Burgos, Spain

²⁶³ Universidad de Sevilla, Seville, Spain

²⁶⁴ Fundación Santa Fe de Bogota, Instituto de servicios medicos de Emergencia y trauma, Bogotá, Colombia

²⁶⁵ Universidad de los Andes, Bogotá, Colombia

²⁶⁶ Quironprevención, A Coruña, Spain

²⁶⁷ Junta de Castilla y León, Consejería de Sanidad, Valladolid, Spain

²⁶⁸ Gerencia Atención Primaria de Burgos, Burgos, Spain

²⁶⁹ Immunogenetics-Histocompatibility group, Servicio de Inmunología, Instituto de Investigación Sanitaria Puerta de Hierro - Segovia de Arana, Madrid, Spain

²⁷⁰ Hospital del Mar, Department of Infectious Diseases, Barcelona, Spain

²⁷¹ IMIM (Hospital del Mar Medical Research Institute, Institut Hospital del Mar d’Investigacions Mediques), Barcelona, Spain

²⁷² Universitat Autònoma de Barcelona, Department of Medicine, Spain

²⁷³ Consejería de Sanidad, Comunidad de Madrid, Madrid, Spain

²⁷⁴ Centro para el Desarrollo de la Investigación Científica, Asunción, Paraguay

²⁷⁵ Internal Medicine Department, Instituto de Investigación Sanitaria-Fundación Jiménez Díaz University Hospital - Universidad Autónoma de Madrid (IIS-FJD, UAM), Madrid, Spain

²⁷⁶ Universidade Federal do Rio Grande do Norte, Programa de Pós Graduação em Nutrição, Natal, Brazil

²⁷⁷ Preventive Medicine Department, Instituto de Investigacion Sanitaria Galicia Sur, Xerencia de Xestion Integrada de Vigo-Servizo Galego de Saúde, Vigo, Spain

²⁷⁸ Universidade Federal do Rio Grande do Norte, Departamento de Infectologia, Natal, Brazil

²⁷⁹ Hospital de Doenças Infecciosas Giselda Trigueiro, Rio Grande do Norte, Natal, Brazil

²⁸⁰ Genomics Division, Instituto Tecnológico y de Energías Renovables, Santa Cruz de Tenerife, Spain