Study population and setting
This case-control meta-analysis included summary statistics from 46 different studies: the sample size consisted of 49,562 cases of European (77%), Middle Eastern (4.9%), East Asian (3.6%), South Asian (3.6%), African (4.9%), and Admixed American (7.0%) ancestry. Three main categories of COVID-19 disease were defined: SARS-Cov-2 infected individuals who were hospitalized for COVID-19 and are either deceased or require respiratory support, cases with lab-confirmed SARS-Cov-2 infection hospitalized with moderate to severe COVID-19, and all cases that had lab-confirmed SARS-CoV-2 infection or physician or self-reported COVID-19. GWAS analysis was run using SAIGE or PLINK, and meta-analyses were performed using the summary statistics from each study. A PheWAS (phenome-wide association study) was conducted to investigate previously reported phenotypes and to investigate 15 index variants associated with risk of developing COVID-19. Finally, GWAS summary statistics for 43 complex disease, behavioral, neuropsychiatric, biomarker, and complex disease phenotypes were chosen for genetic correlation and Mendelian randomization analyses.
Summary of Main Findings
Thirteen distinct loci associated with SARS-CoV-2 infection or COVID-19 were identified. The strongest signal for increased susceptibility to SARS-CoV-2 infection was at the ABO locus, with variants in two additional loci (PPP1R15A and SLC6A20) also demonstrating associations with higher infection susceptibility. Nine of the 13 loci were associated with an increased risk of developing severe COVID-19 symptoms, including variants in DPP9 (OR 1.29, p 2.0×10-12) and FOXP4 (OR 1.2, p 6.0×10-13), which were previously identified as increasing lung disease risk. Previously identified autoimmune disease-protective variants in TYK2 conferred an increased risk for hospitalization due to COVID-19 (OR 1.43, 95% CI: 1.29-1.59, p 9.71×10-12), and a variant in KANSL1 (OR 0.96, p 1.00×10-20) was associated protectively against COVID-19-related hospitalization. Interestingly, heritability of SARS-CoV-2 infection was enriched in genes expressed in the lung (p 5×10-4). Overall, this meta-analysis suggests a polygenic architecture (that is, influenced by more than one gene) of SARS-CoV-2 infection and COVID-19 severity.
The study population is large and drawn from multiple studies with global ancestry representation, albeit primarily European ancestry (77%). The augmented sample size increases the statistical power to identify associations of varying effect size.
Despite the identification of genetic variants associated with infection and disease, untagged genetic variation suggested by linkage disequilibrium structure and physical proximity, particularly at the SLC6A20 locus, may drive association signals in certain regions. Variability in case ascertainment, sample sizes, and case phenotyping among the included 46 studies may bias the associations. Inclusion of untested population controls assumed not to have been infected may bias effect sizes. Lower socioeconomic status and other socio-demographic variables associated with higher SARS-CoV-2 infection risk, COVID-19 disease severity, and study sampling are likely to have introduced selection bias, which may further distort effect sizes.
This study brings together the largest number of COVID-19 host genetics studies to date using standardized methods. This study provides valuable insight on the putative genes that may be involved with infection and severity, and warrants additional gene exploration with refined phenotypes and additional diverse populations.
This review was posted on: 11 September 2021