key: cord-0937957-wx8sm3v2 authors: Wu, L.; Zhu, J.; Wu, C. title: Mendelian randomization analysis to characterize causal association between coronary artery disease and COVID-19 date: 2020-05-30 journal: nan DOI: 10.1101/2020.05.29.20117309 sha: 5b90aef8b35029660f0d50d64afaa051ce663563 doc_id: 937957 cord_uid: wx8sm3v2 Observational studies have suggested that having coronary artery disease increases the risk of Coronavirus disease 2019 (COVID-19) susceptibility and severity, but it remains unclear if this association is causal. Inferring causation is critical to facilitate the development of appropriate policies and/or individual decisions to reduce the incidence and burden of COVID-19. We applied Two-sample Mendelian randomization analysis and found that genetically predicted CAD was significantly associated with higher risk of COVID-19: the odds ratio was 1.29 (95% confidence interval 1.11 to 1.49; P = 0.001) per unit higher log odds of having CAD. Coronavirus disease 2019 (COVID-19) has become a global pandemic. Specific risk factors such as coronary artery disease (CAD) 1 have been reported to be related to susceptibility and severity through conventional observational studies. However, findings from conventional observational studies are susceptible to selection bias, unmeasured confounding, and reverse causation. With these limitations, it is very challenging to establish the causality of reported associations in observational studies. Inferring causation is critical to facilitate the development of appropriate policies and/or individual decisions to reduce the incidence and burden of COVID-19. Here we show that the Mendelian Randomization (MR) design, similar to a "genetic randomized controlled trial", can be adopted to decipher the causality of the association between CAD and COVID-19 2. For evaluation of associations with COVID-19 risk, we used summary statistics data of the most recent version of genome-wide association study (GWAS) analyses from The COVID-19 host genetics initiative (released on May 17, 2020) 3. Detailed information on participating studies, quality control and analyses has been provided on the COVID-19 host genetics initiative website is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 30, 2020. . proportion of other ethnic groups. Only variants with imputation quality > 0.6 were retained. We extracted genetic instruments for CAD risk based on a comprehensive GWAS of ~185,000 CAD cases and controls by Nikpay et al 4. In this study, a majority of the participants (77%) were of European ancestry, and 19% were of Asian ancestry. We selected single nucleotide polymorphisms (SNPs) associated at P<5×10-8 and only retained independent variants from each other (r2 < 0.001). For correlated SNPs, the SNP with the lowest p-value was selected. We applied the widely-used inverse variance weighted (IVW) method 5 to estimate the overall causal association of CAD on COVID-19 susceptibility. To account for potential violations of valid instrumental variable assumptions, we conducted sensitive analyses by applying several methods that are robust to horizontal pleiotropy at the cost of reduced statistical power. These include weighted median MR, MR-Egger regression, and MR-PRESSO test, and leave-one-out analysis. Results are presented as odds ratios per unit higher log odds of CAD. For MR analysis, we used 38 independent variants as the instrument for CAD (data not shown) with an estimated F-statistic of 62. Genetically predicted CAD was significantly associated with higher risk of COVID-19. The odds ratio was 1.29 (95% confidence interval 1.11 to 1.49; P = 0.001) per unit higher log odds of having CAD (Figure 1) . The association remained robust in . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 30, 2020. . https://doi.org/10.1101/2020.05.29.20117309 doi: medRxiv preprint weighted median MR (Figure 1 ) and leave one out analyses (Figure 2) . There was limited evidence of horizontal pleiotropy and heterogeneity based on the MR-Egger intercept test (P=0.83) and MR-PRESSO (P=0.26). In an MR analysis, we observed evidence supporting a potential causal association between CAD and COVID-19. Such a link may be explained by changes of viscosity during febrile illnesses, heightened coagulation systems, endothelial cell dysfunction, or proinflammatory effects 6. This is the first study to characterize potential causality of suspected risk factors for COVID-19 susceptibility using a MR design. Limitations of the current study include 1) the overlap of relatively small samples in both GWAS of CAD and COVID-19; and 2) mixed population composition in both GWAS of CAD and COVID-19. Because BioMe only contributed a relatively small sample size to the COVID-19 GWAS: 20 (1.1%) cases and 10,169 (1.5%) controls, the potential influence of this on the MR estimate should be small. In both GWAS of CAD and COVID-19, a majority of the subjects are Europeans. In summary, an MR study could potentially avoid many biases and confounding issues existing in conventional observational studies and thus help to identify causally related risk factors. Using MR design, we found evidence that having CAD is associated with a higher risk of COVID-19. Therefore, particular attention should be given to individuals with CAD during this pandemic and we expect that this finding will facilitate appropriate policymaking and individual decisions to reduce COVID-19 burden. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 30, 2020. . Cardiovascular Disease, Drug Therapy, and Mortality in Covid-19 Mendelian randomization: genetic anchors for causal inference in epidemiological studies The COVID-19 Host Genetics Initiative, a global initiative to elucidate the role of host genetic factors in susceptibility and severity of the SARS-CoV-2 virus pandemic A comprehensive 1,000 Genomes-based genome-wide association meta-analysis of coronary artery disease Mendelian randomization analysis with multiple genetic variants using summarized data COVID-19 Illness and Heart Failure: A Missing Link? JACC Heart Fail No potential conflicts of interest were disclosed by the authors.