RESUMEN
Background: The association between gut microbiome and coronavirus disease 2019 (COVID-19) has attracted much attention, but its causality remains unclear and requires more direct evidence. Methods: In this study, we conducted the bidirectional Mendelian randomization (MR) analysis to assess the causal association between gut microbiome and COVID-19 based on the summary statistics data of genome-wide association studies (GWASs). Over 1.8 million individuals with three COVID-19 phenotypes (severity, hospitalization and infection) were included. And 196 bacterial taxa from phylum to genus were analyzed. The inverse-variance weighted (IVW) analysis was chosen as the primary method. Besides, false discovery rate (FDR) correction of p-value was used. To test the robustness of the causal relationships with p-FDR < 0.05, sensitivity analyses including the secondary MR analyses, horizontal pleiotropy test, outliers test, and "leave-one-out" analysis were conducted. Results: In the forward MR, we found that 3, 8, and 10 bacterial taxa had suggestive effects on COVID-19 severity, hospitalization and infection, respectively. The genus Alloprevotella [odds ratio (OR) = 1.67; 95% confidence interval (95% CI), 1.32-2.11; p = 1.69×10-5, p-FDR = 2.01×10-3] was causally associated with a higher COVID-19 severity risk. In the reverse MR, COVID-19 severity, hospitalization and infection had suggestive effects on the abundance of 4, 8 and 10 bacterial taxa, respectively. COVID-19 hospitalization causally increased the abundance of the phylum Bacteroidetes (OR = 1.13; 95% CI, 1.04-1.22; p = 3.02×10-3; p-FDR = 2.72×10-2). However, secondary MR analyses indicated that the result of COVID-19 hospitalization on the phylum Bacteroidetes required careful consideration. Conclusion: Our study revealed the causal association between gut microbiome and COVID-19 and highlighted the role of "gut-lung axis" in the progression of COVID-19.