Although many recent studies have examined associations between the gut microbiome and COVID-19 disease severity in individual patient cohorts, questions remain on the robustness across international cohorts of the biomarkers they reported. Here, we performed a meta-analysis of eight shotgun metagenomic studies of COVID-19 patients (comprising 1,023 stool samples) and 23 > 16S rRNA gene amplicon sequencing (16S) cohorts (2,415 total stool samples). We found that disease severity (as defined by the WHO clinical progression scale) was associated with taxonomic and functional microbiome differences. This alteration in gut microbiome configuration peaks at days 7-30 post diagnosis, after which the gut microbiome returns to a configuration that becomes more similar to that of healthy controls over time. Furthermore, we identified a core set of species that were consistently associated with disease severity across shotgun metagenomic and 16S cohorts, and whose abundance can accurately predict disease severity category of SARS-CoV-2 infected subjects, with Actinomyces oris abundance predicting population-level mortality rate of COVID-19. Additionally, we used relational diet-microbiome databases constructed from cohort studies to predict microbiota-targeted diet patterns that would modulate gut microbiota composition toward that of healthy controls. Finally, we demonstrated the association of disease severity with the composition of intestinal archaeal, fungal, viral, and parasitic communities. Collectively, this study has identified robust COVID-19 microbiome biomarkers, established accurate predictive models as a basis for clinical prognostic tests for disease severity, and proposed biomarker-targeted diets for managing COVID-19 infection.
Keywords: COVID; disease severity; eukaryotic microorganisms; gut microbiome; meta-analysis; microbiota-targeted diet.