Evaluation of Bacteriophages As a Signature of Microbiome Health: a Systematic Review and Meta-Analysis.

BACKGROUND:Parasites are foundational to ecosystem health both as indicator species of community productivity and as drivers of diversity. In bacterial communities, bacteriophage viruses can have such roles as they track and modulate the dynamic composition of bacterial hosts within an ecosystem. We aimed to test whether viromes can be used as broad signatures of microbiome health using previously published results across systems. METHODS:In this systematic review and meta-analysis, we searched PubMed, Google Scholar, Scopus, and Web of Science from Jan 22, 2022, to Sept 17, 2024, for peer-reviewed, primary literature published in English, using search terms "phage diversity", "microbiome", "virome", "virus", "phageome", "disease", and "dysbiosis". Inclusion criteria were: a comparison between a dysbiosis state and a healthy state in a human or animal host; a defined host organism and microbiome site; examination of the virome; an obtained measure of virome diversity (α, β, or both); use of statistical analysis to assess whether α or β diversity are changed in dysbiosis; and sufficient methodology description on viral isolation and on virus sequence analysis pipeline. We conducted a qualitative data analysis to assess factors explaining changes to virome diversity in dysbiosis. We then calculated response ratios for each study to test for overall patterns of virome α diversity change under disturbance. Finally, we conducted a quantitative analysis on studies from which we were able to obtain paired virome and bacteriome α diversity data to examine the correlation between these data in defined health compared with defined disturbance conditions. This study was not registered. FINDINGS:We identified a total of 74 studies for inclusion that spanned human (n=61), mouse (n=8), pig (n=3), dog (n=1), and cow (n=1) hosts and a diverse spectrum of infections and diseases. By comparing observed phage and bacterial diversity in microbiomes characterised by dysbiosis with those considered control populations, we were able to identify some key commonalities. Of the 69 studies that investigated changes to α diversity of the virome in dysbiosis, 28 (41%) reported significant changes, but with variable directional change. Of 38 datasets (from 30 studies) for which virome α diversity values were available, 22 (58%) gave a response ratio of less than 1 (α diversity decreases in dysbiosis) and 16 (42%) of more than 1 (α diversity increases in dysbiosis); however, in 27 (71%) datasets, 95% CIs overlapped with 1 (ie, no change in α diversity). We found shifting virome composition to be a more consistent signature of dysbiosis, with 47 (69%) of 68 studies reporting a significant change in viral β diversity with dysbiosis. 62 (89%) of 70 studies reported significant enrichment of system-specific viral taxa under dysbiosis. Our quantitative correlation analysis suggested that bacterial α diversity is a greater predictor of virome α diversity in healthy groups (mean r2=0·380; 95% CI 0·597-0·163) than in dysbiosis (mean r2=0·118, 0·223-0·012; sign test for asymmetric non-parametric data p=4·9 × 10-10). INTERPRETATION:Overall, although specific viral signatures of dysbiosis are likely to be highly disease-specific and condition-specific, we show that existing ecological theory shows promise in predicting the relationship between bacterial and phage diversity and in providing broad signatures of dysbiosis across disease systems. Our observation that the relationship between bacterial and phage diversity breaks down under disturbance suggests that this feature could be a useful signature of dysbiosis and that future studies incorporating the virome could provide opportunity to diagnose, treat, and better understand the causes of microbiome disturbance. FUNDING:There was no funding source for this study.