An integrative analysis of the biological clock hypothesis in human gut microbiome

Abstract

BackgroundWhile previous studies have explored the relationship between aging and the gut microbiome, it remains unclear how consistent and reproducible this association is across different cultures and groups. We performed an integrative analysis with 11 independent datasets from nine different countries to test the idea that the aging gut microbiome can be viewed as a biological clock, in which microbial changes associated with age are consistent and measurable across distinct datasets. ResultsAs expected, our Principal Coordinate Analysis found strong batch effects with study ID by far the strongest signal across datasets. Despite this large batch effect, we found a consistent signal across studies that was largely driven by sample size with only our larger cohorts showing taxa in common associated with age. Likewise, Shannon diversity and richness were not consistently associated with age across the 11 datasets, but some positive correlations with richness and host age were observed among the four largest cohorts. The taxon with the most potential as a biomarker for the aging human gut microbiome is genus Bifidobacterium, with significantly negative associations with host age in three out of the four datasets that had more than 200 samples. ConclusionThe driving force behind low reproducibility of association of age with the microbiome in previous studies appears to be inadequate sample size rather than structural differences in the microbial community based on cohort characteristics. Results from a power analysis suggest that future studies on the aging human gut microbiome consider on the order of 100-300 samples to consistently observe an age signal. With these larger sample sizes, parametric and non-parametric model yield broadly similar power.