Background: The temporal dynamics of the gut microbiome and its associated factors remain poorly understood.
Methods: We analyzed 3,466 shotgun metagenomes that were repeatedly collected over weeks to years in four cohorts: 213 men from the Men’s Lifestyle Validation Study (MLVS, interval: 6 months), 306 women from the Mind-Body Study (MBS, interval: 6 months), 90 women from the Microbiome Among Nurses Study (Micro-N, interval: 3 years), and 130 participants from the Human Microbiome Project 2 (HMP2, interval: 2-52 weeks). Diet was assessed via food frequency questionnaires, while lifestyle and clinical information were collected by questionnaires administered before or at stool collection. Gut microbiomes were profiled for taxonomy and functions using MetaPhlAn4 and HUMAnN4, respectively. Temporal stability of the gut microbiome was assessed by the within-individual Bray-Curtis (BC) distance between repeated samples over time. Using 10-fold cross-validation Random Forest models, we examined, within each cohort, the variation in species- and pathway-level BC distances explained by dietary (up to 37 food groups, 3 dietary patterns, and fiber intake) and non-dietary factors (lifestyle, medications, and stool-related factors), as well as baseline microbial features.
Results: Overall, collection timepoint explained a minimal (0.2–0.5%) variation in the BC of the microbial compositions (detected species number range: 395–423) and pathways (355–443). Although within-individual BC distances increased with longer time intervals, they remained substantially lower than between-individual distances for both species and pathways (Species BC for 3-year interval, mean: 0.47 vs. 0.73). Within-individual pathways showed higher stability than species. Across cohorts, dietary factors, non-dietary factors, microbial species, and functional pathways accounted for 9.0–16.1%, 11.8–29.0%, 9.4–18.2%, and 4.1–32.3% of the variations in the within-individual species-level BC, respectively. For pathway-level BC, these factors explained 5.9–26.9%, 16.9–21.8%, 7.8–22.6%, 5.2–41.6% of the variation, respectively. No factor consistently predicted within-individual BC distances across all four cohorts.
Conclusions: The gut microbiome is relatively stable in epidemiologic settings, providing important context for future population-based research using fecal microbiome data. The overall associations of dietary and lifestyle factors with microbiome stability were weak and inconsistent.