Quantification of Human Microbiome Stability Over 6 Months: Implications for Epidemiologic Studies

Area of Science:

• Microbiome research
• Statistical genetics
• Human health

Background:

• Temporal variation in microbiome measurements can significantly reduce statistical power in research studies.
• Quantifying this variation is crucial for designing effective studies, particularly for chronic diseases.
• Understanding microbiome stability over time informs study design and sample size calculations.

Purpose of the Study:

• To evaluate the temporal stability of microbiome measurements across different body sites.
• To determine the impact of temporal variation on statistical power for microbiome studies.
• To calculate sample sizes needed for detecting microbiome differences in case-control studies.

Main Methods:

• Analysis of 16S ribosomal RNA profiles from paired biological specimens collected 6 months apart.
• Utilized data from three distinct studies: National Cancer Institute colorectal cancer study, Costa Rica study, and the Human Microbiome Project.
• Calculated intraclass correlation coefficients (ICCs) to assess temporal stability and estimated sample sizes for nested case-control designs.

Main Results:

• Phylum-level and alpha-diversity metrics showed ICCs greater than 0.5 across various body sites.
• Fecal beta-diversity estimates also demonstrated ICCs over 0.5, indicating reasonable temporal stability.
• Detecting an odds ratio of 2.0 with a single sample typically requires 300-500 cases; using sequential specimens can reduce this by 40%-50% for low-ICC metrics.

Conclusions:

• Relative abundances of major phyla and alpha-diversity metrics exhibit low temporal stability, necessitating large sample sizes for detecting moderate effect size associations.
• Fecal beta-diversity is reasonably stable, allowing for smaller sample sizes to detect associations with microbial community composition.
• Detecting modest disease associations with specific microbiome metrics requires sequential prediagnostic specimens from thousands of prospectively ascertained cases.