Study of Variation Over Time in the Oral Microbiome

The Human Microbiome Project (HMP), started in 2008, aimed to characterize the microbial communities on the skin, in the oral and nasal cavities, and in the gastrointestinal and urogenital tracts. This project cross-sectionally described the microbial communities in a healthy human population, but did not adequately describe longitudinal stability. Out of the 236 adults included in the HMP, a second saliva sample was collected on a small subset (N=74) with a long interval between collections, an average of 203 days. Community similarity was calculated between two visits; for saliva samples, the Spearman correlation was high (rs=0.8). However, it was unclear which metric of community similarly was being used for this calculation. When we obtained the HMP data and calculated the Spearman correlation for specific diversity metrics in the saliva samples, the correlations were weaker and ranged from 0.225 to 0.524 for the Shannon Index and PD tree, respectively. Other studies with fewer people, but more time points, have had conflicting conclusions on the temporal variation of the oral microbiome. Given this variability, it is vital to establish whether a single measurement is representative of the typical composition of microbiota in a specific body site using more than two time points.

In epidemiological studies, it is often impractical, if not impossible, to collect biological samples or measurements multiple times for each participant during the study period. However, when a biological sample or a measurement is taken only once, there is concern that the single assessment is not representative of the typical value. For example, serum 25(OH)D levels, which are used to assess an individual s vitamin D exposure, have seasonal fluctuations, but studies have established that a single measurement is sufficiently reliable to represent typical exposure. The correlation between samples over time is also used to create more accurate power calculations for future studies.

Since previous work has not adequately addressed temporal variation in the oral microbiome, we propose to: 1) establish a cohort of 40 adults willing to provide oral samples approximately every two months for one year; 2) process all oral samples and evaluate the presence and quantity of microbiota using 16S ribosomal RNA gene sequencing; 3) calculate appropriate reliability measures for the sample over time using the intra-class correlation coefficient (ICC) for and <= diversity metrics; and 4) empirically evaluate the efficiency of different longitudinal designs including two or more sample collections with an outcome such as body mass index (BMI). Because of the complexity and high prevalence of human DNA in saliva, metagenomic analyses of oral samples are not yet feasible, however, the samples obtained from this study will be available for use in future metagenomic analyses when the technology is operationalized. The current study is essential to establish the reliability of a single measure of the oral microbiome to determine appropriate sample sizes for future cohort studies considering the effect of the oral microbiome on the risk of cancer and other health outcomes.