Influential factors of saliva microbiota composition

Abstract

The oral microbiota is emerging as an influential factor of host physiology and disease state. Factors influencing oral microbiota composition have not been well characterised. In particular, there is a lack of population-based studies. We undertook a large hypothesis-free study of the saliva microbiota, considering potential influential factors of host health (frailty; diet; periodontal disease), demographics (age; sex; BMI) and sample processing (storage time), in a sample (n = 679) of the TwinsUK cohort of adult twins. Alpha and beta diversity of the saliva microbiota was associated most strongly with frailty (alpha diversity: β = -0.16, Q = 0.003, Observed; β = -0.16, Q = 0.002, Shannon; β = -0.16, Q = 0.003, Simpson; Beta diversity: Q = 0.002, Bray Curtis dissimilarity) and age (alpha diversity: β = 0.15, Q = 0.006, Shannon; β = 0.12, Q = 0.003, Simpson; beta diversity: Q = 0.002, Bray Curtis dissimilarity; Q = 0.032, Weighted UniFrac) in multivariate models including age, frailty, sex, BMI, frailty and diet, and adjustment for multiple testing. Those with a more advanced age were more likely to be dissimilar in the saliva microbiota composition than younger participants (P = 5.125e-06, ANOVA). In subsample analyses, including consideration of periodontal disease (total n = 138, periodontal disease n = 66), the association with frailty remained for alpha diversity (Q = 0.002, Observed ASVs; Q = 0.04 Shannon Index), but not beta diversity, whilst age was not demonstrated to associate with alpha or beta diversity in this subsample, potentially due to insufficient statistical power. Length of time that samples were stored prior to sequencing was associated with beta diversity (Q = 0.002, Bray Curtis dissimilarity). Six bacterial taxa were associated with age after adjustment for frailty and diet. Of the factors studied, frailty and age emerged as the most influential with regards to saliva microbiota composition. Whilst age and frailty are correlates, the associations were independent of each other, giving precedence to both biological and chronological ageing as processes of potential importance when considering saliva microbiota composition.

Figure 1

Heat tree of phylogenetic relationship and relative abundance of ASVs within the saliva microbiota of TwinsUK participants (n = 679). Colour and size of nodes relate to the taxonomic level and abundance of taxa, respectively. Higher to lower taxonomic levels range from blue to dark green to light green, to grey.

Figure 2

Association of factors with beta diversity of the saliva microbiota. Models using both the full sample (n = 679) and diet subsample (n = 582) are shown. PERMANOVA R2 (marginal) is plotted for each variable. The diet subsample includes only participants for whom dietary data were available. Bars are annotated with FDR adjusted p values (Q values). Age was significantly associated with beta diversity after adjusting for all other factors listed plus sequencing depth when measured using both Bray Curtis and Weighted UniFrac. Frailty was associated with Bray Curtis beta diversity (Q = 0.003). Bray Curtis dissimilarity was also associated with sample storage time (length of time that samples were stored in the freezer (Q = 0.004) and diet (Q = 0.04; subsample analysis).

Figure 3

Bray Curtis beta dispersion of age within full cohort sample. A difference in the variance across age groups was demonstrated (P = 5.125e−06, ANOVA).

Figure 4

PCoA showing Bray Curtis dissimilarity according to frailty, within the full sample. Bray Curtis dissimilarity was associated with frailty index (FI), Q = 0.002, PERMANOVA. Each sample is assigned colour gradient reflecting low (brown) to higher (blue) frailty index.

Figure 5

Saliva microbiota associated with age. Age was associated with six taxa in a multivariate model after adjustment for sex, frailty and diet of participants. Taxa positively associated with age were Comamonas (Q = 0.029), Phocaeicola abscessus (Q = 0.037) and Anaeroglobulus germinatus (Q = 0.004). Taxa inversely associated with age were Veillonella (Q = 0.0002), Haemophilus (Q = 0.0004), Veillonella atypica/dispar (Q = 0.037) and Granulicatella adiacens/para-adiecens (Q = 0.05).