Saliva sampling is simple and non-invasive and this fluid is a rich source of biomarkers thanks to the high diversity of its microbiome, proteome and metabolome. The SALAMANDER project aims at defining salivary signatures indicative of healthy dietary choices (adherence to a Mediterranean diet: MeDi) with a positive long-term health outcome (protection against type 2 diabetes: T2DM).

We first described the link between adherence to MeDi and incidence of T2DM in the UK population. Our main hypothesis was that the benefit of MeDi on the reduced risk for T2DM was partly explained by an impact on weight status. We used nutritional and clinical data from 21612 UK Biobank participants. A nutritional score reflecting the adherence of each participant to a MeDi was computed. Then we identified participants with incident T2DM. Using regression-based mediation analyses, we observed that higher adherence to a MeDi was associated with a significant 14% reduced risk for diabetes (HR 0.86, 95%CI [0.83 – 0.90]) and that this effect was in part explained by an indirect effect mediated by a lower odds (-10%) of being overweight (HR 0.90, 95%CI [0.88 – 0.92]). 4% of the total effect were not explained by overweight (HR 0.96, 95%CI [0.93 – 0.99]). These results add a mechanistic explanation to the well-known association between a healthy dietary pattern and the lower risk of diabetes, and suggest that both overweight and non-overweight people should adopt a healthy dietary pattern to avoid T2DM.

In parallel, 200 saliva samples from older participants (cohorts 3 City-Bordeaux, France and ENRICA-Seniors, Spain) were obtained and analyzed by 16SrDNA analyses (microbiome), ¹H NMR (metabolome) and LC-MS-MS (proteome). For microbiome, there were 25 abundant bacterial genera containing at least 0.5% of the total sequence reads. These were characteristic of oral microbial communities, the most abundant genera being Streptococcus, Veillonella, Prevotella and Neisseria. 450 unique amplicon sequence variants (ASVs) were identified. Alterations in the alpha and beta diversities of the oral microbiome was linked with age, sex and cohort but not with T2DM. Three dietary items (cooking with olive oil, butter and vegetables intake) were correlated with altered microbial diversity. For metabolome, 126 different peaks corresponding to 47 identified and 31 currently unidentified metabolites were quantified. In accordance with microbiome data, no robust model could explain the diabetic status by metabolome. Finally for proteome, 246 proteins were semi-quantified. Approximately 50 proteins were significantly different between diabetic and non-diabetic subjects. These proteins were linked to three dietary items: cooking with olive oil, butter and vegetables intake. 

The three datasets (microbiome, metabolome, proteome) are currently jointly analyzed in order to identify salivary markers of T2DM and/or dietary habits.

/