METABOLOMICS ANALYSIS OF SALIVA COLLECTED FROM PATIENTS WITH PERIODONTAL DISEASE

Aim: This study proposed a novel metabolomic approach based on the thorough investigation of oral metabolites to study the potential use of a metabonomic analysis as a diagnostic periodontal tool. In this validation study we want to supply some evidence whether metabonomic profiling of saliva samples can provide a signature of the disease. Methods: Saliva samples were collected from a cohort of 157 subjects (36 males, 31 females, mean age 45.27± 10.74 years) referred to the Section of Periodontology, University of Torino (Italy). The diagnosis of periodontal diseases was made by an experienced clinician according to the classification proposed by Armitage (1999). 39 healthy subjects (HS) and 118 patients with clinical and radiographic diagnosis of periodontal diseases (PD) were consecutively selected: 16 gingivitis (G), 83 chronic periodontitis (CP) and 19 aggressive periodontitis (AP) patients. The collection of saliva was done in the morning pre-prandial, between 8 a.m. and 10 a.m. for all subjects. The day scheduled for saliva collection the patients had no food or beverage, with exception of water and did not wash their teeth with toothpaste or mouthrinsing. Every subjects was seated comfortably and was advised not to force salivation, then was asked to spit saliva into a sterile graduated tube for 10 min. About 1 ml of saliva was collected from every patient and immediately frozen. For each saliva sample, one-dimensional (1D) NMR (nuclear magnetic resonance) spectrum was acquired with water peak suppression using a standard pulse sequence (Bruker terminology: noesygppr1d.comp), 64 scans, 96 k data points, a spectral width of 18028 Hz, a relaxation delay of 4 s. All metabolites of interest were then manually checked and their NMR signals were assigned on template 1D NMR profiles by using matching routines of AMIX 3.8.4 (Bruker BioSpin) in combination with the BBIOREFCODE (Version 2.0.0; Bruker BioSpin) reference database and published literature when available. Unsupervised Principal Component Analysis (PCA) was run for obtaining a general overview of the variance of NMR profile. Results: The pattern recognition analysis of NMR profiles could discriminate CP patients (n = 83) from HS (n = 39) with an accuracy of 73%. Metabolic profiles of CP patients exhibited higher concentrations of acetate, c-aminobutyrate, n-butyrate, succinate, trimethylamine, propionate, phenylalanine and valine, and decreased concentrations of pyruvate and N-acetyl groups compared with controls (P<0.05). Conclusion: Our results can provide a contribution to the understanding of the biochemical network and pathway in the PD, however at this stage the method can not be extended to the general population as a ready-to-use clinical tool, due to the limited cohort recruited and the exploratory nature of this work. Anyway, a further validation of the statistical model on a larger cohort is in progress with the aim to demonstrate the potential impact on the clinical practice of our findings.