Diagnostic and Prognostic Test for Periodontitis Using Nuclear Magnetic Resonance and Mass Spectrometry

Abstract

Abstract Background: The technological advances in the field of metabolomics have shown promise for the development of a site-specific diagnostic test for periodontitis using the gingival crevicular fluid (GCF). We hypothesize that metabolites, identified through Mass Spectrometry (MS) will distinguish between health and disease. Methods: Ten stage III/IV periodontitis patients were recruited and ten healthy controls and GCF samples from the 6 Ramfjord teeth and 2 randomly selected sites were harvested from subjects with a porous silver disc. The samples were analyzed with Hydrophilic interaction liquid chromatography (HILIC) and reverse-phase liquid chromatography (RPLC) coupled with tandem MS. Statistical analysis was performed using mixed linear models and the p-value was adjusted to control false discovery rate (FDR) using the method of Benjamini and Hochberg. Metaboanalyst 5.0 (www.metaboanalyst.ca) was utilized for pathway analysis. Results: 3,491 metabolites were identified from the sample analysis and 108 could be matched to existing libraries. Those metabolites can be statistically assigned (p<0.05) to 12 human metabolic pathways. Butyric acid was associated healthy sites of healthy subjects, while Thymine, L-Theanine, Cytokinin B, L-Aspartic acid, D-Galactose, Citruline and Stachyose were associated with diseased sites of diseased subjects with an FDR <5%. The remaining identified compounds presented with a higher FDR. More compounds were detected with FDR <5%, but could not be identified due to warranting a targeted approach. Conclusion(s): GCF metabolites identified through MS could be a potential future means to diagnose periodontitis. Butyric acid was associated with periodontal health and thymine, L-theanine, Cytokinin B, L-aspartic acid, D-galactose, citrulline and stachyose were associated with periodontitis. Longitudinal comparisons and a larger sample could show value in the development of a predictive test/model as well.