Diagnostic Accuracy of Novel Protein Biomarkers in Saliva to Detect Periodontitis Using Untargeted 'SWATH' Mass Spectrometry

To discover new salivary biomarkers to diagnose periodontitis and evaluate the impact of age and smoking on predictive capacity. Saliva samples were collected from 44 healthy periodontal individuals and 41 with periodontitis. Samples were analysed by sequential window acquisition of all theoretical mass spectra (SWATH-MS), and proteins were identified by employing the UniProt database. The diagnostic capacity of the molecules was determined with generalized additive models. The models obtained were single-protein unadjusted and adjusted for age and smoking status, besides two-protein combinations. Eight single salivary proteins had a bias-corrected accuracy (bc-ACC) of 78.8%-86.8% (bc-sensitivity/bc-specificity of 62.5%-86.9%/60.9%-98.1%) to diagnose periodontitis. Predictive capacity increased more by adjusting for age (bc-ACC: 94.1%-98.2%; bc-sensitivity/bc-specificity: 90.2%-98.6%/93.6%-97.2%) than smoking (bc-ACC: 83.9%-90.4%; bc-sensitivity/bc-specificity: 73.6%-89.9%/76.2%-96.4%). These proteins were keratin, type II cytoskeletal 1, protein S100-A8, β-2-microglobulin, neutrophil defensin 1, lysozyme C, ubiquitin-60S ribosomal protein L40, isoform 2 of tropomyosin α-3 chain and resistin. Two dual combinations showed bc-sensitivity/bc-specificity of > 90%: β-2-microglobulin with profilin-1, and lysozyme C with zymogen granule protein 16 homologue B. New salivary biomarkers show good or excellent ability to diagnose periodontitis. Age has a more significant influence on the accuracy of the single biomarkers than smoking, with results comparable to two-protein combinations.