Biologic modelling of periodontal disease progression

AIM: To investigate the synergistic role of biologic markers from saliva, serum and plaque in modelling periodontitis disease progression. MATERIAL AND METHODS: This longitudinal study evaluated characteristics of disease progression in 114 patients with generalized moderate to severe periodontitis. The primary outcome was detection of sites with progressing attachment loss sites over 6 months in patients who received scaling and root planing or oral hygiene only. The predictive potential of 27 biomarkers in serum, whole saliva and subgingival plaque was evaluated using three classification algorithms (Support Vector Machines; Naive Bayes Classifier; and Linear Discriminant Analysis) within an ensemble predictive modelling framework. RESULTS: Disease progression occurred in 24.6% of subjects (28/114). Predictive modelling using Naive Bayes Classifier identified progressors best with sensitivity of ~89%. The use of the three classification algorithms revealed the concerted role of salivary matrix metalloproteinase-8, serum biomarkers (serum amyloid P, matrix metalloproteinase 1, bactericidal permeability-increasing protein, isoprostane) along with levels of Porphryomonas gingivalis and Tannerella forsythia in plaque in predicting progressors. CONCLUSIONS: Synergistic utility of baseline bacterial and inflammatory biomarkers from saliva, serum and plaque predicted disease progression.