Investigation of a Novel Predictive Biomarker Profile for the Outcome of Periodontal Treatment.

BACKGROUND: An ability to predict the response to conventional non-surgical treatment of a periodontal site would be advantageous. However, biomarkers or tests devised to achieve this have lacked sensitivity. The aim of this study is to assess the ability of a novel combination of biomarkers to predict treatment outcome of patients with chronic periodontitis. METHODS: Gingival crevicular fluid (GCF) and subgingival plaque were collected from 77 patients at three representative sites, one healthy (probing depth [PD] </=3 mm) and two diseased (PD >/=6 mm), at baseline and at 3 and 6 months after treatment. Patients received standard non-surgical periodontal treatment at each time point as appropriate. The outcome measure was improvement in probing depth of >/=2 mm. Concentrations of active enzymes (matrix metalloproteinase [MMP]-8, elastase, and sialidase) in GCF and subgingival plaque levels of Porphyromonas gingivalis, Tannerella forsythia, and Fusobacterium nucleatum were analyzed for prediction of the outcome measure. RESULTS: Using threshold values of MMP-8 (94 ng/muL), elastase (33 ng/muL), sialidase (23 ng/muL), and levels of P. gingivalis (0.23%) and T. forsythia (0.35%), receiver operating characteristic curves analysis demonstrated that these biomarkers at baseline could differentiate healthy from diseased sites (sensitivity and specificity >/=77%). Furthermore, logistic regression showed that this combination of these biomarkers at baseline provided accurate predictions of treatment outcome (>/=92%). CONCLUSION: The "fingerprint" of GCF enzymes and bacteria described here offers a way to predict the outcome of non-surgical periodontal treatment on a site-specific basis.