Accuracy of Metabolomics in Peri-Implant Crevicular Fluid for Diagnosis and Prognosis of Peri-Implantitis.

Abstract

Aims: Poor accuracy of diagnostic and prognostic tools prevents the prediction of peri-implant disease stability or progression. We analyzed metabolites from peri-implant crevicular fluid (PICF) samples from healthy and diseased implants to identify those diagnostic of health and peri-implant disease and predictive of peri-implant bone loss over time.

Methods: Clinical, radiographic examinations and PICF samples were collected from 59 healthy implants, 33 implants with peri-implantitis, and 38 implants with peri-implant mucositis in 71 subjects. A subset of implants was evaluated at 6, 12, 18, and 24 months. Over time, all initially healthy implants remained stable (Group B, N = 28), whereas 6 initially diseased implants continued to lose bone and 8 did not (Group C). PICF metabolites were measured using proton-nuclear magnetic resonance (1H-NMR) 2-dimensional Total Correlation Spectroscopy. PCA and PLS-DA tested the cross-sectional clustering and importance of each metabolite, while the AUC summarized the accuracy of predicting radiographic bone changes ≥ 1 mm at 6-month intervals.

Results: At baseline, the Cadaverine/Lysine and Putrescine/Lysine signatures diagnosed peri-implantitis (AUC = 0.76 and 0.70; p < 0.000) with good accuracy, while alpha-ketoglutarate diagnosed implant health (AUC = 0.706; p = 0.002). Combining metabolites increased diagnostic accuracy (AUC_{Cadaverine/Lysine+Methionine} = 0.81; p < 0.01). Proline and 1-3-diaminopropane predicted future bone loss (AUC_Proline = 0.917 and AUC_{1-3-diaminopropane} = 0.854). ANOVA post hoc analysis established that biotin and propionate levels were higher in Group C compared to Groups A and B (p < 0.001; AUC_biotin = 0.889; AUC_propionate = 0.87). Valine levels were higher in Groups A and C compared to Group B (p = 0.002; AUC = 0.841).

Conclusions: ¹H-NMR 2-dimensional spectroscopy identified PICF metabolites diagnostic of peri-implantitis with high accuracy. Despite the small number of affected implants, metabolite signatures that predict future bone loss in peri-implantitis appear to be different from those diagnostic of peri-implantitis.