Comparative analysis of gingival crevicular fluid and peri-implant crevicular fluid by mid-infrared spectroscopy: a split mouth study

Abstract

Objectives: This proof-of-concept study aimed to compare the biochemical composition of gingival crevicular fluid (GCF) and peri-implant crevicular fluid (PICF) under healthy conditions, through mid-infrared (MIR) spectroscopy.

Materials and methods: Using a split-mouth design, GCF and PICF samples were collected from 12 participants and analyzed through MIR spectroscopy. Advanced chemometric models, including partial least squares-discriminant analysis, k-nearest neighbors, and support vector machine discriminant analysis, were applied to explore potential biochemical differences between the biofluids.

Results: No cluster formation was observed with PCA, indicating a high degree of similarity between groups. The PLS-DA model didn't effectively discriminate between GCF and PICF with prediction rates of 62.5% (10/16) for calibration, 37.5% (6/16) for cross-validation, and 50% (4/8) for validation. The k-NN model, using k = 3 neighbors showed 25% (4/16) correct classification rates during calibration and a validation set accuracy of 50%. SVM-DA analysis showed a correct prediction rate of 37.5% (6/16) for calibration and 50% for cross-validation 50% (8/16) and 50% (4/8) in the validation phase. Nonetheless, subtle spectral differences were observed in spectral regions R1 (3982-2652 cm⁻¹) and R4 (1180-922 cm⁻¹), suggesting a slightly increased lipidic content and the presence of ethers and glycosidic bonds linked to carbohydrates, in PICF.

Conclusions: The lack of significant biochemical differences between GCF and PICF under healthy conditions, as determined by MIR spectroscopy, suggests that implant-related changes in PICF composition are negligible.

Clinical relevance: The demonstrated biochemical similarity between GCF and PICF under healthy conditions reinforces the potential of PICF as a reliable biofluid for diagnostic applications, including monitoring oral and systemic health biomarkers, without significant influence from implant-related factors.

Keywords: Chemometric models; Gingival crevicular fluid; Mid-infrared spectroscopy; Peri-implant crevicular fluid.

Fig. 1

Flowchart illustrating patient recruitment, eligibility screening, group allocation (healthy gingival crevicular fluid vs healthy peri-implant crevicular fluid), sample collection and spectral data analysis

Fig. 2

Raw MIR spectra of GCF and PICF samples collected from teeth and dental implant crevices, respectively. Five distinct spectral regions were identified and analyzed separately to assess differences between the two groups

Fig. 3

Average raw MIR spectra of GCF PICF samples collected from the crevices of teeth and dental implants, respectively

Fig. 4

a Hotelling’s T² plot (weighed sum of squared scores) and Q residuals (sum of squared residuals) statistics, evaluated using a 95% confidence interval, for outlier detection. B PCA scores plot of Principal Component 1 (PC1) versus Principal Component 2 (PC2), considering the entire spectra using 2 principal components. Spectra were previously mean centered