Multivariate versus univariate spectrum analysis of dentine sialophosphoprotein (DSPP) for root resorption prediction: a clinical trial

• Published in: BMC oral health Vol. 22; no. 1; p. 151
• Main Authors: Mohd Zain, Mohd Norzaliman, Md Yusof, Zalhan, Basri, Katrul Nadia, Yazid, Farinawati, Teh, Yong Xian, Ashari, Asma, Zainal Ariffin, Shahrul Hisham, Megat Abdul Wahab, Rohaya
• Format: Journal Article
• Language: English
• Published: England BioMed Central Ltd 29.04.2022; BioMed Central; BMC
• Subjects: Absorption spectroscopy; Antibodies; Biomarkers; Collagen; Dentin - metabolism; DSPP; ELISA; Enzymes; Extracellular Matrix Proteins - metabolism; Humans; Incisors; Inflammation; Medical research; Methods; Molars; Multivariate analysis; Orthodontics; Phosphoproteins - metabolism; Proteins; Root Resorption; Roots; Sialoglycoproteins - metabolism; Spectrum Analysis; Teeth; Univariate analysis; Wavelength; Malaysia; Absorption spectroscopy; Root resorption; Univariate analysis; Multivariate analysis; DSPP; ELISA
• DOI: 10.1186/s12903-022-02178-2
• ISSN: 1472-6831

A force applied during orthodontic treatment induces inflammation to root area and lead to root resorption known as orthodontically induced inflammatory root resorption (OIIRR). Dentine sialophosphoprotein (DSPP) is one of the most abundant non-collagenous proteins in dentine that was released into gingival crevicular fluid (GCF) during OIIRR. The aim of this research is to compare DSPP detection using the univariate and multivariate analysis in predicting classification level of root resorption. The subjects for this study consisted of 30 patients in 3 group classified as normal, mild, and severe groups of OIIRR. The GCF samples were taken from upper permanent central incisors in the normal and mild group while the upper primary second molars in the severe group. The DSPP qualitative detection limit was determined by analyzing the whole absorption spectrum utilizing multivariate analysis embedded with different preprocessing method. The multivariate analysis represents the multi-wavelength spectrum while univariate analyzes the absorption of a single wavelength. The results showed that the multivariate analysis technique using partial least square-discriminate analysis (PLS-DA) with the preprocess method has successfully improved in classification prediction for the normal and mild group at 0.88 percent accuracy. The multivariate using PLS-DA algorithm with Mean Center preprocess method was able to predict normal and mild tooth resorption classes better than the univariate analysis. The classification parameters have improved in term of the specificity, precision and accuracy. Therefore, the multivariate analysis helps to predict an early detection of tooth resorption complimenting the sensitivity of the univariate analysis. Trial registration NCT05077878 (14/10/2021).