Evaluation of the influence of enamel thickness on the effectiveness of tooth whitening: an in vitro study

Abstract

Introduction. While tooth whitening is widely used to manage dental discoloration, results can vary substantially across teeth and protocols. In this in vitro study, we assessed whether enamel thickness across tooth regions influences whitening efficacy under different peroxide‑based protocols.

Methods. Ninety extracted human premolars were stained and allocated to three protocols (n=30 each): 10% (10 h), 20% (4 h), and 35% (1 h). Enamel thickness was measured on CBCT at the cervical, middle, and occlusal thirds. Color was recorded with a spectrophotometer using one acquisition before whitening and one after whitening; three regional L*a*b* values were extracted from each image. Color change was expressed as CIEDE2000 (ΔE00). Because three regions were analyzed per tooth, the primary analysis used a linear mixed‑effects model (LMM) with a random intercept for tooth (ΔE00 ~ Group×Region + (1|Tooth)). Enamel thickness was added as a covariate in a secondary LMM. Robustness was assessed by GEE with robust standard errors and a tooth‑level sensitivity analysis (mean ΔE00 per tooth).

Results. ΔE00 differed significantly across protocols in the LMM: compared with 35%, ΔE00 was higher for 20% (p=0.003) and 10% (p<0.001). Region was not significant (p>0.05) and the interaction was not supported overall; a localized 10%×occlusal effect was borderline in LMM (p≈0.055) and significant in GEE (p≈0.024). Baseline color differed between groups and was therefore included a priori as a covariate; adjustment for baseline L* did not alter the main conclusions. After adjustment, enamel thickness was not associated with ΔE00 (p≈0.933). Exposure‑dose (concentration × time, % h) showed a positive trend with ΔE00 (r=0.401, p<0.001; R²=0.161). Only 0.37% of measurements showed ΔE00 ≤ 1.8.

Conclusion. Whitening efficacy was primarily governed by the protocol (cumulative exposure) rather than enamel thickness. Regional modulation was modest, and enamel thickness alone was not predictive.