Objective: The biocompatibility of adhesive systems is essential for the long-term success of restorative dental procedures due to their close proximity to dentin and pulpal tissues. This study aimed to evaluate the cytotoxic effects of adhesive systems with different pH values on L929 mouse fibroblast cells under in vitro conditions. Materials and Methods: Four commercially available adhesive systems with different pH values—All-Bond Universal, G-Premio Bond, Tokuyama Bond Force II, and Clearfil Universal Bond Quick—were evaluated. Cytotoxicity was assessed using the MTT assay at four different concentrations (0.1%, 0.01%, 0.001%, and 0.0001%) and three incubation periods (24, 48, and 72 h). Cell viability data were analyzed using two-way analysis of variance followed by Bonferroni post hoc tests. Cytotoxicity was interpreted according to ISO 10993-5 criteria. Results: All adhesive systems exhibited concentration-dependent cytotoxicity, with significant reductions in cell viability observed only at the highest concentration (0.1%). At lower concentrations, no cytotoxic effects were detected. Despite having the highest pH value, All-Bond Universal consistently demonstrated the lowest cell viability. In contrast, Tokuyama Bond Force II showed the most favorable cytocompatibility profile, with relatively higher cell viability values over time. Morphological analysis supported the quantitative findings, revealing pronounced cellular alterations at high concentrations and preserved fibroblastic morphology at lower concentrations. Conclusions: adhesive systems demonstrate cytotoxic effects in a concentration-dependent manner, and pH alone is insufficient to predict their biocompatibility. Monomer composition and formulation characteristics appear to play a more critical role in determining cytotoxic behavior. These findings emphasize the importance of appropriate adhesive handling and isolation techniques to minimize tissue exposure and enhance clinical safety.

The aim of this study was to compare the responses given by the large language models (LLMs) ChatGPT-5, Gemini 2.5 Pro, DeepSeek-V3.2, and Claude Sonnet-4.