Risk factors for periodontal ligament area reduction in maxillary central incisors following orthodontic treatment: a retrospective analysis using a deep learning-based CBCT-PDL segmentation network.

Authors

Abstract

To analyze the relevant risk factors causing the reduction of periodontal ligament area (PDLA) of the maxillary central incisor by measuring the PDLA before and after orthodontic treatment using a previously validated cone-beam computed tomography-periodontal ligament (CBCT-PDL) deep learning segmentation network, providing a theoretical basis for orthodontic clinical treatment design. A retrospective analysis was performed on 63 patients who completed comprehensive orthodontic treatment in the Department of Stomatology, Beijing Friendship Hospital, Capital Medical University, between May 2021 and December 2022. Pre- and post-treatment CBCT images and clinical data were extracted from the hospital's dedicated imaging database. Using the CBCT-PDL segmentation network and orthodontic measurement and analysis software, PDLA and root length (RL) of 126 maxillary central incisors, were quantified before and after orthodontic treatment. Absolute and percentage changes in PDLA (ΔPDLA and ΔPDLA%, respectively) and RL (ΔRL and ΔRL%, respectively) were calculated. Statistical analyses were conducted, including Mann-Whitney U tests, Spearman correlation analysis, Generalized Linear Mixed Model and multiple linear regression, to identify the associated predictors of reduced PDLA in maxillary incisors from risk factors including gender, age, U1-NA angle, Angle's classification, sagittal skeletal pattern (ANB angle), vertical skeletal pattern, history of chronic periodontitis, tooth extraction status, and treatment duration. At the end of active orthodontic treatment, ΔPDLA and ΔPDLA% were significantly higher in the following subgroups compared with the reference group: patients with ANB angle > 3° [Maxillary right central incisor (T11) ΔPDLA: 9.49(13.37) mm, Maxillary left central incisor (T21) ΔPDLA: 11.02(14.45) mm, P < 0.05; ΔPDLA%: T11 6.73(8.20) %, T21 7.05(8.66) %, P < 0.05], tooth extraction treatment group [T11 ΔPDLA: 11.83(14.14) mm, T21 ΔPDLA: 11.82(16.67) mm, P < 0.05; ΔPDLA%: T11 7.54 (10.13) %, T21 8.23(10.32) %, P = 0.058], and treatment duration > 2 years [T11 ΔPDLA: 9.34(14.83) mm, T21 ΔPDLA: 11.94(15.56) mm, P = 0.059; ΔPDLA%: T11 7.00 (10.87) %, T21 8.37(8.73) %, P < 0.05]. Additionally, the U1-NA angle > 28.5° subgroup showed marginally significantly higher ΔPDLA [11.46(25.13) mm, P = 0.082], and patients with skeletal Class II malocclusion exhibited a marginally significantly higher ΔPDLA% [T11 7.53 (12.03) %, T21 6.91(9.24) %, P = 0.059] relative to controls. Significant predictors of root resorption included adult age [T11 ΔRL: 0.63(1.30) mm, T21 ΔRL: 0.98(1.33) mm, P < 0.05; T11 ΔRL%: 5.41(11.20) %, T21 ΔRL%: 8.16(10.20) %, P < 0.05], and tooth extraction treatment [T11 ΔRL: 0.73(1.42) mm, T21 ΔRL: 1.12(1.41) mm, P < 0.01; T11 ΔRL%: 6.33(12.03) %, T21 ΔRL%: 8.20(11.84) %, P < 0.05]. Multiple linear regression revealed that ΔRL%, as a post-orthodontic outcome variable, was independently positively associated with ΔPDLA% (β = 0.536, P < 0.001). Among patient-related and treatment-related factors, the pre-treatment U1-NA angle significantly positively affected ΔPDLA% (β = 0.257, P < 0.01). Pre-treatment ∠ANB (P = 0.055) and treatment duration (P = 0.082) demonstrated marginal statistical significance, while age and tooth extraction status exerted no significant effects (all P > 0.05). Subgroup analyses identified adulthood, pre‑treatment ∠ANB > 3°, pre‑treatment U1‑NA angle > 28.5°, tooth extraction, and treatment duration > 2 years as associated risk factors for supporting tissue loss around the teeth. Multiple linear regression further confirmed the independent effects of pre‑treatment U1‑NA angle, ∠ANB, and treatment duration on periodontal ligament area reduction. A comprehensive pre‑treatment evaluation of combined skeletal‑dental features is critical for individualized orthodontic risk stratification and clinical decision‑making.

Tags

Topics