This is a retrospective cohort study and patients who received orthodontic treatment in the Department of Orthodontics, West China Hospital of Stomatology, Sichuan University (Chengdu, China) from March 2014 to January 2020 were filtered manually in a medical record database of the hospital.
The sample size was calculated according to Lee (14), and by setting the level at 0.05 and power at 0.9, at least 24 samples were needed. Designed with before-after contrast, this study needed at least 12 teeth.
The inclusion criteria were: (1) patients with CBCT images taken before orthodontic treatment (T1) and after the treatment (T2) were available, (2) patients who were 18 to 30 years of age at T1, (3) patients with one or two maxillary premolars extracted during orthodontic treatment, and (4) patients with third molar images in relevant tooth extraction sites existed in both pretreatment and posttreatment CBCT images. The exclusion criteria were: (1) patients with craniofacial syndrome and systemic disease, (2) patients with CBCT images taken 2 weeks before the treatment began or 2 weeks after the treatment finished, and (3) patients with insufficient CBCT image quality, that is, any critical landmark was missing.
Orthodontic diagnoses and treatment characteristics of selected patients were recorded according to the medical records combined with pretreatment CBCT images.
All the CBCT images were taken within 2 weeks before and after the orthodontic treatments. The standards of the end of orthodontic treatment were fully closed space and functional occlusal relationship. All the CBCT images were taken using the same CBCT machine (3D Accuitomo, Morita Group, Japan), which was set according to the manufacturers’ recommendations (10*10 cm FOV, 85 kV, 4 mA, and 360° rotation). The voxel size is 125 μm. During image acquisition, patients were seated statically with the Frankfort plane parallel to the ground. The CBCT data were exported in DICOM multifile format and imported into Invivo software (Version5.3.4; Anatomage, Inc., San Jose, CA, USA) with the plugin “3D analysis” for 3D volume rendering and later evaluation.
A space coordinate system was used with four landmarks as follows (Fig. 1). The horizontal plane (xOy) was defined as the plane passing through bilateral orbitales, while parallel to ANS-PNS. The sagittal plane (yOz) was defined as the plane passing through ANS and PNS while perpendicular to the horizontal plane. The frontal plane (xOz) was defined as the plane perpendicular to both horizontal plane and sagittal plane while passing through ANS. Landmark superimposition with the same four landmarks was performed to overlap three-dimensional reconstructed pre- and post- therapy images to evaluate the stability of the space coordinate system (Fig. 2).
Six other landmarks were located to define the spatial position of the third molar: mesiobuccal and distobuccal cusps of the third molar, root furcation and central pit of the third molar, root furcation and central pit of the second molar. Software calculated the linear and angular dimensions between certain landmarks as follows. The forward, outward, and upward position were defined as positive directions (Fig. 3).
The distances from the mesiobuccal and distobuccal cusp of the third molar to the horizontal (xOy), frontal (xOz), and sagittal planes (yOz) were measured (Fig. 4). The changes in vertical, transverse, and sagittal positions were calculated using the distance differences before and after orthodontic treatment.
The angles between the long axes of second and third molar (root furcation–central pit), and the angles between the long axes of third molar and the horizontal, frontal, and sagittal planes were measured (Fig. 5). Changes of the angulation of third molars in all directions before and after orthodontic treatment were calculated.
The angles between the mesiobuccal-distobuccal cusp (crown axis) and the horizontal, frontal, and sagittal planes were measured (Fig. 6). Changes of the angulation before and after orthodontic treatment were measured to evaluate the rotation of third molars.
All subjects were measured independently by two operators (Y. Z. and P. F.) and both two operators were in good command of orthodontic theory and well trained in pre-experiment. No more than two patients were analyzed per day by each operator, and the whole process of evaluation was finished within 3 weeks. An experienced orthodontist (L. W.) strictly guided the study. The intraclass correlation coefficient (ICC) values ranged from 0.78 to 0.96 for all the angular variables, and from 0.96 to 0.97 for all the positional values, showing the measurement’s reliability.
Paired t-tests were performed to evaluate the changes between pre- and after-treatment CBCT images with SPSS software (Version 21.0; SPSS, Chicago, IL, USA). Multiple regression analysis was later processed to control heterogeneity brought by retrospective design with R software (Version 4.0.0; R, Las Vegas, USA). Statistical significance was set at 0.05.