RESUMEN
PURPOSE: To investigate a predisposition to mandibular angle fractures, a retrospective study was performed in which fractured mandibles were compared with healthy mandibles with no history of fracture. Other investigations of angle fracture risk have exclusively studied patients with existing fractures. In addition, the risk has not been comprehensively explained in conjunction with the specific features of mandibular anatomy. We sought to characterize any anatomic variations between the jaws that had fractured and those that had never fractured. MATERIALS AND METHODS: Healthy mandibles with no history of fracture were physically measured at the William M. Bass Skeletal Collection at the University of Tennessee, Knoxville and compared with fractured mandibles from computed tomography (CT) scans at the Dartmouth Hitchcock Medical Center. A total of 52 healthy mandibles and 44 CT scans were evaluated. MATLAB machine learning algorithms (MathWorks, Natick, MA) were used to compare the study populations and isolate those anatomic features that differed between healthy and fractured mandibles. RESULTS: Machine learning classifiers were able to differentiate between male and female jaws, with the condylion-gnathion distance the most distinguishing feature. The 6 most common anatomic features that differed between healthy and fractured mandibles were the 1) retromolar space, 2) perimeter of the cross-section just proximal to the second molar, 3) breadth of the ramal cross-section, 4) thickness of the oblique ridge, 5) transgonial angle, and 6) location of the ipsilateral mental foramen. The presence of third molars was also related to fracture risk, with third molars present in 72.7% of the fractured mandibles versus 26.9% of unfractured mandibles. Of the fractured mandibles with third molars present, 87.5% had the fracture running directly through the tooth or its socket. CONCLUSIONS: The results from the present study have provided evidence that anatomic differences exist between mandibles that sustain angle fractures and those that do not. Although much of the morphology was found to be interdependent, the fracture risk could be accurately predicted using 6 anatomic features. Understanding these mandibular variations and identifying patients vulnerable to mandibular fracture could provide clinicians with additional objective information. Furthermore, using the methods demonstrated in our study, future research could focus on developing an algorithm that includes these unique anatomic features in the hope of assisting surgeons in providing tailored treatment for mandibular angle fractures according to patient-specific morphology.
