Digital risk prediction model for transoral surgery in patients with atlantoaxial dislocation and sandwich fusion

In patients presenting with the sandwich fusion, characterized by C1 occipitalization and C2-C3 non-segmentation, leading to stress concentration at the atlantoaxial joint, there is an increased likelihood of atlantoaxial dislocation (AAD). The decision to proceed with transoral surgery is contingent upon the outcomes of intraoperative traction assessment. The complexity of intraoperative decision-making introduces a degree of uncertainty in preoperative planning, complicating both the surgical preparation and doctor-patient communication. This study aims to investigate the factors influencing the necessity of transoral surgery in these patients by analyzing atlantoaxial CT parameters and to develop a predictive model for transoral surgery.

A retrospective analysis was conducted on patients with AAD and associated sandwich deformity who were treated at our institution between 2010 and 2020. Patients were categorized into transoral and non-transoral groups. The sagittal inclination angle (SIA) and coronal inclination angle (CIA) of the C1 lateral masses, along with the lateral mass joint slippage distance of C1 relative to C2 (LD), were measured using three-dimensional reconstructed images from atlantoaxial CT scans. SIAmax was defined as the larger of the two SIA angles (left and right sides), and SIAmin as the smaller; SIAavg was the average of these angles. CIA was measured and defined in a similar manner. The ratio of the longer LD to the length of the C2 lateral mass surface was defined as the lateral displacement ratio (LDR). Statistical comparisons of clinical data between the two groups were performed using t-tests and Mann-Whitney U tests. Univariate and multivariate binary logistic regression models were employed to analyze factors influencing the necessity of transoral surgery, culminating in the development of a predictive model.

A total of 155 patients were included in the analysis, with 42 patients undergoing transoral surgery and 113 patients undergoing only posterior cervical surgery. Significant differences were observed between the groups in SIAmax, SIAmin, SIAavg, CIAmin, CIAavg, and LDR (P < 0.05). Multivariate binary logistic regression including SIAmax, SIAmin, CIAmax, CIAmin, and LDR, identified SIAmin as an independent factor influencing the need for transoral surgery (P < 0.001, r = 1.072). The predictive model's ROC analysis yielded an AUC of 0.801, with a model sensitivity of 73.8% and specificity of 70.8% at the selected cut-off value.

Clinical decision-making in patients with the sandwich deformity is closely related to the morphological characteristics of the atlantoaxial joint, with SIAmin, indicative of the sagittal inclination of the bilateral lateral masses, emerging as an independent predictor for the necessity of transoral surgery. The greater the forward inclination angle, the higher the likelihood of requiring transoral surgery. The developed binary logistic regression model shows promise as a tool for assessing the need for transoral surgery in this patient population.