Diagnostic Performance and Reproducibility of Odontoid Retroflexion Angle and Posterior Basion-C2 Line in Type II Basilar Invagination

Objective: To assess the diagnostic performance and interobserver reproducibility of the odontoid retroflexion angle (ORA) and posterior basion-C2 (pB-C2) line in adult patients with type II basilar invagination (BI), using the odontoid apex-Chamberlain line distance (OA-CL) as the reference standard. Methods: This retrospective diagnostic accuracy study included adult patients with type II BI who underwent cranial magnetic resonance imaging between January 2015 and September 2021, along with age- and sex-matched control participants. Two blinded observers performed measurements on midline sagittal images. ORA and pB-C2 were evaluated as index tests, and OA-CL was the reference standard. Interobserver agreement was assessed using intraclass correlation coefficients (ICC). Group comparisons were performed using Student's t-test. The diagnostic performance was evaluated using receiver operating characteristic (ROC) analysis. Results: The study included 33 patients with BI and 34 controls. The OA-CL demonstrated excellent interobserver agreement (ICC=0.98),whereas the ORA showed moderate agreement (ICC=0.69), and pB-C2 showed good agreement (ICC=0.82). ORA was significantly lower and pB-C2 was significantly higher in patients than in controls (p<0.001 for both). ROC analysis revealed moderate diagnostic performance for ORA (AUC=0.79) and pB-C2 (AUC=0.77). Conclusions: Although ORA and pB-C2 differed significantly between patients with type II BI and controls, both parameters demonstrated limited reproducibility and moderate diagnostic performance. These measurements should be considered complementary morphometric descriptors rather than primary diagnostic markers of BI.