¹Department of Orthopaedic Surgery, Inselspital, University of Bern, Bern, Schweiz
²University Institute of Diagnostic, Interventional and Paediatric Radiology, Inselspital, University of Bern, Bern, Schweiz
³Radiology Department, Balgrist University Hospital, Zürich, Schweiz

Text

Objective and research question: Coxa profunda sign has traditionally been used as a radiographic parameter of increased lateral acetabular coverage. However, it is also observed in hip dysplasia, suggesting that it does not reflect acetabular coverage. Artificial intelligence (AI)-powered 3D MRI segmentation now enables precise evaluation of lunate surface size.The objective of this study is to evaluate whether a positive coxa profunda sign reflects increased lunate surface size using an AI-based automatic 3D analysis.

Material and methods: This retrospective study included 98 patients (100 hips) with symptomatic hip deformities who underwent 3T MR arthrography between January 2020 and October 2021 (mean age 30 ± 9 years; 64% female). The imaging protocol included proton-density-weighted turbo spin-echo (PD-w TSE) and axial-oblique 3D T1 MP2RAGE sequences. Anteroposterior pelvic radiographs were assessed for the coxa profunda sign and standard parameters. Patients were stratified into acetabular morphology groups based on the lateral center-edge angle (LCE): dysplasia (LCE < 23°), normal coverage (LCE 23–33°), overcoverage (LCE 33–39°), severe overcoverage (LCE > 39°), and retroversion (retroversion index > 30%). AI-based segmentation evaluated lunate surface size, normalized to femoral head size, with statistical analyses using Fisher’s exact test for prevalence and unpaired t-tests for morphological relationships.

Results: Positive coxa profunda was found in 65% of cases (dysplasia: 68%, normal: 63%, retroversion: 45%, overcoverage: 83%, severe overcoverage: 75%), with no statistically significant difference between groups (p = 0.156). By contrast, the relative size of the lunate surface differed significantly across groups, with dysplastic hips showing the smallest area (28 ± 3%) and severe overcoverage the largest (35 ± 3%; p < 0.001). Lunate surface size (31 ± 3% vs. 31 ± 3%; p = 0.462) and lateral coverage parameters (LCE: 30 ± 10° vs. 29 ± 9°; p = 0.633, acetabular index: 3 ± 6° vs. 4 ± 6°; p = 0.504) showed no difference between coxa profunda-positive and -negative individuals. However, those with positive coxa profundahad significantly higher acetabular version (18 ± 6° vs. 15 ± 6°; p = 0.026) and lower retroversion index (11 ± 16% vs. 21 ± 20%; p = 0.006).

Discussion and coclusion: The coxa profunda sign is highly prevalent and does not reflect the size of the lunate surface. Therefore, it should not be relied upon for diagnosing acetabular coverage. However, it might be an indicator of increased acetabular version.