Pre-operative MR Arthrography Underestimates Arthroscopically Identified Acetabular Chondral Lesions

Literature reports on the ability of MR arthrography to diagnose acetabular chondral lesions in the case of femoroacetabular impingement demonstrate varied results and generally report suboptimal sensitivity. The potential failure of MR imaging to predict these lesions has several implications as cartilage repair techniques, which continue to evolve, often require advance preparation. Pre-operative recognition of these lesions allows for proper operative planning and patient counseling. The purpose of this study is to examine the ability of MR arthrography in predicting acetabular chondral lesions and determine MR sequences preferred in their diagnosis. A series of 47 patients that underwent hip arthroscopy for labral tears and/or bony impingement with a preoperative 1.5T MR arthrogram performed in our hospital system were included in this study. Acetabular chondral lesions were prospectively documented and characterized at the time of arthroscopy. Two academic musculoskeletal radiologists, blinded to the surgical findings and previous MR report, independently examined each arthrogram for acetabular chondral lesions and classified them as partial thickness, full thickness, or delamination. Lesion location and diagnostic MR sequence were also documented. Patients averaged 30 ± 10 years of age and 62% were female. Acetabular chondral lesions were documented in 46.8% of patients at the time of arthroscopy. Pre-flap delaminations were documented in 25.5% of patients, flap-delaminations in 6.4%, and full-thickness grade IV lesions in 6.4%. Overall, MR arthrography predicted acetabular chondral lesions with an accuracy of 70.2%, sensitivity of 77.2%, and a specificity of 64% for observer 1 and 72.3%, 86.4%, and 60% for observer 2, respectively. The kappa coefficient for interobserver reliability was .345. Sagittal T2 fat-saturated sequencing was the preferred diagnostic sequence and was available in 25.5% of cases; coronal T2 fat-saturated sequencing was otherwise preferred. Upon subgroup analysis, studies with available sagittal T2 fat-sat sequencing experienced a decline in specificity to 50% and 33% for observer 1 and 2, respectively, while accuracy, sensitivity, and specificity increased for both observers when coronal T2 fat-sat imaging was the necessary diagnostic sequence. The sensitivity of MR arthrography in diagnosing acetabular chondral lesions is higher than previously reported. However, an inaccuracy rate approaching 30% confirms the necessity for superior imaging techniques in the pre-operative diagnosis of these lesions. Though interobserver reliability was fair, this suggests even experienced musculoskeletal radiologists vary in the reporting of these lesions. While sagittal T2 fat-saturated sequences were preferred for interpretation, the availability of this sequence lead to a decline in specificity, which, when relied upon, may result in a higher false-positive interpretation rate.