Association of intra-operative findings with MRI observations in patients with peroneal tendon pathology: a blinded retrospective case-control observational study.

Objective: Assess the accuracy of MRI for peroneal tendon tears, evaluate inter/intra-rater reliability among musculoskeletal radiologists, and identify MRI findings that may be associated with tendon tears.

Materials and methods: Preoperative MRIs of patients who underwent surgical debridement of peroneal tendon tears were paired with MRIs of a matched control group with medial ankle pain. Randomized MRIs were reviewed by three musculoskeletal radiologists, recording the presence of structural abnormalities of the peroneal tendons and associated morphological abnormalities. Consensus MRI findings were analyzed for associations with surgical diagnoses. Weighted Cohen's and Fleiss' kappa coefficients were used to describe inter/intra-rater agreement.

Results: For peroneus brevis tear, individual sensitivities ranged from 58 to 87%, specificities ranged from 78 to 97%, and accuracies ranged from 79 to 87%, while consensus showed 84% sensitivity, 92% specificity, and 88% accuracy. For peroneus longus tear, individual sensitivities ranged from 14 to 57%, specificities ranged from 76 to 94%, and accuracies ranged from 72 to 79%, while consensus showed 36% sensitivity, 91% specificity, and 79% accuracy. Associations were detected between peroneal tendon tears and numerous secondary MR findings. Tearing of the superior peroneal retinaculum was most diagnostically powerful, present in 55.9% of cases and in none of the controls.

Conclusion: MRI is a valuable tool for assessing for peroneus brevis tears but may be less sensitive for detecting peroneus longus tears. Diagnostic accuracy is similar between radiologists, but there is variability in sensitivity/specificity. No individual radiologist achieved a higher accuracy than that obtained through consensus. This may imply value in additional opinions from multiple radiologists when there is diagnostic uncertainty.