Objective: To assess the utility of MR in detecting surgically induced Stener lesions (displaced thumb ulnar collateral ligaments) in cadaveric models.
Design: Six cadaver thumbs had ulnar collateral ligament (UCL) tears created surgically. MR examinations (2D STIR and 3D GRASS) were performed identically on all specimens both before displacement (non-Stener) and after displacement (Stener lesion) of the UCL. The MR images were then randomly numbered. Each image was evaluated separately in blinded fashion by four musculoskeletal radiologists for the presence or absence of a Stener lesion. Each radiologist reinterpreted the images after an interval of several days. The interpretation was based on previously published criteria for Stener lesion diagnosis by MR.
Results: The sensitivity of GRASS ranged from 0.17 to 0.67 with the most experienced reader scoring the lowest. The specificity of GRASS ranged from 0.33 to 1.0 (most experienced reader 0.67, 0.83). STIR had a sensitivity of 0.00-0.17 and a specificity of 0.53-0.83. The kappa values for inter- and intraobserver agreement were measured. The intraobserver kappa for GRASS was 0.27-0.75 (most experienced reader 0.75).
Conclusions: 2D imaging is probably inadequate for the evaluation of Stener lesions. The most likely reason is that the STIR slice thickness of 3 mm limits resolution of small UCLs. The poor sensitivity and specificity of GRASS as well as poor interobserver agreement suggest that MR may not be sufficiently accurate for Stener lesion evaluation.