Evaluation of ulnar variance on wrist MR imaging: is it a reliable measure?

Abstract

Objective To determine if ulnar variance can be evaluated by magnetic resonance (MR) imaging and if this measure can be used as a reliable indicator when correlated to the gold standard technique, conventional radiography (CR). Materials and methods From January to July 2018, the MR images of 64 participants, comprising 66 wrists (mean age 34.9 years; 33 females; 31 males), were obtained. Among those, 29 were referred for evaluation of the wrist for different medical reasons and 35 were asymptomatic volunteers from our radiology group. All subjects had a plain radiography of the wrist in a posteroanterior view with a mean interval between images of 1 day. Local ethics committee approved the study and written informed consent was obtained from all patients. Two musculoskeletal radiologists evaluated the images. Correlation coefficients and a linear regression model were used for statistical analyses. Results Intra- and inter-observer analyses were performed for both diagnostic methods with results showing concordance (intra-observer: kappa score: MR 0.915/CR 0.931; p < 0.05; inter-observer: kappa score: MR 0.857/CR 0.931; p < 0.05). The intraclass correlations of MR and CR to evaluate agreement between the radiologists was slightly higher for radiologist #1 (0.771) than for radiologist #2 (0.659). A linear regression model showed good model fit indicating that MR does correlate with the ulnar variance as measured by CR (CR = 0.554 + 0.897 x MR, R-2 = 0.665). Conclusion Although CR is the gold standard method for the evaluation of ulnar variance, our study demonstrated that MR can be used as a reliable qualitative option.