Carpal tunnel syndrome: are the MR findings a result of population selection bias?

Abstract

Objective: Previous descriptions of MR imaging of carpal tunnel syndrome used limited study populations and volunteers as controls. We reevaluated these descriptions to determine their sensitivity and specificity when applied to a large consecutive clinical series in which the incidence of carpal tunnel syndrome was small.

Subjects and methods: In 196 consecutive wrists for which supplemental axial conventional spin-echo T1-weighted and fast spin-echo T2-weighted images were obtained at 1.5 T with a dedicated wrist coil, 165 studies were available for review. Previously described signs of carpal tunnel syndrome such as proximally increased size, flattening of the median nerve, increased median nerve signal intensity, flexor tenosynovitis, retinacular bowing, decreased deep tendon fat, and deep palmar bursitis were retrospectively and independently evaluated by two observers who were unaware of patient diagnosis.

Results: None of the previously described signs was sensitive for the diagnosis of carpal tunnel syndrome. However, specificity was high for retinacular bowing (94%), median nerve flattening (97%), and deep palmar bursitis (95%).

Conclusion: Most previously described MR imaging signs of carpal tunnel syndrome are insensitive and nonspecific. Exceptions include retinacular bowing, median nerve flattening, and deep palmar bursitis, which in our study proved to have specificities greater than or equal to 94%.