A comparison of CT arthrography of the wrist to findings during wrist arthroscopy

Purpose: To determine the accuracy of computed tomography after wrist arthrogram (CT arthrogram) compared with findings during wrist arthroscopy as interpreted by a musculoskeletal radiologist and a board-certified hand surgeon. Methods: Seventy-six consecutive patients who had a wrist CT arthrogram and subsequent wrist arthroscopy were identified in the arthroscopic database of four hand surgeons working at a single institution between 2001 and 2006. Operative findings were obtained from the medical record. A single, blinded, musculoskeletal fellowship-trained radiologist prospectively read all CT arthrograms focusing on the following areas: scapholunate interosseous ligament (SLIL), lunotriquetral interosseous ligament (LTIL), peripheral and central triangular fibrocartilage complex (TFCC), and chondral surfaces. Computed tomography arthrograms were also read in a blinded, prospective fashion by a single hand surgeon. The results of both readings were compared with the findings at the time of wrist arthroscopy. Results: Sixteen SLIL tears, 29 LTIL tears, 10 peripheral TFCC tears, and 33 central TFCC tears were identified at the time of wrist arthroscopy. Sensitivity and specificity of the CT arthrogram were calculated using arthroscopy findings as our reference standard. The radiologist's readings (sensitivity, specificity) revealed the following: SLIL (94%, 86%), LTIL (85%, 79%), peripheral TFCC (30%, 94%), and central TFCC (88%, 95%). Results of CT arthrogram readings by the hand surgeon were SLIL (94%, 82%), LTIL (97%, 81%), peripheral TFCC (40%, 97%), and central TFCC (91%, 85%). Conclusions: Computed tomography arthrogram of the wrist is highly accurate at detecting SLIL, LTIL, and central TFCC tears with high sensitivity and specificity. It is not as accurate at identifying peripheral tears.