Abstract
BACKGROUND: Compression ultrasonography (CUS) is the first-line imaging test for diagnosing upper extremity deep vein thrombosis (UEDVT), but often yields inconclusive test results. Contrast venography is still considered the diagnostic standard but is an invasive technique. OBJECTIVES: We aimed to determine the diagnostic accuracy of magnetic resonance noncontrast thrombus imaging (MR-NCTI) for the diagnosis of UEDVT. METHODS: In this international multicenter diagnostic study, we prospectively included patients with clinically suspected UEDVT who were managed according to a diagnostic algorithm that included a clinical decision rule (CDR), D-dimer test, and diagnostic imaging. UEDVT was confirmed by CUS or (computed tomography [CT]) venography. UEDVT was excluded by (1) an unlikely CDR and normal D-dimer, (2) a normal serial CUS or (3) a normal (CT) venography. Within 48 h after the final diagnosis was established, patients underwent MR-NCTI. MR-NCTI images were assessed post hoc by two independent radiologists unaware of the presence or absence of UEDVT. The sensitivity, specificity, and interobserver agreement of MR-NCTI for UEDVT were determined. RESULTS: Magnetic resonance noncontrast thrombus imaging demonstrated UEDVT in 28 of 30 patients with UEDVT and was normal in all 30 patients where UEDVT was ruled out, yielding a sensitivity of 93% (95% CI 78-99) and specificity of 100% (95% CI 88-100). The interobserver agreement of MR-NCTI had a kappa value of 0.83 (95% CI 0.69-0.97). CONCLUSIONS: Magnetic resonance noncontrast thrombus imaging is an accurate and reproducible method for diagnosing UEDVT. Clinical outcome studies should determine whether MR-NCTI can replace venography as the second-line imaging test in case of inconclusive CUS.