MRI Characterization of Vascular Spinal Tumors

Abstract

Randomized trial. This study described practical magnetic resonance imaging (MRI) characteristics to assist in the identification of these tumors including the corresponding statistics. Identification of vascular spinal tumors using MRI allows the opportunity for angiography and preoperative embolization (PE) to improve outcomes and decrease hemorrhage. The MRI characterization reliability of these tumors has not been described. A retrospective comparison of 40 patients, at a single institution, with known spinal tumors was conducted with 20 vascular versus 20 nonvascular cases (based on postoperative pathology). Randomized MRI T1 precontrast/postcontrast and T2 images of these tumors were blinded and reviewed by 7 spine surgeons and 5 musculoskeletal radiologists. Four criteria were reviewed: T2 hyperintensity, contrast enhancement, flow voids, and feeding vessels. The clinical relevance was evaluated by asking if the reviewer recommended PE. The specificity, sensitivity, and accuracy of each characteristic for surgeons were: T2 hyperintensity (65%, 85%, 75%), contrast enhancement (84%, 38%, 61%), flow voids (42%, 86%, 64%), and feeding vessels (35%, 90%, 63%). The results for the radiologists were: T2 hyperintensity (83%, 61%, 72%), contrast enhancement (87%, 32%, 60%), flow voids (44%, 93%, 69%), and feeding vessels (33%, 93%, 63%). Both the groups had low κ and intraclass correlation values. Review of angiography/PE recommendation showed that both surgeons and radiologists had a false-negative rate of 33%. Surgeons and radiologists have similar moderate accuracy of MRI vascularity identification. Radiologists have a higher sensitivity, but lower specificity. Even with simplified, straightforward criteria, the MRI characterization of vascular spinal tumors has low interobserver/intraobserver reliability with a false-negative rate for angiography/PE recommendation of 33%. Angiography may become more standard in the workup of spine tumors.