Reliability of a modified Modic classification of bone marrow changes in lumbar spine MRI

Abstract

Objective The purpose of this study was to determine the intra- and interobserver reliability of a modified Modic classification for bone marrow changes seen on lumbar spine magnetic resonance imaging (MRI), taking into consideration mixed signals. Methods Lumbar MRI scans from 94 patients with low back pain were assessed independently by 2 spine specialists (senior [senior1], junior) and a radiologist (senior2). One reviewer (senior1) assessed the MR images twice at a three-week interval for evaluation of intraobserver reliability. Senior2 and junior reviewers assessed the MR images once. Pure edema endplate signal changes were classified as Modic type I, and pure fatty endplate changes as Modic type II. A mixture of types I and II but predominantly edema signal changes was classified as Modic I-2 and a mixture of types I and II but predominantly fatty changes was classified as Modic II-1. Results The intraobserver agreement was excellent (weighted kappa 0.85). The interobserver agreement was moderate to substantial (weighted kappa range 0.56–0.74). Interobserver reliability depended on the experience of the observer, thus highlighting the importance of a learning curve. Conclusion This study shows that the modified Modic classification is reliable and easy to apply for observers with different clinical experience. The inclusion of mixed marrow changes in the modified classification may have clinical and therapeutic implications.