Background Radiographic assessment of new bone formation in the anterior vertebral corners of the cervical and lumbar spine is the current gold standard for detection of damage and disease progression in ankylosing spondylitis (AS). However, sensitivity to change is limited and radiography cannot assess the thoracic spine reliably. Standardization of terminology for MRI features indicative of new bone formation has been limited and it is unclear whether MRI might offer any advantages over radiography due to its visualization of the thoracic spine.
Objectives To develop standardized definitions for features of new bone formation on MRI, to test the reliability of their detection in patients with AS, to compare this reliability with radiography, and to determine whether availability of radiographs enhances the reliability of detection on MRI.
Methods We generated consensus definitions for bone spurs and ankylosis observed on sagittal images of T1-weighted MRI scans that included lesions at anterior and posterior vertebral corners as well as non-corner lesions in the disc space and lesions in the lateral segments of the thoracic and lumbar spine. A reference image set was generated that included examples of all these lesions as well as variations in normal anatomy. In the MRI-based score, bone spurs and ankylosis are assigned a score of 2 and 3, respectively. The first reading exercise assessed reliability for status and change scores for lesions detected on baseline and 2 year scans in 55 patients with AS by 3 readers scoring in known time sequence. Discrepant scans were reviewed extensively using radiography as a reference. The second exercise was conducted as follows by the same expert readers on 25 AS patients with baseline/2 year pairs of radiographs and MRI scans which were numbered independently from radiographs: 1. Assessment of radiographs alone for syndesmophytes and ankylosis. 2. Assessment of MRI scans alone for new bone. 3. Assessment of radiographs and MRI scans simultaneously. Reliability was assessed by intra-class correlation coefficient (ICC).
Results ICC for 3 readers reading MRI scans in the first exercise were 0.79 and 0.23 for baseline status and 2 year change scores, respectively. In the second exercise, radiography was superior to MRI in reliably detecting new bone (Table 1). Simultaneous availability of radiographs enhanced the reliability of detecting new bone in the C spine by MRI but this was still inferior to radiography. ICC for detection of new bone in the thoracic spine by MRI was 0.48 and 0.36 for baseline status and 2-year change scores, respectively.
Conclusions Extensive standardization of MRI features, scoring methodology, and calibration of expert readers with radiography failed to show any major advantage of MRI over radiography in the reliable detection of new bone in the cervical and lumbar spine of patients with AS. Future efforts should focus on the methodology for assessment of the thoracic spine.
Disclosure of Interest None Declared
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.