OP0250\u2005MRI VERTEBRAL CORNER INFLAMMATION AND FAT DEPOSITION ARE ASSOCIATED WITH WHOLE SPINE LOW DOSE CT DETECTED SYNDESMOPHYTES: A MULTILEVEL ANALYSIS

Abstract

A few studies have shown an association between vertebral corner inflammation (VCI) and vertebral corner fat deposition (VCFD) on MRI and syndesmophyte formation on cervical and lumbar conventional radiography.To investigate whether magnetic resonance imaging (MRI) patterns of VCI, VCFD and a combination of both are associated with the development of new or grown syndesmophytes as detected by whole spine low dose computed tomography (ldCT), thereby studying these associations also in the thoracic spine.Patients in the Sensitive Imaging in Ankylosing Spondylitis cohort underwent MRI at baseline, 1 year and 2 years, and ldCT at baseline and 2 years. MRI lesions were scored by 3 central readers, using the SPARCC method for VCI and the CanDen method for VCFD, and coded as absent or present per timepoint and per reader. MRI patterns over time (Table) were based on patterns studied by Machado et al.1 and deemed present if seen by ≥2 out of 3 readers. The patterns reflect hypothetical associations between presence and absence of VCI and VCFD, independently and combined, on ldCT detected new or grown syndesmophytes. Individual reader change scores were used for ldCT images, scored by 2 central readers with the Computed Tomography Syndesmophyte Score. New (CTSS 0 to 1, 2 or 3) and grown (CTSS 1 to 2 or 3; 2 to 3) syndesmophytes were grouped together to represent bone formation. Corners not at risk for the outcome due to presence of a bridged syndesmophyte at baseline were excluded. Multilevel generalized estimated equations were used, with separate models per MRI pattern, accounting for correlations within patients and between ldCT readers.Table 1.Effect of vertebral corner inflammation and vertebral corner fat deposition on syndesmophyte formationPatterns of lesions over time on MRICorners with VCI/VCFD patternN(%)OR (95% CI)1. VCI at any TP, irrespective of VCFD691 (15.0%)2.37 (1.49-3.78)2. VCFD at any TP, irrespective of VCI1080 (23.5%)2.58 (1.97-3.39)3. VCI on ≥1 TP and absence of VCFD on all TPs372 (8.1%)1.90 (1.15-3.13)4. VCFD on ≥1 TP and absence of VCI on all TPs754 (16.4%)1.87 (1.41-2.48)5. VCI precedes VCFD43 (0.9%)2.20 (0.83-5.86)6. VCI precedes or coincides with VCFD. VCFD does not precede VCI198 (4.3%)2.33 (1.47-3.69)7. Absence of VCI and VCFD on all TPs3108 (67.6%)0.35 (0.25-0.49)VCI, vertebral corner inflammation; VCFD, vertebral corner fat deposition; TP, timepoint.50 patients were included, contributing a total of 4600 vertebral corners. Their mean age was 49.3 years (SD 9.8), 86% were male and 78% were HLA-B27+. Presence of VCI and VCFD patterns ranged from 43 (0.9%) to 3108 (67.6%) corners (Table), with the lowest frequency being for VCI preceding VCFD. Protection against syndesmophyte development was seen in case of absence of both VCI and VCFD (OR 0.35) and positive associations with ORs ranging from 1.87-2.58 were observed for various VCI/VCFD patterns. Nevertheless, out of all corners with a new or grown syndesmophyte, 47.3% of corners according to reader 1 and 43.9% according to reader 2 had neither VCI nor VCFD preceding the bone formation.Presence of VCI or VCFD and combinations of the two, measured yearly on MRI, increased odds of bone formation 2 years later, whereas absence of both VCI and VCFD decreased the odds, showing that VCI and VCFD have some role in the development of syndesmophytes. However, almost half of all bone formation occurred in corners without VCI or VCFD, suggesting the presence of these lesions in yearly MRIs does not fully explain the development of syndesmophytes. This study confirmed that there is an association between VCI and VCFD and bone formation also for the thoracic spine and on ldCT compared to conventional radiography.[1]Machado et al ARD 2016Rosalinde Stal: None declared, Xenofon Baraliakos: None declared, Alexandre Sepriano: None declared, Floris A. van Gaalen Grant/research support from: Novartis, Sofia Ramiro: None declared, Rosaline van den Berg: None declared, Monique Reijnierse: None declared, Juergen Braun: None declared, Robert B.M. Landewé: None declared, Désirée van der Heijde: None declared