Background/Purpose: The relevance of structural lesions in non-radiographic axial SpA (nr-axSpA) is unclear, particularly without signs of MRI inflammation. In a post hoc analysis we evaluated the association between structural lesions in the sacroiliac joints (SIJ) and spinal inflammatory lesions on MRI in early nr-axSpA. We hypothesized that structural lesions indicate a more extensive disease phenotype that includes early spinal involvement visible on MRI.

Methods: The EMBARK study (ClinicalTrials.gov: NCT01258738) enrolled patients 18-49 yrs old with axial SpA per ASAS imaging or clinical criteria without meeting modified New York radiographic criteria. Patients had symptoms for >3 months and <5 yrs, BASDAI score ≥4, and had failed ≥2 NSAIDs. Bone marrow edema (BME) in the SIJ and spine at baseline (BL) was assessed on short tau inversion recovery (STIR) scans by 2 independent readers using Spondyloarthritis Research Consortium of Canada (SPARCC) SIJ and 23-DVU scores, respectively. BL structural lesions were evaluated on T1 weighted spin echo scans blinded to STIR scans, using the SPARCC MRI SIJ structural score (SSS). Univariate analysis evaluated the relationship between BL spinal inflammation and these BL characteristics: gender, presence/absence of any structural MRI lesions in the SIJ (SSS>0 or =0), presence/absence of specific MRI SIJ structural lesions, and SPARCC SIJ ≥2 or <2. Multivariate stepwise regression analysis evaluated the relationship between spinal inflammation and MRI SIJ lesions after including age, gender, and symptom duration.

Results: BL MRI scans were available for 185 patients. Mean (standard deviation [SD]) age was 32.0 (7.8) yrs, 60.5% were male, mean (SD) symptom duration was 2.4 (1.8) yrs, 133/182 (71.9%) patients were human leukocyte antigen B27+ and 152 (82.2%) met ASAS MRI imaging criteria. At BL, mean (SD) SPARCC MRI 23-DVU spinal score was 4.0 (8.0); 128/183 (69.9%) patients had SPARCC SIJ BME scores ≥2 and 55/183 (30.1%) had scores <2. A total of 77/185 (41.6%) patients had ≥1 structural lesion on MRI comprised of erosion (65/185, 35.1%), backfill (26/185, 14.1%), fat metaplasia (15/185, 8.1%) and ankylosis (4/185, 2.2%). Higher spine 23-DVU scores were observed in males, in the presence of definite SIJ inflammation (SPARCC SIJ ≥2, p=0.01), and in the presence of any one of the SIJ structural lesions, Table. Multivariate analysis indicated that erosion and backfill are independent factors associated with spinal inflammation; parameter estimates (SE): erosion: 2.9 (1.3), p=0.03; backfill: 3.9 (1.8), p=0.03.

Conclusion: MRI structural lesions in the SIJ occur in a substantial proportion of patients with nr-axSpA and their presence suggests a more extensive phenotype of disease associated with early spinal involvement.