Differences in Topographical Location of Sacroiliac Joint MRI Lesions in Patients With Early Axial Spondyloarthritis and Mechanical Back Pain

Rosa Marie Kiil; Clara E. Mistegaard; Anne Gitte Loft; Anna Zejden; Oliver Hendricks; Anne Grethe Jurik


Arthritis Res Ther. 2022;24(75) 

In This Article


Study Sample

The study sample is described in detail elsewhere.[21] In brief, 84 patients were included. They were referred from a cohort of 1020 patients with low back pain for 2–12 months not responding adequately to conservative treatment.[22] They were characterized by either positive MRI according to the 2009 ASAS MRI criteria, fulfilling at least the minimum requirement,[4] or a positive HLA-B27 and ≥ 1 clinical spondyloarthritis (SpA) feature according to the ASAS classification criteria.[3] These inclusion criteria were chosen to ensure that all potential axSpA patients in the basic cohort were included in the study sample. The diagnostic progress involved a retrospective evaluation by multidisciplinary team conferences including both expert rheumatologists and radiologists regarding an axSpA diagnosis after a mean follow-up period of 3.5 years. One of rheumatologists (AGL) had personally assessed 50 (60%) of the 84 included patients clinically at baseline. At the MDT, baseline and mean 3.5-year follow-up MRI of the SIJs and spine as well as clinical and biochemical data were available in all patients. After presentation and discussion on each case, the MDT conference resulted in a definite consensus diagnosis of whether the patient had axSpA or not; 25 (30%) patients were diagnosed with axSpA, whereas the remaining 59 (70%) constitute the group of patients with MBP (Table 1).

MRI Procedure

The same MRI scanner and protocols were used at baseline and follow-up. The scan protocols have been published previously.[23] MRI of the spine and SIJs was performed on a 1.5-Tesla MR scanner (Phillips Achiva). Identical sequences were obtained at baseline and follow-up: SIJ—semi-coronal T1-weighted and T1-fat saturated and semi-axial T2-weighted short tau inversion recovery (STIR) sequences; spine—sagittal T1 and STIR of the entire spine with supplementary 3D T2 Vista sequence of the lumbar spine and axial T2-weighted slices at the three lowest intervertebral spaces. Only the 84 baseline scans were assessed in this study. Two radiologists (one senior consultant specialized in musculoskeletal imaging (AGJ) and one junior resident radiologist (RMK) with 4 years of training in musculoskeletal imaging focusing on SIJ diagnostics) independently performed the granular evaluations of all MR examinations of the SIJ and the lumbar spine blinded to all clinical information. In addition, global assessments of the SIJ MRIs were made independently by two experienced senior musculoskeletal radiologists (AGJ and AZ) blinded to all clinical information regarding the presence of axSpA changes on − 5 (definitely not) to + 5 (definitely yes) confidence scale of the axSpA probability.

MRI Reading

The following SIJ MRI findings were assessed according to the 2019 ASAS lesion definitions:[24] subchondral BME (including depth > 1 cm and intensity), fatty marrow deposition (FMD) (including depth > 1 cm), sclerosis and erosions, along with anatomical SIJ variations (including the presence of BME or FMD/sclerosis in relation to the variations). In the lumbar spine, disc degeneration was evaluated on a 0–3 scale: 0, no signs of disc degeneration or herniation; 1, loss of water content and/or disc height; 2, disc protrusion; and 3, disc extrusion, as defined by Fardon.[25] Besides, the presence of BME and/or FMD was registered divided into vertebral corner and endplate changes, respectively.

The SIJ analyses were based on the simultaneous evaluation of the semi-axial and semi-coronal slices. The iliac and sacral joint facets were divided into three portions based on the semi-axial slices: upper, middle and lower joint portion, regarding the location of BME, FMD, sclerosis and erosion. Furthermore, the location of BME was assessed in relation to the anterior and posterior half of the joint facets.

The joint divisions were determined on semi-axial slices assisted by scout lines on concomitant semi-coronal T1 slices to determine anatomical location. The upper- and lowermost semi-axial slices were determined by the first and last slice with clearly visible SIJ cartilaginous compartment, respectively. Furthermore, a tiny bit of subchondral trabecular bone should be visible at both the sacral and iliac joint facet and extend for ≥ 1 cm on ≥ 1 side. The number of semi-axial slices covering the cartilaginous compartment was divided into three, defining the upper, middle and lower joint portions. In case of an unequal number of slices when dividing by three, one extra slice was added to the lower and eventual middle portions.

For BME, the occurrence and the number of slices per region with changes were registered, thereby obtaining an absolute sumscore. Due to the differences in joint size, e.g. due to gender,[26] relative sumscores were used, calculated by dividing the absolute score by the number of slices in the joint portion and for convenience, multiplied by 10.

The presence of FMD, sclerosis and erosions was scored dichotomously as present or absent in each of the region(s). The occurrence of depth > 1 cm of BME and FMD and the intensity of BME (same or almost the same signal intensity as the spinal fluid) were added in each region.

The extent and location of BME were evaluated on semi-axial STIR supported by the semi-coronal T1, whereas the location of FMD, sclerosis and erosions was primarily evaluated on coronal T1 and T1-fat saturated sequences assisted by scout lines on concomitant semi-axial STIR.

BME lesions were registered in both the cartilaginous and the ligamentous compartment, whereas FMD, sclerosis and erosions were only registered in the cartilaginous compartment.

The following atypical SIJ morphologies were reported: accessory SIJ, iliosacral complex, bipartite iliac bony plate and dysmorphic cartilaginous facets. The definitions have been described in detail in a recent paper.[19] Furthermore, the presence of lumbosacral transitional anomaly/anomalies, defined as a vertebra with transverse process (es) articulating with the superior border of the sacrum being either a lumbalization of S1 or a sacralization of L5,[27] was also noted. In addition, BME and structural changes (FMD and/or sclerosis) in relation to the variations were assessed.

For the most prevalent lesions, BME and FMD, the results of the scorings were based on concordant reads, whereas the results regarding the remaining lesions were based on consensus between the readers. BME sumscores were reported as the mean of scores if the readers agreed on the scorings of > 0 in the region. The mean axSpA MRI confidence score was reported.

Statistical Analyses

The interreader reliability on continuous MRI variables was tested by intraclass correlation (ICC) based on a two-way, random effects, single-measure model with the absolute agreements presented, whereas the interreader reliability on binary MRI variables was tested using the kappa test[28]

Demographic and clinical continuous variables were presented as means with standard deviation (SD). Continuous MRI data was presented as mean. The remaining variables were binary and presented as proportions.

The differences between the group prevalence rates were tested with the two-way proportion test for binary variables, whereas the Mann-Whitney rank sum test was used in continuous variables.

Univariate logistic, and when appropriate, linear regressions were performed to investigate whether scoring > 0 uni- or bilaterally for BME, FMD, sclerosis and erosions along with BME scoring > 0 in the ligamentous compartments and the presence of BME and FMD depth were independently associated with an axSpA diagnosis, anatomical variations, lumbar disc extrusion, gender, number of childbirths, age and BMI and were reported as odds ratios (OR) or regression coefficients.