Psoriatic Arthritis (PsA) is an inflammatory joint disease that develops in patients with psoriasis and manifests by highly heterogeneous clinical symptoms (1). Inflammation in the sacroiliac (SI) joints and spine are common. Magnetic resonance (MR) imaging allows direct visualization of inflammation and is the most sensitive imaging modality for assessment of structures critical in evolution of inflammatory disease. Image scoring according to the spondyloarthritis research consortium of Canada (SPARCC) (2) gives a semi-quantitative measure of the manifestation of the disease in SI joints and spine, which is reliable and sensitive to change, but also labor intensive. This study aims to compare a pixel-based morphometric method for accurate quantification of bone marrow inflammatory edema to SPARCC assessment in MR images of SI joints of PsA patients with low or very low inflammatory activity.

Patients diagnosed with PsA were included in the study (N=41). Clinical evaluation, patient global assessment (PGA), disease activity score of 28 joints (DAS28), Bath ankylosing spondylitis disease activity index (BASDAI), quality of life questionnaire, and high-sensitivity C-reactive protein (HS-CRP) provided patient health status. All patients underwent MRI examinations of the sacroiliac joints and the spine based on standardized protocols (2, 3). Examinations were performed on two 1.5T scanners (Avanto, Siemens Healthcare, Germany). A T2-weighted sequence with fat suppression (STIR; time to recovery [TR] 3700 msec, time to echo [TE] 52 msec, time to inversion 145 msec) and a T1-weighted turbo spin-echo sequence (TR 423 msec, TE 11 msec) were used for the examination of the SI joints. Data from the SI joints of 12 patients (ages 24-63 years) have been analyzed so far. A rheumathologist scored the T2-weighted STIR images, according to the SPARCC criteria (2).

Pixel-based morphometric image analysis included manual outlining of the sacrum and the iliac bones, excluding vascular and neural structures (3D slicer, MIT Artificial Intelligence Lab). Signal intensity was normalized to signals from muscle in the same slice (MATLAB, Mathworks). For each set of images, a circular region of interest (ROI) was placed in the first sacral vertebra in one slice. A threshold was defined as $$$μ+2σ$$$, where μ is the mean and σ the standard deviation of signal intensity in that ROI. All voxels with higher signal intensity than the threshold, consistent with inflammation (4), were masked (Figure 1). The volume of subchondral sacroiliac STIR hyperintensity (volume_hyper) and the average signal intensity of subchondral sacroiliac STIR hyperintensity (SI_RelHyper) were calculated for all patients.

The patients have low SPARCC scores, ranging from 0 to 5, with 7 patients having 0 and the maximum possible SPARCC score being 72. Patients also have low calculated volume with a mean of 0,13 cm³, ranging from 0 to 0,74 cm³. SI_RelHyper had a mean of 1.07, ranging from 0 to 5.06. SPARCC scores correlate with statistical significance with volume_hyper (p<0.001) and SI_RelHyper (p<0.001) (Figure 2). Pixel-based morphometric analysis and SPARCC agree in 72 out of 77 (93,5%) of the scored slices.

MR image inflammation scoring (SPARCC MRI Index for scoring inflammation) validates the use of pixel-based morphometry as a way to accurately quantify active bone marrow inflammatory edema in PsA patients with very low inflammation. In a previous study (4), this method was used to measure inflammatory changes in the SI joints of patients with lower back pain. Those patients had larger inflammatory edema (volume_hyper ranging from 0 to 23,1 cm³) than the patients included in this study (volume_hyper ranging from 0 to 0,74 cm³ ).

The number of patients included in the analysis is relatively small, and analysis of the remaining data can further validate the use of pixel-based morphometry. Another limitation is that the sacrum and iliac bone ROIs of the patients were drawn manually. For further data analysis, automated or semi-automated methods will be explored.

Pixel-based morphometry correlates well with SPARCC scores. This method allows accurate quantification of active bone marrow inflammatory edema in PsA patients with low or very low inflammation. The significant correlation for low inflammatory scores suggests pixel-based morphometry as a reliable and sensitive quantitative method for measuring the presence and change of inflammation in bone marrow.