Background/Purpose: EULAR guidelines consider conventional radiograph of sacroiliac joints (SIJs) as the first recommended imaging method in case of suspected axial spondyloarthritis (axSpA). However, it is not clear whether sacroiliac joints can be reliably assessed on anteroposterior (AP) lumbar radiographs, which are often performed as a part of the diagnostic work-up in patients presented with back pain. The aim of the study is to investigate reliability and validity of radiographic sacroiliitis assessment on AP lumbar radiographs as compared to conventional pelvic radiographs in patients with axSpA.

Methods: Patients from the GErman SPondyloarthritis Inception Cohort (GESPIC) were selected based on the availability of sets of pelvic and AP lumbar radiographs with visible SIJs at baseline and after 2 years of follow-up. Two trained readers (ML and VR) scored the images independently and in a random order according to the radiographic system of the modified New York (mNY) criteria (grade 0 to 4). The sacroiliitis sum score (0-8) was calculated as a sum of the mean grades of 2 readers for the right and left SIJ. We assessed intra- and inter-reader reliability using intraclass correlation coefficients (ICC) of the sacroiliitis sum scores.

Results: A total of 226 sets radiographs were scored from 113 patients included in the present study. The intra-observer agreement was good to excellent for the sacroiliitis sum score of pelvic vs. AP lumbar radiographs at baseline (ICC 0.80 for ML and 0.74 for VR) and at year 2 (ICC 0.81 for ML and 0.77 for VR). The inter-observer agreement for pelvic and AP lumbar radiographs was also good to excellent: ICC at baseline: 0.81 and 0.73, respectively, at year 2: 0.76 and 0.79, respectively.

A total of 62 (54.9%) and 55 (48.7%) patients were classified as r-axSpA at baseline based on evaluation of pelvic and AP lumbar radiographs, respectively. The absolute agreement on the classification was 84.9% (figure). A total of 17 patients (12 (10.6%) with nr-axSpA and 5 (4.4%) with r-axSpA) were classified differently based on assessment of AP lumbar as compared to conventional pelvic radiographs (figure).

After 2 years of follow-up, progression from nr- to r-axSpA occurred in 7 patients (6.2%) and 8 patients (7.1%) classified as nr-axSpA at baseline based on pelvic or AP lumbar radiographs assessment, respectively. Regression from r- to nr-axSpA occurred in 4 patients (3.5%) and 3 patients (2.7%) on pelvic or AP lumbar radiographs, respectively, giving respective net progression rates of 2.7% and 4.4%.

Conclusion: Radiographic sacroiliitis can be assessed on AP lumbar radiographs with a similar reliability as on conventional pelvic radiographs.