We enrolled 37 patients with definite and probable AS, and these patients were seen and treated at Ewha Womans University Medical Center. The diagnosis of AS was based on the modified NY criteria. Those patients with probable AS didn't satisfy the radiologic criteria, but they fulfilled the clinical criteria. The clinical assessment included age, gender, the duration of disease, the duration of treatment, the treatment drugs, spine involvement, peripheral joint involvement, enthesopathy, uveitis, the Shober test, chest expansion, the mean ESR and CRP levels during the treatment, the HLA-B27 level and the BASDAI (Bath Ankylosing Spondylitis Disease Activity Index). The duration of disease was defined as the time between the first symptom of back disease and the radiographic evaluation.

Both plain radiography and MDCT (Sensation 16; Siemens Medical System, Erlangen, Germany) of the SI joints were performed withn 20 days. Each patient underwent anteroposterior (AP) radiography of the SI joints in the supine position with a 20 caudal tube angle. The SI joint radiographs were graded according to the modified NY criteria1): grade 0: normal, grade 1: changes suggestive of sacroiliitis, grade 2: minimal abnormality with small localized areas of erosion or sclerosis, and without alteration of the joint width, grade 3: definite abnormality with erosions, sclerosis, joint space widening or narrowing, or partial ankylosis, grade 4: complete ankylosis (Figure 1A). MDCT was performed without contrast enhancement and its parameters were as follows: 150 mA, 120 kVp, a 0.5-sec tube rotation speed and a beam pitch of 1. The transverse and coronal MPR (multiplanar reconstruction) MDCT images were obtained with a slice thickness of 3 or 5 mm and using a bone reconstruction algorithm. The MDCT images were analyzed by using the same grading system as was used for the plain radiography (Figure 1B). Two musculoskeletal radiologists independently reviewed the images without having any knowledge of the clinical information. Each set of images was reviewed again by each observer after 1 week.

The clinical parameters are presented as means±SDs. The reliability of the radiologic assessment and the intra- and inter-observer agreement were evaluated by using Cronbach's alpha test and the kappa test. Comparisons of the clinical parameters and the radiologic grade of sacroiliitis between the plain radiographs and the CT scans were analyzed with using χ²-testing and ANOVA testing. The agreement between the 2 imaging modalities was evaluated with using the kappa test. Statistical analysis was performed using SPSS 12.0 (SPSS Inc, Chicago, IL) software. A p level < 0.05 was considered as a significant difference.

Twenty nine (78.4%) of the 37 AS patients were male with a mean age SD of 36.6±12.1 years (age range: 17-65). The mean disease duration was 103.1 months. Five (13.5%) patients were at an early stage of disease, which was defined as a disease duration of less than 2 years. The mean treatment duration was 21.2 months. Thirty six (97.3%) of the patients were positive for HLA-B27. Twenty (54.1%) of the patients had peripheral arthritis. Limitation of chest expansion was observed in 15 patients (40.5%). The mean ESR and CRP levels during the treatment were 21.4 mm/hr and 1.4 mg/dL, respectively. The BASDAI was evaluated for 27 patients, and the mean value was 3.8 (Table 1).

The intra-observer agreements between the plain radiographs and the MDCT images were good with value of >0.9. The inter-observer agreements were moderately good with a value of >0.5 (p=0.000) for the plain radiographs and >0.4 (p=0.000) for the MDCT image. When there was inter-observer disagreement for the grading, a consensus was reached through a discussion between the radiologists. Grading of the SI joints between the plain radiography and the MDCT showed moderately high agreement (κ=0.454 for the right SI joint, κ=0.612 for the left SI joint) (Table 2). MDCT detected a higher grade of right sacroiliitis in 11 (32.4%) patients who had a grade of 0-1 according to their plain radiography. Similarly, the left sacroiliitis in 9 (24.3%) patients showed a higher grade according to the MDCT than that of the plain radiography. 4 (10.8%) of the patients had a lower grade of right sacroiliitis according to CT, and left sacroiliitis was detected in 2 patients (5.4%) according to the CT as compared to the plain radiography. MDCT detected more bilateral sacroiliitis as compared to that detected via plain radiography, although this was not statistically significant (86.5% vs. 75.7%, respectively, p=0.081). The proportion of patients fulfilling the radiographic criteria of the modified New York criteria was significantly higher for the MDCT images as compared to using the plain radiography (81.1% vs. 54.1%, respectively, p=0.002) (Table 3). There were 10 (27%) patients who did not satisfy the modified NY criteria when using the plain radiography, but they did fulfill the criteria when they were imaged via MDCT. For the details, the sacroiliitis that was graded 0 or 1 by plain radiography was graded 2 or more by the MDCT (Table 3). High grades of 2-4 for the right sacroiliitis were observed more often by employing MDCT than by employing plain radiography (p=0.007), and this was also true for the (p=0.031) (Figure 2).

The clinical parameters failed to show significant correlation with the sacroiliitis grade. Although it was statistically not significant, the correlation between the disease duration and the radiologic grade showed a linear trend curve, and especially for the grades 3 and 4 sacroiliitis (Table 4).