MR images of 63 joints in 62 patients (32 female and 30 male; age range, 4–76 years) that were clinically and pathologically diagnosed as either RA (36 cases in 35 patients) or tuberculous arthritis (27 cases in 27 patients) were evaluated. The cases were retrieved retrospectively by a computer-based database program that searched on patients by final diagnosis of RA or tuberculous arthritis during a period from October 1998 to September 2003. Medical records were reviewed, and all patients had undergone MRI because of monoarticular disease with joint manifestation for the first time with symptom durations varying from 3 to 40 months, so the clinician had no prior knowledge of the disease of the patient. Final diagnoses were made on the basis of clinical history, physical examinations, radiographic survey, synovial biopsy, culture, tuberculosis polymerase chain reaction, and serologic studies. The study design was approved by our institutional review board; informed consent was not required at the time of study.

MRI was performed with various MR units, including 1- and 1.5-T scanners, but in all patients, spin-echo T1-weighted (TR range/TE range, 400–600/10–30) and fast spin-echo T2-weighted (2,500–4,500/60–108) images with or without fat suppression, along with gadolinium-enhanced spin-echo T1-weighted images were acquired [13, 14]. Although imaging planes were variable, at least two orthogonal plane images were available, including transverse images in all patients. The field of view varied from 100 × 100 to 400 × 400 mm and the matrix from 128 × 128 to 512 × 512; various coils were used, including wrist, shoulder, knee, head, and body surface coils.

MR images were analyzed retrospectively and independently by three experienced musculoskeletal radiologists without prior information about patient history, findings of other imaging studies, and final diagnosis. MR images were analyzed regarding the following findings: uniformity and degree of synovial thickening, size of and enhancement around bone erosion, degree of marrow edema, degree of soft-tissue edema, and the presence of extraarticular cystic masses. One of the three observers determined the number of extraarticular cystic masses, which were defined as any cystic masses located outside the joint regardless of their contiguity with the joint. The degree of synovial thickening after IV gadolinium administration was classified into four grades according to the maximal thickness of enhancing synovium: grade 0, 0∼3 mm; grade 1, 3.1∼6 mm; grade 2, 6.1∼9 mm; and grade 3, > 9 mm. Uniform synovial thickening was considered to be present when the maximal thickness of synovium measured less than two times the minimal thickness after IV gadolinium administration. Size of the largest bone erosion in longest diameter was documented. The size of the bone erosions was classified into five grades: grade 0, no erosion; grade 1, < 6 mm; grade 2, 6∼10 mm; grade 3, 11 mm; grade 3, 1115 mm; and grade 4, > 15 mm on T1-weighted images. Enhancement around bone erosion was determined on gadolinium-enhanced images as either present or absent. The number of erosions was not counted. Bone marrow edema was classified into four grades after calculation of the extent of marrow edema for the joint as ([maximal distance from articular margin to outer margin of signal change of bone marrow on sagittal or coronal image / maximal diameter of articular surface on axial image] × 100[%]) into the following four grades: grade 0, no edema; grade 1, 1 mm; grade 3, 1125%; grade 2, 26 mm; grade 3, 1150%; and grade 3, > 50%. Soft-tissue edema was classified into three grades after measuring the maximal vertical distance from the outer margin of joint capsule to the outer margin of periarticular soft-tissue edema on coronal or sagittal image as follows: grade 0, no edema; grade 1, 0.1 mm; grade 3, 111 cm; and grade 2, > 1 cm. Bone marrow and soft-tissue edema were evaluated on T2-weighted sequences.

Statistical analysis was performed using a commercially available software (SPSS for Windows, version 10.0.5). Using kappa statistics, interobserver agreement was evaluated for uniformity of synovial thickening, degree of synovial thickening, rim enhancement around bone erosions, bone marrow edema, soft-tissue edema, and the presence of extraarticular cystic masses. Interobserver agreement was considered poor for kappa values equal to or less than 0.200, fair for values of 0.201–0.400, moderate for values of 0.401–0.600, substantial for values of 0.601–0.800, or almost perfect for values of 0.801–1.000. If the kappa value of an item was less than 0.2 in one or more pairs between the three observers, the item was considered to be improper for use as an objective sign for differential evaluation of the joint [15, 16].

The Mann-Whitney U test was used for comparing two groups for the degree of synovial thickening, size of bone erosion, and bone marrow edema. Linear-by-linear association was calculated for the statistically significant items for evaluation of likelihood. The chi-square test was used for comparing two groups for nonuniform synovial thickening, rim enhancement of bone erosion, and presence of extraarticular cystic mass. A p value of 0.05 was considered to indicate statistical significance.