Patients were recruited to participate in a natural history study of symptomatic knee OA, the Boston Osteoarthritis of the Knee Study. The recruitment for this study has been described in detail elsewhere.10
Briefly, patients were recruited from two prospective studies, one in men and one in women, of quality of life among veterans; from clinics at Boston Medical Center in Boston, Massachusetts; and from advertisements in local newspapers. Potential participants were asked two questions: “Do you have pain, aching, or stiffness in one or both knees on most days?” and “Has a doctor ever told you that you have knee arthritis?” For patients who answered yes to both questions, we conducted a follow‐up interview in which we asked about other types of arthritis that could cause knee symptoms. If no other forms of arthritis were identified, then the individual was eligible for recruitment. A series of knee radiographs (PA, lateral and skyline) were obtained for each patient to determine whether radiographic OA was present. If patients had a definite osteophyte on any view in the symptomatic knee, they were eligible for the study. Because they had frequent knee symptoms and radiographic OA, all patients met American College of Rheumatology criteria for symptomatic knee OA.11
The study included a baseline and follow‐up examinations at 15 and 30 months. At baseline, patients who did not have contraindications to MRIs had an MRI of the more symptomatic knee. MRIs of the same knee were also performed at follow‐up visits. At each visit, pain in the imaged knee over the past week was assessed using a visual analogue scale (VAS, 0–100), and subjects were weighed with shoes off on a balance‐beam scale. The Institutional Review Boards of Boston University Medical Center and the Veterans Administration Boston Health Care System approved the study.
All MRIs were performed with a Signa 1.5T system (General Electric, Milwaukee, WI) using a phased‐array knee coil. A positioning device was used to ensure uniformity among patients. Coronal, sagittal and axial images were obtained. Fat‐suppressed spin‐echo proton density and T2‐weighted images (repetition time, 2200 ms; echo time, 20/80 ms) with a slice thickness of 3 mm, a 1‐mm interslice gap, one excitation, a field of view of 11–12 cm, and a matrix of 256×128 pixels were obtained.
Cartilage morphology was assessed by a musculoskeletal radiologist (AG) using a semiquantitative, multi‐feature scoring method (whole‐organ magnetic resonance imaging score, WORMS) for whole‐organ evaluation of the knee that is applicable to conventional MRI techniques.12
Intraclass correlation coefficients of agreement among the readers for cartilage readings ranged from 0.72 to 0.97.
Tibiofemoral cartilage on MRI was scored on all 5 plates (central and posterior femur; anterior, central and posterior tibia) in each of the medial and lateral tibiofemoral joints. The anterior femur was not included in this analysis, as this is part of the patellofemoral joint. Patellofemoral cartilage was scored on 4 plates (medial and lateral patella, and medial and lateral anterior femur). These were read using the fat‐suppressed T2‐weighted FSE images on a 7‐point scale: 0
normal thickness and signal; 1
normal thickness but increased signal on T2‐weighted images; 2
partial‐thickness focal defect <1 cm in greatest width; 3
multiple areas of partial‐thickness (Grade 2) defects intermixed with areas of normal thickness, or a Grade 2 defect wider than 1 cm but <75% of the region; 4
75% of the region) partial‐thickness loss; 5
multiple areas of full‐thickness loss wider than 1 cm but <75% of the region; 6
75% of the region) full‐thickness loss.
In WORMS, grade 1 does not represent a morphological abnormality but rather represents a change in signal in cartilage of otherwise normal morphology. Grades 2 and 3 represent similar types of abnormality of the cartilage, focal defects without overall thinning. Scores of 1 and 2 were exceedingly unusual. Therefore, to create a consistent and logical scale for evaluation of cartilage morphological change, we collapsed the WORMS cartilage score to a 0–4 scale, where the original WORMS scores of 0 and 1 were collapsed to 0, the original scores of 2 and 3 were collapsed to 1, and the original scores of 4, 5 and 6 were considered 2, 3 and 4, respectively, in the new scale.13
The intraobserver agreement for reading of cartilage morphology ranged from 0.65 to 0.78 (kappa). We defined a lesion as occurring in either the medial or lateral tibiofemoral compartment if it was present in the femur or tibia of that compartment. While we conducted analyses using this collapsed WORMS cartilage scale, analyses using the original scale yielded the same results.
On the baseline and follow‐up MRIs, effusion was scored 0–3 based on volume. Bone‐marrow lesions were scored only on the baseline MRIs using the WORMS scale also in which lesions are scored according to their size (0–3) within quadrants of the femur and tibia.
Synovial thickening on MRI using sagittal T2‐weighted and proton‐density sequences was scored separately at 3 locations (infrapatellar fat pad, suprapatellar and intercondylar regions) using a semiquantitative scale (0–3) at all 3 time points (fig 1). Given the confirmation that these MRI findings connote inflammation in the synovium, we shall label these findings as synovitis.
Figure 1(A) T2‐weighted MR image, sagittal view, with soft tissue density and surrounding synovitis in intercondylar region. (B) T2‐weighted MR image, sagittal view, with synovitis (grade 2) in infrapatellar region.
One reader (CLH) read synovitis on MRI's. For each subject, MRIs were blinded to subject's identity and read paired and in sequential order. The intraobserver agreement (kappa) for infrapatellar fat pad synovitis score was 0.63, intercondylar 0.49 and suprapatellar 0.20.
Validation of non‐gadolinium synovitis scoring
To validate non‐gadolinium scoring, 20 subjects with knee OA at University of Leeds underwent MRI using a sagittal T2 weighted fat suppressed sequence and a gadolinium enhanced T1 weighted fat‐suppressed sequence. A trained reader scored the infrapatellar fat pad for synovial thickening on a 0–3 scale using just the T2 sequence without reference to other sequences (as above). The same reader then scored the infrapatellar synovitis using the same scoring system using just the postgadolinium sagittal sequence again without reference to other sequences. The films were blinded and presented to the reader in random order for the 2 reads, which took place 1 week apart. Of the 20 knees, 13 showed contrast (gadolinium) enhanced and non‐contrast enhanced scores that were identical (ranging from 0 to 3). As expected, in 6 knees, the non‐gadolinium images underestimated the amount of synovial thickening seen on the contrast enhanced image. Only one knee showed an over‐reading of synovial thickening on the non‐contrast image (score of 2 vs 1).
Synovitis scores at each location were added to give a summary synovitis score at each time point (0–9). Changes in synovitis score were calculated at each time point. In addition, an analysis was carried out for synovitis scores at each individual site.
To examine whether differences in VAS pain score can be explained by differences in synovitis both cross‐sectionally and longitudinally, we applied the generalised estimating equation to test this hypothesis with the following statistical model:
where Yit is the pain score assessed at baseline, at 15 months and at 30 months. Xi0 is the synovitis assessed at baseline, and Xit is the corresponding measure of synovitis assessed at time t, that is, baseline, 15 months and 30 months, respectively. The coefficient β1 measures the cross‐sectional association of the synovitis at baseline and VAS pain score, and β2 measures the effect of changes in synovitis on changes of VAS pain score.
The interpretation of the estimate from the model is the expected change in pain over time (from baseline to follow‐up including both 15 and 30 months) per unit change in synovitis score in the corresponding follow‐up period for a given subject.
To examine whether cartilage loss can be explained by baseline synovitis, we used cartilage loss in each compartment (medial tibiofemoral, lateral tibiofemoral, and patellofemoral) in 30‐month follow‐up for analyses unless unavailable, in which case cartilage loss in 15 months was used. Cartilage loss took whole number values from 0 (no loss) to 4 (maximum loss) and was analysed as ordered categories using the proportional odds logistic regression model. A generalised estimating equations correction was applied to the regression model to account for the association in the cartilage loss outcome between regions within a joint. For change in pain and cartilage loss, analyses were adjusted for baseline cartilage scores, age, sex, BMI, effusion score (0–3) and bone‐marrow lesion score (using WORMS), and change in both bone‐marrow lesion score and effusion score. For cartilage loss, we adjusted additionally for baseline WORMS based meniscal score.
A similar method was used to examine whether cartilage loss can be explained by synovitis change. The change in synovitis in a 30‐month follow‐up was used for analyses unless unavailable, in which case the change of synovitis in 15 month follow‐up was used.