Background The primary MRI scoring method for assessment of bone marrow lesions (BML) in the knee in OA, MOAKS, relies on a subdivision of the knee into15 subregions and then estimation of the percent volume of subregion with BML using 3 imaging planes (1). We have developed a method for direct semi-quantitative assessment of BML using only a single sagittal imaging view, the Knee Inflammation MRI Scoring System (KIMRISS), and have custom designed an electronic overlay aimed at feasibility and capturing change in size of target lesion.
Objectives To conduct preliminary validation of a new scoring method for BML in OA of the knee based on direct assessment of the lesion using an electronic overlay and to compare this with MOAKS.
Methods MRI scans at enrolment and at 1-year follow-up were selected from the Osteoarthritis Initiative aimed at 40 patients who had received a corticosteroid injection into the knee within 6 months of the follow-up scan. We also selected baseline and 1 year follow up scans for 40 control patients matched for age, sex, and K-L grade. Pairs of scans from each case were scored by two readers blinded to time point, patient demographics, and disease characteristics. For KIMRISS a semitransparent graphic overlay containing a graphic representation of the tibia, femur and patella was fit to each image. The overlay subdivided each bone into 1cm x 1cm regions for each sagittal slice and scoring BML was dichotomous (present/absent). Inter-observer reliability of status and change scores was assessed by intra-class correlation coefficient (ICC3,1). We assessed association with WOMAC pain using correlation and by multivariate regression adjusted for age, sex, Kellgren-Lawrence grade, and baseline scores for WOMAC pain.
Results Both groups comprised 32 (80%) females, mean age 62.3 years, mean K-L grade 2.8. Baseline mean (SD) WOMAC pain scores were significantly higher in the injection group (p=0.02). There were no significant treatment group differences in WOMAC pain score at follow up. Reliability of change scores was greater for KIMRISS (Table 1). WOMAC pain status and change score correlated significantly with KIMRISS BML score (r==0.86, p<0.0001 and r=0.84, p<0.0001 for status and change score, respectively) and in both treatment and control groups (p<0.0001). Significant correlation with MOAKS BML was only seen for WOMAC status score (r=0.28, p=0.01). In multivariate analyses adjusted for age, sex, KL-grade, baseline WOMAC score, change in KIMRISS BML was independently associated with change in WOMAC pain (adjusted R2=0.75, β=0.62 p<0.0001).
Conclusions Direct semi-quantitative assessment of BML on MRI using an electronic overlay may be considerably more feasible and reliable than a volumetric-based method.
Hunter et al. Osteoarthritis Cart 2011; 19: 990
Disclosure of Interest None declared