Mylène Jansen

330 Chapter 16 between medians. Correlations between the changes over 6 weeks for each analyte were assessed by Spearman’s R coefficient (range -1 to 1; where ±1 = strongest positive (or negative) correlation, 0 = no correlation). Association of change in analytes with KOOS 4 The clinical outcome variable was change in KOOS 4 over time (KOOS 4 at either 3, 6 or 12 months respectively - KOOS 4 at baseline). Linear regression was employed to model the relationship between continuous change in analyte levels (concentrations at 6 or 3 weeks - baseline concentrations) and change in KOOS 4 . In a planned secondary analysis, linear regression also assessed change in KOOS 4 by categories of change in analytes. Concentrations of analytes (at baseline and 6 weeks) were classified into normal (≥25th and <75th percentiles), high (≥75th percentile) and low (<25th percentile) categories. The 25th and 75th centiles were calculated from measurements of these molecules in SF from 40 individuals with OA who had undergone either partial or total knee joint replacement (see Comparator ranges). These data were generated at same time as participant data, using the same assay batches. ‘Relevant change’ was defined as a movement between at least 1 category from baseline to 6 weeks (relevant increase, or relevant decrease), or as no relevant change. Change in analytes by responders and non-responders in KOOS 4 Responders (those whose change in KOOS 4 over 6 months (the latest point at which there was clinical change from baseline), (Figure 1C) was ≥10 points, i.e., the minimal clinically important difference (MICD) for KOOS 4 ) 32 ; and non-responders (those whose KOOS 4 change over 6 months was <10 points) were categorized. Differences between molecular changes in these 2 groups were compared by Mann-Whitney U test. Data were stored on a secure database (OpenClinica). Analysis was performed in STATA IC 13.1 and Graphpad Prism 6.03. Results 13/20 (65%) participants were male with mean age 55 ± 5 years (Table II). All had KL grade≥2; 18 (90%) grade 3/4, with substantial knee pain at baseline (KOOS pain 38.6 (SD 16.0); where 100 is no pain, normal function). As expected from previously published studies, there was an improvement in KOOS 4 in the subsequent months following the intervention (Figure 1C).