Mylène Jansen

246 Chapter 12 Discussion Although cross-sectional evaluation provided a statistically significant correlation between plain radiographic mean JSW (bone-to-bone distance) and MRI surface mean cartilage thickness, no statistically significant correlation between these measures was found when evaluating the relatively small changes over 2 years follow-up. Similarly, there was no significant correlation between the 2-year change in plain radiographic mean JSW and 2D CT mean JSW, whereas cross sectional evaluation provided such a correlation. In contrast, the 2-year change in MRI surface mean cartilage thickness correlated strongly with 3D CT surface median bone-to-bone distance. Also, the 3D CT surface median JSW correlated strongly with the 2D mean JSW. From this it is concluded that non-weight-bearing image acquisitions, independent of using evaluation of bone-to-bone distance measurements (CT) or cartilage thickness measurements (MRI), result in significant correlations between outcomes. In contrast, when a weight-bearing imaging technique (plain radiography) is compared to non-weight-bearing imaging techniques (MRI and CT) the correlation is lacking. It can therefore be concluded that weight-bearing image acquisition provides an independent characteristic of cartilage, that is not observed by non-weight-bearing techniques. Deformability of the cartilage (cartilage quality) may be involved in addition to the quantitative measurement of cartilage thickness. The position and morphology of the meniscus may also play a role, although visually scored meniscal extrusion (grade 0 – 3) did not seem to significantly influence the longitudinal correlation in this group of patients (data not shown). The significant correlations found between the different imaging techniques when evaluating cross-sectional data, whereas such correlations are lost in case of relating more subtle changes in cartilage quantitative measures during (2-year) follow-up, fits the inconclusive literature on this topic. 14–21 With the exception of radiographic JSW, the 2-year changes over time in our study were much smaller than the absolute baseline or 2-year values (at least 1 order of magnitude decrease), while the standard deviations stayed roughly the same (Table 1). Apparently, correlations are lost when weight-bearing image acquisition is compared to non-weight-bearing acquisition in case of small changes (over time), whereas they are maintained when bone-to-bone distance is compared to cartilage thickness in a 2D or 3D manner when the image acquisition is non- weight-bearing. This argues for the use of weight-bearing image acquisition, such as weight-bearing CT or weight-bearing MRI. Both these techniques have been researched and have shown positive results, but use of both is mostly limited to research settings. 34–36 To further investigate the objectives of our study, a rotatable MRI scanner would be a valuable tool, since both cartilage thickness and JSW can be measured in weight-bearing and non-weight-bearing position using