The contribution of 3D meniscus and cartilage measures to normal radiographic joint space width – Data from the Osteoarthritis Initiative healthy reference cohort

Abstract

Purpose: Minimum radiographic joint space width (mJSW) represents the current reference standard for demonstrating structural benefits of disease modification in knee osteoarthritis (OA), accepted by regulatory agencies. Loss of mJSW in knee OA is associated with both cartilage thinning and meniscal pathology. Yet, to what extent quantitative meniscal measures contribute to minimum and fixed location JSW in healthy knees without radiographic knee OA is unknown. The aim of this study was therefore to elucidate to what extent 3D morphometric measures of the meniscus and cartilage explain variability in measures of healthy JSW in men and women. Methods: 99 right knees of participants from the healthy reference cohort of the OAI were evaluated. These subjects were free of radiographic OA (bilateral KLG0 in central readings v0.6), asymptomatic, and had no risk factors of knee OA, including malalignment. Medial compartment mJSW as well as medial (JSW225) and lateral (JSW800) fixed location measures were obtained from fixed flexion radiographs from the OAI data base (v0.6). Femorotibial cartilage and meniscal segmentation was available from DESS MRIs (sagittal and coronal reconstructed images, respectively). The mean thickness of the tibial (T.ThCtAB), weight-bearing femoral (cF.ThCtAB) and femorotibial (FTC.ThCtAB) cartilage was calculated, including subregions. Twelve medial and 18 lateral compartments could not be included due to missing JSW values or cartilage/meniscus segmentations. Meniscal parameters included the percent tibial area covered by the meniscus (ACdAB.Cov%), meniscal area extruding from the tibia (TA.Uncov%), mean extrusion distance (Ex.Me), mean meniscal thickness (Th.Me), and mean meniscal width (Wid.Me). Bivariate correlation coefficients (Pearson) were computed between JSW measures and independent variables using the bootstrap method (1000 replications, BCa method, simple sampling). These and partial correlations were used to reduce the number of independent variables to model JSW by stepwise multiple linear regression, with all parameters in Table 1, apart from body weight, used. Results: Cartilage thickness generally displayed higher correlations with all JSW measures than meniscal parameters, except for mJSW in women. The highest correlation of mJSW was noted with anterior tibial ThCtAB (r=0.69), and a similar coefficient with total medial compartment (MFTC).ThCtAB (r=0.65; Fig. 1). The highest correlations of medial (JSW225) and lateral (JSW800) fixed location measures were noted for ispi-compartimental (FTC).ThCtAB (r= 0.75/r=0.72). Amongst meniscal measures, the highest correlation with mJSW and medial fixed location JSW was observed for Th.Me in the posterior partition (r=0.50/r=0.32), and the highest for lateral JSW for Wid.Me (r=0.32). Stepwise regression revealed that meniscal measures added significant information to explain variance in mJSW (adjusted multiple R2=58%), whereas variation in fixed location JSW measures was explained by cartilage measures alone (JSW225: R2=60%; JSW800: R2=51%). Whilst the correlation of fixed location medial JSW with cartilage and meniscal measures was similar between men and women, that of mJSW with cartilage appeared higher for men than women, and that of mJSW with meniscal measures higher for women than men (Table 1). Differences in the correlations between sexes, however, did not reach statistical significance (Fisher’s z-test).Tabled 1Sex-specific correlations of mJSW, JSW225 and JSW800 with quant. meniscal and cartilage parameters* p<0.05** p<0.01ParametermJSW men (n=37)mJSW women (n=50)JSW225 men (n=37)JSW225 women (n=50)JSW800 men (n=34)JSW800 women (n=47)DemographicsBMI−0.270.20−0.36*0.13−0.060.00Height0.250.26&0.51**&0.35*0.070.43**&Weight−0.080.38**0.010.35*0.000.29Age−0.35*−0.08−0.35*−0.12−0.46**&−0.14ClinicalKnee Alignment0.140.23&0.160.10−0.170.10CartilageFTC.ThCtAB0.63**0.38**0.72**&0.66**&0.66**&0.46**&T.ThCtAB0.67**&0.43**&0.69**0.57**0.63**0.42**cF.ThCtAB0.48**0.270.61**0.61**0.46**0.35*MeniscusACdAB.Cov%0.220.34*&−0.14−0.110.02−0.20TA.Uncov%−0.26−0.15−0.160.07−0.060.06Th.Me0.35*0.37**0.120.220.010.17Ex.Me−0.060.06−0.140.12−0.24&−0.07Wid.Me0.120.42**−0.120.050.04−0.02Stepwise linear regression (variance explained)&:included by model44%57%60%42%54%27% Open table in a new tab Conclusions: In healthy reference subjects, mJSW reflects a combination of quantitative cartilage and meniscal measures. Fixed location JSW, in contrast, is dominated by variance in cartilage thickness, particularly in the medial compartment. The correlation of lateral fixed location JSW with lateral cartilage was as high as that of mJSW with medial cartilage. In contrast to medial fixed location JSW, mJSW and lateral fixed location JSW tended to display higher correlations with cartilage thickness in men than in women. A potential explanation for this observation may be the greater knee joint laxity in women. In normally aligned knees, joint loading is theorised to be transmitted mostly through the central medial compartment during fixed flexion radiography, where medial fixed location (225) JSW is measured. In laxer joints, mJSW and lateral fixed location JSW may be more variable, and hence less reflective of cartilage thickness than medial fixed location JSW.