3-D JOINT SPACE WIDTH FROM WEIGHT BEARING COMPUTED TOMOGRAPHY IS BETTER THAN 2-D RADIOGRAPHIC MEASURES AT PREDICTING FUTURE LATERAL JOINT SPACE NARROWING AT THE KNEE: A MULTICENTER OSTEOARTHRITIS STUDY INVESTIGATION

Abstract

3-D joint space width (JSW) measured from weight bearing CT (WBCT) has been suggested to be more sensitive than radiographic 2-D measurement in detecting joint space narrowing (JSN), but its diagnostic value in osteoarthritis is yet to be established. To determine the predictive validity of 3-D JSW measurement from baseline WBCT compared to 2-D radiographic measures in identifying structural disease progression at the knee. WBCT knee imaging was acquired at the 144-month visit of the Multicenter Osteoarthritis Study (MOST) with radiographic Kellgren and Lawrence grade (KLG), medial and lateral tibiofemoral radiographic OARSI JSN grades. Progression in medial and lateral JSN grade was recorded 2 years subsequent. After semi-automatic segmentation of 663 available knees, joint space mapping was performed to create 3-D JSW maps. A template was registered to each individual joint surface, a 3-D statistical shape model created, and JSW measurements transferred to the template. A statistical parametric mapping (SPM) general linear model adjusted for age, sex, BMI, and the first 5 shape modes (controlling for effects of systematic misregistration) was used to test the dependence of baseline 3-D JSW on baseline medial and lateral JSN in turn. One knee per participant was included in the SPM model, when required selecting side according to worst baseline compartmental JSN grade or randomly if equal. Mean JSW values from significantly narrower SPM ROIs were taken for all knees and used in receiver operating characteristic analysis to deliver areas under the curve (AUC) in a leave-one-out cross-validation classifier predictive model for future worsening of medial and lateral JSN at 2 years. This prediction model also tested baseline KLG, medial and lateral JSN grade. All predictive models were controlled by inclusion of age, sex, and BMI and compared against each other. 10 knees did not have radiographic grading, so the study set consisted of 653 knees from 394 individuals. 218 participants were female, mean ± SD age was 63.5 ± 9.6 years, mass 82.7 ± 17.7 kg, height 170 ± 9 cm, and BMI 28. 5 ± 5.0 kg/m 2 . SPM revealed regions of significantly narrower baseline medial JSW up to ∼0.75 mm per increment in medial JSN grade (unmasked medial compartment red zone in fig. 1) and narrower baseline lateral JSW up to ∼1 mm per increment in baseline lateral JSN grade (unmasked lateral compartment red zone in fig. 2). Mean medial and lateral ROI JSW values were taken at all available knees for predictive modelling. AUC values with 95% confidence limits showed that baseline mean medial ROI JSW (0.68, 0.61-0.74) was worse than baseline KLG (0.75, 0.68-0.81) and no better than medial JSN (0.73, 0.65-0.79) or lateral JSN (0.66, 0.6-0.71) in predicting future medial JSN. However, baseline mean lateral ROI JSW (0.84, 0.72-0.91) was better than medial JSN (0.61, 0.49-0.72) and lateral JSN (0.71, 0.54-0.86) in prediction of future lateral JSN. It was also better than baseline KLG (0.77, 0.76-0.86) but not outside the 95% confidence limits, however combining baseline lateral ROI JSW and KLG in the model further improved the AUC (0.87, 0.78-0.92) beyond the limits for just KLG (0.77, 0.76-0.86). Baseline 3-D lateral compartment JSW from WBCT is better than 2-D radiographic measures at predicting progression in lateral compartment JSN, but the same was not seen for medial JSN. This suggests an important role for 3-D weight bearing JSW in assessment of knee osteoarthritis according to disease phenotype. National Institutes of Health, University of Kansas (R01AR071648), University of Iowa (U01AG18832) and University of California-San Francisco (U01AG19069). NS is a consultant for Integra BioLife, Trice Medical and Pacira Biosciences. The authors would like to thank participants and staff of the MOST study. CORRESPONDENCE ADDRESS: tom@turmezei.com