Background An operational case definition for identification of erosions imaged by HR-pQCT was achieved and tested in a first reliability exercise (RELEX-1) using cross-sectional data.
Objectives To test the case definition and define landmarks for measurement of erosion size in 2D in a longitudinal dataset.
Methods Patients meeting classification criteria for RA at various stages of disease duration and severity had HR-pQCT scans of the 2nd and 3rd MCP joints at 0 and 12 month timepoints. Standard image acquisition and segmentation was performed for a 2.7 cm scan area. Images are evaluated for erosions at 8 surfaces per joint (ulnar, radial, dorsal, palmar surfaces of the proximal phalanx and metacarpal head). The erosion case definition requires the presence of a definite interruption in the cortical bone extending over at least 2 consecutive slices, visualized in 2 orthogonal planes, with loss of underlying trabecular bone and being non-linear in shape. The maximal width of the cortical break is identified and measured in the axial plane, with the maximal depth recorded at 90 degrees to this line. This same method is repeated in the perpendicular plane. Five readers blinded to patient identity and timepoint of the scan scored 36 baseline and follow-up images (18 joints from 9 patients). Percent agreement and a kappa score for erosion detection (minimum of 2 readers in agreement) was calculated. The variation in width and depth measurements of erosions in axial and perpendicular planes between readers was calculated using the root-mean-square coefficient of variance (RMSCV). McNemar's test was used to test for a significant change in the number of erosions identified from 0 to 12 months. Paired t-tests were used to calculate the average width and depth changes from 0 to 12 months.
Results Agreement for the presence or absence of an erosion was 92.9% (k=0.711, 95%CI 0.539-0.839). Mean (SD) dimensions of the erosions were: axial width 1.66 (SD 0.99) mm, perpendicular width 1.60 (SD 0.78) mm, axial depth 1.16 (SD 0.66) mm, and perpendicular depth 1.17 (SD 0.57) mm. The respective RMSCV were 36.4%, 30.3%, 15.7% and 27.3%. Seven erosions were detected at both 0 and 12 month timepoints (k=0.755, 95%CI 0.616-0.895) and all readers agreed on a single new erosion developing between 0 and 12 months for 1 subject (McNemar's test p=1.00). Erosion dimensions did not significantly change over 1 year but a trend to reduction in mean size was observed, with an axial width decrease of 0.171 mm (95%CI -0.168, 0.509, p=0.310), perpendicular width decrease of 0.014 mm (95%CI -0.202, 0.230, p=0.898), axial depth increase of 0.007 mm (95%CI -0.147, 0.133, p=0.917), perpendicular depth decrease of 0.133 mm (95%CI -0.075, 0.340, p=0.201).
Conclusions The case definition for erosions imaged by HR-pQCT is valid in longitudinal datasets. Despite variability in measurements, a trend towards changes in erosion size is demonstrated.
Disclosure of Interest None declared