Background/Purpose:

High resolution peripheral quantitative computed tomography (HR-pQCT) allows in vivo 3 D imaging of human joint microstructure and joint space width (JSW) at an isotropic resolution of 82μm voxel size. As JSW is regarded a surrogate measure of arthritic disease activity, the HR-pQCT-derived JSW may qualify as an important rheumatologic outcome measure. To date, JSW was reproducibly computed with an ICC of 0.989 in healthy subjects with maintained metacarpophalangeal (MCP) joints. However, data are still lacking on the performance of HR-pQCT-derived JSW measurements in arthropathies with very narrow joint spaces. One of those arthropathies is hemochromatosis arthropathy (HA) which is characterized by excess iron-induced joint destruction and hook-like osteophyte formation. Objective: We aimed to determine in patients with HA the 1) feasibility of the HR-pQCT-derived JSW algorithm 2) performance of the JSW-algorithm in this narrow-spaced cohort by assessing failure rates in relation to the corresponding joint space widths 3) JSW cut-off value, below which the JSW algorithm is set to fail and would further require manual re-segmentation.

Methods: MCP of 29 HA patients were imaged on one HR-pQCT system (XtremeCT, Scanco Medical AG). From the images, the mean joint space width (JSW) was semi-automatically computed via an in-house developed and well-validated software as the mean of local 3D widths across the joint space. In case the software failed at the first JSW analysis attempt, each MCP was semi-manually segmented by a trained operator to separate the metacarpal head and phalangeal base and a second JSW-analysis was attempted.

Results: 76.2 % of all 84 MCP joint spaces were successfully segmented at the first attempt. 22.6% required semi-manual intervention while in one case the JSW remained undeterminable even after semi-manual correction due to direct contact between the metacarpal and phalangeal bones. MCPs were next subdivided by JSW tertile and failure rates recorded. In MCPs of the lowest tertile software failure rates were highest with 50 % failure in MCP2 joints, 44.4 % in MCP3 joints, and 22.2% failure rates in MCP4 joints. MCPs ranging in the medium JSW tertile showed 22% (MCP2), 33% (MCP3) and 22% (MCP4) failures rates. Joints with a JSW in highest tertile were all successfully segmentable at the first attempt (0% failure rate). When looking at a JSW-cut-off value, below which the JSW software would fail its segmentation, we observed that in general a mean JSW of <1.4 mm was associated with a high rate of software failure. In MCP 3 and 4 a cut-off of a mean JSW of ≥ 1.41 mm resulted in 100% successfully segmented cases.

Conclusion: Our findings suggest that HR-pQCT-derived JSW quantification in MCP joints is feasible even in arthropathies with known narrow joint spaces such as hemochromatosis arthropathy. Moreover, performance analysis indicated that the software performance seems to be best in the joints belonging to the highest JSW tertile, while joints with a mean JSW of 1.4 mm or smaller seem to fail very likely at the first analysis attempt and do require semi-manual corrections. Further work is needed to assess failure rates in other patient cohorts.

« Back to 2018 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/feasibility-and-performance-of-hr-pqct-derived-joint-space-width-measurement-as-outcome-parameter-in-arthropathic-disease-lessons-from-hemochromatosis-arthropathy/