Skip to main content

28-04-2022 | Rheumatology | News | Article

BSR 2022

Semi-automated hip OA classification based on DXA images has potential to predict clinical outcomes

Author: Claire Barnard

medwireNews: A semi-automated method of classifying hip osteoarthritis (OA) into grades based on dual-energy X-ray absorptiometry (DXA) images may be able to predict the need for total hip replacement, suggests research presented at the British Society for Rheumatology 2022 Annual Conference in Glasgow, UK.

This tool could provide a “more objective measure” of radiographic hip OA (rHOA) than subjective methods such as Kellgren–Lawrence scoring of hip X-rays, said presenting author Benjamin Faber (University of Bristol, UK).

The investigators employed a machine-learning approach to categorize hip DXA images from the UK Biobank into rHOA grades 0–4 based on osteophyte and joint space narrowing scores. Faber noted that greater weight was given to osteophytes, which were shaded manually on the images.

Among the 40,340 individuals included in the study (mean age 63.7 years), 81.2% were categorized as rHOA grade 0, while 11.3%, 5.7%, 1.3%, and 0.4% were categorized as rHOA grade 1, 2, 3, and 4, respectively.

Faber said that these grades showed “progressive relationships [with] three clinically relevant outcomes,” with the strongest association seen for total hip replacement.

People with rHOA grade 2 and above had a significantly higher likelihood of requiring total hip replacement than those with rHOA grade 0 after adjustment for potential confounders, with the hazard ratio (HR) increasing from 4.00 in the grade 2 group to 13.39 in the grade 3 group and 57.70 in the grade 4 group.

A similar pattern of results was seen for hip pain, with HRs increasing from 1.57 for people with rHOA grade 2 to 11.82 for those with grade 4, as well as for hospital-diagnosed hip OA, with corresponding HRs of 3.84 and 41.06.

Faber believes that the study results “open up opportunities to diagnose hip OA and screen for total hip replacement using high-resolution DXAs, which are a cheap scanning method” frequently used to assess osteoporosis.

“More work is needed in this area but we feel this is exciting,” he concluded.

medwireNews is an independent medical news service provided by Springer Healthcare Ltd. © 2022 Springer Healthcare Ltd, part of the Springer Nature Group

BSR Annual Conference 2022; Glasgow, UK: 25–27 April

More on this topic

See the research in context now

with trial summaries, expert opinion and congress coverage