Authors
Benjamin G Faber, Mijin Jung, Raja Ebsim, Fiona R Saunders, Asad Hashmi, Sophie Scott, Jennifer S Gregory, Nicholas C Harvey, John P Kemp, George Davey Smith, Andy Judge, Cindy Boer, Richard M Aspden, Claudia Lindner, Timothy Cootes, Jamie E Collins, Jonathan H Tobias
Published in
Osteoarthritis and cartilage. Nov 16, 2026. Epub Nov 16, 2026.
Abstract
Osteoarthritis is a heterogeneous disease, with diverse structural patterns likely reflecting distinct pathomechanisms. However, scalable imaging-driven approaches to define clinically meaningful phenotypes are limited. This study aimed to identify structure-based osteoarthritis phenotypes incorporating demographic features at population scale and evaluate their clinical and genetic profiles.
Machine learning models were applied to knee and hip DXA scans from UK Biobank to quantify osteophyte area, minimum joint space width, and B-scoresDXA. Imaging and demographic features were clustered using k-means to classify individuals with at least one osteoarthritis feature. Associations with joint pain and total joint replacement (TJR) were evaluated relative to structurally normal controls. Genetic correlations and polygenic risk scores were analysed to characterise shared and differing genetic architecture across phenotypes.
Among 59,539 participants, a nine-phenotype clustering solution was derived in 28,953 participants with structural features of osteoarthritis, spanning joint-specific and multi-joint patterns. Across the nine phenotypes, prolonged knee pain ranged from 12.2% (95% CI 11.5-13.0) to 61.2% (52.7-69.0), and during follow-up knee replacement from 0.37% (0.25-0.55) to 14.9% (9.9-21.9), with the lowest proportions in the narrower joint space phenotype and highest in knee end-stage. The adjusted odds ratio for knee replacement in knee end-stage versus structurally normal controls was 98.8 (95% CI 55.5-175.9). Hip replacement ranged from 0.21% (0.12-0.35) in narrower joint space to 13.6% (10.7-17.1) in left hip end-stage (adjusted OR versus controls 72.9 [95% CI 50.3-105.6]). Genetic differences included an inverse genetic correlation between wider and narrower joint space phenotypes (rg -0.46 [95% CI -0.91, -0.01]). End-stage phenotypes demonstrated the highest joint-specific polygenic risk scores.
Population-scale imaging identifies structure-based osteoarthritis phenotypes, encompassing variation in demographics, structural components and severity, with distinct clinical outcomes and genetic profiles. This provides a scalable framework for stratifying osteoarthritis patients. Whether these phenotypes predict treatment response or inform treatment selection remains to be established and further investigation is required.
PMID:
43039212
Bibliographic data and abstract were imported from PubMed on 17 Nov 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 1
- Comments 0