Deep Learning–Based Comparison of Knee Minimum Joint Space Width in Patients With Rheumatoid Arthritis and Osteoarthritis Before Total Knee Arthroplasty
Authors' conclusion
Patients with RA demonstrated diffuse, symmetric cartilage loss that contrasted with medial-predominant narrowing in patients with OA.
Study context
- Population
- Patients with rheumatoid arthritis (RA) or osteoarthritis (OA) undergoing total knee arthroplasty, 2000–2021302 with paired radiographs
- Exposure
- RA vs OA patients undergoing total knee arthroplasty (TKA)
- RAn = 409
- Outcome
- Compartment-specific (medial and lateral) mJSW measured by deep learning (DL) on anteroposterior knee radiographs
- Follow-up
- Early (> 2 years before TKA) and late (≤ 2 years before TKA)
Methodological detailsage- and sex-matched, deep learning–based measurement, mixed-effects models adjusted for BMI and alignment
- Age- and sex-matched
- Deep learning–based measurement
- Mixed-effects models adjusted for BMI and alignment
Main finding
As reported in the abstract
Lateral mJSW · late timepoint (≤ 2 years before TKA)
P < 0.001
Observational data: associations, not effects.
Further findings
mm/year
| Outcome | Seronegative | Seropositive | Estimate | P value |
|---|---|---|---|---|
| Rate of lateral JSN (mean lateral ΔmJSW), faster in RA | mean lateral ΔmJSW 0.11 mm/year | P = 0.01 | ||
| Rate of medial JSN (mean medial ΔmJSW), faster in OA | mean medial ΔmJSW 0.21 mm/year | P < 0.001 | ||
| Lateral ΔmJSW in the surgical knee by serostatus (RA subgroup) | 0.12 | 0.06 | P = 0.03 |
- Rate of lateral joint space narrowing (JSN) (mean lateral ΔmJSW), faster in RA: 302 patients with paired radiographs
Lateral mJSW and RA disease characteristics
- Longer RA duration and higher inflammatory markers negatively associated with lateral mJSW
Original publication
Deep Learning–Based Comparison of Knee Minimum Joint Space Width in Patients With Rheumatoid Arthritis and Osteoarthritis Before Total Knee Arthroplasty. J Rheumatol. Published 1 May 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →