Low Artificial Intelligence-Estimated Bone Mineral Density From Chest Radiographs Is Associated With Delayed Pain Recovery and Lower Quality of Life After Medial Unicompartmental Knee Arthroplasty.
Authors
Affiliations (3)
Affiliations (3)
- Department of Orthopaedic Surgery, Faculty of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan. Electronic address: [email protected].
- Science for Joint Reconstruction, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan.
- Department of Orthopaedic Surgery, Faculty of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-0033, Japan.
Abstract
Whether bone quality influences recovery after medial unicompartmental knee arthroplasty (UKA) remains uncertain, partly because bone mineral density (BMD) is not routinely evaluated using dual-energy X-ray absorptiometry before UKA. This study investigated whether artificial intelligence (AI)-estimated proximal femur BMD derived from preoperative chest radiographs was associated with longitudinal patient-reported outcomes after cemented medial UKA. We retrospectively reviewed 342 consecutive cemented medial UKAs performed from 2012 to 2023. After exclusions, 336 procedures were categorized according to AI-estimated proximal femur young adult mean (YAM), with low estimated BMD defined as YAM less than 80%. Propensity-score matching based on age, sex, body mass index, and diagnosis identified 58 low-BMD procedures and 58 matched controls. Knee injury and Osteoarthritis Outcome Score (KOOS) subscales were evaluated preoperatively and up to three years postoperatively. Linear mixed-effects models assessed group-by-time interactions. After matching, baseline characteristics were comparable between groups. Both groups demonstrated postoperative improvement across all KOOS subscales. Significant group-by-time interactions were observed for KOOS pain (P < 0.001) and quality of life (QOL) (P = 0.010), whereas no significant interactions were identified for symptoms, activities of daily living, or sport/recreation. The low BMD group had lower pain scores at three months, six months, and one year, as well as lower QOL scores at one, two, and three years. The one-year postoperative range of motion, alignment, component positioning, and radiographic migration or subsidence did not significantly differ between groups. Low AI-estimated proximal femur BMD from routine chest radiographs was associated with delayed early pain recovery and statistically lower knee-related QOL from one to three years after cemented medial UKA. Opportunistic AI-based BMD screening may identify candidates for preoperative bone health evaluation and counseling.