Adults with moderate-to-severe knee osteoarthritis had weaker knee extensors, poorer muscle quality, and more fibrillar collagen in the vastus lateralis than controls (12.2% vs 8.4%). Collagen content tracked inversely with isokinetic torque (r = -0.479). Their muscle also held more PDGFRα+ mesenchymal cells and fewer satellite cells. After a single resistance session, both groups raised MMP2, MMP9 and MMP14 gene expression and MMP2/9 activity, but MMP2 and MMP9 protein area rose in controls and was blunted in the OA group. Muscle weakness in knee OA may reflect more than disuse or joint pain: the tissue appears structurally stiffer, and its collagen-clearing machinery responds poorly to a training stimulus. That fits a broader picture of fibrotic, 'anabolically resistant' muscle in older and chronically ill adults, and it hints at why strength gains in OA programmes can lag. Practically, nothing here changes advice. Strength training remains first-line for knee OA, and the finding that gene and activity responses still rose suggests the muscle can respond. Limitations are substantial: the sample is small, the design is cross-sectional so collagen cannot be shown to cause weakness, and only one exercise bout and a 72-hour biopsy were examined, so long-term remodeling is unknown. This is a preprint that has not been peer reviewed, and the findings may change. Mechanistically interesting but incremental; it needs training-intervention data to show whether repeated loading reduces collagen and restores strength.