Early Failure of a Polyvinyl Alcohol Hydrogel Implant With Osteolysis and Foreign Body Reactions in an Ovine Model of Cartilage Repair

Background: Hemiarthroplasty using a polyvinyl alcohol (PVA) hydrogel synthetic implant has been suggested as a good alternative to arthrodesis for the treatment of hallux rigidus. However, failure rates as high as 20% have been recorded. Purpose: To characterize the pathological processes in bone, cartilage, and the synovial membrane after PVA hemiarthroplasty in an ovine model with 6 months of follow-up. Study Design: Controlled laboratory study. Methods: A unilateral osteochondral defect (8-mm diameter × 10-mm depth) was made in the medial femoral condyle in 6 sheep. Animals were randomized to receive a PVA implant (n = 4) or to have an empty defect (n = 2) and were monitored for 6 months. Patellofemoral radiographs were obtained at monthly intervals, and quantitative computed tomography was performed at the end of the study. After death, the joints were macroscopically evaluated and scored. Osteochondral and synovial membrane histological findings were assessed using modified Osteoarthritis Research Society International (OARSI) and aseptic lymphocyte-dominated vasculitis-associated lesion (ALVAL) scoring systems. Immunohistochemistry using Iba1 was performed to evaluate activated macrophage infiltration. Results: Overall, 2 sheep with PVA implants were euthanized at 1 and 5 months because of uncontrollable pain and lameness (failed implants). Quantitative computed tomography showed that sheep with failed implants had 2.1-fold more osteolysis than those with successful implants. The sheep with failed implants had osteoarthritis with extensive glycosaminoglycan loss and cartilage fibrillation of the condyle and opposing tibial surface on histological examination. A foreign body reaction with severe chronic lymphoplasmacytic and granulomatous inflammation with giant cells was detected surrounding the implant. The synovial membrane ALVAL score was 9 of 19 and 14 of 19 in failed implants with synovial hyperplasia and lymphoplasmacytic and macrophage infiltration. In contrast, the synovial membrane in successful implants and empty defects was normal (ALVAL score = 0/19). Immunolabeling for Iba1 in failed implants confirmed extensive and dense macrophage infiltration within the condyle and synovial membrane, with the highest immunoreactive score (9/9). Conclusion: PVA hydrogel implants had a 50% failure rate with uncontrollable pain, severe osteolysis, inflammation, and foreign body reactions. Clinical Relevance: The failure rate and pathological characteristics of the PVA implants suggest that their use should not be continued in human patients without further in vivo safety studies.