Over the past decades, numerous structural changes in implants, medical treatments, and surgical techniques have been made for Malignant Bone Tumors (MBT) around the knee. However, the overall care improvement is still unclear. The method is crucial when analyzing outcomes in surveys involving tumors, and a thorough assessment of the mortality is mandatory because death acts as competing event. The aims of this study were: 1) a comprehensive and longitudinal assessment of the revisions with an extensive follow-up and adequate methods; 2) a complete mortality review to consider competing risks.

The hypothesis was that some prosthesis's structural improvements were made while the surgical toll increased as well as an improvement of mortality was also expected.

Analyses were performed on 248 patients with MBT (mean follow-up was 8.7 years, surgeries between 1972 and 2017). Three prosthesis models were successively used over time: 120 Guepar (older model), 42 Tornier, and 86 Stanmore (more recent model). The primary outcome was the assessment of revisions sorted out according to Henderson: type-1 soft-tissue failures or instability, type-2 aseptic loosening, type-3 structural failures, type-4 periprosthetic infections, type-5 tumoral progression. Death and amputations were considered as competing events. An extensive assessment of mortality was performed by merging the dataset with the French register of Deaths (INSEE). Cumulative probabilities were computed at 2, 5, 10, and 15 years and compared with Gray's tests.

The overall 5-year survival was, 80% (95% CI: 73–87) for Guepar, 69% (95% CI: 56–84) for Tornier, and 71% (95% CI: 62–82) for Stanmore (p=0.4). The 5-year cumulative risks for type-1 were 5% (95% CI: 1–9), 9% (95% CI: 0–18), and 17% (95% CI: 9–25) for Guepar, Tornier, and Stanmore, respectively (p=0.01). The 15-year cumulative risks for type-2 were 22% (95% CI: 15–39), 8% (95% CI: 0–17) and 8% (95% CI: 2–14) for Guepar, Tornier, and Stanmore, respectively (p=0.10). Ten patients had an implant failure, nine Guepar, and one Tornier. The 5-year cumulative risks for type-4 were 7% (95% CI: 2–12), 19% (95% CI: 7–31), and 12% (95% CI: 5–18) for Guepar, Tornier, and Stanmore, respectively (p=0.08). There were 29 tumoral progressions; the 15-year risks were 16% (95% CI: 2–22), 2% (95% CI: 0–7%), and 12% (95% CI: 4–19%) for Guepar, Tornier, and Stanmore, respectively (p=0.08). No difference whatsoever was found between the proximal tibial and distal femur.

There were some improvements in prosthesis design (forged steel instead of cast steel) and probably also in cemented stem fixation, but not in prosthetic joint infection and local recurrence over forty years. The overall mortality did not change significantly over the last 40 years amongst this specific cohort of patients who benefited from a hinge reconstruction prosthesis.

III; comparative case series with sensibility analysis.