Introduction: The risk of prosthetic joint infection (PJI) in mega-prosthesis for malignancy is increased compared with non-tumor cases. While several studies describe PJI in tumor-related arthroplasty, prospective studies comparing infection characteristics among different joints are limited. The present study analyzes mega-arthroplasty for hip, knee, and shoulder malignancy and compares the epidemiology, diagnosis, microbe spectrum, treatments, and outcomes between the different entities. Methods: The retrospective inclusion criteria were as follows: (1) mega-arthroplasty (2) in the hip, knee, or shoulder joint and a total femur arthroplasty (3) following a malignant bone tumor or metastasis (4) between 1996 and 2019. All included patients were prospectively followed and invited for a renewed hospital examination, and their PJI characteristics (if identified) were analyzed using both retrospective as well as newly gained prospective data. A PJI was defined according to the Infectious Disease Society of America (IDSA) and re-infection was defined according to the modified Delphi Consensus criteria. Results: In total, 83 cases of tumor mega-arthroplasty at a mean follow-up of 3.9 years could be included (32 knee, 30 hip, and 19 shoulder cases and 2 cases of total femur arthroplasty). In total, 14 PJIs were identified, with chondrosarcoma in 6 and osteosarcoma in 3 being the leading tumor entities. Knee arthroplasty demonstrated a significantly higher rate of PJI (p = 0.027) compared with hips (28.1% vs. 6.7%), while no significant difference could be found between the knee and shoulder (10.5%) (p = 0.134) or among shoulder and hip cases (p = 0.631). The average time of PJI following primary implantation was 141.4 months in knee patients, 64.6 in hip patients, and 8.2 months in shoulder patients. Age at the time of the primary PJI, as well as the time of the first PJI, did not show significant differences among the groups. Thirteen of the fourteen patients with PJI had a primary bone tumor. Statistical analysis showed a significant difference in the disadvantage of primary bone tumors (p = 0.11). While the overall cancer-related mortality in the knee PJI group (10%) was low, it was 50% in the hip and 100% in the shoulder group. Conclusion: The risk of PJI in knee tumor arthroplasty is significantly increased compared with hips, while cancer-related mortality is significantly higher in hip PJI cases. At the same time, mega-prostheses appear to be associated with a higher risk of infection due to a primary bone tumor compared with metastases. The study confirms existing knowledge concerning PJI in tumor arthroplasty, while, being one of the few studies to compare three different joints concerning PJI characteristics.
