Bone tumors in children are rare and prior to the use of effective chemotherapy, overall patient survival rates were reported to be 15–20% at 2-years following surgical resection and/or radiotherapy [26, 27]. This study demonstrated a patient survival rate of 75.9% over a mean follow up of 4.25 years. Although patient survival is highly dependent on the stage of osteosarcoma at diagnosis, our result is similar to other recent studies who report contemporary 5-year patient survival rates ranging between 60–78% in pediatric patients following limb-salvage and MAP treatment [28–31]. Thus, the challenges of DFR longevity remains a priority as overall survival and activity levels continues to improve however, surgery is challenging in growing children and problems can result in loss of joint function, high-level amputation, and systemic sequelae for the patient . Aseptic loosening in young and physically active patients who place high demands on their prosthesis is a major concern . A study by Unwin et al.  reported a 67.4% probability of a Stanmore® DFR survival at 10 years with a significantly higher risk of ASL (13.6%) in patients < 20-years of age. Further, this study also identified that patients < 20-years of age and with > 60% of bone resection, having the poorest prognosis. In this study, an encouraging overall implant survival rate of 82.8% was found and the incidence of ASL that required revision was 10.3%. No correlation between %bone resection and implant failure was seen although mean levels were less than 60%. To determine load distribution within the intramedullary fixation in adults, a clinical study by Taylor et al.  added strain gauges and telemetric instrumentation to a massive implant. At 100 weeks post-surgery, 60% of the applied load was directed through the cemented stem fixation when compared with 25% in the more immediate post-op period. These findings suggested a progressive mechanical cause of ASL and led to the concept that osseointegration at the shoulder offered more beneficial load distributions. As such, the Stanmore® and CPS® implants were designed to maximize osseous growth at the shoulder, and in this study, none of these implants failed due to ASL. Hydroxyapatite is classified as a bioactive, osseoconductive and osseoinductive material and bone is able to chemically bond with it providing increased interfacial and mechanical coupling, to superior levels when compared with a polished titanium implant surface [35, 36]. Our results showed significantly increased extracortical bone growth and osseointegration to the HA collar in the Stanmore® group when compared with both the GMRS® and Repiphysis® implants. Significantly reduced cortical loss and the progression of radiolucent lines was also evident in Stanmore-given patients over the follow up period. These results are similar to other studies that investigated osseointegration and ASL [10, 13, 14]. This study demonstrated poor performance of the Repiphysis® design where 2 of the 3 implants inserted were revised. Results are similar to other studies that report high rates of ASL as well as mechanical failure in young patients [37–41]. Two recent studies also reported a 100% implant survival rate of the Stanmore® implant in pediatric patients and both demonstrated overall poor survival (79.2% at 2-years and 21% at 5-years) of the Phenix-Rephiphysis® implant at a mean follow-up of 6.2 years  and 32% survival at 6-years .
Infection of massive endoprostheses ranges between 8–40% [44, 45] with CPS® implant infection reported as 14% over a 20-year follow-up . In this study, none of the DFR implants were revised due to infection and 3 patients were successfully treated for implant-associated infection. Two of these 3 patients had received a CPS® implant, however this group of patients also experienced significantly higher %bone resection levels and the increased tissue exposed during surgery may account for the infections observed. Two CPS® implants failed due to fracture of the titanium traction bar. In both patients the implants appeared radiographically well fixed. Traction bar fracture has been reported in the same location in other studies [47, 48] however, the reason for fracture remains unclear. Nevertheless, our results indicate that the CPS® implant continues to be a reliable option for distal femoral limb salvage surgery and the absence of ASL is encouraging. Finally, multidrug chemotherapy impairs bone growth and causes early radiological signs of loosening in DFRs . No significant differences were found when the total dose and length of treatment was compared between implant groups.
Our study had several limitations. First, osteosarcoma is rare and as such, the study is limited by its small sample size as well as loss of follow-up as patients transitioned out of the hospital system and into adult care. Furthermore, the cohort of patients presented individual differences in their activity levels, which would impact prosthetic survival. Because this study was a retrospective study, both AP and ML radiographs were not always available for review and this reduced the number of patients followed-up beyond 6-years post-operatively.
In conclusion, chemotherapy and limb-salvage surgery yield good oncologic outcomes. Results from this study suggest that implant designs modified to augment osseointegration at the bone-implant shoulder may be critical in reducing the initiation and development of ASL in this vulnerable patient group. While the limitations of this study do not allow us to conclude that extracortical bone growth and osseointegration is directly responsible for a lower incidence of ASL, our results do confirm the ability of the HA collar to increase radiographic bone-implant contact at the implant shoulder. Further studies that involve a larger cohort over a longer follow-up are needed to confirm these preliminary findings.