Abstract
BACKGROUND: Proximal femur reconstruction after bone tumor resection in children is a demanding surgery for orthopaedic oncologists because of the small bone size and possible limb-length discrepancy at the end of skeletal growth owing to physis loss. The most commonly used reconstruction types used for the proximal femur are modular prostheses and allograft-prosthesis composites. To our knowledge, there are no previous studies comparing the outcomes after modular prosthesis and allograft-prosthesis composite reconstruction of the proximal femur in children with primary bone tumors. QUESTIONS/PURPOSES: (1) What was the cumulative incidence of reoperation for any reason after allograft-prosthesis composite and modular prosthesis reconstructions of the proximal femur in children with primary bone tumors? (2) What was the cumulative incidence of reconstruction removal or revision arthroplasty in those two treatment groups? (3) What complications occurred in those two treatment groups that were managed without further surgery or with surgery without reconstruction removal? METHODS: Between 2000 and 2021, 54 children with primary bone tumors underwent resection and reconstruction of the proximal femur at a single institution. During that time, allograft-prosthesis composite reconstruction was used in very young children, in whom we prioritize bone stock preservation for future surgeries, and children with good response to chemotherapy, while modular prosthesis reconstruction was used in older children and children with metastatic disease at presentation and poor response to chemotherapy. We excluded three children in whom limb salvage was not possible and 11 children who underwent either reconstruction with free vascularized fibular graft and massive bone allograft (n = 3), an expandable prosthesis (n = 3), a massive bone allograft reconstruction (n = 2), a rotationplasty (n = 1), standard (nonmodular) prosthesis (n = 1), or revision of preexisting reconstruction (n = 1). Further, we excluded two children who were not treated surgically, three children with no medical or imaging records, and three children with no follow-up. All the remaining 32 children with reconstruction of the proximal femur (12 children treated with modular prosthesis and 20 children treated with allograft-prosthesis composite reconstruction) were accounted for at a minimum follow-up time of 2 years. Children in the allograft-prosthesis group were younger at the time of diagnosis than those in the modular prosthesis group (median 8 years [range 1 to 16 years] versus 15 years [range 9 to 17 years]; p = 0.001]), and the follow-up in the allograft-prosthesis composite group was longer (median 5 years [range 1 to 23 years] versus 3 years [range 1 to 15 years]; p = 0.37). Reconstruction with hemiarthroplasty was performed in 19 of 20 children in the allograft-prosthesis composite group and in 9 of 12 children in the modular prosthesis group. A bipolar head was used in 16 of 19 children, and a femoral ceramic head without acetabular cup was used in 3 of 19 children in the allograft-prosthesis composite reconstruction group. All 9 children in the modular prosthesis group were reconstructed with a bipolar hemiarthroplasty. Reconstruction with total arthroplasty was performed in one child in the allograft-prosthesis composite group and in three children in the modular prosthesis group. For both groups, we calculated the cumulative incidence of reoperation for any reason and the cumulative incidence of reconstruction removal or revision arthroplasty; we also reported qualitative descriptions of serious complications treated nonoperatively in both groups. RESULTS: The cumulative incidence of any reoperation at 10 years did not differ between the groups with the numbers available (36% [95% confidence interval 15% to 58%] in the allograft-prosthesis composite group versus 28% [95% CI 5% to 58%] in the modular proximal femoral replacement group). The cumulative incidence of reconstruction removal or revision arthroplasty at 10 years likewise did not differ between the groups with the numbers available (10% [95% CI 2% to 28%] versus 12% [95% CI 0% to 45%], respectively). In the allograft-prosthesis composite group (20 children), hip instability (n = 3), nonunion (n = 2), fracture of the greater trochanter (n = 1), screw loosening (n = 1), limb-length discrepancy (n = 1), and coxalgia due to acetabular wear (n = 1) were treated surgically without reconstruction removal. Complications treated without surgery included resorption of the allograft at the trochanteric region (n = 4), fracture of the greater trochanter (n = 4), limb-length discrepancy (n = 6), and coxalgia due to acetabular wear (n = 2). In the modular prosthesis group (12 children), hip instability (n = 1), coxalgia due to acetabular wear (n = 1), and limb-length discrepancy (n = 1) were treated surgically without reconstruction removal. Complications treated without surgery included hip instability (n = 2), stress shielding (n = 6), infection (n = 1), sciatic nerve palsy (n = 1), and limb-length discrepancy (n = 3). CONCLUSION: Although the two groups of children were not directly comparable due to differences in age and clinical characteristics, both modular prosthesis and allograft-prosthesis composite reconstructions of the proximal femur after bone tumor resection appear to be reasonable options with similar revision-free survival and complications. Therefore, the type of reconstruction following proximal resection in children with bone sarcoma should be chosen taking into consideration factors such as patient age, bone size, implant availability, technical expertise, and the surgeon's preference. Although children treated with expandable prostheses were not included in this study, such prostheses may be useful in bridging the surgical defect while correcting residual limb-length discrepancies even though they face limitations such as small intramedullary diameter, short residual bone segments, as well as stress shielding, loosening, and breakage. LEVEL OF EVIDENCE: Level III, therapeutic study.