Since Bauer  described transplantation of bones stored by refrigeration in 1910, fresh-frozen allografts have been used for reconstruction in patients with tumor resections for limb salvage. Although massive allograft reconstructions have been associated with considerable proportions of infection, fracture, and nonunion [1-5, 7-9, 11-19, 21-23], most published studies have reported that these complications occur in the first few years after reconstruction, and it is generally believed that after that time, allografts become relatively stable and more reliable .
Only two studies have evaluated allografts at a minimum of 10 years [19, 23]. Both studies evaluated osteoarticular allografts, and the results show a 10-year survival of 60% to 63%; the authors in both noted that fracture or joint resurfacing were the main causes of allograft removal. However, these studies were limited by the number of patients (less than 30) and included only osteoarticular allografts. Although one study  described one patient with allograft removal after 10 years, it was not clearly described what happened after 10 years of followup, mainly because of the small numbers of the patients in study. Because it is important to consider implant longevity when choosing reconstruction options, we wished to determine survivorship, complications, and factors associated with survivorship of reconstruction with massive bone allografts after bone tumor resection in patients who were followed for more than 10 years.
Therefore, we asked: (1) What is the survival of allografts free from removal, amputation, or joint replacement in patients treated for bone tumors in the lower limb with a minimum of 10 years of followup? (2) What complications occur after 10 or more years of followup? (3) Are there factors associated with allograft survival, such as age, sex, the affected bone, reconstruction type (intercalary or osteoarticular allograft), tumor type (malignant or benign), failure type, and chemotherapy use?
Patients and Methods
We retrospectively analyzed the records of 398 patients treated in one center by four orthopaedic surgeons (LAAT, MAA, GLF, DLM) for a benign or malignant bone tumor in the femur or tibia between 1986 and 2007. This study received approval from the institutional review board.
During the period in question, our general indications for using allografts (354 patients) included patients with benign or low-grade sarcomas and those with high-grade sarcomas with clinical and imaging response to neoadjuvant chemotherapy. Other approaches such as endoprostheses (44 patients) were used if the patient received radiotherapy, in patients with high-grade sarcomas without clinical and imaging response to neoadjuvant chemotherapy, or in patients with neurovascular tumor involvement. We excluded from the analysis 53 patients treated with allograft-prosthetic composites, 46 patients with hemicondylar osteoarticular allografts, and 57 patients with intercalary hemicylindrical allografts. This left 198 patients treated with segmental massive allografts in the long bones of the lower extremity (132 femurs and 66 tibias) after resection of a primary bone tumor, including 120 patients with osteoarticular and 78 patients with segmental intercalary allografts. Thirty-two (16%) of 198 patients died before 10 years, and the status of the graft was known in all of those patients; we included these patients because patient death is a competing risk for the calculation of the cumulative incidence of allograft removal or amputation (mean followup, 192 months; range 1–370 months). All the remaining 166 patients who were not known to have died before 10 years were accounted for at least 10 years after the allograft procedure (mean, 222 months; range, 120–370 months). No patient was lost to followup.
The mean age of the patients was 22 years (range, 2–55 years); 105 were male (53%) and 93 were female. Patients aged 16 years and younger were treated at the pediatric unit of the hospital, and those older than 16 years were treated in the adult unit. The predominant diagnoses were osteosarcoma (n = 125, 63%), giant cell tumor of bone (n = 27, 14%), and Ewing’s sarcoma (n = 19, 10%) (Table 1). In all, 146 patients (74%) underwent chemotherapy.
Surgical treatment consisted of en bloc resection of the primary bone sarcoma followed by reconstruction with a massive structural allograft. The types of allografts used for reconstruction were intercalary and osteoarticular. All allografts were maintained under sterile conditions and stored frozen at -80 °C in a bone bank that was established at our institution according to a previously described technique and guidelines . We did not attempt to preserve the viability of the articular cartilage, and we performed bacteriologic and viral studies in accordance with the recommendations of the American Association of Tissue Banks  and the tests available at the time. After tumor resection, the grafts were removed from their packaging and placed directly in warm, normal saline solution. After being thawed, the donor bone was cut to the proper length, and soft-tissue structures were prepared for implantation. We did not use intramedullary cement or vascularized fibula grafts in any of our patients. In osteoarticular allografts, the ligaments were reattached to the corresponding allograft tissues to improve stability with an absorbable suture. The allograft tissue was reattached to the host tissue through a direct lateral-lateral continuous absorbable suture. A transverse osteotomy was performed on every patient, and rigid fixation was used for stabilization. We used locking intramedullary nails in 31 patients, and we stabilized the remaining 167 patients with plates (120 with dynamic compression plates and 47 with locking compression plates).
We evaluated survivorship with the endpoint being freedom of allograft removal, joint replacement, or amputation. Orthopaedic surgeons (LAAT, GF) who were involved in clinical care interviewed all patients by telephone or at their latest followup examination. Functional evaluation of the patients was performed using the revised 30-point functional classification system established by the International Society of Limb Salvage and the Musculoskeletal Tumor Society (MSTS) . The functional score measured six parameters: pain, function, emotional acceptance, use of walking supports, walking ability, and gait. Each parameter is given a value ranging from 0 to 5, according to specific criteria. The individual scores are added together to obtain an overall functional score, with a maximum of 30 points, which then is expressed as a percentage of normal, with 30 points being defined as normal function. A score of 23 points or greater is considered an excellent functional result, 15 to 22 points is a good result, 8 to 14 points is a fair result, and less than 8 points is a poor result .
The risk factors for allograft survival such as age, sex, affected bone, type of reconstruction, type of tumor (malignant or benign), type of failure, and use of chemotherapy were calculated with the competitive risk analysis method, with limb salvage reconstruction censored at the time of failure or last followup. The statistical analysis was performed using the R programming language (R Foundation for Statistical Computing, Vienna, Austria) .
Fractures occurred in 29 patients (15%), infection in 27 patients (14%), nonunion in 23 patients (12%), and tumor recurrence in 13 patients (7%). At a mean followup of 192 months, 62 (31%) of 198 reconstructions resulted in allograft removal (43 patients), joint replacement (six patients) or amputation (13 patients). The risk of allograft removal was 36% at 5 years (95% CI, 30–43), 40% at 10 years (95% CI, 33–47), and 44% at 20 years (95% CI, 37–51) (Fig. 1). The predominant reasons for graft removal were fracture (n = 26, 13%), infection (n = 21, 11%), and tumor recurrence (n = 13, 7%); the remaining two allografts were removed because of joint instability and a second malignancy. Twenty-three patients (12%) had a nonunion that did not result in allograft removal; these patients were managed with bone autograft and augmented osteosynthesis. Three fractures in patients with tibial allografts did not result in allograft removal; the loose or broken internal fixation was removed and a new plate with screws and autologous iliac crest graft were added at the fracture site. Six patients with infection were treated with surgical débridement and antibiotics without allograft removal. In an analysis of the time of graft removal, 20 of 21 infections occurred in the first 3 years after surgery; only eight of 26 fractures occurred in the same period. All nonunions occurred before the first year after surgery. Fourteen of the 26 patients with fractures in whom the allograft was removed were reconstructed with a second allograft (an allograft composite prosthesis in one patient, osteoarticular allograft in three patients, and an intercalary segmental allograft in 10 patients), six patients were reconstructed with an endoprosthesis, and six other patients were reconstructed with joint resurfacing. Of the secondary reconstructions due to fracture, two patients with intercalary reconstructions had a second fracture and were reconstructed with an endoprosthesis, two patients with endoprostheses were removed and were revised with a second endoprosthesis, and two patients with joint resurfacing were revised with an endoprosthesis. Among the 21 patients with infections in whom the allograft was removed, one underwent amputation. The remaining 20 patients underwent temporary reconstruction with an antibiotic-loaded cement spacer; after the infections was resolved, they were reconstructed with either an endoprosthesis (seven patients), an allograft composite prosthesis (six patients), an osteoarticular allograft (two patients), an intercalary allograft (two patients), or an intercalary allograft arthrodesis (three patients). Of the 20 patients with infected allografts who had secondary reconstruction with a cement spacer, three had a secondary infection that resulted in amputation, and the remaining patients underwent reconstruction with an endoprosthesis. Eleven of the 13 patients with local recurrence who underwent allograft removal had an amputation. The other two patients were treated with resection of the recurrence and reconstruction with an endoprosthesis; one patient had a second local recurrence and underwent amputation, and the second patient was revised with a second endoprosthesis after the first fractured.
At 10 years after the allograft procedure 46 of the 166 living patients had an allograft removal (36 patients), joint replacement (six patients), or amputation (four patients); however, after 10 years six more patients underwent allograft removal (four for fracture, one for infection and one for joint instability) and another patient had an amputation for second malignancy (fibrosarcoma) and later died of their disease. All patients who underwent allograft removal were reconstructed with an endoprosthesis. Of the 113 surviving patients who retained the original allograft and were available for functional evaluation, we measured a mean functional score of 26 points (range, 14–30 points). Three patients used some kind of support to walk, and the other 110 patients were able to walk unaided.
After controlling for potentially confounding variables including death, we found no differences in the risk of allograft removal, joint replacement, or amputation when we analyzed age (older than 16 years, 117 patients [30%; 95% CI, 21–38]); 16 years and younger, 81 patients [33%, 95% CI, 23–44; p = 0.500], sex (female, 93 patients [28%, 95% CI, 19–37]; male, 105 patients [34%, 95% CI, 25–44; p = 0.280]), affected bone (femur, 132 patients [42%, 95% CI, 34–51]; tibia, 66 patients [47%, 95% CI, 35–59; p = 0.143]), type of reconstruction (intercalary, 78 patients [28%, 95% CI, 18–38]; osteoarticular, 120 patients [33%, 95% CI, 25–42; p = 0.290]), and chemotherapy use (without chemotherapy, 52 patients [21%, 95% CI, 10–33]; with chemotherapy, 146 patients [35%, 95% CI, 27–43; p = 0.067]). We found there was a higher risk of graft removal in reconstructions after resection of malignant tumors (166 patients) (at 5 years, 25% [95% CI, 18–33], 10 years, 30% [95% CI, 22–38], and at 20 years, 34% [95% CI, 26–42] ) than after benign tumors (32 patients) (9% at 5, 10, and 20 years [95% CI, 0–20]; p = 0.008).
We also found that there was a higher risk of removal of osteoarticular tibia allografts (58% at 5, 10, and 20 years [95% CI, 43–73]) than of osteoarticular femur allografts (29% at 5 years [95% CI, 18–39], 30% at 10 years [95% CI, 19–40], and 37% at 20 years [95% CI, 25–48]; p = 0.010) and tibial intercalary allografts (26% at 5, 10, and 20 years [95% CI, 7–45]; p = 0.020) (Fig. 2). Although we did not find differences between the risk of removal of femur intercalary allografts (35% at 5 years [95% CI, 22–48], 45% at 10 years [95% CI, 32–59], and 51% at 20 years [95% CI, 32–65]) and other reconstructions, we noted that the risk of removal increased over time (Fig. 2). Regarding the type of failure, fractures occurred more frequently in the femur (18% [95% CI, 11–25]) than in the tibia (5% [95% CI, 0–10]; p < 0.010), and infections were more common in the tibia (24% [95% CI, 14–35]) than in the femur (4% [95% CI, 0–8]; p < 0.001). With the number of patients we had, we found no difference in the occurrence of local recurrence in the tibia (12% [95% CI, 4–20]) compared with the femur (5% [95% CI, 1–9]; p < 0.053) (Fig. 3).
Massive bone allografts after bone tumor resection have been used for more than a century . The use of allografts to reconstruct tumor defects has been reported to have acceptable results regarding allograft survival and functional scores [3, 5, 12, 14]. However, long-term studies analyzing late complications are scarce [19, 23]. We found that 56% of patients can expect to have their grafts survive for more than 10 years, but complications that result in graft removal do occur, even after 10 years. The most common causes of removal of the grafts were infection, fracture, and local recurrence.
There were several limitations and biases in this study. There are selection biases that could affect the results. In this study, those patients with high-grade sarcomas without clinical and imaging response to neoadjuvant chemotherapy or with neurovascular tumor involvement or receiving radiotherapy, were selected to be reconstructed with endoprosthesis. The benign lesions treated with bone allografts generally underwent less soft tissue resection and did not receive chemotherapy, which could explain the differences in allograft survival compared with the patients with malignant tumors. However, to have a more homogenous group, we excluded small resections (reconstructed with hemicondylar osteoarticular allografts and intercalary hemicylindrical allografts) and upper limb allografts. This allowed us to have massive bone allografts of weightbearing limbs, which are more susceptible to complications. Although we performed functional MSTS evaluation in this study, there are potential assessment biases. We based our analysis on the last followup scores, and we did not analyze whether they were stable or worsening. We also did not evaluate the final function of the patients who were converted to other types of reconstruction after removal (for example, an endoprosthesis) so we do not know whether or not they improved in the functional results. However, we evaluated all patients with preserved grafts, and the mean functional score was satisfactory at last followup. The small numbers in each subgroup of patients analyzed for risk factors (age, sex, the affected bone, reconstruction type (intercalary or osteoarticular allograft), tumor type (malignant or benign), failure type, and chemotherapy use) did not allow sufficient power to explore differences and we did not perform a power analyses; however, in some groups the differences were so remarkable that this will allow future studies to explore other risk factors that were not analyzed in this series (such as fracture in femur). Another limitation is that all procedures were performed at the same center, and we did not compare the grafts we used with other types of reconstructions. However, because of our years of expertise in these techniques, it would be difficult to compare substantially different techniques at the same institution with other types of reconstruction. Lastly, because of the long period of analysis (20 years) the group of patients who were studied have some inherent heterogeneity in terms of diagnosis, the amount of soft-tissue resection, type of internal fixation, and extent of resection, which could have affected the incidence of failures and complications, as well as functional results.
The risk of allograft removal, joint replacement, and amputation was 36% at 5 years, 40% at 10 years, and 44% at 20 years. These allograft removal proportions are similar to those previously reported [5, 17, 14, 19, 23 ]. A systematic review of 31 studies reported allograft removal (partial or complete) or allograft conversion to an allograft-prosthetic composite in 27% of patients . However, most published studies reported series with a minimum followup between 2 and 5 years [5, 7, 14]. Two studies reported allografts with a minimum of 10 years of followup [19, 23]. Ogilvie et al.  described that 60% of the 20 analyzed osteoarticular allografts (14 of the lower limb) were removed. Toy et al.  analyzed 26 distal femur osteoarticular allografts with a minimum of 10 years of followup, and they found that 11 patients retained their original osteoarticular allograft and nine were converted to allograft-prosthetic composites. These studies found that the patients retained a functional limb at last followup and patients with intact grafts were comparable to patients with endoprostheses reported in the evidence .
We observed that six patients underwent allograft removal and one had an amputation after 10 years of followup. The grafts were removed primarily because of mechanical failure (four fractures and one instability). All six of our patients with nononcological complications were reconstructed with endoprostheses when the failure occurred after 10 years; these could be because endoprostheses have replaced allografts as an option for a second reconstruction after allograft removal in our unit in patients with longer followup.
After controlling for potentially confounding variables, we found that there was a higher risk of graft removal in reconstructions after malignant tumor resection (166 patients) than after benign tumors (32 patients). In a series of 718 patients, Mankin et al.  stated that high-grade tumors have more complications than low-grade and benign tumors. This finding could be because more aggressive resection was used in patients with high-grade tumors. We also noted that graft removal due to fracture increased over time (Fig. 3); thus, surgeons should be aware of this complication, especially in the femur (more in the intercalary), when discussing reconstruction options with patients. The addition of a vascularized fibular graft to a massive allograft was described to lessen the impact of this complication [8, 12]. However, in a recent study  of 23 intercalary reconstructions in the femur with a mean followup of 141 months (range, 24–313 months), five patients (21.7%) had a fracture of the reconstruction despite the use of a vascularized fibula graft to augment the allograft. Frisoni et al.  showed that there was little difference in the percentage of fractures when comparing allografts with and without this vascularized graft. In the same study , protecting the allograft with intramedullary cement did not substantially decrease the allograft fracture rate. A systematic review  found that there was a higher fracture risk in the proximal humerus (odds ratio [OR], 4.1; 95% CI, 2.2–7.7; p < 0.001) and proximal tibia (OR, 2.2; 95% CI 1.3–4.4; p < 0.005) than in our study. This difference may be because the minimum followup period for the patients who were analyzed in that study was short; in comparison, most of the patients with fractures in our study occurred after 5 years of followup. Infections occurred more frequently in the tibia than in the femur. However, infections are more likely to occur in the first years after reconstruction. A recent study  analyzing the risk factors of infection in 673 patients who underwent reconstruction with massive bone allografts found that the tibia was associated with a greater risk of infection than other long bone (femur, humerus, or radius) (OR, 3.17; 95% CI, 1.80–5.60; p < 0.001). This finding is similar to the findings of the analysis performed by Bus et al. , in which the infection risk was higher in patients with osteoarticular allografts of the proximal tibia than osteoarticular allografts of distal femur, distal radius, or proximal humerus (OR, 2.2, 95% CI, 1.1–4.3). We found there was a higher risk of graft removal in osteoarticular allografts of the proximal tibia than in osteoarticular allografts of the distal femur and intercalary tibia allografts. Previous studies (9, 12, 18] reported survival rates ranging from 56% to 66%. Currently, we limit the indication of proximal tibia osteoarticular allografts to pediatric patients because of this high complication proportion.
In conclusion, infections were the most common complication associated with allograft removal in the first 2 to 3 years after reconstruction and were more frequently associated with tibial allograft removal; fractures were more commonly associated with graft removal with longer term followup and were more frequently associated with femoral allograft removal. Future studies in femoral allograft with longer followup should be performed to analyze the factors that could explain these findings, such as percent of the length of the bone resected, type and number of plates and screws used, and type of fixation (rods versus plates). We found there was a higher incidence of graft removal in patients with proximal tibia osteoarticular allografts; as a result, we currently use this reconstruction type only in pediatric patients. In our practice, we use endoprostheses for secondary reconstructions due to allograft removal after 10 years of followup. We do not perform a second allograft, but we do not have evidence support this preference.
We thank D. Luis Muscolo MD, of the Department of Orthopedic Surgery, Hospital Italiano de Buenos Aires, Buenos Aires, Argentina, who was the surgeon of cases in this series of patients.
1. Aponte-Tinao LA, Ayerza MA, Muscolo DL, Farfalli GL. What are the risk factors and management options for infection after reconstruction with massive bone allografts? Clin Orthop Relat Res. 2016;474:669-673.
2. Aponte-Tinao L, Farfalli GL, Ritacco LE, Ayerza MA, Muscolo DL. Intercalary femur allografts are an acceptable alternative after tumor resection. Clin Orthop Relat Res. 2012;470:728-734.
3. Aponte-Tinao LA, Ritacco LE, Albergo JI, Ayerza MA, Muscolo DL, Farfalli GL. The principles and applications of fresh frozen allografts to bone and joint reconstruction. Orthop Clin North Am. 2014;45:257-269.
4. Bus MP, Dijkstra PD, van de Sande MA, Taminiau AH, Schreuder HW, Jutte PC, van der Geest IC, Schaap GR, Bramer JA. Intercalary allograft reconstructions following resection of primary bone tumors: a nationwide multicenter study. J Bone Joint Surg Am. 2014;96:e26.
5. Bus MP, van de Sande MA, Taminiau AH, Dijkstra PD. Is there still a role for osteoarticular allograft reconstruction in musculoskeletal tumour surgery? A long-term follow-up study of 38 patients and systematic review of the literature. Bone Joint J. 2017;99:522-530.
6. Bauer H. [About bone grafting] [In German]. Zentralbl Chir.
7. Campanacci L, Manfrini M, Colangeli M, Alí N, Mercuri M. Long-term results in children with massive bone osteoarticular allografts of the knee for high-grade osteosarcoma. J Pediatr Orthop. 2010;30:919-927.
8. Campanacci DA, Totti F, Puccini S, Beltrami G, Scoccianti G, Delcroix L, Innocenti M, Capanna R. Intercalary reconstruction of the femur after tumour resection. Bone Joint J. 2018;100:378-386.
9. Clohisy DR, Mankin HJ. Osteoarticular allografts for reconstruction after resection of a musculoskeletal tumor in the proximal end of the tibia. J Bone Joint Surg Am. 1994;76:549-554.
10. Enneking WF, Dunham W, Gebhardt MC, Malawer M, Pritchard DJ. A system for the functional evaluation of reconstructive procedures after surgical treatment of tumors of the musculoskeletal system. Clin Orthop Relat Res. 1993;286:241–246.
11. Farfalli GL, Aponte-Tinao L, Lopez-Millán L, Ayerza MA, Muscolo DL. Clinical and functional outcomes of tibial intercalary allografts after tumor resection. Orthopedics. 2012;35:e391-396.
12. Frisoni T, Cevolani L, Giorgini A, Dozza B, Donati DM. Factors affecting outcome of massive intercalary bone allografts in the treatment of tumours of the femur. J Bone Joint Surg Br. 2012;94:836-941.
13. Hornicek FJ, Mnaymneh W, Lackman MD, Exner GU, Malinin TI. Limb salvage with osteoarticular allografts after resection of proximal tibia bone tumors. Clin Orthop Relat Res. 1998;352:179-185.
14. Mankin HJ, Gebhardt MC, Jennings LC, Springfield DS, Tomford WW. Long-term results of allograft replacement in the management of bone tumors. Clin Orthop Relat Res. 1996;324:86-97.
15. Mankin HJ, Hornicek FJ, Raskin KA. Infection in massive bone allografts. Clin Orthop Relat Res. 2005;432:210-216.
16. Muscolo DL, Ayerza MA, Aponte-Tinao L, Ranalletta M, Abalo E. Intercalary femur and tibia segmental allografts provide an acceptable alternative in reconstructing tumor resections. Clin Orthop Relat Res. 2004:97-102.
17. Muscolo DL, Ayerza MA, Aponte-Tinao LA, Ranalletta M. The use of distal femoral osteoarticular allograft in limb salvage surgery. J Bone Joint Surg Am. 2005;87:2449-2455.
18. Muscolo DL, Ayerza MA, Farfalli G, Aponte-Tinao LA. Proximal tibia osteoarticular allografts in tumor limb salvage surgery. Clin Orthop Relat Res. 2010;468:1396-1404.
19. Ogilvie CM, Crawford EA, Hosalkar HS, King JJ, Lackman RD. Long-term results for limb salvage with osteoarticular allograft reconstruction. Clin Orthop Relat Res. 2009;467:2685-2690.
20. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria; 2014. Available at: http://www.Rproject.org
. Accessed July 20, 2017.
21. Sorger JI, Hornicek FJ, Zavatta M, Menzner JP, Gebhardt MC, Tomford WW, Mankin HJ. Allograft fractures revisited. Clin Orthop Relat Res. 2001:66-74.
22. Thompson RC Jr, Garg A, Clohisy DR, Cheng EY. Fractures in large-segment allografts. Clin Orthop Relat Res. 2000;370:227-235.
23. Toy PC, White JR, Scarborough MT, Enneking WF, Gibbs CP. Distal femoral osteoarticular allografts: long-term survival, but frequent complications. Clin Orthop Relat Res.