Osseous defects from the resection of a malignant or aggressive extremity tumor typically require reconstruction. Reconstruction options include structural allografts, allo- prosthetic composites, rotationplasty, or megaprostheses. These implants are advantageous because they are widely available, offer a durable and stable reconstruction, allow maintenance of joint range of motion, and permit the patient to resume chemotherapy in a timely fashion.1,6,7,11,12,14-16,19,22,23 Many patients outlive their implant because modern chemotherapy allows a longer survival after treatment.5,8,14 As a result, patients may endure multiple procedures to treat prosthetic complications and failures.
Although limb salvage surgery is an effective procedure with a low risk of local recurrence,9,11,19 the results for implant survival and effectiveness are more challenging. Reported failure rates for tumor megaprosthesis have varied in the literature.3,5,8,9,11,19,22 Roberts et al,15 Malawer et al,11 and Eckardt et al2 described large series of implants for tumors of the distal femur and proximal tibia. Implant failure rates were reported as 25%, 17%, and 20% at 5 years, respectively, and 41% and 33% at 10 years by Roberts et al15 and Malawer et al.11 Reports by Kawai et al8 described a 5-year failure rate of 33% and 10-year failure rate of 52%, and Frink et al3 reported a 14% failure rate at year 5 with no additional failures at year 10 (Table 1). Reports in children by Wilkins et al20 describe an estimated 20% failure rate at 5 years and 30% at 10 years. These publications investigating the incidence of implant failure and revision have cited variable 5-year failure rates of 14% to 33% for implants surrounding the knee7,11,22 and one recent series reporting a failure rate of 36% in children without a described followup interval.21 These incidences of failure are cause for concern because of their magnitude and possible application to the pediatric group. These series have also addressed the causes of failure including implant design (ie, rotating hinge, stem design, and component material wear), operative technique, stem fixation and/or the comorbidities of chemotherapy and radiation therapy.1,3,9 Several important issues not addressed in prior studies include the early incidence and timing of implant failure and whether these failures occur more frequently in children. Aseptic loosening is the most common cause of implant failure in joint implants for reasons other than tumors and we presumed that would be the case with tumors.
Given the variable rate of implant failure in the literature, our primary question was to confirm the rate of implant failure surrounding the knee. We hypothesized that 5- and 10-year implant failure rates are higher than previously described, that the failure rate increases as time passes from the index procedure, and most failures occur within the first 5 years of the index procedure. We also hypothesized aseptic loosening would be the leading cause of implant failure in this series. Lastly, we hypothesized pediatric patients (< 18 years of age at their index procedure) would have a higher failure rate than older patients.
MATERIALS AND METHODS
We retrospectively reviewed 105 patients (index group) who received a distal femoral (DF) or proximal tibial (PT) modular prosthesis from January 1985 to February 2004. Thirty-eight of the 105 patients were under 18 years of age, while 67 were 18 or older. Patients were included if they received their implant after resection of a malignant bone tumor, soft tissue malignancy with osseous involvement, or benign aggressive bone tumor. We included patients (n = 6) who had their initial implant placed at an outside institution, but received the majority of their followup and all of their revision procedures at our institution. Patients who died less than 2 years postoperatively were included because they had intact prostheses at the time of death (n = 16). In the index group, 18% of patients (n = 19) received allo- grafts as their primary procedure and were converted to implants when their allografts failed. Patients were excluded (n = 98) if they received their implant for a nontumorous condition (trauma, failed total knee arthroplasty [TKA], or rheumatoid arthritis [RA]) (n = 7), the location of their implant was not about the knee (n = 64), they were lost to followup (n = 15), or had an index procedure after February 2004 (n = 12) (Fig 1).
Resection procedures were performed by one of two surgeons (EC, JB) through a medial or lateral parapatellar approach for an intraarticular resection. The mean bony resection length was 15 cm (range, 5-39 cm) in the 93 of 105 index patients with documented resection length. The tumor margins were assessed by routine pathologic evaluation of the resected specimen.
All distal femoral and proximal tibial prostheses had a rotating hinge design. Most patients (85%; 88 of 105 patients) received Modular Replacement System (MRS) implants (Howmedica, Rutherford, NJ), 12% (12 of 105 patients) of patients received Orthopaedic Salvage System (OSS) implants (Biomet, Warsaw, IN), and 2% (two of 105 patients) received a Lewis Expandable Adjustable Prosthesis ([LEAP], Wright Medical Technology, Arlington, TN). Two patients received other types of prostheses, including a Repiphysis expandable implant (Wright Medical Technology), and a nonmodular custom rotating hinge implant (formerly Osteonics, now Stryker Orthopaedics, Mahwah, NJ) (Table 2). Femoral components were cemented in all but one 15-year-old female patient whose stem was press-fit. The tibial components were cemented in all patients. Since 1996, our surgical technique has included extracortical bone grafting (n = 70) performed with local autologous bone at the junction of the patient host bone and the porous coated collar of the implant to enhance bony bridging between native bone and the prosthesis.
Most implants were distal femoral (76 of 105 patients; 72%) and 28% were proximal tibial implants (29 of 105 patients). The mean age of patients at resection in the index group was 33 years (9-86 years), and 36% (38 of 105 patients) were pediatric patients (< 18 years of age). Fifty-five percent (58 of 105) of patients were male (Table 2). The mean followup for the index group was 66 months (1-235 months). Patients with less than 24 months of followup were those who died within 2 years after receiving the implant. If this group was excluded the mean followup increased to 81 months (24-235 months). In this index group 13% (14 of 105 patients) had greater than 10 years of followup.
Implant failure was defined as a failure that necessitated a femoral and/or tibial component revision. Two patients underwent revision because of local recurrence, but they were not included in the revision analysis given our focus on failure attributable to mechanical or surgical factors. The mean age of patients requiring a revision was 29 years (9-78 years) and 50% were younger than 18 years of age. Similar to the index group, 56% of the revision group was male (18 of 32 patients). Nine of 32 patients (28%) with implant failure received a distal femoral or proximal tibial allograft before receiving limb salvage reconstruction with an implant (Table 2). The mean followup for patients requiring a revision (32 of 105 patients; 30%) was 95 months (4-235 months). The mean followup of the revision group increased to 103 months if deceased patients (seven of 32 patients; 22%) were excluded from the analysis, and 15% (five of 32) patients whose implants were revised had more than 10 years of followup.
Patients were routinely evaluated clinically and radiographically at 6 weeks (postoperatively), 6 months, and annually thereafter. More frequent followup occurred for the assessment of the implant or extremity pain. Each patient also had pertinent anteroposterior (AP) and lateral radiographs of their affected extremity at 6 weeks, 6 months, and annually after surgery.
Implant survival (time from date of index procedure to date of first revision) was estimated using Kaplan-Meier survival analysis, and the survivorship curves between groups were compared with the log rank test. Two sample Student's t-tests were used to compare quantitative variables where the data represented a normal distribution, and categorical data were compared using the chi square test.
Thirty two patients in the index group of 105 patients underwent 42 revisions for implant failure. The incidence of overall implant failure was 36.5% (42 revised components in 115 implants), requiring 32 primary prosthetic revision procedures. Of the 32 primary failed implants, 10 failed a second time requiring a second revision procedure.
Assessment of the timing of failure (Table 3) demonstrated implant failures as follows: three of 32 failures (9%) were revised within 1 year postoperatively; 15 of 32 failures (47%) in year 2; 18 of 32 failures (56%) in 3 years; and 22 of 32 implant failures (69%) were revised within 5 years of the index procedure.
The Kaplan-Meier prosthetic survivorship of the index group (having no femoral, tibial, or both components revised) was 84% at 2 years, 73% at 5 years, and 59% at 10 years, with a mean prosthetic survival time of 123 months (3-217 months) (Fig 2). Proximal tibial implants had a longer (p = 0.03) survival time than distal femoral implants (67 months versus 47 months) (Fig 3). In the previously revised implant group, 10 of the 32 index failures (31%) failed a second time requiring a second revision procedure. The Kaplan-Meier prosthetic survivorship of this group (the survival of the implant after having one revision and not requiring a second or more revisions) was 86% at 2 years, 58% at 5 years, and 44% at 10 years, with a mean prosthetic survival time of 60 months (0.7-217 months) (Fig 4).
The most common cause of failure was aseptic loosening (18 of 32 patients; 56%) (Table 4), requiring the revision of 12 femoral components and three tibial components. Three patients required the revision of both components. Other causes of implant failures included stem fracture (five of 32 patients; 16%), septic loosening (three of 32 patients; 9%), lower extremity malrotation (two of 32 patients; 6%), axle fracture, femoral condyle fracture, implant disengagement at the stem-body junction, peri- prosthetic fracture, stiffness, and a painful implant each leading to one revision (Table 4). Similar to the patients who only had one revision, aseptic loosening was the most common reason for a second revision (six of 10 secondary failures; 60%). The other reasons for a second revision procedure were septic loosening (two of 10 secondary failures; 20%), body fracture, and implant disengagement (Table 4).
When compared to the entire cohort, young patients (< 18 years) had a higher (p = 0.05) revision rate than older patients (42% versus 24%, respectively). The Kaplan-Meier prosthetic survivorship in children was 89% at 2 years, 69% at 5 years, and 44% at 10 years, with a mean prosthetic survival time of 101 months (8-53 months) (Fig 5). Pediatric patients (< 18 years) also had an increased (p = 0.005) rate of deep infection compared with adult patients (85% versus 14%). Gender, length of resection, diagnosis (osteosarcoma versus other), and the use of allograft for reconstruction before implant placement did not affect index group implant survival. Patient comorbidities included a history of chemotherapy or radiation therapy, which did not affect revision rates. There was no difference identified between the manufacturer types of implant that failed (p = 0.23).
In the index group of 105 patients, 10% of patients (11 of 105 patients) had arthrofibrosis. Three patients had persistent postoperative nerve palsies develop. Patients with patellar instability (two of 32 patients; 6%) required two patellar revisions, and patients with bushing failures (three of 32 patients; 9%) required two bushing-repair procedures. The other patient received a total revision because of mechanical failure of the prosthesis. There was one amputation for a painful, arthrofibrotic knee. There were seven patients in the index group (seven of 105; 7%) who had a deep infection develop, all of whom were in the revision group. In addition, five of the seven patients (71%) required irrigation and débridement procedures and three required staged total prosthetic revisions. The other two patients required chronic suppressive antibiotics. One of these patients died of disease and the other had apparent infection resolution after 3 months of intravenous antibiotics.
Our primary study goal was to evaluate the failure rate for modular tumor implants located in the DF and PT as defined by the need for revision. Our concern regarding this issue resulted from the variable incidence of implant revision in the literature and the observation of implant failures within our practice (Table 1). In addition, we were concerned about the cause and early appearance of failures within the first 2 years (47% at 2 years and 56% at 3 years) and what seemed to be a high failure rate in children (47%) in this series.
We note several limitations. Although the surgery was conducted by primarily one surgeon, surgical technique was not strictly controlled and there was potential selection bias in regard to diagnostic categories and size of resection. We did not control for the use of different implant manufacturers, implant design, and stem-femur geometry in this study. Patient age, body mass, and other comorbidities (such as the use of neoadjuvant treatment) also may affect the validity and reproducibility of these results. Nonetheless, our incidence of implant failure and 5- and 10-year implant survival is consistent with other published series and represents a valid finding.3,9 Further, our study of implants for tumors about the knee represents a relatively large series with intermediate followup.
The overall incidence of failure in our series was 36.5% (42/115). For primary failures the incidence was slightly less at 30% (32/105), and the incidence of second failures was 31% (10/32). Our incidence is higher than some of the more recently reported implant failures in the literature,2,3,23 but is consistent with studies published by Roberts et al,15 Malawer and Chou,11 and Kawai et al.9 We believe this higher incidence is a more accurate reflection of the true incidence of implant failure. Series that have quoted a lower incidence of failure have included other anatomic sites, such as the proximal femur, which has a lower failure rate, or have involved smaller cohorts that may not accurately reflect implant failure. Variations in patient survival in other series may result in falsely elevated implant survival secondary to shorter patient followup. Our previous concern that more recent published series reported a higher implant failure rate was not substantiated.
The primary cause of failure in most series is aseptic loosening (Figure 6), which was also corroborated in this study.8,11,22 Mechanical failure or implant defect occurred in 21% (9/42) failures and stem fractures were the second highest reason for implant defects. Complications from infection were low, but there was a higher incidence in children (85%, 6/7). We suggest this may be due to intense chemotherapy regimens and/or nutritional issues in children while receiving chemotherapy. We remain interested in the etiology of aseptic loosening as a cause of failure and question whether this is attributable to the inadequacy of stem fixation as a consequence of stem design, cementation technique, or rotational stresses throughout the implant or implant interface.5,8,14,18 More information regarding the radiographic assessment of implants may be helpful in addressing these issues.
Our original concern with timing of failure has been validated. The hypothesis the majority of these implants fail within the first 2 to 3 years was confirmed by our review (47% at year 2 and 56% by year 3). Other published series have not discussed the timing of failure. The relatively high rate of aseptic loosening suggest designers should focus on minimizing this complication. Careful radiographic followup is more important for these patients in the first 2 to 3 years postoperatively. Our observation of progressive failure is consistent with previously reported series,4,9,14 and recognized patterns of failure for total hip arthroplasty and total knee arthroplasty.10,17
Our patient series seems to demonstrate a higher implant failure rate in children (42% in children versus 24% in adults; p = 0.05). Although this incidence was higher than that reported by Wilkins et al20 and Wilkins and Miller21 this confirms children have a higher failure rate than adults. These findings have also confirmed the need to focus on implant failure in the pediatric population. This population is living longer because of the success of chemotherapy, and a higher rate of failure would be associated with greater morbidity. A high failure rate in children may be related to greater fixation in a younger, more active population. Poor stem fit or design for the smaller femur or tibia of a child is another possible explanation for a higher failure rate in children. Use of uncemented stems in this population may lower failure rates by providing superior fixation through bony ingrowth.13
We have confirmed the incidence of failure in tumor megaprostheses similar to early published literature, and the incidence of these failures is highest within the first 3 years. The etiology of most failures (aseptic loosening) is unexplained and children are considered at greater risk for failure. Improvement in lowering the failure rate has major implications, especially in the pediatric population.
We would like to acknowledge James Bruckner, MD (JB), who was the primary surgeon on several of the patients included in this analysis.
1. Choong PF, Sim FH, Pritchard DJ, Rock MG, Chao EY. Megapros- theses after resection of distal femoral tumors. A rotating hinge design in 30 patients followed for 2-7 years. Acta Orthop Scand
2. Eckardt JJ, Eilber FR, Dorey FJ, Mirra JM. The UCLA experience in limb salvage surgery for malignant tumors. Orthopedics
. 1985; 8:612-621.
3. Frink SJ, Rutledge J, Lewis VO, Lin PP, Yasko AW. Favorable long-term results of prosthetic arthroplasty of the knee for distal femur neoplasms. Clin Orthop Relat Res
4. Gruen TA, McNeice GM, Amstutz HC. “Modes of failure” of cemented stem-type femoral components: a radiographic analysis of loosening. Clin Orthop Relat Res
5. Horowitz SM, Glasser DB, Lane JM, Healey JH. Prosthetic and extremity survivorship after limb salvage for sarcoma. How long do the reconstructions last? Clin Orthop Relat Res
6. Horowitz SM, Lane JM, Otis JC, Healey JH. Prosthetic arthroplasty of the knee after resection of a sarcoma in the proximal end of the tibia. A report of sixteen cases. J Bone Joint Surg Am
. 1991;73: 286-293.
7. Kawai A, Healey JH, Boland PJ, Athanasian EA, Jeon DG. A rotating-hinge knee replacement for malignant tumors of the femur and tibia. J Arthroplasty
8. Kawai A, Lin PP, Boland PJ, Athanasian EA, Healey JH. Relationship between magnitude of resection, complication, and prosthetic survival after prosthetic knee reconstructions for distal femoral tumors. J Surg Oncol
9. Kawai A, Muschler GF, Lane JM, Otis JC, Healey JH. Prosthetic knee replacement after resection of a malignant tumor of the distal part of the femur. Medium to long-term results. J Bone Joint Surg Am
10. Lazzaro F, Verdoia C, Gorla P, Balbino C, Benvenuti R. Ceramic- ceramic in total hip arthroplasty. An analysis of 7-13 year survival rate. Chir Organi Mov
11. Malawer MM, Chou LB. Prosthetic survival and clinical results with use of large-segment replacements in the treatment of high- grade bone sarcomas. J Bone Joint Surg Am
12. Malo M, Davis AM, Wunder J, Masri BA, Bell RS, Isler MH, Turcotte RE. Functional evaluation in distal femoral endoprosthetic replacement for bone sarcoma. Clin Orthop Relat Res
. 2001;389: 173-180.
13. Mandell JA, Carter DR, Goodman SB, Schurman DJ, Beaupre GS. A conical-collared intramedullary stem can improve stress transfer and limit micromotion. Clin Biomech (Bristol, Avon)
. 2004;19: 695-703.
14. Renard AJ, Veth RP, Schreuder HW, van Loon CJ, Koops HS, van Horn JR. Function and complications after ablative and limb- salvage therapy in lower extremity sarcoma of bone. J Surg Oncol
15. Roberts P, Chan D, Grimer RJ, Sneath RS, Scales JT. Prosthetic replacement of the distal femur for primary bone tumours. J Bone Joint Surg Br
16. Safran MR, Kody MH, Namba RS, Larson KR, Kabo JM, Dorey FJ, Eilber FR, Eckardt JJ. 151 endoprosthetic reconstructions for patients with primary tumors involving bone. Contemp Orthop
. 1994; 29:15-25.
17. Vervest TM, Anderson PG, Van Hout F, Wapstra FH, Louwerse RT, Koetsier JW. Ten to twelve-year results with the Zweymuller cementless total hip prosthesis. J Arthroplasty
18. Ward WG, Eckardt JJ, Johnston-Jones KS. Five to ten year results of custom endoprosthetic replacement for tumors of the distal femur. In: Brown KLB, ed. Complications of Limb Salvage: Prevention, Management and Outcome. ISOLS 6th International Symposium
. Montreal: ISOLS; 1991:483-491.
19. Ward WG, Johnston KS, Dorey FJ, Eckardt JJ. Extramedullary porous coating to prevent diaphyseal osteolysis and radiolucent lines around proximal tibial replacements. A preliminary report. J Bone Joint Surg Am
20. Wilkins RM, Camozzi AB, Gitelis SB. Reconstruction options for pediatric bone tumors about the knee. J Knee Surg
. 2005;18: 305-309.
21. Wilkins RM, Miller CM. Reoperation after limb preservation surgery for sarcomas of the knee in children. Clin Orthop Relat Res
22. Wirganowicz PZ, Eckardt JJ, Dorey FJ, Eilber FR, Kabo JM. Etiology and results of tumor endoprosthesis revision surgery in 64 patients. Clin Orthop Relat Res
23. Wunder JS, Leitch K, Griffin AM, Davis AM, Bell RS. Comparison of two methods of reconstruction for primary malignant tumors at the knee: a sequential cohort study. J Surg Oncol
. 2001;77:89-99. discussion 100.