Abstract
Aims:
Rotating-hinge knee prostheses are the most frequent procedures for distal femoral reconstruction after tumour resection despite the projected long-term burden of reoperation due to complications. Few studies have examined factors influencing prosthetic failure. To our knowledge, no reports have investigated the risk factors for prosthetic failure using competing risk models despite the high mortality. We aimed to determine the risk factors for overall/cause-specific prosthetic failure in patients undergoing rotating-hinge knee distal femoral replacement using the Fine-Gray competing risk model.
Patient and Methods:
We retrospectively reviewed 209 consecutive patients undergoing rotating-hinge knee prosthesis reconstruction after distal femoral resection from 1991–2016. The study endpoint was prosthetic failure defined as: 1) removal of the metal body femoral component, the tibial component, or the bone-implant fixation, 2) major revision (exchange of the metal body femoral component, tibial component, or the bone-implant fixation), or 3) amputation.
Results:
Multivariate Fine-Gray regression analyses revealed different hazards for each Henderson failure mode: percentage of femoral resection (p = 0.001) and extent of quadriceps muscle resection (p = 0.005) for overall prosthetic failure, extent of quadriceps muscle resection (p = 0.002) and fixation of femoral component (p = 0.011) for type 2 failure (aseptic loosening), age (p = 0.009) and percentage of femoral resection (p = 0.019) for type 3 failure (mechanical failure), and type of joint resection (p = 0.037) for type 4 (infection) were independent predictors. Bone stem ratio of 2.5 predicted aseptic loosening.
Conclusions:
We identified independent risk factors for overall/cause-specific prosthetic failure after rotating-hinge knee distal femoral replacement using a competing risk Fine-Gray model. A bone stem ratio > 2.5 is a reproducible cutoff predicting aseptic loosening. Accurate risk of oncologic distal femoral replacement is valuable for surgical planning and managing patient expectations.
Level of evidence:
Prognostic studies, Level II
Keywords: distal femoral replacement, prosthetic failure, risk factors, Fine-Gray model, competing risk model
INTRODUCTION
The most common method of reconstruction for malignant bone tumours is endoprosthetic reconstruction following distal femoral resection(1–17). This technique has advantages, including technical stability, widespread availability, and immediate stability following surgery. Currently, reconstruction using rotating-hinge knee prosthesis is the most popular method for distal femoral reconstruction since its development around 1990(1, 2, 4, 5, 7–10, 18). However, the procedure is still associated with the long-term burden of reoperation due to several complications, including aseptic loosening, structural failure, or infection(1, 2, 4, 5, 7–10, 18). Few studies have examined the factors influencing the risk of prosthetic failure after distal femoral replacement, especially for rotating-hinge knee prosthesis(8, 12, 19).
Additionally, disease-related mortality remains a significant event among patients with a malignant bone tumour undergoing distal femoral replacement, making the accurate interpretation of implant survival difficult. Schuh et al.(20) showed a competing risk model that reveals considerably reduced risks for every complication compared to Kaplan-Meier analysis when death is included as a competing event. They concluded that competing risk models should be used for estimating the risk of implant failure.
There have been no reports to date that have investigated the risk factors for prosthetic failure after distal femoral replacement taking death and other failures into account using competing risk models. Therefore, we aimed to determine the independent risk factors for overall and cause-specific prosthetic failure in patients undergoing rotating-hinge distal femoral replacement for oncologic indication using the Fine-Gray competing risk model.
PATIENTS AND METHODS
Eligibility
Patients who underwent distal femoral rotating-hinge knee replacement using Finn/OSS® rotating-hinge knee prosthesis for oncologic indication were eligible. Patients who received a total femoral replacement as their primary reconstruction were not eligible. Institutional databases were searched to identify patients who had resection of a malignant or aggressive benign bone tumour where the primary reconstruction was with a Finn/OSS/Compress® rotating-hinge knee prosthesis (Biomet, Warsaw, IN, USA). We identified 264 patients from 1991–2016 who met eligibility criteria. Patients lost with < four years of follow-up, without any endpoints or competing events (i.e., death, n = 28), were excluded from analysis. Patients who had a distal femoral replacement performed as a revision surgery of a failed reconstruction after oncologic indications or for the treatment of post-radiotherapy fracture of the distal femur (n = 27) were also excluded. We identified 209 patients for analyses (Figure 1). The Institutional Review Board (IRB) approved this study (Protocol #16-582).
Figure 1.
Flow diagram showing how patients were identified to be eligible for analysis.
Patient demographics and adjuvant therapy details are summarised in Table 1. In our cohort, 124 patients were male, and 85 were female. The median age was 29 years (interquartile range, 38). The most common histologic diagnosis was osteosarcoma (n = 118; 56%). Chemotherapy was given to 129 (62%) and radiotherapy administered to five (2%) patients, respectively. The mean follow-up period was 12.2 years for survivors (standard deviation, 6.1 years).
Table 1.
Patient demographics and adjuvant therapy data.
| Characteristic | Number of patients | (%) |
|---|---|---|
| Overall | 209 | 100 |
| Age, years; mean [SD] | ||
| ≤40 | 124 | 59 |
| ≥41 | 85 | 41 |
| Sex | ||
| Male | 109 | 52 |
| Female | 100 | 48 |
| BMI | ||
| 18.5-24.9 (Normal weight) | 80 | 38 |
| <18.5 (Underweight) | 33 | 16 |
| ≥25 (Overweight, obesity) | 89 | 43 |
| Unknown | 7 | 3 |
| Histologic diagnosis | ||
| Osteosarcoma | 118 | 56 |
| Chondrosarcoma | 22 | 11 |
| Malignant fibrous histiocytoma of bone | 10 | 5 |
| Bone involving soft tissue sarcoma | 8 | 4 |
| GCT of bone | 7 | 3 |
| Ewing’s sarcoma | 5 | 2 |
| Cancer metastasis/hematological malignancy | 22 | 11 |
| Others | 17 | 8 |
| Chemotherapy | 129 | 62 |
| Radiotherapy | 5 | 2 |
SD = standard deviation; BMI = body mass index; GCT = giant cell tumour.
Surgical procedure and prosthesis
Distal femoral resection was performed through a lateral or medial parapatellar approach, including excision of the previous biopsy site. Uninvolved vastus muscle was spared. Finn/OSS® rotating-hinge knee prosthesis was used in all cases. The canal was reamed alternating between straight and flexible reamers until chatter noted for approximately 5% of the stem length. Tapered conical reamers contoured the opening over 2.5 cm. Bowed trials were tested starting 0.5 mm below the maximally reamed diameter. If they could be inserted to within 2 cm of complete, then the uncemented implant was used. Larger reaming was done until this could be achieved. If the canal was patulous or stem insertion was too simple, the stem was cemented with a 2 mm total cement mantle. The prosthesis has a constrained hinge mechanism that also permits distraction movements and axial movements between the undersurface of the polyethylene bearing and the fixed tibial tray. Due to the weight-bearing rotating tibial articulation, weight-bearing is shared throughout the prosthesis, not borne by the axle alone. The femoral component has an anatomic center of rotation and a deep patellar tracking groove to enhance patellar biomechanics. Fixation of the stem is designed to be uncemented if a good press-fit can be achieved intraoperatively. Compress fixation was inserted with 181–363 kilograms of compression.
Young patients with no radiotherapy and localised disease received press fit stem fixation of the femur. Compress® was used for all patients after 2002 for this same indication. Elderly patients, those with metastatic cancer, and those requiring local radiotherapy received cemented components. Tibial components were uncemented whenever a surgeon could achieve a tight fit of the implant in the tibial bone.
Endpoints
The study endpoint was prosthetic failure defined as: 1) removal of the metal body femoral component, the tibial component, or the bone-implant fixation, 2) major revision (exchange of the metal body femoral component, the tibial component, or the bone-implant fixation), or 3) amputation. Prosthetic failures were recorded according to the classification system of Henderson, et al. Briefly, it is a classification for prosthetic failure which categorises prosthetic failure into mechanical and nonmechanical failure. Mechanical failure includes type 1 (soft-tissue failure: instability, tendon rupture, or aseptic wound dehiscence), type 2 (aseptic loosening), and type 3 (structural failure including periprosthetic or prosthetic fracture or deficient osseous supporting structure), and nonmechanical failure includes type 4 (infection necessitating removal of device), and type 5 (tumour progression).
Because: 1) most of type 1 (soft-tissue failure: instability, tendon rupture, or aseptic wound dehiscence) failures were not associated with implant removal/revision, and 2) the total number of type 5 failures was small for multivariate analyses, we focused on implant removal/revision overall and due to type 2 (aseptic loosening), 3, and 4 failure.
Data collection
The data examined included demographic information (patient age at surgery, sex, body mass index [BMI] at surgery), chemotherapy, and surgical details (e.g., percentage of femoral resection, the extent of quadriceps muscle resection, type of joint resection, fixation of the femoral component, or blood transfusion).
The quadriceps muscles were resected when they had been invaded by tumor or violated by the biopsy track. The rectus femoris muscle was removed when the tumor involved the muscle itself or the suprapatellar pouch. The extent of quadriceps muscle resection was defined based on the number of the 4 quadriceps muscles (not a percentage of quadriceps muscle resected).
BMI was calculated as weight in kilograms divided by the square of the height in meters (kg/m2) and categorised into three groups according to the World Health Organization nutritional status: underweight (<18.5 kg/m2), normal weight (18.5–24.9 kg/m2), and overweight/obese (≥25 kg/m2) (21). The percentage of femoral resection was categorised into < 40% and ≥ 40% according to the literature(11). Type of joint resection was categorised into intra-articular and extra-articular resection. Classic (en-bloc resection of the extensor mechanism) and modified (continuity of the extensor mechanism retained by coronal osteotomy of the patella) extra-articular resection were defined as extra-articular resection, while patellar retention with en-bloc total capsulectomy (circumferential arthrotomy of the patella and primary closure of the joint) technique was defined as intra-articular resection. Fixation of the femoral component was categorised into cement, press-fit, and compliant compression fixation. The total volume of blood transfusion was defined as the amount of blood transfused from the intraoperative period until 48 hours after the operation and categorised into ≤4 units and ≥5 units.
Bone stem ratio
Farfalli et al. showed that a diaphyseal bone stem ratio over 2.5 predicted higher rates of aseptic loosening when used press-fit femoral component in distal femoral replacement (22). Later, Bergin et al.(23) showed this cutoff was also a good predictive value when used cemented femoral component in distal femoral replacement. Therefore, we validated these findings using our cohort composed of 47 press-fit and 55 cemented femoral components who were available for radiologic assessment. The method of calculation of bone stem ratio was described previously(22). Figure 2 illustrates the technique to measure the prosthetic bone relationship. Briefly, we measured the outer cortical width and stem thickness ratios using a PACS system (Centricity®; General Electric Medical Systems, New York, NY). The measurements were made independently by two individuals (KO, TF) blinded to the outcomes.
Figure 2.
A. An anterior-posterior radiograph of the femur showing the 26-year follow-up of a distal femoral rotating hinge replacement with uncemented fixation. The technique to measure the prosthetic bone relationship is illustrated, selecting the level corresponding to half of the stem length. Black line = stem diameter. White line = bone outer diameter. Stem to bone ratio is 14.8/31.1= 0.48 exceeding the threshold of 0.4 that predicted good long term stability. B. Anterior-posterior radiograph of the femur 1.5 years after distal rotating hinge replacement with failure of uncemented fixation. The stem to bone ratio is 14.6/38.9 = 0.375.
Statistical analyses
Since 80 of 209 patients (38.3%) were dead at the time of follow-up, and death is a competing event of prosthetic failure, the Kaplan-Meier method overestimates the cumulative incidence of prosthetic failures. Implant removal/revision for those resulting from different reasons can be considered a competing event. Therefore, we performed Fine-Grey competing risk analysis, considering death and other prosthetic failures as competing events to estimate the incidence of prosthetic failure accurately(20, 24). Univariate and multivariable competing risk Fine-Gray regression models were performed to examine the association between each factor and the occurrence of prosthetic failure.
All statistical analyses were conducted using IBM SPSS version 23.0 (IBM SPSS, Armonk, NY, USA) and R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria).
RESULTS
Risk factors for prosthetic failure
Sixty-seven of 209 patients (32.1%) had prosthetic failure overall. Fine-Gray competing risk analysis excluding the influence of death on the cumulative incidence of prosthetic failure for any causes are shown in Figure 3. We found that younger age (p = 0.002), larger percentage of femoral resection (p < 0.001), and larger extent of quadriceps muscle resection (p < 0.001) were significantly associated with prosthetic failure. On multivariate Fine-Gray regression analyses for overall prosthetic failure, the largest percentage of femoral resection (p = 0.001) and largest extent of quadriceps muscle resection (p = 0.005) were significantly associated with prosthetic failure (Table 2).
Figure 3.
Cumulative incidence of prosthetic failure for any causes stratified by each covariate: sex, age, body mass index (BMI), chemotherapy, joint resection, percentage of femur resection, extent of quadriceps resection, fixation of femoral component, and blood transfusion.
Table 2.
Univariate and multivariate Fine-Gray models for prosthetic failure (any cause)
| Number of patients | Number of patients with failure (%) | Univariate analysis | Multivariate analysis | |||
|---|---|---|---|---|---|---|
|
| ||||||
| Hazard ratio (95% CI) | p value | Hazard ratio (95% CI) | p value | |||
| Total number of patients | 209 | 67 (32.1) | ||||
| Sex | ||||||
| Male | 109 | 34 (32.1) | Reference | |||
| Female | 100 | 33 (33.0) | 1.07 (0.66–1.71) | 0.790 | ||
| Age | ||||||
| ≤40 | 124 | 51 (41.1) | Reference | Reference | ||
| ≥41 | 85 | 16 (18.8) | 0.42 (0.24–0.73) | 0.002 | 0.70 (0.37–1.32) | 0.270 |
| Body Mass Index | ||||||
| 18.5–24.9 (Normal weight) | 80 | 23 (28.7) | Reference | |||
| <18.5 (Underweight) | 33 | 14 (42.4) | 1.51 (0.79–2.89) | 0.210 | ||
| ≥25 (Overweight, obesity) | 89 | 30 (33.7) | 1.29 (0.75–2.21) | 0.360 | ||
| Chemotherapy | ||||||
| No | 80 | 20 (25.0) | Reference | |||
| Yes | 129 | 47 (36.4) | 1.54 (0.92–2.58) | 0.100 | ||
| Type of joint resection* | ||||||
| Intra-articular | 183 | 57 (31.1) | Reference | |||
| Extra-articular | 26 | 10 (38.5) | 1.44 (0.70–2.98) | 0.330 | ||
| Percentage of femoral resection | ||||||
| <40% | 133 | 30 (22.6) | Reference | Reference | ||
| ≥40% | 76 | 37 (48.7) | 2.45 (1.52–3.95) | <0.001 | 2.31 (1.39–3.83) | 0.001 |
| Extent of resection of the quadriceps muscles | ||||||
| 0–2 | 168 | 45 (26.8) | Reference | Reference | ||
| 3–4 | 41 | 22 (53.7) | 2.62 (1.57–4.36) | <0.001 | 2.14 (1.26–3.63) | 0.005 |
| Fixation of femoral component | ||||||
| Press-fit | 64 | 26 (40.6) | Reference | Reference | ||
| Cement | 70 | 16 (22.9) | 0.51 (0.28–0.94) | 0.030 | 0.59 (0.30–1.16) | 0.130 |
| Compress | 75 | 25 (33.3) | 0.85 (0.49–1.49) | 0.580 | 0.75 (0.43–1.32) | 0.320 |
| Blood transfusion | ||||||
| −4 units | 158 | 53 (33.5) | Reference | |||
| ≥ 5 units | 41 | 13 (31.7) | 1.03 (0.55–1.95) | 0.920 | ||
In terms of cause-specific prosthetic failure, 30 (14.4%), 27 (12.9%), and 17 (8.1%) of 209 patients had type 2 (aseptic loosening), 3 (structural failure), and 4 (infection) failures, respectively. Fine-Gray competing risk analysis to exclude the influence of death on the cumulative incidence of prosthetic failure are shown in Figure 4, Figure 5, and Figure 6. We found that younger age (p = 0.026), larger extent of quadriceps muscle resection (p < 0.001), and fixation of femoral component (press-fit stem) (p = 0.016) for type 2 failure (aseptic loosening) (Figure 4), sex (female) (p = 0.030), younger age (p < 0.001), and larger percentage of femoral resection (p = 0.001) for type 3 failure (structural failure) (Figure 5), extra-articular joint resection (p < 0.001), and larger extent of quadriceps muscle resection (p = 0.005) for type 4 failure (infection) (Figure 6), were significantly associated with higher rate of prosthetic failure. On multivariate Fine-Gray regression analyses for each type of prosthetic failure, larger extent of quadriceps muscle resection (p = 0.002) and fixation of femoral component (press-fit compared with Compress) (p = 0.011) for type 2 failure (aseptic loosening) (Table 3), younger age (p = 0.009) and larger percentage of femoral resection (p = 0.019) for type 3 failure (structural failure) (Table 4), and extra-articular joint resection (p = 0.037) for type 4 (infection) (Table 5) were independent predictors for prosthetic failure.
Figure 4.
Cumulative incidence of type 2 prosthetic failure (aseptic loosening) stratified by each covariate: sex, age, body mass index (BMI), chemotherapy, joint resection, percentage of femur resection, extent of quadriceps resection, fixation of femoral component, and blood transfusion.
Figure 5.
Cumulative incidence of type 3 prosthetic failure (structural failure) stratified by each covariate: sex, age, body mass index (BMI), chemotherapy, joint resection, percentage of femur resection, extent of quadriceps resection, fixation of femoral component, and blood transfusion.
Figure 6.
Cumulative incidence of type 5 prosthetic failure stratified by each covariate: sex, age, body mass index (BMI), chemotherapy, joint resection, percentage of femur resection, extent of quadriceps resection, fixation of femoral component, and blood transfusion.
Table 3.
Univariate and multivariate Fine-Gray models for type 2 prosthetic failure (aseptic loosening)
| Number of patients | Number of patients with failure (%) | Univariate analysis | Multivariate analysis | |||
|---|---|---|---|---|---|---|
|
| ||||||
| Hazard ratio (95% CI) | p value | Hazard ratio (95% CI) | p value | |||
| Total Number of patients | 209 | 30 (14.4) | ||||
| Sex | ||||||
| Male | 109 | 18 (16.5) | Reference | |||
| Female | 100 | 12 (12.0) | 0.72 (0.35–1.50) | 0.380 | ||
| Age | ||||||
| ≤40 | 124 | 24 (19.4) | Reference | Reference | ||
| ≥41 | 85 | 6 (7.1) | 0.36 (0.15–0.89) | 0.027 | 0.41 (0.12–1.41) | 0.160 |
| Body Mass Index | ||||||
| 18.5–24.9 (Normal weight) | 80 | 13 (16.3) | Reference | |||
| <18.5 (Underweight) | 33 | 6 (18.2) | 1.08 (0.41–2.83) | 0.880 | ||
| ≥25 (Overweight, obesity) | 89 | 11 (12.4) | 0.76 (0.34–1.68) | 0.490 | ||
| Chemotherapy | ||||||
| No | 80 | 9 (11.3) | Reference | |||
| Yes | 129 | 21 (16.3) | 1.46 (0.68–3.17) | 0.330 | ||
| Type of joint resection* | ||||||
| Intra-articular | 183 | 26 (14.2) | Reference | |||
| Extra-articular | 26 | 4 (15.4) | 1.07 (0.37–3.11) | 0.900 | ||
| Percentage of femoral resection | ||||||
| < 40% | 133 | 14 (10.5) | Reference | |||
| ≥ 40% | 76 | 16 (21.1) | 1.99 (0.98–4.06) | 0.059 | ||
| Extent of resection of the quadriceps muscles | ||||||
| 0–2 | 168 | 16 (9.5) | Reference | Reference | ||
| 3–4 | 41 | 14 (34.1) | 4.17 (2.05–8.46) | <0.001 | 3.25 (1.55–6.82) | 0.002 |
| Fixation of femoral component | ||||||
| Press-fit | 64 | 16 (25.0) | Reference | Reference | ||
| Cement | 70 | 8 (11.4) | 0.43 (0.18–0.99) | 0.010 | 0.78 (0.26–2.36) | 0.660 |
| Compress | 75 | 6 (8.0) | 0.29 (0.12–0.75) | 0.047 | 0.30 (0.12–0.77) | 0.011 |
| Blood transfusion | ||||||
| −4 units | 158 | 23 (14.6) | Reference | |||
| ≥ 5 units | 41 | 6 (14.6) | 1.05 (0.42–2.62) | 0.910 | ||
Table 4.
Univariate and multivariate Fine-Gray models for type 3 prosthetic failure (structural failure).
| Number of patients | Number of patients with failure (%) | Univariate analysis | Multivariate analysis | |||
|---|---|---|---|---|---|---|
|
| ||||||
| Hazard ratio (95% CI) | p value | Hazard ratio (95% CI) | p value | |||
| Total number of patients | 209 | 27 (12.9) | ||||
| Sex | ||||||
| Male | 109 | 9 (8.3) | Reference | |||
| Female | 100 | 18 (18.0) | 2.32 (1.04–5.20) | 0.040 | 2.17 (0.99–4.77) | 0.054 |
| Age | ||||||
| ≤ 40 | 124 | 25 (20.2) | Reference | Reference | ||
| ≥ 41 | 85 | 2 (2.4) | 0.11 (0.03–0.47) | 0.003 | 0.14 (0.03–0.62) | 0.009 |
| Body Mass Index | ||||||
| 18.5–24.9 (Normal weight) | 80 | 9 (11.3) | Reference | |||
| <18.5 (Underweight) | 33 | 6 (18.2) | 1.62 (0.59–4.45) | 0.350 | ||
| ≥25 (Overweight, obesity) | 89 | 12 (13.5) | 1.23 (0.52–2.90) | 0.630 | ||
| Chemotherapy | ||||||
| No | 80 | 8 (10.0) | Reference | |||
| Yes | 129 | 19 (14.7) | 1.49 (0.66–3.38) | 0.340 | ||
| Type of joint resection* | ||||||
| Intra-articular | 183 | 25 (13.7) | Reference | |||
| Extra-articular | 26 | 2 (7.7) | 0.53 (0.12–2.29) | 0.400 | ||
| Percentage of femoral resection | ||||||
| < 40% | 133 | 9 (6.8) | Reference | |||
| ≥ 40% | 76 | 18 (23.7) | 3.55 (1.60–7.88) | 0.002 | 2.62 (1.17–5.85) | 0.019 |
| Extent of resection of the quadriceps muscles | ||||||
| 0–2 | 168 | 18 (10.7) | Reference | |||
| 3–4 | 41 | 9 (22.0) | 2.11 (0.97–4.61) | 0.061 | ||
| Fixation of femoral component | ||||||
| Press-fit | 64 | 9 (14.1) | Reference | |||
| Cement | 70 | 5 (7.1) | 0.53 (0.18–1.55) | 0.250 | ||
| Compress | 75 | 13 (17.3) | 1.45 (0.64–3.29) | 0.370 | ||
| Blood transfusion | ||||||
| −4 units | 158 | 22 (13.9) | Reference | |||
| ≥5 units | 41 | 4 (9.8) | 0.70 (0.24–2.07) | 0.520 | ||
Table 5.
Univariate and multivariate Fine-Gray models for type 4 prosthetic failure (infection necessitating removal of device).
| Number of patients | Number. of patients with failure (%) | Univariate analysis | Multivariate analysis | |||
|---|---|---|---|---|---|---|
|
| ||||||
| Hazard ratio (95% CI) | p value | Hazard ratio (95% CI) | p value | |||
| Total number of patients | 209 | 17 (8.1) | ||||
| Sex | ||||||
| Male | 109 | 12 (11.0) | Reference | |||
| Female | 100 | 5 (5.0) | 0.44 (0.16–1.23) | 0.120 | ||
| Age | ||||||
| ≤ 40 | 124 | 9 (7.3) | Reference | |||
| ≥ 41 | 85 | 8 (9.4) | 1.38 (0.52–3.62) | 0.520 | ||
| Body Mass Index | ||||||
| 18.5–24.9 (Normal weight) | 80 | 4 (5.0) | Reference | |||
| <18.5 (Underweight) | 33 | 1 (3.0) | 0.56 (0.06–5.01) | 0.575 | ||
| ≥25 (Overweight, obesity) | 89 | 12 (13.5) | 2.82 (0.90–8.88) | 0.075 | ||
| Chemotherapy | ||||||
| No | 80 | 9 (11.3) | Reference | |||
| Yes | 129 | 8 (6.2) | 0.53 (0.20–1.40) | 0.200 | ||
| Type of joint resection* | ||||||
| Intra-articular | 183 | 10 (5.5) | Reference | Reference | ||
| Extra-articular | 26 | 7 (26.9) | 5.64 (2.15–14.77) | <0.001 | 3.87 (1.14–13.18) | 0.037 |
| Percentage of femoral resection | ||||||
| < 40% | 133 | 10 (7.5) | Reference | |||
| ≥ 40% | 76 | 7 (9.2) | 1.23 (0.46–3.26) | 0.680 | ||
| Extent of resection of the quadriceps muscles | ||||||
| 0–2 | 168 | 9 (5.4) | Reference | Reference | ||
| 3–4 | 41 | 8 (19.5) | 3.85 (1.50–9.90) | 0.005 | 2.36 (0.71–7.89) | 0.160 |
| Fixation of femoral component | ||||||
| Press-fit | 64 | 7 (10.9) | Reference | |||
| Cement | 70 | 5 (7.1) | 0.68 (0.21–2.14) | 0.510 | ||
| Compress | 75 | 5 (6.7) | 0.66 (0.21–2.08) | 0.480 | ||
| Blood transfusion | ||||||
| −4 units | 158 | 11 (7.0) | Reference | |||
| ≥ 5 units | 41 | 6 (14.6) | 2.20 (0.82–5.90) | 0.120 | ||
Validation of “bone stem ratio > 2.5” as a cutoff to predict aseptic loosening
The mean bone stem ratio was 2.27 (standard deviation 0.27) overall. The sensitivity and specificity of bone-stem ratio cutoff of 2.5 for aseptic loosening is shown in Table 6. Bone stem ratio of 2.5 predicted aseptic loosening in the entire cohort as well as those reconstructed with press-fit and cemented femoral component.
Table 6.
Sensitivity and specificity for predicting aseptic loosening using bone-stem ratio cutoff of 2.5
| Number of patients | Bone-stem ratio, mean [standard deviation] | Sensitivity | Specificity | |
|---|---|---|---|---|
| Entire cohort | 102 | 2.27 [0.27] | 80.0% | 95.1% |
| Press-fit femoral component | 47 | 2.28 [0.29] | 76.9% | 97.1% |
| Cemented femoral component | 55 | 2.26 [0.26] | 85.7% | 93.8% |
DISCUSSION
Among the various reconstructive methods available for distal femoral resection, endoprosthetic replacement has been the most frequently used. This is due to several advantages, which include early stability, mobilization, and weight bearing. The first-generation non-rotating-hinge knee prosthesis was shown to attain an acceptable result after distal femoral resection(8, 11, 14–17). However, data defining patients at high risk of prosthetic failure after long-term follow-up are limited. To our knowledge, while there have been several reports analysing a large number of patients undergoing distal femoral replacement, only a few studies have analysed risk factors for prosthetic failure(1, 2, 4, 5, 7–12, 18, 19, 25). Furthermore, few of these investigations utilised a competing risk model which consider death as a competing event despite high mortality during the follow-up due to the nature of the population(25). As Schuh et al. (20) showed in their report, using the Kaplan-Meier analysis to estimate prosthetic survival in oncology patients leads to considerable overestimate of prosthetic failure rate. We used a Fine-Gray competing risk analysis when estimating the prosthetic failure and identifying independent risk factors for prosthetic failure. In addition, in the present study, we only analysed patients with oncologic indication because of the following reasons: 1) a recent study indicated the failure rate and mode after DFR is different between oncologic and non-oncologic patients and indicated these patients should be analysed separately(25), and 2) non-oncologic patients lack comparable risk factors such as intra- or extra-articular resection consideration, quadriceps resection, or chemotherapy exposure.
Kawai et al.(12) analysed 82 patients undergoing distal femoral replacement reconstructed with Lane-Burstein® (n = 51) or Finn® knee prosthesis (n =31) and reported risk factors for prosthetic failure for any causes as endpoints. In their univariate comparison, length of femur resection ≥ 40% and subtotal quadriceps resection were the significant factors for prosthetic failure. Multivariate analysis showed that length of femur resection ≥ 40% was the only independent adverse prognostic factor for prosthetic survival. Myers et al.(8) analysed 162 fixed hinge and 173 rotating-hinge distal femoral replacement and reported risk factors for prosthetic failure for any reason, and aseptic loosening as endpoints. They performed a univariate comparison and identified: 1) rotating-hinge prosthesis had significantly better outcomes than fixed hinge implants when considering prosthetic failure for any causes as an endpoint, and 2) rotating-hinge knee with a hydroxyapatite collar had significantly better outcomes than fixed-hinge/other rotating-hinge prothesis when considering aseptic loosening as an endpoint.
In addition to these reports, prosthetic failure due to aseptic loosening (Henderson type 2) has been well investigated. As mentioned above, bone stem ratio > 2.5 has been reported as a predictive cutoff value to predict higher rate of aseptic loosening in patients undergoing distal femoral replacement using press-fit(22) and cemented femoral components(23). In the present study, we successfully confirmed that bone stem ratio > 2.5 is a robust cutoff value both in press-fit and cemented stem patients to predict aseptic loosening. Although a recent publication suggested that multidrug chemotherapy causes early radiological signs of loosening in distal femoral replacements,(26) were not able to show the correlation between chemotherapy and aseptic loosening.(27)
Risk factors for prosthetic failure due to structural failure, including periprosthetic or prosthetic fracture (Henderson type 3), have not been sufficiently studied, especially for rotating-hinge knee prosthesis. In the present study, we demonstrated younger age ≤ 40 years and longer femur resection ≥ 40% were the significant factors affecting structural failure. Based on our results, changes in biomechanical stresses after extensive resections of bone and greater activity in young patients may be possible explanation of the higher rate of structural failure. Capanna et al.(19) found that breakage of the stem of the Kotz prosthesis was associated with greater excision of quadriceps muscles, suggesting increased torque production out of the line of prosthesis and/or impairment of quadriceps contraction may contribute to these complications. Although patients with greater excision of quadriceps muscles in our series had higher tendency of structural failure, there was no statistically significant difference.
There have been no reports investigating risk factors for prosthetic failure after distal femoral replacement due to infection (Henderson type 4). Morris et al.(28) analysed risk factors for infection in 110 patients undergoing lower extremity oncologic surgery. They identified on multivariate analysis that blood transfusion and obesity were the independent risk factors for the development of wound infection in patients having lower extremity orthopaedic oncologic procedures. In the present study, we demonstrated extra-articular resection was the significant factor for prosthetic failure due to infection. Although patients with overweight/obesity and blood transfusion ≥ 5 units had tendency of higher infection rates in accordance with the report by Morris et al.(28), we were not able to show statistical significance.
Our study has several limitations. First, it was designed as a retrospective observation study and lacked a control group. Comparison to a control group is difficult since other methods of reconstruction were very rarely used at our institution. Second, patients are composed of a heterogeneous population. Even for the same brand implant design (Zimmer Biomet), there were variation in length of femoral component, or short vs. long stemmed tibial trays, or different fixation methods of the femoral component. Although the variation was minimised because the kinematics of the knee articulation are identical and independent of the femoral fixation, this might affect surgical outcomes.
In conclusion, we identified independent risk factors for overall and cause-specific prosthetic failure after Finn/OSS® rotating-hinge knee prosthesis using Fine-Gray models considering death and other failures as competing events. We also demonstrated that bone stem ratio > 2.5 is a reproducible cutoff to predict aseptic loosening. Accurate stratification of patients undergoing distal femoral replacement presented in this study is valuable to take into account when planning surgery, counselling patients, and anticipating revisions.
TAKE HOME MESSAGE.
Independent risk factors exist for each mode of cause-specific failure of rotating-hinge knee distal femoral replacement. This may influence patient management, follow-up, and interpretation of the literature.
A bone stem ratio > 2.5 is a reproducible cutoff predicting aseptic loosening.
Funding:
This research was funded in part by the NIH/NCI Cancer Center Support Grant, P30 CA008748, The Limb Preservation Fund, and The Perlman Research Fund.
Footnotes
Conflicts of interest: The authors declare no conflicts of interest pertinent to this work.
Contributor Information
Koichi Ogura, Department of Surgery, Orthopaedic Service, Memorial Sloan Kettering Cancer Center, 1275 York Ave., New York, NY, 10065.
Tomohiro Fujiwara, Department of Surgery, Orthopaedic Service, Memorial Sloan Kettering Cancer Center, 1275 York Ave., New York, NY 10065.
Carol D. Morris, Division Chief of Orthopaedic Oncology, Department of Orthopaedic Surgery, Johns Hopkins University, 601 N. Caroline St., Baltimore, MD 21287.
Patrick J. Boland, Department of Surgery, Orthopaedic Service, Memorial Sloan Kettering Cancer Center, 1275 York Ave., New York, NY, 10065.
John H. Healey, Department of Surgery, Orthopaedic Service, Memorial Sloan Kettering Cancer Center, 1275 York Ave., New York, NY, 10065.
REFERENCES
- 1.Bus MP, van de Sande MA, Fiocco M, Schaap GR, Bramer JA, Dijkstra PD. What are the long-term results of MUTARS((R)) Modular Endoprostheses for Reconstruction of Tumor Resection of the Distal Femur and Proximal Tibia? Clin Orthop Relat Res. 2017;475(3):708–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Batta V, Coathup MJ, Parratt MT, Pollock RC, Aston WJ, Cannon SR, et al. Uncemented, custom-made, hydroxyapatite-coated collared distal femoral endoprostheses: up to 18 years’ follow-up. Bone Joint J. 2014;96-b(2):263–9. [DOI] [PubMed] [Google Scholar]
- 3.Biau D, Faure F, Katsahian S, Jeanrot C, Tomeno B, Anract P. Survival of total knee replacement with a megaprosthesis after bone tumor resection. J Bone Joint Surg Am. 2006;88(6):1285–93. [DOI] [PubMed] [Google Scholar]
- 4.Bickels J, Wittig JC, Kollender Y, Henshaw RM, Kellar-Graney KL, Meller I, et al. Distal femur resection with endoprosthetic reconstruction: a long-term followup study. Clin Orthop Relat Res. 2002(400):225–35. [DOI] [PubMed] [Google Scholar]
- 5.Coathup MJ, Batta V, Pollock RC, Aston WJ, Cannon SR, Skinner JA, et al. Long-term survival of cemented distal femoral endoprostheses with a hydroxyapatite-coated collar: a histological study and a radiographic follow-up. J Bone Joint Surg Am. 2013;95(17):1569–75. [DOI] [PubMed] [Google Scholar]
- 6.Kinkel S, Lehner B, Kleinhans JA, Jakubowitz E, Ewerbeck V, Heisel C. Medium to long-term results after reconstruction of bone defects at the knee with tumor endoprostheses. J Surg Oncol. 2010;101(2):166–9. [DOI] [PubMed] [Google Scholar]
- 7.Morgan HD, Cizik AM, Leopold SS, Hawkins DS, Conrad EU 3rd., Survival of tumor megaprostheses replacements about the knee. Clin Orthop Relat Res. 2006;450:39–45. [DOI] [PubMed] [Google Scholar]
- 8.Myers GJ, Abudu AT, Carter SR, Tillman RM, Grimer RJ. Endoprosthetic replacement of the distal femur for bone tumours: long-term results. J Bone Joint Surg Br. 2007;89(4):521–6. [DOI] [PubMed] [Google Scholar]
- 9.Pala E, Trovarelli G, Calabro T, Angelini A, Abati CN, Ruggieri P. Survival of modern knee tumor megaprostheses: failures, functional results, and a comparative statistical analysis. Clin Orthop Relat Res. 2015;473(3):891–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Schwartz AJ, Kabo JM, Eilber FC, Eilber FR, Eckardt JJ. Cemented distal femoral endoprostheses for musculoskeletal tumor: improved survival of modular versus custom implants. Clin Orthop Relat Res. 2010;468(8):2198–210. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.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. 1998;80(5):636–47. [DOI] [PubMed] [Google Scholar]
- 12.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. 1999;70(2):109–15. [DOI] [PubMed] [Google Scholar]
- 13.Grimer RJ, Aydin BK, Wafa H, Carter SR, Jeys L, Abudu A, et al. Very long-term outcomes after endoprosthetic replacement for malignant tumours of bone. Bone Joint J. 2016;98-b(6):857–64. [DOI] [PubMed] [Google Scholar]
- 14.Jones GB. Total knee replacement-the Walldius hinge. Clin Orthop Relat Res. 1973(94):50–7. [DOI] [PubMed] [Google Scholar]
- 15.Kotz R, Ritschl P, Trachtenbrodt J. A modular femur-tibia reconstruction system. Orthopedics. 1986;9(12):1639–52. [DOI] [PubMed] [Google Scholar]
- 16.Mascard E, Anract P, Touchene A, Pouillart P, Tomeno B. [Complications from the hinged GUEPAR prosthesis after resection of knee tumor. 102 cases]. Rev Chir Orthop Reparatrice Appar Mot. 1998;84(7):628–37. [PubMed] [Google Scholar]
- 17.Wilson FC, Fajgenbaum DM, Venters GC. Results of knee replacement with the Walldius and geometric prostheses. A comparative study. J Bone Joint Surg Am. 1980;62(4):497–503. [PubMed] [Google Scholar]
- 18.Pala E, Trovarelli G, Angelini A, Ruggieri P. Distal femur reconstruction with modular tumour prostheses: a single Institution analysis of implant survival comparing fixed versus rotating hinge knee prostheses. Int Orthop. 2016;40(10):2171–80. [DOI] [PubMed] [Google Scholar]
- 19.Capanna R, Morris HG, Campanacci D, Del Ben M, Campanacci M. Modular uncemented prosthetic reconstruction after resection of tumours of the distal femur. J Bone Joint Surg Br. 1994;76(2):178–86. [PubMed] [Google Scholar]
- 20.Schuh R, Kaider A, Windhager R, Funovics PT. Does competing risk analysis give useful information about endoprosthetic survival in extremity osteosarcoma? Clin Orthop Relat Res. 2015;473(3):900–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Executive summary of the clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults. Arch Intern Med. 1998;158(17):1855–67. [DOI] [PubMed] [Google Scholar]
- 22.Farfalli GL, Boland PJ, Morris CD, Athanasian EA, Healey JH. Early equivalence of uncemented press-fit and Compress femoral fixation. Clin Orthop Relat Res. 2009;467(11):2792–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Bergin PF, Noveau JB, Jelinek JS, Henshaw RM. Aseptic loosening rates in distal femoral endoprostheses: does stem size matter? Clin Orthop Relat Res. 2012;470(3):743–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Gray R. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat. 1988. 16:1141–54. [Google Scholar]
- 25.Staats K, Vertesich K, Sigmund IK, Sosa B, Kaider A, Funovics PT, et al. Does a competing risk Analysis show differences in the cumulative incidence of revision surgery between patients with oncologic and non-oncologic conditions after distal femur replacement? Clin Orthop Relat Res. 2020;478(5):1062–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Mumith A, Coathup M, Edwards TC, Gikas P, Aston W, Blunn G. Multidrug chemotherapy causes early radiological signs of loosening in distal femoral replacements. Bone Joint Res. 2020;9(7):333–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Elalfy MA, Boland PJ, Healey JH. Chemotherapy curtails bone formation from compliant compression fixation of distal femoral endoprostheses. Clin Orthop Relat Res. 2019;477(1):206–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Morris CD, Sepkowitz K, Fonshell C, Margetson N, Eagan J, Miransky J, et al. Prospective identification of risk factors for wound infection after lower extremity oncologic surgery. Ann Surg Oncol. 2003;10(7):778–82. [DOI] [PubMed] [Google Scholar]