In terms of patient survival it is very difficult to estimate the role of the surgical method or choice of implant. Many cofactors have shown to influence patient survival in metastatic bone disease as age, preoperative general health status, type of cancer, location of metastasis in the femur or solitary versus multiple metastases [11–14]. Apart from these factors actual fracture compared to impending fracture in long bones seems to be another essential cause influencing patient survival [5, 7]. Mavrogenis et al. [5] demonstrated that impending fractures show a significantly better life expectancy with survival rates of approximately 60% at one year, 40% at two years, 30% at three years and 20% after 5 years. Survival rates of actual pathologic femoral fracture are reported to be approximately 45% at six months, 30% at one year, 15% at two years and less than 10% after 3 years [4, 5, 15], corresponding well to our results. Mavrogenis et al. additionally found that the type of surgery is a significant factor in patient survival. This was also demonstrated for pathologic fractures of the proximal femur. Patient survival was significantly higher in 18 patients with proximal femoral resection and modular prosthetic replacement compared to 11 patients with intramedullary nailing [5]. According to these results our primary hypothesis was to confirm superior survival in patients undergoing proximal femoral resection and endoprothetic reconstruction compared to intramedullary osteosynthesis. Although we could see a tendency to better survival in the prosthetic group it was statistically not significant. Besides, preoperative general health performance which is a substantial predictor of survival in skeletal metastases [11, 12] was significantly lower in the osteosynthesis group contributing to an earlier death and putting a better survival by trend in the prosthesis group into perspective. Preoperative general health performance was not reported by Mavrogenis et al. Consequently, the benefit in patient survival after resection and endoprosthesis in their study might be equivocal [5].
None of the patients in our series died because of intraoperative embolic events. It has been shown by others that implantation of a long intramedullary nail or a long-stem cemented femoral component in patients with femoral metastasis increases the risk of an embolic syndrome considerably resulting in catastrophic outcome in up to 8% [16–19]. We believe that cemented stems with a regular length and modular reconstruction bushings bridging the osseous defect as applied in our study may help in reducing the risk of an embolic syndrome und increase patient safety.
Complications requiring reoperation are reported to be as high as 26% for intramedullary nailing and 18% for endoprosthesic replacement in treatment of pathologic fracture of the proximal femur [4, 9]. This compares well to our results with a complication rate of 25% in each group. As for intramedullary nailing complication rates might be substantially higher considering actual pathologic fractures exclusively. The rate of fracture union is considerably less than 50% for common tumor types such as breast and renal carcinoma or even absent as demonstrated for lung carcinoma [6]. On the other hand impending fractures proceed to fracture in only 13% after local irradiation [20], subsequently load induced stress to osteosynthetic devices and potential failure is conspicuously lower. This is confirmed by Harvey et al. [9] demonstrating a significantly higher complication rate for intramedullary nails in actual pathologic fractures than in impending fractures. Other authors report considerably lower complication rates for intramedullary nailing ranging from 2 to 6%, but it must be mentioned that impending fractures account for approximately 60-67% in these studies [5, 7, 8]. Many authors of larger series advocate arthroplasty in favor of osteosynthesis in metastases of the proximal femur because of superior durability and lower complication rates [4, 8, 9]. Our complication rate of 25% in the prosthesis group compared well to the results of Harvey et al. [9]. Nevertheless, this is higher than the 3-10% several others authors have reported [4, 5, 8]. A reason for our higher dislocation rate might be, that 50% received a total arthroplasty which is associated with substantially higher dislocation rates than hemiarthroplasty in this setting [4, 21]. Most other studies use hemiarthroplasty in the vast majority of cases [4, 8, 9], explaining a lower complication rate in this respect. Additionally, regular or long-stemmed revision femoral components are widely used [4, 8, 22] preserving the greater trochanter and subsequently improving hip joint stability. In order to improve safety and reduce the higher dislocation risk of proximal femoral replacements, preservation and repair of the hip capsule as well as applying a bipolar head whenever possible is recommended [9]. If preservation of the capsule is not possible, attachment tubes for soft tissue reconstruction or tripolar cups might help in reducing the risk of dislocation [23]. Both options were not performed in our patients except in one (Figure 3).
Preoperative general health status is an important parameter in predicting survival in patients with skeletal metastases [11, 12]. Unfortunately, many authors do not report on this [4, 5, 9]. As for our study groups we demonstrated a significant difference suggesting a profound selection bias. Correspondingly, most of the patients with a poor preoperative health condition were treated with an intramedullary nail and patients with moderate to good general performance received endoprosthetic reconstruction. It should be mentioned that two out of four patients with a good preoperative general health condition and normal ambulatory capacity in the osteosynthesis group sustained a hardware failure. This is in contrast to the results of Steensma et al. [8]. They reported a preoperative Eastern Cooperative Oncology Group (ECOG) Score of 2 points or less in 88% of patients treated with an intramedullary nail (IM) compared to only 61% in the endoprosthesis group. That means almost all patients in the IM group were ambulatory and had a moderate to very good preoperative health status preoperatively. Nevertheless, a good health status and a subsequently higher activity of patients did not contribute to a higher implant related complication rate after nailing which was 6.1% and is considerably lower than in our IM group. A possible explanation might be the relatively high rate of impending fractures (70%) as discussed above [8].
Despite the relatively small number of patients our study holds several strengths compared to other larger series [4, 7–9]. First, we focused on the actual pathologic fracture and its corresponding characteristic features outlined above, exclusively. Second, we included only patients with multiple osseous metastases and fracture location in the proximal femur, excluding further potential confounding variables. Third, we addressed the preoperative general health performance which is a major prognostic factor in patient survival and correspondingly supported interpretation of survival data which is rarely reflected in other studies. On the other hand, several limitations of our study must be mentioned. First, the retrospective design and relatively small number of patients in the study comprise familiar limitations by itself. Nevertheless, actual pathologic fractures are quite rare and survival of patients is limited qualifying this type of study. Second, many different types of cancer were included. However, distribution of cancer type according to aggressiveness [12] was comparable with approximately 60% of slow to moderate growth types in each group. Third, adjuvant and neoadjuvant therapy was not considered. But a wide divergence of treatment protocols and unclear effectiveness [22] precluded inclusion of this criterion.