Surgical management of metastatic lesions in the proximal femur: a systematic review

Design and study registration

The current study is a systematic review and is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (9). The study protocol was registered prospectively on Prospero (CRD42023408021). Protocol deviations over the course of the study are detailed in the discussion after the 'Limitations' section.

Review questions

i) What is the revision rate and ii) what is the implant failure rate following IF compared to EPR of the proximal femur due to MBD?

Population, intervention and comparator

The population of interest comprises adult patients with a pathologic or impending pathologic fracture of the proximal femur due to cancer dissemination or local hematologic malignancy. We included studies on patients who had undergone either IF (screws, plates or IMN) or EPR (total hip arthroplasty or hemiarthroplasty). We excluded populations with metastases to the diaphysis or distal femur without proximal femur involvement and those undergoing a revision surgery for a failed proximal femur implant, partial pelvic replacement, total femoral replacement or allograft-prosthesis composites, as these patients are ineligible to receive IF. Only studies comparing IF to EPR were included in the quantitative synthesis. Studies with other comparators or no comparators were included in the qualitative synthesis, with results reported descriptively.

Outcomes

This study had two co-primary outcomes: i) the overall revision rate ratio and ii) the overall implant failure rate ratio for IF compared to EPR. Secondary outcomes included time-specific and cause-specific revision rate ratios and implant failure rate ratios at 3 months, 6 months, 1 year, 2 years and 5 years; the overall revision and implant failure rates stratified by treatment group for the entire study period and the same time points; and the change in postoperative European Quality of Life 5 Dimensions (EQ5D) score, Musculoskeletal Tumor Society Score (MSTS) and Toronto Extremity Salvage Score (TESS) at the previously mentioned time points. Revision surgery is defined as any repeated surgery on the operated proximal femur with a skin incision, with or without the exchange of bone-anchored components. Implant failure is defined as a revision surgery involving the exchange of any bone-anchored component of the IF device or endoprosthesis.

Eligibility criteria

We established the following eligibility criteria for assessing studies for inclusion in our review: randomized trials, prospective or retrospective cohorts and case–control studies reporting on the revision and/or implant removal rates of IF devices and EPR in patients treated surgically for MBD or hematologic malignancies of the proximal femur. Articles in English, German, Danish, Swedish or Norwegian published in peer-reviewed journals were eligible for inclusion. We excluded reviews, meta-analyses, abstracts and conference proceedings. Case series involving fewer than 20 patients were excluded, while case series with 20 or more patients were included for the subgroups with a sufficient number of patients. Furthermore, we omitted studies where inclusion was contingent upon conditions other than MBD or hematologic malignancy, where the occurrence or absence of events was an inclusion criterion, where outcomes of interest for the specified population could not be extracted and where data could not be separately extracted for patients undergoing IF or EPR. If the same study population was presented in more than one article, only the article with the newest data was included.

Search strategy and study selection

The search strategy was developed in cooperation with an information specialist from Rigshospitalet’s Research Library. We searched MEDLINE and Embase from their inception until April 27, 2023. We updated the search on October 12, 2024. The search string was developed stepwise, combining medical subject headings (MeSH/Emtree) and free-text terms, with separate search strings integrated through Boolean operators. We employed a validation set comprising 14 articles, previously identified as relevant through preliminary research by MSA. The search strategy was iteratively refined until all 14 articles from the validation set were successfully captured in both databases. We also reviewed the references of all screened full texts for additional sources. The complete search strategy for each database is detailed in the appendix. The search results from each database were imported into Covidence, where duplicates were removed, and titles and abstracts were screened. Two investigators (AI, MSS) independently screened the titles and abstracts. Any discrepancies were resolved through discussion between the two investigators until a consensus was reached.

Data extraction

Two investigators (AI, MSS) extracted data using a pre-piloted data extraction sheet. Data were extracted at the study level and pooled across all participants. We extracted the country of origin, study design, objective, study period, eligibility criteria, interventions and the number of patients stratified by osteosynthesis with plates and/or screws, IMN, EPR with a modular stem or EPR with a non-modular (standard) stem. Furthermore, we extracted the total number of patients stratified by treatment group, age (measure of central tendency reported by the study), proportion of males/females, proportion of complete/impending fractures, range of follow-up in months or years and overall survival. Age, sex, fracture status, follow-up and survival data were pooled if the study presented this separately by treatment group. The outcomes revision surgery and implant removal were extracted as the total number of patients experiencing events by treatment group as previously detailed and overall for any cause stratified by cause if specified and stratified by the above-mentioned time points if specified.

Quality assessment

Study quality was assessed independently by two investigators (AI, MSS) using the Methodological Index for Non-Randomized Studies (MINORS) instrument (10). This instrument is a validated 12-item tool that has demonstrated good inter-rater agreement and internal consistency. It comprises 12 items, eight applicable to non-comparative studies and four additional items for comparative studies. Each item is scored as 0 (not reported), 1 (inadequately reported) or 2 (adequately reported), yielding a maximum score of 16 for comparative studies and 24 for non-comparative studies. Any discrepancies were resolved through discussion until a consensus was reached.

Data synthesis

We were unable to identify any eligible randomized trials, nor a sufficient number of high-quality prospective cohort studies comparing IF to EPR. Consequently, we refrained from conducting the pre-planned meta-analysis or reporting on the comparative outcomes from the protocol. We instead chose to only present our findings descriptively, separately for each type of treatment (we refer the reader to the ‘Protocol deviations’ section). We detail the characteristics of each study and the outcomes for revision and implant removal rates, both overall and according to cause. The results are presented descriptively, without statistical tests, as we determined that statistical analysis would not yield meaningful insights given the data constraints. We present results only for treatment groups comprising ≥20. Results for subgroups with <20 patients within a study are not reported. A majority of the included studies did not apply survival methods to report the risk of revision or implant removal but instead reported these results as the percentage of the study cohort experiencing a revision or implant removal. To present results consistently across studies, we extracted the number of patients experiencing a revision or implant removal for the maximum available observation period and then proceeded to calculate revision rate and implant removal rate as a simple percentage of the total number of patients reported to receive a certain treatment. When possible, this was also extracted by time point. These results were summarized across studies stratified by treatment type and presented as range accompanied by the median and interquartile range (IQR), presented as the 1st and 3rd quantiles.

Study selection and study characteristics

We included 34 studies in the qualitative synthesis (Fig. 1) (11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44). All but two of the included studies were retrospective. Seventeen were comparative, of which 15 reported data for both IF and EPR. However, only 12 studies reported on 20 or more cases in each treatment group. For seven of these 12 studies, the decision for treatment allocation was either not reported or reported at the discretion of the treating physician (Supplementary table 1 (see section on Supplementary materials given at the end of the article)). For the remainder of the comparative studies, there was a tendency toward assigning patients with poorer overall health and prognosis to IMN instead of EPR. Because of this, we determined that the imbalance was too high for any meaningful comparisons of IF and EPR. We therefore restricted the reporting of study outcomes to presenting revision and implant removal rates for each identified treatment, without performing any statistical comparisons between treatment groups.

View Full Size

PRISMA flow chart of the study inclusion process.

Citation: EFORT Open Reviews 10, 2; 10.1530/EOR-24-0138

• Download Figure
• Download figure as PowerPoint slide

Revision

The revision rates ranged between 0 and 12.4% (median: 5.1% and IQR: 2–6.5%) in 12 studies on patients treated with EPR using a modular prosthesis, between 0 and 6.9% (median: 2.4% and IQR: 0.0–4.2%) in eight studies using a standard prosthesis and between 0 and 11.4% (median: 2.4%, IQR: 4.2–7.5%) in seven studies reporting combined results for modular and standard stems (Table 1). The primary reasons for revision were dislocations/instability and infections (Table 1). For IMN, the revision rate varied from 0 to 26.7% (median: 5.8%, IQR: 3.1–10%) according to 18 studies, with nonunion and hardware failure being the predominant reason for revision (Table 2). We only identified six studies reporting on other methods of IF (Table 3). Two studies on dynamic/sliding hip screws (DHS/SHS) reported revision rates of 16.7 and 17.1%, respectively. Three studies on patients treated with percutaneous screw fixation using cannulated or cancellous screws reported revision rates of 8.0, 8.2 and 15.5%. In addition, one study that combined results for DHS and plate and/or screw fixation reported an overall revision rate of 13.3% for these treatments. The main reasons for revision in all these cases were hardware breakage, disease progression and fractures (Table 3).

Implant removal

The implant removal rates ranged between 0 and 4.0% (median: 1.3%, IQR: 0–3.1%) in 11 studies on patients treated with EPR using a modular prosthesis, between 0 and 5.2% (median: 0%, IQR: 0–1.9%) in eight studies using a standard prosthesis and between 0 and 8.3% (median: 0.5%, IQR: 0–3.6%) in six studies reporting combined results for modular and standard stems (Table 1). For IMN, the implant removal rate varied from 0 to 26.7% (median: 5.3%, IQR: 2.2–7.3%) according to 16 studies (Table 2). For the remaining studies concerning IF, implant removal rates were 16.7% for DHS, 5.0, 8.0 and 8.2% for the three studies with cannulated screws and 13.3% for one study on combined results for various methods of open reduction and IF (Table 3). Results concerning time-specific revision/implant removal rates and MSTS scores are detailed in Tables 4 and 5 but are not discussed further.

Quality assessment

The included studies were determined to be of low to moderate quality (Table 6). Scores ranged from 2 to 11 out of a maximum of 16 for non-comparative studies and from 3 to 20 out of 24 for comparative studies. The median scores were 6/16 for non-comparative studies and 10/24 for comparative studies. Most of the included studies adequately reported their aims, follow-up periods and loss to follow-up. However, 19/34 studies did not report or inadequately reported the inclusion of consecutive patients. Similar reporting deficiencies were observed for prospective data collection in 33/34 studies and for the use of appropriate endpoints in 25/34 studies. Notably, all 34 studies did not report on whether endpoints were assessed blinded and did not provide a reason for unblinded assessment, while 32/34 did not report whether a sample size calculation had been performed. In the comparative studies, 18/20 did not report or inadequately reported on the adequacy of the control group, 19/20 did not report or inadequately reported on baseline equivalence between groups, and 19/20 studies did not or inadequately report on the statistical analyses used.

Limitations

Our systematic review offers insight into the proportion of patients undergoing treatment for MBD of the proximal femur that can be anticipated to require revision and implant removal. However, limitations arise from the variability in study quality and methodological heterogeneity among the included studies, thereby constraining the extent of conclusions that can be drawn from this analysis. A primary limitation of our review is the inability to address the pre-specified primary outcomes. Our objective was to compare IF to EPR for treating MBD in the proximal femur. However, treating the proximal femur as a singular entity may oversimplify the complexities involved. For instance, management approaches differ between pertrochanteric and subtrochanteric pathologic fractures compared to fractures involving the femoral head and neck (45), with the consequence that interventions with inherently different and potentially different revision rates were at risk of being pooled together. Cannulated screws and IMN serve as examples, each with distinct surgical indications. Addressing our research question ideally necessitates studies with patients of similar performance status, anticipated survival and metastatic lesions localized to similar and specific regions of the proximal femur, wherein patients are randomized to either specific types of IF or a particular type of EPR. However, such studies were not found, and the comparative retrospective studies either assigned patients with poorer overall health to IF or lacked clarity on treatment allocation (Supplementary table 1). This raises the question of whether randomized trials comparing IF to EPR are necessary or ethical for these patients if IF is considered inherently inferior to EPR.

We also identified a limitation in the reporting of revision and implant removal rates: survival among patients treated for MBD is poor (46), making death a relevant competing risk worth considering when estimating these rates (45). However, only a few of the included studies calculated cumulative incidences while accounting for competing risks. The preferred reporting method was the total number of patients receiving a treatment and the total number of events in each treatment group. Consequently, the revision and implant removal rates in this review do not adjust for competing risks or observation time. As evident from Table 4, the majority of studies did not report times until events. Thus, when comparing outcome rates across studies, it remains unclear whether differences could be attributed to variations in observation time or survival. Finally, we only found six studies on IF with cannulated screws, DHS/SHS or a combination of methods other than IMN. The scarcity of the literature suggests that either the fracture pattern in metastatic lesions does not warrant these interventions or these treatments have become less common and been replaced by other methods, such as hip arthroplasty, which is an alternative for lesions around the femoral neck. It is worth noting that some of the included studies also reported on DHS/SHS and cannulated screws but included <20 cases, which is why those results were not presented (Supplementary table 1). A second look at this group shows revision rates of 3/15 (20%) (19), 3/13 (23%) (33) and staggering 8/19 (42%) (26). Given the infrequent use of these methods and their high revision rates, caution is advised against using these methods for treating MBD of the proximal femur. Nevertheless, it must be acknowledged that there is limited knowledge regarding the revision rate and implant survival for these treatments in patients with MBD.

Protocol deviations

We registered the study protocol before screening and data extraction but made deviations during these phases. We decided to exclude studies published before the year 2000 to reflect current standards. We initially planned to stratify groups as either IF or EPR. However, full-text screening revealed that most IF studies focused exclusively on IMN or reported IMN outcomes separately from other IF methods, while EPR studies often separately reported results for subgroups with modular vs standard stems, without sufficient details on acetabular components. Thus, we extracted and presented outcomes separately for EPR with modular stems, EPR with standard stems, IMN, DHS/SHS and cannulated screws whenever possible. Groups were pooled as EPR or IF only when treatment-specific extraction was not feasible. We abandoned our pre-planned use of the Cochrane risk-of-bias tool for randomized trials, version 2 (RoB 2; https://www.riskofbias.info/welcome/rob-2-0-tool), and the Newcastle–Ottawa Scale (NOS) after full-text screening. We only identified one prospective cohort study and one trial, making RoB 2 obsolete. We chose MINORS over NOS, as none of the studies were case–control studies, and MINORS offers a more detailed assessment of cohort studies. None of the studies were randomized trials comparing EPR to IF. We therefore did not perform the pre-planned meta-analysis or report on the protocoled co-primary or comparative secondary outcomes. We only reported the total number of patients experiencing specific causes over the overall period, as presented in the articles. Although we planned to report changes in EQ5D, MSTS and TESS scores over time, no studies reported EQ5D or TESS at the defined time points, while MSTS was only presented as a mean score at specific time points. Therefore, we reported the mean MSTS at the protocoled time points instead of changes over time.

Revision

We observed that the reasons for revision varied between IF and EPR. Both IMN and other methods of IF were primarily revised due to nonunion, disease progression and hardware breakage, whereas EPR was primarily revised due to instability/dislocation and infections (Tables 1, 2, 3). The majority of studies did not specify whether instability surgery necessitated the removal of the implant or merely involved a closed or open reduction of the prosthesis. Furthermore, there was a tendency for revision rates to peak within one year in studies that reported time-specific revision rates for EPR. Conversely, the revision rate tended to continuously increase up to 5 years for IF (Table 4). These two phenomena could be related – dislocations are known to primarily occur within the first 3 to 6 months (5, 47). Conversely, hardware breakage occurs over a longer period due to repetitive stress causing material fatigue, ultimately leading to breakage. Meanwhile, the progression of a metastatic lesion can be a prolonged process. The lesion must progress to a point where the bone either fractures at a location other than the implant or until the structural integrity of the surrounding bone is compromised to such an extent that the load-sharing becomes disproportionate relative to what the implant can withstand. However, as commented earlier, data on the temporal aspect of revision were scarce and further studies are needed to guide surgeons in choosing the most suitable treatment strategy, given the performance status, expected survival and functional needs of the patients.

Implant removal

We observed that implant removal rates closely corresponded to the revision rates for both IMN and other methods of IF, while the implant removal rates were lower than the revision rates for EPR (Tables 1, 2, 3). This is likely related to the underlying causes, in which dislocation in EPR may be treated with closed or open reduction without the removal of bone-anchored components, while infections can be managed through antibiotics or debridement and irrigation. In contrast, for IF implants, disease progression or nonunion with physical loads across the joint that exceed the capability of the implant will require an exchange to another implant. This suggests that EPR might be more appropriate for patients with MBD, even despite their short life expectancy, if this can avoid the patient undergoing a major revision. However, as noted previously, information on the temporal aspects of implant failure is scarce, which limits interpretation. Moreover, a unilateral focus on implant survival does not convey any information on the needs of the patient. It is just as important to consider rehabilitation, the time it takes to ambulate and to what extent the patient can ambulate. These outcomes are best monitored through patient-reported outcomes. Only four studies reported the MSTS at 6 months, while two studies reported MSTS at 1 year. The remainder of the studies either did not include EQ5D or TESS scores at the protocoled time points or reported them in such a way that data could not be extracted (e.g., graphically or pooled across differing time points). However, it is worth noting that our search strategy was developed to identify revision and implant removal, so we could have overlooked the relevant literature reporting on the progression in patient-reported outcomes and ambulation over time in these patients.