Survivorship and clinical outcomes of proximal femoral replacement in non-neoplastic primary and revision total hip arthroplasty: a systematic review

Background Several studies have evaluated the survivorship and clinical outcomes of proximal femoral replacement (PFR) in complex primary and revision total hip arthroplasty with severe proximal femoral bone loss; however, there remains no consensus on the overall performance of this implant. We therefore performed a systematic review of the literature in order to examine survivorship and complication rates of PFR usage. Methods A systematic review of the literature according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines was performed. A comprehensive search of PubMed, MEDLINE, EMBASE, and the Cochrane Database of Systematic Reviews was conducted for English articles using various combinations of keywords. Results In all, 18 articles met the inclusion criteria. A total of 578 PFR were implanted. The all-cause reoperation-free survivorship was 76.6%. The overall complication rate was 27.2%. Dislocation was the most common complication observed and the most frequent reason for reoperation with an incidence of 12.8 and 7.6%, respectively. Infection after PFR had an incidence of 7.6% and a reoperation rate of 6.4%. The reoperation rate for aseptic loosening of the implant was 5.9%. Overall, patients had improved outcomes as documented by postoperative hip scores. Conclusion PFR usage have a relatively high complication rate, however, it remains an efficacious treatment option in elderly patients with osteoporotic bone affected by severe proximal femoral bone loss. Modular designs have shown reduced dislocations rate and higher survivorship free from dislocation. However, PFR should only be used as salvage procedure when no other reconstruction options are available.

In the setting of femoral revision arthroplasty, significant bone loss continues to be problematic for healthcare professionals, potentially threatening the primary fixation and durability of the reconstruction. Reduced bone stock can be associated with septic and aseptic failure, periprosthetic fracture, osteoporotic fracture in the elderly with severe comminution or failed fracture fixation, and multiple revisions [2][3][4][5][6][7]. In case of severe proximal femoral bone loss multiple treatment options have been described in the literature, including structural allograft-prosthesis composite, impaction allografting, long revision stems, resection arthroplasty, and proximal femoral replacement (PFR) [3,4,8]. Proximal femoral replacement, also known as "megaprosthesis", is a wellestablished limb salvage procedure for reconstruction of bone defects after the oncological resections of malignant bone neoplasms [9] and the encouraging outcomes have broadened the indications to the treatment of severe bone loss in non-oncologic conditions [10,11].
PFR allows a faster recovery especially in elderly patients, and it avoids the disadvantages of bone grafting such as resorption, graft integration, and diseases transmissions [12,13]. However, it is associated with an increased risk of infection and instability, secondary to the difficult healing of the abductor mechanism [14].
Multiple studies have described the outcomes of PFR in oncologic patients [9,10], however, only few have described PFR usage in non-oncologic severe femoral bone loss associated with periprosthetic fracture, septic and aseptic revisions, or failed osteosynthesis. We therefore performed a systematic review of the literature in order to examine survivorship and complication rates of PFRs. Specifically, we aimed to examine (1) what is the survivorship from reoperation when PFRs are used? (2) what complications are most common in PFRs? (3) what is the cumulative incidence of hip dislocation with PFRs? (4) do PFRs provide adequate implant survivorship in line with alternative treatment methods in the management of severe femoral bone loss? and (5) what are their clinical outcomes?

Search strategy
This search was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines [15]. The US National Library of Medicine (PubMed/MEDLINE), EMBASE, and the Cochrane Database of Systematic Reviews were queried for publications utilizing various combinations of the search terms "proximal femur replacement, " "proximal femur megaprosthesis, " "hip megaprosthesis, " "proximal femur arthroplasty, " "non-oncologic, " "bone loss, " "femoral reconstruction, " "bone defect, " "femoral revision, " in combination with the Boolean operators (AND, OR, *) since inception of database to January 2021. No limit was set with regard to the year of publication. Two authors (F.Ma. and V.D.M.) independently conducted all the searches and screened the titles and abstracts to identify relevant studies. Differences were resolved by consulting a third senior reviewer (I.D.M.). Only abstracts that evaluated the clinical outcomes and survivorship of nononcologic patients with PFR following primary or revision THA were reviewed. If the title and abstract of each study contained insufficient information, the full manuscript was reviewed. An additional search was conducted by screening the references list of each selected article.

Inclusion and exclusion criteria
Inclusion criteria were any original study in which a PFR was used in primary or revision THA in non-oncologic patients. Postoperative complications, clinical outcomes using validated patient reported scales and implant survivorship where reported. Exclusion criteria were case reports, surgical technique reports, review articles, expert opinions, letters to editors, biomechanical reports, instructional course lectures, studies on animals, cadaver or in vitro investigations, book chapters, abstracts from scientific meetings, unpublished reports, studies with less than 5 hips, studies with a mean follow-up less than 1 year, studies using the same database of patients, studies reporting the use of PFR in oncologic patients, and studies written in non-English language. If a duplicate population was noticed, the study with the longer mean follow-up was included to avoid including the same patients twice.

Data extraction and collection
Two independent reviewers (F.Ma. and V.D.M.) separately examined all the identified studies and extracted data. During initial review of the data, the following information was collected for each study: title, first author, year of publication, study design, number of patients, patients died and lost at follow-up, age of patients, length of follow-up, indication for index surgery, PFR implant used, complication types, reoperations for any reason, implant loosening, dislocations, deep infections, nerve injuries, and patient-reported outcomes. The level of evidence analysis was determined using the Oxford Centre for Evidence-Based Medicine Levels of Evidence [16]. The methodological quality of each study and the different types of detected bias were assessed independently by each reviewer with the use of Modified Coleman Methodology Score (Fig. 1). The Modified Coleman Methodology score ranges from 0 to 100, with a higher score reflecting higher quality. Final score was categorized as excellent (85-100 points), good (70-84 points), fair (55-69 points), and poor (<55 points). Selective reporting bias was not included in this analysis. Implant failure was defined by need for revision or resection of femoral and/ or acetabular component.
Categorical variables were presented as frequency and percentages. Continuous variables were presented as means. A P-value <0.05 was considered statistically significant.

Study selection
The search query resulted in 2473 abstracts that were then examined to determine if they met the inclusion criteria related to the outcome of PFR for primary and revision THA in non-oncologic patients (Fig. 2). Following elimination of duplicate articles, predetermined inclusion and exclusion criteria were applied. In total, 18 articles met the inclusion criteria and were included in the final analysis [3,4,[17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32] (Table 1). Consensus on which articles would be analyzed in the present study was achieved by discussion between the reviewers based on the predetermined inclusion and exclusion criteria described above.

Quality assessment
The quality of the studies was variable, with the average modified Coleman methodology score of the included studies of 42 points (fair, range 33-50 points), showing that the quality of the studies was low. A meta-analysis was not undertaken due to the general poor quality of the studies. (Table 1)

Demographic data
A total of 626 patients and 632 hips were initially included in this analysis. After excluding 54 hips (8.5%) due to missing data or loss to follow-up, 578 hips with a mean age of 70 years (range, 59-79 years) at the time of surgery were included for the final analysis. The mean follow-up was 4.3 years (range, 1.3-11 years) ( Table 1).
Other indications including osteosynthesis failure, and non-union were reported in 4.8% of the cases (30 hips). Additional information is further outlined in Table 2.

Reoperations
The all-cause reoperation-free survivorship after PFR implantation was 76.6% (445 of 578 hips). The overall reoperation rate was 23.4% (135 of 578 hips) at mean follow-up of 4.3 years (range, 1.3-11 years) ( Table 3). Dislocation and infection were the most common reasons for reoperation with an incidence of 7.6% (44 of 578 hips) and 6.4% (37 of 578 hips), respectively. The reoperation rate for aseptic loosening of the implant was 5.9% (34 of 579 hips). Reoperation rate due to other complication including periprosthetic fracture and hematoma was 3.1% (19 of 578). Among those, periprosthetic fracture was the most frequent complication requiring surgery in 12 hips (2.1%).

Dislocation
Dislocation after PFR insertion was the most frequent post-operative complication observed. The overall prevalence of dislocation was 12.8% (74 of 578 hips) ( Table 3). In case of dislocation, conservative treatment with closed reduction was performed in 40.5% (30 of 74 dislocations), and reoperation was required for 59.5%% of all dislocations (44 of 74 dislocations) ( Table 3). Among those, revision or resection arthroplasty were performed in 81.8% (36 hips). Open reduction was performed in 6 hips (13.6%) [28], and 2 hips were treated with advancement and reattachment of the greater trochanter and by a firm closure of the fascia lata [18].

Aseptic loosening, infection, and other complications
The incidence of implant aseptic loosening for either femoral stem or cup was 5.9% (34 of 578 hips). All cases of aseptic loosening reported required further reoperation.
Among those, revision of the acetabular component was performed in 47.0% of the cases (16 of 34 loose implants), revision of both components was performed in 20.6% (7 of 34 hips), revision of the femoral component in 23.5% (8 of 34 hips), and resection arthroplasty in 8.8% (3 of 34 hips). The overall reoperation rate due to aseptic loosening of the implant was 5.9% (34 of 578 hips).
The overall incidence of infection was 7.6% (44 of 578 hips), of those, further reoperation was required in 84.0% of the cases (37 hips), whilst 15.9% of the cases (7 hips) were treated conservatively with suppressive antibiotic therapy. Of the infected hips that required surgery, revision or resection arthroplasty was performed in 23 cases (of 37 hips, 62.2%), whilst debridement, antibiotics, and implant retention (DAIR) was performed in 14 cases (of 37 hips, 37.9%) ( Table 3). The overall reoperation rate due to infection was 6.4% (37 of 578 hips).
Other complications were reported in 41 cases (of 578, 7.1%). Among those, surgical treatment was required in 46.3% of the cases (19 of 41 complications), and periprosthetic fracture was the most frequent complication that required subsequent surgery (12 hips, 2.1%). In case of periprosthetic fracture, revision of the implant was reported in 75% of the cases (9 out of 12 hips), and osteosynthesis in 25% (3 hips). Complications that did not   [23,24]. It is a patient-reported measure of function designed to assess physical disability for patients after limb-salvage surgery for musculoskeletal tumors. Its lower extremity version consists of 30 questions regarding everyday activities such as dressing, working, mobility and leisure and allows a percentage score to be calculated. The mean postoperative score was 69.5% (range 68-71%). Further information on clinical outcomes is outlined in Table 4.

Discussion
Our review of the literature suggests that PFR implants are an effective way to manage severe femoral bone loss in non-oncologic primary and revision THA in case of elderly and less active patients with multiple comorbidities where an early mobilization and immediate full weight bearing are crucial for a faster recovery [33]. The all-cause reoperation-free survivorship after PFR implantation was 76.6% (447 of 578 hips) at a mean follow-up of 4.3 years. The overall complication rate was high at 27.2% (157 of 578 hips), with dislocation as the most commonly reported (12.8%), followed by infection (7.6%), and implant AL (5.9%), suggesting that despite acceptable short-to mid-term survivorship given the high complexity of these patients, PFR should be considered as a salvage procedure when other reconstruction options are no longer available. In addition, PFR are currently used in multiple settings, the most common indication was periprosthetic fracture in 29.    improved functional outcome and implant survivorship at long-term follow-up. In addition, they allow the reattachment of the gluteus and iliopsoas tendons supporting hip biomechanics and increasing postoperative hip stability. However, infection, non-union, allograft resorption, periprosthetic fracture and risk of disease transmission continue to be major issues, and the final outcome is strictly related on the etiology, soft tissue damage, type of bone defect, method of reconstruction, and preparation of the allograft [12,13]. Despite the effectiveness of PFR in restoring function, overall complications rate is high. Hip dislocation was the most frequent complication with an overall prevalence of 12.8 % (74 of 578 hips), higher than the one usually seen after revision THA with conventional implants at shortto-midterm follow-up [34]. Among those, the 59.5% required subsequent reoperation (44 hips), whilst 40.5% were treated nonoperatively (30 closed reductions). In case of reoperation for recurrent dislocation, the majority (81.8%) required a full revision or resection arthroplasty, suggesting the complex management after dislocation in these kinds of patients. However, if excluding the studies that used monobloc and custom-made implants and considering only the ones that implanted modular PFR implants [3,4,[20][21][22][23][24][25][26][27][28][29][30][31][32] the overall dislocation rate was 10.3% (51 of 497 hips), suggesting that modularity enables the surgeon to restore better offset, limb length and soft-tissue tension, providing better postoperative stability compared to older monobloc implants. In addition, our results showed a slightly reduced dislocation rate compared with the rates reported by Vaishya et al [35] that noted a dislocation rate of 14.6% (out of 245 PFRs) at a mean 44 months follow-up in a critical analysis of proximal and distal femoral replacement, and by Korim et al [33] that noted a dislocation rate of 15.7% (out of 356 PFRs) at mean 45 months follow-up in non-oncologic conditions, suggesting that newer implants may provide increased stability. In fact, if we consider only the recent literature (after 2010) [4,[22][23][24][25][26][27][28][29][30][31][32], the dislocation rate results considerably reduced compared to the previous studies probably related not only to the modular systems but also to the increased usage of constrained acetabular liners and dual mobility cups [36][37][38][39]. Recurrent dislocation is multifactorial and commonly related to multiple previous procedure and inadequate soft tissue envelop with abductor mechanism deficiency, especially of the posterior vertical fibers of the gluteus medius are considered the main actor in providing dynamic hip stability [40,41]. In addition, most of the patients included are relatively old with multiple comorbidities that can increase the risk of dislocation [42][43][44]. Modular implants present a porous-coated proximal surface that promotes osseointegration with the possibility to approximate the retained proximal host bone to the implant enhancing implant's stability [3]. Moreover, a possible solution to reduce the dislocation rate in case of abductor deficiency was proposed by Du et al. [45]. The authors used a band-shaped artificial ligament wrapped spirally around the proximal site of the total femur prosthesis for periacetabular soft tissue reconstruction in a cohort of 48 PFRs implanted for neoplastic reasons and reported a reduced dislocation rate [46]. Infection was the second most frequent complication reported in 7.6% of the cases (44 of 578 hips), in line with what have been previously reported [32,34,47]. Among those, further surgery was required in 84.1% of the cases (37 of 44 infections), while conservative treatment with suppressive antibiotic therapy was reported in 15.9% of the cases (7 of 44 infections). Among the PJIs that underwent subsequent surgery, 14 of them were DAIR (37.8%) and 23 of them were either 2-stage revision to a total femur replacement or resection arthroplasty (62.2%). PJI remains the most challenging complication after PFR because of poor quality soft tissue, poor overall health, and long operative times [24,48,49]. Currently, surface coating with iodine and silver [50] have shown a reduced infection rate, improving implant retention and reducing amputations in case of PJI, however, these implants were not clearly used in all the studies included and it was not possible to stratify the infection rate regarding this characteristic [32,51,52].
Aseptic loosening of the implant was reported with an incidence of 5.9% (34 of 578 hips), comparing favorably with what has been previously reported [35]. All cases of AL required subsequent reoperation, revision of the cup was performed in 47.1% of the cases (16 hips), followed by revision of the femoral component in 23.4% (8 hips), revision of both components in 20.5% (7 hips), and resection arthroplasty in 8.8% (3 hips).
Despite a relatively high level of complications for current treatment option, patients receiving a PFR showed considerable improvement in a variety of functional scores. Specifically, it was associated with an average of a 42.5 HHS increase between preoperative and postoperative periods. On average, patient improved from "poor" health (mean preoperative HHS of 35.7) to "fair" health (mean postoperative HHS of 72.8) at latest follow-up.
These results suggest that PFR remain an efficacious procedure, improving patient functionality and outcomes, especially in complex patients with severe proximal femoral bone loss.
There were a variety of limitations in this study. First, we were limited by the quality of the original studies, the variability in inclusion criteria as well as the methods for reporting the evaluated variables, and number of patients analyzed. Second, our methodology did not allow for identification of unpublished literature on PFR and is limited by potential publication bias. Several different outcome scores were used across the included studies to assess overall hip function. The studies included were heterogeneous, including small sample sizes and different implant used. The studies covered a large time period reflecting the use of variable implant designs from original monobloc to new modular implants. Limited information available on the complications in each cohort did not allow to stratify for indication and provide a better overview on which indication of PFR is associated with better/worse outcomes.

Conclusion
To date, given the designs available and the current literature, PFR should be considered a valuable option in case of salvage procedure in complex patients affected by severe proximal femoral bone loss when other available reconstruction options cannot be utilized. Newer designs have shown improved stability and clinical outcomes; however, dislocation and infection remain major issues after PFR, and long-term survivorship has not been clearly defined. In conclusion, PFR should be considered as a limb salvage option in carefully selected patients when other options are not feasible.