Outcome of different reconstruction options using allografts in revision total hip arthroplasty for severe acetabular bone loss: a systematic review and meta-analysis

Introduction Several studies have reported good to excellent outcomes of revision total hip arthroplasty (rTHA) using allografts for treating severe acetabular bone defects. However, precise information on the impact of allograft type and reconstruction method is not available. Material and methods Systematic literature search was performed in Medline and Web of Science including patients with acetabular bone loss classified according to the Paprosky classification who underwent rTHA involving the use of allografts. Studies with a minimum follow-up of 2 years published between 1990 and 2021 were included. Kendall correlation was applied to determine the relationship between Paprosky grade and allograft type use. Proportion meta-analyses with 95% confidence interval (CI) were performed to summarize the success of various reconstruction options, including allograft type, fixation method, and reconstruction system. Results Twenty-seven studies met the inclusion criteria encompassing 1561 cases from 1491 patients with an average age of 64 years (range 22–95). The average follow-up period was 7.9 years (range 2–22). Structural bulk and morselized grafts were used in equal proportions for all Paprosky acetabular defect types. Their use increased significantly with the type of acetabular defect (r = 0.69, p = 0.049). The overall success rate ranged from 61.3 to 98.3% with a random effect pooled estimate of 90% [95% CI 87–93]. Trabecular metal augments (93% [76–98]) and shells (97% [84–99]) provided the highest success rates. However, no significant differences between reconstruction systems, allograft types and fixation methods were observed (p > 0.05 for all comparisons). Conclusion Our findings highlight the use of bulk or morselized allograft for massive bone loss independent of Paprosky classification type and indicate similar good mid- to long-term outcomes of the different acetabular reconstruction options using allografts. Clinical trial registration PROSPERO: CRD42020223093. Supplementary Information The online version contains supplementary material available at 10.1007/s00402-023-04843-9.


Introduction
Reconstruction of severe acetabular bone defects during revision total hip arthroplasty (rTHA) represents a challenging procedure with a variety of available options [1]. Current projections describe a significant increase in rTHA of 43-70% by 2030, especially in patient groups aged 55-64 and 65-74 years [2]. To ensure the best possible service life of the prosthesis, defects of the acetabular bone stock must be addressed using a suitable reconstruction method. This includes the selection of appropriate implants during surgery planning along with the precise classification of the bone defect [3]. The appropriate type of acetabular bone graft is typically selected depending on the classification of the bone stock deficit. While cavitary defects may be restored with cancellous morselized auto-or allografts, larger segmental bone stock defects are usually addressed with structural (or bulk) cortico-cancellous auto-or allografts [4].
First described in 1994, the Paprosky classification was developed to grade acetabular bone loss [5] and has become the most widely used acetabular defect classification system worldwide [6]. Besides the description of the severity of bone defects, the classification was also developed to predict the required bone graft and implants. Whereas Paprosky types 1 and 2 defects are mostly treated with morselized bone graft, a bulk graft is recommended for type 3 defects [5,7].
Several studies have documented good to excellent clinical and radiological outcomes of rTHA using allografts [8][9][10][11]. Our group has previously demonstrated successful histological incorporation of morselized and bulk allografts in acetabular reconstruction, thereby providing a scientific basis for their successful use in rTHA [12,13]. However, differences in outcomes according to type of allograft, bone defect and reconstruction devices have not been systematically investigated. Current reviews focused on overall treatment options but not on the reconstruction device and bone graft in relation to the defect type [1,14,15]. So far these reviews depicted that trabecular metal (TM) systems, comprising augments and shells, demonstrated the best results for large acetabular defect reconstruction, referring to rerevisions and radiographic loosening [1,14].
Therefore, aim of the present study was to provide a systematic overview of mid-and long-term outcomes of rTHA using morselized or bulk allografts considering different surgical reconstruction options. Specifically, this systematic review and meta-analysis compiled studies with the purpose to (a) investigate the use of allograft types according to the acetabular bone loss classified according to Paprosky and (b) to evaluate the success rates of various cup or reconstruction devices combined with morselized or bulk allografts.

Methods
The findings are described according to MOOSE guidelines for reporting meta-analyses [16]. Study protocol was registered with the international prospective register of systematic reviews (http:// www. crd. york. ac. uk/ PROSP ERO/) and can be accessed under registration code CRD42020223093.

Search strategy
Medline (National Library of Medicine) and Web of Science (Clarivate Analytics) were screened for literature published from 1st January 1990 to 31th December 2021. The search was supplemented by records identified through the bibliographic lists of included references. The detailed search history is listed in Suppl. Table 1.
Titles and abstracts were screened. If title or abstract did not provide sufficient information, full texts were accessed. Two investigators (AS, TR) independently screened titles, abstracts, and full texts according to predefined inclusion and exclusion criteria (Table 1). Incongruities were resolved by consensus discussion. To minimize screening bias, the systematic review tool Rayyan was used [17].

Quality assessment
Two investigators (AS, CKB) independently evaluated the quality of the full texts with consensus discussion in case of discrepancies. Quality assessment was performed with the modified Downs and Black checklist [18] classifying quality according to four quality levels: excellent (28-26 points), good (25-20 points), fair (19-15 points), and poor (≤ 14 points) [19].

Data analysis and meta-analysis
For data synthesis, the main findings are summarized in Table 2. Cage fractures, broken hooks/plates, broken or loose fixation screws, unspecified mechanic failures and unstable cups were summarized as combined implant failure. The infection rate is compounded from early postoperative and late (deep) infections, and the pooled dislocation rate integrates cases with early post-operative and recurrent dislocation. The main outcome "success rate"           was evaluated for each included study by subtracting the number of implant failures, aseptic loosening, and infections from the total number of surgical procedures. Dislocation was not included in the success rate since confounding factors had a significant impact on the occurrence of dislocation [20]. The investigated surgical reconstruction devices comprise reconstruction cages, meshes, cemented and cementless cups, TM acetabular augment, TM shells and customized special cups. Studies that used different allografts or reinforcement devices were excluded from this analysis. SPSS statistical program 26.0 (SPSS, Chicago, IL, USA) and Comprehensive Meta-Analysis software, Version 3 (Biostat, Englewood, NJ, USA) were used for data analysis. Missing mean values and variances were estimated by median, range and sample size [21]. Chi 2 statistics was used to investigate differences in allograft type by bone loss classification. Kendall rank correlation evaluated the association between frequency of allograft use and Paprosky grade. Mann-Whitney U test or Kruskal-Wallis H test was applied to investigate outcome differences between fixation method, allograft type and reconstruction devices.
For the main outcome variable, success rate, the effect size (proportion) was calculated combined in pooled effect proportion meta-analyses. The results were depicted with 95% confidence intervals (CI) using a random-effect model analysis [22]. I 2 -statistic was used to quantify the degree of heterogeneity. Values of 25, 50 and 75% correlate to predefined thresholds for low, medium and high degrees of heterogeneity [23]. All p values are two-sided with an alpha set at 0.05.

Literature search results
A total of 2459 studies obtained in Web of Science and 1973 records in Medline resulted in a list of 4432 possible studies. After excluding duplicates (n = 1074) and screening titles and abstracts for eligibility (n = 3358), 121 articles remained. Following full text review, 26 studies met the inclusion criteria and were included in this review. One additional study was added after identification through citation searching, resulting in 27 articles (Fig. 1).

Study characteristics
All included articles are retrospective follow-up studies published between 2003 and 2021. The Black and Downs quality assessment indicated poor quality in 18 cases and fair quality in 9 cases (Suppl. Table 2). Study characteristics and outcomes are described in Table 2. In total 1561 cases of rTHA (range 20-304 per study) from 1491 patients with an average age of 63.9 years (range 22-95) were included. The average follow-up period of 7.9 years ranged from minimum 2-22 years. The main indications for rTHA were aseptic loosening (n = 1224), infection (n = 35), recurrent dislocation (n = 13), stem fracture (n = 1) or other (e.g., adverse reaction to metal debris, failed bulk allograft) (n = 3). Seven studies did not report the indication for revision surgery [24][25][26][27][28][29][30].
In total, rTHA with allograft use was performed in 31 cases with Paprosky type 1, 135 cases with type 2A, 213 cases with type 2B, 135 cases with type 2C, 502 cases with type 3A and 309 cases with type 3B defects. Three studies did not provide the specific subtype of the classification, instead only the main Paprosky category was provided [26,31,32]. A wide range of different reinforcement components were used for acetabular reconstruction. The most frequently used reconstruction systems were cages applied in 539 cases, followed by meshes (331 cases), cementless cups (317 cases), cemented cups (188 cases), TM cups (78 cases), TM acetabular augments with cemented cup (58 cases) and customized special cups (50 cases). In 18/27 studies cemented acetabular cup fixation was used ( Table 2). Related to the pooled sample size this represents 71% all cases.
Considering bone grafts, morselized allografts (n = 998) were nominally utilized more often than bulk allografts (n = 563) in all type of defects (Chi 2 = 121.1; p < 0.001). However, the percentage distribution suggests that the grafts are used at a similar proportion and independently of the Paprosky grade (Chi 2 = 4.84; p = 0.435; Fig. 2A). Although there was a trend that bulk allografts were more likely used with higher Paprosky grades, there was no statistical 1 3 difference between the use of morselized and bulk allografts in any of the defect types (p Pap. 1-3 > 0.05). Namely, in Paprosky type 2 acetabular defects, morselized grafts were implanted in 39.6% and bulk allografts in 30.6% of the cases. In type 3, morselized grafts were applied in 58.4% and bulk allografts in 67.7%. The overall use of allografts increased significantly with the type of the acetabular defect (r Kendall = 0.69, p = 0.049; Fig. 2B).
We also evaluated the influence of different cup types and reinforcement devices. No significant differences were found between the types of acetabular fixation systems in term of success rate (Fig. 5)

Success rate in relation to allograft type, stratified by acetabular fixation method
Since no global differences were found in the preceding meta-analyses, we further investigated allograft types and reinforcement devices within each specific fixation method. 18/27 studies (66.7%) applied a cemented cup fixation. Within the group of cemented rTHA the pooled overall estimate of success was 90% [80-95]. Level of heterogeneity was medium (I 2 = 55.4%). In cemented rTHA, the success rate was 7% lower when using bulk allografts (85.5%  (Fig. 6 A). Yet, this difference was not statistically significant. In contrast, the pooled overall estimate of success within the group of cementless rTHA was 89% [80-94] with a success rate of 91% [77-97] for bulk and 87% [73-95] for morselized allografts, demonstrating a non-significant trend  for clinical improved results using bulk allografts in cementless rTHA (Fig. 6B).

Discussion
Revision total hip arthroplasty is a complex surgical procedure, which can be particularly challenging in situations of severe acetabular bone loss. Bone grafts and various reconstruction systems are widely used during rTHA to address severe bone loss. The purpose of this systematic review and meta-analysis was to evaluate the mid-and long-term outcome of different reconstruction options using allografts in rTHA. Overall, no differences between allograft types, fixation methods, reconstruction systems, and cup types were observed. Notably, all included studies were retrospective follow-up studies without control groups (evidence level: Type IV). Therefore, no high-level evidence on the use of allografts with different reconstruction devices could be reported, which is also reflected by the mainly poor quality in the Downs and Black quality assessment.
A postulated, clinically important predictor for the outcome is the reconstruction method [53]. Reconstruction rings with cemented cups are considered as reliable reconstruction method. Current systematic reviews and meta-analyses have previously studied the options for managing severe acetabular bone loss [1,14,15]. Trabecular metal augments and shells may demonstrate the lowest revision and radiographic loosening rate while bone grafting with metal mesh has not been recommended for the reconstruction of pelvic discontinuity and type 3B bone loss [14]. Complementary to these previous findings, this review included four studies reporting results on TM reconstruction devices, two of these using impacted morselized grafts with augments and cemented cups in 58 cases [25,38]. The other two studies described the results of 78 cases treated with cementless TM cups [26,35]. TM systems achieved the highest success rates in this meta-analysis, which is why these devices may represent a promising option to address severe acetabular bone defects. In total, studies with TM reconstruction systems report a pooled loosening rate of 2.2%, a dislocation rate of 3.7%, and an implant failure rate of 1.4%. Still, the number of included studies is small, which may be the reason for the lack of significant differences in comparison with other acetabular reconstruction devices.
Regarding the type of allograft, 17 studies used exclusively morselized and ten studies bulk allografts. Studies evaluating bulk allografts reported that gaps between graft and host bone were filled with additional morselized grafts. The long-term results of morselized and bulk allografts did not differ significantly. Morselized allografts, however, had a 2% higher overall success rate. When stratified by fixation, a trend towards improved outcome with cementless fixation and poorer outcome with cemented fixation was observed for bulk compared to morselized allografts. In search of mechanisms for aseptic loosening, we have previously demonstrated histological evidence of more fibrous ingrowth in cases with morselized compared to bulk allografts during rTHA with cemented cup fixation [12,13]. Although this appears somewhat contrary to the results of the present meta-analysis, the extent to which fibrous ingrowth at the cement-allograft or allograft-host bone interface contributes to aseptic loosening remains unknown. It is also likely that other factors (e.g., surgical technique, host-related factors) also influence the outcome.
In most articles, complications were presented in an aggregated format across all bone defect types. Only 2/27 studies categorized complications by Paprosky bone loss type [30,40]. Consequently, revisions and complications that have occurred cannot be related to a specific type of bone defect. It can be suggested that the lower success rate for bulk allografts is attributed to the preference for applying these types of grafts in more severe acetabular defects. However, since both graft types were applied in equal proportion in all Paprosky classifications, it may be concluded that Fig. 6 Random model metaanalysis forest plot for success rate of rTHA compared by allograft type stratified by acetabular fixation method. A Forest plot for allograft type comparison within the group of cemented rTHA. B Forest plot for allograft type comparison within the group of cementless rTHA bulk allografts tend do have a higher complication rate than morselized allografts independent of the applied acetabular reconstruction system.
On average, the clinical outcome of the examined patients, measured with HHS and MdA score, was good. Both scores are established and well validated instruments for the objective assessment of the functionality of the hip and reflect the subjective pain intensity. Overall, the results described good and encouraging mid-to long-term results of acetabular reconstruction with both cages and TM augments or shells using bulk or morselized allografts, supporting their use for severe bone loss. There were no significant differences between the reconstruction systems, fixation methods or graft types but descriptive slightly improved outcomes for the use of TM systems and morselized allografts. The method of acetabular cup fixation solely had no influence on the success or revision rate. Also, no significant differences were observed between cemented and non-cemented cups.

Strength and limitations
This study has clear strengths. While previous systematic reviews on various reconstruction options for acetabular defects were heterogeneous and not comparable, this review is focusing on different reconstruction options that included allograft use with precise inclusion and exclusion criteria. Our search strategy allows the meta-analytical comparison Fig. 7 Random model metaanalysis forest plot for success rate of rTHA compared by implanted reconstruction system stratified by acetabular fixation method. A Forest plot for reconstruction system comparison within the group of cemented rTHA. B Forest plot for reconstruction system comparison within the group of cementless rTHA of different influence factors affecting the outcome of rTHA, e.g., reconstruction device, fixation method, graft type.
The main limitation of this systematic review is that the source studies did not provide high-quality evidence. In view of fact that, contrary to Paprosky's original recommendation [5], morselized allografts were also used for type 3A and 3B bone defects, it should be considered to compare the use of bulk and morselized allografts in a randomized controlled trial to eliminate confounding variables. The differences in surgical approach, postoperative care protocols, age at primary revision, patient demographics, primary disease, number of cases (range 20-304), and time to follow-up (range 2-22 years) created limited comparability. However, the calculated I 2 -statistics for heterogeneity was medium in any of the meta-analyses conducted.
Two of the largest scientific databases for medical literature were systematically searched. When identifying studies, there was the difficulty that all cases had to be treated with an allograft. Often allografts were not primary outcome, resulting in the exclusion of numerous studies that did not clearly demonstrate that all patients received a bone graft. However, this approach has concurrently increased the consistent character of the studies. Furthermore, metanalytical results were affected by publication bias [54,55], meaning the tendency not to publish studies with non-significant or unfavorable results. Overall, the quality of the included study is low. No randomized-controlled trials were found to be included in the meta-analysis. However, for the outcome parameters examined, e.g., complication rate, only follow-up studies were suitable and applicable.

Conclusion
To our knowledge, this is the first systematic review analyzing the outcome of rTHA with allograft use in acetabular defects. TM system and cages appear to have the best success rate. The results further suggest that morselized and bulk allografts are used in equal proportions in all Paprosky grades, with no significant differences in clinical outcome. Since only retrospective studies could be included, additional research is needed to finally address this question.