Mobile-bearing versus fixed-bearing total knee arthroplasty: a meta-analysis of randomized controlled trials.

Objective The purpose of this study was to perform a meta-analysis comparing mobile-bearing with fixed-bearing total knee arthroplasty (TKA) in terms of all-cause revision rates, aspetic loosening, knee functional scores, range of motion and radiographic lucent lines and osteolysis. Methods PubMed, Cochrane Library, Google Scholar and Web of Science were searched up to January 2020. Randomized controlled trials that compared primary mobile-bearing with fixed-bearing TKA, reporting at least one of the outcomes of interest, at a minimum follow-up of 12 months were included. All outcomes of interest were pooled at short-term (< 5 years), mid-term (5 to 9 years) and long-term (> = 10 years) follow-up intervals. Results A total of 70 eligible articles were included in the qualitative and statistical analyses. There was no difference between mobile-bearing or fixed-bearing TKA at short-term, mid-term and long-term follow-ups in all outcome measures including all-cause revision rate, aseptic loosening, oxford knee score, knee society score, Hospital for Special Surgery score, maximum knee flexion, radiographic lucent lines and radiographic osteolysis. Conclusion The current level of evidence demonstrated that both mobile-bearing and fixed-bearing designs achieved excellent outcomes, yet it does not prove the theoretical advantages of the mobile-bearing insert over its fixed-bearing counterpart. The use of either design could therefore be supported based on the outcomes assessed in this study. Level of Evidence: Level II, Therapeutic Supplementary information The online version contains supplementary material available at (10.1007/s00590-021-02999-x).


Introduction
The design of the polyethylene insert has been debated numerously in the literature [22]. Fixed-bearing designs, which provide rigid fixation of the polyethylene insert within the tibial implant, have demonstrated satisfactory outcomes and long-term survival rates [1,38,45,64]. However, implant loosening in fixed-bearing designs was theoretically attributed to higher contact stresses and polyethylene wear rates [20,75], which motivated the pursuit of improved TKA designs. Mobile-bearing polyethylene designs were developed to mitigate the drawbacks of fixed-bearing TKA through improving the conformity, lowering contact stresses with the aim of mimicking the kinematics of the native knee [16]. However, these advantages are theoretical and yet to be fully proven in vivo. Furthermore, mobile-bearing TKA can introduce unique complications such as bearing dislocation [5].
Earlier meta-analyses have reported superior results with the mobile-bearing TKA [13,85]. Subsequent meta-analysis with mid-term follow-up had refuted such findings without any significant difference between mobile-bearing and fixedbearing TKA [55,81]. However, in June 2020 two recent meta-analyses with a limited number of studies presented further contradicting results, with one meta-analysis supporting long-term clinical outcomes in favor of mobile bearing, whereas the other meta-analysis refuted such findings [15,84]. Therefore, controversy continues to exist regarding the superiority of mobile-bearing over fixed-bearing designs.
This study aimed to provide an updated meta-analysis comparing mobile-bearing versus fixed-bearing TKA using a multi-modal method of outcomes to include overall revision rates, aseptic loosening, clinical as well as radiological outcomes. Our hypothesis was that no significant differences exist in all outcomes between the mobile-bearing and the fixed-bearing designs.

Materials and methods
This meta-analysis was conducted with adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [54]. The focus was randomized controlled trials that compared mobile-bearing with fixedbearing TKA. The primary outcome was the all-cause revision rate. The secondary outcomes were aseptic loosening rates, knee functional scores, maximum knee flexion, radiographic lucent lines and osteolysis.

Eligibility criteria
The inclusion criteria were randomized controlled trials that compared primary mobile-bearing with fixed-bearing TKA, reporting at least one of the outcomes of interest, a minimum follow-up of 12 months. Exclusion criteria were inaccessible full-text, abstracts and studies reporting outcomes of interest but with unextractable data for meta-analytic comparisons. Articles published in English were only sought. Studies that reported the same sample population were not excluded if the follow-up intervals were different. The exclusion criteria were non-randomized clinical trials and studies with a population reported in a previous study with an overlapping follow-up interval.

Information sources and search strategy
PubMed, Cochrane Library, Google Scholar and Web of Science were searched till January 2020.
The search strategy involved the use of the following keywords that involved synonyms of "total knee arthroplasty" AND "mobile bearing" AND "fixed bearing" AND "randomized controlled trials." Studies were screened by titles and abstracts. A full-text review was performed if a study matched the eligibility criteria. Furthermore, the references of each eligible article were manually searched to ensure eligible studies were not missed. The search strategy was performed by three authors independently. Any disagreement between the three authors in the search strategy was resolved by the senior author.

Data collection process and data items
The data items that were collected included: the first author's surname, study year, study location, age, sex, number of patients, type of prosthetic bearing used (mobile-bearing or fixed-bearing), the specific type of mobile-bearing prosthesis (rotating platform, rotating platform and gliding, and meniscal bearing), patella resurfacing, follow-up timepoints, all-cause revision rates, Oxford Knee Scores (OKS), Knee Society Scores (KSS), the Hospital for Special Surgery (HSS) knee scores, reported maximum knee flexion, radiographic radiolucent lines, radiographic osteolysis and rates of aseptic loosening. The OKS was transformed into the 0-48 scale to facilitate data synthesis. The Western Ontario and McMaster Universities Arthritis Index was not collected as it was reported variably among studies with the 0-96 Likert scores or the 0-100 visual analog scales. Data collection forms were used independently by three authors, with any arising disagreement in the collected data being resolved by the senior author.

Risk of bias in individual studies
The qualitative analysis was performed with the revised Cochrane risk-of-bias tool for randomized trials (RoB 2) [74]. The tool contains five domains that assesses the randomization, adherence to intended treatments, missing outcomes, measurement bias and reporting bias. Each study was assessed with the RoB 2 by three authors independently, and the final rating of each study was reviewed by the three authors and the senior author to arrive at a consensus.

Statistical analysis
Analysis was performed with the use of Stata/IC (StataCorp. 2019. Stata Statistical Software: Release 16. College Station, TX: StataCorp LLC.). The outcomes were estimated with the use of 95% confidence interval (CI). The risk ratio (RR) was utilized for dichotomous outcomes such as the revision rates and the aseptic loosening rates. The mean difference (MD) was used for expressing continuous outcomes such as the OKS, the KSS and the HSS knee score. The Hedge's G mean difference was used for maximum knee flexion due to potential variability in the range of motion measurements. The outcome measures of interest were pooled at three different follow-up intervals at short term (<5 years), mid-term (5 to 9 years) and long term (>=10 years). The meta-analytic models were based on random effects (RE) with the use of the DerSimonian-Laird method as a heterogeneity variance estimator [17]. The formulas developed by Hozo et al. [31] were used in studies that reported medians instead of means and ranges instead of standard deviations (SD).

Study selection
The search strategy resulted in 581 (569 articles from database search and 12 articles from manual references search) articles, of which 409 articles were excluded due to duplications. Subsequently, a total of 172 articles were screened by titles and abstracts, of which 67 articles were excluded. This resulted in a total of 105 articles that were eligible for full-text reviews, of which 35 articles were excluded. Thus, a total of 70 articles were included in the qualitative and statistical analyses. The PRISMA flowchart is displayed in Fig. 1.

Study characteristics
Among the 70 included studies, 4968 patients underwent mobile-bearing TKA and 5034 patients underwent fixedbearing TKA. The most utilized TKA implant was PFC Sigma® (DePuy) in 34.3% of all studies. A posterior-stabilized (PS) implant was routinely used in 60% of studies, whereas a cruciate-retaining (CR) design was routinely used in 25.7%. The rest of the studies used either CR or PS designs (4.3%) depending on the total knee system utilized, and 10% of studies did not specify whether the posterior cruciate ligament was sacrificed. The mobile-bearing designs used were a rotating platform in 81.4%, rotating and anterior-posterior gliding in 11.4% and meniscal bearing in 2.86%. Patella resurfacing was performed routinely in 48.57% of studies, unresurfaced in 22.86% and selectively resurfaced on a case-by-case basis in 17.14%. Study characteristics are summarized in Table 1.

Quality assessment
Low risk of bias was found in 27 studies, some concern for bias in 28 studies and high risk of bias in the remaining 15 studies. Most studies had a low risk of bias for deviation from intended interventions, missing outcome data, measurement of outcomes and in the selection of reported results.
In terms of randomization, 55.7% of included studies had a low risk of bias, 38.5% had some concern for bias, and 5.7% had a high risk for bias. A graphic summary of the qualitative assessment is displayed in Supplementary Fig. 1.  (Fig. 2).
The HSS knee score was reported in 8 studies at short term, 3 studies at mid-term and 3 studies at long term. The and long-term (MD −0.48; 95% CI −2.9, 1.95; P = 0.7; I 2 = 79.88%) follow-up intervals did not demonstrate any statistically significant difference for the HSS knee scores (Fig. 3).
The range of motion was reported in 27 studies at short term, 12 studies at mid-term and 6 studies at long term. No differences were significant between mobile-bearing and fixed-bearing TKA at any of the three follow-up intervals (Fig. 4).

Discussion
This meta-analysis on randomized controlled trials demonstrated no significant difference between mobile-bearing and fixed-bearing TKA with regard to all outcome measures compared. The revision rates among studies throughout all follow-up intervals were 2.4% in mobile-bearing TKA and 2.2% in fixed-bearing TKA. Furthermore, this metaanalysis did not result in statistically significant differences in revision rates or aseptic loosening between both designs at short-term, mid-term and long-term follow-up intervals. The long-term follow-up interval ranged from 10 to 17 years postoperatively in 12 studies for revision rates and 11 studies for aseptic loosening. Likewise, previous meta-analyses and the vast majority of included randomized trials found similar survivorship when comparing mobile-bearing and fixedbearing TKA [55,81]. In contrast, few non-randomized studies have found contradicting evidence. A registry-based prospective study by Namba et al. [57] on 47,339 knees found that mobile-bearing TKA had a twofold increase in Fig. 4 Random-effect meta-analytic comparison for maximum knee flexion between mobile-bearing versus fixed-bearing total knee arthroplasty. CI: confidence interval Fig. 5 Random-effects meta-analytic comparison for radiolucent lines and osteolysis between mobile-bearing versus fixed-bearing total knee arthroplasty. CI: confidence interval aseptic revision at 6.7 years when compared to fixed-bearing TKA following a multi-variate adjusted regression analysis (P < 0.001). Likewise, Heesterbeek et al. [28] found in a recent multicenter retrospective study that fixed-bearing had superior survivorship at 12 years as opposed to mobilebearing designs. In a randomized trial by Fransen et al. [21], mobile-bearing TKA was found to have a 6-times higher risk for all-cause revision compared to fixed-bearing TKA at 5-year follow-up. This study had major limitations such as a 38% drop-out rate and lack of blinding of those who assessed outcomes.
Assessment of knee functional outcomes demonstrated no clinically significant differences between mobile-bearing and fixed-bearing TKAs. The OKS was only pooled at the short-and the mid-term follow-up intervals without any statistical significance. The KSS knee sub-score was not statistically significant at the short-and the mid-term follow-up intervals; however, at the long-term there was a statistically significant effect in favor of fixed-bearing TKA. It is paramount to acknowledge that this finding was not clinically significant as the minimal clinically important difference (MCID) of the KSS knee sub-score is between 5.3 and 5.9 points [48]. The KSS functional sub-score was statistically insignificant at short-, mid-and long-term follow-ups. The HSS knee score was in favor of mobile-bearing TKA at the short-term follow-up which was statistically significant, however, yet clinically irrelevant as the HSS MCID is 8.29 points [32]. The mid-and the long-term follow-up for the HSS knee score had no statistically significant difference between mobile-bearing and fixed-bearing TKA. Furthermore, there was no statistically significant difference between mobile-bearing and fixed-bearing TKA for the postoperative maximum knee flexion. Most prior meta-analyses and randomized trials have shown similar results without any statistical difference in clinical outcomes. Nonetheless, several studies have had better outcomes with mobilebearing TKA. At 6-10-year follow-up, the randomized trial Baktir et al. [7] resulted in significantly improved pain and KSS knee sub-scores in mobile-bearing TKA. However, the authors found no difference in the functional sub-score of the KSS. In a recent randomized trial by Powell et al. [64], mobile-bearing TKA had superior results with the OKS and the Knee Injury and Osteoarthritis Outcome Score sports and quality of life subscales. This difference was observed at 10-year follow-up which exceeded the MCID threshold. In contrast, a similarly well-designed trial by Abdel et al. [1] refuted such findings without any advantages provided by the mobile-bearing design over fixed-bearing TKA in terms of maximum knee flexion or function at 10-year follow-up.
In terms of radiological outcomes, no significant differences were detected between both mobile-bearing and fixedbearing TKA at the short-, mid-and long-term follow-up intervals for either radiolucent lines or osteolysis. In all randomized trials included except for the study by Bailey et al. [6], there was no statistical difference between mobilebearing and fixed-bearing designs in radiological outcomes. Bailey et al. [6] have reported that radiolucency was higher in the mobile-bearing designs around the tibial component; however, this was clinically insignificant. Furthermore, in a radiostereometric analysis (RSA) by Schotanus et al. [71] both mobile-bearing and fixed-bearing designs had similar implant migration detected by the maximum total point motion at 2 years.
The strengths of this study were the inclusion of the largest number of randomized trials thus far, and the analyzing outcomes measure at the short-, mid-and long-term followup intervals. To the best of our knowledge, this is the most comprehensive recent meta-analysis on the topic. The last systematic review was performed in 2017 by Fransen et al. [22]. In addition, the last two meta-analyses were performed in June 2020 on this topic by Chen et al. [15] and Wang et al. [84]; however, both meta-analyses combined had 16 randomized trials versus 70 randomized trials in our meta-analysis. Furthermore, both meta-analyses had conflicting results as one supported long-term outcomes of mobile-bearing TKA, yet the other found no difference between fixed-bearing and mobile-bearing designs. In contrast, our study found no differences between mobile-and fixed-bearing designs at anytime point; this is mainly due to pooling data from 70 RCTs, thereby demonstrating more valid results. Several limitations to this meta-analysis should be acknowledged. Although we included RCTs, several trials had high risk of bias as evident in our qualitative review. Another limitation was that outcome measures varied among included studies, which prevented measuring the long-term outcome using the OKS and pooling a higher number of patients in other outcome measures. Implant migration using RSA was not analyzed due to the variability in its reporting across RSAbased studies. Another important limitation was that different types of mobile-bearing TKA were used by different trials, in turn this could be a potential source of bias given the mobile-bearing type was not adjusted for.

Conclusion
This meta-analysis on 70 randomized controlled trials demonstrated no clinically significant differences between mobile-bearing and fixed-bearing TKA at short-, mid-and long-term follow-up for revision rates, aseptic loosening rates, knee functional scores, maximum knee flexion and radiographic lucent lines and osteolysis. The current level of evidence demonstrated that both mobile-bearing and fixedbearing designs achieved excellent outcomes, yet it does not prove the theoretical advantages of the mobile-bearing insert over its fixed-bearing counterpart. Given that the use of either design can be supported by this meta-analysis, we recommend that surgeons can use mobile-or fixed-bearing inserts in TKA at their own discretion.
Authors' contribution AH, OZ, AT performed the literature search and collected the data. MS, AE performed the quality assessment. AFA conceived the study and performed the statistical analysis. SH, GO, AA and MAA helped in preparing the manuscript and supervising the whole work.
Funding Open access funding provided by the Qatar National Library. None.

Availability of data and material Not applicable.
Code availability Not applicable.

Conflicts of interest
On behalf of all authors, the corresponding author states that there was no conflict of interest.

Consent for publication Not applicable.
Ethical approval This article does not contain any studies with human participants performed by any of the authors.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.