Improved survival rate and functional recovery of TKA in cementless fixation compared with cemented component: a systematic review and meta-analysis

Background: Whether the cement could be given up in total knee arthroplasty (TKA) was still in controversy. We perform this meta-analysis to compare the rate of revision and functional recovery between two kinds of fixation in TKA. Methods: Randomized controlled trials (RCTs), prospective/retrospective observational studies from PubMed (on 2019 Sep), EMBASE (on 2019 Sep), and the Cochrane Central Register of Controlled Trials (CENTRAL) and Web of Science (on 2019 Sep) were searched. Continuous outcomes were presented as mean difference or standard mean difference with 95% CI and discontinuous outcomes were reported as relative risk (RR) with 95% CI. Random-effects or fixed-effects model was conducted to analyze the extracted data. The PRISMA guidelines and Cochrane Handbook were adopted to assess the quality of the results reported in included studies to ensure that the results of our meta-analysis were reliable and veritable. The continuous and dichotomous outcome were collected in a standard form, and the data were analysed by using Review Manager 5.3 software. Finally, the results were presented in the Forest plots. The rate of revision and reasons caused revision was the primary outcome of our study. Results: Twenty-six studies involving 2369 patients in cementless TKA and 2654 patients in cemented TKA were finally included in our meta-analysis. 26 studies were divided into three subgroups according to the length of follow-up. The studies in the first subgroup followed less than 5.5 years, the second followed less than 10.5 years, and the third followed more than 10.5 years. Cementless fixation significantly decreased the rate of aseptic loosing ( p=0.0002 ) and revision ( p=0.0002 ) in the first subgroup. Other reasons such as periprosthetic joint infection (PJI), instability and polyethylene wear were not significantly different between two groups. Significantly better functional recovery got in cementless TKA in terms of Knee Society Function Score ( p=0.01 ) when followed longer 8.5 years. Significantly less patients in cementless group require manipulation p=0.02 rate

decreased in cementless TKA within 5.5 years. In addition, the cementless TKA seemed performed better in postoperative functional recovery when had a long-term follow-up. However, the rate of complication was not significantly different in two kinds of fixation in TKA.

Background
As the gold standard of fixation method in total knee arthroplasty (TKA), cemented fixation occupied 93.5% implants of TKA in 2010 [1]. There were a series of advantages in conventional cemented fixation in TKA. Firstly, cemented fixation allows for small bone cut defects, which was less technically challenging compared with uncemented fixation [2]. Secondly, cemented fixation could deliver antibiotics into the joint to prevent infection [3]. Thirdly, as an effective barrier, cement could insulate the polyethylene debris from the articular surface and prevent the osteolysis [4]. Therefore, most likely because the greater clinical experience with cemented fixation and better clinical results over cementless fixation, cemented fixation was still most used in TKA. An analysis using New Zealand Joint Registry (NZJR) data revealed that most (91.5%) were fully cemented with 4.8% hybrid and 3.7% uncemented in 96,519 primary TKAs from 1998 to 2017 [5].
However, accompanied with the increasing demanding of TKA, the mean age of patients underwent TKA was decreasing [6]. It was predicted that more than half of patients underwent TKA was contributed by patients younger than 65 years-old by 2030 [6]. This posed a difficulty to the development of TKA, for the more active lifestyle was needed by younger patients. Therefore, the concern of bone resorption in the bone-cement interface would make the dominance of cemented fixation challenging [7]. Although the preliminary results of cementless fixation was proved discouraging, cementless TKA in young patients was found to have comparable midterm results to cemented TKA [8]. With a biologic bone-implant interface, cementless fixation determined to prevent the osteolysis and reduce the risk of aseptic loosening, especially in young patients with enough bone stock. Therefore, with development of prosthesis design and material renovation, cementless TKA has been introduced to extend the life of prosthesis [9].
In recent years, many studies and several meta-analysis have compared cemented with cementless fixation in terms of implant survival, functional recovery and radiological differences. However, there was not still a widely-accepted conclusion formed whether the cement should be used in TKA. To our knowledge, the length of follow-up was an important factor influenced the revision rate of TKA. While there was still no review and meta-analysis divided the studies according to length of follow-up to compare two kinds of fixation. Therefore, we divided the studies included in our meta-analysis into three groups to explore the revision rate and reasons of revision in two groups. We hypothesized that the cementless fixation was not inferior than cemented TKA in terms of revision rate and functional recovery.

Methods
The guidelines listed in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) was the basis of this systematic review and meta-analysis [10]. "Knee, knee replacement, knee arthroplasty, total knee replacement, TKR, total knee arthroplasty, TKA," and "cementless, cemented, uncemented" were used as key words in connection with AND or OR. Meta-analyses were identified and screened out of the search results by the third reviewer. Then, the references of these meta-analyses were screened to find additional relevant studies. Another reviewer tried to contact expert informants by email to search for unpublished studies Finally, two reviewers independently assessed the studies, and any discrepancies were resolved by a discussion with the other reviewers.

Inclusion and Exclusion criteria
Studies were included according to the PICOS criteria: (1) Population: patients experiencing TKA who were demographically alike; (2) Intervention and Control: cementless and cemented fixation in TKA; (3) Outcomes: revision rate, reasons of revision, functional recovery, and rate of complication; (4) Study design: randomized controlled trial (RCT), prospective or retrospective observational studies.
Studies were excluded if: (1) relevant outcomes were missing or (2) the quality assessment was low (RCT<5, non-RCTs<20) [11,12]. (3) the groups in study was not fully-cementless and fully cemented, that the hybrid fixation was not included in this study.

Quality assessment
A modified seven-point JADAD scale was adopted to assess the methodological quality of the RCTs [11]. The scale considers five items, namely, randomization, concealment of allocation, double blinding, withdrawals and dropouts [11]. Based on the Cochrane Handbook, two reviewers independently evaluated the quality of the included RCTs. There was no disagreement between the two reviewers' ratings.
Two reviewers evaluated the quality of non-RCTs by using Methodological Index for Non-Randomized Studies scale (MINORS), which has a range of scores from 0 to 24 [12]. Unified consensus was obtained if there were any different opinions.

Data extraction
The relevant data, including the authors, year of publication, country, baseline information of participants, prosthesis design, revision rate, power analysis, the length of follow-up were independently extracted by two reviewers using a standard data extraction form.
To compare the two kinds of fixation in TKA, the outcomes were summarized in three parts. The first part was the rate of revision and reasons of revision, which was the primary outcome of our study.

Statistical analysis
The Review manager 5.3 (Nordic Cochrane Center, Copenhagen, Denmark) was used to perform the meta-analysis. The final results are shown in Forest plots. Mean differences (MD) or standard mean differences (SMD) were used to weigh the effect size for continuous outcomes, and relative risks (RR) were used for dichotomous outcomes. The I 2 statistic was used to test for heterogeneity across the included studies [11]. A p value≤0.1 or an I 2 >50% was regarded as proof of heterogeneity. A randomeffects model is used to synthesize results with high heterogeneity and is more conservative than a fixed effects model. Therefore, a random-effects model was used to alleviate the effect caused by high heterogeneity, and a fixed effects model was used when statistical evidence showed low heterogeneity.

Search results
As shown in Fig.1, among 1787 articles were obtained from the databases via the search strategy.
After removing duplicates, 767 articles were screened. From among them 722 articles were removed after reading the title and abstract based on the inclusion criteria. Then, 19 studies were excluded on the basis of exclusion criteria (the reasons were offered in the supplementary material).

Baseline information and quality assessment
The JADAD score of 11 RCTs were listed in Table 2, both of them were ≧5, four of them [2,15,21,22] got 7 points. The MINORS scores of 15 non-RCTs were listed in Table 3, both of them were ≧20, only one of them [37] got 24 points.

Rate of revision and reasons of revision
A total of 22 studies involving 2178 patients in cementless group and 2442 patients in cemented group reported the rate of revision during the follow-up. There were 101(4.6%) and 131 (5.3%) patients in two groups underwent the revision surgery for all kinds of reasons. Among of them, aseptic loosing was the most common reason of revision, next was the periprosthetic joint infection (PJI). The specific number of revisions and all kinds of reasons were listed in Table 4 in detail. In addition, we made a heat-map based on the risk ratio of revision caused by all kinds of reasons in every study (Fig.2). We divided 22 studies into three subgroups according to the length of follow-up and reported results as following:  Fig.4).

Only 1 study involving 100 patients in cementlesss group and 100 patients in cemented group
reported the rate of revision caused by periprosthetic joint infection (PJI). There were 0 (0%) and 1 (1%) patients in two groups revised for PJI.

Functional recovery
12 studies involving 652 patients in cementlesss group and 656 patients in cemented group recorded the Knee Society knee score. They were divided into two subgroups according to if the duration of follow-up was >5.5 years. Pooled results revealed there was no significant difference regarding Knee Society knee score between two groups in all subgroups (Fig.6A). 9 studies involving 827 patients in cementlesss group and 819 patients in cemented group recorded the Knee Society function score. There was no significant difference in short duration subgroup.
5 studies involving 230 patients in cementlesss group and 221 patients in cemented group recorded the Oxford knee score, all of them followed shorter than 5.5 years. Pooled results revealed that fixation method didn't make a difference on Oxford knee score in short duration (Fig.6C). (1.4%) in cemented group diagnosed as superficial wound infection or PJI. In all subgroups at three different time periods, the results were not significantly different (Fig.8B).

Discussion
Compared with the published review and meta-analysis [38][39][40][41][42], the most prominent advantage of our study was that we divided studies into three subgroups according to the length of follow-up and reduced the bias for revision rate caused by the length of follow-up.
Survivorship of prosthesis was the most important endpoint in TKA [43]. The rate of revision and reasons caused revision were primary outcomes of our study. Nearly each included study reported the number of revision and reasons caused revision. We found that the rate of revision was not significantly different when all studies were combined, but when data were combined in studies followed less than 5.5 years the cementless fixation TKA presented significant superiority. Although relatively longer follow-up was needed to compare the true difference regarding the rate of revision between two kinds of fixation in TKA. It has been revealed that 3% to 50% primary TKAs underwent revision within the first five years [44,45]. Therefore, we believe that even we could not get significant difference in all three subgroups, the results from this period of follow-up was very meaningful for the future research. In addition, a potential explanation of non-significant results in subgroups followed longer than 5.5 years was that less articles followed longer than 5.5 years in our systematic research.
However, other 8 reasons including PJI was not significantly different between two kinds of fixation ( Fig.5 and supplementary material). Therefore, we could conclude that aseptic loosing was significantly easier happened in bone-cement interface in cemented fixation group and lead to revision within 5.5 years after TKA.
As the secondary outcome of our study, the postoperative functional recovery also showed significant difference between two kinds of fixation in TKA. In this study, we used Knee Society/Function Score, Oxford Knee Score, Western Ontario and McMaster Universities Osteoarthritis Index scores (WOMAC), range of motion (ROM) and manipulation under anesthesia to evaluate the difference on clinical recovery. On the one hand, patients in cementless group had better functional recovery in terms of Knee Function Score compared with cemented group when followed longer than 8.5 years. On the other hand, although ROM was not significantly different, there were significantly less patients in cementless group required manipulation under anesthesia. A possible explanation for the better recovery in cementless TKA was that relevant complications such as osteolysis, anterior knee pain was more common in cemented group. However, rate of complications including DVT and infection was not significantly different between two kinds of fixation in our study.
From the results of previous research, cementless TKA showed inferior clinical outcomes and survivorship compared with cemented component [14,40]. However, along with the development of manufacture and biomaterials including highly porous metals, cross-linked polyethylene, and corrections in initial cementless designs, some recent publications show successful results in longterm follow-up of cementless fixation [46]. Interest on cementless fixation increased as more young patients underwent TKA. Moreover, cementless TKA presented lower revision rates compared with cemented fixation in morbidly obese patients [25]. A possible reason for this phenomenon was that greater stress was placed on the bone-implant interface when patients were more active or obese [47]. Therefore, inferior performance of cemented TKA in younger and obese patients made the advent of cementless an alternative way to offer long-term results.
In addition, a study published in 2019 compared the actual cost of a cemented and cementless total knee arthroplasty procedure concluded that the overall procedural cost of implanting a cementless TKA was less than implanting a cemented TKA [48]. Therefore, cost alone should not be a barrier to using cementless TKA.
There are some weaknesses in this review and meta-analysis. Firstly, there were small number of studies followed >5.5 years included in our study and we did not divide the RCT and non-RCT in our study. Secondly, the prosthesis design used in included studies were not totally same, which might increase bias, while the heterogeneity was not high in our outcomes. Thirdly, in the 26 studies included in our meta-analysis, there were only 11 RCTs included, which decreased the level of evidence of our study. Therefore, more high-level RCTs were needed in futrure research. Fourthly, outcomes regarding complication after TKA was relatively simple, more relevant outcomes such as instability, stiffness, dislocation and osteolysis should be paid attention in future research.

Conclusion
Not only the rate of aseptic loosing was decreased, rate of revision was also significantly decreased in cementless TKA within 5.5 years. In addition, the cementless TKA seemed performed better in postoperative functional recovery when had a long-term follow-up. However, the rate of complication was not significantly different in two kinds of fixation in TKA.     Figure 1 The flowchart of the study selection Heat-map regarding the risk ratio of rate of the revision and reasons in every study Figure 3 The frost blot about the rate of revision The frost blot about the rate of aseptic loosing Figure 5 The frost blot about the rate of periprosthetic joint infection Figure 6 The frost blot about the functional recovery The frost blot about the knee motion