Background

As the gold standard of fixation method in total knee arthroplasty (TKA), cemented fixation occupied 93.5% in 2010 [1]. There were a series of advantages in conventional cemented fixation in TKA. Firstly, the cemented fixation allows for small bone cut defects, which required less technical challenge compared with the uncemented fixation [2]. Secondly, the 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, because of the abundant clinical experience and great clinical results, cemented fixation was still most used in TKA. An analysis using New Zealand Joint Registry (NZJR) data revealed that most (91.5%) of primary TKA 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 demand of TKA, the patients underwent TKA are becoming younger and younger [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 was determined to prevent the osteolysis and decrease the risk of aseptic loosening, especially in young patients. In addition, with the development of prosthesis design and material renovation, cementless TKA has been introduced to extend the life of prosthesis [9].

Therefore, this study was performed to compare the rate of revision, reasons of revision, functional recovery, and rate of complications in two kinds of fixation in TKA. We hypothesized that the cementless fixation was not inferior to the cemented fixation in terms of rate of revision 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 (The PRISMA checklist was shown in the Supplementary Material) [10].

Search strategy

MEDLINE (1950 to date), PubMed (1966 to date), EMBASE (1974 to date), the Cochrane Central Register of Controlled Trials, the Wanfang database (1982 to date), and the Web of Science were systematically searched for studies on cementless fixation in total knee arthroplasty on 30 August 2019. “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 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: patients followed at least 2 years, and rate of revision, reasons of revision, functional recovery, and rate of complication were reported; (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 were 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 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, and 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. The second part was the postoperative functional recovery, including the Knee Society (KSS) knee and function scores, Oxford knee scores, range of motion (ROM), and manipulation under anesthesia. The third part was the rate of complication, including deep vein thrombosis (DVT) and all infection (including superficial wound infection and periprosthetic joint infection).

Statistical analysis

The Review Manager 5.3 (Nordic Cochrane Center, Copenhagen, Denmark) was used to perform the meta-analysis. The final results were 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 I2 statistic was used to test for heterogeneity across the included studies [11]. A p value ≤ 0.1 or an I2 > 50% was regarded as proof of heterogeneity. A random-effects 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.

Results

Search results

As shown in Fig. 1, a total of 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. Finally, 11 RCTs [2, 13,14,15,16,17,18,19,20,21,22] and 15 non-RCTs [23,24,25,26,27,28,29,30,31,32,33,34,35,36,37] were included in this study.

Fig. 1
figure 1

The flowchart of the study selection

Full size image

Baseline information and quality assessment

Twenty-six studies [2, 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] involving 2369 patients in cementless group and 2654 patients in cemented group were included in this review and meta-analysis. The baseline information including study design, demographical data, revision rate, prosthesis, and length of follow-up were clearly listed in Table 1. Especially, the duration of follow-up was same for cemented and cementless groups. And the length of the follow-up in all included studies ranged from 2 years to 16.6 years.

Table 1 The baseline information of studies compared cementless with cement fixation in TKA
Full size table

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 1 of them [37] got 24 points.

Table 2 The quality assessment of RCTs
Full size table
Table 3 The quality assessment of non-RCTs
Full size table

Rate of revision and reasons of revision

A total of 20 studies involving 1925 patients in cementless group and 2203 patients in cemented group reported the rate of revision during the follow-up. There were 95(4.9%) patients in the cementless group and 89 (4%) patients in the cemented group underwent the revision for all kinds of reasons. Pooled results shown that there was no significant difference between the rate of revision during the at least 2 years of follow-up in two groups (RR = 1.09, 95% CI [0.82, 1.44], p = 0.55; Fig. 2).

Fig. 2
figure 2

The frost blot about the rate of revision

Full size image

Among reasons that caused revision, aseptic loosing was the most common, next was the periprosthetic joint infection (PJI). Fifteen studies with 1697 patients in the cementless group and 1999 patients in the cemented group recorded the rate of aseptic loosing caused the revision (49 (2.9%) and 47(2.4%), respectively). Pooled result presented that there was no significant difference in two groups (RR = 1.03, 95% CI [0.7, 1.52], P = 0.88; Fig. 3). Sixteen studies with 1777 patients in the cementless group and 2032 patients in the cemented group reported the rate of PJI caused the revision (20 (1.1%) and 27(1.3%), respectively). Pooled result presented that there was no significant difference in two groups (RR = 0.81, 95% CI [0.47, 1.4], p = 0.45; Fig. 4). The specific number of revisions and other reasons were listed in Table 4 in detail.

Fig. 3
figure 3

The frost blot about the rate of aseptic loosing

Full size image
Fig. 4
figure 4

The frost blot about the rate of periprosthetic joint infection

Full size image
Table 4 The specific rate of revision caused by all kinds of reasons
Full size table

Functional recovery

Twelve studies involving 827 patients in cementless group and 819 patients in cemented group recorded the Knee Society knee score. Pooled results revealed there was no significant difference regarding Knee Society knee score between two groups (MD = 0.69, 95% CI [− 0.97, 2.35], p = 0.42; Fig. 5a). Nine studies involving 652 patients in cementless group and 656 patients in cemented group recorded the Knee Society function score, which was significantly higher in the former group (MD = 1.70, 95% CI [0.53, 2.86], p = 0.004; Fig. 5b). Four studies involving 176 patients in cementless group and 167 patients in cemented group recorded the Oxford knee score. Pooled results revealed that the kind of fixation did not make a difference on Oxford knee score in short duration (MD = − 0.62, 95% CI [− 1.71, 0.47], p = 0.27; Fig. 5c).

Fig. 5
figure 5

The frost blot about the functional recovery

Full size image

Seven studies involving 626 patients in cementless group and 679 patients in cemented group reported the range of motion (ROM) following TKA. Pooled results revealed the ROM was not significantly different in two groups (MD = 0.9, 95% CI [− 0.72, 2.52], P = 0.28; Fig. 6a). However, 7 studies involving 566 patients in the cementless group and 588 patients in the cemented group revealed that the rate of manipulation under anesthesia was significantly more in the cemented group (RR = 0.44, 95% CI [0.24, 0.80], p = 0.007; Fig. 6b).

Fig. 6
figure 6

The frost blot about the knee motion

Full size image

Complications

Five studies involving 369 patients in cementless group and 390 patients in cemented group recorded the rate of deep vein thrombosis (DVT) following TKA. There were 17 (4.6%) in cementless group and 27 (6.9%) in cemented group diagnosed as DVT, while it was not significantly different (RR = 0.81, 95% CI [0.47, 1.39], p = 0.44; Fig. 7a). Twenty studies involving 2048 patients in the cementless group and 2337 patients in the cemented group recorded the rate of all infection following TKA. There were 29 (1.4%) patients in the cementless group and 32 (1.4%) patients in the cemented group diagnosed as superficial 3rwound infection or PJI. The pooled results shown insignificant difference (RR = 0.97, 95% CI [0.61, 1.57], P = 0.92; Fig. 7b).

Fig. 7
figure 7

The frost blot about the rate of complications

Full size image

Discussion

Compared with the published review and meta-analysis [38,39,40,41,42], the most prominent advantage of our study was that a minimum 2 years length of follow-up criteria was used for screening studies and analyzing these together.

Survivorship of prosthesis was the most important endpoint in TKA [43]. Rate of revision and reasons caused revision were primary outcomes in our study. Although the pooled results shown insignificant difference, the cementless fixation presented significant superiority in studies followed more than 2 years and less than 5.5 years. In addition, although a 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 reported that 3 to 50% primary TKAs underwent revision within the first 5 years [44, 45]. More than 8 kinds of reasons that caused revision after TKA were found, aseptic loosing was the most common. Among all causes of revision, rate of aseptic loosing was 2.1%, followed by the periprosthetic joint infection (PJI, 0.97%), instability (0.24%), exchange of tibia polyethylene insert (0.24%), periprosthetic fracture (0.14%), patella dislocation (0.1%), polyethylene wear (0.05%), stiffness (0.02%), and other reasons (0.56%). Consistent with the rate of revision, rate of aseptic loosing was significantly decreased in the cementless fixation in studies followed more than 2 years and less than 5 years. However, other reasons including PJI were not significantly different between two groups. Therefore, it is induced that aseptic loosing was possibly easier happened in bone-cement interface.

In terms of functional recovery after TKA, patients in cementless group had better Knee Function Score compared with patients in cemented group. 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 including osteolysis, anterior knee pain was more common in the cemented group. In addition, it is worth noting that the length of follow-up in studies reported the Knee Society Function score was all more than 8.5 years, and also in the studies reported manipulation under anesthesia mostly followed more than 5 years. It could be concluded that cementless fixation might present long-term advantages regarding the functional recovery.

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 long-term 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 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 fixation. What is more, a study published in 2019 reported that cementless TKA costed much less than cemented TKA [48].

There are several shortcomings in our study. Firstly, studies followed longer than 10 years was not enough. Secondly, the prosthesis design used in included studies was not totally same, which might increase bias of risk. Thirdly, there were only 11 RCTs in 26 studies included in our study decreased the level of evidence.

Conclusion

Cementless fixation did not decrease the rate of revision after the total knee arthroplasty compared with the cemented fixation, while the long-term functional recovery was significantly better in the cementless group.