RSA prediction of high failure rate for the uncoated Interax TKA confirmed by meta-analysis

Background and purpose In a previous radiostereometric (RSA) trial the uncoated, uncemented, Interax tibial components showed excessive migration within 2 years compared to HA-coated and cemented tibial components. It was predicted that this type of fixation would have a high failure rate. The purpose of this systematic review and meta-analysis was to investigate whether this RSA prediction was correct. Materials and methods We performed a systematic review and meta-analysis to determine the revision rate for aseptic loosening of the uncoated and cemented Interax tibial components. Results 3 studies were included, involving 349 Interax total knee arthroplasties (TKAs) for the comparison of uncoated and cemented fixation. There were 30 revisions: 27 uncoated and 3 cemented components. There was a 3-times higher revision rate for the uncoated Interax components than that for cemented Interax components (OR = 3; 95% CI: 1.4–7.2). Interpretation This meta-analysis confirms the prediction of a previous RSA trial. The uncoated Interax components showed the highest migration and turned out to have the highest revision rate for aseptic loosening. RSA appears to enable efficient detection of an inferior design as early as 2 years postoperatively in a small group of patients.


Introduction
Aseptic loosening remains a major reason for revision surgery in Total Knee Arthroplasty (TKA). 1,2 Since revision rates are generally low it is necessary to follow up hundreds if not thousands of patients for a long period of time (10 years) to be able to detect inferior designs. 3 A method for early detection of aseptic loosening exposing as few patients as possible is therefore of value. Radiostereometric analysis (RSA) enables accurate measurement of migration of prosthetic components relative to the bone 4 , which has been shown to be associated with late aseptic loosening. [5][6][7] Although these findings are promising and the number for RSA studies is increasing, few studies have actually researched whether the RSA predictions are correct. [5][6][7][8] In TKA the question thus remains: Do TKA with increased early migration have higher revision rates for aseptic loosening?
We have already shown in a randomized RSA trial that uncoated Interax tibial components have increased early migration compared to HA-coated and cemented tibial components. 9 We predicted that the uncoated components would have a high failure rate. The aim of the present study was therefore to investigate whether this prediction of the previous RSA trial is correct. We performed a meta-analysis to evaluate the failure rate of these components.

Design of the meta-analysis, and rationale
The design is based on the Cochrane standards and reporting of this meta-analysis is according to the PRISMA guidelines. 10 In order to exclude confounding due to differences in prosthesis design, the meta-analysis is restricted to studies comprising exactly the same implant as the previously published RSA-trial 9 : the cruciate retaining (CR) Interax TKA tibial component, (Howmedica / Stryker, Rutherford New Jersey) with two polyethylene halfbearings. The fixation of the components is either by cement or by bone ingrowth on uncoated or hydroxy-apatite (HA) coated prosthetic surfaces. The cemented components had a diamond surface on the side that was within bone, whereas the uncemented components had a mesh-wire surface (2.25 square millimetres corresponding to circular pore diameter of 1690 micrometers) with or without a HA coating.
The outcome of interest is the number of revisions or recommended revisions for aseptic loosening of the tibial component, for each fixation separately. This outcome will be compared to the early migration results of the RSA-trial 9 which showed increased early migration of the uncoated tibial component compared to the cemented and HA coated tibial components (Figure 3.1).
Uncemented components show high initial migration followed by stabilisation. [11][12][13][14][15][16] Thus, we also present the migration rate of MTPM (mm/year) determined on the migration measured with the post-operative RSA examination as reference (Table 3.1).  9 The plot shows the mean migration -expressed as Maximal Total Point Motion (MTPM) -with 95% CI for each type of fixation of the tibial components: red dashed line for uncoated; green dotted line for HA-coated and blue solid line for cement. The uncoated tibial components showed the most migration. * mm = millimetre

Literature search
The literature search is the foundation on which a systematic review and meta-analysis is built. Inadequate search strategies have been shown to give biased results. 17 We therefore adopted The search involved among others the all fields-and fulltext-options to screen if the following component was mentioned anywhere in a manuscript: "Interax" and relevant abbreviations and extensions. Since "Interax" is a registered brand name of a particular TKA model, it was assumed to be spelled out the same way in the text of a manuscript irrespective of the language used. We did not use any language restrictions

Study selection
All studies were subjected to the following inclusion criteria: 1) The study comprises an original patient cohort treated with the Interax TKA (Howmedica, Rutherford, New Jersey).
2) The cruciate retaining Interax prosthesis with halfbearings is used (Posterior stabilised Interax and Interax ISA versions are excluded).
3 Two reviewers, BGP and MJN, independently subjected all studies to these five inclusion criteria.
In cases where the title and abstract were inconclusive, the full text article was obtained. Any disagreement between the reviewers was resolved by re-examination and subsequent discussion to reach a consensus. Randomized Controlled Trials (RCT) as well as observational studies were considered for inclusion.

Quality Assessment and Data extraction
The quality of each included study was independently appraised by two reviewers, BGP and MJN, using the Jadad Scale. 18 The same reviewers independently extracted relevant data for each included study using a standardized form including demographic data, number of TKA in each fixation group, number of revisions for aseptic loosening in each fixation group, and loss to follow up. Any disagreement between the reviewers was resolved by re-examination and subsequent discussion for consensus.

Statistical analysis
Before considering a meta-analysis (pooling of data), we investigated whether it was appropriate to pool the data. Studies should be similar in design and patient population. In addition, the variability in effect size between studies should not exceed those expected from sampling error: low heterogeneity is desirable. Heterogeneity was assessed by calculating the I 2 -statistic, which is appropriate in case of a small number of studies. 19 Publication bias was assessed with a funnel plot. 20 Meta-analysis was performed with Peto Odds Ratio (OR) fixed effect pooling and Mantel-Haenszel random effects pooling for the risk difference (RD) and number needed to treat (NNT). 21 The NNT was defined as the number of cemented tibial components that would have to be implanted in order to prevent 1 revision as compared to when uncoated components were implanted. We used RevMan software.

Study selection & study characteristics
The search strategy resulted in 268 unique hits of which 4 studies could be included (Figure 3.2). [22][23][24][25] Two papers were published in the English language 23,25 , one in German 24 and one in French 22 (Table 3.2). Three studies compared the cemented component to the uncoated one. 22,24,25 One of these studies 24 was part of a thesis 26 , which we used for more details. One of these studies 25

Risk of bias within studies
The sequence of randomization as well as concealment of allocation was described and appropriate in two studies. 22,25 In one study 23 randomization was performed but the method and concealment not adequately described and in another study 24  In all studies blinding was a potential source of bias. Since evaluation of X-rays is essential for the indication of a revision and the presence or absence of cement cannot be masked on the X-ray, blinding -if possible at all-was not performed in any of the studies.
The number of withdrawals and dropouts was adequately described in all studies. The number of lost to follow-up (8 cemented and 12 uncoated) was high in study by Gicquel et al (Table 3.2). 22 All three studies which compared cemented versus uncoated components included all patients consecutively during study inclusion period and thus reduced the possibility of selection bias. 22,24,25

Uncoated versus Cemented Components
Our aim was to investigate whether the predictions of a previous Radiostereometric Analysis (RSA)trial were correct. Since the uncoated Interax components had shown the highest migration, it was predicted that this type of fixation would have a high failure rate. 9 The results of the metaanalysis show a significant 3 times higher revision rate for the uncoated uncemented component for the early detection of inferior TKA designs in a small series of patients. [5][6][7] It is noteworthy that none of the individual traditional clinical studies with large numbers of patients and medium term or long term follow-up reported a significant difference in revision rates between the uncoated uncemented and cemented Interax tibial component. 22,24 Only when the results of these studies were combined in a meta-analysis setting did the high revision rate in the uncoated components became clearly visible.

Uncoated versus HA-coated
One of the selected studies compared the uncoated tibial component to the HA-coated component. 23

Strengths and limitations
Our search strategy was thorough and complete. This is underscored by the fact that we found two studies that have been published in non-English literature. Although our research question was highly specialized, i.e. fixation of a single type of TKA, we were still able to include three studies. This is not uncommon for orthopaedic meta-analysis even in Cochrane reviews. 29 The included studies were of moderate quality mostly due to issues with blinding for the fixation method, which is a general problem of any study comparing cemented with uncemented components and not specific to the present meta-analysis.
Publication bias generally favours the newly introduced treatment 30 : the uncoated uncemented fixation in this case. Since the studies included in this meta-analysis did not find a positive effect for the uncoated components, publication bias was probably not a major factor here. Thus, we are confident that our conclusion is correct: the uncoated tibial component of the Interax has a higher revision rate for aseptic loosening.
The I-statistic was 0%, so there was no indication for statistical heterogeneity. Despite differences in patient demographics, surgical technique or study design all OR's are on the same side, i.e. showed higher -although not individually significantly -revision rates for the uncoated component and this confirms the predictions of the RSA trial.

Future Perspectives
More than a decade ago Liow and Murray 31 and Muirhead-Allwood 32 called for a more evidencebased evaluation and clinical introduction of (new) prosthetic designs and fixations. Malchau 33 proposed a phased evidence based introduction of new designs. Recently, a renewed call for concrete steps has been made towards such a evidence-based clinical introduction. 34,35 A disastrous design can be detected early post-operatively in a small group of patients by RSA, which therefore has the potential to play an important role in the clinical introduction of new models and fixation methods in total knee arthroplasty. For example, in vitro testing machine studies, should be followed by two year RSA studies in small cohorts in different institutions worldwide, followed by larger comparative studies after which introduction to the market can be started. 33 The latter also involving follow-up in national registries. In this way a more phased prosthesis introduction to the market is guarantied, as is currently the standard for pharmacological agents.