Disease progression, aseptic loosening and bearing dislocations are the main revision indications after lateral unicompartmental knee arthroplasty: a systematic review Journal of ISAKOS

Importance: Lateral unicompartmental knee arthroplasty (UKA) is a surgical option for patients with isolated lateral osteoarthritis however, the procedure has higher revision rates than medial UKA. The reason for this re-mains unclear; therefore, a better understanding of the indications for lateral UKA revision is needed. Aim: The primary aim of this systematic review was to identify revision indications for lateral UKA. Secondary aims were to further investigate if revision indications were in ﬂ uenced by implant design and time from surgery. Evidence review: A systematic literature review was performed according to the PRISMA 2020 guidelines. Search was performed in January 2022 in MedLine, EMBASE, CINAHL and the Cochrane Library using the keywords “ knee arthroplasty ” , “ unicompartmental ” , “ reoperation ” , synonyms and abbreviations. Articles published in 2000 – 2021 that were at least level III retrospective cohort studies with at least 10 lateral UKAs and reported all failure modes were included. Risk of bias was assessed using the ROBINS-I tool. Revision indications, patient characteristics, study design, implant types and time to failure were extracted from the selected studies. Collated data were tabulated and differences were tested using Chi-square or Fisher ’ s exact test. total of 29 cohort and 4 registry 7,668 UKAs Studies judged having moderate or severe risk of bias; this with the retrospective nature of studies required to investigate long-term outcomes of knee arthroplasty. The main indications for lateral UKA revision were OA progression (35%), aseptic loosening (17%) and dislocation (14%). The incidence of revision similar for mobile-bearing implants and ﬁ xed-bearing (6.4%). tibial UKA dislocations late with UKA, (41% cf. p malalignment (2.7% cf. 0.8%, p ¼ 0.02), cf. p ¼ and bearing dislocations (20% cf. 10%, p < 0.001) more for lateral UKA. Conclusions OA progression, aseptic loosening and bearing dislocation were the three main revision indications for lateral UKA. Compared to medial UKA, OA progression, malalignment, instability and bearing dislocations were more common revision indications for lateral UKA. Higher survivorship of metal-backed ﬁ xed- bearing implants was found. The ﬁ ndings suggest that the outcomes of lateral UKA may be improved with more optimal alignment, gap balancing and patient selection.


What is already known
Lateral unicompartmental knee arthroplasty (UKA) has lower survivorship than medial UKA and potentially different mechanisms of failure due to anatomical differences between the compartments. The last review on this topic in 2016 reported that osteoarthritis (OA) progression (29%), aseptic loosening (23%) and bearing dislocation (10%) were the main revision indications; however, an update is due with the introduction of implant design changes and an increased number of publications on lateral UKA within the last five years.

What are the new findings
The main indications for lateral UKA revision remain OA progression (35%), aseptic loosening (17%) and bearing dislocation (13%). Mobile-bearing lateral implants had a similar rate of failure (7.6%) compared to fixed-bearing implants (6.4%). For fixedbearing implants, all-polyethylene components had a higher rate of failure (13.9%) compared with metal-backed (1.8%). Early revisions were associated with bearing dislocation, whereas late revisions were associated with OA progression. Compared with medial UKA, OA progression, malalignment, instability and bearing dislocations were more common for lateral UKA, which suggest that achieving optimal alignment and gap balancing may be of more importance with lateral UKA.

Introduction
Knee arthroplasty is an effective treatment for patients with end-stage osteoarthritis (OA) [1,2]. An estimated 50% of patients undergoing knee arthroplasty have disease isolated to one compartment and may be eligible for unicompartmental knee arthroplasty (UKA) instead of a total knee arthroplasty [3,4]. UKA of the medial compartment is most commonly performed, whereas only 2% are performed for the lateral compartment [5][6][7]. Lateral UKA has good outcomes for patients with isolated lateral disease [8]; however, the procedure has higher revision rates compared with medial UKA [9]. A better understanding of why lateral UKA is more frequently revised is necessary to guide orthopaedic services towards improved clinical outcomes for these patients.
The main indications for medial UKA revision are OA progression, aseptic loosening and bearing dislocation [10]. Less is known about the outcomes following lateral UKA. This is important given that the considerations are different; anatomic and biomechanical differences between the medial and lateral compartments, differences in loading, size and ligament laxity [11,12], may lead to different modes of failure. The most recent review of revision indications for lateral UKA was undertaken in 2016 [13], which found that the main revision indications for lateral UKA were also OA progression, aseptic loosening and bearing dislocation. However, for lateral UKA, the main indication for revision was OA progression, in contrast with medial UKA, for which aseptic loosening was the main indication [10]. The authors suggested that alignment and loading differences between the compartments led to the different reasons for revision, demonstrating the importance of separating the outcomes of medial and lateral UKA.
For lateral UKA, there have been several recent improvements in implant design, such as the introduction of cement-less components and the redesign of polyethylene (PE) bearings [14][15][16][17][18]. These were intended to address issues associated with older generation implants, in particular bearing dislocations and aseptic loosening [19,20]. While there is a paucity of studies on lateral UKA, outcomes using the newer implant designs have recently been published. This systematic review was therefore undertaken to provide an update on the revision indications for lateral UKA. Different implant designs and time from surgery have been demonstrated to affect outcomes and revision indications for medial UKA [10]; however, this has not previously been reported for lateral UKA. Therefore, we also aimed to investigate if revision indications for lateral UKA were influenced by the type of implant used, and by time from surgery, as well as to make comparisons to medial UKA. We hypothesised that the main revision indications would be aseptic loosening and OA progression, similar to medial UKA, however, that the rates of revision would be different between the procedures due to anatomical differences between the medial and lateral compartments.

Methods
A systematic review of the literature was performed according to PRISMA 2020 protocol [21]. Each step was performed by one author (MLT), with any uncertainties discussed with the senior author (SWY). A consensus was reached before any data were included.
After the removal of duplicates, articles were evaluated for relevance based on title and abstract. Conference proceedings, newsletters and reviews were excluded. Registry reports were also checked and included if failure modes were reported separately for medial and lateral UKA. The remaining articles were then assessed using the pre-defined inclusion and exclusion criteria presented below.

Inclusion/exclusion criteria
The inclusion criteria were (1) English language articles in humans published between January 2000 and December 2021 (inclusive); (2) minimum level III retrospective cohort studies using the Oxford Centre for Evidence-Based Medicine 2011 Levels of Evidence (OCEBM [22]; (3) primary lateral UKA; (4) included more than ten lateral UKA; (5) reported all failure modes; (6) a majority of the patients had a primary diagnosis of OA. The exclusion criteria were (1) case reports; (2) studies on patients with previous surgery in the same knee, such as high tibial osteotomy or medial UKA; (3) failures not reported separately as medial or lateral; (4) multiple reports on the same study or with overlapping patient cohorts. For the latter, only the most recently published study was included.
Failure was defined as having a revision event, classified as the addition, exchange or removal of one or more prosthetic components or a periprosthetic joint infection. The studies were grouped according to the outcomes of interest, which were the overall indications for lateral UKA revision, mobile-bearing vs. fixed-bearing components, all-PE vs. metalbacked fixed-bearing components and time from primary surgery.

Study quality assessment
Studies were assessed for the risk of bias using the ROBINS-I ("Risk of Bias in Non-randomised Studies -of Interventions") tool [23]. Studies were evaluated for the risk of confounding, participant selection, classification of interventions, deviations from intended interventions, missing data, outcome measurements and selective reporting. The items were scored as "low risk of bias", "moderate risk of bias", "serious risk of bias", "critical risk of bias" or "insufficient information". The overall risk of bias judgement for each study was based on the worst judgement assigned within any one domain for that study.

Data extraction
The data were extracted from the selected studies into tables using Microsoft Excel. The following data were extracted from the studies: author, year, number of UKAs, number of patients, follow-up times (means/medians and range), number of revisions, implant types including bearing type and fixation, time to revision for each case (if available) and counts of each revision indication). The categories of revision indications used were aseptic loosening, OA progression, pain, instability, periprosthetic joint infection, wear, bearing dislocation, malalignment, fracture, tibial subsidence, other and unknown [24]. Additionally, we also extracted means and ranges of patient age, body mass index and gender splits if they were reported.
All studies reported the implants that were used; however, some studies did not report bearing (mobile-bearing vs. fixed-bearing) or tibial component (all-PE vs. metal-backed) type. Bearing type could be inferred from the implants used, however, the tibial component could not; therefore, studies that did not report the latter were excluded from that subgroup analysis. Similarly, some studies did not report time to failure for each event and these were excluded from the "time from primary surgery" subgroup analysis. Studies that did not report patient characteristics or mean time of follow-up were included in the analyses as these were not the main outcomes of this review. As studies were only included in each of the three subgroup analyses if they reported the outcome of interest, no additional data preparation were needed for the following analyses. To avoid the impacts of clustering as a result of repeated reporting of outcomes on the same group of patients, one of the exclusion criteria used was the removal of duplicate reports from the same patient cohort. As the subgroup analyses did not include repeated measurements or missing data, we did not perform additional sensitivity analyses for the data.
Ethical approval was not sought for this review as the study used previously published and publicly available data with no identifiable information.

Statistical analysis
Categorical variables were tabulated as absolute frequencies and percentages. Statistical analyses were performed with IBM SPSS Statistics version 28 (IBM corp., Armonk, NY, USA). Chi-squared tests with Haldane-Anscombe correction or Fisher's exact test and relative risk were used for between-group comparisons. In all cases, the null hypothesis was that groups had equal proportions of revision indications. The tests were two sided and p-values below 0.05 were considered significant.

Study selection and quality
An initial 3,816 records were retrieved from a literature search performed according to PRISMA 2020 guidelines (Fig. 1). Following the removal of duplicates and screening, a total of 117 full-text records were obtained and assessed for eligibility. A total of 29 cohort [8,14,18,, 3 registry studies [51][52][53] and one registry [54], which included 7,668 UKAs, were selected for analysis (Table 1). Whilst a further seven studies met the inclusion and exclusion criteria, these were found to be overlapping reports of the same study cohort and were excluded. Strength of the evidence was moderate as the 27 retrospective and 6 prospective observational cohorts were judged to present either moderate or serious risk of bias (Table 2). Further information on patient characteristics and specific implants used in these studies are available in Appendix B.
Revision indications for lateral UKA Some differences were found between the revision indications reported by cohort and registry studies, therefore these were reported separately (Table 1). From the cohort studies, 222 revisions (7.9% or 1.35 per 100 observed component years (ocy)) were reported from a total of 2,802 lateral UKAs, with a mean follow-up of 5.1 years. The most common revision indication was OA progression to the other compartments (41%), followed by bearing dislocation (20%) and aseptic loosening (10%). From the registry studies, 371 revisions (7.6%, 1.96 per 100 ocy) were reported from a total of 4,866 lateral UKAs, with a mean follow-up of 5 years. Similar to cohort studies, the most common revision indication reported by registry studies was OA progression (30%). However, in contrast with the cohort studies, the second most common indication was aseptic loosening (20%), followed by bearing dislocation (8%). Compared with registry studies, cohort studies reported larger proportions of OA progression (41% cf. 30%, p ¼ 0.004) and bearing dislocation (20% cf. 8%, p < 0.001), but lower proportions of aseptic loosening (10% cf. 20%, p ¼ 0.001), pain (6% cf. 11%, p ¼ 0.03) and "other" reasons (5% cf. 11%, p ¼ 0.02).

Time to failure
A total of 14 studies reported individual time to failure for each revision event, and these were collated to report on outcomes by the time of surgery [8,14,[44][45][46]50,18,26,27,29,33,36,37,41]. Nine studies had a moderate risk of bias [8,18,26,33,37,[44][45][46]50] and five studies had a serious risk of bias [14,27,29,36,41]. From these studies, 76 revisions were reported out of 1,413 lateral UKAs (5.4% revised, 1.07 per 100 ocy, mean follow-up 4.6 years). The proportion of revisions due to OA progression was 0% in the first six months of surgery but increased sequentially over time to account for 100% of all revisions performed after 10 years (p < 0.001, Fig. 2). In contrast, the proportion of revisions due to bearing dislocations decreased sequentially over time, from 69% of all revisions in the first six months of surgery to 0% after 10 years (p < 0.001). There were no identifiable trends over time for the other revision indications.

Discussion
This systematic review has presented an updated and more detailed synthesis of revision indications following lateral UKA. The main indications were OA progression (41% of all failures), aseptic loosening (20%) and bearing dislocation (10%). When bearing type was compared, revision for mobile-bearing UKA was more likely to be associated with OA progression and bearing dislocations. When comparing time to revision, bearing dislocations were the most common reason for early revision, whereas OA progression was most common in late revisions.
There were some differences between what was reported in cohort vs. registry studies. Cohort studies reported larger proportions of OA progression and bearing dislocations, but lower proportions of aseptic loosening, "pain" and "other" reasons compared with registry studies. This may be explained by the difference in the level of detail that can be captured by cohort studies and registries: registries tend to report a higher level of "unexplained pain" and "unknowns" and may have lower capture of certain revision types, such as bearing dislocations [55,56].
The subgroup analyses revealed that both implant survival and revision indications varied according to implant design. The incidence of revision was higher with the use of mobile-bearing and all-PE components than fixed-bearing and metal-backed, respectively. This was associated with an additional risk of bearing dislocations that were not found with fixed-bearing implants. These mobile-bearing implants were also associated with lower proportions of OA progression. This can be explained by implant design; the mobile-bearing design allows for more balanced surface and subsurface contact stresses [57]. There has been relatively a high usage of mobile-bearing implants for lateral UKA [54] ( Table 3). The mobile-bearing implants included in this study were AMC-UKA (Corin, Cirencester, UK), Oxford Domed (Zimmer Biomet, Warsaw, IN) and Preservation (De Puy, Johnson&Johnson, Raynham, MA), with the Oxford Domed being the most common (92%, from studies that reported numbers of each implant type). Given the increased laxity of the lateral compartment and resulting risk of bearing dislocations [18], mobile-bearing implants for lateral UKA may become less common.  For medial UKA, previous studies comparing the use of all-PE and metalbacked tibial components reported a higher incidence of aseptic loosening and tibial subsidence for all-PE implants. This has been suggested to be due to the increased proximal tibial strain with the all-PE component, which can lead to deformation of the tibial component and changes to underlying bone architecture [58,59]. While a higher rate of revision was found with all-PE implants, we did not find evidence for the increased incidence of aseptic loosening or tibial subsidence. However, it should be noted that numbers of collated revisions were low so we could not conclusively identify the reasons for the higher revision incidence found with all-PE implants for lateral UKA.
A comparison of revision indications over time from surgery has not previously been reported. In this review, we found that revisions due to OA progression increased over time, with a highest incidence at 10þ years following surgery. In contrast, revisions due to bearing dislocations decreased over time, with the highest incidence within six months of surgery. Bearing dislocations are considered to be associated with suboptimal surgical technique [4,60], and the events are reported to take place within the first two years of surgery [60,61]. In comparison, OA progression may sometimes be reported as unexplained pain in the early-term because it may be undetectable in routine radiographs [24]. OA progression following UKA has also been associated with suboptimal patient selection [13,62]. Overall, the study findings suggest that there may be potential for improving the outcomes of lateral UKA with improved component positioning and more careful patient selection. .
When the findings from this study were compared with our recent systematic review on revision indications for medial UKA [10], we found that lateral UKAs were more likely to be revised for OA progression (41% cf. 30%, p ¼ 0.004), malalignment (2.7% cf. 0.8%, p ¼ 0.02), bearing dislocations (20% cf. 10%, p < 0.001) and instability (4% cf. 1%, p ¼ 0.02; Fig. 3). In contrast, medial UKAs were more likely to be revised for aseptic loosening (24% cf. 10%, p < 0.01). There are anatomical and kinematic differences between the medial and lateral compartments. The lateral plateau is convex and smaller than the concave medial plateau, and lateral ligaments are more lax in flexion [11]. To explain the higher incidence of OA progression with lateral UKA, van der List et al. proposed that this may be associated with the smaller lateral joint space and potential of higher loading on the medial condyle with a sub-optimally balanced UKA [13]. The aetiology of OA progression following UKA is not well-understood; however, it has also been linked to patient selection [13,62]. The relative laxity of the lateral compartment, particularly in deep flexion, may explain the increased risk of a bearing dislocation with lateral UKA. The "domed" Oxford Partial Knee was introduced to address this issue; however, a significant risk of bearing dislocations remained. The study findings suggest that surgeons may need to pay particular attention to appropriate patient selection and achieving optimal balancing during lateral UKA to avoid an increased risk of revision.
There are some limitations to this review. First, there is a paucity of studies reporting the outcomes of lateral UKA in the literature because the procedure is not commonly performed. This review therefore included studies with many different implant types and in some cases, small cohort sizes, which limited the subgroup analyses that could be conducted. However, the inclusion of a variety of implant types may also enhance generalisability of the study findings. Second, we could not compare outcomes based on implant fixation, as there was a lack of reports on cement-less fixation. This will likely be an area of interest for future research. There were also some limitations with the review process and the available body of evidence. For logistical reasons, study screening and data extraction were mostly performed by one author. However, any uncertainties were discussed with a senior author to reach a consensus, which allowed for confidence in the overall study findings. The strength of the evidence was moderate as the included studies were judged to have moderate to serious risk of bias, particularly with confounding, patient selection and reporting bias. This was because the studies included were observational and mostly retrospective due to the long follow-up times required and low incidence of revision arthroplasty. However, with the reporting of relevant patient and implant characteristics (Appendix B), it is hoped that the findings are still widely transferable, for example, when considering potential bias due to differences in baseline patient demographics. Despite these limitations, this review    was able to present an updated synthesis on revision indications reported for lateral UKA and demonstrated that these indications were associated with implant design and time from surgery. The findings from this study can be used to inform surgeons of the specific revision risks associated with use of lateral UKA.

Conclusion
In conclusion, despite recent introduction of improved component designs, OA progression remained the main revision indication for lateral UKA, followed by aseptic loosening and bearing dislocation. Compared to medial UKA, OA progression, malalignment, instability and bearing dislocations were more common indications for lateral UKA revision. Metal-backed fixed-bearing implants had lower reported revision rates compared with implants with all-PE tibial components. The findings suggest that the outcomes of lateral UKA may be improved with more optimal alignment, gap balancing and patient selection.

Registration
This review was not registered.

Review protocol
We did not prepare a formal review protocol for this study.

Financial disclosure
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conflict of interest
Author 3 has a relationship with Zimmer Biomet, including consulting and research funding. Author 4 has a relationship with Stryker New Zealand and Smith þ Nephew, including consulting and research funding. The funding bodies were not involved at any stage with study design, analysis or manuscript preparation. Authors 1 and 2 have no competing interests to declare.

Data availability statement
All data relevant to the study are included in the article or uploaded as online supplemental information.