Long-term outcomes of the cementless Oxford Unicompartmental Knee Replacement: a 16-year follow-up study

Abstract

Aims

Cementless fixation is an alternative to cemented unicompartmental knee arthroplasty (UKA), with several advantages over cementation. This study reports on the 16-year survival and ten-year clinical and radiological outcomes of the cementless Oxford Unicompartmental Knee Replacement (OUKR).

Methods

This is a prospective study of the first 693 consecutive cementless medial OUKRs implanted in New Zealand.

Results

The 16-year survival was 89.2%, with 46 knees being revised. The most common reason for revision was progression of arthritis, which occurred in 24 knees. The other reasons for revision included ten bearing dislocations, eight of which were for trauma, one ruptured anterior cruciate ligament (ACL), two tibial plateau fractures, three cases of polyethylene wear, three cases of aseptic loosening, one impingement secondary to overhang of the tibial component, one deep infection, and one revision where the reason was not stated. At the 14 to 16 years survey, the mean Oxford Knee Score (OKS) improved from 23.3 (SD 7.4) to 40.59 (SD 6.8). Radiological analysis at ten years demonstrated no evidence of femoral loosening, subsidence, or radiolucent lines. There were 42 complete radiolucent lines in zone 7 around the tibial baseplate, and ten incomplete radiolucent lines seen in other tibial zones with no progression.

Conclusion

The cementless OUKR is a safe and reproducible procedure with excellent 16-year survival, clinical outcomes, and radiological outcomes in the hands of surgeons who are independent of the design centre.

Cite this article: Bone Jt Open 2026;7(3):326–332.

Introduction

The rate of osteoarthritis (OA) of the knee is increasing, largely due to the ageing population and increasing obesity, and the rates of arthroplasty surgery are projected to follow this trend.¹ The results of knee arthroplasty are excellent, with registries showing up to 92% survival at 23 years.²

Unicompartmental knee arthroplasty (UKA) has become an accepted and effective treatment for isolated medial compartment knee osteoarthritis,^3,4 which has been assessed by some centres as comprising up to 50% of all knee arthroplasty patients.^5-7 UKA has shown benefits over total knee arthroplasty (TKA) such as faster recovery, more normal knee kinematics, increased range of motion (ROM), and better patient-reported outcomes.^8-10 Furthermore, patients undergoing UKA are half as likely to suffer a major complication within the first 30 days of surgery.^11,12 The cemented Oxford Unicompartmental Knee Replacement (OUKR; Zimmer Biomet, USA) has demonstrated satisfactory survivorship rates and functional outcomes over the past 20 years, however the cemented OUKR has a higher revision rate than TKA, with aseptic loosening, component migration, and medial-sided pain contributing to higher revision rates.^2,13-18 The cementless OUKR was developed to address these perceived limitations by providing better biological fixation and a more physiological loading of the knee.^19-23 With positive early results, the use of the cementless OUKR is becoming increasingly popular.^2,19-22 However, currently there is only limited evidence available on the long-term outcomes and implant survivals with cementless OUKR from a centre which is independent of the design surgeons.^23,24

The aim of this study is to report the 16-year survival, clinical outcomes, and radiological outcomes of the first consecutive 693 OUKRs performed in an independent centre.

Methods

Selection criteria

A detailed description of selection criteria and patient information has been previously described.¹ In brief, all patients fulfilled the recommended indications by Goodfellow et al²⁵ for a medial compartment arthroplasty: correctable varus deformity, fixed flexion of < 10°, an asymptomatic lateral compartment, intact anterior cruciate ligament (ACL), and no previous high tibial osteotomy (HTO). Patients with ACL deficiency, or those with a previous/concurrent ACL reconstruction and previous HTO, were excluded. Age, level of activity, BMI, chondrocalcinosis, or presence of patellofemoral OA (except for severe lateral OA with bone loss and grooving) were not considered contraindications.

Patient characteristics

There were 691 OUKRs implanted in 578 patients (113 bilateral procedures). Most were male (315 patients, 54.5%) with the mean age at the time of surgery being 66.29 years (SD 9.39) for males and 65.69 years (SD 9.82) for females. The primary indication was anteromedial OA as defined on plain films. At the time of the 16-year follow-up, 98 patients had died from unrelated causes. As these surgeries were performed at a teaching hospital, the primary surgeon was a consultant in 93% of cases and an advanced training registrar under supervision in 7% of cases.

Study design

This was a prospective single-centre observational study of the first 693 consecutive medial cementless OUKRs performed at Burwood Hospital (Christchurch, New Zealand) from the introduction of this implant in our country in June 2004 until December 2011. All patients fulfilled selection criteria, reached a minimum follow-up of five years, and were performed by, or under direct supervision of, one of four high-volume surgeons who undertook between 82 and 106 knee arthroplasties per year. As this was a low-risk observational study and audit of radiological and functional outcome in patients receiving standard treatment and follow-up, ethical approval was not required. Patients were assessed preoperatively and at minimum of one, two, five, seven, ten, and 15 years after surgery. The clinical outcome used in this study was the Oxford Knee Score (OKS).^26,27

Complications or further surgeries were recorded when they occurred, or at each follow-up appointment. Patients were contacted by post or telephone to obtain the OKS, and appointments made for long-term follow-up radiographs. Revision was defined as removal, exchange, or addition of a component and was further confirmed by cross-referencing with the New Zealand Joint Registry (NZJR). The NZJR has a capture rate of > 97% for revision knee procedures, and an independent team followed up the results of all patients in the cohort. Therefore, it is assumed that no one is lost to follow-up.²⁸ Survival was determined with life table analysis.

All radiological data were collected by an independent observer (OLID) who was not involved in the procedures or follow-up assessments. Radiological assessment was undertaken immediately postoperatively, at minimum of six months and one, two, five, and ten years. Any questionable results were reviewed by both senior surgeons involved as authors of the study (GJH, ARM). All knees underwent fluoroscopic screening to achieve a true anteroposterior (AP) radiograph with the tibial base plate at 90° to the X-ray beam to allow the tibial implant-bone interface to be assessed as described by Gulati et al.⁴ Radiolucent lines around the tibial base plate were assessed on the AP radiograph within seven zones of interest (Figure 1), as previously described.⁴ Zone 7 was used to describe the vertical, lateral wall of the tibial tray; this area is hydroxyapatite-coated without titanium and is not thought to undergo significant loading. As it is not considered a weightbearing zone, radiolucent lines in this area are not considered to be of significance.⁴ Osseointegration in this region was analyzed, although it is not considered to affect overall fixation. Lateral radiographs were also obtained and studied, looking for any changes around the femoral component, also within the seven zones of interest (Figure 1).²⁹ Six of these zones are related to the pegs, which are round and less influenced by any obliquity of the radiograph; any gross or progressive lucency, especially around these femoral pegs, was recorded. Alignment of the femoral and tibial components was also assessed in both the coronal and sagittal planes using Inteleviewer software (Intelerad, Canada). Measurements were obtained relative to the anatomical axis of the femur and tibia as described by Sarmah et al.¹³

Fig. 1.

a) Anteroposterior radiograph showing the six zones of the tibial component. b) Lateral cross-table radiograph showing the seven zones of the femoral component.

All four surgeons were previously familiar with the implant, having used the cemented OUKR, however these were the first uncemented implants used and therefore represent the ‘learning curve’ of all four surgeons with this device. The cementless OUKR differs in that the femoral component has a longer extension anteriorly, with an additional femoral peg. The pockets of the twin peg femoral components have a calcium hydroxyapatite coating and a titanium porous plasma spray on their surfaces that contact bone. The vertical wall of the tibial tray has hydroxyapatite without porous titanium. All procedures were performed through a minimally invasive approach. During the course of this study, the technique for tibial preparation was altered with the introduction of a specially designed keel cut saw blade, otherwise the bony preparation was performed similar to the previous cemented technique. All patients were encouraged to weightbear as tolerated immediately following surgery, and were treated with a standard rehabilitation protocol.

Statistical analysis

The joint survivorship is summarized as the revision rate recorded as rate per 100 component years. This is equivalent to the yearly revision rate as a percentage, and is calculated as the number of prostheses revised divided by the total observed component years multiplied by 100. Revision-free survivorship is estimated using the Kaplan–Meier method, with the Greenwood estimates for the CIs. For these estimates, patients were censored at the date of death. A two-tailed p-value < 0.05 was taken to indicate statistical significance, and all analyses were undertaken using SPSS v29.0 (IBM, USA).

Results

Revisions

Considering revisions for any cause, the percentage revision-free at 16 years was 89.2% (95% CI 84.1 to 94.2), with a total of 46 revisions performed (Figure 2).

Fig. 2.

Kaplan-Meier curve demonstrating the revision-free survival at 16 years.

Of the 46 cases that required revision surgery (45 patients, one of whom had bilateral UKAs which both required revision to TKA), progression of arthritis was the most common reason (24 knees), with 19 knees converted to a TKA and four having lateral compartment arthroplasty.

The second most common reason for revision was bearing dislocation. There were ten primary bearing dislocations, of which nine were successfully managed with an upsized polyethylene bearing, while one knee required revision to TKA. All of the bearing dislocations occurred more than two years after primary insertion. Eight of the ten bearing dislocations had a history of a major injury associated with the dislocation event, such as a fall from a bench, or knee injury to the medial collateral ligament while playing tennis. Five of the ten patients had a bearing dislocation in the context of a twisting knee injury and a partial ACL injury, all of which were treated with bearing exchange with no further issues. A further two patients sustained a bearing dislocation in the context of a twisting knee injury with ACL rupture. Both patients were successfully managed with an ACL reconstruction and upsized polyethylene bearing. Two of the ten cases did not have a cause for dislocation listed in their clinical notes.

There were three cases of polyethylene wear: one was managed with a polyethylene bearing exchange, while the other two required revision to a TKA.

Aseptic loosening was rare, with no cases of femoral loosening and three with tibial component loosening requiring revision to a TKA. Two cases had tibial loosening at 12 months postoperatively, and the other had tibial loosening at 13 years postoperatively. The two cases of aseptic loosening occurring within 12 months were due to surgical error, with one case of incomplete seating of the tibial implant and one component placed in valgus and revised to a TKA due to some tibial subsidence. No clear risk factors for aseptic loosening were noted for the case occurring after 13 years.

There were two periprosthetic fractures of the tibial plateau. One occurred eight months and the other at three years following the index procedure. Both were in the context of a fall and were successfully managed with revision to TKA. There were no cases of intraoperative fracture in this cohort.

One acute deep infection occurred one month following the index procedure and was managed successfully with debridement and polyethylene bearing exchange with implant retention. No late or chronic periprosthetic infections were identified.

There was only one revision where the underlying pathology was uncertain, with a subsequent revision to a TKA.

When considering all revision operations, the revision rate for the OUKR was 0.57/100 component-years (95% CI 0.42 to 0.77). Considering conversion to TKA as the end-point, the conversion-revision rate was 0.35/100-component years (95% CI 0.23 to 0.50). When bearing dislocations are excluded, the revision rate was 0.46/100-component years (95% CI 0.32 to 0.64).

Clinical outcome

The mean OKS improved from 22.3 (SD 7.4) preoperatively to 40.72 (SD 6.8) at six to 12 months. The postoperative OKS did not change significantly with time (Figure 3).

Fig. 3.

Pre and postoperative Oxford Knee Scores at different postoperative timepoints.

Radiological analysis

Overall, 575 patients (687 knees) had adequate follow-up for radiological analysis. At ten years there was no evidence of femoral loosening or subsidence in any of the knees, with no radiolucent lines. There were no complete radiolucent lines identified in any zone around the tibial base plate, except for zone 7 (42 lines). There were ten other incomplete lines seen in other zones with no progression.

Implant positioning of the tibial component in the coronal plane related to the tibial anatomical axis showed a mean 2.8° of varus (12.7° of varus to 6.6° of valgus). All implants were found to have been implanted with mean 6.2° of posterior slope (0° to 15.8°). The femoral components had a mean 5.2° of varus (18.6° of varus to 9.3° of valgus) related to the femoral anatomical axis and in the sagittal plane were positioned with a mean 7.5° of flexion (25.4° flexion to 16.4° extension).

Discussion

The cementless OUKR was introduced in an attempt to reduce aseptic loosening, one of the most common causes for revision of the cemented prothesis. Several short- and medium-term studies have demonstrated a reduced incidence of radiolucency, shorter operating time, and better outcomes when compared with the cemented prosthesis.^19,20 However, despite these studies, concern has remained regarding the long-term outcomes of the cementless prosthesis. This large prospective study supports the use of this component by demonstrating excellent survival and functional results greater than 15 years, with a very low incidence of aseptic loosening, and no femoral loosening observed. Furthermore, there were no exclusions for eligibility for using this implant based on patient age, sex, or weight, and as such the results are generalizable to all patients who meet the appropriate indications for an OUKR.

The minimum ten-year radiological analysis did not identify any protheses that were failing, with no cases of progressive radiolucent lines or implant subsidence. The highest rate of radiolucent lines was seen in tibial zone 7. This area is neither titanium-coated nor directly loaded, and as such is believed to contribute little to the overall stability of the prosthesis.^30,3,31 Furthermore, radiolucent lines in zone 7 are not believed to be a reliable indicator of prosthetic loosening or pending failure.⁴

In 2020, Mohammad et al²¹ reported on the ten-year radiological outcomes of the first 1,000 cementless OUKRs performed by the designer surgeons. This cohort had only one revision for femoral component loosening (0.1%), with 13% of tibial components and 1% of femoral components having partial radiolucent lines.¹⁹ This compared favourably to the reported one-third of cemented OUKRs with radiolucent lines, and suggests improved fixation with the cementless prosthesis.²¹ The efficacy of cementless fixation for UKA is likewise supported by the results of the NZJR, which demonstrates an all-cause revision rate of 0.77/100 component-years (95% CI 0.54 to 0.89) for cementless compared with 1.29/100-component years (95% CI 1.21 to 1.37) for cemented UKAs.² Similar patterns are seen on the UK National Joint Register, with the revision rate at 15 years for the cementless OUKA being 11.54 (95% CI 8.58 to 15.43) and the cemented OUKA 16.55 (95% CI 16.12 to 16.99).³²

Excellent long-term survivorship has also been reported by the designer surgeons, however in 2011 Labek et al²⁴ raised concerns regarding the validity of these results. It was noted that the outcomes from their reports deviated significantly from those reported on national registries and independent studies. Conversely, both Campi et al²² and Mohammed et al¹⁸ have demonstrated that similar results can be achieved by both designer and non-designer surgeons. When compared with the New Zealand Joint Registry, this cohort had superior outcomes, with an 89.2% 16-year survival compared with 81.1% in the NZJR.² The observed difference in these results may reflect the high-volume nature of the surgeons involved in this cohort. Surgeon’s volume has been demonstrated to have an important impact on the outcomes of UKA, with higher revision rates in low-volume surgeons.² The NZJR shows a revision rate of 1.38/100 component-years (95% CI 1.27 to 1.50) versus 1.00/100 component-years (95% CI 0.92 to 1.09) for those surgeons performing fewer or more than ten UKAs per year, respectively. These data would tend to suggest that surgeons who use the recommended indications and technique, with a moderate-volume practice (> ten OUKAs per year), should be able to achieve similar results to designer surgeons.² This has been supported in a 2017 meta-analysis.³³

This study demonstrates a significant and durable improvement in the OKS, compared with preoperative scores, with no significant decline in functional outcomes over 16 years. UKA has been thought to provide better functional outcomes than TKA because of the less invasive surgery and more rapid recovery, and our results show a rapid and sustained improvement in OKS, but these results are not clinically different from the six-month and five-year results for TKA recorded on the NZJR.

Periprosthetic fractures are a recognized complication of UKA. A cadaver study has demonstrated a higher risk of periprosthetic fracture in the cementless prosthesis compared with the cemented prosthesis.³⁴ Reported technical errors that increase the risk of tibial plateau fracture include tibial re-cutting, a medial sagittal cut, deep tibial resection, and damage to the posterior cortex while performing the vertical cut or preparing the keel slot.^34,35 Careful adherence to the surgical technique can minimize the risk of this complication. It is recommended that after the keel saw cut in the tibial component is made, the slot should be enlarged so as to easily accommodate the trial component, especially in smaller-sized tibial trays. Furthermore, the tibial component should be gently impacted. Despite this cohort including all cases from the introduction of the cementless prosthesis and therefore the learning curve for cementless fixation, the rate of periprosthetic fracture was very low (0.2%) with only two such cases. Furthermore, the fractures occurring in this series are not typical of those associated with a UKA, which tend to occur early within the postoperative period. One fracture occurred eight months postoperatively and the other three years postoperatively. Both fractures were associated with trauma with no clear risk factors for fracture identified.

Revision rates for UKAs are universally higher than for a total knee arthroplasty. The revision rate for a primary knee arthroplasty in the NZJR is 0.48/100 component-years (95% CI 0.46 to 0.49) compared with a revision rate of 0.77/100 component-years (95% CI 0.54 to 0.89) for a cementless UKA.² A similar trend is seen in the UK National Joint Registry, with the revision rate for a cementless UKA being 2.3 and three times higher than a cemented TKA at ten and 15 years, respectively.³² Surgeons have been considered as having a lower threshold to revising a UKA than a TKA, and consequently have been more prepared to revise a patient with an unsatisfactory outcome following UKA compared with TKA, despite no known diagnosis for that poor outcome.³⁶ Tay et al³⁶ demonstrated that the relative risk of revision for a UKA was 2.5 times higher for unexplained pain and ‘other unknown’ reasons compared with a TKA. This is further supported in the literature, with data showing that for any given OKS, a UKA is five times more likely to undergo a revision compared with a TKA.³⁷

The revision rate in our series is considerably closer to that of a primary knee arthroplasty, with similar revision rates when bearing dislocations are excluded (0.46/100 component-years for cementless OUKR). Bearing dislocation was the second highest cause for revision and may be directly related to the cementless implant. After implantation of a cementless implant we commonly expect some ‘settling’ of that implant before osseointegration occurs, and this has been demonstrated with radiostereometric analysis performed by the designing centre where femoral and tibial subsidence was often observed, with more subsidence seen in cementless tibial components compared with cemented.¹⁹ This has the effect of increasing the laxity of the medial compartment and may be a factor in bearing dislocation. This has prompted some surgeons to insert larger bearings than previously recommended to maintain stability while this subsidence occurs. Currently, there are no data to confirm whether this practice reduces bearing dislocation, or is associated with other adverse effects such as lateral disease progression.

The most common cause for revision was lateral compartment disease progression which occurred several years after the index procedure. In these cases the OUKR was still functioning satisfactorily with no evidence of loosening, and an argument could be made that the cause of the revision was not in fact related to the implant, but was one of patient selection. In fact, if we eliminate ‘progression of disease’ as a cause of implant failure, then the OUKR performs better than TKA.

This study has some limitations. First, it is an observational study using prospectively collected data; there was no control group, as all surgeons were using the cementless OUKA as their sole UKA implant throughout the study. This disadvantage is offset by the fact that all patients were recruited consecutively, with no patients being excluded because of age or bone quality.

Second, a radiological analysis at 16 years was not performed. However, the analysis of the minimum ten-year radiological outcomes for the cementless OUKR did not demonstrate any complete radiolucent lines or failing implants, and the low rate of radiolucent lines with a lack of progression is encouraging for this prosthesis. Third, these are the results of four high-volume surgeons, who committed to using UKA in all appropriate cases. As mentioned earlier, the outcomes of OUKR are influenced by the number of cases performed by the surgeons each year, but we believe that by using the same indications and surgical technique proposed by the design surgeons, other surgeons should be able to achieve similar results.

A further limitation is the lack of long-leg weightbearing views to assess overall alignment postoperatively and at subsequent follow-up. However, all implants were screened fluoroscopically to obtain reproducible and reliable images centred on the tibial base plate, which was a considerable advantage in assessing for radiolucent lines. However, it is accepted that obtaining a perfect image of the tibial base plate is not always possible and, as previously mentioned, is likely to be responsible for the difference in reporting of radiolucent lines between the two study times. The lack of radiological assessment of alignment was also thought to be unnecessary, as alignment of all implants was assessed clinically at time of follow-up, with no reported instances of ongoing varus or valgus malalignment. As part of the surgical procedure, it was considered important not to overcorrect, and no attempt was made to position the knee in physiological valgus.

In conclusion, cementless fixation for the medial OUKR is a safe and effective procedure for all patients who are eligible candidates, with excellent 16-year survival and long-term clinical outcomes in the hands of non-designer surgeons. In particular, this cohort had a very low rate of aseptic loosening. Surgeons are able to achieve excellent results with this prosthesis if they adhere to the recommended indications and surgical technique.

Take home message

- The cementless medial Oxford Unicompartmental Knee Replacement (OUKR) has excellent 16-year survival and clinical outcomes in the hands of non-designer surgeons, when there is careful patient selection and adherence to the correct surgical technique.

- The medial OUKR is associated with a very low rate of aseptic loosening.

Data Availability

The datasets generated and analyzed in the current study are not publicly available due to data protection regulations. Access to data is limited to the researchers who have obtained permission for data processing. Further inquiries can be made to the corresponding author.