Unicompartmental Knee Arthroplasty

We aim to restore the kinematics of the knee in unicompartmental arthroplasty as it was being used by that individual before the arthrosis developed. The knee is used in compression when standing and squatting and in swing phase, where a competent anterior and posterior cruciate ligament complex allows efficient and congruent flexion after toe off, followed by extension leading to heel strike. This combination of ligament tension and joint congruence is the key to a natural and efficient gait at varying speeds and gradients. Following UKA, which restores both stability and congruence, this state can be approached, but it is very hard to achieve following TKA which inevitably involves ACL sacrifice [1].


INTRODUCTION
Debate continues regarding the optimal surgical treatment for isolated medial compartment osteoarthritis (OA) of the knee. Although high tibial osteotomy is an option, most patients are treated with either a unicompartmental knee arthroplasty (UKA) or total knee arthroplasty (TKA). UKA has been around since the 1950s; however, initial designs were fraught with complications. 1 As component designs and instrumentation have evolved, the survivorship of UKA improved greatly. 2-5 UKA has many advantages over TKA, including improved knee kinematics, range of motions, and functional outcomes. 6-10 Furthermore, medical complications with UKA are significantly less than TKA. [11][12][13] US and global use of UKA has varied over the years. In the United States, there was a steady increase in UKA use from 2002 to 2008, after which there has been a decline. 14 Data from the 2018 Australian National Joint Replacement Registry demonstrate that partial knee replacement represented 8.6% of primary knee arthroplasties in 2017, which was down from 16.9% in 2003. 15 A similar incidence of use of UKA in 2017 was reported from the National Joint Registry of England and Wales (NJREW) at 8.9%, which has remained consistent over the past decade. 16 UKA is not performed by all arthroplasty surgeons, and a small percentage of surgeons perform most procedures. 17 Surgeons tend to be pretty rigid on their impression of UKA, with a handful of zealots as well as vocal detractors. This article presents US and global perspective on UKA.

INDICATIONS FOR MEDIAL UNICOMPARTMENTAL KNEE ARTHROPLASTY
The classic restrictive inclusion criteria of Kozinn and Soctt 18 have been greatly expanded with modern research demonstrating successful outcomes with UKA in younger patients, 19 obese patients, 12 patients with patellofemoral a JIS Orthopedics, 7277 Smith's Mill Road, Suite 200, New Albany, OH 43054, USA; b Cliniques universitaires Saint Luc, Avenue Hippocrate 10, Brussels 1200, Belgium disease, 20 and those who are very active. 21 However, proper patient selection is still vital to ensuring a successful outcome with UKA. Medial UKA could be considered in all patients with anteromedial OA (Fig. 1) with correctable deformity (Fig. 2), intact knee ligaments, and preserved knee range of motion with less than 15 flexion contracture. UKA should be avoided in patients with inflammatory arthropathy and used cautiously in those who have previously undergone a high tibial osteotomy. Patients should have full thickness cartilage loss and/or avascular necrosis, because those with partial thickness loss have inconsistent pain relief and 6 times higher revision rate. 22 Studies have estimated that between 25% and 48% of patients presenting with knee OA are candidates for UKA. 23,24 UNICOMPARTMENTAL KNEE ARTHROPLASTY VERSUS TOTAL KNEE ARTHROPLASTY: FUNCTION Although TKA achieves excellent outcomes on a range of measures, there remains a portion of patients who are not completely satisfied with their outcome and suffer continued impaired functional activity 25 and persistent postsurgical pain. 26,27 In a multicenter study, Nam and colleagues 26 reported that, although 90% of patients after TKA had overall satisfaction with the functioning of their knee, only 66% felt that their knee was "normal," with nearly half conveying residual symptoms and functional problems. The persistence of symptoms lends itself for other implant concepts that can improve functional outcomes with a more normal feeling knee.
UKA allows for more closely matched knee kinematics of the native knee due to its cruciatepreserving nature, and its intact contralateral compartment and patellofemoral anatomy. [6][7][8] This results in a more normal gait, as well as reduced perioperative trauma, greater range of motion, and faster rehabilitation. 9,10 In a series of 23 patients with a UKA in 1 knee and a TKA in the contralateral knee, patients more often reported the UKA as feeling normal. 28 In a 10-year minimum propensity matched analysis of 519 UKA to 519 TKA, Burn and colleagues 29 found that UKA was associated with better Oxford Knee Scores and EQ-5D. UKA was also associated with an increased likelihood of successful outcome and chance of obtaining minimally clinically important improvements in Oxford Knee Scores and EQ-5D.

UNICOMPARTMENTAL KNEE ARTHROPLASTY VERSUS TOTAL KNEE ARTHROPLASTY: PERIOPERATIVE COMPLICATIONS
UKA is a less-invasive procedure than TKA, and as such the risk profiles of these surgeries differ as shown in both US and global research publications. In a 2:1 matched cohort study of obese patients undergoing TKA versus UKA in the United States, the UKA cohort had significantly less blood loss (P 5 .004) as well as a lower infection rate (0% vs 0.5%, P 5 .016). Furthermore, the risk of manipulation was significantly higher in the TKA group (6.5%) compared with the UKA group (0.5%) (P<.001). 12 Another US study by Hansen and colleagues 13 compared complications and outcomes between UKA and TKA by analyzing a 5% Medicare sample from 2004 to 2012. Compared with TKA, UKA was found to have significantly lower wound complications, pulmonary embolisms, infection, myocardial infarction, readmission, and death.
In analysis of the NJREW, Liddle and colleagues 30 found that the average length of stay, rate of medical complications, such as thromboembolism, myocardial infarction, stroke, and rate of readmission, were all higher for TKA than for UKA. Furthermore, 30-day and 8-year mortality was lower with UKA than with TKA, at hazard ratios of 0.23 and 0.85. Hunt and colleagues, 31 in an analysis of 467,779 primary knee replacements from the same registry, found that UKA was associated with substantially lower 45-day mortality than TKA, at a hazard ratio 0.32 (P<.0005). There is global consensus, by way of studies out of India, England, Canada, and the United States, that UKA has a shorter length of stay and decreased rate of readmission compared with TKA. 13,[31][32][33] UNICOMPARTMENTAL KNEE ARTHROPLASTY VERSUS TOTAL KNEE ARTHROPLASTY: SURVIVAL Surgeons must balance the improved function and lower medical complications of UKA with the consistently reported lower survival compared with TKA. For example, the Australian Orthopedic Association National Joint Replacement Registry report found the 10-year survival rates with TKA and UKA in primary OA were 94.4% and 84.7%, respectively. 15 This disparity is also reflected in the NJREW and Swedish Knee Arthroplasty Register annual reports. 16,34 The 15th annual NJR report noted that the 14year cumulative revision rate for TKA was 6%, whereas that for UKA was 16.9%. 16 In a study evaluating the survivorship of UKA versus TKA over an 8-year period in US Medicare patients, Hansen and colleagues 13 reported the 7-year survivorship of UKA to be 80.9% versus 95.7% for TKA. A systematic review of cumulative data from 6 national registries and clinical studies found that the overall 10-year revision rate for TKA was 6.2% compared with 16.5% for UKA. 35 Liddle and colleagues 36 used propensity matching to adjust for confounding baseline patient demographics to compare 25,334 UKAs and 75,996 TKAs from the NJREW. At 8-year follow-up they found that UKA had worse implant survival in terms of both revision and revision/reoperation than TKA.
Some have suggested that the discrepancy in revision rates between UKA and TKA may be due a lower surgeon threshold to revise a painful UKA. In their analysis of the New Zealand Joint Registry, which includes outcome data, Goodfellow and colleagues 37 found a significantly greater percentage of patients reporting good or excellent results after UKA than TKA and fewer patients reporting poor or fair outcomes. Although both groups had an increase in revision rates with lower outcome scores, patients with UKA had a 6 times greater revision rate than those with TKA for the same score category of Oxford Knee Score of less than 20. Analysis of the Norwegian Arthroplasty Register found that UKA is more than 11 times more likely to be revised for pain than TKA, at a relative risk of 11.3 (P<.001). 38 Ultimately, the choice between UKA and TKA according to available literature is still a choice between function and survivorship. 39 Better preoperative analysis of the patient's disease process leading to arthritis progression, 39 and better understanding of pain after knee arthroplasty 40 could reduce the number of necessary revisions to TKA.

CEMENTED VERSUS CEMENTLESS UNICOMPARTMENTAL KNEE ARTHROPLASTY
Although most UKA designs use cemented fixation, 41 the minimally invasive nature of the procedure can make the insertion and extrusion of cement challenging. [41][42][43] Cementation errors can lead to failure of fixation and aseptic loosening, which is the leading cause of revisions in UKA. 2,42,44,45 Extruded cement can break off and become a loose body creating third body wear on both the replaced surface as well as normal articular cartilage in the unresurfaced portions of the knee. Cementless UKA could be a way of reducing these failures and achieving more durable long-term fixation. Cementless UKA has consequently become increasingly popular in recent years. 42 Early data from the Finish Arthroplasty Registry found that the 5-year survival of the cementless Oxford UKA was 92.3% compared with 88.9% for the cemented Oxford UKA. 46 In the 2018 NJREW report, cementless UKA knees had a slightly higher 10-year survival at 87.3% comparted with cemented knees at 85.1%. 16 In a comparison of cemented versus cementless Oxford Partial Knee, Pandit and colleagues 44 found significantly more radiolucencies in the cemented group at 20/30 knees versus 2/27 knees in the cementless (P<.001). Furthermore, there were 9 complete radiolucencies in cemented knees, as opposed to none in the cementless group (P 5 .01) [Pandit]. 44 These findings were further supported by Kendrick and colleagues, 47 who noted significantly less tibial radiolucencies in cementless UKA compared with cemented UKA at 2 years (P 5 .02).
Some rare complications have been reported for cementless UKA, including early subsidence of the tibial component into a valgus position. 48 A cadaveric study has also suggested that cementless implants may be more susceptible to periprosthetic tibial plateau fracture (PTPF) [Seeger], 49 although this may be due to implantation errors known to dispose toward PTPF, such as a deep posterior cortical cut in the tibia and perforating the posterior cortex perforation during keel preparation. 42 Overall, cementless UKA is a promising technology, although more research is required and cemented fixation remains the gold standard.

FIXED VERSUS MOBILE UNICOMPARTMENTAL KNEE ARTHROPLASTY
Fixed bearing (FB) (Fig. 3) and mobile bearing (MB) (Fig. 4) are the 2 main design concepts in UKA. [50][51][52] Although the theoretic advantages of MB prostheses over FB designs have made it increasingly popular, 53 advances in polyethene manufacturing have significantly decreased the wear in the FB design. 54 With polyethene wear no longer a major issue with FB designs, the  Crawford et al choice of design for UKA therefore remains controversial. 55 Several meta-analyses have compared FB with MB designs. Peersman and colleagues 56 reviewed 44 comparative and noncomparative studies, involving 9463 knees. After stratification by age and follow-up time, there were no major differences in survival rates between FB and MB implants. However, the mean time to revision was shorter for MB knees, at 2.5 AE 1.8 years and 6.7 AE 2.5 years, respectively (P<.001). 56 In another meta-analysis comparing FB and MB UKA, Cheng and colleagues 53 examined 9 studies, involving 915 knees. They also found no significant differences between the implants in terms of clinical outcome scores, range of motion, or revision rates. Similar to Peersman and colleagues, 56 they found the time to revision was significantly sooner in MB (5.0 vs 6.3 years for FB implants, P 5 .016), with early failures in MB patients due to bearing dislocations; later failures in the FB group were more commonly due to polyethylene wear. Smith and colleagues 52 identified 5 studies comparing FB and MB implants for medial and lateral UKA, involving a total of 165 and 159 knees, respectively. There were no significant differences between the implant types in medical UKA in terms of clinical outcome, and no differences in complication rates.
In summary, any decision between FB and MB designs in UKA is hampered by a lack of robust and well-designed studies, and an evidence base limited to observational studies and small, randomized controlled trials. Nevertheless, it seems that there are no major differences between the 2 implant types.

FAILURE ANALYSIS OF UNICOMPARTMENTAL KNEE ARTHROPLASTY Aseptic Loosening
The most common cause of early UKA failure is aseptic loosening, representing 28% to 59.2% of all UKA revisions. 2,57-59 This trend has been shown in multiple international registries, including Sweden, England/Wales, Australia, and Italy, 57-59 along with US institutional case series. [60][61][62] However, the overall incidence of aseptic loosening in large reports is low with rates reported between 1.5% and 2.7% in midterm follow-up. 57,63 There are specific UKA designs that have much higher reported incidences of aseptic loosening. In a randomized trial between all-polyethylene and metal-backed tibial trays, the all-polyethylene design had a 37% incidence of aseptic loosening. 64 This may be attributed to the more focal tibial loading seen in these designs. 65 Surgeons must be cautious of interpreting all radiolucencies as aseptic loosening, which is discussed below.

Arthritic Progression
With appropriately selected patients and proper surgical technique, the risk of arthritic progression following UKA (Fig. 5) is low. At minimum 10-year follow-up, Emerson and colleagues 66 found an incidence of 4.2% for revision due to lateral progression. Progression of arthritis, however, is the most common reason for mid-term (5-10 year) and long-term (>10 year) failure in UKA. 2 In a meta-analysis Van der List and colleagues 2 showed that 40% of revisions at greater Unicompartmental Knee Arthroplasty than 10 years were for progression of arthritis. In the Australian Orthopedic Association National Joint Replacement Registry, 43.7% of UKA failures were for disease progression. 15 The NJREW 2018 report noted that 32.3% of UKA revisions were for arthritic progression. 16 The most significant predictor of progression of OA is the arthritic grade of the lateral compartment at the time of surgery. 44 This emphasizes the importance of proper patient selection. If full thickness lateral disease is identified intraoperatively, TKA should be performed instead of UKA.

Pain
Residual medial tibial pain can persist for up to 1-year after UKA. More chronic persistent pain is the cause of 8% of early failures and 10% of late failures after UKA. 2 Even in well-aligned UKA, the proximal medial tibial stress is increased up to 50% of the native bone. 67 Surgeons should be cautious about revising a UKA for unexplained pain because results are significantly worse than when a cause of pain can be identified. 68

Radiolucent Lines
Radiolucencies can be an indication of component loosening; however, "physiologic" tibial radiolucencies (Fig.  6) are well-defined, nonprogressive radiolucencies that can be seen in up to 62% of UKA with the Oxford UKA. 69,70 These physiologic radiolucencies do not correlate with clinical outcome or component failure. 70 Surgeons who are relatively unfamiliar with radiolucencies may attribute medial knee pain to the presence of a radiolucent line, and convert the UKA to a TKA for aseptic loosening. This is often unnecessary, as a fine, well-defined radiolucent line may be present at the bonecement interface even in well-functioning cemented UKAs. 71 Other Failure Modes Bearing dislocation is a unique complication to the MB designs, with incidence reported between 2% and 3%. 37,72 The cause of bearing dislocation is multifactorial, but advances in the Oxford instrumentation have reduced the risk. 73 It is critical to ensure there is no remaining posterior femoral osteophyte at the time of surgery, which can lead to impingement of the bearing resulting in anterior dislocation. If components are properly aligned, bearing often dislocation can be managed by removing any impingement and replacing a new bearing.
Medial tibia collapse (Fig. 7) or fracture is a rare complication after UKA, representing only  2% of revisions for failed UKA. 2 These fractures can be related to component alignment as well as surgical technique. Coronal alignment in greater than 6 of varus or any valgus alignment significantly increases the load to the proximal medial tibia. 74 Treatment of these fractures can include open reduction internal fixation or revision to a TKA depending on the stability of component and size of the fracture.
Periprosthetic joint infection (PJI) after UKA is quite uncommon, with incidence reported between 0% and 1%. 5,57,75 Furthermore, PJI after UKA is significantly lower than TKA. 76 Unlike in a TKA, however, with a PJI in an UKA there is native cartilage remaining. Therefore, in these cases of chronic PJI after UKA, a 1-or 2-stage revision to a TKA should be performed to remove the damaged native cartilage from the infection.

REVISION OF UNICOMPARTMENTAL KNEE ARTHROPLASTY TO TOTAL KNEE ARTHROPLASTY
Revision of UKA to TKA (Uni to Total) has been stated to be a relatively easy procedure and might offer advantages over revision of primary TKA. 77,78 However, revision of UKA to TKA are typically more challenging than a primary TKA and approximately 50% of patients will have significant bone defects, and stemmed implants and/or augments are required in 33% of cases. [77][78][79][80][81] The mode by which the UKA fails influences the complexity of the revision, 60 but most UKA revisions can be successfully completed using primary components (Fig. 8). 82 Revision for tibial collapse poses the highest complexity ( Fig. 9) as these cases will more frequently require augments and constraint. 60,82 Using data from the New Zealand Joint Registry from 1999 to 2008, Pearse and colleagues 83 examined 4284 UKAs, of which 236 required revision, and compared those revisions with 34,369 primary TKAs. The authors found that the revision rate for Uni to Total was 4 times higher than that for primary TKA, at rates of 1.97 and 0.48 per 100 component years, respectively (P<.05). The mean Oxford Knee Score was also significantly worse in the Uni to Total group than in the primary TKA group, at 30.02 versus 37.16 (P<.01). 83 Contrary to these findings, Lombardi and colleagues 78 reported on 184 UKA to TKA revisions. At 6-year minimum follow-up, 4.1% of knees required rerevision. This rate was similar to the institution's revision rate of primary TKA's and much lower than their re-revision rate of failed TKA. The mean Knee Society Clinical score in the UKA to TKA revision group was 83.4.

UNICOMPARTMENTAL KNEE ARTHROPLASTY OUTCOME RELATED TO HOSPITAL AND SURGEON VOLUME
If surgeons choose to perform UKA, this should be a dedicated portion of their practice because surgeon experience influences the risk of failures. Liddle and colleagues 84 found that the revision rate following UKA dropped steeply until the annual volume reached 10 cases and plateaued at 30 cases per year. Furthermore, they found that case load more strongly affected risk of revision in UKA than TKA, indicating that UKA is possibly a more technically demanding surgery. Baker and colleagues 17 suggested that 13 surgeries per year should be the minimum threshold for performing UKA in their registry analysis of 23,400 UKAs from the New Zealand Registry and NJREW. Data from the Norwegian Arthroplasty Register suggest that the risk of UKA revision is lower in hospitals that perform more than 40 procedures a year than in those that carry out less than 10 per year, at a risk ratio adjusted for age, diagnosis, and sex of 0.59 (P 5 .01). The main reasons for Unicompartmental Knee Arthroplasty failure in low-volume hospitals are dislocation, instability, malalignment, and fractures. 85 Because it might be difficult for surgeons to increase the referrals to their practice, 1 option to increase UKA volume is to more closely evaluate the indications for UKA as noted previously. 36

COMPONENT POSITIONING AND ENABLING TECHNOLOGY
Correct component positioning is critical to the success and survival of UKA. 74,86 For example, lowering the medial joint space greater than 2 mm compared with the lateral joint space has been associated with more tibial aseptic loosening. 86 In MB designs, studies have found increased rates of bearing dislocation associated with excessive varus or valgus positioning of the tibial component, elevation of the medial joint, inadequate posterior slope, and internal rotation of the tibial component. 87 Excessive varus alignment decreases the contact area of the femoral component on the tibia, thus increasing contact stress and potentially leading to increased wear. 88 Malpositioning errors are more commonly seen in low-volume surgeons 85,89 and during the learning curve of the procedure. 90 Technologies to improve alignment accuracy may therefore benefit UKA even more than they have done for TKA. 91 Current enabling technologies for UKA include patient-specific instruments (PSI), computer navigation and robotic-assisted surgery.
PSI involves the manufacturing of instruments based on computed tomography or MRI to match the individual patient's anatomy. PSI has been shown in some studies to enhance implant alignment, 92,93 which should, theoretically, improve surgical outcomes and reduce the risk of revision. In a sawbones model, PSI technology allowed inexperienced trainee surgeons to have equivalent tibial saw cuts as high-volume experienced surgeons. 94 Computer navigation can further enhance the accuracy of component positioning with a significant reduction in outliers. [95][96][97][98] Suda and colleagues 95 found 100% accuracy within 3 of target alignment with the use of an accelerometer-based portable navigation system, compared with 76.5% coronal and 88.2% sagittal accuracy with conventional instruments. At 5-year minimum follow-up, Chowdhry and colleagues 99 reported a 97.6% survival rate of 252 UKA performed with computer navigation. This survival rate rivals that of TKA.
Robotic technology has been at the forefront of debate in arthroplasty for the past few years. One challenge with interpreting the literature on robotic-assisted UKA is that many of the studies are funded by industry with design surgeons writing the articles. 100 Although there is little argument whether robotics improves component positioning, 92,101-103 the question remains whether this technology improves patient outcomes to justify the cost. As discussed throughout this paper, component positioning is likely more critical to the survival of UKA compared with TKA, and thus this procedure may benefit more greatly from the increased accuracy. With robotics, the surgeon receives continuous, real-time feedback on knee kinematics, range of motion, and implant placement during the procedure with extremely accurate bony resection. In a matched prospective cohort study of robotic assisted versus conventional instrumentation, Kayani and colleagues 104 found that the robotic-arm assisted group had Fig. 9. Postoperative radiograph after revision for a failed UKA due to medial collapse that required a tibial augment and stem. significantly reduced postoperative pain, decreased opiate requirements, shorter time to straight leg raise, decreased therapy sessions, and reduced hospital stay. These early clinical improvements may be related to decreased periarticular soft tissue damage from confines of the stereotactic boundaries with roboticassisted UKA. At present, however, there have been no studies to show that robotic-assisted improves UKA mid-to long-term UKA survivorship. 103,105 However, robotic technology offers lower-volume surgeons to achieve high levels of accuracy with implant positioning.