Patients with unicompartmental osteoarthritis of the knee often face the decision to proceed with unicompartmental knee arthroplasty (UKA) or total knee arthroplasty (TKA). There are a host of considerations, and one of those is the fate of a subsequent revision of the primary surgery, for which there exist conflicting data in the literature, both from small case series and from registries. Leta et al. provide additional insight into this issue with their comparative study of 768 failed primary TKAs revised to TKAs (rev-TKA) and 578 failed primary UKAs revised to TKAs (rev-UKA) reported to the Norwegian Arthroplasty Register between 1994 and 2011.

The methodology was well described and in keeping with that of most registry studies. Importantly, all cases for septic revision were excluded. The authors aimed to evaluate the technical complexity of the procedures, functional outcomes, and survivorship for the two cohorts. Specifically, they tabulated the re-revision rates and the reported indications through a standard survivorship analysis. The study was adequately powered to evaluate survivorship differences with relative risks of at least 2 and to detect clinically important differences in the Knee Injury and Osteoarthritis Outcome Score (KOOS). Propensity analysis was used to match the two revision groups as well as possible for age at revision, sex, primary diagnosis, technique, and duration of time since revision. Scores on patient reported outcome measures (PROMs) were obtained by way of a self-completed mailed questionnaire sent in 2006, with measures including the EuroQol EQ-5D, KOOS, and visual analog scales (VAS) assessing satisfaction and pain. The authors acknowledge that these PROM outcome data were limited to approximately 20% of the total of 1346 revised cases evaluated.

There was a survivorship advantage for rev-UKA in the revision period of 1994 to 2002, but this was not seen for the period of 2003 to 2011. The authors did find a significant difference in survivorship in the age-stratified analysis, with a greater risk of re-revision for rev-TKA among those who underwent revision at an age of more than seventy years. Not surprisingly, rev-TKA procedures took longer (a mean operative time of 150 versus 114 minutes), and more of the procedures required stems (58% versus 19%) and constraint (27% versus 9%), all of which were considered proxies for technical complexity. This may partially explain the overall significant twofold increased risk of failure due to infection in the rev-TKA group. Other differences in the indication for re-revision were exposed, including a higher proportion of revisions for pain alone in the rev-UKA group. The authors hypothesized that, because that cohort was younger, it may have followed that they were more active, with greater expectations, although this was not tested in the present study.

There was a twofold difference in the rate of revision for the indication of tibial component loosening in the rev-UKA group that may seem counterintuitive, although this may also stem from patient age, functional demand, or surgical technique. As with rev-TKA, there is a spectrum of bone loss to address in rev-UKA. On balance, and as referenced in the article, the technical complexity of rev-UKA is generally less than that of rev-TKA. Surgeons who are technically experienced in UKA use conservative bone resection in primary operations that can make revision of primary components more straightforward. This is a major potential issue that, in practice, can separate a simple UKA revision from a technically difficult one and possibly lead to fewer subsequent re-revisions for tibial loosening in particular. Larger bone defects requiring augments should be protected with short cemented or uncemented diaphysis-engaging stems. In this study, fewer stems were used in the rev-UKA group (19% of the procedures) compared with the rev-TKA group (58%). One could hypothesize that stems were underutilized in the rev-UKA group, especially if augments were needed, but those data were not available in the article. The issue of revision due to pain is even more complex and influenced by the extent of arthritis before the primary surgery, psychosocial factors, and patient expectations.

A response rate of 85% (150 patients in the rev-TKA group and 127 patients in the rev-UKA group) was achieved for the evaluation of the mailed PROM survey of the patients who underwent revision between 1994 and 2005. There was no difference between the two groups in the preoperative, postoperative, or change in scores for the EQ-5D and in the KOOS subscales, VAS for pain, and rate of dissatisfaction (25% of patients in the rev-UKA group and 22% in the rev-TKA group). The authors concede that these PROM outcomes relied on memory for preoperative status, so there may be threats to the validity of the change in scores reported. Furthermore, we do not have data to assess the activity levels in these patients, which could have influenced the self-reported outcomes.

The discussion and conclusion are well written and balanced, and the authors refer to the Australian Orthopaedic Association National Joint Replacement Registry, which reported a significantly higher risk of re-revision for rev-TKA cases (hazard ratio of 1.41 in 2013, and more recently, 1.5, as reported in its 2015 hip and knee revision report). Revision rates for UKA in registries with data from high-volume surgeons have been very favorable and, in some cases, have approached those of primary TKA. This may reflect experience with initial patient selection and surgical technique as well as indication for revision. Of note, revisions for pain alone with no other definitive cause have not generally yielded good results. This was somewhat mirrored in the present study, with pain being a relatively more common indication for both revision and re-revision in the rev-UKA group.

Unfortunately, registry studies do not provide the granularity regarding the specific revision technique and the osseous defects that required treatment. Given that the rate of revisions due to tibial component loosening in the rev-UKA group was notably higher, it would be prudent for surgeons who perform UKA to be conservative in tibial bone resection, which will have the dual advantage of allowing seating of the implant in more structurally robust proximal bone while retaining bone stock to allow for a simpler conversion to TKA if and when that should occur.

In summary, I congratulate the authors for this detailed examination with respect to the state of revision UKA and revision TKA among patients in the Norwegian Arthroplasty Register. Patients can be reassured that the functional and survivorship outcomes of rev-TKA do not appear to be much different from those of rev-UKA. Surgeons can expect that rev-TKA will be a lengthier, more complex procedure with a higher risk of infection. Patients facing the initial decision between UKA and TKA should likely focus more on differences in perioperative morbidity, clinical outcomes, and satisfactionand the likelihood of a future revision given their age and level of activity.