Less pain reported 5 years after cementless compared to cemented unicompartmental knee replacement: an analysis of pain, neuropathy, and co-morbidity scores

Unicompartmental knee replacement (UKR) is an alternative to total knee replacement (TKR) for the surgical management of antero-medial osteoarthritis. The UK and other registry data show that UKR offers better function, faster recovery with fewer and less severe complications and is more cost effective [2, 3, 12, 16]. However, UKR has a higher revision rate than TKR, in part due to aseptic loosening [20]. To reduce the incidence of aseptic loosening, the cementless Oxford UKR (Zimmer Biomet), with a porous bone–implant interface for bony ingrowth to improve fixation, was developed. UK registry data have shown that the aseptic loosening rate of the cementless Oxford UKR is half that of the cemented [2].

Small randomised controlled trials designed to assess radiographic outcomes have suggested that patient-reported outcomes (PROMS) might be better following cementless than cemented Oxford UKR; however, these studies were underpowered for assessing clinical outcomes [13, 24]. In an adequately powered 5-year study, it was found that cementless Oxford UKR had better quality-of-life (EQ-5D), Oxford Knee Scores (OKS) and American Knee Society Scores (AKSS) than cemented [17]. Analysis of the sub-scores of these measures suggested that the improvement was due to a reduction in pain with the cementless UKR. However, it is not clear how large the difference is or why it might be present.

Following knee replacement, pain has traditionally been assessed with a few questions that form part of the overall assessment. Examples are the pain scores with the OKS or the AKSS [9, 11]. There are other scores available that are more sensitive and assess pain in much more detail. ICOAP (Intermittent and Constant Osteoarthritis Pain) is a PROM designed to assess pain in knee osteoarthritis [22]. PainDETECT (PD) is a PROM designed to assess likelihood of pain being neuropathic in origin, and also includes three visual analogue scales [6]. It is also possible that knee pain may not actually be arising from the knee and is a manifestation of pathology elsewhere.

This aim of this study is to compare the magnitude and nature of pain five years following cemented and cementless Oxford UKR, with the hypothesis being that there is no difference in pain. This will be done using specific pain scores, and the influence of co-morbidity and neuropathic pain will also be explored.

A longitudinal cohort study comparing 5-year outcomes of cemented and cementless medial mobile-bearing Oxford UKR implanted from 2006 to 2012 was conducted. All procedures were performed by four high-volume knee surgeons at two hospitals in the United Kingdom. During the period of this study, clinicians transitioned from use of the cemented Oxford UKR to the cementless Oxford UKR. The indications, pre-operative care and post-operative care were identical for both procedures. Surgical technique and instrumentation were identical [1] except for method of fixation—wider slots were produced to accept a cement mantle interface in cemented and narrower slots and holes were produced for interference fit in cementless. ‘Hybrid’ implants were not included. Patients were reviewed by independent research orthopaedic physiotherapists pre-operatively and 5 years post-operatively. Patients who were unable to be reviewed in clinic were contacted by post or telephone to obtain patient-reported outcomes. Revision rates were also tracked.

Patients were assessed five years post-operatively using two pain-specific instruments: the Intermittent and Constant Osteoarthritis Pain (ICOAP) instrument and painDETECT instrument. Pain sub-scores for the Oxford Knee Score (OKS) and the American Knee Society Score (AKSS), as well as Charnley classification, were analysed.

The knee-specific ICOAP score quantifies the magnitude and nature of knee pain [22]. ICOAP-A assesses constant pain and ICOAP-B assesses intermittent pain, i.e. pain that “comes and goes”. ICOAP-A is the total score from five questions, and ICOAP-B from six questions. Each question is equally weighted, and scores pain in different contexts from 0 (‘no pain’) to 4 (‘extreme’). ICOAP-Total is the sum of both A and B scores. ICOAP-A, -B, and -Total are scaled and scored between 0 and 100, with a higher score indicating worse pain. Patients scoring 0 on all scores were considered to have “no pain at all”.

The painDETECT instrument includes three visual analogue scales (PD-VAS) assessing ‘Now’, ‘Average’ and ‘Strongest’ pain experienced over the last four weeks and a separate ‘PD-Q’ score to assess likelihood of the pain being neuropathic. The PD-VAS are scored between 0 and 10 (increasing with magnitude of pain), and PD-Q is between 0 and 38, where PD-Q ≤ 12 is negative, 13 ≤ PD-Q ≤ 18 is unclear and PD-Q ≥ 19 is positive for a neuropathic pain component [6].

The OKS-Pain sub-score (0–20, higher is better) [9] and AKSS-Pain sub-score (non-continuous score of 0/10/20/30/40/45/50, higher is better) [11] were also analysed.

Patients were also classified into Charnley classification [5]: a 3-point classification of disease co-morbidity—A: single knee affected, B: both knees affected and C: multiple arthritis or medical infirmity.

Patients were recruited to the study and implanted with the cemented or cementless versions of the Phase 3 Oxford Partial Knee, in a non-blinded manner. Pre-operatively, patient demographics (date-of-birth, sex, height, weight), OKS-Pain and AKSS-Pain were compared between cohorts to ensure an even baseline of patients were recruited into both groups. Five-year post-operative ICOAP, painDETECT, OKS-Pain and AKSS-Pain measures were collected and analysed. Differences in magnitude and nature of pain between cemented and cementless cohorts were assessed, with the added impact of patients’ Charnley classification.

This study provides compelling evidence that patients experience low levels of pain following both cemented and cementless UKR. A large proportion of both cohorts reported no pain, and in cases with more serious pain, much of it did not arise from the knee. The amount of pain following cemented or cementless UKR was markedly less than that reported in the literature following TKR [2, 3, 12, 16]. At five years, the cementless UKR was found in all the scores we used to have significantly less pain than the cemented UKR despite the floor and ceiling effects resulting from the very low levels of pain (Fig. 2).

As mean pain scores were low for both cohorts, the distributions of the scores offer important insights. When compared to cemented UKR, patients with the cementless UKR were 45% more likely to have No pain at all (43% vs 61%). No cementless cases had Severe or Extreme pain, whereas 2.9% of the cemented did. As a result, most cases had No, Very Mild or Mild pain (cementless 98%, cemented 93%). This is corroborated by the revision rate of 0.76% in both cohorts, where there were no revisions for unexplained pain.

The different scores also give insight into differences in the nature of pain. Pain was twice as likely to be intermittent than constant in both cohorts (ICOAP-B > 0 47% vs ICOAP-A > 0 21%), a trend corroborated in literature [4, 15, 19, 26]. In general, patients reporting constant pain, also reported intermittent pain. Both pain scores were significantly, but proportionally, lower for the cementless cohort. The painDETECT VAS results followed the expected trend—‘Strongest’ pain was worse than the ‘Average’ which was worse than pain ‘Now’ (1.97 vs 1.25 vs 0.621, out of 10). The cementless cohort reported less pain across the 3 groups than the cemented. The difference was statistically significant for ‘Strongest’ and ‘Average’, but not for ‘Now’ Pain, which may be explained by the few patients reporting any pain ‘Now’, causing an especially strong floor effect (Fig. 4).

Across both cohorts, where there is pain, it is highly likely to be nociceptive (painDETECT PD-Q ≤ 12 in 87.8%), i.e. with a likely significant physiological origin. However, if the worst pain that patient experienced was Severe/Extreme (PD-VAS-Strongest ≥ 7/10) rather than Very Mild/Mild/Moderate (PD-VAS-Strongest ≥ 1–6), those patients were 3.5–3.8 times more likely to have a neuropathic element to their pain (Fig. 5). This is corroborated by the influence of Charnley grades on pain scores; despite all scores being knee-specific, patients with multiple arthritis or medical infirmity (Charnley C) consistently scored more poorly than those with only arthritis in the knees (Charnley A+B) (Fig. 6). More importantly, patients reporting higher levels of pain in ICOAP are markedly more likely to be Charnley C patients (Fig. 5). Interestingly, these differences were statistically significant in the cementless, but not cemented, cohort. This may be due to the greater “knee-origin” pain with the cemented cohort masking “other-origin” pain, while the cementless cohort experience weaker “knee-origin” pain, thereby allowing the relatively stronger influence of the “other-origin” pain to increase scores in Charnley C patients.

These correlations with neuropathy and non-knee morbidity suggest that pain perceived by patients to be knee-specific may not necessarily be caused by knee pathology, but instead be of external origin, be it biological, neuropathic, or psychosocial, as discussed in literature [10]. These external causes appear to be most likely in patients experiencing higher levels of pain, whilst “knee-origin” pain appears in most cases to be mild.

The origin of pain in UKR and the cause of the differences in pain between cemented and cementless implants remain unclear. The main difference between the implants is the bone–implant interface, which is likely to be responsible for the difference in pain. There are fewer radiolucent lines under cementless than cemented tibial components [14, 21, 23]. However, it has been shown that, following cemented fixation, there is no relationship between pain and radiolucency [7]. An alternative explanation relates to the stress within the tibial condyle, which increases appreciably following UKR and despite remodelling, this might remain elevated and contribute to pain [28]. The main reason for this is removal of the subchondral bone plate, which acts as a tension band supporting the medial condyle. If tension was transmitted between the wall and tibial eminence, this tension band may be at least partially restored and there should be less pain [28]. In this region, there are less radiolucencies with cementless than cemented components [25]. Further study is needed to understand why the pain occurs [18].

Scores from both cemented and cementless UKR cohorts are better than TKR scores in the literature. OKS-Pain in TKR at 5 years is reported to be 15.9 [27], which is poorer than cemented and cementless UKR scores in this study (17.0, 18.2 of 20). The differences are greater with the pain-specific score ICOAP and exceed minimally important clinical difference (MCID 18 of 100) [29]: UKR patients experience 8 times less constant pain (ICOAP-A 5.27 vs 42.3), 5 times less intermittent pain (ICOAP-B 10.8vs52.3), and 6 times less total pain (ICOAP-Total 8.06 vs 47.7) than TKR [4]. These studies are not matched and have different lengths of follow up. Nonetheless, TKR scores are consistent in literature [19, 29], do not change substantially after 6 months [10], and any changes that do occur would likely be small relative to the differences noted.

Score differences between cohorts in this study are not greater than published MCIDs: ICOAP-Total 18/100 [29], PD-VAS 0.9/10 [30]. However, these MCIDs were developed for TKRs, where scores are more normally distributed than the UKR scores encountered in this study. Due to strong floor/ceiling effects (Figs. 3, 5, 6), it is unfeasible to apply them to this study: for example, the cemented ICOAP-Total was 10.8, which cannot be decreased by 18 points (the MCID). MCID for OKS-Pain and AKSS-Pain were not found in literature.

The main limitations of the study are that it is not randomised and the cemented cohort was implanted predominantly before the cementless across the 6-year study period. Therefore, differences in pain observed could be related to improvements in surgical practice over time. However, when comparing the early and late sub-groups within the cemented and cementless cohorts, no significant differences are found. In contrast, when comparing the late-cemented and early-cementless sub-groups, which were implanted at approximately the same time, differences were fully consistent with the overall cemented and cementless groups. This demonstrates little, if any, effect of the date of implantation on outcome. In addition, the instrumentation and implantation procedure are largely identical. Hence, differences between the two cohorts are most likely due to differences in the implant itself, rather than other confounding causes. A further limitation is that the procedures in this study were performed by high-usage surgeons, which limits its generalisability. However, evidence suggests that if surgeons adhere to the recommended indications and surgical techniques, they get similar results [8]. Therefore, the conclusions of this study should relate to all surgeons using the recommended indications and techniques.

This study demonstrates that at 5 years, both cemented and cementless UKR have remarkably low pain levels compared to TKR scores reported in literature. The cementless implant had significantly less pain than the cemented, with most patients experiencing no pain at all and no patients experiencing more than moderate pain. The physiological origin of chronic pain experienced after UKR remains unclear but at least some of the pain may not originate from the knee.