Introduction

Knee osteoarthritis (OA) is a frequent orthopedic disease [14]. Treatment includes both conservative and operative options that depend on the patient’s individual characteristics and disease stage. The disease is generally staged by a radiological classification, as detailed in Kellgren and Lawrence [51]. Today, total knee arthroplasty (TKA) is the method of choice for the treatment for symptomatic late-stage OA. Moderate-grade stages of the disease require an individualized approach. Both high tibial valgus osteotomy (HTO) and unicompartmental medial knee arthroplasty (UKA) are established methods for the treatment for medial moderate OA.

Valgus HTO was first described in 1960s [18, 22, 61, 86]. This method is used for medial unicompartmental knee OA. It aims for a mild valgus correction and a shift of the mean bearing axis into the non-affected lateral compartment.

Nearly in the same period, unicompartmental arthroplasty was established [39]. This method aims to replace only the damaged compartment and thus protects the unaffected joint compartments.

Valgus HTO is more appropriate for younger patients who accept a slight decrease in their physical activity. Medial UKA is appropriate for older patients obtaining sufficient pain relief but with reduced physical activity.

Consent on the best indication was unknown until now. Many studies and partially controversial results have been published. This is the first meta-analysis to compare the impacts of both methods on the treatment for unicompartmental medial knee OA.

Materials and methods

Retrieval of published studies and inclusion and exclusion criteria

Literature search was performed using the following established medical databases: MEDLINE (via PubMed), EMBASE (via OVID) and the Cochrane register. Retrieval in PubMed was conducted using the following combination: Knee [Mesh] AND Osteotomy [Mesh], Knee [Mesh] and Unicondylar [Mesh] OR Unicompartmental [Mesh] AND Knee.

Criteria for inclusion were as follows: English or German papers, a clinical trial with a clear description of survival, an outcome evaluation using a well-described knee score and a long-term follow-up (minimum 5 years).

In the first step, two investigators (the first and senior authors) independently reviewed the titles and abstracts of all of the retrieved studies. Second, both investigators hold a conference and found a consensus about which studies were to be evaluated with a full-text review. Then, a full-text review was performed independently by the two lead investigators. After determining the results of the second review, the studies were indentified for their final inclusion into the meta-analysis. Data extraction was performed by both investigators. Results were entered into an Excel table. The established and well-described scores were normalized to a 0-to-100 scale. Before the statistical evaluation, both investigators fine-tuned the results once again.

Statistical analysis

Statistical analysis was performed using the special meta-analysis software called “Comprehensive Meta Analysis” (version 2.0; Biostat, Englewood, NJ, USA).

All of the effect sizes were calculated using a random-effects model. Dichotomous effect sizes (rate of survival, rate of complications and a semi-quantitative estimate of the results) were expressed as ERs (event rates). Continuous values (estimates determined from the knee scores) were calculated by comparing their means (t test). Scores (baseline to follow-up) were compared by calculating the standard difference of the means (SDM). All of the results were presented as forest plots. Results of the scores were normalized to a 0-to-100 scale. In other words, results were calculated as [x = (points) × 100/(maximum possible score points)].

A 95% confidence interval was given for each effect size.

Heterogeneity was calculated according to the method of Higgins et al. [44]. Heterogeneity is expressed as I 2. This value ranges from 0% (complete consistency) to 100% (complete inconsistency).

Results

Included studies

The deadline for evaluation in the databases was July 31, 2010. Primary search resulted in 1,400 studies about valgus HTO and 895 studies about medial UKA.

Final meta-analysis after the full-text review included 46 studies about valgus HTO and 43 studies about medial UKA. Flowcharts describing the study selection are in Figs. 1, 2. Mean age of patients at operation who had undergone valgus HTO was 56.1 years (95% CI, 53.2–58.7 years). Medial UKA patients were significantly (P < 0.001) older. Mean age at operation was 70.1 years (95% CI, 67.3–72.7 years).

Fig. 1
figure 1

Flowchart for the identification of valgus HTO studies

Full size image
Fig. 2
figure 2

Flowchart for the identification of medial UKA studies

Full size image

Comparative studies

Only one randomized prospective study was identified. Borjesson et al. [12] reported the 5-year results of 18 valgus HTO patients (closed-wedge technique) versus 22 medial UKAs. During the follow-up, no significant difference in the subjective outcome (BOA Score) was found.

Broughton et al. [13] published the 10-year results of a retrospective comparative study. They found a significant better outcome after a medial UKA. Survival was 0.711 for valgus HTO (n = 49) and 0.913 for medial UKA (n = 42). In HSS, 76.1% of the medial UKAs had good or excellent outcomes. This rate was 42.8% in the valgus HTO group.

Stukenborg-Colsman et al. [94] also retrospectively assessed the 7.5-year outcome after 32 valgus HTOs and 28 medial UKAs. The 5- to 10-year survival after a valgus HTO was 0.688 and 0.800 after a medial UKA. In KSS, 71% of the valgus HTO patients and 65% of the medial UKA patients had an excellent outcome.

Survival

Survival was defined as the time to a total arthroplasty revision. ERs for survival were grouped as follows: 5–8 [mean] years, 9–12 years and more than 12 years of follow-up.

At 5–8 years, 91.0% of the valgus HTO patients and 91.5% of the medial UKA patients did not require a TKA revision (Fig. 3). Within 9–12 years after the operation, 84.4% of the valgus HTO patients and 86.9% of the medial UKA patients did not require revision (Fig. 4). After more than 12 years, valgus HTO tended to be revised more frequently (n.s). Survival to revision was 70.1% in the valgus HTO patients and 77.5% in the medial UKA patients (Fig. 5).

Fig. 3
figure 3

Survival to endpoint total knee arthroplasty after 5–8 years of follow-up. The forest plots present the effect size (ER event rate). Each square represents the individual study’s ER with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 30; medial UKA, n = 26. Survival: valgus HTO, 0.910; medial UKA, 0.915. Heterogeneity (I 2): HTO = 71.1, UKA = 82.3. Significance: P = 0.801

Full size image
Fig. 4
figure 4

Survival to endpoint total knee arthroplasty after 9–12 years of follow-up. The forest plots present the effect size (ER event rate). Each square represents the individual study’s ER with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 28; medial UKA, n = 25. Survival: valgus HTO, 0.844; medial UKA, 0.869. Heterogeneity (I 2): HTO = 89.5, UKA = 92.0. Significance: P = 0.458

Full size image
Fig. 5
figure 5

Survival to endpoint total knee arthroplasty after more than 12 years of follow-up. The forest plots present the effect size (ER event rate). Each square represents the individual study’s ER with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 15; medial UKA, n = 9. Survival: valgus HTO, 0.701; medial UKA, 0.775. Heterogeneity (I 2): valgus HTO = 93.7, medial UKA = 95.9, Significance: P = 0.451

Full size image

In 17 studies, the survival time to revision was calculated according to the method of Kaplan–Meier (Figs. 6, 7). Time of the required revision tended to be sooner in the medial UKA patients than in the valgus HTO patients (n.s). Medial UKA patients were revised after a mean of 8.2 years, while the valgus HTO patients had undergone a revision 1 year later at a mean of 9.7 years (Tables 1, 2).

Fig. 6
figure 6

Mean survival (Kaplan–Maier). The forest plots present the mean survival of each study. Each square represents the individual study’s mean survival with a 95% CI indicated by the horizontal lines. Number of included studies: HTO, n = 12; UKA, n = 5. Mean survival: valgus HTO, 9.7 years; medial UKA, 8.2 years. Heterogeneity (I 2): valgus HTO = 96.0, medial UKA 92.1. Significance: P = 0.374

Full size image
Fig. 7
figure 7

Mean survival (Kaplan–Meier) funnel plot

Full size image
Table 1 Studies included in the meta-analysis (valgus HTO)
Full size table
Table 2 Studies included in the meta-analysis (medial UKA)
Full size table

Outcome measurement using established knee scores (continuous values)

The papers that were included in this investigation are listed in Table 3. For outcome evaluation, following different scores were used.

Table 3 Outcome measurements grouped by established knee score
Full size table
Table 4 Outcome estimates grouped by established knee score
Full size table

BOA (British Orthopaedic Association; n = 2) [1]; Bristol or synonymous Baily score (n = 1) [58]; Conventry score (n = 3) [19]; HSS (Hospital for Special Surgery score; n = 30) [77]; JOA (Japanese Orthopaedic Association; n = 8) [82]; KOOS (Knee Injury And Surgery Outcome score; n = 1) [81]; KSS (Knee Society score) or synonymous Insall score (n = 21) [47]; and the Lysholm score (n = 1) [57]. Each study used a self-created 100-point score. This score was well described and similar to the other scores.

Dates at baseline (Fig. 8) were extracted from 19 papers (for valgus HTO group, the mean score was 60.1 and medial UKA group 55.9; n.s.).

Fig. 8
figure 8

Mean normalized knee scores at baseline. The forest plots present the mean score of each study. Each square represents the individual study’s mean score with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 14; medial UKA, n = 5. Mean: valgus HTO, 60.1; medial UKA, 55.9. Heterogeneity (I 2): valgus HTO = 98.5, medial UKA = 99.6, Significance: P = 0.496

Full size image

There was a significant difference between valgus HTO and medial UKA in the 5- to 8-year follow-up (P < 0.001). Mean score was 83.4 in the valgus HTO patients and 91.2 in the medial UKA patients (Fig. 9).

Fig. 9
figure 9

Mean normalized knee score after 5–8 years of follow-up. The forest plots present the mean score of each study. Each square represents the individual study’s mean score with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 7; medial UKA, n = 5. Mean: valgus HTO, 83.4; medial UKA, 91.2. Heterogeneity (I 2): valgus HTO = 94.2, medial UKA = 96.6. Significance: P < 0.001

Full size image

Only one study reported 10-year results after medial UKA. In this study, mean score was 90.0. Mean score in the 9 valgus HTO papers was 79.9. Results from Price et al. [75, 76] were significantly better (P < 0.001) than results from the valgus HTO papers (Fig. 10).

Fig. 10
figure 10

Mean normalized knee score after 9–12 years of follow-up. The forest plots present the mean score of each study. Each square represents the individual study’s mean score with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 9; medial UKA, n = 1. Mean: valgus HTO, 79.9; medial UKA, 90.0. Heterogeneity (I 2): valgus HTO = 85.9; medial UKA is not adjustable. Significance: P < 0.001

Full size image

A total of 5 papers (2 valgus HTO and 3 medial UKA) reported results beyond 12 years. The outcome of medial UKA (mean 65.6) tended to be better than the outcome of valgus HTO (mean, 58.8; n.s). Results are shown in Fig. 11.

Fig. 11
figure 11

Mean normalized knee score after more than 12 years of follow-up. The forest plots present the mean score of each study. Each square represents the individual study’s mean score with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 2; medial UKA, n = 3. Mean valgus HTO, 58.8; medial UKA, 65.6. Heterogeneity (I 2): valgus HTO = 89.2, medial UKA = 92.3. Significance: P = 0.331

Full size image

Whenever possible, the baseline score was compared to the score at the 5- to 8-year follow-up, 9- to 12-year follow-up or the >12-year follow-up. SDM after valgus HTO tended to increase more in the valgus HTO studies (SDM = 5.0) compared with the medial UKA studies (4.1) after a 5- to 8-year follow-up (Fig. 12; n.s.).

Fig. 12
figure 12

Weighted (standardized) mean (SDM) of the score from baseline to 5–8 years of follow-up. The forest plots present the SDMs of each study. Each square represents the individual study’s SDM with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 7; medial UKA, n = 3. SDM: valgus HTO, 5.0; medial UKA, 4.1. Heterogeneity (I 2): valgus HTO = 97.5; medial UKA = 60.1. Significance: P = 0.359

Full size image

Only the study by Price et al. [76] was available to compare the valgus HTO results (n = 8 studies) between baseline and the 9- to 12-year follow-up. In this study, the 9- to 12-year SDM was 10.7, whereas the SDM of the 8 valgus HTO studies was only 1.7 (Fig. 13). This difference was significant (P < 0.001).

Fig. 13
figure 13

Weighted (standardized) mean (SDM) of the score from baseline to 9–12 years of follow-up. The forest plots present the SDMs of each study. Each square represents the individual study’s SDM with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO: n = 8; medial UKA: n = 1. SDM: valgus HTO, 1.7; medial UKA, 10.7. Heterogeneity (I 2): valgus HTO = 95.4. Heterogeneity (I 2) was not adjustable. Significance: P < 0.001

Full size image

Only 2 studies with very long-term results (>12-year follow-up) were identified. Mercier et al. [64] reported an SDM of approximately 1.2 after valgus HTO. Sprenger and Doerzbacher [88] found an SDM of −0.2 after 12 years. The difference between these studies was not significant (n.s; Fig. 14).

Fig. 14
figure 14

Weighted (standardized) mean (SDM) of the score from baseline to more than 12 years of follow-up. The forest plots present the SDMs of each study. Each square represents the individual study’s SDM with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 1; medial UKA, n = 1. SDM: valgus HTO, −0.2; medial UKA, 1.2. Heterogeneity was not adjustable. Significance: P = 0.603

Full size image

Outcome measurement using established knee scores (semi-quantitative estimation)

In some papers, the result was estimated by establishing knee scores, and the results were expressed as semi-quantitative estimates. No significant differences between valgus HTO and medial UKA were observed (Table 3, Fig. 15).

Fig. 15
figure 15

Semi-quantitative estimation of the clinical outcome based on the evaluation using clinical scores. The detailed ESs are given in Table 4

Full size image

Complications

In general, there were more complications after a valgus HTO (13.8%) than after a medial UKA (11.3%; n.s). Forest plots are shown in Fig. 16.

Fig. 16
figure 16

Complication rates. The forest plots present the effect sizes (ER event rate). Each square represents the individual study’s ER with a 95% CI indicated by the horizontal lines. Number of included studies: valgus HTO, n = 31; medial UKA, n = 13. Complications: valgus HTO, 0.138; medial UKA, 0.113. Heterogeneity (I 2): valgus HTO = 82.0, medial UKA = 70.2. Significance: P = 0.369

Full size image

Discussion

The most important finding of this study was both valgus HTO and medial UKA are sufficient operative treatment options for symptomatic medial knee osteoarthritis.

This meta-analysis was undertaken to compare the impacts of valgus HTO and medial UKA on the treatment for symptomatic unicondylar medial knee osteoarthritis. Survival rates to total knee replacement were not different. Over time, both groups exhibited increased revision rates. Clinical outcome of medial UKA is better than that of valgus HTO. After more than 12 postoperative years, patients in both groups exhibited worsening clinical outcomes.

Both methods are clinically very interesting. This is reflected by the large number of published studies on this topic. In this meta-analysis, we excluded studies with follow-ups >5 years and studies that did not enable the evaluation of concrete effect sizes. Furthermore, clear duplicates were excluded from this evaluation. This was also considered for the results of the central arthroplasty registers [28, 53, 55, 60]. An important limitation is the large heterogeneity of the studies (range, 70–95%). Only 3 comparative [12, 13, 94] studies were found on this topic. The survival and outcome results of these studies are contradictory.

Both valgus HTO and medial UKA are, in principle, sufficient for the treatment for medial knee osteoarthritis.

A valgus HTO corrects varus malalignment by transferring the load to the relatively unaffected lateral compartment. This can be accompanied by pain relief and improvements in gait and function. Mechanical release alone or a combination of HTO with arthroscopic measures (debridement, synovectomy or microfracture), chondral resurfacing or meniscal transplantation can also improve these results [25, 83]. The main benefit for the patient is the preservation of the natural joint. The main advantage for the patient is that potential physical loading (professional or sports-related) is almost entirely unaffected. The limits of valgus HTO are failure of medial compartment restoration, progression of medial degeneration, progression of degeneration in the patellofemoral or lateral compartment and development of pathological valgus. However, in most patients, this method is generally sufficient for avoiding a TKA for about 10 years. Moreover, a revision to a TKA generally does not cause problems [33, 97].

With the proviso that very excellent results are seldom achieved, the valgus HTO can be considered to be the method of choice for younger and physically active patients who are suffering from unicompartmental knee osteoarthritis.

Unlike medial UKA, this treatment addresses the reconstruction of the joint surface by replacing the degenerated joint part and preserving the unaffected joint. A correction of axial misalignment is seldom possible. The main limitation of medial UKA is partially similar to that of valgus HTO (progression of osteoarthritis). The method is comparable to valgus HTO but with a slightly lower complication rate and a shorter rehabilitation time. This outcome is associated with a better 12-year knee score outcome. However, this method also has important disadvantages. Medial UKA requires restricted physical activity. This may be a reason for the higher mean age of the patients who undergo medial UKA. If needed, a revision from medial UKA to total arthroplasty causes more problems and worsens the worse result [7].

Conclusions

This meta-analysis examined the advantages and disadvantages of two established methods for treatment for medial compartment knee osteoarthritis. Valgus HTO is more appropriate for younger patients who accept slight decreases in physical activity in their lifestyles. Medial UKA is a good method for older patients who need painless but reduced physical activity.