Introduction

The prevalence of symptomatic osteoarthritis (OA) in Caucasian population is around 10% in men and 20% in women aged 45 years and above [1]. Total knee arthroplasty (TKA) represents the gold standard treatment for end-stage OA, being both functional and cost-effective procedure [2, 3]. However, OA affects in ca 30% of cases only one joint compartment, mostly the medial one [4, 5]. Unicompartmental knee arthroplasty (UKA) is an available option to treat selected patients. The indications to perform a UKA were first outlined in 1989 by Kozinn and Scott [6]: presence of unicompartmental OA and of an efficient anterior cruciate ligament, varus deformity < 5°, range of motion > 90° without flexion contracture, body mass index (BMI) < 30 kg/m2. Compared to the total knee arthroplasty, the UKA preserves the cruciate ligaments, meniscus and tibia plateau of the healthy compartment, thus sparing more bone stock for further revision. In the last decades, a significant amount of published studies compared the UKA and TKA for unicompartmental knee OA [7,8,9,10,11,12,13,14,15,16]. However, there is a lack of recent meta-analysis comparing the two implants. Since the number of performed UKAs is currently increasing [17] and the indications are obsolete [18, 19], it becomes of fundamental importance to update current evidences and outcomes. With these premises, a meta-analysis of clinical trials comparing UKA versus TKA was conducted. We focused on implant’s survivorship, clinical, functional and perioperative outcomes.

Materials and methods

Search strategy

The literature search was independently performed by two authors (FM, JE). This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [20]. Criteria for inclusion were the following:

  • Patient cohort: unicompartmental knee degeneration;

  • Intervention: unicompartmental knee arthroplasty (UKA);

  • Comparison: total knee arthroplasty (TKA);

  • Outcomes: clinical and functional scores, length of stay, ROM, total estimated blood loss, surgical duration, further revisions.

Data extraction

Data extraction was performed by two independent authors (FM, JE). For the initial database search, we accessed the Cochrane Systematic Reviews, Scopus, PubMed and Google Scholar, using the following keywords in isolation or combined: knee, total, unicondylar, partial, unicompartmental, osteoarthritis, replacement, prosthesis, implant, arthroplasty. The search was performed in October 2018. The authors independently read title and eventually the abstract of each paper, accessing the full text of the articles of interest. The bibliographies of the articles of interest were also screened.

Eligibility criteria

All the studies comparing UKA versus TKA for primary knee replacement were considered for inclusion. According to the Oxford Center of Evidence-based Medicine [21], level of evidence articles I to III were included. Biomechanics, animal and in vitro studies were excluded, along with case reports, reviews and meta-analysis, editorials, letters and expert opinions. According to the author language capabilities, studies in English, Italian, German, Spanish and French were considered. Due to the continuous innovations in components design and surgical progresses, the search was limited in a time frame from 2000 to 2018. Studies concerning bi- or tri-compartmental implants were excluded, along with cementless prosthesis. Studies providing navigation systems were included. Only studies reporting quantitative data concerning the outcomes of interest were included. Authors’ disagreements were debated and mutually solved.

Outcomes of interest

First, demographic data of each study were collected: mean age and BMI, percentage of female subjects, number of knees treated, mean follow-up. The primary outcome was to compare the revision’s rate between the implants. Secondary outcomes were: the Oxford Knee Score (OKS) [22], Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) and related subscale [23], the clinical and functional subscales of the Knee Society Rating System (KSS) [24] and the postoperative range of motion (ROM). Duration of surgery and total estimated blood loss were also considered.

Methodological quality assessment

Two independent authors (FM and JE) evaluated the methodological quality assessment using the Review Manager Software 5.3 (The Nordic Cochrane Centre, Copenhagen). For the evaluation, we referred to allocation concealment (selection bias), blinding of outcome assessments (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias) and other possible uncharged/unexplained risks.

Statistical analysis

For the statistical analysis, we referred to the Review Manager Software 5.3 (The Nordic Cochrane Centre, Copenhagen). Dichotomous data were analyzed through the Mantel–Haenszel statistical method with the odd ratio (OR) effect measure. For continuous data, the inverse variance statistical method was used with a mean difference effect measure. A confidence interval (CI) of 95% was considered for the analysis. We performed a forest plot for each outcome of interest. To analyze the heterogeneity, both Chi-square and Higgins (I2) tests were performed. Value of I2 of 25%, 50% and 75% represents, respectively, low, moderate and high grades of heterogeneity. An initial fixed analysis model to the comparisons was used. If high value of heterogeneity was detected, a random analysis model was adopted. To evaluate the risk of publication’s bias, the funnel plot was performed. To evaluate the component survivorship, the Kaplan–Meier curve was performed. Values of P < 0.5 were considered statistically significant.

Results

Search result

A total of 356 studies from database search and cross-references screening were obtained. Of these, 112 were removed because of duplicates. Other 175 articles were rejected because they did not match the topic. Other 66 studies were rejected because they did not report informations under the outcomes of interest. This left 23 articles for this study. The literature search is shown in Fig. 1.

Fig. 1
figure 1

PRISMA flow-chart of the literature search

Full size image

Methodological quality assessment

The methodological quality assessment reported a moderate risk of selection and detection bias, a low to moderate risk of reporting bias and a low level of attrition bias. Other not-specified biases have a low to moderate risk. The methodological quality assessment is negatively affected by the overall low level of evidence of the included studies. Concluding, the methodological quality assessment revealed a moderate quality. The methodological quality assessment is shown in Fig. 2.

Fig. 2
figure 2

Cochrane methodological quality assessment

Full size image

Risk of publication’s bias

To analyze the risk of publication’s bias, the funnel plot of the primary outcome of interest was performed (revision’s rate). All studies are detected in the range of acceptability and are positioned moderately symmetrically close to the no-effect line, indicating a low to moderate risk of publication’s bias. The funnel plot is shown in Fig. 3.

Fig. 3
figure 3

Funnel plot to assess the publication’s bias. SE standard error, OR odd ratio

Full size image

Patients’ demographics

A total of 13,789 patients were enrolled, undergoing a mean follow-up of 42.69 ± 24.76 months. To the UKA cohort belonged 3254 patients: The mean age was 66.75 ± 4.43 years, the mean BMI was 31 ± 2.75 kg/m2 and 57% of the patients were females. The TKA group included 10,649 patients, 61% of which were females, with mean age of 66.00 ± 3.06 years and mean BMI of 29.33 ± 2.87 kg/m2. The demographic baseline is shown in Table 1.

Table 1 Demographic baseline of the included studies
Full size table

Outcomes

A total of 16 studies reported revision’s rate, including 10,629 patients. The UKA reported a statistically significant higher risk compared to the TKA (OR 2.16; CI 1.66–2.81; I2 = 43%; P < 0.0001, Fig. 4).

Fig. 4
figure 4

Forest plot of the comparison revision. SD standard deviation, IV inverse variance, CI confidence interval

Full size image

All the analyzed scores are statistically significant in favor of the UKA: the Oxford Knee Score (EE 1.30; CI − 1.08, 3.67; I2: 85%; P = 0.28), KSS Clinical (EE 1.22; CI 0.25, 2.19; I2: 0%; P = 0.01), KSS Function (EE 5.40; CI − 3.14, 13.94; I2: 91%; P = 0.22), WOMAC overall (EE 5.22; CI − 0.57, 11.01; I2 75%; P = 0.08) and related subscales: pain (EE 5.77; CI − 2.87, 14.42; I2: 88%; P = 0.19), stiffness (EE 3.20; CI 2.79, 9.18; I2: 67%; P = 0.30), function (EE 6.02; CI 3.22, 8.82; I2: 24%; P < 0.0001). The UKA reported also statistically significant longer walking distance (EE 36.80; CI 13.40, 60.20; I2: 66%; P = 0.002), improvement of the joint flexion (EE 11.33; CI 7.92, 14.73; I2: 73%; P < 0.00001) and ROM (EE 6.42; CI 1.84, 11.00; I2: 72%; P = 0.006). Moreover, in the UKA group have been reported statistically significant shorter length of stay (EE − 1.88; CI − 2.84, − 0.92; I2: 98%; P = 0.0001), reduced estimated blood loss (EE − 70.66; CI − 142.27, 0.95; I2: 100%; P = 0.5) and shorter surgical duration (EE − 23.16; CI − 41.51, − 4.81; I2: 98%; P = 0.01). Table 2 summarizes the results of the comparisons.

Table 2 Overall results of the comparisons
Full size table

Discussion

The main findings of this meta-analysis are that UKA reported a reduced survivorship but better clinical and functional performances compared to TKA. Furthermore, even shorter surgical duration, lower estimated blood loss and quicker hospitalization length were observed in the UKA cohort. Current evidences demonstrated also a quicker return to high-level sports after UKA, which is of special interest in younger patients [37,38,39,40]. These results suggest that the UKA implants are more suitable for physically active patients, since they allow better clinical and functional performances. TKA are indicated in patients presenting comorbidities, cognitive impairment and reduced performance status, since a lower revision’s rate is preferable to higher functional performances.

The analysis of the comparison revision (Fig. 4) showed overlapping of the CI and heterogeneous statistical weights distribution. The level of heterogeneity expressed by the Chi-square and Higgins (I2) tests was acceptable, and a fixed analysis model was used. The overall result was statistically significant in favor of the UKA group, reporting a risk doubled (OR 2.16) compared to the TKA. This result is highlighted even by the Kaplan–Meier curve (Fig. 5), which clearly reported a reduction of the survivorship during the follow-up term. This result was coherent with other previous published studies [41,42,43]. Several studies analyzed the causes of revision: implant loosening, tibial collapse, worn-out of the inlay, mechanical failures, fractures and OA progression [44,45,46,47,48]. Since UKA implants are extremely sensible to the components alignment [49,50,51], it is easier to incur into technical errors [52]. A typical example is the overcorrection of the tibiofemoral angle, causing overload of the contralateral joint compartment [53]. The long learning curve characterizing the UKA implants contributes as well to improve the revision’s rate [54]. Furthermore, even the higher athletic performances observed after UKA implants can be related to the survivorship reduction.

Fig. 5
figure 5

Kaplan–Meier survivorship curve

Full size image

The Oxford Knee Score and WOMAC pain subscales were used to analyze the level of pain after surgery. The UKA evidenced statistically significant improvement in both the scores. To analyze the functional performances of the implants, we referred to the walking distance, flexion, ROM, KSFS and WOMAC (functional and stiffness subscales). The UKA scored better results in all the mentioned parameters, providing statistically significant results. We hypothesize that these functional improvements arise from the retention of the cruciate ligaments and the meniscus. These structures are involved in the joint kinematics, improving the flexion by promoting the rollback [55, 56]. Moreover, the cruciate ligaments function as mechanoreceptors [57], detecting body kinesthesia and proprioception [58]. Regarding the clinical scores, the KSCS and the overall WOMAC were both in favor of the UKA group, providing statistically significant results. Furthermore, a statistically significant shorter hospitalization, lower estimated blood loss and shorter surgical duration were evidenced. Overall, these results explain the high satisfaction rate observed in patients after the UKA implants [34, 59, 60].

The present study has some limitations. We included studies with low levels of evidence, since high-quality studies concerning this topic are very limited, increasing the risk of selection bias. Further studies providing samples randomization and longer follow-up are required to improve the evidences regarding UKA implants. Points of strengths of this work were the comprehensive nature of the literature search, the rigorously methodological quality assessment and the good baseline comparability of the samples.

Conclusion

The main findings of this meta-analysis are that UKA reported a reduced survivorship but better clinical and functional performances compared to TKA. Furthermore, shorter surgical duration, lower estimated blood loss and quicker hospitalization length were observed in the UKA cohort.