Patellar resurfacing versus nonresurfacing in total knee arthroplasty: A meta-analysis of randomised controlled trials

Abstract

Purpose

Patella resurfacing or nonresurfacing in total knee arthroplasty remains controversial. The aim of this study was to evaluate the efficacy of patellar resurfacing through an evaluation of the current literature.

Methods

We carried out a meta-analysis of randomised controlled trials comparing total knee arthroplasties performed with and without patellar resurfacing. Outcomes of reoperation, anterior knee pain and knee scores were analysed.

Results

Fourteen trials assessing 1,725 knees were eligible. The absolute risk of reoperation was reduced by 4 % (95 % confidence interval, 2–6 %) in the patellar resurfacing arm (between-study heterogeneity, P = 0.05, I² = 42 %), implying that one would have to resurface 25 patellae (95 % confidence interval, 17–50 patellae) in order to prevent one reoperation. There was no difference between the two groups in terms of anterior knee pain, knee pain score, Knee Society score and knee function score. But in the studies followed up for a mean time of not less than five years, a difference was found between the two arms in Knee Society scores (RR = 2.14, 95 % confidence interval, 0.76–3.52; P = 0.002).

Conclusions

The available evidence indicates that patellar resurfacing reduces the risk of reoperation after total knee arthroplasty. Patellar resurfacing patients may make a difference in long-term follow-up (five or more 5 years) of Knee Society scores. In other aspects, the benefit of patellar resurfacing is limited. Additionally, more carefully and scientifically designed randomised controlled trials are required to further prove the claim.

Introduction

Management of the patella in total knee arthroplasty (TKA) remains controversial. Some surgeons prefer resurfacing because of the increasing rate of secondary patellar revision or other reoperations and anterior knee pain in patients without patellar resurfacing. Other surgeons do not resurface the patella during TKA to avoid complications such as fracture, loosening, instability and patella tendon damage [1]. There are various studies about the patellar resurfacing puzzle, including randomised controlled trials (RCTs), retrospective studies, meta-analysis and systematic review, but no final conclusion has been drawn. Since the latest meta-analysis of patellar resurfacing was done in 2010 and published in 2011 [2, 3], and many RCTs have been conducted since then, it is necessary to engage a new analysis containing these latest RCTs. We therefore conducted a meta-analysis of all available RCTs comparing TKAs with and without patellar resurfacing to evaluate the efficacy of this aspect of the procedure.

Materials and methods

Identification and eligibility of relevant RCTs

We searched MEDLINE (1966 to present), EMBASE and the Cochrane databases for randomised trials comparing total knee arthroplasties performed with and without patella resurfacing. Inclusion criteria were [3]: (1) English literature, (2) adults undergoing primary TKA for osteoarthritis, (3) all available prospective RCTs comparing patellar resurfacing to nonresurfacing in TKA, (4) patellar resurfacing not at the same time as primary TKA, (5) data for reoperation, anterior knee pain and knee scores (Knee Society score, knee pain score, knee function score). Exclusion criteria were: non-English language literature, unpublished data, proceedings of meetings, non-randomised controlled trials, studies of body specimens and studies of TKA but not patellar resurfacing. To avoid double counting, multiple articles of the same patient population were pooled as one study.

Outcomes

The number of reoperations, cases suffering postoperative anterior knee pain, symptom and function scores such as the Hospital for Special Surgery score (HSS) [4], the WOMAC score [5], the Bristol knee score [6] and the Knee Society score (KSS) [7] were sorted successively for further analysis.

Data extraction

The relevant data were extracted from every article by two of the authors independently and were checked by a third researcher against the original information in order to avoid anthropic mistakes. The extracted data included publication information, numbers of reoperations for any reason, follow-up time and numbers lost to follow up, numbers of patients who suffered any grade of postoperative anterior knee pain, and the relevant knee scores.

Assessment of methodological quality

Two of the authors independently assessed the methodological quality of each article with respect to random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data and selective reporting by the computer software developed by the Cochrane Collaboration, RevMan (RevMan 5.2).

Meta-analysis

Data was analysed by the software RevMan (RevMan 5.2). The fixed model and the random-effects model were used to analyse the pooled data. A P value of < 0.1 was considered suggestive of statistical heterogeneity. In case of significant heterogeneity, the random-effects model was used. Referring to the study of Fu et al. [3], 95 % confidence limits and risk differences were calculated for reoperation and the presence of postoperative anterior knee pain (AKP) of any grade, weighted mean differences and 95 % confidence limits calculated for the KSS knee score of relative studies and the studies with a mean length of follow-up not less than five years. The number of knees needed to be treated in order to avoid reoperations or postoperative AKP was also calculated (1/|risk difference|).

Results

After detailed evaluation, 14 independent randomised trials [6, 8–20] with a cumulative sample size of 1,725 knees at final follow-up were included in the overall meta-analysis (Table 1). The methodological quality is shown in Fig. 1 to see the risk of bias. All 14 studies provided data on reoperations during follow-up. Ten provided particular data of the prevalence of postoperative anterior knee pain of any grade [6, 9, 11, 12, 14–18, 20]. Nine studies reported Knee Society score [8, 10–12, 14–16, 18, 20], and five of them had a mean length of follow-up of five or less  years [12, 14, 15, 18, 20]. Seven studies provided details on the knee function score [10–12, 14, 16, 18, 20], and four of them had a length of follow-up five or more years [12, 14, 18, 20]. Only four trials provided details on knee pain scores at the last follow-up [14, 16, 19, 20].

Table 1.

Characteristics of studies

First author	Journal	Year of publication	Number of knees (RS / NRS)		Mean age(year)	Loss to follow-up (knees)	Mean years of follow-up	Outcomes	Level of evidence	Treatment	Number of reoperations	Main reasons for reoperations
Feller [8]	JBJS	1996	19	19	70.8	2	2	HSS, patellar score, patellar pain	I	PCA (HOWMEDICAL)	RS:2/NRS:0	Suprapatellar adhesions, posterolateral pain
Barrack [9, 21]	JBJS	1997, 2001	58	60	66.2	25	5-7	KSS, AKP, patient satisfaction, ROM	I	Miller-Galante II (ZIMMER)	RS:0/NRS:7	AKP
Schroeder [10]	Arch Orthop Trauma	1998	20	20	72.6	0	2	KSS, AKP, climbing staris, radiographs	I	DURACON (HOWMEDICA)	RS:1/NRS:3	AKP, lateral subluxation of patella
Newman [6]	KNEE	2000	42	42	72	0	5	Bristol score, patellar score	I	KINEMATIC (HOWMEDICA)	RS:0/NRS:6	AKP
Wood [11]	JBJS	2002	92	128	73.7	22	4	KSS, AKP, stair climbing, radiographs, patient satisfaction	I	Miller-Galante II (ZIMMER)	RS:9/NRS:15	AKP, maltracking, patella dislocation/loosening/fracture, infection, loose bodies
Waters [12]	JBJS	2003	243	231	69	0	5.3	KSS, AKP, patient satisfaction, function score, ROM, radiographs	I	Press-Fit Condylar prosthesis (Johnson & Johnson)	RS:3/NRS:11	AKP
Mayman [13]	J Arthroplasty	2003	50	50	70	29	10	KSS, AKP	I	AMK (Depuy)	RS:2/NRS:2	Pain, patella fracture and other patellofemoral complications
Burnett [14]	CORR	2004	42	48	78	0	10	KSS, AKP, ROM, patient satisfaction, radiographs	I	AMK (Depuy)	RS:1/NRS:3	AKP, sepsis, tibial base plate PE wear, osteolysis, patella necrosis, quadriceps tendon rupture
Campbell [15]	JBJS	2006	46	54	72	42	10	WOMAC scores, AKP, radiographs	I	Miller-Galante II (ZIMMER)	RS:1/NRS:2	AKP
Smith [16]	JBJS	2008	73	86	71.5	2	4	KSS, AKP, patient satisfaction, radiographs	I	Profix Total Knee System (Smith and Nephew)	RS:1/NRS:1	PF nodule, locked knee, infection, tibial insert exchange
Burnett [17]	CORR	2007	28	28	NR	0	10	KSS, AKP, patient satisfaction	I	Miller-Galante II (ZIMMER)	RS:1/NRS:2	AKP, PF-related aseptic loosening, sepsis, tibial liner exchange
Burnett [18]	JBJS	2009	58	60	NR	40	10	ROM, KSS, AKP, patient satisfaction	II	Miller-Galante II (ZIMMER)	RS:5/NRS:7	AKP
Lauren Beaupre [19]	BMC Research Notes	2012	21	17	63,6	6	10	WOMAC, ROM, RAND 36 score	I	Profix Total Knee System (Smith and Nephew)	RS:1/NRS:3	AKP
Zhong-tang Liu [20]	KNEE	2012	68	64	67.7	12	7	AKP, KSS, ROM, radiographs	I	Press-Fit Condylar prosthesis (DePuy)	RS:0/NRS:1	Infection

RS resurfacing, NRS nonresurfacing, AKP anterior knee pain, HSS hospital for special surgery, ROM range of motion, WOMAC Western Ontario McMaster Osteoarthritis Index, PF patella-femoral, PE polyethylene

Fig. 1.

Risk of bias

The relative risk for reoperation favoured patellar resurfacing (RR = 0.50, 95 % CI, 0.33-0.76; P = 0.001 < 0.05) (Fig. 2). The test for homogeneity was not significant, implying that there was no heterogeneity between these 14 studies (P = 0.60). The absolute risk of reoperation was reduced by 4 % (95 % CI, 2–6 %) in the patellar resurfacing group compared with the nonsurfacing arm (Fig. 3). A funnel plot of the reoperation data of the 14 articles showed low publication bias (Fig. 4). The result implied one would have to resurface 25 patellae (95 % CI, 17–50 patellae) in order to prevent one reoperation.

Fig. 2.

Reoperation

Fig. 3.

Reoperation–risk difference

Fig. 4.

Funnel plot

The relative risk for postoperative anterior knee pain of the ten studies was 0.83 (95 % CI, 0.50–1.37; P = 0.46). The difference was not significant (Figs. 5 and 6). The studies were heterogeneous since the test for homogeneity was significant (P < 0.00001). So a random effects model was used. In our point, the heterogeneity might have a relationship with different length of follow-up or different kinds of implants. But no major differences in the studies could explain the observed heterogeneity convectively.

Fig. 5.

Postoperative anterior knee pain

Fig. 6.

Postoperative anterior knee pain–risk difference

The nine studies [8, 10–12, 14–16, 18, 20] which reported the details of the Knee Society score showed a significant heterogeneity (P = 0.02) (Fig. 7); so we could draw no conclusion from the data. In addition, we analysed Knee Society scores of studies with a long-term follow-up (five or more years) individually. Five studies were recorded [12, 14, 15, 18, 20], and the test for homogeneity was not significant (P = 0.43). A significant difference was found between the two groups (RR = 2.14, 95 % CI, 0.76–3.52; P = 0.002) (Fig. 8). This might imply that during the long-term follow-up (five or more years), patellar resurfacing patients may get a higher Knee Society score than the nonresurfacing patients.

Fig. 7.

Knee score

Fig. 8.

Knee score (≥5 years)

Knee function score and knee pain score are part of the Knee Society score. In the seven studies which provided details on the knee function score [10–12, 14, 16, 18, 20], there was no heterogeneity (P = 0.23), and no difference was shown between the two groups (P = 0.14) (Fig. 9). The same conclusion was drawn in the four long-term follow-up studies [12, 14, 18, 20] (no heterogeneity: P = 0.73; no significant difference: P = 0.27) (Fig. 10) and only four studies provided knee pain scores [14, 16, 19, 20] (no heterogeneity, P = 0.96; no significant difference, P = 0.99) (Fig. 11).

Fig. 9.

Function score

Fig. 10.

Function score (≥5 years)

Fig. 11.

Pain score

Discussion

By this meta-analysis of 1,725 randomised trials, we found that the rate of reoperation was lower following total knee arthroplasty with patellar resurfacing than after total knee arthroplasty without patellar resurfacing. But there was no difference between the two groups in terms of anterior knee pain. Interestingly, we found that during the long-term follow-up (five or more years), the patellar resurfacing group might achieve a higher Knee Society score than the nonresurfacing group. This new conclusion is different from many of the past systematic studies. More high quality and well-designed RCTs with long-term follow-up are needed to support this conclusion. The absolute risk difference for reoperation was the same as the study of Fu et al. [3]. It implies that one would have to resurface 25 patellae in order to prevent one reoperation. The outcome of reoperation was not assessed in this study. In the study of Fu et al. [3], they found that 76 % of the patients who suffered postoperative patella-femoral-related pain benefited from the secondary resurfacing.

When searching for the literature for this meta-analysis, we found many meta-analysis reports or systemic studies of the same theme, as well as prospective studies and some prospective studies that did not mention randomised controlled trials. The results of these studies were sometimes inconsistent. Pakos et al. [22] and Calvisi et al. [23] found that patellar resurfacing might reduce the risks of reoperation and anterior knee pain after AKP and Patel’s cohort study supported this conclusion [24]. But in the studies of Fu et al. [3] and He et al. [2], no difference between the resurfacing group and the nonresurfacing group in terms of anterior knee pain rate was found. They only found significance of reoperation risks between the two groups. And the systematic review of Li et al. had a result similar to the previous two studies [25]. The study of Lindstrand et al. [26] showed that there was no statistical difference in the number of revisions due to patella problems between the two groups. A prospective study from Patil et al. [27] concluded that there were no differences in the improvement of KSS scores, Short-Form 36 scores and satisfaction rates among the patellar resurfacing group, the patellar retention group and the patelloplasty group. A retrospective study conducted by Seo et al. [28] showed that there was no correlation between the degree of patellar articular defect and patellar resurfacing in terms of the clinical and radiological outcomes. And the latest study of Pavlou et al. [29] provided further findings that neither patellar resurfacing nor the prosthetic design affects the clinical results of TKAs. Studies of LCS mobile-bearing total knee arthroplasty [30, 31] showed the importance of soft tissue, especially the ligaments, in the decision for patellar resurfacing or not. Hwang et al. [31] concluded that if soft tissue balancing and a patella-friendly prosthetic design are properly used, patellar retention with a patelloplasty might be viable, even in knees with serious patellofemoral arthritis. Since there have been no final conclusions about resurfacing of patella or not during TKA, more studies are needed to be conducted in the future.

The strengths of this meta-analysis include: (1) this is an update of the present meta-analysis; (2) the design of search strategy, and the inclusion and exclusion criteria are based on the experience of the present meta-analysis; (3) explicit research protocol was developed before the analysis; (4) consensus between two reviewers and checked by a third reviewer to avoid anthropic mistakes; and (5) new finds about long-term follow-up.

Some limitations of our study should be acknowledged: (1) the sample sizes are small; larger and more rigorous prospective randomised controlled studies are needed to strengthen the findings; (2) not all studies provide the details of prevalence of anterior knee pain and knee scores, which may cause bias; (3) the skills of different surgeons may influence the clinical outcomes; we need a standard to assess the surgery skills in order to avoid surgeon bias; and (4) implant designs may affect the outcomes, though a study from Pavlou et al. [29] shows that the prosthetic design may not affect the clinical results of TKAs. There are various prosthetic features in the 14 studies included. This may affect the result of this meta-analysis.

Conclusion

This meta-analysis shows that patellar resurfacing may reduce the risks of reoperation after total knee arthroplasty. It may make a difference in long-term follow-up (five or more years) of Knee Society scores. In other aspects, the benefit of patellar resurfacing is limited. Additionally, more carefully and scientifically designed RCTs with large samples and long-term follow-up are required to further prove the claim. Based on the evidence provided by this study, the authors now prefer not resurfacing the patella during primary TKA operations. Patients with and without patella resurfacing are being followed up carefully in terms of reoperations, anterior knee pain, Knee Society scores and radiographs to assess the long-term outcomes.