Introduction

The ultimate goal of total knee replacement (TKR) is to relieve pain and to improve the functional outcome. Patient expectations play an important role in achieving this goal [3]. Residual pain following TKR is an important reason for patients’ dissatisfaction [9, 52]. Several factors that are associated with residual pain have been identified, both extra- and intra-articular, but often no cause can be found [53]. Moreover, female sex, young age at time of surgery, and a higher than normal depressive or anxiety state are associated with a more painful knee [9, 10].

Anterior knee pain (AKP) is reported to occur in up to one half of all patients following primary TKR. The presence of AKP after TKR is negatively correlated with patient satisfaction and quality of life [4, 11, 44]. With the increasing number of TKRs and the particular importance of patient expectations, addressing the problem of AKP after TKR is of special significance. Although several patient and technique related factors involved in AKP have been identified [5, 11, 16, 46, 54], no systematic evaluation and qualitative assessment of these data has been performed.

The purpose of this systematic review is therefore to identify and assess the different modifiable and non-modifiable determinants that may be associated with the development of AKP in patients following primary TKR.

Materials and methods

Identification of studies

An independent librarian performed the literature search using the following search terms with Boolean operators: anterior knee pain or patellofemoral pain or retropatellar pain and knee arthroplasty or knee arthroplasties or knee replacement or knee replacements or knee prosthesis or knee prostheses or TKA or TKR or TKP. Search queries were limited to title/abstract, and language was restricted to English, French, or German. The electronic search involved the Cochrane Database of Systematic Reviews, the Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE, and Google Scholar, and included articles published until January 2012.

Assessment of study eligibility

Only published full-text peer-reviewed studies of AKP following primary TKR without regard for the underlying pathology were included if the prevalence of AKP was related to distinctive patient and technique related variables. Studies not reporting the prevalence of AKP were excluded, as were publications with incompletely described patient populations, less than 10 included patients, less than 6 months of follow-up, insufficient descriptions of treatment, and studies that failed to describe the method used to assess the prevalence of AKP. Two reviewers (HPWvJ and JMR) independently examined all titles and abstracts and selected the studies for full-text review. Where there were discrepancies in the studies included, the authors were able to reach a consensus. Additionally, the reference lists in the included studies were hand-searched for other relevant studies. The full texts were retrieved and further checked for inclusion and exclusion criteria. If articles described the same series of patients, both studies were included to evaluate for possible time-dependent differences in the prevalence of AKP.

Data abstraction

For all studies selected for full-text review, relevant data were abstracted from the text, figures, and tables using a structured data abstraction form. Data extraction was undertaken by one author (HPWvJ) and validated by a second (JMR). The elements abstracted included type of study, number of patients, details on intervention, outcome measures, follow-up, variables studied in relation to AKP, and the prevalence of AKP including the method used to assess the prevalence. When the reported data were incomplete, corresponding authors were contacted by email for additional data.

Assessment of methodological quality

Two authors (HPWvJ and JMR) assessed the quality of the included studies independently using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach [21, 49]. A recommendation for or against the use of an intervention was given for the modifiable determinants. For the non-modifiable determinants, the strength of the correlation with AKP was graded as no correlation, inconclusive, weak, or a strong correlation.

Statistical analysis

Data from the included studies were tabulated to show the determinants investigated for their possible association with AKP and the reported prevalence of AKP. Data were analysed descriptively by using proportions and frequencies for categorical variables. A meta-analysis and statistical analysis were not feasible due to heterogeneity of the study designs and patient populations, surgical techniques, the type of knee prosthesis used, and the reported outcome measures.

Results

Search strategy and results are summarized in Fig. 1. Of 221 titles reviewed, 108 studies underwent a full review, of which 54 were included. The quality of the included studies according to GRADE is reported in Table 1.

Fig. 1
figure 1

PRISMA flow diagram of included studies

Full size image
Table 1 GRADE evidence profile: non-modifiable and modifiable determinants of anterior knee pain following primary total knee replacement
Full size table

Patient characteristics

Both age and gender were not found to be predictive of postsurgical AKP [11, 14, 16, 24, 54, 60, 61]. Also, a multitude of studies reported no association between postsurgical AKP and weight, height, or BMI [4, 6, 11, 14, 16, 24, 43, 48, 54, 55, 60]. A prospective cohort study found that the frequency and severity of AKP after TKR is related to retained presurgery gait patterns that had higher external flexion moments in the early mid-stance phase, which place higher forces on the patellofemoral joint [54]. A number of studies demonstrated that preoperative AKP was not predictive of postoperative AKP, whether the patella was resurfaced or not [2, 4, 6, 7, 11, 14, 15, 48, 54, 55, 61].

Knee-specific characteristics

Diagnosis (osteoarthritis vs. other) and preoperative radiographic stage are not predictive of postsurgical AKP [16, 43]. The relation between the degree of intra-operatively assessed patellar cartilage involvement and postsurgical AKP is inconclusive, with some studies reporting an association [50, 51], while others did not [4, 6, 7, 14, 15, 40, 43, 61]. Preoperative range of motion and preoperative Knee Society knee scores were not found to be predictive of AKP [61], and no correlation was found between a preoperative deformity and AKP [16].

Prosthetic design

Several studies evaluated the results of improving the ‘patella-friendliness’ of femoral component designs, and found no effect on AKP rates [16, 35, 41, 57]. In a recent meta-analysis on patellar resurfacing, the impact of prosthetic design was evaluated by classifying the TKRs in the included RCTs as patella-friendly or non-patella-friendly according to the conformity of the patella against the femoral component [46]. No differences in the incidence of AKP between groups were found. The use of prostheses with a more posterior centre of rotation may result in lower AKP rates [32]. No differences in the prevalence of AKP were seen between PCL-retaining- and PCL-substituting TKR [16, 57, 60]. With regard to the tibial component, no differences were found in the prevalence of AKP between a posterior-stabilized tibial component and a conforming deep-dish congruent component [28], or between a conventional posterior-stabilized and a high-flex posterior-stabilized polyethylene tibial insert [34]. Although one study reported lower AKP rates with mobile-bearing as compared to fixed-bearing tibial components [11], other studies found comparable proportions of AKP [1, 8, 26]. One study found similar AKP rates for one-peg patellar component fixation compared to three-peg fixation [27].

Operative technique

A retrospective case series demonstrated no differences in AKP rate between the lateral and medial approach [16]. Imageless navigation and computed tomography (CT)-based navigation resulted in an equal occurrence of AKP [33]. Others also found no differences in the AKP rate between computer-navigated and conventional jig-based TKR [56]. An inconclusive relation was found between postsurgical AKP and radiographic measurements of changes in the joint line, the anterior or posterior offset of the tibial component relative to the centre of the tibia (the distance from the neutral axis), the distance between the inferior pole of the patella and the tibial prosthesis, or centralization of the tibial component on the cut surface of the tibia [12, 18]. Barrack et al. [5] used computed tomography to evaluate the relationship between AKP and component rotation after TKR. Patients with combined component internal rotation were more than 5 times as likely to experience AKP compared with those with combined component external rotation [5]. A number of studies found no correlation between the amount of patellar tilt or subluxation and the presence of AKP [5, 16, 24, 47, 61]. Similarly, several studies did not find an association between the development of AKP and patella height [12, 16, 24, 47, 61], or between lateral tibiofemoral angle and AKP [5, 47, 61]. However, one study noted more AKP with greater patella height [37], while another study noted the opposite [17]. No association was found between the prevalence of AKP and whether an uncovered lateral patellar facet was articulating with the femoral component, the percentage cover of the implant of the patella on a lateral radiograph, or tilt of the patellar component [16]. Also, a number of studies demonstrated no relation between lateral release performed for patellar tracking and the occurrence of AKP [14, 16, 55, 60, 61]. Resection of Hoffa’s fat pad may result in a lower prevalence of postsurgical AKP [31]. In case of patellar resurfacing, residual patellar thickness was not found to be predictive of AKP [16, 25].

Treatment of the patella

Several RCTs and observational studies reported on resurfacing versus non-resurfacing of the patella with regard to the prevalence of AKP, with some studies finding no differences [4, 6, 12–15, 23, 48, 55], whereas others have shown a lower prevalence of AKP following patellar resurfacing [20, 38, 43, 45, 59–61]. Some of these studies have given conflicting results even when the same implant has been used [4, 61]. A number of meta-analyses have demonstrated a quantitative estimate of overall effect of patellar resurfacing through sample pooling of different RCTs. The rate of AKP favoured patellar resurfacing in 3 meta-analyses [39, 42, 44], yet between-study heterogeneity existed in all of these three meta-analyses. More recent meta-analyses demonstrated no differences between resurfacing and not resurfacing in terms of AKP [19, 22, 29]. Of note, these recent meta-analysis studies included more recent RCTs that showed no advantage of resurfacing in TKR [13, 55]. In a study evaluating the results of selective patellar resurfacing, no difference in AKP rate was observed between resurfacing and retaining [36]. Resurfacing was performed when patients had preoperative patellofemoral symptoms, patellar instability/malalignment, and Grade III/IV articular cartilage changes with a patellar thickness >15 mm, and the decision not to resurface was made if Grade 0/I/II changes, or a patella thickness of less than 15 mm were observed. Patellar resurfacing and patellar reshaping with partial resection of lateral facet and osteophytes resulted in comparable rates of postsurgical AKP [30]. Electrocautery of the patellar rim in TKR without patellar resurfacing may result in a lower incidence of AKP as compared to TKR without patellar rim electrocautery [58].

Time of assessment

Some studies describe a gradual reduction in the incidence of patellar pain over time [2, 32, 35], others found no change in the incidence between 5 and 10 years postoperatively [12], and two studies noted an increase in the overall prevalence of AKP over time [4, 15].

Discussion

The most important finding of the present study was that a multitude of variables have been researched with regard to the prevalence of AKP, with some clearly demonstrating a correlation with AKP, while others have shown no correlation. Patient characteristics including age, gender, weight, height, BMI, and the presence of preoperative AKP were not found to be predictive of AKP. Knee-specific characteristics including diagnosis, radiographic grade of osteoarthritis, knee scores, range of motion, and deformity were not related to the occurrence of postsurgical AKP. An inconclusive relation was found for the intra-operative degree of patellar cartilage wear. With regard to prosthetic design, no recommendation can be given for the use of ‘patella-friendly’ trochlear designs, specific PCL management strategies, tibial component bearing strategies, design geometry of the tibial insert, and patellar component fixation. In contrast, a weak recommendation can be given for prostheses with a posterior centre of rotation. With regard to operative technique, no recommendation can be given for the use of a specific surgical approach, navigation, lateral release, or residual patellar thickness. However, a weak recommendation can be given for preventing combined component internal rotation and for resecting Hoffa’s fat pad. No recommendation can be given for patellar resurfacing, patellar reshaping is provided with a weak recommendation against, and the use of patellar rim electrocautery is recommended. The relation between AKP and time of assessment is inconclusive.

The main limitation of our study is that the retrieved studies have not been quantitatively assessed because we found important heterogeneity in both study design and outcome assessment. Most of the included studies evaluated one single factor with regard to the prevalence of AKP, although several RCTs presented data on multiple factors potentially influencing AKP [4, 6, 12, 14, 15, 43, 55, 60, 61]. The presence of confounding factors, whether clinical, technique, or prosthesis related, thus reduces the generalizability of the reported findings. Furthermore, the use of different scoring systems for AKP that consider both the presence and the severity of AKP has resulted in variations in the objective assessment of pain and contributes to the observed heterogeneity. Another limitation is that we included patellofemoral pain in our search strategy while AKP may not be necessarily synonymous with patellofemoral pain.

Some investigators found that the occurrence of postoperative AKP is negatively correlated with patient satisfaction [4, 44]. Also, patients with AKP reported lower levels of quality of life than patients without AKP [11]. Considering the implications of AKP following TKR on reoperation rate, patient satisfaction, and quality of life, further study into the causative mechanisms of AKP is needed.

So far, most studies evaluating AKP have focused on the patella as the pain generator. In the majority of the included studies, neither the grade of patellar cartilage damage (radiographically or intra-operatively assessed), nor the patella-friendliness of the femoral component exhibits a clear relation with the occurrence of AKP. In contrast, we found that patellofemoral joint loading is associated with the occurrence of AKP, and this may be an important hypothesis-generating observation. Several factors resulting in abnormal patellofemoral joint loading have been identified in our review, including gait pattern [54], an anterior position of the centre of rotation of the femoral component [32], suboptimal component alignment with changes in joint line height [17, 18, 37], and rotational errors [5]. The presence of these factors may explain the generally unsuccessful results of secondary resurfacing. However, other factors influencing patellofemoral load, including the anterior or posterior offset of the tibial component [12], patella height [12, 16, 24, 47, 61], and patellar thickness [16, 25] demonstrated no or an inconclusive correlation with AKP. Clearly, further study into the causative mechanisms of AKP after TKR is needed.

In order to reduce the prevalence of AKP following primary TKR, we recommend the use of femoral components with a posterior centre of rotation, resection of Hoffa’s fat pad, patellar rim electrocautery, and preventing combined component internal rotation. No recommendation can be given for the use of specific prosthetic designs, PCL management strategies, tibial component bearing strategies, specific surgical techniques, or patellar resurfacing, since the available evidence did not show a correlation with these factors and the occurrence of postsurgical AKP.

Conclusion

AKP following primary TKR is both prevalent and clinically relevant. A multitude of patient and technique related factors have been researched with regard to the prevalence of AKP. No single variable is likely to explain the differences in the reported rates of AKP, although variables resulting in abnormal patellofemoral joint loading appear to be of special significance.