To investigate the relationship between quadriceps strength and the peak knee adduction moment during walking in medial tibiofemoral osteoarthritis (OA), and whether varus malalignment influences this relationship.
To investigate the relationship between quadriceps strength and the peak knee adduction moment during walking in medial tibiofemoral osteoarthritis (OA), and whether varus malalignment influences this relationship.
Maximum isometric quadriceps strength at 60° flexion relative to body mass and the peak knee adduction moment during walking were assessed in 184 community volunteers with medial knee OA. Mechanical knee alignment was determined either directly from full-leg radiograph or extrapolated from anatomic alignment on knee radiograph using regression equations. Pearson's correlation coefficient was used to assess the association between quadriceps strength and peak knee adduction moment. The independent relationship between quadriceps strength and peak knee adduction moment, and the impact of varus malalignment on this relationship, was assessed using multiple regression analyses with and without adjustment for covariates.
Quadriceps strength was not significantly associated with peak knee adduction moment (r = 0.14, P = 0.059). Neither quadriceps strength (b = 0.25, P = 0.142) nor the interaction between quadriceps strength and varus malalignment (b = −0.01, P = 0.693) significantly contributed to the variance in peak knee adduction moment. Results were unchanged with the inclusion of covariates.
No significant association was observed between quadriceps strength and the peak knee adduction moment, and the severity of varus malalignment did not influence the relationship. Results suggest that clinicians should not be concerned that patients with knee OA and stronger quadriceps are more likely to demonstrate a higher knee adduction moment.
Quadriceps weakness is common in people with knee osteoarthritis (OA) (1, 2) and is associated with impaired physical function and increased knee pain (3). Not surprisingly, studies have demonstrated that quadriceps strengthening can reduce pain and improve function in the short term (4–6). Accordingly, quadriceps strengthening is recommended in clinical guidelines for the management of knee OA (7–9). Despite the widespread clinical use of quadriceps strengthening, and in light of the structural joint abnormalities associated with knee OA, it is not clear whether quadriceps strength is associated with important biomechanical parameters such as indices of knee joint loading.
An important biomechanical parameter in tibiofemoral OA is the external knee adduction moment. The knee adduction moment is created by the ground reaction force passing medial to the center of the knee joint during gait, and is primarily derived from the product of the resultant ground reaction force in the frontal plane and its moment arm (the perpendicular distance from the ground reaction force to the knee joint center of rotation) (10). The knee adduction moment is generally higher in people with knee OA compared with healthy populations (11, 12), and it is an important predictor of disease progression in medial knee OA (13). As such, it is a commonly accepted surrogate measure of medial knee compartment loading (10, 14). If treatments aimed at reducing the knee adduction moment, and therefore aimed at reducing the risk of disease progression, are to be developed, research must first determine factors that contribute to an elevated knee adduction moment.
Quadriceps strength is one factor that theoretically may influence the knee adduction moment, yet no research has evaluated this relationship. Greater quadriceps strength is often associated with a faster walking speed, particularly at the lower end of the functional scale (15–17). Faster speed is associated with increased ground reaction force magnitude (18), and therefore the knee adduction moment (11). Furthermore, greater quadriceps strength is associated with less severe knee pain (3), and pain reduction has been shown to increase the knee adduction moment in some studies (19–21). Therefore, it is also possible that greater quadriceps strength could be associated with a greater knee adduction moment by virtue of its association with reduced pain severity. Conversely, there is evidence indicating that the quadriceps is the major muscle group responsible for generating an abduction moment to resist the knee adduction moment (22, 23). Therefore, it is also feasible that greater quadriceps strength could be associated with a lower knee adduction moment.
There are growing data to suggest that the local mechanical environment is important in influencing the clinical status of knee OA, as well as the outcome from treatment (24, 25). Knee malalignment is considered one of the most important local mechanical factors. For example, our recent cross-sectional study revealed that people with knee OA and varus malalignment demonstrate both greater quadriceps strength relative to body mass and a higher peak knee adduction moment when compared with people with more neutral alignment (26). Furthermore, a longitudinal study has demonstrated that greater absolute quadriceps strength at baseline increases the risk of knee OA progression in patients with malaligned, but not neutrally aligned, knees (27). It is not clear how greater absolute quadriceps strength increases the risk of disease progression in individuals with knee malalignment, but it may be via an increased knee adduction moment. It is possible that the concomitant effects of knee malalignment and greater quadriceps strength could increase focal loading across the articular cartilage beyond that observed with either malalignment or elevated quadriceps strength alone. Therefore, the relationship between quadriceps strength and the peak knee adduction moment may differ according to the severity of varus malalignment present.
This study aimed to investigate the relationship between quadriceps strength and the peak knee adduction moment during walking in people with medial tibiofemoral OA, and to examine the impact of varus malalignment on this relationship. It was hypothesized that quadriceps strength would be significantly correlated with the knee adduction moment, and that the relationship between quadriceps strength and the knee adduction moment would be stronger in people with more varus malalignment compared with those with more neutral alignment.
This study presents a cross-sectional analysis of data collected from 184 community volunteers with medial tibiofemoral OA in 3 separate but similar cohorts. The first cohort consisted of 38 participants from a cross-sectional study investigating effects of footwear and gait aids on the knee adduction moment (28), the second cohort consisted of a subset of 39 participants from a randomized controlled trial examining the efficacy of laterally wedged insoles (29), and the third cohort consisted of 107 participants involved in another randomized controlled trial investigating effects of quadriceps strengthening on the knee adduction moment (30).
Participants in all cohorts were recruited from the community in Melbourne, Australia through advertisements in newspapers and local community clubs. Eligibility criteria were similar in all cohorts except that participants with severe radiographic OA were excluded from the second cohort. All participants had tibiofemoral OA in at least 1 knee and fulfilled the American College of Rheumatology classification criteria (31): age >50 years, knee pain most days of the past month, and osteophytes apparent on knee radiograph. In addition, to ensure that participants had medial tibiofemoral joint OA, the following criteria were set: self-reported pain on the medial aspect of the knee, osteophytes in the medial tibiofemoral compartment, and medial joint space narrowing greater than lateral joint space narrowing via visual inspection (32).
Participants were excluded if they had a history of lower extremity joint replacement, had knee surgery or intraarticular steroid or hylan G-F 20 injection in the previous 6 months, had a systemic arthritic condition, were seeking or currently receiving physiotherapy for knee OA, or had a severe medical condition. Participants were screened over the telephone and those eligible underwent weight-bearing radiographic analysis. Participants fulfilling radiographic eligibility criteria were enrolled into the study. Ethical approval was obtained from the University of Melbourne Human Research Ethics Committee and from the Department of Human Services Radiation Advisory Committee. Written informed consent was provided by participants at enrollment.
The most painful eligible knee was deemed the study knee. Where both knees were equally painful and eligible for inclusion, the dominant knee was deemed the study knee. Disease severity was assessed using the Kellgren/Lawrence (K/L) system (33), in which higher grades indicate greater OA severity. Mechanical alignment was measured from a long-extremity radiograph in the first cohort, and in the remaining participants was extrapolated from anatomic alignment measured using published equations (34, 35) and semiflexed (second cohort) or extended (third cohort) knee radiographs.
Participants in the first cohort underwent a single, full-leg, anteroposterior weight-bearing radiograph (28). Participants stood barefoot with the knee extended and the tibial tuberosity facing forward. Knee alignment was defined as the angle of intersection of the femoral and tibial mechanical axes (36). The mechanical axis of the femur was determined by a line from the femoral head center to the center of the femoral intercondylar notch, and the tibial mechanical axis was established using a second line drawn from the center of the tibial spines to the talar center. In this way, a direct measure of mechanical alignment was obtained.
Participants in the second cohort underwent a semiflexed posteroanterior weight-bearing knee radiograph. Participants stood barefoot with the anterior aspect of the thigh, the knee, and the toes touching the radiography frame. The feet were externally rotated 10° by aligning the medial aspects of the feet against a customized frame (37). Anatomic knee alignment was determined using the methods by Moreland et al (38). The anatomic axis of the femur was attained by drawing a line from the center of the tibial spines to a point located 10 cm proximally and bisecting the medial to lateral width of the femur. Similarly, tibial anatomic axis was obtained by drawing a line from the center of the tibial spines to a point located 10 cm distally and midway between the medial and lateral tibial surfaces. Anatomic knee alignment was indicated by the angle subtended at the point where the 2 lines met in the center of the tibial spines. Mechanical knee alignment was extrapolated using the following regression equation from Kraus et al (35):
Participants in the third cohort underwent an anteroposterior extended weight-bearing knee radiograph. Participants stood barefoot with the knee extended and the tibial tuberosity facing forward (36). The procedure for measuring knee alignment was identical to that used with the second cohort. Mechanical knee alignment was extrapolated using the regression equation from Hinman et al (34):
For all participants, knee alignment was reported as the deviation from neutral in the varus direction, and an alignment of 0° indicated a neutrally aligned knee.
Quadriceps strength was assessed isometrically at 60° knee flexion while sitting using a Kin Com 125-AP dynamometer (Chattecx Corporation, Chattanooga, TN). A submaximal warmup was followed by 3 maximal 5-second contractions with a 15-second rest interval in between trials. Participants were given verbal encouragement during the test. The highest peak force of the 3 trials (corrected for lower extremity weight) was multiplied by lever length (in meters) to obtain maximum torque (Nm), which was then normalized by body mass (Nm/kg). This conventional normalization is consistent with predictions of geometric similarity that scale the product of muscle cross-sectional area and moment arm (i.e., torque-generating capability) to body mass (to the power of 1.0) (39). Test–retest reliability for this measure in our laboratory in 10 patients with knee OA tested 1 week apart was excellent (intraclass correlation coefficient [ICC] 0.93).
Peak external knee adduction moment was measured using a Vicon 612 motion analysis system (Vicon, Oxford, UK) with 8 cameras (120 Hz M2). Two force plates (0R6-6-2000; AMTI, Watertown, MA) embedded in the center of an 8-meter walkway were sampled at 1,080 Hz. Twenty retroreflective markers were placed over standardized anatomic landmarks according to the Vicon Plug-in Gait lower extremity model, and inverse dynamics were used to calculate external joint moments about an orthogonal axis system located in the distal segment of the joint (40). Participants walked in their own low-heeled shoes at a self-selected pace until 5 trials with single-leg force platform contacts were recorded. Peak knee adduction moment was normalized to body weight multiplied by height (%Nm/BW×HT), which is the convention in OA literature, where the intention is to yield a dimensionless number (41) that accounts for the influence of body weight and height principally on the ground reaction force and its moment arm (the primary determinants of the knee adduction moment). The mean peak knee adduction moment for the 5 trials was used in the analysis. Test–retest reliability in our laboratory was excellent with this procedure in 11 patients with knee OA tested 1 week apart (ICC 0.97).
Walking speed and knee pain were measured to control for their potential influence on the knee adduction moment (11, 42). Walking speed was measured using photoelectric beams (Jaycar Electronics, Melbourne, Victoria, Australia) attached to a stopwatch and placed 4 meters apart in the middle of the walkway. The pain subscale of the Likert version of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), transformed to a 100-point scale in which higher scores indicated worse pain, was used to assess knee pain over the past 48 hours (43). Knee pain data for 1 participant in the third cohort was missing.
The Statistical Package for the Social Sciences, version 15 (SPSS, Chicago, IL) was used for all analyses, and P values less than 0.05 were considered statistically significant. All data approximated a normal distribution; therefore, parametric tests were used for analyses. Pearson's correlation coefficient assessed the association between quadriceps strength and knee adduction moment in the entire cohort, as well as within each of 3 increasing varus malalignment groups based on the tertiles of the distribution of knee alignment within the cohort. Participants with knee alignment <2° varus were categorized as least varus, those with 2–5° varus inclusive were categorized as moderate varus, and those >5° varus were categorized as most varus.
Multiple regression analyses were performed to assess the independent relationship between quadriceps strength and knee adduction moment, as well as the impact of varus malalignment, using the ENTER method. The dependent variable was the peak knee adduction moment and independent variables were varus malalignment, quadriceps strength, and the interaction variable between quadriceps strength and varus malalignment. This latter variable was included to assess whether varus malalignment modified the relationship between quadriceps strength and knee adduction moment. Regression analyses were then repeated with adjustment for age, sex, walking speed, and knee pain severity. The impact of the independent variables on knee adduction moment was determined from the unstandardized regression coefficient, b.
Characteristics of the 184 participants are presented in Table 1. There were more women (55%) than men, and the mean age of participants was 64.8 years. There was a relatively even spread of disease severity across the cohort, with 37% of participants demonstrating K/L grade 2, 30% with K/L grade 3, and 33% with K/L grade 4. The average WOMAC score for pain was 38 (out of 100), indicating a moderate level of pain across the cohort.
|Age, years||64.8 ± 8.4|
|Female sex, n (%)||101 (55)|
|Height, meters||1.66 ± 0.10|
|Body mass, kg||80.4 ± 14.8|
|BMI, kg/m2||29.2 ± 4.6|
|Knee alignment, degrees||3.9 ± 3.6|
|Peak knee adduction moment, %Nm/BW×HT||3.90 ± 0.92|
|Disease severity, n (%)†|
|K/L grade 2||69 (37)|
|K/L grade 3||55 (30)|
|K/L grade 4||60 (33)|
|WOMAC pain score‡||38 ± 15|
|Quadriceps strength, Nm/kg||1.33 ± 0.52|
|Self-selected walking speed, meters/second||1.20 ± 0.19|
Distribution of knee alignment across participants is presented in Figure 1. The mean ± SD alignment of the entire cohort was 3.9° ± 3.6° in the varus direction, with a range from 4.5° (valgus) to 15.5° (varus). There was no association between quadriceps strength and the peak knee adduction moment in the whole cohort (r = 0.14, P = 0.059) (Figure 2). The relationship remained nonsignificant even when the cohort was divided into tertiles based on alignment. The associations between quadriceps strength and the knee adduction moment in the least varus, moderate varus, and most varus groups were r = 0.12 (P = 0.377), r = 0.24 (P = 0.059), and r = 0.08 (P = 0.532), respectively (Figure 3).
Regression analyses demonstrated that varus malalignment was the only independent predictor of peak knee adduction moment (b = 0.11, P = 0.029). Neither quadriceps strength (b = 0.25, P = 0.142) nor the interaction between quadriceps strength and varus malalignment (b = −0.01, P = 0.693) contributed significantly. This model accounted for 13% of the total variance in peak knee adduction moment (adjusted R2 = 0.13, F = 10.27, P < 0.001).
The inclusion of covariates to the model did not change the results. Varus malalignment remained a significant predictor of peak knee adduction moment (b = 0.12, P = 0.012). Neither quadriceps strength (b = 0.08, P = 0.660) nor the interaction term between quadriceps strength and varus malalignment (b = −0.01, P = 0.718) was significantly associated with the peak knee adduction moment. Of the covariates used, only self-selected walking speed was significantly associated with knee adduction moment (b = 1.82, P < 0.001). The adjusted model explained 22% of the variance in peak knee adduction moment (adjusted R2 = 0.22, F = 8.24, P < 0001).
This study investigated the relationship between quadriceps strength and the peak knee adduction moment during walking in people with symptomatic medial tibiofemoral joint OA, as well as the impact of varus malalignment on this relationship. It was hypothesized that quadriceps strength would be significantly associated with the knee adduction moment, and that this relationship would be stronger in the presence of varus malalignment. Contrary to this hypothesis, there was no significant association between quadriceps strength and the knee adduction moment, and the severity of varus malalignment did not influence the relationship. Given the large number of participants in this study and the spread of the data obtained, the nonsignificant findings are unlikely to result from low statistical power or range effects. Our data suggest that quadriceps strength does not influence the peak knee adduction moment in people with medial knee OA, even when varus malalignment is present.
To the best of our knowledge, this is the first study examining the relationship between quadriceps strength and the peak knee adduction moment, a measure of knee loading. Therefore, there is no similar study with which to compare results. However, the association between quadriceps strength and another loading parameter, impact loading during the initial contact phase of walking, has been examined (44). The authors of that study classified 37 healthy women into strength training or sedentary groups based on their training history in the past year. The strength training group had performed lower extremity training for at least an hour ≥3 times/week. The sedentary group had not engaged in any formal exercise. Sedentary women demonstrated significantly weaker concentric and eccentric quadriceps strength relative to body mass compared with the women who performed strength training exercises. The body-weight–normalized rate of loading measured from the ground reaction force during gait was significantly greater in the sedentary group than in the strength training group. The frequency of heelstrike transients was also higher in the sedentary group, but the difference was not examined statistically. The authors postulated that quadriceps weakness may reduce the ability to dampen rates of loading and, if transmitted to the knee joint, could be injurious to the joint. Therefore, although it appears that quadriceps strength may play a role in attenuating load associated with initial foot contact during walking, our study suggests it has no significant contribution to the magnitude of the peak knee adduction moment experienced during stance phase.
The mean peak knee adduction moment of 3.9% Nm/BW×HT demonstrated by our participants is comparable with the results of other studies (11, 20). Unfortunately, there are relatively few studies that have measured quadriceps strength correctly as torque in OA patients that have also normalized for body mass. Our mean isometric strength at 60° knee flexion was 1.33 Nm/kg, which is very similar to that recorded by Jan et al (45). Other comparable studies have used 60°/second isokinetic testing and, after appropriate unit conversion, their peak torque scores are ∼30% lower than ours for their OA cohorts of varying age, sex, and severities (2, 46–49). Isokinetic strength recorded at slow velocities would be expected to be somewhat lower than isometric strength, given the relatively flat torque–velocity relationship in this slow velocity region (45). Therefore, the knee adduction moment and strength parameters of our cohort appear to be broadly consistent with that measured by other studies.
There are several potential explanations for why there was no association between quadriceps strength and the peak knee adduction moment in our study. The knee adduction moment magnitude is influenced by both the magnitude of the ground reaction force and by its distance from the knee joint center (moment arm). As expected, post hoc analyses revealed that greater quadriceps strength was associated with faster walking speed (r = 0.35, P < 0.001) and decreased knee pain (r = −0.25, P = 0.001) in our study. Given that faster walking speed can increase the magnitude of the ground reaction force (18), it would be expected that a concomitant increase in the knee adduction moment (11) would also occur. Furthermore, as reduced knee pain has been shown to increase the knee adduction moment in knee OA (19–21), it might have been expected that a relationship between quadriceps strength and the knee adduction moment would be observed. However, the relationship between knee pain severity and the knee adduction moment is inconsistent across the literature and remains poorly understood, as greater knee pain has also been associated with a greater knee adduction moment by some authors (13).
Walking does not require maximal levels of quadriceps strength, and therefore maximum isometric quadriceps strength, as measured in our study, might not be reflected proportionally in the submaximal, dynamic strength needed to influence the knee adduction moment during gait. Having a stronger quadriceps may not necessarily cause one to activate that muscle group proportionally during gait for forward locomotion and for resisting the knee adduction moment. There is substantial variability in muscle activation patterns during gait in people with knee OA (50, 51), and those with severe disease may utilize a higher proportion of maximum quadriceps activation or strength compared with those with less severe disease (50). Therefore, variability in muscle activation patterns among participants might have distorted any relationship between maximal quadriceps strength and the knee adduction moment.
It is possible that other lower extremity muscles, such as the hip abductors, play an important role in influencing the knee adduction moment. An 18-month study of 57 patients with mild to moderate knee OA found that a greater internal hip abductor moment (equivalent to an external hip adduction moment, and suggestive of increased use of the hip abductors) during gait was associated with a reduced likelihood of ipsilateral medial knee OA progression (52). The authors suggested that weak hip abductors in the stance extremity may lead to a pelvic drop in the contralateral swing extremity. This could shift the body's center of mass toward the swing extremity and increase the ground reaction force frontal plane moment arm at the knee, and therefore the magnitude of the peak knee adduction moment. However, hip abductor strength was not assessed in our study, and therefore its influence on the peak knee adduction moment could not be established or controlled for.
Although an attempt was made to control for the potential confounding effects of age, sex, walking speed, and knee pain severity on the knee adduction moment, adjustment for these factors did not change results. However, other factors (e.g., type of footwear worn, degree of trunk sway exhibited by participants, and medication use) were not controlled, and these factors could potentially mask a relationship between quadriceps strength and the peak knee adduction moment (53, 54).
Our study found that varus malalignment did not alter the relationship between relative quadriceps strength and the peak knee adduction moment. Previously, greater absolute quadriceps strength in people with malaligned knees has been shown to pose an increased risk of disease progression (27). Although medialization of the patella tendon may occur in knees with greater varus malalignment, this may not be enough to significantly alter the quadriceps line of action as suggested by other authors (27, 55). Van der Esch and colleagues (48, 49) demonstrated that high varus–valgus laxity and poor proprioception significantly increased the impact of weakness of the knee extensors and flexors relative to body mass on functional ability. Therefore, knee laxity and proprioceptive acuity, which we did not assess, may have a stronger impact than varus malalignment on the relationship between quadriceps strength and the peak knee adduction moment. This should be an area of future research.
Increases in absolute quadriceps strength may have deleterious consequences for disease progression if it results in the overloading of damaged cartilage via increased knee adduction moment. Therefore, the absence of a relationship between quadriceps strength and the knee adduction moment suggests that clinicians should be not concerned that patients with knee OA and stronger quadriceps are more likely to demonstrate a higher knee adduction moment, and possibly greater loading in the medial compartment. However, it is possible that greater quadriceps strength may influence other measures of knee joint loading, such as heelstrike transients (56) and joint compressive force (10). It has been suggested that stronger quadriceps may help to attenuate heelstrike transients (57), which was not examined in our study. Contraction of the quadriceps muscle increases axial compressive force across the knee joint (58). Therefore, greater quadriceps strength could increase axial compressive force (10), which was also not assessed in our study. Further research is required in this area.
The pooling of data from 3 different cohorts of participants is both a strength and limitation of our study. It resulted in a robust cross-sectional study with a large number of participants, which increased its statistical power. Because different researchers were involved in each cohort, subtle differences in data collection procedures (such as the nature of verbal encouragement given to participants during muscle strength testing) cannot be excluded. However, quadriceps strength, knee adduction moment, pain, and walking speed were assessed in all 3 cohorts using identical measurement protocols and equipment in our laboratory. Another limitation is the cross-sectional design, which prevents the establishment of a cause and effect relationship between quadriceps strength and the knee adduction moment.
In conclusion, we did not find a cross-sectional relationship between quadriceps strength and the peak knee adduction moment. Furthermore, this relationship was not influenced by the presence of varus malalignment.
Dr. Hinman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Lim, Wrigley, Bennell, Crossley, Hinman.
Acquisition of data. Lim, Kemp, Metcalf.
Analysis and interpretation of data. Lim, Kemp, Metcalf, Wrigley, Bennell, Crossley, Hinman.
Manuscript preparation. Lim, Kemp, Metcalf, Wrigley, Bennell, Crossley, Hinman.
Statistical analysis. Lim.
Acquisition of funding. Bennell, Hinman.
We wish to acknowledge the assistance of Kelly Anne Bowles with data collection on some of the participants in this study.