Subject: Knee stabilization in patients with medial compartment knee osteoarthritis
*Correspondence to Katherine S. Rudolph, 301 McKinly Lab, University of Delaware, Newark, DE 19716
Received: 29 November 2004; Accepted: 18 May 2005
10.1002/art.21237 About DOI
Knee osteoarthritis (OA) involves the progressive destruction of articular cartilage and can cause substantial disability among middle-age and older adults (). Many factors contribute to the development of knee OA, including heredity, biochemical changes in articular cartilage, and biomechanical compressive loads that lead to joint damage. In people with medial knee OA, compressive loads on the medial tibiofemoral joints are increased in the presence of varus knee alignment and may be manifested as elevated knee adduction moments during walking, both of which have become hallmarks of the disease ([3-6]).
Our recent study demonstrated that patients with medial knee OA co-contract the quadriceps and gastrocnemius muscles on the medial side of the knee joint to a greater extent than that in age-matched control subjects, during the early stance phase of walking (). Larger magnitudes of medial co-contraction in the presence of higher knee adduction moments seems counterintuitive, since this process would appear to further increase the compressive load on the painful medial articular surface. We assert, however, that the higher co-contraction of the medial muscle is a direct response to the frontal plane laxity that appears on only the medial side of the knee joint. In fact, patients with isolated medial knee OA and genu varum were shown to have significantly greater laxity of the medial joint, as demonstrated on stress radiographs and depicted in Figure 1, compared with an uninjured control group, whereas no such difference was observed on the lateral side of the knee joint ().
The presence of greater frontal plane laxity in individuals with knee OA has been well established (). The terms joint laxity and instability are often used interchangeably; however, laxity is measured statically and may not have any relation to how the knee functions under dynamic conditions. In the present study, we use the operational definition of joint instability suggested by Irrgang et al, in which joint instability is the patient's perception of the extent to which shifting, buckling, and giving way in the knee interferes with daily activities (). Recently, Fitzgerald et al () reported that a significant proportion of patients with OA report perceiving sensations of joint instability which diminish physical function. Insight into the manner in which patients attempt to control laxity to minimize instability is vital to understanding disease progression and would facilitate the design of treatment programs to improve outcomes in patients with medial knee OA.
Evidence suggests that frontal plane stability may be enhanced by reflexive responses elicited in muscles surrounding the knee by stimulating periarticular medial and lateral joint structures ([11-13]). Palmer () found that mechanical stimulation of the deep portion of the medial collateral ligament results in activation of the semimembranosis, sartorius, and vastus medialis of the cat hindlimb. Kim et al () activated the medial collateral ligament with electrical stimulation of uninjured human knees, thus demonstrating that afferent information from the ligament produces a reflexive response in the medial knee muscles. Buchanan et al () determined that a similar reflexive response from muscles on the medial side of the joint (vastus medialis, semitendinosis, gracilis, and sartorius) was elicited by a rapid valgus movement of the knee.
The presence of excessive medial joint laxity in individuals with medial compartment knee OA may delay or inhibit neuromuscular responses because greater joint excursions are required to activate high-threshold mechanoreceptors (). Thus, when an unexpected perturbation occurs, the unprepared neuromuscular system may be incapable of appropriately activating the correct muscles to stabilize the joint. To counteract this and avoid the perception of instability during routine activities of daily living, individuals with medial knee OA appear to use greater co-contraction of the medial muscles to help stabilize the knee (). This observation of altered muscle-activity patterns was made during level walking (), a relatively low-level task. When individuals with medial knee OA are faced with higher-level tasks that may challenge knee stability, such as changing direction, the knee can be subjected to substantial valgus loads (). Such activities may challenge the lax medial joint structures to a greater extent. Because muscle-activity patterns are seen to be altered even during a simple walking task (), the use of a potentially destabilizing valgus movement, which challenges the lax medial compartment, may provide greater insight into the muscle-activation patterns in patients with medial knee OA.
Higher muscle co-contraction can lead to high joint-compressive forces () that could hasten the progression of OA. As joint destruction progresses, instability may increase, requiring higher muscle co-contraction. In this scenario, the very strategy that was adopted to stabilize the joint could be the strategy that initiates a downward spiral of joint destruction. Without a thorough understanding of muscle stabilization in the knee with medial compartment OA, we cannot develop appropriate training programs to help improve joint stability, lessen disease progression, and improve patient function. In this study we set out to determine the manner in which individuals with medial compartment knee OA and excessive medial joint laxity respond to a potentially destabilizing event at the knee during standing. We hypothesized that the individuals with medial knee OA would stabilize the knee with greater co-contraction of the medial muscle. We also hypothesized that the greater magnitude of muscle co-contraction would be related to joint laxity and instability.
PATIENTS AND METHODS
Forty subjects participated in this study. Twenty-one patients with symptomatic medial knee OA and genu varum (7 female, 14 male, ranging in age from 39 years to 64 years, mean ± SD age 49.3 ± 7.0 years, mean ± SD body mass index [BMI] 30.0 ± 4.3 kg/m2) were referred by an orthopedic surgeon. Skeletal alignment was assessed from a radiograph of the joints in a weight-bearing position (). OA subjects were referred to our study if the weight-bearing line (connecting the center of the femoral head with the center of the ankle mortise) was <35% (calculated as the perpendicular distance from the weight-bearing line to the medial edge of the proximal tibia, divided by the width of the tibia; a ratio of less than 0.5 indicates varus angulation, and a ratio of greater than 0.5 denotes valgus angulation). The OA group had a mean ± SD weight-bearing line of 19.1 ± 12.0%. The diagnosis of OA was made from the clinical history, a physical examination, and radiographic changes observed on a posteroanterior radiograph carried out with knees flexed to 30° in a standing position (). Joint space width was measured using calipers (to the nearest 0.1 mm) at the narrowest location of both the medial and the lateral compartments. All subjects in the OA group showed joint space narrowing in the medial compartment (mean ± SD 2.0 ± 1.4 mm of joint space in the medial compartment and 6.4 ± 1.4 mm in the lateral compartment).
A group of 19 control subjects with neutral knee alignment were assembled for testing and matched to the OA group by age and sex (7 female, 12 male, ranging in age from 38 years to 62 years, mean ± SD 49.3 ± 5.8 years, mean ± SD BMI 28.3 ± 4.9 kg/m2). The control group had a weight-bearing line on the skeletal-alignment radiograph of 44.3 ± 8.1%, and had 4.9 ± 1.0 mm of joint space in the medial compartment and 6.0 ± 1.6 mm in the lateral compartment. Subjects were not included if they were pregnant, had a history of ligament deficiency, neurologic impairment, impaired balance or history of unexplained falls, rheumatoid arthritis, total knee replacement in either knee, or any other orthopedic problems in the hips, ankles, or spine, or had a BMI of 40.0 kg/m2. The test limb was the involved limb for the OA subjects and a randomly chosen limb for the control subjects. All subjects were informed of the purpose of the study and provided their informed consent on forms approved by the Human Subjects Review Board prior to testing.
Measurement of joint laxity.
Methods of measurement of frontal plane joint laxity have been described previously (). Briefly, frontal plane laxity was measured on radiographs of the test limb after application of stress forces, which were obtained from subjects as they lay supine with the knee supported and flexed 20°. A TELOS stress device (Austin & Associates, Fallston, MD) was used to reliably apply a 15-daN (33-lb) force to generate varus and valgus forces (). Joint space width was measured during the application of both varus and valgus stresses, and medial and lateral joint laxity were determined as shown in Figure 1 (). Intraclass correlation coefficients (ICCs) of repeated measurements using these methods on 8 healthy subjects revealed a reliability of 0.95 for measurements of lateral laxity and 0.97 for measurements of medial laxity.
Assessment of knee-joint function and instability.
All subjects rated their knee function using the Knee Outcome Survey Activities of Daily Living Scale (). Instability was assessed on the 6-point scale with the following question: To what degree does giving way, buckling or shifting of the knee affect your level of daily activity? . The reliability of this particular question for assessing knee instability in individuals with knee OA has been reported as an ICC of 0.72, indicating that it is adequate for the determination of the presence of knee instability ().
Muscle activity and motion capture.
Subjects stood with their feet placed shoulder-width apart and the test limb on a movable platform. A positioning actuator (NSK, Tokyo, Japan) translated the platform 5.8 cm laterally in 195 msec, at a peak velocity of 40 cm/second. The subject's contralateral foot was positioned on a 6-component force platform (Bertec, Worthington, OH) to monitor weight bearing. The involved limb's distal thigh was restrained with cuffs and cables that were bolted firmly to the wall, to prevent the thigh from translating laterally (Figure 2). To ensure that the motion was isolated to the knee, an additional strap was secured around the pelvis to restrict motion. Subjects stood comfortably with equal weight bearing on each limb. Five 3-second trials were carried out, in which the plate was triggered to move at a random interval within the 3 seconds. Subjects were unaware of when the plate would be triggered to move.
Motion of markers placed on the thigh and shank were collected at 120 Hz using a 6-camera passive-motion analysis system (VICON 512, Oxford Metrics, Oxford, UK). Electromyography (EMG) data from the test limb were recorded simultaneously at 1,920 Hz with a 16-channel system (Motion Lab Systems, Baton Rouge, LA). Surface electrodes recorded EMG signals from the vastus lateralis (VL), lateral hamstrings (LH), medial hamstrings (MH), the medial and lateral heads of the gastrocnemius (MG and LG, respectively), and the distal portion of the vastus medialis (VM) (). Fine-wire electrodes were inserted 2 cm apart into the sartorius and gracilis muscle along the longitudinal length of the muscle. Electrode placement was confirmed through muscle testing (). Signals were collected with the muscles at rest and from maximum voluntary isometric contractions (MVICs).
Data management and processing.
Marker trajectory and force plate data were filtered with a fourth-order, low-pass Butterworth filter, with cutoff frequencies of 6 Hz and 40 Hz, respectively. Vertical ground reaction force was normalized to body weight. Knee-joint angles were calculated with Euler angles (Move3D; NIH Biomechanics Laboratory, Bethesda, MD). Variables measured included the magnitude of limb loading, the time to unload the test limb, sagittal and frontal plane knee angles, frontal plane knee excursion, and the time (in msec) to reach peak valgus during plate movement (see Figure 3).
EMG data were filtered with a 350-Hz low-pass Butterworth filter, full-wave rectified, and filtered again with a phase-corrected eighth-order, 20-Hz low-pass Butterworth filter to generate a linear envelope (Labview 6i; National Instruments, Austin, TX). The linear envelopes were normalized to the maximum signal obtained during the MVIC trials.
The reflex muscle responses were evaluated over 2 intervals. The baseline interval comprised the 100-msec period prior to the onset of plate movement. The long loop interval was a fixed window of 40-175 msec after the onset of plate movement (). This time frame was chosen to constitute the muscle response that occurs following the monosynaptic stretch reflex, but prior to what would be elicited volitionally. It is thought that this response is mediated subconsciously, with afferent commands being sent to the cerebellum and brain stem (). The magnitude of the activity of individual muscles was evaluated by integrating the individual linear envelopes over the long loop reflex interval.
Muscle co-contraction was also calculated across the baseline and long loop intervals. Co-contraction is defined as the simultaneous activation of antagonistic muscles, with the co-contraction index calculated according to Equation 1 () as follows:
where i is the sample number. This method, developed in our laboratory, incorporates the relative activation of antagonistic muscles with their magnitudes, to evaluate the level of co-contraction that is likely to lead to joint compression. Co-contraction was calculated between the vastus lateralis and lateral hamstrings (VLLH), vastus medialis and medial hamstrings (VMMH), vastus lateralis and lateral gastrocnemius (VLLG), and vastus medialis and medial gastrocnemius (VMMG) muscle groups.
Joint laxity, joint kinematics, and weight-bearing timing and loads were compared between groups using independent-samples t-tests (SPSS version 11.0; Chicago, IL). Group differences in the percentage of trials with an observed muscle response were compared using independent t-tests after arc sine transformation (). Multivariate analysis of variance (ANOVA) was used to determine group differences in muscle integrals and co-contraction indices. The factors included were side of the knee (medial versus lateral), group (OA versus control), and muscle (VMMG, VMMH, VLLG, and VLLH, or VM, VL, MH, LH, MG, and LG). Two-way ANOVAs (side-by-muscle) were used to determine differences in co-contraction and integrated EMG values. For post-hoc testing, t-tests with Bonferroni correction were used. Spearman's rank correlation coefficients were used to assess the relationship between instability and joint laxity and the extent of muscle co-contraction. Measures of joint laxity and muscle co-contraction were converted to ordinal data for the purposes of the Spearman's rank correlation analysis only. Significance was established with P values of less than or equal to 0.05.
Baseline knee-joint function and stability.
The OA group had a mean ± SD 5.0 ± 1.7 mm of medial joint laxity, which was significantly greater than that in the control group, whose medial joint laxity measured 3.3 ± 0.9 mm (P = 0.001). No difference in lateral joint laxity was observed (3.4 ± 1.7 mm in the OA group versus 4.1 ± 1.5 mm in the control group; P = 0.19). No subject in the control group (n = 19) reported having knee instability, whereas 17 of 21 subjects (81%) in the OA group reported that they experienced knee instability. Of the 17 OA subjects who reported having knee instability, 14 reported that the instability affected their ability to perform activities of daily living. Among these subjects, 8 reported that the instability slightly affected their ability to perform daily activities, 3 stated that the instability moderately affected them, 2 reported that the instability severely affected them, and 1 reported that the instability prevented the performance of all daily activities. No significant relationship was observed between medial joint laxity and self-reported knee instability in the OA group (r = -0.098, P = 0.690).
Co-contraction levels prior to translation.
The series of straps used to stabilize the femur were pulled firmly, and in doing so, the knees were set in a more valgus position than that observed when the subjects were standing without the stabilization system. Compared with the knee position in the standing posture before application of the straps, after application of the straps (which elicited joint stress), the knees produced a change in frontal plane position that was similar in both groups; the OA knees were a mean ± SD 2.7 ± 2.3° more valgus, while the control knees were 3.3 ± 2.3° more valgus (P = 0.410). Prior to plate movement, no difference was observed in the knee flexion angle between the groups (OA 4.1 ± 5.4° versus controls 3.6 ± 6.6°; P = 0.783), and the OA knees were less valgus (0.6 ± 4.9°) than the control knees (6.0 ± 5.4°) (P = 0.002).
The OA group demonstrated greater co-contraction of the VMMG muscles (mean ± SD co-contraction index 4.9 ± 5.7) than did the control group (1.4 ± 1.6) (P = 0.014) before plate translation. The 7 OA subjects who reported having no knee instability or having instability that did not affect their daily activities had significantly larger VMMH co-contraction levels prior to plate translation than did the 14 OA subjects who reported having instability that affected their daily activities (mean ± SD co-contraction index 3.80 ± 3.08 in those with stable OA versus 1.67 ± 1.18 in those with unstable OA; P = 0.038). In the OA subjects, higher VMMH co-contraction correlated positively with greater knee stability (r = 0.459, P = 0.042) during the baseline interval.
Co-contraction levels during translation.
As the plate translated laterally, the positioning of the knee of the test limb moved more valgus (Figure 3). No between-group difference in valgus excursion was observed between the OA subjects (2.5 ± 1.0° valgus) and the controls (2.7 ± 0.7° valgus) (P = 0.699), although at the end of the plate movement, the OA group had significantly lower valgus angles (3.2 ± 4.8°) than did the control group (8.7 ± 5.4°) (P = 0.002). All subjects reached peak knee valgus before the end of the plate movement, and the time to reach the peak valgus position was no different between groups (173 ± 17 msec in OA subjects versus 173 ± 16 msec in controls; P = 0.911).
During plate translation, all subjects unloaded the test limb by transferring weight to the contralateral limb. The OA group shifted 17.9 ± 4.9% of body weight off the test limb, which was less than the 22.0 ± 5.0% of body weight that the control group shifted off the test limb (P = 0.013). The OA group also appeared to take a longer time to shift the weight off the test limb (199 ± 44 msec) compared with the control group (181 ± 13 msec) (P = 0.094), but this difference was not statistically significant (P > 0.05).
Not every muscle responded to the valgus perturbation during every trial. Nevertheless, muscle activation was different between the groups. Compared with the control group, the OA group had a significantly higher percentage of trials involving activation of the medial hamstrings (97 ± 7.3% of trials among the OA subjects versus 79 ± 29.4% of trials among the controls; P = 0.021) and activation of the sartorius (72.2 ± 34.1% of trials among the OA subjects versus 41.2 ± 35.0% of trials among the controls; P = 0.015).
When we compared the EMG integrals of the individual medial and lateral muscles, the OA subjects appeared to use more medial muscle activity than lateral muscle activity (side-by-muscle interaction P = 0.010), and post-hoc testing revealed that both the hamstrings (P = 0.001) and gastrocnemius (P = 0.011) muscles were more active on the medial side. The control subjects had higher medial muscle activity in the hamstrings muscles only (P = 0.020) (Figure 4).
The level of muscle co-contraction in the medial and lateral muscles during the long loop interval showed a significant side-by-group interaction (P = 0.039) (Figure 5). Post-hoc testing revealed that the control group showed no difference in the level of co-contraction between the medial and lateral sides (P = 0.106) of the joint, whereas the OA group showed significantly higher VMMH than VLLH co-contraction (P = 0.037), although there was no significant difference between the VMMG and the VLLG (P = 0.079).
This study provides insight into the stabilization strategy used by individuals with medial knee OA and genu varum that could hasten joint destruction. These observations have important implications for the prospect of rehabilitation. As we hypothesized, subjects with knee OA responded to a mechanical perturbation that was designed to stress medial periarticular structures and create a potentially destabilizing sensation, with greater co-contraction on the medial side of the joint. Subjects with OA also showed greater axial loading that could result in greater medial joint compression. The OA subjects in this study had more laxity on the medial side of the joint and the majority reported knee instability that interfered with daily activities. Our data, therefore, suggest that the manner in which individuals with medial knee OA attempt to cope with the threat of knee instability might contribute to further joint destruction and propagate a cycle of cartilage deterioration.
Before the plate translated, the individuals with medial knee OA who reported having greater knee stability during daily activities maintained higher medial muscle co-contraction, which might represent a strategy to augment joint stability by increasing reflex-mediated joint stiffness. Others have demonstrated that in both the upper () and lower () extremities, higher background muscle activity yields increased reflexive joint stiffness, potentially providing greater joint stability.
Greater background muscle activity, as seen on the medial side of the joint in the OA group, could increase the reflexive joint stiffness through increased muscle stiffness. Muscle stiffness is controlled in part by the -motoneuron system, which is influenced by afferent information from the muscles, ligament mechanoreceptor, and joint afferents from other structures, including the joint capsule and skin. This afferent input has a strong effect on the -motoneuron system that provides continuous preparatory adjustments to muscle stiffness ([31-34]). Higher muscle stiffness may increase the responsiveness of muscles, a speculation that is supported by our data indicating significantly greater frequencies of responses in the medial hamstrings and sartorius muscles in the OA group.
One factor that could have influenced the responses to the perturbation is the preload imposed on the medial joint structures as a result of the thigh-stabilization system. Tension of the stabilization straps caused the subjects' knees to be 2-3° more valgus than the knee position during typical standing; however, the change in position with the straps was equal in both groups, and therefore its influence should have been equivalent in both groups. Further research is needed to measure muscle stiffness to confirm its relationship to muscle responsiveness, but greater preactivation of the medial quadriceps and hamstring muscles could be an effective adaptation for minimizing instability through increased muscle stiffness ([35-37]).
Our data demonstrate that people with knee OA and genu varum use higher medial co-contraction when under conditions of potentially destabilizing perturbation. Patients with knee OA also use greater magnitudes of muscle co-contraction than do uninjured subjects during functional activities (), presumably to minimize instability. The purpose of our co-contraction equation was to determine how muscles that are antagonists might contribute to joint compression. The level of muscle activation that contributes to joint compression occurs when the relative activation of the muscles is equal and relatively high in amplitude. It should be noted that there are many other methods of assessing co-contraction, including an assessment of the timing of concurrent activity () or ratios of peak muscle activity (). When ratios are used, muscles that are active at equal, but small, magnitudes will appear to have the same level of co-contraction as muscles with an equal, but high, magnitude of EMG activities. In addition, if the peak magnitudes are found at relatively different points in the experimental cycle, they would not be said to co-contract according to our operational definition of co-contraction that requires the simultaneous activation of 2 muscles.
When timing alone is used as a measure of co-contraction, the co-contraction value of a condition when one muscle might have very-low-magnitude EMG that is active concurrently with a muscle with high-magnitude EMG will be equivalent to a condition when 2 muscles are simultaneously active at very high or very low magnitudes. By combining the relative activation of 2 muscles (ratio) and multiplying it by the sum of the magnitudes, our method characterizes the co-contraction that could result in higher joint compression. Although the concurrent activation of quadriceps and hamstrings muscles is capable of resisting frontal plane forces (), large magnitudes of medial co-contraction, as evident in the OA group, may result in greater damaging compressive loads in the knee. It may therefore be more beneficial to retrain muscles to be activated selectively, rather than simultaneously, to minimize joint compression and, perhaps, increase joint stability.
The presence of medial laxity and knee instability may play an important role in the adaptation of muscle-activity patterns in people with medial knee OA. Our subjects with knee OA had both increased medial joint laxity and a sensation of knee instability and demonstrated altered muscle-activation patterns that are consistent with the results reported by others. Shultz et al () demonstrated that uninjured subjects with more anterior-posterior knee laxity used greater muscle preactivation, greater reflex activation, and delayed responses in the muscles that were antagonistic to the imposed perturbation. Although the direction of laxity measurements and testing was performed in a different plane from the testing in our study, the results of the study by Schultz et al underscore the fact that neuromuscular control of the knee can be influenced by joint laxity.
Unlike other studies of muscle activation in response to varus/valgus perturbations (), our testing paradigm included axial loading of the knee joint through weight bearing. The subjects with knee OA maintained more load on the limb during the perturbation, which was unexpected. None of the subjects reported having discomfort during the test; however, they reported experiencing pain and instability during daily activities. We would have expected the subjects to unload the limb more quickly to avoid potential discomfort or instability, but it appears that these subjects with OA were either incapable of unloading the joint quickly or also used axial loading to help stabilize the joint when it was perturbed. If this strategy is one that individuals with OA adopt during functional activities, the increased and prolonged joint loading, coupled with varus alignment, high adduction moments, and higher medial co-contraction, would be particularly detrimental. Rehabilitation may be more effective if it addresses the role of joint loading along with improved neuromuscular responses to help the OA subjects to improve dynamic knee stability.
This speculation is underscored by a recent case report by Fitzgerald et al (), in which agility and neuromuscular training were added to the rehabilitation regimen of a patient who had knee OA and reported having knee instability. They found that after treatment, the patient experienced no more episodes of instability, had higher knee function, and reported having less knee pain during daily activities. Although the mechanisms behind the improved outcome following neuromuscular training in this patient with OA are not known, studies on neuromuscular training in other patient populations () demonstrate the possibility that such training can improve knee stability and promote more selective recruitment of muscles that may preserve joint integrity. In those cases in which muscles cannot be retrained in such a manner, other investigators have advocated the use of orthoses such as unloading braces () and heel wedges () to minimize medial joint loading, which may reduce pain and allow higher function in some patients.
Muscle strength has long been an important part of the management of people with knee OA ([48-51]), but Sharma et al () recently reported that strong quadriceps muscles are associated with a greater likelihood of progression of knee OA in patients with knee malalignment or excessive frontal plane laxity. This is in contrast to the findings of Slemenda et al (), who found that even relatively small increases in quadriceps femoris strength, particularly in women, was predictive of a 20-30% decrease in the odds of having OA, suggesting that quadriceps strength is an integral component for preventing knee OA. The lack of consensus on the role of the quadriceps muscles may be due to the emphasis on muscle strength rather than the manner in which patients activate the muscles surrounding the knee. Knee instability has only recently been identified as an important aspect of the disease process ().
Our work suggests that when treating patients with knee OA, knee instability is a problem that, if left untreated, might lead to the development of a neuromuscular joint-stabilization strategy that could hasten progression of the disease. Continued research is needed to identify whether increased co-contraction leads to higher muscle stiffness or more joint loading, and to understand the long-term effect of altered muscle activity on the progression of OA. Nevertheless, the results of this study suggest that the relationship between muscle function and knee OA is highly complex. Therefore, knowledge of the strategies that improve knee stability and function is important for the development of treatment approaches that could improve outcomes in individuals with medial knee OA.
The authors wish to thank William Newcomb, MD for his valuable assistance with subject recruitment, Laura C. Schmitt, MPT for her assistance with data collection, and Laurie Andrews, RTR for assistance with the stress radiographs.