|
|
< Previous Article | Next Article > |
Save Article to My Profile Download Citation
Abstract | References | Full Text: HTML , PDF (271k)
|
Research Article
*Correspondence to Annegret Mündermann, Department of Mechanical Engineering, Durand Building 205, Stanford University, Stanford, CA 94305-4038 Funded by:
Received: 24 January 2005; Accepted: 3 June 2005
10.1002/art.21262 About DOI
Osteoarthritis (OA) is a degenerative joint disease that affects an increasing proportion of the population ([1-3]). Although most joints of the lower extremity, including the ankle and hip, may be involved, the knee is the most common site for OA ([4]). Changes related to OA are more frequently observed in the medial compartment than in the lateral compartment of the knee ([5]). Moreover, loads transferred through the medial compartment during walking are substantially higher than loads transferred through the lateral compartment ([6]). The distribution of loads transferred through the medial and lateral compartments during walking can be estimated by the external knee adduction moment ([6]); a higher external knee adduction moment indicates greater loads in the medial than in the lateral compartment. The first peak knee adduction moment during walking has been shown to be a strong predictor of the presence ([7-9]), severity ([10][11]), and rate ([12]) of progression of medial compartment knee OA. However, little attention has been paid to the changes in the mechanical environment of other joints of the affected limb, which presumably occur concomitantly with changes in knee-joint mechanics. Most studies investigating the gait of patients with knee OA have concentrated on kinematics and kinetics at the knee ([7][8][13-16]), ground reaction forces ([7][8][15][17]), and the sagittal plane ([13][17-20]). Moreover, study populations have typically been small ([8][15][20]) and involve a range of OA stages, from early OA ([15]) to severe OA ([18]). In addition, knee OA has been used as a generic inclusion criterion ([13][14][16][17][19][21]); that is, no distinction was made between patients with OA affecting the medial knee compartment and those with knee OA in the lateral compartment. Nevertheless, these previous studies indicate that patients with knee OA seem to experience a smaller range of knee flexion during the stance phase of walking ([13][14][16][18][21]), which is associated with a smaller net quadriceps moment ([14][16]). However, such differences in sagittal-plane knee kinematics and kinetics could be caused primarily by slower walking speeds ([22][23]), as has been reported for patients with knee OA in most gait studies ([8][14][16-18][21]). Similarly, it is likely that the observed reduction in ground reaction forces ([8][17][21]) in patients with knee OA is attributable to slower walking speeds. Hurwitz et al ([24]) reported higher knee adduction moments in patients with medial compartment knee OA who were receiving pain medication. Thus, patients seem to adopt a gait pattern in an attempt to unload the affected structures during walking, possibly by changing moments at the adjacent ankle and/or hip. Moreover, patients with less severe knee OA adopt a gait pattern that differs from that of patients with more severe OA and control subjects, which is indicative of the potential to reduce the adduction moment when walking at slower speeds ([10]). These gait changes appear to be associated with increased loading rates, defined as increased slopes of the ground reaction force curve immediately following heel strike during walking ([15][21]). It is likely that the increased loading rate at the foot is transferred, to some extent, to the joints proximal to the foot-ground interface. If this is the case, then patients with knee OA may experience not only greater relative loads on the medial compartment of the knee, but also a more rapid increase in axial force at the ankle, knee, and hip. A higher loading rate at the tissue level may lead to the initiation ([25]) or propagation ([26]) of surface fissures in cartilage, similar to that seen in OA cartilage, which may ultimately lead to a faster rate of progression of OA. Thus, secondary gait changes in patients with medial compartment knee OA may be associated with increased load not only on knee cartilage, but also on cartilage at the ankle and the hip. The purpose of this study was to investigate the loading environment of the major joints of the lower extremity during walking in patients with medial compartment knee OA of varied severity. We hypothesized that gait changes related to knee OA of varied severity are associated with increased loads at the ankle, knee, and hip, and that most changes will occur in the frontal plane. PATIENTS AND METHODS
Patients. Forty-two patients with bilateral OA in the medial compartment of the knee participated in this study (Table 1) after giving their written consent in accordance with policies of the institutional review board. All subjects fulfilled all inclusion criteria for participation in this study, as follows: definite osteophyte presence in the medial or lateral tibiofemoral compartment, a smaller interbone distance at the narrowest point of the medial compartment compared with the lateral compartment, pain in and around at least 1 knee for most of the days in the previous months, and at least some difficulty with 2 or more items in the Western Ontario and McMaster Universities OA Index physical function scale ([27]). Twenty-two exclusion criteria for patients with knee OA were used: rheumatoid or other systemic inflammatory arthritis, avascular necrosis, periarticular fracture, Paget's disease of bone, villonodular synovitis, chronic knee-joint infection, ochronosis, neuropathic arthropathy, acromegaly, hemachromatosis, Wilson's disease, osteochondromatosis, gout or recurrent pseudogout, osteopetrosis, total knee replacement in either knee, flexion contracture >15° in either knee, OA grade higher than moderate (on a scale of none, mild, moderate, and severe) by examination (history and physical examination) of either ankle or either hip (using the American College of Rheumatology criteria [28]), morbid obesity (body mass index [BMI] >45 kg/m2), intraarticular corticosteroid injection within the previous 2 months, knee surgery within the previous 6 months, plan for total knee replacement within the next year, and hip or spine disease as the major source of disability.
Control subjects. For each patient, an asymptomatic control subject, matched for sex, age (±5 years), height (±5 cm), and mass (±5 kg), was selected and included in the study after their written consent was provided in accordance with the policies of the institutional review board. All asymptomatic control subjects (Table 1) had no clinical diagnosis of OA or rheumatoid arthritis or a history of knee trauma or pain. The patient and control groups did not differ in age, height, mass, or BMI (Table 1). None of the control subjects had previously been treated for any clinical lower back or lower extremity condition and none had any activity-restricting medical or musculoskeletal condition. Clinical assessment. The Kellgren/Lawrence (K/L) radiographic severity grades for both knees of patients with medial compartment knee OA were determined based on clinical and radiographic data (0 = no osteophytes, 1 = possible osteophyte lipping, 2 = definite osteophytes and possible joint space narrowing, 3 = moderate multiple osteophytes and definite joint space narrowing, as well as some sclerosis and possible bone contour deformity, 4 = large osteophytes, marked joint space narrowing, severe sclerosis, and definite bone contour deformity [29]). Patients were classified as having less or more severe knee OA based on the K/L grades of both knees. The less severe OA group consisted of patients with a K/L grade of 2 for either knee. Patients with at least 1 knee assigned a K/L grade 3 were placed in the more severe OA group. The mechanical axes of all knee joints were measured by the same rheumatologist on a single weight-bearing radiograph that included the hip, knee, and ankle; the axes were determined as the angle between a line from the center of the femoral head to the center of the femoral intercondylar notch, and a line from the center of the tips of the tibial spines to the ankle talus ([30]). To minimize measurement variation related to limb rotation, each patient was positioned with the tibial tubercle in the anterior view. Neutral alignment was defined as zero, varus alignment as positive angles, and valgus alignment as negative angles. Gait analysis. Patients and control subjects performed walking trials in their own low-top, comfortable walking shoes. All subjects were instructed to walk at their self-selected normal speed (Table 1). The approach used to collect kinematic and kinetic data is identical to that described in previous investigations ([31][32]). Briefly, reflective markers were placed on the leg along the superior iliac spine, greater trochanter, lateral joint line of the knee, lateral malleolus, lateral aspect of the calcaneus, and head of the fifth metatarsal. Marker data were obtained using 4 high-speed cameras (120 frames/second, MCU240; Qualisys Medical, Gothenburg, Sweden). Data on the ground reaction force were collected using a force platform (sampling frequency 120 Hz; Bertec, Columbus, OH) that was placed in the center of the walkway, level with the ground. Each limb segment (thigh, shank, and foot) was idealized as a rigid body with a local coordinate system that was defined to coincide with a set of anatomic axes. Intersegmental angles, external moments, and forces were calculated from the position of the markers, measurements of the ground reaction force, and properties of the limb segment mass/inertia. The angle, force, and moment at the knee were resolved into a coordinate system fixed in a tibial reference system, with axes defining flexion-extension, abduction-adduction, and internal-external rotation. Similarly, the angle, force, and moment at the ankle were resolved into a coordinate system fixed in a foot reference system, with axes defining dorsiflexion-plantar flexion, eversion-inversion, and adduction-abduction. The angle, force, and moment at the hip were resolved into a coordinate system fixed in a thigh reference system, with axes defining flexion-extension, abduction-adduction, and internal-external rotation. The axial forces at the ankle, knee, and hip were defined as the resultant intersegmental forces at these joints, resolved along the long axis of the distal segment (foot, shank, and thigh, respectively), taking into account the ground reaction force, the weight of the distal segment, and inertial forces ([33]). The mediolateral ground reaction force is related to the acceleration of the center of gravity in the mediolateral direction, which contributes to the external knee adduction moment at the knee. Discrete variables describing peak values of the ground reaction force and kinematics and kinetics for each joint in 3 dimensions (Table 2) were calculated using an in-house algorithm written in Mathematica version 4.1 (Wolfram Research, Champaign, IL). Forces were normalized to body weight (with results expressed as %Bw), and moments were normalized to body weight and height (with results expressed as %Bw·Ht) to allow for comparison between subjects. The average values from 3 trials at the self-selected walking speed were calculated for each joint, which then allowed for comparison of the average values for each patient and control subject.
Statistical analysis. Multivariate analysis of variance (MANOVA) was used to detect an overall significant difference in gait patterns between patients with medial compartment knee OA and matched control subjects and between patients with less severe (K/L grades 2) and more severe (K/L grades 3) knee OA (defined from the more severely affected knee of each patient). When a significant result was obtained by MANOVA, separate mixed-factor univariate analysis of variance (ANOVA) models, with severity as the intersubject factor (less severe versus more severe OA) and presence of OA as the intrasubject factor (repeated measure to account for the fact that patients and control subjects were matched 1:1), were used to detect significant differences in discrete variables that describe the 3-dimensional intersegmental angles, moments, and forces between patients with medial compartment knee OA and matched control subjects and between patients with less severe (K/L grades 2) and more severe (K/L grades 3) knee OA. The significance level for all ANOVAs was set at 5%. Student's paired and independent-sample t-tests were used for post hoc analysis to detect significant differences between patients with knee OA and control subjects and between patients with less severe and more severe knee OA, respectively. The significance level for the Student's t-tests was adjusted to account for multiple comparisons ( = 0.025). RESULTS
Gait and alignment. The general gait pattern differed between patients with knee OA and matched control subjects (P < 0.001 by MANOVA, >99.9% power) despite the fact that they walked at similar speeds (Table 1). Patients with more severe knee OA had a 6.0° greater varus mechanical axis alignment than did patients with less severe knee OA (5.7° versus 0.3°; P < 0.001). Kinematics. The angle of knee flexion at heel strike was different between patients and matched control subjects. All patients with knee OA made initial contact with the ground with the knee in a 5.3° more extended position than that of the control subjects. This difference was more pronounced in patients with less severe knee OA (for patients versus controls, 1.8° versus 5.3°) than in patients with more severe knee OA (for patients versus controls, 0.9° versus 4.6°) (P = 0.003 for both). All other sagittal plane angles at the ankle, knee, and hip were similar between all groups. Joint moments. All patients with knee OA walked with 18.1% greater hip flexion moments during terminal stance compared with that of their matched control subjects (for patients with less severe OA versus controls, 4.0% Bw·Ht versus 3.1% Bw·Ht; for patients with more severe OA versus controls, 3.8% Bw·Ht versus 3.5% Bw·Ht [P = 0.011 for both]). All other sagittal plane moments at the ankle, knee, and hip were similar between all groups. Following heel strike, maximum abduction moments at the knee and at the hip were increased 93.3% and 100.7%, respectively, in all patients with knee OA compared with their matched controls (P < 0.001) (Figures 1 and 2). At the knee, the difference in maximum abduction moment compared with that in matched control subjects was greater in patients with less severe knee OA (+162.9%) than in patients with more severe knee OA (+45.5%).
During the mid-stance and terminal-stance phases, patients with more severe knee OA had significantly greater first peak adduction moments at the knee than their matched control subjects (+11.4%; P = 0.039) and than patients with less severe knee OA (+27.9%; P < 0.001) (Figure 1). Patients with less severe knee OA had significantly lower second peak adduction moments at the knee than their matched control subjects (-32.8%; P = 0.001) and than patients with more severe knee OA (-37.8%; P < 0.001). Patients with more severe knee OA had significantly lower first and second peak adduction moments at the hip compared with their matched control subjects (first peak -20.3%; P = 0.024; second peak -29.0%; P < 0.001) (Figure 2). In terminal stance, all patients with knee OA had 18.2% smaller maximum inversion moments at the ankle (for patients with less severe OA versus controls, 1.6% Bw·Ht versus 1.9% Bw·Ht; for patients with more severe OA versus controls, 1.7% Bw·Ht versus 2.0% Bw·Ht [P = 0.017 for both]). Ground reaction and intersegmental axial forces. Shortly after heel strike, the lateral ground reaction force in all patients with knee OA was 54.0% higher than that in control subjects. In addition, the magnitude of the increase was related to severity (for patients with less severe OA versus controls, 5.7% Bw versus 3.8% Bw; for patients with more severe OA versus controls, 7.2% Bw versus 4.6% Bw [P < 0.001 for both]). The vertical loading rate, or the slope of the ground reaction force, in all patients with knee OA was elevated by 50.1% compared with that in the matched control subjects (for patients with less severe OA versus controls, 1,176.5% Bw/second versus 746.0% Bw/second; for patients with more severe OA versus controls, 1,158.7% Bw/second versus 794.8% Bw/second [P < 0.001 for both]). The increased vertical loading rate was consistent with increased intersegmental axial loading rates at all joints of the lower extremity, with a 64.4%, 55.5%, and 59.2% increase in the axial loading rate at the ankle, knee, and hip, respectively (P < 0.001) (Figure 3).
Maximum axial forces at the ankle, knee, and hip were only slightly different between patients with knee OA and matched control subjects (at the ankle +4.8% [P = 0.029]; at the knee +4.0% [P = 0.060]; at the hip 9.0% [P = 0.020]). Moreover, the maximum axial forces at the ankle, knee, and hip were unaffected by the severity of the disease (P > 0.867 between patients with less severe and more severe knee OA). DISCUSSION
This study demonstrated that gait changes related to medial compartment knee OA are dependent on the severity of the disease. We hypothesized that gait changes related to knee OA of varied severity are associated with increased axial loading at the ankle, knee, and hip, and that most changes will occur in the frontal plane. The results of this study support our hypothesis, since most changes were observed in the frontal plane moments and forces in the joints of the lower extremity. Patients with medial compartment knee OA and their matched control subjects walked at similar speeds (Table 1), and thus differences in gait patterns cannot be attributed to differences in walking speeds. Similarly, pain ratings and patellofemoral joint space narrowing were similar among patients with less severe knee OA and patients with more severe knee OA (Table 1), and thus differences in gait patterns between patient groups cannot be attributed to differences in pain level or patellofemoral OA. All patients with knee OA landed with the knee in a more extended position and experienced a more rapid increase in the ground reaction force that was reflected in greater vertical loading rates. A more rapid increase in ground reaction force indicates a more rapid shift of the body's weight from the contralateral limb to the support limb. The greater knee and hip abduction moments immediately following heel strike in all patients with knee OA (Figures 1 and 2) suggest that patients exert greater hip adductor muscle forces during this period to move their trunk laterally. To accomplish a more lateral trunk motion, the foot would exert a medial force on the ground, which would be reflected in a greater subsequent ground reaction force during this period, as observed in this study. This change in loading pattern is a potential mechanism of gait compensation used by patients with knee OA to reduce the mediolateral distance between the center of mass and the knee joint center, thereby reducing the moment arm of the ground reaction force and supposedly reducing the knee adduction moment at a later point in the stance phase. Differences in the ankle inversion moment between groups were much smaller than differences in the knee and hip adduction moments. Moreover, the differences in the ankle inversion moment occurred during the second half of the stance phase, while the peak adduction moment at the knee occurs in the first half of the stance phase. These 2 observations suggest that the potential mechanism of gait compensation is more likely driven by changes in trunk acceleration than by changes in ankle-joint kinetics. Patients with less severe knee OA (K/L grades 2) had hip adduction moments throughout the stance phase that were similar to their matched control subjects (Figure 2). The external hip adduction moment is balanced by the hip abductor muscle moment, suggesting that patients with less severe knee OA have sufficiently strong hip abductor muscles to maintain the altered position of the trunk, resulting in similar first peak knee adduction moments and reduced second peak knee adduction moments compared with their matched control subjects (Figure 1). Thus, patients with less severe knee OA seem to adopt a strategy of gait compensation that allows them to control and lower the load at the medial compartment of the knee (Figure 4), thereby reducing their risk for progression of their disease ([12]).
In contrast, patients with more severe knee OA (K/L grades 3) had substantially lower hip adduction moments throughout the stance phase compared with their matched control subjects (Figure 2). Thus, it is likely that patients with more severe knee OA lack the hip abductor muscle strength required to maintain the altered position of the trunk. In this patient group, the contralateral hip may drop throughout the swing phase, leading to a lateral movement of the trunk away from the support limb, which is similar to what is clinically referred to as Trendelenburg gait. The increased mediolateral distance between the center of mass and the knee-joint center and the greater varus alignment in these patients presumably result in higher first peak knee adduction moments (Figure 1). It has been suggested that hip adductor muscles may have the capability to stabilize knees with medial compartment OA or knees with varus alignment ([34]). Yamada et al showed that hip adductor muscles are stronger with increasing OA severity, and with greater varus alignment with similar hip abductor muscle strength ([34]). Thus, although patients with more severe knee OA may adopt a strategy of gait compensation similar to that demonstrated by patients with less severe knee OA, those with more severe knee OA may require a net hip abductor moment not only for maintaining a level pelvis, but also for knee varus control. In that case, even with gross abductor moments similar to those in patients with less severe disease and to those in control subjects, they may not be able to maintain a level pelvis, so that patients with more severe knee OA are unable to control or lower the load at the medial compartment of the knee (Figure 4), thereby increasing their risk for progression of their disease ([12]). This proposed strategy of gait compensation in patients with medial compartment knee OA is supported by the fact that the magnitude of the first peak knee adduction did depend, in part, on the magnitude of the hip abduction moments (R2 = 0.249, P < 0.001 by linear regression analysis). This strategy may come at the cost of increased load at adjacent joints; that is, all patients with knee OA walked with substantially increased vertical loading rates (ground reaction force) and axial loading rates (at the ankle, knee, and hip) (Figure 3). Radin et al ([15]) reported increases in the vertical loading rate in patients with presumably pre-OA changes because of mild, intermittent, and activity-related knee pain, and yet the reported increases were smaller than the increases observed in the present study. Radin et al referred to these observed differences in gait between patients with knee OA and control subjects as microklutziness and compared the sequence of consecutive steps with this pattern to repetitive impulse loading. In the present study, intersegmental resultant joint forces were estimated based on the ground reaction force, the weight of the distal segment, and inertial forces. The results of this analysis show that the increase in loading rate immediately following heel strike is transferred through all joints of the lower extremity and can also be observed at the hip. The axial hip loading rates reported in this study are similar to values reported by Bergmann et al using instrumented hip endoprostheses ([35]), even though Bergmann et al measured the stem-cup contact force at the hip that also takes into account muscle forces. Moreover, the percentage differences in axial hip loading rate between patients and matched control subjects in the present study are only slightly larger than those between different footwear conditions, measured with an instrumented hip endoprosthesis ([35]). Similar to the findings by Bergmann et al, the percentage changes in loading rate in this study were much greater than those in joint force. Thus, although the joint forces reported in the present study were measured indirectly and represent resultant joint forces, the axial loading rates appear to be good estimates of the actual loading rates at the joint. Changes in the axial loading rate at the ankle, knee, and hip may be harmful to cartilage. Indeed, higher loading rates have been shown to generate more surface fissuring of cartilage than lower loading rates ([25]), and surface fissures in the cartilage can propagate mechanically if the joint surface is subjected to rigorous repetitive loading ([26]). Thus, increased loading rates may not only accelerate fibrillation of already damaged cartilage, as observed in severe knee OA, but may also lead to the initiation of OA at adjacent joints, including the hip. However, although 37.6% of patients with radiographic knee OA also have radiographic hip OA ([36]), information is lacking on the relationship between gait changes associated with medial compartment knee OA and the development of secondary hip OA. The anatomy of the tibiofemoral joint is uniquely disparate from that of other synovial joints, including the ankle and hip, because the tibia and femur articulate in 2 compartments (medial and lateral), the menisci increase the contact areas, the ligaments mainly guide joint motion, and the muscles stabilize the joint in static and dynamic situations. This bicondylar configuration makes the knee particularly susceptible to acute and chronic changes in the mechanical environment associated with injury (e.g., rupture of the anterior cruciate ligament) and with aging or menopause ([37]), since applying a greater moment in abduction or adduction direction to a bicondylar joint will merely shift the net contact point toward one compartment, and thus always increase the load in one compartment. In contrast, the joints of the ankle-joint complex (subtalar and talocrural joints) and the hip of a healthy person have highly congruent articulating surfaces. Articulating bones contact in one area, and thus can be idealized as hinge or ball-and-socket joints. Joint moments can be described by a force couple, and applying a greater moment to a ball-and-socket joint or about the sole axis of rotation of a hinge joint does not necessarily increase contact forces. Thus, a change in joint moments at the hip may not be as critical as a change in joint forces or axial loading rates. Therefore, from a biologic and mechanical perspective, the increases in the loading rate of the ground reaction force and the axial loading rate at the ankle, knee, and hip cannot be neglected. Current interventions for the treatment of medial compartment knee OA aim to reduce the first peak knee adduction moment to reduce the load transferred through the medial compartment of the knee. These load-modifying interventions include bracing ([38][39]), footwear modifications ([40-42]), gait training ([10]), and quadriceps muscle strengthening ([43-47]). However, it has been shown that footwear modification may also alter the loading rate of the ground reaction force during walking ([48][49]). Thus, in order to develop effective strategies for the treatment of medial compartment knee OA, interventions that solely focus on reducing the load distribution at the knee should be evaluated not only for their primary effects on knee adduction moments, but also for their secondary effects. The most successful intervention for slowing the rate of progression of medial compartment knee OA may be an intervention that assists the natural compensation strategy used by patients with medial compartment knee OA. To date, the role of core- and hip-muscle strength in the treatment and prevention of knee OA has not received any attention. The major kinematic and kinetic gait differences between patients with medial compartment knee OA and control subjects are generally observed in the frontal plane ([7-12]). However, physical therapy ([43-47]) has concentrated on strengthening the quadriceps muscle, which, while stabilizing the joint, acts primarily in the sagittal plane. The observation that patients with more severe knee OA have smaller hip adduction moments and greater knee adduction moments during walking suggests that strengthening hip abductor muscles, especially in patients with more severe knee OA, may be an effective, yet noninvasive, treatment for medial compartment knee OA. In summary, patients with medial compartment knee OA walk with a different gait pattern than do matched control subjects. All patients landed with the knee in a more extended position, experienced a more rapid increase in the ground reaction force, had greater knee and hip abduction moments, and had greater lateral ground reaction force, indicating a more rapid shift of the body's weight from the contralateral limb to the support limb and a lateral shift of the trunk. This change in loading pattern is a possible strategy of gait compensation used by patients with knee OA to reduce the mediolateral distance between the center of mass and the knee, thereby reducing the moment arm of the ground reaction force and supposedly reducing the knee adduction moment at a subsequent point during stance. This strategy appears to be successful only in patients with less severe knee OA. Gait changes appear to be initiated before heel strike and this not only affects the load distribution between the medial and lateral compartments of the knee, but also leads to increased axial loading rates at all joints of the lower extremity. More rapidly increasing joint forces may lead to a more rapid progression of existing OA and to the initiation of OA at joints adjacent to the knee. Thus, interventions that are aimed at slowing the rate of progression of OA should be assessed for their effects on the joint mechanics not only of the knee, but also of all joints of the lower extremity. Acknowledgements
The authors express their appreciation to Drs. Debra Hurwitz and Leena Sharma for their contribution to the collection of gait and clinical data, and to the NIH for partial funding of this project.
|
|