Effect of baseline quadriceps activation on changes in quadriceps strength after exercise therapy in subjects with knee osteoarthritis
To examine whether pretreatment magnitude of quadriceps activation (QA) helps predict changes in quadriceps strength after exercise therapy in subjects with knee osteoarthritis (OA). We hypothesized that subjects with lower magnitudes of QA (greater failure of muscle activation) would have smaller gains in strength compared with those with higher magnitudes of QA following exercise therapy.
One hundred eleven subjects with knee OA (70 women) participated. Baseline measures included demographic information, quadriceps muscle strength, and QA using a burst-superimposition isometric torque test. Following baseline testing, subjects underwent a 6-week supervised exercise program designed to improve strength, range of motion, balance and agility, and physical function. On completion of the program, quadriceps strength and QA were reassessed. Multiple regression analysis was used to determine whether baseline QA predicted quadriceps strength scores at the 2-month followup.
Bivariate correlations demonstrated that baseline QA was significantly associated with quadriceps strength at baseline (ρ = 0.30, P < 0.01) and 2-month followup (ρ = 0.23, P = 0.01). Greater magnitude of baseline QA correlated with higher strength. While controlling for baseline quadriceps strength and type of exercise therapy, the level of QA did not predict quadriceps strength at the 2-month followup (β = −0.04, P = 0.18).
Baseline QA did not predict changes in quadriceps strength following exercise therapy. Measurement of QA using the central activation ratio method does not appear to be helpful in identifying subjects with knee OA who will have difficulty improving quadriceps strength with exercise therapy.
Knee osteoarthritis (OA) is a common chronic condition affecting more than 4.3 million older adults in the US (1). It is a major cause of pain and functional impairment, including difficulty with several activities of daily living (1–3). Weakness of the quadriceps muscle is well documented in subjects with knee OA (4–6), is strongly associated with pain, and is an important determinant of disability (4, 7). Quadriceps strength is an important factor to target because weakened quadriceps muscles may increase joint stresses from decreased ability to attenuate loads across the joint (8). Additionally, quadriceps weakness may play a role in the etiology and progression of OA (6, 8, 9).
Multiple factors play a role in the etiology of muscle weakness. In addition to pain and disuse atrophy, reduced quadriceps activation (QA) has been suggested to contribute (4, 8, 10). The presence of reduced QA (failure to fully activate the muscle) is well established in subjects with knee joint effusions (11, 12) and traumatic injuries (10, 13–17), and also can be present in subjects with knee OA even in the absence of pain and effusion (4, 5, 7, 18). The proposed mechanism for the etiology of reduced QA in subjects with knee OA is a damaging cycle. Degenerative changes in the joint may damage articular mechanoreceptors, which leads to abnormal processing of sensory information and inhibition of muscle activation. This, in turn, predisposes the quadriceps to weaken, which increases the risk for more joint damage (4, 8). Recently, it has been suggested that reduced QA may have an impact on physical function in subjects with knee OA by moderating the relationship of quadriceps strength to function (19). Subjects with low strength and reduced QA had lower function than those with similar strength but high QA (19).
Although exercise leads to increased quadriceps strength along with improved pain and function (20, 21), strength deficits remain and overall effect sizes have been variable (10, 22). It is possible that the presence of reduced QA could contribute to this lack of robust response. A few studies have suggested some improvement in QA with exercise alone, but results have been inconclusive, and there has not been an attempt to examine how an individual's pre-therapy level of QA might affect or predict their degree of response to therapeutic exercise (10, 23, 24). It is possible that reduced QA might not allow an individual the capacity to produce enough tension in the muscle during exercise to achieve an exercise training effect, which in turn may limit their responsiveness to an exercise program.
We are specifically interested in looking at how pre-therapy QA levels may affect subjects' response to therapeutic exercise in terms of quadriceps strength, which is a key determinant of function in subjects with knee OA. Therefore, the aim of this study was to examine whether pretreatment magnitude of QA helps predict changes in quadriceps strength after exercise therapy in subjects with knee OA. We hypothesized that subjects with lower magnitudes of QA (greater failure of muscle activation) would have smaller gains in strength compared with those with higher magnitudes of QA following exercise therapy.
SUBJECTS AND METHODS
The data reported here were drawn from an ongoing randomized clinical trial comparing 2 exercise rehabilitation regimens for subjects with knee OA. The current longitudinal study looks at associations using baseline and 2-month followup data.
One hundred forty-two subjects with knee OA were eligible for the study. Subjects were eligible if they completed baseline measurement of quadriceps strength and QA, 12 sessions of exercise therapy over 6 weeks, and a strength measurement 2 months after baseline. Subjects were included in the study if they were ≥40 years of age, met the 1986 American College of Rheumatology clinical criteria for knee OA (25), and had grade ≥2 Kellgren/Lawrence radiographic changes in the tibiofemoral joint (26). Subjects were excluded from the study if they 1) had conditions that would place them at risk for injury during the exercise training program (e.g., required an assistive device for ambulation, had a history of ≥2 falls in the previous year, or were unable to ambulate 100 feet independently), 2) had undergone total knee arthroplasty, 3) exhibited uncontrolled hypertension, 4) had a history of cardiovascular disease, 5) had history of neurologic disorders that affect lower extremity function (e.g., stroke, peripheral neuropathy), 6) had conditions that would place them at risk for reinjury during quadriceps strength testing (e.g., recent corticosteroid injection to the quadriceps or patellar tendons, quadriceps or patellar tendon rupture, patellar fracture), or 7) reported vision problems that affected performance of basic mobility tasks. All subjects signed an informed consent form approved by the University of Pittsburgh Institutional Review Board prior to participation in the study.
Exercise therapy intervention.
Subjects were randomized to receive 1 of 2 types of exercise therapies. The standard exercise therapy consisted of lower extremity stretching, range of motion and strengthening exercises, and aerobic exercise. The experimental exercise therapy included all of the above plus dynamic agility and balance training activities (see Supplementary Table 1, available on the Arthritis Care & Research Web site at http://www3.interscience.wiley.com/journal/77005015/home). Physical therapists supervised 12 sessions of therapy over 6 weeks.
Data for each subject were collected during 2 testing sessions. Baseline testing included completion of demographic and health history questionnaires, knee radiographs, and measurement of quadriceps muscle strength and QA. Two months from baseline, testing consisted of measurement of quadriceps muscle strength.
Quadriceps strength and the magnitude of QA were measured using a burst-superimposition maximum isometric quadriceps torque test. This procedure has been shown to yield reliable quadriceps muscle torque measurements by other investigators (27) and in our own laboratory (intraclass correlation coefficient [ICC] 0.97 for intratester reliability between 1 and 3 days, and ICC 0.82 for same-day intertester reliability). Because most subjects had bilateral knee involvement, an involved-to-uninvolved limb comparison was not possible. Therefore, we elected to test the limb that subjects reported as being the most symptomatic limb with regard to pain and functional limitation.
Subjects sat on an isokinetic dynamometer (Biodex System 3 Pro, Shirley, NY) with the test knee positioned in 60° of flexion. Electrodes were placed proximally over the vastus lateralis muscle belly and distally over the vastus medialis muscle belly. A thigh strap, a waist strap, and 2 chest straps were then secured to stabilize the subject in the dynamometer chair.
Once the subject was prepared for testing, we employed a process of potentiating the quadriceps muscles to maximize the subject's ability to produce maximum torque output (28). In addition, this process familiarized the subjects with both the electrical stimulus to be used during testing and the maximum voluntary isometric torque test procedure, which would help minimize the potential for learning effects on the test results. During the first step in the process, subjects practiced producing 3–5-second voluntary isometric quadriceps contractions against the force arm of the dynamometer at 50%, 75%, and 100% effort. Next, subjects received 3 successive trains of electrical stimulation (pulse duration = 600 μsec, pulse interval = 10 msec, train duration = 100 msec), separated by 30-second time intervals, applied to the resting muscle at amplitudes of 40V, 60V, and 100V. If the 100V stimulus did not produce ≥25% of the practice maximum voluntary torque, we increased the stimulus to 130V to ensure the stimulus would be adequate to show failure of muscle activation if it existed. If the resting electrically stimulated torque did not achieve a value of ≥25% of the practice maximum voluntary torque after increasing the stimulus amplitude to 130V, we concluded the stimulus was not adequate for accurate QA measurement and did not include the subject's data in the final analysis.
Following the series of electrical stimuli, formal measurements of maximum voluntary isometric quadriceps strength and QA were initiated. Subjects were asked to exert as much force as possible while extending the knee against the fixed force arm of the dynamometer and to hold each maximal contraction for 3–5 seconds. During the maximal contraction, the train of electrical stimuli (amplitude = 100V or 130V, pulse duration = 600 μsec, pulse interval = 10 msec, train duration = 100 msec) was applied to determine the extent of muscle activation. To maximize their ability to produce maximum torque output during the test, the examiner provided intense verbal encouragement to subjects and provided them with real-time visual feedback of the torque trace displayed on a computer monitor. A torque target line, placed at a torque level slightly greater than the peak torque produced during the practice maximum voluntary isometric contraction, was also visible on the computer monitor. If subjects exceeded this torque target during a given trial, the target was reset at a higher level for the next trial. To optimize the measurement of a maximum contraction, subjects completed 3–6 trials with 1–1.5 minutes of rest between trials until the level of voluntary torque produced decreased compared with the prior trials.
The magnitude of QA was calculated using the quadriceps central activation ratio (CAR), which is a ratio of the highest maximum voluntary torque produced prior to delivery of the electrical stimulus divided by the highest torque produced when the electrical stimulus was superimposed on the maximum voluntary contraction (29). Full QA is represented by a CAR equal to 1 because the superimposed electrical stimulus does not result in a further increase in torque compared with the maximum voluntary torque. When failure of full QA is present, the electrical stimulus will recruit previously inactive muscle fibers to fire, and the torque produced with the superimposed electrical stimulus will surpass that of the maximum voluntary torque, yielding a CAR <1.
To ensure adequate measurement of QA, the data had to meet certain criteria. In addition to recording the maximum torque produced before and during the delivery of electrical stimulus, we recorded the magnitude of the torque just prior to the time the stimulus was delivered to determine whether there was a drop from the maximum torque. A significant drop in torque before the electrical stimulus was delivered could affect the accuracy of the QA calculation. We used the data from the trial with the highest maximum voluntary torque prior to the delivery of the electrical stimulus, as long as the subject was able to maintain that torque within 5% at the time that the electrical stimulus was applied. If the drop in torque was >5% at the time that the electrical stimulus was delivered, we used values from the next highest maximum voluntary torque trial with a drop ≤5%, as long as the voluntary torque achieved was ≥95% of the overall maximum voluntary torque. If there were no trials with a maximum voluntary torque within 95% of the overall voluntary maximum and with a drop in torque ≤5% when the electrical stimulus was delivered, we concluded that the data were not appropriate for accurate QA measurement and excluded that subject from the analysis. Our outcome measure of strength at the 2-month followup was simply the highest maximum voluntary torque produced out of all of the trials performed at the 2-month followup testing session.
Other factors potentially associated with QA and quadriceps strength.
We recorded several other factors that could potentially be associated with the level of QA and quadriceps strength for consideration as covariates. Demographic factors included sex, age, height, weight, and number of years since a diagnosis of knee OA. Clinical factors included knee pain during the burst-superimposition testing procedure, the severity of radiographic knee OA, and medication use. Knee pain during the test was measured using a verbal 0–10 numeric pain scale, with 0 representing no pain and 10 representing the worst pain imaginable. The severity of radiographic knee OA was rated by an experienced rheumatologist, using the method described by Kellgren and Lawrence (26). Medication use related to the management of knee OA was determined by the information collected on the demographic and health history questionnaire. Additionally, we recorded the torque produced when the electrical stimulus was applied to the resting muscle (resting e-stim torque).
We first examined descriptive statistics to assess for outliers and data distributions. We then calculated bivariate correlation coefficients to look at associations among quadriceps strength, QA, and potential covariates. Pearson's correlation coefficients were used between normally distributed continuous variables and Spearman's rho coefficients were used for categorical and non-normally distributed continuous variables. We planned to use the variables that were concomitantly associated with QA and quadriceps strength as covariates in the multivariable regression analysis.
Next we performed multivariable regression to test the hypothesis that the magnitude of pre-therapy QA might affect changes in quadriceps strength following a regimen of exercise therapy. Quadriceps strength at the 2-month followup was our outcome variable. In the first step of the regression we controlled for baseline quadriceps strength and type of exercise therapy. Next we entered our predictor variable, the CAR, as a measure of QA. Statistical significance was determined using an alpha level of 0.05. Regression coefficients and standardized β coefficients for each variable in the final model were calculated, and the significance of each was tested under the null hypothesis that the coefficient was not different from zero. Regression diagnostics were performed to assess whether our model met the linear regression assumptions of normality, homoscedasticity, and linearity.
Of the initial 142 subjects eligible for the study, 24 subjects were excluded because their QA data did not meet our criteria of an adequate QA measurement (stated above). In addition, 7 subjects were excluded because they could not achieve a resting electrically stimulated torque measurement of ≥25% of the practice maximum voluntary contraction. Therefore, data from 111 subjects were included in the final analysis. The 31 subjects excluded from the analysis did not differ in demographic characteristics from the remaining subjects included in the analysis, based on Wilcoxon's test comparison between groups for continuous data and chi-square test for nominal data, at α = 0.05. Descriptive statistics for the 111 subjects are listed in Table 1.
Table 1. Descriptive statistics for the 111 subjects*
|Female sex||70 (63.1)||N/A|
|Age, mean ± SD years||63.7 ± 9.1||40–85|
|Height, mean ± SD cm||167.9 ± 9.9||148–196|
|Weight, mean ± SD kg||84.0 ± 18.8||48–160|
|Time since diagnosis|| ||N/A|
| <1 year||17 (15.3)|| |
| 1–2 years||22 (19.8)|| |
| 3–5 years||21 (18.9)|| |
| 5–10 years||19 (17.1)|| |
| >10 years||32 (28.8)|| |
|Tibiofemoral radiograph grade|| ||N/A|
| 2||14 (12.6)|| |
| 3||62 (55.9)|| |
| 4||35 (31.5)|| |
|Patellofemoral radiograph grade|| ||N/A|
| 0||2 (1.8)|| |
| 1||20 (18.0)|| |
| 2||47 (42.3)|| |
| 3||29 (26.1)|| |
| 4||11 (9.9)|| |
|Medication use†|| ||N/A|
| Yes||75 (67.6)|| |
| No||36 (32.4)|| |
|Exercise therapy type|| || |
| Standard||55 (49.5)|| |
| Perturbation||56 (50.5)|| |
|Pain during test, mean ± SD||0.50 ± 1.4||0–7|
|Resting e-stim torque, mean ± SD percentage of practice MVC||53.8 ± 14.9||26.4–107.8|
|Baseline maximum quadriceps strength, mean ± SD Nm||139.5 ± 53.3||30–280|
|Baseline CAR, mean ± SD||0.93 ± 0.06||0.62–1.00|
|2-month maximum quadriceps strength, mean ± SD Nm||144.7 ± 57.0||25–302|
Factors related to QA and quadriceps strength.
Bivariate correlations among potential covariates, QA, and quadriceps strength are shown in Table 2. Sex, age, height, weight, number of years since diagnosis of knee OA, numeric pain rating during the burst-superimposition test, severity of radiographic knee OA, medication use, and resting e-stim torque were not significantly correlated with the magnitude of QA as measured by the CAR. Quadriceps strength was associated with several of these factors. Men exhibited higher strength than women, younger subjects had higher strength than older subjects, taller subjects had higher strength than shorter subjects, and heavier subjects had higher strength than lighter subjects. Time since diagnosis of knee OA was associated with quadriceps strength, indicating that subjects more recently diagnosed tended to have higher strength than those with longstanding disease. A higher level of resting e-stim torque was also related to higher quadriceps strength. There were no significant associations between quadriceps strength and pain during the burst-superimposition test, severity of radiographic knee OA, or medication use.
Table 2. Bivariate correlations between CAR, quadriceps strength, and potential covariates*
|Time since knee OA diagnosis, ρ||0.05||−0.22§||−0.20§|
|Pain during testing, ρ||−0.17||−0.09||−0.05|
|Tibiofemoral radiographic grade, ρ||0.14||−0.03||−0.01|
|Patellofemoral radiographic grade, ρ||0.08||−0.12||−0.13|
|Medication use, ρ||−0.02||0.10||0.09|
|Resting e-stim torque||−0.09||0.76‡||0.74‡|
|Exercise therapy type, ρ||−0.07||0.13||0.10|
|Baseline CAR, ρ||–||0.30‡||0.23§|
|Baseline quadriceps strength||–||–||0.96‡|
Relationship of QA magnitude and strength after exercise therapy.
Bivariate correlations demonstrated that QA was associated with quadriceps strength at baseline (ρ = 0.30, P < 0.01) and 2-month followup (ρ = 0.23, P = 0.01). Greater QA correlated with higher strength. The results of the linear regression on quadriceps strength at the 2-month followup are shown in Table 3. While controlling for baseline quadriceps strength and type of exercise therapy, the level of QA did not add to the prediction of quadriceps strength outcome (β = −0.04, P = 0.18). We did not include any of the other potential factors as covariates because none of them were associated with QA. Variance inflation factors were <10, indicating no multicollinearity. Observation of residual plots demonstrated that our model fit the linear regression model assumptions of normality, homoscedasticity, and linearity.
Table 3. Linear regression model on quadriceps strength at 2-month followup*
|Overall model†||0.92||0.92||3, 107|| || || |
|Constant|| || || ||32.14||–||0.15|
|Exercise type|| || || ||−2.68||−0.02||0.38|
|Baseline quadriceps strength|| || || ||1.04||0.98||< 0.01|
|Baseline CAR|| || || ||−33.74||−0.04||0.18|
Consistent with prior research (4), we found that lower QA is associated with lower quadriceps strength. Interestingly, our results indicated that the pretreatment level of QA did not predict quadriceps strength after exercise therapy in subjects with knee OA. Although greater reduction of QA was related to weaker quadriceps, the reduction did not appear to affect how subjects responded to exercise therapy in terms of quadriceps strength. Our sample had a wide range in changes in strength (−54 to 52 Nm, mean ± SD change of 5.2 ± 15.9 Nm). However, QA did not contribute to predicting the variance of this change. This leads us to believe that other factors may be more important in determining who responds best to strengthening regimens. Further study is needed to examine other potential factors such as subject effort, motivation, and compliance with the exercise therapy.
Our subjects' mean CAR was slightly lower (0.93 versus 0.94–0.95) and their prevalence of reduced QA was slightly higher (50% versus 40–44%) than values reported for healthy elderly adults using similar testing protocols (30, 31). In comparison, in previous studies healthy middle-aged individuals had a 25% incidence of reduced QA (5), and ∼10% of healthy younger subjects did not achieve a CAR >0.95 (30, 31). Knee OA may increase the likelihood of muscle activation deficits, although the magnitude of QA is not substantially lower than estimates due to age-related changes. It is possible that activation deficits first occur early in the disease course when their clinical impact is minimal, but can predispose to quadriceps weakness, further joint damage from a less stable joint, and further decrease in activation as time and disease progress. A recent study reported that both the level of QA and the lean muscle cross-sectional area of the quadriceps are significantly associated with quadriceps strength in limbs with and without end-stage knee OA, but relative contributions differ. The level of QA explained a greater percentage of variance in quadriceps strength in limbs with end-stage knee OA, whereas lean muscle cross-sectional area explained a larger proportion of the variance in quadriceps strength in uninvolved limbs (32). However, that study utilized a cross-sectional design, so it is uncertain how QA left untreated affects quadriceps strength, joint damage, or joint instability over time. The data from our study seem to indicate that QA does not have an influence on strength gains when subjects participate in an exercise program.
The mean ± SD QA in our sample was 0.93 ± 0.06, with a median of 0.95 and a range of 0.62–1.00. This is higher than in several prior studies, which reported means of 0.81 (10) and 0.66 (18), and a median of 0.73 (4). The difference in mean QA in our study compared with others may be related to differences in methodology. Some studies used different techniques (4, 10, 18, 23), and not all describe a procedure with sufficient practice, motivation, and feedback (10, 18), which is important for eliciting maximum voluntary effort and accurate estimation of QA. One study that did use the same burst-superimposition technique with CARs, as well as procedures to optimize accurate QA measurement, found QA values similar to ours (with a mean CAR of 0.93) (5). In addition, the prevalence of failure of full QA, as defined by a CAR <0.95 (5, 30, 31), was 50% in that study as well as in our current study.
Overall, the difference in values for QA among studies implies that the precise magnitude cannot be directly compared across studies using different stimulus parameters and testing procedures. Some believe that the burst-superimposition method may be the best method for measuring QA (29, 33), but others have contested this viewpoint in favor of twitch interpolation (34, 35). It should also be acknowledged that our finding that baseline QA using the CAR method does not predict change in strength after exercise may not necessarily apply to other methods of measuring QA, such as the twitch-interpolation method. To our knowledge, examining whether baseline QA predicts changes in strength following rehabilitation in subjects with knee OA has not been studied.
The experience of pain during testing could affect measurements of QA and quadriceps strength if it inhibited a subject from exerting full volitional effort. However, we found that the pain reported by subjects during testing was not correlated with either QA or quadriceps strength. Most subjects reported no pain during testing and only 4 subjects reported pain of ≥5 on the 0–10 numeric pain rating scale. This suggests that knee pain was not a significant confounder in our study.
Knee effusions have been shown to prevent full muscle activation (11, 12). We did not have data on the presence or size of knee effusions in our sample, so we were not able to examine this factor as a potential confounder. However, a prior study did not find an association between joint effusion and QA or quadriceps strength in subjects with knee OA (19). It is plausible that the low prevalence and small size of effusions in subjects with knee OA are not enough to significantly induce activation failure as is observed in other populations. Therefore, we do not believe our results were confounded by this factor.
Although the burst-superimposition test has been shown to be sensitive to detect failure of muscle activation (29, 33) and has been widely used (5, 15, 17, 30, 31, 36), there remain limitations to the test. Recently, some investigators have suggested that the CAR overestimates the actual degree of muscle activation because the relationship between percentage of maximum voluntary effort and CAR appears curvilinear and not strictly linear (34, 37). This poses complexities for the analysis and comparison of QA data. We optimized our measurement of QA by providing adequate practice, verbal encouragement, and visual feedback to the subjects, using an adequate stimulus and taking the best of several trials. But can we really know whether all the subjects were exerting their maximum effort? Low CAR may be due to submaximal voluntary effort in addition to intrinsic inhibition of muscle activation. Additionally, to further ensure accurate measurement of QA, we only included data that met strict criteria. A number of potential subjects were not included in the study because they were not able to maintain a maximum voluntary contraction long enough for delivery of the electrical stimulus to measure QA, despite practice trials, verbal encouragement, and visual feedback. Given the time, effort, and difficulties associated with administering this test, the utility may be limited in a knee OA population where the magnitude of activation deficits appears modest. However, in populations with significant trauma to the joint and greater activation deficits (e.g., following total knee arthroplasty) (36, 38), the utility may be greater.
In conclusion, pretreatment magnitude of QA using the CAR method did not help predict quadriceps strength following exercise therapy. Measurement of QA using the CAR method does not appear to be helpful in identifying subjects with knee OA who will have difficulty improving quadriceps strength with exercise therapy. It is not known whether these results would apply when QA is measured using twitch-interpolation methods. Investigation of other factors that may affect response to exercise therapy is warranted.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Fitzgerald 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 conception and design. Scopaz, Piva, Gil, Woollard, Oddis, Fitzgerald.
Acquisition of data. Scopaz, Piva, Gil, Woollard, Oddis, Fitzgerald.
Analysis and interpretation of data. Scopaz, Piva, Gil, Woollard, Oddis, Fitzgerald.