To examine the relationship between knee osteoarthritis (OA) and muscle parameters in a biracial cohort of older adults.
To examine the relationship between knee osteoarthritis (OA) and muscle parameters in a biracial cohort of older adults.
Participants in the Health, Aging and Body Composition Study (n = 858) were included in this cross-sectional analysis. Computed tomography was used to measure muscle area, and quadriceps strength was measured isokinetically. Muscle quality (specific torque) was defined as strength per unit of muscle area for both the entire thigh and quadriceps. Knee OA was assessed based on radiographic features and knee pain. We compared muscle parameters between those with and without radiographic knee OA (+RKOA group and −RKOA group, respectively) and among 4 groups defined by +RKOA and −RKOA with and without pain.
The mean ± SD age was 73.5 ± 2.9 years and the mean ± SD body mass index (BMI) was 27.9 ± 4.8 kg/m2. Fifty-eight percent of participants were women and 44% were African American. Compared to the −RKOA participants, +RKOA participants had a higher BMI (30.2 versus 26.8 kg/m2), larger thigh muscles (117.9 versus 108.9 cm2), and a greater amount of intermuscular fat (12.5 versus 9.9 cm2; all P < 0.0001). In adjusted models, the +RKOA participants had significantly lower specific torque (P < 0.001), indicating poorer muscle quality, than −RKOA participants, but there was no difference between groups in quadriceps specific torque. The +RKOA without pain (P < 0.05) and the +RKOA with pain (P < 0.001) participants had lower specific torque compared to the −RKOA without pain group. There were no significant differences in quadriceps specific torque among groups.
Muscle quality was significantly poorer in participants with RKOA regardless of pain status. Future studies should address how lifestyle interventions might affect muscle quality and progression of knee OA.
Osteoarthritis (OA) of the knee is a major cause of disability among the elderly (1–3), and its prevalence is projected to grow substantially with the increasing age of the US population (2, 4). Despite the high prevalence of this condition and the high personal and economic impact of OA-related disability, there are currently no specific therapies that can prevent the onset or progression of joint damage caused by OA. Medical therapies designed to control pain and swelling can be associated with side effects, and the withdrawal of certain cyclooxygenase 2 inhibitors from the market due to safety concerns has further limited therapeutic options (5). With limited therapeutic modalities and concerns about long-term safety of existing modalities, it is imperative to better understand the physiologic and biologic relationships underlying knee OA in order to devise and promote better preventive and therapeutic options.
Muscle weakness, and in particular quadriceps weakness, has been implicated in knee OA, and quadriceps strengthening exercises are a therapeutic approach that has been recommended by some experts (6–9). However, reports on the relationship between muscle strength and knee OA have presented conflicting data, due in part, perhaps, to an emphasis on gross measurements of strength that do not adequately control for other intrinsic muscle properties. A generalized loss of skeletal muscle mass with aging, known as sarcopenia, contributes to muscle weakness. In OA patients, as well as in aging, the decrement in strength is greater than would be expected for reduction in size of the muscle (10). This suggests that alterations in the intrinsic characteristics of muscle, reflecting a diminished muscle quality, may also contribute to impaired muscle function of the elderly and of those with knee OA. However, there is little information concerning the association of skeletal muscle composition with the prevalence of OA.
Lower extremity muscle composition and quality is likely related to overall body composition, and obesity has also been implicated as a risk factor for OA, particularly in women (10, 11). Many patients with knee OA are obese, and will be stronger than those who are less obese because they have a higher muscle mass as part of a higher total mass. Strength is highly correlated with lean or muscle mass (12, 13). Therefore, strength differences in those with and without OA cannot be adequately examined without adjusting for lean mass. One of the most intriguing reports relating body composition and muscle strength to knee OA was that of Slemenda and colleagues (10), who showed that reduced quadriceps strength relative to body weight and lean muscle mass (measured by dual x-ray absorptiometry [DXA]) was a risk factor for development of knee OA in women. These results suggest that the quality of muscle is impaired; however, there has been little recent information in the literature to follow up or confirm these provocative findings.
A more recent report by Sharma and colleagues (14) suggested that in some cases greater quadriceps strength may be associated with higher risk of progressive knee OA in patients with greater laxity in their knee. Unfortunately, this study did not include any measures of muscle mass or composition and controlled for body composition only through body mass index (BMI). However, these results emphasize that the relationship of muscle strength and knee OA may be modified by other factors.
Advanced imaging techniques such as computed tomography (CT) can be used to give accurate estimates of lower extremity muscle mass. Lower extremity strength per unit of muscle area, or specific torque, is a measure of muscle quality. In addition, CT can be used to measure muscle attenuation, which is suggestive of an increased muscle fat content (15) and is a marker of muscle quality that contributes to muscle weakness independent of the age-associated loss in muscle mass (16).
In the current study, we examined the association of radiographic and symptomatic knee OA with muscle strength, muscle area, intermuscular fat infiltration, and muscle quality using CT in a large, biracial cohort of elderly participants. We hypothesized that both radiographic and symptomatic knee OA would be associated with poorer muscle quality and higher muscle fat content.
This study explored the relationships between muscle strength, mass, and quality conducted in a large, biracial population.
This study used detailed and sophisticated computed tomography measures to assess muscle area and fat infiltration, which are lacking in most other studies in this area.
This study supports the idea that knee osteoarthritis is associated with lower specific torque, indicating poorer muscle quality.
The Health, Aging and Body Composition (Health ABC) Study is a longitudinal cohort study of factors that contribute to disability in the elderly, with special focus on body composition and weight-related health conditions. The Health ABC Study was approved by the Institutional Review Boards at the University of Tennessee and the University of Pittsburgh, which are the 2 clinical sites for the study. Participants were recruited to the Health ABC Study from a random sample of Medicare beneficiaries ages 70–79 years residing in zip codes from the metropolitan areas surrounding Pittsburgh, PA and Memphis, TN. All participants signed an informed consent document. To be eligible, participants had to report that they had no difficulty walking at least one-quarter of a mile and climbing 10 stairs without resting. A total of 3,075 men and women were enrolled in Memphis (n = 1,548) and Pittsburgh (n = 1,527).
In a knee OA substudy of the Health ABC Study, individuals with and without knee pain had knee radiographs obtained during the second and third annual visits. A qualifying case of knee pain was defined as having at least either knee pain on most days of the last month, knee pain lasting ≥1 month in past 12 months, or moderate or severe pain performing any of the activities in the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain scale (i.e., walking on a flat surface, going up or down stairs, at night while in bed, standing upright, getting in or out of a chair, and getting in or out of a car) in the past 30 days. Individuals with qualifying knee pain and a random sample of those without qualifying knee pain were invited to be a part of the knee OA substudy and undergo knee imaging. Knee radiographs were obtained for 864 knee pain cases and 273 controls.
All participants in the Health ABC Study underwent a total body DXA scan and a CT scan. The total body DXA scan was performed using pencil beam technology (Hologic QDR 1500, enhanced whole body, software version 5.71). A standard soft tissue examination included total body and regional measurements of the trunk, arms, and legs to analyze body composition according to a 3-compartment model (fat mass, lean tissue, and bone mineral content).
The CT scan of the thigh was acquired using a Somatom Plus 4 (Siemens) or a Picker PQ 2000S (Marconi Medical Systems) in Memphis, and in Pittsburgh using a 9800 Advantage (General Electric). Slice thickness was set at 10 mm. Muscle and fat areas were calculated by multiplying the number of pixels of a given tissue type by the pixel area. Density values were determined by averaging the CT number (pixel density) values of the regions outlined on the images. CT numbers were defined on a Hounsfield unit (HU) scale, where 0 equals the HU of water and −1,000 equals the HU of air. The HU scale was used to define tissue types as well as a measure of muscle density. For each participant, the determination of soft tissue type was made using the bimodal image distribution histogram resulting from the distribution numbers in adipose tissue and nonadipose soft tissue (17). Using this method, we were able to quantify the proportion of intermuscular fat in the total thigh muscle as well as the area of the total thigh and quadriceps muscles.
The strength of the participants' dominant leg was evaluated with an isokinetic dynamometer (Kin Com, model 125AP). All participants were given the same instructions before testing and the same prompting during testing. Concentric peak torque of the knee extensors was assessed; the participants' extremity was weighed for gravity correction and start–stop angles were set at 90° and 30°. Participants were allowed several submaximal practice efforts, followed by the recording of the best of 3 maximal efforts. Strength testing was performed by research technicians trained and certified in the Kin Com protocol and primarily measured the quadriceps. Full details of the strength testing protocol have been described in detail elsewhere (18).
Specific torque was calculated as the peak strength divided by the total thigh muscle area of the same leg for which strength testing was performed. Quadriceps specific torque was calculated as the peak strength divided by the quadriceps muscle area of the same leg for which strength testing was performed. Intermuscular fat was not included in the muscle cross-sectional area used to define specific torque or quadriceps specific torque.
Bilateral, standing, and semiflexed views of the tibiofemoral compartment of the knee joint were obtained using the fixed flexion technique. For the tibiofemoral compartment, both knees were assessed radiographically with a posteroanterior projection using a SynaFlexer (Synarc) positioning frame in order to fix knee flexion between 20° and 30° and external rotation of the feet at 10° for each participant. For our analyses, we defined a case of radiographic knee OA (RKOA) as any knee having a Kellgren/Lawrence (K/L) score ≥2.
Physical activity level was determined using a standardized questionnaire designed specifically for the Health ABC Study and modeled after commonly used physical activity assessments, including the Leisure-Time Physical Activity Questionnaire (19). Physical activity level was calculated as kcal per week by multiplying the appropriate kcal score for each of the activities by the amount of time spent during the week doing the activity (20). Antiinflammatory drug use was self-reported and included nonsteroidal antiinflammatory drugs (NSAIDs), salicylates, and oral steroids.
We conducted 2 primary sets of analyses to address our research questions. In the first, we used t-tests to examine the differences in body composition, muscle mass, muscle strength, and muscle quality between participants who did and did not have radiographic knee OA as defined by a K/L score of ≥2. In the second, we used analysis of variance to test the differences in the same outcome measures between 4 groups: no RKOA and no knee pain (−RKOA/−pain), no RKOA and knee pain (−RKOA/+pain), RKOA and no knee pain (+RKOA/−pain), and both RKOA and knee pain (+RKOA/+pain). We stratified our unadjusted results by both sex and race to determine whether the effects differed between either men and women or African Americans and whites.
Univariate analyses were followed by multivariable linear regression models that controlled for 1) age, sex, race, physical activity, arthritis medication use (any NSAID, antiinflammatory, or steroid) and 2) all of the above plus BMI. In each of the multivariable analyses, the body composition, muscle mass, muscle strength, and muscle quality variables were the outcome (dependent) variables, while the RKOA and RKOA/pain groups were the main predictor (independent) variables. In the RKOA/pain group models, separate models were run to test for a trend across the 4 categories of the predictor variable for each outcome variable.
Since lower extremity strength data were only collected in 1 leg for each participant, for the primary analyses, we considered the muscle mass, knee radiograph, and knee pain data for the same leg in which strength data were collected. A secondary set of analyses considered evidence of RKOA or knee pain in either leg. For each set of primary analyses, we also conducted analyses stratified by sex, given prior reports showing sex-related differences in the relationships among body composition, muscle mass and strength, and knee OA.
A total of 858 participants in the Health ABC Study with complete knee pain, RKOA, CT, and leg strength data were included in the present analyses; the characteristics of this group are similar to those of the overall cohort in the Health ABC Study. At baseline, the mean ± SD age was 73.5 ± 2.9 years. Fifty-eight percent of participants were women; 56% were white and 44% were African American. The mean ± SD BMI was 27.9 ± 4.8 kg/m2; 52% were fairly sedentary with physical activity levels of <500 kcal per week, and 57% were taking an antiinflammatory drug, including NSAIDs.
Participants with RKOA (+RKOA) were heavier (83.2 versus 73.6 kg; P < 0.0001) and had greater amounts of both total body lean tissue (51.5 versus 47.4 kg; P < 0.0001) and fat tissue (31.9 versus 26.3 kg; P < 0.0001) as measured by DXA than those without RKOA (−RKOA). Participants in the +RKOA group also had larger thigh muscles (117.9 versus 108.9 cm2; P < 0.0001) and a greater amount of intermuscular fat (12.5 versus 9.9 cm2; P < 0.0001) as measured by CT. There were no significant differences in either quadriceps muscle size (P = 0.11) or strength (P = 0.61) between the 2 groups. However, both specific torque (0.86 versus 0.94 strength per unit of thigh area; P < 0.0001) and quadriceps specific torque (1.92 versus 2.02 strength per unit of quadriceps area; P = 0.002) were significantly lower in the +RKOA group, indicating poorer muscle quality than the −RKOA group. The differences between OA groups were similar both in men and women and in African Americans and whites (data not shown), and similar results were seen when radiographic OA in either knee was considered (data not shown).
We also examined differences in measures of body composition and muscle strength, mass, and quality when accounting for knee pain in addition to RKOA. Table 1 shows how these measurements varied across 4 distinct groups (−RKOA/−pain, −RKOA/+pain, +RKOA/−pain, and +RKOA/+pain). The participants in the +RKOA groups with or without knee pain were slightly older and more likely to be African American than participants in the −RKOA groups with or without knee pain. When compared to the −RKOA/−pain group, participants in the +RKOA/+pain group were heavier and had larger thigh muscles with a greater amount of intermuscular fat. However, quadriceps muscle size did not vary by group. Absolute strength also did not differ by group. Similar to the findings above, participants in the +RKOA/+pain group had poorer muscle quality, reflected in both lower muscle attenuation and lower specific torque than participants in the −RKOA/−pain group.
|−pain (n = 334)||+pain (n = 263)||−pain (n = 91)||+pain (n = 170)|
|Age, years||73.3 ± 2.7||73.3 ± 2.9||73.7 ± 2.9||74.1 ± 3.1||0.008|
|Sex, no. men (%)||140 (42)||110 (42)||38 (42)||69 (41)||0.99|
|Race, no. African American (%)||145 (43)||100 (38)||52 (57)||84 (49)||0.007|
|No. taking NSAIDs (%)||181 (55)||143 (54)||56 (62)||105 (62)||0.22|
|Physical activity, no. (%)||0.15|
|<500 kcal/week||159 (48)||151 (57)||49 (54)||91 (54)|
|500–999 kcal/week||58 (17)||47 (18)||10 (11)||27 (16)|
|1,000–1,499 kcal/week||35 (10)||23 (9)||10 (11)||22 (13)|
|≥1500 kcal/week||82 (25)||42 (16)||22 (24)||30 (18)|
|BMI, kg/m2||26.4 ± 4.3||27.3 ± 4.4||30.5 ± 5.0||30.1 ± 4.8||< 0.0001|
|Total mass, kg||72.2 ± 13.3||75.4 ± 14.9||83.8 ± 13.0||83.0 ± 15.5||< 0.0001|
|Total body lean tissue, kg||47.1 ± 10.4||47.8 ± 10.6||52.1 ± 9.4||51.2 ± 9.7||< 0.0001|
|Total body fat tissue, kg||25.2 ± 7.6||27.6 ± 8.3||31.7 ± 9.3||32.0 ± 10.0||< 0.0001|
|Thigh muscle area, cm2||108.8 ± 27.7||108.9 ± 27.7||120.5 ± 27.3||116.5 ± 26.1||< 0.0001|
|Quadriceps muscle area, cm2||51.0 ± 13.3||50.7 ± 13.0||53.6 ± 13.2||51.8 ± 12.7||0.29|
|Muscle attenuation, HU||36.3 ± 6.9||35.0 ± 6.6||33.9 ± 8.2||32.3 ± 7.1||< 0.0001|
|Intermuscular fat, cm2||9.5 ± 6.2||10.5 ± 5.8||12.4 ± 6.7||12.5 ± 7.4||< 0.0001|
|Absolute strength/peak torque, Nm||103.2 ± 34.6||102.6 ± 35.9||104.6 ± 38.3||100.0 ± 35.4||0.74|
|Specific torque, strength/thigh area||0.94 ± 0.20||0.94 ± 0.20||0.87 ± 0.24||0.85 ± 0.19||< 0.0001|
|Quadriceps specific torque, strength/quadriceps area||2.02 ± 0.41||2.01 ± 0.43||1.94 ± 0.47||1.91 ± 0.41||0.02|
As with the RKOA analyses, differences between groups were not significantly different in women and men (data not shown), and similar results were seen when knee pain or radiographic OA in either knee was considered (data not shown). We also performed analyses using an alternate definition of knee pain as WOMAC score >0 and also found similar results (data not shown).
Table 2 shows the results of multivariable linear regression models of body composition and muscle mass, strength, and quality in −RKOA and +RKOA participants, where the −RKOA group served as the referent group. After adjusting for sex, age, race, physical activity, and antiinflammatory medication use (model 2), +RKOA participants remained heavier, with more lean and fat tissue, and with larger thigh and quadriceps muscle area, whereas muscle attenuation, specific torque, and quadriceps specific torque remained significantly lower. After adding BMI (model 3), significant differences remained between groups in more lean body tissue, greater strength, and lower specific torque; however, there was no longer a significant difference in quadriceps specific torque between groups after adjusting for BMI.
|Body mass index, kg/m2||0.33§||0.32§||NA|
|Total body lean tissue, kg||0.18§||0.18§||0.04¶|
|Total body fat tissue, kg||0.29§||0.29§||0.002|
|Thigh muscle area, cm2||0.15§||0.15§||0.022|
|Quadriceps muscle area, cm2||0.05||0.05¶||−0.06#|
|Muscle attenuation, HU||−0.19§||−0.17§||−0.03|
|Specific torque, strength/thigh area||−0.19§||−0.17§||−0.13§|
|Quadriceps specific torque, strength/quadriceps area||−0.10#||−0.08¶||−0.05|
Table 3 shows similar multivariable results for the analyses of body composition and muscle strength, mass, and quality when accounting for knee pain in addition to RKOA. Here, the −RKOA/−pain group was the referent group and the model is adjusted for sex, age, race, physical activity, antiinflammatory medication use, and BMI. When compared to the referent group, +RKOA participants had lower specific torque (indicating poorer muscle quality) regardless of their pain symptom status. There was no significant difference among the groups in quadriceps specific torque. +RKOA/+pain participants also had more fat mass and lower muscle attenuation and strength than those in the referent group.
|−pain†||+pain||−pain||+pain||P for trend|
|Total body lean tissue, kg||NA||−0.001||0.032||0.033||0.03|
|Total body fat tissue, kg||NA||0.05‡||−0.004||0.034§||0.07|
|Thigh muscle area, cm2||NA||−0.023||0.019||0.002||0.73|
|Quadriceps muscle area, cm2||NA||−0.031||−0.035||−0.069‡||0.002|
|Muscle attenuation, HU||NA||−0.053||0.009||−0.077§||0.05|
|Specific torque, strength/thigh area||NA||−0.016||−0.071§||−0.13¶||0.0004|
|Quadriceps specific torque, strength/quadriceps area||NA||−0.009||−0.03||−0.06||0.12|
We examined the relationship between knee OA and muscle mass, strength, and quality, measured by both radiographic findings and knee pain, in a relatively large biracial cohort of older men and women. The key finding from these analyses is that while absolute strength did not differ by knee OA group, measures of muscle quality, i.e., muscle attenuation and specific torque (strength/muscle area), were significantly poorer in participants with knee OA. However, there were no differences in quadriceps specific torque in fully adjusted models.
Our findings are consistent with those of Slemenda and colleagues (10) in showing that reduced leg strength relative to thigh muscle mass may be associated with knee OA. In our population, we found similar relationships in men as well. There are a few important differences in our population and measures that should be considered when comparing our results to those of Slemenda and colleagues. The age of the cohort in Slemenda and colleagues was similar to the Health ABC Study, but their cohort was predominantly white, whereas ours was biracial. Also, while similar measures of strength (i.e., extensor strength measured via Kin Com) were employed, we used CT, which is a more precise means of measuring thigh and quadriceps muscle area that can also measure and account for factors such as intermuscular fat. Last, Slemenda and colleagues examined incident knee OA, whereas our current study is cross-sectional. Our finding that the lower-quality muscle of participants with knee OA could be related to higher amounts of intermuscular fat infiltration extends this previous report on biomechanical mechanisms linking reduced lower extremity strength relative to body mass and knee OA.
It is somewhat more challenging to compare our results to those of other recent studies due to somewhat limited data about lower extremity muscle mass in those reports. In a slightly younger cohort (mean age 64 years), Sharma and colleagues (14) found that greater lower extremity strength may be associated with an increased risk of knee OA in those with lax or malaligned knees. While this study controlled for subject BMI, no other data on body composition or muscle mass were collected. Another study in a notably younger cohort (mean age 45 years) (21) found that decreased lower extremity strength and poorer functional performance predicted development of knee OA. Again, BMI was the only measure reported to put this into the context of body and/or muscle composition. Segal and colleagues (22) examined the relationship between incident knee OA and thigh strength and found thigh strength predicted incident symptomatic knee OA but not radiographic OA. Our findings seem to be more robust in relation to RKOA findings. However, we believe that a consistent message from these other studies and our current study is that other host factors, be they biomechanical, body composition, or functionality variables, should be considered when looking at the relationship between lower extremity strength and knee OA.
The results of our study should be interpreted within the context of our study's limitations. Because we consider only baseline results from the Health ABC knee OA substudy, we cannot make any comment about the causal or directional nature of the relationships among muscle strength, mass, and quality and knee OA. Poorer muscle quality could be an intermediary variable between obesity and knee OA. We chose to control for BMI in our fully adjusted multivariable models because we wished to see whether knee OA groups and muscle quality parameters were related independent of BMI, since higher BMI is already a well-known risk factor for knee OA. However, these final models could be overadjusted since specific torque and quadriceps specific torque by definition already contain an estimate of body size. Although we did not see a difference in physical activity levels between the RKOA groups in our study, physical activity levels may play in important role in muscle quality for those with knee OA. We were limited by a self-reported measure, and future studies may be enhanced by inclusion of an objective physical activity assessment such as accelerometry.
Another potential concern is the stability of our measure of knee pain, a subjective measure that may vary greatly depending on the timing of the survey and numerous other individual factors. Unfortunately, we do not have any test–retest data of this particular pain variable within our cohort. We attempted to control for any possible misclassification by examining the WOMAC pain variable as well, and found that results were similar to a definition of knee pain as a WOMAC score >0, and also that there was a linear trend in increased WOMAC scores through the 4 groups (−RKOA/−pain, −RKOA/+pain, +RKOA/−pain, and +RKOA/+pain) (data not shown). We do not hypothesize that there would be any directionality to misclassification of pain symptoms, i.e., participants would probably be equally likely to underreport as overreport.
A limitation of our measurement of muscle strength and quality is that we cannot distinguish neuromuscular recruitment limitations versus intrinsic muscle tissue or cellular limitations. Future studies could include testing modalities such as electromyogram to better discriminate these potential defects. Last, we do not account for other individual characteristics, such as knee alignment and gait speed, that might help to account for some of the variability in the relationships among muscle strength, mass, and quality, and knee OA.
Despite these limitations, our current study adds to the literature on the role of lower extremity weakness in knee OA and has several strengths. We studied a fairly large population that includes more African American subjects than most previous reports. We also had the opportunity to look at specific, high-quality muscle mass and composition data that have been, for the most part, missing from prior reports in this area. CT is a particularly useful modality for examining lower extremity muscle in detail and has been validated against muscle biopsy (15) as a tool for measuring intermuscular fat. Using longitudinal CT data or looking at CT measures in a younger cohort would be two areas for future study that could help to further illuminate the causal relationships among muscle strength, mass, and quality, and knee OA.
Last, our findings of the relationship between measures of lower extremity muscle quality and knee OA may have important treatment implications. Apart from medical therapy, practitioners tend to focus on weight loss and quadriceps strengthening as methods to either prevent or treat knee OA. While these may be sound recommendations, it is unknown how they would impact muscle quality and are often given in a blanket fashion without full consideration of the patient characteristics that may also influence knee OA. Future intervention studies should address how various weight loss and strengthening programs might affect muscle quality and the progression of OA.
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 published. Dr. Conroy 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. Conroy, Kwoh, Nevitt, Newman, Goodpaster.
Acquisition of data. Nevitt, Harris, Newman, Goodpaster.
Analysis and interpretation of data. Conroy, Kwoh, Krishnan, Nevitt, Boudreau, Carbone, Chen, Newman, Goodpaster.