Subject: Physical Activity and Knee Structural Change: A Longitudinal Study
To Print: Click your browser's PRINT button.
NOTE: To view the article with Web enhancements, go to:
Physical Activity and Knee Structural Change: A Longitudinal Study Using MRI
Stella Foley;1 Changhai Ding;1 Flavia Cicuttini;2 Graeme Jones1
Med Sci Sports Exerc. 2007;39(3):426-434. ©2007 American College of Sports Medicine
Introduction: Exercise therapy is effective in improving symptoms of knee osteoarthritis, but its effect on structural change remains unclear.
Osteoarthritis (OA) is a condition characterized by changes to the integrity of articular cartilage and subchondral bone. The knee is the most frequently affected joint, with a prevalence of 30% in people aged 65 yr and over . Although exercise therapy is effective in improving symptoms of knee OA , the relationship between physical activity and structural change of the knee remains unclear.
Observational studies have suggested a higher risk of radiographic knee OA with repetitive, high-impact sports, and this risk is most strongly associated with joint injury . A history of regular sports participation also has been shown to increase the odds of incident but not progressive radiographic OA (ROA) [8,14]. In contrast, moderate recreational physical exercise has been associated with decreased risk of knee OA requiring arthroplasty . Most of these studies, however, are retrospective in nature and are subject to many biases. In randomized trials on animals, it has been repeatedly shown that exercise decreases the risk of developing OA of the weight-bearing joints [20,24,25]. Likewise, Slemenda et al.  have shown that increased quadriceps strength may be a protective mechanism guarding against the onset of medial-tibiofemoral OA.
Recently, novel imaging modalities such as magnetic resonance imaging (MRI) have made important contributions to our understanding of OA. We have reported both cross-sectional and longitudinal positive associations between physical activity and cartilage development in children . Our group also has shown that increased muscle mass is strongly associated with medial tibial cartilage volume and a reduction in the loss of tibial cartilage . On a similar note, glycosaminoglycan (GAG) content, a measure that may reflect cartilage quality, increased after 4 months of moderate exercise .
The controversy surrounding this issue warrants urgent attention because muscle-strengthening interventions are now a major component of the usual treatment program for patients with knee OA, despite the little information regarding their effects on disease prevention and/or progression. This is particularly relevant because both bone size and rate of cartilage loss have been identified as independent predictors of knee replacement in a longitudinal study . Furthermore, there have been no studies in adults that have investigated the role of physical fitness on knee structural changes. The aim of this longitudinal study was to describe the associations between strength, endurance fitness and self-reported physical activity, and structural change of the knee joint in a convenience sample of adult male and female subjects.
Materials and Methods
The study was carried out in Southern Tasmania, primarily in the capital city of Hobart, from June 2000 until December 2001, with the follow-up data collected approximately 2 yr later. Subjects were selected from two sources. Half the subjects were the adult children of people who had had knee replacement surgery performed for primary knee OA at any Hobart hospital in the years 1996-2000. This diagnosis was confirmed by reference to the medical records of the orthopedic surgeon and the original radiograph whenever possible. The other half were randomly selected controls. The controls were selected by computer-generated random numbers from the most recent version of the electoral role (2000). Subjects from either group were excluded in cases of contradiction to MR imaging (including metal sutures, presence of shrapnel, iron filling in eye, and claustrophobia). No women were on hormone therapy at the time of the study. The Southern Tasmanian health and medical human research ethics committee approved the study, and all subjects gave written informed consent.
Weight was measured to the nearest 0.1 kg (with shoes, socks, and bulky clothing removed) using a single pair of electronic scales (Seca Delta Model 707). Height was measured to the nearest 0.1 cm (with shoes and socks removed) using a stadiometer. Body mass index (BMI) was calculated as kilograms per square meters. Knee pain was determined by self-administered questionnaire if subjects answered yes to the following question: Have you had knee pain for more than 24 h in the past 12 months, or daily pain on more than 30 d in the last year?
Physical Activity Measures
Physical activity measures included physical work capacity, lower-limb muscle strength, and questionnaire items. Lower-limb muscle strength was measured to the nearest 1.0 kg using a dynamometer (TTM Muscle Meter, Tokyo, Japan). In our opinion, the muscle measures with this technique were primarily the quadriceps and hip flexors, as evidenced by the high correlation of 0.78 between this test and a specific test of quadriceps function (Foley et al., unpublished data). Subjects were asked to stand on the dynamometer with a straight back, flat against the wall, holding a hand bar with an overhand grip. Subjects' knees were flexed until an angle of 115° was obtained at which the bar was attached to the dynamometer via a chain. Subjects kept a firm grip on the bar and pulled upward with their legs only as far as possible. Subjects were instructed in each technique before testing, and each measure was performed twice. Repeatability estimates (intraclass correlation coefficient (ICC)) for lower-limb muscle strength were 0.91.
Physical work capacity was assessed by use of a bicycle ergometer . Subjects were asked to cycle at a constant rate of 60 rpm for 3 min each with three successively increasing, but submaximal, workloads. Heart rate was recorded at 1-min intervals at each workload, using an electronic heart rate monitor. Work capacity at 170 bpm (PWC170) was assessed by linear regression with extrapolation of the line of best fit to a heart rate of 170 bpm. The PWC170 was not considered a technically adequately measure unless subjects had spent a minimum of 2 min at each workload and the pulse rate increased by at least 5 bpm with increasing workloads. Repeatability was not assessed in our subjects but has previously been reported as 0.92 .
Physical activity was measured retrospectively using a questionnaire  that was modified after piloting to include popular Australian sports. The test-retest Spearman correlation of overall leisure physical activity in hours per week during the last year was found to be 0.66. This questionnaire has demonstrated predictive validity in our hands for children and adolescents . The questionnaire has items on days of either strenuous activity or light activity for more than 20 min in the last 2 wk (1: 0; 2: 1-2 d; 3: 3-5 d; 4: 4-5 d; 4: 6-8 d; 5: ≥ 9 d), daily television watching in the last week (1: 0; 2: ≤ 1 h; 3: 2-3 h; 4: 4-5 h; 4: ≥ 6), number of competitive sports played in the past 12 months (1: 0; 2: 1; 3: 2; 4: 3; 5: 4 or more), and activities done at least 10 times in the last month.
A standing anteroposterior semiflexed view of the right knee was performed in all subjects. Radiographs were then assessed using the Altman atlas . Each of the following was assessed on a scale of 0-3: medial joint space narrowing (JSN), lateral JSN, medial femoral osteophytes, medial tibial osteophytes, lateral femoral osteophytes, and lateral tibial osteophytes. Each score was arrived at by consensus, with two readers (G.J., F.S.) simultaneously assessing the radiograph with immediate reference to the atlas. Reproducibility was assessed in 50 radiographs, 2 wk apart, yielding an ICC of 0.99 for osteophytes and 0.98 for JSN. This may represent an overestimate of the actual agreement because of the high proportion of normal radiographs. However, this method also has very high reproducibility in our group for ROA of the hands, with ICC of 0.94-0.98 .
Cartilage Volume Assessment
MRI scans of the right knee were performed on two occasions at baseline and follow-up. Knee cartilage volume was determined by means of image processing on an independent workstation using the software program Osiris, as previously described . Knees were imaged in the sagittal plane on a 1.5-T whole-body magnetic resonance unit (Picker, Cleveland, OH) with use of a commercial transmit-receive extremity coil. The following image sequence was used: a T1-weighted fat saturation 3D gradient recall acquisition in the steady state; flip angle 55°; repetition time 58 ms; echo time 12 ms; field of view 16 cm; 60 partitions; 512 × 512 matrix; acquisition time 11 min 56 s; one acquisition. Sagittal images were obtained at a partition thickness of 1.5 mm and an in-plane resolution of 0.31 × 0.31 (512 × 512 pixels). The volume of individual cartilage plates (medial tibial, lateral tibial, and patella) was isolated from the total volume by manually drawing disarticulation contours around the cartilage boundaries on a section-by-section basis. These data were then resampled by means of bilinear and cubic interpolation (area of 312 and 312 µm and 1.5-mm thickness, continuous sections) for the final 3D rendering. The volume of the particular cartilage plate was then determined by summing all the pertinent voxels within the resultant binary volume. Femoral cartilage volume was not assessed because we had previously established that the two tibial sites and the patella site correlated strongly with this site . Using this method, we had high intra- and interobserver reproducibility. The coefficient of variation (CV) for cartilage volume measures was 2.1% for medial tibial and 2.2% for lateral tibial .
Cartilage Defect Assessment
Knee cartilage defects were determined by means of image processing on an independent workstation by one observer (C.D.) using Osiris. The cartilage defects were graded (10-12) on both occasions at the medial tibial, medial femoral, lateral tibial, lateral femoral, and patellar sites as follows: grade 0, normal cartilage; grade 1, focal blistering and intracartilaginous low-signal intensity area with an intact surface and bottom; grade 2, irregularities on the surface or bottom and loss of thickness of less than 50%; grade 3, deep ulceration with loss of thickness of more than 50%; grade 4, full-thickness chondral wear with exposure of subchondral bone. A cartilage defect also had to be present in at least two consecutive slices. The reader was unaware of the initial result at the time of the second reading. The cartilage was considered to be normal if the band of intermediate signal intensity had a uniform thickness. The cartilage defects were regraded 1 month later, and the average scores of cartilage defects at medial tibiofemoral (0-8) and lateral tibiofemoral (0-8) were used in analysis. Intraobserver reliability (expressed as ICC) was 0.89-0.94, and interobserver reliability was 0.85-0.90 (10-12).
Tibial Plateau Area Assessment
Knee tibial plateau bone area was determined by means of image processing on an independent work station by one observer using Osiris as previously described . To transform the images to the axial plane, the analyses software package developed by the Mayo Clinic was employed. Medial and lateral tibial plateau bone area was determined by creating a composite measure from the three input images closest to the knee joint after reformatting in the axial plane. The areas of the medial and lateral tibial plateau were then directly measured from these images. The CV for these measures in our group are 2.2-2.6% .
One-way ANOVA tests were used for comparisons of means. The chi-square test was used to compare nominal characteristics between groups. Rate of change in cartilage volume were calculated both as the absolute change per annum, (v1 - v0)/t and as the percentage change per annum, 100(v1 - v0)/v0t, where v0 is cartilage volume at baseline, v1 is cartilage volume at follow-up, and t is the time in years between scans. Change in tibial bone area was calculated in the same manner. Changes in cartilage defects were calculated by subtracting cartilage defect scores at baseline from cartilage defect scores at follow-up. A change in cartilage defects of at least 1 was defined as an increase in cartilage defects, and a change in cartilage defects of no more than -1 was defined as a decrease in cartilage defects. Multiple linear-regression techniques were used to explore the possible physical activity measures affecting the rate of change in cartilage volume and tibial bone area, with nonstandardized regression coefficients presented. For changes in cartilage defect scores in individual compartments, the range and distribution were such that basic assumptions for applying linear regression did not hold. Logistic regression analysis was used to examine the associations between progression of knee cartilage defects and physical activity variables. Although associations did not differ in offspring and controls, we adjusted all associations for offspring control status because of the convenience nature of this sample. Furthermore, multivariate results were age- and sex adjusted (as well as ROA and weight where appropriate) because of possible confounding by these factors. A model also was constructed containing the physical activity variable, sex, and their interaction term (physical activity × sex). Statistical significance was determined on the basis of the P value for the interaction term. A P value less than 0.05 (two tailed) or a 95% confidence interval not including the null point was considered statistically significant. All statistical analyses were performed on Intercooled Stata 8.2 for Windows (StataCorp LP).
A total of 325 subjects completed the study (87% of those originally studied). The mean interval between measurements was 2.3 yr (range 1.8-2.6). Demographic and study factors are presented in Table 1 . This was a young sample, with an average age of 45 yr at baseline (range 26-61). When subjects were examined by categories of PWC170 and change in lower-limb muscle strength, there was a significant difference in the proportion of males and females. In addition, the height and weight of subjects differed between groups, with those who had significant decreases in lower-limb muscle strength and PWC170 being taller and heavier. BMI and the percentage of subjects who had progression of lateral defects were significantly different between categories of PWC170, whereas changes in medial tibial plateau area differed significantly between categories of lower-limb muscle strength.
Analysis of the effect of baseline work capacity on the rate of annual change of cartilage volume (%) revealed a significant interaction between sex and PWC170 (P = 0.09-0.004); therefore, the results are presented separately for males and females ( Table 2 ). In univariate analysis, for females, the higher the PWC170 at baseline, the greater the rate of loss in cartilage volume (%) in the medial, lateral, and total compartments. After adjustment for confounders, this relationship persisted for lateral and total compartments only. Conversely, in both sexes, lower-limb muscle strength was positively associated with annual cartilage change in the total compartment, with no other significant associations noted.
For tibial plateau bone area, there was again a significant interaction between sex and PWC170 ( Table 3 ). In univariate analysis, there was a significant negative association between PWC170 and annual percent change in lateral and total tibial plateau area for males (P = 0.03 and 0.04, respectively) that did not persist after adjusting for confounders. Likewise, number of competitive sports was negatively associated with annual change of lateral tibial plateau area (P = 0.04), but this association also did not persist after adjustment for age, sex, case/control status, lateral JSN, and initial bone area. In multivariate analysis, in both sexes, lower-limb muscle strength was significantly positively associated with annual percent change of lateral and total tibial plateau area, with borderline significance in the medial compartment. In addition, there was a positive relationship between PWC170 and annual change of the lateral and total tibial plateau area for females.
There was also a significant beneficial association between annual changes in PWC170 and percent change in medial, lateral, and total cartilage volume per year ( Table 4 ; Fig. 1). Changes in PWC170 were not related to annual percent change in tibial plateau area. On the contrary, annual changes in lower-limb muscle strength were negatively associated with annual percent changes in lateral tibial and total tibial plateau area (Fig. 2); however, when baseline muscle strength was added to the model, the results were weakened (total tibial P = 0.01; medial P = 0.07; lateral P = 0.07). Changes in muscle strength were not related to annual percent change in cartilage volume.
Association between annual change in cartilage volume (%) and change in PWC at 170 bpm during 2.3 yr. r was adjusted for age, sex, change in BMI, and case/control status. There was a modest but significant positive association between annual percent cartilage volume change and change in physical work capacity at 170 bpm in the total sample in the medial, lateral, and total compartments. A, total; B, medial; C, lateral.
Association between annual change in bone area (%) and change in lower-limb muscle strength (kg) during 2.3 yr. r was adjusted for age, sex, change in BMI, and case/control status. There was a negative association between annual percent bone area change and change in lower-limb muscle strength in the lateral and total tibial compartments in the total sample. A, total; B, medial; C, lateral.
For both univariate and multivariate logistic analyses (age, sex, BMI, case/control status, baseline defect score, baseline cartilage volume, and bone area adjusted) of cartilage defects, sessions of strenuous exercise were significantly associated, with a reduction in the odds of lateral knee cartilage defects progressing during 2.3 yr (odds ratio (OR) = 0.73, P = 0.039). Although not significant, there was a consistent trend for medial cartilage defects decreasing with strenuous activity (OR = 0.86, P = 0.24). No other physical activity measures demonstrated a significant association with progression of knee cartilage defects (data not shown).
There were no significant differences in the associations between physical activity and knee structural changes when cases (offspring) and controls were examined separately (data not shown); thus, both groups were combined for all analyses. These results were also independent of knee pain and/or past knee injury (data not shown).
To our knowledge, this is the first study to describe the associations between fitness/physical activity and knee structural change. In this longitudinal study, we demonstrated that lower-limb muscle strength was positively associated with both total cartilage volume and tibial plateau area change per year. In females, there was a deleterious relationship between PWC170 and annual change in lateral and total cartilage volume and tibial bone area. Furthermore, annual change in lower-limb muscle strength was negatively associated with change in lateral and total tibial plateau area annually, whereas changes in PWC170 were positively associated with annual changes in medial, lateral, and total cartilage volume. Strenuous exercise was associated with a reduction in the odds of defects progressing. These associations were modest in magnitude, typically explaining less than 4% of the variance in the data, but they indicate that the relationship between physical activity and change in knee structure is complex and may have differential effects on both bone and cartilage.
The clinical significance of structural change in the knee is gradually emerging. Every 25-mm increase in tibial bone size at baseline increases the odds of knee replacement by 5% per annum . In our study population, subjects in the lowest quartile of change in muscle strength had an average gain of bone area of 26 mm·yr-1, whereas people in the highest quartile had a loss of -1.6 mm·yr-1. This difference would lead to a yearly increase of 5.6% in the odds of knee replacement. Similarly, cartilage volume loss is also a predictor of joint replacement , with a 1% increase in the rate of cartilage loss increasing the risk of undergoing knee replacement by 5% per annum . Again, subjects in the lowest quartile of baseline muscle strength would have increased odds of 3.8% per year of undergoing knee replacement compared with subjects in the highest muscle strength quartile. Furthermore, given that an end-stage knee has lost 60% of it cartilage volume [3,4], a person in the lowest quartile of muscle strength would reach end-stage loss 5 yr earlier than those in the highest quartile (given the overall rate of loss in controls in this sample). Although these changes are modest in magnitude, these effects would be important at a population level.
Strength is largely a reflection of muscle size. As such, the significant association between strength and annual change in total cartilage volume is in accordance with the results of Cicuttini et al. , who recently showed that muscle mass was associated with a reduction in the rate of loss of tibial cartilage in a healthy population. The quadriceps attenuate loads across the knee joint and are important in providing anterior-posterior stability. The muscle-dysfunction hypothesis suggests that when the muscles cannot contract properly (i.e., because of age, disuse, atrophy, or injury-induced weakness), more force is transmitted to the bone, leading to microtrabecular damage and, eventually, sclerosis that could alter the stresses across the articular cartilage . The unfavorable relationship between lower-limb muscle strength at baseline and increases in bone area may be consistent with this because larger bone may alter cartilage loading, thus, representing a maladaptive response to load. However, the inverse association between change in muscle strength and change in bone area was beneficial. The reasons behind this discrepancy are unclear. Baseline muscle strength and change in muscle strength are under both genetic and environmental control, and there may be differential genetic effects such that change in muscle strength may more accurately reflect environmental change. The negative association between baseline muscle strength and change in muscle strength (attributable to ceiling/floor effects) may also explain the opposite effects on joint structure. Overall, the data suggest that programs aimed at increasing muscle strength may still have good effects on tibial bone area.
The PWC170 test is essentially a test of fitness. This also has both genetic and environmental components. The baseline association most likely reflects both (with a genetic preponderance), whereas the association between change in PWC170 and annual percent change in cartilage volume is likely to be mainly environmental, reflecting changes in fitness attributable to changes in activity, because there was no significant genetic component to change in PWC170 in our sample (G Zhai et al., unpublished data). This could explain the apparent contradictions between the two analyses, with women with higher PWC170 having higher rates of loss, but those who increased PWC170 having higher rates of increase in cartilage volume. The latter relationship is consistent with mouse studies in which wheel running led to strengthening of the ligaments and muscles around the knee joint, building up improved dynamic stability and shock-absorbing capacity needed during loading movements. Furthermore, Lapvetelainen et al.  have hypothesized that the locomotion also may have affected the structure and strength of the collagen network of articular cartilage. In a randomized trial, human cartilage responded to moderate exercise by increasing its GAG content, which might improve the biomechanical properties of the cartilage . On the contrary, increased cartilage volume may be a result of aggrecan loss from the matrix and increased water content, which is an early event in the onset of OA in some subjects .
In females, there was a positive association between PWC170 at baseline (but not change) and annual percent change in tibial plateau area. Peak knee adduction moment has been shown to positively correlate with medial tibial plateau bone area , supporting the notion that mechanical load plays a role in the regulation of bone remodeling. Oettmeier et al.  have shown in dogs that below the intact articular surface, the articular cartilage and subchondral bone may become thicker after mechanical loading. This process, however, may not be available when sclerosis is further advanced and impairing the biomechanical function of the surrounding tissues. Our results add to this and indicate a possible deleterious effect on the joint.
It is interesting to note that we observed no significant associations between PWC170, strength, and chondral defects, despite increased bone area being a predictor of chondral defect progression . This observation still remains largely unexplained, although the finding that baseline strenuous exercise had a protective effect independent of injury is reassuring. The lack of consistent association with other questionnaire items most likely reflects the well-known deficiencies of questionnaires compared with objective measures. It is possible that more accurate measures of actual activity, such as pedometers, may give different results.
The physical activity associations are complex, but the observation that changes in PWC170 and lower-limb muscle strength are both associated with knee structural changes is important for a number of reasons. First, this indicates that articular cartilage and subchondral bone are dynamic structures that can respond to environmental manipulation, despite being under strong genetic control . Further, there is the possibility that exercise programs could be designed and implemented in those predisposed to OA, to delay or even prevent structural changes. Our results suggest that muscle strengthening is more likely to be beneficial than fitness training.
The current study has a number of potential limitations. First, assessment of physical activity is problematic in epidemiological studies. Although the questionnaire we used has good test-retest reliability, there is still a considerable margin for measurement error. The likely effect of this is to decrease the strength of associations between physical activity and structural change, leading to lower power to find moderate associations. Similarly, the questionnaire assesses leisure time physical activity and does not take into account occupation-related physical activity. As such, the few associations noted with regard to questionnaire items may be the results of exercise that does not constitute leisure-time physical activity. The questionnaire also has limitations when focusing on specific sports; thus, it would be worthwhile to study inception cohorts of those participating in sports with different weight-bearing properties to examine whether there is variation in effect with different sports. Secondly, the current study was primarily designed to look at genetic mechanisms of knee OA, using a matched design. The matching was broken for the current study, but adjustment for case-control status did not alter the results. Indeed, although there was a reduction in power, the results differed very little when offspring and controls were examined separately. While the sample is a convenience sample, Miettinen  states that for these observations to be generalizable to other populations, three key criteria need to be met regarding selection, sample size, and adequate distribution of sample size-all of which are met by the current study. Lastly, measurement error associated with MRI may have weakened the associations. However, our assessment technique has high reproducibility in our group, suggesting that this is not of major concern.
In conclusion, this study suggests that knee cartilage volume and tibial plateau area are dynamic structures that can respond to physical stimuli. Greater muscle strength and endurance fitness, especially in women, may be protective against cartilage loss; however, it also may result in maladaptive enlargement of subchondral bone in both sexes, suggesting that physical activity may have both good and bad effects on the knee.
The authors thank the subjects and orthopedic surgeons who made this study possible. The assistance from C. Boon in coordinating the study, and from S. Quinn and L. Blizzard for statistical support, is gratefully acknowledged. We also would like to thank M. Rush, who performed the MRI scans, and K. Morris, for technical support.
This study was supported by the National Health and Medical Research Council of Australia, Masonic Centenary Medical Research Foundation.
What else is published in Medicine & Science in Sports & Exercise®? Visit www.acsm-msse.org.
Table 1. Characteristics of Participants by Change in PWC170 and Lower-limb Muscle Strength During 2.3 yr.
Table 2. Relationship Between Annual Percent Change of Cartilage Volume and Physical Activity Measures.
Table 3. Relationship Between Annual Percent Change of Tibial Plateau Area and Physical Activity Measures.
Table 4. Association Between Change in Physical Activity Measures and Annual Percent Change of Cartilage Volume and Tibial Plateau Area.
Address for correspondence: Stella Foley, Menzies Research Institute, University of Tasmania, Private Bag 23, Hobart, Tasmania, 7001, Australia; E-mail: firstname.lastname@example.org .
Stella Foley,1 Changhai Ding,1 Flavia Cicuttini,2 and Graeme Jones1
1Menzies Research Institute, University of Tasmania, Hobart, Australia
2Department of Epidemiology and Preventive Medicine, Monash University Medical School, Melbourne, Australia