Relationships amongst osteoarthritis biomarkers, dynamic knee joint load, and exercise: results from a randomized controlled pilot study

Osteoarthritis (OA) is a common chronic health condition resulting in significant personal and economic burdens. In the absence of a cure, the main areas of research to-date have centred on developing treatments that can improve symptoms of pain and physical dysfunction as well as improving methods to diagnose the disease and to monitor progression. Most commonly affecting the knee, much of the current literature in these two research areas has been conducted in populations with knee OA.

Plain radiographs remain the most commonly used clinical method of assessing joint structure for the purposes of OA diagnosis and classification of severity. Magnetic resonance imaging (MRI) has also been used to assess changes in the morphology of the bone and cartilage to supplement radiographical findings [1]. Though these approaches are reliable and well-established clinically, their use for the monitoring of changes in cartilage over time is limited by the fact that these changes take place over long periods of time and simply detail soft tissue damage that has already occurred [2]. Indeed, a recent systematic review showed greater responsiveness of radiographic joint space width measurements in studies using follow-up periods of greater than 2 years [3]. Given that significant changes in symptoms and joint structure can occur over the course of 2 years, coupled with the fact that most non-surgical and non-pharmacological interventions are conducted over months instead of years, improved methods of assessing changes in cartilage structure and outcomes following shorter-terms treatments over shorter periods of time are needed.

Joint tissue-related biomarkers in the blood and urine have been used to further the understanding of the pathogenesis of knee OA. Many biomarkers are produced during the synthesis or degradation of articular cartilage and are found in different concentrations based on the presence and severity of knee OA [4, 5]. For example, the biomarkers C-propeptide of type II procollagen (CPII), hyaluronic acid (HA), and cartilage oligomeric matrix protein (COMP) in serum, as well as C-telopeptide of type II collagen (CTX-II) and type II collagen cleavage neopeptide (C2C) in the urine have all been shown to be elevated based on the presence and severity of knee OA [6]. Importantly, given that their concentrations reflect processes directly implicated in the synthesis or degradation of articular cartilage, analysis of biomarker concentrations may represent an effective method of assessing cartilage structure over shorter periods of time than conventional methods such as radiography or MRI. However, much is still unknown about these biomarkers, including factors involved in their production and responses to non-pharmacological interventions.

Excessive joint loading is a recognized risk factor for breakdown of articular cartilage based on early in vitro studies [7]. These findings have been supported by later gait analysis studies showing a significantly higher rate of knee OA progression over six years in people with high baseline peak external knee adduction moment (KAM) values [8] – a valid and reliable measure of medial compartment knee joint load during walking [9‐11] – as well as a relationship between knee cartilage volume loss over twelve months and the baseline KAM impulse [12] – the time integral of the KAM during stance [13]. A link between joint load and biomarker concentrations has also been reported. Piscoya et al. [14] have shown increases in COMP production in response to dynamic mechanical load in cartilage explants, and acute bouts of moderately intense physical activity have been shown to temporarily increase the concentration of sCOMP in healthy individuals [15] as well as those with knee OA [16, 17]. However, the relationships between other cartilage biomarkers and measures of everyday joint loading, such as the KAM, are less well known. Further, the effects of common non-pharmacological interventions on biomarker concentrations are also not known.

Given the paucity of research on the relationships of OA biomarkers with joint loading and changes following exercise, the purpose of the present exploratory, pilot study was to address these two knowledge gaps in a sample of individuals with knee OA. Specifically, the primary objective of this study was to determine the relationships between different biomarker concentrations and knee joint load, as measured by the KAM. The secondary objective was to examine changes in biomarker concentrations following a 10-week muscle strengthening intervention.

Seventeen participants (8 males, 9 females; mean (SD) age = 66.1 (11.3) years, BMI = 27.0 (4.5) kg/m²) were enrolled in the study. Ten participants had mild OA (KL 2), five had moderate OA (KL 3), and two had severe OA (KL 4). Baseline demographic and clinical data were similar between the two groups. Sixteen participants (five males and four females from the exercise group and three males and five females from the control group) returned for the follow-up assessment a mean (SD) of 76.3 (5.3) days after the baseline assessment. Blinding of group allocation to the assessor was maintained for all participants. The time of sample collection was consistent between baseline and follow-up for each participant with the mean (SD) difference being 20 (21) minutes (maximum = 65 minutes). Most (13/17) participants provided samples and underwent biomechanical testing in the morning prior to 11:00 am. Attendance at the supervised physiotherapy sessions was high in the exercise group with participants attending a mean (SD) of 91% (14%) of the sessions. Home exercise adherence rates were also high with a mean (SD) of 83% (17%) of the exercise days completed.

Across all participants at baseline, and while adjusting for age and sex, KAM impulse predicted significant variation in uCTX-II levels (β = 1.19, 95% CI = 0.16, 2.21; p = 0.05) as well as the uCTX-II:sCPII ratio (β = 1.50, 95% CI = 0.72, 2.28; p < 0.01). However, when KL grade and walking speed were added to the regression models, the relationship with uCTX-II became non-significant (β = 0.58, 95% CI = -0.53, 1.68, p = 0.33), while the significant relationship with uCTX-II:sCPII remained (β = 1.11, 95% CI = 0.15, 2.07; p = 0.04). In contrast, peak KAM was not able to explain any significant amount of variation in any biomarker or ratio when accounting for age and sex or when adding KL grade and walking speed to the models (p > 0.34). No other significant findings were found for either measure of KAM in any biomarker regression model.

When comparing changes between groups following the intervention (Table 1), significantly greater reductions in sCOMP (β = 0.16, 95% CI = 0.02, 0.30; p = 0.04) as well as slightly greater, non-significant reductions in uCTX-II (β = 0.33, 95% CI = 0.04, 0.71; p = 0.11) were observed in the exercise group compared to those in the control group. No other significant between-group differences existed in any single biomarker outcome or biomarker ratio. Finally, no significant between-group differences were observed in any gait or strength outcome (p > 0.11).

This pilot study provides new information regarding the characteristics of articular cartilage biomarkers relevant to the study and treatment of knee OA. Specifically, this study provides the first data detailing the relationship between multiple OA biomarkers and a measure of dynamic knee joint load – a potential mechanism of OA biomarker production – as well as changes in biomarker concentrations following exercise. Both study objectives taken together, these data provide some support to further explore the utility of uCTX-II and sCOMP as biomarkers relevant to knee OA.

Previous animal studies have shown a direct relationship between load magnitude and articular cartilage degradation [22]. The data from the present study provide some evidence that higher musculoskeletal loading can be associated with increased circulating levels of uCTX-II, though a causative relationship cannot be claimed based on the current data. Furthermore, recent evidence suggests that uCTX-II levels may arise from bone as well as cartilage [23]; however, higher loading applied to the bone would presumably increase uCTX-II levels under the same mechanism as uCTX-II derived from cartilage. Though the KAM is a well-accepted and valid measure of medial compartment joint load [10] with significant relationships with many clinical outcomes specific to knee OA in the medial compartment [8, 24‐29], it does not represent the total load within the knee joint [30] nor does it account for any potential changes in joint contact force that may result from increased muscle activity [31]. Further, though a significant relationship was observed between KAM impulse and uCTX-II at baseline, this relationship became non-significant when adjusting for disease severity and walking speed, and no relationship was found when examining the peak KAM value. Although dynamic knee joint load may play a role in the production of biomarkers – in particular, uCTX-II – it is not the only factor involved in this process and evaluation of a single component of overall joint load (KAM) does not necessarily represent the forces experienced by the cartilage in their entirety. It is clear that more research needs to be done to understand better the relationship between knee joint biomechanics and OA biomarkers.

The second objective of this study was to examine changes in biomarker concentrations following an exercise intervention. Results provide some evidence of a beneficial role of exercise on joint integrity at the cartilage level. Specifically, those in the exercise group demonstrated reduced mean sCOMP levels following the intervention compared to those in the control group. Previous studies have shown reductions in sCOMP following a single muscle strengthening session [32], though increases in sCOMP levels immediately following moderate exercise have also been reported [17]. Given that no significant between-group differences in muscle strength or KAM were observed, it is unlikely that the reductions in sCOMP were due to reductions in joint load. Indeed, baseline results from this study did not provide evidence of a relationship between dynamic joint load during walking and sCOMP levels. This is in contrast to uCTX-II and may suggest that production of uCTX-II and sCOMP occurs due to different mechanisms. This hypothesis cannot be tested using the current study design or data and requires further research.

Though the current results provide new information, this study does have some limitations. First, the small sample size in this pilot study may have reduced statistical power and the ability to make more definitive conclusions. Nevertheless, despite the small sample size, we did find some statistically significant results. Also, the biomarker concentrations measured in this study represent systemic levels that could theoretically have arisen from any number of joints in the body. However, the normal turnover of type II collagen in the body is relatively low, suggesting that significant changes in systemic levels may be expected to be due to pathological turnover from a single joint [33]. Finally, we chose a BMI cut-off of 35 kg/m² to decrease skin movement artifact during walking, as is commonly used in motion analysis studies measuring the KAM [34‐36] to maximize gait data accuracy. However, given that many people with knee OA are overweight or obese, the results of this study cannot necessarily be generalized to the entire knee OA population, and our findings must be viewed in light of this limitation.

This study provides initial evidence of a potential relationship between loading in the knee joint during walking and circulating levels of biomarkers associated with articular cartilage degradation, specifically uCTX-II. A beneficial effect of strengthening exercises on cartilage health as evidenced by reduced levels of circulating sCOMP was also concluded from the results, though the mechanism of this finding is unknown. Further research with more subjects and a longer intervention period would provide verification of these findings and enhance our understanding of the utility of biomarkers in the diagnosis of knee OA as well as their potential as outcome measures following treatment.