Mechanically Stimulated Biomarkers Signal Cartilage Changes Over 5 years Consistent With Disease Progression in Medial Knee Osteoarthritis Patients

Abstract

Using serum biomarkers to assess osteoarthritis (OA) disease state and risks of progression remain challenging. This study tested the hypothesis that changes to serum biomarkers in response to a mechanical stimulus in patients with medial knee OA signal cartilage thickness changes 5 years later. Specifically, serum concentrations of a collagen degradation marker (C1,2C) and a chondroitin sulfate synthesis marker (CS846) were measured 0.5 and 5.5 hours after a 30-min walk in 16 patients. Regional cartilage thickness changes measured from magnetic resonance images obtained at study entry and at 5-year follow-up were tested for correlations with baseline biomarker changes after mechanical stimulus, and for differences between groups stratified based on whether biomarker levels increased or decreased. Results showed that an increase in the degradation biomarker C1,2C correlated with cartilage thinning of the lateral tibia (R = −0.63, p = 0.009), whereas an increase in the synthesis marker CS846 correlated with cartilage thickening of the lateral femur (R = 0.76, p = 0.001). Changes in C1,2C and CS846 were correlated (R² = 0.28, p = 0.037). Subjects with increased C1,2C had greater (p = 0.05) medial tibial cartilage thinning than those with decreased C1,2C. In conclusion, the mechanical stimulus appeared to metabolically link the biomarker responses where biomarker increases signaled more active OA disease states. The findings of medial cartilage thinning for patients with increases in the degradation marker and correlation of cartilage thickening in the less involved lateral femur with increases in the synthetic marker were consistent with progression of medial compartment OA. Thus, the mechanical stimulus facilitated assessing OA disease states using serum biomarkers.

Osteoarthritis (OA) is a leading cause of disability for which efficient methods to diagnose early disease and to assess disease-modifying treatments remain lacking. Challenges to early diagnosis and treatment of OA reflect the complexity of a lengthy disease process that is influenced by mechanical, biological, and structural components.^1,2 It is the interaction of these diverse components that support the need to treat OA development as a complex system. As such, if a perturbation (e.g., injury) to one component of the system elicits an adaptive response in the other components then cartilage homeostasis and health can be maintained, whereas, the lack of an adequate adaptive response can lead to cartilage degradation and OA development. Thus, analyzing only a single component (e.g., biological) of the system may lack sensitivity to determine risks for disease progression without assessing adaption in other components. The paucity³ of sensitive methods for OA risk analysis, disease stratification, and assessment of treatment responses supports employing a more comprehensive method to developing markers for OA that accounts for potential interactions between multiple components of the disease.

There are several observations that should be considered in developing a comprehensive approach to assessing markers of cartilage health. First, it has been shown that the chondrocyte is mechanosensitive and responds to its local mechanical environment⁴ in a manner that creates regional variations in the expression of markers of synthesis and degradation.⁵ Thus, concentrations of specific biomarkers can depend on the status of the local mechanical environment of the cell. Second, cartilage is metabolically active in early OA^6–8 and there are dynamic changes in macromolecule turnover⁶ that likely increase with a mechanical stimulus. Finally, there is evidence that ambulatory mechanics^9,10 can influence a biological response detected in serum concentrations of cartilage oligomeric matrix protein (COMP) in healthy subjects¹⁰ and in patients with knee OA.¹¹ Most importantly changes in COMP following a mechanical stimulus (30 min walking) were associated with cartilage structural changes 5 years later¹¹ in patients with knee OA. This particular stimulus-response protocol^11,12 has demonstrated that introducing a mechanical stimulus can overcome the large inter-subject variability^13–15 in the nominal resting levels^16–18 of serum biomarkers that has hampered the sensitivity of biomarkers for predicting prospective cartilage changes^13,19 in patients with knee OA. As COMP is a general marker of matrix turnover it would be useful to consider the information that can be obtained from testing the response of specific markers of degradation and synthesis to a mechanical stimulus, and if that response provides new insight into the pathogenesis of OA.

A report by Cibere et al.¹³ provides a useful basis for considering both synthesis and degradation markers for applying the stimulus-response protocol to complement the COMP study.¹¹ Specifically it has been reported¹³ that serum levels of C1,2C²⁵ (neo-epitope of both type I and type II collagen cleavage products) were elevated in patients with radiographic knee OA, a finding consistent with previous studies.^21,22 Similarly, it has been shown^6,13 that CS846, an aggrecan synthesis marker and an epitope found only on chondroitin sulfate chains, is increased in OA patients. Also the regional thickening in cartilage morphology^23,24 commonly observed in OA is consistent with reports^6,13 of increased serum concentrations of CS846 in patients with knee OA.

It is important to note that the above observations on C1,2C and CS846 were based on resting serum concentrations and raise questions regarding the response of these markers to a mechanical stimulus. The use of walking as a mechanical stimulus is supported by the previous study on COMP¹¹ as well as several studies^24–26 that indicate specific features of gait mechanics are common to subjects at risk for developing knee OA. As such, understanding of the pathogenesis of OA could be enhanced by further exploration of the relationships between prospective regional cartilage changes and baseline changes in C1,2C and CS846 using the stimulus-response protocol previously used for the COMP study.¹¹

Thus, the purpose of this study was to explore the response of synthesis⁶ (CS846) and degradation²⁵ (C1,2C) biomarkers to a mechanical stimulus, and whether the magnitude and direction of the stimulus induced biomarker responses signal prospective cartilage changes over 5 years. Specifically, the following hypotheses were tested in patients with medial compartment knee OA: (i) Changes in biomarker concentrations after mechanical stimulus will correlate with regional cartilage thickness changes over 5 years; (ii) medial knee OA patients showing increases in C1,2C and CS846 biomarker concentrations after mechanical stimulus will have greater medial compartment cartilage thinning over 5 years; and (iii) The magnitude and direction of the changes in the synthesis and degradation marker following the mechanical stimulus will be correlated.

METHODS

Study Population

In accordance with IRB approved protocols, banked serum samples and knee magnetic resonance imaging (MRI) from subjects with medial knee OA were obtained from cohorts described in prior studies.^10,11 The study population included six males and 10 females (age 60.3 ± 10.2 yrs; BMI 28.3 ± 4.4 kg/m²); with mean baseline KL grade 2.1 ± 0.8.

Subjects in this cohort formed a convenience sample comprised of individuals from an original cross-sectional study¹⁰ who subsequently agreed to and were able to return at 5 years for a second MRI of the knee,^10,11 and who had sufficient serum from the original assessment for a new analysis. Due to the cross-sectional nature of the original study, a new IRB was obtained to recall subjects for a second MRI approximately 5 years later (mean time to follow-up of 56.5 ± 6.3 months) to assess for cartilage changes.

The inclusion criteria for the original cross-sectional study¹⁰ (baseline) were: Age greater than 40 years and medial compartmental knee OA (Kellgren–Lawrence (KL) grade ≥1) in at least one knee. Exclusion criteria were: Serious lower extremity (foot, ankle, knee, back, hips) injury or surgery; diagnosed or symptomatic OA in other lower extremity joints; radiographic evidence of isolated lateral compartment OA; gout or recurrent pseudogout; age greater than 85 years; body mass index (BMI) greater than 35 kg/m²; total knee or hip replacement in either leg; and inability to have a magnetic resonance imaging (MRI) scan. Bilateral anterior-posterior standing full leg radiographs were available for all OA participants. Radiographs underwent blinded grading by an experienced orthopedic surgeon (CRC) and a musculoskeletal radiologist (BHD) using the Kellgren–Lawrence (KL) radiographic scoring system.²⁷ Radiographs were also reviewed for evidence of isolated lateral compartment knee OA. While a KL grade was recorded no additional information was available for this convenience cohort. Thus radiographic scores were used for screening purposes only.

Serum Collection Protocol

Serum was collected before and after a 30 min walk during the baseline study.¹⁰ Briefly, subjects were asked to limit their physical activity 36 h prior to testing. On the day of the experiment, subjects were asked to limit their activity between the time they woke up and the initiation of study procedures which began within 2 h of waking. Following a 30 min period where subjects were asked to stay seated while completing the informed consent, MRI scans of both knees were collected. During MR collection, subjects remained in a prone position for 2 h. Subjects then walked at their self-selected pace for 30 min on a level walking track. Subjects were then returned to a seated position with minimal physical activity during which five-milliliter blood samples were collected by venous puncture at 0.5 and 5.5 h after the 30-min walking exercise. Serum collected 0.5 and 5.5 h (h) after a 30-min walking exercise at baseline were available for the current study.

Enzyme-Linked Immunosorbent Assay (ELISA)

Following collection, blood was allowed to clot for 30 min. The blood specimens were then centrifuged to separate serum, which were aliquoted, and frozen to −20°C within 1 h of collection, following which the specimens were transferred to −80°C freezers for storage until analysis. Serum concentrations of C1,2C and CS846 were measured in triplicate and in random order using commercial competitive ELISA kits according to manufacturer’s directions (IBEX, Montreal, QC, Canada). All serum samples were diluted 1:2.5, and all samples for any single subject were tested on the same plate.

Serum concentrations at the 0.5 h post-activity time point were used as a reference to assess biomarker changes following the mechanical stimulus. The use of the 0.5h time point was motivated by the availability of serum from medial knee OA patients at this time point and supported by a previous study¹⁰ that demonstrated serum levels of COMP at 0.5h post-activity did not differ from pre-activity levels. This finding with COMP was confirmed for the CS846 and C1,2C markers in a cohort of patients with knee OA from the baseline study where pre-activity serum was available (n= 16). Specifically the pre-activity serum concentrations of C1,2C (0.352±0.134μg/ml) vs. 0.5h post-activity (0.332±0.096μg/ml) were not significantly (p= 0.24) different and pre-activity concentration of CS846 (0.165±0.077μg/ml) vs. post-activity (0.158±0.72μg/ml) were not significantly different (p= 0.10). The changes in each biomarker were normalized to the mean of the cohort and expressed as a percent change relative to the 0.5h reference to depict where concentrations increased or decreased following the stimulus.

MRI Acquisition and Cartilage Thickness Measurements

MRI exams were performed at baseline and at a 5-year follow-up^10,11 on the more affected knee, as determined by KL grade, or in the case of equal KL grade between knees with more severe self-reported pain. All scans were performed on the same General Electric Signa 1.5 Tesla MRI machine (GE Medical Systems, Milwaukee, WI) with a standard transmit-receive extremity coil. Sagittal plane images of the knee were obtained using a fat-suppressed three-dimensional spoiled gradient recalled echo (SPGR) sequence with repetition time (TR) = 60 ms, echo time (TE) = 5 ms, flip angle = 40°, field of view 140 × 140 mm, slice thickness of 1.5 mm, 60 slices, and matrix 256 × 256.

The boundaries of the femoral and tibial cartilage were defined on the two-dimensional sagittal slices of the MRI scans using a semi-automatic segmentation method by a single experienced operator using custom software.²⁸ The coefficients of variation for this segmentation method have previously been shown to be less than 3%.²⁸ Three-dimensional models of the femoral and tibial cartilages were then constructed, and mean cartilage thicknesses were calculated over the medial tibial, medial femoral, lateral tibial, and lateral femoral regions as previously described²⁹ (Fig. 1). The medial and lateral femoral areas were partitioned into three sub-regions (central, internal, and external). Similarly, the total tibial medial and lateral areas were each partitioned into five sub-regions (central, internal, external, anterior, and posterior).²⁹ Change in cartilage thickness was calculated as the mean cartilage thickness at follow-up minus the mean cartilage thickness at baseline.

Figure 1.

Cartilage sub-regions of the femur and tibia. An illustration of the definitions of the subregions located on the femur and tibia³⁰ are overlaid on the corresponding areas of knee articular cartilage to the femur and tibia from a healthy donor.

Statistical Analysis

Changes in biomarker concentrations between 0.5 and 5.5 h were assessed by paired two-tailed Student’s t-tests. Associations between changes in the synthesis and degradation marker following the stimulus were assessed by calculation of Pearson correlation coefficients. Relations between changes in cartilage thickness over time and changes in biomarker levels, with 5.5 h post-activity concentrations expressed as a percentage of 0.5 h levels, were assessed by the calculation of Pearson correlation coefficients and by multiple linear regression analysis using backward stepwise elimination with additional variables of age, sex, and BMI. Given this population had primary medial compartment disease we tested for medial compartment thinning based on whether the serum concentrations increased or decreased following the mechanical stimulus using a two sample one-tailed Student’s t-test. SPSS software was used for analysis, version 20.0 (IBM SPSS Statistics).

RESULTS

Biomarker Resting Concentrations and Response to Stimulus

Biomarker changes were assessed following a mechanical stimulus in the form of a 30 min walk. The average walking speed for the 30 min walk was 1.27±0.19 m/s. The reference (0.5 h) concentrations of C1,2C and CS846 were 0.45±0.14 and 0.17±0.05 μg/ml, respectively. The 5.5 h sample showed the mean C1,2C concentration (0.53±0.20 μg/ml) significantly (p = 0.03) increased relative to reference whereas the mean CS846 concentration (0.17±0.05 μg/ml) was not significantly changed (p = 0.75).

Cartilage Changes and Biomarker Response to Stimulus

Cartilage thickness at the 5-year follow-up averaged over the full region of the lateral tibial surface was negatively correlated (R = −0.63, p = 0.009) with changes in the degradation biomarker C1,2C following the mechanical stimulus at baseline (Fig. 2a). In contrast, mean cartilage thickness over the full region of the lateral femoral condyle (Fig. 2b) was positively correlated (R = 0.76, p = 0.001) with changes to the synthesis biomarker CS846 concentration following the stimulus.

Figure 2.

(a) The cartilage thickness of the full region on the lateral tibia (red) significantly (p = 0.009) declined with increasing concentration of degradation marker (C1,2C) whereas (b) the thickness in the full regions on the lateral femoral condyle (blue) significantly (p = 0.001) increased with increasing concentration of the synthesis marker (CS846). Note: The concentration changes were expressed as a % change relative to the mean of the cohort at the 0.5 h time point.

There were also a number of sub-regions of the lateral compartments of the femur and tibia that showed significant correlations between thickness changes and changes in biomarker concentrations (Figs. 3 and 4). A common finding for sub-regions on the lateral tibia (Fig. 3) was a negative correlation between thickness changes and changes in the C1,2C marker. Significant negative correlations were found in the anterior sub-region vs. C1,2C (R = −0.68, p= 0.004) and the external sub-region vs. C1,2C (R = −0.53, p= 0.036). The nature of the thickness change versus biomarker changes for sub-regions of the lateral femur (Fig. 4) was different than the tibia as all the correlations for the femur were positive in relationship to the synthesis marker CS846. Specifically, there were significant positive correlations in the internal subregion versus CS846 (R = 0.77, p= 0.001) and the central sub-region versus CS846 (R = 0.76, P= 0.001). All results remained significant after the analyses were adjusted for age, sex, and BMI, with only BMI entering the regression for the internal sub-region of the lateral tibia vs. CS846. There were no significant correlations found for the medial compartment.

Figure 3.

The biomarker-thickness relationships on the lateral tibial articular surface had significant negative correlations for the (a) anterior (p = 0.004) and (b) external (p = 0.036) sub-regions. Note: The concentration changes were expressed as a % change relative to the mean of the cohort at the 0.5 h time point.

Figure 4.

The biomarker-thickness relationships on the lateral femoral articular surface had significant positive correlations for the (a) internal (p = 0.001) and (b) central (p = 0.001) sub-regions with CS846. Note: The concentration changes were expressed as a % change relative to the mean of the cohort at the 0.5 h time point.

Biomarker Concentration Changes After Mechanical Stimulus Differed in Magnitude and Direction

There were substantial variations (Fig. 5a) among different patients in the magnitude and direction of the biomarker changes following the stimulus where some patients had substantial increases reaching nearly 50% while others showed substantial decreases reaching 40%. The number of patients showing increases and decreases were equal (eight each) for the C1,2C marker, while for the CS846 marker more patients (11 of 16) showed decreased concentrations following the stimulus (Fig. 5a). Interestingly, the change in the synthesis (CS846) and degradation (C1,2C) biomarkers following the mechanical stimulus were significantly correlated (R² = 0.28, P = 0.037, Fig. 5b).

Figure 5.

(a) There were individual variations in C1,2C and CS846 biomarkers changes following the stimulus where some patients had substantial increases and some patients had substantial decreases relative to the mean of the sample. (b) There was a significant (p = 0.04) correlation between the synthesis marker (CS846) and the degradation marker (C1,2C).

Increased Levels of C1,2C After Mechanical Stimulus Signal Greater Medial Tibial Cartilage Thinning

Patients with an increase in C1,2C concentration following the mechanical stimulus had cartilage thinning in the medial tibial compartment 5 years later. Specifically, the group averages showed significantly more thinning in the anterior (P = 0.048) and posterior (P = 0.049) regions of the medial tibia for patients (n = 8) with an increased serum concentration of C1,2C following the mechanical stimulus (Fig. 6) relative to patients (n = 8) with a decreased concentration as illustrated in Figure 5a. Patients grouped based on baseline changes in CS846 where only five of 16 patients showed increased concentration did not show significant thickness changes in the medial compartment.

Figure 6.

Cartilage thinning over five years was significantly greater in the anterior (p = 0.05) and posterior (p = 0.05) regions of the medial tibia for patients grouped with an increase in the degradation marker (C1,2C) relative to thickness changes for the group with a decreased level of C1,2C following the stimulus.

DISCUSSION

The results of this study further support the potential utility of introducing a mechanical stress test^11,12 to enhance the use of biomarkers for OA. Specifically, testing Hypothesis 1 showed a mechanical stimulus in the form of a 30-min walk elicited changes in serum biomarkers of cartilage degeneration (C1,2C) and synthesis (CS846) that were associated with cartilage thickness changes 5 years later. A unique finding of this study (Hypothesis 2) was that the synthesis and degradation biomarkers provided different information regarding the nature of cartilage change with progressive knee OA. For example, increases in the synthesis marker (CS846) following stimulus signaled cartilage thickening in the lateral femoral condyle whereas elevations to the degradation marker C1,2C were associated with greater cartilage thinning to the lateral tibial plateau. The results showing these differential associations found in the less involved lateral compartment of patients with medial OA are consistent with regional cartilage changes commonly reported^23,30–32 in early OA. Finally the correlation (Hypothesis 3) between changes in the synthesis and degradation biomarkers is consistent with a metabolic linkage^6–8 among biomarkers that is enhanced by the mechanical stimulus. Further, taken together with the finding that there was increased average thinning in the medial compartment for patients with increased C1,2C relative to patients with no increase in C1,2C, these results suggests that an increase in biomarker concentration following stimulus reflects advancing disease.

The C1,2C and CS846 results of the current study along with previous findings of a similar study¹¹ with COMP are important because the three markers represent distinctly different cartilage properties including matrix turnover²⁵ (COMP), degradation³³ (C1,2C) and synthesis⁶ (CS846). Specifically, the finding that it took several hours after a mechanical stimulus to detect meaningful serum concentration changes for all three markers¹¹ suggests that the markers are linked by metabolic activity occurring in response to the walking task. The correlation between C1,2C and CS846 found in this study also supports a metabolic linkage between markers in a knee OA population. The fact that the direction of the change of each of the three markers was associated with different regions of cartilage thickening or thinning supports the conclusion that the different markers reflect differing disease states and activities occurring in different regions of the joint. The results showing that patients with increased C1,2C had greater thinning in the anterior medial region of the medial tibia (Fig. 6) was consistent with the previous study¹¹ of COMP showing that increased COMP was also associated with changes in this region.

The nature of the relationship between the C1,2C degradation marker and medial compartment thickness changes raises some interesting points regarding the influence of gait mechanics on patterns of cartilage changes in patients with medial knee OA. Specifically, cartilage thicknesses in the medial compartment have been associated with individual variations in gait mechanics.^1,25,34–36 Thus, gait differences between patients with medial compartment knee OA would preclude finding a proportional relationship between biomarkers and cartilage changes. However, the finding that when C1,2C increased there was average cartilage thinning in the anterior and medial regions of the medial tibial (Fig. 6) was consistent with gait studies^25,35,36 reporting sagittal plane kinematics changes common to knee OA that can alter the location of the anterior and posterior regions of tibial-femoral contact. While it was beyond the scope of this exploratory study to account for gait variations, the results suggest that the direction (increase vs. decrease) of the biomarker changes should be considered for assessing medial compartment thinning.

The relationship between the C1,2C marker and medial compartment thinning found in this study (Fig. 6) suggests that it is important to consider the possibility that the direction (increase or decrease) of the biomarker concentration change is as important as the magnitude of the change. Of note, patients with an increase in C1,2C, irrespective of the magnitude, after mechanical stimulus had increased medial compartment thinning consistent with progression of medial OA. Whereas, if C1,2C decreased, no significant thinning was observed. There is likely a threshold effect relative to the level of increase or decrease. However, these findings suggest that the central tendencies (average) of groupings based on directional biomarker changes appear to reflect directional differences in cartilage changes, even with a relatively small population. A similar bifurcation between biomarker increases and decreases can also be seen in the relatively less involved lateral compartment (Fig. 2b). Specifically, the result showing thickening in the lateral compartment when the synthesis marker (CS846) increased, taken together with prior studies^6–8 reporting elevated metabolism in early OA, suggests evidence of early OA. Thus, it appears that the mechanical stimulus can produce changes in biomarker concentrations that reflect different OA disease states depending on the direction of the change.

In evaluating the results of this study, one must consider that the relatively small cohort limits generalization of these results. Further, the nature of the convenience sampling from a larger cohort might impose selection bias, and also limit the ability to test additional markers and assess follow-up radiographs. The exploratory nature of this study precluded corrections for multiple comparison and raises the risk of type 1 errors. Clearly, other biomarkers may show a response to the mechanical stimulus. However, considering this was the first study of markers other than COMP previously evaluated for OA¹³ using the stimulus-response protocol, the selection of C1,2C and CS846 was a logical first step. In addition, despite the small numbers, significant correlations were found between changes in cartilage thickness over 5 years and the changes in cartilage biomarkers following a mechanical stimulus obtained at study entry 5 years earlier. That the significant correlations were limited to the lateral compartment in a cohort with medial knee OA might be related to the differential disease states within the medial and lateral compartments as well as the fact that medial compartment cartilage change can be substantially influenced by gait mechanics.^25,35,36

In conclusion, a mechanical stimulus can produce measurable changes in both markers of synthesis and degradation that signal regional changes (thickening and thinning) in cartilage thickness 5 years later. The walking stimulus appears to activate a metabolic response that links synthesis and degradation markers reflecting different stages of the disease where a biomarker increase suggests a more active disease state. Specifically, the biomarker changes appear to signal different stages of the disease where increased concentrations of C1,2C indicated cartilage loss (thinning), whereas an increase in the synthesis marker CS846 signaled cartilage thickening consistent with early disease. Finally, these results indicate that the response of serum biomarkers to a mechanical stimulus provides signals consistent with disease progression over time, as well as new insight into multidisciplinary approaches that can overcome many of the barriers to the use of biomarkers for OA.