Abstract
Objective
To evaluate the ability of the macrophage markers CD163 and CD14 to predict different osteoarthritis (OA) phenotypes defined by severity of joint inflammation, radiographic features and progression, and joint pain.
Methods
We evaluated 2 different cohorts totaling 184 patients with radiographic knee OA. These included 25 patients from a cross-sectional imaging study for whom there were data on activated macrophages in the knee joint, and 159 patients (134 with 3-year longitudinal data) from the longitudinal Prediction of Osteoarthritis Progression study. Multivariable linear regression models with generalized estimating equations were used to assess the association of CD163 and CD14 in synovial fluid (SF) and blood with OA phenotypic outcomes. Models were adjusted for age, sex, and body mass index. P values less than or equal to 0.05 were considered significant.
Results
SF CD14, SF CD163, and serum CD163 were associated with the abundance of activated macrophages in the knee joint capsule and synovium. SF CD14 was positively associated with severity of joint space narrowing and osteophytes in both cohorts. SF and plasma CD14 were positively associated with self-reported knee pain severity in the imaging study. Both SF CD14 and SF CD163 were positively associated with osteophyte progression.
Conclusion
Soluble macrophage biomarkers reflected the abundance of activated macrophages and appeared to mediate structural progression (CD163 and CD14) and pain (CD14) in OA knees. These data support the central role of inflammation as a determinant of OA severity, progression risk, and clinical symptoms, and they suggest a means of readily identifying a subset of patients with an active inflammatory state and worse prognosis.
With the advent of biologic therapies for rheumatic diseases, there is a growing armamentarium of effective therapeutics targeting specific inflammatory mediators. With the growing consensus that osteoarthritis (OA) is an inflammatory condition (1,2), many of these new therapies might be brought to bear on OA if a means could be developed to readily identify individuals with inflammatory disease phenotypes. One means of identifying inflammatory phenotypes in vivo, etarfolatide (EC20) imaging, has recently become available (3). This imaging technique is dependent on detecting the selective expression of folate receptor (FR) on activated, but not resting, macrophages (3–5). In a recent study, we showed that the quantity of FRβ+ macrophages in joints, based on EC20 imaging, correlated with radiographic knee OA severity and symptoms (6).
Although it is a powerful means of phenotyping patients, EC20 entails radiation exposure; it would therefore be desirable to identify soluble biomarkers correlating with EC20 imaging that could be used adjunctively or independently of imaging techniques to identify patient subsets. The goal of this study was therefore to identify soluble biomarkers of activated macrophages that could provide a means of identifying inflammatory OA phenotypes in a clinical setting through analysis of biologic specimens (blood or synovial fluid [SF]). We hypothesize that soluble macrophage markers, shed into the systemic circulation upon local macrophage activation in the joint, correlate with the level of joint tissue inflammation as quantified by EC20 imaging. We also hypothesize that macrophage markers may be used as predictors of OA progression. We focused on 2 macrophage markers, CD163 and CD14. Both CD163 (7) and CD14 (8) have been observed to be coexpressed with FRβ on tumor-associated macrophages (9). Both of these macrophage markers exist in soluble form and can be measured in serum/plasma and SF.
CD163 cell surface expression is limited to the monocyte/macrophage lineage (10) and is activated by antiinflammatory signals such as glucocorticoids (11) and certain cytokines (interleukin-6 [IL-6] and IL-10) (12,13). CD163 cell surface and gene expression is down-regulated by proinflammatory mediators such as lipopolysaccharide (LPS), tumor necrosis factor α (TNFα), and interferon-γ (IFNγ) (12). In 1 study, CD163+FRβ+ mononuclear cell numbers were increased in OA synovial tissue compared to rheumatoid arthritis (RA) synovial tissue (14). Several signals, including phorbol 12-myristate 13-acetate (PMA) and LPS, induce metalloprotease-dependent cleavage of ectodomain of CD163, which subsequently results in the shedding of soluble CD163 from the surface of macrophages (10).
CD14 is found on various cell types, predominantly monocytes and macrophages (15). CD14 serves as a receptor for the bacterial LPS–LPS binding protein complex (16). The CD14–LPS–LPS binding protein complex then binds to cell surface Toll-like receptor 4 (TLR-4)–myeloid differentiation protein 2 complexes (17), leading to activation of innate host defense mechanisms, including production of the inflammatory mediators TNFα, IL-1, IL-6, IL-8, IL-10, and IL-12 (18). Like CD163, various stimuli (PMA, IFNγ, and LPS) induce shedding of soluble CD14 via membrane-associated serine proteases (19). The soluble forms of CD163 and CD14 can therefore be considered a reflection of the inflammatory activation status of macrophages.
PATIENTS AND METHODS
Study cohorts
EC20 scan cohort
A total of 25 participants with radiographic knee OA (unilateral or bilateral Kellgren/Lawrence [K/L] grade 1–4 [20]) underwent bilateral knee EC20 imaging. Briefly, after participants were injected intravenously with 99mTc-EC20, both knees were imaged with single-photon–emission computed tomography (SPECT) combined with conventional CT (SPECT/CT) (6). The study included 7 men and 18 women ages 30–89 years (mean ± SD 62.4 ± 15.8 years) with a mean ± SD body mass index (BMI) of 29.2 ± 4.7 kg/m2 (range 22.5–38.4). The percentages of each K/L grade for all 50 knees were as follows: grade 1 24%, grade 2 26%, grade 3 32%, and grade 4 18%. Of the 50 knees, SF was obtainable by direct aspiration from 28 knees and by lavage (when it was not possible to obtain fluid directly) from 19 knees; SF was unobtainable from 2 knees, and arthrocentesis was refused for 1 knee. SF volumes were sufficient for quantification of CD163 concentrations in 47 samples and CD14 concentrations in 44 samples (Figure 1A). Marker concentrations in lavage fluid were corrected for dilutional effects of lavage based on an established urea method (21).
Figure 1.
Description of data available and used for each cohort. Shown for each cohort are sample size, imaging data (abundance of macrophages by single-photon–emission computed tomography [SPECT] combined with conventional CT [SPECT/CT] and radiographic knee osteoarthritis [OA] data), and knee symptom data. A, Etarfolatide (EC20) scan cohort. B, Prediction of Osteoarthritis Progression (POP) cohort. K/L = Kellgren/Lawrence; SF = synovial fluid; OST = osteophyte; JSN = joint space narrowing; TF = tibiofemoral.
Prediction of Osteoarthritis Progression (POP) cohort
The POP cohort provided baseline SF and serum samples and baseline and longitudinal (3-year) followup knee radiographic data to assess the association of CD163 and CD14 with knee OA radiographic severity and structural progression. A total of 159 participants (41 men and 118 women ages 35–85 years [mean ± SD 63.7 ± 11.8 years]) with K/L grade 1–3 severity radiographic OA of the index knee (22) were enrolled at baseline; participants met the American College of Rheumatology criteria for symptomatic knee OA (23). Their mean ± SD BMI was 31.2 ± 6.7 kg/m2 (range 18.3–53.1). The percentages of each K/L grade for the 308 nonreplaced knees were as follows: grade 0 2.3%, grade 1 22.4%, grade 2 20.1%, grade 3 45.5%, and grade 4 9.7%. Baseline SF was obtainable by direct aspiration from 111 knees and by lavage from 72 knees. SF volumes were sufficient for quantification of CD163 concentrations in 171 samples and CD14 concentrations in 172 samples (Figure 1B). A total of 134 participants returned for followup (36 men and 98 women). Because both knees were studied, a total of 308 (nonreplaced) knees provided baseline data and 246 (nonreplaced) knees were available at followup for computing progression data.
Participants in both cohorts were excluded on the basis of inflammatory joint diseases, endocrinopathies, Paget’s disease, neuropathic disorders, avascular necrosis, use of corticosteroids within the previous 3 months, and knee arthroscopy within the previous 6 months. In addition to SF, baseline sera and EDTA plasma were obtained from all study participants. All samples were stored at −80°C until analyzed, and all study procedures were approved by the Duke University Institutional Review Board.
Image scoring
The intensity of the scintigraphic radiolabel on EC20 SPECT/CT scans of the knee was scored semiquantitatively as normal (score 0), mild (score 1), moderate (score 2), or intense (score 3) by an experienced nuclear medicine radiologist with high intrarater reliability (6). Scores were summed across medial and lateral knee compartments by site of tissue localization (tibiofemoral knee capsule or synovium) (6).
Subjects in both cohorts underwent posteroanterior fixed-flexion weight-bearing knee radiography with a Syna-Flexer lower limb positioning frame (Synarc) and a 10° caudal x-ray beam angle (24). Each knee radiograph was scored for K/L grade (0-4) and individual radiographic OA features of joint space narrowing (JSN; score 0-3) and osteophyte severity (score 0-3) in the medial and lateral compartments using the Osteoarthritis Research Society International standardized atlas (25). Total JSN scores of 0–6 were possible for each knee and total osteophyte scores of 0–12 were possible as all 4 margins of the knee joint were scored. Patients from the POP cohort were categorized as “control,” “nonprogressor,” or “progressor” for osteophytes and for JSN based on the knee with the greatest change in osteophyte or JSN scores, respectively, between baseline and the 3-year followup visit. Seventeen patients had undergone total knee replacement for OA between visits and were categorized as “progressors.” A total of 29% of knee radiographs were reread blinded to the original scores, and intraclass correlation coefficients (ICCs) were high (JSN ICC 0.86, osteophyte ICC 0.84) (22).
Pain assessment
In both cohorts, knee symptoms were ascertained by the First National Health and Nutrition Examination Survey (NHANES-I) criterion (26) of pain, aching, or stiffness in participants’ individual knees on most days of any 1 month in the last year. Pain was graded as none (score 0), mild (score 1), moderate (score 2), or severe (score 3).
Measurement of soluble CD163 and CD14
CD163 and CD14 were analyzed in baseline serum and plasma, respectively, and both were analyzed in SF. Immediately before enzyme-linked immunosorbent assay (ELISA) measurements, SF was treated for 1 hour at 37°C with 5 units/ml hyaluronidase (27) from Streptomyces hyalurolyticus (Sigma-Aldrich) in 20 mM sodium phosphate, 77 mM sodium chloride, pH 7, 0.01% bovine serum albumin.
CD163 was measured using a Quantikine Human sCD163 Immunoassay (R&D Systems) according to the manufacturer’s protocol. Sera were diluted 10-fold into Calibrator Diluent RD5-24 (1×). SF aspirated directly from the knee joint without lavage was diluted 50-fold, while lavage SF samples (aspirated with 10 ml saline “lavage” when a direct aspirate was not obtained) were diluted up to 12-fold. The optical density of each well was measured using a microplate reader (Infinite 200 PRO; Tecan) set to read the absorbance at 450 nm with a wavelength correction set to 540 nm. Samples were tested in duplicate. A standard curve was plotted and a best-fit curve was generated by regression analysis of the log–log data (Prism version 5.0a; GraphPad Software). The interassay coefficient of variation (CV) of the standards was 10.0%. The intraassay CV ranged from 3.3% to 6.8%. The minimum detectable concentration reported by the manufacturer was 0.177 ng/ml. The concentrations of 12 lavage SF samples from the POP cohort were below the minimum detectable level and were excluded from the analyses.
CD14 was measured using a Quantikine ELISA Human sCD14 Immunoassay (R&D Systems) according to the manufacturer’s protocol. Plasma samples were diluted 200-fold into Calibrator Diluent RD5P (1×). Direct SF was diluted 1,000-fold, while lavage samples were diluted up to 40-fold. The optical density measurements and subsequent analysis steps were performed as described above. The interassay CV of the standards was 12.1%. The intraassay CV ranged from 1.2% to 9.2%. The minimum detectable concentration reported by the manufacturer was 0.125 ng/ml. The concentrations of 11 lavage SF samples from the POP cohort were below the minimum detectable level and were excluded from the analyses.
Statistical analysis
By assessment of skewness and kurtosis, biomarker data were determined to be normally distributed and therefore were not log-transformed. Biomarker concentrations are reported in ng/ml, but all parameter estimates were generated with CD163 and CD14 concentrations in µg/ml. The correlation of CD163 and CD14 was assessed using Pearson’s correlation. Paired SF and blood samples were analyzed by paired t-test. P values less than or equal to 0.05 were considered significant. Since the analyses were exploratory, no adjustment for multiple testing was performed.
Multivariable linear regression models using generalized estimating equations (GEEs), which account for correlation within knees, were performed to evaluate associations of CD163 and CD14 concentrations with the following: intensity of folate uptake in the tibiofemoral knee joint capsule and synovium (representing abundance of activated macrophages reflecting synovitis), JSN and osteophyte severities, JSN and osteophyte progression over 3 years, and joint symptom scores. Models testing the associations of CD163 and CD14 with JSN and osteophyte progression were additionally adjusted for baseline JSN or osteophyte severity, respectively. The parameter estimate (β), which represents the estimated mean for the slope, was reported along with the 95% confidence interval and P value. The data were adjusted for age, sex, and BMI. To determine if differences in OA severity between the cohorts affected the results, we evaluated the association of macrophage markers with OA phenotypes and joint symptoms in the data sets without K/L grade 4 knee samples.
The combinatorial effect of CD163 and CD14 was evaluated using the quasi-likelihood information criterion (QIC) (28) with the assumption that CD163 and CD14 concentrations had a linear relationship with the outcomes. The QICu values, which are defined as the quasi-likelihood (Q) + double the number of parameters (2p) in the model, were compared to evaluate the individual and joint contributions of the biomarkers in each GEE model with CD163, with CD14, and with both CD163 and CD14. The QICu statistic is used to measure the nonredundant variance in a model.
RESULTS
Soluble macrophage marker concentrations
SF CD163 concentrations measured in the combined cohorts ranged from 4.0 ng/ml to 20,106.8 ng/ml, with a mean value of 964.7 ng/ml (n = 218). Serum CD163 concentrations ranged from 178.4 ng/ml to 1,929.0 ng/ml, with a mean value of 706.6 ng/ml (n = 183); this mean value was greater than the mean ± SD concentration reported by the ELISA manufacturer (472 ± 186 ng/ml; n = 36) for a sample of individuals with unknown medical history. SF CD14 concentrations measured in the combined cohorts ranged from 3.5 ng/ml to 7,876.5 ng/ml, with a mean value of 1,199 ng/ml (n = 216). Plasma CD14 concentrations ranged from 975.5 ng/ml to 3,839.2 ng/ml, with a mean value of 2,200 ng/ml (n = 184); this mean concentration was greater than the mean concentration reported by the ELISA manufacturer for healthy individuals of unknown OA status (1,800 ng/ml, range 1,200–2,600 ng/ml; n = 34). CD163 concentrations were significantly (P = 0.009) greater in SF than in sera between paired biologic specimens (Figure 2A). Conversely, CD14 concentrations were significantly lower (P < 0.0001) in SF than in paired plasma (Figure 2B). SF CD14 and CD163 concentrations in the combined cohorts were highly correlated (r = 0.59, P < 0.0001) (Figure 2C); however, plasma CD14 and serum CD163 concentrations were not correlated (Figure 2D). The correlations among the measured biomarkers within each of the cohorts are summarized in Table 1.
Figure 2.
Concentrations of soluble macrophage-associated markers, CD163 and CD14. A, CD163 concentrations in serum and synovial fluid (SF) samples (n = 218) were measured simultaneously. The mean concentration in SF (964.7 ng/ml) was significantly greater than that in serum (706.6 ng/ml). B, CD14 concentrations in plasma and SF samples (n = 216) were measured simultaneously. The mean concentration in SF (1,199 ng/ml) was significantly lower than that in plasma (2,200 ng/ml). C, SF CD163 and CD14 concentrations were highly correlated. D, Serum CD163 and plasma CD14 concentrations were not correlated. Symbols represent individual samples; bars in A and B show the mean ± SD. P values in A and B were determined by paired t-test.
Table 1.
Correlations among the measured biomarkers in the etarfolatide (EC20) scan cohort and Prediction of Osteoarthritis Progression (POP)cohort*
| Synovial fluid CD163 | Synovial fluid CD14 | Serum CD163 | Plasma CD14 | |
|---|---|---|---|---|
| EC20 cohort | ||||
| Synovial fluid CD163 | 1 | – | – | – |
| Synovial fluid CD14 | 0.87 (<0.0001) | 1 | – | – |
| Serum CD163 | 0.38 (0.009) | 0.34 (0.02) | 1 | – |
| Plasma CD14 | 0.19 (0.2) | 0.24 (0.1) | 0.48 (0.0004) | 1 |
| POP cohort | ||||
| Synovial fluid CD163 | 1 | – | – | – |
| Synovial fluid CD14 | 0.57 (<0.0001) | 1 | – | – |
| Serum CD163 | −0.11 (0.2) | 0.10 (0.2) | 1 | – |
| Plasma CD14 | −0.009 (0.9) | 0.07 (0.38) | 0.14 (0.01) | 1 |
Values are Pearson’s r (P).
Association of soluble macrophage markers with the presence of activated macrophages in OA
Both biomarkers were significantly associated with activated macrophages in the joints of patients with knee OA in the EC20 scan cohort (Table 2). SF CD163 and SF CD14 were significantly associated with the presence of macrophages in the knee joint capsule (β = 0.989, P = 0.005 and β = 1.293, P = 0.002, respectively) and synovium (β = 1.442, P = 0.002 and β = 1.806, P = 0.0005, respectively). Systemically, only serum CD163 was significantly associated with the presence of activated macrophages in the knee joint capsule (β = 2.308, P = 0.006) and synovium (β = 3.225, P = 0.002) (Table 3).
Table 2.
Association of synovial fluid CD163 and CD14 concentrations with OA phenotypes*
| Synovial fluid CD163 | Synovial fluid CD14 | |||
|---|---|---|---|---|
| OA phenotype | β (95% CI)† | P | β (95% CI)† | P |
| Activated macrophages in joint capsule‡ | 0.989 (0.368, 1.610) | 0.005 | 1.293 (0.569, 2.016) | 0.002 |
| Activated macrophages in synovium‡ | 1.442 (0.625, 2.259) | 0.002 | 1.806 (0.955, 2.657) | 0.0005 |
| Osteophyte severity‡ | 2.820 (1.524, 4.116) | 0.0003 | 3.473 (2.121, 4.825) | <0.0001 |
| Osteophyte severity§ | 0.093 (−0.132, 0.319) | 0.418 | 0.686 (0.433, 0.939) | <0.0001 |
| JSN severity‡ | 0.422 (−0.048, 0.892) | 0.093 | 0.608 (0.116, 1.099) | 0.025 |
| JSN severity§ | 0.042 (−0.003, 0.087) | 0.066 | 0.098 (0.007, 0.189) | 0.035 |
| Osteophyte progression§ | 0.072 (0.020, 0.123) | 0.007 | 0.096 (0.039, 0.185) | 0.015 |
| Osteophyte progression§¶ | 0.056 (0.027, 0.085) | <0.0001 | 0.043 (−0.030, 0.115) | 0.246 |
| JSN progression§ | 0.006 (−0.013, 0.026) | 0.528 | 0.006 (−0.040, 0.052) | 0.802 |
| JSN progression§# | 0 (−0.017, 0.016) | 0.975 | −0.018 (−0.061, 0.025) | 0.411 |
OA = osteoarthritis; β = parameter estimate (based on CD163 and CD14 concentrations in µg/ml); 95% CI = 95% confidence interval.
Adjusted for age, sex, and body mass index.
Etarfolatide scan cohort.
Prediction of Osteoarthritis Progression cohort.
Adjusted for age, sex, body mass index, and baseline osteophyte severity.
Adjusted for age, sex, body mass index, and baseline joint space narrowing (JSN) severity.
Table 3.
Association of serum CD163 and plasma CD14 concentrations with OA phenotypes*
| Serum CD163 | Plasma CD14 | |||
|---|---|---|---|---|
| OA phenotype | β (95% CI)† | P | β (95% CI)† | P |
| Activated macrophages in joint capsule‡ | 2.308 (0.824, 3.791) | 0.006 | 0.859 (−0.128, 1.846) | 0.103 |
| Activated macrophages in synovium‡ | 3.225 (1.475, 4.975) | 0.002 | 1.088 (−0.147, 2.323) | 0.099 |
| Osteophyte severity‡ | 3.663 (−0.526, 7.851) | 0.102 | 0.359 (−2.256, 2.974) | 0.791 |
| Osteophyte severity§ | −1.236 (−2.051, −0.421) | 0.003 | 0.199 (−0.419, 0.818) | 0.528 |
| JSN severity‡ | 1.470 (0.0529, 2.887) | 0.055 | 0.259 (−0.640, 1.158) | 0.578 |
| JSN severity§ | −0.031 (−0.376, 0.314) | 0.860 | −0.077 (−0.292, 0.138) | 0.485 |
| Osteophyte progression§ | −0.106 (−0.368, 0.158) | 0.437 | 0.060 (−0.157, 0.277) | 0.586 |
| Osteophyte progression§¶ | 0.017 (−0.220, 0.253) | 0.889 | 0.046 (−0.141, 0.234) | 0.628 |
| JSN progression§ | −0.017 (−0.138, 0.104) | 0.786 | 0.097 (0.018, 0.211) | 0.097 |
| JSN progression§# | −0.057 (−0.176, 0.063) | 0.351 | 0.117 (0.010, 0.225) | 0.032 |
See Table 2 for definitions.
Adjusted for age, sex, and body mass index.
Etarfolatide scan cohort.
Prediction of Osteoarthritis Progression cohort.
Adjusted for age, sex, body mass index, and baseline osteophyte severity.
Adjusted for age, sex, body mass index, and baseline JSN severity.
Association of soluble macrophage markers with radiographic knee OA severity and progression
SF CD163 was significantly associated with osteophyte severity (β = 2.820, P = 0.0003) and serum CD163 showed a trend toward association with JSN severity (β = 1.470, P = 0.055) in the EC20 scan cohort (Tables 2 and 3). SF CD14 was significantly associated with osteophyte severity in both the EC20 scan cohort (β = 3.473, P < 0.0001) and the POP cohort (β = 0.686, P < 0.0001) (Table 2). SF CD14 was also significantly associated with JSN severity in both the EC20 scan cohort (β = 0.608, P = 0.025) and the POP cohort (β = 0.098, P = 0.035). SF CD163 (β = 0.072, P = 0.007) and SF CD14 (β = 0.096, P = 0.015) were both significantly positively associated with osteophyte progression over 3 years in patients with knee OA in the POP cohort (Table 2). Systemically, only serum CD163 was significantly associated with osteophyte severity in the POP cohort, but this was an inverse association (β = −1.236, P = 0.003) (Table 3).
Independent association of macrophage markers with OA progression
SF CD163 was significantly associated with osteophyte progression (β = 0.056, P < 0.0001) (Table 2) after adjustment for baseline osteophyte severity, with osteophyte severity having a significant contribution in the model (β = 0.076, P = 0.002). SF CD14 was not associated with osteophyte progression (β = 0.043, P = 0.246) after adjustment for baseline osteophyte severity, with osteophyte severity having a significant contribution in the model (β = 0.082, P = 0.002). An association between plasma CD14 and JSN progression was unmasked (β = 0.117, P = 0.032) (Table 3) after adjustment for baseline JSN severity, with JSN severity having a significant contribution in the model (β = 0.187, P < 0.0001).
Association of soluble macrophage marker CD14 with joint symptoms
Self-reported knee joint symptom severity ranged from 0 to 3 in both cohorts, with a higher mean ± SD value observed in the EC20 scan cohort (1.58 ± 0.86) than in the POP cohort (1.40 ± 0.82). Concentrations of CD163 and CD14 were evaluated for association with self-reported joint symptoms. Only CD14, in both the SF (β = 0.773, P = 0.003) and plasma (β = 0.641, P = 0.031) from patients in the EC20 scan cohort, exhibited a correlation with joint symptoms (Table 4).
Table 4.
Association of CD163 and CD14 with joint symptoms*
| CD163 | CD14 | |||
|---|---|---|---|---|
| Sample type | β (95% CI)† | P | β (95% CI)† | P |
| Synovial fluid‡ | 0.341 (−0.065, 0.747) | 0.116 | 0.773 (0.334, 1.212) | 0.003 |
| Synovial fluid§ | 0.067 (−0.004, 0.138) | 0.063 | 0.031 (−0.085, 0.146) | 0.602 |
| Serum/plasma‡ | 0.387 (−0.603, 1.377) | 0.452 | 0.641 (0.101, 1.182) | 0.031 |
| Serum/plasma§ | 0.112 (−0.142, 0.366) | 0.389 | 0.102 (−0.144, 0.349) | 0.415 |
See Table 2 for definitions.
Adjusted for age, sex, and body mass index.
Etarfolatide scan cohort.
Prediction of Osteoarthritis Progression cohort.
Differences in results between cohorts are independent of OA severity
To determine if differences in OA severity between the cohorts could account for the observed cohort effect, we evaluated the association of the biomarkers with OA phenotypes and joint symptoms in the data sets without K/L grade 4 knee samples. Indeed, the exclusion of K/L grade 4 knee samples from the data set generally resulted in lower parameter estimates, but results for the EC20 cohort still remained significant (further information is available at http://dmpi.duke.edu/faculty/virginia-kraus-md-phd).
Association of OA phenotypic outcomes with the combination of CD163 and CD14
The association of the combination of CD163 and CD14 with OA phenotypic outcomes was assessed by relative comparison of the QICu values from each GEE model with CD163, with CD14, and with the combination of CD163 and CD14. A lower QICu value indicates a more suitable model, with a difference >10 indicating a substantial improvement in the model (29). Only the combination of SF CD163 and SF CD14 in the POP cohort provided a substantially better model than the individual biomarkers alone (further information is available at http://dmpi.duke.edu/faculty/virginia-kraus-md-phd).
DISCUSSION
The abundance of activated macrophages in the knee joint, as indicated by the intensity of 99mTc-EC20 signal on SPECT/CT images, was significantly associated with soluble SF CD163, serum CD163, and SF CD14. These results demonstrate that soluble macrophage markers, which reflect shedding of these proteins from activated macrophages, are strongly indicative of the active and dynamic inflammatory state of the knee joint with OA. Moreover, the strong correlation of SF CD163 with SF CD14 indicates the coordinated shedding of both markers from joint tissue macrophages. The significant positive association of baseline SF macrophage marker concentrations with OA progression status, defined by osteophyte formation after 3 years, implicates macrophage-related synovitis as a driver of OA structural damage and progression. This was further supported by the significant independent association of shed CD163 with osteophyte progression after controlling the statistical model for baseline osteophyte severity. These data are consistent with animal models showing the involvement of synovial macrophages in osteophyte formation and growth (30). Our results are also consistent with RA studies, in which SF CD163 was associated with level of disease activity (31) and radiographic progression (32).
Phenotyping patients by detecting a marker in the blood, rather than by subjecting them to arthrocentesis, would obviously be more conducive to their cooperation. These results are therefore especially remarkable for the ability of soluble serum CD163 to reflect the local state of inflammation in the joint. Ectodomain shedding and the release of soluble CD163 can increase severalfold during various pathologic conditions (for review, see ref. 10), such as autoimmune disorders, sepsis, liver disease, Gaucher’s disease, reactive hemophagocytic syndrome, multiple sclerosis, and malaria. Our results demonstrate that OA can now be added to the list of pathologic conditions associated with the release of soluble CD163. The significantly higher concentrations of soluble CD163 in SF than in paired serum is consistent with local shedding of CD163 from activated macrophages in inflamed joints; these results are also consistent with results in RA (31) and spondyloarthritis (33). Our results therefore suggest that serum CD163 may be useful for OA phenotyping to identify a subset of patients with an active inflammatory state.
CD14 concentrations in the SF, but not in plasma, of patients with knee OA were associated with activated macrophages in the knee joint. However, unlike CD163, the mean concentration of CD14 in SF was significantly lower than that in plasma between matched biospecimen pairs. This suggests that there are peripheral sources of soluble CD14, such as the dorsal root ganglia (DRGs) in association with joint disease (as described below), or that soluble CD14 may be cleared rapidly from the joint fluid into plasma, where it might have a longer residence time. In addition to being significantly associated with OA status defined by osteophyte severity, SF CD14 correlated with JSN severity; this is intriguing because investigators have rarely been able to identify a marker for JSN severity. Interestingly, adjustment for baseline JSN severity unmasked a significant association of plasma CD14 with progression of JSN over 3 years. Thus, plasma CD14 was linked with both pain and JSN progression. Elevated concentrations of serum CD14 have been documented in a number of inflammatory diseases, including RA (34), systemic lupus erythematosus (35), and atopic dermatitis (36). In addition to binding LPS, CD14 mediates the recognition and phagocytosis of cells undergoing apoptosis (37) and may be an indication of the increased catabolic processes occurring in OA joints.
Interestingly, EC20 uptake in OA joints correlated with self-reported pain (6). Here, we report that soluble CD14, both in SF and plasma, was associated with self-reported knee pain in the EC20 scan cohort. Since most OA treatment plans focus on inhibiting the debilitating joint pain, the diminution of pain is typically used as a guide for evaluating the effectiveness of treatment. However, pain is a relativistic measurement and can be difficult to compare among different individuals with varying levels of pain tolerance. For this reason, these results suggest the attractive possibility of using a serologic marker, CD14, to assist with objectively quantifying pain mediators in patients with OA or in animal models.
Studies of an in vivo mouse model of neuropathic pain have suggested that CD14 mediates nociception via TLR-4–dependent pathways in microglia (38), the resident macrophages of the brain and spinal cord. TLR-dependent pathways also lead to the production of inflammatory mediators that activate a number of pathologic and pain signals in OA (39). Macrophages are activated and generate proinflammatory mediators in response to TLR signaling (40). TLR expression has been found to be increased in OA cartilage (41), and TLR ligation increases activation of RA synovial macrophages (42). In addition, a recent study has demonstrated increased expression of the activated macrophage marker F4/80 in the L2–5 DRGs of mice in an antigen-induced arthritis model of the knee (43). Cultured DRG cells from these mice also produced increased concentrations of monocyte chemoattractant protein 1 in comparison to cells cultured from naive mice (43). Taken together, these data suggest that the pain in OA may be due in part to the release of CD14 from activated macrophages in inflamed joint tissue and from infiltrated macrophages in the DRG. Macrophage infiltration in the DRG may also be the primary peripheral source of CD14 that would explain our observed higher concentrations of CD14 in the plasma than in the SF of patients with OA.
Activated macrophages have been classified into 2 divergent phenotypes: proinflammatory M1 macrophages and antiinflammatory M2 macrophages. M2 macrophages are further subdivided into 3 subtypes (M2a, b, and c) depending on the activation mechanism (44). The wide spectrum of macrophage phenotypes works in a concerted, yet still poorly characterized, manner to promote clearance of injured or invaded tissue, cell proliferation, and tissue remodeling (45). The transition between phenotypes is believed to be the result of macrophages adapting to their surrounding microenvironment (46). Age-dependent macrophage dysfunction (46), characterized by the inability to properly respond to environmental factors, has been implicated in the chronic inflammatory state of diseases such as atherosclerosis (47). CD163 is a known M2 marker (7), and, more recently, CD14 expression was demonstrated to be higher in M2 macrophages than in M1 macrophages (48,49). The association of shed CD163 and CD14 with the abundance of activated macrophages suggests a prominent role of M2 macrophages in OA pathogenesis. However, there is evidence that proinflammatory cytokines, which are typically produced by M1 macrophages (44), are elevated in patients with OA (2,50). Taken together, these results support the notion that synovial macrophages are dysfunctional in OA and are trapped in transition, resulting in concurrent activation of M1 and M2 macrophage phenotypes, leading to a chronic inflammatory state.
We also examined the utility of combining CD163 and CD14 biomarkers to enhance the prediction of OA phenotypic outcomes. The lack of change in QICu values with the combination of CD163 and CD14 biomarkers indicated that the biomarkers do not strengthen the models when used conjointly. Only in the POP cohort did the combination of SF CD163 and SF CD14 have a substantial benefit in the models over the use of the biomarkers individually. Future studies with larger cohorts may provide a clearer indication of the potential utility of the combinatorial approach.
A limitation of this study was the observed cohort effect, possibly due to differences in OA severity between the cohorts. Whereas the EC20 scan cohort lacked K/L grade 0 knees, the POP cohort had a lower percentage of patients with K/L grade 4 knees than the EC20 scan cohort (9.7% versus 18%). Even though the exclusion of K/L grade 4 knee samples generally resulted in lower parameter estimates, there still remained differences between the cohorts. Although we were able to show an association of the macrophage markers with macrophage activity in knees, radiographic OA severity, OA progression, and knee joint symptoms, these cohorts lacked individuals completely without knee OA. However, we were able to circumvent this limitation, in part, by including both knees (index and contralateral) in the analyses and by using GEEs to account for the association between the 2 knees of an individual.
In summary, our findings indicate the existence of a subset of patients with OA in whom joint tissue macrophages are activated. Moreover, we report the correlation of soluble markers of activated macrophages with severity of knee joint inflammation and risk of OA progression. These data suggest that macrophage-related inflammation is a driver of OA structural damage and progression. The ability to identify inflammatory knee OA phenotypes by measuring these markers on readily obtained biologic specimens, without a requirement for imaging that entails radiation exposure, offers the prospect that these markers may have clinical applicability and utility in the future for sensitively differentiating and characterizing OA disease subsets, and predicting and monitoring disease progression. These data certainly suggest that macrophage markers merit further validation as prognostic biomarkers in patients with knee OA. These results also suggest that targeting macrophages and macrophage-associated inflammatory pathways may be a way to decrease OA symptoms and joint structural deterioration.
ACKNOWLEDGMENTS
The authors thank Gary E. McDaniel, Janet L. Huebner, Thomas V. Stabler, Jonathan B. Catterall, and Sunil Suchindran for their contributions to this project.
Supported by the NIH (grants 5-T32-AI-007217-30 to Dr. Daghestani and P01-AR-050245 to Dr. Kraus). EC20 scan samples were obtained with the support of a grant from Lilly USA, Inc.
Footnotes
AUTHOR CONTRIBUTIONS
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. Kraus 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. Daghestani, Kraus.
Acquisition of data. Daghestani, Kraus.
Analysis and interpretation of data. Daghestani, Pieper, Kraus.
REFERENCES
- 1.Sellam J, Berenbaum F. The role of synovitis in pathophysiology and clinical symptoms of osteoarthritis. Nat Rev Rheumatol. 2010;6:625–635. doi: 10.1038/nrrheum.2010.159. [DOI] [PubMed] [Google Scholar]
- 2.Denoble AE, Huffman KM, Stabler TV, Kelly SJ, Hershfield MS, McDaniel GE, et al. Uric acid is a danger signal of increasing risk for osteoarthritis through inflammasome activation. Proc Natl Acad Sci U S A. 2011;108:2088–2093. doi: 10.1073/pnas.1012743108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Low PS, Kularatne SA. Folate-targeted therapeutic and imaging agents for cancer. Curr Opin Chem Biol. 2009;13:256–262. doi: 10.1016/j.cbpa.2009.03.022. [DOI] [PubMed] [Google Scholar]
- 4.Nakashima-Matsushita N, Homma T, Yu S, Matsuda T, Sunahara N, Nakamura T, et al. Selective expression of folate receptor β and its possible role in methotrexate transport in synovial macrophages from patients with rheumatoid arthritis. Arthritis Rheum. 1999;42:1609–1616. doi: 10.1002/1529-0131(199908)42:8<1609::AID-ANR7>3.0.CO;2-L. [DOI] [PubMed] [Google Scholar]
- 5.Lu Y, Low PS. Folate-mediated delivery of macromolecular anticancer therapeutic agents. Adv Drug Deliv Rev. 2002;54:675–693. doi: 10.1016/s0169-409x(02)00042-x. [DOI] [PubMed] [Google Scholar]
- 6.Kraus VB, McDaniel G, Huebner JL, Stabler T, Pieper C, Coleman RE, et al. Direct in vivo evidence of activated macrophages in human osteoarthritis. Osteoarthritis Cartilage. 2013;21(Suppl):S42. doi: 10.1016/j.joca.2016.04.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Komohara Y, Hirahara J, Horikawa T, Kawamura K, Kiyota E, Sakashita N, et al. AM-3K, an anti-macrophage antibody, recognizes CD163, a molecule associated with an anti-inflammatory macrophage phenotype. J Histochem Cytochem. 2006;54:763–771. doi: 10.1369/jhc.5A6871.2006. [DOI] [PubMed] [Google Scholar]
- 8.Beyer M, Mallmann MR, Xue J, Staratschek-Jox A, Vorholt D, Krebs W, et al. High-resolution transcriptome of human macrophages. PLoS One. 2012;7:e45466. doi: 10.1371/journal.pone.0045466. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Puig-Kroger A, Sierra-Filardi E, Dominguez-Soto A, Samaniego R, Corcuera MT, Gomez-Aguado F, et al. Folate receptor β is expressed by tumor-associated macrophages and constitutes a marker for M2 anti-inflammatory/regulatory macrophages. Cancer Res. 2009;69:9395–9403. doi: 10.1158/0008-5472.CAN-09-2050. [DOI] [PubMed] [Google Scholar]
- 10.Van Gorp H, Delputte PL, Nauwynck HJ. Scavenger receptor CD163, a Jack-of-all-trades and potential target for cell-directed therapy. Mol Immunol. 2010;47:1650–1660. doi: 10.1016/j.molimm.2010.02.008. [DOI] [PubMed] [Google Scholar]
- 11.Hogger P, Dreier J, Droste A, Buck F, Sorg C. Identification of the integral membrane protein RM3/1 on human monocytes as a glucocorticoid-inducible member of the scavenger receptor cysteine-rich family (CD163) J Immunol. 1998;161:1883–1890. [PubMed] [Google Scholar]
- 12.Buechler C, Ritter M, Orso E, Langmann T, Klucken J, Schmitz G. Regulation of scavenger receptor CD163 expression in human monocytes and macrophages by pro- and antiinflammatory stimuli. J Leukoc Biol. 2000;67:97–103. [PubMed] [Google Scholar]
- 13.Sulahian TH, Hogger P, Wahner AE, Wardwell K, Goulding NJ, Sorg C, et al. Human monocytes express CD163, which is upregulated by IL-10 and identical to p155. Cytokine. 2000;12:1312–1321. doi: 10.1006/cyto.2000.0720. [DOI] [PubMed] [Google Scholar]
- 14.Tsuneyoshi Y, Tanaka M, Nagai T, Sunahara N, Matsuda T, Sonoda T, et al. Functional folate receptor β-expressing macrophages in osteoarthritis synovium and their M1/M2 expression profiles. Scand J Rheumatol. 2012;41:132–140. doi: 10.3109/03009742.2011.605391. [DOI] [PubMed] [Google Scholar]
- 15.Landmann R, Muller B, Zimmerli W. CD14, new aspects of ligand and signal diversity. Microbes Infect. 2000;2:295–304. doi: 10.1016/s1286-4579(00)00298-7. [DOI] [PubMed] [Google Scholar]
- 16.Wright SD, Ramos RA, Tobias PS, Ulevitch RJ, Mathison JC. CD14, a receptor for complexes of lipopolysaccharide (LPS) and LPS binding protein. Science. 1990;249:1431–1433. doi: 10.1126/science.1698311. [DOI] [PubMed] [Google Scholar]
- 17.Akashi S, Ogata H, Kirikae F, Kirikae T, Kawasaki K, Nishijima M, et al. Regulatory roles for CD14 and phosphatidylinositol in the signaling via Toll-like receptor 4-MD-2. Biochem Biophys Res Commun. 2000;268:172–177. doi: 10.1006/bbrc.2000.2089. [DOI] [PubMed] [Google Scholar]
- 18.Krutzik SR, Sieling PA, Modlin RL. The role of Toll-like receptors in host defense against microbial infection. Curr Opin Immunol. 2001;13:104–108. doi: 10.1016/s0952-7915(00)00189-8. [DOI] [PubMed] [Google Scholar]
- 19.Bazil V, Strominger JL. Shedding as a mechanism of down-modulation of CD14 on stimulated human monocytes. J Immunol. 1991;147:1567–1574. [PubMed] [Google Scholar]
- 20.Kellgren JH, Lawrence JS. Radiological assessment of osteoarthrosis. Ann Rheum Dis. 1957;16:494–502. doi: 10.1136/ard.16.4.494. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kraus VB, Stabler TV, Kong SY, Varju G, McDaniel G. Measurement of synovial fluid volume using urea. Osteoarthritis Cartilage. 2007;15:1217–1220. doi: 10.1016/j.joca.2007.03.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kraus VB, McDaniel G, Worrell TW, Feng S, Vail TP, Varju G, et al. Association of bone scintigraphic abnormalities with knee malalignment and pain. Ann Rheum Dis. 2009;68:1673–1679. doi: 10.1136/ard.2008.094722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. Development of criteria for the classification and reporting of osteoarthritis: classification of osteoarthritis of the knee. Arthritis Rheum. 1986;29:1039–1049. doi: 10.1002/art.1780290816. [DOI] [PubMed] [Google Scholar]
- 24.Peterfy C, Li J, Zaim S, Duryea J, Lynch J, Miaux Y, et al. Comparison of fixed-flexion positioning with fluoroscopic semi-flexed positioning for quantifying radiographic joint-space width in the knee: test-retest reproducibility. Skeletal Radiol. 2003;32:128–132. doi: 10.1007/s00256-002-0603-z. [DOI] [PubMed] [Google Scholar]
- 25.Altman RD, Gold GE. Atlas of individual radiographic features in osteoarthritis, revised. Osteoarthritis Cartilage. 2007;15(Suppl A):A1–A56. doi: 10.1016/j.joca.2006.11.009. [DOI] [PubMed] [Google Scholar]
- 26.Davis MA, Ettinger WH, Neuhaus JM. Obesity and osteoarthritis of the knee: evidence from the National Health and Nutrition Examination Survey (NHANES I) Semin Arthritis Rheum. 1990;20:34–41. doi: 10.1016/0049-0172(90)90045-h. [DOI] [PubMed] [Google Scholar]
- 27.Jayadev C, Rout R, Price A, Hulley P, Mahoney D. Hyaluronidase treatment of synovial fluid to improve assay precision for biomarker research using multiplex immunoassay platforms. J Immunol Methods. 2012;386:22–30. doi: 10.1016/j.jim.2012.08.012. [DOI] [PubMed] [Google Scholar]
- 28.Pan W. Akaike’s information criterion in generalized estimating equations. Biometrics. 2001;57:120–125. doi: 10.1111/j.0006-341x.2001.00120.x. [DOI] [PubMed] [Google Scholar]
- 29.Burnham KP, Anderson DR. Multimodel inference: understanding AIC and BIC in model selection. Sociol Methods Res. 2004;33:261–304. [Google Scholar]
- 30.Blom AB, van Lent PL, Holthuysen AE, van der Kraan PM, Roth J, van Rooijen N, et al. Synovial lining macrophages mediate osteophyte formation during experimental osteoarthritis. Osteoarthritis Cartilage. 2004;12:627–635. doi: 10.1016/j.joca.2004.03.003. [DOI] [PubMed] [Google Scholar]
- 31.Matsushita N, Kashiwagi M, Wait R, Nagayoshi R, Nakamura M, Matsuda T, et al. Elevated levels of soluble CD163 in sera and fluids from rheumatoid arthritis patients and inhibition of the shedding of CD163 by TIMP-3. Clin Exp Immunol. 2002;130:156–161. doi: 10.1046/j.1365-2249.2002.01963.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Greisen SR, Moller HJ, Stengaard-Pedersen K, Hetland ML, Horslev-Petersen K, Jorgensen A, et al. Soluble macrophage-derived CD163 is a marker of disease activity and progression in early rheumatoid arthritis. Clin Exp Rheumatol. 2011;29:689–692. [PubMed] [Google Scholar]
- 33.Baeten D, Demetter P, Cuvelier CA, Kruithof E, Van Damme N, De Vos M, et al. Macrophages expressing the scavenger receptor CD163: a link between immune alterations of the gut and synovial inflammation in spondyloarthropathy. J Pathol. 2002;196:343–350. doi: 10.1002/path.1044. [DOI] [PubMed] [Google Scholar]
- 34.Mikuls TR, LeVan TD, Sayles H, Yu F, Caplan L, Cannon GW, et al. Soluble CD14 and CD14 polymorphisms in rheumatoid arthritis. J Rheumatol. 2011;38:2509–2516. doi: 10.3899/jrheum.110378. [DOI] [PubMed] [Google Scholar]
- 35.Nockher WA, Wigand R, Schoeppe W, Scherberich JE. Elevated levels of soluble CD14 in serum of patients with systemic lupus erythematosus. Clin Exp Immunol. 1994;96:15–19. doi: 10.1111/j.1365-2249.1994.tb06222.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Wuthrich B, Kagi MK, Joller-Jemelka H. Soluble CD14 but not interleukin-6 is a new marker for clinical activity in atopic dermatitis. Arch Dermatol Res. 1992;284:339–342. doi: 10.1007/BF00372036. [DOI] [PubMed] [Google Scholar]
- 37.Devitt A, Moffatt OD, Raykundalia C, Capra JD, Simmons DL, Gregory CD. Human CD14 mediates recognition and phagocytosis of apoptotic cells. Nature. 1998;392:505–509. doi: 10.1038/33169. [DOI] [PubMed] [Google Scholar]
- 38.Cao L, Tanga FY, Deleo JA. The contributing role of CD14 in Toll-like receptor 4 dependent neuropathic pain. Neuroscience. 2009;158:896–903. doi: 10.1016/j.neuroscience.2008.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Wu Q, Henry JL. Functional changes in muscle afferent neurones in an osteoarthritis model: implications for impaired proprioceptive performance. PLoS One. 2012;7:e36854. doi: 10.1371/journal.pone.0036854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Hu X, Chakravarty SD, Ivashkiv LB. Regulation of interferon and Toll-like receptor signaling during macrophage activation by opposing feedforward and feedback inhibition mechanisms. Immunol Rev. 2008;226:41–56. doi: 10.1111/j.1600-065X.2008.00707.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Kim HA, Cho ML, Choi HY, Yoon CS, Jhun JY, Oh HJ, et al. The catabolic pathway mediated by Toll-like receptors in human osteoarthritic chondrocytes. Arthritis Rheum. 2006;54:2152–2163. doi: 10.1002/art.21951. [DOI] [PubMed] [Google Scholar]
- 42.Huang Q, Ma Y, Adebayo A, Pope RM. Increased macrophage activation mediated through Toll-like receptors in rheumatoid arthritis. Arthritis Rheum. 2007;56:2192–2201. doi: 10.1002/art.22707. [DOI] [PubMed] [Google Scholar]
- 43.Zaki S, Miller RE, Malfait AM, Smith S, Tran PB, Ishihara S, et al. Characterization of pain-related behaviors in association with joint pathology in an 8-week antigen-induced arthritis model. Osteoarthritis Cartilage. 2014;22:S7–S56. [Google Scholar]
- 44.Laskin DL. Macrophages and inflammatory mediators in chemical toxicity: a battle of forces. Chem Res Toxicol. 2009;22:1376–1385. doi: 10.1021/tx900086v. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Novak ML, Koh TJ. Phenotypic transitions of macrophages orchestrate tissue repair. Am J Pathol. 2013;183:1352–1363. doi: 10.1016/j.ajpath.2013.06.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Stout RD, Suttles J. Immunosenescence and macrophage functional plasticity: dysregulation of macrophage function by age-associated microenvironmental changes. Immunol Rev. 2005;205:60–71. doi: 10.1111/j.0105-2896.2005.00260.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Leitinger N, Schulman IG. Phenotypic polarization of macrophages in atherosclerosis. Arterioscler Thromb Vasc Biol. 2013;33:1120–1126. doi: 10.1161/ATVBAHA.112.300173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Samaniego R, Palacios BS, Domiguez-Soto A, Vidal C, Salas A, Matsuyama T, et al. Macrophage uptake and accumulation of folates are polarization-dependent in vitro and in vivo and are regulated by activin A. J Leukoc Biol. 2014 doi: 10.1189/jlb.0613345. E-pub ahead of print. [DOI] [PubMed] [Google Scholar]
- 49.Foucher ED, Blanchard S, Preisser L, Garo E, Ifrah N, Guardiola P, et al. IL-34 induces the differentiation of human monocytes into immunosuppressive macrophages: antagonistic effects of GM-CSF and IFNγ. PLoS One. 2013;8:e56045. doi: 10.1371/journal.pone.0056045. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Hulejova H, Baresova V, Klezl Z, Polanska M, Adam M, Senolt L. Increased level of cytokines and matrix metalloproteinases in osteoarthritic subchondral bone. Cytokine. 2007;38:151–156. doi: 10.1016/j.cyto.2007.06.001. [DOI] [PubMed] [Google Scholar]