Synovial fluid biochemical markers (or, biomarkers) are molecules that enter the joint fluid, presumably as a result of cartilage and synovium metabolism. These proteins are thought to reflect the metabolic state of the articular cartilage and may serve as surrogate markers for the severity of arthritis in humans and animals (1, 2). Monitoring of a single index joint is most directly done by analysis of joint fluid. However, direct aspiration of joint fluid can be challenging, particularly in nonpathologic control joints, where joint fluid is usually scarce and highly viscous. The majority of studies in animals have utilized the technique of joint lavage to reliably obtain a sample of joint fluid from joints that do not have effusions (for review, see ref. 2). Joint lavage with small volumes of saline has also been performed in at least 2 studies in humans (3, 4).
The technique of joint lavage involves either injecting 3–5 ml of saline into a joint and then aspirating the fluid after moving the joint several times through its range of motion or aspirating after repeated compression of the infra- and suprapatellar regions. The latter technique, which was used in 1 of the human studies (3), has the advantage of requiring only a single intraarticular injection. Joint lavage, as currently performed for biomarker analysis, does not impose a confounding therapeutic effect in longitudinal studies of patients (3) but does contribute significant errors to synovial fluid biomarker analyses due to the inability to predict the sample dilution introduced by this technique.
In the present study, we developed a method of quantifying joint fluid biomarker concentrations in a manner that corrected for dilutional effects of the lavage procedure. We initially explored the use of biocompatible tracer dyes, such as those used for the measurement of plasma volume, Evans blue (5) and indocyanine green (6, 7). We discovered several disadvantages related to the use of dyes, including the inability of the dyes to completely clear from the joint, the heterogeneity of dye sampling due to incomplete mixing of dye and synovial fluid within the joint, and the absorption and metabolism of the dyes by intraarticular cells.
We hypothesized that native molecules that maintain a fixed relationship between synovial fluid and serum could be used to correct for the dilutional effect introduced by joint lavage and would not possess the disadvantages inherent in the dye techniques. Urea is a small molecule that is neither synthesized nor metabolized by joint tissues. The transport of urea across synovium in humans is consistent with passive diffusion (8). In the present study, we evaluated the relationship between serum and synovial fluid concentrations of urea in various joint sites, in normal joints, and in abnormal joints with acute inflammation. Furthermore, we utilized this technique to compare concentrations of various cartilage biomarkers (glucose, lactate, cartilage oligomeric matrix protein [COMP], and keratan sulfate [KS]) in synovial fluid obtained by direct aspiration of joint fluid as well as by joint lavage. Glucose and lactate have been shown to be markers of synovial inflammation associated with both rheumatoid arthritis and osteoarthritis (9). COMP is a cartilage matrix component whose synovial fluid levels are altered in osteoarthritis in humans and in animal models (for review, see refs. 10 and 11). KS is the most abundant glycosaminoglycan constituent of mature cartilage matrix (12). Levels of these matrix components are generally considered to reflect cartilage catabolism. These measurements were made in dogs and are potentially applicable to other animal and human studies in which validation of a surrogate marker for arthritis is undertaken.
MATERIALS AND METHODS
- Top of page
- MATERIALS AND METHODS
Study design. Serum and joint fluid from 32 adult mongrel dogs (55 joints) were acquired immediately before the animals were euthanized for other, unrelated studies. Procedures were approved by the Institutional Animal Care and Use Committee. Joint fluid samples were obtained by direct aspiration (neat samples) or by joint lavage.
Urea was measured in the sera and neat joint fluids obtained from normal joints (28 dogs; 41 joints) to evaluate the relationship between circulating and joint fluid urea concentrations in the absence of joint pathology. A normal joint was defined as one that had not undergone previous joint procedures of any kind and that had no clinical evidence of effusion or functional impairment. Joint radiography was not routinely performed as part of this study. Knees (21 joints) and elbows (20 joints) were sampled neat to evaluate urea concentrations from various joint sites within the same animal.
The contralateral normal knees of 10 animals (10 joints) were lavaged in order to develop and test the accuracy of using the concentration of urea to correct joint fluid biomarker measures for dilutional effects introduced by lavage. Joint fluid urea concentrations were also evaluated in the context of acute joint injury by intraarticular injection of chymopapain (4 dogs; 4 joints), which induced acute inflammation characterized by joint effusion (13). The concentrations of the biomarkers, glucose, lactate, COMP, and KS (5-D-4 epitope) were measured and corrected for dilution due to lavage, based on urea concentrations in the lavage fluid and serum.
Samples. Food was withheld from all dogs for 12 hours (overnight) prior to the collection of samples. Samples were collected from animals that had been anesthetized to affect with 5% sodium pentothal. Blood was collected from the left antecubital vein of 28 dogs into SST brand Gel and Clot Activator tubes (VWR, Morrisville, NC). Sera were removed and frozen at −80°C until analyzed. Joint fluid was collected by needle aspiration within 10 minutes after venipuncture and stored in aliquots at −80°C until analyzed. Lavage was carried out in 10 dogs by intraarticular injection of 3 ml of normal saline, withdrawal of the needle, and manipulation of the joint through its full range of motion 12 times. Following this procedure, all obtainable fluid was aspirated from the joint through a lateral patellar approach (2).
Four additional dogs underwent intraarticular injection of 0.5 ml of 4 mg/ml chymopapain in buffer (100 mM sodium phosphate, 50 mM EDTA, and 10 mM cysteine hydrochloride, pH 6.5) under anesthesia (medetomidine and atipamezole). Intraarticular chymopapain injections have been shown to result in acute alterations of joint tissue metabolism within 24 hours of injection. This model is characterized by a severe loss of proteoglycan from cartilage and an acute rise in proteoglycan concentrations in synovial fluid and serum (for review, see ref. 2). Joint effusions were induced in all chymopapain-injected knees, as determined by clinical examination. Fasting blood and joint fluid were obtained 24 hours after chymopapain injection. Joint fluid was aspirated directly from the inflamed knees, centrifuged for 10 minutes at 3,000g, and the supernatants were stored at −80°C until analyzed.
Biomarker assays. Concentrations of urea, glucose, and lactate were determined by a CMA model 600 microdialysis analyzer (CMA Microdialysis, Solna, Sweden) in 5-μl samples of serum and joint fluid. Concentrations of urea, glucose, and lactate were obtained by completely automated enzymatic reaction techniques. The assays consisted of standard enzymatic phosphorylation and/or oxidation reaction steps and spectrophotometric detection. Briefly, quantification of urea depends on the rate of utilization of NAD and was measured at 365 nm. For glucose and lactate, the rate of formation of a colored substance (quinoneimine) was measured photometrically at 546 nm. A measurement cycle was completed in less than 2 minutes. All reagents were obtained from CMA Microdialysis. Although the CMA 600 provides a means of analyzing small volumes of fluid with sensitivity and high precision, when access to a microanalyzer is not feasible, urea concentrations are also readily measured with a commercially available kit (catalog no. 63-25; Sigma, St. Louis, MO), which involves the same chemistry as the CMA. Thus, this method should be accessible to most laboratories and useful for a broad range of arthritis biomarker studies.
Concentrations of COMP and KS, two measures of cartilage catabolism, were determined by inhibition enzyme-linked immunosorbent assay. COMP is one of the most abundant glycoproteins in cartilage, but it is also found in synovium, meniscus, and cranial cruciate ligament (11, 14). COMP was quantified in this study with monoclonal antibody (mAb) 17-C10, against human COMP (10), which has been shown to cross-react with canine COMP (2). The 17-C10 epitope has been mapped to an epidermal growth factor–like domain of the COMP subunit (15).
The assay was performed as described previously (16), with the following modifications. Samples and standards were incubated with 17-C10 antibody at a dilution of 1:40,000 overnight at 4°C on a low-binding 96-well plate (microwell plates; Nunc, Naperville, IL). Neat and lavage joint fluid samples were diluted serially in phosphate buffered saline (PBS)–Tween 20, pH 7.3. Samples were assayed in at least 5 dilution steps, ranging from 1:200 to 1:3,200. The intra- and interassay variations were <4% and <5%, respectively. COMP is expressed in μg/ml of a purified human COMP standard kindly provided by Dr. Vladimir Vilim (Institute of Rheumatology, Prague, Czech Republic).
The mAb 5-D-4 reacts against several repeats of a highly sulfated heptasaccharide in KS, a glycosaminoglycan linked to cartilage proteoglycan. The assay used to quantify the concentration of KS in the joint fluid samples was performed as described previously (2). The mAb 5-D-4 was diluted in PBS–Tween 20 at pH 5.3. Samples and standards were incubated with the 5-D-4 antibody at a dilution of 1:250,000 overnight at 4°C on a low-binding transfer plate (microwell plates; Nunc). Neat and lavage samples were assayed in up to 9 dilution steps, ranging from 1:1,500 to 1:38,443. Joint fluid samples from inflamed joints were assayed in up to 8 dilution steps, ranging from 1:50,000 to 1:6,400,000. The concentration of KS was expressed in μg/ml as a KS-2 international standard equivalent (kindly provided by Dr. Eugene Thonar, Rush–Presbyterian–St. Luke's Medical Center, Chicago, IL).
Osmolality of joint fluids. The osmolality of synovial fluids was measured with a vapor pressure osmometer (Wescor, Logan, UT). Briefly, 10-μl samples of joint fluid from 4 control knees, 4 control elbows, and 4 knees injected with chymopapain were analyzed. The instrument was calibrated using 2 calibration standards at 290 mOsmoles/kg and 1,000 mOsmoles/kg, and each sample was analyzed in triplicate. Vapor pressure osmometry, unlike the freezing-point method, permits precise and accurate measurements of osmotic pressure of small amounts of biologic material at physiologic temperature.
Statistical analysis. Serum and joint fluid urea concentrations were analyzed by linear regression. The biomarker data from paired samples (neat and lavage) were analyzed by paired 2-tailed t-test. Results of joint fluid biomarker analyses from normal and inflamed joints were analyzed by analysis of variance. Using the Bonferroni correction, a P value less than 0.0167 was considered significant. All statistical analyses were performed on a Macintosh PowerPC with StatView software (Abacus Concepts, Berkeley, CA).
- Top of page
- MATERIALS AND METHODS
Levels of urea in normal joints. Joint fluid was aspirated directly from 41 normal joints, yielding a volume of 100–300 μl. Linear regression analysis of the concentrations of urea in joint fluid and sera revealed a strong linear relationship: R2 = 0.96, where joint fluid urea = (1.05[serum urea] + 0.52) (Figure 1). There was no significant difference between the values obtained for the knees compared with the values obtained for the elbows (Figure 1). The wide range of serum urea concentrations was considered to be a result of the varying hydration status of the animals at the time of sampling and was below levels that are suggestive of renal impairment (<50 mM).
Figure 1. Comparison of urea concentrations in joint fluid and serum obtained from normal joints of 41 dogs. Joint fluid was obtained by direct aspiration (neat) of 21 knees (⧫) and 20 elbows (▪). Urea concentrations were measured by a CMA model 600 microdialysis analyzer (CMA Microdialysis, Solna, Sweden).
Download figure to PowerPoint
Joint fluid was obtained by lavage from the contralateral knee or elbow joints of 10 animals from which neat joint fluid was obtained. All attempts at joint lavage were successful, with an average yield of 1.8 ml. Urea concentrations in the lavage fluid were measured and were used to correct for the dilution of the synovial fluid due to lavage by determining the dilution factor (DF): DF = (1.05[serum urea] + 0.52)/[joint fluid lavage urea].
Levels of biomarkers in normal joints. In order to obtain stable glucose data, food was withheld from all animals for 12 hours (overnight) prior to sample collection. The expected fasting glucose level in nondiabetic dogs is ≤120 mg/dl (17). The fasting serum glucose in our sample population was within this range for all but 1 animal (range 54.59–131.52 mg/dl), with a mean ± SD value of 95.93 ± 17.45 mg/dl, which verified a nondiabetic and fasting state in the majority of animals.
Joint fluid glucose and lactate concentrations were measured in normal joint fluids drawn neat and by lavage. As expected, significantly lower levels of glucose and lactate were found in lavage fluid compared with neat fluid (P < 0.0001 for each comparison) (Figures 2A and B). Levels of these two molecules in lavage joint fluids were expected to be similar to those in neat fluids when the concentration in lavage fluid was corrected using the calculated dilution factor based on urea. Consistent with this hypothesis, after correction of values with the dilution factor derived from urea concentrations, glucose and lactate levels in lavage fluid (lavage corrected) showed no significant difference from those in neat samples (P = 0.21 and P = 0.90, respectively).
Figure 2. Comparison of A, glucose, B, lactate, C, cartilage oligomeric matrix protein (COMP), and D, keratan sulfate (KS) concentrations in normal canine knee and elbow joint fluids. Joint fluid was obtained by direct aspiration (neat) or by saline lavage. A and B, Concentrations of glucose and lactate were measured in 10 pairs of joints (6 pairs of knees and 4 pairs of elbows) with a CMA model 600 microdialysis analyzer (CMA Microdialysis, Solna, Sweden). Glucose and lactate concentrations in lavage samples were corrected for dilutional effects of lavage (lavage corrected) based on a dilution factor (DF) derived from the urea concentrations: DF = (1.0513[serum urea] + 0.5249)/[joint fluid lavage urea]. C and D, Concentrations of COMP and KS, two biomarkers of cartilage matrix catabolism, were measured in 5 joint pairs and 7 joint pairs, respectively. COMP and KS concentrations in lavage samples were corrected for dilutional effects of lavage based on a dilution factor derived from the urea concentrations, as described for A and B. COMP was measured by monoclonal antibody (mAb) 17-C10 and was expressed in μg/ml of a purified human COMP standard (from Dr. Vladimir Vilim). KS was measured by mAb 5-D-4 and was expressed in μg/ml of a KS-2 international standard (from Dr. Eugene Thonar).
Download figure to PowerPoint
Sufficient neat joint fluid was available to measure COMP in 5 joint pairs and to measure KS in 7 joint pairs. As expected, higher levels of COMP and KS were detected in neat fluid than in lavage fluid (P = 0.008 and P = 0.003, respectively) (Figures 2C and D). After correction for dilution, levels of these biomarkers were not significantly different from those in neat samples (P = 0.87 and P = 0.39, respectively).
Levels of urea in inflamed joints. Intraarticular injection of chymopapain produced knee effusions in all 4 animals within 24 hours. The urea concentration in these acutely injured joints was significantly lower (mean 12%) than that in serum. Based on the assumption that urea is neither synthesized nor metabolized by joint tissues, we hypothesized that urea concentrations would provide a means by which to calculate the change in a biomarker concentration attributable to alterations of intraarticular fluid water volume in the presence of acute inflammation characterized by effusion.
Levels of biomarkers in inflamed joints. Relatively large volumes of joint fluid were readily obtained from the 4 chymopapain-treated joints 24 hours after intraarticular injection (range 2–6 ml). Thus, under these circumstances, joint lavage was not required, and all samples were acquired neat. Compared with normal joint fluids, those from chemically injured joints had lower glucose concentrations but higher lactate concentrations (P = 0.005 and P < 0.0001, respectively) (Figures 3A and B). Glucose and lactate biomarker concentrations were adjusted for the apparent dilutional effect of acute inflammation based on the joint fluid urea concentration, as follows: DFapparent = (1.05[serum urea] + 0.52)/[inflamed joint fluid urea]; and ([biomarker]inflamed adjusted = [biomarker]Inflamed × DFapparent).
Figure 3. Comparison of A, glucose, B, lactate, and C, keratan sulfate (KS) concentrations in noninflamed and inflamed canine joints. Joint fluid was aspirated directly from noninflamed or inflamed joints. Joint fluid for glucose and lactate measurements was available from 41 noninflamed joints and for KS measurements from 7 joints. Acute inflammation was induced in 4 joints by intraarticular injection of chymopapain. Biomarker levels in inflamed joint fluids were adjusted (inflamed adjusted) for the volume of distribution calculated from the apparent dilution factor as described in Materials and Methods.
Download figure to PowerPoint
After adjustment, joint fluid glucose concentrations in acutely inflamed joints were not significantly different from those in noninflamed joints (P = 0.65). In contrast, after adjustment, joint fluid lactate concentrations in acutely inflamed joints remained significantly higher than those in normal joints (P < 0.0001).
COMP levels in fluid from chymopapain-treated joints were 4 times lower than those from normal joints (data not shown). Proteolysis of COMP, with loss of 17-C10 immunoreactivity as a result of chymopapain exposure, was confirmed by Western blotting with mAb 17-C10 (data not shown). The 5-D-4 epitope reflects the presence of KS, a glycosaminoglycan that is resistant to chymopapain digestion. Compared with normal joint fluids, those from acutely inflamed joints had markedly higher KS concentrations (P < 0.0001) (Figure 3C). The KS levels in inflamed joints were a mean of 26 times higher than those in normal joints. After adjusting for the dilutional effect introduced by the increase in the joint fluid water fraction in the presence of acute inflammation, the KS level was even further elevated.
Osmolality of joint fluids. The mean ± SD osmolality of normal joint fluid was 308.68 ± 16.97 mOsmoles/kg. There was no appreciable difference between the osmolality of joint fluid obtained from normal knees and elbows (P = 0.63). The mean ± SD osmolality of acutely inflamed joint fluid was 300.79 ± 30.44 mOsmoles/kg. There was no significant difference between the osmolality of normal and inflamed joint fluids (P = 0.57).
- Top of page
- MATERIALS AND METHODS
These studies demonstrate that the concentration of urea in joint fluids can provide a valuable means of accurately quantifying levels of biomarkers in joint fluids that have been unpredictably diluted due to acute inflammation or joint lavage. Intraarticular chymopapain injection is a chemical model of acute joint injury, and therefore, the findings are most analogous to acute joint injury in humans. Studies of joint fluid urea in the setting of chronic arthritis in humans may show different results from this acute injury model. More studies are therefore necessary to evaluate and optimize the use of urea measurements in the setting of chronic inflammation.
In a previous study, the ratio of the concentration of urea in joint fluid to that in serum was reported to be 0.95 ± 0.12 (mean ± SD), based on an analysis of 15 bovine joint fluids and sera (18). The original characterization of joint fluid urea and other small molecules led to the conclusion that joint fluid is a simple filtrate, or “dialysate,” of blood plasma (18). In the present study, we confirmed the existence of a fixed relationship between joint fluid and serum concentrations of urea in normal joints. Arginine hydrolysis is the source of urea in the body. Although arginine is ubiquitous in animal cells, urea formation is reportedly “almost entirely limited to the liver, the only tissue that contains arginase” (19). The finding of a fixed relationship between serum and joint fluid urea levels supports the assumption that joint tissues neither synthesize nor metabolize urea, and we were aware of no published reports specifically evaluating chondrocytes for urea production. In addition, we have detected no urea production by chondrocytes cultured in serum-free medium as cartilage explants for 24 hours (Kraus VB, et al: unpublished observations).
Urea crosses the synovium in both directions by a process of unrestricted diffusion, the rate of which is determined by the diffusion path between synovial lining cells (8). Factors governing the concentration of intraarticular macromolecules include the input rate of the macromolecule, the flow of water across the synovial lining, the properties of the synovial intercellular matrix that affect the clearance rate of the macromolecule, and the osmotic gradient between the synovial fluid and the serum (20).
The average osmotic pressure of serum is greater than that of normal synovial fluid, with a difference of 250 mm of water (18). The osmotic pressure effect exerted by a gram of proteoglycan (formerly referred to as mucin) was estimated by Ropes et al to be 9 times greater than the effect of albumin. The addition of charged glycosaminoglycan molecules to the joint fluid as a result of cartilage degradation increases pericapillary osmotic pressure in the synovial intercellular matrix and promotes capillary filtration and the flow of water into the joint (20). Counteracting this effect is the effect of increased intraarticular pressure in the inflamed joint, which promotes trans-synovial absorption of fluid from the joint (20).
We found that in response to chymopapain, there was a marked elevation of joint fluid levels of KS, a negatively charged glycosaminoglycan that is capable of exerting an osmotic effect. Lower urea concentrations were found in fluid from acutely inflamed joints compared with normal joint fluid. This finding may be interpreted to be the result of a net increase in the water fraction of the joint fluid with acute inflammation. Thus, in this model of acute joint inflammation, a net flow of water into the joint predominates. Under these conditions, joint fluid urea provides a means of estimating the change in a biomarker caused by a change in the volume of distribution (Vd) due to an altered water fraction of joint fluid in the presence of inflammation. It has been demonstrated that even low-grade synovitis can significantly alter the Vd (21). Despite lower glucose and urea concentrations, and therefore an increased water fraction in fluid from acutely inflamed joints, the measured joint fluid osmolality in fluid from acutely inflamed joints was equivalent to that in fluid from normal joints. These results suggest that the marked elevation of joint fluid glycosaminoglycan fragments in response to chymopapain exerts a significant osmotic effect that contributes to the change in the joint fluid Vd of small molecules.
Interestingly, we observed lower levels of joint fluid glucose in acutely inflamed joints than in normal joints. The increased Vd of the joint fluid compartment with inflammation was sufficient to explain the magnitude of the change in concentration. In contrast, we identified 2 biomarkers with higher levels in acutely inflamed joint fluids: lactate and KS. After adjustment for the increase in the water fraction of fluid from the acutely inflamed joint, lactate and KS epitope joint fluid concentrations were even further elevated.
It is not possible from these data to determine whether elevations of lactate and KS with acute inflammation are a result of increased intraarticular production, decreased clearance, or both. At high transcapillary water velocities characteristic of the inflamed joint, water molecules are sieved more efficiently than large macromolecules (20). In previous studies of dogs, however, there was an increase in proteoglycan clearance from joints that were inflamed as a result of anterior cruciate ligament resection (22). Therefore, the concentrations of lactate and KS epitope measured in the present study may potentially underestimate the true production of these biomarkers during the first 24 hours after chymopapain injection. These results suggest a shift toward a more anaerobic glycolysis in response to chymopapain and demonstrate that lactate, but not glucose, is a marker of acute joint inflammation in this experimental model system.
An important advantage of this analysis is that it significantly reduces the variability of biomarker determination when samples are obtained by lavage. When samples are obtained from acutely inflamed joints with effusions, normalization of biomarker levels to urea, a passive diffusion marker, could provide a means of identifying the relative contribution of cartilage and joint tissue metabolism to the level of the marker versus the change due to the flux of water, and may provide insights into the metabolic alterations of joint tissues.
Based on results of our study, we recommend the following procedure for studies involving joint biomarker evaluation. First, when joint fluid is obtained, serum should also be obtained to be used for correction or adjustment of joint fluid biomarker concentrations based upon urea. For biomarkers that are affected by the postprandial state, such as glucose, collect a fasting serum sample. Second, proceed with joint lavage in joints without effusions, and correct the biomarker levels based on the dilution factor calculated from joint fluid and serum urea concentrations as described in Materials and Methods. The single-injection method of lavage is feasible and is preferred for human studies, while either method can be performed ethically in anesthetized animals. Third, obtain joint fluid from effused joints by aspiration without lavage. We would recommend reporting unadjusted and adjusted (based on the apparent dilution factor as described in Materials and Methods) biomarker values to depict the effect of a change in the Vd due to an altered water fraction of joint fluid in the presence of acute inflammation. For longitudinal studies necessitating serial measurements, subsequent lavage of the effused joint (discarding lavage fluid) is recommended to control for the potential effects of lavage on cartilage metabolism in cases where contralateral joints have undergone lavage.