Synovial chemokine expression and relationship with knee symptoms in patients with meniscal tears

Abstract

Objective

In patients with knee OA, synovitis is associated with knee pain and symptoms. We previously identified synovial mRNA expression of a set of chemokines (CCL19, IL-8, CCL5, XCL-1, CCR7) associated with synovitis in patients with meniscal tears but without radiographic OA. CCL19 and CCR7 were also associated with knee symptoms. This study sought to validate expression of these chemokines and association with knee symptoms in more typical patients presenting for meniscal arthroscopy, many who have pre-existing OA.

Design

Synovial biopsies and fluid (SF) were collected from patients undergoing meniscal arthroscopy. Synovial mRNA expression was measured using quantitative RT-PCR. The Knee Injury and Osteoarthritis Outcome Score (KOOS) was administered preoperatively. Regression analyses determined if associations between chemokine mRNA levels and KOOS scores were independent of other factors including radiographic OA. CCL19 in SF was measured by ELISA, and compared to patients with advanced knee OA and asymptomatic organ donors.

Results

90% of patients had intra-operative evidence of early cartilage degeneration. CCL19, IL-8, CCL5, XCL1, CCR7 transcripts were detected in all patients. Synovial CCL19 mRNA levels independently correlated with KOOS Activities of Daily Living scores (95% CI [-8.071, -0.331], p= 0.036), indicating higher expression was associated with more knee-related dysfunction. SF CCL19 was detected in 7 of 10 patients, compared to 4 of 10 asymptomatic donors.

Conclusion

In typical patients presenting for meniscal arthroscopy, synovial CCL19 mRNA expression was associated with knee-related difficulty with activities of daily living, independent of other factors including presence of radiographic knee OA.

Introduction

Osteoarthritis (OA) is a common joint disease characterized by degeneration of cartilage and other joint structures and remodeling of subchondral bone. The knee is the most common joint affected by OA. Joint injuries such as meniscal tears are a strong risk factor for both incidence and progression of knee OA (1, 2). The Multicenter OA Study (MOST), a longitudinal study of the natural history of OA progression, demonstrated that meniscal damage is associated with a 6-fold increased risk of developing radiographic changes of OA (3). Although meniscal damage in patients with knee OA is clearly associated with disease progression, it is not clear whether meniscal damage is always associated with knee pain or joint dysfunction (3, 6, 7). In addition, population based studies have shown that there is a discordance between radiographic findings of structural joint changes and knee symptoms in patients with OA (4, 5).

In patients with OA, inflammation of the synovial membrane is a variable characteristic of disease but is potentially involved in the arthritic process. A number of studies have shown that synovial inflammation (synovitis) is associated with both risk of progression of cartilage loss (8, 9) and knee pain (10, 11). There is a need for a greater understanding of the role of synovitis in patients with meniscal tears and OA, and for the development of better biomarkers of synovitis to identify patients at risk for development and progression of symptomatic OA. We recently reported an association between synovitis and preoperative knee symptoms (12) in patients undergoing arthroscopic partial meniscectomy with a history of traumatic knee injury but no radiographic evidence of OA. Synovitis was already present in 40% of these patients and was associated with worse preoperative pain and function measured using the Lysholm score (13). Microarray analysis of synovial gene expression patterns identified a set of chemokines (IL-8, CCL5, XCL1, and CCL19 and its receptor CCR7) associated with the presence of synovitis. Expression levels of CCL19 and its receptor CCR7 were also associated with worse Lysholm scores, suggesting these inflammatory mediators may be useful in identifying patients with a detectable synovial inflammatory response after meniscal injury. Our findings using both histologic analysis and mRNA analysis of synovial inflammatory markers established a relationship between synovitis and worse symptoms in these meniscal injury patients who are at risk for development of OA.

Many patients referred for arthroscopic meniscal surgeries already have evidence of early radiographic OA, and don't always have a clear history of joint injury (7). Therefore it was not clear whether findings from our previous study could be generalized to typical patients with meniscal damage undergoing arthroscopic procedures, or whether the relationship between inflammatory chemokines and symptoms would be independent of pre-existing OA. Our current study was designed to validate the relationship between synovial chemokine expression and knee symptoms in more representative patients, in order to determine whether the inflammatory chemokines identified previously might have utility as markers of inflammation in a more typical clinical population.

Materials and Methods

Patients and tissue collection

Biospecimens (synovial tissues and fluids) for these investigations were obtained from three different tissue repositories maintained at Rush University Medical Center: Specimens from patients with meniscal tears were obtained from the Knee Injury and Arthritis repository; specimens from advanced knee OA patients were obtained from the Orthopedic Tissue and Implant Repository; and specimens from asymptomatic controls were obtained from the Repository of Joint Tissues from the Gift of Hope Tissue and Organ Donor Network.

Meniscal tear patients

Patients 18 years and older scheduled for arthroscopic knee procedures were recruited to participate in the Knee Injury and Arthritis repository. Patients undergoing ligament procedures were excluded. The Institutional Review Board approved the study and all patients gave written informed consent. For the current study we chose patients with meniscal tears undergoing arthroscopic partial meniscectomy. Synovial biopsies were collected from the suprapatellar pouch during surgery and stored at -80°C in RNAlater (Life Technologies, CA) for gene expression analysis. Synovial fluids from these patients were obtained by needle arthrocentesis, prior to establishment of the arthroscopic portals. Fluids were centrifuged at 1000 g for 10 minutes to remove cells and debris, then stored in aliquots at -80°C prior to batch analysis.

Advanced OA patients and asymptomatic controls

For comparative evaluation of chemokine protein levels, biospecimens were also obtained from asymptomatic donors and from patients with advanced OA. Synovial tissue and fluid specimens were obtained from patients with advanced OA undergoing total knee replacement, who were participants in the Orthopedic Tissue and Implant Repository at Rush University Medical Center. All specimens from this repository came from patients with a diagnosis of OA from the operating surgeon, and had clear radiographic evidence of knee OA on pre-operative knee x-rays (Kellgren-Lawrence score >2). Synovial tissue and fluid was collected and processed as described above. Synovial fluids from asymptomatic donors (based on no history of arthritis or chronic joint pain) were obtained through the Repository of Joint Tissues from the Gift of Hope Tissue and Organ Donor Network, maintained at Rush University Medical Center. These de-identified biospecimens were obtained within 24 hours of death, and processed as described for the meniscal tear and advanced OA patients.

Collection of clinical data

Clinical and demographic data for arthroscopy and arthroplasty patients were obtained by chart review. Cartilage integrity was assessed intra-operatively using the Outerbridge scale (14), and radiographic changes were scored on pre-operative x-rays using the Kellgren-Lawrence (K-L) classification(15). For arthroscopy patients, the pattern of meniscal injury (degenerative versus traumatic) was also evaluated intra-operatively by the surgeon where 0= no tear, 1= radial or longitudinal including bucket handle tear, 2= flap or oblique tear, 3= horizontal tear (likely degenerative), 4= complex tear with multiple cleavage planes (degenerative) and 5= complete maceration or destruction. In asymptomatic organ donors, no x-rays were available but a pathologist assessed gross degenerative joint changes at the time of dissection using the modified Collins grading system (16).

Knee Symptom Assessment

In the meniscal tear patients only, the Knee Injury and Osteoarthritis Outcome Score (KOOS), English version LK1.0 (17) was used to measure preoperative knee symptoms. Questionnaires, scoring manual and user's guide were obtained from http://www.koos.nu. The KOOS measures symptoms and disability on five separately scored sub scales: Pain, other Symptoms, Activities of Daily Living (ADL), Sports and Recreation and Quality of Life (QOL). Patients completed the KOOS during routine pre-surgical clinical evaluations, and questionnaires were obtained by chart review after enrollment into the repository.

Gene expression analysis

Prior to evaluating inflammatory chemokine gene expression, the presence of synovial inflammation in the meniscectomy patients was confirmed in hematoxylin and eosin (H&E) stained thin sections from a subset of 19 repository patients. Using a previously reported grading system (12, 18), synovitis (grade 1 or above) was confirmed in 55% (n= 11), similar to our previous report in a separate group of meniscal arthroscopy patients (12). Total RNA was then extracted as described (12) from synovial biopsies available from a separate set of 19 meniscal tear patients, as well as 16 advanced OA patients. RNA integrity was confirmed on a microfluidics-based system (Agilent Bioanalyzer, Agilent Technologies, Santa Clara, CA). All RNA was DNase treated, and cDNA synthesized using the iScript cDNA Synthesis kit (Bio-rad, Hercules, CA). Expression levels of four chemokines and one chemokine receptor (CCL19, IL-8, XCL1, CCL5 and CCR7) identified in a previous study (12) was measured by real-time quantitative PCR using specific primers and iQ SYBR Green Supermix (Bio-rad, Hercules, CA). After normalizing Ct values to GAPDH, expression levels in meniscal tear patients were calculated relative to the mean of the advanced OA group (19,20).

Synovial Fluid (SF) CCL19 measurement

SF CCL19 was measured by ELISA (R&D Systems, Minneapolis, MN) in a set of specimens from 10 meniscal arthroscopy patients, 10 advanced OA patients, and 10 asymptomatic donors. The lower limit of detection for CCL19 was 15.6 pg/mL, and the ELISA was performed according to the manufacturer's instructions using freshly thawed aliquots of SF in duplicate.

Statistical analysis

SPSS version 19.0 and GraphPad Prism 5.0 were used for statistical analyses. Between group differences were evaluated using Kruskal-Wallis and results confirmed by Dunns post-hoc tests to compare individual groups. For categorical variables we used Fisher exact or Chi-square tests, and unpaired t-tests or Mann-Whitney U tests for continuous variables. KOOS Pain, ADL and QOL surveys were incomplete for one patient each, and questionnaires were missing for two patients. To calculate KOOS subscores when there were incomplete responses to the questionnaires, missing values were calculated according to the KOOS scoring guide (available from http://www.koos.nu/). Hot deck imputation methods (21) were then used to calculate scores for the two missing questionnaires. To determine whether chemokine levels correlated with KOOS subscores in the meniscal tear patients, Pearson's correlations were examined. Significant correlations were then further tested using hierarchical linear regression (22)-in order to determine whether relationships between KOOS subscores (dependent variables) and mRNA levels of chemokines (independent variables) remained significant after adjusting for other variables. In this hierarchical approach, demographic variables (age, gender, BMI), Outerbridge scores and Kellgren-Lawrence grades were entered into the regression first, followed by chemokine level. It was assumed that the residuals of the regression were normally distributed, and this was confirmed by residual analysis of the final regression model. All tests were conducted using a 2-tailed alpha of 0.05.

Results

Chemokine mRNA levels in Meniscal and Advanced OA patients

19 meniscal tear patients and 16 advanced OA patients had synovial biopsies available for gene expression analysis. Synovial tissues from asymptomatic donors were obtained up to 24 hours post-mortem, and were not included in this analysis as RNA was not consistently of sufficient quality to provide reliable results. The characteristics of the patients included in this analysis are presented in Table 1. Meniscal patients were significantly younger than advanced OA patients (p=0.0002). Median BMI was in the overweight range for meniscal patients and obese range in advanced OA patients (p =0.0004). Gender distributions were statistically different between the two groups (Fisher's exact p = 0.04). Median radiographic OA severity (K-L grades) differed significantly between meniscal tear and advanced OA patients as expected (p<0.0001). All but 2 of the advanced OA patients had Outerbridge grade 4 cartilage damage (exposed subchondral bone). 2 of 19 (10%) of meniscal patients had normal (K-L= 0) x-rays, and only one patient (5%) had normal appearing cartilage surfaces assessed by the Outerbridge score; the majority exhibited grade 2 (n=5) and grade 3 (n=7) lesions. Meniscal scores indicated that 18 of 19 patients exhibited degenerative type tears and 1 patient had a radial tear. The predominance of degenerative-type tears and cartilage integrity loss in the majority of these patients is indicative of early to intermediate-stage knee OA. Meniscal scores were not obtained for advanced OA patients.

Table 1.

Characteristics of patients with meniscal tears, and patients with advanced OA included in the gene expression analysis.

	Meniscal Tear Patients N= 19	Advanced OA Patients N =16	p-value
Age, median (IQR), years	49 (42.2-55.5)	64 (58-72)	0.0002a
BMI, median (IQR), kg/m2	27.9 (23.7-31.8)	37.5 (23.3-42.8)	0.0004a
Male/Female N, (%)	14/5 (74/26)	6/10 (37.5/62.5)	0.04b
Kellgren-Lawrence Score, median (IQR)	2 (1-2)	4 (4-4)	<0.0001c
Outerbridge Score (N, % Of Patients)
Grade 0	1 (5)	0 (0)
Grade 1	3 (15)	0 (0)	0.0019c
Grade 2	5 (25)	0 (0)
Grade 3	7 (35)	2 (12.5)
Grade 4	4 (20)	14 (87.5)

^aUnpaired t-test

^bFisher's Exact test

^cFisher's Exact test, K-L 0-2 vs. 3-4, or Outerbridge 0-2 vs. 3-4

IQR = Interquartile Range

CCL19, IL-8, CCL5, XCL1 and CCR7 were detected in all 19 meniscal tear patients and > 90% of advanced OA patients (Figure 1). Relative expression (RE) of IL-8 and CCR7 was significantly higher in advanced OA patients ([Mean RE, +/- SD] IL-8 = 1.416, +/- 0.913, <0.001; CCR7 = 1.659, +/- 2.209, p=0.004) versus the meniscal group IL-8 = 0.219, +/- 0.171; CCR7 = 0.278, +/- 0.134). Relative expression levels of CCL19, CCL5 and XCL1 were similar in the two patient groups.

Figure 1. Synovial mRNA relative expression levels in patients with meniscal tears and advanced OA.

mRNA expression levels of CCL-19, CCR7, IL-8, XCL-1 and CCL5 were measured using Real-Time PCR as described in Materials and Methods. Expression was calculated relative to the mean in the advanced OA group after normalizing to GAPDH. **p = 0.004, *** p < 0.001

Relationship between chemokine levels and knee symptoms/dysfunction in patients undergoing meniscal arthroscopy

In the meniscal patients, CCL19 mRNA RE was associated with KOOS ADL (Pearson r = -0.620, p = 0.005), KOOS Pain (r =-0.547, p=0.015), and KOOS QOL scores (r=-0.479, 0.038), indicating that higher expression levels are associated with worse KOOS scores. No significant correlations between CCR7, CCL5, or IL-8 and KOOS subscores were observed (data not shown).

Linear regression analysis revealed that CCL19 RE levels were associated with KOOS ADL scores (estimate (B) = -0.602, 95% CI [-8.071, -0.331], p = 0.036) and were independent of age, gender, BMI, degree of cartilage damage (Outerbridge Score) and radiographic OA severity (K-L Score) in meniscal patients. Our final regression model including CCL19 RE levels explained 50% of the variance in KOOS ADL scores (F = 1.988, R²= 0.500, p = 0.145) (Table 2A). Regression analysis using Pain and QOL scores did not demonstrate statistically significant associations with CCL19 expression, after adjusting for covariates (KOOS Pain (B) =-0.536, p = 0.065; KOOS QOL (B) =-0.610, p = 0.070) (Table 2B and 2C).

Table 2A.

Regression models to determine associations between KOOS ADL scores and CCL19 relative expression levels.

Variable	Estimate (SE)	95% CI	p- value
Age	0.431 (0.621)	-0.921, 1.783	0.501
Gender	-28.927 (21.416)	-75.588, 17.735	0.202
BMI	-2.178 (1.77)	-6.051, 1.694	0.244
Outerbridge Score	0.854 (5.699)	-11.564, 13.271	0.883
K/L score	10.037 (9.529)	-10.725, 30.799	0.313
CCL-19	-4.201 (1.776)	-8.071, -0.331	0.036*
R2		0.500
F for R2 change		1.998
p-value for F			0.145

^*p < 0.05

Table 2B.

Regression models to determine associations between KOOS Pain scores and CCL19 relative expression levels.

Variable	Estimate (SE)	95% CI	p- value
Age	0.624 (0.560)	-0.597, 1.845	0.288
Gender	-23.097 (19.343)	-65.241, 19.048	0.256
BMI	-1.977 (1.605)	-5.474, 1.521	0.242
Outerbridge Score	-0.457 (5.147)	-11.673, 10.758	0.931
K/L score	8.941 (8.607)	-9.811, 27.693	0.319
CCL-19	-3.252 (1.604)	-6.748, 0.243	0.065
R2		0.461
F for R2 change		1.771
p-value for F			0.202

Table 2C.

Regression models to determine associations between KOOS QOL scores and CCL19 relative expression levels.

Variable	Estimate (SE)	95% CI	p- value
Age	-0.030 (0.652)	-1.451, 1.391	0.965
Gender	-0.276 (22.509)	-49.319, 48.768	0.990
BMI	-0.029 (1.868)	-4.041, 4.099	0.988
Outerbridge Score	-0.841 (5.990)	-13.893, 12.210	0.891
K/L score	6.935 (10.015)	-14.887, 28.756	0.502
CCL-19	-3.719 (1.867)	-7.786, 0.349	0.070
R2		0.277
F for R2 change		0.766
p-value for F			0.610

Synovial Fluid CCL19 levels

To determine whether CCL19 protein is produced within the joint, CCL19 levels in SF specimens from a separate set of 10 meniscal tear patients, 10 advanced OA patients, and 10 asymptomatic donors were measured by ELISA. Due to limitations of synovial fluid amounts in the repositories, this analysis only included 2 of the meniscal patients and 3 of the advanced OA patients included in the gene expression analysis. Characteristics of the subset of patients and donors contributing SF are shown in Table 3. Arthroscopy and arthroplasty patients included in this analysis were well matched for age, gender and BMI, although both patient groups tended to be younger than the asymptomatic donors. Again, K-L scores and Outerbridge scores were higher in the advanced OA patients compared to the meniscal tear patients, as expected. The arthroscopy patients contributing SF to this analysis were similar to those included in the gene expression analysis, as there was a predominance of degenerative-type tears and cartilage integrity loss, indicating that the majority had early to intermediate-stage knee OA.

Table 3.

Characteristics of patients with meniscal tears, advanced OA, and asymptomatic donors that contributed SF for CCL19 analysis.

	Meniscal Tear Patients N= 9	Advanced OA Patients N =10	Asymptomatic Donors N = 10	p-value
Age, median (IQR), years	53.5 (38.7-60.25)	55.5 (46.7-59.2)	69.5 (63.2-73.5)	0.004a
BMI, median (IQR), kg/m2	34.4 (27.2-37.4)	37.6 (34.7-42.2)	N/A	0.08b
Male/Female N, (%)	5/5 (50/50)	6/4 (60/40)	6/4 (60/40)	0.87c
Kellgren-Lawrence Score, median (IQR)	2 (1-2.7)	4 (3-4)	ND	0.007d
Outerbridge Score (N, % Of Patients)
Grade 0	1 (11)	0 (0)
Grade 1	1 (11)	0 (0)	ND	0.032d
Grade 2	2 (22)	0 (0)
Grade 3	2 (22)	2 (20)
Grade 4	3 (33)	8 (80)
SF CCL19 pg/ml Median (IQR)	62.5 (0-117.9)	67.8 (0-200)	0 (0-59.3)	0.25e

^aOne-way ANOVA followed by Holm-Sidak's multiple comparison test. Meniscal vs. asymptomatic, p=0.004; Advanced OA vs. asymptomatic, p=0.02.

^bUnpaired t-test.

^cχ² test

^dFisher's Exact test, K-L 0-2 vs. 3-4, or Outerbridge 0-2 vs. 3-4

^eKruskall-Wallis

N/A: Notavailable

ND: Not done

CCL19 was detectable in SF of 7/10 patients with meniscal tears, 7/10 advanced OA patients, and 4/10 asymptomatic donors (χ², p=0.28). Median (interquartile range) levels in meniscal tear patients were 62.5 (0-117.9), in advanced OA 67.8 (0-200), and in the asymptomatic donors 0 (0-59.2) (Table 3; Kruskall-Wallis p = 0.25).

Discussion

We previously reported a relationship between chemokine markers of synovitis, and increased knee symptoms and dysfunction in patients undergoing arthroscopic meniscectomy without radiographic evidence of OA (12). In the current study, we sought to determine whether this relationship between chemokine expression and knee symptoms could be replicated in a more typical population of patients indicated for meniscal arthroscopy. The majority of arthroscopy patients in this study demonstrated Outerbridge grade 1-4 cartilage lesions, had K-L radiographic scores >/= 2, consistent with published reports demonstrating pre-radiographic or radiographic evidence of OA in many patients presenting with symptomatic meniscal tears (14,15). This is in contrast to the select group in our previous pilot study, 20% of whom had no cartilage abnormality, and all of whom had normal radiographs (K-L grade 0). Therefore, the current study allowed us to validate our previous findings with the goal of identifying markers of inflammation in a more heterogeneous but representative clinical population.

We focused our analysis on four inflammatory chemokines and one receptor identified previously: CCL19, IL-8, CCL5, XCL-1 and CCR7. These genes are involved in the recruitment and trafficking of leukocytes, including monocytes, lymphocytes, dendritic cells, neutrophils, and eosinophils to sites of inflammation (23). All five transcripts were expressed by the majority of patients presenting for meniscal surgeries and patients with advanced OA (Figure 1). Levels of CCR7 and IL-8 transcripts were significantly higher in the advanced OA group, while levels of CCL19, XCL1, and CCL5 were similar in the two groups. Expression of various chemotactic factors has been previously reported in knee injury and OA patients. A previous study by Cuellar et.al. (24) showed increased mean synovial fluid concentrations of monocyte chemotactic protein-1 (MCP-1) and macrophage inflammatory protein-1 (MIP-1β) in patients with acute symptomatic meniscal injury compared with asymptomatic control subjects. Another recent study detected CCL5 protein levels in synovial fluid, but reported lower levels in post-traumatic arthroscopy patients compared to OA patients undergoing total knee replacements (25). Elevated gene expression levels of proinflammatory cytokines and chemokines in the meniscus have also been reported in young patients with an isolated meniscal tear (26) suggesting the meniscus itself may be a source of these mediators. These studies along with our observations give further support to the hypothesis that low-grade inflammation is a characteristic of OA disease even in early stages. In contrast to other studies, our analysis focused solely on chemotactic factors specifically identified based on their association with histologic synovitis (12).

Next we investigated whether these markers of synovitis were associated with pre-operative symptoms measured using the KOOS outcome score. We found that of the five mediators measured, only synovial CCL19 mRNA expression levels were associated with KOOS scores. Unadjusted associations were observed with KOOS-ADL, KOOS- Pain, and KOOS – QOL subscores, but the association with KOOS-ADL was independent of age, gender, BMI, K-L and Outerbridge scores. As we had limited numbers of synovial tissues from arthroscopy patients with a full set of clinical data in the repository, we used an exploratory analytic approach without adjustment for multiple tests of hypotheses. We chose this approach to balance concerns of obtaining a spurious relationship against the danger of making a Type II error (accepting the null hypothesis when it is false). Despite its limitations, this approach replicated, strengthened and expanded our previous findings (12), and demonstrated that CCL19 mRNA expression in the synovium is associated with more severe symptoms in more typical meniscal arthroscopy patients, despite numerous potential confounding factors such as BMI, and varying degrees of pre-existing OA. Although a cause and effect relationship between the activity of CCL19 and knee-related dysfunction cannot be extrapolated from this study, this association is intriguing, and should be explored in future mechanistic studies. Of note, gene expression levels of CCL19 along with other OA related genes have been shown to be up-regulated in post-traumatic mouse models of OA, suggestive of the importance of these genes in an early inflammatory response to joint injury and progression of OA (27, 28).

The use of the KOOS instrument in this study allowed us to further define the relationship between CCL19 mRNA expression and specific knee-related symptoms. The Lysholm score used previously is a clinician-administered survey, which measures pain, other symptoms and knee related dysfunction in a variety of tasks on a single scale, and therefore cannot discriminate between specific domains of symptoms and functioning. In contrast, the KOOS is a patient- administered knee instrument which measures pain, other symptoms, function and quality of life on 5 independent subscales (17) and has been validated in knee injury populations including patients undergoing meniscal arthroscopies (29). Using the KOOS, we identified an independent association with the KOOS ADL subscore, which measures knee-related difficulty/dysfunction with activities of daily living. Our results are consistent with those of Sowers et al, who found an association between MRI-defined synovitis and knee-related dysfunction in OA patients, using objective measurement of stair-climbing and walking times (30). The specific reasons for a relationship between levels of this inflammatory mediator and knee dysfunction are as yet unclear, but we speculate may be related to inflammation-associated symptoms such as stiffness. Further studies using animal models are necessary to understand whether a mechanistic link between CCL19 and knee related symptoms and dysfunction in early stages of OA, exists.

mRNA expression levels of CCL19 were similar in the advanced knee OA and meniscal patients, most of whom had signs of early stage OA either on x-ray or intra-operatively. Our SF ELISA measurement confirmed that CCL19 is indeed expressed at the protein level in the joint. CCL19 was detectable in 7 of 10 SF from meniscal arthroscopy patients, and median levels were 62.5 (0-117.9). SF CCL19 was also detectable in 7 of 10 patients with advanced knee OA and median levels were similar (Table 3). In contrast, CCL19 was detectable in SF from 4 of 10 asymptomatic donors, and median levels were 0 (0-59.2) (Table 3; Kruskall-Wallis p = 0.25). Although not statistically significant given the small numbers of samples, this comparative analysis indicates that levels may be higher in disease compared to asymptomatic donors. However, we cannot rule-out variable post-mortem degradation of the donor specimens which is a limitation of this comparison.

Additional limitations of this study include small numbers as well as the cross-sectional nature of the design. Longitudinal studies are needed to determine a relationship between expression of these chemokines with post-operative prognosis, and to further elucidate longitudinal changes in gene expression patterns. Problems with using regression analyses are well known (31), including the possibility of finding a significant result that is instead an artifact of co-linearity among variables. We attempted to minimize this chance by using a hierarchical approach to entry of variables based on theoretical importance (22). Although causation cannot be inferred by this approach alone, it can be used to develop possible associations between interrelated variables such as the ones examined in the current analysis (32). As the regression models used in this report were exploratory in nature, validation of these markers in larger populations is needed, as well as in other joint injury and at-risk patients to determine if CCL19 has value in predicting outcomes including risk of OA. Even with these limitations, we were able to validate the relationship between synovial CCL19 mRNA expression and knee symptoms/dysfunction that we previously identified, and demonstrate that CCL19 protein was detectable in a more heterogeneous but representative clinical population. Our study provides support that synovial CCL19 expression may be a signature chemokine marker associated with inflammatory infiltration and knee-related dysfunction in the setting of meniscal degeneration and early stage knee OA.