You are here

Article Text

Article menu
Clinical and epidemiological research
Extended report
Circulating C reactive protein in osteoarthritis: a systematic review and meta-analysis
  1. Xingzhong Jin1,
  2. Julieta Ruiz Beguerie2,
  3. Weiya Zhang3,
  4. Leigh Blizzard1,
  5. Petr Otahal1,
  6. Graeme Jones1,
  7. Changhai Ding1,4,5
  1. 1Menzies Research Institute Tasmania, University of Tasmania, Hobart, Tasmania, Australia
  2. 2Department of Dermatology, Austral University Hospital, Buenos Aires, Argentina
  3. 3Academic Rheumatology, University of Nottingham, Nottingham, UK
  4. 4Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Victoria, Australia
  5. 5Arthritis Research Institute, 1st Affiliated Hospital of Anhui Medical University, Hefei, Anhui, China
  1. Correspondence to Prof Changhai Ding, Menzies Research Institute Tasmania, University of Tasmania, Private Bag 23, Hobart, TAS 7000, Australia; Changhai.Ding{at}utas.edu.au

Abstract

Background There is emerging evidence that the development and progression of osteoarthritis (OA) is associated with inflammation. C reactive protein (CRP), a systemic marker for inflammation, may be elevated in OA patients but the evidence is conflicting.

Objective To systematically review the literature for the relationship between serum CRP levels measured by a high sensitivity method (high sensitive CRP (hs-CRP)) and OA, as well as the correlation between circulating CRP levels and OA phenotypes.

Methods MEDLINE, EMBASE and CINAHL databases were systematically searched from January 1992 to December 2012. Studies were included when they met the inclusion criteria and data from studies were extracted. Two independent reviewers assessed study quality using a modified Newcastle-Ottawa Quality Assessment Scale. Meta-analyses were performed to pool available data from included studies.

Results 32 studies met the inclusion criteria. Serum hs-CRP levels in OA were modestly but statistically significantly higher than controls (mean difference=1.19 mg/L, 95% CI 0.64 to 1.73, p<0.001) with significant heterogeneity between studies. Levels were significantly associated with pain (r=0.14, 95% CI 0.09 to 0.20, p<0.001) and decreased physical function (r=0.25, 95% CI 0.13 to 0.39, p<0.001). No significant associations were found between hs-CRP levels and radiographic OA.

Conclusions Low-grade systemic inflammation may play a greater role in symptoms rather than radiographic changes in OA.

  • Osteoarthritis
  • Inflammation
  • Epidemiology

https://doi.org/10.1136/annrheumdis-2013-204494

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Background

    Osteoarthritis (OA) is the most common form of arthritis in the world.

    Although OA has generally been perceived as a ‘non-inflammatory’ arthropathy, recent studies have suggested that local inflammation plays a prominent role in its pathogenesis.1 Pro-inflammatory cytokines, including interleukin (IL)-1β, tumour necrosis factor-α, IL-6 and others, are produced by synovium and chondrocytes and contribute to the progression of cartilage degradation.2 A number of studies suggest that this local inflammation may be observed systemically.3–6

    C reactive protein (CRP) is an acute phase protein that is produced by hepatocytes and adipocytes and regulated by pro-inflammatory cytokines.7 An immunoassay was introduced in 1992 by Montagne et al8 to perform high sensitive CRP (hs-CRP) measurement. In 1997, a study by Loose et al showed that OA patients had higher serum hs-CRP levels than age-matched controls.9 Since then, numerous studies have been conducted to explore the association between serum CRP levels and OA. While some studies reported that serum CRP levels were significantly increased in OA,10–12 others showed no association between hs-CRP levels and OA after the adjustment for body mass.13–15 The aim of this study, therefore, was to perform a systematic review of the associations between hs-CRP levels and OA and the individual features of OA.

    Objectives

    1. To determine whether hs-CRP levels are elevated in OA patients compared with control.

    2. To determine the correlation between serum hs-CRP levels and the following OA features:

      • Prevalence or progression of radiographic features of OA (eg, joint space width, osteophyte, classified Kellgren–Lawrence score).

      • Prevalence or progression of joint symptoms (joint pain, stiffness or physical dysfunction).

    Methods

    Literature search

    A systematic search on literature from January 1992 to December 2012 was performed on electronic databases including MEDLINE, EMBASE and CINAHL. MeSH terms ‘osteoarthritis’, ‘inflammation’, ‘c-reactive protein’ and related free text terms were used for the search. Search filters designed by Scottish Intercollegiate Guidelines Network for observational studies were incorporated in the electronic database search strategies.16 Results were then limited to human epidemiological and clinical studies in English language. The search strategy for each electronic database is detailed in the online supplementary appendix 1. We tried to identify ongoing clinical trials by electronically searching ClinicalTrials.gov, WHO International Clinical Trial Registration Platform Search Portal, and Australian and New Zealand Clinical Trial Registry. The reference lists of obtained studies from the initial electronic search were scanned for further unidentified relevant studies. Conference abstracts from the American College of Rheumatology (ACR) and the European League Against Rheumatism (EULAR) were also searched.

    Selection of studies

    Two independent investigators (XJ and JRB) were assigned for the selection of studies. Titles, abstracts, keywords and information of identified studies were entered in an Inclusion/Exclusion form (see online supplementary appendix 2). The initial screening was set to be relatively open-ended to retain as many relevant studies as possible. Full text was then further examined if the collected information of a primary study suggested that it might meet the inclusion criteria for this review. When information from the published article was not sufficient to make a judgment, correspondent authors were contacted to obtain further information. Discrepancies between two investigators were addressed by consensus after discussion.

    Inclusion criteria

    Studies that fulfilled the following criteria were included in this systematic review.

    1. Studies included patients with OA

    2. Serum CRP levels were measured using high sensitivity methodology

    3. Study compared OA patients with healthy subjects or associated serum hs-CRP levels with phenotypes of OA (eg, radiological grading, joint space narrowing, pain score and dysfunction score)

    4. The article represented original data

    5. Human study

    6. Studies published in English.

    Exclusion criteria

    1. Studies included patients with inflammatory joint diseases and other acute inflammatory conditions

    2. Review article

    3. In vitro, animal or ex vivo study.

    Data extraction

    One investigator (XJ) extracted the data from included studies using a predesigned data extraction form (see online supplementary appendix 3). The accuracy of the data was verified by a second investigator (JRB). Study characteristics were recorded including publication information, study design, origin of study, study setting, time frame of study, age, gender split, body mass index (BMI), definition of OA, affected joints, hs-CRP measuring method, serum hs-CRP levels, relative measures and correlational data with OA.

    Quality assessment

    Two investigators (XJ and JRB) independently evaluated the methodological quality of all included studies. The assessment was based on the Newcastle-Ottawa Quality Assessment Scale (NOS) for case-control studies17 with modifications to accommodate the topic of this review (see online supplementary appendix 4). The NOS was identified to be one of the two useful tools to assess the quality of non-randomised studies in a systematic review of 182 tools.18 The total quality score was not used in the meta-analyses, as we believed it was more appropriate to assess different aspects of methodological quality of a study in a separate manner.

    Assessment of heterogeneity

    Heterogeneity across included studies was examined using Cochran Q test and I2 test. A result of χ2>25% and p<0.10 was defined as evidence of significant heterogeneity across studies. To further analyse heterogeneity, the I2 test was used to estimate the extent of heterogeneity, for example, the percentage of variation across studies that is not caused by chance. A I2 value higher than 30% would indicate moderate heterogeneity and a value higher than 50% would represent substantial heterogeneity.19 Possible sources of heterogeneity and their effects on the results were explored by subgroup analyses and sensitivity analyses.

    Assessment of publication bias

    Publication bias and ‘small-study effects’ were evaluated using a funnel plot. Asymmetry identified on the funnel plot would imply possible publication bias. A modified Egger's regression test was performed to detect the publication bias. When p value equalled or was less than 0.10, significant publication bias was considered.20

    Data synthesis and analysis

    For dichotomous data, a pooled OR and a 95% CI were computed by the Mantel–Haenszel method. For studies that reported relative risks (RR), we made an attempt to reconstruct a 2×2 table with information provided in the text and calculated the OR.

    Because crude hs-CRP levels are frequently skewed, some individual studies had normalised the hs-CRP by logarithmic transformation in the statistical analysis while others reported the results on original scale. In order to allow meta-analysis to be conducted on a common scale, we adopted the methods proposed by Higgins et al to transform data from a logarithmic scale to a raw scale.21 The unit of hs-CRP measurement was uniformly converted to mg/L in the meta-analysis. Difference in means and its 95% CI was calculated to estimate the difference in hs-CRP levels between OA patients and healthy controls. Generic inverse variance method on the random-effects model was used for the statistical pooling as we expected the true effects would vary across individual studies.

    For correlational data, we obtained correlation coefficients (r) and calculated the corresponding SD by computing the square root of sample variance as below.Embedded Image

    Correlation coefficients were combined using generic inverse variance method and the random-effects model.

    When data were sufficient and appropriate, prespecified subgroup analyses stratified by age, BMI, joints of OA, definition of OA, hs-CRP measuring methods and study designs were performed to assess the influence of the above parameters. Meta-regression analysis was performed to assess the influence of age, BMI and female sex using the random-effects model.

    Statistical analyses were performed using Revman V.5.2 (The Nordic Cochrane Centre, The Cochrane Collaboration, 2012) and STATA (Release V.12. College Station, Texas, USA. StataCorp LP). All reported p values were two sided and p<0.05 was considered to be statistically significant.

    Results

    Literature search

    A total of 1198 records were retrieved from initial search on electronic databases, with 523 from MEDLINE, 346 from EMBASE and 329 from CINAHL. In all, 20 relevant abstracts were identified from ACR and 40 from EULAR. Reference mining identified another 14 articles. Overall, 1272 records were identified. A total of 185 duplicated records were removed. After title and abstract screening, 1025 articles were excluded. In all, 67 full texts were examined for the eligibility of inclusion. Overall, 35 articles were further excluded because they were ex vivo studies, in vitro studies, review articles, and studies in which OA was not the disease of interest, CRP levels were not measured or analysis data on CRP were not performed or not reported in the article. Two articles5 ,22 were also excluded because the same data were reported in another article. One study23 which used the number of OA-affected knees instead of number of OA patients in the analysis was excluded. Finally, 32 studies met the inclusion criteria (figure 1).

    Figure 1
    Figure 1

    Study selection flow chart.

    Included studies

    Ten case-control,24–33 15 cross-sectional,15 ,34–47 four longitudinal studies48–51 and three clinical trials were included.9 ,52 ,53 The three clinical trials had data comparing CRP levels in OA patients with healthy controls at baseline; therefore, their data were included and assessed as a case-control design. The characteristics of included studies are summarised in table 1 (information for individual studies is detailed in online supplementary table S2). The 32 included studies provided data of 17 090 participants (6440 OA cases and 10 650 controls) in 12 countries.

    View this table:
    Table 1

    Characteristics of included studies

    Excluded studies

    Excluded studies after full text assessment and reasons for exclusion are listed in online supplementary table S1.

    Quality assessment

    The results of methodological quality assessment for each individual study were summarised in online supplementary table S3. Discrepancies between two reviewers were resolved by consensus after discussion. The average methodological quality across included studies was satisfactory. Seventeen studies were scored above 6 out of 9. However, two studies, Keenan et al (2008)42 and Sharif et al (2000),49 were graded 2 and 3 respectively, suggesting a higher risk of bias. Another study9 was only published in abstract and scored in only four items due to limited information. All studies but two9 ,42 provided clear definitions of OA and description of CRP measurement (table 1).

    View this table:
    Table 2

    Summary of results of meta-analyses

    Meta-analyses

    The results of all meta-analyses are summarised in table 2.

    Comparison of hs-CRP levels between OA and non-OA

    We made efforts to retain all available data from included studies in the meta-analysis. Correspondent authors were contacted by email for studies that had missing data or insufficient information. Eventually, data from 15 primary studies were available for this component of the meta-analysis. Two studies36 ,37 reported hs-CRP levels in different OA subgroups; therefore, we separately entered the subgroup data in the meta-analysis. Six studies9 ,24 ,29 ,39 ,41 ,45 had either missing data or insufficient information for transforming data that allow to be pooled in the meta-analysis. Data from one study48 were also not included because observations in the control group were not completely free of OA. A summary of results from these seven studies is presented in online supplementary table S4.

    Fifteen studies were included in the meta-analysis of the difference in means of hs-CRP levels between OA patients and healthy controls (figure 2). Except for one study28 showing no difference between OA and controls, all other studies revealed that circulating levels of hs-CRP were higher in OA patients than in healthy controls. The pooled mean difference (MD) showed that hs-CRP level was significantly higher in OA than in controls, with an average increase in value of 1.19 mg/L (95% CI 0.64 to 1.73, p<0.001). There was significant variation across the included studies in terms of the MD of hs-CRP levels, which ranged from 0.06 to 3.43 mg/L. The test for heterogeneity was statistically significant (χ2=76.23, p<0.001; I2=79%), indicating substantial inter-study variation. Possible explanations for the variation were explored in subgroup analyses and sensitivity analyses.

    Figure 2
    Figure 2

    Comparison of C reactive protein levels between osteoarthritis (OA) and non-OA.

    Publication bias was examined by the evaluation of the symmetry of a funnel plot (see online supplementary figure S8) as well as the Egger's test. The funnel plot suggested that there were potential publication bias and small-study effects in the results. The existence of small-study effects was confirmed by the Egger's test. Therefore, we subsequently performed a sensitivity analysis to examine the impacts of small studies.

    Subgroup and sensitivity analyses

    Subgroup analyses based on OA joint sites (hip and knee) and diagnosis (clinical and radiological) were performed. Sensitivity analyses based on CRP measurement techniques (ELISA and nephelometry), study size (small and large), study design (case-control and cross-section) and adjustment for BMI were performed (see online supplementary results and figures). A summary of results is presented in table 3. The increase in hs-CRP level was more considerable in hip OA than in knee OA (3.37 vs 1.15 mg/L). Elevated hs-CRP levels were observed both in clinical OA and radiographic OA, 1.76 and 1.07 mgL higher than non-OA, respectively. Compared with nephelometry, high sensitivity ELISA was the more sensitive measuring technique. Small studies (N<100) were more likely to report significant differences. The increase in hs-CRP was reported to be more moderate in a case-control design than in a cross sectional design (0.91 vs 1.82 mg/L). Control for BMI in a study or adjustment for BMI in an analysis did not significantly change the increase of hs-CRP in OA.

    View this table:
    Table 3

    Summary of subgroup and sensitivity analyses of difference in hs-CRP between OA and non-OA

    CRP levels and OA progression

    Three studies reported relative measures, either RR or OR (see online supplementary figure S7), on the predictive value of hs-CRP levels for progression of OA as either exacerbation of joint space narrowing or total joint replacement. The pooled OR showed no significant predictive value of hs-CRP level in the progression of OA (OR=0.99, 95% CI 0.81 to 1.21, p=0.93). The test for heterogeneity was borderline significant (χ2=7.02, p=0.07; I2=57%).

    Correlations between hs-CRP levels and OA phenotypes

    hs-CRP and radiographic OA

    The links between serum hs-CRP levels and knee radiographic OA were investigated in four studies (figure 3). The pooled correlation coefficient showed that the link between serum hs-CRP levels and knee radiographic OA was weak and was not statistically significant (r=0.11, 95% CI −0.03 to 0.26, p=0.13). The correlation coefficients remained non-significant when we separated studies into Kellgren–Lawrence grading subgroup (r=0.13, 95% CI −0.08 to 0.35, p=0.22) and joint space narrowing subgroup (r=0.10, 95% CI −0.10 to 0.29, p=0.33). There was no significant difference between the two subgroups (χ2=0.07, p=0.79; I2=0%).

    Figure 3
    Figure 3

    Correlation between high sensitive C reactive protein levels and radiographic osteoarthritis.

    hs-CRP and symptoms of OA

    The correlation coefficients between hs-CRP levels and pain in OA patients were available in six studies. A total of 840 OA patients were included in the meta-analysis (figure 4). Brenner study34 reported a negative correlation between hs-CRP levels and visual analog scale (VAS) for pain score, while the other four all reported a positive correlation. The pooled result of the meta-analysis showed that there was a weak but statistically significant correlation between hs-CRP levels and pain scale score (r=0.14, 95% CI 0.08 to 0.20, p<0.001). There was no significant heterogeneity observed across the studies (χ2=4.12, p=0.39; I2=3%).

    Figure 4
    Figure 4

    Correlation between high sensitive C reactive protein levels and symptoms of osteoarthritis.

    Two studies reported the correlation coefficient of hs-CRP with physical function (figure 4).29 ,46 The results from both studies were consistent with each other, indicating a correlation between increased hs-CRP levels and worsening physical function. The pooled correlation coefficient was statistically significant (r=0.26, 95% CI 0.13 to 0.39, p<0.001).

    The pooled results of correlation data for symptoms of OA were statistically significant (r=0.16, 95% CI 0.11 to 0.22, p<0.001). There was no significant difference between correlation with pain and correlation with physical function loss (χ2=2.48, p=0.12; I2=59.7%).

    Discussion

    This study is a comprehensive systematic review of serum hs-CRP levels in OA. A related meta-analysis of three studies previously performed in 2010 by Kerkhof et al48 did not provide evidence of an association between serum CRP levels and knee OA. In the present systematic review, we included all available data in our meta-analysis on the difference of serum hs-CRP levels between OA population and healthy controls. In addition, the association between hs-CRP levels and different OA phenotypes was also examined.

    The major findings of this systematic review are: (1) serum hs-CRP level is modestly elevated in the OA population, as reported in most studies (19 out of 21) (MD=1.19 mg/L, p<0.001); (2) serum hs-CRP level is associated with symptoms of OA, such as pain and loss of physical function; and (3) serum hs-CRP level is not significantly associated with Kellgren–Lawrence scores and joint space narrowing.

    The first result above should be regarded with caution due to the significant heterogeneity. We identified some reasons for this with the apparent difference being greater in studies using ELISA as compared with nephelometry, smaller studies, cross-sectional studies and those restricted to hip or radiographic OA. This suggests variation between studies is due to both study quality issues and varying methodology. Cross-sectional studies should not be used for causal inference and there were only four longitudinal studies which are stronger methodologically.

    Obesity may influence serum hs-CRP levels. Adipose tissue mediates the secretion of pro-inflammatory cytokines, such as IL-6, which in turn triggers the hepatic synthesis of CRP.54–56 A previous meta-analysis did not show an association between serum CRP levels and knee OA independent of BMI;48 however, the meta-analysis included only three studies. In contrast, Sower and colleagues reported that difference in CRP levels between participants with knee OA and without knee OA remained statistically significant in both BMI ≤30 and BMI >30 subgroups.37 Our subgroup analysis, which compares studies adjusted for BMI with those did not, suggests that BMI does not explain the variation between studies and the association appears to be independent of BMI. Chronic inflammation, reflected by increased hs-CRP levels, may directly contribute to the occurrence of OA.

    With regard to OA phenotypes, our results show a consistent and homogeneous association with pain levels and activity limitations. The magnitude of association is small suggesting a minor explanatory proportion but would be consistent with a recent paper from our group where CRP predicted the development of knee pain over 5 years.11 In comparison, the associations between hs-CRP and radiographic OA are not significant.

    There are also statistical challenges for data transformation and synthesis. The distribution of hs-CRP values is skewed in most studies. The use of means and SDs to calculate MD in the meta-analysis requires an assumption of normal distribution. Because the SD is larger in a skewed distribution, this may weaken the statistical significance of our results. Furthermore, the majority of studies report crude values of hs-CRP levels while others present the data on geometric scale. In order to pool data on a common scale, logarithmic back-transformation to the raw scale was required. The method requires small effect sizes and similar distributional shapes in the two groups, which may not be the case in some studies.

    Future researches are therefore suggested based on this meta-analysis: (1) cohort studies with adequate adjustment for body mass and other common confounding factors such as age, sex and smoking to examine causal relationship between OA structure damage and subsequent elevation of the CRP and other inflammation markers; (2) studies to examine the effects of inflammation alone and in combination with structure abnormalities on OA symptoms; (3) studies using more sensitive OA imagine technologies such as MRI are also desired.

    Conclusions

    This systematic review suggests that low-grade systemic inflammation may play a greater role in symptoms rather than radiographic changes in OA.

    Acknowledgments

    We thank the following correspondence authors for their support and disclosure of unpublished data to make the meta-analysis feasible in this systematic review: Professor Virginia B Kraus (Duke University School of Medicine), Professor Til Stürmer (University of North Carolina) and Yvonne C Lee, MD (Brigham and Women's Hospital).

    References

    1. ↵
      1. Rainbow R,
      2. Ren W,
      3. Zeng L
      . Inflammation and Joint Tissue Interactions in OA: Implications for Potential Therapeutic Approaches. Arthritis 2012;2012:8.
      OpenUrl
    2. ↵
      1. Scanzello CR,
      2. Goldring SR
      . The role of synovitis in osteoarthritis pathogenesis. Bone 2012;51:249–57.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Filkova M,
      2. Liskova M,
      3. Hulejova H,
      4. et al
      . Increased serum adiponectin levels in female patients with erosive compared with non-erosive osteoarthritis. Ann Rheum Dis 2009;68:295–6.
      OpenUrlFREE Full Text
    4. ↵
      1. Lavigne P,
      2. Benderdour M,
      3. Lajeunesse D,
      4. et al
      . Expression of ICAM-1 by osteoblasts in healthy individuals and in patients suffering from osteoarthritis and osteoporosis. Bone 2004;35:463–70.
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Otterness IG,
      2. Weiner E,
      3. Swindell AC,
      4. et al
      . An analysis of 14 molecular markers for monitoring osteoarthritis. Relationship of the markers to clinical end-points. Osteoarthritis Cartilage 2001;9:224–31.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Toncheva A,
      2. Remichkova M,
      3. Ikonomova K,
      4. et al
      . Inflammatory response in patients with active and inactive osteoarthritis. Rheumatol Int 2009;29:1197–203.
      OpenUrlCrossRefPubMed
    7. ↵
      1. Calabro P,
      2. Chang DW,
      3. Willerson JT,
      4. et al
      . Release of C-reactive protein in response to inflammatory cytokines by human adipocytes: linking obesity to vascular inflammation. J Am Coll Cardiol 2005;46:1112–13.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Montagne P,
      2. Laroche P,
      3. Cuillière ML,
      4. et al
      . Microparticle-enhanced nephelometric immunoassay for human C-reactive protein. J Clin Lab Anal 1992;6:24–9.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Loose L,
      2. de Oliveira R,
      3. Sipe J,
      4. et al
      . A possible systemic component of osteoarthritis: Elevated concentrations (by ELISA) of C-reactive protein in the serum of OA patients and modulataion by tenidap (abstr.). Arthritis Rheum 1996;36(Suppl):S166.
      OpenUrl
    10. ↵
      1. Pearle AD,
      2. Scanzello CR,
      3. George S,
      4. et al
      . Elevated high-sensitivity C-reactive protein levels are associated with local inflammatory findings in patients with osteoarthritis. Osteoarthritis Cartilage 2007;15:516–23.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Stannus OP,
      2. Jones G,
      3. Blizzard L,
      4. et al
      . Associations between serum levels of inflammatory markers and change in knee pain over 5 years in older adults: a prospective cohort study. Ann Rheum Dis 2013;72:53540.
      OpenUrlAbstract/FREE Full Text
    12. ↵
      1. Sturmer T,
      2. Brenner H,
      3. Koenig W,
      4. et al
      . Severity and extent of osteoarthritis and low grade systemic inflammation as assessed by high sensitivity C reactive protein. Ann Rheum Dis 2004;63:200–5.
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Davis CR,
      2. Karl J,
      3. Granell R,
      4. et al
      . Can biochemical markers serve as surrogates for imaging in knee osteoarthritis? Arthritis Rheum 2007;56:4038–47.
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Nevitt M,
      2. Felson D,
      3. Peterfy C,
      4. et al
      . Inflammation markers (CRP, TNF-alpha, IL-6) are not associated with radiographic or MRI findings of knee OA in the elderly: The Health ABC Study [Abstract]. Arthritis Rheum 2002;46(Suppl):S372–3.
      OpenUrl
    15. ↵
      1. Vlad SC,
      2. Neogi T,
      3. Aliabadi P,
      4. et al
      . No association between markers of inflammation and osteoarthritis of the hands and knees. J Rheumatol 2011;38:1665–70.
      OpenUrlAbstract/FREE Full Text
    16. ↵
      Scottish Intercollegiate Guidelines Network. SIGN filters for observational studies. Secondary SIGN filters for observational studies 2012. http://www.sign.ac.uk/methodology/filters.html#obs
    17. ↵
      1. Wells GA,
      2. Shea B,
      3. O'Connell D,
      4. et al
      . The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Secondary The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 2012. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp
    18. ↵
      1. Deeks JJ,
      2. Dinnes J,
      3. D'Amico R,
      4. et al
      . Evaluating non-randomised intervention studies. Health Technol Assess 2003;7:iii–x, 1–173.
      OpenUrlPubMed
    19. ↵
      1. Higgins JP,
      2. Thompson SG,
      3. Deeks JJ,
      4. et al
      . Measuring inconsistency in meta-analyses. BMJ 2003;327:557–60.
      OpenUrlFREE Full Text
    20. ↵
      1. Harbord RM,
      2. Egger M,
      3. Sterne JA
      . A modified test for small-study effects in meta-analyses of controlled trials with binary endpoints. Stat Med 2006;25:3443–57.
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      1. Higgins JP,
      2. White IR,
      3. Anzures-Cabrera J
      . Meta-analysis of skewed data: combining results reported on log-transformed or raw scales. Stat Med 2008;27:6072–92.
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Otterness IG,
      2. Zimmerer RO,
      3. Swindell AC,
      4. et al
      . An examination of some molecular markers in blood and urine for discriminating patients with osteoarthritis from healthy individuals. Acta Orthopaedica 1995;66:148–50.
      OpenUrlWeb of Science
    23. ↵
      1. Golightly YM,
      2. Marshall SW,
      3. Kraus VB,
      4. et al
      . Biomarkers of incident radiographic knee osteoarthritis: do they vary by chronic knee symptoms? Arthritis Rheum 2011;63:2276–83.
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Bos SD,
      2. Suchiman HED,
      3. Kloppenburg M,
      4. et al
      . Allelic variation at the C-reactive protein gene associates to both hand osteoarthritis severity and serum high sensitive C-reactive protein levels in the GARP study. Ann Rheum Dis 2008;67:877–9.
      OpenUrlAbstract/FREE Full Text
    25. ↵
      1. Garnero P,
      2. Piperno M,
      3. Gineyts E,
      4. et al
      . Cross sectional evaluation of biochemical markers of bone, cartilage, and synovial tissue metabolism in patients with knee osteoarthritis: relations with disease activity and joint damage. Ann Rheum Dis 2001;60:619–26.
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Hulejova H,
      2. Baresova V,
      3. Klezl Z,
      4. et al
      . Increased level of cytokines and matrix metalloproteinases in osteoarthritic subchondral bone. Cytokine 2007;38:151–6.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Hussein MR,
      2. Fathi NA,
      3. El-Din AME,
      4. et al
      . Alterations of the CD4(+), CD8 (+) T cell subsets, interleukins-1beta, IL-10, IL-17, tumor necrosis factor-alpha and soluble intercellular adhesion molecule-1 in rheumatoid arthritis and osteoarthritis: preliminary observations. Pathol Oncol Res 2008;14:321–8.
      OpenUrlCrossRefPubMed
    28. ↵
      1. Melikoglu MA,
      2. Yildirim K,
      3. Senel K
      . Relationship between radiographic grading of osteoarthritis and serum beta-2 microglobulin. Ir J Med Sci 2009;178:151–4.
      OpenUrlPubMedWeb of Science
    29. ↵
      1. Punzi L,
      2. Ramonda R,
      3. Oliviero F,
      4. et al
      . Value of C reactive protein in the assessment of erosive osteoarthritis of the hand. Ann Rheum Dis 2005;64:955–7.
      OpenUrlAbstract/FREE Full Text
    30. ↵
      1. Otterness IG,
      2. Swindell AC,
      3. Zimmerer RO,
      4. et al
      . An analysis of 14 molecular markers for monitoring osteoarthritis: segregation of the markers into clusters and distinguishing osteoarthritis at baseline. Osteoarthritis Cartilage 2000;8:180–5.
      OpenUrlCrossRefPubMedWeb of Science
    31. ↵
      1. Sharif M,
      2. Elson CJ,
      3. Dieppe PA,
      4. et al
      . Elevated serum C-reactive protein levels in osteoarthritis. Br J Rheumatol 1997;36:140–1.
      OpenUrlFREE Full Text
    32. ↵
      1. So AK,
      2. Varisco PA,
      3. Kemkes-Matthes B,
      4. et al
      . Arthritis is linked to local and systemic activation of coagulation and fibrinolysis pathways. J Thromb Haemost 2003;1:2510–15.
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Takahashi M,
      2. Naito K,
      3. Abe M,
      4. et al
      . Relationship between radiographic grading of osteoarthritis and the biochemical markers for arthritis in knee osteoarthritis. Arthritis Res Ther 2004;6:R208–12.
      OpenUrlCrossRefPubMedWeb of Science
    34. ↵
      1. Brenner SS,
      2. Klotz U,
      3. Alscher DM,
      4. et al
      . Osteoarthritis of the knee--clinical assessments and inflammatory markers. Osteoarthritis Cartilage 2004;12:469–75.
      OpenUrlCrossRefPubMedWeb of Science
    35. ↵
      Biochemical markers are associated with radiographic subtypes of osteoarthris (OA) in subjects with familial OA at multiple sites. The GARP study. 2006.
    36. ↵
      1. Chen HC,
      2. Shah S,
      3. Stabler TV,
      4. et al
      . Biomarkers associated with clinical phenotypes of hand osteoarthritis in a large multigenerational family: the CARRIAGE family study. Osteoarthritis Cartilage 2008;16:1054–9.
      OpenUrlCrossRefPubMedWeb of Science
    37. ↵
      1. Sowers M,
      2. Jannausch M,
      3. Stein E,
      4. et al
      . C-reactive protein as a biomarker of emergent osteoarthritis. Osteoarthritis Cartilage 2002;10:595–601.
      OpenUrlCrossRefPubMedWeb of Science
    38. ↵
      1. Wolfe F
      . The C-reactive protein but not erythrocyte sedimentation rate is associated with clinical severity in patients with osteoarthritis of the knee or hip. J Rheumatol 1997;24:1486–8.
      OpenUrlPubMedWeb of Science
    39. ↵
      1. Engstrom G,
      2. Gerhardsson de Verdier M,
      3. Rollof J,
      4. et al
      . C-reactive protein, metabolic syndrome and incidence of severe hip and knee osteoarthritis. A population-based cohort study. Osteoarthritis Cartilage 2009;17:168–73.
      OpenUrlCrossRefPubMedWeb of Science
    40. ↵
      1. Garnero P,
      2. Mazieres B,
      3. Gueguen A,
      4. et al
      . Cross-sectional association of 10 molecular markers of bone, cartilage, and synovium with disease activity and radiological joint damage in patients with hip osteoarthritis: the ECHODIAH cohort. Journal of Rheumatology 2005;32:697–703.
      OpenUrlAbstract/FREE Full Text
    41. ↵
      1. Pearle AD,
      2. Scanzello CR,
      3. George S,
      4. et al
      . Elevated high-sensitivity C-reactive protein levels are associated with local inflammatory findings in patients with osteoarthritis. Osteoarthritis Cartilage 2007;15:516–23.
      OpenUrlCrossRefPubMedWeb of Science
    42. ↵
      1. Keenan RT,
      2. Swearingen CJ,
      3. Yazici Y
      . Erythrocyte sedimentation rate and C-reactive protein levels are poorly correlated with clinical measures of disease activity in rheumatoid arthritis, systemic lupus erythematosus and osteoarthritis patients. Clin Exp Rheumatol 2008;26:814–19.
      OpenUrlPubMedWeb of Science
    43. ↵
      1. Conrozier T,
      2. Chappuis-Cellier C,
      3. Richard M,
      4. et al
      . Increased serum C-reactive protein levels by immunonephelometry in patients with rapidly destructive hip osteoarthritis. Revue du Rhumatisme (English Edition) 1998;65:759–65.
      OpenUrl
    44. ↵
      1. Kraus VB,
      2. Stabler TV,
      3. Luta G,
      4. et al
      . Interpretation of serum C-reactive protein (CRP) levels for cardiovascular disease risk is complicated by race, pulmonary disease, body mass index, gender, and osteoarthritis. Osteoarthritis Cartilage 2007;15: 966–71.
      OpenUrlCrossRefPubMedWeb of Science
    45. ↵
      1. Spector TD,
      2. Hart DJ,
      3. Nandra D,
      4. et al
      . Low-level increases in serum C-reactive protein are present in early osteoarthritis of the knee and predict progressive disease. Arthritis Rheum 1997;40:723–7.
      OpenUrlCrossRefPubMedWeb of Science
    46. ↵
      1. Conrozier T,
      2. Carlier MC,
      3. Mathieu P,
      4. et al
      . Serum levels of YKL-40 and C reactive protein in patients with hip osteoarthritis and healthy subjects: a cross sectional study. Ann Rheum Dis 2000;59:828–31.
      OpenUrlAbstract/FREE Full Text
    47. ↵
      1. Sturmer T,
      2. Brenner H,
      3. Koenig W,
      4. et al
      . Severity and extent of osteoarthritis and low grade systemic inflammation as assessed by high sensitivity C reactive protein. Ann Rheum Dis 2004;63:200–5.
      OpenUrlAbstract/FREE Full Text
    48. ↵
      1. Kerkhof HJ,
      2. Bierma-Zeinstra SM,
      3. Castano-Betancourt MC,
      4. et al
      . Serum C reactive protein levels and genetic variation in the CRP gene are not associated with the prevalence, incidence or progression of osteoarthritis independent of body mass index. Ann Rheum Dis 2010;69:1976–82.
      OpenUrlAbstract/FREE Full Text
    49. ↵
      1. Sharif M,
      2. Shepstone L,
      3. Elson CJ,
      4. et al
      . Increased serum C reactive protein may reflect events that precede radiographic progression in osteoarthritis of the knee. Ann Rheum Dis 2000;59:71–4.
      OpenUrlAbstract/FREE Full Text
    50. ↵
      1. Livshits G,
      2. Zhai G,
      3. Hart DJ,
      4. et al
      . Interleukin-6 is a significant predictor of radiographic knee osteoarthritis: The Chingford Study. Arthritis Rheum 2009;60:2037–45.
      OpenUrlCrossRefPubMedWeb of Science
    51. ↵
      1. Mazieres B,
      2. Garnero P,
      3. Gueguen A,
      4. et al
      . Molecular markers of cartilage breakdown and synovitis at baseline as predictors of structural progression of hip osteoarthritis. The ECHODIAH Cohort. Ann Rheum Dis 2006;65:354–9.
      OpenUrlAbstract/FREE Full Text
    52. ↵
      1. Gungen G,
      2. Ardic F,
      3. Fiotandikoglu G,
      4. et al
      . The effect of mud pack therapy on serum YKL-40 and hsCRP levels in patients with knee osteoarthritis. Rheumatol Int 2012;32:1235–44.
      OpenUrlCrossRefPubMed
    53. ↵
      1. Lee YC,
      2. Lu B,
      3. Bathon JM,
      4. et al
      . Pain sensitivity and pain reactivity in osteoarthritis. Arthritis Care Res 2011;63:320–7.
      OpenUrlWeb of Science
      1. Kao TW,
      2. Lu IS,
      3. Liao KC,
      4. et al
      . Associations between body mass index and serum levels of C-reactive protein. S Afr Med J 2009;99:326–30.
      OpenUrlPubMedWeb of Science
      1. Visser M,
      2. Bouter LM,
      3. McQuillan GM,
      4. et al
      . Elevated C-reactive protein levels in overweight and obese adults. JAMA 1999;282:2131–5.
      OpenUrlCrossRefPubMedWeb of Science
      1. Choi J,
      2. Joseph L,
      3. Pilote L,
      4. et al
      . Obesity and C-reactive protein in various populations: a systematic review and meta-analysis. Obesity Reviews 2013;14:232–44.
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    • Lay summary

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • Handling editor Tore K Kvien

    • Transparency declaration The lead author affirms that this manuscript is an honest, accurate and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained.

    • Contributors XJ and CD designed the review protocol. XJ carried out the literature search. XJ and JRB contributed to study selection, data extraction and quality assessment. LB and PO provided statistical supports for meta-analysis. XJ performed the analyses and drafted the manuscript. All the authors contributed to the preparation of the manuscript.

    • Competing interests None.

    • Provenance and peer review Not commissioned; externally peer reviewed.