Familial clustering of the serum cytokine profile ... - Wiley Online Library

ARTHRITIS & RHEUMATISM Vol. 64, No. 6, June 2012, pp 1720–1729 DOI 10.1002/art.34449 © 2012, American College of Rheumatology

Familial Clustering of the Serum Cytokine Profile in the Relatives of Rheumatoid Arthritis Patients Hani S. El-Gabalawy,1 David B. Robinson,1 Irene Smolik,1 Donna Hart,1 Brenda Elias,1 Keng Wong,1 Christine A. Peschken,1 Carol A. Hitchon,1 Xuan Li,1 Charles N. Bernstein,1 Marianna M. Newkirk,2 and Marvin J. Fritzler3 Discriminant analysis and logistic regression classified individuals based on their cytokine profile. Results. The prevalence of RF (cutoff level predetermined to include 5% of Caucasian controls) and anti-CCP (cutoff level of >40 units) was, respectively, 88% and 81% in the RA patients, 34% and 9% in first-degree relatives, and 9% and 4% in North American Native controls; the prevalence of anti-CCP was 0% in Caucasian controls. Levels of most cytokines were highest in RA patients; 17 of 40 cytokines (43%) were significantly higher in first-degree relatives than in controls, including multiple proinflammatory cytokines. Discriminant analysis showed a notable distinction between the groups, with 85% classification accuracy. First-degree relatives had markedly higher MCP-1 and hsCRP levels than North American Native controls, but there was no consistent association with RA autoantibodies. Conclusion. Our findings indicate that levels of multiple cytokines and hsCRP are higher in first-degree relatives of North American Native patients with RA compared to individuals from a nonautoimmune background. These data suggest that elevated baseline cytokine levels may be part of the risk profile for developing RA.

Objective. Rheumatoid arthritis (RA) is prevalent in North American Native populations, with a high frequency of multicase families and seropositivity in first-degree relatives. This study was undertaken to determine whether the serum cytokine profile of firstdegree relatives of North American Native patients with RA differed from that of individuals with no family history of autoimmunity and whether there was an association with RA autoantibodies. Methods. North American Native patients with RA (n ⴝ 105), their first-degree relatives (n ⴝ 273), healthy North American Native controls (n ⴝ 200), and Caucasian controls (n ⴝ 150) were studied. Serum levels of 42 cytokines were tested using a multiplex laser bead assay. Rheumatoid factor (RF), anti–cyclic citrullinated peptide 2 (anti–CCP-2), monocyte chemotactic protein 1 (MCP-l), and high-sensitivity C-reactive protein (hsCRP) were tested by enzyme-linked immunosorbent assay, and HLA–DRB1 alleles by specific primers. Supported by the Canadian Institutes of Health Research (grants MOP 7770 and IIN 84040). 1 Hani S. El-Gabalawy, MD, FRCPC, David B. Robinson, MD, FRCPC, Irene Smolik, PhD, Donna Hart, RN, Brenda Elias, PhD, Keng Wong, BSc (Hons), Christine A. Peschken, MD, Carol A. Hitchon, MD, Xuan Li, MSc, Charles N. Bernstein, MD: University of Manitoba, Winnipeg, Manitoba, Canada; 2Marianna M. Newkirk, PhD: McGill University, Montreal, Quebec, Canada; 3Marvin J. Fritzler, MD, PhD: University of Calgary, Calgary, Alberta, Canada. Dr. Bernstein has received consulting fees, speaking fees, and/or honoraria from Abbott Canada, Janssen Canada, and Shire Canada (less than $10,000 each); owns stock or stock options in Pfizer; and has served as an expert witness on behalf of Mylan Pharmaceuticals, Barr Pharmaceuticals, Ranbaxy Pharmaceuticals, and Cardinal Health. Dr. Fritzler has received consulting fees, speaking fees, and/or honoraria from Inova Diagnostics and Bio-Rad (less than $10,000 each) and from Immuno Concepts (more than $10,000). Address correspondence to Hani El-Gabalawy, MD, FRCPC, Arthritis Centre, University of Manitoba, RR149, 800 Sherbrook Street, Winnipeg, Manitoba R3A 1M4, Canada. E-mail: elgabalh@ cc.umanitoba.ca. Submitted for publication September 5, 2011; accepted in revised form February 16, 2012.

Rheumatoid arthritis (RA) is a prevalent chronic inflammatory disorder with a substantial genetic contribution; there is an increased risk of RA development in the first-degree relatives of RA patients (1–7). In most populations, the development of RA occurs primarily in a sporadic manner, but previous studies have indicated that North American Native populations, including the Pima, Tlingit, and Cree/Ojibway, have a marked tendency toward familial clustering of disease (2,6,7). Such familial clustering might be based on genetic, epigenetic, or shared environmental influences. 1720

SERUM CYTOKINE PROFILE IN RELATIVES OF RA PATIENTS

Our previous studies of RA in the Cree/Ojibway population of Central Canada have demonstrated a high estimated prevalence of disease and a high frequency of multicase families (6,8). Based on administrative data from the Province of Manitoba, estimates of RA prevalence in this North American Native population are twice that for all other Manitobans; the RA prevalence is 3% in the North American Native population, based on self-report (8). Moreover, we have shown that approximately one-third of the first-degree relatives of North American Native patients with RA are seropositive for anti–citrullinated protein antibodies (ACPAs) and/or rheumatoid factor (RF) (9). Since it has been shown in several other populations that these two RA-associated autoantibodies are often present in the serum of individuals months to years prior to the onset of clinically detectable RA (10–12), first-degree relatives from this population appear to be at risk of future development of the disease. Cytokines are known to play a key role in the pathogenesis of RA, although the expanding complexity of the cytokine networks involved has challenged investigators to determine informative patterns of cytokine expression, both systemically and within the joints. This approach has been greatly facilitated by the availability of multiplex assays that allow the simultaneous determination of multiple cytokine levels (13–15). Using this approach, analysis of available preclinical serum samples from individuals who subsequently developed RA has shown that the levels of multiple cytokines rise immediately prior to disease onset, although the patterns detected have differed in the various studies (16–18). Despite these differences, a consistent theme that has emerged from these preclinical studies is the association between elevated cytokine levels and the presence of RA autoantibodies. Because of the high prevalence of ACPAs and RF in the first-degree relatives of North American Native patients with RA, and the strong familial clustering of disease in this population, we studied the serum cytokine profile of a cohort of first-degree relatives and compared the profile to that of the RA probands and a control population of North American Native and Caucasian individuals with no family history of autoimmune disease. Our findings indicate that the serum cytokine profile of the first-degree relatives resembles that of the RA patients, and can readily be discriminated from that of the control populations, with the chemokine monocyte chemotactic protein 1 (MCP-1) being particularly important in this distinction.

1721

PATIENTS AND METHODS Study subject recruitment. Cree/Ojibway patients with RA were recruited from clinic populations in several urban and rural locations in Central Canada, including Winnipeg, Manitoba and Saskatoon, Saskatchewan (urban), and Norway House and St. Theresa Point, Manitoba (rural). All patients met the American College of Rheumatology 1987 revised criteria for RA (19). RA patients were asked to approach their first-degree relatives regarding study participation. Individuals who were willing to participate were enrolled at rheumatology and community health clinics in their respective areas. North American Native and Caucasian controls were recruited from the same geographic areas as the RA patients and first-degree relatives by advertising in the local media and at health fairs. All controls were specifically questioned about a personal or family history of autoimmune diseases and were excluded from the study if they had a personal or family history of RA, systemic lupus erythematosus, scleroderma, polymyositis, vasculitis, spondylarthritides, multiple sclerosis, inflammatory bowel disease, type 1 diabetes mellitus, or thyroid disease. All North American Native study subjects, including controls, were included only if 3 of their 4 grandparents were of North American Native origin (by self-report). All study subjects provided informed consent in their language of choice, and all aspects of the study were approved by the Research Ethics Board of the University of Manitoba and by the Band Councils of the individual study communities. Study protocol and procedures. Study subjects completed a detailed study questionnaire, with assistance from a translator if necessary. Information obtained included demographic features, medical history, musculoskeletal and connective tissue disease symptoms, medication use, and family history. All study subjects then underwent a systematic examination by a rheumatologist (HEG or DBR) to detect the presence of synovitis or other stigmata of inflammatory arthritides. Serum samples were obtained by centrifugation at the time of the clinical evaluation and were stored at ⫺80°C until tested, with no freeze–thaw cycles involved. Samples obtained from RA patients, first-degree relatives, North American Native controls, and Caucasian controls were all processed according to a standard operating procedure. Anti–cyclic citrullinated peptide (anti-CCP) testing. Serum samples from RA patients, first-degree relatives, and controls were tested for the presence of anti-CCP antibodies using a standard commercially available anti–CCP-2 enzymelinked immunosorbent assay (ELISA; Inova Diagnostics). In determining anti-CCP seropositivity, a cutoff of ⱖ40 units was chosen to maximize specificity. Based on this cutoff, and using the Caucasian controls as a reference population, the sensitivity and specificity for the anti–CCP-2 test in the North American Native RA patient group were 81% and 100%, respectively. Had the manufacturer’s suggested cutoff of ⱖ20 units been used in the RA cohort, the test would have yielded a sensitivity and specificity of 86% and 92%, respectively. A number of samples were tested in duplicate to ensure the consistency of data between plates. Rheumatoid factor testing. IgM-RF was determined in the laboratory of one of us (MMN) by an ELISA calibrated with a standard of known IU measured by nephelometry. Because cutoff levels for RF have not been established in this North American Native population and in view of the high

1722

prevalence rate of smoking in the population, a factor known to increase the prevalence of RF, the cutoff for RF seropositivity was set at ⱖ50 IU, based on a level where 95% of the Caucasian controls were seronegative. Using the Caucasian controls as a reference population, this cutoff level gave a sensitivity and specificity of 88% and 73%, respectively, in the North American Native RA patient group. Multiplex addressable laser bead assay (MALBA) for serum cytokine testing. Testing for cytokine, chemokine, and growth factor levels was performed in the laboratory of one of us (MJF) using a 42-plex addressable laser bead assay with a Luminex 200 flow fluorometer (Eve Technologies). Each assay run included control normal and positive samples as well as a set of calibrators, which allowed determination of whether a single test result was within or out of range. Intertest variability was monitored by continuous tracking of control data values over time. Stability of intratest and intertest variability was set to ⬍5% of the mean values of preceding tests. Intratest variability was also addressed by routinely running random samples in duplicate or triplicate at the beginning and end of the run. The frequency of variability of ⬎5% in 1 year was ⬍2% of all runs. The effect of confounding factors (immune complexes, RF, cryoglobulins, protein precipitates, and hemolysis) was also monitored regularly and rarely accounted for variability that exceeded the 5% threshold. Moreover, we conducted runs with and runs without various commercial nonspecific blocking agents and did not find a consistent beneficial or deleterious effect of those agents. The samples included in the present study were split into 2 different assay runs, with overlap of selected samples to ensure consistency. Each individual run used a single lot of analyte-coupled beads and reagents. The first sample run included the RA patients (n ⫽ 104), first-degree relatives (n ⫽ 124), North American Native controls (n ⫽ 100), and Caucasian controls (n ⫽ 100). The second sample run included first-degree relatives (n ⫽ 149), North American Native controls (n ⫽ 100), and Caucasian controls (n ⫽ 50). There were no significant differences (⬎5%) between the 2 runs in terms of the internal negative and positive controls. MCP-1 testing. As indicated in the Results below, we found that serum MCP-1 levels as detected by multiplex cytokine testing were an important discriminator between the study groups. Thus, in order to validate these findings using an alternative platform, MCP-1 levels were also tested using a commercially available ELISA (R&D Systems). High-sensitivity C-reactive protein (hsCRP) testing. Levels of hsCRP were determined in the serum samples using a commercially available ELISA, according to the recommendations of the manufacturer (R&D Systems). HLA–DRB1 allele testing. HLA–DRB1 typing was performed by polymerase chain reaction using sequencespecific oligonucleotide primers in a subset of North American Native patients with RA, first-degree relatives, and controls. The following DRB1 alleles were included as shared epitope– bearing alleles: DRB1*0101, 0102, 0401, 0404, 0405, 0408, 0410, 1001, and 1402, as previously described (20). Statistical analysis. Statistical analysis was performed using SPSS version 18 software. The raw cytokine data were normalized by log transformation for analysis and comparisons between groups. Patients with RA, first-degree relatives, and North American Native and Caucasian controls were initially compared using analysis of variance on the log-transformed

EL-GABALAWY ET AL

data, with a post hoc Bonferroni adjustment to correct for multiple comparisons. Thus, based on the 42 cytokines and chemokines tested by MALBA, the significance level for differences between groups was adjusted to P ⫽ 0.001 rather than P ⫽ 0.05. Discriminant analysis was used to classify the North American Native groups (RA patients, first-degree relatives, and controls) based on the normalized 42-plex cytokine data. In this analysis, 2 discriminant functions were generated that were based on nonoverlapping contributions from each cytokine, where the contribution of the cytokine to the discriminant function was based on the variability exhibited within and between groups, and was quantified using a standardized coefficient (21). First-degree relatives were specifically compared to North American Native controls using a multivariate logistic regression model where demographic data, log-transformed cytokine levels, hsCRP levels, autoantibody seropositivity, smoking status, and shared epitope status were used as input variables.

RESULTS The demographic characteristics, smoking rates, anti-CCP and RF prevalence, and shared epitope frequency in the study groups are shown in Table 1. Of the 105 RA patients recruited, 76 (72%) had ⱖ1 firstdegree relative enrolled in the study. There were no significant clinical differences between the RA patients with and those without first-degree relatives enrolled in the study (data not shown). In total, the 273 first-degree relatives were recruited from 105 families. Of these 273 first-degree relatives, 59 (22%) were recruited from 29 families in whom the RA proband was not included in the study for a variety of reasons, including lack of availability, death, etc. The median number of firstdegree relatives per family was 3 (range 1–14). The distribution of first-degree relatives relative to family size is available online at www.proteome.ca. This distribution indicated that, although there were several large families included in the analysis, 151 (55%) of the 273 first-degree relatives were recruited from smaller families of ⱕ3 members. The data in Table 1 demonstrate a high prevalence of smoking in the North American Native groups, with a slightly higher frequency of ever smokers in first-degree relatives compared to North American Native controls. The frequency of shared epitope alleles in the North American Native groups was comparable to what we have previously demonstrated in larger immunogenetic studies (20) and emphasizes the high frequency of these alleles, particularly *1402 and *0404, in the background population. The estimated prevalences of RF and anti-CCP are also shown in Table 1. On the basis of the selected cutoff levels described in Patients and Methods, 91% of the RA patients were either anti-CCP or RF positive,


1723

Table 1. Characteristics of the RA patients, first-degree relatives, and controls*

Sex, % female Age, mean ⫾ SD years Urban/rural residence, % History of smoking Ever, % Current, % Pack-years, mean ⫾ SD Anti–CCP-2 positive (ⱖ40 units), % RF positive (ⱖ50 units), % Anti–CCP-2 and RF positive, % Shared epitope status, no. (%)‡ Any shared epitope 2 copies shared epitope

RA patients (n ⫽ 105)

First-degree relatives (n ⫽ 273)

North American Native controls (n ⫽ 200)

Caucasian controls (n ⫽ 150)

87 46.5 ⫾ 14.5 48/52

69 35.1 ⫾ 12.5 47/53

60 35.5 ⫾ 9.1 68/32

75 39.9 ⫾ 9.5 85/15

71 55 13.7 ⫾ 10.5 80.8 87.5 76.9

81 65 9.4 ⫾ 10.1 8.5 33.5 3.3

62 62 10.0 ⫾ 7.5 4 8.5 1.5

39 17 9.0 ⫾ 8.9 0 4.7† 0

61 (90) 21 (31)

169 (82) 50 (24)

96 (61) 32 (20)

NA NA

* Anti–CCP-2 ⫽ anti–cyclic citrullinated peptide 2; NA ⫽ not available. † Predetermined based on a cutoff level for rheumatoid factor (RF) that excluded 95% of the Caucasian controls. ‡ Information on shared epitope status was available for 68 rheumatoid arthritis (RA) patients, 207 first-degree relatives, and 158 North American controls.

with 77% being positive for both. Of the first-degree relatives, 8.5% were anti-CCP positive and 33.5% were RF positive, but only 3.3% were positive for both. In contrast, 4% and 8.5% of the North American Native controls were positive for anti-CCP and RF, respectively. In comparing first-degree relatives belonging to small families of 1–3 members (n ⫽ 151) and those belonging to large families of ⱖ4 members (n ⫽ 122), we found no significant differences in the prevalence of anti-CCP (10% in small families versus 7% in large families, P not significant) or the prevalence of RF (30% in small families versus 38% in large families, P not significant), indicating that the increased prevalence of RA autoantibodies in the first-degree relatives compared to North American Native controls was not skewed by seropositive individuals originating from a handful of large families. Furthermore, there were no differences in the prevalence of RA autoantibodies in first-degree relatives from families in which the RA proband was included in the study and those from families in which the RA proband was not included in the study (data not shown). Table 2 shows a comparison of the mean ⫾ SD of the log-transformed cytokine data in RA patients, first-degree relatives, and North American Native controls. Levels of IL-3 were excluded because they were undetectable in most of the samples, while levels of platelet-derived growth factor AB/BB were excluded because they were above the detection range in most of the samples. Of the 40 remaining cytokines included in the multiplex analysis, 36 (90%) were significantly higher in RA patients compared to first-degree relatives and North American Native controls, with P ⬍ 0.001

after Bonferroni correction for multiple comparisons in most cases. As shown in Table 2, the levels of 17 (43%) of the cytokines were significantly higher in first-degree relatives compared to the North American Native controls, including the proinflammatory cytokines/ chemokines interleukin-1␤ (IL-1␤), IL-6, tumor necrosis factor ␣ (TNF␣), IL-12, IL-8, MCP-1, macrophage inflammatory protein 1␣ (MIP-1␣), and vascular endothelial growth factor. In contrast, 3 (8%) of the 40 MALBA analytes tested were higher in North American Native controls compared to first-degree relatives. These were interferon-␥–inducible 10-kd protein, growth-related oncogene ␣ (GRO␣; CXCL1), and CD40L. Of note, the cytokine levels seen in the Caucasian controls were, for the most part, similar to those seen in the North American Native controls, and where there were differences, these were modest (data not shown). We used the discriminant analysis model to classify the entire North American Native study population on the basis of their cytokine values. Using this model, 2 major discriminant functions were generated based on the variances seen in the full panel of cytokines. These 2 functions, each of which was based on contributions from the cytokines, accounted for 82% of the total variance seen in the data set. These 2 discriminant functions were then assigned to the x-axis and y-axis of a scatterplot, and the values of the discriminant functions for each study subject were plotted individually in this matrix, as shown in Figure 1. The RA patients, firstdegree relatives, and North American Native controls were readily discriminated from each other based on these 2 functions, with function 1 discriminating most effectively between the first-degree relatives and the

1724

EL-GABALAWY ET AL

Table 2. Comparison of log-transformed multiplex cytokine levels between RA patients, first-degree relatives, and North American Native controls*

General activation IL-1␤ IL-1Ra sIL-2R␣ TNF␣ IL-6 IL-2 IL-15 IFN␣ IL-1␣ Th1 related IL-12p40 IL-12p70 IFN␥ Th2 related IL-4 IL-5 IL-9 IL-13 Eotaxin Th17 related IL-17 Treg cell related IL-10 Bone marrow derived IL-7 GM-CSF G-CSF Stromal, angiogenic Basic FGF-2 PDGF-AA VEGF EGF Chemokines IL-8 IP-10 MCP-1 MIP-1␣ MIP-1␤ GRO␣ MCP-3 MDC RANTES Others Flt-3 ligand Fractalkine CD40L TGF␣ TNF␤

RA patients

First-degree relatives

North American Native controls

4.4 ⫾ 2.6† 4.8 ⫾ 2.6† 4.8 ⫾ 2.4† 2.6 ⫾ 1.7† 2.8 ⫾ 1.7† 4.4 ⫾ 2.5† 4.8 ⫾ 2.4† 4.4 ⫾ 2.1† 3.0 ⫾ 1.8†

1.3 ⫾ 1.6‡ 2.2 ⫾ 1.5 2.3 ⫾ 1.5§ 1.1 ⫾ 0.9§ 0.8 ⫾ 1.0§ 1.4 ⫾ 1.7§ 1.8 ⫾ 1.6 2.2 ⫾ 1.7 1.0 ⫾ 1.3

0.9 ⫾ 1.3 1.8 ⫾ 1.6 1.6 ⫾ 1.4 0.7 ⫾ 0.8 0.3 ⫾ 0.6 0.8 ⫾ 1.1 1.2 ⫾ 1.4 1.6 ⫾ 1.6 0.7 ⫾ 1.0

6.0 ⫾ 2.5† 3.7 ⫾ 2.7† 2.7 ⫾ 1.6†

2.7 ⫾ 1.8§ 0.8 ⫾ 1.6 1.2 ⫾ 1.3

1.8 ⫾ 1.7 0.7 ⫾ 1.0 1.0 ⫾ 1.1

3.9 ⫾ 2.4† 0.6 ⫾ 1.1† 1.9 ⫾ 2.3† 2.0 ⫾ 1.8† 4.6 ⫾ 1.3

0.9 ⫾ 1.4 0.1 ⫾ 0.5 0.2 ⫾ 1.0 0.5 ⫾ 1.1‡ 3.5 ⫾ 1.4

0.6 ⫾ 1.0 0.1 ⫾ 0.2 0.1 ⫾ 0.4 0.2 ⫾ 0.6 3.3 ⫾ 1.1

1.6 ⫾ 1.5†

0.7 ⫾ 1.0

0.5 ⫾ 0.9

2.3 ⫾ 1.9†

0.8 ⫾ 1.0

0.7 ⫾ 0.9

2.4 ⫾ 1.3† 4.7 ⫾ 2.3† 3.0 ⫾ 1.4†

1.4 ⫾ 0.9§ 2.3 ⫾ 1.7§ 2.1 ⫾ 0.7

0.6 ⫾ 0.7 1.7 ⫾ 1.4 2.0 ⫾ 0.8

4.9 ⫾ 1.4† 7.8 ⫾ 2.6† 3.9 ⫾ 1.3† 2.3 ⫾ 0.2†

3.1 ⫾ 1.2 7.3 ⫾ 1.8 2.8 ⫾ 1.4§ 2.2 ⫾ 1.7

3.0 ⫾ 1.1 6.0 ⫾ 1.2 1.8 ⫾ 1.5 2.4 ⫾ 2.0

2.5 ⫾ 1.0† 5.9 ⫾ 1.4† 6.0 ⫾ 0.8§ 3.3 ⫾ 1.9† 5.8 ⫾ 1.8† 6.8 ⫾ 1.3† 4.5 ⫾ 1.6† 6.8 ⫾ 0.9 6.8 ⫾ 0.9

2.1 ⫾0.8§ 4.5 ⫾ 0.8 6.0 ⫾ 0.6§ 1.6 ⫾ 1.8§ 3.7 ⫾ 1.4 6.2 ⫾ 0.7 3.0 ⫾ 1.1‡ 6.9 ⫾ 0.7 6.7 ⫾ 0.5

1.5 ⫾ 1.0 4.9 ⫾ 0.7¶ 5.4 ⫾ 0.5 0.9 ⫾ 1.5 3.4 ⫾ 1.4 6.4 ⫾ 0.9¶ 2.6 ⫾ 1.2 7.1 ⫾ 0.7 6.7 ⫾ 0.7

4.2 ⫾ 1.6† 5.8 ⫾ 1.8† 6.2 ⫾ 2.0# 3.4 ⫾ 1.6† 2.3 ⫾ 2.3†

3.1 ⫾ 0.9§ 3.8 ⫾ 1.5 5.3 ⫾ 2.3 2.1 ⫾ 1.0§ 1.0 ⫾ 1.4‡

2.7 ⫾ 0.8 3.8 ⫾ 1.8 6.6 ⫾ 1.9** 1.0 ⫾ 1.1 0.6 ⫾ 1.0

* Values are the mean ⫾ SD. RA ⫽ rheumatoid arthritis; IL-1␤ ⫽ interleukin-1␤; IL-1Ra ⫽ IL-1 receptor antagonist; sIL-2R␣ ⫽ soluble IL-2R␣; TNF␣ ⫽ tumor necrosis factor ␣; IFN␣ ⫽ interferon-␣; GM-CSF ⫽ granulocyte–macrophage colony-stimulating factor; G-CSF ⫽ granulocyte colony-stimulating factor; FGF-2 ⫽ fibroblast growth factor 2; PDGF-AA ⫽ platelet-derived growth factor AA; VEGF ⫽ vascular endothelial growth factor; EGF ⫽ epidermal growth factor; IP-10 ⫽ IFN␥-inducible 10-kd protein; MCP-1 ⫽ monocyte chemotactic protein 1; MIP-1␣ ⫽ macrophage inflammatory protein 1␣; GRO␣ ⫽ growth-related oncogene ␣; MDC ⫽ macrophage-derived chemokine; TGF␣ ⫽ transforming growth factor ␣. † P ⬍ 0.001 versus first-degree relatives and North American Native controls. ‡ P ⬍ 0.05 versus North American Native controls. § P ⬍ 0.001 versus North American Native controls. ¶ P ⬍ 0.05 versus first-degree relatives. # P ⬍ 0.05 versus first-degree relatives and North American Native controls. ** P ⬍ 0.001 versus first-degree relatives.


Figure 1. Discriminant analysis using cytokine levels to classify North American Native (NAN) patients with rheumatoid arthritis (RA; solid circles), disease-free first-degree relatives (shaded circles), and controls with no personal or family history of autoimmunity (open circles). Two canonical discriminant functions, function 1 and function 2, were generated from the multiplex panel of cytokines, based on their individual standardized coefficients. There is clear discrimination between the 3 groups, and the model predicts group membership with 85% accuracy. Open squares represent the group centroid.

controls. The model predicted group membership based on these discriminant functions with an overall accuracy of 85%. The correlation coefficients for the individual cytokines and the discriminant functions are available online at www.proteome.ca. Of note, when the data from the Caucasian control group was added to the discriminant analysis, the centroid for the Caucasian controls was very close to that for the North American Native controls, indicating no major ethnic differences in the absence of a family history of autoimmunity (data not shown). The RA patients had higher hsCRP levels than all of the other groups. Consistent with the data obtained in the MALBA analysis, hsCRP levels in the first-degree relatives were significantly higher than those in the North American Native controls (mean ⫾ SD 4.9 ⫾ 4.7 versus 1.6 ⫾ 1.6 ␮g/ml; P ⬍ 0.0001). A logistic regression model that incorporated cytokine, chemokine, and hsCRP levels, demographic characteristics, seropositivity for anti-CCP and RF, smoking status, and shared epitope status was used to

1725

determine the best independent discriminators between the first-degree relatives and the North American Native controls. The data shown in Table 3 indicate that the MCP-1 level as detected by the MALBA array was the strongest independent predictor of group membership and accounted for 74% of the variability, with progressively smaller contributions from other cytokines/ chemokines, hsCRP, and RF seropositivity. In contrast, demographic characteristics, anti-CCP seropositivity, smoking status, and shared epitope status did not contribute to the model. We further confirmed the differences in MCP-1 levels between the groups using a standard MCP-1 ELISA. The data demonstrated good correlation between the MCP-1 levels detected by ELISA and those detected by MALBA (Spearman’s ␳ ⫽ 0.56, P ⬍ 0.001). Moreover, there was a modest overall correlation between MCP-1 and hsCRP across all of the study groups (Spearman’s ␳ ⫽ 0.35, P ⬍ 0.001), although there was no correlation between these 2 parameters within the group of first-degree relatives (Spearman’s ␳ ⫽ 0.09, P not significant). Table 4 summarizes comparisons between the study groups with respect to serum levels of MCP-1 and hsCRP as detected by ELISA. MCP-1 and hsCRP levels were significantly higher in first-degree relatives compared to North American Native controls, and these differences were not explained by a higher prevalence of RF-positive first-degree relatives. To assess whether or not the differences between first-degree relatives and North American Native controls in hsCRP and MCP-1 levels were skewed by large families, we compared first-degree relatives from small families and those from large families to the North American Native control group. As shown in Table 4, MCP-1 and hsCRP levels were markedly higher in the first-degree relatives irrespective of whether they came

Table 3. Independent variables in a logistic regression model discriminating first-degree relatives from North American Native controls Cytokine/chemokine* MCP-1 CD40L IL-7 TGF␣ IP-10 hsCRP PDGF-AA IgM-RF ⱖ50 IU IL-8 GRO␣

Individual ␹2

Incremental correct classification, %

135 116 93 36 25 33 21 13 6 6

74 83 86 88 91 91 91 93 93 93

* hsCRP ⫽ high-sensitivity C-reactive protein; IgM-RF ⫽ IgM rheumatoid factor (see Table 2 for other definitions).

1726

EL-GABALAWY ET AL

Table 4. Comparison of MCP-1 and hsCRP levels between subsets of first-degree relatives and North American Native controls* MCP-1, pg/ml

hsCRP, ␮g/ml

536 ⫾ 421 594 ⫾ 482 NS 219 ⫾ 106 190 ⫾ 100 NS 536 ⫾ 421 219 ⫾ 106 ⬍0.0001 484 ⫾ 275 219 ⫾ 106 ⬍0.0001 599 ⫾ 545 219 ⫾ 106 ⬍0.0001 512 ⫾ 402 220 ⫾ 107 ⬍0.0001 621 ⫾ 608 238 ⫾ 263 ⬍0.0001 621 ⫾ 608 494 ⫾ 277 0.02

4.9 ⫾ 4.6 9.1 ⫾ 5.6 ⬍0.0001 1.6 ⫾ 1.6 2.2 ⫾ 2.1 0.004 4.9 ⫾ 4.6 1.6 ⫾ 1.6 ⬍0.0001 4.4 ⫾ 4.2 1.6 ⫾ 1.6 ⬍0.0001 5.6 ⫾ 5.1 1.6 ⫾ 1.6 ⬍0.0001 5.0 ⫾ 4.8 1.6 ⫾ 1.6 ⬍0.0001 4.8 ⫾ 4.5 1.6 ⫾ 1.5 ⬍0.0001 4.8 ⫾ 4.5 5.0 ⫾ 4.7 NS

All first-degree relatives (n ⫽ 273) RA patients (n ⫽ 104) P North American Native controls (n ⫽ 200) Caucasian controls (n ⫽ 150) P All first-degree relatives (n ⫽ 273) North American Native controls (n ⫽ 200) P First-degree relatives from small families (1–3 members) (n ⫽ 151) North American Native controls (n ⫽ 200) P First-degree relatives from large families (⬎3 members) (n ⫽ 122) North American Native controls (n ⫽ 200) P RF/anti-CCP–negative first-degree relatives (n ⫽ 168) RF/anti-CCP–negative North American Native controls (n ⫽ 178) P RF-positive first-degree relatives (n ⫽ 91) RF-positive North American Native controls (n ⫽ 17) P RF-positive first-degree relatives (n ⫽ 91) RF-negative first-degree relatives (n ⫽ 182) P

* Values are the mean ⫾ SD. MCP-1 ⫽ monocyte chemotactic protein 1; hsCRP ⫽ high-sensitivity C-reactive protein; NS ⫽ not significant (see Table 1 for other definitions).

from small or large families. We also performed analyses where one first-degree relative per family was randomly selected from the 105 families and then groups of the first-degree relatives that were randomly selected were compared to the North American Native controls. In all cases, the group levels were significantly higher in the first-degree relatives, although there was variability within each family. Similar data were obtained when the

levels of other cytokines/chemokines were tested relative to family size (data not shown). Since RA autoantibodies were prevalent in the first-degree relatives, we compared the level of each of the cytokines/chemokines in anti-CCP–positive (n ⫽ 14), RF-positive (n ⫽ 82), or double-positive (n ⫽ 9) firstdegree relatives to the levels detected in the remaining, autoantibody-negative first-degree relatives (n ⫽ 168).

Table 5. Cytokines demonstrating differences between subsets of rheumatoid arthritis autoantibody– positive first-degree relatives compared to autoantibody-negative first-degree relatives* Anti-CCP⫹ only (n ⫽ 14)

EGF GRO␣ IL-1␣ IL-4 IP-10 MDC MIP-1␣ RANTES

RF⫹ only (n ⫽ 82)

Anti-CCP⫹/RF⫹ (n ⫽ 9)

Mean difference

P†

Mean difference

P†

Mean difference

P†

1.4 – 1.1 1.1 – ⫺0.8 1.4 –

0.03 – 0.009 0.04 – 0.001 0.04 –

– ⫺0.6 0.5 – 0.3 – 0.7 ⫺0.3

– 0.03 0.03 – 0.02 – 0.02 0.002

1.6 – – – – – 1.7 –

0.04 – – – – – 0.03 –

* Values are the mean difference between the log-transformed levels in the anti–cyclic citrullinated peptide (anti-CCP)–positive, rheumatoid factor (RF)–positive, or anti-CCP/RF–positive first-degree relatives and the autoantibody-negative first-degree relatives. See Table 2 for other definitions. † Versus autoantibody-negative first-degree relatives after post hoc Bonferroni adjustment for multiple comparisons.


These data are shown in Table 5 and indicate modest differences in some cytokine levels after correction for multiple comparisons, although no consistent pattern of cytokine differences was detected across the autoantibodypositive groups. Moreover, there was no significant difference in hsCRP between the autoantibody-negative and -positive first-degree relatives (data not shown). DISCUSSION We studied the serum cytokine profiles of North American Native patients with RA, their first-degree relatives, and a group of North American Native and Caucasian controls with no family history of autoimmune disease. We have previously demonstrated that the Cree/Ojibway North American Native population, like several other North American Native populations, has a high prevalence of RA and multicase families and a high background frequency of predisposing HLA– DRB1 shared epitope alleles (6,9,20,22–24). In the present study, our multiplex analyses of 42 proinflammatory and antiinflammatory cytokines, chemokines, and growth factors indicated that, as a group, the serum cytokine profile of the disease-free first-degree relatives more closely resembled that of the RA patients than did the cytokine profiles of North American Native and Caucasian controls. Patterns of circulating cytokines have been systematically analyzed in retrospective studies of the preclinical stages of RA in several distinct populations (16,17,25). A study of a Swedish population compared the preclinical plasma samples of individuals who subsequently developed RA to samples obtained after disease onset and to samples obtained from matched population controls at the time of blood donation (17). That study showed that multiple cytokines, monokines, chemokines, and growth factors discriminated between the preclinical RA samples and the samples from the matched controls, including IL-1␤, IL-6, IL-1Ra, TNF␣, MCP-1, MIP-1␣, interferon-␥, IL-12, IL-4, eotaxin, IL-10, IL-7, granulocyte–macrophage colony-stimulating factor (GM-CSF), and granulocyte colony-stimulating factor. The cytokine elevations were particularly prominent in autoantibody-positive individuals, and in most cases, the levels increased further with disease onset. A similar study from Norway comparing serum cytokine levels in preclinical RA samples to those of matched controls also demonstrated elevated levels of multiple monokines (25). Consistent with the data obtained in European populations, a study of the stored samples from 73 US military personnel who subsequently developed RA

1727

demonstrated preclinical increases in levels of IL-1␣, IL-1␤, IL-6, IL-10, IL-12, IL-15, TNF␣, GM-CSF, and CRP (16). These increases were most evident immediately prior to disease onset. Taken together, the findings of those previous studies provide compelling evidence of broad-based preclinical immune activation that progressively culminates in clinically detectable disease, particularly in individuals who are positive for ACPA and RF. An unanswered question is whether the elevations in cytokine levels arise as a result of the break in immune tolerance or whether these processes evolve independently, albeit in parallel. As in the previous studies of preclinical RA samples, the present study used a multiplex methodology to evaluate the levels of a broad array of cytokines, chemokines, and growth factors in RA patients, their first-degree relatives, and controls with no family history of autoimmunity. Interestingly, the broad-based elevations in cytokine levels seen in the first-degree relatives closely paralleled the elevations demonstrated in the previous studies of preclinical RA samples (16,17,25). Indeed, the levels of almost half of the cytokines tested, as well as the hsCRP levels, were significantly higher in the first-degree relatives compared to controls. In contrast, the difference in the cytokine profile between the North American Native controls and Caucasian controls was comparatively quite modest, suggesting that ethnicity did not play a major role in determining these profiles. The findings of the present study indicated that the autoantibody-positive first-degree relatives, whether they were anti-CCP positive, RF positive, or double positive, could not consistently be differentiated from their autoantibody-negative first-degree relative counterparts based on cytokine or hsCRP levels. As shown in Table 5, there were modest differences in the serum levels of a handful of cytokines in each autoantibody group compared to negative first-degree relatives, and perhaps with larger numbers of positive individuals, stronger trends would emerge. Nevertheless, these data provide evidence against the hypothesis that the development of RA autoantibodies in asymptomatic individuals is in itself associated with deviations in the circulating cytokine profile. Longitudinal observation of this population will clarify how the cytokine profile evolves in seropositive individuals who develop RA compared to those who do not. Interestingly, although elevation in hsCRP level was a predictor of future RA development in a US military population (16), it was not a predictor of future RA in the Women’s Health Study (26) or the Nurses’ Health Study (27). We noted that serum MCP-1 levels were par-

1728

ticularly effective in discriminating between the firstdegree relatives and controls, and indeed were comparable in the first-degree relatives and RA patients. This chemokine has been shown to play a key role in the pathogenesis of a spectrum of chronic immunemediated inflammatory disorders, including atherosclerosis, glomerulonephritis, and inflammatory bowel disease, as well as RA. Some (17,18), but not all (16), of the available preclinical RA cytokine studies demonstrated elevated MCP-1 levels in the preclinical RA samples compared to controls. Moreover, there was a further increase in MCP-1 levels with disease onset (17). The elevated MCP-1 levels detected in the North American Native patients with RA compared to controls in this study are consistent with these observations. It is more difficult to explain why the first-degree relatives had levels comparable to those of the RA patients, which were almost twice those detected in both control groups. Interestingly, hsCRP levels were also substantially higher in this population of first-degree relatives. Although there was a modest correlation between hsCRP and MCP-1 levels across all of the groups we studied, there was no correlation between these proinflammatory biomarkers within the group of firstdegree relatives. MCP-1 clustering in RA families could be based on genetic factors since polymorphisms in the CCL2 (MCP-1) promoter region are known to be associated with increased serum levels of the protein, as well as an increased risk of cardiovascular events such as myocardial infarction (28). Thus, if such polymorphisms were present in the families in this study, it is possible that they increased the risk of developing several chronic inflammatory disorders, such as RA and atherosclerosis, in parallel. Such a risk could be further increased by the presence of elevated hsCRP levels, possibly related to obesity, smoking, periodontal disease, or other environmental factors. To address this question further, studies of CCL2 polymorphisms, along with measures of obesity, are ongoing in this high-risk population. There are a number of important technical and analytical considerations that are relevant to the interpretation of the data presented in this study, and in other studies that have used multiplexed cytokine methodologies to compare various groups of individuals. In the present study, we demonstrated a good correlation between MCP-1 levels detected by MALBA and those detected by ELISA. This may not be the case for all of the cytokines and chemokines tested, particularly those present at low levels (29,30). Furthermore, it has recently been pointed out that the presence of RF can be a potential confounder in the analysis of cytokine data

EL-GABALAWY ET AL

sets by MALBA (31). Although we did not address this question directly in the present study, our laboratory has found that commercial blocking agents used to address this potential problem rarely change the levels of any of the analytes by ⬎5%, which was the threshold used in our quality controls. Importantly, the data presented in this study clearly demonstrate that the differences in MCP-1 levels between first-degree relatives and North American Native controls are not based on a higher prevalence of RF in the first-degree relatives. In summary, we have studied the serum cytokine profile of a North American Native population at high risk of RA and shown that there is an unexpected clustering of the cytokine profile in the disease-free family members of RA patients compared to North American Native and Caucasian controls with no family history of autoimmune disease. MCP-1 levels were particularly prominent in this clustering, and effectively discriminated the first-degree relatives of the RA patients from controls. These findings also suggest that the circulating cytokine profile may be an independent risk factor for RA development, although longitudinal followup of this high-risk population is needed to address this question directly. Moreover, it remains to be determined whether these findings can be generalized to other at-risk populations where familial clustering of RA is less frequent. ACKNOWLEDGMENTS The authors wish to thank the Assembly of Manitoba Chiefs and the Chiefs and Band Councils of the Norway House and St. Theresa Point communities for their role in facilitating this study. We also wish to thank Dr. Janet Markland and her staff at the University of Saskatchewan, Saskatoon, Saskatchewan, Canada, for their assistance in recruiting study subjects. AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. El-Gabalawy had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design. El-Gabalawy, Elias, Peschken, Hitchon, Li, Bernstein, Newkirk, Fritzler. Acquisition of data. El-Gabalawy, Robinson, Smolik, Hart, Wong, Peschken, Bernstein, Newkirk, Fritzler. Analysis and interpretation of data. El-Gabalawy, Hitchon, Li, Bernstein, Newkirk, Fritzler.

REFERENCES 1. Del Junco D, Luthra HS, Annegers JF, Worthington JW, Kurland LT. The familial aggregation of rheumatoid arthritis and its relationship to the HLA-DR4 association. Am J Epidemiol 1984; 119:813–29.


2. Hirsch R, Lin JP, Scott WW Jr, Ma LD, Pillemer SR, Kastner DL, et al. Rheumatoid arthritis in the Pima Indians: the intersection of epidemiologic, demographic, and genealogic data. Arthritis Rheum 1998;41:1464–9. 3. Jones MA, Silman AJ, Whiting S, Barrett EM, Symmons DP. Occurrence of rheumatoid arthritis is not increased in the first degree relatives of a population based inception cohort of inflammatory polyarthritis. Ann Rheum Dis 1996;55:89–93. 4. Koumantaki Y, Giziaki E, Linos A, Kontomerkos A, Kaklamanis P, Vaiopoulos G, et al. Family history as a risk factor for rheumatoid arthritis: a case-control study. J Rheumatol 1997;24: 1522–6. 5. Kwoh CK, Venglish C, Lynn AH, Whitley DM, Young E, Chakravarti A. Age, sex, and the familial risk of rheumatoid arthritis. Am J Epidemiol 1996;144:15–24. 6. Oen K, Robinson DB, Nickerson P, Katz SJ, Cheang M, Peschken CA, et al. Familial seropositive rheumatoid arthritis in North American Native families: effects of shared epitope and cytokine genotypes. J Rheumatol 2005;32:983–91. 7. Templin DW, Boyer GS, Lanier AP, Nelson JL, Barrington RA, Hansen JA, et al. Rheumatoid arthritis in Tlingit Indians: clinical characterization and HLA associations. J Rheumatol 1994;21: 1238–44. 8. Barnabe C, Elias B, Bartlett J, Roos L, Peschken C. Arthritis in Aboriginal Manitobans: evidence for a high burden of disease. J Rheumatol 2008;35:1145–50. 9. Ioan-Facsinay A, Willemze A, Robinson DB, Peschken CA, Markland J, van der Woude D, et al. Marked differences in fine specificity and isotype usage of the anti–citrullinated protein antibody in health and disease. Arthritis Rheum 2008;58:3000–8. 10. Majka DS, Deane KD, Parrish LA, Lazar AA, Baron AE, Walker CW, et al. Duration of preclinical rheumatoid arthritis-related autoantibody positivity increases in subjects with older age at time of disease diagnosis. Ann Rheum Dis 2008;67:801–7. 11. Nielen MM, van Schaardenburg D, Reesink HW, van de Stadt RJ, van der Horst-Bruinsma IE, de Koning MH, et al. Specific autoantibodies precede the symptoms of rheumatoid arthritis: a study of serial measurements in blood donors. Arthritis Rheum 2004;50:380–6. 12. Rantapaa-Dahlqvist S, de Jong BA, Berglin E, Hallmans G, Wadell G, Stenlund H, et al. Antibodies against cyclic citrullinated peptide and IgA rheumatoid factor predict the development of rheumatoid arthritis. Arthritis Rheum 2003;48:2741–9. 13. Zhou X, Fragala MS, McElhaney JE, Kuchel GA. Conceptual and methodological issues relevant to cytokine and inflammatory marker measurements in clinical research. Curr Opin Clin Nutr Metab Care 2010;13:541–7. 14. Codullo V, Baldwin HM, Singh MD, Fraser AR, Wilson C, Gilmour A, et al. An investigation of the inflammatory cytokine and chemokine network in systemic sclerosis. Ann Rheum Dis 2011;70:1115–21. 15. Lu LD, Stump KL, Seavey MM. Novel method of monitoring trace cytokines and activated STAT molecules in the paws of arthritic mice using multiplex bead technology. BMC Immunol 2010;11:55. 16. Deane KD, O’Donnell CI, Hueber W, Majka DS, Lazar AA, Derber LA, et al. The number of elevated cytokines and chemokines in preclinical seropositive rheumatoid arthritis predicts time to diagnosis in an age-dependent manner. Arthritis Rheum 2010; 62:3161–72. 17. Kokkonen H, Soderstrom I, Rocklov J, Hallmans G, Lejon K, Rantapaa Dahlqvist S. Up-regulation of cytokines and chemokines predates the onset of rheumatoid arthritis. Arthritis Rheum 2010;62:383–91.

1729

18. Rantapaa-Dahlqvist S, Boman K, Tarkowski A, Hallmans G. Up regulation of monocyte chemoattractant protein-1 expression in anti-citrulline antibody and immunoglobulin M rheumatoid factor positive subjects precedes onset of inflammatory response and development of overt rheumatoid arthritis. Ann Rheum Dis 2006;66:121–3. 19. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24. 20. El-Gabalawy HS, Robinson DB, Hart D, Elias B, Markland J, Peschken CA, et al. Immunogenetic risks of anti-cyclical citrullinated peptide antibodies in a North American Native population with rheumatoid arthritis and their first-degree relatives. J Rheumatol 2009;36:1130–5. 21. Khondoker MR, Bachmann TT, Mewissen M, Dickinson P, Dobrzelecki B, Campbell CJ, et al. Multi-factorial analysis of class prediction error: estimating optimal number of biomarkers for various classification rules. J Bioinform Comput Biol 2010;8: 945–65. 22. Hitchon CA, Chandad F, Ferucci ED, Willemze A, Ioan-Facsinay A, van der Woude D, et al. Antibodies to porphyromonas gingivalis are associated with anticitrullinated protein antibodies in patients with rheumatoid arthritis and their relatives. J Rheumatol 2010;37:1105–12. 23. Peschken CA, Hitchon CA, Robinson DB, Smolik I, Barnabe CR, Prematilake S, et al. Rheumatoid arthritis in a North American Native population: longitudinal followup and comparison with a white population. J Rheumatol 2010;37:1589–95. 24. Willemze A, Ioan-Facsinay A, El-Gabalawy H. Anti-citrullinated protein antibody response associated with synovial immune deposits in a patient with suspected early rheumatoid arthritis. J Rheumatol 2008;35:2282–4. 25. Jorgensen KT, Wiik A, Pedersen M, Hedegaard CJ, Vestergaard BF, Gislefoss RE, et al. Cytokines, autoantibodies and viral antibodies in premorbid and postdiagnostic sera from patients with rheumatoid arthritis: case-control study nested in a cohort of Norwegian blood donors. Ann Rheum Dis 2008;67:860–6. 26. Shadick NA, Cook NR, Karlson EW, Ridker PM, Maher NE, Manson JE, et al. C-reactive protein in the prediction of rheumatoid arthritis in women. Arch Intern Med 2006;166:2490–4. 27. Karlson EW, Chibnik LB, Tworoger SS, Lee IM, Buring JE, Shadick NA, et al. Biomarkers of inflammation and development of rheumatoid arthritis in women from two prospective cohort studies. Arthritis Rheum 2009;60:641–52. 28. McDermott DH. CCL2 polymorphisms are associated with serum monocyte chemoattractant protein-1 levels and myocardial infarction in the Framingham Heart Study. Circulation 2005;112: 1113–20. 29. Bomert M, Kollisch G, Roponen M, Lauener R, Renz H, Pfefferle PI, et al. Analytical performance of a multiplexed, bead-based cytokine detection system in small volume samples. Clin Chem Lab Med 2011;49:1691–3. 30. Dossus L, Becker S, Achaintre D, Kaaks R, Rinaldi S. Validity of multiplex-based assays for cytokine measurements in serum and plasma from “non-diseased” subjects: comparison with ELISA. J Immunol Methods 2009;350:125–32. 31. Todd DJ, Knowlton N, Amato M, Frank MB, Schur PH, Izmailova ES, et al. Erroneous augmentation of multiplex assay measurements in patients with rheumatoid arthritis due to heterophilic binding by serum rheumatoid factor. Arthritis Rheum 2011;63: 894–903.