To describe a large, multicenter prospective cohort study of first-degree relatives (FDRs) of probands with rheumatoid arthritis (RA), and outline the use of such a study in investigating the natural history of RA development.
To describe a large, multicenter prospective cohort study of first-degree relatives (FDRs) of probands with rheumatoid arthritis (RA), and outline the use of such a study in investigating the natural history of RA development.
A total of 1,058 FDRs, none of whom met the American College of Rheumatology criteria for RA, were enrolled in a prospective study investigating genetic and environmental influences on the development of RA-related autoimmunity. Demographic, epidemiologic, genetic, autoantibody, and physical examination data from the initial study enrollment visit were described for these FDRs, and the relationship was examined between genetic factors, autoantibodies, inflammation, and joint disease.
Fifty-five percent of the FDRs had ≥1 copy of the shared epitope, 20% had ≥1 copy of the PTPN22 polymorphism, and ∼16% were positive for rheumatoid factor (RF; including isotypes) and/or anti–cyclic citrullinated peptide antibody. IgM-RF positivity is associated with ≥1 tender joint on examination (odds ratio [OR] 2.50, 95% confidence interval [95% CI] 1.27–4.89; P < 0.01) and elevated C-reactive protein (CRP) levels (OR 5.31, 95% CI 1.45–19.52; P = 0.01).
FDRs without RA demonstrate high prevalences of genetic risk factors and RA-related autoantibodies. Additionally, an RF association with tender joints and elevated CRP levels suggests that autoantibodies are a valid intermediate marker of RA-related autoimmunity in this cohort. This prospective FDR cohort will be a valuable resource for evaluating the relationship between genetic and epidemiologic factors and the development of RA-related autoimmunity.
Rheumatoid arthritis (RA) is a systemic inflammatory disease of unknown etiology characterized by destructive joint disease, significant morbidity, and increased mortality (1). There are numerous genetic factors associated with increased risk for RA, including HLA alleles containing the shared epitope, the PTPN22 polymorphism, and others (2–6). These genetic factors are thought to predispose individuals to the development of autoimmunity likely due to genetic and environmental interactions (Figure 1). Epidemiologic studies have identified several potential risk factors for disease, including cigarette smoke, hormones, and infections, but it is not known at what point during disease evolution these factors are important (7–13). However, elevated autoantibodies, cytokines, and inflammatory markers occur years prior to clinical onset of joint symptoms, suggesting that epidemiologic factors may play an early role in the development of RA (14–18). As such, the investigation of subjects during the preclinical stage is important to furthering our understanding of disease pathogenesis.
In this study, we used the first-degree relative (FDR) cohort of the Studies of the Etiology of Rheumatoid Arthritis. The basis for our study design comes from preclinical studies of type 1 diabetes mellitus (DM) at the University of Colorado and elsewhere (19–22). By utilizing autoantibodies as surrogate markers of disease-related autoimmunity, these type 1 DM studies have identified potential environmental exposures contributing to disease pathogenesis and evolution (19, 20). We have created a prospective cohort using a model similar to that of type 1 DM studies, with the intent to investigate the natural history of RA development. Our cohort consists of FDRs of probands with RA, and was designed to examine the role of genetic and environmental factors in the development of RA-related autoimmunity and to explore preclinical immunologic changes within this population. FDRs will be a valuable resource to the research community for following the natural history of preclinical RA because they 1) are enriched with genetic and possibly environmental risk factors for RA, 2) have a higher prevalence of autoantibodies, and 3) have an increased risk for the development of RA (23–31). Our discussion herein will provide a detailed description of 1,058 FDRs from this unique cohort that have been enrolled as of September 2008.
Probands with RA were identified from academic centers, veterans' hospitals, and private and public sector rheumatology clinics at sites based in New York, Chicago, Omaha (the center of the Rheumatoid Arthritis Investigational Network), Denver, Seattle, and Los Angeles. For inclusion, probands must meet ≥4 American College of Rheumatology (ACR; formerly the American Rheumatism Association) classification criteria for RA based on clinical and laboratory findings present on chart review, or have a diagnosis of RA from a board-certified rheumatologist (32). The latter definition was included to prevent exclusion of patients who no longer fulfill the ACR criteria due to well-controlled disease. Additionally, probands must be diagnosed with RA after age 16 years to avoid inclusion of those with juvenile inflammatory arthritis, which likely has different genetic factors and autoantibody profiles than adult RA. Probands were excluded if no FDRs were alive or interested in participation.
For this study, the definition of FDR included the parent, full sibling, or offspring of a proband. FDRs were recruited through their probands or through responses to advertising. An FDR was eligible to participate in the study if they did not have a diagnosis of RA and were age ≥18 years. If an FDR self-reported a diagnosis of RA, their medical record was reviewed, and they were excluded if ≥4 ACR classification criteria were met or if they were diagnosed with RA by a board-certified rheumatologist. Once enrolled, FDRs came to a study site to undergo the initial clinical research visit. At the initial visit, 2 FDRs were found to have previously undiagnosed RA after evaluation by a study physician and were excluded from entry.
Research protocols and consent process for this study were approved by the Institutional Review Boards at each participating site. In addition, an observational study monitoring board was created to provide oversight for this study. The board consisted of an individual with RA, an FDR, a geneticist, an ethicist, and a community-based rheumatologist to assist with ongoing discussion regarding the ethics of obtaining and reporting genetic and autoantibody data.
Each proband was evaluated at a single clinical research visit where the following data were obtained: 1) demographic information, 2) a detailed history of RA (including treatment and severity of disease), and 3) blood and urine collection (see below). Probands with verified RA who declined such an evaluation were still permitted to contribute FDRs to the cohort.
Once enrolled, FDRs were evaluated in a clinical research visit and the following data were obtained: 1) demographic information, 2) a medical history, including prior diagnoses of autoimmune or infectious diseases and current medications and supplements, 3) epidemiologic questionnaires with assessment of hormonal and environmental exposures, 4) the Connective Tissue Disease Screening Questionnaire, a 30-item questionnaire that can assess for RA or other connective tissue diseases (33), 5) a standardized interview and 68-count joint examination by a trained study physician or nurse, and 6) blood and urine collection for testing for genetic factors, autoantibodies, inflammatory markers, nutritional factors, measurements of oxidative stress, and assessment of other biomarkers. Additionally, samples were stored for future studies. FDRs that could not come to a study site were evaluated with mailed questionnaires, joint symptoms were ascertained via a phone interview, and blood samples were collected at local laboratories.
All FDRs were invited for longitudinal followup; FDRs that were positive for any RA-related autoantibodies were seen annually, and autoantibody-negative FDRs were seen every other year. At these followup visits, FDRs completed interval-assessment questionnaires, underwent joint interview and examination, and had blood drawn for studies as indicated above. Additionally, FDRs were instructed to contact study personnel if they developed signs/symptoms of RA in the intervening periods. FDRs with an abnormal joint evaluation at their initial or followup visit returned 6 weeks later for hand and wrist radiographs, as well as a repeat interview, examination, and blood draw.
Testing for RA-related autoantibodies was performed at the University of Colorado Division of Rheumatology Clinical Research Laboratory (Clinical Laboratory Improvement Amendments certified). Testing was performed for the rheumatoid factor (RF) isotypes IgM, IgG, and IgA by enzyme-linked immunosorbent assays (ELISAs) using QUANTA Lite kits (Inova Diagnostics, San Diego, CA), and results are reported in units/ml. RF was also measured by nephelometry according to the manufacturer's specifications (Dade Behring, Newark, DE). A positive RF (ELISA isotypes or by nephelometry) was defined as the level present in ≤5% of healthy controls according to the ACR RA criteria (32). Cutoffs for RF positivity have been established using 490 randomly selected healthy blood donors from the Denver area.
Antibodies against citrullinated peptides were tested by ELISA using the anti–cyclic citrullinated peptide 2 (anti-CCP2) kit (Diastat; Axis-Shield, Dundee, UK). Per the manufacturer's specifications, a positive test was defined as ≥5 units/ml. For all of the autoantibody assays, 5% of the antibody-negative samples as well as all of the positive results were retested and confirmed by blinded duplicate analysis.
Genetic testing in FDRs to date has been limited to the shared epitope and PTPN22 polymorphism, performed at the Benaroya Research Institute at Virginia Mason in Seattle, Washington, although DNA and RNA are stored for future analyses. All of the probands were tested for the shared epitope as well; however, PTPN22 testing has been performed only in a limited number of subjects during early enrollment (n = 78), as well as ongoing evaluation at the Seattle site (n = 133). Complete subtyping for HLA–DR4 alleles was done via a modification of a real-time polymerase chain reaction (PCR) approach, as described previously (34). In addition to the primers and probes described, one additional probe was added to allow resolution of DRB1*0403 and *0406, allowing for identification of the major DRB1 polymorphisms and accurate resolution of DRB1*0401 to *0421. DR4 subtypes that were considered shared epitope positive include DRB1*0401, 0404, 0405, 0408, 0409, 0410, 0413, 0416, 0419, and 0421. A real-time low-resolution PCR analysis was also performed to identify the presence of shared epitope–containing DR1 alleles, including DR1*0101, 0102, 0104, 0105, 0107, 0108, and 0111.
The single nucleotide polymorphism within the PTPN22 gene (1858C→T) was tested for with a fluorescently labeled minor groove binder Eclipse probe (Nanogen, Bothell, WA), with a followup PCR using the Applied Biosystems 7900HT (Foster City, CA).
High-sensitivity CRP testing was performed by nephelometry by the Clinical Translational Research Center at The Children's Hospital in Denver, Colorado. CRP level results were divided into the following categories: 0 to ≤3 mg/liter, >3 to ≤10 mg/liter, >10 to ≤20 mg/liter, and >20 mg/liter.
A cross-sectional analysis was performed using data obtained at the FDR initial visit (n = 1,058) from July 2003 through September 2008. The relationship between autoantibodies and 1) genetic factors (shared epitope and PTPN22), 2) joint findings, and 3) CRP level were evaluated using a chi-square test of proportions. Sparse data categories were addressed with Fisher's exact test. For analysis with outcome variables consisting of multiple categories such as the CRP level analysis, logistic regression was performed using a reference cell design model (35).
In addition to assessing the relationship between single autoantibodies and various factors listed above, several combinations of antibodies were also evaluated: any positive antibody and ≥2 positive RF isotypes and/or positive anti-CCP antibody. The latter grouping was decided on because prior research has suggested that positivity for ≥2 RF assays and/or anti-CCP antibodies is highly specific for RA, and may be predictive of future disease (14, 17, 36–38).
For analysis of the association between autoantibodies and joint disease on examination, our definition of joint involvement did not include isolated knee tenderness or swelling, given the potential for a noninflammatory etiology. These subjects were not considered to have an inflammatory arthritis, and were retained in the analysis. All statistical analyses were performed using SAS, version 9.2 (SAS Institute, Cary, NC).
A total of 674 probands with RA are included in the current description, with a mean age at diagnosis of 45 years. Based on laboratory testing done at their enrollment visit, 69.6% were positive for at least 1 RF isotype (IgM, IgG, IgA), 66.5% were positive for anti-CCP, and 60.2% were positive for anti-CCP and RF (by any assay). A significant association exists between the shared epitope and anti-CCP–positive RA (odds ratio 3.84, 95% confidence interval 2.75–5.35; P = 0.0001), replicating the findings of previous studies (39). Additional information on the probands can be found in Table 1.
|Age at diagnosis of RA, mean years||45|
|Non-Hispanic white†||559 (83)|
|Shared epitope present (n = 642 with testing performed)‡|
|≥1 allele||411 (64)|
|1 allele||296 (46)|
|2 alleles||115 (18)|
|PTPN22 polymorphism present (n = 211 with testing performed)§|
|≥1 allele||54 (25.6)|
|1 allele||47 (22)|
|2 alleles||7 (3)|
|Autoantibody positivity, % total|
|Anti-CCP and any RF assay||406 (60.2)|
|DMARD therapy (including prednisone)||650 (96.4)|
|Erosive disease at the time of diagnosis of RA (n = 509 with report of radiographs at diagnosis)||253 (49.7)|
|Mean HAQ score at time of enrollment||0.79|
A total of 1,058 FDRs were included in this analysis, with a mean ± SD age of 47 ± 16 years. Seventy-one percent are women, and 81.7% are non-Hispanic white. Forty-two percent reported ever smoking, with 9% reporting ≥20 pack-years of use. Of the women, 82.2% reported ever use of oral contraceptives (Table 2).
|Los Angeles||248 (23.4)|
|New York||75 (7.1)|
|Age, mean ± SD years||47 ± 16|
|Non-Hispanic white‡||864 (81.7)|
|Income level per year|
|Missing data||84 (7.9)|
|High school||182 (17.4)|
|Missing data||28 (2.6)|
|0.5–19.5 pack-years||349 (33)|
|≥20 pack-years||97 (9)|
|Oral contraceptive, ever use (among female FDRs)||619 (82.2)|
|No. of FDRs per family (n = 696 families), mean||1.5|
Of the 1,058 FDRs, 813 underwent a joint examination: 14.8% of these FDRs had ≥1 swollen joint and 25% had ≥1 tender joint (Table 3). The remaining subjects were evaluated offsite and therefore had no examination.
|FDR characteristics||No. (%)|
|Shared epitope present (n = 982 with testing performed)†|
|≥1 allele||538 (55)|
|1 allele||451 (46)|
|2 alleles||87 (9)|
|PTPN22 polymorphism present (n = 980 with testing performed)|
|≥1 allele||196 (20)|
|1 allele||183 (19)|
|2 alleles||13 (1)|
|Any autoantibody||167 (15.9)|
|Anti-CCP and/or ≥2 RF assays||51 (4.9)|
|CRP level quartiles (reference range 0–3 mg/liter)|
|0 to ≤3||753 (71.2)|
|>3 to ≤10||250 (23.7)|
|>10 to ≤20||36 (3.4)|
|Swollen joint count ≥1‡||120 (14.8)|
|Excluding knees||98 (12.1)|
|Tender joint count ≥1‡||203 (25.0)|
|Excluding knees||177 (21.8)|
Fifty-five percent of the FDRs had ≥1 allele containing the shared epitope, with 9% having 2 copies. The PTPN22 polymorphism (1 or 2 copies) was present in 20% of the FDRs. In comparison, the general population prevalence of the shared epitope and the PTPN22 polymorphism (1 or 2 copies) is 43% and 16%, respectively (40).
The presence of any RA-related autoantibody (RF assays and/or anti-CCP) was present in 15.9% of the FDRs, with 1.7% positive for anti-CCP. Further analysis of FDR autoantibody positivity is shown in Table 3.
In univariate analysis, there were no significant associations between autoantibody positivity and age, sex, or race. However, in univariate analysis, IgM-RF positivity was significantly associated with joint tenderness (P = 0.0078) and CRP levels >20 mg/liter (P = 0.012). Additional results are shown in Table 4.
|Clinical parameter (outcome)||Selected definitions of RA-related autoimmunity, OR (95% CI)|
|Any autoantibody (n = 167)†||Anti-CCP (n = 18)||IgM-RF (n = 50)||IgG-RF (n = 65)||IgA-RF (n = 40)|
|CRP level, mg/liter|
|0 to ≤3||1.00 (ref.)||1.00||1.00||1.00||1.00|
|>3 to ≤10||1.32 (0.91–1.92)||0.93 (0.3–2.88)||1.70 (0.91–3.19)||1.24 (0.7–2.19)||1.93 (1.00–3.75)|
|>10 to ≤20||0.91 (0.35–2.4)||1.63 (0.21–12.79)||1.46 (0.33–6.37)||0.94 (0.22–4.02)||0.86 (0.11–6.56)|
|>20||1.22 (0.34–4.30)||−‡||5.31 (1.45–19.52)§||0.99 (0.13–7.67)||−‡|
|≥1 swollen joint (n = 98)¶||1.42 (0.84–2.43)||0.56 (0.07–4.30)||0.87 (0.3–2.51)||1.14 (0.5–2.60)||1.51 (0.57–4.05)|
|≥1 tender joint (n = 177)¶||1.29 (0.83–1.99)||0.59 (0.13–2.68)||2.50 (1.27–4.89)§||0.84 (0.41–1.70)||0.90 (0.36–2.24)|
No significant association between autoantibodies and the presence of the shared epitope or the PTPN22 polymorphism was present in this interim analysis (Table 5).
|RA-related autoantibodies||Shared epitope ≥1 alleles||PTPN22 ≥1 alleles|
|No.||OR (95% CI)||No.||OR (95% CI)|
|Anti-CCP||11||1.83 (0.58–6.78)||5||1.99 (0.58–2.67)|
|IgM-RF||22||0.81 (0.42–1.56)||6||0.63 (0.21–1.54)|
|IgG-RF||27||0.73 (0.41–1.29)||12||1.06 (0.5–2.09)|
|IgA-RF||22||1.07 (0.53–2.17)||9||1.29 (0.52–2.86)|
|Any autoantibody†||85||1.04 (0.72–1.49)||30||1.00 (0.63–1.57)|
Evaluation of the preclinical phase is crucial to a more thorough understanding of RA development. Epidemiologic research thus far has been largely retrospective, and therefore unable to identify the timing of the effects of etiologic factors during the prolonged phases of early disease development. In addition, many epidemiologic studies have been performed using subjects with established disease; therefore, the effects from recall bias, chronic immunologic changes associated with advanced disease, and interval treatment are unknown. The use of a preclinical cohort to address these issues has been successful in fields such as type 1 DM, and we believe that using a similar methodology in the study of RA development will result in higher quality data collection and a more complete understanding of early immunologic changes.
The preclinical period of RA can be difficult to study prospectively in general populations due to the low prevalence of disease (∼1%). In order to increase our yield for identifying individuals with preclinical RA, we have targeted a cohort with a higher risk of disease based on family history, likely due to enrichment of genetic and environmental risk factors. Due to these factors, we expect autoantibody positivity and seroconversion rates as well as the development of incident joint disease to be higher in the FDRs than in the general population, increasing the yield of study end points. In support of this, the prevalence of ≥1 allele containing the shared epitope or the PTPN22 polymorphism in the FDRs (55% and 20%, respectively) is higher than that reported in North American historic controls (43% and 16%, respectively) (40). In addition, the prevalence of anti-CCP antibody positivity in FDRs is 1.7% compared with a control prevalence of 0% among 182 Denver blood donors. Of great interest are the FDRs in this cohort with positivity for 2 or more RF isotypes and/or anti-CCP, who based on prior literature are likely to develop symptomatic RA in the future (14, 17).
We plan to continue our study of this cohort using autoantibody positivity as an intermediate outcome in the development of RA. Existing data support the relationship of preclinical RA-related autoantibodies to the development of future disease, especially in relatives of probands with RA (14, 15, 17, 23). Furthermore, the association of IgM-RF with tender joints and CRP level elevations reported here supports the biologic relevance of autoantibody positivity in subjects without overt RA. We must be careful, however, not to overlook the importance of antibody-positive subjects without signs of joint disease or inflammation. Potential genetic and environmental factors associated with antibody formation may be easier to elucidate from this population compared with retrospective cohorts. Of note, our group has already demonstrated several interesting associations with RF positivity among asymptomatic cohorts: 1) a negative association with oral contraceptive use, 2) a positive association with cigarette exposure, and 3) lack of an association with vitamin D levels (41, 42).
In addition to the epidemiologic analysis, the FDR cohort provides us the opportunity to study in detail individuals with high-risk autoantibody profiles for RA, and the potential discovery of additional biomarkers, pathogenic mechanisms, and sites of inflammation important to the development of RA. For example, in preliminary studies we have shown an association between autoantibody positivity and elevations of several cytokines in members of our cohort, as well as identifying a potential link between autoantibody positivity and asymptomatic lung disease (43, 44). As we continue to move forward, the prospective, longitudinal design of our study will prove to be its greatest asset. Data collection is ongoing, with 237 FDRs to date seen in followup visits (mean followup 2.5 years, range 7 months to 5 years). Study personnel have maintained contact with research participants, with a followup rate of 84.5% across study sites. Four FDRs have developed incident RA from this cohort in ∼600 person-years of followup (∼6.7 cases per 1,000 person-years), similar to incidence rates for RA in previously evaluated FDR cohorts (23).
Our proband source is not derived from a true population- based sample, and as such, there may be concerns as to the applicability to the general RA population. However, comparison with other cohorts in the literature reveals that our probands are similar to the overall RA population in terms of the age at diagnosis, proportion of women, prevalence of erosive disease, and rates of antibody and genetic marker positivity (6, 45–48). In addition, there is a possibility that FDRs enrolled in our cohort are healthier, on average, than the general FDR population by virtue of their participation in the study. Alternatively, it is possible that the FDRs have chosen to enter the study based on existing joint symptoms or concern for undiagnosed RA. We recognize that both of these issues represent potential participation bias, but the inability to collect information on nonenrolled family members prevents systematic comparison with our FDRs. Although this may potentially affect the external validity of our results, we believe our study design, specifically the creation of an internal control population derived from the same pool of FDRs as our cases, will provide excellent internal validity and therefore high-quality data.
Interestingly, IgM-RF positivity is associated with CRP level elevation and joint tenderness despite being less prevalent than IgG-RF. This may be due to the biologic activity of IgM-RF in its ability to form immune complexes more readily than IgG-RF. Alternatively, because IgM-RF may be less specific for RA, its association with CRP level elevation and joint tenderness may reflect another inflammatory condition not detected in our evaluation (38). Two RF-positive FDRs reported concurrent hepatitis C virus; however, neither had the IgM-RF isotype.
Our study is limited in this interim analysis by lack of power for several association studies between autoantibodies and clinical outcomes, as well as between the shared epitope or PTPN22 polymorphism. This will need further exploration when the cohort reaches full enrollment (the goal is 2,100 FDRs by 2010).
Preclinical studies are crucial to understanding the relationship between genetic and environmental factors in the development of RA. This FDR cohort is a valuable resource for such studies, allowing for high-quality prospective data collection and genetic and epidemiologic investigations using autoantibody positivity as an intermediate outcome. The association of RF positivity with abnormal joint examination and elevated CRP levels lends validity to the hypothesis that autoantibodies represent part of a continuum of immunologic changes leading to the development of symptomatic RA.
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 submitted for publication. Dr. Holers 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. Kolfenbach, Deane, Derber, O'Dell, Gregersen, Norris, Holers.
Acquisition of data. Kolfenbach, Deane, Derber, Weisman, Buckner, Gersuk, Wei, Mikuls, O'Dell, Gregerson, Keating, Norris.
Analysis and interpretation of data. Kolfenbach, Deane, Derber, O'Donnell, Weisman, Gersuk, Wei, O'Dell, Norris, Holers.
We thank the individuals who participated in the Studies of the Etiology of Rheumatoid Arthritis. We are also indebted to the coordinators of the recruitment sites: Elaine Hamburger, Marie Feser, Amber Matheny, Cynthia Marr, Monica Choi, and Marlena Kern; and those involved in sample processing: Mark Parish and Whitney Hilton.