Abstract
Objectives
Little is known about the likelihood of developing inflammatory arthritis (IA) in individuals who screen autoantibody positive (aAb+) in a non-clinical research setting.
Methods
We screened for serum cyclic citrullinated peptide antibody (anti-CCP) and rheumatoid factor (RF) isotype aAbs in subjects who were at increased risk for rheumatoid arthritis (RA) because they are a first-degree relative of an individual with classified RA (n=1780). We evaluated combinations of aAbs and high titer aAbs, as defined by 2-times (2x) the standard cut-off and an optimal cut-off, as predictors of our two outcomes, aAb+ persistence and incident IA.
Results
304 subjects (17.1%) tested aAb+; of those, 131 were IA-free and had at least one follow-up visit. Sixty-four percent of these tested aAb+ again on their next visit. Anti-CCP+ at levels ≥2x the standard cut-off was associated with 13-fold higher likelihood of aAb+ persistence. During a median of 4.4 years (IQR: 2.2-7.2), 20 subjects (15.3%) developed IA. Among subjects that screened anti-CCP+ at ≥ 2x or ≥ an optimal cut-off, 32% and 26% had developed IA within 5-years respectively. Both anti-CCP cut-offs conferred an approximate 4-fold increased risk of future IA (HR 4.09 and HR 3.95, p<0.01).
Conclusions
These findings support that aAb screening in a non-clinical setting can identify RA-related aAb+ individuals, as well as levels and combinations of aAbs that are associated with higher risk for future IA. Monitoring for the development of IA in aAb+ individuals and similar aAb testing approaches in at-risk populations may identify candidates for prevention studies in RA.
Keywords: Rheumatoid Arthritis, Synovitis, Rheumatoid Factor, Anti-CCP, Epidemiology
INTRODUCTION
Seropositive rheumatoid arthritis (RA) is characterized by immune system dysregulation prior to any signs of inflammatory arthritis (IA) and classifiable disease.1,2 The development of RA is thought to occur in several phases, with the earliest phase including genetic and environmental relationships that trigger autoimmunity and may initially occur in the absence of clinically-apparent IA. 3-7 Individuals may be identified in this pre-clinical phase by the presence of circulating autoantibodies (aAbs), specifically antibodies to citrullinated protein/peptide antigens (ACPAs) and rheumatoid factor (RF).1,2,7,8 ACPA and RF have been shown to appear an average of 3-5 years before clinically-apparent IA and classifiable RA in a period termed pre-RA.1,2 Many epidemiologic factors associated with systemic autoimmunity and RA have been identified, including older age, female sex, race/ethnicity, smoking, and lower omega-3 fatty acids. 6
Our current understanding of RA-related aAbs and the timing of IA/RA development comes largely from the study of subjects with banked blood samples prior to RA diagnosis1,2,9,10, subjects with undifferentiated arthritis who presented to the health care system in rheumatology clinics11-16, or subjects who were tested for aAbs due to a clinical indication of arthralgia.17 From these studies, the reported positive predictive value (PPV) for the future development of classified RA in the presence of aAbs has varied, depending on the specific cohort design, aAbs assessed, and analytic approach. In a study using pre-RA diagnosis blood bank samples from patients with established RA and controls, the PPV for future RA was 82% among those positive for anti-CCP and 83-87% when combined with a RF isotype.1 In a study among patients with arthralgia, the risk of developing RA was nearly 9-fold among those positive for both RF IgM and anti-CCP and 3-5-fold for those only positive for anti-CCP (depending upon anti-CCP level) compared to those RF IgM positive only.18 While informative, these studies do not represent the likelihood of developing RA in those who screen aAb+ outside of the clinical setting. Of the two studies that screened first-degree relatives (FDRs) of RA patients for aAbs, both showed the risk of developing IA/RA was highest for subjects positive for both anti-CCP and RF (64%19 and 38%20 after 5-years of follow-up), although the latter study found that aAb reversion was equally as likely.20
The ability to identify individuals with elevated aAbs in absence of clinically-apparent IA has led to an increased focus on preventing the development of future IA.21-23 Determining factors associated with persistent aAb positivity and the development of IA is a crucial component in the design of RA prevention strategies. Importantly, screening for aAbs in populations at-risk for the development of future IA outside of clinical care settings, is a means to assemble cohorts for epidemiologic, mechanistic and interventional studies. Our goal is to identify factors related to persistent aAb+ and future development of IA/RA in a prospective non-clinical population that is at-risk based on family history of disease.
METHODS
Study Population
The multicenter Studies of the Etiology of Rheumatoid Arthritis (SERA) is a prospective study designed to examine the environmental and genetic factors leading to the development of RA among those who are at increased risk for developing RA. The SERA population of FDRs of RA probands has been described previously.24
At study entry, a 68-count joint examination was performed to confirm subjects do not have RA by 1987 American College of Rheumatology (ACR) criteria.25 At each follow-up visit, a joint exam was performed to assess IA/RA status, and questionnaires capturing socioeconomic status, self-reported joint symptoms and environmental exposures, were collected. And, blood samples for biomarker and genetic studies were obtained.
As of March 2019, SERA has enrolled 1780 FDR participants. Subjects were included in current analyses based on having at least one visit where they were aAb+ for one or more of the aAbs listed below, had at least 1 additional follow-up visit after their first aAb+ visit, underwent joint examinations during the study, and did not have IA at their baseline visit (figure 1).
Figure 1.
Flow chart of study subject inclusion
Measurement of RA-Related biomarkers
Autoantibody assays were performed in the clinical and research lab at the University of Colorado.
Standard cut-offs for aAb positivity:
Anti-CCP2 (IgG) was measured in serum using ELISA kits according to manufacturer’s instructions (Diastat; Axis-Shield, Dundee, UK: cut-off >5 U/mL, reported specificity of 99.0%). Anti-CCP3.1 (IgG/IgA) was measured using ELISA kits (Inova Diagnostics, Inc., San Diego, California: cut-off ≥20 U/mL, reported specificity of 97.8%). RF isotypes IgM and IgA were measured using ELISA assays (QUANTA Lite™: IgM positive cut-off >13.6 IU/mL, IgA positive cut-off >10.5 IU/mL). Based on recommendations included in the 1987 ACR RA Classification Criteria, positivity for RF isotypes was established using a cut-off level higher than that observed in 95% of 491 randomly selected blood donor controls.25 In our analyses, we examined type of aAb+ (anti-CCP+ and RF isotype+/anti-CCP+ only/RF isotype+ only based on the standard cut-off), as well as whether the subject tested positive for the aAb at >2x the standard cut-off (anti-CCP ≥2x cut-off and RF isotype ≥2x cut-off).
Optimal cut-offs for anti-CCP positivity:
In addition to analyzing standard cut-offs, we assessed optimal cut-offs for anti-CCP2 and anti-CCP3.1 positivity with the IA outcome using methods developed by Contal and O’Quigley for time-to-event outcomes with the SAS macro %findcut.26 The optimal cut-offs were defined by the value of anti-CCP whose split is most significantly associated with incident IA using the log-rank test with a False Discovery Rate correction for multiple comparisons. The optimal cut-off for anti-CCP2 was ≥5 U/mL and for anti-CCP3.1 was ≥30 U/mL. Combining these two cut-offs we analyzed anti-CCP+ ≥ optimal cut-off (yes/no).
98th percentile cut-offs for aAb positivity:
Positivity for anti-CCP2, anti-CCP3.1, RF IgM and RF IgA was established using a cut-off level higher than that observed in 98% of 200 randomly selected blood donor controls from the Denver, CO area. Results using this cut-off are presented in the supplementary materials.
C-reactive protein (CRP):
Serum was tested for high sensitivity C-reactive protein (CRP) by nephelometric assay (BN II Nephelometer, Dade Behring, Deerfield, Illinois, USA) and dichotomized using an elevated cut-off of >3mg/L.27
Shared Epitope
The presence of shared epitope (SE) alleles, HLA-DR4 and HLA-DR1, was tested and described previously.24 A subject was considered SE positive if one or more alleles contained the following SE subtypes: DRB1*0401, *0404, *0405, *0408, *0409, *0410, and *0413; DRB1*0101, *0102; DRB1*1001.
Assessment of Risk Factors
Risk factors, which were assessed at the screened aAb+ visit, are listed in Table 1. We dichotomized the following factors based either upon small sample size or previously published associations with IA/RA6,28,29: race (non-Hispanic white (NHW)/other), SE (present/absent), >10 smoking pack-years as of the screening visit (yes/no: calculated as years of smoking multiplied by packs of cigarettes per day and dichotomized at >10 years), CRP (elevated/normal), tender joint on exam (yes/no), self-reported joint symptoms of pain, stiffness, or swelling within the past week (yes/no).
Table 1.
Characteristics of study population
| N=131 | |
|---|---|
| At Screened aAb+ Visit | |
| Age (year) : mean ± SD | 47.7 ± 15.0 |
| Sex: % female | 77.9 |
| Race: % NHW | 79.4 |
| Shared Epitope: % positive | 54.2 |
| Cigarette pack-years: % >10 | 13.7 |
| * CRP: % elevated (≥3 mg/L) | 34.6 |
| Type of CCP | |
| † CCP2: % positive (>5 U/mL) | 3.1 |
| ‡ CCP3.1: % positive (≥20 U/mL) | 41.5 |
| CCP2 and CCP3.1: % positive | 9.9 |
| Type of RF isotype+ | |
| RF IgM: % positive (>13.6 IU/mL) | 35.9 |
| RF IgA: % positive (>10.5 IU/mL) | 13.0 |
| RF IgM and RF IgA: % positive | 5.3 |
| Type of aAb+ | |
| Anti-CCP- and RF isotype+ | 45.8 |
| Anti-CCP+ and RF isotype− | 45.8 |
| Anti-CCP+ and RF isotype+ | 8.4 |
| § RF+ isotype ≥ 2x cut-off: % yes | 26.7 |
| ∣ Anti-CCP+ ≥ 2x cut-off: % yes | 27.5 |
| ¶ Anti-CCP+ ≥ optimal cut-off: %yes | 32.8 |
| Tender joint on exam: % yes | 28.8 |
| # Self-reported joint symptoms (pain, stiffness, swelling): % yes | 63.6 |
| At Follow-Up Visit | |
| Δ Years from aAb+ screening visit to next follow-up visit: median (IQR) | 1.6 (1.1-2.2) |
| Δ Follow-up aAb+ status: % Persistent | 64.1 |
| Number of visits (screening visit to IA or last visit): median (IQR) | 3 (2-5) |
| Years from aAb+ screening visit to last study visit or IA: median (IQR) | 4.4 (2.2-7.2) |
| Incident IA: % yes | 15.3 |
1 subject missing CRP
1 subject missing anti-CCP2
1 subject missing anti-CCP3.1
RF IgM or RF IgA
Anti-CCP2 or anti-CCP3.1
The optimal cut-off was calculated as ≥5 U/mL for anti-CCP2, and ≥30 U/mL for anti-CCP3.1
Subjects report presence of symptoms within the past week: 3 subjects missing data
3 subjects developed IA/RA prior to the follow-up visit and are removed from analyses assessing predictors of aAb+ persistence
Assessment of Outcomes
We classified subjects as aAb persistent, ie, those who were aAb+ (for any aAb) at the next visit and aAb non-persistent, ie, those who tested aAb- (for all aAbs) at the next visit. We defined IA as the presence of at least 1 swollen joint consistent with RA-like synovitis. This was assessed using the 68-count joint examination performed by a study rheumatologist or trained study nurse, or through medical record review if IA was identified outside of our study. All research study records and relevant medical records were reviewed by a single board-certified rheumatologist applying similar criteria to those classified from within and outside of the study.
Statistical Analysis
To examine factors associated with aAb+ persistence, we conducted logistic regression analysis for the likelihood of aAb+ persistence.
To examine what predicts progression to IA, we first created Kaplan-Meier curves to assess IA-free survival after the screened aAb+ visit by aAb classifications determined at the screened aAb+ visit. Univariable and multivariable Cox proportional hazards models were then used to determine risk factors associated with the risk of progression to IA. Years from first aAb+ visit to IA or last study visit was used as the time scale. The proportional hazards assumption was checked using methods developed by Lin et al.30 Both the additive and multiplicative interaction between SE and pack-years >10 was assessed, given associations found in prior literature.31
The predictive ability of the risk models was compared by generating time-dependent area under the curve (AUC) statistics and Uno’s concordance statistic (C-statistic) using the inverse probability of censoring weighting (IPCW) method.33 Pairwise differences in the C-statistic were made between each consecutive model to determine whether one model was a better predictor of incident IA. A p-value for the difference in Uno’s C-statistic >0.05 indicates no difference in the predictive ability of the models. The time-dependent AUC was calculated and represents the average of the AUC statistics over time. Analyses were conducted in SAS (SAS Version 9.4, Cary, North Carolina).
RESULTS
Study Population
Of the 1780 subjects that were screened for aAbs, 304 (17.1%) subjects were aAb+. Of those, 164 subjects had at least 1 follow-up research visit. Twenty-six subjects did not have an exam and were removed from analyses. Of the 138 subjects remaining, 131 were IA-free at the aAb+ screening visit (figure 1). There were no significant differences between the screened aAb+ population (n=304) and the study population (n=131) (Table S1). The study population was 78% female and 79% NHW with a mean age of 48 years at the screened aAb+ visit (table 1).
Analysis of Persistent aAb+
Three subjects developed IA/RA prior to the follow-up visit and were removed from analyses assessing predictors of aAb+ persistence. Of the 128 subjects analyzed for persistent aAb+, 82 subjects (64.1%) tested aAb+ again with a median of 2.0 (IQR: 1.1-3.1) years between visits. Most subjects (96.3 %) were positive for the same aAb at follow-up. All subjects positive for both anti-CCP and RF isotype(s) at screening remained aAb+ at the follow-up visit. In addition, subjects who tested anti-CCP+ at levels ≥2x the standard cut-off at screening were significantly more likely to remain aAb+ at the next follow-up visit compared to subjects who were negative for anti-CCP or tested positive at levels <2x the standard cut-off (OR 13.37, 95% CI 3.03 to 59.06, p=0.001). Testing positive for an RF isotype at ≥2x the standard cut off was not associated with testing positive again (OR 1.70 95% CI 0.71 to 4.06, p = 0.23).
Analysis of Incident IA
Over a median follow-up time of 4.4 years (IQR 2.2-7.2), 20 subjects (15.2%) developed IA. Of these, 12 IA subjects were identified as a result of a study visit exam, and in the remaining 8 subjects, IA was determined by a physician outside of the study, which was confirmed through medical record review. Overall sixteen (80.0%) of the 20 IA subjects met 2010 ACR/EULAR RA criteria34 at some point during the study; 12 subjects met criteria at the time of incident IA and 4 subjects met criteria by the next study visit.
Kaplan-Meier curves illustrating IA-free survival after screening aAb+ by different aAb levels and aAb type are presented in figure 2; a summary of absolute risks of IA and 95% CI is presented in table 2. Of subjects who screened aAb+ for both anti-CCP and at least 1 RF isotype, 38.0% developed IA within 5 years, whereas 15.0% of anti-CCP+ only subjects and 9.0% of RF isotype+ only subjects developed IA within 5 years of follow-up (figure 2A). Among subjects with anti-CCP ≥2x the standard cut-off at screening, 32.0% developed IA within 5 years and had a shorter time to IA (p<0.01) (figure 2B). The absolute risk using the optimal cut-off for anti-CCP+ at screening produces a slightly lower absolute risk for IA (26.0%) (figure 2C). The absolute risk for IA at 5 years was highest among subjects with anti-CCP2+ or both CCP2+ and CCP3.1+ at the standard cut-off regardless of RF isotype status (figure 2D). Additional Kaplan-Meier curves by individual aAb are presented in supplementary figure S1.
Figure 2.
Probability of IA-free survival over 12 years of follow-up by aAb levels and type of aAb at screening visit. A) Progression to IA by type of aAb+ at the screening visit; B) Progression to IA by whether the anti-CCP at the screening visit was above or below 2x the standard cut-off for positivity (note that the latter group includes those that were anti-CCP negative by the standard cut-off at the screening visit); C) Progression to IA by whether the anti-CCP at the screening visit was above or below the optimal cut-off for positivity at the screening visit. The optimal cut-off was calculated as ≥5 U/mL for anti-CCP2, and ≥30 U/mL for anti-CCP3.1; D) Progression to IA by type of anti-CCP+ (anti-CCP2 or anti-CCP3.1) at the screening visit regardless of RF isotype status; E) Progression to IA by type of RF isotype+ (RF IgA or RF IgM) at the screening visit regardless of anti-CCP status.
Table 2.
Absolute risk of progression to IA among aAb+ subjects (N=131)
| Characteristic | 1-Year Risk (95% CI) |
3-Year Risk (95% CI) |
5-Year Risk (95% CI) |
|---|---|---|---|
| All Subjects | 3% (2%-8%) | 8% (5%-15%) | 14% (8%-23%) |
| Type of aAb+ | |||
| Anti-CCP- and RF isotype+ | 2% (1%-12%) | 6% (2%-16%) | 9% (4%-24%) |
| Anti-CCP+ and RF isotype− | 5% (2%-15%) | 9% (4%-20%) | 15% (5%-28%) |
| Anti-CCP+ and RF isotype+ | 9% (1%-49%) | 22% (6%-65%) | 38% (13%-79%) |
| Anti-CCP+ ≥ 2x cut-off: Yes | 8% (3%-24%) | 18% (8%-36%) | 32% (18%-54%) |
| *Anti-CCP+ ≥ optimal cut-off | 7% (2%-20%) | 15% (7%-31%) | 26% (14%-46%) |
| †Type of anti-CCP+ | |||
| Anti-CCP2+ | 0% (0%- 0%) | 50% (9%-99%) | 50% (9%-99%) |
| Anti-CCP3.1+ | 2% (13%-30%) | 4% (1%-15%) | 4% (1%-15%) |
| Anti-CCP2+ and Anti-CCP3.1+ | 23% (8%-56%) | 32% (13%-64%) | 57% (29%-88%) |
| ‡Type of RF isotype+ | |||
| RF IgA+ | 0% (0%-0%) | 0% (0%-0%) | 17% (2%-73%) |
| RF IgM+ | 2% (0.3%-15%) | 10% (4%-24%) | 14% (6%-30%) |
| RF IgA+ and RF IgM+ | 14% (2%-67%) | 14% (2%-67%) | 14% (2%-67%) |
The optimal cut-off was calculated as ≥5 U/mL for anti-CCP2, and ≥30 U/mL for anti-CCP3.1
Type of anti-CCP+ regardless of RF isotype status, using standard cut-off.
Type of RF isotype+ regardless of anti-CCP status
Subjects who report NHW race were less likely to develop incident IA (table 3). There was no evidence of a multiplicative or additive interaction between cigarette pack-years (> 10 years) and SE (interaction p=0.37 and p=0.45, respectively). Anti-CCP ≥2x the standard and optimal cut-off at screening were associated with about a 4-fold increased risk of developing incident IA. Anti-CCP2+ and anti-CCP3.1 ≥ the optimal cut-off are associated with increased risk of IA (table3). We note that levels of anti-CCP2 are higher among subjects anti-CCP3.1+ at ≥ the optimal and 2x the standard cut-off (table 4).
Table 3.
Factors associated with progression to inflammatory arthritis since screened aAb+ visit (n=131)
| Characteristic* | Incident IA: No N=111 |
Incident IA: Yes N=20 |
HR (95% CI) | p-value |
|---|---|---|---|---|
| Age: years | 0.98 (0.95-1.01) | 0.24 | ||
| Sex: Female | 87 (73.4%) | 15 (75.0%) | 0.69 (0.25-1.90) | 0.47 |
| Race: NHW | 91 (82.0%) | 13 (65.0%) | 0.39 (0.16-0.98) | 0.04 |
| Shared Epitope: Present | 58 (52.3%) | 13 (65.0%) | 1.24 (0.49-3.14) | 0.65 |
| BMI: ≥25 | 60 (54.1%) | 11 (55.0%) | 1.31 (0.54-3.17) | 0.56 |
| Cigarette Pack-years > 10: Yes | 17 (15.3%) | 1 (5.0%) | 0.32 (0.04-2.43) | 0.27 |
| Tender joint on exam: Yes | 26 (23.4%) | 8 (40.0%) | 2.05 (0.79-5.35) | 0.14 |
| †Self-report joint symptoms: Yes | 67 (60.9%) | 15 (79.0%) | 1.76 (0.58-5.34) | 0.32 |
| ‡CRP: elevated | 38 (34.2%) | 7 (36.8%) | 1.05 (0.41-2.67) | 0.92 |
| Type of aAb+ | ||||
| Anti-CCP- and RF isotype+ | 54 (48.7%) | 6 (30.0%) | ref | ref |
| Anti-CCP+ and RF isotype− | 49 (44.1%) | 11 (55.0%) | 1.67 (0.62-4.52) | 0.31 |
| Anti-CCP+ and RF isotype+ | 8 (7.2%) | 3 (15.0%) | 3.69 (0.92-14.83) | 0.07 |
| RF+ isotype (IgM and/or IgA) ≥ 2x cut-off: Yes | 29 (26.1%) | 6 (30.0%) | 1.69 (0.64-4.48) | 0.29 |
| Anti-CCP+ ≥ 2x cut-off: Yes | 24 (21.6%) | 12 (60.0%) | 4.09 (1.67-10.04) | 0.002 |
| §Anti-CCP+ ≥ optimal cut-off: Yes | 30 (27.0%) | 13 (65.0%) | 3.95 (1.57-9.91) | 0.003 |
| ∣aAb+ Persistence: Yes | 42 (37.8%) | 4 (23.5%) | 1.92 (0.63-5.91) | 0.25 |
| ∣Anti-CCP Persistence: Yes | 37 (33.3%) | 9 (52.9%) | 1.95 (0.75-5.05) | 0.17 |
| Individual aAb+ | ||||
| Anti-CCP2: % positive standard cut-off | 8 (7.2%) | 9 (47.4%) | 11.51 (4.39-30.18) | <0.001 |
| Anti-CCP2: % ≥ 2x standard cut-off | 6 (5.4%) | 8 (40.0%) | 8.88 (3.48-22.64) | <0.001 |
| Anti-CCP2: % positive optimal cut-off (same as standard cut-off) | 11.51 (4.39-30.18) | <0.001 | ||
| Anti-CCP3.1: % positive standard cut-off | 54 (49.1%) | 13 (65.0%) | 1.61 (0.64-4.04) | 0.31 |
| Anti-CCP3.1: % ≥ 2x standard cut-off | 21 (18.9%) | 10 (50.0%) | 3.04 (1.26-7.33) | 0.01 |
| Anti-CCP3.1: % ≥ optimal cut-off | 27 (24.6%) | 11 (55.0%) | 2.84 (1.17-6.86) | 0.02 |
| RF IgM: % positive standard cut-off | 46 (41.4%) | 8 (40.0%) | 1.00 (0.41-2.44) | 0.99 |
| RF IgM: % ≥ 2x standard cut-off | 22 (19.8%) | 6 (30.0%) | 2.34 (0.88-6.21) | 0.09 |
| RF IgA: % positive standard cut-off | 22 (19.8%) | 2 (10.0%) | 0.57 (0.13-2.45) | 0.45 |
| RF IgA: % ≥ 2x standard cut-off | 7 (6.3%) | 1 (5.0%) | 1.05 (0.14-7.86) | 0.97 |
Separate unadjusted models for each risk factor were run
2 subjects did not complete self-reported joint symptom questionnaire
1 subject missing CRP
The optimal cut-off was calculated as ≥5 U/mL for anti-CCP2, and ≥30 U/mL for anti-CCP3.1
3 subjects removed from the analysis of aAb+ persistence because they developed IA prior to the follow-up visit at which aAb+ persistence would have been assessed
Table 4.
Comparing levels of anti-CCP2 by anti-CCP3.1+ status
| ≥ standard cut-off | |||
|---|---|---|---|
| Anti-CCP3.1 ≥20 U/ml |
Anti-CCP3.1 <20 U/ml |
p-value | |
| N | 66 | 63 | |
| Anti-CCP2: median (IQR) |
0.28 (0.14-0.68) | 0.32 (0.14-0.53) | 0.87 |
| ≥ optimal cut-off | |||
| Anti-CCP3.1 ≥30 U/ml |
Anti-CCP3.1 <30 U/ml |
p-value | |
| N | 37 | 92 | |
| Anti-CCP2: median (IQR) |
0.37 (0.20-9.93) | 0.30 (0.12-0.48) | 0.04 |
| ≥2x standard cut-off | |||
| Anti-CCP3.1 ≥40 U/ml |
Anti-CCP3.1 <40 U/ml |
p-value | |
| N | 30 | 99 | |
| Anti-CCP2: median (IQR) |
0.47 (0.20-41.91) | 0.30 (0.12-0.47) | 0.02 |
The use of standard cut-offs for anti-CCP and RF isotypes may be problematic because they are set to different specificities, i.e., the 98%ile for the former and 95%ile for the latter. To determine whether the stronger association of anti-CCP with IA than the RF isotypes was due to the higher specificity of the anti-CCP cut-off, we re-set all aAb+ cut-offs to be the 98%ile of a control population, in this situation, anti-CCP was associated with IA risk and RF isotypes were not (see supplementary tables S2 and S3).
We compared risk models to investigate combinations of factors that predict future development of IA using data available at the screened aAb+ visit. Anti-CCP2+ at the standard cut-off was the strongest predictor of incident IA. However, the presence of either anti-CCP2 or anti-CCP3.1 above the optimal cut-off or ≥2x the standard cut-off was also predictive of incident IA, and this combination identified a larger at-risk population. Therefore, we used the combined anti-CCP for the risk models. We first compared the AUCs of any anti-CCP+ at the standard cut-off and at the optimal cut-off and selected anti-CCP+ ≥ optimal cut-off as a better descriptor of aAb status based on the higher AUC. The addition of NHW Race to model 2 resulted in marginal improvement in the predictive ability over Model 1b, although this was not significant (difference in C-statistics p=0.21). Adding tender joint signs on exam to Model 3 did not improve prediction of IA in this cohort (AUC=0.73, difference in C-statistics p=0.45) (table 5).
Table 5.
Progression to inflammatory arthritis (IA) since aAb+ screening visit (n=131)
| Model | Variable | HR (95% CI) | P-value | Integrated Time-Dependent AUC |
|---|---|---|---|---|
| Model 1a | Anti-CCP+ ≥ 2x cut-off at screening | 4.09 (1.67-10.04) | 0.002 | 0.66 |
| *Model 1b | Anti-CCP+ ≥ optimal cut-off at screening | 3.95 (1.57-9.91) | 0.003 | 0.67 |
| Model 2 | Anti-CCP+ ≥ optimal cut-off at screening | 3.52 (1.37-9.03) | 0.01 | 0.70 |
| Race (NHW) | 0.53 (0.21-1.36) | 0.18 | ||
| Model 3 | Anti-CCP+ ≥ optimal cut-off at screening | 3.23 (1.13-9.23) | 0.03 | 0.73 |
| Race (NHW) | 0.48 (0.16-1.40) | 0.18 | ||
| Tender joint on exam (yes) | 1.61 (0.59-4.42) | 0.37 |
The optimal cut-off was calculated as ≥5 U/mL for anti-CCP2 (the same as the standard cut-off), and ≥30 U/mL for anti-CCP3.1
Differences in Uno’s C-statistic
Model 1b v. Model 2: p=0.21 Model 2 v. Model 3: p=0.45
DISCUSSION
While screening a RA-free at-risk FDR population in a non-clinical setting, 17% tested positive for a RA-related autoantibody, and 15% of these aAb+ individuals developed incident IA over a median of 5 years of follow-up. The strongest predictor of incident IA was screening anti-CCP2+ at or above the standard cut-off and was present in 13% of the study population. Levels of anti-CCP3.1 above an ‘optimal’ cut-off calculated from our data were also predictive of incident IA likely because anti-CCP2 levels were also high or because this indicated epitope spreading.
Prevention trials in preclinical RA 21,23,35,36 rely on the ability to identify at-risk individuals. Two trials use anti-CCP at either >2x or >3x the standard cut-off as inclusion criteria,35,36 and our study confirms that anti-CCP ≥2x cut-off is predictive of future IA. However, we also found that an optimal cut-off for anti-CCP, which included the standard anti-CCP2 cut-off (≥5 U/mL) or 1.5x the standard cut-off for anti-CCP3.1 (≥30 U/mL) was just as predictive of IA. This may inform prevention studies to expand their recruited populations by lowering their anti-CCP inclusion cut-offs.
We found that 64% of our aAb+ study population maintained at least one of their aAbs from the screened positive visit to the follow-up visit. However, we did not find that aAb+ persistence was predictive of incident IA. This is in some contrast to findings in indigenous North American people who have a high prevalence of RA where conversion from aAb+ to aAb- status was associated with decreased risk of IA/RA.20 We hypothesize that we may not have seen an association between aAb persistence and IA because of the speed by which some of the aAb+ subjects progress. For example, we excluded 3 subjects from the analysis examining aAb persistence as a risk factor for IA because they developed IA before the follow-up visit when they would have been tested for aAb again.
The high risk of progression to classifiable RA among those who are anti-CCP positive with joint signs in a clinic setting has previously been reported.11,15,16 Of RA-free subjects who tested anti-CCP+ in tertiary care clinics, 46% with the highest levels of anti-CCP progressed to RA within 5 years.17 The lower proportion of our study subjects with high anti-CCP levels that developed IA (29.0%) may be because our subjects were potentially earlier in the development of RA, perhaps because they were tested outside of routine clinical care.
Our study adds to the literature on the importance of RA-related aAbs on the timing and progression to IA in a non-clinical research setting. Factors thought to be associated with risk of established disease, including older age, smoking, presence of the SE allele(s), the smoking*SE interaction, high levels of CRP, and higher BMI were not associated with IA, perhaps due to their primary influence at other stages of disease, including the initial development of RA-related aAbs. Alternatively, sample size may have limited our ability to detect some of these associations. Even though our population was ascertained outside of a clinical setting, 64.0% of subjects still report some joint symptoms including pain, stiffness, and swelling and 29.0% percent of subjects were found to have joint tenderness on exam. However, these joint signs and symptoms were not associated with either aAb+ persistence or incident IA. This finding is not unexpected because joint symptoms are common.37 Because we could only ascertain self-reported joint symptoms within the past week or joint tenderness on the same day we may be missing the importance of these subtle fluctuations in the prediction of IA.
We have assembled and followed a large cohort of at-risk aAb+ subjects outside of the clinic setting, allowing us the best opportunity to identify factors specifically related to the evolution of systemic autoimmunity and later phases of RA development. However, follow-up studies are needed to validate the results of this study and usefulness of screening for subjects likely to have persistent aAb+ for selecting a targeted high-risk population for targeted epidemiologic, mechanistic and intervention studies.
Supplementary Material
KEY MESSAGES.
What is already known about this subject?
-The presence of anti-CCP and/or RF antibodies are associated with the development of rheumatoid arthritis in banked blood studies and in clinic settings where subjects present with joint symptoms
What does this study add?
-High risk subjects can be identified, recruited, and followed for development of aAb+ persistence and incident IA outside of clinical settings.
- In a prospective study, subjects without RA-like synovitis and with higher levels of anti-CCP are more likely to remain anti-CCP positive and develop IA within 5 years of screening
- A prediction model including factors easy to assess at a study visit may be an efficient means to assemble an at-risk cohort ideal for future prevention, epidemiologic, and mechanistic studies
How might this impact on clinical practice or future developments?
- Identification of individuals at-risk for future RA in a non-clinic setting will improve our capability to study preclinical RA at earlier stages and help to find and verify factors related to disease development
Acknowledgments
This work was previously presented at 2019 American College of Rheumatology/Association of Rheumatology Professionals Annual Meeting. Abstract available at https://acrabstracts.org/abstract/progression-to-inflammatory-arthritis-after-screening-autoantibody-positive-in-a-non-clinical-setting/ (accessed 2 Dec 2019).
Funding This work is supported by the NIH Autoimmunity Prevention Center U19 AI050864 and U01 AI101981; the NIH (grants R01 AR051394, M01 RR00069, M01 RR00425, K23 AR051461, and T32 AR007534); the General Clinical Research Centers Program, National Center for Research Resources, NIH; National Center for Research Resources (grant UL1RR033176), now the National Center for Advancing Translational Sciences (grant UL1TR000124), the Walter S. and Lucienne Driskill Foundation, the Research Support Fund grant from the Nebraska Medical Center, and the University of Nebraska Medical Center.
Footnotes
Competing interests Dr. Demoruelle reports grants from Pfizer, Inc., outside the submitted work; Dr. Buckner reports grants from NIH, during the conduct of the study; in addition, Dr. Buckner reports grants from Jannsen and from BMS, outside the submitted work; Dr. Gregersen reports grants from NIH, during the conduct of the study; Dr. Deane reports non-financial support from Inova Diagnostic, Inc, outside the submitted work; in addition, Dr. Deane has a patent on biomarkers in rheumatoid arthritis with royalties paid; Dr. Holers reports grants from NIH, a grant from Pfizer, Inc, and grants and personal fees from Janssen Research and Development and personal fees from Celgene, all outside of the submitted work. Dr. Norris reports a grant from Pfizer, Inc., outside the submitted work; in addition, Dr. Norris reports grants from NIH, during the conduct of the study. All other authors have nothing to disclose.
Patient consent Obtained.
Patient and public involvement Research was performed without patient involvement in study design, analysis, interpretation, or writing of the manuscript.
Ethics approval The study protocol was approved by the following institutional review boards (IRBs) at each SERA site: Colorado Multiple IRB, University of Nebraska Medical Center IRB, Benaroya Research Institute at Virginia Mason IRB, Cedars-Sinai Medical Center’s IRB, North Shore-LIJ IRB and the Chicago Biomedicine IRB.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information. Data are available on reasonable request.
REFERENCES
- 1.Rantapää-Dahlqvist S, de Jong BA, Berglin E, et al. Antibodies against cyclic citrullinated peptide and IgA rheumatoid factor predict the development of rheumatoid arthritis. Arthritis Rheum 2003;48(10):2741–9. doi: 10.1002/art.11223 [DOI] [PubMed] [Google Scholar]
- 2.Nielen MM, van Schaardenburg D, Reesink HW, et al. Specific autoantibodies precede the symptoms of rheumatoid arthritis: a study of serial measurements in blood donors. Arthritis Rheum 2004;50(2):380–6. doi: 10.1002/art.20018 [DOI] [PubMed] [Google Scholar]
- 3.Malmström V, Catrina AI, Klareskog L. The immunopathogenesis of seropositive rheumatoid arthritis: from triggering to targeting. Nat Rev Immunol 2017;17(1):60–75. doi: 10.1038/nri.2016.124 [published Online First: 2016/12/05] [DOI] [PubMed] [Google Scholar]
- 4.del Puente A, Knowler WC, Pettitt DJ, et al. The incidence of rheumatoid arthritis is predicted by rheumatoid factor titer in a longitudinal population study. Arthritis Rheum 1988;31(10):1239–44. doi: 10.1002/art.1780311004 [DOI] [PubMed] [Google Scholar]
- 5.Silman AJ, Hennessy E, Ollier B. Incidence of rheumatoid arthritis in a genetically predisposed population. Br J Rheumatol 1992;31(6):365–8. doi: 10.1093/rheumatology/31.6.365 [DOI] [PubMed] [Google Scholar]
- 6.Deane KD, Demoruelle MK, Kelmenson LB, et al. Genetic and environmental risk factors for rheumatoid arthritis. Best Pract Res Clin Rheumatol 2017;31(1):3–18. doi: 10.1016/j.berh.2017.08.003 [published Online First: 2017/09/18] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Majka DS, Deane KD, Parrish LA, 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(6):801–7. doi: 10.1136/ard.2007.076679 [published Online First: 2007/11/01] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ärlestig L, Mullazehi M, Kokkonen H, et al. Antibodies against cyclic citrullinated peptides of IgG, IgA and IgM isotype and rheumatoid factor of IgM and IgA isotype are increased in unaffected members of multicase rheumatoid arthritis families from northern Sweden. Ann Rheum Dis 2012;71(6):825–9. doi: 10.1136/annrheumdis-2011-200668 [published Online First: 2011/11/29] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Sokolove J, Bromberg R, Deane KD, et al. Autoantibody epitope spreading in the pre-clinical phase predicts progression to rheumatoid arthritis. PLoS One 2012;7(5):e35296. doi: 10.1371/journal.pone.0035296 [published Online First: 2012/05/25] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Brink M, Hansson M, Mathsson L, et al. Multiplex analyses of antibodies against citrullinated peptides in individuals prior to development of rheumatoid arthritis. Arthritis Rheum 2013;65(4):899–910. doi: 10.1002/art.37835 [DOI] [PubMed] [Google Scholar]
- 11.van Gaalen FA, Linn-Rasker SP, van Venrooij WJ, et al. Autoantibodies to cyclic citrullinated peptides predict progression to rheumatoid arthritis in patients with undifferentiated arthritis: a prospective cohort study. Arthritis Rheum 2004;50(3):709–15. doi: 10.1002/art.20044 [published Online First: 2004/03/17] [DOI] [PubMed] [Google Scholar]
- 12.van der Linden MP, van der Woude D, Ioan-Facsinay A, et al. Value of anti-modified citrullinated vimentin and third-generation anti-cyclic citrullinated peptide compared with second-generation anti-cyclic citrullinated peptide and rheumatoid factor in predicting disease outcome in undifferentiated arthritis and rheumatoid arthritis. Arthritis Rheum 2009;60(8):2232–41. doi: 10.1002/art.24716 [DOI] [PubMed] [Google Scholar]
- 13.Bos WH, Wolbink GJ, Boers M, et al. Arthritis development in patients with arthralgia is strongly associated with anti-citrullinated protein antibody status: a prospective cohort study. Annals of the rheumatic diseases 2010;69(3):490–4. doi: 10.1136/ard.2008.105759 [published Online First: 2009/04/14] [DOI] [PubMed] [Google Scholar]
- 14.van de Stadt LA, de Koning MH, van de Stadt RJ, et al. Development of the anti-citrullinated protein antibody repertoire prior to the onset of rheumatoid arthritis. Arthritis Rheum 2011;63(11):3226–33. doi: 10.1002/art.30537 [DOI] [PubMed] [Google Scholar]
- 15.Rakieh C, Nam JL, Hunt L, et al. Predicting the development of clinical arthritis in anti-CCP positive individuals with non-specific musculoskeletal symptoms: a prospective observational cohort study. Ann Rheum Dis 2015;74(9):1659–66. doi: 10.1136/annrheumdis-2014-205227 [published Online First: 2014/04/15] [DOI] [PubMed] [Google Scholar]
- 16.Bizzaro N, Bartoloni E, Morozzi G, et al. Anti-cyclic citrullinated peptide antibody titer predicts time to rheumatoid arthritis onset in patients with undifferentiated arthritis: results from a 2-year prospective study. Arthritis Res Ther 2013;15(1):R16. doi: 10.1186/ar4148 [published Online First: 2013/01/22] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Ford JA, Liu X, Marshall AA, et al. Impact of Cyclic Citrullinated Peptide Antibody Level on Progression to Rheumatoid Arthritis in Clinically Tested Cyclic Citrullinated Peptide Antibody-Positive Patients Without Rheumatoid Arthritis. Arthritis Care Res (Hoboken) 2019;71(12):1583–92. doi: 10.1002/acr.23820 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.van de Stadt LA, Witte BI, Bos WH, et al. A prediction rule for the development of arthritis in seropositive arthralgia patients. Ann Rheum Dis 2013;72(12):1920–6. doi: 10.1136/annrheumdis-2012-202127 [published Online First: 2012/11/23] [DOI] [PubMed] [Google Scholar]
- 19.Ramos-Remus C, Castillo-Ortiz JD, Aguilar-Lozano L, et al. Autoantibodies in prediction of the development of rheumatoid arthritis among healthy relatives of patients with the disease. Arthritis Rheumatol 2015;67(11):2837–44. doi: 10.1002/art.39297 [DOI] [PubMed] [Google Scholar]
- 20.Tanner S, Dufault B, Smolik I, et al. A Prospective Study of the Development of Inflammatory Arthritis in the Family Members of Indigenous North American People With Rheumatoid Arthritis. Arthritis Rheumatol 2019;71(9):1494–503. doi: 10.1002/art.40880 [published Online First: 2019/08/01] [DOI] [PubMed] [Google Scholar]
- 21.Gerlag DM, Safy M, Maijer KI, et al. Effects of B-cell directed therapy on the preclinical stage of rheumatoid arthritis: the PRAIRI study. Ann Rheum Dis 2019;78(2):179–85. doi: 10.1136/annrheumdis-2017-212763 [published Online First: 2018/12/01] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.van Aken J, Heimans L, Gillet-van Dongen H, et al. Five-year outcomes of probable rheumatoid arthritis treated with methotrexate or placebo during the first year (the PROMPT study). Ann Rheum Dis 2014;73(2):396–400. doi: 10.1136/annrheumdis-2012-202967 [published Online First: 2013/01/22] [DOI] [PubMed] [Google Scholar]
- 23.Deane KD, Striebich CC, Holers VM. Editorial: Prevention of Rheumatoid Arthritis: Now Is the Time, but How to Proceed? Arthritis Rheumatol 2017;69(5):873–77. doi: 10.1002/art.40061 [published Online First: 2017/02/22] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kolfenbach JR, Deane KD, Derber LA, et al. A prospective approach to investigating the natural history of preclinical rheumatoid arthritis (RA) using first-degree relatives of probands with RA. Arthritis Rheum 2009;61(12):1735–42. doi: 10.1002/art.24833 [published Online First: 2009/12/02] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Arnett FC, Edworthy SM, Bloch DA, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31(3):315–24. doi: 10.1002/art.1780310302 [DOI] [PubMed] [Google Scholar]
- 26.Meyers JP, Mandrekar JR. Cutpoint Determination Methods in Survival Analysis using SAS®: Updated %FINDCUT macro. Proceedings of the SAS Global Forum 2015 Conference Cary, NC: SAS Institute Inc., 2015. [Google Scholar]
- 27.Pearson TA, Mensah GA, Alexander RW, et al. Markers of inflammation and cardiovascular disease: application to clinical and public health practice: A statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation 2003;107(3):499–511. doi: 10.1161/01.cir.0000052939.59093.45 [DOI] [PubMed] [Google Scholar]
- 28.Sparks JA, Chang SC, Deane KD, et al. Associations of Smoking and Age With Inflammatory Joint Signs Among Unaffected First-Degree Relatives of Rheumatoid Arthritis Patients: Results From Studies of the Etiology of Rheumatoid Arthritis. Arthritis Rheumatol 2016;68(8):1828–38. doi: 10.1002/art.39630 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Gan RW, Bemis EA, Demoruelle MK, et al. The association between omega-3 fatty acid biomarkers and inflammatory arthritis in an anti-citrullinated protein antibody positive population. Rheumatology (Oxford) 2017;56(12):2229–36. doi: 10.1093/rheumatology/kex360 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Lin DY, Wei LJ, Ying Z. Checking the Cox Model with Cumulative Sums of Martingale-Based Residuals. Biometrika 1993;80(3):557–72. doi: 10.2307/2337177 [DOI] [Google Scholar]
- 31.Karlson EW, Chang SC, Cui J, et al. Gene-environment interaction between HLA-DRB1 shared epitope and heavy cigarette smoking in predicting incident rheumatoid arthritis. Ann Rheum Dis 2010;69(1):54–60. doi: 10.1136/ard.2008.102962 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Li R, Chambless L. Test for additive interaction in proportional hazards models. Ann Epidemiol 2007;17(3):227–36. doi: 10.1016/j.annepidem.2006.10.009 [DOI] [PubMed] [Google Scholar]
- 33.Chambless LE, Diao G. Estimation of time-dependent area under the ROC curve for long-term risk prediction. Stat Med 2006;25(20):3474–86. doi: 10.1002/sim.2299 [published Online First: 2005/10/13] [DOI] [PubMed] [Google Scholar]
- 34.Aletaha D, Neogi T, Silman AJ, et al. 2010 Rheumatoid arthritis classification criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Arthritis Rheum 2010;62(9):2569–81. doi: 10.1002/art.27584 [DOI] [PubMed] [Google Scholar]
- 35.Al-Laith M, Jasenecova M, Abraham S, et al. Arthritis prevention in the pre-clinical phase of RA with abatacept (the APIPPRA study): a multi-centre, randomised, double-blind, parallel-group, placebo-controlled clinical trial protocol. Trials 2019;20(1):429. doi: 10.1186/s13063-019-3403-7 [published Online First: 2019/07/15] [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Strategy to Prevent the Onset of Clinically-Apparent Rheumatoid Arthritis: https://ClinicalTrials.gov/show/NCT02603146.
- 37.Hider SL, Muller S, Helliwell T, et al. Symptoms associated with inflammatory arthritis are common in the primary care population: results from the joint symptoms survey. Rheumatology (Oxford) 2019;58(11):2009–14. doi: 10.1093/rheumatology/kez184 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
All data relevant to the study are included in the article or uploaded as supplementary information. Data are available on reasonable request.