Multi‐Autoantibody Testing Identifies Expansion of Reactivity to Targeted Antigens Before a Diagnosis of Rheumatoid Arthritis

Abstract

Objective

Rheumatoid arthritis (RA) has a “pre‐RA” period in which multiple autoantibodies, including antibodies to citrullinated (cit) proteins (ACPA), rheumatoid factor (RF), anti–peptidyl arginine deiminase (anti‐PAD), among others, have been described; however, few studies have tested all autoantibodies in a single pre‐RA cohort. This study aims to evaluate the prevalence of multiple autoantibodies in pre‐RA and potentially identify an autoantibody profile in pre‐RA that indicates imminent onset of clinical RA.

Methods

We evaluated 148 individuals with two pre‐ and one post‐RA diagnosis samples available from the Department of Defense Serum Repository and matched controls. Samples were tested for immuglobulin (Ig) G anti–cyclic cit peptide‐3 (anti‐CCP3), five ACPA fine specificities, five anti‐PAD isoforms, as well as RF IgA and RF IgM using commercial platforms; cutoffs were determined using levels present in <1% of controls.

Results

Positivity of anti‐CCP3, RF IgA and RF IgM, anti‐PAD1, anti–cit‐vimentin 2, anti–cit‐fibrinogen, and anti–cit‐histone 1 increased over time in pre‐RA, although anti‐PAD and ACPA fine specificities were predominately present within anti‐CCP3–positive individuals. Within anti‐CCP3–positive samples from the pre‐RA period, positivity for RFs as well as anti‐PAD and ACPA fine specificities classified samples as being closer to the time of RA diagnosis.

Conclusion

Multiple autoantibodies are present in pre‐RA and increase in positivity as the time of RA diagnosis approaches. These results confirm previous findings predicting imminent RA and provide a pathway using commercial‐grade assays to assess the risk for and timing of development of clinical RA.

INTRODUCTION

Serum elevations of autoantibodies, including anti-citrullinated (cit) protein antibodies (ACPA) and rheumatoid factor (RF), are associated with and often predate the development of clinically-apparent rheumatoid arthritis (RA) (ie clinical RA).¹ Research in this “pre-RA” state is necessary to understand the biology of RA development as well as to better identify individuals who may benefit from early intervention and develop strategies for prevention trials.

ACPA and RF may first appear years before disease onset, which may limit their utility in identifying individuals at risk for an imminent RA diagnosis.^2,3 The presence of ACPA is commonly assessed by solid-phase assays containing a set of cyclic cit peptides (CCP). Several generations of these anti-CCP assays have been developed (eg, CCP2, CCP3) and are commercially available. In addition, although a clear consensus about the performance difference between the second (CCP2) and third generations (CCP3) is still pending, some studies have indicated that CCP3 might provide benefits in pre-RA settings. In particular, it was demonstrated that anti-CCP3 improves prediction of RA development in anti-CCP2–positive individuals with musculoskeletal symptoms.⁴ However, the antigen specificity of ACPA and their associated repertoire (ie, ACPA fine specificities) are considered heterogenous, and these antibodies are known to be reactive to multiple cit proteins, such as cit-fibrinogen or cit-vimentin.⁵ These ACPA fine specificities (ACPA FS) can be analyzed using specific multiplex assays, and studies have shown that in pre-RA, the number of targets that ACPA recognize can increase, and higher levels are potentially associated with a more imminent onset of RA.^6–8

In addition to ACPA, other autoantibody systems have been shown to be present in pre-RA development. These systems include antibodies to carbamylated proteins (anti-CarP),^9,10 acetylated proteins,¹¹ and malondialdehyde-acetaldehyde,¹² as well as various types of peptidyl arginine deiminases (anti-PAD). Specifically, anti-PAD4 antibodies have been found in up to 45% of patients with established clinical RA (ie, clinical RA) and are highly specific to the disease.^13,14 However, only a single study to date has analyzed anti-PAD4 in the pre-RA population, and in this study, ~18% of individuals who developed RA had anti-PAD4 antibodies before their RA diagnosis.¹⁵

However, despite these autoantibodies being evaluated in pre-RA, to date there are limited data of systematic testing of multiple autoantibody systems in a single pre-RA cohort. Furthermore, historically, many autoantibody tests have not been performed on commercial-grade platforms, which may limit reproducibility and comparisons among studies. As such, this current study evaluated a repertoire of ACPA FS and anti-PAD isoforms along with anti-CCP3 and RF IgA and RF IgM isotypes in pre-RA in a large and well-characterized single cohort of individuals with RA and matched controls obtained from the US Department of Defense Serum Repository (DoDSR) to determine the relative prevalence of positivity of these autoantibodies and whether these autoantibodies can be used in combination to predict time of onset of RA.

PATIENTS AND METHODS

Study population

The individuals with RA as well as controls without RA were created as a retrospective case-control cohort from the DoDSR, which has been described previously.^12,16,17 The DoDSR is part of a program to monitor the health of United States’ uniformed services and military personnel.^18–20 In brief, a total of ~340 individuals who had a diagnosis of clinical RA were identified based on documentation in the medical record and at least one rheumatologist encounter, and their diagnoses were confirmed by a medical chart review by a rheumatologist or trained rheumatology nurse from Walter Reed National Military Medical Center. Furthermore, controls without RA were selected by medical chart review and matched to the participants with RA on age, sex, race, and duration of storage of serum samples, with approximately four controls selected per participant with RA.

For these analyses, a subset of 148 individuals with RA were selected for autoantibody testing because they had three serial serum samples available for testing as follows: one sample >3 years before a diagnosis of RA (termed “pre-RA period 1”), one sample from an interval of ≤3 years and >60 days before a diagnosis of RA (termed “pre-RA period 2”), and a post-RA diagnosis sample. The rationale for this sample set selection was that these samples allowed for a comparison of evolutions of autoantibodies within individuals across multiple time points pre- and post-RA diagnosis.

To set cutoff levels for positivity for each of the autoantibodies tested, a control population was established (see “Autoantibody testing”). For this, we selected a single serum sample per person from 309 controls who were frequency matched to the participants with RA on age, sex, race, and duration of sample storage. In addition, to compare rates of positivity in participants with RA, we selected a separate set of serial samples from 308 controls who were frequency matched to the participants with RA on age, sex, race, and duration of sample storage. Of note, there was a previous evaluation of anti-PAD antibodies in a similarly derived DoDSR cohort^15,21; however, this is a separate study with no overlap among participants. All aspects of this study were approved by institutional review boards at the participating institutions and were done in compliance with the ethical guidelines of the Declaration of Helsinki.

Autoantibody testing

Anti-CCP and RF

Anti-CCP was detected using an automated chemiluminescent assay (CIA) and 3rd generation CCP antigen (QUANTA Flash immunoglobulin G [IgG] CCP3, Inova Diagnostics, Inc.) according to the manufacturer's specifications. RF isotypes (IgA and IgM) were measured using the CIA QUANTA Flash Rheumatoid Factor IgA and IgM (Inova Diagnostics., Inc.) according to the manufacturer's specifications.

ACPA fine specificities and anti-PAD antibodies (IgG)

All samples were tested using a novel fully automated particle-based multianalyte technology (Inova Diagnostics) and the Aptiva system, which uses paramagnetic particles with unique signatures and a digital interpretation system, as described previously.²² The Aptiva RA research use only panel leverages recombinant PAD1, 2, 3, 4, and 6 as well as cit peptides (two versions of cit-vimentin, two versions of cit-histone, and cit-fibrinogen) coupled to microparticles. These particles are first incubated with diluted sera from study participants. After 9.5 minutes of incubation at 37°C, particles are then washed and incubated for another 9.5 minutes at 37°C with anti-human IgG conjugated to phycoerythrin to label the bound IgG autoantibodies from the serum. After a final wash cycle, median fluorescence intensity on the particles is captured using a digital imager and analyzed using proprietary algorithms to extract meaningful information for each analyte.

Cutoffs for each antibody were determined at a level that was present in ≤1% of 309 controls selected from the DoDSR and frequency matched to the participants with RA by age, sex, race, and duration of sample storage. All autoantibody platforms are from Inova Diagnostics Inc, and samples were tested in the Inova Diagnostics laboratory, with all sample RA and control status masked to the operator.

Statistical analyses

Comparisons of demographic features and autoantibody positivity between participants and controls were performed using chi-squared testing (including Fisher's exact test) and t-tests as appropriate for the data (dichotomous or continuous). In addition, for comparisons within the cases among samples over time, when appropriate, paired analyses were performed using McNemar's test for dichotomous outcomes and Wilcoxon signed-rank test was performed for continuous variables. Sensitivity, specificity, positive predictive values (PPVs), negative predictive values (NPVs), and odds ratios (ORs) were calculated using 2 × 2 table analyses, chi-squared testing, and Cochran's and Mantel-Haenszel statistics, as well as matched paired ORs when appropriate. Area under the receiver operator curve (AUC) was also calculated. P values <0.05 were considered significant. All analyses were performed using SPSS version 29.0.2.0 (IBM, Inc.).

RESULTS

Characteristics of participants and controls

The characteristics of the 148 participants with RA and 308 controls are presented in Table 1. The participants had a mean age at diagnosis of RA of approximately 37 years; furthermore, 98.6% of participants met the 1987 American College of Rheumatology (ACR) RA criteria (with the remainder diagnosed with RA by a board-certified rheumatologist), and 55.4% of the participants were women. The controls were similar in age and race to the participants with RA; however, they had a lower percentage of women compared with those with RA, although this difference was not statistically significant.

Table 1 Characteristics of 148 individuals who developed RA and 308 controls*

Characteristics	Earliest pre-RA sample	Immediate pre-RA diagnosis sample	Post-RA diagnosis sample	Comparison controls (first sample)	P value of pre-RA diagnosis 1 compared with controls’ first sample
n	148	148	148	308	–
RA period	Pre-RA diagnosis 1	Pre-RA diagnosis 2	Post-RA diagnosis
Days to diagnosis of RA, mean ± SD [mean y] (negative numbers indicate sample from before RA diagnosis)	−3,711 ± 1,915 [~10.1]	−437 ± 252 [~1.2]	+409 ± 288 [~1.1]	–	–
Age at sample, mean ± SD, y	26.7 ± 7.2	35.5 ± 8.1	37.9 ± 7.9	24.8 ± 6.5	0.314
Age at diagnosis of RA, mean ± SD, y	36.7 ± 8.0	–	–	–	–
Female, n (%)	82 (55.4)	–	–	156 (50.6)	0.423
Race, n (%)
White	80 (54.1)	–	–	163 (52.9)	0.752
Black	40 (27.0)			79 (25.6)	0.752
Hispanic	14 (9.5)			37 (12.0)	0.752
Meeting 1987 ACR RA criteria, n (%)	146/148 (98.6)	–	–	–	–

Autoantibody positivity rates in participants with RA and controls

The prevalence rates of positivity of autoantibodies pre- and post-RA diagnosis in the 148 participants with RA and in the 308 controls are presented in Table 2. In addition, cumulative prevalence rates, as well as levels over time, are presented in Figure 1. Of the individuals with RA, in their post-RA diagnosis sample, 61.5% were positive for anti-CCP3, 45.3% for RF IgA, and 45.9% for RF IgM. When comparing pre-RA diagnosis periods 1 and 2, the positivity rates of anti-CCP3, anti-PAD1, anti–cit-vimentin 2, anti–cit-fibrinogen, anti–cit-histone 1, and RF IgA and RF IgM significantly increased. Furthermore, the positivity rates of all these autoantibodies were higher in the participants with RA compared with the controls. However, although there were mild increases in positivity rates for the individual autoantibodies between the samples from pre-RA diagnosis period 2 and post-RA diagnosis, none of those changes were significant, indicating that for these individuals, most autoantibody positivity developed before the diagnosis of RA. In contrast, the positivity rates of anti-PAD3, anti-PAD4, anti-PAD6, anti-histone 2, and anti-vimentin 1 were not significantly different over time within those who developed RA, and with the exception of anti-PAD4, they were not significantly different between participants with RA and controls.

Table 2 Autoantibody positivity over time in the 148 individuals who developed RA and 308 controls*

RA period, days to diagnosis of RA and autoantibody tested	Autoantibody positivity in earliest sample before RA diagnosis (>3 y)	Autoantibody positivity in the immediate pre-RA diagnosis sample (≤3 y, >60 d)	Autoantibody positivity in the post-RA diagnosis sample	P value of comparison between pre-RA periods 1 and 2	P value of comparison between pre-RA period 2 and post-RA diagnosis	Autoantibody positivity in controls (first sample)	P value of pre-RA diagnosis 1 to controls’ first sample	Autoantibody positivity in controls (last sample)	P value of post-RA diagnosis to controls’ last sample
n	148	148	148			308		307
RA period	Pre-RA period 1	Pre-RA period 2	Post-RA diagnosis
Days to diagnosis of RA, mean ± SD [mean y] (negative numbers indicate sample from before RA diagnosis)	−3,711 (1,915) [~10.1]	−437 (252) [~1.2]	+409 (288) [~1.1]
Anti-CCP3 IgG, n (%)	28 (18.9)	88 (59.5)	91 (61.5)	<0.001	1.000	5 (1.6)	<0.001	3 (1.0)	<0.001
RF IgA, n (%)	16 (10.8)	67 (45.3)	68 (45.9)	<0.001	1.000	5 (1.6)	<0.001	6 (2.0)	<0.001
RF IgM, n (%)	19 (12.8)	68 (45.9)	73 (49.3)	<0.001	0.472	1 (0.3)	<0.001	5 (1.6)	<0.001
Anti-PAD1 IgG, n (%)	3 (2.1)	9 (6.1)	15 (10.1)	0.031	0.146	4 (1.3)	1.000	0 (0)	<0.001
Anti-PAD2 IgG, n (%)	0 (0)	2 (1.4)	0 (0)	0.500	0.500	1 (0.3)	1.000	0 (0)	–
Anti–PAD3 IgG, n (%)	0 (0)	1 (0.7)	0 (0)	1.000	1.000	1 (0.3)	1.000	0 (0)	–
Anti-PAD4 IgG, n (%)	4 (2.7)	9 (6.1)	13 (8.8)	0.063	0.219	4 (1.3)	1.000	2 (0.7)	<0.001
Anti-PAD6 IgG, n (%)	1 (0.7)	1 (0.7)	1 (0.7)	1.000	1.000	2 (0.6)	1.000	2 (0.7)	1.000
Anti–cit-vimentin 1 IgG, n (%)	0 (0)	1 (0.7)	1 (0.7)	1.000	1.000	3 (1.0)	0.554	1 (0.3)	0.541
Anti–cit-vimentin 2 IgG, n (%)	12 (8.1)	63 (42.6)	65 (43.9)	<0.001	0.678	3 (1.0)	0.015	2 (0.7)	<0.001
Anti–cit-histone 1 IgG, n (%)	9 (6.1)	25 (16.9)	30 (20.3)	<0.001	0.383	4 (1.3)	0.043	3 (1.0)	<0.001
Anti–cit-histone 2 IgG, n (%)	0 (0)	2 (1.4)	2 (1.4)	0.500	1.000	3 (1.0)	0.554	1 (0.3)	0.244
Anti–cit-fibrinogen IgG, n (%)	12 (8.1)	63 (42.6)	65 (43.9)	<0.001	0.832	3 (1.0)	0.003	1 (0.3)	<0.001

[IMAGE OMITTED. SEE PDF]

Autoantibodies in participants with RA who are anti-CCP3 positive or negative

We evaluated the profiles of autoantibody positivity in the subset of 100 participants with RA who were anti-CCP3 positive at any point pre- or post-RA diagnosis (Table 3). Similar to what was seen in the analyses that included all participants with RA, there was a significant increase in positivity over time in pre-RA diagnosis periods 1 and 2 for the autoantibodies anti-CCP3, RF IgA and RF IgM, and anti–cit-vimentin 2, anti–cit-histone 1, and anti–cit-fibrinogen. Of note, of the 100 participants with RA who were positive for anti-CCP3 at any time point pre- or post-RA, 91 were positive in their post-RA diagnosis sample, demonstrating that ~9% of individuals lost anti-CCP3 positivity after a diagnosis of RA.

Table 3 Comparisons of autoantibody prevalence rates pre- and post-RA diagnosis in participants who were anti-CCP3 positive or negative*

	Participants (n = 100) who were anti-CCP3 positive at any point pre- or post-RA diagnosis	Participants (n = 48) who were anti-CCP3 negative at any point pre- or post-RA diagnosis
Days to diagnosis of RA, participants per interval and autoantibody tested	Pre-RA period 1	Pre-RA period 2	Post-RA diagnosis	P value of comparison between pre-RA periods 1 and 2	P value of comparison between pre-RA period 2 and post-RA diagnosis	Pre-RA period 1	Pre-RA period 2	Post-RA diagnosis	P value of comparison between pre-RA periods 1 and 2	P value of comparison between pre-RA period 2 and post-RA diagnosis
Days to diagnosis of RA, mean ± SD [mean y] (negative numbers indicate sample from before RA diagnosis)	−3,699 (1,941) [−10.1]	−407 (238) [−1.1]	+421 (297) [+1.2]			−3,736 (1,881) [−10.2]	−497 (272) [−1.4]	+384 (271) [+1.1]
n per time interval	100	100	100			48	48	48
Anti-CCP3, n (%)	28 (28.0)	88 (88.0)	91 (91.0)	<0.001	0.648	0	0	0	–	–
RF IgA, n (%)	14 (14.0)	59 (59.0)	58 (58.0)	<0.001	1.000	2 (4.2)	8 (16.7)	10 (20.8)	0.091	0.794
RF IgM, n (%)	13 (13.0)	60 (60.0)	61 (61.0)	<0.001	1.000	6 (12.5)	8 (16.7)	12 (25.0)	0.773	0.452
Anti-PAD1, n (%)	2 (2.0)	8 (8.0)	14 (14)	0.031	0.146	1 (2.1)a	1 (2.1)a	1 (2.1)a	1.000	1.00
Anti-PAD2, n (%)	0 (0)	2 (2.0)	0 (0)	0.500	0.500	0	0	0	–	–
Anti-PAD3, n (%)	0 (0)	1 (1.0)	0 (0)	1.000	1.000	0	0	0	–	–
Anti-PAD4, n (%)	4 (4.0)	8 (8.0)	12 (12.0)	0.125	0.125	0	1 (2.1)a	1 (2.1)a	1.000	1.000
Anti-PAD6, n (%)	1 (1.0)	1 (1.0)	1 (1.0)	1.000	1.000	0	0	0	–	–
Anti–cit-vimentin 1, n (%)	0 (0)	1 (1.0)	1 (1.0)	1.000	1.000	0	0	0	–	–
Anti–cit-vimentin 2, n (%)	11 (11.0)	62 (62.0)	64 (64.0)	<0.001	0.832	1 (2.1)a	1 (2.1)a	1 (2.7)a	1.000	1.000
Anti–cit-histone 1, n (%)	9 (9.0)	24 (24.0)	30 (30.0)	<0.001	0.263	0	1 (2.1)a	1 (1.8)a	1.000	1.000
Anti–cit-histone 2, n (%)	0 (0)	2 (2.0)	2 (2.0)	0.500	1.000	0	0	0	–	–
Anti–cit-fibrinogen, n (%)	12 (12.0)	43 (43.0)	55 (55.0)	<0.001	0.017	0	1 (1.7)a	0	1.000	1.000

Among the 48 participants with RA who were negative for anti-CCP3 pre- and post-RA diagnosis, RF IgA and RF IgM were the most prevalent autoantibodies (20.8% and 25.0% in post-RA diagnosis samples, respectively). Although the prevalence of positivity for these autoantibodies increased in the pre-RA samples, this was not statistically significant. In contrast, positivity for any of the anti-PAD or ACPA FS occurred in only ~2% of the 48 participants with RA who were anti-CCP3 negative post-RA diagnosis and furthermore did not significantly increase in rates of positivity in the pre-RA period.

Autoantibody combinations and the timing of diagnosis of future RA

A potential approach to identifying at what point in the pre-RA time period an individual may be is to use biomarkers in a two-step method, in which an initial autoantibody (eg, anti-CCP3) is used to determine overall risk, and additional biomarkers are then tested to refine risk and also to estimate timing of onset of future clinical RA.^9,17,23,24 With this as the background, we used anti-CCP3–positive samples from the time points before a diagnosis of RA to evaluate the role of additional autoantibody positivity to estimate the timing to RA diagnosis. There were 28 anti-CCP3–positive samples from pre-RA period 1 and 88 anti-CCP3–positive samples from pre-RA period 2; these were analyzed to test the hypothesis that the prevalence of positivity of anti-CCP3 plus additional autoantibodies would be highest in the time period immediately preceding an RA diagnosis (ie, pre-RA period 2). Furthermore, because RF assays are already commonly used in RA, we first evaluated the diagnostic accuracy of anti-CCP3 and positivity for RFs for the immediate pre-RA period, and then evaluated the diagnostic accuracy of anti-CCP3 positivity plus combinations of the six autoantibodies (besides anti-CCP3) that were found in univariate analyses to significantly increase in positivity in the pre-RA period (ie, RF IgA and RF IgM, anti-PAD1, anti–cit-vimentin 2, anti–cit-histone 1, and anti–cit-fibrinogen).

Positivity for anti-CCP3 plus any RF positivity, or both RF IgA and RF IgM positivity, was significantly greater in pre-RA period 2 compared with the earlier period (Table 4). Furthermore, positivity for anti-CCP3 and one, two, or more of the six additional autoantibodies was also significantly greater in pre-RA period 2. However, although there were trends for combinations of positivity for anti-CCP3 and three or more autoantibodies to be of higher prevalence in pre-RA period 2 compared with the earlier period, these were not statistically significant. In addition, the highest odds ratio (~5.6) was seen with positivity for anti-CCP3 and one or more additional autoantibody (of the six), although the specificity of this result, as well as the AUC results, was less than that seen for the combination of anti-CCP3 plus any RF. Notably, though there were some trends for a greater prevalence of positivity of combinations of anti-CCP3 and other autoantibodies in the post-RA diagnosis period, these were not significant, suggesting that with these systems, most positivity is already present in the immediate pre-RA diagnosis period.

Table 4 Diagnostic accuracy of autoantibodies and combinations of autoantibodies for an anti-CCP3 positive sample that is three years or fewer before a diagnosis of RA*

RA period, days to diangnosis of RA, autoantibodies tested and counts	Pre-RA period 1	Pre-RA period 2	Post-RA diagnosis	P value comparing period 1 and 2	P value comparing period 2 and post-RA diagnosis	Sens, %a	Spec, %a	PPV, %a	NPV, %a	OR, (95% CI)a	AUC, (95% CI)a
Days to diagnosis of RA, mean ± SD [mean y] (negative numbers indicate sample from before RA diagnosis)	−3,699 ± 1,941 [−10.1]	−407 ± 238 [−1.1]	+421 ± 297 [+1.2]
Anti-CCP3 and any RF positive, n/N samples positive for anti-CCP3 (%)	11/28 (39.2)	64/88 (72.7)	64/91 (70.3)	0.003	0.743	72.7	60.7	85.3	41.5	4.121 (1.689–10.054); P = 0.002	0.667 (0.548–0.786); P = 0.006
Anti-CCP3 and both RF isotypes positive, n/N samples positive for anti-CCP3 (%)	8/28 (28.5)	46/88 (52.3)	47/91 (51.6)	0.032	1.000	52.3	71.4	85.2	32.3	2.738 (1.091– 6.874); P = 0.032	0.619 (0.502–0.735); P = 0.047
Median counts of 6 antibodies in anti-CCP3 positive sample, median (range).	1.0 (0–6)	3 (0–6)	3 (0–6)	0.005	1.000	–	–	–	–	–	–
Anti-CCP3 and ≥1 additional autoantibody, n/N samples positive for anti-CCP3 (%)b	17/28 (60.7)	79/88 (89.8)	87/91 (95.6)	0.001	0.158	89.8	39.3	82.3	55.0	5.680 (2.038–15.830); P < 0.001	0.645 (0.518–0.773); 0.025
Anti-CCP3 and ≥2 additional autoantibodies, n/N samples positive for anti-CCP3 (%)b	13/28 (46.4)	70/88 (79.5)	77/91 (84.6)	0.001	0.437	79.5	53.6	84.3	45.5	4.487 (1.814–11.097); P = 0.001	0.666 (0.544–0.788); P = 0.008
Anti-CCP3 and ≥3 additional autoantibodies, n/N samples positive for anti-CCP3 (%)b	9/28 (32.1)	47/88 (53.4)	54/91 (59.3)	0.055	0.454	53.4	67.9	83.9	31.7	2.400 (0.987–5.934); P = 0.053	0.606 (0.488–0.725); P = 0.060
Anti-CCP3 and ≥4 additional autoantibodies, n/N samples positive for anti-CCP3 (%)b	6/28 (21.4)	33/88 (37.5)	35/91 (38.5)	0.168	1.000	37.5	78.6	84.6	28.6	2.200 (0.809–5.984); P = 0.122	0.580 (0.463–698); P = 0.060
Anti-CCP3 and ≥5 additional autoantibodies, n/N samples positive for anti-CCP3 (%)b	3/28 (10.7)	13/88 (14.8)	17/91 (18.7)	0.758	0.551	14.8	89.3	81.3	25.0	1.444 (0.380–5.486); P = 0.589	0.520 (0.399–0.642); P = 0.062
Anti-CCP3 and ≥6 additional autoantibodies, n/N samples positive for anti-CCP3 (%)b	1/28 (3.6)	2/88 (2.3)	3/91 (3.3)	0.567	1.000	2.3	96.4	66.7	23.9	0.628 (0.055–7.197); P = 0.708	0.494 (0.548–0.786); P = 0.063

Finally, because some studies have shown that high levels of anti-CCP are predictive of a high likelihood of imminent onset of RA,²⁴ following a definition of “high” autoantibody level put forward in the 2010 ACR/EULAR RA classification criteria,²⁵ we performed analyses in samples in which anti-CCP3 was present at three or more times the upper limit of normal (ie, anti-CCP3high). In these analyses, we saw similar findings as before, in which positivity of anti-CCP3high plus any RF, both RF IgA and IgM, and one, two, or more of the six autoantibodies was significantly higher in the pre-RA period 2 compared with the earlier time period, although OR and AUCs were somewhat lower in comparison with using anti-CCP3 positivity set at one time or greater than the upper limit of normal (Table 5).

Table 5 Diagnostic accuracy within anti-CCP3 high positive samples (three times or more than the upper limit of normal) of combinations of autoantibodies for a sample that is three years or fewer before a diagnosis of RA*

Days to diagnosis of RA, autoantibodies tested and counts	Pre-RA period 1	Pre-RA period 2	Post-RA diagnosis	P value comparing period 1 and 2	P value comparing period 2 and post-RA diagnosis	Sens, %a	Spec, %a	PPV, %a	NPV, %a	OR, (95% CI)a	AUC, (95% CI)a
Days to diagnosis of RA, mean ± SD [mean y] (negative numbers indicate sample from before RA diagnosis)	−2,640 (1,567) [−7.2]	−377 (224) [1.0]	+423 (305) [1.2]
Anti-CCP3high positive, n/N samples positive for anti-CCP3 (%)	19/28 (67.9)	75/88 (85.2)	76/91 (83.5)	0.054	0.838	85.2	32.1	79.8	40.9	2.733 (1.018–7.338); P = 0.046	0.587 (0.460–0.714); P = 0.181
Anti-CCP3high and any RF isotype (IgA, M) positive, n/N samples positive for anti-CCP3high (%).	9/19 (47.4)	55/75 (73.3)	55/76 (72.4)	0.051	1.000	73.3	52.6	85.9	33.3	3.056 (1.085–8.809); P = 0.035	0.630 (0.484–0.776); P = 0.081
Anti-CCP3high and both RF isotypes (IgA, M) positive, n/N samples positive for anti-CCP3high (%).	6/19 (31.6)	39/75 (52.0)	42/76 (55.3)	0.130	0.745	52.0	68.4	86.7	26.5	2.347 (0.807–6.831); P = 0.117	0.602 (0.462–0.742); P = 0.153
Anti-CCP3high and ≥1 additional autoantibodies, n/N samples positive for anti-CCP3high (%).b	12/19 (63.2)	69/75 (92.0)	75/76 (98.7)	0.004	0.063	92.0	36.8	85.2	53.8	6.078 (1.920–23.436); P = 0.003	0.644 (0.491–0.798); P = 0.066
Anti-CCP3high and ≥2 additional autoantibodies, n/N samples positive for anti-CCP3high (%).b	11/19 (57.9)	62/75 (82.7)	70/76 (92.1)	0.031	0.091	82.7	42.1	84.9	38.1	3.469 (1.167–10.311); P = 0.025	0.624 (0.474–0.774); P = 0.106
Anti-CCP3high and ≥3 additional autoantibodies, n/N samples positive for anti-CCP3high (%).b	9/19 (47.4)	46/75 (61.3)	52/76 (68.4)	0.305	0.397	61.3	52.6	83.6	25.6	1.762 (0.640–4.855); P = 0.273	0.570 (0.424–0.716); P = 0.698
Anti-CCP3high and ≥4 additional autoantibodies, n/N samples positive for anti-CCP3high (%).b	6/19 (31.6)	32/75 (42.7)	35/76 (46.1)	0.441	0.744	42.7	68.4	84.2	23.2	1.612 (0.553–0.701); P = 0.382	0.555 (0.413–0.698); P = 0.446
Anti-CCP3high and ≥5 additional autoantibodies, n/N samples positive for anti-CCP3high (%).b	3/19 (15.8)	13/75 (17.3)	17/76 (22.4)	1.000	0.541	17.3	84.2	81.3	20.5	1.118 (0.284–4.403); P = 0.873	0.508 (0.363–0.653); P = 0.917
Anti-CCP3high and ≥6 additional autoantibodies, n/N samples positive for anti-CCP3high (%).b	1/19 (5.3)	2/75 (2.7)	3/76 (4.0)	0.496	1.000	2.7	94.7	66.7	19.8	0.493 (0.042–5.745); P = 0.573	0.487 (0.339–0.635); P = 0.864

DISCUSSION

Herein, we have identified that multiple autoantibody systems are abnormal in pre-RA. In particular, in this sample set, anti-CCP3 was the most commonly positive autoantibody, followed by RF IgA and RF IgM, then anti–cit-vimentin 2, anti–cit-fibrinogen, anti–cit-histone 1, and anti-PAD1. In addition, positivity for the anti-PAD and ACPA FS autoantibodies was most common in individuals who were also positive for anti-CCP3. Furthermore, in pre-RA samples, combinations of autoantibodies were able to classify a sample as being closer to the time of diagnosis. Notably, however, several autoantibodies tested (anti–PAD2–4 and 6, anti–cit-vimentin 1, and anti–cit-histone 2) did not significantly increase over time in pre-RA and were not significantly different from controls.

Overall, this study validates that certain systems that include anti-CCP, certain ACPA FS and anti-PAD1, and RF IgA and RF IgM can be present in pre-RA. These findings can be used to support further investigations into the biology of the development of RA. For example, anti-PAD antibodies may affect the function of the enzymes and therefore play a role in the generation/expansion of autoimmunity.¹³ In addition, and building on findings from other studies,⁹ the finding of multiple autoantibodies classifying samples as being within three years of a diagnosis of RA may be useful in developing future prediction models for inclusion of individuals in prevention studies in RA.^26–29 Moreover, in the abatacept in Individuals at high risk of rheumatoid arthritis  study, which was a randomized controlled trial evaluating the efficacy of abatacept to delay future RA in individuals with ACPA(+), findings from exploratory analyses that have been presented in abstract form demonstrated that individuals with ACPA(+) with additional autoantibody positivity (RF and others) had the highest rates of RA development as well as the highest response rates to the study drug^29,30; this could indicate that multi-antibody testing will identify a subset of individuals who will be most likely to respond to certain preventive interventions. Furthermore, the commercial nature of the testing platform can be leveraged in future studies to optimize standardization of testing and translation into other populations. Notably, the sensitivities, specificities, PPVs, NPVs, ORs, and AUCs of a combination of anti-CCP3 and RF IgA and RF IgM, which are commonly available in most clinical laboratories, were fairly comparable with the results using the extended autoantibody profile (ie anti-PADs and ACPA FS) to identify whether a sample was from a time point that was close (eg, three years or fewer) or far (eg, three years or more) from a diagnosis of RA. This suggests that anti-CCP3 and RFs alone may be useful in studies of pre-RA and, in particular, to identify individuals who are at risk of imminent onset of clinical RA. However, the extended profile allows for additional refinement of classification if a sample is from a time point close to the onset of clinical RA. As such, extended profile testing may have an important role in selecting individuals to include for future studies. For example, positivity for anti-CCP3 plus six other autoantibodies is highly specific (96.4%, Table 4) for that sample being three years or fewer before a diagnosis of clinical RA, and a future trial could consider that as an inclusion criteria if individuals with high-risk for imminent onset of RA were deemed necessary.

Several findings are of particular interest. When evaluating the anti-CCP3 positive cases (Table 3), the prevalence of positivity for anti–cit-fibrinogen significantly increased from pre- to post-RA diagnosis. However, in most cases, the positivity rate of autoantibodies did not significantly increase between pre- and post-RA diagnosis samples—a finding that has been observed in other studies.³¹ A hypothesis has been that a factor relating to the transition from pre-RA to clinical RA is an expansion of autoantibody reactivity. These findings could indicate that an increase in reactivity to cit-fibrinogen in some individuals may be related to a transition to clinical RA. In particular, antibodies to cit-fibrinogen may lead to increased inflammatory responses in established RA,³² and therefore, increased prevalence of positivity may help drive a transition from pre- to clinical RA. However, given that the majority of autoantibodies did not increase, these findings also support that there may be other factors besides autoantibody reactivity to these specific antigens that are related to a transition from pre- to clinical RA. These factors could be expansion of autoantibodies (or isotypes) to antigens that were not tested herein or other processes such as evolving cellular activity. In addition, as studies have shown that antibody glycosylation may change in pre-RA,^33–35 it is possible that glycosylation changes are part of the transition from pre-RA to clinical RA. These points will need further investigation.

It is also of interest that in this sample set, ACPA FS and anti-PAD antibodies were largely only present in individuals who were anti-CCP3 positive. Previous studies have shown that ACPA FS and anti-PAD, and other autoantibodies, may be present in individuals with RA who are anti-CCP negative.^5,14 Our findings may be related to this specific DoDSR sample set or could also imply that these autoantibodies are more common in more advanced clinical RA which may have been present in other studies. Furthermore, the technologies used for autoantibody testing, or the specific antigens used in the assays, may differ in this study compared with others. Our approach to establishing test positivity (ie, levels ≤1% in controls) may also result in differences in findings from other studies because of the high specificity of this cutoff for RA. These issues will need exploration in future studies that may include additional evaluation of molecular signatures, such as relationships between levels of autoantibodies rather than autoantibody positivity rates. Notably, anti-CarP is a biomarker that has been described in both seropositive and seronegative clinical RA.³⁶ Additionally, anti-CarP positivity, along with ACPA and RF positivity, has been shown to be highly specific for future RA.¹⁰ We did not include anti-CarP testing, but this, as well as a growing number of autoantibodies, will be systems to explore in the future.

There are several limitations to this study. This DoDSR population is younger and has more men than many published cohorts of RA and pre-RA. That may make these results less generalizable to other populations. Importantly, there are a growing number of studies in RA evaluating the natural history of RA as well as preventive interventions in individuals who are anti-CCP positive. However, in many of these studies, only 30% to 50% of individuals who are anti-CCP positive develop clinical RA.^{23,26,27,37,38} As such, identifying biomarkers that could help differentiate which individuals who are anti-CCP positive will get future RA would improve prediction. The findings herein are from a case-control study and thus limit our ability to create prediction models to help differentiate which individuals who are anti-CCP positive may develop future RA. However, the concepts of using autoantibody counts with the autoantibodies we tested or additional biomarkers (eg, antibodies to carbamylated and acetylated proteins and others^39,40) could be applied to other studies in the future to gain further insights into which individuals with ACPA(+) will or will not develop clinical RA, as well as using these markers to understand the potential timing of onset of clinical RA. In addition, at the earliest pre-RA time point, anti-CCP3 has the highest positivity of the autoantibodies we tested in this study. This could suggest that anti-CCP3 appears earliest. We did not formally evaluate the timing of appearance of each autoantibody; however, future research efforts will investigate this further.

In summary, our findings suggest that there is expansion of autoantibodies to the antigen targets measured herein in pre-RA. These findings support and extend previous work and set the stage for additional studies of the biology of evolution of pre-RA as well as models to predict more imminent onset of RA.

ACKNOWLEDGMENTS

The authors thank the Department of Defense, the DoDSR, and the members of the United States Armed Forces for providing the data and samples.

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 Deane 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

Goff, Bergstedt, Feser, Moss, Mikuls, Edison, Holers, Martinez-Prat, Aure, Mahler, Deane.

Acquisition of data

Feser, Moss, Edison, Martinez-Prat, Aure, Mahler, Deane.

Analysis and interpretation of data

Goff, Bergstedt, Feser, Moss, Mikuls, Edison, Holers, Martinez-Prat, Aure, Mahler, Deane.

References

Deane KD, Holers VM. Rheumatoid arthritis pathogenesis, prediction, and prevention: an emerging paradigm shift. Arthritis Rheumatol 2021;73(2):181–193.

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):380–386.

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):2741–2749.

Di Matteo A, Mankia K, Duquenne L, et al. Third‐generation anti‐cyclic citrullinated peptide antibodies improve prediction of clinical arthritis in individuals at risk of rheumatoid arthritis. Arthritis Rheumatol 2020;72(11):1820–1828.

Ioan‐Facsinay A, Willemze A, Robinson DB, et al. Marked differences in fine specificity and isotype usage of the anti‐citrullinated protein antibody in health and disease. Arthritis Rheum 2008;58(10):3000–3008.

van der Woude D, Rantapää‐Dahlqvist S, Ioan‐Facsinay A, et al. Epitope spreading of the anti‐citrullinated protein antibody response occurs before disease onset and is associated with the disease course of early arthritis. Ann Rheum Dis 2010;69(8):1554–1561.

van de Stadt LA, van der Horst AR, de Koning MH, et al. The extent of the anti‐citrullinated protein antibody repertoire is associated with arthritis development in patients with seropositive arthralgia. Ann Rheum Dis 2011;70(1):128–133.

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): [eLocator: e35296].

Ponchel F, Duquenne L, Xie X, et al. Added value of multiple autoantibody testing for predicting progression to inflammatory arthritis in at‐risk individuals. RMD Open 2022;8(2):8.

Verheul MK, Böhringer S, van Delft MAM, et al. Triple positivity for anti‐citrullinated protein autoantibodies, rheumatoid factor, and anti‐carbamylated protein antibodies conferring high specificity for rheumatoid arthritis: implications for very early identification of at‐risk individuals. Arthritis Rheumatol 2018;70(11):1721–1731.

Kissel T, Reijm S, Slot LM, et al. Antibodies and B cells recognising citrullinated proteins display a broad cross‐reactivity towards other post‐translational modifications. Ann Rheum Dis 2020;79(4):472–480.

Mikuls TR, Edison J, Meeshaw E, et al. Autoantibodies to malondialdehyde‐acetaldehyde are detected prior to rheumatoid arthritis diagnosis and after other disease specific autoantibodies. Arthritis Rheumatol 2020;72(12):2025–2029.

Curran AM, Naik P, Giles JT, et al. PAD enzymes in rheumatoid arthritis: pathogenic effectors and autoimmune targets. Nat Rev Rheumatol 2020;16(6):301–315.

Martinez‐Prat L, Palterer B, Vitiello G, et al. Autoantibodies to protein‐arginine deiminase (PAD) 4 in rheumatoid arthritis: immunological and clinical significance, and potential for precision medicine. Expert Rev Clin Immunol 2019;15(10):1073–1087.

Kolfenbach JR, Deane KD, Derber LA, et al. Autoimmunity to peptidyl arginine deiminase type 4 precedes clinical onset of rheumatoid arthritis. Arthritis Rheum 2010;62(9):2633–2639.

Kelmenson LB, Wagner BD, McNair BK, et al. Timing of elevations of autoantibody isotypes prior to diagnosis of rheumatoid arthritis. Arthritis Rheumatol 2020;72(2):251–261.

Bettner LF, Peterson RA, Bergstedt DT, et al. Combinations of anticyclic citrullinated protein antibody, rheumatoid factor, and serum calprotectin positivity are associated with the diagnosis of rheumatoid arthritis within 3 years. ACR Open Rheumatol 2021;3(10):684–689.

Perdue CL, Cost AA, Rubertone MV, et al. Description and utilization of the United States Department of Defense Serum Repository: a review of published studies, 1985–2012. PLoS One 2015;10(2): [eLocator: e0114857].

Perdue CL, Eick‐Cost AA, Rubertone MV. A brief description of the operation of the DoD serum repository. Mil Med 2015;180(10 Suppl):10–12.

Rubertone MV, Brundage JF. The Defense Medical Surveillance System and the Department of Defense Serum Repository: glimpses of the future of public health surveillance. Am J Public Health 2002;92(12):1900–1904.

Gan RW, Trouw LA, Shi J, et al. Anti‐carbamylated protein antibodies are present prior to rheumatoid arthritis and are associated with its future diagnosis. J Rheumatol 2015;42(4):572–579.

Palterer B, Vitiello G, Del Carria M, et al. Anti‐protein arginine deiminase antibodies are distinctly associated with joint and lung involvement in rheumatoid arthritis. Rheumatology (Oxford) 2023;62(7):2410–2417.

Bergstedt DT, Tarter WJ, Peterson RA, et al. Antibodies to citrullinated protein antigens, rheumatoid factor isotypes and the shared epitope and the near‐term development of clinically‐apparent rheumatoid arthritis. Front Immunol 2022;13: [eLocator: 916277].

Duquenne L, Hensor EM, Wilson M, et al. Predicting inflammatory arthritis in at‐risk persons: development of scores for risk stratification. Ann Intern Med 2023;176(8):1027–1036.

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):2569–2581.

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):179–185.

Krijbolder DI, Verstappen M, van Dijk BT, et al. Intervention with methotrexate in patients with arthralgia at risk of rheumatoid arthritis to reduce the development of persistent arthritis and its disease burden (TREAT EARLIER): a randomised, double‐blind, placebo‐controlled, proof‐of‐concept trial. Lancet 2022;400(10348):283–294.

Rech J, Tascilar K, Hagen M, et al. Abatacept inhibits inflammation and onset of rheumatoid arthritis in individuals at high risk (ARIAA): a randomised, international, multicentre, double‐blind, placebo‐controlled trial. Lancet 2024;403(10429):850–859.

Cope AP, Jasenecova M, Vasconcelos JC, et al; APIPPRA study investigators. Abatacept in individuals at high risk of rheumatoid arthritis (APIPPRA): a randomised, double‐blind, multicentre, parallel, placebo‐controlled, phase 2b clinical trial. Lancet 2024;403(10429):838–849.

Cope A, Jasenecova M, Vasconcelos J, et al. Abatacept in individuals at risk of developing rheumatoid arthritis: results from the arthritis prevention in the preclinical phase of RA with abatacept (Apippra) trial. [Abstract taken from EULAR 2023]. Ann Rheum Dis 2023;82:86.

Heutz JW, Rogier C, Niemantsverdriet E, et al. The course of cytokine and chemokine gene expression in clinically suspect arthralgia patients during progression to inflammatory arthritis. Rheumatology (Oxford) 2024;63(2):563–570.

Boman A, Brink M, Lundquist A, et al. Antibodies against citrullinated peptides are associated with clinical and radiological outcomes in patients with early rheumatoid arthritis: a prospective longitudinal inception cohort study. RMD Open 2019;5(2): [eLocator: e000946].

Ercan A, Cui J, Chatterton DE, et al. Aberrant IgG galactosylation precedes disease onset, correlates with disease activity, and is prevalent in autoantibodies in rheumatoid arthritis. Arthritis Rheum 2010;62(8):2239–2248.

Hafkenscheid L, de Moel E, Smolik I, et al. N‐linked glycans in the variable domain of IgG anti‐citrullinated protein antibodies predict the development of rheumatoid arthritis. Arthritis Rheumatol 2019;71(10):1626–1633.

Rombouts Y, Willemze A, van Beers JJ, et al. Extensive glycosylation of ACPA‐IgG variable domains modulates binding to citrullinated antigens in rheumatoid arthritis. Ann Rheum Dis 2016;75(3):578–585.

Shi J, Knevel R, Suwannalai P, et al. Autoantibodies recognizing carbamylated proteins are present in sera of patients with rheumatoid arthritis and predict joint damage. Proc Natl Acad Sci USA 2011;108(42):17372–17377.

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):1659–1666.

van Boheemen L, Turk S, Beers‐Tas MV, et al. Atorvastatin is unlikely to prevent rheumatoid arthritis in high risk individuals: results from the prematurely stopped STAtins to Prevent Rheumatoid Arthritis (STAPRA) trial. RMD Open 2021;7(1):7.

Volkov M, Kampstra ASB, van Schie KA, et al. Evolution of anti‐modified protein antibody responses can be driven by consecutive exposure to different post‐translational modifications. Arthritis Res Ther 2021;23(1):298.

Kampstra ASB, Dekkers JS, Volkov M, et al. Different classes of anti‐modified protein antibodies are induced on exposure to antigens expressing only one type of modification. Ann Rheum Dis 2019;78(7):908–916.