Evidence of Diagnostic and Treatment Delay in Seronegative Rheumatoid Arthritis: Missing the Window of Opportunity

Abstract

Objectives

To compare the time to fulfillment of ACR/EULAR classification criteria between patients with seropositive and seronegative rheumatoid arthritis (RA), and to assess the impact of seronegative status on the time to DMARD therapy and achievement of remission.

Methods

Time from first provider-documented joint swelling to fulfillment of 1987 and 2010 ACR/EULAR criteria and clinical diagnosis of RA were measured in a population-based cohort of adults with incident RA between January 1, 2009 and December 31, 2014. Disease characteristics and achievement of remission were compared between seropositive (RF+ and/or ACPA+) or seronegative (RF−/ACPA−) patients.

Results

Median time from first joint swelling to fulfillment of 1987 (48 vs 2 days, P=.001) and 2010 (14 vs 0 days, P=.004) classification criteria and to clinical diagnosis of RA (187 vs 11 days, P<.001) were significantly longer in seronegative versus seropositive patients. The median time in days from first joint swelling to first DMARD was significantly longer in seronegative patients (40 vs 14 days, P=.01). Patients with seronegative RA were less likely to achieve remission (28% vs 50% at 5 years after fulfillment of 2010 criteria; P=.007), but there was no difference when patient global score was removed from the remission definition.

Conclusions

Patients with seronegative RA experienced a delay in diagnosis, according to both 1987 and 2010 classification criteria, and delay in initiation of DMARD therapy. Patients with seronegative RA also were less likely to attain remission, suggesting that the window of opportunity for intervention may be more frequently missed in this group.

Introduction

Rheumatoid arthritis (RA) is an autoimmune, inflammatory joint disease characterized by swelling, pain, and destruction of synovial joints. Joint damage accumulates over time with disease, leading to disability and mortality (1). Serologic status according to rheumatoid factor (RF) and anti-citrullinated peptide antibody (ACPA) has become an important diagnostic and prognostic factor. An estimated 20-25% of cases of RA are seronegative, meaning that patients do not express RF or ACPA in the serum despite meeting clinical classification criteria for RA. Furthermore, an estimated 50% of patients are seronegative in early disease and become seropositive (2).

Early diagnosis and initiation of therapy correlates with better outcomes, higher rates of remission, and reduced joint damage and disability for both seropositive and seronegative RA patients (1,3–5). Evidence supports a therapeutic window of opportunity, during which initiation of disease-modifying antirheumatic drug (DMARD) therapy most effectively improves clinical outcomes and prevents joint damage (5). However, early RA may have subtle features and thus go undiagnosed in the early stages during which treatment may be most beneficial. In seropositive patients, ACPA and RF can often be detected before clinical disease onset (6). The 2010 ACR/EULAR RA classification criteria were developed with the goal of earlier classification; leaving out criteria for manifestations of chronic, erosive disease such as rheumatoid nodules included in 1987 criteria (7), and placing greater emphasis on serologic biomarkers. As such, current 2010 ACR/EULAR criteria require that seronegative patients have more joint involvement than seropositive patients in order to meet criteria for RA classification (1).

Several studies have shown that the 2010 criteria are fulfilled earlier than the 1987 criteria (8,9). Recently, the 2010 criteria performed better in the seropositive population in the Leiden early arthritis and ESPOIR cohorts, suggesting that 49-75% of seronegative patients miss early classification by 2010 criteria (10). To date, the delay that seronegative patients experience in meeting classification criteria has not been defined. The impact of this delay on clinical outcomes such as pain, function, and achievement of remission remains unknown. We hypothesized that seronegative patients experience a delay in meeting classification criteria and receiving a clinical diagnosis of RA from time of symptom onset, and thus, a delay in treatment initiation when compared with seropositive patients, potentially missing the optimal window of opportunity for intervention.

Methods

Study Population

The Mayo Clinic and Olmsted Medical Center Institutional Review Boards approved this retrospective cohort study. Subjects were identified using the Rochester Epidemiology Project, a geographically based collaboration of healthcare facilities allowing access to complete medical records across institutions. Subjects were adult residents of Olmsted County, MN who developed incident RA between January 1, 2009 and December 31, 2014. Eligibility required age ≥ 18 years and earliest fulfillment of either the 1987 or 2010 ACR/EULAR classification criteria for RA in 2009-2014. All subjects were followed longitudinally until last medical visit, death or December 31, 2017.

Data Collection and Study Variables

Retrospective review of medical records and diagnoses was performed by a trained nurse abstractor. Additional record review was performed by study personnel (CMC, CSC). All subjects were classified either as seropositive, defined as RF positive and/or ACPA positive, or seronegative, defined as RF negative and ACPA negative, at the time of fulfillment of RA criteria. First joint swelling was defined as the first clinically recognized synovitis as documented by a medical provider in any specialty. Times from first joint swelling to meeting 1987 and 2010 ACR/EULAR classification criteria, dates of clinical diagnosis of “rheumatoid arthritis” or “inflammatory arthritis” as recorded in the medical chart or provider note, and dates of first prescribed DMARD therapy were calculated in days. Outliers, defined as fulfillment of criteria >365 days from first joint swelling, were reviewed by the first author (CMC) and any errors corrected, with verified outliers included in the analysis.

All available disease activity measures were electronically retrieved from electronic health records. These measures, routinely collected as part of clinical care since 2011, included the Health Assessment Questionnaire (HAQ), patient and provider global assessments (0-100) of disease activity, self-reported pain on a visual analogue scale, tender (TJC, 0-28) and swollen joint counts (SJC, 0-28), C-reactive protein (CRP), Disease Activity Score (DAS28-CRP), and Clinical (CDAI) and Simplified (SDAI) Disease Activity Indices. The Boolean definition of remission was used, defined as SJC ≤1, TJC ≤1, CRP ≤10 mg/L, and patient global assessment ≤10 per 2010 ACR/EULAR criteria.

Statistical Analysis

Descriptive statistics were used to summarize the data. Comparisons between groups were made using Chi-square and rank sum tests. Trends in disease activity measures over time were examined using generalized linear models, adjusted for age, sex, and time since fulfillment of 2010 criteria with random subject effects to account for multiple measurements per patient. Smoothing splines were used for examination of non-linear trends. Time to remission was estimated using Kaplan-Meier methods. Cox proportional hazards models adjusted for age and sex were used to compare time to remission between groups. Analyses were performed using SAS version 9.4 (SAS Institute, Cary, North Carolina, U.S.) and R 3.4.2 (R Foundation for Statistical Computing, Vienna, Austria).

Results

There were 37 patients who first fulfilled 1987 criteria for RA in 2009-2014, but never fulfilled 2010 criteria. Of these, 34 were seronegative and 3 were seropositive. A total of 214 patients with RA who fulfilled both the 1987 and 2010 criteria were included; of these 145 were seropositive and 69 seronegative at the time of fulfillment of the latter of the 1987 and 2010 criteria. Two of the seronegative patients became seropositive more than 1 year after criteria fulfillment; these were considered as seronegative for analyses. The baseline characteristics of patients by antibody status are shown in Table 1. The mean age was 55.1 years in the seropositive group and 55.8 years in the seronegative group. Age, sex, smoking status, and obesity did not differ significantly between groups. Seronegative patients had more extensive joint involvement than seropositive patients (83% vs 64% had >10 joints involved, P=.009) at the time of meeting 2010 classification criteria. 19% of seropositive and 28% seronegative patients had erosive disease at time of fulfilling 2010 criteria (P=.17).

Table 1.

Baseline characteristics of 214 patients with incident RA between 2009-2014 fulfilling both the 1987 and 2010 classification criteria for RA.

Characteristic	RF+ and/or ACPA+	RF− / ACPA−	P-value
Number (%) of patients	145 (68%)	69 (32%)
Age, years, mean (±SD)	55.1 (±14.7)	55.8 (±14.6)	.75
Sex, female	99 (68%)	49 (71%)	.68
Smoking status			.43
Never	80 (55%)	34 (49%)
Former	38 (26%)	24 (35%)
Current	27 (19%)	11 (16%)
Obesity (BMI≥30 kg/m2)	59 (41%)	26 (38%)
Number of joints involved			.009
1-3 joints	7 (5%)	4 (6%)
4-10 joints	45 (31%)	8 (12%)
>10 joints	93 (64%)	57 (83%)
Abnormal ESR and/or CRP	108 (74%)	45 (65%)	.16
Erosions	29 (19%)	19 (28%)	.17

^aACPA = anti-citrullinated peptide antibodies; BMI = body mass index; CRP = C-reactive protein; HAQ = health assessment questionnaire; RA = rheumatoid arthritis; RF = rheumatoid factor; SD = standard deviation; SJC = swollen joint count; TJC = tender joint count.

Among 109 patients with available disease activity measures prior to or within 1 year after fulfillment of 2010 criteria, the median follow-up was 4.1 years (range: 1.0 to 7.4 years), during which disease activity measures were available for a median of 8 visits per patient, with no significant differences between groups. 43% of seronegative and 53% of seropositive patients had first synovitis documented by a rheumatologist. Those not on DMARD therapies at time of last follow-up were reviewed to ensure that a diagnosis of RA was still valid. A total of 25 patients were off of DMARDs; reasons included patient preference, alternate treatment with biologic medications, or complications from other health conditions. Two seropositive patients in this group had a diagnosis change, to limited scleroderma. All were included in analysis.

Median time from first joint swelling to fulfillment of the 1987 (48 vs 2 days, P=.001) and 2010 (14 vs 0 days, P=.004) classification criteria was significantly longer in seronegative versus seropositive patients (Table 2). There was no significant difference in days from fulfillment of 2010 to 1987 criteria between the groups. 41% of seronegative vs 17% of seropositive patients experienced a delay of ≥12 weeks to meeting 1987 criteria (P<.001), while 38% of seronegative vs 16% of seropositive patients experienced a delay ≥16 weeks (P<.001). 40% of seronegative vs 14% of seropositive met 2010 criteria ≥ 12 weeks from first joint swelling (P<.001) and 33% of seronegative vs 12% of seropositive patients met 2010 criteria ≥16 weeks from first joint swelling (P<.001). 178 patients received a clinical diagnosis of RA and 197 patients received a clinical diagnosis of RA and/or inflammatory arthritis. The median time from first joint swelling to clinical diagnosis of RA was longer in seronegative vs seropositive patients (187 vs 11 days, P<.001), though median time to clinical diagnosis of either RA or inflammatory arthritis was not significantly different (12 vs 1 days, P=.11). Time from first joint swelling to first rheumatologist visit was similar for seronegative and seropositive patients (median [IQR]: 1 [0-31] vs 0 [0-20], P=.34). The median time in days to first DMARD was significantly longer in seronegative patients (40 vs 14 days, P=.01).

Table 2.

Comparison of time from first documented joint swelling to fulfillment of the 1987 or 2010 criteria and to time of clinical diagnosis between seropositive and seronegative patients with incident RA between 2009-2014.

	RF+ and/or ACPA+	RF− / ACPA−	P-value
Nb	139	63
Median time from first joint swelling to 1987 criteria, days (IQR)	2 (0, 45)	48 (0, 300)	.001
Median time from first joint swelling to 2010 criteria, days (IQR)	0 (0, 29)	14 (0, 196)	.004
Time from first joint swelling to DMARD initiation, days (IQR)	14 (0, 73)	40 (5, 199)	.01
Patients with ≥12 weeks from first joint swelling to 1987 criteria, n (%)	24 (17%)	26 (41%)	<.001
Patients with ≥12 weeks from first joint swelling to 2010 criteria, n (%)	19 (14%)	25 (40%)	<.001
Time from first joint swelling to clinical diagnosis of RA, days (IQR)	11 (0, 76)	187 (13, 503)	<.001

^aACPA, anti-citrullinated peptide antibodies; DMARD, disease-modifying anti-rheumatic drug; IQR, interquartile range; RF, rheumatoid factor.

^bDate of documented first joint swelling was indeterminate for 6 patients in each group, and date of clinical diagnosis of RA was indeterminate for 15 seropositive and 21 seronegative patients.

Analysis comparing the three seropositive subgroups (RF+/ACPA+; RF+/ACPA−; RF−/ACPA+) did not show significant differences in time to meeting 1987 or 2010 classification criteria. When the RF+/ACPA− and RF−/ACPA+ groups were combined and compared with double-positive patients, 20% of single-positive vs 9% of double-positive patients met 2010 criteria ≥12 weeks from first joint swelling (P=.067). 19% of single-positive vs 8% of double positive patients met 2010 criteria ≥16 weeks from first joint swelling (P=.071).

Methotrexate was used as first DMARD in 133 of all 214 patients (62%) and was used as first DMARD therapy in 66% of seropositive vs 55% of seronegative patients (P=.14). Hydroxychloroquine was used first in 67 patients (31%) and more commonly used in seronegative patients though this difference was not significant (36% vs 29%, P=.37). Leflunomide, sulfasalazine, or other DMARDs were used first rarely; with similar frequency between groups. The starting dose of methotrexate was not different between groups (data not shown), nor was the time from first joint swelling to first glucocorticoid use (median 20 vs 51 days, P=.16).

Figure 1 compares achievement of remission between seronegative and seropositive groups over time. Patients with seropositive RA were more likely to achieve remission. By 5 years after fulfillment of 2010 criteria, 50% (95% confidence interval [CI]: 36-61%) of seropositive compared with 28% (95% CI: 8-43%) of seronegative patients achieved remission at least once. This corresponded with a nearly 3 fold difference in achieving remission at least once (hazard ratio: 2.91; 95% CI 1.34-6.31) for seropositive vs. seronegative patients. Trends in SJC, TJC, DAS28-CRP, CDAI, HAQ, SDAI, and pain during the follow-up period did not differ significantly over time between groups (data not shown). If the Boolean remission definition was modified to exclude patient global score, there was no apparent difference in time to achieve remission for seropositive compared to seronegative patients (HR: 1.15;95% CI: 0.71-1.86).

Figure 1:

Time to remission for seropositive (solid line) and seronegative (dashed line) patients with rheumatoid arthritis. RA = rheumatoid arthritis.

Discussion

Early diagnosis and treatment initiation is crucial for improving patient outcomes in RA. To our knowledge, this is the first study to quantify the delay to meeting criteria as well as demonstrate a delay to treatment initiation in the seronegative RA population from the time of first documented joint swelling. The present study demonstrates that in our population, patients with seronegative RA experience a delay in meeting both 1987 and 2010 ACR/EULAR criteria. The median delay was 46 days longer for the 1987 and 14 days longer for the 2010 criteria among seronegative patients. This finding agrees with existing literature showing that in European early arthritis cohorts, seropositive patients are more likely to fulfill criteria early in the disease process (10). Furthermore, in our study population, seronegative patients experience a longer time to clinical diagnosis of RA, as well as to first DMARD treatment by a median of 40 days from the time of first clinically recognized synovitis, as compared to 14 days for seropositive patients.

Prior work has shown evidence for a window of opportunity, during which initiation of treatment is most beneficial for long-term outcomes (5). This window has been estimated to comprise the first 12-20 weeks of symptoms (5, 11–12). In a United Kingdom early arthritis cohort, delay from specialist assessment to start of DMARD therapy was reported, and attributed to diagnostic uncertainty and greater likelihood diagnosing undifferentiated arthritis in seronegative patients (13). Our findings that more seronegative than seropositive patients took ≥12 and ≥16 weeks to meet 1987 and 2010 criteria after first synovitis, and that the median time to a clinical RA diagnosis was over 26 weeks from first synovitis for seronegative patients, suggest that a clinically meaningful delay in meeting classification criteria may occur more frequently in this group. Among seropositive patients, a higher percentage of single-positive patients fulfills 2010 criteria ≥12 and ≥16 weeks from first joint swelling suggesting more delay in this group compared with double-positive patients, however this finding did not reach statistical significance. This result is consistent with previous work (14) which showed a higher likelihood of double-positive patients meeting classification criteria early in the disease process.

In our study, the median delay in initiating DMARD therapy was less than four weeks longer for seronegative versus seropositive patients and, despite this delay being shorter than the previously described window of opportunity, seronegative patients were less likely to achieve remission during the follow-up interval. Though this study did not address underlying reasons for treatment delay, we conclude that delay in meeting classification criteria and in receiving a clinical diagnosis of RA may reflect diagnostic uncertainty in the seronegative group and affect initiation of DMARD therapy. Methotrexate was used as first DMARD in a smaller percentage of seronegative RA patients despite meeting classification criteria, though this difference was not significant. Time from first joint swelling to first rheumatologist visit was not different between groups, suggesting that referral is not contributing to the delay in this population.

Data is mixed regarding disease activity at the time of diagnosis when comparing seronegative and seropositive RA patients, though recent data suggest a greater extent of disease in seronegative patients despite similar time from symptom onset (15). This may be a consequence of the 2010 criteria’s stronger weighting of autoantibodies than other domains. In our study, seronegative patients had more extensive joint involvement in order to fulfill the 2010 classification criteria. However patients were no different with regard to prevalence of elevated inflammatory markers, erosive disease, or duration of symptoms before meeting classification criteria, and disease activity markers such as HAQ, CDAI, SDAI, and DAS28-CRP were comparable between groups over time. This is in contrast to data from the Canadian Early Arthritis Cohort, in which seronegative subjects with early arthritis higher DAS28 scores and prevalence of radiographic erosive disease in addition to higher mean swollen and tender joint counts, despite shorter disease duration from symptom onset (14).

Previous research suggests that seronegative patients respond better to therapy and are less likely to develop new destructive joint disease at two years (14). Seronegative patients in the Canadian Early Arthritis Cohort had greater improvement in disease activity measures than seropositive patients over the follow-up interval (14). Our findings differ in that seronegative patients had similar core measures of disease throughout the follow-up interval, as indicated by SJC, TJC, DAS28-CRP(3) and pain and HAQ scores, and yet were less likely to achieve remission when the Boolean definition of remission was used. This is in contrast to the Canadian cohort, in which autoantibody status had no significant impact on achievement of remission, though remission was defined differently (as DAS28 <2.6) and the follow-up interval shorter (14). The current results suggest that the patient global score, a reflection of impact of disease on the patient, is the component of Boolean remission that is adversely affected more in the seronegative group in our population. A “dual target” for therapy in RA has been proposed, suggesting that disease activity, as indicated by TJC, SJC, acute phase reactants can be addressed separately from impact of disease on the patient, as indicated by patient global (16). The lower likelihood of achieving remission during the 5-year follow-up interval in this cohort, though with comparable measures of disease activity (SJC, TJC, CRP) over time, suggests that the delay in classification, clinical diagnosis, and/or treatment initiation observed in seronegative patients may affect their experience of disease impact despite no obvious difference in joint involvement or inflammatory markers over time. Further work is needed to determine whether earlier diagnosis and treatment in seronegative RA patients would reduce the persistent impact of disease. Additionally, achievement of near-remission (16) was not examined in this study and would be an interesting topic of future research.

Strengths of this study include median 4.1 years of follow-up and a geographically based cohort, in which patients have local access to both primary and specialty care making long delay to evaluation uncommon. Use of the Rochester Epidemiology Project allowed for a comprehensive review of documented patient visits amongst local medical facilities, minimizing loss of follow up data. Use of provider-documented synovitis as a baseline may minimize potential impact of recall bias, which could occur when using patient-reported symptom onset as in other early arthritis cohorts (4, 14).

This study has several limitations, including a sample size of 214 patients, retrospective design, and a primarily Caucasian population which may limit generalizability to other more diverse groups. Achievement of remission based on clinical documentation from return visits may introduce bias, as patients with fewer symptoms may return less frequently, leading to delay in documentation of remission. Enrollment was based upon meeting ACR/EULAR classification criteria, potentially including patients whose diagnosis changed in the future, though careful review at time of last follow-up showed diagnosis of RA remained valid for 212 of 214 patients. Additionally data from 2009, prior to publication of the 2010 ACR/EULAR classification criteria were included. A sensitivity analysis removing the 24 patients who fulfilled criteria in 2009 revealed no changes in the results, and the lack of effect of the publication date of the 2010 criteria is also supported by the observed delay in meeting 1987 criteria, and the delay in clinical diagnosis. Another potential limitation includes the use of a baseline of first synovitis recognized by a provider, rather than first symptoms, which could underestimate the time between the fulfillment of criteria and initiation of DMARDs in those delayed in seeking medical attention. The fact that first synovitis was as documented by any provider, and not only a rheumatologist, was used in order to minimize impact of potential referral delay, however could potentially introduce error particularly in those with subtle findings which may be more accurately detected by a trained rheumatologist. Similar proportions of each of the seronegative and seropositive groups had first synovitis documented by a rheumatologist, suggesting no important effect of healthcare provider on first documented synovitis. The provider’s degree or level of experience was not recorded for the purposes of this study and this would be of interest for future projects.

Conclusion

This study demonstrates that seronegative patients experience a delay in meeting ACR/EULAR classification criteria, receiving a clinical diagnosis of RA, and initiating treatment compared with seropositive patients. Seronegative patients are also less likely to achieve remission in the first five years after meeting classification criteria, though the two groups are similar with regard to achievement of remission when the patient global score is removed from the Boolean definition of remission. These findings are in agreement with existing literature demonstrating that the 2010 ACR/EULAR criteria are more sensitive in identifying early seropositive disease. Further work is necessary to identify factors that contribute to early diagnosis and treatment initiation in RA, particularly for patients with seronegative RA, in order to improve patient outcomes for all subgroups.

Abbreviations and Acronyms

ACPA

Anti-citrullinated peptide antibody

ACR

American College of Rheumatology

CRP

C-reactive protein

DAS28

Disease activity score

DMARD

Disease-modifying antirheumatic drug

EULAR

European League Against Rheumatism

HAQ

health assessment questionnaire

RA

rheumatoid arthritis

RF

rheumatoid factor

SJC

swollen joint count

TJC

tender joint count