Prevalence and characteristics of adults with difficult-to-treat rheumatoid arthritis in a large patient registry

Abstract

Objectives

An estimated 5–20% of patients with rheumatoid arthritis (RA) fail multiple treatments and are considered ‘difficult-to-treat’ (D2T), posing a substantial clinical challenge for rheumatologists. A European League Against Rheumatism (EULAR) task force proposed a definition of D2T-RA in 2021. We applied EULAR’s D2T definition in a cohort of patients with established RA to assess prevalence, and we compared clinical characteristics of participants with D2T-RA with matched comparisons.

Methods

Data from the longitudinal Brigham and Women’s Hospital Rheumatoid Arthritis Sequential Study (BRASS) registry were used. Participants were classified as D2T if they met EULAR’s definition. A comparison group of non-D2T-RA patients were matched 2:1 to every D2T patient, and differences in characteristics were evaluated in descriptive analyses. Prevalence rates of D2T were estimated using Poisson regression.

Results

We estimated the prevalence of D2T-RA to be 14.4 (95% CI: 12.8, 16.3) per 100 persons among 1581 participants with RA, and 22.3 (95% CI: 19.9, 25.0) per 100 persons among 1021 who were biologic/targeted synthetic DMARD experienced. We observed several differences in demographics, comorbidities and RA disease activity between D2T-RA and non-D2T-RA comparisons. Varying EULAR sub-criteria among all participants in BRASS resulted in a range of D2T-RA prevalence rates, from 0.6 to 17.5 per 100 persons.

Conclusion

EULAR’s proposed definition of D2T-RA identifies patients with RA who have not achieved treatment targets. Future research should explore heterogeneity in these patients and evaluate outcomes to inform the design of future studies aimed at developing more effective RA management protocols.

Rheumatology key messages.

• This study uses a robust longitudinal RA registry to identify patients with difficult-to-treat RA.

• EULAR’s definition identifies RA patients who have not achieved targets of low disease activity.

• Difficult-to-treat RA patients have evidence of disease activity and inflammation, though substantial heterogeneity exists.

Introduction

Rheumatoid arthritis (RA) is a chronic inflammatory disease that impacts 0.3–1.0% of people worldwide, and ∼1.3 million adults in the USA [1, 2]. The past two to three decades have observed several advancements in RA management, in part due to the increasing availability of new biologic and/or targeted synthetic (b/ts)DMARDs [3]. Despite these advancements, a subset of patients with RA remains persistently disabled with chronic pain and functional limitations, resulting in cycling through multiple b/tsDMARD therapies without substantial improvement in symptoms or disease activity [4]. Approximately one-sixth of patients with RA have been treated with five classes of medication with different mechanisms of action [5].

Patients who are non-responsive to different treatments pose a substantial clinical challenge for rheumatologists, and there is a paucity of evidence on how to identify and manage their condition more effectively. Historically, these patients were referred to as having refractory RA, or treatment resistant RA. The phrase ‘difficult-to-treat’ (D2T) first appeared in 2018 and was defined as persistent signs and symptoms and treatment with at least two bDMARDs [6]. To better define D2T-RA, a survey of 410 rheumatologists from 33 countries was conducted in 2017–2018 to identify key characteristics for classifying patients [7]. After convening a task force and reviewing results of the international survey and existing literature, EULAR provided guidance in 2021 [8] to identify three main criteria for D2T-RA: DMARD failure, signs suggestive of active/progressive disease, and management of symptoms viewed as problematic according to the rheumatologist and/or patient. In addition, EULAR noted that heterogeneity in patients classified as D2T-RA may arise from persistent inflammatory activity and/or from non-inflammatory symptoms. Non-inflammatory conditions could include fibromyalgia, osteoarthritis and psychosocial factors related to coping may lead to symptoms such as pain, that may mimic inflammatory processes. EULAR’s proposed definition of D2T-RA includes patients with both inflammatory and non-inflammatory processes.

This work is a much-needed advancement towards standardizing a definition of D2T-RA, and will facilitate future research into new insights for individuals with RA. Several studies have attempted to apply portions of EULAR’s proposed definition and have observed a broad range of prevalence of D2T-RA of 5–34% [5, 6, 9–24]. It is important to note that these studies have defined D2T-RA using only subcomponents of EULAR’s definition, which may explain the broad amount of variability observed in prevalence.

Therefore, our first objective was to fully implement EULAR’s D2T-RA definition using validated and widely available measures in the field of rheumatology and evaluate prevalence and characteristics compared with matched comparisons. Our second objective was to evaluate prevalence of D2T-RA using sub-components of EULAR’s definition to observe whether varying criteria may explain the broad range of prevalence reported in the literature. To address this objective, we used data from the Brigham and Women’s Rheumatoid Arthritis Sequential Study (BRASS) registry, a rich and robust longitudinal observational cohort of patients with RA with comprehensive assessments of disease activity.

Methods

Participants

The BRASS registry is a single-centre, prospective observational cohort of adult patients with RA. Enrolment began in 2003 for patients of the Brigham and Women’s Hospital (BWH) Arthritis Center in Boston, MA who had a diagnosis of RA using American College of Rheumatology RA classification criteria that was current at the time or based on the opinion of their rheumatologist. Participants completed annual visits with their rheumatologists and interim 6-month questionnaires. Details related to the BRASS registry have been reported elsewhere [25]. All patients provided informed consent, and approval was obtained by the Institutional Review Board of Brigham and Women's Hospital (2002P001762). The BRASS clinical trial registration number is NCT01793103. The present study included participants who enrolled in BRASS between 2003 and 2019, with the follow-up period extending through 2022.

RA medication use

Medication use, including conventional synthetic (cs)DMARD and b/tsDMARD therapies, and glucocorticoid use, were collected from participants every 6 months during study follow-up. We classified b/tsDMARD use as TNF inhibitors (etanercept, adalimumab, infliximab, golimumab, certolizumab pegol), anti T cell co-stimulator (abatacept), anti-IL-1 inhibitors (anakinra), B cell depletion (rituximab), anti-IL-6 inhibitors (tocilizumab, sarilumab), and Janus kinase inhibitors (tofacitinib, baricitinib, upadacitinib). Use of glucocorticoids, such as prednisone or methylprednisolone, was assessed at each visit and expressed as the most common daily dosage in the past 6 months and recorded in categories of prednisone equivalents: 1–5 mg, 6–10 mg, 11–20 mg and >20 mg.

Disease activity

Rheumatologists completed an annual 28-joint DAS–CRP-3 (DAS28-CRP3) [26] including a tender joint count and swollen joint count, a composite score of physician-reported 28 joint count and CRP laboratory values. A Clinical Disease Activity Index (CDAI) was assessed annually with scores >10 indicative of active RA.

Bilateral hand and wrist radiographs were performed at enrolment and every 2 years during follow-up, and were scored by four BWH radiologists according to the Sharp–van der Heijde method [27, 28]. Erosions were scored for 16 joints on each side of the body ranging from 0 = no erosions to 5 = complete collapse of bone (total erosion score range: 0–160). A total erosion score summed erosions in hands and wrists. Joint space narrowing (JSN) was scored on 15 joints on each side of the body in each hand and wrist and ranged from 0 = normal to 4 = bony ankylosis. A total JSN score summed across hands and wrists with range 0–120 [28, 29]. The Sharp–van der Heijde score (SHS) was computed as total erosion score + total JSN score; range 0–280. The intra-class correlation coefficient for the SHS in our study was 0.93 for baseline and 0.85 for change scores and was calculated from the scores of two Brigham radiologists who read 90% of the radiographs [30].

Extra-articular manifestations of RA experienced by the participant in the past year were recorded by rheumatologists, and included pericarditis, pleuritis, pulmonary fibrosis, bronchiectasis, pulmonary nodules, bronchiolitis obliterans with organizing pneumonia, subcutaneous rheumatoid nodules, Sjögren’s syndrome/keratoconjunctivitis sicca, large granular lymphocytic leukaemia, cervical myelopathy, neuropathy, Felty’s syndrome, cutaneous vasculitis, pyoderma gangrenosum, scleritis/episcleritis, glomerulonephritis, vasculitis of other organs, amyloidosis and lymphadenopathy. Routine Assessment of Patient Index Data (RAPID3) [31] is a patient-reported measure of RA severity that includes three subcomponents on the multidimensional Health Assessment Questionnaire (mHAQ) [32] for function, pain and global estimate of status. The mHAQ was assessed every 6 months. Standard cut-offs of disease activity on the RAPID3 were remission (≤3.1), low disease activity (3.1–6), moderate disease activity (6.1–12) and high disease activity (>12).

Patient and provider global RA severity

Participants provided a global score of arthritis activity over the prior 6 months. Global scores were assessed every 6 months during study follow-up and ranged from zero (arthritis was not at all active) to 10 (arthritis was extremely active) [33, 34]. Scores of 3 or higher were indicative of patient perception of active RA disease. In addition, rheumatologists completed an annual global assessment of disease activity that ranged from 0 (no arthritis activity) to 10 (severe arthritis activity) [35]. Scores of 3 or higher were indicative of provider perception of active RA disease.

Covariates

Demographics (age, sex, education, ethnicity), RA disease duration, comorbid medical conditions (diabetes mellitus, hypertension, cardiovascular disease [angina, heart attack and heart failure], and depression), age at RA onset, rheumatoid factor and anti-CCP positivity were assessed at enrolment. BMI and smoking status were assessed at annual visits, and we used a last value carry forward approach to fill in missing values at interim 6-month and missed visits.

D2T-RA and non-D2T-RA comparisons

We mapped available study variables to EULAR D2T-RA criteria and identified a subset of participants who met all three components of DMARD failure, persistent disease activity and perception of a persistent problem within a study visit. A minimum of 3 months of b/tsDMARD treatment duration was required. Details of EULAR’s D2T-RA algorithm and mapped BRASS variables are provided in Table 1. We matched each D2T-RA participant to two participants with RA who were b/tsDMARD experienced but did not meet all three D2T criteria at the same study visit using the MatchIt package [36]. Exact matching on study visit was used to ensure balance in duration of enrolment in the study and temporality given that the study period spanned 20 years, and index date for matched comparisons was defined as the visit date when the D2T-RA participant met EULAR criteria. We performed matching without replacement within the study visit; however, non-D2T-RA comparisons could be matched to a different D2T-RA participant at subsequent visits. Matched non-D2T-RA comparisons could partially meet D2T-RA criteria because we considered any b/tsDMARD experienced participants not meeting all three of EULAR’s criterions within a study visit to serve as a potential comparison.

Table 1.

Description of EULAR criteria for difficult-to-treat rheumatoid arthritis [8]

Criterion	EULAR criterion	BRASS definition
DMARD failure	Criterion 1: failure of ≥2 b/tsDMARDs with different mechanisms of action after failing conventional synthetic DMARD therapy	Patient reported treatment with ≥2 b/tsDMARDs with different mechanisms of action for at least three months
Persistent disease activity	Criterion 2: at least one of the following conditions (a–e) are present:
	(a) At least moderate disease activity such as DAS28-ESR >3.2 or CDAI >10	DAS28-CRP >3.2 or CDAI >10
	(b) Signs (acute phase reactants and imaging) and/or symptoms suggestive of active disease	Presence of extra-articular manifestations in the past year
	(c) Inability to taper glucocorticoid treatment below 7.5 mg/day prednisone or equivalent	Glucocorticoid most frequent average dose in past 6 months ≥6 mg
	(d) Rapid radiographic progression (with or without signs of active disease)	Change in Sharp score of 5 or more points
	(e) Well-controlled disease according to above standards but still having RA symptoms that are causing a reduction in quality-of-life	RAPID3 score >3 (moderate/high disease severity) and does not meet criteria 2a–d
Perception of a persistent problem	Criterion 3: management of RA and signs/symptoms are perceived as problematic by the rheumatologist and/or the patient	Patient Global Disease Activity score >3 or Provider Global Arthritis Activity Scale >3

b/ts DMARD: biologic and/or targeted synthetic DMARD; BRASS: Brigham and Women’s Hospital Rheumatoid Arthritis Sequential Study; CDAI: Clinical Disease Activity Index; DAS28: 28-joint DAS.

Statistical analysis

Differences in demographic and clinical characteristics between D2T-RA and non-D2T-RA comparisons were evaluated using χ² tests or Fisher’s exact tests for categorical variables or Student’s t-test or Wilcoxon’s rank sum tests for continuous variables. We constructed an UpSet plot for visualizing patterns in the second EULAR sub-criterion using the ‘UpSetR’ package [37]. An UpSet plot is used to visualize patterns of cross tabulations, called intersections, across multiple variables. Prevalence rates and 95% confidence intervals for D2T criteria were computed using Poisson regression with robust standard errors in the ‘sandwich’ package (version 3.1–0) [38]. Results are presented per 100 persons. The cumulative proportion of participants classified as D2T and using any b/tsDMARD was computed with the denominator defined as the total number of enrolled participants across all years. All analyses were conducted in R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria) [39].

Results

A total of 1581 patients with RA were enrolled in BRASS between 2003 and 2019; of these, 1021 (65%) were b/tsDMARD experienced at enrolment or used one of these agents during follow-up. We identified 228 participants who met EULAR D2T-RA criteria, representing 14% of all BRASS enrolees and 22% of those with any b/tsDMARD experience.

Demographic and clinical characteristics of 228 participants with RA classified as D2T, and 456 matched non-D2T comparisons are presented in Table 2. The mean age at enrolment among those with D2T-RA was 53 years, 90% were female, 93% White, 60% graduated from college or graduate school, 71% were RF positive and 70% were anti-CCP positive. The median time from study enrolment until developing D2T-RA was 48 months (about 4 years). Participants classified as D2T-RA were younger at RA diagnosis, had lower education, a greater proportion were female, a greater proportion had a history of hypertension, cardiovascular disease and depression, and had a greater BMI and significantly more disease activity than non-D2T-RA participants across multiple metrics. A greater proportion of participants classified as D2T had been treated with two DMARD classes compared with non-D2T-RA participants. While differences in the proportion of participants who were TNFi experienced in D2T-RA and non-D2T comparisons were not statistically significant, a greater proportion of D2T-RA were experienced across other b/tsDMARDs and csDMARDs except for methotrexate use, which was not statistically significant between the groups.

Table 2.

Demographic and clinical characteristics of BRASS participants classified as difficult-to-treat rheumatoid arthritis and matched non-difficult-to-treat rheumatoid arthritis comparisons

Characteristics	D2T-RA, EULAR (N = 228)	RA, non-D2T (N = 456)	P-value
Characteristics at baseline
Age at enrolment, mean (s.d.), years	53.3 (11.8)	52·4 (12.5)	0.571
Gender, n (%)			0.006
Male	23 (10%)	83 (18%)
Female	205 (90%)	373 (82%)
Race, n (%)
White	211 (93%)	425 (93%)	0.058
Black	12 (5%)	11 (2%)
Othera	5 (2%)	20 (5%)
Hispanic	6 (3%)	7 (2%)	0.376
Education categories, n (%)			0.075
High school diploma or less	37 (16%)	65 (14%)
At least some tech college or prof school	54 (24%)	75 (16%)
Graduated college	70 (31%)	152 (33%)
Graduate school	67 (29%)	164 (36%)
Comorbidities, n (%)/N
Diabetes mellitus	12 (5%)/228	18 (4%)/452	0.443
Hypertension	84 (37%)/228	121 (27%)/452	0.007
Cardiovascular disease (angina, heart attack, heart failure)	28 (12%)/228	30 (7%)/451	0.013
Depression	55 (24%)/228	65 (14%)/454	0.002
Age at RA diagnosis, mean (s.d.), years	39.8 (13.5)	42.3 (13.5)	0.033
Rheumatoid factor positivity, n (%)/N	147 (71%)/208	276 (67%)/414	0.312
ACPA positivity, n (%)/N	150 (70%)/213	275 (65%)/424	0.160
Characteristics at time of D2T
Time in months from study enrolment until index date, median (IQR)	48 (24–96)	48 (24–96)	1.000
Cumulative number of b/tsDMARD classes taken, n (%)			<0.001
One	0	439 (96%)
Two	196 (86%)	15 (3%)
Three	27 (12%)	2 (>1%)
Four	5 (2%)	0
Prior b/tsDMARD classes taken, n (%)
TNF inhibitors	224 (98%)	439 (96%)	0.158
Anti-T cell-co-stimulation	149 (65%)	15 (3%)	<0.001
B cell depletion	39 (17%)	5 (1%)	<0.001
Anti-IL-1	9 (4%)	0	<0.001
Anti-IL-6	36 (16%)	9 (2%)	<0.001
JAK inhibitors	36 (16%)	9 (2%)	<0.001
Methotrexate experienced, n (%)	208 (91%)	408 (89%)	0.470
Leflunomide experienced, n (%)	88 (39%)	91 (20%)	<0.001
Sulfasalazine experienced, n (%)	74 (32%)	144 (32%)	0.816
Hydroxychloroquine experienced, n (%)	151 (66%)	266 (58%)	0.046
DAS28-CRP score, n (%)			<0.001
≤3.2	72 (40%)	277 (76%)
3.3–5.1	82 (45%)	77 (21%)
>5.1	28 (15%)	11 (3%)
Missing, n	46	91
CRP, median (IQR)	2.9 (1.0–7.0)	1.4 (0.5–3.7)	<0.001
CDAI categories, n (%)			<0.001
≤10	45 (23%)	270 (70%)
>10	151 (77%)	115 (30%)
Missing, n	32	71
Average daily glucocorticoid use in the past 6 months, n (%)			<0.001
<6 mg/day	126 (64%)	359 (92%)
≥6 mg/day	70 (36%)	31 (8%)
Missing, n	32	66
Extra-articular manifestations in the past yearb, n (%)	50 (22%)	65 (14%)	0.011
RAPID3 score categories, n (%)			<0.001
≤3 (Remission)	22 (10%)	173 (41%)
>3 (low/moderate/high disease severity)	196 (90%)	251 (59%)
Missing, n	10	32
Tender joint count, median (IQR)	4.0 (1.0–11.0)	0 (0–3.0)	<0.001
Swollen joint count, median (IQR)	3.0 (0–9·0)	0 (0–2.0)	<0.001
Physician global arthritis activity score, category, n (%)
0–2	73 (36%)	311 (76%)	<0.001
≥3	132 (64%)	100 (24%)
Missing, n	23	45
Patient arthritis activity score in past 6 months, category, n (%)			<0.001
0–2	17 (8%)	208 (49%)
≥3	199 (92%)	213 (51%)
Missing, n	12	35
BMI, mean (s.d.), kg/m2	28.0 (6.6)	26.6 (6.1)	0.003
Missing BMI, n	6	11
Smoking status, n (%)			0.285
Never	117 (55%)	250 (60%)
Former	86 (40%)	144 (34%)
Current	10 (5%)	26 (6%)
Missing, n	15	36

^aOther includes Asian, Pacific Islander, Native American or any other background, including multiple races.

^bIncludes pericarditis, pleuritis, pulmonary fibrosis, bronchiectasis, pulmonary nodules, bronchiolitis obliterans with organizing pneumonia, subcutaneous rheumatoid nodules, Sjögren’s syndrome/keratoconjunctivitis sicca, large granular lymphocytic leukaemia, cervical myelopathy, neuropathy, Felty’s syndrome, cutaneous vasculitis, pyoderma gangrenosum, scleritis/episcleritis, glomerulonephritis, vasculitis of other organs, amyloidosis, and lymphadenopathy. BRASS: Brigham and Women’s Hospital Rheumatoid Arthritis Sequential Study; DAS28: disease activity score in 28 joints; D2T: difficult-to-treat; IQR: interquartile range; JAK: Janus kinase; RAPID3: routine assessment of patient index data.

Comparisons of current RA-related treatments between D2T-RA and non-D2T-RA groups are presented in Table 3. A greater proportion of participants with D2T-RA were currently treated with abatacept, rituximab, tocilizumab, tofacitinib or upadacitinib compared with non-D2T-RA participants. Additionally, a lower proportion of D2T-RA participants were concurrently treated with methotrexate, and a greater proportion were concurrently treated with prednisone than non-D2T-RA comparisons.

Table 3.

Current rheumatoid arthritis treatments of BRASS participants classified as difficult-to-treat rheumatoid arthritis and matched non-difficult-to-treat rheumatoid arthritis comparisons

Characteristic	D2T-RA, EULAR (n = 228)	RA, Non-D2T (n = 456)	P-value
Current b/tsDMARD therapy			<0.001
Adalimumab	9 (4)	157 (34)
Etanercept	6 (3.0)	216 (47)
Infliximab	5 (2)	41 (9)
Golimumab	1 (<1)	4 (1)
Certolizumab pegol	0	5 (1)
Abatacept	116 (51)	12 (3)
Rituximab	33 (14)	5 (1)
Anakinra	1 (<1)	0
Tocilizumab	24 (11)	9 (2)
Sarilumab	0	0
Tofacitinib	27 (12)	5 (1)
Baricitinib	0	0
Upadacitinib	6 (2.6)	2 (<1)
Current csDMARD therapy
Methotrexate	97 (43)	246 (54)	0.005
Leflunomide	13 (6)	18 (4)	0.298
Sulfasalazine	6 (3)	7 (2)	0.376
Currently taking prednisone	102 (45)	79 (17)	<0.001

Results are expressed as n (%). BRASS: Brigham and Women’s Hospital Rheumatoid Arthritis Sequential Study; b/tsDMARD: biologic/targeted synthetic DMARD; csDMARD: conventional synthetic DMARD; D2T: difficult-to-treat.

The prevalence of varying D2T criteria among all patients enrolled in BRASS and those treated with at least one b/tsDMARD during study follow-up is presented in Table 4. We observed a prevalence for D2T-EULAR of 14.4 per 100 persons (95% CI: 12.8, 16.3) among 1581 participants, and 22.3 per 100 persons (95% CI: 19.9, 25.0) among 1021 b/tsDMARD experienced participants. Applications of varying EULAR criteria resulted in a prevalence of D2T-RA that ranged from 4.7 to 17.5 per 100 persons among all participants, and 2.4 to 27.0 per 100 persons among b/tsDMARD experienced. A substantial number of participants were missed when definitions did not include DAS-28 or CDAI criteria. Additionally, 40 (3%) participants were classified as D2T based on DMARD only criteria but did not demonstrate evidence of both persistent disease activity and provider/patient perception of a problem.

Table 4.

Prevalence rates (95% CI) of difficult-to-treat rheumatoid arthritis overall definition and sub-criteria among participants with established RA

D2T definition by EULAR D2T-RA sub-criteria	Prevalent cases	Prevalence per 100 persons (95% CI)
		Participants with RA (n = 1581)	Participants with RA treated with a b/tsDMARD (n = 1021)
D2T EULAR (EULAR criteria [EC] 1, 2 and 3 in Table 1)	228	14.4 (12.8, 16.3)	22.3 (19.9, 25.0)
DMARD criteria only (EC 1)	276	17.5 (15.7, 19.4)	27.0 (24.4, 29.9)
DMARD + persistent disease activity (EC 1 and 2)	243	15.4 (13.7, 17.3)	23.8 (21.3, 26.6)
DMARD + DAS-28 or CDAI (EC 1 and 2a)	202	12.8 (11.2, 14.5)	19.8 (17.5, 22.4)
DMARD + EAM (EC 1 and 2b)	119	7.5 (6.3, 8.9)	11.7 (9.8, 13.8)
DMARD + glucocorticoid (EC 1 and 2c)	147	9.3 (8.0, 10.8)	14.4 (12.4, 16.7)
DMARD + radiologic progression (EC 1 and 2d)	25	1.6 (1.1, 2.3)	2.4 (1.7, 3.6)
DMARD + RA symptoms (EC 1 and 2e)	74	4.7 (3.7, 5.8)	7.2 (6.8, 9.0)
DMARD + perception of a problem (EC 1 and 3)	260	16.4 (14.7, 18.4)	25.5 (22.9, 28.3)

Prevalence and 95% CI assessed using Poisson regression with robust standard errors. EULAR criteria and sub-criteria definitions are presented in Table 1. b/tsDMARD: biologic or targeted synthetic DMARD; CDAI: Clinical Disease Activity Index; DAS-28: disease activity score in 28 joints; D2T: difficult-to-treat; EAM: extra-articular manifestations; EC: EULAR criteria (see Table 1).

We evaluated the prevalence in D2T-RA and b/tsDMARD use over time. The proportion of participants classified as D2T was <1% in the early years of the study (2003–2006) which increased steadily to 14% in 2022. The proportion of participants treated with a b/tsDMARD increased from 43% to 61% from 2003 to 2009, eventually reaching 65% in 2022.

As EULAR’s criteria for persistent disease activity had multiple subcomponents, we were interested in understanding which subcomponents were the strongest drivers of D2T-RA (Fig. 1). We observed that sub-criteria 2a, or moderate disease activity via DAS-28 or CDAI scales had the greatest prevalence (n = 161/228, 71%). Additionally, 90 (43%) patients qualified as D2T based on DAS and/or CDAI criteria only. Additional sub-criteria 2 b–d were less prevalent. Finally, 18 (8%) qualified as D2T based on sub-criterion 2e (RAPID3 score > 3).

Figure 1.

Upset plot of the 15 most frequent intersections in EULAR sub-criterion 2. This Plot demonstrates the 15 most prevalent mutually exclusive combinations of sub-criteria to demonstrate how patients satisfied the EULAR requirement for persistent disease activity. The denominator for this figure is 228 participants meeting EULAR D2T-RA criteria. Black dots depict which criteria were being met, and black bars indicate the number of participants meeting each combination of sub-criteria. Sub-criteria included: a: moderate disease activity using Clinical Disease Activity Index (CDAI) >10 or DAS28-CRP > 3.2; b: signs of active disease using presence of extra-articular manifestations; c: inability to taper glucocorticoid dose of >6 mg/day prednisone equivalents; d: rapid radiographic progression of Sharp score change of 5 or more points; and e: RAPID3 > 3 and does not meet criteria a–d

Discussion

We implemented EULAR’s proposed algorithm to identify D2T-RA participants and observed a prevalence of 14.2 per 100 persons in a general RA population, and 22.3 per 100 persons when subset to b/tsDMARD experienced RA participants. Additionally, we observed several differences in demographics, comorbidities and RA disease activity between D2T-RA and non-D2T-RA comparisons. Therefore, EULAR’s proposed definition can successfully identify a subset of RA patients who have not achieved targets of low disease activity despite multiple treatments.

This study aligns with prior literature that has observed a prevalence of D2T-RA of 5–34% [5, 6, 9–24]. We observed a similarly broad range of prevalence of 5–26% across varying sub-criteria of D2T-RA, with the largest prevalence for DMARD failure criteria only (EULAR component 1) among participants with RA (17%) and b/tsDMARD experienced (26%). Conversely, we observed the lowest prevalence when we applied DMARD failure and presence of extra-articular manifestations, glucocorticoid use, or RA symptoms despite no evidence of disease activity (EULAR components 1 and 2b–e; prevalence 5–12%). Given these results, we feel that the variability in prevalence observed across studies can likely be explained by population characteristics and varying definitions of D2T. DMARD failure serves as the cornerstone of EULAR’s D2T-RA definition and is defined as ‘two or more b/tsDMARDs with different mechanisms of action’. Using this definition alone, we observed a prevalence of 17.1 per 100 persons, which aligns with other studies such as Leon et al. [10], where 15% of participants met similar criteria. However, unlike our study, 23% of participants with DMARD failure met all three EULAR criterions in Leon et al. whereas we observed 83%. Differences in population characteristics could partially explain this, as the population in Leon et al. was early RA, and our participants were a mix of early and late RA. It is reasonable to expect that DMARD failure criteria may be a stronger indicator of D2T-RA in an established RA cohort. Other studies only applying DMARD failure criteria to identify D2T-RA observed prevalence of 13–34%, which likely overestimates the true prevalence of D2T-RA [9, 11, 12, 15]. Therefore, we feel that DMARD failure criteria alone are insufficient for identifying a D2T-RA population, resulting in excessive false positives.

There were aspects of EULAR’s criteria that required interpretation, and we attempted to translate these into measurable constructs using validated measures that are widely available in the field of rheumatology research. For example, component 2b, ‘signs (acute phase reactants and imaging) and/or symptoms suggestive of active disease’, is not concretely defined, although EULAR did specify that non-joint related complications, such as extra-articular manifestations, should be considered [8]. Some studies have interpreted this criterion as ESR ≥ 60 or CRP ≥ 0.5 but did not include assessments of extra-articular manifestations [10]. We used an extensive list of extra-articular manifestations assessed by a rheumatologist. Furthermore, we did not have access to ESR values in this study and did not include CRP as it was already incorporated in criterion 2a as part of the DAS28-CRP score. Criterion 2e, ‘well-controlled disease according to above standards but still having RA symptoms that are causing a reduction in quality-of-life’, also required some interpretation on our part. EULAR stated that the intention was to identify patients with non-inflammatory complaints [8]. Therefore, we interpreted this as participants who did not meet criteria 2a–2d and had a RAPID3 score greater than 3 indicating moderate/high disease severity. We chose to use RAPID3 as it is a composite of function, pain and personal assessment of well-being. Other studies have defined this as HAQ ≥1.2 [10]. We observed a prevalence of 4.7 per 100 persons for this criterion, suggesting that few participants with D2T-RA had evidence of non-inflammatory disease activity. EULAR should provide more detailed guidance on criterions 2b and 2e and future studies should attempt to validate alternative definitions.

Portions of EULAR’s D2T-RA algorithm rely on patient-reported outcomes of disease activity and symptoms, such as criterion 3. Subjective complaints of symptoms without objective evidence of active inflammatory disease could impact disease management and complicate care in patients with RA, especially patients with comorbid depression or anxiety, obesity, fibromyalgia, and pain or fatigue [40]. It is important for future research in D2T-RA to focus on mechanisms and sub-phenotypes in order to better identify more effective treatment and management options.

Some researchers have advocated for more stringent definitions of D2T-RA, such as failure in at least four classes of b/tsDMARDs [40]. According to our study, if a more stringent definition were applied, the prevalence of D2T-RA would be 1.6%. We feel this definition overlooks a large proportion of patients with RA with less b/tsDMARD experience who are not obtaining sufficient benefit from their current therapy. Some criticisms of EULAR’s definition are that a treat-to-target management strategy may drive more patients to be defined as D2T. However, the strength of EULAR’s definition is in the triad of DMARD failure, persistent disease activity and perception of a problem. Therefore, patients who switch therapies and achieve low disease activity targets are not likely to qualify as D2T. Future studies should continue to explore heterogeneity among D2T-RA, including patients who exhaust all current b/tsDMARD classes. We also encourage future research to apply EULAR’s definition in earnest, and if subcomponents are not available or feasible to implement, the results of our study provide some indication on the impact that these alternative definitions may have on the prevalence of D2T-RA. EULAR’s definition is notably broad and likely identifies a heterogeneous population of patients with RA, but is a starting point to guide future research towards gaining a better understanding of this subset of RA patients and will continue to evolve over time.

Our study has several limitations to note. There were missing data across several variables, due in part to the data collection schedule (annual vs semi-annual) and missed visits. We used a last value carry forward method for BMI and smoking status variables only and did not use imputation methods for variables included in the D2T-RA definition. Carrying forward approaches for components of the D2T-RA definition would not be appropriate as they would ignore potentially important changes in values that may have occurred between visits. Therefore, it is possible that some patients may have been misclassified as non-D2T due to missing data, although we feel this risk is minimal given the length of BRASS follow-up. Some criteria for D2T-RA relies on subjective, or patient-reported outcomes of symptoms and disease activity, and while these measures are widely used in the field, they are not objective assessments and care should be taken to interpret results. We dichotomized the patient visual analogue scale global arthritis activity scale using a cut point of ≥3, although to our knowledge, this cut point has not been validated in the literature. We chose this cut point to align with the provider global scale, under the assumption that there should be equivalency. Future studies should validate cut points in the patient global scale and assess whether there is comparability with the provider global scale. Also, we used DAS28-CRP in our definition of D2T-RA because EULAR specifically recommended using a validated disease activity measure (specifically DAS28-ESR) and we did not have ESR assessed in our study. Alternative definitions should be explored. Finally, this study was conducted in a cohort of patients with RA seen at the Brigham and Women’s Hospital Arthritis Center in the Boston area and results may not be generalizable to other patient populations and geographic areas.

In conclusion, among a cohort of 1581 participants with RA, we applied EULAR’s D2T-RA criteria and observed a prevalence of 14.2 per 100 persons. D2T-RA participants differed from non-D2T-RA comparisons across several demographic, comorbid and disease activity measures. These results suggest that EULAR’s proposed definition can successfully identify a subset of RA patients who have not achieved targets of low disease activity despite multiple treatments. Our study serves as a necessary next step in translating EULAR’s definition into a measurable definition that uses validated and widely available measures in the field of rheumatology research. Future studies should continue to validate this definition, explore alternative definitions, and explore heterogeneity in this important subset of patients with RA.

Funding

This work was supported by NIH P30-AR072577. The BRASS cohort is supported by Bristol Myers Squibb, Sanofi, Aqtual, and by Janssen. The funders of this study had no role in the design, data collection, data analysis, data interpretation, or writing of this manuscript.

Disclosure statement: M.L.P., R.L. and C.N.: no conflicts; D.H.S. receives salary support from research contracts to Brigham and Women’s Hospital from CorEvitas, Janssen and Novartis. He also receives royalties for unrelated UpToDate chapters on NSAIDs. M.E.W. has received research support from Abbvie, Aqtual, Bristol Myers Squibb and Janssen and is a consultant to Abbvie, Amgen, Aclaris, Bristol Myers Squibb, Glaxo Smith Kline, Horizon, Johnson and Johnson, Novartis, Pfizer, Sanofi, Scipher, Set Point, Tremeau, Rani Therapeutics, Revolo, and Sci Rhom. Stock options with Canfite, Scipher and Inmedix; N.S. received research grant support from BMS, Amgen, Eli Lilly, Mallinckrodt, Sanofi-Regeneron and Crescendo Biosciences.