High-Titer Rheumatoid Factor is Associated with Worse Clinical Outcomes and Higher Needs for Advanced Therapies in Rheumatoid Arthritis Under Real-Life Conditions

Abstract

Introduction

Rheumatoid factor (RF) plays an important role in rheumatoid arthritis (RA) pathophysiology, yet the differential effects of varying RF titers remain understudied. We evaluated associations between different RF titers and clinical outcomes in long-standing RA.

Methods

This multicenter, cross-sectional study included adults meeting ACR/EULAR (2010) criteria for RA. Circulating RF titers and clinical-epidemiological characteristics were evaluated. Bivariate (Student’s t and chi-squared tests) tests and multiple logistic and linear regression analyses were conducted.

Results

We included 1097 participants; 78.7% had positive RF, with high titers (≥ 3 × the upper limit of normality) in 56.2%. Negative vs. low-positive RF groups performed similarly concerning all clinical outcomes, being subsequently aggregated as "non-high" RF group. High RF titers (compared to "non-high") were associated with tobacco use (odds ratio, OR [95% confidence interval, CI]: 2.04 [1.35, 3.08]; p < 0.001), multiraciality (OR [95% CI] 1.31 [1.03, 1.67]; p = 0.028, compared to White race), and higher body mass index (mean difference [95% CI] 0.69 [0.05, 1.33] kg/m²; p = 0.033). In multivariate analyses, high-titer RF was independently associated with higher disease activity (Clinical Disease Activity Index, CDAI: β = 2.44 [0.89, 3.99], p = 0.002), worse functional capacity (Health Assessment Questionnaire Disability Index, HAQ-DI: β = 0.112 [0.018, 0.205], p = 0.020); extra-articular manifestations (OR 1.48 [1.09, 2.00], p = 0.011); increased corticosteroid (OR 1.53 [1.19, 1.96], p = 0.001) and biological disease-modifying antirheumatic drugs (bDMARD) use (OR 1.41 [1.08, 1.84], p = 0.011).

Conclusions

High RF titers in long-standing RA were associated with worse disease activity, lower physical functionality, increased extra-articular manifestations, and higher usage of corticosteroids and bDMARDs. Comparing high vs. non-high RF titers (rather than positive vs. negative RF) seems more useful for evaluating the clinical effects of RF in RA. This approach should be considered in future studies of RF.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40744-024-00730-w.

Key Summary Points

Why carry out the study?

Few studies have explored the effects of high-titer rheumatoid factor (RF) on clinical outcomes of rheumatoid arthritis (RA) in real-life context.

The present study aimed to evaluate RF titers and their effects on patients with long-standing RA under real-life conditions.

What was learned from the study?

Using a large sample of participants, we found that high-titer RF is associated with worse disease activity, lower physical functionality, increased extra-articular manifestations, and greater use of corticosteroids and biological disease-modifying antirheumatic drugs. Meanwhile, negative and low-titer RF performed similarly.

Dichotomizing RF titers into high (≥ 3 × the upper limit of normality, ULN) and non-high appears more clinically relevant and discriminative than merely using the ULN. This strategy may be useful in the design of further clinical studies and in therapeutic considerations of the management of patients with RA.

Introduction

Rheumatoid arthritis (RA) is a chronic autoimmune disease characterized by synovial joints inflammation with a prominent humoral response disorder distinguished by anti-citrullinated protein antibodies (ACPA) and against immunoglobulin G (IgG) Fc fragment, known as rheumatoid factor (RF). These autoantibodies may form immunocomplexes inside the joints and trigger complement activation, immune cell chemotaxis, and response amplification [1, 2].

The rheumatoid factor (RF) was the first autoantibody type found in RA. RF can be present in other autoimmune diseases (Sjögren’s syndrome, mixed connective tissue diseases, systemic lupus erythematosus…) but is also found in healthy individuals (1.3–4% in White population) or as a normal response to antigenic stimuli (endocarditis, tuberculosis, viral infections…) [2, 3].

Despite its low specificity, RF positivity is reliable in a high pretest probability context. Therefore, RF positivity has been part of RA classification criteria since the 1987 American College for Rheumatology (ACR) through to the 2010 criteria by ACR/European League Against Rheumatism (EULAR) [2, 4, 5].

Furthermore, RF has also proven its value as a drug response predictor. As autoantibody positivity denotes a B cell-driven response, RF, along with ACPA, have been associated with good rituximab response in some studies. There is also data to support better response in patients using abatacept or tocilizumab, however no association has yet been made between serological status and tumoral necrosis factor (TNF) inhibitor utilization [6–8].

In addition to its value in diagnosis and drug response, RF positivity is also widely perceived as a predictor of poor prognosis. The inclusion criteria for many randomized controlled trials list RF positivity as a poor prognostic factor when aiming for RA remission or non-progression [9]. EULAR 2022 recommendations for RA management endorse RF, especially at high titers, as an unfavorable predictor and use it as a marker to guide clinical decisions, i.e., whether to add a second disease-modifying antirheumatic drug (DMARD), change the first conventional synthetic DMARD (csDMARD) or to step up directly to a biological or targeted synthetic DMARD (bDMARD or tsDMARD) when RA is not controlled [10].

Most data regarding RF positivity as a poor prognosis predictor derives from scenarios where most variables were controlled and isolated, but external validation in real-life contexts still lacks. There is international interest in promoting studies to evaluate prognostic features across different populations in real-life conditions. For instance, this tendency can be noticed in many studies in America, Europe, Asia, and Africa [11–19]. However, most of these studies reported only if RF was positive or not, and there is a lack of studies reporting different RF titers and evaluating its association with clinical outcomes.

Therefore, the present study aimed to evaluate RF titers and their effects on patients with long-standing RA under real-life conditions.

Methods

This study was part of the Rheumatoid Arthritis in Real Life (REAL) study, a prospective multicenter cohort study, conducted in 11 Brazilian university hospitals [20]. The patients were enrolled from August 2015 to April 2016. Inclusion criteria were (1) meeting ACR/EULAR 2010 and/or ARA 1987 [4, 5] classification criteria for RA; (2) age ≥ 18; and (3) a minimum of 6 months follow-up in the healthcare center prior to data gathering.

The present study was cross-sectional, as it was derived from the baseline assessment of the REAL study only. Patients underwent a review of medical records, a structured clinical interview, and a physical examination.

Patients were assessed as to socioeconomic and demographic characteristics, current smoking, body mass index (BMI), RA treatment, activity, complications, disease duration and time from symptom onset to the first DMARD prescription. Race was self-reported. Black and Pardo Brazilians (a Brazilian multiethnic denomination category) share common racial, ethnic, and socioeconomic backgrounds. Given the similarity of these two groups and the underrepresentation of other races in our sample, these participants were grouped into a multiracial category for statistical analysis.

Erosive disease was defined as the presence of erosions (cortical breaks) in at least three separate joints on radiographs of both hands and feet, according to the definition proposed by the EULAR task force [21].

To assess disease activity, the chosen scores were Disease Activity Score-28 with ESR (DAS28-ESR), DAS28 with C-reactive protein (CRP), Clinical Disease Activity Index (CDAI). The Health Assessment Questionnaire Disability Index (HAQ-DI) was used to evaluate functional capacity. The Short Form-12 (SF-12) physical and mental subscales were used to assess quality of life (QoL).

The outcomes of interest (dependent variables) for the analytical tests comprised treatment (bDMARDs and corticosteroids usage), documented erosive disease, extra-articular manifestations, disease activity scores (DAS28-ESR, DAS28-CRP and CDAI), HAQ-DI and QoL scores (SF-12 physical and mental).

RF titers were initially grouped as negative (≤ upper limit of normality, ULN), low titers (positive, but < 3 × ULN) and high titers (≥ 3 × ULN), as defined by ACR/EULAR 2010 [5]. Statistical analyses were initially performed comparing negative and low positive titer groups. Afterwards, since the low-titer and negative RF groups performed similarly as to the outcomes of interest (see Results), RF titers were regrouped in high titers and non-high titers (merging negative and low-titer groups) for the subsequent analyses.

Concerning inferential statistics, bivariate analyses were initially performed. Continuous variables were compared using Student’s t test with Welch’s correction when appropriate. Categorical variables were compared using the chi-square test. Although many variables were non-normally distributed, parametric tests were adopted given the large sample size (based on the central limit theorem).

Categorical variables’ effect size measures were calculated using odds ratio (OR). As to continuous variables, effect sizes were expressed as differences in means (DM).

Following the bivariate analysis, multiple binary logistic and linear regression models were performed to evaluate the association of RF titers (high vs. non-high)—as the main predictor—and the outcomes of interest. The effects of RF titers were adjusted for covariates with a p value < 0.10 in the bivariate analysis. Statistical analysis was performed using SPSS 25. There was no missing data imputation. A p value < 0.05 was considered statistically significant.

Ethics

The study was approved by the National Commission of Ethics in Research (CONEP-Comissão Nacional de Ética em Pesquisa, Brazilian Ministry of Health; protocol number CAAE 45781015.8.1001.5259) and by institutional review boards of each participating center (listed in Supplementary Table 1). All patients provided written informed consent prior to inclusion in the study. The study was conducted in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments.

Results

In total, 1097 participants were included; female subjects: 89.4%; White race: 56.8%; current smokers: 10.9%. RF was positive in 78.7% with high titers in 56.2%. The median [interquartile range, IQR] disease duration was 152 [84–241] months and the delay from first symptoms to first DMARD was 12 [6–42] months. Erosive disease was found in 54.9%. General characteristics of the sample are shown in Table 1. Extra-articular manifestations present in the sample and their frequencies are listed in Supplementary Table 2.

Table 1.

General characteristics of the patients with rheumatoid arthritis

Characteristics		% (n) or Median [IQR]	n
Sex	Female	89.4% (981)	1097
	Male	10.6% (116)
RF titers	Negativea	21.3% (234)	1097
	Lowb	22.5% (247)
	Highc	56.2% (616)
Race	White	56.8% (623)	1097
	Pardod	31,3% (343)
	Black	10,9% (120)
	Others	1,0% (11)
Socioeconomic classe	A/B (higher)f	23.4% (253)	1083
	C/D/E (lower)g	76.6% (830)
Schooling (years)h		8 [4–11]	1058
Age (years)		58 [50–65]	1097
BMI (kg/m2)		26.60 [23.70–30.45]	1029
Tobaccoi		10.9% (120)	1097
Disease duration (months)j		152 [84–241]	1096
Time to first DMARD (months)j		12 [6–42]	994
Corticosteroid use		47.0 (516)	1097
bDMARD use		35.6 (391)	1097
Erosive diseasek		54.5 (589)	1081
Extra-articular disease		23.2 (253)	1090
HAQ-DI		0.875 [0.250–1.500]	1093
DAS28-ESR		3.52 [2.55–4.49]	911
DAS28-CPR		3.02 [2.22–4.21]	927
CDAI		9.0 [3.7–18.9]	1095
SF-12 physical		36.100 [31.660–41.740]	1061
SF-12 mental		47.180 [36.305–56.740]	1061

^a≤ upper limit of normality. ^bPositive, but < 3 × upper limit of normality. ^c≥ 3 × upper limit of normality. ^dBrazilian multiethnic individuals. ^eBrazilian economic classification criterion. Each stratum corresponds to an estimated household monthly income: ^fUS $1357.00 or higher. ^gLess than US$1357.00. Conversion of Brazilian reais into US dollars made in accordance with the exchange rate of April 16, 2016—US $1.00 = R$ 3.5376. ^hTotal formal education time. ⁱCurrent smoker. ^jFrom symptoms onset. ^kAccording to EULAR definition

RF rheumatoid factor, BMI body mass index, DMARD disease-modifying antirheumatic drug, bDMARD biological DMARD, HAQ-DI Health Assessment Questionnaire Disability Index, DAS28-ESR and -CRP scores Disease Activity Scores, 28 joints, based on erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP), respectively, CDAI Clinical Disease Activity Index, SF-12 Short-Form 12 subscales, IQR interquartile range [Q1–Q3]

Negative and low-positive RF titer groups showed no statistically significant differences across all outcomes of interest: disease activity (DAS28-ESR, DAS28-CRP, CDAI), functional capacity (HAQ-DI), QoL (SF-12 physical and mental subscales), bDMARD use, corticosteroids, erosive disease, or extra-articular manifestations (Supplementary Table 3).

Bivariate analysis revealed that high and non-high RF titer groups were comparable across most background characteristics: sex, age, disease duration, delay from first symptom to first DMARD, schooling years and economic class (p > 0.05 for all comparisons). Current smoking and belonging to the multiethnic group were significantly more likely in the high titer group [OR 2.04 (95% CI 1.35–3.08) and OR 1.31 (95% CI 1.03–1.67) respectively]. There was also a significantly higher BMI in the high titer group [DM 0.69 (95% CI 0.05–1.33)] (Table 2).

Table 2.

Bivariate comparisons between high and non-high RF titer groups as to general characteristics among patients with rheumatoid arthritis

Characteristics	RF high titers n = 616 % (n) or mean (SD)	RF non-high titers n = 481 % (n) or mean (SD)	Effect sizes [95% CI]a	p valuesb
Female sex	89.4% (551)	89.4% (430)	1.01[0.68–1.48]†	0.978
Multiracialc	46.1% (284)	39.5% (190)	1.31 [1.03–1.67]†	0.028*
Tobaccod	13.8% (85)	7.3% (35)	2.04 [1.35–3.08]†	0.001*
Lower socioeconomic classe	78.0% (476)	74.8% (354)	1.19 [0.9–1.58]†	0.218
Schooling (years)f	7.87 (4.21)	8.27 (4.34)	− 0.39 [− 0.91–0.13]‡	0.138
Age (years)	57.56 (10.92)	56.26 (12.11)	1.30 [− 0.09–2.68]‡	0.066
BMI (kg/m2)	27.69 (5.32)	27.00 (4.96)	0.69 [0.05–1.33]‡	0.033*
Time to first DMARD (months)g	37.89 (54.88)	37.77 (64.11)	−  0.12 [− 7.75–7.51]‡	0.976
Disease duration (months)g	172.09 (114.08)	175.60 (112.39)	− 3.51 [− 17.05–10.02]‡	0.611

^aExpressed as odds ratios† or difference in means‡ for categorical or continuous variables respectively. ^bOn Pearson’s chi-squared or Student’s t tests. ^cBlack and Pardo Brazilians and other races, compared to White race. ^dCurrent smoker. ^eClasses C, D, or E according to Brazilian economic classification criterion. Equivalent to a monthly household income lower than US $1357.00. Conversion of Brazilian reais into US dollars made in accordance with the exchange rate of April 16, 2016—US $1.00 = R$ 3.5376 ^fTotal formal education time. ^gFrom symptoms onset

RF rheumatoid factor, BMI body mass index, DMARD disease-modifying antirheumatic drug, SD standard deviation, 95% CI 95% confidence interval

^*p values significant at < 0.05

High RF titers were associated with greater use of corticosteroids and bDMARDs, higher frequency of extra-articular disease, and higher DAS28-ESR, DAS28-CRP, CDAI, and HAQ scores (Table 3). All these bivariate associations remained significant in the multiple regression models after adjusted for the covariates: BMI, age, race, and tobacco exposure. The effect size and significance of each variable in each regression model are listed in Table 4.

Table 3.

Bivariate comparisons between the high and non-high RF titer groups as to clinical outcomes in patients with rheumatoid arthritis

Outcomes	RF high titers n = 616 % (n) or mean (SD)	RF non-high titers n = 481 % (n) or mean (SD)	Effect sizes [95% CI]a	p valuesb
Corticosteroid use	52.4% (323)	40.1% (193)	1.65 [1.29–2.09]†	< 0.001*
bDMARD use	39.3% (242)	31.0% (149)	1.44 [1.21–1.86]†	0.004*
Erosive disease	56.1% (340)	52.4% (249)	1.16 [0.91–1.48]†	0.227
Extra-articular disease	26.0% (159)	19.6% (94)	1.44 [1.08–1.92]†	0.013*
HAQ-DI	0.999 (0.796)	0.872 (0.728)	0.127 [0.036–0.218]‡	0.006*
DAS28-ESR	3.80 (1.52)	3.37 (3.37)	0.43 [0.23–0.62]‡	< 0.001*
DAS 28-CPR	3.41 (1.40)	3.12 (1.32)	0.09 [0.11–0.47]‡	0.001*
CDAI	14.09 (13.21)	11.63 (11.04)	2.46 [1.02–3.90]‡	0.001*
SF-12 physical	36.446 (6.526)	36.253 (7.102)	0.194 [− 0.639–1.027]‡	0.648
SF-12 mental	45.299 (12.464)	46.591 (12.096)	− 1.292 [− 2.786–0.203]‡	0.090

^aExpressed as odds ratios† or difference in means‡ for categorical or continuous variables respectively. ^bOn Pearson’s chi-squared or Student’s t tests

RF rheumatoid factor, bDMARD biological disease-modifying antirheumatic drug, HAQ-DI Health Assessment Questionnaire Disability Index, DAS28-ESR and -CRP scores Disease Activity Scores, 28 joints, based on erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP), respectively, CDAI Clinical Disease Activity Index, SF-12 Short-Form 12 subscales, SD standard deviation, 95% CI 95% confidence interval. *p values significant at < 0.05

Table 4.

Adjusted effects of high-titer RF on clinical outcomes in multivariate analysis among patients with rheumatoid arthritis

Outcomes†	Main predictor	Covariates
	High-titer RFa	Age (years)	Multiracialb	Tobaccoc	BMI (kg/m2)
Corticosteroid use
OR [95% CI]	1.53 [1.19–1.97]	0.99 [0.98–1.00]	1.24 [0.97–1.60]	1.42 [0.94–2.14]	1.03 [1.00—1.05]
p value	0.001*	0.162	0.093	0.098	0.047*
bDMARD use
OR [95% CI]	1.41 [1.08–1.84]	0.99 [0.98–1.00]	0.93 [0.72–1.22]	1.41 [0.93–2.14]	1.04 [1.02–1.07]
p value	0.011*	0.088	0.608	0.109	0.001*
Extra-articular disease
OR [95% CI]	1.48 [1.10–2.00]	1.02 [1.00–1.04]	1.39 [1.03–1.86]	0.93 [0.57–1.52]	0.98 [0.95–1.01]
p value	0.011*	0.001*	0.030*	0.775	0.136
HAQ-DI
ß [95% CI]	0.112 [0.018–0.207]	0.009 [0.005–0.013]	− 0.010 [− 0.104–0.085]	− 0.78 [− 0.228–0.072]	0.012 [0.003–0.021]
zß	0.073	0.139	− 0.006	− 0.032	0.081
p value	0.020*	< 0.001*	0.843	0.310	0.009*
DAS28-ESR
ß [95% CI]	0.40 [0.19–0.60]	0.01 [0.00–0.02]	0.37 [0.16–0.57]	− 0.16 [− 0.49–0.16]	0.00 [− 0.02–0.02]
zß	0.13	0.08	0.12	− 0.03	0.01
p value	< 0.001*	0.026*	< 0.001*	0.326	0.700
DAS28-CRP
ß [95% CI]	0.27 [0.08–0.45]	0.00 [− 0.01–0.01]	0.25 [0.07–0.44]	0.02 [− 0.28–0.31]	0.01 [− 0.01–0.03]
zß	0.10	0.00	0.09	0.00	0.03
p value	0.005*	0.949	0.008*	0.917	0.318
CDAI
ß [95% CI]	2.44 [0.91–3.97]	0.01 [− 0.06–0.07]	2.77 [1.25–4.29]	− 2.01 [− 4.48–0.37]	− 0.6 [− 0.20–0.09]
zß	0.10	0.01	0.11	− 0.05	− 0.02
p value	0.002*	0.818	< 0.001*	0.096	0.437

^†A multiple binary logistic or linear regression model was fit for each of the categorical or continuous outcomes, respectively. Each model was adjusted for the predictor variables above. Effect sizes are expressed in OR or ß coefficients. ^aHigh-titer RF defined as ≥ 3 × upper limit of normality, compared to non-high titers (< 3 × upper limit of normality). ^bBlack and Pardo Brazilians and other races, compared to White race. ^cCurrent smoker

RF rheumatoid factor, BMI body mass index, bDMARD biological disease-modifying antirheumatic drug, HAQ-DI Health Assessment Questionnaire Disability Index, DAS28-ESR and -CRP scores Disease Activity Scores, 28 joints, based on erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP), respectively, CDAI Clinical Disease Activity Index, OR odds ratio, ß unstandardized beta-coefficient, zß standardized beta-coefficient *p values significant at < 0.05

Discussion

The present work consisted of a large transversal study that was able to show associations between high titers of RF and various clinical aspects within a long-standing RA population undergoing treatment in a real-life setting.

RF positivity in our study (78.7%) was similar to many real-life studies [11, 14, 17]. Regarding the prevalence of high-titer RF (56,2%), our findings were similar to the literature [18, 22].

The initial bivariate analysis showed positive associations between high-titer RF and the multiracial group, current tobacco usage, and higher BMI. The present study is consistent with the established association between tobacco exposure and higher RF titers in the literature [23]. It is also already established that obesity is related to RA disease activity and prognosis, and influences treatment response [24].

However, the impact of higher BMI on RA serological profile is still understudied. Hannech et al. showed similar prevalence of obesity in both positive and negative RF titers [25]. Wesley et al. showed that obesity was linked to a higher chance of developing ACPA-negative RA in women and ACPA-positive RA in men [26]. We found no studies examining the association between BMI and high RF titers. It is possible that the statistical significance (p < 0.05) assigned to the association between high RF titer and a slightly higher BMI in bivariate analysis (mean difference of only 0.69 kg/m² compared to the non-high group, as shown in Table 2) was due to sampling variability (type I error), given the power conferred by our large sample size. Moreover, it is possible that the higher usage of corticosteroids in the high-titer RF group could, at least in part, account for the higher BMI in this group, as weight gain is a well-known side effect of corticosteroids. Anyhow, please notice that the final associations in multivariate analyses between high-titer RF and all the clinical outcomes shown in Table 4 are adjusted for the effects of BMI, i.e., high-titer RF remained an independent predictor of all those outcomes even after accounting for any imbalances in BMI across the groups.

There is conflicting data regarding the associations between serum autoantibodies, disease activity, and race and ethnicity in RA. For instance, Cawley et al. reported Hispanic ethnicity as a predictor of fewer complications, while Greenberg et al. showed higher disease activity in African Americans, Hispanics, and Asians [27, 28]. Likewise, Greenberg et al. linked these to RF positivity, while Mikuls et al. found no difference when comparing African Americans to White Americans [28, 29]. There is even less evidence concerning race and ethnicity and RF titers since most works report only whether the RF is positive or not. In our study, through a large sample size, we were able to find an association between high-titer RF and Black and Pardo Brazilians when compared to White Brazilians.

The present study showed a positive association between high titers of RF and all the reported disease activity scores (DAS28-ESR, DAS28-CRP, and CDAI). This is consonant with RA pathophysiology—how RF amplifies and perpetuates immune response—and has already been demonstrated in the literature, including some real-life studies [10, 12, 30].

Besides the disease activity scores, the use of corticosteroids and bDMARDs was also more frequent among the high-titer RF group. These findings were interpreted as indirect signs of more severe RA in this population, since the inference that bDMARDs could induce higher RF titers is not supported by current evidence [31, 32]. Many other studies confirm our observed association with corticosteroids usage [18, 33].

Regarding the association of RF titers and the need for bDMARDs, there were conflicting data in the literature as some studies showed higher usage of bDMARDs and others were unable to pose this association [34, 35]. Nevertheless, international consensuses have encouraged a more liberal bDMARD prescription in a high RF titer scenario [10, 36]. Most of the recent studies have focused on RF as a predictor of treatment response. The performance of bDMARDs in different RF titers is a current field of research [37].

We also found a higher risk of extra-articular manifestations among those with high RF titers. This is compatible with current literature [38, 39].

Regarding quality of life, neither SF-12 mental nor physical subscales were associated with high-titer RF. Many authors report no association between RF presence and QoL scales, even though these scales are frequently related to higher disease activity and functional capacity scores [40–43].

High titers of RF also implied risk for functional disability (measured by HAQ-DI score). Sobhy et al. tested the same hypothesis but demonstrated only a tendency for higher HAQ scores at high RF titers (p 0.058) [18]. In other populations, there are conflicting results regarding the presence of RF and its association with functional capacity scores, without considering the different RF titers during analysis [33, 35, 42, 44].

This study found no difference in erosive disease between the high and non-high RF titer groups. Slimani et al., in a real-life study, showed results similar to ours when evaluating the association between the prevalence of erosive disease and the serological status regarding RF [44]. At first glance, this seems unexpected since high RF titer is a well-known risk for worse radiographic pattern in RA [45–47]. However, it should be noted that our study elected as the outcome variable the presence (prevalence) of erosive disease—according to the EULAR definition [21]—rather than the extent of cumulative erosions, which would have required different tools like the Larsen’s or the modified Sharp/van der Heijde’s scores.

Furthermore, our sample had considerably longer disease duration than most real-life studies [13, 14, 17, 44]. Additionally, there was a long delay from the onset of symptoms to the first DMARD prescription among our participants [48]. Therefore, the long-standing RA and the loss of opportunities for early treatment in most participants might have attenuated the difference between high and non-high titer groups in meeting the erosive disease definition, for the presence of erosions was common in both groups. Had we assessed the extent of erosive disease (instead of its mere presence), it is conceivable that the high-titer group could have shown higher radiographic scores. Therefore, our findings do not exclude an association between high RF titers and worse erosive profile. Nevertheless, it is possible that, in real-life settings, other variables may exert greater influence on the cumulative radiographic damage than the RF serological status. Future studies are welcome to evaluate the relative importance of high-titer RF and other clinical predictors on the extent of erosive disease in long-standing RA, under real-life conditions.

This work was cross-sectional and based on a convenience sample. As such, it is potentially susceptible to the presence of unobserved confounding and selection bias, partially compensated by our large sample size. Since the participants were recruited from tertiary hospitals, our sample could be biased toward more severe RA profiles. Participants with various therapeutic regimens and disease durations were included in the study, thus affecting the homogeneity of the sample. However, this variability is just as expected in the real-life practice at tertiary centers. This variability, along with the multicentric nature of this work and our large sample, strengthens the generalizability of our results to real-life scenarios.

Another possible issue is that only current smoking status (excluding past or passive smoking) was considered in the study. This could have attenuated the difference between the current and non-current smoker groups. However, this limitation would not change the direction of our finding, i.e., tobacco exposure could account for even greater effect sizes regarding high-titer RF prevalence.

Although fluctuations in RF titers may occur during follow-up and may not be captured in a cross-sectional design, these generally minor variations are unlikely to alter the classification of RA serological profiles among participants—particularly considering the long mean disease duration in the study [49, 50]. Moreover, our large sample size makes the study less susceptible to minor individual variations.

The REAL study protocol envisioned measuring anti-cyclic citrullinated peptide (anti-CCP) as part of the baseline laboratorial assessment, and we believe it would have indeed strengthened our findings. However, only few participating centers had the test readily available. Therefore, just a small fraction of the sample could be tested, not enough to conduct robust analyses or draw sound conclusions regarding ACPA. Trying to assess ACPA (along with RF) under such limitations could be misleading. Hence, the present study chose to focus on RF and its associations with clinical outcomes in RA. We do hypothesize high-titer ACPA should behave in a similar manner as we observed for high-titer RF, but we could not test such a hypothesis in this study. Further studies would be welcome to test for the clinical effects of high-titer anti-CCP.

Despite the wide agreement that RF positivity is associated to worse outcomes, most studies dichotomizing RF as positive or negative were unable to establish as many clinical associations at once as ours. Although high RF titers are already perceived as more strongly associated with clinical outcomes, categorizing RF as high and non-high titers remains uncommon in the literature. Another issue is the variation among studies in categorization strategies for defining high RF titers: ≥ 3 × ULN, above median concentration, or different quartiles [18, 22, 37]. This limits comparisons as it increases data heterogeneity.

Negative and low-positive titers performed similarly in this study, while many clinical associations emerged in the high vs. non-high titer analysis. Hence, when choosing to dichotomize RF titers for RA prognosis, the high-titer threshold (≥ 3 × ULN) appears as more clinically relevant and more statically powerful than merely the upper limit of normality. This strategy may explain why our study was able to detect that many associations.

Moreover, the present study is consonant with the EULAR 2022 recommendations, which advise earlier prescription of bDMARDs or tsDMARDs for patients with RA with high-titer RF [10]. The broad associations between worse RA clinical profile and RF titers above the 3 × ULN suggest these patients would likely benefit from more intensive treatment strategies.

Conclusions

The present study found that RF in high titers was associated with higher disease activity scores, lower physical functionality, the occurrence of extra-articular manifestations, and greater use of corticosteroids and bDMARDs. In addition, high-titer RF was associated with current tobacco use, higher BMI, and multiracial background. Therefore, in a real-life scenario, high-titer RF is associated with a worse RA clinical profile. Dichotomizing RF titers as high (≥ 3 × ULN) and non-high titer appeared more clinically relevant and statistically powerful than merely using the ULN. This strategy may be useful in the design of further clinical studies and in therapeutic considerations of the management of patients with RA.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contribution

All authors contributed to the study conception and design. Data collection was performed by Ana Paula Monteiro Gomides Reis, Claiton Viegas Brenol, Ivânio Alves Pereira, Karina Rossi Bonfiglioli, Letícia Rocha Pereira, Manoel Barros Bértolo, Maria de Fátima Lobato da Cunha Sauma, Maria Fernanda Brandão de Resende Guimarães, Paulo Louzada Júnior, Rina Dalva Neubarth Giorgi, Sebastiao Cezar Radominski, Licia Maria Henrique da Mota, Cleandro Pires de Albuquerque and Geraldo da Rocha Castelar-Pinheiro. Material preparation and analysis were performed by Victor Davi Rosa e Silva de Oliveira, Cleandro Pires de Albuquerque and Licia Maria Henrique da Mota. The first draft of the manuscript was written by Victor Davi Rosa e Silva de Oliveira and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Dr. Cleandro Pires de Albuquerque and Dr. Geraldo da Rocha Castelar-Pinheiro contributed equally to this work.

Funding

The study was supported by the Brazilian Society of Rheumatology. The funder had no role in the study design, data collection, analysis, interpretation, or manuscript preparation. The journal’s Rapid Service Fee was also funded by the Brazilian Society of Rheumatology.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of Interest

The authors Victor Davi Rosa e Silva de Oliveira, Letícia Rocha Pereira, Manoel Barros Bértolo, Maria de Fátima Lobato da Cunha Sauma, Paulo Louzada Júnior, Cleandro Pires de Albuquerque, and Geraldo da Rocha Castelar-Pinheiro declare that they have no conflicts of interest. Ana Paula Monteiro Gomides Reis: Has received speaking fees and support for educational activities from AbbVie, Pfizer, and Janssen. Has participated in advisory boards for AbbVie and Pfizer. Claiton Viegas Brenol: Has participated in clinical and/or experimental studies related to this work and sponsored by AbbVie, BMS, Janssen, Pfizer, and Roche; has received personal or institutional support from AbbVie, BMS, Janssen, Pfizer, and Roche; has delivered speeches at events related to this work and sponsored by AbbVie, Janssen, Pfizer, and Roche. Ivânio Alves Pereira: Has received consulting fees, speaking fees, and support for international congresses from Roche, Pfizer, UCB Pharma, Eli-Lilly, AbbVie, and Janssen. Following completion of this work, Ivânio Alves Pereira has changed affiliation to Universidade do Sul de Santa Catarina. Karina Rossi Bonfiglioli: Has received speaking fees and support for educational activities from AbbVie, UCB, Janssen, Pfizer. Maria Fernanda Brandão de Resende Guimarães: Has received speaking fees and support for international congresses from Amgen, AbbVie, UCB, Lilly, and Janssen. Rina Dalva Neubarth Giorgi: Has participated in advisory boards and received speaking fees and support for international congresses from Eli Lilly, Biocon, Novartis, Janssen, AbbVie, and Fresenius Kabi. Sebastião Cezar Radominski: Has received speaking fees from AbbVie, Janssen, Novartis, and Pfizer; has participated in clinical studies sponsored by AbbVie and Novartis. Lícia Maria Henrique da Mota: Has received personal or institutional support from AbbVie, Janssen, Pfizer, and Roche; has delivered speeches at events related to this work and sponsored by AbbVie, Boehringer Ingelheim, GSK, Janssen, Libbs, Lilly, Novartis, Pfizer, Roche, Sandoz, and UCB.

Ethical Approval

The study was approved by the National Commission of Ethics in Research (CONEP—Comissão Nacional de Ética em Pesquisa, Brazilian Ministry of Health) and by institutional review boards of each participating center (protocol number CAAE 45781015.8.1001.5259). All patients provided written informed consent prior to inclusion in the study. The study was conducted in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments.