Introduction

Rheumatoid arthritis (RA) is a chronic autoimmune disease associated with systemic inflammation and erosive and symmetrical polyarthritis.¹ RA is more common in women and is predominantly in middle-aged individuals.² The number of new cases of RA reported is increasing by years, and the mortality associated with RA is also elevating.^3,4 Highly active RA results in the destruction of joints; in addition, RA is associated with systemic inflammation that may involve other tissues and organs and result in extra-articular effects, such as secondary Sjogren’s syndrome,⁵ interstitial pneumonia,⁶ and rheumatoid nodules,⁷ that are associated with poor prognosis.⁸

Rheumatoid factors (RFs) are defined as autoantibodies that recognize the Fc fragments of other immunoglobulins.⁹ The most common RFs are IgM and IgA RFs that bind to the Fc fragment of IgG.⁹ RFs are clearly detectable and often present in the early stages of RA, years before the clinical symptoms appear in some patients.¹⁰ The RF titer is considered as a diagnostic and prognostic marker of RA. RF still has an irreplaceable value in the classification standard of RA. The combination of RF and anti-cyclic citrullinated protein/peptide antibodies (ACPA) can improve the diagnostic specificity of RA.¹¹ In a Danish cohort study, healthy individuals with elevated RF levels had an increased risk of developing RA disease.¹² Studies have reported that high levels of RF have a good response to B cell-depleting therapy.¹³ Moreover, RF is associated with the activity of RA.^14–16 A decrease in RF titers was found to be consistent with a decrease in activity in RA patients after treatment.¹⁷ However, until now, no researchers have explored the specific relationship between RF titers and RA activity.

Restricted cubic spline (RCS) is a data analysis model that can combine continuous variables with outcomes. It can reveal the influence of independent variables on outcome risk in the form of continuous curves, and it is an important method to analyze the dose-response relationship between continuous variables and outcomes.¹⁸ At present, it has been applied in many researches.^19,20 This study sought to fill in this research gap by using the restricted cubic spline (RCS) function to perform a dose–effect analysis and quantify the association of RF titer with RA activity.

Materials and Methods

Study Participants

The cross-sectional study recruited 2044 patients with RA, which was diagnosed and classified according to the 2010 criteria of the American College of Rheumatology and the European League Against Rheumatism (ACR-EULAR).²¹ The patients were recruited between February 2022 and October 2023 from the First Affiliated Hospital of Anhui Medical University. Potential participants were first screened for their eligibility by two rheumatologists, who interviewed and assessed disease activity in the patients by calculating the disease activity score 28 (DAS28)-erythrocyte sedimentation rate (ESR) (DAS28-ESR) and DAS28-C reactive protein (CRP).^22–24 Those with severe infections, severe trauma, and other inflammatory and autoimmune diseases were excluded. The Ethics Committee of the Anhui Medical University gave its approval for the study (Approval no.20121090) and all processes were in accordance with the 1964 Declaration of Helsinki.

Demographic and Clinical Data

Data on the following variables were obtained: sex, age, duration of RA, body mass index (BMI), presence of hypertension and diabetes, history of cigarette smoking and consumption of drugs within the past 3 months, including glucocorticoids (GCs), non-steroidal antiinflammatory drugs (NSAIDs), conventional synthetic disease-modifying antirheumatic drugs (csDMARDs), biological DMARDs (bDMARDs), and targeted synthetic DMARDs (tsDMARDs).

Laboratory Assessment

Standard laboratory tests were performed to measure ESR (mm/h), CRP (mg/l), RF (IU/mL), white blood cell count (WBC, ×10⁹/l), hemoglobin (HB, g/l), lymphocyte percentage (LY%), blood platelet count (BPC, ×10⁹/l), albumin (ALB, g/l), alanine aminotransferase (ALT, U/l), aspartate aminotransferase (AST, U/l), blood urea nitrogen (BUN, mmol/l), uric acid (UA, µmol/l), and creatinine (CR, µmol/l). ESR was evaluated with the Wintrobe method, and the CRP level was determined with a turbidimetric assay (BioSystems). RF (isotype IgM) was assayed with an automated turbidimetric assay system (BioSystems), with the cut-off for RF detection set at 14 IU/mL.

Statistical Methods

SPSS 26.0 and R 4.4.0 were used to perform all the data analyses. Clinical variables were evaluated with descriptive statistics. The Mann–Whitney U-test was employed to analyze non-parametric data, and the chi-square test, to analyze frequencies. The RCS model was used to determine the non‑linear relationship of RF titer with RA activity. The factors that were significantly associated with the risk of RA activity were identified by univariate and unconditional multivariate logistic regression analyses, with odds ratios (ORs) and 95% confidence intervals (CIs) used to describe the risk ratio. A p-value <0.05 was assumed to indicate statistical significance.

Results

Characteristics of the Cohort

Of the 2044 participants with RA, 85.91% were female and 14.09% were male. The participants were divided into RF-positive (RF > 14 IU/mL, n = 1792) and RF-negative (RF ≤ 14 IU/mL, n = 252) groups. The groups showed significant differences in duration of RA, BMI, DAS28-ESR, DAS28-CRP, use of NSAIDs and csDMARDs, number of affected joints (indicated by tenderness and swelling), ESR, CRP, WBC, LY%, and ALB (P < 0.05 for all). The detailed clinical characteristics are presented in Table 1.

Table 1 Characteristics of the Study Cohort (N = 2044)

Logistic Regression Analysis of Factors Associated with the Risk of RA Activity

DAS28 greater than 3.2 was considered to indicate active RA, and DAS28 less than 3.2 was considered to indicate low RA activity.²⁵ According to the results of univariate logistic regression analysis, the following factors were identified as significant predictors of the risk of RA activity according to DAS28-ESR: sex; age; duration of RA; diabetes; use of GCs, NSAIDs, csDMARDs, bDMARDs, and tsDMARDs; WBC; HB; LY%; BPC; ALB; ALT; AST; and RF. Further, the following variables emerged as significant predictors of the risk of RA activity assessed by DAS28-CRP: age; duration of RA; use of GCs, NSAIDs, csDMARDs, bDMARDs, and tsDMARDs; WBC count; HB, LY%, BPC, ALB, AST, Cr, UA, and RF (Table 2).

Table 2 Univariate Logistic Regression Analysis of Significant Factors Associated with the Risk of RA Activity

The variables identified in the univariate analysis were entered into the multivariate logistic regression model to determine which ones were independently related with the risk of RA activity. As shown in Table 3, the use of csDMARDs, bDMARDs and tsDMARDs; WBC count, LY%; HB; BPC; ALB; and RF titers were identified as being independently associated with the risk of RA activity according to DAS28. In addition, age was an independent predictor of the risk of RA activity according to DAS28-ESR, and duration of RA was an independent predictor of the risk of RA activity according to DAS28-CRP.

Table 3 Multivariate Logistic Regression Analysis of Independent Factors Associated with the Risk of RA Activity

Sensitivity Analysis

The RF titer was transformed from a continuity variable to a quartile categorical variable. As can be seen from Table 4, the risk of RA activity determined by DAS28-ESR for the lowest category of RF was 1.117 in the crude model, and it increased to 2.457 and 3.619 for the median and highest category, respectively. In the adjusted model, its value was 1.088 for the lowest category and increased to 1.904 and 2.020 in the median and highest category, respectively. The risk of RA activity assessed by DAS28-CRP for the lowest category of RF was 0.978 in the crude model and increased to 2.199 and 3.096 for the median and highest category, respectively. In the adjusted model, it was 0.880 for the lowest category of RF and increased to 1.519 and 1.526 for the median and highest category, respectively.

Table 4 Logistic Regression Analyses of the Correlation Between RF and the Risk of RA Activity

Non‑Linear Relationship Between RF Titers and RA Activity

Considering the skewed distribution of RF titers (Figure 1A), the values was log2-transformed for RCS analysis (Figure 1B). When the RCS function was used with 3 knots, a non-linear association between RF titers and RA activity was observed in the crude and adjusted models (Figures 2 and 3) (P < 0.05). In addition, a threshold effect was noted at a log2RF value of 6.04 (RF = 65.80 IU/mL). That is, when RF was less than 65.80 IU/mL, the risk of RA activity was almost unaffected or even reduced, but when RF was greater than 65.80 IU/mL, there was a rapid rise in the risk of RA activity.

Figure 1 Distribution of RF titers. The figures show the distribution of (A) crude and (B) log2‑transformed RF titers. Abbreviation: RF, rheumatoid factor.

Figure 2 Association of RF titers with RA activity in the crude model. (A) Association of RF titers with RA activity according to DAS28-ESR. (B) Association of RF titers with RA activity according to DAS28-CRP. Abbreviations: RA, rheumatoid arthritis; RF, rheumatoid factor; DAS28, disease activity score 28; ESR, erythrocyte sedimentation rate; CRP, C reactive protein; OR, odds ratio; CI, confidence interval.

Figure 3 Association of RF titers with RA activity in the adjusted model. (A) Association of RF titers with RA activity according to DAS28-ESR. (B) Association of RF titers with RA activity according to DAS28-CRP. Abbreviations: RA, rheumatoid arthritis; RF, rheumatoid factor; DAS28, disease activity score 28; ESR, erythrocyte sedimentation rate; CRP, C reactive protein; OR, odds ratio; CI, confidence interval.

RF titers were classified according to the cutoff value of 65.80 IU/mL. In the adjusted model, when RF exceeded the cutoff value, the risk of RA activity assessed by DAS28-ESR increased by 99.2% and the risk of RA activity assessed DAS28-CRP increased by 62.8% (Table 5 and Figure 4).

Table 5 Relationship Between RF Levels and the Risk of RA Activity Stratified Based on a Cutoff Value of 65.80 IU/Ml

Figure 4 Risk of RA activity based on an RF cutoff value of 65.80 IU/mL in the adjusted model. (A) Risk of RA activity assessed by DAS28-ESR. (B) Risk of RA activity assessed by DAS28-CRP. Abbreviations: RA, rheumatoid arthritis; RF, rheumatoid factor; DAS28, disease activity score 28; ESR, erythrocyte sedimentation rate; CRP, C reactive protein; csDMARDs, conventional synthetic disease-modifying antirheumatic drugs; bDMARDs, biological DMARDs; tsDMARDs, targeted synthetic DMARDs; WBC, white blood cell; HB, hemoglobin; LY%, lymphocyte percentage; BPC, blood platelet count; ALB, albumin; CI, confidence interval.

Discussion

The present study demonstrates a significant, non-linear relationship between RF titer and RA activity in patients with RA after adjusting for all covariates. In addition, based on the data, we found that an RF cutoff of 65.80 IU/mL could be used to determine the risk of RA activity. In the adjusted model, when the RF value was greater than 65.80 IU/mL, there was a 99.2% increase in the risk of RA activity based on DAS28-ESR and a 62.8% increase in the risk of RA activity based on DAS28-CRP. In line with this observation, it has previously been reported that assessing RA activity with DAS28-CRP leads to significant underestimation of disease activity in comparison with assessing RA activity based on DAS28-ESR.^26,27

As early as the 1980s, studies have found that RF positivity, especially high RF titers, is associated with a higher degree of inflammation in patients with RA.^28,29 Further, fluctuations in RF levels are considered biomarkers of disease activity and treatment responsiveness in patients with RA.^30–32 However, there is limited analysis of the continuous changes in RF titers with RA activity. Most studies analyze RF as a categorical variable and classify patients based on RF positivity or quartiles.^31,33–35 For example, high RF levels (≥156.4 IU/mL) were found to be predictive of the development of refractory RA.³⁶ Further, a systematic review found that RF was correlated to bone erosion in RA as determined by ultrasound observation.³⁷ However, these approaches to classifying RF could mask important clinical features or prognostic outcomes.

This study seeks to overcome the limitation of previous studies by exploring the dose-effect relationship between RF titers and RA activity. To this end, we used the RCS function, which is a useful tool for analyzing dose-effect associations between successive exposures and outcomes.^38–40 We also conducted logistic regression analyses and adjusted for other covariables to improve the accuracy and reliability of the results. Moreover, the non-linear relationship was explored by RCS and smooth curve fitting, and the critical point was at which RF value was indicative of RA activity risk was also calculated. Our threshold RF value of 65.80IU/mL may be low compared to other literature,³⁶ possibly due to the different materials used to measure RF in different laboratories, furthermore, the range of RF normal value in our lab is no more than 14 IU/mL, and it is necessary to establish a uniform standard.¹¹

Another important finding of the study was the identification of factors that influence RA activity. The results show that the use of csDMARDs, bDMARDs and tsDMARDs; WBC; LY%; HB; BPC; and ALB were associated with RA activity assessed with DAS28. This is consistent with previous research reports.^39,41–47 In addition, age was associated with DAS28-ESR, and duration of RA was associated with DAS28-CRP. This difference might be attributable to differences in the reference indicators. That is, ESR is an indicator of disease activity in recent weeks and is affected by sex, age, anemia, and other factors,⁴⁸ while CRP is indicative of a shorter inflammatory response time.⁴⁴

A major limitation of the present study is its cross-sectional design, as a result of which it is difficult to establish a causal link between RF and RA activity. Moreover, while the regression model was adjusted for some variables, we could not account for covariables for which data were not collected or were unavailable. This is another limitation to be considered when interpreting the study findings. Finally, some research variables from the questionnaires and self-reports may be biased. Given these limits, more research needs to be conducted to verify the present observations. Nonetheless, we believe that this study serves as a critical reference point for future research on the subject.

Conclusion

This is the first study to elucidate the non-linear significant relationship between the risk of RA activity and RF titers. The results show that the risk of RA activity increased non-linearly along with the continuous change in RF titers. Thus, RF may be useful as a clinical marker of inflammation and risk of RA activity in patients with RA. Given its low cost and simplicity of measurement, we believe that RF as a predictor of RA disease activity might have immense clinical potential and impact in terms of providing meaningful data for timely intervention and treatment of RA.