Introduction

Methotrexate (MTX) is the anchor drug in the treatment of rheumatoid arthritis (RA), and its effectiveness is well established [1]. However, it is still unclear how MTX affects the complicated cytokine network involved in the pathology of RA [2, 3], and few studies have investigated changes in plasma cytokine levels during MTX treatment [4].

Tumor necrosis factor (TNF)-α and interleukin (IL)-6 play central roles in the pathogenesis of RA and are the main targets of biological agents [5]. MTX inhibits cytokine production induced by T cell activation in vitro [6, 7]. However, there are no reports regarding the effects of MTX on circulating cytokine levels in treatment-naive early RA patients. We performed longitudinal measurements of plasma TNF-α and IL-6 in patients in such a cohort and analyzed the relationship between cytokine levels and clinical effectiveness and radiographic progression.

Patients and methods

Patients

Systematic Cohort Analysis in Keio University–Rheumatoid Arthritis (SAKURA) is a prospective cohort consisting of 150 consecutive newly diagnosed RA patients at Keio University Hospital between August 2008 and March 2011. At inclusion in SAKURA, those patients had never been treated with any disease-modifying antirheumatic drugs (DMARDs) or corticosteroids, and each rheumatologist in charge started treatment. Since some patients in SAKURA had already severe joint destruction with a long history of arthralgia at the time of diagnosis, we selected patients with disease duration (which we defined here as a time period from symptom onset to diagnosis) of up to 36 months in this current study in order to catch small changes in joint destruction. Concomitant use of classic DMARDs or low-dose prednisolone at a dose of no more than 10 mg/day was allowed. Clinical, radiographic, and plasma cytokine data was collected at baseline and 1 year later. Since the interval of the patient’s visit was 1–3 months, 1-year data was obtained with 3-month allowance. Patients in whom a biological agent was started within 1 year were analyzed with data just before the commencement of it using last observation carried forward (LOCF) method. Although the time points of data collection at baseline were not necessarily exactly the same with the time of the MTX commencement, all the patients had been treated with MTX for more than 5 months during this study. Six patients assessed at 16 to 18 months due to patient’s personal reasons were exceptionally included. This study was approved by the ethics committee at Keio University School of Medicine. Written informed consent was obtained from all patients.

Clinical and radiological assessments

Clinical characteristics including matrix metalloproteinase-3 (MMP-3) which was routinely measured in our clinics were obtained from the medical charts. Radiographs of the hands and feet were taken at baseline and 1 year later (at the same time as the blood samples). Two blinded and independent readers (NN and YK) scored the images using the van der Heijde modified Total Sharp Score (mTSS) [8]. The smallest detectable change of the two in mTSS was 1.32. The mean of the two scores was used in the analysis. Estimated yearly progression of mTSS (∆mTSS/year) was adjusted by the exact months from baseline to the second time of x-ray.

Plasma assays

Plasma samples were collected at baseline and 1 year later. All samples were stored immediately after collection at −30 °C in our laboratory. Plasma TNF-α and IL-6 levels were assessed using chemiluminescent enzyme immunoassays (R&D Systems Inc., Minneapolis, MN, USA) that had a threshold of detection of 0.55 and 0.30 pg/ml, respectively. The details of the measurement methods were reported previously [9].

Statistical analysis

Continuous data are presented as the median with interquartile range (IQR), mean with standard deviation, or as a number with percentage value as appropriate. The Wilcoxon signed rank test was used to examine changes between baseline and 1 year later. Correlation of two continuous variables was analyzed using Pearson’s correlation. Receiver operating characteristics (ROC) curve analysis was used to evaluate the performance of the indices as a method for identifying clinically relevant radiographic progression (CRRP), defined as ∆mTSS/year >3 [10]. Multivariate analysis was performed using multiple regression models by parameters with p value <0.20 in preceding univariate analysis. All statistical analyses were performed with JMP 9 (SAS Institute Inc., Cary, NC, USA).

Results

Patient characteristics

Sixty-two patients were analyzed. Of the SAKURA cohort, 88 patients were excluded because of one of the following reasons: disease duration >36 months, a biological agent was started within 5 months, MTX was not used, or lost to follow-up. The median duration of observation was 11 (7–14) months. The median age was 57 (49–66) years, 49 (79.0 %) patients were female, and the median disease duration was 3 (2–7) months. Patient characteristics and the treatment during this study are shown in Table 1. Nineteen (31 %) patients did not have 1-year data because of the commencement of a biologic agent or patient’s personal reasons and their data were analyzed using LOCF methods.

Table 1 Baseline characteristics and treatments during this study of rheumatoid arthritis patients (n = 62)
Full size table

Clinical assessments

The median Disease Activity Score 28 (DAS28) and the Health Assessment Questionnaire–Disability Index (HAQ-DI) significantly (p <0.001) decreased from baseline to 1 year later from 4.42 to 2.58 and from 0.625 to 0.125, respectively (Table 2). Changes in other parameters are shown in Table 2. Despite the clinical effectiveness of MTX, the mTSS showed a statistically significant increase (p <0.001) from 4 (0.75–8.25) to 7 (2–11.25). CRRP was observed in 12 (19.4 %) patients. Median ∆mTSS/year was 0.5 (0–2.5). Thirty-one (50 %) patients showed no radiographic progression (∆mTSS/year <0.5).

Table 2 Comparison of the clinical and biological characteristics of early rheumatoid arthritis patients at baseline and 1 year later
Full size table

Plasma cytokine concentrations

Plasma IL-6 and TNF-α were detectable in 59 (95.2 %) and 36 (58.1 %) patients at baseline, respectively, and in 57 (91.9 %) and 37 (59.7 %) patients 1 year later, respectively. The median plasma IL-6 level decreased from 4.72 to 1.04 pg/ml (p <0.001), while there was no significant difference in the level of TNF-α at baseline and 1 year later (0.87 and 0.83 pg/ml, respectively; p = 0.14) (Table 2).

Association of radiographic progression with clinical parameters

The ∆mTSS/year correlated with DAS28 (r = 0.469, p <0.01); simplified disease activity index (SDAI) (r = 0.422, p <0.0006); MMP-3 (r = 0.581, p <0.01); and IL-6 (r = 0.673, p <0.01). However, it did not correlate with C-reactive protein (CRP); (r = 0.221, p = 0.08) or with TNF-α (r = 0.158, p = 0.22) (Fig. 1a–f). Plasma IL-6 level at 1 year discriminated between patients with CRRP and those without CRRP at a cut-off level of 4.03 pg/ml with a sensitivity of 91.7 % and a specificity of 88.0 %, while plasma TNF-α level at 1 year did not (Fig. 1k and l). DAS28, SDAI, CRP, and MMP-3 discriminated between these groups but less effectively than IL-6 (areas under the ROC curve were 0.810, 0.813, 0.792, and 0.675, respectively; Fig. 1g–j). We also performed the ROC analyses in the same way on the Sharp units of 0.5, 2, and 5 instead of 3 (CRRP), and got the almost same results (data not shown). The proportion of patients with CRRP paralleled with specified categories of disease activity or with parameter values, except for TNF-α (Fig. 1m–r). Moreover, we divided the patients by their DAS28 at 1 year into “remission and low disease activity (LDA)” group and “moderate disease activity (MDA) and high disease activity (HDA)” group, and examined the association between cytokines and radiographic changes. While in remission and LDA group, patients with the high IL-6 level were disposed to show CRRP, in MDA and HDA group no patients with low IL-6 level showed CRRP. In regard to TNF-α, we could not find any relationships (data not shown).

Fig. 1
figure 1

Association of disease activity and biomarkers after methotrexate (MTX) treatment with radiographic progression in early rheumatoid arthritis patients. Association of a DAS28, b SDAI, c CRP, d MMP-3, e IL-6, and f TNF-α with ∆mTSS/year. The ROC curve discriminated between ∆mTSS/year >3 (CRRP) and ∆mTSS/year ≤3 for g DAS28, h SDAI, i CRP, j MMP-3, k IL-6, and l TNF-α. The percentage of patients who had CRRP according to disease activity category: m DAS28 and n SDAI or quartiles of o CRP, p MMP-3, q IL-6, and r TNF-α. CRP C-reactive protein; CRRP clinically relevant radiographic progression; DAS disease activity score; HDA high disease activity; IL interleukin; LDA low disease activity; MDA moderate disease activity; MMP matrix metalloproteinase; pts patients; Q quartile; REM remission; ROC receiver operating characteristics; SDAI simplified disease activity index; TNF tumor necrosis factor; ∆mTSS/year, estimated yearly progression of modified Total Sharp Score

Full size image

Baseline characteristics for prediction of RA patients with radiographic progression

Among the parameters, sex and the plasma level of TNF-α at baseline were statistically significant (p <0.05) but weakly correlated with ∆mTSS/year in multivariate analysis (Table 3). Thus, female patients with higher plasma TNF-α level at baseline may be predisposed to worse radiographic progression.

Table 3 Association of baseline parameters of early rheumatoid arthritis patients with ∆mTSS/year
Full size table

Discussion

This study demonstrated that in early RA patients, the plasma IL-6 level significantly decreased during MTX treatment, while the plasma TNF-α level did not change. Radiographic progression correlated with higher levels of plasma IL-6 after MTX treatment, suggesting that inhibition of plasma IL-6 may be critical for preventing radiographic progression.

Most of the patients analyzed in this study were in the early stages of RA and had mild to moderate disease activity. Although the overall clinical response to MTX was excellent, one-fifth of the patients exhibited CRRP, which is one of the criteria that typically prompt consideration of treatment changes [10]. In fact, there was a significant correlation between the DAS28 score at 1 year and radiographic progression. However, it should be noted that one-fourth of the patients with LDA resulted in CRRP according to DAS28 category. In this regard, the plasma IL-6 level after MTX treatment showed the strongest association with radiographic progression. For these patients, the measurement of plasma IL-6 might be able to provide additional clinical information that we may as well consider the treatment strategy more carefully in those patients with plasma IL-6 levels greater than 4.03 pg/ml to suppress joint destruction even if they were in remission or LDA.

MTX appeared to have a greater effect in terms of suppressing circulating IL-6 than on suppressing circulating TNF-α. It is likely that the effect of MTX on cytokines in RA is achieved mainly via inhibition of IL-6 production. In this context, it makes sense that TNF inhibitors have a greater effect when used in combination with MTX than when used as monotherapy [11] which is not the case with the IL-6 receptor inhibitor tocilizumab [12, 13]. Moreover, the result that the patients with higher TNF-α at baseline showed worse radiographic progression indicated that TNF-α could be used as a biomarker to consider the treatment strategy. Since our study showed that MTX have little effect on TNF-α, we may be able to suggest that anti-TNF agents be an option at an early stage in patients with high TNF-α.

Our study has several limitations. First, since the study population was small, the results of this study may not be valid for all RA patients. For example, the results showing that IL-6 was affected by MTX might be different if the patients showed an inadequate response to MTX. Second, we examined only two cytokines. Since multiple cytokines are involved in the pathogenesis of RA [14], other cytokines need to be investigated as well. Third, the patients showed a good response to MTX when a lower dose of MTX was used rather than the global recommended dose [15]. Ethnic differences between Caucasians and Japanese might account for the difference in the effectiveness of MTX.

In conclusion, this study is the first to demonstrate that MTX reduces IL-6 but not TNF-α in vivo and to show that the plasma IL-6 concentration after MTX was the parameter that was most associated with radiographic progression. Further investigations in a larger population are needed to confirm these findings.