Brief Report: M-Ficolin Levels Reflect Disease Activity and Predict Remission in Early Rheumatoid Arthritis

Authors


Abstract

Objective

To assess plasma M-ficolin concentrations in disease-modifying antirheumatic drug (DMARD)–naive patients with early rheumatoid arthritis (RA), to investigate the correlation of M-ficolin concentrations with disease activity markers, and to determine the predictive value of M-ficolin with respect to the Disease Activity Score in 28 joints (DAS28).

Methods

The study group included 180 DMARD-naive patients with early RA who participated in a randomized controlled trial of methotrexate and intraarticular glucocorticoids plus either adalimumab or placebo/adalimumab. One hundred healthy control subjects and 51 patients with chronic RA were also assessed. A sandwich-type time-resolved fluorometric immunoassay was used for quantification of plasma M-ficolin.

Results

At baseline, M-ficolin levels were highest in the group of DMARD-naive patients with newly diagnosed active RA, and the level in these patients decreased 26% after 1 year of aggressive treatment. The baseline M-ficolin level correlated with 5 of 7 disease activity markers, including the DAS28 and the Health Assessment Questionnaire (HAQ), and a similar pattern of correlations was observed at 1 year. Multiple logistic regression analysis showed that an elevated M-ficolin level at baseline was the strongest predictor of not achieving either DAS28 remission (odds ratio [OR] 4.18, 95% confidence interval [95% CI] 2.02–8.63) or low disease activity (OR 2.45, 95% CI 1.13–5.28) at 1 year. The presence of a baseline M-ficolin level in the lowest quartile resulted in sensitivity of 29%, specificity of 93%, and positive predictive value of 95% for low disease activity at 1 year.

Conclusion

In patients with early RA, elevated plasma M-ficolin levels correlated with a high DAS28 and a high HAQ score at baseline and 1 year. A low M-ficolin level was the strongest predictor of remission and low disease activity in a multivariate analysis.

The complement system plays a role in the development of rheumatoid arthritis (RA), as judged by increased consumption of complement factors and elevated levels of activated complement products in the synovial fluid of inflamed joints. Furthermore, there are correlations between the degree of complement activation and the severity of joint disease ([1]).

The complement system is activated by the classical pathway, the alternative pathway, and the lectin pathway. The lectin pathway is activated by 4 soluble pattern-recognition molecules: mannon-binding lectin and the 3 ficolins (H-ficolin, L-ficolin, and M-ficolin). Mannan-binding lectin recognizes patterns of carbohydrates, and the ficolins recognize patterns of acetylated molecules, including acetylated carbohydrates, thereby sensing foreign and altered self antigens. The pattern-recognition molecules circulate in complexes with mannon-binding lectin–associated serine proteases (MASPs). Upon binding to targets, the associated MASP-2 cleaves complement proteins C2 and C4, thus generating C3 convertase. The lectin pathway protects against invading pathogens and participates in regulation of the adaptive immune system and clearance of apoptotic host cells ([2]).

M-ficolin is synthesized by monocytes and granulocytes and secreted upon stimulation, but also is expressed as a membrane-associated protein on the surface of these cells ([3-5]). Exposure to proinflammatory stimuli such as gram-negative bacteria and interleukin-8 leads to increased secretion of M-ficolin by monocytes and granulocytes ([5, 6]). M-ficolin levels correlate with the number of neutrophils in the blood of healthy blood donors and patients with RA ([7]). Furthermore, associations between 2 single-nucleotide polymorphisms in the M-ficolin gene and susceptibility to RA have been described ([8]). We recently reported a 30-fold increase in synovial fluid M-ficolin concentrations in patients with chronic active RA compared with the concentrations in patients with osteoarthritis, suggesting that M-ficolin is involved in the pathogenesis of RA ([7]).

The aim of this study was to investigate the plasma levels of M-ficolin in patients with early RA who had not received disease-modifying antirheumatic drugs (DMARDs), at baseline and after 1 year. Moreover, we investigated the correlations between M-ficolin and disease activity markers, and the predictive capacity of M-ficolin with respect to the Disease Activity Score in 28 joints (DAS28) ([9]).

PATIENTS AND METHODS

Patients

One hundred eighty DMARD- and steroid-naive patients with early RA (disease duration <6 months) were included in the OPERA study ([10]), an investigator-initiated, randomized, double-blind, placebo-controlled trial. Briefly, patients were randomized 1:1 to receive methotrexate and intraarticular triamcinolone injections plus either adalimumab or placebo. Triamcinolone was injected intraarticularly at all visits. Presence of joint erosions on baseline radiography as assessed by a Sharp/van der Heijde erosion score of ≥1 ([11]), the Stanford Health Assessment Questionnaire (HAQ) score ([12]), and the DAS28 using the C-reactive protein (CRP) level were registered. Paired plasma samples obtained at baseline and year 1 were analyzed. In addition, 51 patients with chronic RA (disease duration >5 years) in remission (DAS28 <2.6) who were receiving tumor necrosis factor inhibitor treatment were included (providing 1 sample each), as were 100 healthy adults. The characteristics of the participants are shown in Table 1. All participants provided written informed consent. The protocol was approved by the local ethics committee (M-2009-0191) and was performed in accordance with the Declaration of Helsinki.

Table 1. Patient characteristics and M-ficolin concentrations*
 Healthy controls (n = 100)Chronic RA (n = 51)OPERA study, placebo (n = 91)OPERA study, adalimumab (n = 89)OPERA study, total (n = 180)
  1. Radiographic erosions were defined as a Sharp/van der Heijde erosion score of ≥1. Except where indicated otherwise, values are the median (5–95% percentile). RA = rheumatoid arthritis; anti-CCP = anti–cyclic citrullinated peptide; IgM-RF = IgM rheumatoid factor; DAS28 = Disease Activity Score in 28 joints; VAS = visual analog scale; HAQ = Health Assessment Questionnaire; CRP = C-reactive protein; 95% CI = 95% confidence interval.
Female sex, %7075696366
Age, years50 (26–77)62 (38–75)54 (28–77)56 (26–78)55 (27–78)
Neutrophil count3.0 (1.5–6.3)5.6 (3.1–8.5)6.1 (2.7–9.1)5.9 (3–9.1)
Monocyte count0.5 (0.3–1.2)0.5 (0.3–0.9)0.3 (0.3–1.0)0.5 (0.3–1.0)
Disease duration, years15.7 (5.2–38.6)0.23 (0.11–0.41)0.24 (0.11–0.44)0.23 (0.11–0.44)
Anti-CCP positive, %96706065
IgM-RF positive, %96747072
DAS281.8 (1.3–2.3)5.6 (3.8–7.3)5.5 (3.8–7.8)5.6 (4.8–6.2)
Tender joint count in 28 joints0 (0–1)11 (3–24)10 (3–27)11 (3–26)
Swollen joint count in 28 joints0 (0–1)8 (2–22)8 (2–26)8 (2–23)
Pain VAS score, mm12 (0–53)58 (13–92)63 (13–98)61 (12–96)
Patient's global assessment VAS score, mm16 (0–47)65 (17–96)70 (12–100)67 (13–98)
Physician's global assessment VAS score, mm2 (0–7)51 (22–86)57 (27–89)55 (21–90)
HAQ score0.50 (0–1.90)1.00 (0.25–2.31)1.13 (0.17–2.58)1.10 (0.10–2.50)
CRP, mg/liter1.4 (0.4–11.3)15 (7–109)15 (7–133)15 (7–132)
Baseline radiographic erosions, %98525453
Disease activity at year 1     
DAS282.6 (1.7–4.7)2.0 (1.7–5.2)
DAS28 <2.6, %4974
DAS28 <3.2, %7680
M-ficolin concentration, median (95% CI)     
Baseline, μg/ml1.89 (1.72–2.07)2.17 (1.94–2.42)2.93 (2.63–3.26)2.77 (2.45–3.14)2.85 (2.63–3.09)
Year 1, μg/ml2.59 (2.34–2.86)2.05 (1.81–2.32)2.30 (2.12–2.49)

M-ficolin assessment

Plasma M-ficolin concentrations were measured using a validated in-house time-resolved fluorometric immunoassay with monoclonal anti-human M-ficolin antibodies ([13]). The principle of the assay is identical to that of a sandwich enzyme-linked immunosorbent assay, only the reporter is the fluorescence of the europium ion rather than enzyme activity. The coefficients of variation were determined, from 3 internal controls included on each plate, to be 17% at 0.12 μg/ml, 9% at 0.62 μg/ml, and 5% at 1.60 μg/ml.

Statistical analysis

Because M-ficolin concentrations were nonparametrically distributed, the analysis was performed on logarithmically transformed M-ficolin concentrations (reported as medians and 95% confidence intervals [95% CIs]). Correlations were calculated using Spearman's rho. Student's t-test and one-way analysis of variance were used for comparisons between groups, and temporal changes in M-ficolin levels were analyzed with Student's paired t-test.

Possible predictors of DAS28 remission and low disease activity were initially in tested in univariate analyses. The M-ficolin concentration was correlated with the number of neutrophils (P < 0.001, ρ = 0.43) and monocytes (P < 0.001, ρ = 0.28) in peripheral blood, and both variables were included in the univariate model. To ensure robustness of the analysis, M-ficolin was entered as both a continuous and a dichotomized variable (the arbitrarily chosen cutoffs were 25%, 33%, 50%), dividing the patients into 2 groups according to the level of M-ficolin at baseline.

Afterward, multiple logistic regression analyses were performed with backward selection, using the criterion of P ≥ 0.05 for removal from the model. In addition to the significant (P < 0.05) variables in the univariate analysis (M-ficolin, neutrophils, sex, and treatment), DAS28, baseline radiographic erosions, age, and anti–cyclic citrullinated peptide (anti-CCP) antibodies were forced into the model because of the previously reported significance of these variables. An intent-to-treat analysis was performed (n = 170), and a completer analysis (n = 161) yielded similar results. Analysis with Stata version 12.1 was performed in cooperation with a professional statistician (NSK).

RESULTS

M-ficolin concentrations in the patient groups

We observed significant differences between M-ficolin concentrations across the groups included in this study. The highest M-ficolin levels were measured in the DMARD-naive patients with newly diagnosed active RA; in this group, M-ficolin levels declined 26% (95% CI 11–40%, P < 0.001) after 1 year of clinically effective treatment. The lowest M-ficolin concentrations were observed in healthy adults, and these concentrations were significantly lower than those in the OPERA cohort at both baseline and 1 year, as well as those in patients with chronic RA in remission (Table 1).

Because there was a significant difference in the DAS28 between the 2 OPERA treatment groups at year 1, we analyzed for a potential difference in M-ficolin levels between these groups. No differences were observed at baseline; at 1 year, however, the M-ficolin concentration in the adalimumab group was 21% lower (95% CI 8–48, P = 0.001) than that in the placebo group. At 1 year, there was no difference between the adalimumab group compared with healthy adults and patients with chronic RA in remission.

The presence of anti-CCP, IgM–rheumatoid factor, or baseline radiographic erosions was associated with a more aggressive RA phenotype. We analyzed whether M-ficolin levels were associated with the presence of 1 of these 3 prognostic markers but observed no differences at baseline or 1 year (P > 0.49).

Correlations between M-ficolin concentrations and disease activity markers

Because a pattern of higher M-ficolin levels was associated with higher disease activity, we analyzed whether M-ficolin concentrations correlated with markers of disease activity in RA. As shown in Table 2, the M-ficolin level correlated with a wide range of disease activity markers, each of which reflected different aspects of the disease. The most striking correlations at baseline were with the DAS28 and 3 of the 4 variables constituting the DAS28 (CRP level, swollen joint count, and physician's global assessment). At 1 year, the correlations between the M-ficolin level and disease activity markers were almost identical to those observed at baseline. In addition, a strong correlation between the M-ficolin level and the HAQ score was observed at both baseline and 1 year (Table 2). The paired samples available from the OPERA cohort enabled calculation of the relative change in the M-ficolin concentration expressed as a ratio. The M-ficolin ratio correlated uniformly with changes in the disease activity markers, except the tender joint count in 28 joints (Table 2).

Table 2. Correlations between the M-ficolin concentration and disease activity markers*
Disease activity markerBaseline (n = 180)Year 1 (n = 170)Change from baseline to year 1
ρPρPρP
  1. DAS28 = Disease Activity Score in 28 joints; VAS = visual analog scale; HAQ = Health Assessment Questionnaire.
DAS280.25<0.0010.220.0040.240.002
C-reactive protein0.36<0.0010.150.050.39<0.001
Tender joint count in 28 joints0.220.220.240.0020.120.13
Swollen joint count in 28 joints0.140.050.160.040.220.004
Pain VAS score0.100.170.120.130.200.009
Physician's global assessment VAS score0.180.010.180.020.230.002
HAQ score0.250.0030.230.0030.32<0.001

Effect of M-ficolin levels in predicting the DAS28.

An elevated M-ficolin level, as both a continuous variable and a dichotomized variable, had the highest explanatory power in the univariate model of an unfavorable DAS28 after 1 year, with odds ratios (ORs) ranging from 3.11 to 6.50 for not achieving remission and from 2.45 to 5.60 for not achieving low disease activity (Table 3). Furthermore, neutrophil count, sex, and treatment group were significantly associated with remission, while no variables except M-ficolin were associated with low disease activity. Multiple logistic regression analysis with backward selection showed that elevated M-ficolin levels were the strongest predictor of not achieving DAS28 remission after 1 year (OR 4.18, 95% CI 2.02–8.63), followed by treatment group (Table 3). Elevated M-ficolin levels were predictive of not achieving low disease activity (OR 2.45, 95% CI 1.13–5.28), while none of the other variables were predictive. We investigated and observed no interaction between the M-ficolin level and treatment with regard to disease activity. A completers analysis and use of a Simplified Disease Activity Index ([14]) of <3.3 as the end point showed similar results (data not shown). Thus, in all models, low M-ficolin levels were the strongest predictor of a favorable DAS28 after 1 year.

Table 3. Univariate and multiple regression analyses*
VariableRemission, DAS28 <2.6 not achievedLow disease activity, DAS28 <3.2 not achieved
OR (95% CI)POR (95% CI)P
  1. Radiographic erosions were defined as a Sharp/van der Heijde erosion score of ≥1. DAS28 = Disease Activity Score in 28 joints; OR = odds ratio; 95% CI = 95% confidence interval; anti-CCP = anti–cyclic citrullinated peptide; IgM-RF = IgM rheumatoid factor.
Univariate analyses    
M-ficolin3.77 (1.91–7.45<0.0012.45 (1.13–5.28)0.02
Dichotomized at 25%6.50 (2.19–19.31)0.0015.60 (1.27–24.62)0.02
Dichotomized at 33%3.11 (1.43–6.78)0.0043.84 (1.27–11.62)0.02
Dichotomized at 50%3.83 (1.92–7.65)<0.0012.77 (1.19–6.47)0.02
Sex0.43 (0.21–0.91)0.030.45 (0.17–1.17)0.10
Age1.00 (0.97–1.01)0.671.00 (0.97–1.02)0.97
Neutrophil count1.28 (1.07–1.52)0.0071.11 (0.91–1.36)0.30
Monocyte count2.01 (0.50–8.04)0.321.64 (0.32–8.38)0.55
C-reactive protein1.00 (0.99–1.00)0.320.99 (0.98–1.00)0.22
DAS281.17 (0.88–1.57)0.291.12 (0.79–1.60)0.52
Anti-CCP1.28 (0.65–2.52)0.480.81 (0.36–1.81)0.61
IgM-RF0.90 (0.44–1.80)0.760.93 (0.39–2.21)0.88
Treatment group3.12 (1.59–6.12)0.0011.21 (0.55–2.67)0.64
Baseline radiographic erosions1.19 (0.61–2.32)0.601.9 (0.82–4.38)0.13
Initial models of multiple logistic regression analyses with backward selection    
M-ficolin as continuous variable3.01 (1.34–6.76)0.0082.21 (0.91–5.38)0.08
Sex0.49 (0.21–1.12)0.090.47 (0.17–1.30)0.15
Age0.99 (0.96–1.010.280.99 (0.96–1.02)0.47
Neutrophil count1.23 (0.98–1.54)0.071.05 (0.81–1.35)0.72
DAS280.97 (0.68–1.40)0.871.00 (0.67–1.52)0.98
Anti-CCP0.91 (0.40–2.06)0.830.65 (0.26–1.63)0.36
Treatment group0.29 (0.14–0.61)0.0010.79 (0.34–1.83)0.59
Baseline radiographic erosions1.19 (0.57–2.49)0.651.99 (0.83–4.72)0.12
Final models of multiple logistic regression analyses with backward selection    
M-ficolin as continuous variable4.18 (2.02–8.63)<0.0012.45 (1.13–5.28)0.02
Treatment3.52 (1.69–7.36)0.001

Because a low M-ficolin level at baseline was the only variable associated with low disease activity at 1 year, M-ficolin was analyzed as a prognostic marker. Patients in the OPERA cohort were divided into 2 groups. One of the groups included the 25% of patients with the lowest M-ficolin levels (cutoff 2.00 μg/ml), and the second group comprised the remaining 75% of patients. Only 2 of the 42 patients in the group with the lowest M-ficolin levels failed to achieve a DAS28 of <3.2 at 1 year, compared with 28 of the 128 patients in the other group. Thus, the presence of an M-ficolin level in the lowest quartile at baseline resulted in low sensitivity (29%), high specificity (93%), a high positive predictive value (95%), and a positive likelihood ratio of 4.3. Therefore, a patient whose baseline M-ficolin level was among the 25% lowest levels had a 95% chance of achieving a DAS28 of <3.2 at 1 year, irrespective of the treatment given, anti-CCP status, and other well-known prognostic factors.

DISCUSSION

Our data demonstrated that increased circulating M-ficolin levels were associated with higher disease activity in patients with RA, when analyzed at both the group level and at an individual level. M-ficolin was associated with numerous disease activity markers in DMARD-naive patients with early RA, most notably the DAS28 and the HAQ, at both baseline (high disease activity) and 1 year (low disease activity). This suggests that M-ficolin is a biomarker that reflects essential aspects of RA disease activity. We also demonstrated that M-ficolin levels at baseline were the strongest predictor of remission and low disease activity, and that baseline M-ficolin levels in the lowest 25% resulted in a 95% chance of achieving low disease activity at 1 year.

M-ficolin is synthesized by peripheral blood leukocytes and bone marrow cells, and synthesis can be induced by proinflammatory stimuli ([6]). M-ficolin is exocytosed through granules localized in the cytoplasm of monocytes and neutrophils, and plasma M-ficolin concentrations in patients with RA are correlated with the number of neutrophils and monocytes ([7]). Neutrophils are the first and most abundant inflammatory cells present in inflamed synovial fluid and have a very important role in the establishment of inflammation. When activated, neutrophils can perform most of the functions of macrophages, i.e., produce and secrete proinflammatory cytokines and chemokines, and express class II major histocompatibility complex, which allows presentation of antigen to and activation of T cells ([15]). Furthermore, evidence is accumulating from animal models of arthritis that neutrophils contribute to both the initiation and the progression of disease ([15]). One could assume that M-ficolin indicates the sum of the “activated neutrophils” and thereby acts as a novel inflammatory marker reflecting disease activity in patients with RA.

H-ficolin and L-ficolin have the ability to bind to apoptotic cells and mediate their clearance through complement activation ([16]). Because M-ficolin and L-ficolin are very similar, with amino acid sequences that are 77% identical ([2]), one could hypothesize that M-ficolin could evoke similar activation. This creates secondary inflammatory damage and augments systemic inflammation, implying a direct mode of action for M-ficolin in the pathogenesis of RA.

The CRP level reflects disease activity in a manner that is fundamentally different from that of M-ficolin. In contrast to M-ficolin, CRP is synthesized by the liver and, in the current study, was not predictive of disease activity. One known caveat with respect to CRP and its role as an inflammatory marker in RA is that despite having obvious clinical signs of systemic inflammation, ∼40% of patients with newly diagnosed RA have a normal CRP level ([17]).

A strength of this study is the inclusion of a large number of patients with early RA and an immune system not exposed to DMARDs and steroids, for whom paired followup samples were available after 1 year ([10]). M-ficolin should be evaluated against other disease activity markers, including joint destruction, and analysis of more time points in the OPERA cohort would add to our understanding. Furthermore, studies of additional cohorts are needed to validate our findings in a broader selection of patients with RA before M-ficolin can be used as a biomarker in the clinical setting.

M-ficolin has unique prognostic abilities to pinpoint a group of patients with RA who will achieve low disease activity with a very high certainty, irrespective of other prognostic factors, although we are well aware of the limitations in terms of the low specificity. Previous studies and previously developed prediction rules have failed to identify a single prognostic marker that can be used on an individual basis in daily clinical practice ([18]). Due to the very diverse nature of the disease, it is unlikely that a single marker will have predictive value at the individual level in a cohort of patients with RA.

We find it very promising that M-ficolin was the strongest predictor of remission and low disease activity in both the univariate and multivariate regression models. We are conducting additional studies to elucidate the clinical potential of M-ficolin.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Ammitzbøll had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Ammitzbøll, Thiel, Jensenius, Ellingsen, Hørslev-Petersen, Junker, Krogh, Østergaard, Stengaard-Pedersen.

Acquisition of data. Ammitzbøll, Ellingsen, Hørslev-Petersen, Hetland, Junker, Krogh, Stengaard-Pedersen.

Analysis and interpretation of data. Ammitzbøll, Thiel, Jensenius, Ellingsen, Hørslev-Petersen, Hetland, Krogh, Stengaard-Pedersen.

Acknowledgments

We thank the OPERA Study Group. We also thank personnel at Arrhus University Hospital (Department of Clinical Immunology) and Julia Johansen (Copenhagen University Hospital, Glostrup, Denmark) for the handling of samples.

Ancillary