Increased galectin-3 may serve as a serologic signature of pre-rheumatoid arthritis while markers of synovitis and cartilage do not differ between early undifferentiated arthritis subsets

Undifferentiated arthritis (UA) is a diagnostic label which is used to categorize patients with arthritis complaints which cannot be classified according to current criteria [1]. In a review on inception cohorts of UA patients with monoarthritis or polyarthritis, in which arthritis was required to be present at inclusion, 17–32% progressed to rheumatoid arthritis (RA) after 1 year of follow-up [2]. Among UA patients with at least two swollen joints enrolled in the Norfolk Arthritis Register, 25–28% were in remission after 2 years [3, 4]. Although a number of risk factors for the development of RA have been identified, including female gender, smoking, socioeconomic status [5], and seropositivity for antibodies to cyclic citrullinated peptides (anti-CCP) [6] even in asymptomatic individuals [7], it is still not possible to predict with adequate certainty how patients with UA will ultimately segregate into distinctive disease entities. Several studies have investigated serological markers including acute phase reactants, e.g., C-reactive protein (CRP), erythrocyte sedimentation rate (ESR) [8], and inflammation modulators, e.g., matrix metalloproteinases (MMPs) [9], as potential predictors for RA development in UA. A number of prediction models for the clinical course of UA based on disease phenotype, serologic autoantibodies, and imaging data have been proposed in order to improve identification of UA patients at high risk for RA development [10, 11]. Recently, Duer-Jensen et al. reported that bone marrow edema of the wrist and metatarsophalangeal (MTP) joints detected by magnetic resonance imaging (MRI) is an independent predictor of future RA in patients with UA of short duration [12]. Combining MRI findings with clinical parameters and autoantibodies yielded additional discriminant capacity. In UA, there is a particular need to develop serological disease markers reflecting distinctive metabolic and inflammatory disease pathways to facilitate early diagnosis and to improve monitoring.

Several lines of evidence have implicated galectin-3 as a pro-inflammatory mediator in RA. Synovial inflammation and structural joint damage was lower in galectin-3^–/– mice with antigen-induced arthritis as compared with wild-type controls [13]. Ohshima et al. [14] reported that galectin-3 in the serum and synovial fluid was increased in patients with long-standing RA compared with osteoarthritis (OA) and controls. These authors also reported that galectin-3 mRNA and galectin-3 binding protein were expressed in particular at sites of cartilage and bone destruction in RA joints [14]. Moreover, they found that, although galectin-3 was only expressed in a few cells in OA synovial tissue, leucocyte infiltration was associated with upregulation of galectin-3 expression. In a 1-year prospective study on patients with newly diagnosed RA before treatment initiation, we recently reported that galectin-3 was increased, particularly in anti-CCP-positive smokers. Furthermore, galectin-3 was associated with bone marrow edema score of the wrist at baseline and remained elevated during 12 months of follow-up despite a targeted, synovitis-suppressive treatment strategy [15]. Conversely, we have previously reported that a seromarker of cartilage regeneration (type IIA collagen N-terminal propeptide (PIIANP)) is decreased in newly diagnosed RA [16‐18] while hyaluronan (HYA), a marker of synovial swellings, is increased [19, 20].

Based on these observations, and the frequent occurrence of bone erosions in early RA before treatment [21, 22], we hypothesized that galectin-3 in the serum may be a pre-diagnostic signature of future RA. The aim of this investigation was to study the serum levels of galectin-3, PIIANP, and HYA in a cohort of patients with early UA (the TUDAR study [12]) and possible associations with disease profile, MRI findings, and molecular markers of cartilage remodeling and synovitis. In addition, we studied the capability of baseline galectin-3 to distinguish between pre-RA and non-RA according to diagnostic reassessment at 1 year follow-up.

One hundred and sixteen patients participated after exclusion of patients with other rheumatic diseases at baseline, an incomplete dataset, or lack of follow-up. Among the 111 patients with available baseline sera, 23 were subsequently diagnosed with RA (pre-RA), while 88 had persistent UA, remitted, or had developed alternative classifiable conditions (non-RA), e.g., OA or psoriatic arthritis (Fig. 1). The median follow-up time was 17 months (12–23 months) [12]. At baseline, the pre-RA patients and non-RA group were comparable with regard to age and gender distribution and symptom duration. However, patients classified as having RA at follow-up had overall higher CRP (p < 0.01), more swollen (p = 0.03) and tender joints (p = 0.05), as well as higher VAS, HAQ, and DAS28-CRP scores at the time of inclusion (Table 1).

When stratifying according to MRI findings in the two UA subgroups, pre-RA had higher wrist MRI synovitis scores (p < 0.01), bone marrow edema scores (p < 0.01) and erosion scores (p = 0.01). Regression analyses with either MRI MTP or wrist scores or MRI wrist + MTP scores yielded similar results. Accordingly, only MRI wrist data were included.

Outliers were identified based on z scores and included in the calculations. Concerning galectin-3, three outliers in the UA study population and two in the controls were identified. One PIIANP outlier was found in both the UA and control population. Similarly, two and five HYA outliers were identified in the UA and control population, respectively. Baseline serum levels of galectin-3, PIIANP, and HYA are presented in Table 2.

Patients presenting with joint complaints of uncertain origin are a common source of concern to clinicians with regard to diagnostic delay and missed therapeutic opportunities. In this first study on galectin-3 and selected seromarkers of joint tissue remodeling in patients with early, undifferentiated arthritis, we report that galectin-3 in serum was increased in pre-RA versus non-RA and a healthy control population. PIIANP, a marker of regenerative cartilage collagen IIA synthesis, was equally depressed in pre- and non-RA, while HYA was in the normal range in either subset. In the entire UA cohort, galectin-3 correlated positively with MRI bone marrow edema score, whereas PIIANP was associated with MRI erosion score and HYA with both MRI synovitis and erosion scores. Galectin-3 discriminated well between pre-RA and non-RA.

The present finding of increased galectin-3 levels in serum of patients with UA who subsequently developed RA as compared with other UA subsets adds to the experimental and clinical evidence that galectin-3 plays a critical role in RA development, e.g., by regulating RA synovial fibroblast functions. Thus, galectin-3 induces distinctive pro-inflammatory cytokine and chemokine expression profiles in RA synovial fibroblasts as compared with dermal fibroblasts [27]. In addition, Ohshima et al. reported that galectin-3 and its binding protein is overexpressed in the RA synovial membrane, particularly at the invading pannus front, and that galectin-3 in the serum and synovial fluid is increased in patients with long-standing RA compared with osteoarthritis and healthy controls [14]. Recently, in a cohort of patients with early, untreated RA we found persistently increased levels of circulating galectin-3 in serum despite subsequent introduction of a targeted synovitis suppressive treatment protocol [15]. Notably, baseline galectin-3 did not correlate with clinical disease activity measures such as tender and swollen joint counts or DAS28-CRP, but correlated positively with MRI erosion and anti-CCP, which is a strong predictor of erosive progression in RA [28]. In the present study on early UA, galectin-3 correlated with CRP but not with joint counts or DAS28-CRP score (Table 3). Galectin-3 also correlated with bone marrow edema, however, reaching statistical significance in the non-RA subset only. This finding may reflect that joint degradative pathways are shared at early stages of different arthritides. Also, the relatively small number of pre-RA patients should be considered. Altogether, these observations concord with evidence from studies using modern imaging technologies that erosive damage in RA may proceed in the absence of clinically detectable synovitis [29‐31]. Recently, this has also been reported to take place in joints with persistent bone marrow edema [32]. The abovementioned experimental and clinical studies and the association between increased galectin-3 levels in the serum and bone marrow edema score in UA and early RA independent of current synovitis suggest that galectin-3 may be an important mediator of joint destruction in arthritis.

Galectin-3 expression by RA synovial fibroblasts is stimulated by binding to cartilage oligomeric matrix protein (COMP), a structural cartilage matrix protein belonging to the thrombospondin family [33]. In order to characterize possible links between the increased galectin-3 levels and cartilage and synovial tissue remodeling, we performed additional measurements of serological markers of cartilage collagen regeneration (PIIANP) [16‐18] and synovitis-related swellings (hyaluronan) [20]. PIIANP was equally depressed in pre-RA and non-RA and correlated with MRI erosion score in the entire UA population, but not with the occurrence of anti-CCP (Table 3). Previously, we reported that circulating PIIANP was decreased in a large cohort of newly diagnosed, treatment-naive patients with RA classified according to the 1987 criteria, particularly in the anti-CCP subset. In that study, PIIANP did not associate with disease activity, but correlated inversely with anti-CCP titer [18] suggesting a chondrocyte suppressive effect by these autoantibodies [34‐36]. More recently, we reported that PIIANP was increased in active, untreated axial spondyloarthritis, particularly in those testing positive for human leukocyte antigen (HLA)-B27, probably reflecting enhanced enthesopathic tissue proliferation in axial joint disease [37]. In the present study on UA, PIIANP was not associated with anti-CCP status, which may be due to the small sample size of anti-CCP-positive patients. However, as in early RA, PIIANP was decreased in the present early UA cohort regardless of the follow-up diagnosis, suggesting that early synovitis of any origin may reduce cartilage regeneration and repair. This is supported by the correlation between decreased baseline PIIANP and MRI erosion score. The inverse correlation between PIIANP and VAS global assessment score probably reflects the link between deep cartilage remodeling and disease activity in UA.

HYA is a well-documented seromarker of synovitis mass in RA [19, 20]. Although the swollen joint count in the entire UA population and the two major subsets was low, wrist MRI synovitis scores ranged between 2–5, 3.5–5.5, and 2–5, respectively. This indicates that a threshold volume of synovial inflammation must be exceeded to be reflected in the circulation and that subclinical synovitis may go unnoticed if tracked by HYA recordings only. Interestingly, HYA correlated inversely with anti-CCP positivity using multivariate regression analysis (Table 3). The unanticipated higher levels of HYA in anti-CCP-negative UA patients may pertain to the report by van Oosterhout et al. that the synovial lining layer is thicker in anti-CCP-negative as compared with anti-CCP-positive RA patients [38].

The ability of galectin-3 to discriminate between pre-RA and non-RA was calculated by ROC analysis (Fig. 2). Galectin-3 alone distinguished fairly well between UA patients who later developed RA and those who did not, with an AUC of 0.64 and, for anti-CCP alone, 0.63. When combining galectin-3 and anti-CCP, AUC increased to 0.71, and with the further addition of bone marrow edema to 0.74. These quite similar figures indicate that galectin-3 may qualify as a future serological marker for prediction of RA development in UA.

Some issues of concern should be addressed. In this study we adopted the 1987 criteria for RA classification [23]. This implies that some individuals would have been categorized as RA according to the 2010 criteria [39], thereby qualifying for exclusion from the study. This potential source of categorization error is underscored by the finding that 7/81 of the non-RA participants tested positive for anti-CCP. On the other hand, this is a study aiming at identifying biologically plausible new seromarkers for clinical use. It was therefore essential to apply these candidate markers to patient subsets diagnosed with optimal specificity. Some patients with UA may only differentiate into RA beyond 12 months [2]. The number of anti-CCP-positive patients was rather small. Accordingly, regression subanalysis with stratification into sero-negatives and sero-positives was not feasible. Smoking status was not available. Major strengths include the collection of a large sample of carefully characterized patients with UA and complete follow-up at 1 year, which allows for separation of patients who did from those who did not develop RA.

Galectin-3 is increased at the pre-diagnostic stage of RA versus non-RA, and discriminates well between UA patients who will and those who will not develop RA within 12–23 months. UA is associated with a decreased cartilage collagen regenerative response irrespective of synovitis origin.