Proliferative Synovitis, an Ultrasound Pattern Associated with ACPA positive Patients, Erosive Disease and Enhanced Need to Change Therapy in Rheumatoid Arthritis

Objectives: To analyze ultrasound (US) differences between rheumatoid arthritis (RA) patients according to their autoantibody status and characterize the clinical, immunological and radiological features associated with the US pattern of seropositive patients. Methods: We collected clinical and immunological data along with bilateral hand US images of RA patients. Serum biomarkers, MRI of dominant hand and immunostaining of synovial biopsies were performed. Results: Two hundred and ve RA patients were collected (84.8% seropositive). No signicant differences in disease activity/therapy were found according to autoantibodies status. An extreme proliferative US pattern, encompassing synovial hypertrophy grade II-III with Power Doppler signal that we called US Proliferative Synovitis (US PS) was present in 55.5% of seropositive and 16.1% of seronegative patients, (p=0.0001). In the multivariate analysis, erosions [OR 4.90 CI 95% (2.17-11.07). p=0.0001] and ACPA [OR 3.5 CI 95% (1.39-10.7), p=0.09] but not RF status [OR 0.74 CI 95% (0.31-1.71), p=0.483] were independently associated with the presence of US PS. Ninety-four per cent of joints with US PS scored 2-3 in RAMRIS synovitis sub-index. At synovial level, US PS was signicantly associated with higher density of vessels (p=0.042). Moreover, signicantly higher serum levels of angiogenic and inammatory cytokines were found in patients with US PS. After a mean of 46 months of follow-up, US PS was independently associated with change of therapy (OR 2.63, 95% CI 1.20-5.77, p=0.016). Conclusions: ACPA+ RA was associated with US PS. This US pattern signicantly detected erosive disease and an enhanced need to change therapy in the long-term.


Introduction
Ultrasound (US) is a non-invasive imaging technique playing an important role for the assessment of patients with rheumatoid arthritis (RA). The use of US is relevant along the different stages of the disease, more sensitive and reliable than physical examination and has a potential role to guide therapeutic interventions [1].
US ndings in patients with chronic in ammatory arthritis include synovial proliferation or synovial hypertrophy (SH), neovessels or Power Doppler (PD) signal, tenosynovitis or soft tissue oedema. However, these imaging ndings are not always present and may differ depending on the speci c diagnosis. Different US patterns have been described for early RA and psoriatic arthritis (PsA) [2]. Whereas synovial involvement was the hallmark of RA patients, soft tissue changes were more frequently found in early PsA.

Page 4/17
Seropositive (sero+) RA is a very well-de ned disease. Its genetic, synovial pathology and immunological pro le are well known. A worse clinical prognosis and more radiological progression seem to characterize sero + RA patients [3][4]. However, up to 30% of RA patients are negative for rheumatoid factor (RF) and anti-citrullinated protein antibodies (ACPA). These patients are more heterogeneous and not so well de ned as sero + RA [5].
Furthermore, we and others have previously reported by arthroscopy a different synovial vascular pattern for sero + RA, characterized by straight vessels, and seronegative (sero-) RA, de ned by tortuous vessels, similar to PsA pattern [6][7]. Interestingly, those different synovial vascular patterns were associated with different radiographic damage, with straight and mixed patterns having signi cantly more erosive disease. However, another immunohistological study in our center could not nd differences in synovial vessel density according to ACPA status [8].
Few imaging studies have analyzed differences in RA according to autoantibody status [9][10]. In a recent study using Magnetic Resonance imaging (MRI), bone marrow oedema was associated with the combination of RF and ACPA. However, no association between autoantibodies serum levels and osteitis score was found [9].
We hypothesize that, similarly to as it is visualized by arthroscopy, sero-RA patients could have a different US pattern as compared to sero + RA, characterized by less synovial proliferation. The aim of this study was to analyze differences in the US synovial pattern between patients with sero + and sero-RA. Speci cally, we aimed to characterize the clinical, immunological and radiological features of the US pattern associated with sero + RA.

Patients
Observational study. Consecutive patients from the Rheumatology Department (Hospital Clinic, Barcelona, Spain) meeting American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) 2010 criteria for RA [11] were enrolled. Clinical, immunological, demographic and US images of patients were collected. Seropositivity was de ned as the presence of RF (>50 UI/ml) and/or ACPA (>50 UI/ml) at least twice at any time of the disease course. Probable RA or overlap syndromes were excluded. Before undergoing US, MRI, serological test or synovial biopsies, informed consent was signed. This study was conducted in accordance with the principles of the Declaration of Helsinki.

Ultrasound Assessment
All sonographic assessments were performed using high-sensitivity US equipment (Acuson Antares®, Siemens AG, Erlangen, Germany) with a linear probe of frequency range from 8 to 14 MHz. Intraarticular in ltrations were not allowed one month previous to the US assessment. US ndings were de ned according to published OMERACT de nitions [12]. An experienced sonographer evaluated 6 joints of each hand (including MCP joints and wrists) for SH and intra-articular PD signals according to EULAR guidelines [13]. All evaluations were scanned on the dorsal aspect using longitudinal midline and transversal planes. PD calibrations and SH/PD assessment have been previously described [14].
We evaluated the morphology of synovial tissue. Speci cally, we looked to identify an extreme proliferative pattern de ned as the marked thickness of synovial membrane with PD signal inside the edges of the joint (Figure 1. A). Synovial thickening should bulge over the line linking tops of the periarticular bones and with/without extension to one of the bone diaphysis, similarly to the de nition of synovial proliferation grade II or III adopted by Szkudlarek et al [15]. We adopted the term "US proliferative synovitis" (US PS) to de ne these features. On the contrary, the at growth of synovial lling the angle between the periarticular bones with PD signal inside the joint, was de ned as " at synovitis", a concept equivalent to SH grade I (Figure 1.B). Both US patterns included the presence of PD signal.
Four rheumatologists, blinded to clinical data and autoantibody status of patients, scored all the images.
Interobserver reliability was evaluated before patients' inclusion by scoring SH/PD in 50-recorded images of joints from 20 RA patients. Interobserver correlations were all good to excellent (range 0.608-0.831).

Histological and immunohistochemical assessments
All the US-guided synovial biopsies were performed in a case day surgery according to the technique previously described by Kelly et al [16]. Biopsies were taken from selected patients who signed the informed consent. 6-8 synovial biopsies were taken per procedure.

Digital image analysis
The stained slides were scored by digital image analysis by an independent observer. Each stained slide was scored by dividing it in different regions. Within each region, the number of stained cells per area and the percentage of stained cells were measured in at least 20 high-power elds using the AnalySIS ® Imaging processing program (Olympus ® ) as previously described [18].

MRI Assessment
Selected patients who signed informed consent were scanned on a 1.5 Tesla system (Siemens Aera, Siemens Medical, Erlangen, Germany) using a dedicated wrist coil. MRI protocol was previously reported [19].
Images were reviewed on a standard Dicom (Digital Imaging and Communication in Medicine) compliance workstation. Images were scored by two independent radiologists blinded to image time point and patient identity using the Rheumatoid Arthritis Magnetic Resonance Imaging Scoring (RAMRIS) system [20].
Quanti cation of biomarkers of in ammation/angiogenesis.
Cytokines and angiogenic mediators were analyzed using Quantibody® Human Custom Array (RayBiotech, Norcross, GA, USA)1 [8]. Detection limits for cytokines are displayed on the manufacturer's website [21]. After sample dilution, the effect of RF on the nal results was estimated to be around 1% [22].
Calprotectin serum levels were determined using an ELISA Test Kit [CALPROLAB Calprotectin ELISA (ALP) CALPRO AS, Norway] in accordance with the manufacturer's protocol [23].

Clinical Follow-up
At baseline and every fourth months patients underwent complete clinical and biological assessment.
Treatment change was de ned as the change in any conventional synthetic Disease-modifying drugs (csDMARDs) or biological synthetic Disease-modifying drugs (bDMARDS) (but not glucocorticoids) across the observational period.

Statistical analysis
Epidemiological, clinical and US variables were compared between patients with or without RF or ACPA and with US PS or at synovitis. Numerical variables were described as mean and standard deviation (SD) and categorical variables as frequencies and percentages. T-student test was used to compare the distribution of numerical variables between groups. Chi-quared test was used to compare categorical variables. To ascertain independent associations between variables we used multivariate analysis. For all tests, p values ≤ 0.05 were considered signi cant. All analyses were performed using the SPSS 18.00 (SPSS Inc., Chicago, IL, USA).

Results
Clinical characteristics of the cohort Two hundred and ve RA patients were collected. Demographic and clinical characteristics are shown in Table 1. All patients had predominant hand involvement at US assessment. Mean age was 57.1 (SD 14.1) years and the mean time of disease evolution was 113.3 (SD 105.7) months. Mean Disease Activity Score 28-joint count C reactive protein (DAS28-CRP) was 2.55 (SD 1.03). Overall, 173 out of 205 (84.8%) patients were sero + for either RF or ACPA autoantibodies. One hundred and eight patients (53.7%) had radiographic erosions in hand/feet. At baseline, 75.6% were taking csDMARDs, 34.3% bDMARDS, and 49.3% low doses of glucocorticoids (≤ 5 mg of prednisone or equivalent). US proliferative synovitis is associated with higher density of synovial vessels and synovial cell in ltrates To better characterize US PS, we performed synovial biopsies in a subgroup of 23 patients with either proliferative (13 patients) or at (10 patients) synovitis. Synovium of patients with US PS had signi cantly higher density of vessels (p = 0.042) and higher, but non-signi cant, trend in density of B, T, mast cells and macrophages (Fig. 2).

MRI assessment
To verify the concordance between MRI and US on the concept of proliferative and at synovitis, 42 patients underwent MRI of the dominant hand, 17 of them (40.4%) with US PS. At patient level, no signi cant differences regarding RAMRIS and its individual components (erosion, synovitis and bone oedema) were found according to the presence of US PS.
After analysing the speci c joints with US PS, all but one scored 2 (58.8%) or 3 (35%) in the MRI synovitis sub-index. On the contrary, joints with US at synovitis predominantly scored 1 in MRI synovitis sub-index, 10 patients scored 2 and none of the joints with US at synovitis scored 3. No remarkable ndings were found on MRI erosion or MRI bone oedema sub-indexes between joints with either US PS or at synovitis. Interestingly, serum levels of calprotectin at the end of the follow-up were numerically higher in patients with US PS (2.5 mg/dl versus 1.9 mg/dl), although signi cant differences were not reached (p = 0.093).

Discussion
This study is focused on US differences according to autoantibody status in RA patients. After performing over 400 hands US in RA patients, we observed that ACPA+ (but not RF) had a signi cantly more proliferative pattern at synovial level that we called US PS, equivalent to SH grade II-III, whereas sero-patients had predominantly a atter US pattern, equivalent to SH grade I. The presence of US PS also identi ed a subgroup of RA patients with erosive disease, more synovial vessels, higher serum levels of both angiogenic and in ammatory biomarkers and an enhanced need for changing the baseline therapy in the long-term.
In ammation and angiogenesis of the synovial membrane are the hallmarks of arthritis. Initially, synovitis was thought to be non-speci c, with no differences among different subtypes of arthritis. However, immunohistochemistry studies have shown that vessels morphology and cell in ltration may differ between RA and spondyloarthritis (SpA) patients. Whereas straight vessels and lining hyperplasia have been related to RA, tortuous vascular pattern and in ltrates with predominant innate immune cells such as neutrophils and mast cells have been related to PsA and SpA [24][25]. Comparing PsA and sero-RA, histological analysis of synovial tissue composition has revealed similarities in cell distribution. Synovial tissue of PsA patients has been found to be enriched in CD117 + cells in the sublining area.
Conversely, synovial tissue of sero-RA patients has been found to be enriched in CD138 + cells [26]. Therefore, the study of synovial tissue could be relevant to provide clues for a de nitive diagnosis in undifferentiated arthritis [27].
Few studies have focused on imaging differences in RA according to autoantibody status. A recent study using US found that PD perfusion patterns were different in sero + and sero-RA. The difference appeared to be ACPA, but not RF dependent, suggesting that the pathophysiological process is different in ACPA + and ACPA-RA [10].
Gatehold et al found that erosion load differed signi cantly between sero + and sero-RA. Joint space narrowing scores were greater in sero + RA. The qualitative comparison showed that sero-RA patients displayed periarticular ossi cations, carpal shortening, and sparing of the carpo-metacarpal joints, whereas sero + RA patients had more carpo-metacarpal damage and less shortening [28].
In this study, we observed that US PS was independently associated with ACPA status and not in uenced by disease duration, therapy or disease activity. The concept of US PS makes reference to the upper outgrowth of synovial with a convex upper edge and corresponds to SH grade II and III (always with PD). In most cases, US PS adopted a characteristic morphology, with capsular distension and convex shape on the upper top. Conversely, sero-RA patients generally showed a atter outgrowth of synovial tissue similar to what can be seen in PsA, a concept closer to SH grade I with PD.
We identi ed an independent association between US PS with erosive disease, even after analyzing only sero + patients (data not shown). In addition, more of them had to change the baseline therapy due to lack of response in the long-term. Therefore, US PS might be considered as a bad prognosis nding to considerate when making therapeutic decisions, although this issue should be de nitely addressed in prospective studies.
After verifying the independent association of US PS with both ACPA status and erosive disease, we sought to better characterize this US pattern using a double approach with MRI and immunochemistry. First, we veri ed that US PS at joint level corresponded almost unanimously to grade 2-3 MRI synovitis whereas US at synovitis corresponded predominantly to grade 1 MRI synovitis. Second, we observed a strong trend to higher density of B, T, mast cells and macrophages, and a signi cantly higher density of vessels in synovial biopsies from US PS, re ecting substantial divergences between these two patterns of synovitis beyond the US appearance.
Few studies have analyzed synovial tissue differences in ACPA+/-patients. Gómez-Puerta et al did not nd signi cant differences in synovial cell in ltrate or lymphoid neogenesis according to ACPA status [8].
Conversely, Orr C et al demonstrated that synovial B cell in ltrates and lymphoid aggregates were signi cantly higher in ACPA + patients [29]. We observed that patients with US PS also had a different synovial pattern and showed a worse response to the therapy. However, neither ACPA nor RF status were relevant for synovial in ltrates or determining clinical outcomes (data not shown). This data suggests that the presence of US PS, better than autoantibodies status, re ects a differential aspect at synovial level and a worse prognosis in the long-term.
Moreover, higher serum levels of angiogenic and in ammatory biomarkers were found in patients with US PS, con rming differences not only at local but also at peripheral level. We had previously seen higher levels of serum biomarkers in RA versus PsA patients in clinical remission [30]. Our RA patients with US at synovitis, most of them RF and ACPA-, showed lower levels of serum biomarkers, resembling PsA patients [30]. Therefore, sero-RA could represent a different entity, standing between sero + RA and PsA, as it has also been suggested in recent studies analyzing serum metabolomes and lipidomes [31].
This study has some limitations. First, four observers scored US images, which could enter too much variability, although intercorrelation ratio was good to excellent. Second, only wrist and MCP US images were evaluated. Considering that hand involvement is the hallmark of RA, this protocol should be enough to capture the disease activity. We do not know if these results could be reproduced in big joints, although the presence of synovial uid (barely seen in small joints) would hamper a proper evaluation of synovial proliferation.

Conclusions
The presence of US PS, encompassing SH grade II and III with PD, was independently associated with ACPA + RA. This US pattern identi ed a subgroup of RA patients with erosive disease, higher vessels density at synovial level and a greater need for changing the therapy in the long-term.

Consent for publication
Signed consent for publication was obtained from all patients.