Rheumatoid arthritis (RA) is the most frequent chronic autoimmune rheumatism (prevalence of nearly 1% in the adult population) affecting synovial tissue in the small and large joints, leading to severe pain and disability. The synovial membrane of affected joints undergoes a series of alterations that result in the formation of a proliferative tissue (i.e. synovitis), responsible for definitive osteoarticular destructions. Although the pathological factors that initiate the inflammatory cascade responsible for rheumatoid synovitis have yet not been fully identified, the first event could be a non-specific inflammatory response leading to the activation of monocytes and the recruitment of macrophages in the synovial membrane. The cellular composition of rheumatoid synovitis includes resident hyperplastic synoviocytes, endothelial cells, T and B cells, dendritic cells and macrophages [1]. Although the process of tissue infiltration is still not fully understood, infiltrating cells in RA are often organized into distinct B-cell areas surrounded by T cells and follicular dendritic cells [2,3]. This cell organization reflects the T and B cells distribution that is found in secondary lymphoid organs, such as lymph nodes. Ectopic lymphoid structures are not specific for RA but are associated with chronic inflammatory autoimmune diseases, since they have also been identified in Sjögren’s syndrome [4] or in Hashimoto thyroiditis [5]. In RA, the synovial tissue is further characterized by hyperplasia of the intimal lining cell layer and infiltration of the sublining layer (stroma) by macrophages, B and T cells, and other inflammatory cells that promote inflammation and local tissue destruction [6].

RA is a heterogeneous disease with various phenotypes: with or without erosion, autoantibodies (such as anti-citrullinated peptide antibodies (ACPA)), or extra articular manifestations. Furthermore, the clinical presentation of RA can be quite heterogeneous with a spectrum ranging from mild to severe disease [7]. This heterogeneity is also found as a feature of synovial inflammation [6-9]. The variations in response to different types of anti-rheumatic treatment in RA further support this notion [10].

RA is a systemic disease but small joints injuries are the most visible manifestations. The preferential targeting of small joints in RA does not seem to be due to specific tissue autoantigen expression [11]. In fact it seems that the rheumatoid synovium provides a breeding ground for abnormal autoimmune responses leading to tissue destructions [2]. Formation of secondary lymphoid structures in lymph nodes shows how tissue microarchitecture determines the immune response [1,2]. This concept has been less studied in nonlymphoid tissues, while it could explain the pathogenesis of chronic inflammatory rheumatisms [2,12]. In RA, some studies have emphasized the importance of the synovium’s microarchitecture responsible for the interactions between cells, leading to an immune inflammatory response [13].

In rheumatoid synovitis, cell organization is complex [14]. Indeed the inflammatory infiltrate can assume well defined yet different forms between individuals with distinct molecular and cellular signatures [7,8,15]. A synovitis score allowing the differentiation between lowgrade and high-grade synovitis was developed by Krenn, and coworkers, to standardize histopathological description of the synovial membrane [16]. Three form of rheumatoid synovitis’s architecture can be observed. In the first form (25% of patients), the synovitis’ microarchitecture shows all characteristics of ectopic lymphoid tissue with germinal center- like formations. In this group, cells show a welldefined organization with distinct B and T cells areas and a network of follicular dendritic cells [17]. The existence of germinal center-like structure in the synovial tissue is indicative of a high level of organization allowing cell-cell contact and therefore cell activation and antibodies’ production [18,19]. In the second group, representing 25% of patients, the synovium shows signs of lymphocyte aggregates without follicular dendritic cells [1]. In the most frequent type of rheumatoid synovitis, the inflammatory infiltrate is diffuse with low level of organization [17]. Finally, granuloma formation can rarely be found in the RA synovium. Several groups have studied the effect of the different types of anti-rheumatic treatment on synovium tissue architecture and showed that there are various cellular responses depending on the type of drugs, suggesting that the synovitis microarchitecture contributes to the clinical phenotype of RA and to the therapeutic effects of immunomodulatory drugs [20-26].

From all these data, the histopathological analysis of synovial membrane is the only effective way to study cell distribution and synovial microarchitecture in RA even if is not the only disease showing synovial lesions. Indeed some other diseases such as degenerative, metabolic or other autoimmune diseases can lead to synovial alterations, the majority being forms of synovitis. Although a histological analysis is not done systematically, because it is not required for RA diagnosis, it may provide complementary information, which could guide the diagnosis and influence the therapeutic decision.

In this study, histopathological features of synovial tissue specimens from RA patients were compared to the synovial samples from patients with osteoarthritis (OA) in order to (1) Highlight the histological specificities of RA synovium and (2) ind subsets of RA patients according to their clinical and histological phenotypes.

Finally we discussed the current interest of histopathological analysis of RA synovium, considering actual data on the pathophysiological mechanisms and therapeutic targets in RA.

Patients and clinical material

Synovial tissue specimens were selected randomly from a bank of specimen including biopsies from RA (n=17) and OA (n=9) patients performed between 2001 and 2013. In the OA group, synovial tissue specimens were obtained from knee or hip joints during joint replacement surgery. In the RA group, synovial tissue samples were obtained thanks to a systematic biopsy during surgical synovectomie. In two cases, synovial tissue specimens were obtained during joint replacement. Clinical diagnosis for RA was based on ACR/EULAR 2010 criteria. Tissue specimens from patients who did not fit these criteria were excluded. Age (median, range), sex and articular localisation of the biopsy were collected in both groups. In the RA group, for each patient the presence (Yes/No) of erosion was recorded at the time of the biopsy as well as autoantibodies (rheumatoid factor and/or ACPA).

Histology

Histological features of the synovial membranes were carried out on routine haematoxylin-eosin-safran–stained (HES) slides. Each synovial tissue specimen was analysed by two different observers and analysis was done at the synovial site showing the strongest histological alterations. Krenn’s synovitis score was established for each sample [16]. In this score three items are analysed: the enlargement of the synovial lining cell layer, the density of the resident cells and the type of the inflammatory infiltrate. Each item is graded from 0 to 3 points, giving a maximal score of 9 points. A score between 2 to 4 points amounts to a low-grade synovitis and a score from 5 to 9 points to a high-grade synovitis. Then, we further reported synovial architecture for each synovial tissue specimen classifying tissue architecture according to the inflammatory infiltrate’s organization: no infiltrate, diffuse infiltrate, infiltrate forming lymphocyte aggregates and infiltrate forming ectopic germinal center-like structures.

Immunohistochemistry

Cell distribution in the inflammatory infiltrate in synovial tissue specimens was analysed using cell lineage markers by immunostaining. We used a monoclonal mouse anti-human CD3 antibody for T cells [27] at a dilution rate of 1/50 (clone F7.2.38, DakoCytomation, Copenhagen, Denmark), a monoclonal mouse anti-human CD20 antibody for B cells [28] at a dilution rate of 1/50 (clone L26, DakoCytomation, Copenhagen, Denmark) and a monoclonal mouse anti-human CD68 antibody for macrophages [29] at a dilution rate of 1/300 (clone KP1, DakoCytomation, Copenhagen, Denmark) using staining procedures described in the DakoCytomation specification sheet of each antibody. Each sample was analysed by two observers, using a semi quantitative method. Immunostaining was graded from 0 to 3 points: 0 point, for no immunostaining, 1 point for weak immunostaining, 2 points for moderate immunostaining and 3 points for intense immunostaining. Moreover, we described the shape of the stained-infiltrate, whether diffuse or aggregates-like.

Statistics

For statistical evaluation, frequencies and median values were chosen for descriptive statistics. The Mann-Whitney test was used to compare quantitative variables between two groups. Pearson correlation was used to investigate the association between the synovitis score and clinical features of the disease in the RA group. The Fisher test was used to find an association between the type of synovial architecture and the disease (OA or RA). It was also used to find an association in the RA group between the types of synovial architecture and clinical features of the disease. A p value less than or equal to 0.05 was considered statistically significant. Statistics were calculated using GraphPad Prism version 5.01.

Nine synovial tissue specimens were obtained from hip (n=3) and knee (n=6) joints from patients with a clinical diagnosis of OA. OA patients (55.5% of women) had a mean age of 65 years-old (56 to 85 years-old). In the RA group, 17 synovial tissue specimens were selected. Among them, 12 tissue specimens (70.5%) came from patients with erosive RA and 14 tissue specimens (82.3%) from patients with auto-antibodies (rheumatoid factor and/or ACPA). Synovial tissue specimens were obtained from wrist (n=11), knee (n=2) and shoulder (n=2). In 2 cases, the localisation of the biopsy was unknown. Patients (70.5% of women) in the RA group had a mean age of 59.1 years-old (38 to 78 years-old).

In the RA group, a high-grade synovitis score (≥5 points) was observed in 13/17 (76.5%) of synovial tissue specimens while in the remaining 4 samples, synovitis scores ranged from 2 to 4 points corresponding to low-grade synovitis (Table 1). In the OA group, 5/9 (55.5%) specimens were graded low, 3/9 (33.3%) high and in the last specimen there was no evidence of synovitis observed (Table 2). In the RA group, 50% of synovial tissue specimens were graded between 5 and 7 points. Despite outlier values in both groups and wide ranges (3 to 9 in RA and 1 to 8 in OA) a significant difference was observed between the 2 groups (p=0.05) (Figure 1). In the RA group, we then investigated whether synovitis score was associated with clinical features of RA such as osteoarticular destruction or auto-antibodies. We could not find significant association between Krenn’s synovitis score and the erosive profile of RA or the presence of auto-antibodies; although a trend for an association between high-grade synovitis score and erosive RA was observed (p=0.07).

Table 1: Results for Krenn’s synovitis score of synovium tissue specimens in the RA group.

Table 2: Results for Krenn’s synovitis score of synovium tissue specimens in the OA group.

Figure 1: Box plot of the synovitis scores for RA and OA groups (*p=0.05).

Tissue architecture types were not distributed as we would have expected in RA samples with 12/17 (70.5%) synovial tissue specimens showing lymphocyte aggregates, 4/17 (23.5%) specimens characterized by diffuse infiltration and 1 sample with no inflammatory infiltrate. No germinal center-like structure was observed. In 52.9% of tissue samples, the synovial lesions observed were heterogeneous in size and distribution of the infiltrate. In the OA group, 8/9 (88.8%) samples were displaying lymphocyte aggregates, which in most cases (77.7%) were homogeneous in size and distribution. We did not find any statistical difference between the 2 groups with respect to synovial architecture (p=0,34). There was also no association between synovial architecture in the RA group and the erosive profile of RA or positivity of auto-antibodies; although a trend was observed (p=0.11) between erosive RA and an inflammatory infiltrate well organized in lymphocytes aggregates.

In the immuno-histochemistry study, we could analysed cell distribution of macrophages and T and B cells in the synovial stroma and in the synovial lining cell layer in 9 synovial tissue specimens from the RA group and in 8 specimens from the OA group (Figure 2). We observed a significant difference between the RA and OA groups regarding the distribution of macrophages in the synovial stroma (p=0.02), with an increased number of macrophages in the RA group. In contrast, no difference were observed for B and T cells’ distribution, neither in the lining cell layer (B cells p=0.09, T cells p=1) nor in the synovial stroma (B cells p=0.1, T cells p=0.07).

Figure 2: Distribution of T and B cells, macrophages in the synovial lining cell layer and in the synovial stroma of synovial tissue specimens in RA and OA groups

In this study, the analysis of synovial tissue specimens according to Krenn’s synovitis score strengthens that it could be used to evaluate in a standardized way the inflammatory infiltrate lesions found in the synovium of patients with RA [16]. Indeed we found a significant statistical difference between the synovitis scores of RA and OA patients, which highlights the importance of inflammatory infiltrate in RA’s pathophysiology. However, in some tissue specimens of OA patients we found the same histological synovial lesions as in RA patients. Therefore Krenn’s synovitis score is not pathognomonic of RA, as described by Krenn himself who found a sensitivity of 61.7% and a specificity of 96.1% of his score [16]. One of the issues of histological analysis of synovial tissue specimens is the heterogeneity of the lesions observed on one sample, making difficult its interpretation. In this study, we chose to analyse the sites showing the strongest histopathological alterations in all tissue specimens, a method chosen by pathologists in histopathological tumour diagnosis, in order to standardize our analysis but, nevertheless, synovial inflammatory lesions can be difficult to associate with one disease [8,30]. Indeed, there are quite a number of inflammatory and non-inflammatory diseases that can give histological synovitis but very few, if not none, diseases show pathognomonic changes in the synovial membrane.

Regarding synovial architecture, our results are different from those found in the literature. Indeed, we did not find any germinal centerlike structure in tissue specimens from RA patients whereas in the literature, in approximately 25% of synovium from RA patients, infiltrates of T and B cells are organized into structures resembling lymph nodes with germinal centers [17]. Furthermore, we observed in 70% of synovial tissue specimens from RA patients an inflammatory infiltrate forming lymphocyte aggregates whereas it should only represent 25% of the tissue specimens. The explanation for those differences between our cohort and other studies could be the occurrence of biological drugs in the treatment of RA in the 2000s, which could have resulted in modifications of synovial inflammatory infiltrate in more recent tissue specimen from treated patients as the ones in this study. In support of this hypothesis, Hirohata et al. have demonstrated that TNF inhibitors alter the histological characteristics of the synovium from RA patients [21]. Another result of our study is the absence of difference in synovial architecture between RA and OA patients. This goes against the hypothesis according to which the type of synovial architecture could determine the local inflammatory response [1,11], but the small number of samples in our study does not allow us to conclude on this particular point.

We then tried to find an association between the type of synovial architecture and characteristics of the disease of RA patients. We observed a trend for association between the synovitis score and the erosive profile of RA that showed that the higher the synovitis score was the higher risk there was to have an erosive disease. However this result was not statistically significant, most probably because of our low number of patients. For many years now, other teams have tried to find a score that could correlate synovial lesions and the severity of RA [31,32]. Overall, they did not find any correlation between clinical features of RA patients and histological synovial lesions; with the exception of Koizumi et al. who found a correlation between his synovitis score and rheumatoid factor’s rate and the number of affected joints [33]. In the immuno-histochemistry study, we observed a significant difference between RA and OA patients regarding the distribution of macrophages in the synovial stroma. Indeed we observed an increased number of macrophages in the RA group, hence the conclusion that macrophages are less present in OA synovial infiltrate, even in a group of inflammatory OA patients with 88.8% samples displaying lymphocyte aggregates. This result reinforces the importance of the role of macrophages in RA synovial infiltrate while other studies emphasize that synovial macrophages are appropriate biomarkers for response to treatment in RA patients [34-36].

This study reinforces our knowledge upon the specificity of synovial architecture and cell distribution of inflammatory infiltrate in synovium of patients with RA. Therefore in RA, a systematic histological analysis of synovial biopsy could be interesting for early diagnosis, to establish a prognosis and for treated patient’s follow-up [21,37,38]. Indeed, even if a synovial biopsy guided by echography or arthroscopy is an invasive act, it can provide synovial tissue specimen of good quality from the most representative lesions of the joint [39,40]. Histological analysis could ease the early diagnosis of some cases in which RA diagnosis is difficult, such as in cases of no positive auto-antibody, in order to initiate a suited treatment as soon as possible. Furthermore, the immuno-histochemistry analysis of synovial tissue specimen could allow targeting more effectively the type of biological drug to initiate for each patient [8,41]. Indeed, for a patient presenting with a synovial inflammatory infiltrate formed by a majority of B cells, we would be more inclined to choose a treatment targeting B cells such as rituximab. Such data regarding the prediction of patients’ response to treatment are still missing and need to be developed. The problem bound to synovium’s invasive obtaining could be overlooked only if systematic histological analysis of synovial biopsy became a key element in early proactive management of patients with RA.

In conclusion, even if the histological analysis of synovial tissue is not part of the current criteria for RA diagnosis or prognosis, it may be helpful in case of difficult early diagnosis and should be further studied to better understand patients’ response to treatment.

The authors declare no conflict of interest.