Rheumatoid arthritis (RA) is a complex inflammatory autoimmune disease characterized by disturbances in T-cell and B-cell functions. Clinical and animal studies highlight the multiple roles of B cells in the development and severity of RA, including production of autoantibodies, inflammatory cytokines such as TNF and IL-6, and aberrant antigen presentation [1]. Recent data in animal models suggest that, among these effects, the antigen-presenting capacity of B cells may be of particular importance in RA pathogenesis [2].

Rituximab is a chimeric mAb against CD20 that induces a profound depletion of B cells in the peripheral blood of RA patients [3]; however, little is known about the qualitative and quantitative aspects of deletion achieved in tissues including the bone marrow (BM) and lymph nodes. Initial studies in humans with RA suggest that B cells are also depleted in the BM as well as in the synovium [4]; nevertheless, the depletion is rather incomplete [5].

The BM is important for the biology of B cells as it represents a site of B-cell differentiation and maturation. The BM is an immunologically privileged site, where stroma promote B-cell survival and thus may protect B cells from depleting therapies [6]. We have previously described quantitative and qualitative changes in the BM of RA patients, which could affect a variety of BM resident cells including the B cells [7, 8]. Furthermore, based on gene expression studies of lupus patients, we have reported that the BM may be more informative than peripheral blood (PB) in differentiating active from inactive lupus patients and in differentiating patients from control individuals [9]. In the present article we explored the effect of rituximab treatment on B-cell subpopulations in the periphery and in the BM in a cohort of RA patients with resistant disease.

In the present study, we investigated the effects of rituximab on PB and BM lymphocyte subsets in patients with active RA. Rituximab effectively reduced CD19⁺ B cells in the PB of patients; a less pronounced effect was observed in the BM. Importantly, rituximab therapy preferentially depleted activated CD19⁺HLA-DR⁺ B cells in both compartments. The relative low clinical response rate in our population probably reflects the fact that these patients represent the most severe and refractory subset among our cohort, with several of them having failed multiple biologic therapies.

Expression and signaling through MHC class II molecules is important for effective antigen presentation and induction of co-stimulatory molecules on B cells [12]. Animal models of inflammatory arthritis support an important role of B cells in disease pathogenesis through a variety of mechanisms, including aberrant antigen presentation and activation of autoreactive T cells in the joint [13]. HLA-DR⁺ B cells have been suggested to be important for B-cell-mediated T-cell activation; depletion of HLA-DR⁺ cells could therefore represent an additional mechanism for the beneficial effects of rituximab in RA.

The BM is a primary lymphoid organ where B-cell differentiation and maturation occurs; BM stroma promote B-cell survival and protect B cells from depleting therapies [14, 15]. In accordance with this observation, we found that rituximab therapy efficiently reduced PB CD19⁺ B cells but had only a weak effect on BM CD19⁺ B cells. The change in BM B cells did not correlate with clinical response to therapy or other disease parameters in the small patient number tested.

Previous studies have suggested the depleting effect of rituximab may be more pronounced in the PB than in inflamed tissues or lymphoid organs. Kavanaugh and colleagues found that - unlike PB B cells, which were profoundly (>95%) depleted after rituximab therapy - synovial B cells decreased but were not eliminated [16]. Similarly, in 24 RA patients who received rituximab, PB B cells were depleted at 4 weeks whereas synovial B cells were only moderately reduced and persisted in five patients [5]. Local tissue expression of B-cell survival factors (for example, B-cell activation factor belonging to the TNF family, stromal cell-derived factor-1, macrophage migration inhibitory factor) [17–19] might explain the relative resistance of B cells against the depleting effect of rituximab. Moreover, data both from animal models [20] and from humans [4, 21] have shown that the effect of B-cell depleting therapies in lymphoid organs is variable. To this end, Teng and colleagues found that only 32% of RA patients had complete depletion of CD19⁺ cells within the BM as compared with 100% in the PB [4]. Leandro and colleagues reported a variable degree of B-cell depletion within the BM [21].

In our cohort, four out of 11 patients had incomplete B-cell depletion in PB - a relatively high proportion compared with that observed by other studies. Nevertheless, resistance to rituximab has been described in autoimmune-prone animals. In MRL/lpr mice and in NOD mice it has therefore been shown that B cells are relatively refractory to depletion compared with nonautoimmune-prone strains [20, 22]. In lupus-prone mice, high doses and extended duration of rituximab has been shown to overcome resistance to depletion; while in NOD mice, deficient FcγRI binding to IgG_2a CD20 mAbs and reduced splenic monocyte numbers have been implicated in impaired B-cell depletion. The importance of FcγR in antibody-dependent cell-mediated cytotoxicity as a mechanism for B-cell depletion has been also shown in lupus patients treated with rituximab [23]. Whether these mechanisms are involved in reduced B-cell depletion in our patients needs to be addressed.

The significance of incomplete B-cell depletion at the sites of inflammation has not been fully determined, but investigators have suggested that disease progression - despite PB B-cell depletion - is probably due to survival of memory B cells, which expand in secondary lymphoid tissues and produce autoantibodies. In line with this hypothesis, Leandro and colleagues reported that patients who relapsed early after repopulation of PB B cells had a trend (P = 0.170) for a higher percentage of CD19⁺CD27⁺ memory B cells compared with those patients that relapse later [24]. Roll and colleagues also performed immunophenotyping in the PB of RA patients who received rituximab, and reported a significantly higher proportion of IgD⁺CD27⁺ memory B cells during B-cell recovery in nonresponders [25]. Moreover, Thurlings and colleagues found that the reduction of synovial tissue plasma cells, which are normally derived by memory B cells, between weeks 4 and 16 after treatment predicted clinical improvement at 24 weeks [5]. Accordingly, Möller and colleagues reported that low numbers of peripheral blood memory B cells (IgD^-CD27⁺) correlated with good clinical responses in RA patients treated with rituximab [26].

Our data further support these findings since we observed a significant decrease in CD19⁺CD27⁺ memory B cells not only in the PB but also in the BM of RA patients who subsequently responded to rituximab. In contrast, memory B cells in nonresponders remained stable or even increased in both compartments. Together these data suggest that analysis of memory B-cell subsets may provide important information on the efficacy and response to rituximab therapy, consistent with the notion that targeting cells with the memory phenotype is a key determinant for its efficacy. Whether persistence of PB and/or BM memory B cells is related to the lack of response after rituximab retreatment observed in RA patients who did not exhibit clinical improvement after the first treatment course [27] remains to be determined.

Previous studies have indicated that rituximab may affect T-cell responses in systemic lupus erythematosus patients through downregulation of activation and co-stimulatory molecules (for example, CD69, CD154). This effect was attributed to decreased antigen-presenting function of B cells after rituximab treatment. We assessed the population of activated CD4⁺CD25⁺ and CD4⁺HLA-DR⁺ T cells, but we found no consistent changes according to response to therapy. Similarly, Thurlings and colleagues found no effect of rituximab therapy on PB CD4⁺ or CD8⁺ T cells of RA patients [5], whereas Leandro and colleagues reported a decrease only in a small subpopulation of CD3⁺CD20⁺ cells while the total CD4⁺ subset and the activated CD4⁺ subset remained unaffected [24]. Nevertheless, an effect on T-cell activation/function cannot be excluded. Therefore, although B-cell depletion in an animal study had no effect on T-cell subsets and activation markers, it impaired adaptive and autoreactive CD4⁺ T-cell activation [28].

Of note in small uncontrolled trials of systemic lupus erythematosus patients, it has been reported that rituximab reduces activated CD4⁺ cells in parallel with B cells. To this end, Sfikakis and colleagues reported that lupus patients responding to rituximab downregulated CD40 ligand, CD69 and HLA-DR expression on CD4⁺ cells [29], while Tokunaga and colleagues reported downregulation of CD40 ligand, inducible costimulatory molecule (ICOS) and CD69 on CD4⁺-positive cells in patients (n = 3) with active systemic lupus erythematosus [30, 31]. Whether the effect of rituximab on activated T cells is disease specific remains to be seen.

Response rates according the European League Against Rheumatism criteria (45% good and moderate responders) were moderate in our cohort. Nevertheless, this was a group of RA patients who were highly active refractory to previous treatment. All but one patient had failed in treatment with anti-TNFα agents (up to three agents). Although we found no correlation between clinical response and the degree of peripheral B-cell reduction, incomplete B-cell depletion (four out of 11 patients) could potentially be one of the reasons for lower clinical responses.

Rituximab effectively reduced PB CD19⁺ B cells in RA patients whereas its effect was less pronounced in the BM. Activated CD19⁺HLA-DR⁺ cells were significantly reduced in both compartments. Importantly, clinical response was associated with the persistence of memory B cells in both compartments.