Prospects for B-cell-targeted therapy in autoimmune disease

Reasons for targeting B cells in autoimmune disease date back to the discovery of autoantibodies over 50 yr ago [1]. The idea became of practical interest when anti-B cell monoclonal antibodies were developed in the early 1990s [2, 3]. Fortuitously, at about the same time it became clear that B cells are not simply the subordinate foot soldiers of an immune response but may be as important as T cells in its genesis and regulation. Moreover, it seemed possible that B cells might actually be the driving force behind human autoimmunity. The concept of therapeutic B-lymphocyte depletion (BLyD) emerged subsequently in the pages of this journal [4]. Concept was transformed into reality with the use of the anti-CD20 (i.e. anti-B cell) monoclonal antibody rituximab [5–19]. BLyD has provided clear evidence that B-cell targeting has therapeutic potential [5–19]. The evidence is most firmly established in rheumatoid arthritis (RA) [8] and much of the discussion below will focus on the use of BLyD in RA as a model for other autoimmune conditions. Nevertheless, it is important to consider B-cell targeting on a wider front, and how its application may be different for different disorders. BLyD therapy as designed for RA is quite different in approach from that for agents such as methotrexate or etanercept, which suppress inflammation only for as long as they are administered. BLyD was designed to induce sustained remission [4], and is perhaps closer to a traditional 500 mg course of intramuscular gold or high-dose cytotoxic therapy with stem cell rescue. This concept of ‘remission induction’ is familiar in the context of intravenous cyclophosphamide in systemic lupus erythematosus (SLE) or Wegener's granulomatosis. It has important implications for the use of the agent, and in particular dosage and timing. Trials of BLyD in RA have shown that it can induce major clinical responses in a significant proportion of patients [5–10]. A short course can reduce a C-reactive protein level of 90 to 1.5 and not infrequently leaves the patient with minimal joint swelling or tenderness. Initial studies in other autoantibody-associated diseases, such as immune thrombocytopenia and SLE, have suggested similar efficacy [11–14]. Severe cytopenias and progressive glomerular disease have in some cases resolved for at least 2 yr. Benefits have been reported in Wegener's granulomatosis, autoantibody-associated neuropathies, myositis, haemolytic anaemia and myasthenia gravis [15–19]. These responses have often been sustained way beyond the period of depletion, sometimes for 3 or 4 yr. However, relapse will probably occur in the great majority with the protocols currently in use, and for about half of the patients benefit is not sustained beyond the period of depletion (∼8 months). Chronically maintained depletion with frequently repeated therapy is, however, unlikely to be a viable option as even in cases with sustained benefit repeated use may induce hypogammaglobulinaemia. Partial depletion has no theoretical basis for use and indeed is not clinically effective [6, 9]. Thus, although BLyD shows great promise and is revealing mechanisms of disease, there is not yet a long-term strategy for routine use. Three reasons might be given for the use of BLyD in RA and other autoantibody-associated diseases. This is the most obvious reason, as removing B cells should remove the precursors of pathogenic antibody-producing plasma cells and therefore ameliorate disease. The first problem with this idea is that BLyD using anti-CD20 does not remove these plasma cells. On the positive side, although some lymphoma studies showed that anti-CD20 therapy was not associated with a major fall in immunoglobulin levels, in patients with autoimmune disease, our and other studies have found that autoantibody levels often do fall significantly. The conclusion is therefore that some autoantibody-producing plasma cells are relatively short-lived and BLyD prevents replenishment from B cells [20]. The positive practical implications are therefore obvious but we are left with an even more urgent need to understand what controls the production of the pathogenic antibodies. The more serious problem with this rationale, therefore, is that if antibody is simply an effector mechanism for tissue damage, with its production controlled by T cells, then autoantibodies will return as soon as B cells return. Thus, chronic depletion would seem to be required. Practically, this would make BLyD a not very attractive proposition in the long-term. B cells have a powerful and highly selective antigen-presenting capacity. However, other powerful antigen-presenting cells, such as dendritic cells, are readily available in most tissues. Even if B cells were to have a dominant role in antigen presentation, this suggestion suffers from the same problem as the first suggestion. Disease would be expected to reappear as soon as B cells return, so chronic depletion will be necessary. In theory, removal of the antigen-presenting capacity of B cells might induce putative autoreactive T cells to die off and allow ‘restoration of tolerance’. Unfortunately, this is difficult to verify experimentally because of the paucity of reproducible tests for autoreactive T cells [21, 22]. We embarked on BLyD treatment in RA on the basis of a third rationale, which suggested that response might be long-lasting [23]. The concept is that autoantibodies and their parent B cells are involved in a self-perpetuating vicious cycle which, if broken by removal of specific pathogenic B-cell clones, may collapse permanently [4, 23]. Although the idea of a vicious cycle as the basis of autoimmunity is not new, it has received relatively little attention. We have therefore included a brief outline of the basic concepts involved. All antibody production is a vicious cycle. B lymphocytes proliferate because they receive growth signals derived from the interaction of their own antibodies with antigen. One of these signals comes via T cells, to which both the B cell and other cells, such as dendritic cells, present antigen that has been picked up by cell surface antibody (attached to an immunoglobulin receptor for non-B cells). The other signal comes from an interaction between an immune complex, carrying the complement fragment C3d, and the B cell, via both B-cell surface antibody and complement receptor 2 [23]. Encounter of antibody expressed on the B cell with antigen therefore stimulates antibody production. It is designed to be an explosive chain reaction which only stops when antigen has been removed from the body. In autoimmunity the antigen cannot be removed [23]. Thus, the idea of autoimmunity as a vicious cycle of B-cell proliferation and antibody production is no more than standard dogma. The critical question is how the cycle can engage for an autoantibody when the system is supposed to be designed so that damaging autoreactivity cannot occur. Prior to about 1995, it was widely believed that as T cells drive (most) antibody production, autoantibody production must arise because autoreactive B cells are under the control of specific autoreactive helper T cells. (Either Th1 or Th2 cells will do, the main difference being that Th2 cells support IgE production.) However, for the major rheumatological autoantigens, including IgG Fc, DNA, Ro/La and topoisomerase-1, T-cell responses are hard to detect. Rheumatoid factor (RF) production is not, as far as is known, supported by anti-IgG T cells but by T cells recognizing foreign antigens [22]. T-cell responses to La do not appear to determine anti-La production [24] and twins discordant for scleroderma and anti-topoisomerase antibodies had the same anti-topoisomerase T-cell responses [25]. In myasthenia, T-cell autoreactivity may occur in the thymus but there appear to be no autoreactive T cells capable of getting into the muscle [26]. There is also the unanswered question: how could functionally significant numbers of autoreactive T cells arise in the first place? No convincing arguments have been given. There is, nevertheless, an alternative reason why certain autoreactive B cells should be able to engage the vicious cycle normally reserved for cells recognizing foreign antigen [23]. It relates to the fact that just as antibody production tends to be informed by T-cell responses, T-cell responses are informed by antibody—part of the vicious cycle concept. The vicious cycle may be engaged because the interaction between antibody and antigen has an unusual effect. How an interaction between a T-cell receptor and a processed peptide could have an unusual effect is difficult to conceive because the interaction occurs in a highly regulated microenvironment, but it is very easy to see how an interaction between an antibody and an antigen could have an unusual effect. Once generated, a free autoantibody molecule can behave how it likes, including as a pseudohormone or a neurotransmitter antagonist in thyrotoxicosis and myasthenia, respectively. What would therefore explain autoantibody production is an interaction between a particular antibody and self-antigen which results in the subversion of the normal immune response, i.e. interactions that are immunomodulatory. Such immunomodulatory interactions could only work for a few autoantigens, which would explain why most autoimmunity is directed at no more than about 40 of 40 000 or so potential self antigens. The most obvious candidate for this is the interaction between IgG RF and itself, in which the rules of antibody–antigen interaction break down. The precise signalling aberrations that arise are complex, but effectively the tail and the dog are wagging each other. Once an ‘unusual’ RF has been generated (e.g. by class switching or hypermutation), normal safeguards controlling its production and the survival of parent RF producing B cells are able to be bypassed [23]. Anti-C1q antibodies in lupus can similarly disturb immune complex-based signalling by activating complement in an inappropriate way [27]. The overproduction of antibody product from daughter plasma cells of these self-perpetuating clones is ultimately responsible for the pathology. Another interesting example of aberrant immune complex-based signalling is the binding of DNA-containing immune complexes to Toll-like receptor 9 on RF-specific B cells, providing an explanation for RF production as a by-product of anti-DNA antibody production [28]. Anti-acetylcholine receptor antibodies may modulate the thymic environment by binding to myoid cells [26]. Mechanisms for other autoantibodies have been suggested but are as yet more difficult to define [23]. The crucial implication of such immunomodulatory signals is that they provide a means whereby autoreactive B cells can engage a vicious cycle of expansion in the context of completely normal T cell responses. Abnormal T-cell responses may help initiate the autoreactive response and still contribute to the action of the immunomodulatory antibodies, but as yet the evidence of their being the major driving force behind the chronicity of the response is sparse. So what do clinical and immunological changes during BLyD tell us about the forces underlying autoantibody production? Is there evidence of an aberrant cycle that might be breakable, or is BLyD therapy of limited potential because of some unknown underlying T-cell-driven response? The essential practical question is why patients relapse, in half of cases only after up to 4 yr of more or less complete remission. Before addressing this question it is necessary to take a step back and reconsider the relationship between antibody and clinical disease in RA, because this has in the past been less obvious than in most other autoantibody-associated disorders. Initial scepticism about B-cell-targeted therapy in RA may have related to the fact that it invokes not one but two paradigm shifts. Perhaps surprisingly, the idea that B cells might play a role in immunoregulatory drive was in some ways the less controversial. The importance of both afferent feedback signals from antibody (feeding a vicious cycle) and of B cells in directing T-cell responses had been emphasized by Carson for some years [29], and more recently by Shlomchik [30], even if the concept of immunomodulatory autoantibody had not been enunciated as such. The more radical shift appeared to be the suggestion, based on our studies of immunoglobulin Fc receptors, that inflammation in RA was antibody-mediated after all [31–33]. As indicated above, it is theoretically possible for BLyD to benefit a condition with a T-cell effector response by removal of antigen-presenting cells. It was therefore important to see whether improvement with BLyD correlated with the absence of B cells themselves (in the role of antigen presenters) or the relative absence of antibody. Kinetic studies are still very much in progress. We addressed this question in our studies of BLyD in 22 patients with RA [20]. Two important observations emerged. First, clinical improvement followed the decline in autoantibody levels more closely than the fall in B-cell numbers; on many occasions the correlation with autoantibody levels was very close [20]. B cells disappear within days but clinical improvement and autoantibody decline may progress over as long as 9 months. Substantial clinical improvement was seen immediately in patients treated with protocols involving high doses of steroid and cyclophosphamide, but this is less evident with lower levels of steroid and no cyclophosphamide. Secondly, B-cell return was only associated with clinical relapse in half of the cases. The other half relapsed anything up to 2 yr later. Whether or not relapse was temporally associated with B-cell return, it was invariably associated with a rise in autoantibodies to levels comparable with those present before treatment [20]. As previously observed following high-dose chemotherapy with bone marrow rescue, RF levels appear to be the of clinical The same of response have recently been in a patients with Another is of patients to the with arthritis but often positive for clinically to be RA have received has shown a clinical response [6, 9]. Moreover, evidence to date is that such as and disease, no response It is as if B cells are only important in a human disease if the clinical condition is associated with The most obvious conclusion is that is because the antibodies are In although of of autoantibodies in RA may be the evidence from BLyD a pathogenic role for a of and very the IgG or RF to the immune complexes to be the of production RF is to have a if more This conclusion has the that it for autoantibodies to be pathogenic in all in which they Moreover, of a role for pathogenic autoantibodies in RA the of B-cell-targeted therapy much both because there is a practical and because antibody levels provide a means of as antibodies are pathogenic in RA, as in other autoimmune then the reason for clinical relapse following BLyD in all these would appear to be autoantibody return. It then to why the autoantibodies return when they In fact there are two why do autoantibodies return at the time of B-cell return in some patients and why is return for many in On the basis of the vicious cycle mechanism for autoantibody production, autoantibodies might return for one of two reasons [20]. First, B-cell clones to production of autoantibodies capable of via afferent signals may not have been and may Secondly, these B-cell clones may have been but their daughter plasma cells may be able to antibody via the same afferent survival of B-cell clones to take in a self-perpetuating cycle. The of autoantibodies in some patients at the time of B-cell return may not help these There is about B-cell return following anti-CD20 in that it does not occur for about much than the This What the seem to be us is that the of autoantibody to induce clinical relapse the of B cells. What it does not tell us is whether the expansion of these B cells the survival of the B-cell The problem is that the vicious cycle not only that pathogenic autoantibody-producing cells will but so will clones producing antibodies of the same whether or not they are in themselves might be the with antibodies with the pathogenic The is that a proportion of autoantibodies will be and pathogenic may be hard to This with the fact that autoantibodies often occur in the absence of clinical disease. these in the following for the basis of relapse are suggested by the observations in patients and by the vicious cycle patients in relapse occurs as soon as B-cell numbers may have of disease because of the of pathogenic B-cell in relapse is may have been of clones, and of a vicious cycle may the of clones autoantibodies with immunomodulatory capacity. of B cells in are in progress and will on these The of BLyD in RA would not have were it not for the evidence of few with with use of anti-CD20 in The would have been that BLyD would be because efficacy would be associated with hypogammaglobulinaemia. Even with available from there was a that BLyD to be a remission therapy to be this seemed to be the in the context of RA as was therapy with an such as cyclophosphamide would be to In the early results with BLyD in RA suggested not only that there was a selective effect on with of antibody levels to and but also that immunoglobulin levels to However, that over a period of yr the repeated use of with or more is in some cases associated with significant in immunoglobulin IgG levels have not to an in the context of However, in cases levels have The clinical of this is However, in the context levels have also been seen and have been to be associated with of There has been a suggestion in the cases of RA treated at that is more than but not in to immunoglobulin levels A significant proportion of have within days of rituximab that they may not be but may be some of The problem has resolved in all RA cases but have been seen in one of Severe are One of in the to be The evidence so far the after BLyD are no more of an than the with agents are probably less of an but will be The implication of these would seem to be that BLyD as currently may be of in disease control in patients in the but there is a major question over the long-term repeated use of much as This the question of what alternative to B-cell targeting might the long-term remission long-term of The first alternative to the BLyD to consider is antibody that might have and perhaps sustained responses in those cases currently for a or monoclonal will be or from carrying human antibody may be different anti-CD20 antibodies have widely different in It is not certain which is the most mechanism in or of other studies that rituximab may complement least one other anti-CD20 antibody is being developed that to have may have potential at which so far have not been in in autoimmune disease. It may also be to for more agents to cyclophosphamide. with methotrexate probably does not the effect on autoantibody levels does but is for use in is a but may not be selective and a that could be in a short course of treatment would be antibodies directed at other B-cell surface have been in for some antibodies have been but because has a high of in the an antibody is to be of a or This may be in the treatment of disease but is difficult to in autoimmune disorders. antibodies are also available but their potential is There is a that just to be a for antibodies and that it is relatively unlikely that a B-cell will be This does not the that therapy with antibodies to more than one surface molecule may not There is also a that the of antibodies might be by of complement or other that cell Another approach to BLyD is the of growth such as B-lymphocyte and perhaps molecule of the same targeting and are into using antibodies or What is not yet clear is whether of the system can have as an effect as is to be less However, of and or may have a selective effect on autoimmune B cells, at least in some conditions. This the of a more B-cell may have much potential than BLyD with In many BLyD is using a to a It would be to induce the of autoimmune B cells B cells recognizing foreign antigens. One strategy is to use which but of binding to antibody to a capable of a parent B cell are designed to a signal to the autoreactive B cell This strategy has been under but has yet to clinically The main of the approach is that it is necessary to which drive the pathogenic mechanism in each which has to be a very What may be more is to to a way of the survival signals on which autoreactive B cells are The of such signals is autoreactive B cells may be to of signalling complement receptor or T-cell help or or to of that The on such signals may be for all autoimmune cells but it may be Thus, one could that of complement receptor 2 might induce the of RF-specific B cells but might the survival of B cells in This for the of explain why such as or gold can reduce RF levels in RA but have a to induce of far it may be that the for each autoimmune disease will be a molecule that the survival of autoreactive B cells only of a found in the disease. its gold is the for RA because it can induce complete and remission. problem is that it this so and more often There is no reason to that if we how and work it should not be possible to efficacy from A of how these agents work might great A few other to B-cell targeting that may into the are of B-cell proliferation by of specific to B-cell and survival might be of B are on specific at different of and these or their might be There may be arguments for targeting plasma cells as as B cells, with the of removing autoantibodies capable of a vicious cycle. it is relatively unlikely that autoreactive plasma cells can be this strategy may a major of immune with present there are no effective agents for removing plasma cells In more it may be that for some of the autoimmune a more approach may still be even if it is less cytotoxic therapy with has very sustained in It might for that a could be by a highly with a high-dose cytotoxic perhaps the same benefit the of found with high-dose cytotoxic In to all these there is an need for an of disease and then can therapy be the use of BLyD and related will allow us to this over the few years and to the of capable of long-term has been in of support from and The other have no of