Wegener's granulomatosis (WG) is a form of systemic vasculitis characterized by a necrotizing, granulomatous inflammation of the respiratory tract that is associated with glomerulonephritis. A disseminated vasculitis involving the small vessels often occurs as the disease progresses. Microscopic polyangiitis is a related small-vessel vasculitis that shares clinical and pathologic features with WG. Both diseases are highly associated with autoantibodies that react with specific neutrophil granular enzymes (antineutrophil cytoplasmic antibody [ANCA]). The vast majority of patients with WG have autoantibodies that react with proteinase 3 (cytoplasmic ANCA [cANCA]). Patients with microscopic polyangiitis may also have cANCA but more commonly have autoantibodies that react with myeloperoxidase (perinuclear ANCA). Because they share serologic and clinicopathologic features, WG and microscopic polyangiitis are sometimes collectively referred to as ANCA-associated vasculitis. However, the presence of granulomatous inflammation is unique to WG, and it remains to be determined whether these 2 diseases differ with regard to prognosis and response to therapy.

The vast majority of information about the treatment of ANCA-associated vasculitis comes from studies of patients with WG. If it is left untreated, generalized WG usually follows a rapidly progressive and lethal course. In one series of untreated patients, the mean survival time was 5 months, and >90% of patients died within 2 years of diagnosis (1). The effect of treatment with glucocorticoids alone on the course of WG has never been fully defined. The most widely quoted article on this subject is a literature review of 26 patients with WG who were treated with glucocorticoids in the 1950s and early 1960s (2). Many of these patients were treated with adrenocorticotropic hormone or moderate doses of cortisone, and, in most cases, treatment was not initiated until patients were moribund and near death.

In the late 1960s, Fauci and Wolff at the National Institutes of Health (NIH) began to use low-dose cyclophosphamide (administered daily) combined with glucocorticoids to treat patients with generalized WG. In 1973, they reported the induction of sustained disease remission in 12 patients with WG who were treated with daily cyclophosphamide and glucocorticoids (3). This cyclophosphamide regimen was soon adopted as standard therapy for patients with generalized WG, and subsequent studies of a large cohort of patients treated at the NIH confirmed its efficacy (4). Through extended followup of the NIH cohort, it became apparent that relapse of disease was far more common than was previously thought (5). Furthermore, the use of repeated or prolonged courses of cyclophosphamide to prevent or treat disease relapses resulted in a high rate of serious drug-related toxicity, including major infections, infertility, myeloproliferative disorders, cystitis, and transitional cell carcinoma of the bladder (5, 6). This degree of treatment-related morbidity limits the utility of the NIH cyclophosphamide regimen as therapy for what is now recognized as a chronic, relapsing disease.

In the past 10 years, progress has been made in developing effective approaches to the treatment of WG that eliminate or reduce the exposure to cyclophosphamides (7). Current cyclophosphamide-sparing approaches are well tolerated. Such regimens can induce remission of disease in 80–100% of patients, and relapse rates are comparable with those observed in patients treated with the NIH cyclophosphamide regimen (8–10).

The development of effective cyclophosphamide-sparing regimens represents an advance in the treatment of WG. However, these regimens are still associated with significant treatment-related morbidity and all too often fail to prevent disease relapse. Such relapses place patients at risk for morbidity and mortality from irreversible disease-related organ damage and the cumulative toxicity of repeated courses of immunosuppressive therapy. Thus, there continues to be a need for effective, less toxic forms of therapy to treat WG and related systemic vasculitis syndromes. To address this need, clinical investigators are beginning to explore therapeutic regimens using biologic agents to target specific elements of the inflammatory response. The hope is that such agents will prove effective without causing global immunosuppression or other deleterious effects typically associated with conventional immunosuppressive drugs. In this issue of Arthritis & Rheumatism, Keogh and colleagues describe the outcome of rituximab treatment in 11 patients with ANCA-associated vasculitis, all but 1 of whom had WG (11). Rituximab is a chimeric monoclonal antibody that binds to CD20, a surface glycoprotein expressed by B cells.

To better appreciate the rationale behind the use of rituximab in the treatment of WG, it is useful to consider some relevant aspects of B cell biology and the pathogenesis of WG. Morbidity in WG results from pathogenic granulomatous inflammation in the respiratory tract, glomerulonephritis, and small-vessel vasculitis. The glomerulonephritis and small-vessel vasculitis observed in WG are thought to be mediated by ANCA (12). In contrast, granulomatous inflammation is a process mediated by CD4 T cells producing Th1 cytokines, and there is little evidence from animal or human studies that either B cells or ANCA are involved in the pathogenesis of this feature of WG (13, 14). Thus, the rationale behind the use of rituximab in the treatment of WG is based on the following 3 assumptions: ANCA play an important role in the pathogenesis of WG; treatment with rituximab can effectively deplete CD20-expressing precursors of ANCA-producing plasma cells; and plasma cells producing ANCA are short-lived, and transient depletion of their CD20+ precursors will abrogate ANCA production.

The scientific basis for these 3 assumptions comes from a variety of data. Thus, there is a significant body of evidence supporting an important role for ANCA in the pathogenesis of WG and related syndromes. A detailed discussion of this evidence is beyond the scope of this commentary, and the reader is referred to recent reviews on this subject (12, 15). Evidence supporting a key role for ANCA in the pathogenesis of WG is summarized in Table 1. Despite these data, several clinical and laboratory observations argue against a primary pathogenic role for ANCA. Patients may have active WG in the absence of ANCA, the absolute height of the antibody titers does not correlate well with disease activity, and patients with WG in remission may continue to have high ANCA titers for years without experiencing a recurrence of disease (16). In addition, it is not clear how ANCA could give rise to granulomatous inflammation, the hallmark of WG. The precise role of ANCA in the pathogenesis of WG will not be defined until a therapeutic agent can be found that selectively inhibits ANCA production and the effects of this inhibition can be observed.

Table 1. Evidence supporting a role for ANCA in the pathogenesis of WG*
  • *

    ANCA = antineutrophil cytoplasmic antibodies; WG = Wegener's granulomatosis.

1. ANCA are highly associated with WG and are rarely observed in other autoimmune disorders.
2. In vitro observations suggest mechanisms whereby ANCA can produce vasculitis in vivo.
 A. Exposure of cytokine-activated neutrophils or monocytes to ANCA results in degranulation and the production of reactive oxygen species.
 B. Activation of neutrophils and monocytes by ANCA induces the release of proinflammatory cytokines.
3. Adoptive transfer experiments in genetically engineered mice provide evidence for a direct pathogenic role of ANCA in vivo.

As indicated above, rituximab is a monoclonal antibody that recognizes the B cell–specific surface protein CD20. An important property of rituximab is its ability to induce death of both malignant and normal CD20+ B cells. The death of B cells appears to result from induction of complement-mediated events and through antibody-dependent cellular cytotoxicity (for review, see ref.17). Rituximab has efficacy in the treatment of non-Hodgkin's lymphoma, and early observations revealed that in addition to killing malignant B cells, treatment also resulted in the disappearance of normal B cells from the peripheral blood for a period of ∼6 months, after which time B cell numbers gradually returned to pretreatment levels (18). This transient disappearance of B cells from the peripheral blood following rituximab treatment is often referred to as B cell depletion. However, it should be noted that the vast majority of B cells reside not in the peripheral blood but in the bone marrow and lymphoid tissue. Studies in nonhuman primates expressing cross-reactive CD20 indicate that, although rituximab induces 98% depletion of peripheral blood B cells, only 40–70% of lymph node B cells are depleted (19). To what degree the currently used rituximab regimens deplete normal B cells in human lymphoid tissues is not known. Even if complete or near-complete killing of CD20+ B cells is possible with rituximab, this may not affect ANCA production by plasma cells that do not express CD20.

Traditionally, it had been thought that production of antibodies is the result of continuous activation of antigen-specific B cells that differentiate into short-lived plasma cells. However, more recent studies indicate that long-lived plasma cells, with life spans measured in months to years, are responsible for a significant portion of antibody production and B cell memory (20–22). These long-lived plasma cells are derived from B cells that have undergone repeated cycles of proliferation, somatic mutation, isotype switching, and apoptosis in splenic and lymph node germinal centers. This germinal center reaction selects for B cells that produce high-affinity antibodies for cognate antigen. Two types of B cells arise from the germinal center reaction: nonsecreting memory B cells and plasmablasts. Some of the plasmablasts migrate to the bone marrow, where they differentiate into long-lived plasma cells that secrete high-affinity antibodies for months to years (21, 23). Long-lived plasma cells do not express CD20, do not divide, and are not sensitive to irradiation or cytotoxic drugs. The presence of long-lived plasma cells may explain why rituximab-induced depletion of CD20+ B cells does not result in significant declines in either total serum IgG levels or titers of antibodies to recall antigens (24).

There is mounting evidence that long-lived plasma cells are also an important source of autoantibody production (for review, see ref.25). For example, a recent study by Hoyer et al indicates that long-lived plasma cells are a major source of anti-DNA antibodies in the NZB mouse model of systemic lupus erythematosus (26). Analogous to observations with antibodies to recall antigens, recent studies in humans have shown that rituximab-induced B cell depletion does not lead to a decline in the level of anti-DNA antibodies (27). This observation suggests that long-lived plasma cells resistant to the effects of rituximab constitute a significant source of anti-DNA autoantibodies in humans as well as in mice.

Several observations suggest that long-lived plasma cells may account for a significant portion of ANCA production. The fact that ANCA titers in patients with WG can remain elevated for years despite treatment with cytotoxic drugs that target dividing cells implies the existence of long-lived ANCA-producing plasma cells. Pathogenic ANCA are of the IgG isotype and have been shown to exhibit somatic mutations (15, 28); both the IgG isotype and somatic mutations are hallmarks of maturation in the germinal center reaction that gives rise to long-lived plasma cells. Taken together, the above data suggest skepticism about whether transient depletion of CD20+ B cells by rituximab can result in inhibition of ANCA production and associated clinical benefit.

The report by Keogh et al attempts to directly address the effect of B cell depletion with rituximab on ANCA production and clinical parameters in 10 patients with WG and 1 patient with microscopic polyangiitis. These patients were selected by the authors for off-label rituximab treatment, because they had evidence of active disease and either could not tolerate optimal doses of cyclophosphamide or had contraindications to the use of that drug. The clinical outcomes of the patients treated by Keogh et al were generally good. However, 3 of the 5 patients with glomerulonephritis failed to have improvement in their renal function following rituximab therapy. All patients received concomitant therapy with high-dose glucocorticoids, and 3 patients were also treated with plasma exchange. In addition, 1 patient underwent renal transplantation following rituximab treatment and presumably received standard posttransplantation immunosuppressive therapy. Thus, it is difficult to determine what role concomitantly administered immunosuppressive treatments might have played in the observed clinical responses.

Because the major premise behind this study is that ANCA are pathogenic, and that inhibition of their production by rituximab will result in clinical benefit, it is important to closely examine the reported effect of the treatment regimen on ANCA titers. The ANCA titers rapidly declined in all 11 patients within 3 months of initial treatment. The authors conclude that the rapid decline in ANCA titers following rituximab treatment indicates that short-lived plasma cells or B cells are the major source of ANCA. However, there is another interpretation of these data. In addition to being treated with rituximab, all patients received high doses of prednisone daily, and 8 of the 11 patients also received multiple 1-gm “pulses” of methylprednisolone. Pharmacologic doses of glucocorticoids have significant effects on humoral immunity and antibody production. This effect was clearly demonstrated more than 30 years ago, when Butler and Rossen observed that normal volunteers treated for 3–5 days with oral methylprednisolone exhibited significant suppression of serum IgG levels for 2–4 weeks (29). Subsequent studies in patients with autoimmune diseases confirmed that pharmacologic doses of glucocorticoids have a suppressive effect on antibody and autoantibody production (30–32). In a report relevant to the current study, Smith et al observed that administration of 1 gm of methylprednisolone on 3 successive days to patients with rheumatoid arthritis resulted in decreases in the serum rheumatoid factor level that persisted for up to 6 months (31). These and other studies demonstrate that treatment with high doses of glucocorticoids alone can result in significant suppression of antibody and autoantibody production, by mechanisms that have yet to be defined.

Given the suppressive effects of glucocorticoids on immunoglobulin production, it is certainly possible that some or all of the rapid decline in the ANCA titer reported by Keogh et al may have been attributable to the high-dose glucocorticoid treatment that was administered with rituximab. In this regard, it is of interest to examine the course of 3 patients who received a second course of rituximab because of asymptomatic rises in the ANCA titer. In these 3 patients the disease was in clinical remission; therefore, they did not receive high-dose glucocorticoids with their second course of rituximab. If CD20+ B cells or short-lived plasma cells are the source of ANCA, then treatment with rituximab alone should result in a rapid fall in ANCA titers similar to that observed when rituximab was given in combination with high-dose glucocorticoids. Alternatively, if the rapid fall in ANCA titers was the result of glucocorticoid-induced suppression of immunoglobulin production, then ANCA titers would not rapidly decline after treatment with rituximab alone. A close examination of Figure 1C indicates that re-treatment with rituximab alone failed to produce a decline in ANCA titers in 2 of these 3 patients. In the third patient, it took nearly a year after treatment for the ANCA titer to fall and become negative. It should be noted that when this latter patient received rituximab combined with high-dose glucocorticoids, the ANCA titer fell to undetectable levels within 4 weeks.

The effects observed in the 3 patients re-treated with rituximab alone are difficult to reconcile with the authors' contention that B cells or short-lived plasma cells (which would be susceptible to depletion by rituximab) are the cellular source of ANCA. Rather, these results suggest that it was treatment with highdose glucocorticoids that produced the rapid fall in ANCA titers observed following initial treatment. The authors acknowledge this possibility but also offer another explanation. They speculate that during periods of active disease, consumption of ANCA may be increased by as-yet-undefined mechanisms. If ANCA production were inhibited during a period of active disease, titers would rapidly fall because of this increased consumption. If ANCA production is inhibited during periods of disease remission, when ANCA consumption is not increased, then titers should decline more slowly, in accordance with the normal half-life of IgG. The problem with this explanation is that the normal half-life of IgG is 18–23 days (33). Thus, if ANCA production had been inhibited in the 3 patients re-treated with rituximab alone, one would expect to see declines in ANCA titers within 6–8 weeks of re-treatment (2–3 IgG half-lives), which is not what was observed. In addition, it is well established that ANCA titers generally rise during periods of active disease, an observation not easily explained by inflammation-induced increases in ANCA consumption. Thus, the question of whether rituximab-induced depletion of CD20+ B cells can inhibit ANCA production remains open and requires further study.

Although preliminary, the findings reported by Keogh et al are provocative and merit further study in a prospective trial. The design of such a trial requires careful consideration. WG is a clinically aggressive autoimmune disease that, if inadequately treated, can rapidly cause irreversible major organ damage. This fact makes it difficult to withhold therapy of known benefit in a trial setting. The data in the report by Keogh et al are suggestive but are too preliminary to justify substituting rituximab for standard therapy in patients with severe manifestations of WG who could otherwise tolerate optimal doses of cyclophosphamide.

One possible study design that could determine efficacy while minimizing risk would be to enroll patients whose WG is in remission and who experience a significant rise in the ANCA titer. As a group, these patients would be expected to have a 40–50% chance of experiencing a disease relapse within 12–18 months of the titer rise (34). A trial in which half of such patients are randomized to have rituximab added to their maintenance regimen would have several advantages. First, no patient would be denied standard therapy of known benefit. Second, because the addition of rituximab would be the only therapeutic intervention, this would answer the question of whether depletion of CD20+ B cells can lower ANCA titers. Finally, efficacy could be assessed by the ability of rituximab treatment to prevent relapse of disease. If efficacy were demonstrated in the absence of an effect on ANCA, this would provide important clues regarding a possible ANCA-independent role of B cells in the pathogenesis of WG. Hopefully, studies of this kind will be performed soon, and the role of B cells in ANCA-associated vasculitis will become better understood.


  1. Top of page
  2. Acknowledgements

The author is grateful to Dr. Pasha Sarraf for his thoughtful review of the manuscript.


  1. Top of page
  2. Acknowledgements