Serologic changes following B lymphocyte depletion therapy for rheumatoid arthritis




To explore the changes in serologic variables and clinical disease activity following B lymphocyte depletion in 22 patients with rheumatoid arthritis (RA).


B lymphocyte depletion was attained using combination therapy based on the monoclonal anti-CD20 antibody rituximab. Levels of a serologic indicator of inflammation, C-reactive protein (CRP), of antimicrobial antibodies, of autoantibodies including IgA-, IgM-, and IgG-class rheumatoid factors (RF), and of antibodies to cyclic citrullinated peptide (anti-CCP) were assayed.


The majority of patients showed a marked clinical improvement after treatment with rituximab, with benefit lasting up to 33 months. Levels of total serum immunoglobulins fell, although the mean values each remained within the normal range. Whereas the IgM-RF response paralleled the changes in total serum IgM levels, the levels of IgA-RF, IgG-RF, and IgG and anti-CCP antibodies decreased significantly more than did those of their corresponding total serum immunoglobulin classes. The kinetics for the reduction in CRP levels also paralleled the decreases in autoantibody levels. In contrast, levels of antimicrobial antibodies did not change significantly. B lymphocyte return occurred up to 21 months posttreatment. The time to relapse after B lymphocyte return was often long and unpredictable (range 0–17 months). Relapse was, however, closely correlated with rises in the level of at least one autoantibody. Increased autoantibody levels were rarely observed in the absence of clinical change.


Following B lymphocyte depletion in patients with RA, a positive clinical response occurred in correlation with a significant drop in the levels of CRP and autoantibodies. Antibacterial antibody levels were relatively well maintained. B lymphocyte return preceded relapse in all patients. There was also a temporal relationship between clinical relapse and rises in autoantibody levels. Although these observations are consistent with a role for B lymphocytes in the pathogenesis of RA, the precise mechanisms involved remain unclear.

The relative contribution of the different elements of the immune system to the pathogenesis of rheumatoid arthritis (RA) remains controversial (1, 2). The genetic link with HLA–DR4 (3) implies a T cell–dependent process. It is now evident that T cell–B cell interaction is a complex 2-way process, in which the function of each cell type is dependent on the other (4). Moreover, this interaction may, in special cases, be atypical, as for example, with rheumatoid factor (RF) B cells, which can potentially present any T cell with its cognate antigen and in return receive a positive survival signal (5).

A specific subset of autoreactive B lymphocyte clones, capable of self-perpetuation, has recently been proposed to be involved in disease persistence in RA (2). In this model, a dual role for B lymphocytes, and in particular, those committed to producing RF, has been suggested. First, they may differentiate into plasma cells that produce autoantibodies capable of forming small immune complexes. Interaction of such small immune complexes with the immunoglobulin receptor Fcγ receptor type IIIa (FcγRIIIa) on macrophages, in joints, and in other tissues may be responsible for the production of proinflammatory cytokines (6, 7). The FcγRIII-dependent arthritis of the K/BxN transgenic mouse may be a useful model for this mechanism (8). Second, daughter plasma cells may perpetuate the survival of parent RF B lymphocytes by providing a constant supply of self-complexed IgG (2).

Therapeutic B lymphocyte depletion provides a new opportunity to assess the roles of B lymphocytes in the pathogenesis of RA and other autoimmune diseases. B lymphocyte depletion has recently been introduced as a therapy for a range of autoantibody-associated disorders, including RA, IgM-associated neuropathies, immune thrombocytopenic purpura, autoimmune hemolytic anemia, systemic lupus erythematosus, and dermatomyositis (9–15). Encouraging results have been reported, and a randomized controlled trial of this therapy in RA is currently in progress.

Selective B lymphocyte depletion has been made possible by the availability of the chimeric anti-CD20 monoclonal antibody rituximab (16–21). CD20 is a B lymphocyte–restricted antigen that is expressed on B lymphocyte precursors and mature B lymphocytes. It is lost during differentiation into plasma cells. Rituximab has been proven to be very effective in depleting normal and malignant B lymphocytes in vivo. In humans, peripheral B cell depletion occurs within days, and studies in primates have shown that up to 70% of B cells in lymphoid organs are also rapidly cleared (16, 17, 21). In autoimmune diseases, rituximab has been used in many cases as monotherapy (10–12). However, B lymphocyte depletion with rituximab alone is probably incomplete, and in patients with lymphoma, long-term benefit from rituximab is increased by combination with other drugs (20). For these reasons, in the treatment of RA, it has initially been used in combination with cyclophosphamide and corticosteroid (9).

We have treated 23 patients with RA with rituximab, alone or in combination with other agents. The detailed clinical responses in these patients have been reported previously (9, 15). The majority of patients showed substantial clinical improvement, which was sustained for up to 33 months and up to 17 months after the return of circulating B lymphocytes. However, in all but 2 patients, the disease subsequently relapsed. The key objective of the present study is to understand the mechanism of this relapse. Critically, this requires an understanding of whether disease persistence depends on the existence of specific T cell clones, specific B cell clones, or the continued production of antibodies by long-lived plasma cells.

Circulating antibody levels are, at present, the only practical indices for monitoring the autoimmune response in RA. The most prevalent autoantibody reactivity in RA is RF, with IgG-RF being particularly implicated in the formation of small immune complexes (2, 22), but a significant proportion of patients also have raised levels of antibodies to cyclic citrullinated peptides (anti-CCP) (23) and to proteins such as the immunoglobulin heavy-chain binding protein (anti-BiP) (24). In the absence of T cell responses to IgG, there is particular interest in the possible roles of T cell responses to these other antigens and their interrelationships with RF B cells. For 22 of our 23 patients with RA (described in refs.9 and15), we have serial immunologic data covering the period of B lymphocyte depletion, during which no other therapy was introduced. We report herein serial observations of the total serum immunoglobulin and specific autoantibody levels together with the levels of protective antibodies to pathogens in these patients, and we report the relationships of those changes to regression and relapse of inflammation following B lymphocyte depletion.



Twenty-two patients with active RA in whom total circulating B lymphocyte depletion was achieved for ≥3 months were studied. All patients satisfied the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) diagnostic criteria for RA (25). Two patients were male and 20 were female. The mean age was 58 years (range 33–81 years) and the mean disease duration was 18 years (range 5–40 years). The study was approved by the local hospital ethics committee. All patients gave informed consent. Serum samples were kept at −80°C until tested for autoantibodies and antimicrobial antibodies. Relapse was determined clinically and was defined as an increase in pain of at least 20% on visual analog scale and an increase in the swollen joint count among >2 joints on 2 consecutive visits.

Treatment protocol

Patients were treated in 5 cohorts with a range of doses of rituximab, with or without intravenous cyclophosphamide and/or oral prednisolone, over a 2–4-week period (9, 15). Disease-modifying antirheumatic drugs (excluding steroids) were discontinued from day 0. All patients received nonsteroidal antiinflammatory agents or analgesics as needed.


Patients were assessed prior to treatment, monthly for the first 6 months after treatment, and then every 2 months until 1 year, and subsequently every 2–3 months. Circulating B lymphocytes (CD19-positive cells, determined by flow cytometry; normal range 0.03–0.40 × 109/liter) and total lymphocytes were measured before treatment, monthly after treatment in the first 6 months, and then every 2 months until they returned to normal levels. Serial laboratory measurements in sera included the C-reactive protein (CRP) concentration, total immunoglobulin levels, IgM-, IgG-, and IgA-RF levels, and levels of anti-CCP antibodies. The CRP concentration was measured by nephelometry (normal range 0–5 mg/liter). Total serum immunoglobulin levels were measured by immunoturbidometry (normal range for IgA 0.7–4.0 gm/liter, IgG 7.0–16.0 gm/liter, IgM 0.4–2.3 gm/liter).

Measurement of autoantibodies

Serial samples from each patient were analyzed together on the same day. IgM-RF, IgG-RF, and IgA-RF were measured by enzyme-linked immunosorbent assay (ELISA) using the RF-isotype detection kit developed by Dr. M. Teodorescu (26) and supplied by TheraTest Laboratories (catalog no. EL-RF/3; Chicago, IL). This test is based on the binding of RFs to rabbit IgG. The TheraTest ELISA utilizes horseradish peroxidase (HRP)–conjugated rabbit anti-human IgG (Fab′)2 to detect IgG-RF, and prior to testing for IgG-RF, samples are digested with pepsin to avoid interference from IgM-RF and IgA-RF (26). Completeness of pepsin digestion was checked by confirming the absence of binding of anti–IgM-HRP to pepsin-digested samples and controls. Results are expressed in IU/ml. The cutoff values for normal control sera were <25 IU/ml, <20 IU/ml, and <35 IU/ml for IgM-RF, IgG-RF, and IgA-RF, respectively. IgG anti-CCP antibodies were measured by ELISA (Immunoscan RA; Euro-Diagnostica, Arnhem, The Netherlands), and results are expressed as units equivalent to the standard serum, as determined by reading on a calibration curve.

Measurement of antimicrobial antibodies

Specific IgG antibodies to tetanus toxoid (anti-TT) and to pneumococcal capsular polysaccharides (anti-PCP; combination of 23 common serotypes) were measured by ELISA (Binding Site Limited, Birmingham, UK). Levels >0.1 IU/ml for anti-TT antibodies were considered to be optimally protective; 90% of the population have anti-PCP levels >35 mg/liter.

Statistical analysis

For comparison of absolute values before and after treatment, the Wilcoxon signed rank test for paired data was used (1-tailed). Wilcoxon rank sum analysis was applied when comparing median levels of parameters for responders and nonresponders. When normalized values were compared, Student's paired t-test was applied.


Pretreatment parameters and analysis of results.

Twenty of the 22 patients with RA were seropositive for RF; all of these patients were found by ELISA to have IgM-RF, 18 (82%) had IgA-RF, and 13 (59%) had IgG-RF (Table 1). Thirteen patients had anti-CCP antibodies, including 1 of the RF-seronegative patients. The other seronegative patient was positive for anti-BiP antibodies. Since all patients treated with the different protocols achieved B lymphocyte depletion in the peripheral blood, results comparing the changes in CRP and total serum immunoglobulin levels with autoantibody and antimicrobial antibody levels were combined and analyzed together (Figures 1–3 and Figure 5). For comparison of the serologic response to treatment, patients were classified as responders (those achieving and maintaining an ACR level of improvement [27] of ≥20% at 6 months) (15 of 22 patients) or nonresponders (those attaining an ACR level of improvement <20% at 6 months) (Tables 1 and 2 and Figures 1 and 4).

Table 1. Relationship between B lymphocyte return, levels of circulating autoantibodies, and relapse in responder and nonresponder patients with rheumatoid arthritis*
PatientAutoantibodies present at baselineACR response at 6 months posttreatment, %Time to B cell return, monthsTime to clinical relapse, monthsTime to autoantibody rise, months (responding antibody)
  • *

    American College of Rheumatology (ACR) responders were those who achieved ≥20% improvement at 6 months posttreatment. M = IgM–rheumatoid factor (RF); A = IgA-RF; CCP = cyclic citrullinated peptide; G = IgG-RF; NYK = not yet known; NA = not applicable; BiP = immunoglobulin heavy-chain binding protein.

  • Patient numbers are described in ref.15. Serum from patient 15 was not available for study.

1M, A, CCP7020.833.530 (M, A, CCP)
2M, A, CCP706.96.96 (M, A, CCP)
3M, G708.825.625.6 (M)
4M, G, A, CCP507.47.86 (A, CCP, M)
5M, G, A, CCP5011.121.221.2 (M, G, A)
7M, A208.37.26.5 (M)
9M, G, A706.77.45.8 (M, A)
10M, G, A, CCP708.59.08.3 (A, M, G)
11M, G, A7010.4>24.0NYK
12M, CCP506.511.59.9 (M, CCP)
13M, A, CCP7012.0>19.0NYK
14M, G, A, CCP5010.212.212.2 (M)
16M, G, A209.517.3
17M, G, A506.013.214.3 (M, A)
18M, G, A, CCP705.87.46 (M, G, A, CCP)
19M, G, A, CCP<208.5NANA
6M, A, CCP<204.6NANA
21M, G, A<209.2NANA
22M, G, A<205.1NANA
23M, A, CCP<207.0NANA
Figure 1.

Time course of changes in IgA–, IgM–, and IgG–rheumatoid factor (RF) as well as anti–cyclic citrullinated peptide (anti-CCP) antibodies and C-reactive protein (CRP). Bars show the mean ± SD percentage of pretreatment (before rituximab combination therapy) values for 6 months (26 weeks) following B lymphocyte depletion in responder (A) and nonresponder (B) patient groups.

Figure 2.

Minimum (nadir) levels reached for IgA-RF, IgM-RF, and IgG-RF, anti-CCP antibodies, and total IgA, IgM, and IgG serum immunoglobulins, expressed as the mean and SD percentage of pretreatment values. P values were calculated using Student's paired t-test. See Figure 1 for definitions.

Figure 3.

Time taken (weeks) following treatment for levels of each parameter to decrease by 50% (T50) and 80% (T80) of initial pretreatment values. Bars show the median and SEM. See Figure 1 for definitions.

Table 2. Serologic parameters in patients before treatment and at nadir*
Parameter, response to therapy (number of patients per group)Pretreatment valueLowest level attained after treatmentP
  • *

    Responders were defined as having American College of Rheumatology ≥20% improvement at 6 months. Values are the median ± SEM IU/ml, except values for C-reactive protein (CRP), which are the median ± SEM mg/liter. NS = not significant (see Table 1 for other definitions).

  • Pretreatment values versus values at nadir for each group of patients tested, using Wilcoxon signed rank test with significance level of 0.1%.

  • P = 0.009 versus nonresponders, by Wilcoxon rank sum test.

 Responders (15)126 ± 78.250 ± 38.20.00006
 Nonresponders (5)122 ± 16.890 ± 15.4NS
 Responders (10)77 ± 25.915.5 ± 12.60.002
 Nonresponders (3)50 ± 24.124.5 ± 13.5NS
 Responders (13)147 ± 194.378 ± 86.10.0002
 Nonresponders (5)158 ± 28.2118 ± 18.5NS
 Responders (8)950 ± 340.9236 ± 122.10.008
 Nonresponders (4)2,282 ± 1,1451,350 ± 894NS
 Responders (15)37.9 ± 8.42 ± 1.30.00006
 Nonresponders (7)39.7 ± 1813.7 ± 5.4NS
Figure 4.

Levels of IgG-RF (A), IgM-RF (B), IgA-RF (C), and anti-CCP antibodies (D) in responding patients at pretreatment, when each patient had reached the maximum attained level of improvement in the American College of Rheumatology response (ACRmax), and at or shortly after relapse. See Figure 1 for other definitions.

Serologic parameters in responder versus nonresponder patient groups.

When the pretreatment median values of the serologic parameters in responders and nonresponders were compared (Table 2), there was no significant difference between the groups (P > 0.01, by Wilcoxon rank sum). Similarly, there was no difference between the median nadir levels in responders and nonresponders, except, as may be expected, when CRP levels were compared (P < 0.01). There was, however, a highly significant difference between the drop in autoantibody levels and the drop in CRP levels in the responding patients, but not in the nonresponding patients. In each patient in the responder group, serum levels of all classes of RF and of anti-CCP antibodies decreased relative to their pretreatment values (Table 2), with all achieving statistical significance at the 0.1% level (by Wilcoxon signed rank paired test).

For comparison of the kinetics of the fall in serologic parameters following B lymphocyte depletion, posttreatment values at each time point were expressed as a percentage of pretreatment values. The results are shown in responding (Figure 1A) and nonresponding (Figure 1B) patients. In the responder group, a gradual decline in the CRP and autoantibody levels was observed, with an apparent plateau reached at ∼12 weeks posttreatment (Figure 1A). In the nonresponder group (Figure 1B), the course of CRP and autoantibody decline was more erratic, although the levels of IgA-RF and IgM-RF did reach an apparent plateau, also at ∼12 weeks posttreatment. Comparison of these percentages of pretreatment values for any of the autoantibodies and CRP at the specific posttreatment time points revealed no significant difference between the 2 groups of patients, except for the changes in IgA-RF at 4 weeks (P = 0.03) and in CRP at 26 weeks (P = 0.04) posttreatment. Similarly, as shown in Table 2, the only significant difference in the absolute values of all measured parameters between the responder group and the nonresponder group was in the CRP levels (at nadir).

Comparison of the effect of B lymphocyte depletion on the levels of total serum immunoglobulins and autoantibodies.

Following treatment, B lymphocytes were undetectable (<0.005 × 109/liter) in the peripheral blood of patients for a mean of 8.4 months (range 3.5–20.8 months) (Table 1). Total serum immunoglobulins decreased moderately, with IgM dropping from a mean of 1.5 ± 1.03 gm/liter to 0.9 ± 0.9 gm/liter, IgA from a mean of 2.7 ± 1.5 gm/liter to 2.0 ± 1.1 gm/liter, and IgG from a mean of 11.9 ± 3.3 gm/liter to 8.8 ± 2.3 gm/liter. In only 3 patients did the IgG levels drop below the lower limit of normal, whereas in 8 patients the IgM levels dropped below the normal range (data not shown).

In order to compare the effect of B lymphocyte depletion on the levels of individual autoantibodies and their respective classes of serum immunoglobulins, the data were normalized by expressing the lowest value attained for each parameter as a percentage of the pretreatment value. Differences between the percentage drop in the 3 classes of RF measured became apparent (Figure 2). The percentage drop in IgG-RF levels was greatest and was also significantly greater than that in IgA-RF levels (P = 0.007, by Student's t-test), but not when compared with that in IgM-RF levels (P = 0.09). There was no significant difference between the decrease in IgA-RF and the decrease in IgM-RF, and no difference was found between the decreases in any of the RF classes and the decrease in anti-CCP antibodies. When decreases in the different autoantibodies and in their respective serum immunoglobulin classes were compared (Figure 2), the mean percentage decrease in IgA-RF (P = 0.00002) and IgG-RF (P = 0.000009), but not in IgM-RF (P > 0.05), was significantly greater than that found in the corresponding serum immunoglobulin-class levels. IgG anti-CCP also decreased significantly more than the drop in levels of circulating IgG (P = 0.001).

Temporal relationship of serologic parameters following B lymphocyte depletion.

Figure 3 compares the T50 (median time taken for serum levels to fall by 50% of the pretreatment levels) and T80 (the equivalent for 80% of pretreatment levels) for each autoantibody and for CRP. The T50 for IgG-RF, anti-CCP, and CRP was within 5 weeks of treatment. The T80 for the fall in all autoantibody levels was longer than that for the CRP levels, although the T80 for IgG-RF and anti-CCP was achieved sooner than that for IgA-RF and IgM-RF.

Effect of B lymphocyte depletion on protective antibody levels.

Anti-TT and anti-PCP antibodies were measured at 3 time points: baseline, 3 months after treatment, and at or shortly after the time of B lymphocyte return. Before treatment, 7 patients had subprotective levels of anti-TT antibodies (<0.1 IU/ml) and 11 patients had subprotective levels of anti-PCP antibodies (<50 mg/liter). At 3 months, anti-TT antibody levels had decreased a mean of 22.7% with no further decrease evident at the time of B lymphocyte return (Figure 5). Comparison with the percentage decrease in serum IgG at the same time points showed no difference. In those subjects whose baseline antibody levels were within the protective range, the levels remained protective at the second and third time points. The levels of anti-PCP antibodies did not change significantly with treatment (Figure 5). However, in 3 patients with baseline levels within the lower protective range, the anti-PCP levels had decreased to subprotective values at or shortly after B lymphocyte repopulation (data not shown). In 1 patient, anti-PCP levels increased 4-fold at 3 months, at the time of a respiratory infection.

Figure 5.

Percentage of pretreatment levels of anti–tetanus toxoid and anti–pneumococcal capsular polysaccharide antibodies at 3 months after B lymphocyte depletion and at or near B lymphocyte return. Bars show the mean and SD. For comparison, levels of serum total IgG at the same time points are also shown.

Factors associated with relapse.

At the time of submission, all except 2 patients (patients 11 and 13 in Table 1) had undergone clinical relapse. All patients who have relapsed have done so at or up to 17 months after B lymphocytes were again detectable in the circulation (>0.005 × 109/liter). In all patients except 1 (patient 16 in Table 1), clinical relapse was preceded by a detectable rise in autoantibody levels. Rises in autoantibody levels, usually IgM-RF, in the absence of relapse were rarely seen (on 4 occasions only) and were transient in nature (data not shown). Figures 4A–D show the absolute values of IgA-, IgM-, and IgG-RF and anti-CCP antibodies in sera from individual responding patients at the pretreatment time point, at the time when patients had reached their maximum ACR levels of improvement, and at relapse. Levels of all autoantibodies were found to decrease following treatment and, in most cases, to show a relative increase in titer at or close to the time of relapse. There was no clear preference for rises in any particular class of RF or anti-CCP antibodies in association with relapse. In addition, no new autoantibody classes or specificities were found to emerge (Table 1).

Serial studies of 2 responding patients.

Figures 6A and B show 2 examples of the change in serologic variables during the time following B lymphocyte depletion. The patient shown in Figure 6A (patient 1 in Table 1) remained well for more than 12 months after B lymphocyte repopulation. The levels of IgA-RF and IgM-RF were seen to rise before relapse, with a rise in anti-CCP antibodies also occurring coincident with relapse. In the patient shown in Figure 6B (patient 4 in Table 1), detectable circulating B lymphocytes together with a rise in autoantibodies coincided with clinical relapse. As exemplified by the time courses of serologic responses in these 2 representative patients and the responder group data shown in Figure 1A, the reduction in inflammatory scores, as measured by a fall in CRP levels, had the general appearance of an exponential decay curve. The pattern of changes in autoantibody levels generally paralleled the CRP profiles in each patient. Although the slopes of the curves differed from patient to patient, there tended to be an early drop in autoantibody levels and CRP levels to a plateau, which was sustained until relapse.

Figure 6.

Serial studies of autoantibody (IgA-RF, IgM-RF, IgG-RF, anti-CCP), antimicrobial antibody (anti–tetanus toxoid [anti-TT] and anti–pneumococcal capsular polysaccharide [anti-PCP]), and CRP levels in 2 representative patients (patients 1 [A] and 4 [B] in Table 1). Results for each parameter are expressed as a percentage of the pretreatment values. See Figure 1 for other definitions.


We found that B lymphocyte depletion had a selective effect on different circulating antibody populations in patients with RA. Sera from patients with a positive clinical response to treatment, in which an ACR response of ≥20% was achieved at 6 months, showed a significant fall in autoantibody and CRP levels, and this was not observed in nonresponders. Although total serum immunoglobulin levels also fell, the percentage decrease in IgA-RF, IgG-RF, and IgG anti-CCP antibodies was significantly greater in all patients. This was not the case for antimicrobial antibodies. Our findings extend the preliminary observations in patients with other autoimmune conditions, in whom a positive clinical response was associated with a significant fall in autoantibody levels following B lymphocyte depletion (9–11, 28).

In patients with lymphoma who were treated with rituximab, either alone or in combination with other drugs, the mean total immunoglobulin levels had been found to remain within the normal range (18, 19, 29, 30). This was taken as an indication that B lymphocyte depletion therapy would not produce significant benefit to autoimmunity if its effect were to be through reduction of autoantibody levels. Rituximab treatment in other autoimmune diseases has also not been found to result in a major drop in circulating immunoglobulin levels, with the possible exception of hemolytic anemia in children (31). In our group of patients, we did see a significant decrease in serum immunoglobulin levels, particularly in IgM, following B lymphocyte depletion, even though the mean values for all classes remained within the normal range. It is possible that in the patients studied herein, the greater drop in serum total immunoglobulins might reflect a large contribution of autoantibodies to circulating immunoglobulin levels.

Treon and Anderson (32) have suggested that only autoimmune conditions associated with pathogenic antibodies of the IgM class, e.g., polyneuropathies, might respond to rituximab treatment. In such conditions and in lymphoma, they suggested that IgM-class antibodies may be derived from CD20-positive plasmablasts. In our studies of patients with RA, we observed that the levels of all autoantibodies measured decreased, and for IgA-RF, IgG-RF, and IgG anti-CCP, the decrease was proportionately greater than the decrease in their respective total immunoglobulin classes. In addition, despite the half-life of IgG being longer than that of other immunoglobulin classes, IgG-RF and IgG anti-CCP antibodies decreased more rapidly than did IgA-RF or IgM-RF. In contrast to what was observed with autoantibodies, the levels of anti-PCP antibodies did not change significantly following treatment, and the anti-TT response closely corresponded to the changes in total serum IgG levels.

Such a selective effect on autoantibody production suggests that their production may be more dependent on the constant generation of new plasma cells from CD20-positive B lymphocytes. It is now known that some plasma cells have short lifespans, but other plasma cells may be able to live for extended periods of time (33). It is also possible that the clones responsible for antimicrobial antibodies are resident in the spleen, where they may be slowly turning over into plasma cells. Autoreactive clones are possibly in a more dynamic situation because of their constant generation and, consequently, many more may be entering the circulation. Their location in tissues other than secondary lymphoid organs may also render autoantibody-committed clones more susceptible to B lymphocyte depletion.

After rituximab administration, B lymphocyte depletion in patients occurs rapidly (within days) in the peripheral blood (17). Animal studies have shown that depletion of B cells in lymphoid tissue is rapid and unlikely to continue beyond 2 weeks (16). In this study, although patients showed differences in the timing and degree of their clinical response as measured by the ACR (15), the drop in CRP, which is an indicator of cytokine production, followed similar patterns in responding patients. This consisted of a gradual decrease in the levels of CRP over weeks to months that then reached a plateau until relapse, which was usually preceded by a rise in autoantibody levels. The reduction in autoantibody levels also occurred gradually and was reflected in the median times taken for them to decrease to 50% and 80% of pretreatment levels. These results are consistent with the effect of B lymphocyte depletion on reducing the progenitors of daughter plasma cells and thereby reducing autoantibody production. This contrasts with the rapid clinical and serologic (CRP) response of patients with RA to anti–tumor necrosis factor α treatment, which reflects the swift removal of a key effector in the inflammatory process (34). In our study, when responders were compared with nonresponders, only the former showed significant drops in autoantibody levels (as well as in CRP) in response to B lymphocyte depletion. Whether this simply reflected differences in the relative degree of B lymphocyte depletion in lymphoid organs in nonresponder patients when compared with responders could not be determined, since no tissue samples were obtained.

As shown in the representative serial studies and by the cumulative data from responding patients, the kinetics of the serologic (CRP) response to treatment paralleled the decline in circulating autoantibody levels. However, B lymphocyte return always preceded relapse. Relapse was often preceded by or coincided with a rise in 1 or more autoantibodies. On only 4 occasions were rises in autoantibodies (usually, small, transient rises in IgM-RF) detected without associated clinical manifestations of more active disease. The often long gap (up to 17 months) between B lymphocyte repopulation and relapse also suggested that relapse was not solely related to the presence of B cells.

The long period between B lymphocyte return and relapse seen in some patients treated with B lymphocyte depletion suggests that generation of new B cell clones capable of engaging in a vicious cycle of expansion may be a rate-limiting step in the recrudescence of disease. This may involve the generation of sufficient pathogenic B cell clones able to either restimulate autoreactive T cells or be precursors of autoantibody-producing plasma cells. It remains uncertain whether the reappearance of autoantibodies following B lymphocyte depletion represents the re-expansion of preexisting B lymphocyte clones or whether new clones are generated. The exact degree of B lymphocyte depletion achieved in solid lymphoid tissues in our patients remains unknown. There is also no information on what might happen to surviving B lymphocytes and to plasma cells during a period of B lymphocyte depletion. Therapeutic B lymphocyte depletion provides an excellent opportunity to learn more about these and other aspects of B lymphocyte and plasma cell biology and their roles in disease pathogenesis.


The authors want to thank Dr. Chris Bunn and the Immunopathology Laboratory, Royal Free Hospital for doing the antimicrobial antibody tests, and the Arthritis Research Campaign, which generously supported this work.