SEARCH

SEARCH BY CITATION

Keywords:

  • anti-CD20;
  • anti-platelet antibodies;
  • neutrophils;
  • Fc receptors;
  • T cells;
  • CD8

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Authorship contribution
  8. Conflicts of interest
  9. References

Rituximab is widely used in autoimmune diseases including immune thrombocytopenia (ITP), although the mechanism of effect remains unclear. This study describes the effects of rituximab on platelet-associated antibodies (PA-APAs), B and T cell counts and clonality (IGHV and TRG@ gene rearrangements), FCGR3A (FcγRIIIa) and FCGR2A (FcγRIIa) polymorphisms and correlation to anti-CD40 ligand (CD40L) response. PA-APA levels fell more frequently in responders (6/8) than in non-responders (2/10: P = 0·08–0·15). Two responders had no PA-APAs. Two non-responders with a fall in PA-APAs had very high CD8 levels. One non-responder had a B cell clone, one responder and one non-responder had a T cell clone. 15/16 patients had the same responses to rituximab and antiCD40L. Patients with FCGR3A V/V polymorphisms were more likely to respond to rituximab (P = 0·03). In summary, the fall in PA-APAs in responders confirms the humoural effect of rituximab. Failure to respond in patients with very high CD8 levels, despite PA-APA fall indicates a role for T cell-mediated platelet/megakaryocyte destruction. Concordance of response to anti-CD40L suggests autoantibody-producing cells are under T cell control. Finally, the effect of FCGR polymorphisms on response confirms the importance of FCGR-mediated depletion of B cells in autoimmunity. This has implications on the pathology of ITP as well as the immunological effect of B cell depletion.

Immune thrombocytopenia (ITP) is an autoimmune disease in which the thrombocytopenia results from both accelerated destruction and sub-optimal production of platelets (Cines et al, 2009). Initial studies demonstrated a plasma-derived destructive effect, later identified as an anti-platelet antibody (APA) (Harrington et al, 1951). More recently, defects in cellular responses have been identified, including abnormalities in T-helper cells, defects in regulatory T cells, evidence that cytotoxic T cells can cause thrombocytopenia (Semple et al, 1996; Semple, 2002; Olsson et al, 2003, 2005; Yu et al, 2008).

The use of rituximab in autoimmunity developed because of its ability to deplete circulating B cells (Protheroe et al, 1999; Edwards et al, 2005). However, why B cell depletion is effective is not completely clear for a number of reasons. When rituximab is successful, resolution of autoimmunity occurs despite no change in immunoglobulin G (IgG) levels (Cambridge et al, 2006; Ferraro et al, 2008). Furthermore, changes in T cell number and function, seen in patients responding to rituximab (Stasi et al, 2007, 2008) and in mice models of B cell depletion (Bouaziz et al, 2007), imply that auto-reactive B cells contribute to the pathology either as an antigen-presenting cell (APC) or via cytokine effects stimulating auto-reactive, cytotoxic T cells.

This study takes the approach of extensive analysis in a small number of patients. Reported here are the laboratory findings from the original clinical cohort (Cooper et al, 2004) including changes in APA levels after rituximab and the influence of the T cell compartment and the FCGR3A V/F polymorphism on responses to treatment. In addition, in view of the impact on T cells on response to rituximab and the nature of CD40-CD40L in the T-B cell interaction, responses to rituximab were compared with results of treatment with anti-CD40 ligand (anti-CD40L).

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Authorship contribution
  8. Conflicts of interest
  9. References

Patients

Patients had been enrolled in two previously published clinical treatment studies at the New York Presbyterian Hospital, Cornell campus, New York, NY, USA (n = 23) and Regina Apostolorum Hospital, Albano Laziale and the University of Rome, Italy (n = 32) between February 1999 and April 2002 (Cooper et al, 2004). Patients were required to have primary ITP and a platelet count <30 × 109/l; all but one patient had a >6-month history of ITP. Most of the data reported here derived from the patients enrolled in New York, some also originated from patients enrolled in Rome (see analysis of data). Not all patients enrolled in the clinical studies contributed data to these laboratory sub-studies.

Treatment

Patients received rituximab (Rituxan; Genentech, Inc, San Francisco, CA, USA and Mabthera; Roche, Basel, Switzerland) 375 mg/m2 intravenously weekly for four consecutive weeks (Cooper et al, 2004). Patients also received 25–50 mg diphenhydramine and 650 mg acetaminophen orally as premedication. Seventeen patients at the Cornell site received prednisone (60 mg with the first infusion, 20 mg with the second infusion, and none subsequently) after fever-chill reactions were seen in the first six patients who received only acetaminophen and diphenhydramine. During the initial treatment period, according to protocol, patients could receive short-term ‘rescue’ therapy with intravenous immunoglobulin (IVIG), intravenous (IV) anti-D, and/or IV methylprednisolone or platelet transfusion for bleeding and/or platelet count <10–20 × 109/l.

Anti-CD40L

Sixteen patients were also treated with anti-CD40L: 15 before receiving rituximab and one after receiving rituximab. Two different antibodies were used in open-label Phase I–II pilot studies between 1997 and 2002. Nine rituximab-treated patients received hu5c8 (BG9588, Antova; Biogen, Cambridge, MA, USA) and 7 received IDEC-131 (toralizumab; IDEC, San Diego, CA, USA). Details of infusions and response rates to these antibodies have been reported (Patel et al, 2008). The responses to anti-CD40L and to rituximab in patients receiving both therapies in separate single agent studies were compared.

Responses

As previously reported (Cooper et al, 2004), a complete response (CR) was defined as an increase in platelet counts to >150 × 109/l on two consecutive occasions 1 week apart. A partial response (PR) was defined as an increase in the platelet count to between 50 and 150 × 109/l on two consecutive occasions, 1 week apart. Duration of response was considered from the day of the initial infusion to the first time of relapse (platelet count <30 × 109/l) or to time of analysis (Cooper et al, 2004).

Laboratory investigations

Complete blood counts, including absolute neutrophil counts, were performed weekly on the routine Coulter counter for the first 5 weeks of study, then every other week until 16 weeks and every 3 months until the end of study (week 52).

Immunoglobulins

IgG and IgM levels were measured using nephelometry in the routine chemistry laboratory before and 12 weeks after treatment.

Immunophenotyping

Immunophenotyping was performed to assess B cell, T cell and Natural Killer (NK) cell numbers and as a percentage of lymphocytes on freshly drawn heparinized peripheral blood pre-treatment, at 5, 12, 26, 36 and 52 weeks in all patients in whom APAs were assessed. Flow cytometry (FACSCalibur; Becton-Dickinson-Pharmingen, San Jose, CA, USA) was used to assess expression of cell surface markers on whole blood. Two- and three-colour staining was carried out for 15 min at room temperature and samples were fixed with 1·5% paraformaldehyde. Specific lymphocyte subsets were identified using combinations of monoclonal antibodies to lymphocyte surface antigens directly conjugated to the fluorochromes fluorescein isothiocyanate (FITC, BD), phycoerythrin (PE, BD), and PE- Cy-5 (TRI-COLOR; CALTAG, Burlingame, CA, USA). For each analysis 15 000 events were acquired in the lymphocyte-gating region using Cell Quest software. T lymphocytes were assessed as CD3+CD4+ and CD3+CD8+; B lymphocytes were CD3−/CD19+; and NK cells were CD16/CD56.

Anti-platelet antibodies

Samples in this study were tested in an active clinical diagnostic laboratory using routine sample requirements. Whole blood samples were collected in acid-citrate-dextrose, solution A, kept refrigerated at (4°C) temperature and tested within 4 d of being taken. Samples of 40 ml were required for patients with platelet counts <100 × 109/l and 10 ml for those with platelet counts >100 × 109/l. Eluate preparation was performed with the GTI Pak Auto kit according to the manufacturers' instructions (GTI Diagnostics, Brookfield, WI, USA). The kit utilizes a low Ph eluate step to ‘strip’ bound autoantibodies off washed, intact patient platelets, followed by subsequent neutralization and then plating of the eluate onto a microtitre plate containing three different immobilized glycoprotein (GP) complexes – GPIIB/IIIA, IB/IX, AND IA/IIA. Antibody bound to the immobilized GPs is detected via a standard enzyme-linked immunosorbent assay (ELISA).

The sensitivity/specificity of the Pak Auto method has been reported (Davoren et al, 2005). Using a well-characterized cohort of ITP patients, we found that the sensitivity of the direct method was 91% and of the indirect (plasma phase) was 40%.

Both plasma and eluate (platelet-associated) IgG and IgM antibodies to platelet GPIIB/IIIA, IB/IX and IA/IIA complexes were assessed in 20 (plasma) and 18 (eluate) patients both pre-treatment and at week 5; samples from 13 of these patients were also measured at week 26. Test results showing optical density (OD) values ≥ twice the value obtained for the mean of the negative control for the corresponding GP complex were regarded as positive results. Positive reactions were graded from 1+ to 4+ based on multiples of the control value (Davoren et al, 2005).

IGHV and TRG@ gene rearrangements

To assess B and T cell clonality, IGHV gene rearrangements using semi-nested polymerase chain reaction (PCR) extraction with primers targeting IGHV FR3/IGHJ and TRG@ gene rearrangements using separate PCRs for each sample with primers targeting multiple TRGV and TRGJ regions were assessed. Sixteen patients were tested pre-treatment: 13 patients were tested again at weeks 5 and 12 of study.

Thrombopoietin

Thrombopoietin levels were measured before treatment and at weeks 1, 5 and 12 using a sandwich ELISA in 23 patients as described (Folman et al, 1997).

FCGR polymorphisms

FCGR2A-131H/R and FCGR3A-158V/F polymorphisms were assessed using a PCR-based allele-specific restriction analysis assay in 23 and 48 patients respectively, as previously described (Koene et al, 1997).

Statistical analysis

B cells, T cells, NK cells, CD4/8 ratios and thrombopoietin levels before, and at various time points after treatment were compared between responders and non-responders using the T test. Changes over time for the variables, including CD4, CD8, NK and APAs within responders or non-responders were assessed using the paired T test. Differences in the degree of change of APA levels after treatment were assessed between responders and non-responders using the Mann–Whitney U-test. Differences in responses for the FCGR polymorphisms were assessed using the Chi squared test. Significance was determined with a P value of <0·05. Analysis of differences between early and late responders was descriptive only (the numbers were too small for statistical analysis).

Certain sub-analyses were restricted to the first 21 patients treated at the Cornell site and included: comparison of clinical responses to anti-CD40L and to rituximab; platelet-associated GP-specific antibodies; studies of clonality; CD4, CD8 and NK levels and thrombopoietin levels.

B cell depletion and reappearance of B cells were described for 42 patients (USA and Italy). FCGR2A and FCGR3A polymorphisms were examined in 23 (USA alone) and 48 patients (USA and Italy), respectively.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Authorship contribution
  8. Conflicts of interest
  9. References

The response rate to rituximab for the overall study has been previously reported: 55% of 57 patients responded to rituximab (with 32% achieving a CR) (Cooper et al, 2004). There were eight responders (one PR and seven CRs) and 10 non-responders within the group of patients who were tested for APAs,.

Six of eight responders and 8 of 10 non-responders received IVIG rescue therapy within the 5 weeks between APA testing. Two responders and two non-responders received weekly IVIG therapy. As the numbers were small, the influence of IVIG therapy on APAs has not been evaluated.

Responses to anti CD40L and anti CD20

Sixteen patients were treated with both an antibody to CD40L [Biogen (n = 9) and IDEC (n = 7)], and with rituximab (anti-CD20); in 15, rituximab was the second therapy (Table 1). Responses to treatment are shown in Table 1. Fifteen patients (94%) had identical responses to anti-CD40L and to rituximab (four CR, one PR and 10 NR). One patient had a PR to anti-CD40L and a subsequent CR to rituximab.

Table 1. Responses to anti-CD20 and anti-CD40L
PatientAnti-CD40L (Agent) responseAnti-CD20 responseTime between CD40L and R (months)
  1. B, Biogen; I, IDEC; NR, no response (platelet count <50 × 109/l); PR, partial response (platelet count >50 × 109/l); CR, complete response (platelet count >150 × 109/l); UNK, unknown date for either anti-CDL or anti-CD20.

  2. a

    Patient received rituximab first and CD40L second.

1(B) CRCRUNK
2(B) NRNR21
3(B) CRCR17
4(B) NRNR62
5(B) NRNR14
6(B) CRCR18
7(B) NRNR14
8(B) NRNR7
9(B) NRNR8
10(I) CRCR17
11(I) PRPR31
12(I) NRNR12
13(I) NRNR18
14(I) NRNR25
15(I) PRCR16
16a(I) NRNRUNK

Anti-platelet antibodies

Twenty patients had evaluation of plasma-derived APAs before and after treatment with rituximab. Eighteen patients had eluate, or PA-APAs evaluated before and after rituximab.

Antibodies detected in plasma

Responders (n = 10)

Five of 10 responders had measureable plasma-derived anti-platelet glycoprotein antibodies. Four had a fall in plasma IgG APA after rituximab, with two having no detectable antibody by week 5. The fifth patient with unchanged circulating APA was a late responder.

Non-responders (n = 10)

Five of 10 non-responders had detectable APAs in their plasma. Four patients showed no change in the antibody levels at week 5 and the 5th patient had an increased level of plasma APA at week 5.

Eluate/platelet associated antibodies

Results of platelet-associated, glycoprotein-specific antibody (PA-APA) testing are shown before rituximab and 5 weeks after rituximab in responders and in non-responders (Fig. 1A–F). Six of eight responders and 8 of 10 non-responders had APA levels of >1+ to one or more of three platelet GP before treatment.

image

Figure 1. Platelet-associated antiplatelet antibodies (PA-APAs) in responders and non-responders: Anti- GPIIB/IIIA(A, B), -IB/IX(C, D) and -IA/IIA(E, F) levels are shown before and 5 weeks after treatment with rituximab in patients who responded to rituximab (1: A, C, E) and those who did not respond (1: B, D, F). There was a trend, but no significant difference in the changes in PA-APA levels at 5 weeks between responders (R) and non-responders (NR) (anti-GPIIB/IIIA P = 0·08, anti-GPIB/IX P = 0·15, and anti-IGPA/IIA P = 0·06).

Download figure to PowerPoint

Responders (n = 8)

The six responders with high levels of PA-APAs before rituximab, had a fall in at least 2 (anti-GPIIB/IIIA, anti-GPIB/IX and anti-GPIA/IIA) of the PA-APAs after B cell depletion (Fig. 1A,C,E). Two late responders, one with high antibody levels (4+) to all three glycoproteins pre-infusion and a second with only anti-GPIIB/IIIA (4+), had no PA-APA response at week 5. Their platelet increases did not occur until 16 and 8 weeks after rituximab respectively; both had a fall in their PA-APA by week 26.

Two responders had little or no measurable antibodies before or after treatment.

Non-responders (n = 10)

The 10 non-responders had variable changes in PA-APA levels (Fig. 1B,D,F): eight had very high levels before treatment (3+ or 4+), only two of these patients had a fall in all three PA-APAs to 0 or 1. These two patients had particularly high CD8 levels, which did not change after rituximab (see below).

There was no difference between responders and non-responders in the levels of platelet-derived APA before rituximab. There was a trend to a greater decrease in PA-APAs at week 5 in responders when compared to non-responders (anti-GPIIB/IIIA P = 0·08, anti-GPIB/IX P = 0·15 and anti-GPIA/IIA P = 0·06).

Platelet-associated IgM APAs were only detectable in a minority of patients and there was no significant difference between responders and non-responders.

Week 26 PA-APAs

PA-APAs were also measured at 26 weeks in six responders and six non-responders. In all six responders PA-APAs were lower at week 26 than before rituximab. This included the two late responders. In contrast, in four of six non-responders, PA-APA levels at week 26 remained high (3+ or 4+).

Immunoglobulin levels

Twenty-five of 41 patients had a fall in IgG levels at week 12, although the difference was small. While four patients had a fall >20% from baseline, only two patients had an IgG level below normal at week 12. Twenty-eight patients had a fall in IgM levels, with a fall of >20% in seven patients; only three patients had levels below normal. There was no difference between responders and non-responders in immunoglobulin levels or their changes with rituximab treatment.

Peripheral B cells

As previously reported, there was a faster return of B cells, with a higher number of B cells at week 26 in patients who did not respond to rituximab when compared to those who did respond (P = 0·05). There was also a trend to a faster return of B cells in patients with a PR than in those with a CR (Cooper et al, 2004).

T and B cell clonality

One of 13 patients had evidence of a B cell clone by PCR gene re-arrangement before starting treatment. This patient did not respond to rituximab therapy and at no point displayed evidence of a lymphoproliferative disorder. The PCR 3 months after rituximab showed that there had been a marked decrease in oligoclonal B cell bands and numbers in all patients.

Two patients showed a T cell band consistent with clonality. One patient had a complete platelet response to rituximab while one had no response, even though there was no evident change in the T cell clonality following rituximab treatment in either patient.

CD4+ CD8+ T cell and NK cell numbers (n = 18)

There were no overall differences between responders and non-responders regarding CD3, CD4, CD8 or NK cell numbers before treatment or at week 26. There was a trend towards a decrease in CD8 counts in responders at 26 weeks (P = 0·18), which did not occur in non-responders.

The two non-responders with a substantial fall in APA (from 4+ and 3+ to 0 or 1+) at week 5 but no platelet response had very high CD8 counts, 3·55 × 109/l and 1·735 × 109/l, which did not change substantively following treatment (Fig. 2). These two patients had the FCGR3A V/V polymorphism associated with good responses to rituximab (see below). The CD8 levels in the remaining patients ranged from 0·076 to 0·814 × 109/l pre-treatment (Fig. 2).

image

Figure 2. CD8 counts in patients treated with rituximab, depending on responses and APA changes: The two patients who did not respond to rituximab, despite a fall in APAs had much higher CD8 levels than the remaining patients. The red dotted line and red triangles represent the APAs and CD8 counts respectively, in these two patients.

Download figure to PowerPoint

Of the two responders who had no (or very minimal) PA-APAs before rituximab, both had essentially normal CD3, CD4, CD8 and NK cell levels before and after rituximab (one had slightly low CD4 and CD8 counts before treatment, with an increase after rituximab).

Neutrophil counts

There was no significant change in neutrophil counts in the 57 patients treated with rituximab over the 12 weeks of follow up (Fig. 3). Even when the few patients on concomitant steroids were removed from the analysis, non-responders had a higher neutrophil count than responders at most time points.

image

Figure 3. Neutrophil counts over time in responders and non-responders. Median absolute neutrophil counts (ANC) are shown over time in patients who responded (Resp., n = 36) or did not respond (Non Resp., n = 21) to rituximab. In order to avoid confounding factors, patients who received prednisolone treatment (Pred) during the study period were separated from those who had not (No Pred). Patients who responded had persistently lower neutrophil counts.

Download figure to PowerPoint

Thrombopoietin levels

There was no significant difference between responders and non-responders regarding thrombopoietin levels before treatment or at 12 weeks (Fig. 4). The levels tended to be higher in non-responders at all time points and to fall in both groups during the 12 week follow-up period.

image

Figure 4. Thrombopoietin levels over time after rituximab. Thrombopoietin (TPO) levels (arbritrary units/ml) in responders and non-responders are shown over time. The numbers within the figure indicate the number of patients in whom TPO was measured at each time point. Non-responders (diamonds) appear to have a higher TPO level throughout the study when compared to responders (squares). Significance was not assessed due to small numbers at different time points. Inf 1, infusion 1 of rituximab.

Download figure to PowerPoint

FCGR polymorphisms

Patients who had a platelet response to rituximab were significantly more likely to have the FCGR3A V/V polymorphism compared to the V/F or FF polymorphism (P = 0·03) (Fig. 5). There was no difference between responders and non-responders regarding the FCGR2A H/R polymorphism (n = 23).

image

Figure 5. Response to rituximab according to (A) FCGR3A and (B) FCGR2A polymorphisms: Representation of the number of patients with each polymorphism for responders and non-responders are shown in the bar charts. Patients who responded to rituximab were more likely to have the FCGR3A V/V polymorphism than the F/F or V/F polymorphism (P = 0·03) (A). There was no difference between responders and non-responders regarding FCGR2A polymorphisms (B).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Authorship contribution
  8. Conflicts of interest
  9. References

Despite almost universal B cell depletion, not all patients with ITP have a platelet response after rituximab. Differences in the occurrence, timing and quality of the response cause us to question our current understanding of how B cell depletion restores tolerance in patients with autoimmunity. Recent studies in patients with ITP imply that responses to rituximab are related to changes in T cells, in particular, resolution of TRBV skewing, changes in T-helper subsets, and improvement in regulatory T cell function (Stasi et al, 2007, 2008). Assessment of APAs after rituximab has not previously been reported. By intensively measuring both T cell numbers and anti-platelet antibodies in a small number of patients and their relationship to subsequent differences in responses to B cell depletion, we can speculate about the mechanism of effect of rituximab in ITP.

Despite well-known limitations (depending upon the assay) of the sensitivity and specificity of anti-platelet antibodies, reduction in PA-APA levels was associated with an increase in the platelet count in the majority of patients. Six of the eight patients with a fall in PA-APAs had an increase in their platelet counts. Furthermore, the timing of the PA-APA fall corresponded to that of the platelet increment; PA-APA fall was seen at 5 weeks in early responders compared to 26 weeks in late responders. Consistent with patients with systemic lupus erythematosus (SLE) (Cambridge et al, 2006; Ferraro et al, 2008), the autoantibody responses occurred despite no clinically significant fall in overall immunoglobulin levels, suggesting that rituximab targets autoantibody-producing B cells out of proportion to other B cells and long-lived plasma cells. This presumably reflects production of autoantibodies by B cells or short-lived plasma cells (Stasi, 2010).

In the two responding patients without detectable PA-APA, the antibody may not have been optimally measured. Alternatively, responses to rituximab in these patients may relate to removal of B cell activation of T cells. No obvious changes in T or NK cells were seen in these patients, although regulatory T cells and TRBV repertoire were not assessed.

In those patients who did not respond, the persistence of PA-APAs suggests that rituximab, despite peripheral blood B cell depletion, had not removed the antibody-producing cell. Further analysis also reveals the potential role of CD8 cell-mediated platelet or megakaryocyte destruction in non-responders. Two patients who did not respond to rituximab, despite a fall in PA-APAs, had very high CD8 levels, much higher than those of the remaining patients (Fig. 2). These patients presumably had both antibody and cellular elements contributing to platelet/megakaryocyte destruction. Removal of the antibody-mediated arm was not sufficient to cause a significant platelet increment. These patients may require dual therapy, attacking both the cytotoxic T cell and autoantibody production. This is consistent with a recent mouse model of ITP in which both cellular and antibody-mediated disease is described, and where responses to treatment are different (Chow et al, 2010). It is also consistent with data showing the direct effect of cytotoxic T cells in ITP (Olsson et al, 2003).

The similar responses to anti-CD40L and anti-CD20 (rituximab) implies that the autoantibody-producing cells in patients who respond to rituximab are driven by T cells, i.e., the interruption of the T-B cell interaction has the same effect as B cell depletion, suggesting the prime pathology remains in the autoreactive T cell. Alternatively, stimulation may be bidirectional, in which case, removal of the autoreactive B cell prevents further T cell activation. This would explain the responses seen in patients who did not have APAs before treatment and is consistent with earlier findings described above (Stasi et al, 2007, 2008). It is also consistent with the findings that only one patient (who did not respond) had a detectable B cell clone – implicating the importance of reducing overall B cell numbers – and that the presence of T cell clones did not predict response to rituximab. Descriptions of abnormal T cell responses in patients who are congenitally without B cells further highlight the importance of the B to T cell stimulation, a concept that is relatively under-examined (Martini et al, 2011).

The data also demonstrates that patients who responded to rituximab were more likely to have the FCGR3A V/V polymorphism. Although the numbers are small, the results are the same as previously reports in both lymphoma and SLE (Anolik et al, 2003). This is consistent with data from this patient population in which responders had a longer duration of B cell depletion (Cooper et al, 2004) and confirms an important role for FCGR3A in depleting B cells coated with antibody. Interestingly, using a double dose of rituximab in a trial for relapsed responders resulted in only a marginally longer B cell depletion and patients relapsed again in the same time window, suggesting a handling effect of antibody-coated B cells rather than a rituximab dose-effect (Hasan et al, 2009). The more efficient B cell depletion in V/V responders may also represent a more profound B cell depletion in the lymph nodes and bone marrow, interrupting antigen presentation and autoantibody production, hence the better platelet responses in these individuals. Of note, FCGR polymorphisms did not affect the timing (early versus late) of response.

Finally, thrombopoietin levels had no particular relevance to responses to rituximab. The lower neutrophil counts in non-responders may relate to complicated, poorly understood features of pathophysiology of patients with ITP.

The primary limitation of this study is the relatively the small numbers of ITP patients investigated in relation to treatment. However, by extensively studying these patients, we can begin to see patterns, which will help in the design of studies to further explore the nature of loss of tolerance in autoimmunity and its restoration with different therapies. Overall, the autoantibody data and associated T cell changes presented here, in limited numbers of patients, demonstrate four potential pathophysiological mechanisms of ITP: (i) Patients with predominantly B cell autoantibody-mediated disease, (ii) Patients with antibody-mediated disease and cytotoxic T cell disease, (iii) Patients with T cell-mediated diseases responding to B cell depletion; and (iv) Patients with plasma cell-producing autoantibodies, which are not destroyed by rituximab.

Another limitation is that these patients received other therapies, both before and during the study. All the therapies could potentially influence autoantibody measurement, such as IVIG therapy during the first 5 weeks given to both responders and non-responders, or the earlier CD40L therapy. The numbers in this study remain too small to evaluate all possibilities, but the almost significant difference between responders and non-responders, despite all these limitations and the time interval between these earlier therapies and rituximab, suggest that these are important preliminary findings.

The emerging data both from clinical trials of B cell depletion therapy and intriguing new mice and human data, including the in vivo data described here, show a prominent role for the B cell in modulating the immune system in autoimmunity. These data also re-validate the importance of measurement of PA-APA. Further investigation of T cell-B cell dynamics in ITP could promote development of novel therapeutic strategies or at least improve the ability to individualize therapy. It also suggests that ideal treatment of particularly difficult cases may require ameliorating both cytotoxic T cells and APA. Optimizing this approach would potentially involve serial testing of each pathway with modification of therapy according to the results.

Acknowledgements

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Authorship contribution
  8. Conflicts of interest
  9. References

This study was partially supported by Genentech (South San Francisco, California) and by the Children's Cancer and Blood Foundation. NC receives support from the Biomedical Research Centre funding scheme. We would like to thank Richard Szydlo at the Hammersmith Hospital for his statistical support and Masja De Haas for her assistance with the FcR polymorphism determinations and Leendert Porcelijn and Claudia Folman for their measurement of thrombopoietin levels.

Authorship contribution

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Authorship contribution
  8. Conflicts of interest
  9. References

NC and JB designed, analysed and wrote the manuscript. RS provided patient data and samples from the Italian site and helped with design, analysis and writing. SCR performed the immunophenotyping, EC performed the T cell and B cell gene re-arrangements, and JGM performed the antiplatelet antibody analysis. All authors approved the manuscript in the form of submission.

Conflicts of interest

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Authorship contribution
  8. Conflicts of interest
  9. References

Nichola Cooper has received honoraria for participation on advisory boards and as a speaker at medical education events supported by GlaxoSmithKline and Amgen. Roberto Stasi has received honoraria for participation on advisory boards and as a speaker at medical education events supported by GlaxoSmithKline and Amgen. Jim Bussel currently receives clinical research support from the following companies: Amgen, Cangene, GlaxoSmithKline, Genzyme, IgG of America, Immunomedics, Ligand, Eisai, Inc, Shionogi and Sysmex and has participated in Advisory Boards for Amgen, GlaxoSmithKline, Ligand, Shionogi, Eisai and Portola. Jim Bussel's family owns stock in Amgen and GlaxoSmithKline.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. Authorship contribution
  8. Conflicts of interest
  9. References
  • Anolik, J.H. , Campbell, D. , Felgar, R.E. , Young, F. , Sanz, I. , Rosenblatt, J. & Looney, R.J. (2003) The relationship of FcgammaRIIIa genotype to degree of B cell depletion by rituximab in the treatment of systemic lupus erythematosus. Arthritis and Rheumatism, 48, 455459.
  • Bouaziz, J.D. , Yanaba, K. , Venturi, G.M. , Wang, Y. , Tisch, R.M. , Poe, J.C. & Tedder, T.F. (2007) Therapeutic B cell depletion impairs adaptive and autoreactive CD4+ T cell activation in mice. Proceedings of the National Academy of Sciences of the United States of America, 104, 2087820883.
  • Cambridge, G. , Leandro, M.J. , Teodorescu, M. , Manson, J. , Rahman, A. , Isenberg, D.A. & Edwards, J.C. (2006) B cell depletion therapy in systemic lupus erythematosus: effect on autoantibody and antimicrobial antibody profiles. Arthritis and Rheumatism, 54, 36123622.
  • Chow, L. , Aslam, R. , Speck, E.R. , Kim, M. , Cridland, N. , Webster, M.L. , Chen, P. , Sahib, K. , Ni, H. , Lazarus, A.H. , Garvey, M.B. , Freedman, J. & Semple, J.W. (2010) A murine model of severe immune thrombocytopenia is induced by antibody- and CD8+ T cell-mediated responses that are differentially sensitive to therapy. Blood, 115, 12471253.
  • Cines, D.B. , Bussel, J.B. , Liebman, H.A. & Luning Prak, E.T. (2009) The ITP syndrome: pathogenic and clinical diversity. Blood, 113, 65116521.
  • Cooper, N. , Stasi, R. , Cunningham-Rundles, S. , Feuerstein, M.A. , Leonard, J.P. , Amadori, S. & Bussel, J.B. (2004) The efficacy and safety of B-cell depletion with anti-CD20 monoclonal antibody in adults with chronic immune thrombocytopenic purpura. British Journal of Haematology, 125, 232239.
  • Davoren, A. , Bussel, J. , Curtis, B.R. , Moghaddam, M. , Aster, R.H. & McFarland, J.G. (2005) Prospective evaluation of a new platelet glycoprotein (GP)-specific assay (PakAuto) in the diagnosis of autoimmune thrombocytopenia (AITP). American Journal of Hematology, 78, 193197.
  • Edwards, J.C. , Leandro, M.J. & Cambridge, G. (2005) B lymphocyte depletion in rheumatoid arthritis: targeting of CD20. Current Directions in Autoimmunity, 8, 175192.
  • Ferraro, A.J. , Drayson, M.T. , Savage, C.O. & MacLennan, I.C. (2008) Levels of autoantibodies, unlike antibodies to all extrinsic antigen groups, fall following B cell depletion with Rituximab. European Journal of Immunology, 38, 292298.
  • Folman, C.C. , von dem Borne, A.E. , Rensink, I.H. , Gerritsen, W. , van der Schoot, C.E. , de Haas, M. & Aarden, L. (1997) Sensitive measurement of thrombopoietin by a monoclonal antibody based sandwich enzyme-linked immunosorbent assay. Thrombosis and Haemostasis, 78, 12621267.
  • Harrington, W.J. , Minnich, V. , Hollingsworth, J.W. & Moore, C.V. (1951) Demonstration of a thrombocytopenic factor in the blood of patients with thrombocytopenic purpura. Journal of Laboratory and Clinical Medicine, 38, 110.
  • Hasan, A. , Michel, M. , Patel, V. , Stasi, R. , Cunningham-Rundles, S. , Leonard, J.P. & Bussel, J. (2009) Repeated courses of rituximab in chronic ITP: three different regimens. American Journal of Hematology, 84, 661665.
  • Koene, H.R. , Kleijer, M. , Algra, J. , Roos, D. , von dem Borne, A.E. & de Haas, M. (1997) Fc gammaRIIIa-158V/F polymorphism influences the binding of IgG by natural killer cell Fc gammaRIIIa, independently of the Fc gammaRIIIa-48L/R/H phenotype. Blood, 90, 11091114.
  • Martini, H. , Enright, V. , Perro, M. , Workman, S. , Birmelin, J. , Giorda, E. , Quinti, I. , Lougaris, V. , Baronio, M. , Warnatz, K. & Grimbacher, B. (2011) Importance of B cell co-stimulation in CD4(+) T cell differentiation: X-linked agammaglobulinaemia, a human model. Clinical Experimental Immunology, 164, 381387.
  • Olsson, B. , Andersson, P.O. , Margareta Jernås, M. , Jacobsson, S. , Carlsson, B. , Carlsson, L.M.S. & Wadenvik, H. (2003) T-cell-mediated cytotoxicity toward platelets in chronic idiopathic thrombocytopenic purpura. Nature Medicine, 9, 11231124.
  • Olsson, B. , Andersson, P.O. , Jacobsson, S. , Carlsson, L. & Wadenvik, H. (2005) Disturbed apoptosis of T-cells in patients with active idiopathic thrombocytopenic purpura. Thrombosis and Haemostasis, 93, 139144.
  • Patel, V.L. , Schwartz, J. & Bussel, J.B. (2008) The effect of anti-CD40 ligand in immune thrombocytopenic purpura. British Journal of Haematology, 141, 545548.
  • Protheroe, A. , Edwards, J.C. , Simmons, A. , Maclennan, K. & Selby, P. (1999) Remission of inflammatory arthropathy in association with anti-CD20 therapy for non-Hodgkin's lymphoma. Rheumatology (Oxford), 38, 11501152.
  • Semple, J.W. (2002) Immune pathophysiology of autoimmune thrombocytopenic purpura. Blood Reviews, 16, 912.
  • Semple, J.W. , Milev, Y. , Cosgrave, D. , Mody, M. , Hornstein, A. , Blanchette, V. & Freedman, J. (1996) Differences in serum cytokine levels in acute and chronic autoimmune thrombocytopenic purpura: relationship to platelet phenotype and antiplatelet T-cell reactivity. Blood, 87, 42454254.
  • Stasi, R. (2010) Rituximab in autoimmune hematologic diseases: not just a matter of B cells. Seminars in hematology, 47, 170179.
  • Stasi, R. , Del Poeta, G. , Stipa, E. , Evangelista, M.L. , Trawinska, M.M. , Cooper, N. & Amadori, S. (2007) Response to B-cell depleting therapy with rituximab reverts the abnormalities of T-cell subsets in patients with idiopathic thrombocytopenic purpura. Blood, 110, 29242930.
  • Stasi, R. , Cooper, N. , Del Poeta, G. , Stipa, E. , Laura Evangelista, M. , Abruzzese, E. & Amadori, S. (2008) Analysis of regulatory T-cell changes in patients with idiopathic thrombocytopenic purpura receiving B cell-depleting therapy with rituximab. Blood, 112, 11471150.
  • Yu, J. , Heck, S. , Patel, V. , Levan, J. , Yu, Y. , Bussel, J.B. & Yazdanbakhsh, K. (2008) Defective circulating CD25 regulatory T cells in patients with chronic immune thrombocytopenic purpura. Blood, 112, 13251328.