Reconstitution of peripheral blood B cells after depletion with rituximab in patients with rheumatoid arthritis

Authors


Abstract

Objective

To study the quantitative and phenotypic reconstitution of peripheral blood B cells and its relationship to the dynamics of clinical response in patients with rheumatoid arthritis (RA) following B cell depletion with rituximab.

Methods

Twenty-four patients with active RA treated with rituximab were studied. Flow cytometry with combinations of monoclonal antibodies to B cell and T cell subsets was used.

Results

The frequency and total number of CD19+ cells in the peripheral blood decreased a mean of 97% for more than 3 months in all but 1 patient following rituximab therapy. All B cell populations were depleted. More than 80% of residual B cells showed a memory or plasma cell precursor phenotype. B cell repopulation occurred a mean of 8 months after treatment and was dependent on the formation of naive B cells, which showed an increased expression of CD38 and CD5. During repopulation, increased numbers of circulating immature B cells, CD19+,IgD+,CD38high,CD10low,CD24high cells, were identified. Patients who experienced a relapse of RA on return of B cells tended to show repopulation with higher numbers of memory B cells. A small number of T cells and natural killer cells expressed low levels of CD20. These cells were depleted following rituximab therapy and returned to the circulation a mean of 5 months after treatment. No other significant changes were detected in the T cell populations studied.

Conclusion

Rituximab induced a profound depletion of all peripheral blood B cell populations in patients with RA. Repopulation occurred mainly with naive mature and immature B cells. Patients whose RA relapsed on return of B cells tended to show repopulation with higher numbers of memory B cells.

Rituximab is a chimeric monoclonal antibody directed toward CD20, a pan–B cell surface marker that has proved very effective in depleting normal and malignant B lymphocytes in vivo and is widely used in the treatment of B cell malignancies, particularly B cell non-Hodgkin's lymphoma (1). In the last 6 years, rituximab has also been used in the treatment of several autoimmune diseases, and the results have been promising. In rheumatoid arthritis (RA), findings of earlier open-label trials suggested that B cell depletion protocols based on rituximab could be an effective therapy in seropositive patients, and this has now been confirmed in a phase II trial (2–5).

In lymphoma patients, rituximab induces almost complete depletion of normal B lymphocytes in the peripheral blood, which usually lasts 6–9 months. To our knowledge, there is no published information on the immunophenotype of the depleted and repopulated normal B cell populations in these patients (1). Data on the degree of depletion of normal B cells achieved in solid tissues, including bone marrow and secondary lymphoid tissues (lymph nodes and spleen), are very limited (6). Results of animal studies suggest that depletion in solid tissues is significant, but not complete, and that it varies from site to site and from individual to individual even if the same doses are used (7, 8). More recently, changes in B cell immunophenotype following treatment with rituximab in systemic lupus erythematosus patients and in dialysis patients awaiting renal transplantation have been described (9).

Since 1998, we have been using B cell depletion based on rituximab to treat patients with active, refractory RA (10). Patients in our cohort have not experienced clinical relapse before B cell repopulation of the peripheral blood has occurred. Approximately 50% experienced a relapse at B cell return (within 2 months), and the rest have relapsed at various times thereafter, but often, after an extended delay (up to 33 months) (3, 10). Clinical relapse was associated with a rise in autoantibody levels, particularly, rheumatoid factor of the IgM isotype (11). The purpose of the present study was to determine the degree of B cell depletion in patients with RA treated with rituximab, the variation between patients, evidence of different sensitivities of subsets of circulating B cells to rituximab treatment, the dynamics of repopulation of the peripheral blood (whether it occurred from naive B cells, memory B cells, or a mixture of both), and whether any of the findings correlated with clinical response to treatment, in particular, with the time of relapse.

PATIENTS AND METHODS

Patients.

Twenty-four patients with active, refractory RA entered the study, having received a total of 32 treatments. All patients were treated in the Department of Rheumatology at the University College London Hospitals (UCLH), according to clinical need. The study was approved by the Hospital Ethics Committee. All patients gave informed consent before entering the study.

Patients were assessed before treatment (baseline), at 1 month after the first rituximab infusion, every 2 months until B cell repopulation occurred, and every 2–3 months thereafter. In 20 patients (22 treatments), data were available for the baseline, 1-month, and 3-month assessments. Twenty-one patients (27 treatments) were studied during depletion and repopulation, and baseline data were available for 14 of these 21 patients (15 treatments).

Of the 24 patients enrolled into the study, 15 were women, and 9 were men. Their mean age at study entry was 59 years (range 32–80 years), and their mean disease duration was 20 years (range 5–42 years). All patients had active, seropositive RA. Patients had failed a mean of 4 disease-modifying antirheumatic drugs (DMARDs) (range 2–7). Eight patients were receiving oral methotrexate at baseline (mean dosage 14 mg/week; range 7.5–20). One patient was taking azathioprine (100 mg/day). A total of 6 patients were taking oral prednisolone at baseline (mean dosage 8 mg/day; range 5–10). Three of these 6 patients were also receiving a DMARD. Fifteen patients (including 2 of the patients taking methotrexate) had previously been treated with rituximab. The previous course of rituximab had been administered a median of 20 months (range 8–46 months) before the current course of rituximab was initiated.

Rituximab treatment.

Patients were treated with 2 infusions of 1,000 mg of rituximab given 2 weeks apart and accompanied by intravenous steroid as premedication (1 100-mg dose of methylprednisolone with each rituximab infusion). No patient received concomitant cyclophosphamide. Patients who were taking oral methotrexate and/or oral prednisolone at baseline continued these medications, and the dosages were reduced during the study according to clinical response to treatment. All other immunosuppressive agents were stopped.

Some of these patients have been the subjects of previous publications. The clinical response to a previous course of rituximab in 6 of the patients included in this study has been described (3). Preliminary data on peripheral blood B cell immunophenotype changes following treatment with rituximab were available for 5 of these patients in May 2003 and were included in a published abstract (12).

Monoclonal antibodies.

Three-color immunophenotyping of B and T lymphocytes and natural killer (NK) cells in peripheral blood was performed using matched combinations of anti-human murine monoclonal antibodies directly conjugated to fluorescein isothiocyanate (FITC), phycoerythrin (PE), or PE-Cy5 (or, Cy-Chrome). For analysis of B cells, combinations of anti-IgD (FITC), anti-CD5 (PE-Cy5), anti-CD19 (PE), anti-CD20 (FITC), anti-CD27 (FITC), and anti-CD38 (PE-Cy5) were used. For analysis of T cells and NK cells, anti-CD3 (PE), anti-CD4 (FITC), anti-CD8 (FITC), anti-CD20 (FITC), anti-CD25 (PE-Cy5), anti-CD45RA (PE or PE-Cy5), anti-CD45RO (PE), and anti-CD56 (PE-Cy5) were used. Four-color immunophenotyping of B lymphocytes upon repopulation was performed using matched combinations of anti-human murine monoclonal antibodies that had been directly or indirectly (through the biotin–streptavidin system) conjugated to FITC, PE, peridinin chlorophyll A protein (PerCP), or allophycocyanin (APC). Antibodies used included anti-IgD (PerCP), anti-CD10 (APC), anti-CD19 (PE or APC), anti-CD24 (PE), and anti-CD38 (FITC). All antibodies were purchased from BD PharMingen (San Diego, CA).

Preparation of samples.

Six milliliters of whole blood was collected by venipuncture into tubes containing EDTA. Samples were prepared within 7 hours after collection using a lysed whole-blood technique. Red cells were lysed by adding 30 ml of red cell lysis buffer (BD PharMingen) to 6 ml of whole blood. The mixture was gently mixed and incubated for 6 minutes at room temperature. After centrifugation, the cell pellet was washed twice in phosphate buffered saline (PBS) by centrifugation at 300g for 5 minutes, and then resuspended in cold PBS with 2% heat-inactivated fetal calf serum. The cells were incubated with each monoclonal antibody combination for 20 minutes at 4°C (50 μl of cell suspension with 20 μl of each antibody or equivalent amount for 106 cells). The samples were subsequently washed twice in cold PBS by centrifugation at 300g for 5 minutes. The cells were then fixed by incubation with 50 μl of PBS with 2% paraformaldehyde for 5 minutes at room temperature. The cells were washed twice by centrifugation at 300g for 5 minutes, and resuspended in 200 μl of cold PBS. Samples were protected from light and maintained at a temperature of 4°C until analyzed by flow cytometry. Analysis was performed either on the same day or on the day after the sample collection and preparation. Cell viability was evaluated before incubation with antibodies and was verified to be >90% by trypan blue assay.

Flow cytometry analysis.

Data were acquired with a FACSCalibur flow cytometer using CellQuest software (BD Biosciences Immunocytometry Systems). Analysis was performed after manual gating around a “lymphocyte” population with low to moderate forward-angle and low right-angle light-scatter properties. A sample with unstained cells was used as a negative control to compensate for autofluorescence of cells. Samples incubated with only 1 antibody were used to compensate for overlap between the spectra of the different fluorochromes. During the period of B cell depletion, each patient sample was tested in parallel with another sample in which B cells were present (from either another patient with B cells or a normal control). For B cell subpopulation analysis, a minimum of 20,000 events were collected in the lymphocyte gate. For T cell subpopulation analysis, a minimum of 10,000 events were collected in the same gate. Results were expressed as the percentage of B or T lymphocytes that were positive for each marker. An absolute lymphocyte count was performed at the UCLH Haematology Laboratory to determine the absolute numbers of the cell subpopulations.

Statistical analysis.

Wilcoxon's 2-sample signed rank test was used to compare different groups of patients. Wilcoxon's matched pairs signed rank test was used to compare frequencies and total numbers of the different lymphocyte subpopulations before depletion and at repopulation.

RESULTS

Efficient depletion of all subpopulations of peripheral blood B cells by rituximab.

All patients except one (patient 10) showed significant depletion of all subpopulations of B cells for >3 months, with CD19 counts decreasing a mean of 97% to less than 2.0 × 106/liter. All patients whose B cells depleted well improved clinically (achieving at least a 20% improvement according to the American College of Rheumatology criteria) following treatment. The total number of CD19+ cells in the peripheral blood decreased significantly, from a median of 65 × 106/liter (range 6.8–331) at baseline to a median of 0.7 × 106/liter (range 0.2–5.1; P < 0.00006) at 1 month after treatment. The frequency of CD19+ cells in the peripheral blood decreased significantly from a median of 4.56% (range 0.48–19.49) at baseline to a median of 0.04% (range 0.01–0.34; P < 0.00006) at 1 month after treatment. The total number and frequency of CD19+ cells did not change significantly between months 1 and 3 (P = 0.77 and P = 0.98, respectively). Before treatment, the frequency and total number of CD19+ cells in the peripheral blood were not significantly different between the patients who were being treated with rituximab for the first time and the patients who were being retreated (P = 0.6 and P = 0.3, respectively). Depletion was similar in both groups of patients.

All subpopulations of B cells were depleted following treatment with rituximab, including cells with plasma cell precursor phenotype (CD19+,CD20–, CD38+++). During the period of depletion, a very small number of CD19+ cells were detected. Approximately 80% of these cells were IgD– and CD27+/++ (memory B cell or plasma cell precursor phenotype).

Repopulation mainly from naive B cells, with an increased expression of CD38 and CD5.

Repopulation occurred a mean of 8 months after treatment (range 5–13 months). Repopulation occurred mainly with naive B cells (IgD+,CD27–), with increased expression of CD38 and CD5 (Figure 1). Compared with baseline, the frequency of CD19+,IgD+,CD38++ cells increased from a median of 7.23% (range 0.16–17.33%) to a median of 51.10% (range 15.25–79.27%; P < 0.0001), and the frequency of CD19+,CD5+ cells increased from a median of 33.73% (range 20.39–67.10%) to 78.70% (range 46.21–93.68%; P < 0.001), at the time of repopulation (Figures 2A and B). In the patient whose B cells did not deplete as expected and started to repopulate at 9 weeks, the same pattern of increased expression of CD38 and CD5 was observed.

Figure 1.

Expression of different surface phenotype markers on CD19+ cells (B cells) in the lymphocyte gate in peripheral blood cells from rheumatoid arthritis patients A, at baseline, B, during B cell depletion with rituximab, and C, at repopulation of B cells following rituximab treatment. Shown is IgD versus CD38 (gated on CD19+ cells; boxed areas indicate IgD+,CD38++ cells), CD5, and CD27 expression on CD19+ cells in the lymphocyte gate (boxed areas indicate IgD+,CD38++ cells).

Figure 2.

Frequencies and total numbers of B cell subsets in rheumatoid arthritis patients treated with rituximab. A and B, Comparison of the frequencies of specific subsets of CD19+ cells (B cells) in the lymphocyte gate at baseline and at repopulation. Shown are IgD+,CD38++ B cells (A) and CD5+ B cells (B). C and D, Comparison of the frequency and total number of CD27+ B cells (memory B cells) in patients who experienced a relapse at the time of B cell repopulation (early relapse) and after B cell repopulation (late relapse). Shown are the frequency (C) and the total number (D) of CD27+ B cells. Bars in AC show the median.

No patient experienced a relapse of RA before repopulation of the peripheral blood with CD19+ cells. In half of the patients, relapse occurred at the time B cell repopulation was first observed in the peripheral blood or shortly thereafter (within 2 months). In the other half, clinical relapse occurred at a variable time after B cell repopulation of the peripheral blood (mean 8 months [range 5–12 months]; by the end of the study, 3 of the 12 patients in this group had not relapsed). Patients who relapsed at the time of B cell repopulation tended to repopulate with a higher frequency of B cells with a memory phenotype (Figure 2C). Patients who relapsed earlier showed a median of 17.25% CD27+ B cells (range 4.98–31.76%) at repopulation, whereas patients who relapsed later showed a median of 9.07% CD27+ B cells (range 3.3–18.75%) (P = 0.17). All patients who showed memory B cell numbers higher than 3 × 106/liter at repopulation experienced a relapse at the time of B cell return (Figure 2D).

When patients being treated with rituximab for the first time were compared with those being retreated, and when patients receiving concomitant methotrexate were compared with those who were not receiving methotrexate, no differences in either the pattern of clinical relapse or the pattern of B cell repopulation of the peripheral blood were observed. Also, no influence of previous treatments was seen.

CD19+,IgD+,CD38++ cells were identified as being immature naive B cells based on their expression of low levels of CD10 and high levels of CD24 (Figure 3). In the majority of these cells, the level of IgD expression was similar to that found on mature naive B cells (CD19+,IgD+,CD38+ cells) (Figure 3).

Figure 3.

Phenotype of immature B cells in A, a rheumatoid arthritis patient at the time of repopulation following rituximab treatment and B, a normal control subject. Shown is the expression of IgD, CD10, and CD24 versus CD38 (gated on CD19+ cells [B cells]; boxed areas indicate IgD+,CD38++,CD10+,CD24high [immature B cells]).

The total numbers and frequencies of CD19+ cells in the lymphocyte gate returned to the normal range very quickly (within 1 month) in a few patients, but more frequently took several months (Figure 4). The pattern of increased expression of CD38 by naive B cells (IgD+,CD19+ cells) returned to a baseline pattern earlier than the increased expression of CD5 by circulating B cells (CD5+,CD19+ cells). The frequency and total numbers of memory B cells (CD27+,CD19+ cells) remained below baseline values for longer.

Figure 4.

Total numbers of CD19+ cells (B cells) in rheumatoid arthritis patients before and after treatment with rituximab.

Expression of low levels of CD20 by a small proportion of T cells and NK cells and their depletion by rituximab.

A small proportion of T cells (CD3+ cells) and of NK cells (CD56+,CD3– cells) in the peripheral blood expressed low levels of CD20 (Figure 5). Treatment with rituximab depleted both of these subpopulations. CD3+,CD20low cells decreased from a median of 3.15% (range 0.38–20.8%) at baseline to a median of 0.02% (range 0–0.8%; P < 0.00006) at 1 month after treatment. CD56+,CD3–,CD20low cells decreased from a median of 1.71% (range 0.22–14.15%) at baseline to a median of 0.1% (range 0–1.5%; P < 0.00006) at 1 month after treatment. Repopulation of both of these subpopulations occurred before B cell repopulation of the peripheral blood, at a mean of 5 months. Studies in a small number of patients after repopulation showed that CD3+,CD20low cells could express CD4 (more frequently), CD8, or the NK marker CD56 (less frequently) (data not shown).

Figure 5.

Depletion of CD3+ and CD56+,CD3− cells expressing low levels of CD20 in peripheral blood cells from rheumatoid arthritis patients A, at baseline and B, 1 month after treatment with rituximab. Boxed areas at top indicate CD3+,CD20+ cells; those at bottom indicate CD56+,CD3−,CD20+ cells.

Lack of significant change in other T cell populations and NK cells following treatment with rituximab.

The frequencies and absolute numbers of T cells (CD3+) and NK cells (CD56+,CD3–) did not change significantly at 1 month and 3 months after rituximab (data not shown). No significant differences were detected in the different subpopulations of T cells (CD4+ T cells, CD8+ T cells, CD4+ T cells expressing CD25 [CD25+], and CD4+ T cells expressing CD45RA, CD45RO, or both markers), except for the subpopulation that expressed low levels of CD20 described above. The frequency of CD25++,CD4+ T cells decreased 1 month after treatment (P < 0.05), but the total numbers did not, and no significant differences were found at 3 months.

DISCUSSION

In this cohort of RA patients, treatment with rituximab induced a profound depletion of all peripheral blood B cell subpopulations for at least 5 months in all patients except 1 (patient 10). Patient 10 experienced a 97% depletion at 1 month, but at 2 months, repopulation had already begun, with a B cell count that was 90% of baseline. This was the only patient who did not improve following treatment, which suggests that a quantitative threshold for B cell depletion may need to be reached for such a therapy to be effective in RA. The same patient had been treated with half-dose rituximab 40 months before, and it is possible that she had developed human antichimeric antibodies to rituximab.

During the period of maximal depletion, a few B cells were detected in the peripheral blood. It is interesting, nevertheless, that repopulation did not occur from these cells. In all patients, repopulation occurred predominantly with naive B cells, indicating that repopulation was dependent on a resumption of the production of naive B cells in the bone marrow. Residual B cells did not appear to be able to expand and repopulate the peripheral blood.

As we had previously observed in our cohort, clinical relapse did not occur before the return of B cells to the peripheral blood, and patients either relapsed at the time of B cell return or several months to years thereafter (3, 11). Patients who relapsed at the time of B cell repopulation tended to show a higher frequency of memory B cells at repopulation, as compared with patients who relapsed later. In particular, all patients with more than 3 × 106/liter memory B cells at repopulation relapsed at the time of B cell return. This suggests that the depletion of memory B cells in solid lymphoid tissues may have been less complete in patients who relapsed earlier, and that it may be possible to prolong the response to treatment by inducing a more extensive B cell depletion. However, we cannot exclude the possibility that the memory B cells circulating at the time of B cell repopulation were “young” B cells that had already differentiated into memory cells and were not related to residual memory B cells.

Our study shows that changes resembling a second round of ontogeny occurred upon B cell repopulation after depletion with rituximab, similar to that described upon repopulation after bone marrow transplantation (13–15). Specifically, the pattern of qualitative B cell reconstitution with a high proportion of naive B cells expressing IgD and overexpressing CD38 and CD5 and a decreased frequency of memory B cells is characteristically seen after bone marrow transplantation in patients without chronic graft-versus-host disease, and it is similar to what is found in cord blood and in very young children.

We believe that the IgD+,CD38high B cells observed were probably young naive B cells exiting the bone marrow, rather than germinal center founder cells as they have previously been classified (16–18). These cells expressed low levels of CD10 but high levels of CD24. CD24 expression decreases when cells enter the germinal center and is absent in the majority of germinal center B cells (19, 20). If circulating IgD+,CD38high B cells were germinal center founder cells, we would expect to find lower, or at least normal, but not higher, levels of CD24 expression compared with mature naive circulating B cells. This pattern of increased expression of CD38 by IgD+ B cells was seen even in the patient whose B cells did not deplete well and could be wrongly interpreted as an increase in circulating germinal center founder cells (9).

Circulating B cells with this phenotype have recently been described in normal adults, with functional studies confirming their immaturity (21, 22). It is possible that this is actually the phenotype of all “young” B cells when they exit the bone marrow. In normal adults, these cells are probably not easily detected since they are diluted in the much bigger naive mature recirculating B cell population that is IgD+ but CD38low. This would imply that B cells need to mature in solid lymphoid tissues other than the bone marrow before becoming naive mature recirculating B cells, a phenomenon previously not known to occur in humans but extensively described in mice (23).

In our study, repopulation was associated with increased expression of CD5 by circulating B cells. Interestingly, this increase seems to be in some way independent of the increased expression of CD38 on naive B cells. After bone marrow transplantation, not all patients who show increased numbers of IgD+,CD38++ B cells also show increased expression of CD5, and in our patients, the return to baseline frequencies of IgD+,CD38++ preceded the return of CD5 expression to baseline levels (13). These changes in the expression of CD5 by circulating B cells associated with B cell depletion suggest that unlike in mice, CD5+ B cells in humans are not a defined population with specific phenotypical and functional characteristics (21, 24).

Expression of the CD20 antigen is not exclusive to B cells. We found that there was a small population of circulating T cells and NK cells that expressed low levels of CD20. These cells disappeared from the circulation for a mean of 5 months following treatment with rituximab. Previous reports had described the presence of T cells expressing low levels of CD20 in the peripheral blood (25–27). We did not observe significant changes in the total number or frequency of the other peripheral blood T cell populations studied, except for a transitory decrease in the frequency of CD4+,CD25++ T cells (regulatory T cells) at 1 month. A recent report by Sfikakis et al (28) described a decrease in the expression of activation markers (CD40 ligand, CD69, and HLA–DR) on circulating CD4+ T cells in patients with lupus nephritis who responded to rituximab.

In conclusion, we found that rituximab treatment induces an almost complete depletion of all peripheral blood B cell populations in patients with RA. Failure to achieve 97% depletion of circulating B cells for at least 3 months in 1 patient was associated with a lack of response to treatment. B cell repopulation of the peripheral blood was dependent upon the formation of naive B cells, rather than the expansion of memory B cells. At repopulation, IgD+,CD38++ B cells were abundant, and the pattern of CD10 and CD24 expression on these cells suggested that they were immature B cells and not germinal center founder cells. Patients who experienced an earlier disease relapse, at the time of B cell return, tended to show a higher number of circulating memory B cells at repopulation, as compared with patients whose relapse occurred later. This suggests that less extensive B cell depletion in solid tissues may be associated with an earlier relapse of RA.

Ancillary