Efficacy of Anti-CD20 Treatment in Patients with Rheumatoid Arthritis Resistant to a Combination of Methotrexate/Anti-TNF Therapy

Authors


M. Bokarewa, Department of Rheumatology and Inflammation Research, Gulhedsgatan 10, S-413 46 Göteborg, Sweden.
E-mail: maria.bokarewa@rheuma.gu.se

Abstract

Rheumatoid arthritis (RA) is characterized by chronic joint inflammation and destruction. B cells play important role in modulating immune responses in RA. In the present study we assessed the impact of the B cell targeting as a third line treatment option. Forty-six patients with established erosive RA non-responding to combination treatment with DMARDs and TNF-α inhibitors were treated with anti-CD20 antibodies (rituximab). Rituximab was given intravenously once weekly on four occasions. All patients continued with the previous DMARD. Patients were followed by DAS28, levels of circulating B cells, frequency of immunoglobulin-producing cells, immunoglobulins, and rheumatoid factor levels during the period of 12–58 months. Clinical improvement was achieved in 34 of 46 patients (73%) supported by a significant reduction in DAS28 (from 6.04 to 4.64, P < 0.001). Infusion of rituximab resulted in the elimination of circulating B cells in all but one patient. Within 12 months follow-up, B cells returned to circulation in 86% of patients. Fifty-three percent of the patients were successfully retreated with rituximab or re-started with anti-TNF-α treatment. Of the 11 non-responders, five were retreated with anti-CD20 within 2 months, four of them with success, four patients received TNF-α inhibitors, the remaining two patients received an additional DMARD. Most of the RA patients resistant to TNF-α inhibitors may be effectively treated with anti-CD20 antibodies. The treatment is well tolerated and may be used repeatedly in the same patient and potentially increase sensitivity to previously inefficient treatment modalities.

Introduction

Rheumatoid arthritis (RA) is characterized by chronic synovial inflammation leading to the destruction of cartilage and subchondral bone. Even though several studies have shown that T cells, B cells, and components of the innate immune system all play a role in the disease, the exact pathogenic mechanisms are far from fully understood [1–3]. The synovial inflammation is largely tumor necrosis factor- α (TNF-α) dependent and treatment with TNF-α inhibitors has proved to be efficient in approximately two-thirds of RA patients. However, in a substantial proportion of RA patients, anti-TNF-α treatment does not lead to a satisfactory clinical improvement or fail to sustain a clinical response over time [4–6]. Seemingly the contribution of different components of the immune system varies among different RA patients and there is an urgent need to find new treatment strategies for patients with anti-TNF-α-resistant RA.

B cells may play a role in autoimmune diseases including RA in several different ways, not only as the source of autoantibodies, but also as antigen-presenting cells and as a source of disease-promoting cytokines [7, 8]. Rituximab, a chimeric monoclonal antibody which selectively depletes human CD20-positive B cells, is an efficient treatment in non-Hodgkin’s lymphoma [9]. Infusion with rituximab leads to depletion of CD20-positive B cells in the blood remaining low for 6–12 months after treatment [10, 11]. Stem cells and plasma cells are spared, enabling generation of naïve B cells and maintaining serum levels of immunoglobulins [11, 12]. Even though the effect of rituximab on circulating immunoglobulin levels has been studied, the impact of rituximab on circulating antibody-producing B cells is unknown. Recent clinical trials of rituximab in RA have demonstrated its significant efficacy in patients with DMARD and/or anti-TNF-α-resistant disease [13–16]. It was recently shown that glucocorticoids had no impact on the ACR20 response 24 weeks after rituximab treatment [14].

In the present study we evaluated the efficacy of rituximab treatment and long-time follow-up of 48 RA patients with established, erosive disease refractory to combined treatment with DMARDs and TNF-α inhibitors. We also show that rituximab may be repeatedly used in the same patient and potentially increase sensitivity to previously inefficient treatment modalities such as TNF-α inhibitors.

Material and methods

Patients.  Forty-eight patients (38 women and 10 men) with established RA (mean disease duration 10 years, range 2–31 years) were treated with monoclonal anti-CD-20 antibodies (rituximab, MabTera; F. Hoffman-La Roche Ltd, Basel, Switzerland) at the Rheumatology Clinics, Sahlgrenska University Hospital, Göteborg, Sweden, during the period between February 2002 and December 2005. Clinical and demographic characteristics of the patients and their previous treatment modalities are presented in Table 1. Previous treatment modalities included different DMARDs (mean 4.2, range 3–8) used as monotherapy and in combinations with parenteral infusions of corticosteroids in high doses, different biologic drugs including anti-thymocyte globulin treatment (n = 3), anti-TNF-α antibodies (infuximab 34, adalimumab 13, etanercept 18), IL-1 receptor antagonist (n = 3). Most of the patients (64%) had experienced more than one biologic drug prior to rituximab treatment. For the treatment with rituximab, RA patients failing on a combination of DMARD and anti-TNF-α antibodies, thus having high activity of RA, were selected. The biologic drugs were discontinued at least 4 weeks before rituximab treatment. At the start of treatment with anti-CD20 antibodies all the patients had a stable treatment with NSAID and DMARD preparations. Thirty-seven patients (77%) used methotrexate (MTX) (mean 17.5 mg/week, range 10–25 mg/week), while the remaining patients had parenteral cyclophosphamide (n = 6), leflunomide (n = 1), azathioprine (n = 1) and mycophenolate mofetil (n = 1). Twenty of forty-eight patients had low doses of oral corticosteroids (range 5–15, mean 6.3 mg). All the patients were informed about the aim of treatment, potential complications that could appear during rituximab infusion and after it. All patients gave informed consent before participation in the study.

Table 1.   Clinical and demographic characteristics of patients with rheumatoid arthritis.
 RA patients (n = 48)
Age, years (range)58 ± 11 (28–76)
Sex, male/female10/38
Radiological data, erosive/non-erosive48/0
Rheumatoid factor, +/−40/8
Duration of the disease, years ± SD10 ± 7
Observation after rituximab treatment, months ± SD (range)28 ± 13 (12–58)

Rituximab treatment regimen.  Rituximab was given as four consecutive infusions once weekly at a dose of 375 mg/m2. One hour before the infusion, patients received paracetamol (1 g orally) and anti-histamine drug (Tavegyl, 2 mg parenterally). No corticosteroids were provided at the time of rituximab infusions. The efficacy of the given treatment was assessed clinically by the count of swollen and tender joints, and the changes in inflammation intensity as measured by erythrocyte sedimentation rate (ESR), C-reactive protein, blood leukocyte (WBC) and platelet counts, hemoglobin. DAS28, a composite measure based on 28 tender and swollen joint counts, and ESR, was determined at baseline and during follow-up after 1–3, 6, 12 and 24 months. The response to rituximab treatment was evaluated on the basis of EULAR response criteria [17]. The reduction in DAS28 equal or above 1.2 during the first 3 months following rituximab treatment was set as the cut-off level for clinical response.

Assessment of B cell numbers and function.  Peripheral blood samples were obtained before the start of rituximab infusion and 1–3, 6, 12 and 24 months after the first infusion. The influence of rituximab on B cells was assessed by the determination of the number of circulating B cells (defined as CD19 and CD20-positive) using flow cytometry, total levels of immunoglobulins of IgG, IgM, IgA classes (using nephelometry), the number of immunoglobulin-producing cells in the peripheral blood (using ELISPOT, providing the number of spot forming cells/106 mononuclear cells), changes in the levels of rheumatoid factor of IgG, IgM and IgA classes (using an ELISA).

Statistical evaluation.  The clinical parameters of disease activity such as the number of tender and swollen joints as well as DAS28, and the laboratory data such as ESR, WBC count and hemoglobin levels following treatment with rituximab were expressed as mean ± SD. A reduction in DAS28 ≥ 1.2 indicated patient response to rituximab treatment. For further comparison, patients were stratified into responders and non-responders to rituximab treatment. Differences in clinical and laboratory parameters of RA activity as well as levels of RF, total immunoglobulin levels and the amount of immunoglobulin producing cells between the groups were compared by a paired t-test and Mann–Whitney U-test. For the evaluation of possible influence of MTX treatment on the duration of rituximab effect, patients were stratified according to DMARD treatment into MTX-treated versus untreated. For all the statistical evaluations of the results, P-values below 0.05 were considered to indicate significant differences.

Results

Safety of anti-CD20 treatment

Treatment with rituximab was well tolerated by the majority of the RA patients. No allergic reactions, as cutaneous rash or anaphylactic reactions, requiring corticosteroids were observed following infusions of rituximab. All patients but one received infusions of rituximab at four occasions. The mean total dose of rituximab was 2550 mg (range 1000–3200 mg). One patient reported severe headache and stomach pain the day after rituximab infusion, and the treatment with rituximab was discontinued after the second infusion. One patient reported nausea directly after rituximab infusions. Other complications that occurred 3 months after rituximab treatment included pneumonia, requiring hospitalization, in one patient. Two patients died of myocardial infarction, one patient died within the first month after rituximab treatment, another patient died of myocardial infarction 13 months after rituximab treatment. Both patients had a medical history of chronic myocardial ischemia based on atherosclerosis and aortic valve stenosis, respectively. None of these patients exhibited signs of myocardial ischemia or arrythmias during rituximab infusions. One patient received breast cancer diagnosis 2 months after the final rituximab infusion and underwent surgery and radiation therapy. These patients were discharged from further evaluation of treatment effect.

Impact of anti-CD20 treatment on clinical manifestations and activity of RA

Clinical characteristics of RA patients before the rituximab treatment are presented in Table 1. Before the treatment all patients displayed highly active RA having mean DAS28 6.1 (range 4.04–7.83). DAS28 showed a significant decrease following rituximab treatment (Fig. 1A–C). The reduction in disease activity was most pronounced during the first 3 month following treatment and was sustained at this level after 6 months (mean 6.1 to 4.7 and 4.6, respectively). The reduction in DAS28 was accompanied with changes in other inflammatory parameters such as a decrease in ESR and C-reactive protein (Fig. 2A,B), and blood platelet count (Fig. 3C) as well as a rise in hemoglobin levels (Fig. 3A).

Figure 1.

 Dynamics of clinical parameters of RA activity following rituximab treatment in the responders (n = 34) and non-responders (n = 12). Values given are means. Statistical analysis of changes of the given parameters with time was assessed by the paired t-test. Comparison between the treatment responders and non-responders was performed by Mann–Whitney statistics. P-values below 0.05 were considered to indicate significant differences. Significant differences between the groups of responders and non-responders are indicates as asterisk (**). (A) Disease activity score (DAS28). (B) Number of swollen joints. (C) Number of tender joints.

Figure 2.

 Dynamics of inflammation markers in RA patients following rituximab treatment. Values are given as means. Statistical analysis of changes of the given parameters with time was assessed by the paired t-test. Comparison between the treatment responders and non-responders was performed by Mann–Whitney statistics. P-values below 0.05 were considered to indicate significant differences. Significant differences between the groups of responders and non-responders are indicates as asterisk (**). (A) Erythrocyte sedimentation rate (ESR). (B) C-reactive protein.

Figure 3.

 Dynamics of blood cell counts in RA patients following rituximab treatment. Values are given as means. Statistical analysis of changes of the given parameters with time was assessed by the paired t-test. Comparison between the treatment responders and non-responders was performed by Mann–Whitney statistics. P-values below 0.05 were considered to indicate significant differences. Significant differences between the groups of responders and non-responders are indicated as asterisk (**). (A) Hemoglobin levels. (B) White blood cell count, WBC. (C) Platelet count.

The clinical outcome of rituximab treatment was assessed using the EULAR response criteria [17] where a decrease in DAS28 ≥ 1.2 distinguished patients who clinically responded to rituximab treatment. Three months after the treatment, 34 of 46 RA patients (73%) had a reduction in DAS28 ≥ 1.2 (mean 2.0). Among them, eight (23%) patients had a decrease in DAS28 < 3.2 (changes in mean from 5.74 ± 0.6 to 2.9 ± 0.23, P = 0.022) and were considered as good responders. Twelve of 46 RA patients (27%) had a decrease in DAS28 less than 1.2 (mean 0.9, range 0.24–1.1), these patients were considered as non-responders. When comparing clinical parameters of the RA patients who responded and not responded to rituximab treatment, significant difference was observed in the number of swollen joints 3 and 6 months after treatment (Fig. 1B). In contrast, inflammatory markers such as ESR, C-reactive protein and blood cell counts did not differ significantly (Figs. 2A,B and 3A–C).

Six of the twelve non-responders were re-treated with the same dose of rituximab within 3 months after the first treatment (Table 2). Four of them had a good effect (mean DAS28 reduction 1.8). Other three non-responders started with anti-TNF-α treatment (adalimumab). Two more patients received an addition of cyclosporin A, and the remaining non-responder patient underwent autologous stem cell transplantation 8 months after rituximab treatment.

Table 2.   Treatment modalities used in RA patients prior to and following rituximab treatment.
 At start (n = 48)12 months after (n = 46)*24 months after (n = 25)
  1. Two patients died within 12 months after rituximab treatment. MTX, methotrexate; ATG, anti-thymocyte globulin; SCT, autologous stem cell transplantation

MTX353717
Cyclophosphamide621
Azathioprine110
Leflunomide120
Mycophenolate mofetil101
Chlorambucil041
MTX + cyclosporin2375
TNF-α inhibitors
 Infliximab 26
 Adalimumab 35
 Etanercept 11
IL-1R antagonist 10
ATG 00
SCT 10
Rituximab re-treatment 76

Long-time clinical outcome of rituximab treatment

The clinical improvement in RA, as assessed by the number of swollen and tender joints and followed by DAS28, was sustained after 6 month of rituximab treatment in all of the 34 early responders (Fig. 1A–C). Between 6 and 12 months after the treatment, 9 of 34 patients (26%) relapsed demonstrating an increase in DAS28. Those patients who were retreated with rituximab were excluded from further statistical analyses.

Of 26 patients with a follow-up period of more than 12 months, one died of myocardial infarction 13 months after rituximab treatment. Seven of the remaining twenty-five patients (35%) had sustained improvement in RA disease activity lasting over 24 months. Another 18 patients relapsed within 24 months following rituximab treatment. The relapse of RA activity in five patients was associated with discontinuation of MTX treatment. During the 24-month period after rituximab treatment, 10 patients underwent major orthopedic surgery (five hip prostheses, two knee prostheses, three ankles, one cervical stabilizing surgery) that could have triggered the relapses. At the relapse, six patients were re-treated with rituximab and eight patients restarted with TNF-α inhibitors, while the remaining four patients continued conventional DMARD treatment. Six patients received re-treatment with rituximab within 24 months (range 8–22 months after the first treatment). Re-treatment doses and infusion regimen were identical to the primary treatment. No adverse effects were registered at re-treatment. Following re-treatment all the patients showed a decrease in DAS28, associated with clinical improvement, and decrease in laboratory inflammation markers such as ESR and C-reactive protein. Four of six patients had a decrease in DAS28 > 1.2 (range 1.2–1.8). Among eight patients who had a new start of anti-TNF-α treatment, six had clinical effect despite a previous history of therapy failure with TNF-α inhibitors followed by a decrease in DAS28 from 5.55 ± 0.72 to 3.86 ± 0.27 (P = 0.033).

Effect of treatment with anti-CD20 antibodies on the function of B cells in RA patients

Before rituximab treatment, circulating B cells, defined as CD19 and CD20-positive lymphocytes by flow cytometry analysis, were detected in circulation of 39 of 46 patients (mean level 8%, range 2–30%). The remaining six patients, three responders and three non-responders, had no detectable B cells in the peripheral blood. At 1–3 months after rituximab treatment B cells could not be detected in peripheral blood by flow cytometry in any but one patient. The return of B cells (>1% of lymphocytes) to the circulation occurred in 37% of cases by 6 months and in 86% by 12 months after rituximab treatment. No difference between responders and non-responders was observed regarding the amount of B cells in circulation (data not shown). The re-appearance of B cells to peripheral blood was not related to a relapse of RA activity. However, B cells were present in circulation of all the patients at the time of relapse.

Forty-three of forty-six patients had Ig-producing B cells in circulation prior to rituximab treatment. At baseline, the responders (DAS28 reduction >1.2) had significantly lower levels of IgM-producing cells in circulation compared with non-responders (283 ± 76 spot-forming cells [sfc] versus 704 ± 294 sfc, respectively, P = 0.047; Fig. 4B), while the amount of IgG-producing cells was higher in responders (904 ± 438 sfc versus 484 ± 184 sfc, respectively, not significant; Fig. 4A). The levels of IgA-producing cells in circulation were similar (512 ± 140 sfc versus 466 ± 152 sfc). Rituximab treatment and elimination of B cells from circulation was associated with a significant reduction in Ig-producing cells within 3 months after the treatment as determined by ELISPOT (Fig. 4A–C). The reduction was more pronounced for IgA and IgM-producing cells (88% and 78% reduction, respectively) and 45% for the IgG-producing cells. The reduction in Ig-producing cells was similar between the responders and non-responders.

Figure 4.

 Changes in the number of immunoglobulin-producing cells in RA patients following rituximab treatment. Values are given as means. Statistical analysis of changes of the given parameters with time was assessed by the paired t-test. Comparison between the treatment responders and non-responders was performed by Mann–Whitney statistics. P-values below 0.05 were considered to indicate significant differences. Significant differences between the groups of responders and non-responders are indicated as asterisk (**). (A) Number of cells producing IgG. (B) Number of cells producing IgM. (C) Number of cells producing IgA.

Following rituximab treatment the total levels of serum Ig showed a significant decrease (Fig. 5A–C). This decrease was significant regarding IgG and IgM classes of immunoglobulins, while circulating IgA levels remained unchanged. This decrease in IgG and IgM levels was sustained by 12 months after treatment despite the return of B cells into circulation. The decrease of circulating Ig was similar both in treatment responders and non-responders.

Figure 5.

 Dynamics of serum immunoglobulin levels in RA patients following rituximab treatment. Values are given as means. Statistical analysis of changes of the given parameters with time was assessed by the paired t-test. Comparison between the treatment responders and non-responders was performed by Mann–Whitney statistics. P-values below 0.05 were considered to indicate significant differences. Significant differences between the groups of responders and non-responders are indicated as asterisk (**). (A) IgG. (B) IgM. (C) IgA.

Forty of forty-six patients were RF-positive before rituximab treatment. Following rituximab infusions RF decreased to undetectable levels in 2 of 46 patients. Rituximab treatment induced a transient decrease in the levels of RF of IgG (60–39 U/ml by month 1–3, Fig. 6A) and IgM (103–83 U/ml by month 1–3, Fig. 6B) isotypes, while for the RF of IgA the isotype did not change following treatment (Fig. 6C). Although responders had higher levels of RF before rituximab treatment, no significant differences between the responders and non-responders were observed.

Figure 6.

 Dynamics of serum rheumatoid factor levels in patients with RA following rituximab treatment. Values are given as means. Statistical analysis of changes of the given parameters with time was assessed by the paired t-test. Comparison between the treatment responders and non-responders was performed by Mann–Whitney statistics. P-values below 0.05 were considered to indicate significant differences. Significant differences between the groups of responders and non-responders are indicated as asterisk (**). (A) RF-IgG. (B) RF-IgM. (C) RF-IgA.

Discussion

In the present study we retrospectively assessed the effect of B-cell depletion using anti-CD20 antibodies (rituximab, Mabthera) on the disease activity in 48 RA patients. We observed that rituximab efficiently reduced the activity of RA in more than 70% of the patients treated. These results are in agreement with previous reports of rituximab treatment in RA [13–15]. Several aspects differ in our study from those previously reported. First, our patient group consisted of RA patients refractory to the available treatment modalities including high-dose MTX in combination with biologic drugs (anti-TNF-α, IL-1 receptor antagonist, anti-thymocyte globulin). Most of the patients (64%) have tested more than two biologic drugs before initiation of rituximab treatment. Although no washout period was kept between anti-TNF-α treatment and rituximab infusions, the activity of RA in the patient group prior to rituximab infusion was high (mean DAS28 6.1 ± 0.8), indicating severe cases of therapy-refractory RA included in the study. Second, the treatment regimen used in our study consisted of four consecutive infusions and the dose of rituximab was adjusted individually. The reason for this schedule is that at the time of initiation of rituximab therapy (April 2002), results of the DANCER study [14] were not available. However, the total dose of rituximab provided differed only by 25% from the present recommendations for the RA treatment, suggesting that the large proportion of patients who responded to rituximab infusion is not related to the somewhat higher dose of rituximab used. It is possible that the higher number of rituximab infusions and the dosing interval of 1 week are favorable for more efficient depletion of B cells providing a longer relapse-free period in a significant number (23%) of our patients. The third aspect is the long follow-up time in our study, which in all the patients was at least 12 months. We observed that the effect of rituximab treatment was time-restricted, and a relapse in RA activity was seen in most of the patients despite continuous treatment with conventional DMARDs. The requirement of re-treatment with rituximab is thus obvious; however, the appropriate time point requires further studies. At present no results of regular re-treatment with rituximab have been reported. In our study, no restrictions were made for follow-up treatment and patients were free to choose between re-treatment with rituximab or a re-start with anti-TNF-α preparations at relapse of RA following the first rituximab treatment. Some patients clearly preferred re-treatment with rituximab, while the others switched back to TNF-α inhibitors. This suggests that (i) rituximab was equally effective for re-treatment and could be used even in non-responders; (ii) in some cases, a re-start with a previously inefficient TNF-α inhibitor could significantly decrease the activity of RA. The efficacy of re-treatment with rituximab and its advantages compared to a change between the anti-TNF-α preparations has been recently reported [18]. The important clinical question to be answered in the future is the strategy for the use of biologic treatment for RA patients resistant to conventional DMARDs. Should one continue using TNF-α inhibitors as the first choice? Does the use of rituximab give advantages with respect to progression of disease activity, joint destructions or the socio-economical situation of the RA patients?

To identify clinical and laboratory parameters helpful in predicting potential responders to rituximab treatment and to distinguish the early signs of RA relapse, we performed the analysis of rituximab responders versus non-responders. Responders and non-responders were similar with respect to circulating immunoglobulin levels, levels of RF and total number of circulating B cells before the treatment. The effect of rituximab regarding the changes in the B-cell-dependent parameters after the treatment such as the B-cell free period, levels of immunoglobulins and changes in RF levels were similar between the responders and non-responders. These data are in agreement with reports of Cambridge et al. [11] and Leandro et al. [12]. Interestingly, we observed that high levels of circulating B cells producing IgM before rituximab treatment were associated with deficient treatment response. In contrast, numbers of IgM-producing B cells decreased to low levels following rituximab irrespective of the clinical outcome. Scanty information is available about the tissue-specific depletion of B cells from bone marrow or from synovia of RA patients. Neither antigen-presenting properties of remaining B cells/macrophages have been a subject of investigation following anti-CD20 treatment. These parameters may be important for the identification of potential rituximab responders prior to treatment.

In conclusion, we observed that B-cell depletion with anti-CD20 antibodies (rituximab) may be successfully used for the treatment of RA in most of the patients refractory to conventional DMARDs/anti-TNF-α treatment. Treatment with rituximab was well tolerated and may be used repeatedly in the same patient and can potentially increase the sensitivity to previously inefficient treatment modalities, such as MTX/TNF-α inhibitors.

Acknowledgment

This work was supported by the Göteborg Medical Society, the Swedish Association against Rheumatism, King Gustaf V’s Foundation, the Swedish Medical Research Council, Nanna Svartz Foundation, Roche AB and the University of Göteborg.

Ancillary