Article Text

This article has a correction. Please see:


Treatment of refractory SLE with rituximab plus cyclophosphamide: clinical effects, serological changes, and predictors of response
  1. T Jónsdóttir,
  2. I Gunnarsson,
  3. A Risselada,
  4. E W Henriksson,
  5. L Klareskog,
  6. R F van Vollenhoven
  1. 1
    Department of Rheumatology, Karolinska University Hospital, Stockholm, Sweden
  1. Dr Ronald F van Vollenhoven, Department of Rheumatology, D2:01, Karolinska University Hospital, Solna, S-171 76 Stockholm, Sweden; ronald.van.vollenhoven{at}


Objective: To evaluate efficacy, serological responses, and predictors of response in patients with severe and refractory systemic lupus erythematosus (SLE) treated with rituximab plus cyclophosphamide.

Methods: 16 patients entered a treatment protocol using rituximab plus cyclophosphamide. Disease activity was assessed by the SLE disease activity index (SLEDAI) and by the British Isles Lupus Assessment Group (BILAG) index.

Results: At six months follow up, mean SLEDAI values decreased significantly from (mean (SD)) 12.1 (2.2) to 4.7 (1.1). Clinical improvement (50% reduction in SLEDAI) occurred in all but three patients. All but one patient responded according to BILAG. Remission defined as SLEDAI <3 was achieved in nine of 16 patients. Isotype analysis of anti-dsDNA antibodies revealed preferential decreases of IgG and IgA, but not IgM. Higher absolute numbers of CD19+ cells at baseline were correlated with shorter depletion time (r = −0.6).

Conclusions: The majority of patients improved following rituximab plus cyclophosphamide. The differential downregulation of anti-DNA of the IgG and IgA but not the IgM isotypes supports the hypothesis that cells producing pathogenic autoantibodies are preferentially targeted by the treatment. The fact that greater absolute numbers of CD19+ cells at baseline predict a less impressive clinical and serological response suggests that more flexible dosing could be advantageous.

Statistics from

Systemic lupus erythematosus (SLE) is characterised by multiple immunological abnormalities, of which B cell hyper-reactivity may be of central importance.1 Polyclonal B cell hyper-reactivity may lead to hypergammaglobulinaemia, autoantibody production, and immune complex formation resulting in tissue injury and organ damage.

Despite immunosuppressive treatment, mortality and morbidity in patients with SLE are still of great concern. Thus the development of more targeted and effective treatments is important. An attractive step in this direction could be the use of a B cell specific agent.

B cell depleting therapy with the monoclonal antibody rituximab, directed against the B cell specific antigen CD20, has in recent years shown encouraging results in patients with SLE.25 In this paper we provide data on clinical efficacy, serological changes, and predictors of clinical response in patients with severe SLE treated with a combination of rituximab and cyclophosphamide.



Sixteen female SLE patients with a mean age of 37 years (range 19 to 56) and a mean disease duration of 10.5 years (1 to 24) were included in a clinical trial with rituximab and cyclophosphamide. All patients fulfilled the American College of Rheumatology (ACR) criteria for SLE.6 Global disease activity was measured by the SLE disease activity index (SLEDAI).7 Assessment on an individual organ system level was undertaken by the British Isles Lupus Assessment Group (BILAG) system.8 The patients were selected when they had a flare classified as one BILAG A or two BILAG B in any organ system and when not responding to conventional immunosuppressive treatment including cyclophosphamide. In cases when the actual flare was not treated with cyclophosphamide before rituximab, cyclophosphamide had been considered inappropriate owing to earlier negative reactions or a high cumulative dose (table 1).

Table 1 Patient characteristics at baseline and six months after treatment

The regional ethics committee approved the study, and all subjects gave their written informed consent before entering.

Treatment protocol and assessment

The study protocol has been described previously in detail.10 In brief, patients were treated with infusions of rituximab 375 mg/m2 once weekly for four weeks. The first and last infusion was combined with cyclophosphamide 0.5 g/m2; glucocorticoid infusions (methylprednisolone 250 mg) were given in combination with cyclophosphamide, followed by oral glucocorticoids (0.5 mg/kg) during the four weeks of treatment, with a rapid taper down to the lowest possible dose. No other immunosuppressive treatment was allowed.

Treatment efficacy was evaluated every one to two months on the basis of improvement in both clinical and laboratory indices of active disease. For evaluation of clinical relapses, an extended longitudinal follow up for up to four years was carried out. The median follow up time was 27 months (range 6 to 59).

Response criteria

Clinical response to treatment was analysed in two ways. First, a global clinical response was defined as a reduction of 50% or more in the SLEDAI score from baseline (SLEDAI-50). Second, by using the BILAG definitions, we defined a reduction in the score of any organ system from A to B, or from B to C, to represent a clinically meaningful improvement.

Remission and relapse criteria

Remission was defined as a SLEDAI score <3, or as the absence of any BILAG A or B. A relapse was defined as an occurrence of a new BILAG A or at least two new BILAG B in any organ system.

Serological and laboratory parameters

Assessment of serum IgG anti-dsDNA antibodies were done by fluorescent enzyme immunoassay (FEIA) (Pharmacia, Uppsala Sweden), normal range < 15 IU/ml. Titres of IgM and IgA anti-dsDNA were assessed by a commercial enzyme linked immunosorbent assay (ELISA) (Orgentec, Germany), normal range <20 U/ml. Total serum IgG, IgM, and IgA were measured by ELISA.

Numbers of CD19+ cells were measured by flow cytometry. Levels of <0.01×109 cells/l were defined as undetectable (B cell depletion).


Statistical analysis was done using Stat view 5.0.1 software (SAS Corporation, Cary, North Carolina, USA). Correlations were evaluated using the Student t test. Probability (p) values less than 0.05 were considered significant.


Clinical response

At six months after treatment, the mean (SD) SLEDAI values decreased from 12.1 (2.2) to 4.7 (1.1), and 13 of 16 patients achieved a SLEDAI-50 clinical response. Median time to response was three months (range 1 to 12).

At baseline all patients had at least one BILAG A or two BILAG B for a disease manifestation in any organ system. Assessment at six months revealed that, of a total of 20 organ systems with a BILAG A at baseline, five improved to a B, 11 to a C, and four to a D. Of the organ systems with a BILAG B at baseline, all but three improved (table 1).

Despite higher glucocorticoid (GC) doses during the treatment period the dosage after six months was significantly reduced compared with baseline (p = 0.01). The treatment was generally well tolerated and no severe adverse events were noticed.

Clinical remission and relapse

A SLEDAI <3 was achieved in nine of 16 patients after a mean of three months (range 0 to 6). A remission defined as the absence of any BILAG A or B was also noted in nine of 16 patients after six months.

A relapse occurred in seven patients after a median of 18 months (range 10 to 40). Relapse was always preceded by B cell reconstitution. The patients who relapsed had a significantly shorter depletion time than those who did not (p = 0.05).

Serological changes

Following rituximab treatment, the mean total serum IgM concentrations decreased significantly while serum IgG and IgA values were only slightly and non-significantly decreased (fig 1).

Figure 1 (A, B) A significant reduction in IgG anti-dsDNA was seen at all time points of follow up (A) (p<0.01). No changes were seen in the total serum IgG (B). Comparing the reduction of IgG anti-dsDNA and total serum IgG levels at six months, a significant difference was observed (p = 0.0006). (C, D) IgA anti-dsDNA decreased from baseline to six months, though not significantly (NS) (C). No change in serum IgA was noted (D). A significant difference between changes of IgA anti-dsDNA and total serum IgA was noted after six months (p = 0.01). (E, F) A decrease in IgM anti-dsDNA was observed but was not statistically significant at six months (E). For total serum IgM a significant reduction was seen at one, two, and four months of follow up compared with baseline (p<0.05) (F). No difference was noted comparing changes in IgM anti-dsDNA and total serum IgM levels (NS).

Twelve of 16 patients were IgG anti-dsDNA positive at baseline with a mean titre of 206 IU/ml (range 35 to 600). In these patients, the levels of anti-dsDNA were significantly reduced at all time points of follow up (p⩽0.01). Titres of IgA and IgM anti-dsDNA decreased, but not significantly. After six months of follow up, IgM anti-dsDNA showed a reduction of the same magnitude as overall serum IgM, but anti-dsDNA of the IgG and IgA isotypes were more reduced from baseline than their respective classes of serum immunoglobulins (p = 0.0006 and p = 0.01, respectively) (fig 1).

B cell repopulation occurred at a median of seven months (range 2 to 12). One patient has not yet achieved B cell repopulation four years after therapy.

Predictors of response

The absolute number of CD19+ lymphocytes in the peripheral blood at baseline was found to be significantly higher in the patients who had not achieved a SLEDAI <3 at six months than in those who had (0.096 (0.037) vs 0.034 (0.008) ×109 cells/l; p = 0.05 (fig 2A)). Likewise, larger numbers of CD19+ B lymphocytes at baseline were correlated with shorter depletion time (r = −0.7, p<0.0001; fig 2B) and a shorter time to reach SLEDAI-50 (r = −0.4, p = 0.0006).

Figure 2 (A) Significantly higher absolute numbers of CD19+ lymphocytes at baseline were observed in patients not achieving a SLEDAI score of <3 (remission) at six months (p = 0.05). (B) Larger numbers of CD19+ cells at baseline correlate with shorter depletion time (r = −0.7, p<0.0001).


A clinical response following treatment with rituximab and cyclophosphamide was observed in a majority of the patients with severe SLE which had been refractory to conventional treatment. Clinical remission was noted in at least half the patients after six months. Relapse, occurring in almost half the patients, was consistently preceded by B cell repopulation. The interval between B cell repopulation and clinical relapse was highly variable and in some patients persistence for longer periods after repopulation occurred without the experience of relapse. Therefore, clinical benefits may well persist despite repopulation. The explanation for this finding might be the normalisation of the disturbances in peripheral B lymphocyte homeostasis characteristic of active SLE.2

The serological changes seen in our patients revealed some new findings. The changes in anti-dsDNA of the IgA isotype was here found to follow the same pattern as for IgG anti-dsDNA—that is, a decrease that was more profound than the change in overall serum IgA/IgG. In contrast, we found that IgM anti-dsDNA decreased only to the same degree as overall serum IgM. Earlier studies have demonstrated the association of IgA anti-dsDNA with indices of disease activity, such as a raised ESR and low complement levels, as well as with kidney and joint abnormalities,11 12 suggesting a pathogenic role. The role of IgM anti-dsDNA in the pathogenesis of SLE is somewhat obscure, however. Our finding that the IgM anti-dsDNA isotype showed a reduction only to the same magnitude as overall serum IgM is in agreement with the hypothesis that IgM anti-dsDNA may be less pathogenic and possibly produced by less pathological plasma cells/plasmacytoid cells.

Hypothetically the pathogenic anti-dsDNA antibodies of the IgG and IgA isotype may be produced by CD20+ plasmacytoid cells, whereas IgM anti-DNA could be produced by the same type of long lived plasma cells as are responsible for long lasting immunological memory.13 14 Thus, treatment with rituximab plus cyclophosphamide may have a certain degree of selectivity for pathogenic autoreactive B cells.

The finding that larger absolute numbers of baseline CD19+ B lymphocytes in peripheral blood were associated with a lower likelihood of achieving a beneficial clinical response, as well as a longer time to achieve a clinical improvement, was somewhat surprising. It is possible to speculate that the shorter depletion period observed in patients with higher baseline CD19+ lymphocytes may be caused by persistence of B cells in the lymphoid tissues. In that case, and in view of the correlation between duration of depletion and clinical response, more intensive treatment for some patients might be justifiable. These observations also underscore the fact that dose–response relations for rituximab in autoimmune diseases are poorly understood and cannot be assumed to be identical to those in oncology.

They also raise the question of whether B cell depletion in the peripheral blood is associated with B cell depletion in the lymphoid tissue. In our recent observation of SLE patients with refractory proliferative nephritis treated with rituximab a temporary disappearance of B cells in the renal tissue was observed, with reoccurrence of B cells in the tissue after repopulation in the peripheral blood.10 Similar findings have been reported in the spleen and bone marrow of one patient with ITP treated with rituximab.15

Our data suggest that, for patients with severe SLE who have failed to respond to conventional treatment, the combination of rituximab and cyclophosphamide can provide a new therapeutic alternative. However, we strongly believe that these data cannot yet be extrapolated to imply that even milder forms of SLE, or patients not yet treated with conventional immunosuppressive agents, should be treated with rituximab.

The serological changes observed suggest that the clinical efficacy of rituximab treatment may be related to a greater sensitivity to treatment for those cells producing pathogenic autoantibodies compared with cells involved in normal adaptive immune responses.


We thank Eva Jemseby for management of blood samples and Eleanor Gullström for assistance with the patients.


View Abstract


  • TJ and IG contributed equally to the study.

  • Funding: Supported by a grant from King Gustaf V’s 80th Birthday Fund and funds from the Karolinska Institute and the Swedish fund for the Renal Research. The fee to register the study with the Swedish Medical Product Agency was paid for by Roche AB, Stockholm, Sweden.

  • Competing interests: None declared

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Correction
    BMJ Publishing Group Ltd and European League Against Rheumatism