Complement activation plays a key role in the side-effects of rituximab treatment

Authors


M. H. J. van Oers, MD PhD, Academic Medical Centre, Department of Internal Medicine F4-224, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. E-mail: M.H.vanOers@ amc.uva.nl

Abstract

Treatment with rituximab, a chimaeric anti-CD20 monoclonal antibody, can be associated with moderate to severe first-dose side-effects, notably in patients with high numbers of circulating tumour cells. The aim of this study was to elucidate the mechanism of these side-effects. At multiple early time points during the first infusion of rituximab, complement activation products (C3b/c and C4b/c) and cytokines [tumour necrosis factor alpha (TNF-α), interleukin 6 (IL-6) and IL-8] were measured in five relapsed low-grade non-Hodgkin's lymphoma (NHL) patients. Infusion of rituximab induced rapid complement activation, preceding the release of TNF-α, IL-6 and IL-8. Although the study group was small, the level of complement activation appeared to be correlated both with the number of circulating B cells prior to the infusion (r = 0·85; P = 0·07) and with the severity of the side-effects. We conclude that complement plays a pivotal role in the pathogenesis of side-effects of rituximab treatment. As complement activation can not be prevented by corticosteroids, it might be relevant to study the possible role of complement inhibitors during the first administration of rituximab.

The chimaeric anti-CD20 monoclonal antibody (mAb) IDEC-C2B8 (rituximab) has become an important treatment modality in low-grade non-Hodgkin's lymphoma (NHL) (McLaughlin et al, 1998; Davis et al, 1999), and its application in other CD20-positive B-cell malignancies [e.g. aggressive lymphoma's (Coiffier et al, 1998; Winkler et al, 1999a), post-transplant lymphoma (Faye et al, 1999; Milpied et al, 1999), or chronic lymphocytic leukaemia (CLL) (Byrd et al, 1999a; Emmerich et al, 1999; Kleinman et al, 1999; Winkler et al, 1999b, c)], is rapidly expanding.

In patients with low-grade follicular lymphoma and low numbers of circulating CD20-positive tumour cells, treatment with rituximab was shown to be safe and well-tolerated (McLaughlin et al, 1998; Davis et al, 1999). However, in patients with high numbers of circulating tumour cells, rituximab treatment may be complicated by severe first-dose side-effects, which can be ameliorated but not prevented by the usual prophylactic medication (i.e. acetaminophen, antihistamine and/or corticosteroids) (Jensen et al, 1998; Byrd et al, 1999b; Winkler et al, 1999b). To gain insight into the pathogenesis of the infusion-related side-effects of rituximab treatment, complement activation products (C3b/c and C4b/c) and cytokines [tumour necrosis factor alpha (TNF-α), interleukin 6 (IL-6) and IL-8] were measured in serial samples obtained during the first infusion of rituximab in five low-grade NHL patients.

Patients and methods

Patients Relapsed low-grade NHL patients were treated in an ongoing study evaluating the safety and efficacy of the combination of rituximab (375 mg/m2 weekly × 4) and granulocyte colony-stimulating factor (G-CSF, filgrastim, Neupogen; 5 µg/kg/d, administered on three consecutive days starting 2 d before each infusion) (Fig 1) (van der Kolk et al, 1998). Inclusion criteria for this study were: measurable progression of histologically confirmed CD20-positive B-cell lymphoma (Working Formulation A–C) after at least one and no more than three prior systemic therapies; expected survival of > 3 months; prestudy performance status of 0–2 according to the World Health Organization (WHO) scale; haemoglobin (Hb) > 8·0 g/dl; white blood cell count (WBC) > 3·0 × 109/l; absolute granulocyte count > 1·5 × 109/l; platelet count (Plt) > 75 × 109/l; circulating tumour cells < 0·5 × 109/l; seronegative for human immunodeficiency virus (HIV) and hepatitis-B surface antigen (HBsAg); serum IgG > 6 g/l. The study was approved by the institutional ethics committee and performed according to the guidelines of the Declaration of Helsinki. All patients gave written informed consent before treatment was initiated. Rituximab was administered according to the guidelines described in the investigational drug brochure (IDEC Pharmaceuticals Corporation, 1997). Toxicity was evaluated according to the National Cancer Institute's Adult's Toxicity Criteria.

Figure 1.

Treatment schedule. Rituximab was given weekly for 4 weeks. G-CSF (5 µg/kg/d) was administered for 3 d, starting 2 d before each infusion of rituximab (i.e. the third injection of G-CSF was administered several hours before rituximab infusion). Follow-up was at 1, 2 and 3 months.

Determination of complement activation and cytokine levels Patients received acetaminophen (1000 mg orally) and clemastine (2 mg intravenously) as prophylactic medication. Corticosteroids were never given as premedication. Complement activation products and cytokine levels were measured before start of rituximab treatment (t = 0) and at 30, 60, 90, 180 and 300 min after onset of the infusion. For measurement of complement activation, freshly drawn blood (5 ml) was collected in siliconized vacutainer tubes containing 0·1 mg/ml soybean trypsin inhibitor (SBTI), 10 mmol/l EDTA and 20 mmol/l benzamidine (final concentrations) to prevent any in vitro complement activation. Plasma was collected immediately by centrifugation and stored at −70°C. Complement activation products (C3b/c and C4b/c) were measured by enzyme-linked immunosorbent assay (ELISA) as described previously (Wolbink et al, 1993). Serum levels of IL-6, IL-8 and TNF-α were measured using a commercially available ELISA [Pelikane Compact ELISA kit, Central Laboratory of the Netherlands Red Cross Blood Transfusion Service (CLB), Amsterdam, the Netherlands], according to the manufacturer's recommendations.

Statistical analysis Correlations were determined using Pearson's correlation coefficient. A P-value of < 0·05 was considered significant.

Results

Complement activation products and cytokine levels were measured in five patients. Patient characteristics are listed in Table I. Side-effects observed during the first infusion were fever and chills in two patients, requiring temporary discontinuation of the infusion and administration of corticosteroids (prednisone 25 mg intravenously) (Table II). One patient experienced dyspnoea (grade 3) and flushes, starting 15 min after onset of the infusion. These side-effects resolved after administration of corticosteroids, and infusion was completed without further problems.

Table I.  Patient characteristics.
PatientHistology (WF)Sex/age (years)Bulky disease*BM involvementExtranodal disease
  • *

    Bulky disease is defined as lesions > 7 cm.

  • WF, working formulation; BM, bone marrow.

1AMale/49NoYesNo
2BMale/55NoYesNo
3BMale/75YesYesYes
4BFemale/70NoNoNo
5BFemale/28NoYesNo
Table II.  Clinical characteristics, complement and cytokine levels.
PatientCirculating
B cells
PMNSide-effectsToxicity
grade
Minutes after start
of infusion
Medication
required*
C3b/cIL-6IL-8TNF-α
  • *

    Prednisone 25 mg intravenously.

  • Circulating B-cells and neutrophils (PMN) (counts × 109/l) at start of infusion, side-effects and maximum levels of complement activation product C3b/c (nmol/l) and cytokines (pg/ml) reached during the first administration of rituximab in five low-grade NHL patients.

10·0227Fever chills160n1302313440
20·4514Fever chills250y483613366131
30·3817Fever, chills145y564610329710
40·0217Fever1225n1756376
50·518Dyspnoea, flushes, fever315y1189513134

Levels of C3b/c, C4b/c and cytokines were within normal limits before start of the infusion of rituximab (Fig 2). An increase in both C3b/c and C4b/c was observed 30 min after onset of the infusion. C3b/c reached a maximum at 180 min, while C4b/c reached a plateau at 90 min. Levels of TNF-α started to increase after 30 min, followed by IL-6 and IL-8. TNF-α reached a maximum at 60 min (177 ± 135 pg/ml, range 6–710), IL-6 and IL-8 after 90 min (IL-6 298 ± 131 pg/ml; IL-8 150 ± 81 pg/ml).

Figure 2.

Complement activation products (C3b/c ◊ and C4b/c □) and cytokines (TNF-α U25CF and IL-6 ▴) in five low-grade NHL-patients during the first infusion of rituximab. Values are depicted as mean ± SEM. Normal values: C3b/c < 57 nmol/l; C4b/c < 8 nmol/l; TNF-α < 10 pg/ml; IL-6 < 20 pg/ml.

The severity of the side-effects observed in this small study group, as reflected by both toxicity grade and the requirement to administer corticosteroids, was related to the number of circulating B cells prior to the first infusion of rituximab. Furthermore, the maximum levels of C3b/c were correlated with the number of circulating B cells prior to the infusion (r = 0·85; P = 0·07) (Fig 3). There was no correlation between the levels of the measured cytokines and the number of circulating B cells.

Figure 3.

Correlation between the number of circulating B cells prior to the first infusion of rituximab and the maximum levels of C3b/c in five patients (r = 0,85; P = 0·07).

Prior to the infusion, neutrophil counts had largely increased (Table II). This was not accompanied by an increase in cytokines or complement activation products (Fig 2). Furthermore, there was no correlation between the number of neutrophils before onset of the infusion and the maximum levels of cytokines or complement activation products during the infusion.

Discussion

In the present study we have demonstrated that administration of the chimaeric anti-CD20 mAb rituximab resulted in a rapid complement activation, followed by cytokine release. Because complement activation products are known to activate macrophages and mast cells, which are important sources of cytokines, and complement activation products (e.g. C3a and C5a) can function as anaphylatoxins, our data suggest that, during rituximab treatment, complement initiates cytokine release and is also directly responsible for some of the side-effects (Fig 4). Thus, complement may be the central factor in the pathogenesis of the side-effects of rituximab treatment. This hypothesis is supported by our observation – in agreement with other studies (Jensen et al, 1998; Byrd et al, 1999b; Winkler et al, 1999b) – that the severity of the side-effects of rituximab treatment appeared to be correlated to the number of circulating B cells prior to the infusion, which in turn correlated with the maximum levels of C3b/c. Such a correlation was not found for the levels of cytokines.

Figure 4.

Proposal for a mechanism explaining the pathogenesis of side-effects occurring during treatment with rituximab. Complement activation products can activate macrophages and mast cells, which are important sources of cytokines and vasoactive mediators. Furthermore, complement activation products (e.g. C3a and C5a) can function as anaphylatoxins.

In line with our results, in a recent study complement activation was only found to occur in patients with clinical side-effects (Schwaner et al, 2000). Interestingly, complement activation was only observed in patients with bone marrow involvement. Unfortunately, in this study the number of peripheral blood B cells was not mentioned. However, it is conceivable that those patients with bone marrow involvement have higher levels of circulating tumour cells than patients without bone marrow involvement. This might explain the relationship between complement activation and bone marrow involvement. In agreement with our study, no relationship between IL-6, complement activation and side-effects was found.

Studies on the infusion-related side-effects of OKT3 (a murine anti-CD3 mAb) support our hypothesis of a pivotal role for complement in the pathogenesis of rituximab-induced side-effects. During treatment with OKT3, a profile of complement activation and cytokine release similar to our findings with rituximab has been described (Raasveld et al, 1993; Bemelman et al, 1994; Buysmann et al, 1997). Interestingly, Buysmann et al (1997) showed that by administering OKT3 as a 2-h infusion instead of an intravenous bolus injection, complement activation and side-effects significantly decreased, whereas the cytokine release was not changed. Furthermore, Raasveld et al (1993) demonstrated complement activation without the release of cytokines in one OKT-3 treated patient, and in this patient clinical side-effects were similar to the patients with increased cytokine levels (M. H. Raasveld, personal communication). Thus, these results in OKT3-treated patients support the importance of complement activation in the pathogenesis of side-effects observed during treatment with monoclonal antibodies.

In a recent study in CLL patients, treatment with rituximab was associated with severe side-effects, which were described as a ‘cytokine–release syndrome’ (Winkler et al, 1999b). However, in these CLL patients, peak levels of cytokines were similar to levels in our study patients, who had only mild side-effects. This suggests that other factors, in addition to cytokines, may have played an important role in the pathogenesis of the side-effects in these CLL patients. Although not measured in the described study (Winkler et al, 1999b), high levels of complement activation products may have been responsible for the severity of the side-effects in these patients.

In the present study, patients were treated with a combination of rituximab and G-CSF. Importantly, although the three injections of G-CSF induced a significant increase in the number of neutrophils prior to onset of rituximab infusion, this was not accompanied by complement activation or increase in cytokines (Fig 2). Furthermore, side-effects observed during the present study (van der Kolk et al, 1998) were similar to those reported in low-grade NHL patients treated with rituximab monotherapy (McLaughlin et al, 1998). Therefore, it is unlikely that the administration of G-CSF essentially altered complement activation and cytokine release during the infusion of rituximab. Moreover, we think it is justified to conclude that complement activation plays an important role in the pathogenesis of side-effects of rituximab monotherapy as well.

In contrast to cytokine release, complement activation cannot be prevented by corticosteroids (Jansen et al, 1991; Raasveld et al, 1993; Bemelman et al, 1994). One possibility to decrease complement activation during mAb infusion is by lowering the infusion rate of the antibody (Buysmann et al, 1997; IDEC Pharmaceuticals Corporation & Genentech Corporation, 1999), as has been applied to CLL patients treated with rituximab (Winkler et al, 1999b). However, despite this strategy, first-dose side-effects during rituximab treatment can still be life threatening in patients with high numbers of circulating tumour cells. Hence, it may be relevant to investigate the role of complement inhibitors (Hack et al, 1993) during the first rituximab infusion, especially in patients with high numbers of circulating tumour cells, as they are at risk of severe first-dose side-effects.

Recently, an important role for complement-dependent cytotoxicity in the efficacy of rituximab was claimed, leading to the suggestion of enhancing the efficacy of rituximab by enhancing complement lysis (Golay et al, 2000; Harjunpaa et al, 2000). However, other studies showed the importance of FcγR mechanisms for the efficacy of rituximab (Clynes et al, 2000). Thus, although the most important mechanism of action of rituximab in vivo has not yet been established, it is likely that not only its safety but also its efficacy is to some extent affected by complement. Therefore, studies investigating complement inhibitors in order to improve safety should carefully analyse the potential effects on clinical efficacy. Alternatively, studies exploring strategies to enhance complement activation should be aware of the possibility of increased toxicity.

Ancillary