- Top of page
- Materials and methods
- Authorship and disclosures
- Supporting Information
The effects of two CD20 antibodies, namely rituximab, the current standard for treatment of chronic lymphocytic leukaemia (CLL) in combination with chemotherapy, and GA101, a glyco-engineered type II antibody were compared on CLL cells ex vivo. Antibody-induced phosphatidylserine exposure was examined in isolated CLL cells. For a more comprehensive assessment of antibody-mediated cell killing including Fc-mediated mechanisms, B cell depletion from whole blood samples was monitored. Treatment with rituximab or GA101 reduced the average viability of isolated CLL cells by 6% or 11%, and the ratio of B to T cells in whole blood samples by 12% or 33%, respectively. Combination with GA101 enhanced the cytotoxicity of the chemotherapeutic agent chlorambucil on isolated CLL cells. CD20 surface expression on CLL cells correlated with GA101-induced B cell depletion, but not with direct cell death induction. Treatment of whole blood samples from CLL patients with a CpG-containing oligonucleotide increased CD20 expression on CLL cells and GA101-dependent B cell depletion. Despite the variable responses of individual CLL samples, the CLL cell depletion from whole blood by GA101 was consistently much stronger than by rituximab, which argues for clinical investigation of GA101 in CLL patients.
Chronic lymphocytic leukaemia (CLL) is the most common leukaemia in the Western world and is characterized by the accumulation of long-lived B lymphocytes. Despite recent advances in its therapy, the disease still remains incurable and new treatment options need to be developed. Among targeted therapy approaches treatment with monoclonal antibodies has the advantage of specifically attacking tumour cells and thus is usually associated with mild side effects.
As a common cell surface antigen of all B cells except stem or plasma cells, CD20 has become the most effective antibody target for the treatment of B cell malignancies (Molina, 2008) and, despite variable surface expression on CLL cells, has also been considered for CLL therapy. Thus, rituximab treatment of CLL patients leads to efficient tumour cell depletion alone (Byrd et al, 2001; Huhn et al, 2001; O’Brien et al, 2001) and particularly when combined with chemotherapeutic agents (Byrd et al, 2003; Keating et al, 2005). Together with the monoclonal anti-CD52 antibody alemtuzumab, rituximab thus may be counted among the most efficient targeted treatment options for CLL reported so far. In a recent phase III trial the inclusion of rituximab was shown to substantially improve the established fludarabine/cyclophosphamide chemotherapy regimen (Hallek et al, 2010).
Treatment with anti-CD20 antibodies leads to both, direct cell death (DCD) induction by receptor engagement and cell killing by Fc-mediated mechanisms, namely antibody-dependent cell-mediated cytotoxicity (ADCC) and complement-dependent cytotoxicity (CDC) (Glennie et al, 2007; Lim et al, 2010). According to the epitope-binding mode two types of anti-CD20 antibodies are distinguished (Cragg et al, 2003), type I, e.g. rituximab, and type II, e.g. GA101. Treatment with type I antibodies leads to CD20 accumulation in lipid rafts and to efficient CDC induction. In turn, type II antibodies potently induce DCD and lead to homotypic aggregation.
The anti-CD20 antibody GA101 (afutuzumab, RO5072759) was developed from the murine antibody B-ly1 and found to exhibit far better pre-clinical overall anti-tumour activity than rituximab in various B cell malignancies including one CLL sample (Mössner et al, 2010). During humanization, variants at the elbow hinge region with superior antigen binding were selected, which recognize a type II epitope. In addition the Fc region was glyco-engineered in order to enhance the affinity for Fcγ receptors and thus to improve ADCC (Ferrara et al, 2006).
In contrast to its in vivo efficacy in terms of B cell depletion, rituximab shows only minimal direct cytotoxic effects on CLL cells in vitro in the absence of cross-linking (Stanglmaier et al, 2004; Zent et al, 2008). Therefore it was of interest to compare, on isolated CLL cells, the action of rituximab and the novel type II CD20 antibody GA101 with greatly improved DCD induction in various lymphoma cell lines (Mössner et al, 2010). As GA101 is a promising new therapeutic reagent for CLL and entering into clinical trials for CLL therapy (Morschhauser et al, 2009), we investigated specifically for CLL the size and distribution among individual samples of GA101-induced anti-tumour effects. For this purpose we assessed whole blood samples from 37 CLL patients for their response to CD20 antibodies in an autologous assay that indicated B cell depletion due to direct as well as Fc-mediated killing mechanisms. In most samples, GA101-dependent depletion of CLL cells was sufficient for attempting a dissection of contributing mechanisms by specific interference. For this purpose we specifically targeted FcγRIIIa, also known as CD16, which is displayed on natural killer (NK) cells and macrophages, because ADCC activity in peripheral blood monocytes is dominated by NK cells (Taylor & Lindorfer, 2008). Other reasons for choosing CD16 among the Fcγ receptors displayed on various effector cell populations include the occurrence of extensively explored low and high affinity variants of FcγRIIIa (Cartron et al, 2002) and its importance for Fc glyco-engineering (Bowles et al, 2006; de Romeuf et al, 2008). As GA101-dependent B cell depletion showed high inter-individual variation we stratified the investigated CLL samples according to prognostic markers and investigated the possibility of correlations between these molecular features of CLL samples and antibody effects with the aim of identifying the patient subgroups which benefit most from GA101 treatment. Given that B cell depletion correlated with CD20 expression on CLL cells, we rationalized that upregulation of CD20 expression via engagement of TLR9 by CpG-containing oligonucleotides (CpG-ODN) might improve antibody-mediated B cell depletion and tested this hypothesis. In addition we investigated in vitro the combination treatment with anti-CD20 antibodies and chemotherapeutic agents, which currently is the predominant clinical application of rituximab for the treatment of CLL.
- Top of page
- Materials and methods
- Authorship and disclosures
- Supporting Information
The present in vitro assessment of anti-CD20 antibodies found superior cytotoxicity against CLL cells by the novel type II glyco-engineered CD20 antibody GA101 than by rituximab, the latter representing a crucial backbone in the current therapy for CLL. For this comparison we either performed cytotoxicity assays on isolated CLL cells in uniform assay matrices or observed antibody-dependent B cell depletion from whole blood samples, which supply individual host-specific immune functions.
In agreement with previous reports (Bellosillo et al, 2001; Stanglmaier et al, 2004; Zent et al, 2008) DCD induction in freshly isolated CLL cells by rituximab was variable, considerably less than in lymphoma cell lines and not reaching a statistically valid difference to untreated controls. With GA101, however, the DCD induction for both, Mec1 cells and CLL patient samples, was significantly higher than with rituximab, i.e. for primary CLL cells DCD induction was, on average, about twice as high. GA101-induced phosphatidylserine exposure on un-stimulated, resting CLL cells was less than on Mec1 cells and B cell lymphoma cell lines, but still significant. Better DCD induction in primary CLL cells and the cell line Mec1 with the type II antibody GA101 compared to the type I antibody rituximab is in agreement with observations in various lymphoma cell lines and murine xenograft models treated with the prototype type II mouse antibody B1 (tositumumab) (Cardarelli et al, 2002; Chan et al, 2003; Beers et al, 2008). Due to a lack of major antibody-mediated induction of membrane disintegration, our evaluation of cell death induction was based on phosphatidylserine exposure alone, similar to the preceding characterization of GA101 effects on B cell lymphoma cell lines (Mössner et al, 2010). Despite the observation of homotypic aggregation by GA101 in CLL cells we did not find the associated extent of membrane disintegration typical of lysosome-mediated membrane damage induced by type II CD20 antibodies in lymphoma cell lines (Ivanov et al, 2009).
Given the significantly improved but still moderate DCD induction by GA101 in primary CLL cells, we used an assay encompassing effector cell-mediated mechanisms that constitute the main contribution to rituximab-induced B cell depletion in vitro (Voso et al, 2002). In the B cell depletion assay from whole blood samples from CLL patients and healthy donors the B cell depletion by GA101 greatly surpassed that induced by rituximab. In contrast, the ex vivo depletion of B lymphocytes and externally added Raji cells from whole blood samples from healthy donors by the second generation type I anti-CD20 antibody veltuzumab (hA20) was in the same order as that of rituximab (Goldenberg et al, 2009).
GA101-induced B cell depletion was highly variable, with a range of 3–88% B cell depletion by 10 μg/ml of GA101 over 24 h. This can be partly explained by the dependence of Fc-dependent killing mechanisms on antigen density, because GA101 and rituximab-induced B cell depletion, but not DCD induction, correlated with CD20 expression on the target cells. This is in agreement with the previously observed correlation of rituximab-induced CDC with CD20 expression on CLL cells (Bellosillo et al, 2001; Golay et al, 2001). The less efficient B cell depletion by type I as compared to type II antibodies has recently been attributed to CD20 internalization, leading to reduced macrophage recruitment and to degradation of cell-bound antibodies (Beers et al, 2010).
Dissection of the contributions of DCD, CDC and ADCC to the observed B cell depletion from whole blood samples is complicated by the divergent read-out used for the assessment of B cell depletion and DCD, but mainly by numerous factors influencing host-dependent complement and immune effector cell function, most prominently inhibition of Fc-mediated antibody effects by endogenous human IgG (Lefebvre et al, 2006; Preithner et al, 2006). A substantial contribution of ADCC to overall GA101-dependent B cell depletion was shown in whole blood from healthy donors by blocking the interaction of FcγIIIa on NK cells and macrophages and the Fc exposed on antibody-coated target cells by incubation with anti-CD16 antibody. In blood samples from CLL patients the same treatment did not substantially decrease GA101-dependent B cell depletion, owing to dramatically increased amounts of target cells resulting in unfavourable ratios of effector to target cells. This effect may be enhanced by impaired NK cell levels and additional deficiencies in immune effector cell function. In the present assay matrix, the CD16-mediated ADCC of GA101 against CLL cells appeared to be much less prominent than reported for lymphoma cells (Mössner et al, 2010). A similar relationship is suggested by observations about the Phe158Val polymorphism of FcγIIIa (Cartron et al, 2002). Rituximab-induced ADCC is enhanced in about 15% of Europeans who are homozygous for the valine form expressing the high affinity variant of FcγIIIa. While the Phe158Val polymorphism of FcγIIIa predicts response to rituximab treatment for non-Hodgkin lymphoma (Cartron et al, 2002), this is not the case for CLL (Farag et al, 2004). For healthy donors our data confirm superior ADCC induction by GA101 than by rituximab (Mössner et al, 2010), which was also reported for the type II antibody mAB 1.5.3 (Bornstein et al, 2010). Quantitative comparison of ADCC elicited by different antibodies requires application of a uniform assay system, e.g. co-culture with the NK cell line NK-92 (Weitzman et al, 2009). On the other hand the B cell depletion assay from whole blood described here has the advantage of reflecting both, the efficiency of antibody-induced B cell depletion and the potential to supply host-dependent immune functions and thus should be able to predict at the individual level the clinical efficiency of therapeutics assayed in vitro.
In a similar approach as for ADCC we tried to determine the contribution of CDC to antibody-mediated B-cell depletion from whole blood samples, namely by inhibiting complement function by incubation with CVF. Possibly due to the rare occurrence of sufficiently high rituximab-induced overall B cell depletion we failed to detect its inhibition upon addition of CVF. This is in agreement with the low rituximab-mediated CDC observed in vitro in CLL cells (Bellosillo et al, 2001; Golay et al, 2001), which is considerably lower than in most B cell lymphomas (Mishima et al, 2009), and negligible in comparison with alemtuzumab-induced CDC in CLL cells (Zent et al, 2008). We confirmed the greatly differing CDC exerted by alemtuzumab and CD20-antibodies on isolated CLL cells for GA101. In Raji cells the presence of complement enhanced the cytotoxicity mediated by rituximab, but not by the anti-CD20 antibody B1 (Stanglmaier et al, 2004). Similarly, in contrast to GA101, a trend for better efficiency in the presence of functional complement was observed for rituximab in the cell line Mec1. Interestingly, the type II antibody mAB 1.5.3 has CDC comparable to rituximab for B cell lymphoma cells other than CLL (Bornstein et al, 2010).
The variability of GA101-induced B cell depletion among CLL samples poses the challenge to define molecular features that predict which patients will benefit most from this new therapeutic agent. In this context, stratification of CLL samples according to cytogenetic aberrations revealed that CD20 expression is below average for del 11q and above for trisomy 12 (Tam et al, 2008), with potential consequences for GA101 efficacy due to its relationship with CD20 expression. Indeed three of five samples in our collection with known del 11q are ranked within the quartile with lowest GA101-induced B cell depletion, while one of the three cases with known trisomy 12 exhibited the highest GA101-dependent depletion of CLL cells encountered in this study. In addition we stratified the investigated CLL samples according to CD38 and ZAP-70 expression status, both of which define prognostic subgroups with aggressive or indolent course (Rassenti et al, 2008). For both markers and, as far as can be concluded from the limited number of characterized samples, also for IGHV hypermutation, we observed a trend for stronger GA101-induced B cell depletion in the subgroups with worse prognosis. The reverse tendency, namely lower response to GA101 in the subgroup with worse prognosis, was found for cases with del 11q or del 17p as compared to samples with a different known karyotype. This also appears to apply separately to samples with del 17q, which predicts p53-deficiency, since the GA101 responses of three among the four samples with del 17p examined ranked in the lowest quartile of the investigated samples. While resistance to chemotherapeutic agents is frequently linked to previous treatment, as was the case for our samples subjected to in vitro treatment with combinations of chemotherapeutic agents and CD20 antibodies, the treatment status of the donors was not correlated with the response of whole blood samples to CD20 antibodies. This predicts that previously treated and untreated patients alike might benefit from treatment with GA101.
Since Fc-mediated killing mechanisms depend on the antigen density on the target cells, both the variability and the modulation of CD20 expression on CLL cells influenced the efficacy of rituximab and GA101. Accordingly combination therapies have been suggested that include anti-CD20 antibodies and treatments that lead to upregulation of CD20 expression, e.g. addition of the cytokines granulocyte-macrophage colony-stimulating factor and interleukin-4. Our results extend to GA101 the previously observed CpG-ODN-induced enhancement of CD20 expression and of antibody-mediated cell killing, which can be clinically beneficial, especially in conjunction with concomitant long-term apoptosis induction (Jahrsdorfer et al, 2001). For rituximab efficacy the reverse effect due to CD20 downregulation by lenalidomide was observed (Lapalombella et al, 2008).
The predominant clinical use of anti-CD20 antibodies for treatment of CLL is in combination with chemotherapeutic agents. Therefore we assessed combinations of rituximab or GA101 with fludarabine or chlorambucil. In agreement with a previous report (Chow et al, 2002), we found enhanced action of chemotherapeutic agents on CLL cells ex vivo when combined with rituximab. The superior DCD of GA101 as compared to rituximab also translated into greater enhancements of the cytotoxicity due to treatment with fludarabine or chlorambucil. Interestingly the mutual enhancement was significant for the combination of GA101 and chlorambucil, which is under clinical investigation in the CLL11 trial of the German CLL Study Group (http://clinicaltrials.gov/show/NCT01010061).
In conclusion, the present comparison of rituximab and the novel type II anti-CD20 antibody GA101 shows better DCD-induction by GA101 in isolated CLL cells and noticeably improved antibody-dependent B cell depletion from whole blood samples. Although the improved DCD induction and ADCC reported for GA101 in B cell malignancies in general (Mössner et al, 2010) were not fully attained in CLL cells ex vivo, the size and frequency of clearly enhanced overall CLL cell depletion make GA101 a promising candidate for potential treatment strategies as a single agent or in combinations. As to the observed variability of GA101-induced effects among CLL samples, it will be interesting to compare, on an individual basis, the B cell depletion in vitro with the treatment response in larger clinical trials, with the goals of predicting treatment outcome and defining the patient subgroup that might benefit most from treatment with GA101.