The modes of action of MT201 and trastuzumab should be highly related. Both antibodies recognise targets that are overexpressed on tumour cells relative to normal tissue, and overexpression of both targets was found to correlate negatively with overall patient survival. Both antibodies share the same human Fcγ
1 (constant region of Ig type 1) portion that, among all human Fcγ
subtypes, is best suited to mediate ADCC (Trinchieri and Valiante, 1993
; Yokoyama and Plougastel, 2003
). Differences between MT201 and trastuzumab may therefore largely relate to their target binding affinities and the biology and quality of the recognised antigens.
Ep-CAM has no known signalling functions and to date no antibody against the molecule has been reported to significantly affect cell proliferation or survival. In contrast, antibody binding to HER-2 is thought to induce receptor-mediated intracellular signalling leading to antiproliferative effects (Chazin et al, 1992
). However, we failed to observe any significant cytotoxic activity by trastuzumab against a panel of nine different breast cell lines in the absence of effector cells (data not shown). While we cannot exclude that incubation times longer than those used in our assays may be needed in order to detect immune effector-independent activities of trastuzumab, the only measurable cytotoxic activity that could be observed for trastuzumab in our study was ADCC. It might be possible that the somewhat higher ADCC activity of trastuzumab compared to MT201 was due to sensitisation of target cells via a receptor-mediated cytotoxic mechanism. Such a sensitisation did, however, not manifest in the presence of complement. A simpler explanation for the lower EC50
values of ADCC of trastuzumab compared to MT201 is that the anti-HER-2 antibody has an approximately 100-fold higher binding affinity than MT201, leading to higher surface density of bound antibody. Similarly, higher levels of ADCC were achieved by MT201 over trastuzumab with cell lines expressing high levels of Ep-CAM and low levels of HER-2. Hence, the lower target binding affinity of MT201 may be compensated by a higher expression level and prevalence of the Ep-CAM over the HER-2 target. Moreover, in this study, we have selected ADCC assay conditions suitable to identify fine differences in ADCC activity between trastuzumab and MT201. In a clinical setting, exposure times of tumour cells to effector cells and antibody will not be 4
h but several weeks. As a consequence, such differences may diminish and much higher levels of ADCC might be observed for both antibodies.
An interesting observation was that six breast cancer cell lines differed only by a factor of two in Ep-CAM expression, namely between 1.24 and 2.22 × 105
sites/cell. Despite a very similar target density, these six cell lines drastically differed with respect to ADCC susceptibility, namely between 6.7 and 44.3% specific lysis. This suggests that at an intermediate Ep-CAM target density other factors that can impact on susceptibility to cellular cytotoxic may become dominant. Such factors could include overexpressed protease inhibitors (serpins) or antiapoptotic proteins involved in immune evasion (Trowsdale and Parham, 2004
). Likewise, the correlation of ADCC induced by trastuzumab with HER-2 expression levels was not very strong.
The absence of any CDC activity by trastuzumab suggests that the antibody cannot promote assembly of the entire complement cascade. Likewise, it is possible that CDC requires a rather high target antigen density and that a cell line such as SKBR3 with the highest density for HER-2 (similar to that of Ep-CAM on MT-3 cells) also happened to express high levels of complement resistance factor CD59, whereas MT-3 cells with a comparably high Ep-CAM target density happened to express low levels of CD59. Ep-CAM, which forms a tetramer (Balzar et al, 2001
), may also allow for a closer binding of two IgG1 molecules and can thereby better mediate C1q assembly and CDC than trastuzumab binding to HER-2. The contribution of CDC to the cytotoxic activity of IgG1 in solid tumours may generally be underestimated when studying cultured cell lines. We have here observed a high expression level of complement resistance factors CD46, CD55 and CD59 on almost all cell lines. IHC studies of tumour samples from various stages of colorectal cancer, however, showed that CD59 and CD55 expression is far less frequent than seen with our tumour cell lines (Koretz et al, 1993
; Hosch et al, 2001
). As CD59 can interact with CD2 on T cells, there may even be a selective pressure against CD59 expression on tumour cells to avoid interaction with T cells (Hahn et al, 1992
; Koretz et al, 1993
In designing a novel antibody against Ep-CAM, experiences with other Ep-CAM antibodies and IgG1 mAb therapies need to be considered. Unlike the murine antibody edrecolomab (Panorex®), which rapidly loses efficacy in man due to neutralisation by anti-edrecolomab antibodies, a novel anti-Ep-CAM antibody should be a human IgG1 and of lowest possible immunogenicity. Only these two features can guarantee maximal serum half-life and efficacy through optimal compatibility with human immune effector mechanisms (ADCC and CDC). This goal was achieved with MT201 by isolating Ep-CAM-specific VH and VL domains from a human IgD-positive B-cell repertoire essentially free of somatic mutations (Raum et al, 2001
), and their fusion to human IgG1 constant domains. Moreover, from a safety standpoint, an anti-Ep-CAM antibody must be able to discriminate between the low levels of Ep-CAM found on normal tissues and the Ep-CAM overexpressed on malignant tissues. There is evidence from clinical trials with high-affinity anti-Ep-CAM antibodies, which caused acute pancreatitis with an increase in serum levels of amylase and lipase (Better et al, 2002
), that Ep-CAM on normal tissue can be recognised by these antibodies. On the other hand, edrecolomab with a moderate affinity in the range of MT201 (Naundorf et al, 2002
) showed in more than 3000 treated patients a relatively benign safety profile (Riethmuller et al, 1994
; Punt et al, 2002
). An additional therapeutic window for MT201 may come from an engagement and sequestration of Ep-CAM within homotypic cell adhesion zones present between normal epithelial cells. Antibody binding to the few disengaged Ep-CAM molecules on the basolateral side of normal epithelial cells may not suffice to form cytolytic synapse with effector cells nor for effective complement fixation. Moreover, in vitro
cell culture reaggregation assays have shown that high-affinity anti-Ep-CAM antibodies such as 323/A3 can block cell adhesion at concentrations of 10μ
(Litvinov et al, 1994
). Therefore, high-affinity but not intermediate affinity anti-Ep-CAM mAbs may be able to damage directly the insulating function of normal (pancreatic) epithelia by neutralising Ep-CAM's epithelial cell adhesion function.
In light of the previous experience with anti-Ep-CAM antibodies, we selected a moderate Ep-CAM binding affinity for development of MT201. This required to compromise on a reduced ADCC activity at low antibody concentrations. The reduced target affinity of MT201 may, however, be effectively counter-balanced by two factors. One is the high-level overexpression of Ep-CAM on many human carcinomas and the high prevalence of Ep-CAM overexpression. The other is that IgG1 therapies for cancer treatment require relatively high serum trough levels, typically in excess of 10μ
, at which concentration target binding is in saturation even for low-affinity antibodies. High trough levels are counter-intuitive in light of the relatively high target binding affinities of most IgG1 mAb therapies. They may relate to the fact that efficacy is mostly mediated by effector cells bearing the low-affinity Fcγ
receptor type III (CD16) (Hazenbos et al, 1996
; Clynes et al, 2000
). Therefore, it may be largely the affinity of IgG1 for CD16 and not for the target antigen that is rate limiting for efficacy. Furthermore, the quenching of this interaction by extra IgG in serum will require higher in vivo
mAb concentrations for competition (Naundorf et al, 2002
). A key role for CD16 in ADCC is supported by the observation that a CD16 polymorphism had a profound impact on the clinical efficacy of rituximab (Cartron et al, 2002
). This polymorphism affects the affinity of CD16 for IgG1 by a single amino-acid difference.
The therapeutic potential of MT201 and trastuzumab in patients will be affected by several other parameters. Tumour penetration of antibodies was found to be improved with reduced target affinity (Weiner and Thakur, 2001
), which could be an advantage for MT201. Likewise, beneficial pharmacokinetic properties such as long serum half-life and low immunogenicity of MT201 and low internalisation of the antibody-bound target can positively impact the efficacy of an antibody in man. The prevalence of Ep-CAM in metastatic breast cancer and its prognostic relevance (Spizzo et al, 2002
) make antibodies against this target very attractive for patients that are not eligible for trastuzumab treatment. According to a study by the Gastl group (Gastl et al, 2000
; Spizzo et al, 2002
), approximately 25% of patient samples analysed had high levels of Ep-CAM expression but were HER-2 negative. A total of 90% of all metastatic breast cancer patients had Ep-CAM-positive tumours, of which 42% showed high-level and 48% medium to low-level Ep-CAM expression. Not only tumours that express high levels of Ep-CAM molecules on their surface but also certain tumours that express low to intermediate levels may be susceptible to ADCC by MT201, as suggested by the data presented in this report (see ). Treatment with MT201 for several months at high trough levels may provide ample time for MT201 to recruit immune effector cells for elimination of tumour cells with a broad spectrum of Ep-CAM expression levels.