The entry of antibody-based drugs into mainstream medicine has provided the oncologist with new therapeutic tools that have begun to transform the treatment of cancer (Scott et al, 2012; Lee et al, 2013). Toxicity profiles and cell killing mechanism(s) exhibited by antibody and small-molecule cytotoxic therapies are largely non-overlapping, thus providing positive clinical advantages relating to drug resistance and therapy side effects. Additionally, in some instances, combinations of antibody given together with conventional chemotherapy show synergistic activity manifest as significant improvements in treatment outcomes (Hallek et al, 2010). However, not all naked antibodies possess therapeutic activity per se, a property that is determined in part by the antibody’s particular target molecule. In instances where the antibody has only weak or no observable therapeutic activity, the strategy has been to ‘arm’ the antibody with a protein toxin or a highly potent small-molecule cytotoxic drug to create two closely related classes of therapeutic, namely immunotoxins (IT) and antibody drug conjugates, respectively (Kreitman, 2006; Lambert, 2013).

Immunogenicity presents challenges for all antibody-based therapies where multiple administrations of drug over a protracted period are required to achieve a therapeutic effect (Kuus-Reichel et al, 1994). Immunogenicity in this context is defined as the property of the therapeutic to elicit an unwanted immune response when the patient’s immune system recognises epitopes displayed by the drug as non-self. The first generation of therapeutic antibodies were based on mouse monoclonal antibodies whose non-human peptide sequences proved highly immunogenic, provoking an immune response in the patient and the resultant production of human anti-mouse antibodies (HAMA) (Norman et al, 1993). The clinical consequences of HAMA responses are variable. HAMA can be inconsequential, but can also lead to a reduction in antibody therapeutic efficacy due to the generation of neutralising antibodies by the host that block the antigen binding site of the therapeutic antibody and/or an increase blood clearance rates. HAMA responses may also have immunopathological consequences that contribute to the drug’s dose-limiting toxicity with immune hypersensitivity reactions being frequently observed (Baldo, 2013). Devising strategies that avoid these problems are a necessary prerequisite to the successful development of any antibody-based therapy.

Humanisation of murine antibodies is one way of overcoming the HAMA problem where murine polypeptide sequences are replaced with human ones. This was achieved initially through the generation of chimeric antibodies in which the entire constant region of the murine antibody was replaced with the corresponding human constant region while the entire murine variable domain (Fv) was retained (Morrison et al, 1984). This still left significant amounts of murine sequence in the Fv region that remained immunogenic, and so this prompted the next evolutionary step with generation of ‘humanised’ antibody molecules where only the murine complementarity-determining regions (CDRs) are grafted onto a wholly human immunoglobulin framework (Jones et al, 1986). Further advances came with the production of entirely human antibodies using phage display technology (McCafferty et al, 1990) or transgenic mice carrying human immunoglobulin genes (Lonberg et al, 1994). While the full humanisation of therapeutic antibodies has succeeded in dramatically reducing the incidence and extent of human anti-human antibody (HAHA) responses (Foon et al, 2004), idiotypic paratopes that are unique to each and every antibody of defined specificity reside in the CDR regions of the antigen binding site and are still capable of provoking a humoral anti-idiotype immune response (Harding et al, 2010). It is therefore general opinion that complete elimination of HAHA responses will be difficult to achieve through antibody engineering alone and that other strategies will be needed.

As a class of antibody-based drug, ITs present additional special challenges when it comes to immunogenicity. Being comprised of two separate protein components, an antibody and a toxin linked covalently either through a chemical bond or as a genetic fusion means that there are potentially two macromolecular structures displaying epitopes that may be recognised as non-self. The resultant highly immunogenic nature of IT severely limits their clinical usefulness where repeated treatments may be necessary to achieve a therapeutic effect. Various approaches have been adopted to overcome this and include a variety of immunosuppressive agents given concomitantly with IT. These include cyclophosphamide (Oratz et al, 1990), cyclosporine A (Selvaggi et al, 1993), anti-CTLA4 Ig (Siegall et al, 1997), 15-deoxyspergulin (Pai et al, 1990), anti-CD4 antibody (Jin et al, 1991) and rituximab (Saleh et al, 2002). Site-specific modification of IT with polyethylene glycol has also been shown to reduce immunogenicity and increase the plasma half-life (Tsutsumi et al, 2000). More recently de-immunisation of ITs based on either Pseudomonas exotoxin or diphtheria toxin by engineering out epitopes identified as likely to provoke an immune response (Liu et al, 2012; Schmohl et al, 2015) have proven largely successful and this approach would seem the most likely way ahead in the further clinical development of these chimeric therapeutic molecules.

Andersson et al (2015) describe the effects of oral or i.v. administered cyclosporine (Sandimmune) on the development of neutralising antibodies in cancer patients receiving the anti-EpCAM IT MOC31PE. MOC31PE is based on the anti-EpCAM (CD326) IgG1 murine monoclonal antibody MOC31 linked covalently via a thioether bond to Pseudomonas exotoxin A and is designed to target EpCAM-positive epithelial cancers. The MOC31PE molecule utilises unmodified murine antibody and PE, and is therefore expected to be highly immunogenic in humans, affected only by the immune status of the individual patients. Attempts to use immunosuppressant doses of CsA to reduce the immune response against IT is not new, having first been used in combination with the ricin-based anti-melanoma IT XOMAZYME-MEL, but with little success in this instance (Selvaggi et al, 1993). Andersson et al (2009) have previously shown in preclinical studies that cyclosporine also augments the cytotoxic and therapeutic activity of MOC31PE in preclinical models of human cervical cancer. The phase I dose-escalation study reported by Andersson et al (2015) was undertaken to firstly establish how well the combination of MOC31PE plus CsA was tolerated in patients with EpCAM-positive epithelial cancers and secondly to evaluate the extent to which CsA inhibited the generation of neutralising anti-MOC31PE antibodies. In this dose-escalation study, a total of 34 patients received MOC31PE alone in escalating doses, while 29 received a combination of MOC31PE plus CsA (23 patients received CsA i.v. and 6 orally). Within the patient groups, there were a total of ten grade 3 and three grade 4 adverse events (AE) in the MOC31PE monotherapy-treated group compared with twelve grade 3 and six grade 4 AEs in the MOC31PE plus CsA combination group. The authors used an in vitro MTS cytotoxicity assay to determine whether serum from MOC31PE-treated patients contained antibodies that neutralised the IT. Their assay was not capable of discriminating between anti-MOC31PE antibodies that blocked MOC31PE binding to its EpCAM target or those which neutralised the catalytic activity of PE. After three cycles of treatment (each cycle given every 14 days), 93% of patients treated with MOC31PE alone developed neutralising antibodies compared with only 50% treated with the combination of MOC31PE plus CsA given on a 5-day schedule commencing 1 day before MOC31PE administration. The authors therefore conclude that CsA given on this schedule had an acceptable toxicity/safety profile and was capable of reducing the incidence of anti-IT antibody responses that should allow for multiple doses of IT to be given effectively. While the relatively small patient numbers reported in this dose-escalation study with MOC31PE IT do not allow for any meaningful analysis of any therapeutic benefits of using CsA in combination with the IT, it has provided important safety information that paves the way ahead for subsequent phase II and III trials to explore this very issue.