«Application of Recombinant Antibodies 27 Application of Recombinant Antibodies in Cancer Patients Jürgen Krauss, Michaela Arndt, and Michael ...»
Application of Recombinant Antibodies 27
Application of Recombinant Antibodies
in Cancer Patients
Jürgen Krauss, Michaela Arndt, and Michael Pfreundschuh
As a consequence of the invention of the hybridoma technology by Köhler and
Milstein (1), many monoclonal antibodies (MAbs) have been evaluated in clinical
trials since the early 1980s. Clinical outcomes were generally poor (2–5), with the
notable exception of marked tumor responses, including long-term remissions of patients with malignant B-cell lymphoma who were treated with patient-specific antiidiotypic antibodies (6–8). The main factors responsible for these initial shortcomings were related to the immunogenicity of the murine protein, to modulation of targeted antigens, and to the poor ability of these antibodies to sufficiently mediate antibody-dependent effector functions in humans.
The advent of recombinant antibody technology led to an enormous revival in the use of antibodies as therapeutic agents in cancer therapy. This review provides a brief historical sketch of the development of recombinant antibodies for immunotherapy of cancer, which is followed by the most significant clinical data, as exemplified by the two clinically most established recombinant antibodies to date. Finally, we will focus on future prospects for antibody-based therapeutic concepts in oncology.
2. The Development of Recombinant Antibodies for Cancer Therapy
2.1. Chimeric Antibodies The first reports of the successful cloning of immunoglobulin gene segments were published in 1977 (9,10), nearly one decade after the discovery of the existence of restriction endonucleases, which enable microorganisms to cleave foreign DNA in a highly specific manner (11). It took another several years until the first recombinant antibodies were constructed as “chimeric” molecules by fusing the rearranged murine variable V(D)J gene segments of a mouse MAb to human constant domains (12,13) or were generated as a recombinant Fab fusion protein by replacing the Fc fragment with From: Methods in Molecular Biology, vol. 207: Recombinant Antibodies for Cancer Therapy: Methods and Protocols Edited by: M. Welschof and J. Krauss © Humana Press Inc., Totowa, NJ 28 Krauss et al.
an enzyme moiety (14). Chimerized antibodies retained the specificity of the monoclonal ancestor and proved to be immunogenic in only a very small subset of patients when administered in clinical trials (15–19). Half lives of chimeric antibodies in human serum were shown to be significantly longer compared to the respective parental murine MAbs (15,16,18,20,21) and even increased after repetitive administrations (18,20,21). Moreover, chimeric antibodies were capable of mediating antibodydependent cellular cytotoxicity (ADCC) with human effector cells and/or to activate the complement cascade very efficiently, both in vitro (22–25) and in vivo (26,27).
2.2. Humanized Antibodies In order to further decrease the immunogenicity of murine antibodies, the first monoclonal antibody was “humanized” in 1986 by grafting the gene segments coding for the antigen binding loops onto human framework regions. Although the expressed antibody retained its full specificity, a substantial decrease of affinity was observed (28). Subsequent cristallographic X-ray diffractions of many antibody variable region binding domains and computer modeling studies based on these crystal structures, allowed, with exception of loop H3, the identification of a small number of “key residues” located either in the loops itself or in the framework regions. These residues determine the main chain conformation (“canonical structure”) of the antigen binding loops (29,30). Based on these fundamental insights into antibody structure, antibodies were successfully modified by retaining murine residues within the acceptor framework regions (31–33) or by secondary directed mutagenesis to restore observed decreases in affinity after humanization (34,35). More recently, antibodies were humanized by “resurfacing” the variable domains. In this case, only accessible residues are of human origin whereas buried, structure-maintaining backbone residues remain murine (36).
Many chimeric and humanized antibodies have been employed in clinical trials (reviewed in ref. 37) and, as a result of these studies, two cancer-specific reagents have been approved by the American Food and Drug Administration (FDA) for the treatment of non-Hodgkin’s lymphoma and metastatic breast cancer, respectively.
In order to extend effector functions, chimeric or humanized antibodies were conjugated to radionuclides and drugs and successfully employed in Phase I/II clinical trials (38–43). As a consequence, the first humanized antibody-drug conjugate (Gemtuzumab Ozogamicin = Myelotarg™) was recently FDA-approved as an orphan drug for the treatment of acute myeloid leukemia of patients 60 yr or older and who are not considered candidates for cytotoxic chemotherapy (44).
To retarget effector cells of the immune system, bispecific chimeric or humanized antibody molecules have been developed to activate cytotoxic T cells (45–48) or myeloid effector cells (49). The latter construct has been administered in clinical Phase I trials to patients with a variety of solid tumors (50–52).
2.3. Recombinant Antibody Fragments The successful expression of functional antigen-binding domains in E. coli (53,54) provided the basis for the rapid development of a new generation of antibody based Application of Recombinant Antibodies 29 molecules with potentially great therapeutic impact. Noncovalently linked VH and VL domains tend to dissociate from each other, particularly at low protein concentrations (55). In order to stabilize the assoziation of the two domains, a synthetic linker peptide has been introduced connecting both variable domains (56). These “single-chain” molecules were shown to retain their full binding specificity and affinity (57). To further enhance the stability of these fragments, intermolecular disulfide bonds were generated by introducing cystein residues in the VH and VL framework regions, respectively, and were shown to increase the stability markedly while retaining the full antigen binding properties (55,58,59). ScFv fragments were engineered as fusion molecules to employ artificial effector functions, since they are unable to mediate natural effector functions owing to the lack of the Fc portion of whole antibodies. This was accomplished by linking them to toxins (60–64) cytotoxic ribonucleases (65–67), enzymes for activation of prodrugs (68–72), radionuclides (reviewed in ref. 73), cytokines (74,75), or chemokines (76). Recombinant antibody fragments have been generated as bispecific molecules by various techniques (77–81) to retarget human cytotoxic T cells (77,80,82–85) or natural killer cells (86). Several methods were employed to increase the avidity of bispecific antibody fragments by constructing them as tetravalent bispecific molecules (87–89).
One approach for combining antibody targeting and activation of cellular effector cells is the construction of chimeric receptor molecules (“T-Body”), consisting of a tumor specific single chain antibody and a signal domain for activation of a cytotoxic effector cell. Engrafting of the constructs into cytotoxic T cells results in the MHC independent destruction of scFv-targeted tumor cells (90–94).
From these third generation antibodies, recombinant immunotoxins are now beginning to enter clinical trials and initial data confirm them as very potent (95,96).
2.4. Phage Display-Derived Antibodies Parallel to the rapid development of engineering the described variants of functional hybridoma derived MAbs, new methods were developed to eventually bypass hybridoma technology. In 1985 was shown that peptides could be expressed on the surface of filamentous bacteriophage. The gene fragments encoding the foreign DNA were inserted into the filamentous phage gene III in order to encode a fusion protein displayed on the surface of the phage without disrupting its capability of infection upon binding of pIII to the F pilus of the bacteria. These phages could be enriched more than 1000-fold after a single round of selection through binding of the displayed peptide to a MAb (97). In 1990 McCafferty and colleagues successfully expressed the variable domains of an antibody on the surface of filamentous phage. The phagederived antibody retained its full binding and specificity to its antigen (98). Antibody variable gene segments of different subgroups could be amplified by the polymerase chain reaction (PCR) (99), using either degenerated primers (100,101), a set of familyspecific oligonucleotides (102), or primers based on the amino acid sequences of immunoglobulin variable domains (103). This allowed the construction of antibody libraries by cloning the PCR-amplified VH and VL repertoire of B lymphocytes into suitable phagemid vectors (104–106) for expression and screening of randomly assoKrauss et al.
ciated variable domain fragments on the phage surface. Binding phage antibodies were isolated from a large number of nonbinders by enrichment of the particles through multiple rounds of in vitro-panning against the antigen of choice and extensive washing steps to remove nonbinders. Phage antibody technology has since become the most powerful tool for isolating highly specific antibodies with high affinities to predefined antigens. Antibody libraries have been constructed from patients (107–109), immunized mice (110–112), as naive libraries from (multiple) healthy donors (113–116), and as (semi)synthetic libraries by randomizing sequences in one or more hypervariable regions (117–123). Antibody repertoires were recently also displayed on ribosomes (124,125), allowing for the generation of very large libraries to be screened within a short period of time.
The natural antibody repertoire of camels and other camelid species contains a large number of functional antibodies devoid of a light chain (126). A phage display library from the VH repertoire of an immunized camel has been constructed (127) and single VH domain antibodies with subnanomolar affinities were isolated (128).
Functional single VH domains with high affinities have recently also been isolated from a human VH repertoire phage display library (129).
Phage display-derived antibody fragments have begun to be introduced in clinical trials as radioimaging reagents in cancer patients (130,131).
2.5. Recombinant Antibodies from Transgenic Mice Many attempts to generate human antibodies by employing the hybridoma technology have been unsuccessful, mainly from the lack of a suitable human myeloma cell line to immortilize B cells (reviewed in ref. 132). Alternately, human antibodies were produced in transgenic mice by replacing the murine immunoglobulin loci of the host genome with the respective human counterpart (133 – 137). Hyperimmunization of the transgenic animals with (tumor) antigens of choice results in the clonal activation of B lymphocytes producing human antibodies. Upon rechallenge of the mice with the antigen of interest, affinity maturated antibodies can be generated in vivo. Immortalization of B cells expressing these antibodies can be achieved by standard hybridoma technology, resulting in the production of entirely human antibodies from established cell lines.
3. Clinical Data It took more than 10 years from the initial development of the first generation of recombinant antibodies to become an integrated part of treatment concepts in oncology today.
3.1. Rituximab (Rituxan™, Mabthera™) The chimeric antibody Rituximab (Rituxan™, Mabthera™) binds to the transmembrane antigen CD20, which is strongly overexpressed in most B cell lymphomas (138).
In two independent Phase II clinical multicenter studies, Rituximab has been administered to more than 200 patients with refractory or relapsed low grade B-cell nonHodgkin’s lymphoma (B-NHL) in four weekly doses of 375 mg/m2. Overall response Application of Recombinant Antibodies 31 rates in 185 evaluable patients were around 50% with complete remissions of 9% and 6%, respectively, by a medium time to progression of 10.2 and 13 mo, respectively (18,139). A favorable tumor response was associated with a histology of follicular NHL, sustained high serum levels of antibody after the first infusion, and a longer remission after prior chemotherapy (18). Treatment-related side effects, mostly observed in the first course of treatment, were low and reversible, and in most cases consisted of fever, chills and headache. These side effects were usually reversible by merely lowering the infusion rate. Only two patients developed an antibody response against the chimeric antibody. Most patients exhibited increasing serum concentrations of the chimeric antibody throughout the treatment courses, associated with progressively longer half lives from 76.3 h to 205.8 h after the fourth infusion (21). The impressive results of these clinical trials led to the FDA approval of Rituximab in 1997 as the first recombinant antibody for tumor therapy.
Rituximab proved its potency also on patients with intermediate and high grade B-NHL. In a Phase II study of 54 patients with relapsing intermediate- and high-grade lymphomas single-agent therapy with Rituximab achieved 5 complete and 12 partial responses. Patients with diffuse large B-cell lymphoma achieved a favorable response rate of 37%, and the median time to progression exceeded 246 d for the responding patients (140). These results formed the basis of a randomized trial in elderly patients (60–80 yr of age) with diffuse large B-cell lymphoma, which compared 8 cycles of a three-weekly CHOP regimen with the same chemotherapy plus Rituximab 375 mg/m2 given on day one of each CHOP cycle. The combination of CHOP and Rituximab reduced the rate of primary progressions by 17%. After a median time of observation of only 12 mo in 328 evaluable patients, event-free and overall-survival achieved with CHOP + Rituximab was significantly better than that after chemotherapy only (141).