A D V A N C E S IN IhlMUNOI.OCY, VOL 5fi
`Chemoimmunoconjugates for the Treatment of Cancer
`
`GEOFFREY A. PIETERSZ, APRIL ROWLAND, MARK J. SMYTH, AND
`IAN F. C. MCKENflE
`
`Austin Research Institute, Austin Hospital, Studley Road, Heidelberg 3084, Victoria, Australia
`
`1. Introduction: Concept of Targeted Chemotherapy
`
`Targeted chemotherapy involves the specific carrier-mediated deliv-
`ery of chemotherapeutic agents to tumors or other target tissues. This
`approach presumes the existence of some molecular, genetic or meta-
`bolic characteristic that differs between target and nontarget cells such
`as a structural membrane protein, a cell-surface receptor, an intracellu-
`lar enzyme, or an altered sequence in the genome. Until recently, a
`problem existed in establishing a discrete and accessible difference
`between neoplastic and normal cells; however, the isolation of some
`oncogenes and their products and the production of monoclonal and
`polyclonal antibodies to tumor-associated antigens indicate that it is
`possible to biochemically distinguish normal and tumor cells (1, 2).
`In parallel with definition of differences between normal and neoplas-
`tic cells is the development ofreagents with a high degree of selectivity
`for targets on the surface and within neoplastic cells. Over the past
`20 years, considerable interest has been focused on targeting systems
`designed to permit selective delivery of drugs, radioisotopes, and tox-
`ins to tumors for both diagnosis and therapy and a great deal of this
`research has been performed utilizing antibodies as carriers (Table I).
`As vehicles for carrying cytotoxic agents to tumors, antibodies have
`the greatest potential; however, a number of other possible carriers
`have been investigated (Table 11). The advantages of antibodies and
`other carriers include: (i) the selective delivery of the cytotoxic agent
`to the tumor cells; (ii) the slow release of the cytotoxic agent from
`the conjugate enabling prolonged exposure of the tumor cells to the
`cytotoxic agent; (iii) the preferential uptake of the cytotoxic agent-car-
`rier conjugate by tumor cells; (iv) the use of extremely cytotoxic agents
`which cannot be used alone because of toxicity; (v) the binding of
`cytotoxic agents to carriers, which may protect the agent from enzy-
`matic degradation and rapid excretion. Evidently then, the use of
`carriers to target cytotoxic agents is an attractive and provocative area
`of research; however, for the drug-targeting concept to succeed, both
`the cytotoxic agent and the carrier when conjugated must retain their
`
`301
`
`Copyright 0 1YY4 hy AcadriiilL P r a % I n <
`All rights of reproduction in m y form rebrrvrd
`
`IMMUNOGEN 2140, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`302
`
`GEOFFREY A. PIETERSZ ET AL.
`
`TABLE I
`AGENTS CONJUGATED TO MONOCLONAL ANTIBODIES
`
`Agent
`
`Examples
`
`Toxins (3)
`
`Anticancer drugs
`Enzymes (5)
`
`Chemotactic factors (6)
`Cytokines (7)
`Isotopes (8)
`Radiosensitizers (9)
`Photosensitizers (10)
`Liposomes (1 1)
`Nuclear magnetic resonance
`contrast agents (12)
`Plasminogenactivators (13)
`Carborane cages (14)
`Iron oxide particles (15)
`Other (16)
`
`Ricin, Pseudomonas
`exotoxin
`see Table 111
`Cytosine deaminase,
`carboxypeptidase
`fMLP
`IL-2
`Wy 1311
`Misonidazole
`Chlorin-e
`
`Gadolinium
`
`Muramyl dipeptide
`
`Note. References are in parentheses.
`
`activity in uiuo. For this and many other reasons outlined below, the
`development of the hybridoma technique to produce monoclonal anti-
`bodies (MAbs) has led to the production of more refined cytotoxic
`agent-carrier conjugates (33).
`As indicated in Tables I and 11, there are many carriers and many
`“bullets” which could be targeted. This review focuses on drug-anti-
`body conjugates; the use of toxins, isotopes, and enzymes are exten-
`reference to them is included
`sively reviewed elsewhere (34)-some
`for comparative purposes.
`
`II. Monoclonal Antibodies as Carriers
`A. DEVELOPMENT
`The use of antibodies as carriers for cytotoxic agents has been under
`consideration since the first recorded suggestions for targeting (35).
`The earliest studies made use of antisera raised by immunizing mice,
`rabbits, sheep, horses, and goats with tumor cells or their subcellular
`fractions (36-38). Antibodies reacting with normal tissue antigens
`were removed by absorption with normal tissue homogenates, thereby
`rendering the antisera relatively “tumor specific.” These approaches
`
`IMMUNOGEN 2140, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`CHEMOIMMUNOCONJUGATES IN CANCER TREATMENT
`
`303
`
`TABLE I1
`NONANTIBODY CARRIERS FOR CYTOTOXIC DRUGS,
`TOXINS AND RADIOISOTOPES
`
`Macromolecules
`DNA (17)
`Bovine serum albumin (18)
`Polyamino acid carriers (19)
`Dextrans (20)
`Lectins
`Concanavalin A (21)
`Hormones
`Insulin (22)
`Melanotropin (23)
`Thyrotropin (24)
`Microparticulate carriers
`Liposomes (25)
`Cells (26)
`Microspheres (27)
`Genetically engineered cytokines
`IL2-PE (28)
`IL6-PE (28)
`IL4-PE (28)
`TCFa-PE (28)
`ICF-PE (28)
`CD4-PE (29)
`IL2-DAB4M (30)
`Miscellaneous
`Arachidonic acid (31)
`Epidermal growth factor (32)
`
`Note. References are in parentheses
`
`were limited, principally because the reagents still lacked specificity
`for tumor antigens; however, many preparations were of value in formu-
`lating procedures for coupling antibodies to cytotoxic drugs (38-40).
`The desire for monospecific antibody reagents and some of the earlier
`difficulties with cell-mediated immunity to detect human tumor anti-
`gens provided some of the impetus for developing MAbs (4 1) and the
`advent of the hybridoma technology and MAbs represented a real
`advance in the field of tumor immunology (33). As a result of this
`technology, the production of many MAbs and the subsequent identifi-
`cation of tumor-associated antigens have considerably extended the
`possibilities of targeting cytotoxic agents to tumors. MAbs, by virtue
`of their unique specificity, the ability to select for the desired affinity,
`and ease of production, have surpassed polyclonal preparations as
`carriers for targeted delivery of cytotoxic agents to tumors. Indeed,
`
`IMMUNOGEN 2140, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`304
`
`GEOFFREY A. PIETERSZ ET AL
`
`the prospect of using antibodies as vehicles for isotopes, drugs, and
`toxins only became a reality with the development of MAbs with some
`degree of specificity for tumors. Reexploration of this approach using
`MAbs has been strengthened by studies which demonstrated that xeno-
`geneic MAbs could not only be safely administered to patients and
`localize in tumors (42) but could also have a therapeutic effect of their
`own in xenograft models (43, 44) and in patients with leukemia and
`lymphoma (45, 46). Although therapeutic effects against tumors have
`been obtained using MAbs alone, and these responses have involved
`complement-mediated effects or modulation of effector macrophages
`and natural killer cells (48), clinical responses to serotherapy have
`been variable (49,50), and animal studies indicate there are limitations
`to this approach (51). The variable antitumor effects of MAbs, however,
`may well be improved by conjugation to cytotoxic agents, given that
`the cytotoxic potential and mechanism of action of many drugs and
`toxins are well understood as many have already been used in the
`clinic.
`
`B. ANTIBODIES ALONE
`Why not antibodies alone? They clearly function in vivo after active
`or passive immunisation, particularly for infectious disease. In prac-
`tice, the use of passively administered antibodies, in cancer, has rarely
`been successful. With regard to antibodies only OKT3 (52) and Cam-
`path 1 (53) appear to be active in transplantation (both) and in lym-
`phoma-leukemia (Campath-1). The reasons have been discussed else-
`where, but essentially there are three major problems: (a) amount of
`antibody bound; (b) poor mobilization of effector mechanisms by
`mouse antibodies; and (c) the development of immune responses to the
`foreign immunoglobulin-refered
`to as human antimouse antibodies
`(HAMA). Recombinant monoclonal antibodies consisting of murine
`variable sequences and human constant domains are now available
`and some have been tested in Phase I clinical trials (54). These recombi-
`nant antibodies where the variable domains of the mouse antibodies
`are engineered onto human constant domains, binds complement and
`have antibody-dependent cellular cytotoxicity (ADCC) activity and
`therefore may activate effector function in man. Alternatively, antibody
`constant domains have been modified (e.g., altered hinged region)
`to improve various functional activities (55) for improved therapy in
`humans. How these modifications effect the HAMA response is dis-
`cussed below. More recently, additional approaches to increase the
`antitumor activity of monoclonal antibodies in uivo have been studied
`by administering biological response modifiers such as interferons (56),
`interleukins (57), and colony-stimulating factors (58).
`
`IMMUNOGEN 2140, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`CHEMOIMMUNOCONJUCATES IN CANCER TREATMENT
`
`305
`
`C. SPECIFICITY AND LOCALIZATION
`1. Targets
`
`To increase the selective targeting of cytotoxic agents to neoplastic
`cells, it is desirable to have clearly defined targets which ideally are
`expressed on the cell surface of tumor cells but not on normal cells.
`Despite the repertoire of murine MAbs reacting with antigens associ-
`ated with human tumors (59), there is no conclusive evidence for
`the existence of human tumor-specific antigens detected by murine
`MAbs-with
`the possible exception of the idiotype of surface immuno-
`globulin on B cell lymphomas (60). While the search continues for
`specific antitumor MAbs produced by murine and more by human
`hybridomas, the targeting of cytotoxic agents with MAbs of absolute
`specificity may not be necessary. For example, an antigen which has
`a higher expression on tumor than normal cells or is absent on vital
`normal cells (e.g., hemopoietic stem cells) may be a suitable target for
`the delivery of cytotoxic agents. Many potential antigens have been
`found to be highly tumor-associated, three of the best known examples
`being a-fetoprotein (AFP) (61), carcinoembryonic antigen (CEA) (62),
`and common acute lymphoblastic leukemia antigen (CALLA) (63).
`A better definition of the known tumor-associated antigens, such as
`CEA and AFP, has been possible using MAbs recognizing different
`epitopes (64-66). CEA is representative of many tumor-associated
`antigens and is one of the most widely studied tumor markers. It is
`immunologically a complex macromolecule, expressing both protein
`and carbohydrate determinants on colon carcinoma cells (67) and has
`been reported to be cross-reactive with NCA-1 (68), NCA-2 (69), nor-
`mal biliary glycoprotein (70), and some circulating cells (71). These
`types of cross-reactivities with normal tissues, displayed by many
`MAbs-binding tumor-associated antigens, make it necessary to clearly
`define the properties of the MAbs both biochemically and by immuno-
`histochemical techniques before they are used as carriers for cytotoxic
`agents. Epitope analysis and immunohistology has allowed a number
`of CEA-specific and cross-reactive antibodies to be identified, prvvid-
`ing the opportunity of using different mixtures of antibodies to over-
`come heterogeneity of CEA epitope expression found within individ-
`ual tumors and between different patients (72). The isolation and
`characterization of cDNA clones encoding CEA reveal a highly con-
`served repeating structure (73). Antibodies to various parts of the CEA
`molecule have been made as an effort to obtain more specific antibod-
`ies (74). MAbs against CEA have proved to be of value for the radioim-
`munoassay of human circulating CEA (75), for the radioimmunolocal-
`ization of tumors (76), and as carriers of cytotoxic drugs such as
`
`IMMUNOGEN 2140, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`306
`
`GEOFFREY A. PIETERSZ ET AL.
`
`vindesine (VDS) (77). Studies using a VDS-anti-CEA MAb conjugate
`demonstrated that the in vivo efficacy against lung and colorectal carci-
`noma xenografts in nude mice was dependent on target antigen expres-
`sion on the tumor xenograft cells. This not only demonstrated the
`specificity of MAbs as carriers for cytotoxic agents, but also showed
`that the amount of antibody binding the tumor cells was limited by
`the number of antigen receptors expressed.
`With the production and characterization of MAbs, there has been
`an identification of many other tumor-associated membrane markers
`of potential value in the diagnosis and therapy of tumors such as
`melanoma (78), lung carcinoma (79), breast carcinoma (80), leukemia,
`and lymphoma (81-83). Some of the antigens identified are carbohy-
`drate structures present on glycoproteins and glycolipids (84), while
`other MAbs have been shown to recognize noncarbohydrate deter-
`minants on glycoproteins (85). Indeed, it has been demonstrated
`that a protooncogene HER-e/neu, an epidermal growth factor (EGF)
`receptor-like member of the tyrosine-specific protein kinase family,
`was amplified in 30% of subjects with human breast cancer, and the
`presence of HER-2 was a significant bad prognostic factor (86). The
`cell-surface membrane protein encoded by this protooncogene is the
`type of target that may be a good candidate for targeted chemotherapy,
`particularly as the target antigen has an integral function in the growth
`of tumor cell. To date, however, MAbs to protooncogene products do
`not appear to be suitable vectors for targeting cytotoxic agents such
`as methotrexate (MTX) to tumor cells (87), possibly due to their poor
`access to oncoproteins in tumor cells as most are expressed intracellu-
`larly and not on the surface (88). However, totally intracellular and
`intranuclear oncogene proteins have been found to be expressed on
`the cell surface as small peptides-in
`association with MHC class I
`molecules (89). In the future, it may be possible to make antibodies
`to these peptides to enable cell-surface detection. At present, they are
`only detected by cytotoxic T cells.
`The cDNAs coding for several other tumor-associated antigens have
`been cloned, CALLA (go), the protein core of the human milk fat
`globule antigen (HMFG) (91) and human prostatic acid phosphatase
`(PAP) (92). cDNAs have been characterized and have led to the produc-
`tion of second-generation antibodies raised to peptide sequences
`based on translation of cDNA (93). These antibodies could be more
`specific than monoclonal antibodies.
`
`2. LocalizationlHeterogeneity
`While highly expressing tumor antigens are required for the specific
`targeting of cytotoxic agents with MAbs, other important factors are
`
`IMMUNOGEN 2140, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`CHEMOIMMUNOCONJUGATES IN CANCER TREATMENT
`
`307
`
`the localization of MAbs to tumor cells in uiuo and their uniform
`penetration into the tumor. Quantitative analysis of the biodistribution
`of radiolabeled antibodies in patients is required as this provides an
`assessment of their tumor : normal tissue localization ratio and ade-
`quately defines the potential of that MAb to localize a particular cyto-
`toxic agent to a particular tumor. MAbs that have been shown to local-
`ize to tumors in uiuo include anti-CEA antibodies to colorectal cancer
`(76,94), anti-PAP antibodies to prostatic cancer (95), anti-HMFG anti-
`bodies to breast cancer (96), MAb p97 to melanoma (97), and 791T/
`36 to bone and soft tissue sarcomas (98). However, radioimmunolocal-
`ization has revealed marked variation in the localization of antitumor
`antibodies in animal and human tumors (99, loo), some MAbs localiz-
`ing in tumors at concentrations fivefold greater than that of normal
`tissues, others failing to selectively localize tumors at all in vivo.
`It is clearly important to coat many antigen-binding sites on the
`tumor with the drug-antibody conjugate and therefore deliver the
`maximum dose of drug to the tumor. Within a given tumor mass not
`all tumor cells are alike and individual cells may not express every
`tumor antigen and result in antigenic heterogeneity. Antitumor MAbs
`have not been demonstrated to bind to all the individual tumor cells
`in human tumor xenografts or patients, and autoradiography of tissue
`sections after localization of MAb 791T/36 in human osteogenic sar-
`coma xenografts revealed peripheral localization and low levels of
`penetration (98). The nonuniform distribution of anti-CEA antibodies
`and their F(ab‘), in colon carcinoma xenografts has been noted (94),
`and in patients l3lI-1abeled 791T/36 did not bind many colon carci-
`noma cells, but rather bound to tumor-pseudoacini and stroma (101).
`Thus the limitation ofany one MAb not only is due to the heterogeneity
`of antigen expression, but is also a reflection of its inability to bind
`every tumor cell due to poor access. Poor tumor vascularity and the
`inefficient transport of MAb across the capillary endothelium into the
`tumor may prevent MAbs from reaching every neoplastic cell in a
`solid tumor. Whether a mixture of several antitumor MAbs can over-
`come these difficulties is yet to be determined, although “cocktails”
`of several MAbs should reduce the more fundamental problem of
`antigen heterogeneity. In this regard, immunohistological staining and
`flow cytometry with MAbs have demonstrated the heterogeneity of
`tumor cell populations (both with regard to antigen density as well as
`the presence or absence of antigen) and emphasized that mixtures of
`several anticolon carcinoma antibodies can react with almost all colon
`carcinomas (102, 103). Of relevance is the study by Ceriani et al.
`which demonstrated an increased therapeutic effect of a cocktail of 1311-
`labeled monoclonal antibodies against a breast xenograft compared
`
`IMMUNOGEN 2140, pg. 7
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`308
`
`GEOFFREY A. PIETERSZ ET AL
`
`to individual labeled antibodies (104). The use of such radionuclide
`antibody conjugates may overcome tumor heterogeneity by killing
`bystander tumor cells lacking antigen and provides a clear advantage
`for using isotopes in immunoconjugates rather than toxin or drugs.
`Cocktails of drug-antibody conjugates have also been used to investi-
`gate the potential of overcoming problems due to antigenic heterogene-
`ity. A cocktail of up to three antibodies conjugated to Idarubicin was
`used against a human colon cancer xenograft in nude mice and it
`was found that combinations of the Ida-MAb conjugates were more
`effective antitumor agents than each conjugate alone (105).
`
`D. INTERNALIZATION
`AND MODULATION
`Once localized at the tumor cell surface an immunoconjugate may
`exert its cytotoxic effect in several ways. First, the cytotoxic agent
`need not be released from the MAb to act; in macromolecular form it
`may exert a cytotoxic action either at the plasma membrane (e.g.,
`phospholipase C) or following internalization. Alternatively, the drug
`or active drug-containing fragment of the conjugate may have to be
`released to be active. This dissociation could occur in the extracellular
`space, at the cell surface, or intracellularly due to degradation by
`lysosomal enzymes. The latter mechanism of action requiring internal-
`ization of the drug-MAb conjugate and delivery to the lysosomes
`constitutes the “lysosomotropic” approach to drug targeting (106).
`The present status of MAbs, as carriers of cytotoxic agents, suggests
`that these reagents should ideally be internalized by tumor cells (100,
`107). The pathways that MAbs use to enter different cell types have
`not been extensively studied, although the internalization of MAbs
`by leukemic, melanoma, and breast carcinoma cells has been examined
`(108-1 11). The fact that some toxin-MAb conjugates have poor cytotox-
`icity (112-114) may be due to the inability of certain antibodies to be
`endocytosed, either due to a low binding affinity or because the target
`antigen is not readily internalized. Indeed, a toxin-F(ab’)2 conjugate
`has been shown to be more cytotoxic than the corresponding Fab’
`conjugate in uitro, suggesting that crosslinking of MAb-antigen com-
`plexes and endocytosis was important for this conjugate’s cytotoxic
`effect (115). The cell type may also influence internalization, as MAbs
`directed against different epitopes of the same target antigen on differ-
`ent cells induce different internalization responses (116). Also given
`that the A-chain fragments of toxin molecules such as ricin require
`translocation from an endosomal compartment to avoid deactivation
`by lysosomal enzymes, the proximity of the MAb-bound epitope to
`
`IMMUNOGEN 2140, pg. 8
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`CHEMOIMMUNOCONJUGATES IN CANCER TREATMENT
`
`309
`
`the plasma membrane may have some bearing on the cytotoxic effect
`of some A-chain toxin-MAb conjugates.
`The availability of large panels of antitumor MAbs to different
`epitopes on target antigens is no guarantee of successful targeted
`chemotherapy, as antigenic modulation and/or immunoselection of
`antigen-negative tumor cells may prevent effective therapy. Antigenic
`modulation is defined as the redistribution of surface antigen after
`binding of antibody (117) and may involve internalization and degrada-
`tion of the antigen or shedding of the antibody-antigen complex from
`the cell surface. The factors regulating antigenic modulation are not
`clear. Some antigens modulate rapidly, slowly, or not at all, others
`modulate when exposed to multiple antibodies directed against unre-
`lated antigens (118). Loss ofantigen from the cell surface by internaliza-
`tion has posed problems for the use of’ MAbs aIone in clinical therapy;
`however, the toxicity of some agents and the intracellular nature of
`their mechanism of action suggest that MAb-targeted cytotoxic agents
`could eradicate tumor cells which modulate by antigen internalization
`(1 19). In targeting surface antigens that are modulated by shedding,
`effective therapy may require that the cytotoxic agent be released from
`the surface-bound antibody before shedding occurs and that the agent
`be cytotoxic extracellularly, or intracellularly independent ofthe MAb.
`Selectivity may be compromised, however, and, therefore, target anti-
`gens that are modulated by cell-surface shedding are not good candi-
`dates for MAb-targeted chemotherapy. Immunoconjugates prepared
`using a noninternalizing antibody have also shown antitumor effects
`in mice. These conjugates used a vinca alkaloid derivative linked to
`antibody via a hydrazide linkage and antitumor efficacy was attributed
`to the release of drug at the tumor site (120).
`Another possible drawback for therapy by immunoconjugates is the
`presence of circulating tumor-associated antigens resulting from cell-
`surface shedding or tumor cell destruction. These bind conjugates in
`the serum before they have an opportunity to eradicate tumor cells.
`As an example, the presence of serum CEA has made MAbs against
`CEA difficult to use as diagnostic agents in vivo (121). Similarly, in B
`cell tumors with circulating monoclonal globulins and where the MAb
`is commonly directed against the idiotype of the antibody secreted by
`the tumor cell little may reach the tumor; some reduction in circulating
`antigen can be obtained by plasmaphoresis but the reduction is short-
`lived (122).
`Several novel strategies for targeting have been described and are
`based on the enzymatic activation of “prodrugs,” i.e., inactive forms
`of the drug which become active only when the drug is released. In
`
`IMMUNOGEN 2140, pg. 9
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`310
`
`GEOFFREY A. PIETERSZ ET AL.
`
`these studies enzymes such as alkaline phosphatase (123), or cytidine
`deaminase (124), can be coupled to monoclonal antibody and, after
`the complex localizes to the surface of tumor cells, a nontoxic prodrug
`is administered. These prodrugs are activated at the target site and
`diffuse into the tumor or act on the surface. For such therapy, MAbs
`to noninternalizing antigens are required.
`
`E. SIZE
`One of the factors that controls the uptake of MAb by the tumor is
`the permeability across the capillary wall of tumor blood vessels which
`may have varying types of endothelium, depending on the site and
`origin of the tumor. The capillaries in tumors have been shown to be
`“leaky,” at least to proteins, and it is therefore assumed that they are
`relatively permeable to MAb; this may not apply to all parts of all
`tumors, and therefore measures which increase permeability, reduce
`the size of the cytotoxic agent-MAb conjugate, and define its shape
`and charge must all be considered. Some emphasis has been placed
`on reducing conjugate size by using antibody fragments: F(ab’), (MW
`-110 kDa) and Fab’ (MW -55 kDa) are smaller than intact immuno-
`globulin IgG (MW -150 kDa) and therefore may permeate into tumors
`more easily. Radioimmunolocalization studies demonstrate that anti-
`body fragments can give earlier and superior localization to intact
`MAbs (125); however, this is probably due to faster clearance of frag-
`ments from the blood, thereby reducing the background rather than
`giving absolutely higher levels of antibody in the tumor. In addition,
`cleaving the Fc portion from the antibody molecule should decrease
`nonspecific binding to nontumor cells possessing Fc receptors and
`reduce the immunogenicity of xenogeneic MAbs. Not only are frag-
`ments cleared more rapidly than intact IgG (126), but the lower affinity
`associated with univalent Fab’ binding reduced cytotoxicity of
`toxin-Fab’ conjugates compared with their corresponding divalent
`F(ab’), and intact IgG conjugates in uitro (127). The preparation of
`fully active fragments by current techniques (128) is not always easy,
`each antibody requiring optimization of fragmentation and activity
`testing in uitro and in uiuo. Therefore MAbs will vary with respect to
`the in uivo localization of their Fab’ and F(ab’), fragments depending
`on the particular behavior of the MAb after fragmentation or coupling
`to cytotoxic agents. Using a nontoxic derivative of the alkylating agent
`melphalan N-acetylmelphalan (N-AcMEL), fewer molecules of N-
`AcMeL were coupled to F(ab’)z than intact MAb and the antitumor
`activity against a murine thymoma was only marginally better (129,
`130). The use of F(ab‘), immunoconjugates may be useful and neces-
`
`IMMUNOGEN 2140, pg. 10
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`CHEMOIMMUNOCONJUGATES IN CANCER TREATMENT
`
`31 1
`
`sary for drug conjugates of methotrexate, aminopterin, and other ex-
`tremely toxic drugs. Immunoconjugates are sometimes more toxic than
`fragments are used as carriers (131, 132).
`unbound drug unless F ( a l ~ ’ ) ~
`Advances in the production of genetically engineered antibodies and
`antibody fragments may reduce some of the problems associated with
`antibody fragments produced by enzymatic methods (133). Fab and
`F(ab’), fragments have been expressed in transfected cells and more
`importantly chimeric fragments produced in this way may decrease
`the immunogenicity (see below) associated with the administration of
`murine antibody in humans (134, 135). Single-chain antigen-binding
`proteins (SCA) have also been engineered. These proteins consist of
`regions linked via a peptide (136). The SCAs
`antibody V, and V,
`retain full antigen-binding capacity with some decrease in binding
`affinity (137). In a study by Yokota et d., the penetration of various
`immunoglobulin forms, IgG, F(ab‘),, Fab’, and sFv were compared in
`uiuo using autoradiography (138). The main conclusion from this study
`is that maximum penetration for sFv was at 0.5 hr while IgG reached
`maximum at 48-96 hr postinjection. The percentage injected dose per
`gram for IgG, F(ab‘)2, Fab’, and sFv were 27.2, 19.2, 3.7, and 1.7,
`respectively. Furthermore, this study showed that sFv penetrated
`more deeply (more distal from blood vessels) into the tumour than
`intact IgG. The smaller size and as a result the rapid clearance from
`the blood make sFv’s exceptional for coupling radioisotopes for immu-
`noscintigraphy (139); their use as carriers of drugs still remains to be
`examined.
`
`F. IMMUNOGENICITY AND TOXICITY OF IMMUNOCONJUGATES
`
`An antibody response develops in humans after the administration
`of mouse or rat MAb to tumors (140, 141), resulting in the presence
`of circulating human anti-immunoglobulin, HAMA. These human anti-
`bodies may, on repeated exposure to the originally administered xeno-
`geneic antibodies, form immune complexes leading to hypersensitivity
`reactions (142), which preclude further immunotherapy. Surprisingly,
`however, a large amount (1.5-3 g) of foreign MAb has been injected
`into human subjects without signs of toxicity (50, 122). Immunogenic
`responses have been noted in more than half the patients receiving
`mouse antibodies, but few instances of a severe anaphylactic reaction
`have been recorded. Other relevant observations include the follow-
`ing: in one patient the therapeutic effect of the MAb was neutralized
`(143) and patients initially developed antibodies against the Fc deter-
`minants on mouse MAbs (142). In a Phase I study where a cocktail of
`three N-AcMEL-MAbs were administered using hepatic artery infu-
`
`IMMUNOGEN 2140, pg. 11
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`312
`
`GEOFFREY A. PIETERSZ ET AL.
`
`sion to patients with colorectal metastases, all patients produced a
`human anti-mouse response (144). Xenogeneic antibodies given to
`patients may have a serious limitation to the duration and effectiveness
`of targeted chemotherapy.
`Different measures have been used to overcome this problem includ-
`ing: chemical engineering or murine monoclonal antibodies by adding
`polyethylene glycol residues (145); the development of immunosup-
`pressive treatment schedules with drugs or anti-T cell antibodies (146);
`the use of genetic engineering to produce stable and functional anti-
`body from the fusion of mouse variable and human constant regions
`of immunoglobulin genes (147-149) or CDR grafting onto human
`framework regions (150); the prior administration of irrelevant MAbs
`to direct antiglobulin responses away from antibodies used to target
`the cytotoxic agent; and the use oftotally human MAbs. The production
`of human MAbs has had some success (151) and improved techniques
`should bring more encouraging results (152-154). Human MAbs
`should not elicit the same intensity of immune response as murine
`antibodies as there will be no protein from a foreign species. However,
`if patients receive (say) 1-2 g of antibody of a unique idiotype, then
`antiidiotype responses may develop. Nevertheless such antibodies
`should have enhanced therapeutic potential as carriers of cytotoxic
`agents .
`
`111. Conjugation Chemistry
`A. INTRODUCTION
`While MAbs with the appropriate immunologic specificity are neces-
`sary for successful targeting of cytotoxic agents to tumors, equally
`important is the method of conjugation (155,156). Prior to conjugating
`cytotoxic agents to MAbs, it is essential to know the available reactive
`groups on both the MAb and the cytotoxic agent. Antineoplastic drugs
`are of many different types and include antimetabolites, alkylating
`agents, DNA intercalators, and antimitotic agents, all of which have
`structures that can interfere with cancer cells, leading to cell death
`(157) (Table 111). The cytotoxic part of anticancer drugs is not usually
`amenable to chemical modification due to its binding to the target,
`and therefore close attention must be focused on the structure-activity
`of cytotoxic drugs when coupling them to MAbs. Reactive groups found
`on drug molecules that can be used to link to antibodies include amino
`groups, keto groups, vicinal hydroxyl groups, phenols, free hydroxyl
`groups, and side-chain carboxyl groups (Table IV). MAbs and interme-
`
`IMMUNOGEN 2140, pg. 12
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`CHEMOIMMUNOCONJUGATES IN CANCER TREATMENT
`
`313
`
`TABLE 111
`ANTINEOPLASTIC DRUGS USED IN
`DRUG-ANTIBOVY IMMUNOCONJUCATES
`
`Antimetabolites
`
`Alkylating agents
`
`Antimitotic agents
`
`DNA intercalating agents
`
`Miscellaneous agents
`
`Methotrexate (158)
`5-Fluorouracil (159)
`Cytosine arabinoside (159)
`Aminopterin (160)
`5-Fluoro-2’-deoxyuridine (161)
`Chlorambucil (162)
`Melphalan (129)
`Mitoinycin C (163)
`Cisplatinum (164)
`Trenimon (165)
`Phenylenediamine mustard (166)
`Verrucarin (167)
`T2-toxin (168)
`Vinca alkaloids (169)
`Deacetyl colchicine (170)
`Podophyllotoxin (171)
`Daunomycin