TEACHING EDITORIAL
`
`Bioconjugate Chem. 1990, 1, 13-23
`
`13
`
`Recent Advances with Monoclonal Antibody Drug Targeting for the
`Treatment of Human Cancer’
`
`Gary A. Koppel
`Lilly Research Laboratories, Indianapolis, Indiana 46285. Received August 21, 1989
`
`I. INTRODUCTION
`Drug targeting had its inception almost a century ago
`when the late Paul Ehrlich proposed that chemothera-
`peutic agents might be covalently joined to ligand sub-
`strates which had affinity for and selectivity to a target
`tissue such as malignant tumors. In addition, he sug-
`gested that antibodies or “magic bullets” as he described
`them might, in fact, be candidates for ligand substrates
`for drug targeting (2). This vision remained dormant for
`almost a century until the attendant technologies and
`support systems would be in place to begin to express
`this vision into a 20th century therapeutic modality.
`During the past 15 years, there has been an exponen-
`tial growth in the area of drug targeting as a result of the
`integration, interfacing, and coordination of the scien-
`tific disciplines represented by cell biology, recombinant
`technology, and chemistry (Figure 1)- Within the area
`of cell biology, major advances have occurred in genet-
`ics, hybridoma technology, screening, and testing. In 1980,
`Benacerraf, Dausset, and Snell received the Nobel Prize
`in Medicine and Physiology for their pioneering efforts
`in elucidating the immune response gene network, that
`is the family of genes that dictate the ability of the mam-
`malian immune system to respond to and process all immu-
`nogens (3). In the course of these important discover-
`ies, the mouse became the representative immune sys-
`tem because of its prolific procreative behavior, physical
`size, and its brief gestation period. As a result, Snell had
`created the world’s most sophisticated genetic library of
`inbred and recombinant strains of mice (4).
`This family of mice became the instrument of knowl-
`edge that was utilized by Kohler and Milstein in their
`discovery of hybridoma technology. The discovery was
`of such magnitude that Kohler and Milstein, along with
`Jerne, were awarded the Nobel Prize in Medicine and
`Physiology in 1984, 9 years after their breakthrough (5,
`6). Hybridoma technology allows the fusion of a normal
`immunological B cell committed to making antibody with
`a malignant myeloma partner, thus affording a hybrid
`cell with the genetic information of both immortality and
`antibody synthesis. Each cell is thus empowered to pro-
`duce unlimited amounts of a single, or monoclonal, anti-
`body (moab) (7). This discovery together with auto-
`mated methods of screening and testing for the selective
`immunoreactivity of derived moabs accelerated the growth
`of the drug-targeting discipline.
`Recombinant DNA technology has complemented and
`facilitated the growth of cell biology. The ability to rap-
`idly identify, sequence, and clone genes of an antibody
`has led to the elucidation of the underlying mechanisms
`of antibody diversity (8). In addition, this powerful tool
`has allowed the construction of chimeric and humanized
`monoclonal antibodies which may reduce their immuno-
`genicity in humans, either as native moab’s or in the con-
`text of drug conjugates (9).
`
`1043-1 802/90/290 1-00 13$02.50/0
`
`Finally, chemistry has catalyzed the growth rate of both
`cell and molecular biology. Through chemistry, the abil-
`ity to rapidly synthesize DNA, peptides, linkers, and phar-
`macologic agents and to characterize and purify biocon-
`jugates has been achieved. What has become apparent
`is the importance of chemistry in the design and synthe-
`sis of bioconjugates; indeed, the rate-determining step
`for the evolution of this program has been the “new chem-
`istry’’ of drugs and linkages compatible with proteins and
`the incorporation of the drugllinkage onto the moab.
`In summary, the three disciplines interface in creating
`a new dimension in the expression of biotechnology that
`has facilitated the emergence of monoclonal antibody drug
`targeting in the treatment of human cancer.
`Conceptually, the process of drug targeting as pro-
`posed by Paul Ehrlich is illustrated in Figure 2. Since
`Ehrlich’s vision of targeting, many investigators in the
`biological fields have tried to translate his dream into a
`reality. In order to build a foundation of understanding
`for moab-based targeting of drugs, this article will review
`the following in turn: (1) the immune system and anti-
`body synthesis, (2) hybridoma technology and the gen-
`eration of monoclonal antibodies, (3) chemical design and
`synthesis of modified targeting agents for attachment to
`moab’s, and (4) the chemistry and biological activity of
`moab-drug conjugates.
`
`11. THE IMMUNE SYSTEM AND ANTIBODY
`SYNTHESIS
`The cast of characters and the sequence of events that
`facilitate the activation of the immune response are illus-
`trated in Figure 3. The macrophage or presenting cell
`takes up the antigen and presents it on its surface in the
`context of class I1 immune response gene products (IO).
`This, in turn, determines the ability of the immune sys-
`tem to respond to the given antigen. Those clones of T
`cells (thymus-derived cells) designated as helpers (Th)
`which express receptors for the antigen in the context of
`the class I1 self-determinant become activated through
`the synthesis and secretion of IL-1 (interleukin 1) (IO).
`This cytokine, in turn, activates those clones of T, cells
`to produce and express receptors for IL-2 (interleukin
`21, a T cell growth factor which supports the growth of
`the autocrine T cell network. This highly sophisticated
`central pathway facilitates and supports the growth of
`both the humoral response represented by B cells (bone
`marrow derived cells committed to producing antibody)
`and the cell-mediated response represented by T cell medi-
`ated delayed type hypersensitivity (TDTH) (IO). The cen-
`tral pathway sustains the cell-mediated response by pro-
`ducing factors which support the growth and differenti-
`ation of the T cell mediating DTH, thus moving the process
`to the end stage effector function. Similarly, the central
`pathway activates the B cell response via the production
`0 1990 American Chemical Society
`
`IMMUNOGEN 2100, pg. 1
`Phigenix v. Immnogen
`IPR2014-00676
`
`

`
`14 Bioconlugate Chem., Vol. 1, No. 1, 1990
`
`Koppel
`
`‘
`
`Drug Targeting I
`I Human Disease Therapy 1
`
`Figure 1.
`
`Normal.
`
`Cancer
`cell
`.... ...... “..
`.
`.#
`c
`i ..... -.. .._-. .
`Figure 2. Immunoconjugate-mediated site-directed therapy.
`Reproduced with the permission of R. John Collier and Don-
`ald A. Kaplan [(1984) Sci. Am. 251 (l), 561.
`
`.-
`‘I 4
`
`Antigen
`
`““I
`
`f
`
`l
`
`I
`
`l
`
`t c l
`
`( 4 4 l c c
`
`clone positive culture @
`
`Assay for antibody,
`
`Assay for antibody,
`propagate positive clone
`
`Figure 4. Reproduced with the permission of R. John Collier
`and Donald A. Kaplan [(1984) Sci. Am. 251 (l), 561.
`
`111. HYBRIDOMA TECHNOLOGY AND THE
`GENERATION OF MONOCLONAL ANTIBODIES
`The process of producing a monoclonal antibody is illus-
`trated in Figure 4. The human cancer tissue is pre-
`sented as an antigen to the mammalian immune system.
`The mouse is the system of choice because hybridoma
`technology was developed within the context of mouse
`genetics (7). The immune system processes the cancer
`tissue and begins to make antibody as has been described.
`Monitoring to determine the reactivity against the can-
`cer tissue is done by assaying serum against original can-
`cer tissue (7). Once the animal is making polyclonal anti-
`body against the target, plasma B cells are obtained by
`excising the spleen, harvesting the B cells, and fusing
`them with malignant myeloma cells (5, 7). The cells are
`then propagated in a medium in which only fused hybri-
`doma cells can survive (e.g., a medium such as
`hypoxanthine/aminopterin/ thymidine, which selects only
`for a survival pathway of fused hybridoma cells). The
`moab’s derived from the surviving hybridoma cells are
`screened in a high-speed automated selection process
`against malignant and normal tissue. Those moab’s that
`have good immunoreactivity against malignant tissue and
`minimal reactivity against normal tissue are selected and
`further evaluated as targeting ligands in the context of
`drug conjugates (5, 7).
`Pursuant to understanding the chemistry of designing
`and developing conjugates, it is important to review the
`structure and the attendant biochemical characteristics
`of the antibody. A representation of an IgG class of anti-
`body is illustrated in Figure 5 (11). The antibody is com-
`posed of identical heavy chains denoted by the subscript
`H which are joined by two disulfide linkages located in
`the “hinge region” of the antibody. Two identical light
`chains denoted by the subscript L are joined to the heavy
`chains by disulphide bonds connecting the constant por-
`tion of the light chain (C,) to the heavy chain first con-
`stant region (CHI). The amino terminus of each chain
`is located at the variable portions of both the heavy and
`light chains, Vy and V,,
`respectively. The C terminus
`of the light chain is located at the CL domain and that
`of the heavy chain is at the cH3 domain (11). The immu-
`noreactivity of the moab is controlled by the variable
`
`I
`I
`
`* Afferent
`
`-
`- I -
`-
`
`--Lymphokines
`
`Efferent
`
`\ <
`I
`I
`I
`I
`
`Figure 3.
`of growth and differentiating factors. As a result, the B
`cell recognizing the antigenic epitope matures to the end
`stage effector cell, the plasma B cell (PC), which secretes
`the antibody specific for the antigen at the incipient stage
`(10).
`In addition to the utility of antibody in maintaining
`the survival of the organism, the antibody was recog-
`nized as an attractive candidate for ligand targeting. If
`one had the capability to intercept a specific plasma B
`cell clone and produce unlimited sources of moab, the
`moab’s potential for recognizing and binding, selec-
`tively, to a given epitope, would make it the “universal
`ligand” in targeting. As a result of the Kohler/Milstein
`hybridoma breakthrough, the ability to select for and pro-
`duce unlimited quantities of moab became a reality and
`thus fueled the research in the targeting program (5, 7).
`
`IMMUNOGEN 2100, pg. 2
`Phigenix v. Immnogen
`IPR2014-00676
`
`

`
`Teaching Editorial
`
`Bioconjugate Chem., Vol. 1, No. 1, 1990 15
`
`L Chain
`Hypervariable
`Regions
`
`H Chain
`Hypervariable
`Regions
`Hinge Region
`Complement Binding Region
`Carbohydrate
`
`u
`CH3t.l
`
`lntrachain
`
`Figure 5.
`
`Agent
`
`Covalent
`Bioconjugate
`Ligand
`Linkage
`-0 D ----+ (-)xD
`
`tissue
`
`Release of
`free drug
`
`Figure 6. Design and synthesis of agent-linkage-ligand con-
`jugate: (1) Agent-linkage-ligand construction expresses inher-
`ent biological activity or facilitates its release at the target site.
`(2) Agent and/or linkage amenable to stoichiometric determi-
`nation within the context of the bioconjugate.
`domains and is comprised of three peptide sequences in
`the hypervariable region of the light chain and four
`sequences in the heavy chain (11). The species charac-
`teristics of the moab are expressed both in the frame-
`work region of the variable domain as well as specific
`sequences of the CH and CL regions. In addition, classes
`of moab’s such as IgM and IgE differ from IgA, IgD, and
`IgG by the addition of cH4 domain at the C terminus
`(12). Finally, within species, classes and subclasses are
`characterized by subtle differences in C region sequences.
`The complement binding receptor and an N-glyco-
`sylated carbohydrate is located within the cH2 domain.
`This fortuitous location of the sugar substrate provides
`a unique functionality for regioselectively incorporating
`drugs outside of the antigen-binding region. The impor-
`tance of this linkage will be discussed later.
`IV. THE CHEMICAL DESIGN AND SYNTHESIS OF
`MODIFIED TARGETING AGENTS FOR ATTACHMENT
`TO MOAB’S
`In designing the bioconjugate, it is important to dis-
`tinguish the integrity of the three components of a bio-
`conjugate: the agent to be delivered, the covalent link-
`age, and the moab. The linkage must not diminish the
`biological activity of the modified agent nor compromise
`the moab’s ability to target. Two general strategies for
`conjugating an agent to a moab are represented in Fig-
`ure 6. In the first strategy, the modified agent is reacted
`directly with functional groups on the surface of the moab.
`In this process, the loading of the drug would be deter-
`mined by the number of available attachment sites on
`the moab. Alternatively, one can engraft the agent onto
`a matrix substrate and then react this unit with the moab.
`In the second process, the amount of agent delivered can
`be increased without having to increase the number of
`covalent bonds to the moab (13). Regardless of which
`strategy is chosen, the chemistry of bioconjugate con-
`struction must be guided by the following tenets: (1) the
`
`Drug:
`
`Linker for covalent attachment: 0
`Figure 7. Moab sites of agent attachment and chemical meth-
`ods for moab agent modification. Goals: (1) attach agent to
`moab without altering serological activity, (2) construct conju-
`gate with retention of biological properties of agent, (3) con-
`struct conjugate with linkage that facilitates release of free drug
`at target site, (4) design of agent/linkage compatible with cova-
`lent attachment to F(ab’), fragment, (5) construct conjugate
`with minimal immunogenicity, and (6) synthesis of conjugate
`amenable to large-scale production.
`
`attachment of agent to moab must be achieved without
`altering its immunoreactivity, (2) the conjugate must be
`constructed with a chemical linkage that will either allow
`the retention of the biologic properties of the agent or
`facilitate the release of the free drug at the target site,
`(3) the design of the agent/linkage chemistry should be
`compatible with covalent attachment to either the intact
`moab or its fragments, (4) the synthesis of the conjugate
`should be done in a manner that would minimize immu-
`nogenicity, and (5) the construction of the conjugate must
`be amenable to large-scale production (Figure 7).
`The moab is attractive as a targeting ligand because
`(1) it has many potential sites of drug attachment and
`(2) it can maintain immunoreactivity even as subfrag-
`ments. Represented in Figure 7 are the potential sites
`of covalent attachment on the intact moab and the pro-
`teolytic fragment, F(ab’), (14). The most accessible sites
`for drug attachment on the polypeptide chains are the e
`amino groups of the lysine residues (approximately 90
`lysines in a moab) and the carbohydrate moiety of the
`cH2 domain (15). One can imagine oxidizing the carbo-
`hydrate to generate aldehyde functions from the uic-di-
`ols, which can react with various drug functionality such
`as hydrazides (16,17). Drugs can be incorporated at the
`lysine residue through the construction of stable amide
`linkages. In addition, the lysine residues are important
`sites for drug attachment in the F(ab’), fragments since
`these have no carbohydrate.
`Having reviewed the required characteristics of the bio-
`conjugate and the sites of covalent attachment to the moab,
`it is appropriate to identify the representative chemical
`linkages that have been utilized in drug targeting. The
`succinate linkage has been employed in joining des-
`acetylvinblastine to the moab via an amide bond to the
`lysine amine (18,19). This, of course, discourages release
`of the free drug from the conjugate (see Figure 8, entry
`1).
`The sulfhydryl-bearing A chains of the toxins ricin, diph-
`theria toxin, and abrin have been joined to the moab via
`disulfide linkages. For example, the moab has been reacted
`with the N-hydroxysuccinimide ester of 4-(2-pyri-
`dyldithio) butyric acid to introduce several latent thiols
`onto moab lysines. This, in turn, is reacted with the sulf-
`hydryl-containing toxin to yield the moab conjugate as
`the disulfide. The lability of the disulfide bond and its
`resultant short half-life has encouraged the construction
`of hindered disulfides such as the a-methyl butyrate (see
`Figure 8, entry 2, R = CH,). This minor change in the
`linkage enhanced the circulation half-life of the corre-
`spondingly linked moab-ricin A significantly (20,21).
`
`IMMUNOGEN 2100, pg. 3
`Phigenix v. Immnogen
`IPR2014-00676
`
`

`
`16 Bioconjugafe Chem., Vol. 1, No. 1, 1990
`
`Linker
`
`4. MoAB-Carbohydrate {C=N-N-C Agent
`
`ai
`
`Agent
`Vinblastine
`
`Ricin A. Diphtheria toxin A,
`Abrin A
`
`Vinblastine hydrazide,
`Methotrexate hydrazide
`
`Anthracycline
`
`Indium and Yttrium chelates
`
`Metal chelates
`
`Anthracycline
`
`8.MoAB-carbohydrate C-N-N=F-Agent
`
`faH R
`
`Figure 8. Chemical linkages for covalent attachment of agent
`to ligand.
`
`An even more stable sulfhydryl-based moab/agent link-
`age has been achieved through formation of a thio-ether
`bond. The substrates ricin and alkaline phosphatase have
`been linked to moab in the following two-step proce-
`dure: (1) the lysine residue of either the substrate or the
`moab has been reacted with thiolane hydrochloride to
`give the 4-sulfhydrylbutyrimidate derivative and (2) sulf-
`hydrylbutyrimidate has been joined to the complemen-
`tary protein that bears a 4-(methylenema1eimido)cyclo-
`hexylcarboxamide via a conjugate addition of sulfur to
`maleimide (see Figure 8, entry 3) (22). The application
`of this linkage in the targeting of alkaline phosphatase
`for prodrug activation will be described in a later discus-
`sion (22).
`Investigators at Cytogen and Lilly Research Labora-
`tories have reported the oxidation of moab-carbohy-
`drate and use of the resultant aldehydes for linkage of
`drug hydrazides (16, 17). Cytogen scientists have pre-
`pared the methotrexate hydrazone conjugate and Lilly
`Research investigators have synthesized the vinblastine
`hydrazone conjugate (see Figure 8, entry 4). The vin-
`blastine hydrazone-moab conjugate will be described later
`in the review as an example of designing and developing
`conjugates in a structure-activity relationship based on
`human clinical feedback.
`The anthracyclines, exemplified by adriamycin and
`daunomycin, are a family of oncolytics that have been
`very challenging as candidates for conjugation. The need
`for the release of the free drug in order to express its
`DNA-binding activity has required the development of
`labile linkages compatible with the functionality of the
`anthracyclines. Reisfeld and others have constructed the
`acid-labile aconitate amide linkage through the lysine
`amine of the moab and the amine group of the glycoside
`of the anthracycline (see Figure 8, entry 5) (23).
`The importance of chemical linkers is most effectively
`highlighted in the targeting of radionuclides for both imag-
`ing and therapy. Historically, 12'1 and 1311 have been
`incorporated onto moab by iodination of the tyrosine res-
`idues (24). As imaging and therapy conjugates, the clin-
`ical data thus far have not been encouraging because of
`the rapid dehydrohalogenation of iodine. More recently,
`Meares has helped pioneer the construction of indium
`and yttrium conjugates through the development of nuclide
`chelate linkages (25). Investigators from Hybritech have
`utilized this technology in advancing the use of indium
`in imaging of solid tumors and then using the matched
`yttrium conjugate for therapy (see Figure 8, entry 6)
`(26). Most recently, investigators at NeoRx have reported
`
`Koppel
`
`\
`
`CHANGE MOA B:
`CHIMERIC, HUMANIZED,
`FRAGMENT, BIFU NCTlO NA L,
`MOLECULAR ENGINEERING
`
`I
`
`-Insrrl;I \
`
`CHANGE LINKER
`
`CHANGE DRUGS
`(POTENCY/ACTION)
`CLUSTER DRUGS,
`COMBINA TlON 0 F
`DRUGS
`Figure 9. Structure-activity relationship: modification for moab-
`drug conjugates.
`
`a radionuclide matched pair of technitium and rhenium
`for imaging and therapy, respectively. Thus, dosimetry
`data collected from the indium or technitium moab con-
`jugate has allowed the therapy with the corresponding
`yttrium or rhenium moab conjugate. This has resulted
`in the first clinical response of a human solid tumor with
`a rhenium conjugate (27).
`Offord and Rose at the University of Geneva as well
`as King at Rockefeller have reported the use of carbox-
`ybenzaldehyde as a linker for incorporating a ketone or
`aldehyde functionality onto the lysine amine of the pro-
`tein (see Figure 8, entry 7) (28, 29). Offord and Rose
`have reported the construction of hydrazone conjugates
`via this linkage (29).
`A complementary approach for modifying the carbo-
`hydrate aldehyde linkage has been one in which a
`hydrazide has been incorporated onto the moab for reac-
`tion with drugs containing carbonyl functions. Investi-
`gators at Cytogen have built in adipic dihydrazide link-
`age to join the oxidized carbohydrate of the moab to the
`ketone of an anthracycline via hydrazone linkages (30).
`Barton et al. at Lilly have utilized a reductive amination
`of the moab carbohydrate aldehyde with glutamic
`hydrazide and subsequently constructed a releasable
`attachment to an anthracycline via its ketone (see Fig-
`ure 8, entry 8) (31).
`As one begins to correlate the biological with the chem-
`ical components of a moab drug conjugate, it is clear that
`developing a medicinal chemistry structure-activity rela-
`tionship (SAR) becomes a multidimensional challenge
`(illustrated in Figure 9) (32). As has been emphasized
`in earlier portions of this review, there are a variety of
`opportunities for independent structural modifications
`of the conjugate that are expressed in the framework of
`the moab, the linker, and the drug.
`The quality of targeting of the conjugate to selective
`tissue can be achieved by changing either the specificity
`of the moab or simply its affinity (i.e., its on/off rate). If
`a human antibody response to the murine moab drug
`conjugate becomes a problem, the constant regions and
`the framework portion of the variable region can be
`replaced by human sequences through molecular engi-
`neering (8). In addition, if the intact moab with its atten-
`dant effector function domains such as complement bind-
`ing present an innocent tissue bystander liability, frag-
`ments of the moab such as F(ab'), that no longer carry
`these domains can be synthesized and conjugated with
`drug (see Figure 9) (14).
`The chemical linker is the heart of the conjugate. It
`determines the ability of the drug to express its activity
`either as an integral part of the conjugate or allows its
`release at a rate that is dictated by the chemistry. The
`
`IMMUNOGEN 2100, pg. 4
`Phigenix v. Immnogen
`IPR2014-00676
`
`

`
`Teaching Editorial
`choice of a releasable or nonreleasable linkage as it impacts
`on the drug toxicity will be discussed specifically with
`respect to the moab-vinca conjugate in the next portion
`of this review.
`V. T H E CHEMISTRY AND BIOLOGICAL ACTIVITY OF
`MOAB-DRUG CONJUGATES
`The oncolytic drug can be selected on the basis of a
`combination of its clinical effectiveness, its mode of action,
`and its potency against the tumor target. It is most impor-
`tant to understand that the parameters of biodistribu-
`tion and local drug concentration may dramatically change
`a drug's profile as a conjugate compared to its unmodi-
`fied form. As discussed before, the SAR that unfolds
`for a targeted drug is one that is multidimensional and
`dependent on a variety of structural changes, Le., moab,
`linker, and drug. The ability to enhance the quality of
`the conjugate depends on the transmission of informa-
`tion from the clinic to the laboratory and applying this
`information to the preclinical model (see Figure 9) (32).
`As an example of the design, the development, and
`the evaluation of a drug conjugate, I would like to describe
`a program that I have participated in, one that is gener-
`ically representative of the drug-targeting efforts occur-
`ring at the many academic and industrial institutions
`throughout the world today. In this example, the impor-
`tance of human clinical feedback is emphasized for adjust-
`ing the SAR and enhancing the efficacy of the drug con-
`jugate.
`Our investigators at Lilly chose a vinca alkaloid for
`the designing of a moab-drug conjugate, partly because
`of our vast prior experience in chemical modification of
`the vincas and partly because of its biological potency
`for treatment of human cancer (33). The vinca sub-
`strate for conjugation was prepared by the reaction of
`vinblastine with acid or base effecting the deesterifica-
`tion of the 4-position of the vindoline component. This,
`in turn, was reacted with succinic anhydride to afford
`the 4-succinate of desacetylvinblastine (DAVLB), a sub-
`strate bearing the nonreleasing linker, succinate (see
`Scheme I, parts a and b) (18,19). The choice of the suc-
`cinate linker was made, initially, in order to evaluate the
`inherent activity of the vinca in the context of the con-
`jugate, and to minimize the liability of free vinca.
`The moab selected for targeting was identified as
`KS1/4, a murine moab developed by Walker in Reisfeld's
`laboratory at Scripps (34). The moab KS1/4 recognizes
`a tumor/epithelial associated antigen (40 KD) found in
`high epitope density on human adenocarcinomas (35).
`The primary target of this moab is lung and colorectal
`adenocarcinoma (36).
`Pursuant to attaching the vinca to KS1/4, the DAVLB
`hemisuccinate was converted to its N-hydroxysuccin-
`imide active ester. This, in turn, was reacted with KS1/4
`in aqueous borate buffer in pH 8.6 at room temperature
`to afford, after a series of chromatographies, a 50% yield
`of the conjugate (see Scheme I, parts a-c) (16, 30, 45).
`Stoichiometric evaluation of the conjugate by ultravio-
`let spectrophotometry indicated a conjugation ratio of
`4-6 drugs/moab (36). This chemical process and the cor-
`responding biochemical purification proved to be reli-
`able, reproducible, and amenable to large-scale produc-
`tion (36). The analytical profile that was developed for
`monitoring the quality of KS1/4-DAVLB and all our other
`drug conjugates is illustrated in Figure 10.
`The difficulty with selecting a relevant in vivo animal
`model for evaluating the biological potential of a new
`moab-drug conjugate is that the targeting substrate rec-
`ognizes "human epitopes". Consequently, a model had
`
`Bioconjugate Chem., Vol. 1, No. 1, 1990 17
`
`bOOCH3
`
`Desacetyl Vinblastine
`
`Vinblastine
`
`b *
`
`PY
`
`Desacelyl Vlnblastlnr
`
`4Succinoyl Vinblastine
`
`KS 114
`P
`DM Fl0.34M Borate
`pH 8.6, R.T.
`
`!
`'qOCOCH2CH2CO-NH-lyS-
`COOCH3
`
`I
`
`(
`
`KS%
`
`to be created which would accommodate the pharmacol-
`ogy of the drug as part of a conjugate as well as the abil-
`ity of the conjugate to target the human tumor tissue.
`The in vivo experimental system which has been widely
`utilized in evaluating drug conjugates has been the "athy-
`mic nude mouse" xenograft model. The inbred athymic
`mouse does not express a thymus; consequently, it is immu-
`nologically impotent and unable to reject tissue grafts
`such as malignant tumors from another species (38). The
`in vivo nude mouse human xenograft models that have
`been selected for our moab-drug conjugate evaluations
`are described in Figure 11 (32).
`The traditional path for tracking antitumor potency
`of standard oncolytics has been in an in vitro potency
`assay measuring the ability of the agent to inhibit tumor
`growth. With respect to the drug conjugate KS1/4-
`DAVLB, which was constructed with a nonreleasing linker
`to minimize free drug, the in vitro assay measured the
`potency of the vinca in the context of the conjugate. Pre-
`dictably, the in vitro assay showed the conjugate to be
`of a low potency, one whose IC, was about 200-fold lower
`than that of vinblastine (Figure 12) (18b, 39). Clearly,
`the utility of targeting a low-potency conjugate could only
`be evaluated in an in vivo system, one that would allow
`the dramatic change in biodistribution to be expressed
`in tumor inhibition or regression.
`In an in vivo nude mouse tumor xenograft model, which
`measures the effect of an agent against a PSUCLA human
`
`IMMUNOGEN 2100, pg. 5
`Phigenix v. Immnogen
`IPR2014-00676
`
`

`
`18 Bioconjugafe Chem., Vol. 1, No. 1, 1990
`
`SDS-PAGE(Reducing and Nonreducing Conditions)
`
`lsoelectricfocusing Analysis
`
`2000
`
`(D
`
`FPLC Chromatography( Superosel2,MonoQ,MonoS)
`
`Functional ELISA- Immunoreactivity
`
`Functional Flow Cytometry- Immunoreactlvity
`
`I-
`
`500
`
`-
`-
`
`Control
`0-0 KS1/4.S2 255mglkg
`)--I KSlICS2.DAVLB (10mglkg)
`
`VLB 10mglkg
`
`Koppel
`
`0
`8
`
`Laser Nephelometry-Total Mouse lg
`
`In Vitro Potency Analysis
`
`0 1
`I
`I
`I
`30
`22
`34
`16
`Figure 13. Effect of KS1/4-DAVLB on the growth of estab-
`lished xenografts ( 3 9 ~ ) .
`
`Stability Studies
`Figure 10. Analytical profile of moab-drug conjugates.
`
`TUMOR INITIATION MODEL
`LOW TUMOR BURDEN-SUBCUTANEOUS TUMOR
`- -1-1-
`Rx DAYS 2.5.8 POST IMPLANT
`- -
`TUMOR MEASURMENTS DAYS i4,21,28
`
`ESTABLISHED TUMOR MODEL
`HIGH TUMOR BURDEN-SUBCUTANEOUS TUMOR
`VARIED Rx SCHEDULESilBD OR GREATER POST IMPLANT
`TUMOR MEASURMENTS FOR 3-6 WEEKS
`
`INTRAVENOUS EXPERIMENTAL METASTASES
`LOW TUMOR BURDEN-MULTI-ORGAN LESIONS
`VARIED Ax SCHEDULES
`SURVIVAL MONITORED
`
`OVCAR3 I.P. OVARIAN XENOGRAFT
`LOW-HIGH TUMOR BURDEN
`VARIED Rx SCHEDULES
`SURVIVAL MONITORED
`Figure 11. Prelinical pharmacology: nude mice efficacy mod-
`els.
`
`% Inhibition vs.
`Concentration
`(pglml)
`Drug
`Control
`VLB
`0
`,0001 5
`VLB
`80
`,0015
`VLB
`a7
`,015
`a
`KSl14-DAVLB
`.1
`3a
`KS114-DAVLB
`.25
`.5
`KS114-DAVLB
`71
`KSlI4-DAVLB
`84
`2.5
`Figure 12. In vitro potency analysis of vinblastine (VLB) and
`KS1/4-DAVLB (39u), as determined by a proliferation assay
`utilizing the P3AUCLA human lung adenocarcinoma cell line
`with a 72-h exposure to drug or monoclonal antibody-drug con-
`jugate. All concentrations are in terms of vinca alkaloid con-
`tent.
`
`lung adenocarcinoma cell line, the KS1/4-DAVLB con-
`jugate showed regression and inhibition of tumor growth
`at 10 and 5 mg/kg of vinca equivalent, respectively (see
`Figure 13) (39). This contrasts with the free drug vin-
`blastine, which at 10 mg/kg was toxic to the mice. More
`importantly, the free KS1/4 showed no activity as did
`an irrelevant moab-DAVLB conjugate (data not shown)
`(39). Moreover, the vinca alkaloid mechanism of action
`is to inhibit the mitotic cell division in the G,+VM phase
`by binding to tubulin and inhibiting its assembly (40).
`The KS1/4-DAVLB was equipotent to vinblastine with
`respect to tubulin binding in a cell-free tubulin assay car-
`ried out by Wilson (40).
`As had been suggested by various investigators in the
`field of targeting, efficacy with a low-potency drug con-
`jugate could be achieved if the concentration of drug in
`the tumor is enhanced by virtue of targeting. In order
`to more clearly define the mechanism of efficacy as well
`as establishing the “proof of targeting”, a nude mouse
`was implanted with a P3UCLA tumor and treated with
`[3H]DAVLB at 5 mg/kg and KSl/4-13H]DAVLB at 5
`
`20 -
`
`15-
`
`10-
`
`KSll4.DAVLB (5 mglkg DAVLB)
`0 DAVLB (5 mglkg)
`
`Hours After Dosing
`Figure 14. Analysis of the biodistribution of tumor tissue of
`[3H]DAVLB and KSl/4-[3H]DAVLB in P3UCLA human lung
`zenograft bearing nude mica (41). The animals were sacrificed
`at the time points indicated for both free drug and conjugate-
`treated animals.
`
`mg/kg of vinca equivalent. After 96 h postinjection, up
`to 7 4 % of the total administered dose was found in the
`tumor tissue after dosing with KS1/4-DAVLB. In con-
`trast, less than 0.3% of the dose was found in tumor tis-
`sue after dosing with free DAVLB (see Figure 14) (41).
`After completing the preclinical biological evaluations
`in the mouse xenograft models and the preclinical toxi-
`cology evaluations in rodents and primates, the KS1/4-
`DAVLB conjugate was evaluated in the clinic. Twenty-
`two patients received KS1/4-DAVLB in a phase I trial.
`Of those, 13 received single iv infusions of conjugate at
`doses ranging from 40 to 240 mg. Nine patients received
`multiple iv infusions at a dose of 1.5 mg/kg every 2-3
`days for up to nine doses (42).
`A surprising and unpredicted dose-limiting duodenal
`toxicity presented itself at 250 mg during single-dose stud-
`ies and at a cummulative dose of 400 mg during multi-
`ple-dose studies. The toxicity was reversible and no resid-
`ual damage to the duodenum was observed. The toxic-
`ity did not appear to be vinca related. Studies by Schneck,
`Petersen, and Zimmerman suggested complement depo-
`sition in the inflamed duodenal tissue (43). Further stud-
`ies are underway to more clearly define the mechanism
`of