1 1 8
`
`Applications and Engineering of Monoclonal Antibodies
`
`of PMN recruitment. to the site. Up-regulation of the selectins allows rolling of the Pl\1N
`along the endothelial cells and up-regulation ofintercellular adhesion molecule-I (ICAM-
`1) allows the PMN to become firmly attached to the endothelial cell and then to migrate
`into the interstitium. Imaging sites of inflammation by antibodies to these endothelial cell
`adhesion molecules may therefore give additional selectivity. 1 1 1In-labelled antibodies to
`both E-selectin and ICAM- 1 have been investigated in animal models with promising
`results (Keelan et al., 1 994; Sasso et al. , 1 9 96).
`The abundance of CD4 on inflammatory cells present in arthritic joints has led to
`interest in targeting CD4 to image arthritic joints. Antibodies to CD4 have been used to
`assess targeting in animal models of arthritis. Design of the reagent is crucial for success,
`as intact IgG demonstrated no benefit over a non-specific antibody, whereas a 99mTc­
`labelled Fab' led to improved imaging (Kinne et al. , 1 995). Anti-CD4 reagents may also
`be useful in monitoring the distribution of CD4 positive lymphocytes, which could be
`useful in assessing patients with HIV infection (Rubin et al., 1 996). RAID has also been
`investigated for detection of atherosc!erotic plaques using an antibody towards proliferat­
`ing smooth muscle cells (Narula et al., 19 95) and lesions in Alzheimer's disease using a
`cationised antibody towards beta A4 protein (Bickel et al., 1 994) . Cationisation through
`chemical modification of the antibody with charged groups is believed to aid transport of
`antibodies through the blood:brain barrier, which is obviously helpful to image brain
`lesions in Alzheimer' s disease.
`
`HOSPIRA EX. 1029
`Page 1 26
`
`
`
`126 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`4
`
`Monoclonal Antibodies in
`Therapeutic Applications
`
`4.1
`
`I ntroduction
`
`The idea of targeted therapy using antibodies dates from the beginning of the twentieth
`century when it was proposed by Paul Ehrlich, and indeed antibodies have been used in
`humans for many years as polyclonal antisera, particularly for passive immunisation.
`Since the discovery of the technology for production of MAbs there has been intense
`interest in their development as therapeutic agents for human disease. However, until
`recently progression of many MAb-based therapeutics was limited by the human immune
`response generated by the administration of murine or rat antibodies. Since the introduc­
`tion ofrecombinant chimeric and humanised antibodies, and ultimately human antibodies,
`interest in developing therapeutics has been revived. There are now five antibody-based
`products licensed for human therapeutic use, and many more under clinical investigation
`(Table 4. 1).
`MAbs can be used in several different modes, depending on the required therapeutic
`effect. In some cases a simple blocking or neutralising effect may be required, for example
`in the neutralisation of an inflammatory cytokine or the blocking of a specific receptor.
`In other applications therapy may require an active role for the antibody by the targeting
`of an effector function. Therapeutics can utilise natural antibody effector functions such
`as complement activation, phagocytosis or ADCC, or entirely novel functions can be
`introduced, as in the case of targeting radioisotopes, drugs or toxins to kill tumour cells.
`In some cases antibodies may also find a role through their ability to mediate signal
`transduction from binding to cell surface receptors (Vitetta and Uhr, 1 994). Available
`mechanisms for generation of therapeutic effects can be considered as a range of anti­
`body effector functions (Table 4.2). Many of these effector functions can be designed
`into, or out of, antibody molecules for specific therapeutic purposes as described in
`Chapter 2. In this chapter the basis of targeted therapy is illustrated using examples of
`the various modes of antibody-based therapy in different disease states. The emphasis is
`placed on the design of antibody-based therapeutics rather than a catalogue of antibodies
`used in therapeutic studies.
`
`HOSPIRA EX. 1 029
`Page 1 27
`
`
`
`127 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`Table 4.1 Com mercially ava i lable monoclonal antibody therapeutics
`
`1 20
`
`Applications and Engineering of Monoclonal A ntibodies
`
`Therapeutic agent Antibody
`
`Disease
`
`Orthoclone OKT3 Murine OKT3
`
`ReoPro
`
`Chimeric 7E3 Fab
`
`Acute transplant rej ection
`(kidney, liver and heart)
`
`Complications of post-coronary
`angioplasty
`
`Panorex
`
`Rituxan
`
`Zenapax
`
`Table 4.2
`
`Murine 1 7- l A
`
`Colorectal cancer
`
`Chimeric 288
`
`Lymphoma
`
`Humanised anti-tac Acute kidney transplant rejection
`
`Effector functions for antibody targeted therapy
`
`Company
`
`Ortho Biotech
`
`Centocor
`
`Centocor
`
`IDEC/Genentech
`
`Protein Design
`Labs/ Roche
`
`Blocking/neutralising
`
`Natural - Fe mediated effects
`
`Cell signalling
`
`Natural inunune responses
`
`Artificial effectors
`
`B ifunctional
`
`4.2 Cancer
`
`Antigen binding
`
`Complement fixation
`ADCC
`Phagocytosis
`
`Receptor cross-linking
`
`Generation of anti-idiotype response
`Other 'vaccination' approaches
`
`Radioisotopes
`Toxins - bacterial
`- plant
`Cytotoxic drugs
`Cytokines
`Enzymes - prodrug activation
`- direct toxicity
`
`Cross-linking cytotoxic effector cells
`Two-step targeting strategies for
`radioisotopes, toxins, etc.
`
`One of the maj or targets of antibody-based therapeutics has been in the development of
`anti-cancer agents. Initially MAbs to tumour-associated antigens were raised and invest­
`igated without modification, in attempts to target natural antibody effector mechanisms to
`tumour cells. Results of such studies were initially disappointing in many cases, particu­
`larly with the conunon solid tumour types. However, beneficial effects were seen when
`treating patients with minimal disease and in treating lymphomas and leukemias, suggest­
`ing that further research to investigate the role of antibodies as anti-cancer agents may
`lead to improved results. In addition, the use of antibodies to target traditional anti-cancer
`agents such as cytotoxic drugs and radioisotopes has made steady progress, particularly
`since the introduction of recombinant antibodies and improved methods of coupling
`
`HOSPIRA EX. 1 029
`Page 128
`
`
`
`128 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`Monoclonal antibodies in therapeutic applications
`
`1 2 1
`
`cytotoxic agents (Chapter 2). A s described in Chapter 3, many tumour-associated anti­
`gens representative of different tumour types are now characterised (see Table 3 .5) which
`allow therapeutic molecules to be designed.
`
`4.2.1 Cancer therapy with unmodifed (naked) antibodies
`
`The use of umnodified, or naked, antibodies for tumour therapy initially resulted from
`attempts to harness the immune system through the natural antibody effector fonctions,
`such as ADCC or complement-mediated lysis. More recently it has become apparent that
`cell signalling mechanisms may be involved in many of the anti-tumour effects observed,
`through arrest of the cell cycle or inducing apoptosis. Design of antibodies to elicit
`natural effector functions has largely been a process of choosing the optimal isotype to
`harness human ADCC and complement effects, although as described in Section 2.6. 1 the
`selection of a suitable isotype alone is not sufficient to ensure good cell killing, and
`suitable antigen targets must be individually tested. Mouse IgG2a is the best of the
`murine isotypes for eliciting human ADCC, with rat IgG2b also potent. Of the human
`isotypes, IgG 1 and IgG3 are the most potent in cell killing studies. Of these IgG 1 is most
`commonly used for construction of humanised antibodies, partly because IgG3 antibodies
`are more difficult to purify and to handle in vitro due to a tendency to aggregate. The
`preparation of chimeric or humanised antibodies from rodent antibodies thus has the dual
`benefit of reduced immunogenicity and the ability to use a constant region best suited to
`recruitment of human effector functions. Similarly human antibodies isolated from phage
`display, or other means, can be reconstructed and expressed with the desired isotype
`constant regions. Expression of the antibody in mammalian cells is required as the
`glycosylation of the Cm region is required for maintainence of the ability to elicit
`effector functions. However, cell killing effects are dependent not only on the constant
`region, but also on the disposition of antigenic sites and other poorly understood mech­
`anisms. Many cells also have protective mechanisms against attack from the immune
`system, and thus the in vivo effects of antibodies capable of eliciting cell killing in vitro
`are often difficult or impossible to predict.
`Several alternative mechanisms have also been suggested for tumour cell killing by
`umnodified antibodies. Antibodies against cell surface markers on many types of tumour
`cell can act as ligands eliciting anti-tumour effects by signal transduction (Vitetta and
`Uhr, 1 9 94). Signal transduction can result in arrest of the cell cycle, hence preventing
`tumour cell growth, or in some cases the induction of programmed cell death known as
`apoptosis. In addition, antibodies to growth factors or their receptors may exert anti­
`tumour effects through blocking binding of growth factors needed for tumour cell growth.
`Growth factors such as EGF (epidern1al growth factor) and IL-6 have been implicated in
`the growth of a number of tumour types, and target tumour cells overexpress large
`numbers of molecules of the growth factor receptors. Receptor blocking antibodies pre­
`vent interaction with the ligand, and can lead to down-regulation of the number of
`receptor molecules present on the tumour cell such that growth is inhibited. However,
`unravelling the contribution of each of these factors to effective tumour cell killing is not
`straightforward and often several mechanisms may operate concurrently. Future research
`may allow the selection of antibodies based on an improved understanding of the import­
`ance of eliciting each of these anti-tumour effects. For example, it may be possible to
`select combinations of MAbs capable of different mechanisms of cell killing.
`
`I
`
`HOSPIRA EX. 1 029
`Page 1 29
`
`
`
`129 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`1 22
`
`Applications and Engineering of Monoclonal Antibodies
`
`As described in Chapter 2, dimeric or polymeric antibodies have been produced either
`by chemical cross-linking or by recombinant means which have increased activity for
`both complement activation and ADCC. Chemically cross-linked constructs comprising
`Fab fragments linked to two Fe regions FabFc2 have also been produced in attempts to
`increase therapeutic efficacy through effector functions. However, studies comparing
`such constructs with a humanised antibody for immunotherapy of multiple myeloma
`showed no advantage in complement activation or ADCC (Ellis et al., 1 995).
`Homodimerisation of IgG has also been reported to be able to confer signalling activ­
`ity, leading to cell cycle arrest or apoptosis, to antibodies which are unable to do so in
`monomeric form (Ghctie et al., I 997a). Chemically cross-linked IgG homodimers were
`produced for several antibodies and shown to be capable of specific anti-tumour effects,
`presumably due to cross-linking the antigen and/or a lower dissociation rate. In the case
`of an anti-CD 1 9 antibody the Fe region was not required for anti-tumour activity, leading
`to the possibility that small multimeric fragments may be capable of potent anti-tumour
`effects.
`Although many studies with unmodified antibodies for cancer therapy have been
`disappointing, with no evidence of anti-tumour effects, several antibodies have shown
`promising effects in the clinic. It is only in the past few years that recombinant chimeric
`and humanised antibodies have reached clinical evaluation. In these studies it has been
`apparent that the ability of the human constant regions to interact with the human im­
`mune system plus the ability to re-treat without the generation of a prohibitive immune
`response is leading to more effective therapeutic agents.
`The first humanised (CDR-graftcd) antibody to be used clinically, CAMPATH-I H,
`recognises the CAMPATH-1 antigen, also known as CDw52, which is present on most
`human lymphocytes and monocytes but not stem cells. CAMPATH- l H has been invest­
`igated as a potential therapeutic agent in non-Hodgkin lymphoma (B-ce!l lymphoma) as
`well as in several inflammatory diseases (Section 4.5.3). Although only limited studies
`have been reported, use of CAMPA TH-I H for the treatment of B-ccll lymphoma has Jed
`to tumour remissions in some patients (Hale et al., 1 988). Studies with different isotypes
`of the rat parent antibody suggest that human effector cells arc involved, as the IgM and
`rat IgG2a versions, which can activate complement but not ADCC, led to only transient
`falls in blood cell counts. In contrast, the rat IgG2b version which could bind to and
`
`activate human effector cells led to much more efficient depletion of tumour cells in vivo
`
`(Dyer et al., 1 989).
`Another antibody, which has shown promising effects in B-cell lymphoma, recognises
`CD20, a phosphoprotein present on the surface of 8 cells. A chimeric IgG 1 version of the
`antibody 288, tenned C288, has been shown in in vitro assays to be an effective media­
`tor of both complement-mediated effects and ADCC using human effector cells (Reff
`et al., 1 994). In addition, this antibody has been shown to induce transmembrane signal­
`ling leading to cell cycle arrest and occasionally apoptosis of CD20 positive cells, and
`thus several mechanisms may be important in the anti-tumour effects observed (Demiden
`et al., 1 995). Phase II and III clinical studies for relapsed B-cell lymphoma have shown
`tumour shrinkage of 50% or more in approximately half of the patients studied (Maloney
`et al., 1 997). In addition, C2B8 was found to sensitise resistant lymphoma cells to certain
`cytotoxic drugs, which may lead to synergistic effects in combination therapy. In a phase
`II clinical study of C2B8 with standard chemotherapeutic treatment, responses were seen
`in all patients studied (Czuczman et al., 1 995). Although as a single agent, tumour
`responses were not as impressive as those observed with radiolabelled antibodies to Co20
`(Section 4.2.5), treatment with this unmodified antibody was much less toxic and is more
`
`HOSPIRA EX. 1 029
`Page 1 30
`
`
`
`130 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`lvfonoclonal antibodies in therapeutic applications
`
`1 23
`
`straightforward than handling large amounts of radioisotope. Therefore, it is likely that this
`agent will find clinical application alongside chemotherapy for B-cell lymphoma treatment.
`The HER2 gene (also known as neu or c-erb-B2) is a proto-oncogcnc which encodes
`a transmembrane growth factor receptor (pl85 HER2) which is ovcrexprcsscd in 25-30% of
`patients with breast cancer, particularly those with a poor prognosis. A humanised anti­
`body to p l 85 11ER1 has been developed which has shown significant growth inhibition of
`cells overexpressing this receptor (Carter et al., 1 992b). The humanised antibody was
`also shown to be more efficient at mediating ADCC with human effector cells than the
`parent murine antibody. In a phase II clinical study in patients with metastatic breast
`cancer that overcxpressed HER2, objective responses were seen in 5 of 43 patients
`including one complete remission and 4 partial remissions (Baselga et al., 1 996). The
`mechanisms of action of this agent are also not clear at present. The antibody induced a
`clear down-regulation of the growth factor receptor which may reverse the malignant
`phenotype. This antibody is also known to be capable of activation of a signal transduction
`pathway that leads to inhibition of tumour cell proliferation and possibly cell death, and
`also elicits ADCC.
`
`4.2.2 Anti-idiotype antibodies
`
`MAbs may also be used as vaccines to generate an anti-idiotype response. The idotype of
`an antibody is made up of a cluster of cpitopes at the antigen-binding site. As each MAb
`has a different antigen-binding site, the idiotype of each MAb is unique. Antibodies can
`be raised through an immune response to the idiotype which have a conformation that
`resembles the antigen of the original antibody (Figure 4.1 ). These anti-idiotype antibodies
`arc also known as anti-id or ab2 antibodies. Immunisation with an ab2 antibody can in
`tum lead to the elicitation of an anti-idiotype to the ab2 antibody, an anti-anti-id, known
`as ab3, a proportion of which will have the same or overlapping specificity as the original
`ab 1 antibody (Figure 4. 1 ). Structural studies suggest that anti-idiotype antibodies carry an
`' internal image' of the original antigen which allows the development of ab3 responses
`that recognise the original antigen. The crystallographic structure of an anti-idiotype
`antibody in complex to an antibody raised against lysozyme showed that the same antigen­
`binding residues used to bind lysozyme were used for binding to the ab2, and that the ab2
`mimicked lysozyme (Fields et al., 1 995).
`Vaccination with ab2 has been investigated in attempts to raise a human ab3 response
`to target antigens which may be poorly immunogenic or difficult to prepare or use in
`immunisation. ab2 may also be able to raise antibodies not raised to the original antigen
`through breaking immunological tolerance and also give rise to T cell responses which
`may have therapeutic significance. Many studies, both preclinical and clinical, have been
`conducted in attempts to achieve therapeutic effects with ab2 vaccines for cancer (re­
`viewed by Herlyn et al., 1 996). In several cases antigen-specific immune responses have
`been achieved, including both antibody and cellular responses. For example, the antibody
`1 0 5AD7 is an anti-idiotype antibody that mimics a colorectal tumour-associated antigen.
`Phase I clinical studies demonstrated that a T cell response to tumour was induced on
`treatment with I 05AD7 and this resulted in delayed tumour growth and an increase in
`survival time compared with patients at the same stage of disease that did not receive the
`antibody (Buckley et al., 1 995).
`It is, of course, also possible for antibodies to tumour-associated antigens administered
`for tumour therapy (ab l ) to elicit anti-idiotype responses. The murine MAb 1 7- l A has
`
`HOSPIRA EX. 1 029
`Page 1 3 1
`
`
`
`131 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`1 24
`
`Applications and Engineering of lvfonoclonal Antibodies
`
`(a)
`
`(b)
`
`(c)
`
`Figure 4.1 Representation of anti-idiotype antibodies: (a) in itial antibodies (abl ) recognise
`
`the antigen to which they were raised; (b) anti-id iotype antibod ies (ab2) recognise the
`antigen-b inding region of ab1 and thus can be considered as contain ing an 'internal
`image' of the antigen; (c) anti-anti-idiotype anti bodies (ab3) raised against ab2 have
`similar specificity to the original antibody (ab 1 ) and can thus recogn ise the same antigen
`
`been extensively studied. It recognises the epithelial membrane antigen (EMA) present
`on the majority of colorectal tumour cells. Originally developed for its ability as a murine
`IgG2a to elicit ADCC in humans, early clinical studies revealed occasional responses in
`patients with advanced disease which were delayed following antibody administration,
`implying an active immune response. Most patients developed an anti-idiotype response
`to 1 7-lA which may have led to the observed therapeutic effects (Fagerberg et al., 1 996).
`In a phase III trial of 1 7-l A for the treatment of minimal residual disease in colorectal
`cancer patients following surgical removal of the primary tumour, 1 7- 1 A treatment re­
`sulted in a statistically significant improval in survival (Riethmuller et al., 1 994 ). After
`following patients for five years, 17-1 A therapy reduced the overall death rate by 30%
`and decreased the recurrence rate by 27%. These results led to the approval of 1 7- l A as
`a therapeutic agent, sold under the name ' Panorex' (Table 4. 1). Analysis has identified
`patients with high levels of ab3 as those that live longer, suggesting that the induction of
`an anti-idiotype response is important in the mechanism of the therapeutic effect of this
`antibody (Fagerberg et al., 1 996).
`
`HOSPIRA EX. 1029
`Page 1 32
`
`
`
`132 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`Monoclonal antibodies in therapeutic applications
`
`1 25
`
`(a)
`
`(b)
`
`figure 4.2 Mechanisms of targeting effector cells to tumours: (a) act ive isotypes which
`can bind to Fe receptors on the effector cel l can be used to target the cel l to a tumou r­
`
`associated antigen; (b) alternatively, bispeci fic a ntibodies may be used which bind to both
`the tumour cell and the effector cel l through the antigen-binding arms of the antibody
`(for further details see text)
`
`4.2.3 Bispecific antibody-mediated effector cell targeting
`
`The limited success that has been achieved in using antibodies to elicit ADCC responses
`has prompted the search for other approaches to recruit effector cells, such as cytotoxic T
`cells, to the tumour site. One such approach is to use bispecific antibodies with one
`specificity for a tumour-associated antigen and one for the effector cell such that the
`effector cell is linked directly to the tumour cell (Figure 4.2). A surface 'trigger molecule'
`is used as the antigen on the effector cell such that the cell is activated to lead to a
`cytotoxic response. Several different types of effector cell and trigger molecule have been
`investigated, including Fey receptors and molecules of the CD3ff cell receptor complex
`(Table 4.3). The advantages and disadvantages of each of these have been reviewed
`(Renner and Pfreundschuh, 1 995). In addition to requiring triggering, cytotoxic effector
`mechanisms may also require the involvement of accessory molecules and activation via
`cytokine release.
`On monocytes, macrophages, eosinophils and polymorphonuclear neutrophils (PMNs),
`activation is achieved through linking to Fey receptors. Natural killer cells are activated
`through FcyRIII and CD2, whereas activation of T cells is usually achieved through the
`
`I
`
`HOSPIRA EX. I 029
`Page 1 33
`
`
`
`133 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`1 26
`
`Table 4.3 Cytotoxic effector cells and trigger molecules
`
`Applications and Engineering of Monoclonal Antibodies
`
`Effector cell
`
`T cells
`
`Natural killer cells
`
`Monocytes/macrophages
`
`Granulocytes ( eosinophils, PMNs)
`
`PMNs activated with interferon-y,
`GM-CSF or G-CSF
`
`Trigger molecules
`
`CD3
`T cell receptor (y/o)
`CD2
`CD28
`
`Fey receptor III (CD 1 6)
`CD2
`
`F cy receptor I ( CD64)
`Fey receptor II (CD32)
`F cy receptor III (CD 1 6)
`
`Fey receptor II (CD32)
`
`Fey receptor I (CD64)
`
`CD3/T-cell receptor complex. The CD3/T-cell receptor complex is normally responsible
`for antigen-specific T cell responses, but the use of bispecific antibodies allows activation
`through this complex which can be redirected to the antigen recognised by the other arm
`of the bispecific antibody. Activation of effector cells is greatly increased by cytokines
`such as interleukin-2 or interlcukin-7, or by co-stimulatory signals. CD2 and CD28 arc
`co-stimulatory molecules which when stimulated increase the activation of T cells. For
`example, combinations of triggering CD3 and CD28 allow particularly potent anti­
`tumour effects. Cell killing by a bispecific antibody cross-linking the Hodgkin lymphoma
`antigen CD30 to CD3 resulted in relatively little tumour cell killing unless co-administered
`with CF30-CD28 bispecific antibody (Renner et al., 1994). The combination was also far
`more potent in treatment of xenografted human tumour in immunodeficient mice
`repopulated with human T cells. Cures of established tumours could only be achieved
`with the combination of both bispecific molecules (Renner et al., 1 994 ). Similar effects
`can be achieved by co-administration of anti-CD28 IgG with anti-CD3/anti-tumour
`
`bispecific antibodies (Demanet et al., 1996). An alternative approach has been the use of
`interesting results, although these have been limited by the need to activate T cells ex vivo
`
`a trispecific antibody, made by cross-linking three Fab' fragments, with specificity for
`tumour antigen as well as two different T cell antigens (Tutt et al., 1 99 1 b ).
`Several phase I clinical studies have been carried out with bispecific antibodies with
`
`and re-infuse to the patient at the time of bispecific antibody therapy. IL-2 activated cells
`targeted by an anti-CD3/anti-tumour bispecific antibody administered intracranially to
`glioma patients resulted in survival of 76% of patients over two years, compared to 33%
`of those treated with activated cells alone (Nitta et al., 1990). Anti-FcyRIII/anti-tumour
`antibodies designed to activate natural killer cells and mononuclear phagocytic cells have
`also been used clinically, with minor responses observed (Weiner et al., 1 995).
`The production of bispecific antibodies can be achieved using a variety of approaches,
`including the production of hybrid hybridomas, chemical cross-linking and recombinant
`approaches (Section 2.5). The affinity of the antibody for the triggering molecule may
`also be important. Humanised Fab' fragments to the breast tumour antigen pl 85HER2 and
`
`HOSPIRA EX. 1029
`Page 1 34
`
`
`
`134 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`Monoclonal antibodies in therapeutic applications
`
`127
`
`to CD3 have been expressed and cross-linked to form a bispecific antibody capable of
`efficient lysis of tumour cells (Zhu et al. , 1 995). In these studies a high affinity version of
`the humanised anti-CD3 Fab' was found to be more effective at tumour cell lysis than a
`lower affinity version. The same antibody specificities have been used to produce a
`bispecific diabody which was equally active in tumour cell lysis but was much simpler to
`produce as it was directly expressed by secretion from E. coli at high yield (Zhu et al.,
`1 996). Under some circumstances diabodies may be more effective than bispecific IgG at
`achieving tumour cell lysis, possibly due to the smaller size of the diabody bringing
`target and effector cell closer together (Holliger et al., 1 996).
`
`4.2.4 Other approaches to recruit the immune system using MAbs
`
`Superantigens are bacterial or viral proteins, such as the exotoxin produced by
`Staphylococcus aureus, that stimulate T cells by cross-linking the T cell receptor to MHC
`class II molecules. Superantigens bind to the T cell receptor outside the antigen recogni­
`tion site and to MHC class II molecules outside the peptide binding groove normally
`involved in antigenic recognition, bypassing normal immunological specificity. Super­
`antigens are thus the most potent known activators of T cells. Fab'-superantigen fusion
`proteins have been produced which stimulate T cells and target to tumour cells. The
`fusion proteins are intended to replace MHC class II binding with tumour cell binding
`through the Fab'. Such fusion proteins have been shown to be capable of inhibiting
`tumour growth in animal models through causing activated T cells to infiltrate and attack
`the tumour (Dohlstein et al., 1995). One problem with the use of such agents, however,
`is residual MHC class II binding which leads to relatively high toxicity through systemic
`T cell activation and accumulation of inflammatory cytokines in serum. Attempts to
`improve Fab'-superantigen fusion proteins for tumour therapy are thus being made by
`mutations in the superantigen binding site for MHC class II which prevent binding and
`reduce systemic toxicity (Hansson et al., 1 997).
`Other approaches in attempts to improve the recruitment of the immune system by
`antibodies have included the use of antibody-cytokine fusion proteins. Certain cytokines,
`such as inter!eukin-2 (IL-2) are able to activate multiple- immune mechanisms, although
`attempts at tumour therapy with cytokine alone have led to severe systemic toxicity. Fusion
`proteins have thus been designed to target a high local concentration of cytokine to the
`tumour site, minimising systemic toxicity. Several different cytokines have been investig­
`ated. IL-2 stimulates T cell proliferation and T cell mediated killing and has thus been
`
`widely investigated, but tumour necrosis factor a (TNFa), TNF�, GM-CSF, IL-5, IL-8
`and interferon a have also been tested. Fusion proteins have been produced with several
`
`antibodies and encouraging pre-clinical results obtained. Chimeric 14. 1 8, an antibody that
`recognises the GD2 ganglioside expressed on neuroblastoma, melanoma and certain other
`tumours, has been used to produce fusions with several cytokines, the most promising of
`which appears to be IL-2 (Hank et al., 1 996). The fusion protein maintained antigen bind­
`ing and IL-2 activity and was able to exert anti-tumour effects in mice bearing human
`tumour xenografts of neuroblastoma and melanoma. Activation of human effector cells
`has also been demonstrated (Hank et al., 1 996). Cytokine fusion proteins have also been
`produced with small antibody fragments such as scFv, which may allow more effective
`tumour penetration (Dorai et al., 1 994 ). However, whether such fusion proteins will be
`sufficiently less toxic than free IL-2 when administered to patients remains to be seen.
`
`HOSPIRA EX. 1 029
`Page 1 3 5
`
`
`
`135 of 254
`
`Celltrion, Inc. 1195
`Celltrion v. Genentech
`IPR2017-01374
`
`

`

`1 28
`
`Applications and Engineering of Monoclonal Antibodies
`
`4.2.5 Radioimmunotherapy
`As with diagnostic tumour imaging (Section 3.9. 1), the design of antibodies for
`radioimmunotherapy (RIT) requires careful consideration of the individual components
`of the immunoconjugate, including the fom1 of the antibody, the radioisotope and the
`means of attachment of the radioisotope to the antibody. In addition, the choice of a
`suitable tumour-associated antigen becomes more critical with therapeutic doses of iso­
`tope, as the presence of antigen on normal tissues will result in the deposition of activity,
`which may lead to unacceptable toxicity. Many factors influence the degree of tumour
`localisation in patients. These include heterogeneous expression of antigen, presence of
`circulating antigen, tumour vascularisation and penetration of antibody into tumour tissue
`(Boxer et al., 1992). However, there may not be a need to achieve localisation to each
`individual tumour cell, as a radioisotope can be chosen with a pathlength of many cell
`diameters, allowing for 'bystander cell' killing at the tumour site. Nevertheless, a high
`proportion of tumour cells expressing the antigen is desirable. This is because the major
`limitation to RIT is the dose of isotope which can be delivered at the tumour site without
`resulting in unacceptable toxicity. In general, only a small proportion of the dose admin­
`istered localises to the tumour site in patients. Typically tumour-localisation levels of
`1 0% injected dose per kg of tumour (0.0 1 % injected dose per gram of tumour) are
`observed in human studies where good targeting is achieved (van Hof et al., 1 996). The
`total administered dose is usually dependent on the toxicity to normal body tissues. Bone
`marrow is the most sensitive normal tissue to radiation and therefore doses are usually
`limited by the dose to bone marrow from radiolabelled antibody circulating in the blood
`(Badger, 1990). Several strategies have been investigated in attempts to reduce bone
`marrow toxicity, including the use of short-range radioisotopes which require cell inter­
`nalisation for cytotoxicity (Auger emitters), the use of specific clearing mechanisms to
`remove circulating activity, two-step targeting strategies and the use of rapidly clearing
`antibody fragments.
`Careful optimisation of an immunoconjugate for RIT, through antibody engineering
`and development of suitable chemistry for radioisotope attachment, can lead to improved
`properties both in vitro and in vivo (King et al., 1 994), and it is likely that such optimised
`immunoconjugates will provide the next generation of molecules of clinical utility for
`RIT. Some of the parameters which need to