R E V I E W S
`
`IMPROVING THE EFFICACY OF
`ANTIBODY-BASED CANCER
`THERAPIES
`
`Paul Carter
`
`A quarter of a century after their advent, monoclonal antibodies have become the most rapidly
`expanding class of pharmaceuticals for treating a wide variety of human diseases, including
`cancer. Although antibodies have yet to achieve the ultimate goal of curing cancer, many
`innovative approaches stand poised to improve the efficacy of antibody-based therapies.
`
`PHAGE DISPLAY
`Technology for displaying a
`protein (or peptide) on the
`surface of a bacteriophage,
`which contains the gene(s)
`that encodes the displayed
`protein(s), thereby physically
`linking the genotype and
`phenotype.
`
`VALENCY
`For antibody-derived
`molecules, this refers to the
`number of binding sites for the
`cognate antigen(s).
`
`COMPLETE RESPONSE
`No remaining tumour can be
`detected by visual inspection or
`by clinical imaging technologies.
`This does not mean that the
`disease has been cured.
`
`PARTIAL RESPONSE
`≥ 50% reduction in tumour with
`no new lesions or increase in size
`of an existing lesion.
`
`Immunex, 51 University
`Street, Seattle, Washington
`98101, USA.
`e-mail:
`carterp@immunex.com
`
`Antibodies are finally realizing their potential as anti-
`cancer therapeutics: since 1995, five antibodies have
`been approved for the treatment of cancer (TABLE 1).
`Additional approvals will surely follow from among
`the 20 or so antibodies now in oncology trials1,2,
`including 10 that have advanced to Phase III trials or
`further (TABLE 1).
`The emergence of antibodies as therapeutics was
`made possible by the advent of technologies designed to
`overcome the main limitations of mouse monoclonal
`antibodies (mAb) — immunogenicity of these foreign
`proteins in patients, inefficient effector functions (see
`below) and half-lives that are typically less than 20
`hours1–4. These core technologies, in historical order of
`development, are chimerization and humanization of
`mouse antibodies, and direct routes to high-affinity
`human antibodies using PHAGE DISPLAY libraries or trans-
`genic mice (BOX 1). Beyond these core technologies,
`remarkable progress has been made in engineering anti-
`bodies with modified properties — for example, molecu-
`lar size, antigen-binding affinity, specificity and VALENCY1,3–5.
`Tumour targeting by antibodies with engineered prop-
`erties is in its infancy, but holds much promise for
`enhancing the antitumour activity of antibodies (BOX 2).
`Important advances in antibody technologies
`notwithstanding (BOXES 1 and 2), patient tumour-
`response data show the urgent need to enhance the effi-
`cacy of the current generation of anticancer antibodies.
`For example, in a Phase II study of the chimeric anti-
`CD20 antibody6 rituximab (Rituxan), in patients with
`
`relapsed low-grade non-Hodgkin’s lymphoma, only
`about half of the patients responded7. This included
`6% COMPLETE and 42% PARTIAL RESPONSES from 166
`patients, similar results to those obtained with a single-
`agent cytotoxic chemotherapeutic in this group of
`patients. These data, combined with the mild toxicity
`profile of Rituxan, led the United States Food and Drug
`Administration (FDA) to approve Rituxan for relapsed
`indolent lymphoma. Unfortunately, the median
`RESPONSE DURATION was only about 12 months7. In a Phase
`III study of trastuzumab (Herceptin) — a humanized
`mAb against the receptor tyrosine kinase ERBB2 (also
`known as HER2/NEU)8 — in metastatic breast cancer,
`the OVERALL RESPONSE RATE was only 15%: 8 complete and
`26 partial responses were observed in 222 patients9. The
`median response duration and survival were 9.1 and 13
`months, respectively9.
`All clinically approved and most experimental anti-
`body drugs directly target tumour cells. Several strategies
`are being explored to increase the efficacy of such antibod-
`ies, including enhancement of effector functions, direct
`and indirect arming, and pre-targeting of prodrugs or
`radionuclides (FIG. 1). In addition, potent antitumour
`activity might be achieved with antibodies that prevent
`soluble growth factors from binding to their cognate
`receptors, such as the epidermal-growth-factor receptor
`(EGFR)10 and ERBB211,12. Promising and potentially com-
`plementary alternative strategies to direct tumour target-
`ing include targeting tumour vasculature, angiogenic
`growth factors and their receptors (BOX 3)5.
`
`118 | NOVEMBER 2001 | VOLUME 1
`
`© 2001 Macmillan Magazines Ltd
`
`www.nature.com/reviews/cancer
`
`Lassen - Exhibit 1047, p. 1
`
`

`

`Table 1 | Antibodies in advanced oncology trials*
`Antibody trade
`Antigen target
`Antibody type
`name
`(generic name)
`Avastin
`(bevacizumab)
`BEC2
`(mitumomab)
`
`VEGF
`
`hu IgG1
`
`Anti-idiotypic mAb, mu IgG
`GD3 ganglioside
`mimic
`CD20
`
`mu IgG2a
`
`Bexxar
`(tositumomab)
`
`Campath
`(alemtuzumab)
`
`CD52
`
`CeaVac
`
`Anti-idiotypic
`mAb, CEA mimic
`
`ERBB2
`
`EGFR
`
`hu IgG1
`
`mu IgG
`
`hu IgG1
`
`ch IgG
`
`Tumour target
`
`Status
`
`Metastatic NSCLC,
`metastatic CRC
`SCLC, malignant
`melanoma
`
`Phase III,
`Phase III
`Phase III,
`Phase II
`
`NHL
`
`B-cell CLL
`
`CRC
`NSCLC
`
`BLA filed
`with US FDA
`
`Phase III,
`Phase II
`
`R E V I E W S
`
`Corporate
`sponsors
`
`Genentech
`
`mClone
`Systems
`
`ImClone
`Systems,
`Merck KGaA
`Corixa,
`GlaxoSmith-
`Kline
`Approved in US Millennium
`May 2001
`and
`ILEX Partners
`Titan
`Pharma-
`ceuticals
`Approved in US Genentech
`Metastatic breast
`cancer overexpressing September 1998
`ERBB2
`CRC, locally advanced BLA filing in
`or metastatic head
`progress,
`and neck
`Phase III
`
`Strategy to enhance
`activity of naked
`antibody
`Combination with
`chemotherapy
`Vaccine
`
`131iodine
`
`None
`
`Vaccine in combination
`with chemotherapy for
`CRC or TriAb for NSCLC
`Combination with
`chemotherapy
`
`Combination with
`chemotherapy or
`external beam
`radiation
`None
`
`Herceptin
`(trastuzumab)
`
`IMC-C225
`(centuximab)
`
`LymphoCide
`(epratuzumab)
`MDX-210
`
`Mylotarg
`(gemtuzumab
`ozogamicin)
`Panorex
`(edrecolomab)
`
`Rituxan
`(rituximab)
`
`Theragyn
`(pemtumomab)
`Zamyl
`
`CD22
`
`hu IgG
`
`NHL
`
`Phase III
`
`Immunomedics
`
`ERBB2
`X CD64 (FcγRI)
`
`mu F(ab’)2
`
`Bispecific
`
`Ovarian cancer
`overexpressing
`ERBB2
`
`CD33
`
`EpCam
`
`CD20
`
`hu IgG4
`
`mu IgG2a
`
`ch IgG1
`
`Calicheamicin
`conjugate
`
`AML
`
`Combination with
`chemotherapy
`
`Dukes’ C CRC
`
`Combination with
`chemotherapy
`
`NHL
`
`PEM
`
`mu IgG1
`
`90yttrium
`
`CD33
`
`CD20
`
`hu IgG1
`
`mu IgG1
`
`Combination with
`chemotherapy
`90yttrium
`
`Ovarian cancer,
`gastric cancer
`AML
`
`NHL
`
`Phase III
`
`Medarex,
`Immuno
`Designed
`Molecules
`Approved in US Wyeth
`May 2000
`Laboratories
`
`GlaxoSmith-
`Approved in
`Kline,
`Germany
`Centocor
`January 1995
`Approved in US IDEC
`November 1997 Pharma-
`ceuticals,
`Genentech
`Antisoma
`
`Phase III,
`Phase II
`Phase III
`
`BLA filed with
`US FDA
`
`Protein Design
`Labs
`Zevalin
`IDEC
`Pharma
`(ibritumomab
`ceuticals
`tituxetan)
`* Phase III clinical trials or later. Not included are ongoing trials with marketed antibody products. Every effort has been made to obtain reliable data from several sources
`(company and industry web sites, and REFS 1,2), but accuracy cannot be guaranteed. AML, acute myelogenous leukaemia; BLA, Biologics License Application; CEA,
`carcino-embryonic antigen; ch, chimeric; CLL, chronic lymphocytic leukaemia; CRC, colorectal cancer; EpCam, epithelial cellular-adhesion molecule; FDA, Federal Drug
`Administration; EGFR, epidermal-growth-factor receptor; hu, humanized; mAb, monoclonal antibody; Ig, immunoglobulin; mu, murine; NHL, non-Hodgkin’s lymphoma;
`NSCLC, non-small-cell lung cancer; PEM, polymorphic epithelial mucin; SCLC, small-cell lung cancer; VEGF, vascular endothelial growth factor.
`
`Clinical strategies
`Combination with cytotoxic drugs. Combining different
`cytotoxic drugs is a widely and successfully used clinical
`strategy in oncology that increases the response rate and
`duration of individual drugs. The use of antibodies in
`conjunction with chemotherapeutics is a natural exten-
`sion of this approach, and is strongly supported by pre-
`clinical TUMOUR XENOGRAFT studies that show improved
`efficacy of antibody and chemotherapeutic combinations
`compared with each drug used in isolation. For example,
`Herceptin has synergistic antitumour activity when
`
`used in combination with cisplatin and carboplatin13,14,
`and additive benefit when used in conjunction with
`doxorubicin, cyclophosphamide, methotrexate, taxol or
`the selective cyclooxygenase-2 inhibitor, celecoxib14–18.
`The addition of Herceptin to a cytotoxic chemotherapy
`regimen was associated with statistically significant bene-
`fits in a Phase III trial in ERBB2-overexpressing meta-
`static breast cancer19. These gains included longer median
`duration of response (9.1 versus 6.1 months), higher
`overall response rate (50% versus 32%) and lower death
`rate at one year (22% versus 33%).
`
`RESPONSE DURATION
`Time from the first response
`until disease progression or
`death.
`
`OVERALL RESPONSE RATE
`Sum of partial and complete
`responses.
`
`NATURE REVIEWS | CANCER
`
`© 2001 Macmillan Magazines Ltd
`
`VOLUME 1 | NOVEMBER 2001 | 1 1 9
`
`Lassen - Exhibit 1047, p. 2
`
`

`

`R E V I E W S
`
`Box 1 | Key therapeutic antibody technologies
`
`Murine antibodies
`Derived by hybridoma technology99 following immunization of mice or, less commonly, rats.
`
`Chimeric antibodies
`Obtained by joining the antigen-binding variable domains of a mouse monoclonal antibody (mAb) to human constant domains: mouse VL to human
`CL and mouse VH to human CH1–CH2–CH3 for light and heavy chains, respectively100,101.
`Humanized antibodies
`In the simplest case, these are created by grafting the antigen-binding loops, known as complementarity-determining regions (CDRs), from a mouse
`mAb into a human IgG102–104. The generation of high-affinity humanized antibodies generally requires the transfer of one or more additional residues
`from the so-called framework regions (FRs) of the mouse parent mAb. Several variants of the humanization technology have been developed105.
`
`VL
`
`CL
`
`VH
`
`Heavy
`chain
`CH1
`
`CH2
`CH3
`
`Light
`chain
`
`Human antibodies
`These have high affinity for their respective antigens and are routinely obtained from very large, single-chain variable fragments (scFvs) or Fab phage
`display libraries106–110. Moreover, the difficulty in obtaining antibodies to self-antigens that are highly conserved between mouse and humans using
`hybridoma technology is readily overcome using phage display technology111. High-affinity human antibodies have also been obtained from
`transgenic mice that contain some, or preferably many, human immunoglobulin genes
`and genetically disrupted endogenous immunoglobulin loci. Immunization elicits the
`production of human antibodies recoverable using standard hybridoma
`technology10,59,112–114. A human anti-epidermal growth factor (EGF) receptor mAb
`obtained using transgenic mice eradicates large, established tumours in some
`preclinical xenograft models10, auguring well for ongoing oncology trials.
`
`Clinical experience
`Chimeric, humanized and human antibody evidence indicates that these types of
`antibody are less immunogenic than those of mice115. Humanized antibodies contain
`less foreign sequence than their chimeric counterparts and are presumed to be less
`immunogenic. Similar arguments have been made about humanized and human
`antibodies, despite the lack of substantiating clinical data. Other factors that affect the
`immunogenicity of antibodies include the method and frequency of administration,
`dose, patients’ disease and immune status, antigen specificity of the antibody and
`immune complex formation with antigen115.
`
`Choice of antibody technology
`Humanization and human antibodies are now the preferred technologies for developing
`antibodies as therapeutics. Humanization is a clinically well-validated technology that
`might be favoured if a well-characterized mouse mAb is available. By contrast, direct
`routes to human antibodies offer faster preclinical development in cases with no
`existing mouse mAbs. The choice of different human antibody technologies will depend
`on their availability, local expertise and commercial considerations.
`
`Mouse
`
`Humanized
`
`Chimeric
`
`Human
`
`Mouse sequences
`
`Glycosylation
`
`Human sequences
`Complementarity
`determining regions
`
`Unfortunately, these tangible clinical benefits of com-
`bining Herceptin with chemotherapy come at the price
`of greater toxicity. Herceptin alone is generally very well
`tolerated and typically associated with only very minor
`adverse events9,20, whereas Herceptin plus chemotherapy
`is associated with additional adverse events that are com-
`parable to, or worse than, chemotherapy alone19,21,22.
`Adding Herceptin to a regimen of an anthracycline drug
`plus cyclophosphamide, was associated with a signifi-
`cant increase in cardiotoxicity, although the symptoms
`generally improved with standard medical care19.
`At present, Herceptin is used as a single agent for
`patients with metastatic breast cancer whose tumours
`overexpress ERBB2 and who have received at least one
`regimen of chemotherapy for their metastatic disease.
`Herceptin is also used in combination with taxol for
`patients with metastatic breast cancer whose tumours
`overexpress ERBB2 and who have not received previous
`chemotherapy for their metastatic disease.
`The anti-CD20 antibody Rituxan sensitizes some
`drug-resistant human B-cell lines to the cytotoxic effects
`of etoposide, cisplatin and doxorubicin23. Single-arm
`
`Phase II clinical studies with Rituxan plus chemotherapy
`(cyclophosphamide, doxorubicin, vincristine and pred-
`nisone) in low-grade24 and high-grade25 B-cell non-
`Hodgkin’s lymphoma indicate that there is an additive24
`or at least comparable25 clinical benefit of Rituxan plus
`chemotherapy versus chemotherapy alone, with no
`additional toxicity. Ongoing randomized Phase III trials
`are anticipated to establish the statistical significance of
`improved response rates or duration from combining
`Rituxan with chemotherapy.
`In addition to Herceptin and Rituxan, at least five
`other anticancer antibodies are being tested in combi-
`nation with chemotherapy (TABLE 1)1. The increased
`number of these studies probably reflects several factors,
`including the need to test an experimental antibody
`drug in the context of current standard treatment —
`commonly chemotherapy — and the desire to improve
`on the modest antitumour activity of many naked anti-
`bodies. Two key factors are preclinical data showing
`the benefit of combining the specific antibody in ques-
`tion with chemotherapy, and the clinical success of
`combining Herceptin and Rituxan with chemotherapy.
`
`TUMOUR XENOGRAFT
`Commonly refers to the growth
`of human tumour cells as
`tumours in immuno-
`compromised mice.
`
`120 | NOVEMBER 2001 | VOLUME 1
`
`© 2001 Macmillan Magazines Ltd
`
`www.nature.com/reviews/cancer
`
`Lassen - Exhibit 1047, p. 3
`
`

`

`R E V I E W S
`
`Box 2 | Engineering antibodies for enhanced antitumour activity
`
`Antibodies can be engineered with altered properties, such as antigen-binding affinity, molecular architecture and
`dimerization state1, 3–5, that can enhance their tumour targeting and potency.
`The antigen-binding affinity (Kd values) of antibodies has historically been increased by selection from phage display
`libraries116,117. This laborious task of affinity maturation has been facilitated through the use of RIBOSOME DISPLAY118 in
`conjunction with DNA SHUFFLING119, or YEAST DISPLAY with DNA shuffling120. The affinity of antibodies can profoundly
`affect their ability to localize to tumours, as shown with a panel of single-chain variable fragments (scFvs) that bind to
`the same epitope on ERBB2 with affinities ranging from 10–7–10–11 M121,122. A threshold affinity of 10–7–10–8 M was
`required for specific tumour localization of scFv, and uptake reached a plateau at 10–9–10–11 M. These data support the
`concept of a ‘binding-site’ barrier that can impair penetration of the tumour for very high-affinity antibodies123.
`The impact of the antigen-binding affinity of IgG molecules on their tumour-localization ability remains to be
`determined. Selectivity might be crucial as many tumour-associated antigens are expressed on some normal cells at
`lower levels. It might be possible to achieve the greatest targeting selectivity by using a low-affinity IgG that relies on a
`high surface density of antigen on tumour cells for efficient binding.
`The molecular architecture of antibodies is readily modified to create non-natural antibody formats that vary in size
`and valency, with significant effects on tumour targeting ability. For example, a dimeric anti-ERBB2 antibody
`fragment, (scFv′)2 , showed increased tumour uptake compared with a monomeric Fab fragment of similar size
`(~50 kDa), and presumed similar tumour penetration and pharmacokinetic properties124. Increasing the valency of
`antibodies is a simple way to increase their avidity for cell-surface antigens and might lessen the need for affinity
`maturation.
`Tumour targeting by IgG is impaired by poor tumour penetration, which results from their large size, and uptake by
`Fc receptors on reticulo-endothelial cells5. By contrast, accrual of IgG within tumours is favoured by their long half-
`life, which reflects their size and their recycling by the neonatal Fc receptor, FcRn. Small antibody fragments, such as
`scFv, EXTRAVASATE more efficiently than IgG, and diffuse more readily within tumours, but are also rapidly eliminated
`through the kidney. These opposing factors limit the accrual of scFv within tumours to levels that are better suited for
`imaging than for therapy125.
`IgG homodimerization by chemical coupling increases the antitumour activity of antibodies against several different
`tumour antigens by mechanisms that include more potent antibody-dependent cellular cytotoxicity or complement-
`dependent cytotoxicity137, direct killing126,138,139, growth arrest126 and synergy with chemotherapy or immunotoxins138.
`IgG homodimerization can also increase the apparent affinity for binding to the cognate antigen on cells, which results
`in more rapid and/or more extensive internalization126,137,139. Significantly, IgG homodimerization might enhance
`in vivo antitumour activity126,139. Enhancement of the in vitro antitumour activity of Rituxan by homodimerization138
`should encourage additional preclinical and perhaps clinical assessment of this strategy.
`
`Targeting minimal residual disease. Using antibodies to
`target MINIMAL RESIDUAL DISEASE as well as MICROMETASTATIC
`DISEASE, following surgery, chemotherapy or radiotherapy,
`is a strategy26,27 that attempts to address the difficulties
`of poor accessibility and limited penetration of solid
`tumours by antibodies1,28. Indeed, the anti-epithelial
`cellular adhesion molecule (EpCam) mAb, 17-1A29,
`now known as edrecolomab (Panorex), has been report-
`ed to be more efficacious in the treatment of micro-
`metastases and minimal residual disease26,27 than bulky
`metastatic disease30. Panorex has been approved for the
`treatment of DUKES’ STAGE C COLORECTAL CANCER in Germany
`based on tangible patient benefit in a 189-patient
`Phase III trial26. It remains to be seen whether Panorex
`alone, or in combination with chemotherapy, will
`prove efficacious in Phase III trials that are ongoing in
`the United States (TABLE 1). Nevertheless, the concept of
`treating minimal residual disease warrants evaluation
`in the context of other antitumour antibodies. Clinical
`trials with Herceptin are now doing just that22.
`Several factors might limit the use of antibodies in
`treating minimal residual disease to those that have
`already been approved as anticancer therapeutics, or at
`least have shown benefit to cancer patients. First, trials
`of many thousands of patients might be needed to allow
`statistical assessment of treatment outcome. Second,
`follow-up periods of several years are required to track
`
`any differences in time to disease progression or death.
`Third, the large size and long duration of such trials
`makes them exceedingly expensive to conduct.
`
`Enhancing effector functions
`Human antibodies of the IgG1 and IgG3 isotypes can
`potentially support the effector functions of antibody-
`dependent cellular cytotoxicity (ADCC) and comple-
`ment-dependent cytotoxicity (CDC) (FIG. 2). Tumour-
`cell killing by ADCC is triggered by the interaction
`between the FC REGION of an antibody bound to a
`tumour cell, and the Fcγ receptors on immune effector
`cells, such as neutrophils, macrophages and natural
`killer cells. CDC is initiated by complement component
`C1q binding to the Fc region of IgG, which is bound to
`the surface of a tumour cell. Subsequent target-cell
`killing can occur in a cell-dependent or cell-independent
`manner (FIG. 2).
`The first demonstration that the Fc–Fcγ receptor
`interaction is important for the antitumour activity of
`an antibody in vivo came with the development of
`mice that lack FcγRI and FcγRIII31. An anti-melanoma
`antibody had potent antitumour activity in a mouse
`model of metastasis, but this benefit was lost in mice
`that lack FcγRI and FcγRIII receptors. ADCC is likely
`to be the mechanism underlying the antitumour
`effects of the Fc–Fcγ receptor interaction. Alternatively,
`
`RIBOSOME DISPLAY
`Technology for displaying
`a nascent protein, which is
`physically linked to its encoding
`mRNA, that relies on in vitro
`transcription and translation.
`
`DNA SHUFFLING
`Process for creating molecular
`diversity by homologous
`recombination of DNA in vitro.
`
`YEAST DISPLAY
`Technology for displaying a
`protein on the surface of a
`yeast cell that contains the
`gene(s) that encodes the
`displayed protein(s).
`
`EXTRAVASATION
`Movement out of the
`vasculature compartment into
`interstitial spaces.
`
`MINIMAL RESIDUAL DISEASE
`Tumour remaining in patients
`following debulking by surgery
`and/or chemotherapy and/or
`radiotherapy.
`
`MICROMETASTATIC DISEASE
`Metastatic disease that can be
`detected by immuno-
`histochemistry of tissue
`biopsies, but involves too few
`tumour cells to be directly
`imaged in patients.
`
`DUKE’S STAGE C COLORECTAL
`CANCER
`Cancer that has spread from
`the colon to nearby lymph
`nodes, but not to other parts of
`the body.
`
`FC REGION
`For an IgG, this comprises
`the CH2 and CH3 domains
`(BOX 1).
`
`NATURE REVIEWS | CANCER
`
`© 2001 Macmillan Magazines Ltd
`
`VOLUME 1 | NOVEMBER 2001 | 1 2 1
`
`Lassen - Exhibit 1047, p. 4
`
`

`

`R E V I E W S
`
`a Enhancing effector
`functions
`
`Complement-dependent
`cytotoxicity
`
`Point mutations
`and/or modified
`glycosylation
`
`d Pre-targeting
`
`Biotin–chelator–
`radionuclide
`
`Streptavidin
`
`Antibody-dependent
`cellular cytotoxicity
`
`Prodrug
`
`scFv–enzyme
`
`Cytokine
`
`Immunocytokine
`
`Small molecule
`or protein toxin
`
`Tumour cell
`
`Drug
`
`scFv
`fragment
`
`Sterically stabilized
`immunoliposomes
`
`Radionuclide
`
`Bispecific
`antibody
`
`Radionuclide, toxin
`or immunological
`effector cell
`
`b Direct arming
`
`c Indirect arming
`
`Figure 1 | Strategies for enhancing the potency of antitumour antibodies. Numerous
`strategies for improving the efficacy of antitumour antibodies are now being tested, including the
`representative examples shown here and described in BOX 2. a | Enhancing effector functions
`involve improving antibody-dependent cellular cytotoxicity and/or complement-dependent
`cytotoxicity by means of site-directed mutations or manipulation of antibody glycosylation.
`b | Direct arming of antibodies entails their covalent linkage to killing machinery, such as
`radionuclides or toxins (for example, small molecules or proteins). Alternatively, arming antibodies
`with cytokines is intended to create high intratumour concentrations of cytokines to stimulate the
`antitumour immune response (T cells, B cells or natural killer cells), while avoiding the toxicities
`associated with systemic cytokine delivery. c | Indirect arming of antibodies can be achieved by
`attaching engineered antibody fragments to the surface of liposomes loaded with drugs or toxins
`for tumour-specific delivery. Bispecific antibodies that bind to two different antigens can be
`preloaded with the cytotoxic machinery before administration (indirect arming) or alternatively
`pre-targeted to the tumour before delivery of the cytotoxic payload. d | Pre-targeting strategies
`aim for the selective delivery of radionuclides to tumours or selective intratumour activation of
`prodrugs, thereby diminishing the systemic toxicities of these cytotoxic agents. For prodrug pre-
`targeting, an antibody-fragment–enzyme fusion protein is typically allowed to localize to a tumour
`and be cleared from the system. A prodrug is then administered and ideally converted to an
`active drug solely within the tumour. For radionuclide pre-targeting, an antibody–streptavidin
`conjugate is allowed to accrue within a tumour and is then used to capture a
`biotin–chelator–radionuclide complex. scFv, single-chain variable fragment.
`
`induction of apoptosis by crosslinking of effector and
`target cells could explain these observations.
`The importance of the Fc–Fcγ receptor interaction
`for antitumour activity was subsequently shown for the
`clinically important antibodies Herceptin and Rituxan32,
`as well as for intracerebral therapy with an anti-EGF
`receptor antibody in a brain tumour model33. The anti-
`tumour activity of Herceptin and Rituxan was greatly
`reduced in mice that lack the activation receptors FcγRI
`and FcγRIII, whereas disruption of the gene that
`encodes the inhibitory receptor FcγRIIB substantially
`enhanced antitumour activity32. The antitumour activity
`of Herceptin was also attenuated by a mutation
`(D265A) that impairs binding to FcγRIII and FcγRIIB.
`These studies indicate the potential for increasing the
`
`KABAT NUMBERING SCHEME
`Immunoglobulin amino-acid
`residue numbering scheme,
`devised by the late Elvin Kabat,
`that accommodates sequence
`insertions and deletions.
`
`BISECTED COMPLEX
`OLIGOSACCHARIDES
`Branched carbohydrate that
`might include several different
`kinds of monosaccharides,
`including N-acetylglucosamine
`between main branches.
`
`antitumour activity of an antibody by manipulating the
`Fc region to increase its affinity for the activation recep-
`tor(s) and/or by abrogating its ability to bind to the
`inhibitory receptor. Indeed, point mutations in the Fc
`region, which result in improved binding to FcγRIII,
`yielded up to a twofold enhancement in ADCC in vitro34.
`The in vivo and clinical significance of this in vitro
`improvement is unknown.
`Glycosylation of IgG molecules at Asn297 (KABAT
`NUMBERING SCHEME35) helps to maintain the tertiary
`structure of their CH2 domains36 (BOX1), and is neces-
`sary for effector functions37. The cells producing the
`antibody (the ‘production host’) and, to a lesser extent,
`culture conditions, can significantly affect the resulting
`antibody glycoforms which, in turn, can influence the
`ability of the antibody to participate in ADCC38.
`Increasing the level of BISECTED COMPLEX OLIGOSACCHARIDES
`attached to the Fc region of an antibody can enhance
`its ability to support ADCC39. This modification of
`glycosylation was accomplished by cellular engineer-
`ing of the production host — Chinese hamster ovary
`cells — by transfection with β-(1,4)-N-acetylglucos-
`aminyltransferase III (REF. 39). Ongoing studies will
`address whether these antibodies have enhanced anti-
`tumour activity in vivo. Effector cell populations might
`be suppressed or diminished in cancer patients, for
`example, by previous chemotherapy, which could curtail
`the effector functions of antitumour antibodies.
`Antibodies with improved ability to support CDC
`have been created by mapping binding sites for the
`complement component C1q40 on human IgG1 Fc and
`then creating site-directed mutants with enhanced C1q
`binding41. It remains to be seen if these improvements
`in CDC in vitro translate into more potent antitumour
`activity in vivo. One potential problem with this approach
`is that many malignant cells, as well as normal cells,
`express membrane-bound proteins, such as CD46,
`CD55 and CD59, that interfere with CDC and allow
`them to escape complement attack42.
`
`Direct arming
`The most widely explored strategy for enhancing the
`efficacy of antitumour antibodies is direct arming by
`covalent linkage to toxins or radionuclides1,28 (FIG. 1).
`Armed antibodies typically show more potent anti-
`tumour activity in preclinical tumour xenograft stud-
`ies than their ‘naked’ parents. Unfortunately, clinical
`evaluation of armed antibodies has been beset by
`unacceptably high levels of toxicity in several clinical
`trials28, leading many to abandon this approach.
`Nevertheless, antibody arming is enjoying a renais-
`sance with the approval of the first armed antibody,
`gemtuzumab ozogamicin (Mylotarg), and with two
`others, tositumomab (Bexxar) and ibritumomab
`tituxetan (Zevalin), on the cusp of regulatory
`approval (see below).
`A judicious choice of both target antigen and anti-
`body is likely to be crucial to the success of all arming
`strategies. Ideally, the target antigen should be univer-
`sally found on tumour cells of a given type, but absent
`or present at much lower levels on normal cells. The
`
`122 | NOVEMBER 2001 | VOLUME 1
`
`© 2001 Macmillan Magazines Ltd
`
`www.nature.com/reviews/cancer
`
`Lassen - Exhibit 1047, p. 5
`
`

`

`R E V I E W S
`
`Box 3 | Alternative targets for anticancer antibodies
`
`Most approved and experimental anticancer antibodies directly target tumour cells. Alternative strategies include
`inhibiting angiogenesis or directly targeting tumour neovasculature5.
`Angiogenesis is the process by which tumours become vascularized by the proliferation of new blood vessels from
`existing ones, thereby allowing the tumours to grow beyond a minimal size. An antibody that neutralizes the
`angiogenic factor vascular endothelial growth factor (VEGF) has potent antitumour activity in vivo127,128, and non-
`invasive imaging indicates that this is due to angiogenesis inhibition129. This has encouraged the humanization130 of an
`anti-VEGF antibody, now known as bevacizumab (Avastin) that is now in Phase III clinical trials for metastatic cancers
`(TABLE 1). The alternative strategy — targeting VEGF receptors — is at present being tested in Phase I trials for patients
`with colorectal cancer and liver metastases with the IMC-1C11 anti-VEGFR2 antibody.
`Clinically significant questions include whether anti-VEGF treatment will benefit patients with established tumours
`and whether tumours will recruit other angiogenic growth factors to escape anti-VEGF blockade. The combination of
`anti-VEGF and anti-ERBB2 (Herceptin) antibodies has proved more efficacious in tumour xenograft models than
`either antibody alone, encouraging clinical evaluation of this combination (M. Pegram, personal communication).
`Targeting the tumour vasculature has several potential advantages over direct tumour targeting131–133: the vasculature
`is more accessible to antibodies; vasculature damage has a multiplicative effect as many tumour cells are dependent on
`each capillary; and as vascular endothelial cells are not transformed, they seem less likely to become resistant to
`antibody therapy.
`The therapeutic potential of targeting the tumour vasculature has been shown in tumour xenograft studies132,133.
`Tissue factor has recently been targeted to the ED-B domain of fibronectin, which is a natural marker of angiogenesis
`that is present in many solid tumours but not most normal blood vessels and tissues134. Tumours were eradicated by
`intratumour thrombosis in ≤ 30% of the mice treated with no obvious side effects134, encouraging further preclinical
`studies. Unfortunately, tumour regrowth following tumour vasculature targeting might occur from a thin layer of
`tumour cells close to blood vessels132,134. Encouragingly, synergistic antitumour activity was observed for
`immunotoxins directed to the tumour vasculature and the tumour itself. Coagulation at non-tumour sites is a
`potentially significant safety issue to be addressed with tumour-targeted thrombosis.
`Unfortunately, tumour vasculature targeting is limited at present by the lack of blood-vessel-specific target antigens.
`It might be possible to exploit antigens on tumour cells for tumour vasculature targeting by the judicious choice of
`antibody drugs. Tumour targeting with rapidly clearing antibody fragments (for example, Fab and scFv) results in
`de facto vasculature targeting5, as these fragments accumulate primarily on perivascular tumour cells122,135,136.
`
`antigens most successfully targeted with armed anti-
`bodies so far — CD20 and CD33 — closely match
`these criteria. CD20 is a tetra-spanning membrane
`protein that is found on mature B cells, including
`>90% of B-cell lymphomas, but is absent from stem
`cells, plasma cells and nonlymphoid tissue. CD33 is a
`sialic-acid-binding Ig-like lectin (siglec) that is found
`on the surface of virtually all acute myelogenous
`leukaemia cells, as well as myeloid progenitor cells and
`committed precursors, but not on stem cells or non-
`lymphoid tissue. A prerequisite for antibody arming
`with small molecule toxins, but not radionuclides, is
`that the antibody should be efficiently internalized.
`Direct selection for antibodies th