Cancer therapy with bispecific antibodies: Clinical experience

Abstract

The binding of at least two molecular targets simultaneously with a single bispecific antibody is an attractive concept. The use of bispecific antibodies as possible therapeutic agents for cancer treatment was proposed in the mid-1980s. The design and production of bispecific antibodies using antibody- and/or receptor-based platform technology has improved significantly with advances in the knowledge of molecular manipulations, protein engineering techniques, and the expression of antigens and receptors on healthy and malignant cells. The common strategy for making bispecific antibodies involves combining the variable domains of the desired mAbs into a single bispecific structure. Many different formats of bispecific antibodies have been generated within the research field of bispecific immunotherapeutics, including the chemical heteroconjugation of two complete molecules or fragments of mAbs, quadromas, F(ab’)₂, diabodies, tandem diabodies and single-chain antibodies. This review describes key modifications in the development of bispecific antibodies that can improve their efficacy and stability, and provides a clinical perspective on the application of bispecific antibodies for the treatment of solid and liquid tumors, including the promises and research limitations of this approach.

Introduction

Developing improved treatment strategies for cancer is a challenging task wherein the balance between increasing clinical efficacy without increasing systemic toxicity determines the success of the drug. mAbs have become an important class of protein-based drugs (> 20 mAbs have been approved by the FDA) for the treatment of cancer and other diseases [1-7]. Multiple novel approaches are being explored to enhance the efficacy and targeting of mAbs to tumor antigens. Early efforts were focused on increasing the efficiency of mAbs by their direct conjugation with various effector compounds, such as toxins, radionucleotides and cytotoxic drugs. However, the majority of these chemical manipulations resulted in the inactivation of antibody-binding sites or in alterations to the functions of effector agents. The next step in this process was to use combinations of whole mAbs, fragments of mAbs, or dual-targeting antibodies referred to as bispecific antibodies (biAbs). BiAbs have two distinct binding specificities and represent a promising approach to improving the effectiveness of antitumor therapy. BiAbs can be infused alone or after coating effector cells to redirect their cytolytic activity. Protein engineering has produced a variety of biAb formats (Figure 1), among which quadroma, F(ab')₂, diabodies, tandem diabodies and single-chain antibody (scFv)-based constructs are best characterized [8-11]. Trifunctional (tri)biAbs were developed to improve serum stability and in vivo efficacy. In tribiAbs, the two halves of the Fab (fragment antigen-binding) segment have different specificities, and engineered heterologous Fc (fragment crystallizable) variants facilitate enhanced serum stability and cytotoxicity [11,12]. The next modification led to the development of a multivalent and multifunctional dock-and-lock (DNL) tribiAb [13]. This review highlights the key developmental steps that lead to biAb-based therapies, either alone or in combination with effector cells armed with biAbs.

Figure 1. Bispecific antibody formats.

Genetically engineered antibody fragments or the heteroconjugation of intact antibodies to generate bispecific antibodies (biAbs) in various formats are shown: (A) fragment antigen-binding (Fab) format, (B) quadroma (IgG) formats constructed by fusing two hybridomas secreting antibodies of different specificities, (C) single-chain antibody (scFv [single-chain variable fragment])-based formats, (D) diabody formats or (E) chemical heteroconjugation of two IgG molecules [(IgG)₂] of different antigen specificities. Hetero-F(ab)₂ Heterogeneous fragment antigen-binding, TAA tumor-associated antigen

Combining cellular and humoral immunity

Both cellular- and antibody-based therapies exhibit antitumor activity, but do not engage each other because of the lack of Fc receptors on T-cells. Thus, a strategy that can combine cellular and humoral effectors will not only offer a potent anticancer response, but also a targeted and non-toxic therapeutic anticancer approach. The importance of cellular immunotherapy in cancer was first documented by Southam et al in 1966 [14]. This study demonstrated that subcutaneous growth of human tumor autografts to patients bearing advanced cancers was inhibited by the cotransfer of autologous leukocytes in approximately half of the patients [14]. Both allogeneic and autologous T-cells obtained from several anatomical sites were tested for cell-mediated antitumor activity. However, the effectiveness of cell therapy was compromised by multiple factors, such as quantity, in vivo proliferative potential, specificity, homing to the tumor targets and effector cell functions [15-23]. Many early clinical trials using autologous lymphokine-activated killer (LAK) cells or tumor-infiltrating lymphocytes (TILs) did not contribute greatly to the routine treatment of human cancer [15-23]. After numerous manipulations and efforts, Rosenberg and colleagues were the first to report an objective response rate of 34% among patients with metastatic malignant melanoma (MM) who were treated with TILs [24,25]. Results improved with further manipulations, such as treatment with adoptive transfer of T-cell receptor (TCR) gene-modified T-cells and high-dose IL-2 after non-myeloablative lymphodepleting chemotherapy, achieving an objective response rate of up to 49% [26]. These trials demonstrated that the adoptive transfer of TILs after non-myeloablative conditioning can be an effective treatment for patients with MM. Another example of immunotherapy successfully inducing cytogenetic and molecular remissions in chronic myelogenous leukemia (CML) was observed as a result of donor lymphocyte infusions (DLIs) after relapse of CML following allogeneic stem cell transplant [27-29]. Infusions of EBV-specific CTLs in EBV-driven lymphoproliferative disease (LPD) were also successful [29]. Nonetheless, the promising responses observed with CTL immunotherapy in MM and LPD have not been reproduced in other solid tumors or hematological malignancies, suggesting that new clinical treatment approaches are needed.

The development of most biAb constructs focuses on recruiting T-cells by targeting the CD3 complex, a potent signal transducer associated with polymorphic TCRs expressed on all T-cell subpopulations. Thus, biAbs offer an effective bridging between cellular and humoral effectors that can redirect and harness the cytotoxic potential of T-lymphocytes to kill tumor cells in a non-MHC-restricted manner.ssn

Bispecific antibodies

BiAbs are able to crosslink different target antigens either on the same cell or on two different cells. Because of their dual specificity, biAbs can activate immune effector cells by targeting the CD3 complex, costimulatory CD28 molecule (supra-agonistic) on T-cells or Fcγ receptors [FcγRs] on accessory cells, and can bind simultaneously to tumor-associated antigens (TAAs) on tumor cells [30-34]. Since Milstein and Cuello first described the hybrid hybridoma technique, numerous biAbs that target antigens on tumor cells and receptors on effector cells have been developed and demonstrated to redirect cellular cytotoxicity [35]. The concept of using such biAbs to engage CTLs for cancer cell lysis was first reported by Staerz and Bevan in 1986 [36,37]. This concept was based on the observation that anti-idiotypic and anti-allotypic mAbs coupled covalently to any cell surface will render that cell sensitive to lysis by a CTL expressing a receptor recognized by the mAb.

BiAb constructs can be produced by the co-expression of two antibodies in one cell, via the hybrid hybridoma technique or by DNA cotransfection, or via protein engineering methods. One antibody of the biAb construct binds to a trigger (receptor) molecule on an immune effector cell and a second antibody binds to tumor antigens, such as HER2/neu [38], epithelial cell adhesion molecule (EpCAM) [39], the EGFR [40] and the folate receptor [41].

Hybrid hybridoma (also referred to as quadroma) was the earliest technology used to produce biAbs [36]. This technology is based on the somatic fusion of two different hybridoma cell lines expressing murine mAbs of desired specificities. However, because of the random pairing between two different Ig heavy and light chains, the quality of the desired biAbs was poor, resulting in 80 to 90% of mispaired byproducts. The next improvement was the development of a mouse-rat quadroma expressing mAbs with two IgG subclasses selected for their preferential pairing [11]. The yields and purity of the biAbs were further improved by differential elution techniques that reduced contamination with parental antibodies [11]. Ertumaxomab (anti-HER2 × anti-CD3; Fresenius Biotech GmbH; Table 1) and catumaxomab (Removab, anti-EpCAM × anti-CD3; Table 1) are tribiAbs developed using the quadroma technology [42]. These biAbs are directed against TAA and CD3 and bind selectively to type I/III FcγRs, forming a complex between tumor cells, T-cells and accessory cells [11,42].

Table 1.

Selected clinical trials with bispecifc antibodies.

BiAb (alternative name; developing company)	Target or cancer type	Effector cell	Antibody format	Reference/ClinicalTrials.gov identifer
Anti-CD3 x anti-EpCAM (catumaxomab, Removab)	Ovarian, gastric, colon and breast (malignant ascites) cancers	T-cells	IgG	[70,73]
	NSCLC	T-cells	IgG	[72]
	Ovarian carcinoma	T-cells	IgG	NCT00189345
Anti-CD3 x anti-CD19 (blinatumomab; Micromet AG/Micromet Inc)	NHL	T-cells	BiTE	NCT00274742
	B-precursor acute lymphoblastic leukemia	T-cells	BiTE	NCT00560794
Anti-CD3 x anti-CA19-9 (OKT3/NSI19-9)	CA19-9-positive tumors	T-cells	IgG	[97]
Anti-CD3 x anti-CEA	Human CEA-expressing cells	T-cells	F(ab’)2	[98]
Anti-CD3 x anti-CEA	Ovarian carcinoma	T-cells	Diabody	[99]
Anti-CD3 x anti-CD19 (SHR-1)	NHL	T-cells	IgG	[55]
Anti-OCAA x anti-CD3 (OC/TR)	Ovarian carcinoma	T-cells	F(ab’)2	[92,100]
Anti-CD3 x anti-CD20 (CD20 biAb)	NHL	T-cells	(IgG)2	NCT00521261
Anti-CD3 x anti-HER2 (HER-2Bi-armed ATCs; Transtarget Inc)	Metastatic breast cancer	T-cells	(IgG)2	NCT01022138
Anti-CD3 x anti-HER2 (ertumaxomab; Fresenius Biotech GmbH)	Metastatic breast cancer	T-cells	IgG	NCT00452140
Anti-CD28 x MAPG (rM28)	Metastatic melanoma	T-cells	Tandem scFv	NCT00204594
Anti-EpCAM x anti-CD3 (MT-110; Micromet Inc)	Adenocarcinoma of the lung, small cell lung cancer, gastric and colorectal cancer	T-cells	BiTE	NCT00635596
Anti-CD16 x anti-CD30 (HRS-3/A9)	Hodgkin’s disease	NK cells	IgG	[101]
Anti-CD16 x anti-HER2 (2B1)	HER2-positive tumors	NK cells	IgG	[57,102]
Anti-CD3 x anti-glioma	Human glioma	LAK cells	F(ab’)2	[53]
Anti-CD64 x anti-HER2 (MDX-210 and MDX-H210)	Breast, ovarian and prostate cancer	Monocytes/ macrophages	F(ab’)2	[59,60]
Anti-CD64 x anti-EGFR (MDX-447)	Solid tumors	Monocytes/ macrophages	F(ab’)2	[103]

ATCs Activated T-cells, biAb bispecifc antibody, BiTE bispecifc T-cell engager, CA19-9 cancer (gastrointestinal)-associated antigen CEA carcino-embryonic antigen, EpCAM epithelial cell adhesion molecule, F(ab’)₂ two fragment antigen binding, (IgG)₂ biAb generated by chemical heteroconjugation of two IgG molecules of different specifcities, LAK lymphokine-activated killer, NHL Non-Hodgkin’s lymphoma, MAPG melanoma-associated proteoglycan, OCAA ovarian cancer associated antigen, OC/TR chimeric biAb against human folate-binding protein overexpressed in ovarian carcinoma and T-cell receptors (CD3), scFv single-chain variable fragment

The DNL method described by Rossi and colleagues is another well-designed approach to generate biAbs with multivalency and multifunctionality [13,43]. This method uses the self-assembling dimerization-and-docking domain (DDD) mAb attached to a DDD moiety, wherein two copies of the DDD moiety form a dimer that binds to the anchoring domain (AD) moiety, resulting in the formation of the DNL complex. The combining of the respective DDD- and AD-modules generates stably-tethered structures of defined composition with retained bioactivity. The main advantage of this system is that an antibody fragment of any specificity can be produced separately, as fusion proteins with either a DDD or a AD and, when needed, a simple mixing of the two purified products will yield the desired biAb [43].

Continued efforts to improve the therapeutic efficacy of biAbs led to the development of a new class of biAbs – BiTE (bispecific T-cell engager) and diabodies. Constructs in which linked variable heavy (V_H)-variable light (V_L) pairs are from the same antibody are referred to as scFvs, with two such scFvs connected by a linker to yield a single polypeptide chain [10]. Diabodies are arranged such that linked V_L–V_H pairs are from two different antibodies. BiTE biAbs can bind monovalently to all T-cells, but do so with low affinity [9], and will not trigger T-cell signaling by CD3 unless the BiTE antibody is presented to the T-cell in a multivalent manner by a target cell [44]. This mode of action is referred to as BiTE technology. An approach using the multimerization of antibody fragments was developed to improve the serum half-life and binding affinity of BiTE biAbs. Similarly, divalent scFvs were developed from the monovalent scFvs, either through the spontaneous non-covalent association of two scFv molecules (ie, diabodies) or through a covalent disulfide bond. Divalent and multivalent scFvs were also generated by linking two or more scFv molecules using a peptide linker [10].

A variety of immune effector cells, such as CTLs, NK cells, neutrophils, monocytes/macrophages, or drugs, toxins, enzymes, prodrugs, DNA, antivascular agents, viral vectors and radionuclides are redirected using biAbs. For T-cells, cytotoxicity is redirected to tumor targets, bypassing MHC restrictions by targeting CD3 on T-cells. A wide variety of effector responses can be derived against tumor cells by changing the specificities of the anti-effector and anti-target components of the biAb [45-48].

Clinical trials using bispecific antibodies Whole IgG-based bispecific antibodies

Several clinical trials have been conducted with chemically linked or genetically engineered biAbs targeting TCRs or FcγRs and TAA, either alone or in combination with IFNγ, GM-CSF or G-CSF [49-51]. The trials using local [52,53] or intravenous injections [54] of biAbs alone were conducted in the early 1990s. Local injections of biAbs caused antitumor activity and acute-phase inflammatory responses. However, the effectiveness of the biAbs was restricted to the tumor injection site with little impact on distant metastatic sites of disease. Subsequent trials tested whether biAbs can arm and redirect T-cells to tumor sites. A phase I trial conducted in patients with CD19+ non-Hodgkin's lymphoma (NHL) using SHR-1 biAbs (anti-CD3 × anti-CD19; Table 1) demonstrated that infusions of SHR-1 could arm endogenous T-cells and retarget them to tumor sites [55]. Similarly, a phase I trial conducted by Kroesen et al using BIS-1 biAbs (anti-CD3 × anti-EGP-2) confirmed that endogenous T-cells could be armed and redirected to tumor sites [56]. However, in both of these trials DLT was observed. Trials in solid tumors using 2B1 (anti-HER2 × anti-FcγRIII; Table 1), a murine IgG quadroma, to target HER2/neu-positive tumors did not reveal any antitumor responses [57,58]. Treatment resulted in significant increases in TNFα, IL-2, and IL-8, with 14 out of 15 patients developing human anti-mouse antibody (HAMA) responses; however, DLT limited the clinical use of this biAb [57]. The results from these trials suggest that whole IgG-based biAb infusions cause the activation of immune cells, leading to unmanageable cytokine storm, and prompting the redesign and modification of biAb constructs to overcome DLTs.

MDX bispecific antibodies based on the heterogeneous F(ab′)₂ molecule

Using the same platform as 2B1 and targeting the same epitope on HER2, MDX-210 (Table 1), a heterogeneous (hetero)-F(ab')₂ molecule, was produced by chemically conjugating a humanized anti-CD64 Fab' with a murine anti-HER2/neu Fab' [59]. This biAb was engineered to delete Fc domains to decrease adverse reactions. Patients tolerated higher doses of MDX-210 than the intact IgG-based biAb 2B1. In addition, the deletion of the Fc domains decreased the cytokine storm-related toxicities observed in the 2B1 clinical trials [59]. Phase I trials using the MDX-210 biAb revealed potent in vitro cytotoxicity, phagocytosis and induction of anti-HER2 antibodies. However, trials with either 2B1 or MDX-210 demonstrated no consistent pattern of antitumor activity [57,59]. In a phase II trial, James et al evaluated MDX-H210 (a semi-humanized antibody; Table 1) in combination with GM-CSF and reported that this combination is active in hormone-refractory prostate carcinoma with acceptable toxicity [60]. In a multidose trial conducted by Posey et al, GM-CSF was administered on days 1 to 4 and MDX-H210 was administered on day 4 weekly for 4 consecutive weeks, with ten patients completing 4 weeks of treatment [61]. One patient had a 48% reduction in lesions and six patients had stable disease at the time of evaluation; three patients exhibited disease progression before the fourth week. The therapy was generally well tolerated with toxicity limited primarily to the days of treatment [61].

BiTE-bispecific antibodies

Bispecific immunotherapeutics involving CD3-binding activity, and in particular the scFv-based BiTE format, have demonstrated promising results in directing CTL responses against tumor cells. BiTE antibodies are able to induce potent redirected lysis of target antigen-expressing cells at pico- to femtomolar concentrations that is accompanied by highly conditional T-cell activation [44,62]. Phase I and II clinical trials are being conducted with the CD19/CD3-bispecific blinatumomab (MT-103; Micromet AG/Micromet Inc; Table 1) and the EpCAM/CD3-bispecific MT-110 (Micromet) BiTE antibodies [9,63,64]. Results with blinatumomab from the phase I trial have confirmed that T-cells can be potently engaged for redirected tumor cell lysis in patients with cancer [65]. Among patients with NHL (n = 38) receiving blinatumomab (doses ranging from 0.0005 to 0.06 mg/m²/day), 11 patients exhibited objective responses. In addition, four patients demonstrated complete responses (CRs), and seven patients exhibited partial tumor regressions (PRs) at doses of ≥ 0.015 mg/m²/day. Seven patients treated at the dose level of 0.06 mg/m²/day displayed objective responses. Tumor regressions were observed in patients with follicular lymphoma, mantle cell lymphoma and chronic lymphocytic leukemia. Blinatumomab doses of ≥ 0.015 mg/m²/day led to the depletion of tumor cells not only in blood, lymph node lesions and spleen, but also in the bone marrow. In 9 out of 11 cases of bone marrow infiltration, immunohistochemical and flow cytometric analysis of patient biopsies revealed either complete (n = 6) or partial (n = 3) elimination of tumor cells from the bone marrow [65].

Data from a phase II clinical trial (ClinicalTrials.gov identifier: NCT00560794) in patients with B-precursor acute lymphoblastic leukemia indicated that T-cells engaged by blinatumomab are able to locate and eradicate rare disseminated tumor cells in the bone marrow that can only be detected by quantitative PCR assays, which identify tumor cell-specific genomic aberrations [66]. Data from these ongoing trials demonstrate the highly effective cytotoxic activity of T-cells against both bulky and minimal residual disease [67].

Trifunctional antibodies

Trifunctional antibodies (triAb) were also designed and developed to improve the effectiveness of biAbs further [68,69]. A triAb is composed of the halves of two distinct full-size antibodies, a tumor-specific mouse antibody and a T-cell specific rat antibody, combined in one molecule. TriAbs therefore have two different specific binding sites, while the Fc portion binds to accessory cells, providing a trifunctionality to this antibody. Burges et al conducted a phase I/II clinical trial using the anti-CD3 × anti-EpCAM triAb catumaxomab, administered intraperitoneally to patients (n = 23) with recurrent malignant ascites from ovarian cancer [70]. A 5-log reduction in EpCAM-positive tumor cells in the ascites was observed after therapy with intraperitoneal injections of catumaxomab, and direct injections of the antibody demonstrated clinical promise, but was limited by DLTs when administered intravenously. Kiewe et al reported a phase I trial of ertumaxomab, which is a tri-antibody directed at CD3 and HER2/neu with a Fcγ type I/III receptor that establishes a tri-cell complex between T-cells, Fc receptor-positive cells and tumor cells [71]. The majority of patients (15 out of 17) completed the trial (MTD of 100 µg) with mild and transient side effects. Antitumor responses were observed in 5 out of 15 patients, with one CR, two PRs and 2 patients with stable disease. Testing for serum cytokines revealed a Th1 pattern [71]. Another phase I trial using a triAb for the treatment of NSCLC demonstrated that catumaxomab can be infused intravenously in combination with dexamethasone with substantially reduced side effects [72]. The MTD was 5 µg of catumaxomab administered after 50 mg of dexamethasone. Results from this trial suggested that clinical responses can be obtained without DLTs using infusions of triAbs. Catumaxomab is the most advanced triAb based on TRION Pharma GmbH's TriomAb technology, and received EU approval for the intraperitoneal treatment of malignant ascites in patients with EpCAM-positive carcinomas in 2009 [73]. This triAb is not only the first drug indicated for the treatment of malignant ascites, but is also the first approved bispecific triAb.

The activation and targeting of immune effector cells

Studies have revealed that the in vitro activation of T-cells prior to arming enhances antitumor activity [74-76]. Similarly, the higher lytic activity of in vitro-activated PBMCs compared with resting PBMCs against tumor cells has been demonstrated [77-80]. While the in vitro activation and expansion of T-cells is straightforward and is typically achieved using the CD3-recognizing mAb (OKT3) that is present on all T-lymphocytes together with low doses of IL-2 [80-82], the in vivo activation of CTLs is not as simple. Clinical trials aimed at activating CTLs using OKT3 were unsuccessful [83-86]. Consistent with these results, Janssen et al demonstrated that whole blood cultures exhibited a poor proliferative response when OKT3 was added, whereas isolated PBMCs displayed brisk proliferative responses to OKT3 [87].

BiAb-induced effector cell-mediated cytotoxicity not only results in cell death, but also leads to cytokine and chemokine release through biAb-induced effector-target interactions. The release of multiple cytokines and chemokines, such as IFNγ, GM-CSF, IL-12, MIP-1α and MIP-1β, may recruit and activate endogenous immune cells to enhance the local antitumor response [88]. Cytotoxicity can induce cell death, either through necrosis or apoptosis, that can facilitate tumor antigen release and cytokine/chemokine production, all of which augment the recruitment and uptake of TAAs. Studies have revealed that arming ex vivo-expanded activated T-cells (ATCs) with biAbs prior to infusion can enhance cytotoxicity to several TAAs; this strategy converts every ATC into a specific CTL [89-91].

Clinical trials using bispecific antibody-armed/coated effector cells

The first clinical trial using biAb-armed LAK cells, which were co-injected into brain tumors with heteroconjugated anti-CD3 × anti-glioma biAbs, was reported in 1990 [53]. This trial demonstrated encouraging results, with four out of ten patients exhibiting regressed tumors and improved survival. A subsequent trial by Canevari et al evaluated the antitumor effect of the anti-ovarian cancer × anti-CD3 antibody (OC/TR; Table 1), a F(ab')₂ of a quadroma that recognizes a folate-binding protein on cells with one arm and anti-CD3 with the other arm [92]. In this trial, ex vivo-expanded T-cells were armed (coated) with OC/TR biAbs and injected regionally (intraperitoneally) together with additional soluble biAbs and, in some cases, low-dose IL-2. The arming dose of OC/TR was 1 mg/10⁹ viable cells (~ 1000 ng/10⁶ cells). Despite high tumor burdens, 27% of patients demonstrated clinical responses. Mild side effects were observed with intraperitoneal administration; however, intravenous injection of even low doses of OC/TR F(ab')₂ resulted in non-specific T-cell activation, TNFα secretion and toxicity [92]. Subsequent to this trial, Link et al reported that non-specific T-cell activation from biAbs can occur in an antigen-independent manner because of the Fc/Fc receptor interaction with whole IgG-based biAbs, or in an antigen-dependent manner when antigen is expressed on healthy or tumor cells with F(ab')₂-biAbs [93]. Both mechanisms may be responsible for the toxicities observed in prior trials. Table 1 summarizes selected biAbs that have been evaluated in trials.

In a pilot clinical trial, Riechelmann et al used antibody-coated immune cells in patients with head and neck squamous cell carcinoma to determine the MTD of the loading dose of catumaxomab onto PBMCs [94]. PBMCs were collected from patients (n = 4) by leukapheresis, incubated with catumaxomab ex vivo for 24 h and cleared from released cytokines prior to infusion into patients at twice-weekly intervals. This strategy was undertaken to remove large amounts of IFNγ and TNFα that were released with the binding of catumaxomab to PBMCs prior to administration. Ex vivo loading with triAb and the removal of potent cytokines and chemokines was concluded to decrease the severity of the cytokine storm [94].

The ongoing phase I/II dose-escalation clinical trial of HER-2Bi-armed ATCs in women (n = 20) with stage IV breast cancer is being conducted to determine the MTD or technically feasible dose and, thus far, has revealed no DLTs at total dose levels of 40, 80 and 160 billion total cells per patient [88]. Eight infusions are administered twice weekly for 4 weeks along with IL-2 (300,000 IU/m²/day sc) and GM-CSF (250 µg/m² twice weekly) beginning 3 days before the first armed ATC infusion and ending 1 week after the last infusion. The highest dose is 4.0 × 10¹⁰ cells per infusion for a total dose of 3.20 × 10¹¹ HER-2Bi-armed ATCs. Armed ATC infusions of up to 20 × 10⁹ cells have been well tolerated, with chills, fever and transient hypotension managed with antihistamines, meperidine, antipyretics and hydration. After four infusions, investigations using the peripheral blood from several patients revealed cytotoxicity directed at the breast cancer cell line SK-BR-3, and serum cytokine profiles that were shifted toward a Th1 pattern. IFNγ, GM-CSF, TNFα and IL-12 were detected in the serum of patients 2 weeks after the initiation of immunotherapy [88].

In vitro studies by Guo et al also suggest that antitumor responses can be generated by in vitro targeting with biAbs and the modification of tumor cells with cytokines to generate immunogenic tumor cells [95]. The major difference of low-grade toxicity observed by Lum et al [88,96] compared with other trials using biAb-armed ATC therapy in which high-grade toxicity has been observed, may be a result of the ex vivo arming of the T-cells prior to infusion, such that unbound biAbs are not infused into the patients. In other words, the biAb effect is limited to the interactions of the armed effector cells, as opposed to biAb infusions that provide much larger amounts of circulating biAbs to interact with immune cells in the patient leading to cytokine storm. The infusions of targeted ATCs may induce vaccine responses by creating a local environment that contains tumor lysates and high local concentrations of Th1 cytokines and chemokines, inducing the development of endogenous dendritic cells, which in turn, would facilitate tumor antigen (lysate) presentation to endogenous naïve T-cells. These trials demonstrate promising results using biAbs that can target tumor cells selectively and may immunize patients against their own tumors [88,96].

Conclusion

Several biAbs display significant in vitro cytotoxicity and antitumor effects in animal models, but fail to demonstrate consistent clinical activity. In most clinical trials in which clinical response was observed, cellular immunotherapy was administered after debulking chemotherapy, surgery and/or radiation, highlighting the importance of lymphodepletion, decreased tumor burden and/or regulatory T-cell depletion. Failure to obtain clinical responses may reflect limitations in tumor targeting, inadequate lymphocyte migration to tumor sites, defective in vivo activation of lymphocytes, and in vivo suppression of memory and effector CTLs. More recently, the concept of multiple receptor targeting and fusion molecules has been applied to the development of triAb or trispecific antibodies. Advances in the understanding of tumor antigens, antibody engineering and immunology, and their application in the appropriate clinical models may help to resolve some of these challenges and improve the effectiveness of biAb-targeted immunotherapy.