RESURGENCE OF BISPECIFIC ANTIBODIES - Fusion Protein Technologies for Biopharmaceuticals: Applications and Challenges - page 517

Schering Pharma, and ImmunGene. Cytokines can be fused
to an antibody fragment or to a complete IgG molecule
(Figure 35.7A). The fused cytokines will bind their high-
affinity receptors on immune cells and this way transiently
connect immune cells to those cells targeted by the antibody
or fragment. The same time, cytokines stimulate the cyto-
kine receptor to elicit an immune effector function. Fre-
quently selected cytokines for this approach are IL-2, IL-12,
and IFN- a 2. A concern of immunocytokines could be that
their cytokine receptor binding may be of much higher
affinity than cancer target binding. This would trigger
receptors on effector cells before bridging of effector and
target cells occurs. Another concern is that cytokines by their
combination with antibodies become long-lived and thereby
more toxic than short-lived cytokines. A number of immu-
nocytokines are in clinical development. An anti-GD2
IgG/IL-2 immunocytokine has shown high clinical activity
in juvenile neuroblastoma with 5 complete responses among
23 patients [81]. Intriguingly, this antibody has only a very
short serum half-life of only 4 h. Disease stabilization has
been observed in cancer patients in a Phase II trial treated
with a single-chain anti-fibronectin EBD domain antibody
fused to IL-2 [82].
Two fusion proteins shown in Figure 35.7 have been
specifically developed for T-cell engagement. One from
Active Biotech is a fusion between an anti-5T4-specific
Fab fragment and an engineered bacterial superantigen,
called naptumomab estafenatox or ANYARA 1 (Figure
35.7B) [83]. It can connect cancer cells expressing the fetal
oncoprotein 5T4 to cytotoxic T cells by specific binding of
the superantigen part to the TCR v b -chain. Point mutations
have much reduced the second specificity of the super-
antigen for MHC class molecules. An issue of the format
still is the high immunogenicity of the bacterial superanti-
gen. Pre-existing antibodies against the streptococcal anti-
gen in patients require that doses have to be individually
adjusted to match the level of anti-superantigen antibody
titers [84]. Naptumomab estafenatox has completed several
clinical studies showing signs of antitumor activity. A Phase
III trial in renal cell cancer patients in combination with
IFN- a treatment versus IFN- a alone is expecting readout in
the first half of 2011.
Another T-cell-engaging format is called “Immune Mobi-
lising mTCR Against Cancer” or “ImmTAC.” The bispecific
construct is a fusion between an anti-CD3-specific scFv and
a high-affinity, disulfide bond-stabilized TCR [85].
ImmTACs are being developed by Immunocore and
designed to engage T cells for lysis of cancer cells. Their
lead product recognizes melanoma cells expressing a gp100-
derived peptide antigen by HLA-A2. Other ImmTACs spe-
cific for other peptide antigens are under preclinical devel-
opment. The gp100 peptide-specific ImmTAC has just
commenced clinical Phase III

One issue of this format and the one featured in
Figure 35.6E is that a particular HLA/peptide complex
will be present on cancer cells only at very low copy number.
High affinity binding by TCR-like scFvs or single-chain
TCRs in combination with high drug concentrations will be
required for efficient lysis. Another problem arises if cancer
cells become selected in patients for loss of MHC class I, b 2
microglobulin or TAP transporter expression, or for altered
proteasome activity. Such “immune-escaped” cancer cells
can no longer be recognized by bispecific antibody con-
structs with TCR-like specificity. Finally, their activity will
be HLA-restricted thereby limiting the patient population,
and no relevant animal species will be available for pre-
clinical toxicity assessment.

35.9 BISPECIFIC ANTIBODIES FOR VARIOUS
FUNCTIONS: HOW TO SELECT THE RIGHT
FORMAT?

This review demonstrates that numerous antibody formats
are currently available for making bispecific antibodies.
However, not every format will support a desired function
and performance. For neutralizing two soluble ligands, for
example, cytokines, an IgG-like bispecific antibody appears
most desirable. This should neutralize ligands with high
affinity, be of low immunogenicity and otherwise share
properties with established ligand-neutralizing commercial
antibodies such as anti-TNF human IgG1 adalimumab
(Humira 1 ) or anti-VEGF humanized IgG1 bevacizumab
(Avastin 1 ). Theoretically, all antibodies formats shown in
Figures 35.1-35.4 qualify for this task. They all retain the
Fc g part mediating a long-serum half-life and clearance of
antibody/ligand complexes via the reticulo-endothelial sys-
tem. Bispecific antibody formats may however generically
differ in target tissue penetration, distribution volume, com-
patibility with subcutaneous administration, clearance kinet-
ics, aggregation behavior, protease resistance, productivity,
immunogenicity, and other parameters. It is hard to predict
which format will suit a neutralizing function best also,
because biophysical properties of antibodies will largely
depend on variable domains, which are likely to dominate
the characteristic of any fusion protein.
For dual receptor blockade, other formats may be optimal
than for ligand neutralization. Because anti-receptor anti-
bodies bind to cell surfaces, it must be decided whether the
antibody should confer as a third activity ADCC, which may
worsen the safety profile but, on the other hand, add cyto-
toxicity to the dual receptor-blocking mode of action. In case
immune effector mechanisms are not desired, the human
IgG4 format, an ADCC/CDC-inactivated Fc g 1 mutant IgG,
or a fusion of antibody fragments to HSA can be used. Good
target tissue penetration, high potency of receptor blockade,
and long serum half-life appear favorable. Certain formats or

testing in the United

Kingdom.