Ex Parte 7612181 et al

Patent Trials and Appeals Board

Mar 21, 2014

95001380 - (S) (P.T.A.B. Mar. 21, 2014)

UNITED STATES PATENT AND TRADEMARK OFFICE
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O. Box 1450
Alexandria, Virginia 22313-1450
www.uspto.gov
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO.
95/001,380 06/24/2010 7612181 ABT-110.2 USRX 4816
74943 7590 04/03/2017
YANKWICH & ASSOCIATES, P.C.
(AND ABBVIE BIORESEARCH CENTER)
201 BROADWAY
CAMBRIDGE, MA 02139
EXAMINER
TURNER, SHARON L
ART UNIT PAPER NUMBER
3991
MAIL DATE DELIVERY MODE
04/03/2017 PAPER
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
____________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________

SANOFI-AVENTIS, SA
Requester and Appellant

v.

ABBOTT LABORATORIES
Patent Owner and Respondent
____________

Appeal 2017-001183
Reexamination Control 95/001,380
Patent 7,612,181 B2
Technology Center 3900
____________

Before JAMES T. MOORE, RICHARD M. LEBOVITZ, and RAE LYNN
GUEST, Administrative Patent Judges.

LEBOVITZ, Administrative Patent Judge.

DECISION UNDER § 41.77(f)
This is a final decision under 37 C.F.R. § 41.77(f). The above-
identified inter partes reexamination of U.S. Patent No. 7,612,181 B2 (“the
’181 patent) was decided by the Patent Trial and Appeal Board Decision on
March 24, 2014 (“Decision” or “Dec.”). In the Decision, the Examiner’s
determination not to adopt Rejections 2 and 3 of claims 1–3, 5–7, 31, and 32
under 35 U.S.C. § 103(a) was reversed. A reversal of an Examiner’s
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

2
determination not to adopt a rejection constitutes a new ground of rejection.
See 37 C.F.R. § 41.77(b).
Patent Owner (“Abbott”) requested that prosecution be re-opened
under 37 C.F.R. § 41.77(b)(1). The Request included new evidence from
Abbott constituting a second declaration under 37 C.F.R. § 1.132 by Dr.
Roland Kontermann (“Kontermann 2 Decl.”) and a declaration by Dr. Janice
Reichert (“Reichert Decl.”). The Declarations were entered and the Request
to Reopen Prosecution granted. Order Granting Request to Reopen
Prosecution and Remand to Examiner (“Order”; mailed Nov. 20, 2014).
Requester (“Sanofi”) filed comments under 37 C.F.R. § 41.77(c) (“Sanofi
41.77c Comments”) accompanied by new evidence constituting a second
declaration of Drs. Davis and Verhoeyen (“Davis-Verhoeyen 2 Decl.”).
Upon remand to the Examiner, the Examiner made a determination
under 37 C.F.R. § 41.77(d) (“Determination”; mailed June 2, 2015)
upholding the Board’s Decision to reinstate Rejections 2 and 3. Abbott filed
comments under § 41.77(e) on the Examiner’s determination (“Abbott
41.77e Comments”). Sanofi filed comments after the Examiner’s
determination under § 41.77(e) (“Sanofi 41.77e Comments”). The appeal
has now been returned to the Board for a final decision under 37 C.F.R. §
41.77(f).
After considering the entirety of the evidence before us, we have
concluded that a preponderance of the evidence does not support the
obviousness of the claimed subject matter as set forth in Rejections 2 and 3.
As explained in more detail below, this conclusion is based on the lack of a
reason to have utilized the CH1-CL2 domain in the claimed engineered
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

3
antibody to have arrived at the claimed structure. Specifically, Dr.
Kontermann in his second declaration identified a misunderstanding in the
Decision which merited reconsideration of the evidence and the conclusions
based upon it.
CLAIM 1
The claimed subject matter is directed to a binding protein comprised
of four polypeptide chains and four functional antigen binding sites. An
exemplary embodiment is DVD-Ig (Dual Variable Domain) which is
depicted below:

DVD-Ig is reproduced above. The heavy shaded lines represent
heavy chains; the lighter shaded lines represent light chains. Vb and Va are
variable domains, each capable of having a different antigen binding
specificity and each composed of a heavy and light chain. When the antigen
specificities of Vb and Va are different, the binding protein (also referred to
herein as an antibody) is “bispecific” because it is specific for two different
antigens. CH1 and CL are constant domains comprising heavy and light
chains, respectively. CH2 and CH3 are constant domains that comprise only
heavy chains and which are identified as an antibody “Fc” region. The Fc
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

4
portion of an antibody mediates several effector functions, such as cytokine
induction and complement-dependent cytotoxicity. ’181 patent, col. 17, ll.
26–30. As shown in the figure reproduced above, the heavy chains are on
one peptide and the light chains are on another shorter peptide. The heavy
and light chains are, therefore, “segregated” from each other.
Bispecific antibodies were known in the prior art (’181 patent, cols.
1–3). Bispecific antibodies are described in the ’181 patent as useful to treat
cancer, inflammatory conditions, and other diseases (id. at col. 1, 14–17).
Claim 1, the only independent claim on appeal, reads as follows
(bracketed letters [A] and [B] have been added to highlight the two different
polypeptides; indentations and paragraph breaks have been added to clarify
the composition of each polypeptide):
1. A binding protein comprising four polypeptide chains,
[A] wherein two polypeptide chains comprise VD1-
(X1)n-VD2-C-(X2)n,
wherein VD1 is a first heavy chain variable domain,
VD2 is a second heavy chain variable domain,
C is a heavy chain constant domain,
X1 is a linker with the proviso that it is not CH1, and
X2 is an Fc region; and

[B] two polypeptide chains comprise VD1-(X1)n-VD2-
C-(X2)n, wherein VD1 is a first light chain variable
domain,
VD2 is a second light chain variable domain,
C is a light chain constant domain,
X1 is a linker with the proviso that it is not CH1, and
X2 does not comprise an Fc region; and
n is 0 or 1;
wherein said four polypeptide chains of said binding
protein form four functional antigen binding sites.
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

5

The polypeptides labeled as [A] comprise the heavy chains. When X2
is 1, the binding protein has an Fc domain. When X2 is 0 for polypeptide
[A], the binding protein lacks an Fc region. The polypeptide labeled as [B]
comprises the light chains. [A] and [B] assemble to form four functional
antigen binding sites. The latter are shown above in DVD-Ig, reproduced
above, as two VA binding sites and two VB binding sites. The claim does
not require VA and VB to bind to different antigens and thus are not limited
to bispecific antibodies.
Rejections 2 and 3 focused on the embodiment in which the claimed
binding protein comprises the Fc region in which VA and VB bind to
different antigen targets. Both Abbott and Sanofi addressed this
embodiment in their briefings.

REJECTIONS
The following two rejections were adopted in the Decision. The full
citations to the publications can be found in the Decision and are not
repeated here.
2. Claim 1–3, 5–7, 31, and 32 of the ’181 patent as obvious under 35
U.S.C. § 103(a) in view of the ’830 patent, Müller, and Lu 2005.
3. Claim 1–3, 5–7, 31, and 32 of the ’181 patent as obvious under 35
U.S.C. § 103(a) in view of the ’830 patent, Miller (2003), Müller, Lu (2005),
Lu (2003), Lu (2004), and Zuo (2000).
CH1-CHL
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

6
Claim 1 comprises [A] a heavy chain comprising the VD1 and VD2
heavy chain domains with a heavy chain constant domain C joined to heavy
chain VD2 (VD1-VD2-C heavy); and [B] a light chain comprising the VD1
and VD2 light chain domains with a light chain constant domain C joined to
light chain VD2 (VD1-VD2-C light). The prior art heavy chain constant
domain C and light chain constant domain C are referred to as CH1 and CL,
respectively. Since the CH1 and CL domains pair with each other, they are
referred to collectively throughout this decision as CH1-CL. The CH1-CL
domains are depicted in DVD-Ig reproduced above as attached to the heavy
and light variable domains Va.
In the Decision, we discussed the CH1-CL domains as follows:
With respect to the inclusion of the CH1-CL domains in
the claimed binding protein, Dr. Kontermann acknowledged that
Müller utilized such domains and that this “approach can also be
used to produce tetravalent IgG-like constructs as described in
Lu (2004) and Zuo [(]2000[)].” (Kontermann [1] Decl. ¶ 56).
Thus, Dr. Kontermann does not appear to dispute the
obviousness of utilizing a CH1-CL domain to have made the
claimed binding protein. However, Dr. Kontermann’s argues
that the latter publications teach that crossing-over does not
occur when CH1-CL is fused to scFv [single chain variable
antigen fragment] units and thus the skilled worker would not
had reason to have made the claimed configuration with CH1-
CL attached to both the heavy and light chain (id. at ¶¶ 56-57).
Dr. Kontermann’s testimony is speculative since both Zuo
[(]2000[)] and Lu [(]2004[)] utilized CH1-CL domains for each
of the scFv binding sites.
Dec. 18.
Dr. Kontermann responded in his Declaration prepared after the
Decision that his “prior testimony has been misunderstood.”
Kontermann 2 Decl. ¶ 42. Dr. Kontermann clarified that the
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

7
paragraph in his first Declaration cited in the Decision referred to the
use of CH1-CL to hold two separate scFv binding units together of
different specificities. Kontermann 2 Decl. ¶ 42; Kontermann 1 Decl.
¶ 56. A scFv is antibody binding domain made of a VH and VL
domain linked by a flexible peptide. Kontermann 2 Decl. ¶ 32. In
contrast, Dr. Kontermann explained that the CH1-CL domain in the
claimed antibody is used to pair the light and heavy chains of a single
antigen binding site, as shown in the DVD-Ig reproduced above.
Kontermann 1 Decl. ¶ 56. Dr. Kontermann also pointed out that
paragraph 42 in his second declaration explained that the results of Lu
(2004) and Zuo (2000) teach “that two different chains composed of
complementary variable domains (VH-VL in the examples) are not
expected to reliably pair when fused to a CH1 and CL domain or to a
Fe region.” Kontermann 2 Decl. ¶ 42. We agree with Dr.
Kontermann that we erred in stating that “Dr. Kontermann does not
appear to dispute the obviousness of utilizing a CH1-CL domain to
have made the claimed binding protein.” Dec. 18. As discussed in
more detail below, upon reconsideration in view of the additional
evidence adduced during the return to prosecution, we find that the
evidence does not support the obviousness of utilizing a CH1-CL
domain to have made the claimed binding protein having a VD1-
VD2-CH1 heavy chain and a VD1-VD2-CL light chain.
DISCUSSION
The claimed binding protein comprises [A] a heavy chain with two
variable heavy chain domains linked together and [B] a light chain with two
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

8
variable light chain domains linked together. Chains [A] and [B] pair to
form to functional antigen binding domains. The light and heavy chains are
segregated from each other.
The ’830 patent is cited in Rejections 2 and 3 for its teaching of
variable light domains (VL) and variable heavy domains (VH) on separate
chains which associate to form two antigen binding sites, the same
configuration recited in the claimed antibody and as depicted in the drawing
of DVD-Ig reproduced above.1 This configuration is reproduced below and
is referred to as the “double head complex”:

1 The configuration of the ’830 patent, with the light and heavy domains
segregated from each other, is mentioned in several of the publications cited
by Abbott as praising the claimed invention. For example, Presta (2008)
wrote: “An elegant solution to the problem was reported recently in which
one VL domain was fused at its C-terminus to the N-terminal portion of a
second, different VL domain (i.e.[,] VL1-VL2-CL) and the partner VH1
domain fused in a similar manner to VH2 (i.e.[,] VH1-VH2-CH1).” Presta
468. We note that, while Presta clearly praised the configuration described
in the ’830 patent, Presta additionally describes the CH1-CL domains as part
of the construct, domains which are not described in the ’830 patent.

Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

9

The drawing above, reproduced from the ’830 patent, shows a double
head antibody fragment (middle drawing) comprised of two variable light
chain domains joined together (leftmost drawing) and two variable heavy
chains joined together (rightmost drawing). When the separate chains
associate, they form a double-head antibody with two functional antigen
binding sites as in claim 1 of this appeal.
The double-head antibody fragment of the ’830 patent lacks 1) CH1-
CL domains, with CH1 added to a heavy chain and CL added to a light
chain; and 2) the optional constant region CH2/CH3. To meet this
deficiency, Miller (2003), Müller, Lu (2005), Lu (2003), Lu (2004), and Zuo
(2000) were cited. Miller (2003), Müller, Lu (2004), and Zuo (2000)
describe constructs with the CH1-CL domains. Lu (2003) and Lu (2005)
describe constructs with the CH3 and CH2/CH3 domains, respectively. Dec.
8, 10. The discussion below focuses on the cited prior art publications
which teach CH1-CL in an engineered antibody construct.

Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

10
Zuo (2000) and Lu (2004)
Each of Zuo (2000) and Lu (2004) show the same basic antibody
configuration: 1) single chain antibodies formed by a single chain
comprising the light and heavy chains fused together (“scFv”) to form an
antigen binding site (two constructs have four scFv antigen binding sites and
one construct has two scFv binding sites); and 2) the CH1-CL constant
region which dimerizes two single chain scFV antibodies together. The
claimed antibody, as exemplified by DVD-Ig, however, does not use the
CH1-CL region to dimerize two antigen binding sites together as in Zuo
(2000) and Lu (2004), but instead uses the CH1-CL region to pair the heavy
and light chains of Vb-Va together.
Lu (2004) and Zuo (2000) do not teach utilizing the CH1-CL domain
to pair heavy and light chains together (i.e., a VH-CH1/VL-CL structure).
Instead, Lu (2004) and Zuo (2000) teach dimerizing two antigen binding
units, each of which comprising both a heavy and light chain (i.e., a VH-VL-
CH1/VH-VL-CL structure). Neither the Examiner nor Sanofi articulated an
adequate reason that would have led the skilled worker at the time of the
invention to modify the CH1-CL domain in Zuo (2000) and Lu (2004) to
pair the light and heavy chains together instead of dimerizing two different
antigen binding sites. To illustrate this difference, the drawings from Lu
(2004) and Zuo (2000) are reproduced below, along with DVD-Ig.
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

11

As shown above, Lu (2004) (on top) produced a construct with two
single chain antibodies dimerized with each CH1-CL domain and with a
constant CH2/CH3 region. The CH1-CL domain dimerizes two complete
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

12
antigen binding sites, each having a heavy and light chain. Zuo (2000)
(middle) also produced an antibody with single chain antibodies (scFv
p4G7; scFv p1C11) dimerized by each CH1-CL constant domain such the
CH1-CL domain dimerizes complete antigen binding sites, each consisting
of a heavy and light chain. In contrast, DVD-Ig uses the CH1-CL domain to
pair the heavy and light chains of the antigen binding site (bottom).
Abbott’s analysis is supported by this evidence because the only way
CH1-CL is utilized in these two cited publications is to dimerize single chain
antibodies (“scFv”). As illustrated by DVD-Ig, CH1-CL is utilized in a
structurally different way in the claimed binding protein. Lu (2004) and Zuo
(2000) would not have suggested to the skilled artisan a structure in which
the CH1 is connected to a heavy chain and CL is connected to a light chain
to pair the segregated light and heavy chains of an antigen binding site, as
recited in the claims.
In contrast, Drs. Davis and Verhoeyen testified that “[k]nowing that
the ’830 approach works, it is not much of a leap for a skilled researcher to
conceive of substituting the scFvs in Zuo 2000 and Lu 2004 (also known to
work) with the ’830 VH-VH and VL-VL constructs and have a reasonable
expectation that it will work.” Davis-Verhoeyen 1 Decl. ¶ 25. However, the
expectation that a construct would work does not provide a reason as to why
it would be made in the first place by one of ordinary skill.

Müller
Müller was also cited as providing a reason to utilize the CH1 and CL
domains “to direct dimerization of polypeptides to which those domains are
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

13
fused.” Decision 9. However, as shown below, Müller utilizes the CH1-CL
domain in the same way as in Lu (2004) and Zuo (2000).

Müller, reproduced above, shows two single-chain antibodies (each
with a heavy and light chain (“scFv”)) dimerized together by CH1 and CL
domains, the same function they served in Lu (2004) and Zuo (2000). The
domains do not pair the light and heavy chains as they do in the claimed
binding protein.

Summary
While we concluded in the Decision that “[t]he evidence shows that
the constant regions of an antibody and additional antigen binding domains
were known to be useful in engineered proteins” (Dec. 26), upon further
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

14
consideration, in view of Dr. Kontermann’s testimony in paragraph 42 of his
second Declaration, we conclude that the CH1-CL constant domains were
not used in the cited prior art engineered antibodies (as they are in the
claimed binding protein) to pair segregated heavy and light chains.
Thus, we do not agree with Sanofi that the domains function in the
same way in the claimed binding protein as in Müller, Lu (2004), and Zuo
(2000). Sanofi 41.77c Comments 13. Sanofi’s statement is not supported
by a preponderance of the evidence in the record presently before us. We
also do not agree that predictability of the accessibility of domains and light
and heavy chain assembly (id. at 13–15) are dispositive because there first
must be a rationale to further modify the CH1-CL domain to segregate the
CH1 and CL with the heavy and light chains of the double-head antibody to
achieve the claimed structure. As reviewed above, Lu (2004), Zuo (2000),
and Müller do not provided a reason to have utilized the CH1-CL domains
as they are used in the claimed binding protein.

Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

15
Miller (2003)
Miller (2003) attaches a constant region in a different way than Zuo
(2000), Luo (2004), and Müller. Drawings from Miller (2003) are
reproduced below.

As shown in the figures reproduced above, Miller (2003) produced a
construct in which the constant region (CH3/CH3-CH2/CH2) is attached to
two CH1-CL regions, where each CH1-CL is attached to an antigen binding
site to dimerize the light (VL) and heavy (VH) chains together. This
structure is similar to how CH1-Cl is used in the claimed antibody (as
illustrated in DVD-Ig above).
There is one key different between Miller (2003) and DVD-Ig. These
include: Miller (2003)’s construct comprises a CH1-CL region for each VL-
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

16
VH antigen binding site (1 and 2 in Miller (2003)’s figure reproduced
above)2 while DVD-Ig has only one CH1-CL region associated with only
one of the two VL-VH antigen binding sites. The question is whether there
is an adequate reason, in view of Müller, Lu (2004), and Zuo (2000), to have
adopted using Miller (2003)’s strategy when adding constant regions to the
double-head antibody of the ’830 patent and whether, in doing so, the skilled
artisan would have arrived at the claimed invention.
In Sanofi’s Appeal Brief, it was argued that it would have been
obvious to have eliminated the second CH1-CL region from Miller (2003):
One of ordinary skill in the art would have been motivated to
remove the outer CH1 and CL domains to reduce the size and
complexity of the TA structure and more importantly because the
'830 patent teaches that that two variable regions from the heavy
or light chain can be linked in series via an amino acid linker (or
directly connected without a linker) without affecting their
binding function. Sanofi also contends that it would have been
obvious to one of ordinary skill in the art, in view of the teachings
of the '830 patent, to link the outer VL domain to the VL domain
of the inner Fab fragment of the TA construct. One of ordinary
skill in the art would have been motivated to link the outer and
inner VL domain because the '830 patent teaches that two VL
domains can be linked in series via an amino acid linker (or
directly connected without a linker) without affecting their
binding function, and because the skilled artisan would recognize
that in the absence of the CH1 and CL domains of the outer Fab
fragment, the outer VH and VL domains would be unlikely to
properly associate.
Sanofi Appeal Br. 20–21.

2 Accordingly, the structure of Miller (2003) has a CH1 linkage on the heavy
chain between the two VH domains (i.e., the VD1 and VD2 domains recited
in the claims), which is expressly excluded in claim 1 (i.e., “X1 is a linker
with the proviso that it is not CH1”).
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

17
The Declarants in the Davis-Verhoeyen 2 Declaration further
explained:
Moreover, we contend that if Miller (2003) had removed the
CH1 and CL domains from the TA construct’s outer Fab
fragment, given the teachings in the '830 patent that tandem
variable domains properly associate and yield two functional
antigen binding sites despite lacking any adjacent constant
domains, the authors would have been motivated to join the outer
VL domain to the inner light chain Fab producing a light chain
with tandem variable domains. Thus, it is our opinion that Miller
(2003) shows that two binding sites (in the form of Fabs in this
case) can be concatenated, and when linked to an Fc region, they
result in a multivalent Ig-like molecule, having an “inner”
binding site that is unaffected by the presence of an extra “outer”
binding site.
Davis-Verhoeyen 2 Decl. ¶ 27.
Sanofi’s reasoning is not supported by a preponderance of the
evidence. As discussed in Dr. Kontermann’s second Declaration, all the
constructs described in Miller (2003) are Fab multimers in which identical
Fab fragments are linked to each other. Kontermann 2 Decl. ¶ 24. A Fab
fragment is the antigen binding portion of an Ig antibody comprising the VL,
VH, CL, and CH1 domains. ’181 patent, col. 18, ll. 12–15; Miller (2003)
4856 (Fig. 1A). Dr. Kontermann testified that the Decision did not
appreciate that Miller (2003) “teaches to multimerize a Fab fragment; the
same Fab moiety is preserved throughout as an intact binding unit, and
multimerization is achieved by connecting multiple Fabs through the heavy
chains only.” Kontermann 2 Decl. ¶ 25.
Dr. Kontermann’s testimony is consistent with the teachings in Miller
(2003). Miller (2003) specifically discloses:
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

18
Based on the observation that cross-linking increases the efficacy
of some mAbs [(citations omitted)], a set of three types of
multivalent mAbs was designed: linear with tri- and tetravalent
Fab repeats, a tetravalent F(ab’)2, and a tetravalent full-length
IgG1. All three classes can be built from a basic linear Fab repeat.
Any VH/CH1 domain of interest can be incorporated into the
system through the use of unique restriction sites situated
between the Fab repeats, i.e., separate PCR fragments of the
VH/CH1 domain with the various restriction sites at the 5’ and 3’
ends allow for construction of two, three, four, or more tandem
repeats of the VH-CH1. This can be expressed with a single L
chain (VL-CL) to generate the multivalent Ab forms.
Miller (2003) 4858 (“Discussion”).
Thus, Miller (2003) describes the versatility of using a single repeated
VH/CH1 domain to produce tandem antigen binding sites assembled with a
single VL-CL chain. Because Miller (2003) teaches that its approach of
using “a basic linear Fab repeat” (VH-CH1 and VL-CL) in tandem (Miller
(2003), Abstract & 4858 (“Discussion”), eliminating a CH1 domain from
one tandem domain but not the other tandem domain would be contrary to
the express purpose of Miller (2003)’s “Fab repeat” strategy for constructing
multimeric antibodies. The claimed binding protein, in contrast, uses VLA-
VLB-CL to achieve an antibody with two antigen binding sites and using
this one chain enables making an antibody with the same or different antigen
binding sites.
Sanofi states that the skilled worker would have eliminated the “outer
CH1 and CL domains to reduce the size and complexity of the TA structure”
of Miller (2003)’s construct, but did not point to scientific or declaration
evidence to support this statement. Sanofi Appeal Br. 20. Sanofi also did
not address the specific teachings in Miller (2003), as described by Dr.
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

19
Kontermann, that “the same Fab moiety is preserved throughout as an intact
binding unit, and multimerization is achieved by connecting multiple Fabs
through the heavy chains only.” Kontermann 2 Decl. ¶ 25. Sanofi did not
explain the obviousness of applying Miller (2003) in the context of the
additionally cited Müller, Lu (2004), and Zuo (2000) publications which use
the CH1-CL domains to dimerize single chain Fab fragments.
Sanofi also argues that the ordinary skilled worker
would have been motivated to link the outer and inner VL
domain because the ’830 patent teaches that two VL domains can
be linked in series via an amino acid linker (or directly connected
without a linker) without affecting their binding function, and
because the skilled artisan would recognize that in the absence of
the CH1 and CL domains of the outer Fab fragment, the outer
VH and VL domains would be unlikely to properly associate.
Sanofi Appeal Br. 20–21.
This argument is not persuasive. First, the fact that the VL domains
can be linked in series and still retain activity is not a sufficient rationale to
have done it. As discussed above, Miller (2003) describes specific reasons
for utilizing a CH1 domain with each VH region, namely to achieve binding
site multivalency with Fab repeats and a single VL-CL light chain. Second,
Sanofi has not directed us to compelling evidence that “the outer VH and VL
domains would be unlikely to properly associate” in the absence of CH1/CL
and why this would have made it obvious, rather than non-obvious, to have
eliminated the CH1/CL domain from the outer Fab fragment.

OBVIOUSNESS IN VIEW OF NATURAL IgG
Sanofi contends that the claimed binding protein is obvious because it
“is essentially a natural IgG molecule that includes an extension of both
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

20
heavy and light chains to include an additional heavy chain variable domain
on the heavy chain and an additional complementary light chain variable
domain on the light chain.” Davis-Verhoeyen 2 Decl. ¶ 10. See also Sanofi
Appeal Br. 10 (Jun. 6, 2012).3 We consider this argument in detail below.

The person of ordinary skill in the art
When making a patentability determination under 35 U.S.C., we
consider the claims and the scope and content of the prior art to be directed
to one of ordinary skill in the art. Thus, a determination as to whether the
claims conform to the patentability requirements of 35 U.S.C., including the
requirements of 35 U.S.C. § 103, is made from the perspective of one of
ordinary skill in the art. In this case, the claimed subject matter involves an
engineered binding protein which can be produced using molecular biology
techniques (e.g., ’181 patent, col. 78). The binding protein utilizes
immunoglobulin domains, including heavy and light chains and the Fc
region. The cited prior art includes patents and peer-reviewed scientific
journal publications. The latter is the forum for scientists who typically have
Ph.D. or are working towards a Ph.D. or master’s degree in the pertinent
field. Here, the pertinent fields are immunology and molecular biology,
evidenced by the subject matter of cited prior art publications. Accordingly,

3 “Sanofi contends that it would have been obvious to one of ordinary skill
in the art at the time the ’181 patent was filed, in view of the ’830 patent and
Miller (2003), to add the constant region backbone of a natural IgG molecule
(CH l-CH2-CH3/CL) to the tandemly arranged variable region molecules of
the ’830 patent in order to accomplish the claimed invention of the ’181
patent.”
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

21
the prior art reflects that the person of ordinary skill in the art is a scientist,
who publishes work in peer-reviewed scientific publications, has at least a
post-graduate level understanding of immunology and molecular biology,
and is able to perform engineering on proteins to produce a protein with
desired binding activity.

Natural IgG
To support the statement that DVD-IgG is essentially a natural IgG
with variable light and heavy chain extensions, Sanofi cited a book chapter
by Dr. Kontermann in which he stated the DVD-Ig antibodies “‘look like
normal IgG antibodies with the exception that a second VH and VL domain
is fused N-terminal of the VH and VL domain of the heavy and light chain,
respectively.’” Davis-Verhoeyen 2 Decl. ¶¶ 10, 25.
Drs. Davis and Verhoeyen also testified:
Moreover, by adding constant regions to an antibody fragment,
disulphide covalent bridges that normally occur between the
constant region domains would have formed as in a natural IgG
molecule and, hence, disulphide covalent bridges were also not a
novel feature of DVD-Ig molecules. Finally, when Dr.
Kontermann discusses “non-covalent linkages that were, by
design, not present in either the diabodies or the double heads of
the ’830 Patent,” Second Kontermann Declaration ¶ 15, we
believe that he must be referring to non-covalent interactions
between the constant regions, which are an integral feature of
constant regions in Ig molecules and therefore also not novel.
Id. ¶ 11.
Based on the knowledge of antibody structure, Drs. Davis and
Verhoeyen further testified:
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

22
. . . an awareness of antibody structure and how this structure
naturally facilitates correct assembly and concatenation of
multiple domains, it is our opinion that the concept of adding a
constant domain to tandem VL-VL domains and CH1-CH2-CH3
domains to tandem VH-VH domains does not require as great a
leap of the imagination as AbbVie [Abbott] argues.
Id. ¶ 20. Drs. Davis and Verhoeyen stated that because of the similarities to
natural Ig and the teachings in the ’830 patent, it would have been
predictable that interchain pairing would take place to produce a functional
binding protein. Id. ¶¶ 29, 30.
As indicated by Drs. Davis and Verhoeyen, there are strong
similarities between the structure of DVD-Ig and a natural antibody. The
drawings below show the structure of DVD-IgG (as illustrative of the
claimed binding protein) and of a natural IgG (taken from Miller (2003)):

The drawing on the left is of DVD-Ig; the drawing on the right is of
natural IgG. As indicated in the second declaration of Davis and Verhoeyen,
DVD-Ig differs from a natural IgG in having a second VH-VL antigen
binding site (“Vb”) at the end of the antibody. Vb in DVD-Ig is not attached
to the CH1-CL region normally associated with a variable antigen binding
domain.
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

23
In her Declaration, Dr. Reichert asserts that the skilled artisan “would
not have manipulated the double heads in the ’830 Patent as proposed in the
Decision, namely by adding structure to it ... because doing so would have
been counterintuitive and against the very purpose for using antibody
fragments.” Reichert Decl. ¶ 16. Dr. Reichert further stated “a person
having ordinary skill in the art in August 2005 would have thought that they
would be diminishing the supposed benefits of using fragments in the first
place – their less complex nature and their ability to reach areas that the
larger, full-length antibodies could not.” Id. Dr. Kontermann also expressed
a similar opinion. Kontermann 2 Decl. ¶ 14.
However, Drs. Davis and Verhoeyen responded in their second
declaration:
Dr. Reichert’s assertion is clearly undercut . . . by Alt et al., FEBS
Letters 454: 90-94 (1999) (“Alt”) (Exhibit C), which lists Dr.
Kontermann as the paper’s corresponding author, and which
describes single-chain diabodies (scDbs) fused to constant
domains. Specifically, the paper describes the benefits of
engineered antibody fragments and then goes on to explain the
reasons for adding extra structures to antibody fragments, stating
that:
In lacking the constant regions, recombinant
antibody fragments also lack the associated effector
functions and are rapidly cleared from serum due to
their small size. However, certain therapeutic
applications require an IgG molecule with its Fc-
mediated effector functions and its long serum half-
life.
We sought to combine the advantages of
single-chain diabodies with those of the IgG-Fc
portion [by] creating IgG-like molecules. These
molecules should possess the immune response-
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

24
mediating and pharmacokinetic properties of
immunoglobulin molecules as well as four antigen
binding sites.
Alt, p. 90 (internal citations omitted).
Davis-Verhoeyen 2 Decl. ¶ 15.
Fig. 1 of Alt illustrates this concept:

Alt’s Fig. 1 illustrates a single-chain bispecific construct on the
far left, and then two such single-chain bispecific constructs linked by
constant regions in the middle and on the right (CH2CH3 is Fc). The
engineered antibody on the right has two bispecific antibodies and is
therefore tetravalent (four binding sites) as is DVD-Ig.
Thus, the second Davis-Verhoeyen declaration, with a specific
citation to Alt, provides evidence of a reason to have utilized two
bispecific variable regions linked together by constant regions to
achieve Fc functionality in a tetravalent bispecific antibody.
In the previous Decision, we had determined that it would have
been obvious to have utilized the Fc region on the ’830 patent double-
head antibody for the reasons set forth in the Request for
Reexamination, that is, “to promote secondary immune function, lead
to longer serum half-life, allow for simple purification by affinity
chromatography, and permit efficient production in mammalian
cells.” DOA 9 (quoting from Request for Reexamination 41). The
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

25
further evidence provided by Requester of adding an Fc region to an
engineered bispecific antibody as described in Alt, co-authored by
Patent Owner’s own declarant, provides strong evidence of the
obviousness of utilizing Fc on a bispecific antibody fragment, such as
the bispecific double-head antibody of the ’830 patent.
CH-CL1 domain
Sanofi provides several reasons as to why one of ordinary skill in the
art would have further utilized the natural IgG CH1-CL domain when
attaching the Fc region to the bispecific antibody of the ’830 patent.
First, as indicated by Sanofi and illustrated by the drawings above, it
is not new to utilize a CH1-CL domain in an antibody. The CH1-CL domain
is part of the natural IgG antibody where it is located between the variable
and Fc regions. Consequently, the skilled worker would have had reason to
use it to produce an engineered antibody that mimicked the natural antibody
as closely as possible. Consistently, Alt also expressly described producing
“IgG-like antibodies.” Alt, p. 90.
Second, Drs. Davis and Verhoeyen testified that the CH1-CL region
would provide stability:
20. If a normally skilled antibody engineer (aware of the
implications of the modular nature of antibodies) wanted to
utilize all means to ensure that a bispecific or bivalent antibody
construct was as stable as it could possibly be, then they would
immediately think of adding the appropriate CH1 and CL
domains. This would introduce an appropriately positioned
disulphide bond between the two chains and add an extra set of
non-covalent inter-chain association forces. Such a strategy
would not be a creative leap, it would merely be following well-
trodden paths of protein engineering and copying construction
principles utilized naturally in the highly evolved vertebrate
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

26
immune system. There is no better way to enhance stable
association between the two chains than to add constant domains.
Davis-Verhoeyen 1 Decl. ¶ 20. Thus, the addition of the CH1-CL domain
would add a disulfide bond and provide better stability to the construct in the
same way it does for the natural IgG.4 Accordingly, one of ordinary skill in
the art would have had reason to use CH1-CL domain in combination with
the double-head antibody described in the ’830 patent for its expected
benefit.
In addition to these two reasons, Sanofi also stated that “the more
pairing domains there are between two chains, the stronger the association
will be between these two chains.” Davis-Verhoeyen 2 Decl. ¶ 22. Thus,
providing a CH1-CL would be seen as increasing the strength of association
between the chains, providing a rationale to use it in addition to the benefit
of an additional disulfide bond.
Sanofi further cited a statement in a review article published by
Pluckthun (Andreas Pluckthun & Peter Pack, New protein engineering
approaches to multivalent and bispecific antibody fragments, 3
Immunotechnology, 83–105 (1997)) of using a constant CL region to
achieve dimerization:
Beside these minimal motives based on helical structures,
fusions to dimerizing enzymes and Ig domains as dimerization
devices have also been investigated, but will not be discussed in
detail. McGregor et al. have reported the use of whole k chain in

4 As indicated by Drs. Davis and Verhoeyen, the role of CH1 and CL in
antibody assembly was well known at the time of the invention: “IgG
assembly is stabilized by an interchain disulfide between CL, the LC [light
chain] constant domain, and CH1, the first HC [heavy chain] constant
domain.” Elkabetz et al., J. Biol. Chem. 280(15):14402-12, April. 15, 2005.
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

27
a VH-linker-VL-CL molecule to achieve dimerization, while the
CH3 domain has been fused directly to a scFv antibody or via a
hinge region as discussed above.
Pluckthun 93.
As indicated by this statement, a constant region CL was used to
achieve dimerization in an engineered antibody, namely by the non-covalent
interactions between the natural CL domains. While the publication cited by
Pluckthun did not utilize CL and CH1 domains, nonetheless, it demonstrated
that the natural light chain constant region had an apparent benefit in
promoting dimerization between antibody chains, supplying another reason
to have mimicked the natural antibody by retaining the CH1-CL region
when attaching the constant Fc region to the variable regions of the double-
head antibody of the ’830 patent.
While we had discussed the non-obviousness of utilizing the CH1-CL
constant regions based on the teachings in Müller, Lu (2004), and Zuo
(2000), in each of these cases, the regions were utilized to pair or dimerize
protein chains together. See particularly, Zuo (2000) and Müller in which
the CH1-CL domains pair single chain variable regions. This provides
further evidence that the function of CH1-CL to stabilize protein chains and
promote dimerization was known at the time of the claimed invention.
Abbott’s principal argument against employing the CH1-CL region is
lack of predictability of achieving the correct protein folding. For example,
Dr. Kontermann stated:
In the DVD-Ig format, the dynamics of association and
dissociation of variable domains would be understood by a
person skilled in the art to be incomparably different from the
dynamics of an isolated double head fragment. For this reason, a
person skilled in the art could not know whether both inner and
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

28
outer binding sites would properly form within the constraints of
the novel DVD-Ig format, whether both binding sites could be
functional, whether the outer binding sites would exhibit
stability, or whether the inner binding sites constrained at both
C-terminus and N-terminus would be both functional and
accessible to antigen.
Kontermann 2 Decl. ¶ 29.
Dr. Kontermann also stated “it must also be considered that adding
CH1 and CL constant domains adjacent selected variable domains would be
expected to change the nature of interchain interactions and possibly hinder
the ability of upstream domains to associate.” Id. at ¶ 30.
Dr. Kontermann did not provide adequate evidence to support this
opinion. As discussed by Sanofi, the prior art showed numerous engineered
antibodies, having multiple binding sites and constant regions which were
still capable of binding antigen. See, e.g., the drawings of Lu (2004), Zuo
(2000), Müller, and Miller (2003) reproduced above. Indeed, these prior art
antibodies utilized similar domains as in the claimed antibody in even more
foreign ways to a natural IgG then they are used in DVD-Ig, yet assembly
into at least some functional binding protein was observed. Thus, there is
significant objective evidence before us of a plethora of engineered antibody
structures, each which folded properly and was capable of some degree of
antigen binding activity.
Without a showing by Dr. Kontermann of substantial failure in the
prior art, we do not have adequate evidence to credit his testimony over the
conflicting testimony of Drs. David and Verhoeyen. For this reason, we do
not find Dr. Kontermann’s testimony persuasive that, because the prior art
engineered antibodies are different from DVD-Ig, they would not provide a
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

29
reasonable expectation of success that the structure of DVD-Ig would fold
properly. Kontermann 2 Decl. ¶¶ 37, 38.
Dr. Kontermann’s testimony that intrachain pairing is favored over
interchain pairing is simply a reflection of the fact that the cited Müller
(1998), Lu (2004), Lu (2005), and Zuo (2000) publications were utilizing
single chain antibodies that required intrachain pairing to reconstitute the
antigen binding site. Kontermann 2 Decl. ¶¶ 32, 33, 34, 44. However, in
view of the fact the DVD-IgG utilizes a structure similar to the native IgG
and interchain pairing was established to occur in the ’830 patent’s double-
head antibody to produce functional antigen binding sites, we do not
consider this argument persuasive as to the issue of unpredictability.
Paragraph 35 of the second Kontermann declaration describes the
production of non-functional dimers in Lu (2005). See also Kontermann 2
Decl. ¶ 55. However, Lu (2005) proposed a solution to this problem, and
did not consider it a deterrent to making functional antibody.5 Drs. Davis
and Verhoeyen responded that, nonetheless, fully functional engineered
antibodies were obtained by Lu (2005). Davis-Verhoeyen 2 Decl. ¶ 31.
Thus, while a specific combination of antibody chains in Lu (2005) may not
have been functional according to Dr. Kontermann, functional antibodies
assembled from the same chains did result. Lu (2005) thus is not evidence
of a lack of expected success.

5 “Careful control of the balance of production of both the scFv-Fc fusion
and the partnering cross-over scFv, for example via genetic manipulation of
the expression vector, to favor the formation of the tetravalent molecule may
reduce the secretion of the nonactive component.” Lu (2005) 19671.
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

30
Dr. Kontermann asserted that the “addition of a new N-terminal
domain would be expected to reduce flexibility of both chains, thereby
calling into question whether the modified heavy and light chains would be
capable of interacting as desired.” Kontermann 2 Decl. ¶ 19. However, the
’830 patent already teaches association of the heavy and light chains to form
a functional bispecific double-head antibody. As concluded in the original
Decision on Appeal, both double-head configurations described in the ’830
patent worked. Dec. 18–19. Thus, Dr. Kontermann’s argument about the
lack of flexibility hindering interactions is not supported by the evidence in
this record and inconsistent with the success described in the ’830 patent.

The ’830 patent
In the Decision on Appeal, we considered arguments about the
unpredictability of the folding of the heavy and light chain segregants to
make a tetravalent DVD-Ig. DOA 14–19. Dr. Kontermann provided
additional testimony in his second declaration about the possibility of non-
functional mispairing and multi-chain aggregates. Kontermann 2 Decl. ¶ 17.
Dr. Kontermann also described examples in the ’830 patent in which he
stated showed that changing the order of domains eliminated binding
activity. Id. at ¶ 18. We have considered these arguments but, for the
reasons already discussed in the previous Decision, are not persuaded by the
evidence that there would not have been a reasonable expectation that a
functional tetravalent antibody would be made. See DOA 14–19. For
example, the cited prior art (Lu (2004), Zuo (2000), Miller (2003)) shows
that even where there are multiple possible combinations between separate
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

31
chains and where only one combination results in the desired antibody, the
desired antibody nonetheless forms in detectable and isolatable amounts.
The disclosure cited by Dr. Kontermann in the ’830 patent that certain
configurations would be less active or inactive (id. at ¶ 19–21) must be
considered in light of the success reported in the ’830 patent of producing a
bispecific antibody fragment with the same basic antigen binding structure
as claimed, albeit without the CH1-CL and Fc regions. The disclosure in the
’830 patent that certain configurations are less active is not a teaching away
because operable embodiments are described, including the one which
serves as the basis of the rejection of the claim. Instead, such disclosure
would be seen as guidance in how to make a functional bispecific antibody
(e.g., switching domain order when a construct is inactive). Routine
experimentation does not undermine obviousness.
Drs. Davis and Verhoeyen provided further evidence of predictability,
citing their own patent in which three functional antigen binding sites
assembled to form a trivalent binding protein. Davis-Verhoeyen Decl. 2 ¶
17–18.
Dr. Kontermann cited EP 2 050 764 (EP ’764) as evidence that the
double-heads of the ’830 patent were prone to incorrect pairing and
aggregates. Kontermann 2 Decl. ¶ 20. However, EP ’764 was filed Oct. 15,
2007 and published Apr. 22, 2009, both dates which are after the filings
dates of the ’181 patent (provisional applications filed Aug. 19, 2005 and
Nov, 2, 2005; regular application filed Aug. 18, 2006). EP ’764 is therefore
not prior art to the ’181 patent. It has not been established that the
procedures followed in EP ’764 were indicative of the state of the art on the
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

32
filing date of the ’181 patent. Consequently, we have not given the
statements in EP’764 significant weight.
While we do not agree with Dr. Kontermann’s opinion that any
unpredictability that DVD-Ig antibody could be made would have
established a lack of a reasonable expectation of success, we acknowledge
that the DVD-Ig was praised for its expression and stability, as discussed in
detail below. Industry praise has been found to be probative evidence that
one of ordinary skill in that art would not have expected success. Institut
Pasteur & Universite Pierre Et Marie Curie v. Focarino, 738 F.3d 1337,
1347 (2013). We address this praise, and other evidence of secondary
considerations, in more detail below.

SECONDARY CONSIDERATIONS
To establish the non-obviousness of the claimed DVD-Ig, Abbott
provided evidence of copying, industry praise, and long-felt need. While
industry praise, copying, and long-felt need are probative evidence of the
patentability of a claimed invention, “[t]o be afforded substantial weight, the
objective indicia of non-obviousness must be tied to the novel elements of
the claim at issue.” Institut Pasteur & Universite Pierre Et Marie Curie v.
Focarino, 738 F.3d 1337, 1346–47 (2013). In addition to this, there are
numerous cases in which objective indicia of non-obviousness were
insufficient to overcome a “strong” case of obviousness. For this reason, we
are compelled to take a rigorous look at the invention through the lens of the
prior art to determine whether the objective indicia of nonobviousness,
which we acknowledge, are present here, are due to the merits of invention.
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

33
Because the parts of an invention are almost always described in the prior
art, and it’s their combination which constitutes the invention, the analysis
must focus on whether the objective indicia of non-obviousness result from
individual known parts functioning in their known way as expected or from
the combination, itself.

Copying
Abbott asserted that Sanofi copied the claimed DVD-Ig. Specifically,
Abbott provided evidence that Sanofi is developing a DVD-Ig binding
protein as a new drug. Abbott’s Comments after Examiner’s Determination
19. However, Abbott acknowledged statements made by Sanofi that such
binding protein was independently derived. Id. A declaration by Ercole
Rao, Ph.D. (dated Nov. 24, 2011) was provided by Sanofi as evidence of this
independent derivation. Abbott did not provide arguments or evidence to
rebut Dr. Rao’s declaration.
Not every competing product that arguably falls within the scope
of a patent is evidence of copying. Otherwise every infringement
suit would automatically confirm the nonobviousness of the
patent. Rather, copying requires the replication of a specific
product. This may be demonstrated either through internal
documents . . . ; direct evidence such as disassembling a patented
prototype, photographing its features, and using the photograph
as a blueprint to build a virtually identical replica.
Iron Grip Barbell Co., Inc. v. USA Sports, Inc., 392 F.3d 1317, 1325 (Fed.
Cir. 2004). As copying has not been established here, we do not find
Abbott’s argument supported by sufficient evidence.

Praise
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

34
We have focused on seven different publications provided by Abbott
as evidence that its DVD-Ig antibody had been praised in scientific
publications. See Appendix (“Praise”) excerpting pertinent passages from
the publications. See also Patent Owner’s Response after Board Decision
11–13. Each of the publications, authored by scientists other than the
inventors of the DVD-Ig technology, discussed the results obtained by the
inventors as reported in a scientific publication of their work, Wu et al.,
Simultaneous targeting of multiple disease mediators by dual-variable-
domain immunoglobulin, Nature Biotech., 25(11): 1290– 1297, Nov. 2007
(“Wu (2007)”).
As discussed in the original Decision (DOA 22-23), in several of the
publications, the authors specifically identified the segregation of the heavy
and light variable domains into separate chains as part of the “solution” and
“breakthrough” to problems encountered in bispecific antibodies. See Presta
(2008), Fischer (2008), Morrison (2007), Beck (2010), and Michaelson
(2009).
However segregation of the heavy and light variable domains on
separate chains (heavy chain segregant; light chain segregant) is described
for the double-head antibodies of the ’830 Patent.
The present invention provides a bispecific or bivalent antibody
fragment analogue, which comprises a binding complex
containing two polypeptide chains, one of which comprises two
times a variable domain of a heavy chain (VH) in series and the
other comprises two times a variable domain of a light chain (VL)
in series.
’830 patent, col. 4, ll. 60–65 (emphasis added).
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

35
Presta (2008) also states the DVD-Ig solved the “light chain
problem.” The so-called “light chain problem” described in Presta (2008)
occurs when “one links two heavy chains (be they VH, VH-CH1, or full-
length IgG) either chemically or by gene engineering, one is left with the
difficulty of assuring that the two light chains partner with their correct
heavy chains.” Presta (2008) 468. Presta (2008) stated that DVD-Ig solved
this problem by fusing the variable light chain domains on a first chain and
fusing the variable heavy domains on a second chain. Id. However, the
fusion of light and heavy chain variable regions to produce light and heavy
chain segregants had been described in the ’830 patent for a single bispecific
antibody fragment.
While the ’830 patent did not produce a tetravalent antibody with two
bispecific binding fragments as claimed, Dr. Kontermann recognized that
Miller (2003) “achieved a tetravalent binding protein but avoided the light
chain problem by using the same VH-CH1 moiety and the same light chain
(VL-CL) throughout.” Kontermann 2 Decl. ¶ 23. DVD-Ig uses the same
heavy chain VHA-VHB and light chain VLA and VLB to make the
tetravalent antibody, the same principle described in Miller (2003).
Consequently, we do not find that Dr. Kontermann’s statement that “the
teachings of the ’830 Patent and Miller (2003) are not relevant to each other”
to be supported by the evidence in this record. Kontermann 2 Decl. ¶ 23.
Several of the publications recognized that DVD-Ig had good yields
and addressed poor expression encountered with other bispecific antibodies.
See Presta (2008), Fischer (2008), Morrison (2007), Beck (2010), and
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

36
Gavrilyuk (2009). Gavrilyuk (2009) described DVD-Ig as a “promising
breakthrough.” Gavrilyuk (2009) 3716–17.
It is not evident from these publications or the record before us
whether the praised stability/expression levels is a function of solving the
light chain problem by following the principles in the ’830 patent and Miller
(2003) and/or whether the addition of CH1-CL/Fc is also responsible for the
praised results. However, as discussed above, CH1-CL was expected to
improve assembly and stability.
Dr. Reichert’s declaration was not helpful in showing that the praise
was for a novel aspect of the invention. Dr. Reichert denied that the ’830
patent’s double-head antibody was “ground breaking” (Reichert Decl. ¶ 40)
and opined that they were a “marginal” and “abandoned approach that failed
to create a therapeutic antibody.” Id. at ¶ 39. However, we find Dr.
Reichert’s declaration to lack credibility because, as discussed above,
several of the publications cited by Abbott expressly describe the
segregation of the variable regions into light and heavy chains as part of the
elegant solution provided by DVD-Ig, and this aspect is specifically
described in the ’830 patent. Dr. Reichert’s denial of the contribution of the
’830 patent technology is therefore inconsistent with the same publications
cited by Abbott as praising DVD-Ig. Furthermore, as discussed above, the
concept of using one type of heavy chain and one type of light claim to solve
the “light chain” problem is described in Miller (2003). Yet, Dr. Reichert,
who was fully aware of Miller (2003) (Reichert Decl., p. 2, indicating Miller
(2003) was reviewed in preparing her declaration), opines (quoting from
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

37
Presta) that DVD-Ig “provides an ‘elegant solution’ to the long-felt light
chain problem.” Id. at ¶ 27.

Long-felt need
In the previous Decision, we rejected arguments made by Patent
Owner that the claimed DVD-IgG “solved a long-felt need for ‘the
production of a bispecific, multivalent antibody-like construct, producible in
sufficient quality and quantity, which would overcome recognized obstacles
of complexity and lack of efficacy.’ (Patent Owner Resp’t Br. 12;
Kontermann [1] Decl. ¶¶ 69-72).” DOA 24. Specifically, the Decision
stated:
By Dr. Kontermann’s own admission, another type of biospecific [sic,
bispecific] antibody has been approved for therapeutic use
(Kontermann Decl. ¶ 72). Thus, it cannot be said that others had failed
in making biospecific [sic, bispecific] antibodies. Indeed, the prior art
is replete with such examples, e.g., Lu 2005, . . , Müller, Zuo 2000, etc.
Id.
Dr. Kontermann in his second declaration responded:
[T]he long-felt need in the art was not merely for a bispecific molecule,
but for a bispecific molecule that was stable both in vitro and in vivo,
easily expressed in mammalian cells, expressed in quantities enabling
preclinical and clinical studies, easily purifiable, and which hopefully
would solve some of the well-documented drawbacks of other
structures of the prior art, e.g., the low efficacy, severe side effects, and
immunogenicity mentioned by Muller and Kontermann in BioDrugs
(2010), the ‘light chain problem’ mentioned by Dr. Presta in Presta
(2008), and the aggregation problem mentioned by the Requester,
sanofi, in EP 2 050 764 at ¶ 18. Many of these long-felt problems are
solved at once by the DVD-Ig format.
Kontermann 2 Decl. ¶ 53.
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

38
As for the statement in the Decision that another bispecific antibody
had been approved for use. Dr. Kontermann stated this antibody –
catumaxomab – is a “quadroma generated by fusing mouse and rat
hybridomas” which suffers from strong immunogenicity in human patients.”
Id. Dr. Reichert further noted that catumaxomab was only approved for use
in 2009 and has an inherent manufacturability problem in that it is the
product of a quadroma (fusion of two different hybridomas), which leads to
production of numerous non-functional isoforms.” Reichert Decl. ¶ 20.
Dr. Reichert also stated that “it is clear that the DVD-Ig format
provided a solution to at least two persistent and well-documented problems
in the art relating to manufacturability and immunogenicity. Id. at ¶ 25.
While catumaxomab was available as early as 1999 (Davis-Verhoeyen
2 Decl. ¶ 38), as discussed by both Drs. Kontermann and Reichert, the
technology utilized to produce it had the major disadvantages of being made
from non-human hybridomas resulting in strong immunogenicity when
administered to humans and of problems with its manufacture. Even Drs.
Davis and Verhoeyen acknowledged shortcomings in quadroma technology.
Id. Thus, we agree that the problem of a bispecific antibody was not solved
by catumaxomab.
With respect to immunogenicity, however, there is no evidence that
DVD-Ig solved that problem. Davis-Verhoeyen 2 Decl. ¶ 38. As testified
by Drs. Davis and Verhoeyen, the use of human sequences was not new to
DVD-Ig. Id. This testimony is supported by evidence in this record. For
example:
We have seen the development of increasingly sophisticated
techniques to obtain human antibodies, ranging from
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

39
‘humanizing’ . . . to the cloning and expression of genuine human
antibody fragments . . . Clinical trials of several ‘humanized’
antibodies are in various stages of progress, and the results
indicate that HAMA [human anti-mouse antibody] can indeed be
overcome by this approach.
Verhoeyen (1995) 1068.
Dr. Reichert’s testimony that “it is clear that the DVD-Ig format
provided a solution to” a “persistent and well-documented” problem relating
to “immunogenicity” (Reichert Decl. ¶ 25) is inconsistent with the evidence
in this record that the problem had indeed been solved by antibody
humanizing technology, and casts doubt on the credibility of her declaration,
particularly in view of her testimony that as “Editor-in-Chief of mAbs and
President of the Antibody Society, I am uniquely qualified to provide an
opinion on the state of the relevant art as of August 2005, long-felt needs in
the field that the claimed DVD-Ig construct solved.” Id. at ¶ 9. In other
words, such qualifications would mean that she was aware of antibody
humanizing technology, but yet she did not explain how DVD-Ig had
advanced a solution to the immunogenicity problem beyond what was
provided by the prior art humanizing technology. We have not been directed
to any features or limitations in claim 1 which address the so-called
persistent immunogenicity problem.
The Decision pointed to the prior art bispecific antibodies cited in the
Decision as evidence that there were bispecific antibodies in the art, and
therefore there was neither failure nor an “unmet” need for one. DOA 25.
Dr. Kontermann addressed the Zuo (2000), Lu (2004), and Lu (2005)
publications, citing evidence in Lu (2005) of difficulty in their production.
However, as noted by Drs. Verhoeyen and Davis, although issues of
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

40
dimerization were noted, functional and high expression levels were
reported by Luo (2005). Verhoeyen-Davis 2 Decl. ¶ 31. Abbott has not
provided comparative data to determine whether the production levels
observed with DVD-Ig were better than those in, for example, Luo (2005),
or Alt (1999). Thus, we do not agree that Abbott established that the
claimed invention met a long-felt need.
Dr. Reichert stated that the original Decision misunderstood
Morrison’s comments on the expression levels of DVD-Ig. Reichert Decl.
¶ 36. Dr. Reichert also appeared to believe that we had misunderstand that
cited publications were praising DVD-Ig. Id. at ¶¶ 34–37.
We clarify that at no time did we misunderstand that the praise in
Morrison and the other cited publications was to anything other than DVD-
Ig. Rather, the original Decision reached the conclusion based on the
evidence that it was features already in the prior art which accounted for the
successful expression of DVD-Ig. Dr. Reichert quoted from the Decision
that the expression levels of DVD-Ig “’were facilitated by features that
[were] . . . already known in the prior art to express well’” (id.), but did not
identify a defect in this statement. Dr. Reichert merely explained that
Morrison’s comments regarding expression levels represented a recognition
that a complex immunoglobulin presents potential expression problems, but
such problems were overcome by DVD-Ig. Id. We do not deny this.
Instead, based on the evidence in this record, we conclude that the
expression problems were overcome, or avoided, using structures known in
the prior art to facilitate antibody expression, and that there were strong
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

41
reasons to have combined such prior art structures to have arrived at a
binding protein with the scope of the claims.

Claim scope
We further note that the DVD-IgG used in Wu (2007) had specific
linkers between the variable regions; Abbott has not demonstrated that its
results with this construct is indicative of other constructs having different
linkers (see claim 1, “X1 is a linker with the proviso that it is not CH1)”).
Michaelson specifically referred the linker in DVD-Ig as allowing for stable
assembly. See Praise: Appendix. However, such linker is not recited in
claim 1.

Discussion of secondary considerations
Abbott has established that DVD-Ig was praised by the industry.
However, as discussed above, the praise in part was of the light and heavy
chain segregants, a feature described in the prior art ’830 patent. Also, the
idea of using one type of chain to make a tetravalent antibody was described
by Miller (2003). Thus, we do not find the praise with respect to this feature
compelling evidence of non-obviousness.
DVD-Ig also comprises constant domains CH1-CL and the Fc
domain. Sanofi provided fact-based evidence of the obviousness of utilizing
the CH1-CL domain in combination with the double-head antibody of the
’830 patent, namely: 1) added stability due to the presence of an additional
disulfide bond (Davis-Verhoeyen 1 Decl. ¶ 20), 2) increased strength due to
the association of CH1 and CL (Davis-Verhoeyen 2 Decl. ¶ 22); and 3)
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

42
dimerization domains to bring the variable heavy and light chains together
assisting assembly of the antibody (Pluckthun, Müller, Lu (2004), and Zuo
(2000). The CH1 and CL domains serve the same function they do in the
native IgG in stabilizing the antibody with an interchain disulfide bond. Dr.
Reichert referred to Presta’s reference to disulfide bond formation in DVD-
Ig, but did not acknowledge that such bonds were also characteristic of a
natural IgG. Reichert Decl. ¶ 33. Thus, even if the addition of the CH1-CL
domains contributed to the praise, a preponderance of the evidence
establishes that it was obvious to have included them and that the known and
expected properties of these protein domains are responsible for the results.
Thus, while we agree that stability and expression was praised, we have not
been directed to evidence that it was anything more than the expected
improvement from the obvious use of CH1-CL and Fc on the double-head
antibody of the ’830 patent.
We conclude that there is a strong showing of obviousness because
the CH1 and CL domains, as well as the Fc region, were utilized in
combination with the known double-head antibody of the ’830 patent for
their known and expected properties. See e.g., Leapfrog Enters. Inc. v.
Fisher-Price, Inc., 485 F.3d 1157, 1162 (Fed. Cir. 2007); Iron Grip Barbell
Co., Inc. v. USA Sports, Inc., 392 F.3d 1317, 1324 (Fed. Cir. 2004);
Richardson-Vicks Inc. v. Upjohn Co., 122 F.3d 1476, 1483-1484 (Fed. Cir.
1997) (Asyst Techs., Inc. v. Emtrak, Inc., 544 F.3d 1310, 1316 (Fed. Cir.
2008) (“[E]vidence of secondary considerations does not always overcome a
strong prima facie showing of obviousness.”).
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

43
Abbott did not provide adequate evidence linking the praise to the
merits of the invention beyond which was readily available in the prior art.
See Ormco Corp. v. Align Technology Inc., 463 F.3d 1299, 1311–1312 (Fed.
Cir. 2006) (in the context of commercial success). See also In re Baxter
Travenol Labs., 952 F.2d 388, 392 (Fed. Cir. 1991) where it was held that
unexpected results could not be based on the property of a material known in
the prior art for the same purpose. Specifically, DVD-Ig was praised for
having light and heavy chain segregants, features of the ’830 Patent’s
double-head antibody. For this reason, it cannot be said that despite the
teaching in the prior art, the industry still praised the invention; rather, the
praise was directed, at least in part, to a feature that already been described
in the prior art, namely in the ’830 patent and Miller (2003). The benefit of
the CH1 and CL domains in improving stability and antibody assembly was
a well-known function of these domains.
At the same time, we recognize that DVD-IgG comprising
combination of heavy and light segregants and CH1-CL and Fc domains has
been praised in numerous publications, providing strong evidence of the
non-obviousness of the claimed DVD-Ig. See Pasteur, 738 F.3d at 1347.
Abbott had the opportunity to establish that it was the unique combination
that produced the praised features of stability and expression, and not the
expected benefits of the Fc, CH1-CL, and heavy and light chain segregants.
But they did not. In our opinion, because each one of these structures was
utilized for its known and expected benefit, and there were strong reasons
for assembling them together, the praise of the invention is not adequate to
establish non-obviousness
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

44
In sum, we conclude, that despite the industry praise, it has not been
established that the praise was due to the merits of the claimed invention,
rather than of the known and expected advantages of utilizing heavy and
light segregants joined to the constant regions of a natural IgG. It would
have been obvious to one of ordinary skill in the art to have attached the Fc
domain to gain its known effector function as described in Alt. The further
use of the CH1-CL region would have been obvious based on its use in other
engineered antibodies, its known stabilizing function in natural IgG, and the
fact that it is configured in DVD-Ig in the same way it is exists in the natural
IgG, i.e., adjacent to the Fc region. When the solution was so readily
available in the prior art, and the person of skill in the art was a scientist
highly skilled in protein engineering and immunology, in our opinion, the
broad claims to a bispecific antibody, comprising heavy and light chain
segregants arranged with constant regions as they are in nature is not
warranted.

SUMMARY
For the reasons discussed above, the Examiner’s Decision in the
Answer and Right of Appeal Notice not to adopt Rejections 2 and 3 is
affirmed. However, based on new evidence and arguments, we find the
claims obvious in view the ’830 patent and the evidence of Alt, Miller
(2003), Müller, Lu (2005), Lu (2003), Lu (2004), and Zuo (2000).
This 37 C.F.R. § 41.77(f) Decision is final.

Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

45
TIME PERIOD FOR RESPONSE
In accordance with 37 C.F.R. § 41.79(a)(1), the “[p]arties to the
appeal may file a request for rehearing of the decision within one month of
the date of: . . . [t]he original decision of the Board under § 41.77(a).” A
request for rehearing must be in compliance with 37 C.F.R. § 41.79(b).
Comments in opposition to the request and additional requests for rehearing
must be in accordance with 37 C.F.R. § 41.79(c)–(d), respectively. Under
37 C.F.R. § 41.79(e), the times for requesting rehearing under paragraph (a)
of this section, for requesting further rehearing under paragraph (d) of this
section, and for submitting comments under paragraph (c) of this section
may not be extended.
An appeal to the United States Court of Appeals for the Federal
Circuit under 35 U.S.C. §§ 141–44 and 315 and 37 C.F.R. § 1.983 for an
inter partes reexamination proceeding “commenced” on or after November
2, 2002 may not be taken “until all parties’ rights to request rehearing have
been exhausted, at which time the decision of the Board is final and
appealable by any party to the appeal to the Board.” 37 C.F.R. § 41.81. See
also Manual of Patent Examining Procedure § 2682 (9th ed., Rev. 07.2015,
Nov. 2015).
In the event neither party files a request for rehearing within the time
provided in 37 C.F.R. § 41.79, and this decision becomes final and
appealable under 37 C.F.R. § 41.81, a party seeking judicial review must
timely serve notice on the Director of the United States Patent and
Trademark Office. See 37 C.F.R. §§ 90.1 and 1.983.

AFFIRMED; 37 C.F.R. § 41.77(f )
Appeal 2017-001183
Patent 7,612,181 B2
Reexamination Control 95/001,380

46

PATENT OWNER:

YANKWICH & ASSOCIATES, P.C.
(AND ABBOTT BIORESEARCH CENTER)
201 BROADWAY
CAMBRIDGE, MA 02139

THIRD PARTY REQUESTER:

McDONNELL, BOEHNEN, HULBERT & BERGHOFF, LLP
300 S. WACKER DRIVE
32nd FLOOR
CHICAGO, IL 60606