`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`PFIZER, INC., and
`SAMSUNG BIOEPIS CO., LTD.,
`Petitioners,
`
`v.
`
`GENENTECH, INC.,
`Patent Owner.
`____________
`
`Case IPR2017-01489
`Patent 6,407,213
`____________
`
`
`
`REPLY DECLARATION OF JEFFERSON FOOTE, PH.D.
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 1
`
`
`
`B.
`
`TABLE OF CONTENTS
`INTRODUCTION ........................................................................................... 1
`I.
`QUALIFICATIONS AND BACKGROUND ...............................................14
`II.
`III. MATERIALS CONSIDERED ......................................................................14
`IV. LEGAL STANDARDS .................................................................................15
`V.
`PERSON OF ORDINARY SKILL IN THE ART ........................................15
`VI. RESPONSE TO THE WILSON DECLARATION ......................................18
`A.
`Bakground of the Technology .............................................................18
`1.
`State of the art of antibody humanization .................................19
`2.
`State of the art of HER2-positive breast cancer and
`development of anti-HER2 antibodies ......................................32
`The ’213 Patent ...................................................................................36
`1.
`Dr. Wilson’s description of “the invention” .............................36
`2.
`The challenged claims ...............................................................43
`3.
`Dr. Wilson shows no advantages of the ’213 patent ................46
`4.
`The ’213 patent claims are not supported by the ’272
`application .................................................................................51
`The Carter and Presta declarations, and related
`documents, do not show invention of the subject matter
`of the challenged claims ............................................................57
`Claim Construction..............................................................................61
`1.
`“Consensus human variable domain” .......................................61
`2.
`“Lacks immunogenicity compared to a non-human
`parent” .......................................................................................62
`D. Dr. Wilson’s Summary of the Prior Art ..............................................62
`
`5.
`
`C.
`
`i
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 2
`
`
`
`1.
`Queen 1989 ...............................................................................62
`Queen 1990 ...............................................................................64
`2.
`Kurrle ........................................................................................65
`3.
`Chothia & Lesk .........................................................................67
`4.
`Chothia 1985 .............................................................................67
`5.
`Furey .........................................................................................68
`6.
`Hudziak .....................................................................................69
`7.
`Tramontano ...............................................................................69
`8.
`Protein data bank (PDB) ...........................................................71
`9.
`10. Kabat 1987 ................................................................................72
`Response to Dr. Wilson’s Opinions Regarding The Asserted
`Prior Art ...............................................................................................73
`1.
`IPR2017-01488 Grounds 1–3: Kurrle and Queen 1990
`teach non-human CDRs “which bind antigen
`incorporated into a human antibody variable domain”
`(claims 66-67, 71-72, 75-76 and 78) .........................................73
`IPR2017-01489 Grounds 1–7: The challenged claims
`would have been obvious based on the PDB Database
`combined with either Queen 1989 or Queen 1990 (claims
`4, 12, 33, 42, 60, 62, 64-67, 69, and 71-79) ..............................79
`IPR2017-01489 Grounds 1–4: The asserted references
`disclose or suggest the recited substitutions (claims 65,
`75-77, and 79) ...........................................................................85
`IPR2017-01488 Grounds 1–3 and -01489 Grounds 1–2:
`The asserted references disclose or suggest the “lacks
`immunogenicity” limitation (claim 63) ....................................86
`IPR2017-01488 Grounds 2-3 and 8 and IPR2017-01489
`Grounds 2, 5 and 7: The asserted references disclose or
`
`5.
`
`3.
`
`4.
`
`E.
`
`2.
`
`ii
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 3
`
`
`
`6.
`
`7.
`
`suggest the “consensus” limitations (claims 4, 33, 62, 64
`and 69) .......................................................................................89
`IPR2017-01488 Grounds 3-10: POSITAs would have
`selected the recited framework substitutions from the
`prior art candidates as a matter of course (claims 12, 42,
`60, 65-67 and 71-79) .................................................................94
`IPR2017-01488 Grounds 4-7: Kurrle and/or Queen in
`combination with additional prior art would lead
`POSITAs to the recited claim limitations (claims 12, 73,
`74, 77, 79 and 65)......................................................................97
`IPR2017-01488 Ground 7 and IPR2017-01489 Grounds
`1-4: The asserted references disclose or render obvious
`the “up to 3-fold more” binding affinity limitation (claim
`65) ...........................................................................................101
`IPR2017-01488 Grounds 8-10 and IPR2017-01489
`Grounds 6-7: The asserted references disclose or suggest
`humanized antibodies with the recited framework
`substitutions that bind p185HER2 (claims 30-31, 33, 42
`and 60) .....................................................................................102
`No Secondary Considerations Support Non-Obviousness Of
`The Challenged Claims .....................................................................104
`VII. CONCLUSION ............................................................................................110
`
`8.
`
`9.
`
`F.
`
`
`
`
`
`iii
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 4
`
`
`
`
`
`I, Jefferson Foote, declare under penalty of perjury as follows:
`INTRODUCTION
`I.
`
`1.
`
`Counsel for Pfizer Inc. (“Pfizer”) retained me to provide my opinions
`
`regarding U.S. Patent No. 6,407,213 (“the ’213 patent”) (Ex. 1502), which is
`
`assigned to Genentech, Inc., in these inter partes review proceedings. I previously
`
`prepared and submitted a Declaration in support of the Petition in this proceeding,
`
`dated May 23, 2017. (Ex. 1503.) I continue to receive $800 per hour for my services
`
`in connection with these proceedings; no part of my compensation is dependent upon
`
`my opinions given or the outcome of this case.
`
`2.
`
`Since preparing my first Declaration, I have reviewed the Expert
`
`Declaration of Dr. Ian A. Wilson (“Wilson Declaration”), which was submitted by
`
`Genentech in response to my initial Declaration. (Ex. 2041.) Dr. Wilson concludes
`
`that the challenged claims of the ’213 patent I addressed in my first Declaration
`
`would not have been invalid as anticipated or obvious in light of the prior art.
`
`3.
`
`For the reasons discussed in my first Declaration and further below, I
`
`disagree with Dr. Wilson. I have reviewed the evidence that has been submitted in
`
`these proceedings since my first Declaration, including but not limited to, the Wilson
`
`Declaration, and declarations and testimony from named inventors Dr. Paul Carter,
`
`and Dr. Leonard Presta, and it remains my opinion that the challenged claims of the
`
`’213 patent are anticipated by and/or obvious over the prior art. Indeed, the
`
`1
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 5
`
`
`
`
`
`humanization process described in the ’213 patent is indistinguishable from that
`
`described in the prior art, and led the inventors to the humanized variants of 4D5 and
`
`others described in the patent as a matter of course.
`
`4.
`
`As a general matter, prior art investigators, including those in the
`
`Winter Lab at Cambridge University’s MRC Laboratory of Molecular Biology
`
`(including (now) Sir Gregory Winter, Peter Jones, Lutz Riechmann, Martine
`
`Verhoeyen, and myself), Queen and Kurrle all had essentially the same
`
`understanding of the need to humanize murine antibodies if they were to be used as
`
`human therapeutics (in which there was much interest and focus at the time), and did
`
`essentially the same thing to address them as the named inventors on the ’213 patent.
`
`Winter, before any of the others, first attempted to humanize antibodies by
`
`transferring the entire complementarity determining regions (CDRs) from a murine
`
`antibody into a human framework (FR), and discovered that this could result in lost
`
`affinity. (Ex. 1627 (Jones); Ex. 1569 (Riechmann).) He attributed that loss to a
`
`change in the conformation of the CDRs that arose from a clash with the new
`
`(human) FR. No other reasonable conclusion was possible, because the CDRs were
`
`known to form the contact surface that bound antigen and the CDRs in the murine
`
`antibody (functional) and humanized antibody (showing impaired function) were the
`
`same: the FR was the changed element.
`
`2
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 6
`
`
`
`
`
`5. Winter therefore studied the CDR–FR interfaces of crystallographically
`
`solved immunoglobulin structures available at that time, making substitutions on a
`
`molecular graphics system to assess what effect a mouse/human interchange might
`
`have, identifying possible clashes. He tested several possibilities experimentally
`
`through point mutations in which a human FR residue was changed back to the
`
`residue in the corresponding murine FR. A single mutation at H27 was sufficient to
`
`restore affinity. (I myself continued these studies under Winter, developing a
`
`humanized anti-lysozyme antibody, designated “HuLys,” as a structurally well-
`
`defined model for understanding and improving the humanization process, using a
`
`“consensus” sequence as the framework for the light chain. Winter’s colleagues at
`
`the MRC, Anna Tramontano, Cyrus Chothia and Arthur Lesk, studied interface
`
`positions including H71 for the sake of a general understanding of antibody
`
`architecture, but also pointed out the relevance of their findings to humanization.)
`
`6.
`
`Queen subsequently studied the same 3D antibody crystal structures for
`
`the same reason. His proposal, to use human FR sequences that were highly
`
`homologous to the mouse FR (an approach colloquially called “best fit”), once again
`
`supplemented with additional back-mutations to the mouse sequence, achieved the
`
`same goal: maintain key mouse FR structures needed to support the CDRs in a
`
`functional conformation. Queen also made clear that any “unusual” residues in the
`
`3
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 7
`
`
`
`
`
`FR should be avoided, and in his 1990 patent application, explicitly taught the
`
`alternative of using a “consensus” sequence for the human FR acceptor.
`
`7.
`
`Kurrle used a staged approach, and the first stage was identical to
`
`Queen: begin with the most homologous human sequence. The next two stages
`
`introduced back-mutations in FR positions on either side of each CDR. That returned
`
`a humanized molecule that “competed effectively with both the murine BMA 031
`
`and the previously constructed chimeric” in cell binding assays. However, Kurrle
`
`went still further, making clear a desire to push the starting Eu framework toward a
`
`human consensus sequence, where possible. Kurrle used sequence database
`
`comparisons and graphic molecular modeling to reconcile these changes with strong
`
`antigen binding. A long and detailed passage (Ex. 1571 (Kurrle) 8:44 –9:7) gives
`
`individual rationales for ruling in or ruling out particular back-mutations that affect
`
`the consensus. The upshot is that any FR residue in their next stage construct that
`
`was not consensus had been changed for a very specific reason.
`
`4
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 8
`
`
`
`
`
`
`
`(Ex.1571_8.)
`
`8.
`
`The ’213 patent inventors brought similar techniques and knowledge to
`
`humanize the 4D5 antibody that had been developed by others, yielding the variants
`
`described in the patent, with the substitutions recited in the claims, as a matter of
`
`course. Presta, like Winter, Queen, and Kurrle, studied FR-CDR interfaces in known
`
`crystal structures, made molecular models on a graphics system, and–again like the
`
`others–arrived at heavy and light chain FR sequences that placed mouse residues at
`
`points susceptible to clashes with the CDRs. Carter, the experimentalist of this
`
`5
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 9
`
`
`
`
`
`partnership, produced a humanized 4D5 embodying Presta’s basic design, as well as
`
`a series of back-mutated humanized 4D5 versions with improved affinity.
`
`9.
`
`All four research groups (Winter, Queen, Kurrle, Genentech) followed
`
`similar approaches to humanization and arrived at similar results. Winter is
`
`unambiguously the originator of CDR grafting. He also unambiguously originated
`
`back-mutation of human FR residues to restore lost affinity, in the course of which
`
`he explicitly articulated a principle for choosing candidate FR residues to back-
`
`mutate: proximal ones with high potential to alter the conformation of CDRs, as
`
`revealed by molecular modeling or other evidence. The three groups after Winter
`
`sought to systematize that last principle, each devising a combination of both general
`
`criteria and explicitly named sequence positions to guide back-mutation. All three
`
`tested a small number of these mutations experimentally in the exemplary projects
`
`of their patent applications, but claimed a far larger number of untested sequence
`
`positions as uniquely critical to successful humanization, based primarily on
`
`modeling studies. (I myself published such a list with the comprehensive HuLys
`
`paper (Ex. 1706, Foote & Winter, Antibody Framework Residues Affecting the
`
`Conformation of the Hypervariable Loops, 224 J. MOLECULAR BIOLOGY 487, 497
`
`(1991)), but I'm not aware that the Medical Research Council ever sought patent
`
`protection for it.).
`
`6
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 10
`
`
`
`
`
`10. Genentech asserts that a key distinguishing feature of the ’213 patent is
`
`the use in humanization of a “consensus human framework” sequence–a FR
`
`comprising at each position the residue most commonly encountered in all human
`
`immunoglobulins of a particular subgroup. In implementing this approach, Presta
`
`used the most common residues in the subgroup listings in Kabat (1987), while
`
`applying some unstated criteria for choosing between residues where no clear most
`
`common residue was identified.
`
`11. But the concept of a human consensus FR preceded the ’213 filing date
`
`by at least 20 years. As an initial matter, the determination of “variability” on which
`
`the distinction between CDR and FR residues rest requires compiling a consensus
`
`sequence, since its formula requires knowing for each position the number of
`
`different amino acids occurring at the position and the frequency of the most
`
`common amino acid at that position. The very first printed Kabat sequence
`
`compilation, in 1979, disclosed a consensus FR composition for each human and
`
`mouse sequence subgroup, as did every subsequent edition through the last printed
`
`one in 1991. Every table of variable region sequences, for every subgroup in every
`
`Kabat edition, begins with a column of residue numberings, and ends with parallel
`
`columns listing “occurrences of most common amino acid” and variability at each
`
`of those positions.
`
`7
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 11
`
`
`
`
`
`12. Genentech and Dr. Wilson do not appear to be suggesting that
`
`consensus sequences were not known, but only that it was not known to use them as
`
`the FR in humanization projects. But that is clearly not the case. As I understand
`
`was acknowledged during prosecution of the ’213 patent, I used a consensus
`
`sequence derived from Kabat (1983) to humanize the light chain of an anti-lysozyme
`
`antibody starting in 1986 (HuLys, supra ¶ 5). While the full HuLys sequence was
`
`not published until 1992, I described it in intermediate publications, including a
`
`paper for a 1988 symposium, published in 1989 (Ex. 1693, Foote, J., Humanized
`
`Antibodies, NOVA ACTA LEOPOLDINA NF 61 Nr. 269, 103-110 (1989)), which stated
`
`that the HuLys light chain comprised “consensus human kappa frameworks.” (Id. at
`
`106.) Moreover, the identical consensus structure was used in the humanization of
`
`Riechmann’s CAMPATH-1H antibody, and was published the same year as an
`
`explicit sequence. (Id. at 108.) Thus the very first antibody humanized in both chains,
`
`which was also the first humanized antibody administered to a patient, had a
`
`consensus light chain. The “consensus” approach is also explicitly described as an
`
`option by Queen in his 1990 patent application.
`
`13. Another criticism Dr. Wilson makes of the invalidity analysis in my
`
`first Declaration is that the lists in the prior art of framework residues to be
`
`considered for back-mutation is intractably large, and when one considers the
`
`multiplicative expansion of combining some mutations with others, “there are
`
`8
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 12
`
`
`
`
`
`literally millions of potential combinations and permutations of framework
`
`substitutions” based upon the cited prior art, in contrast to the specific substitutions
`
`recited in the claims.
`
`14. But Dr. Wilson ignores the contrary accomplishments by prior art
`
`investigators. Winter (Riechmann), in the CAMPATH-1H project, did not make
`
`“literally millions” of FR mutants, he made two. Kurrle made four. Queen reported
`
`only a single construct. Against these numbers are the eight huMAb-4D5 variants
`
`in the ’213 specification. No POSITA using CDR grafting following the prior art
`
`teachings required millions, thousands, hundreds, or even 20 variants to make a
`
`working humanized antibody.
`
`15. The disparity between Dr. Wilson’s inflated estimate and actual
`
`findings arises because of a departure from basic scientific method, as well as the
`
`patent’s teachings. The latter first: as Dr. Presta testified, narrowing the list of
`
`candidates to a manageable size was first a matter of “simple” sequence alignment
`
`to determine at which candidate positions the mouse donor and human acceptor
`
`differ. At many positions, mouse and human will have the same residue, hence no
`
`substitution need be considered. The departure from scientific method concerns Dr.
`
`Wilson’s suggestion of the way the remaining candidates would be assessed. A basic
`
`principle of search (much of science is a search) is to look for the sought item in the
`
`most likely place first. A POSITA would not begin a search for a corrective back-
`
`9
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 13
`
`
`
`
`
`mutation by making a vast matrix of recombinant plasmids representing every
`
`possible framework mutation in combination with every other possible framework
`
`mutation. Particularly if faced with limited time and resources, a POSITA would
`
`test single substitutions, beginning at residue positions most likely to have a positive
`
`effect, then later, if necessary, combine mutations that had yielded improvements.
`
`As the ’213 patent itself acknowledges in justifying its claims to millions of antibody
`
`variants with up to 47 different possible FR substitutions while only describing a
`
`handful of variants with only a few of the recited positions represented, this was “per
`
`se routine, and well within the ordinary skill in the art.” (Ex. 1502 (’213 patent) at
`
`at 10:28–34.) And in doing so, as Dr. Wilson acknowledged, based on prior results,
`
`the POSITA would expect to be able to achieve about the same binding affinity,
`
`more or less, as the parent. Dr. Presta himself testified that, for any criteria known
`
`to identify candidate FR residues for substitution, this would require testing only
`
`around ten variants to achieve a satisfactory result. (Ex. 1699 (Presta Tr.) at 98:25–
`
`99:5.) At base, the ’213 patent provides no more guidance in selecting from the
`
`candidate residues than the prior art.
`
`16. Nor is “lacking immunogenicity” compared to the parent in any way
`
`inventive. Dr. Wilson rails against “...statements of intended result without any real
`
`assurances that the described humanized antibodies in fact lack immunogenicity…”
`
`(Ex. 2041, ¶ 201), yet the fact is that the ’213 patent provides no immunogenicity
`
`10
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 14
`
`
`
`
`
`data for any claimed antibody. These data-less claims instead fall back on the
`
`inherent properties of humanized antibodies and/or the reasonable expectations of
`
`POSITAs in the field. Given this reliance on prior art and the knowledge of skilled
`
`artisans, Genentech could hardly assert that immunogenicity data is somehow
`
`necessary to demonstrate expectation of meeting the claims’ immunogenicity
`
`requirements. In that regard, the reduced immunogenicity of CDR-grafted
`
`antibodies was widely accepted by 1991. Multiple research groups by then were
`
`preparing new humanized antibodies as possible therapeutic agents, and they would
`
`not take on the expense, time, and risk of preparing a CDR-grafted antibody if they
`
`believed a simple chimeric would work just as well.
`
`17. The expectation of reduced immunogenicity was based on both theory
`
`and experiment. The theory, which I detailed in my earlier declaration, is none other
`
`than the rock-solid foundation of immunology: the immune system tolerates self and
`
`attacks foreign. A critic might fret about the CDRs of a humanized antibody being
`
`a target, or framework back-mutations, but no reasonable critic could deny that
`
`converting two-thirds of each variable domain from mouse to human would remove
`
`otherwise immunogenic target surfaces.
`
`18. Theory aside, Dr. Wilson’s characterization of prior immunogenicity
`
`reduction by humanization as “...aspirational statements...proven wrong in clinical
`
`testing...” (Ex. 2041, ¶ 204) collides most unfavorably with another fact: by the ’213
`
`11
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 15
`
`
`
`
`
`patent priority date a substantial body of experimental data had been published on
`
`the immune response of trial subjects to administered humanized antibodies and the
`
`data all pointed to an unprecedented reduction in immunogenicity. As of 1991,
`
`the humanized CAMPATH-1H antibody had been tested and shown to induce no
`
`immune response in treatment of lymphoma (Ex. 1707, Hale et al., Remission
`
`Induction in Non-Hodgkin Lymphoma with Reshaped Human Monoclonal Antibody
`
`Campath-1H, 332 LANCET 1394-1399, 1398 (1988) (“...failed to detect any
`
`antiglobulin to CAMPATH-1H.”)), vasculitis (Ex. 1708, Mathieson et al.,
`
`Monoclonal Antibody Therapy in Systemic Vasculitis, 323(4) NEW ENG. J. MED.
`
`250-254, 252 (1990) (“no detectable antiglobulin response.”), and rheumatoid
`
`arthritis (Ex. 1709, Kyle et al., Humanized Monoclonal Antibody Treatment in
`
`Rheumatoid Arthritis, 18(11) J. RHEUMATOLOGY 1737-1738, 1737 (1991) (“No anti-
`
`Campath 1H response was detected”).) Similarly, Queen’s humanized anti-Tac
`
`antibody was reported in April, 1991 to be “much less immunogenic [than the
`
`murine parent] when administered to cynomolgus monkeys,” a species widely used,
`
`even in the present day, as a model for drug safety studies of therapeutic antibodies,
`
`owing to its immunological closeness to humans. (Ex. 1710, Brown, Jr. et al., Anti-
`
`Tac-H, a Humanized Antibody to the Interleukin 2 Receptor Prolongs Primate
`
`Cardiac Allograft Survival, 88 PROC. NAT’L. ACAD. SCI. U.S. 2663-2667, 2663,
`
`2666 (1991).)
`
`12
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 16
`
`
`
`
`
`19. Patent Owner asserts that with Campath-1H “3 out of 4 patients
`
`suffered immunogenic response.” (POR at 66; Ex. 2041, ¶ 204.) However, the
`
`implication that this number invalidates the therapy is in disconnect with the actual
`
`conclusions of the 1992 paper from which these data were cherry-picked (Ex. 2025,
`
`Isaacs et al, Humanized monoclonal antibody therapy for rheumatoid arthritis, 340
`
`THE LANCET 748-752 (1992)). In this study, eight patients had ten injections of
`
`CAMPATH-1H and none of the eight showed an anti-CAMPATH response (Table
`
`III column 2). That result was a night-and-day contrast to earlier experience with
`
`murine CAMPATH-1G, to which a rapid response appeared in 11/14 patients. (Id.
`
`at 748–751.) Four patients entered a second round of five injections, and three did
`
`indeed develop anti-CAMPATH antibodies (Table III column 3). (Id. at 750–751.)
`
`However, a meticulous analysis by the authors showed reactions in two of the three
`
`were directed purely to the CDRs, and a more tentative assessment of the third was
`
`of a reaction to the constant region. (Id. at 751.) Evidence of a reaction to variable
`
`domain frameworks, a potential sign of a flawed humanization approach, was not
`
`found. What’s more, the patients’ anti-CAMPATH-1H antibodies did not appear
`
`until 6-10 days after conclusion of treatment, caused no clinical problems, and the
`
`treatment itself was considered highly successful.
`
`20. Given the preponderance of published clinical results, a POSITA in
`
`1991 would see that with CDR grafting, theory and experiment were aligned with
`
`13
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 17
`
`
`
`
`
`the expected result: reduced immunogenicity. The ’213 patent brings no
`
`immunogenicity data to this discussion and would give a POSITA no reason to
`
`believe that humanization projects ongoing at that time were doomed to failure.
`
`21. At base, therefore, as described in my first Declaration, it is my opinion
`
`that the ’213 patent “invention” is nothing more than the result of following the same
`
`humanization processes in the prior art, and achieving the same results. I do not see
`
`any valid invention in the patent’s claims.
`
`22.
`
`I have described my opinions in further detail below.
`
`II. QUALIFICATIONS AND BACKGROUND
`
`23.
`
`I described my qualifications and background in my first Declaration.
`
`(Ex. 1503, ¶¶ 2–9.) I also attached a copy of my curriculum vitae at Exhibit A to
`
`my initial Declaration.
`
`III. MATERIALS CONSIDERED
`
`24. Exhibit B includes a list of the materials I considered in providing my
`
`opinions cited herein. Specifically, I have reviewed the materials cited in this
`
`Declaration, including the Wilson Declaration, and the materials listed in Exhibit B.
`
`I have also considered my own knowledge and experience including my work over
`
`decades in humanizing antibodies.
`
`25. As noted in my first Declaration (Ex. 1503, ¶ 11), that Declaration was
`
`prepared based on the declaration of Dr. Eduardo Padlan, which was submitted in
`
`14
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 18
`
`
`
`
`
`support of petitions challenging certain claims of the ’213 patent by third party
`
`Mylan. As I explained there, I applied my own judgment and expertise, as well as
`
`conducting my own fact checking and consideration of potential counterarguments,
`
`before presenting the opinions previously submitted by Dr. Padlan. I have continued
`
`to apply my own judgment and expertise in preparing this Declaration, which is not
`
`based on opinions prepared by any other expert.
`
`IV. LEGAL STANDARDS
`
`26.
`
`In my first Declaration, I set forth the legal standards that I was
`
`informed of that apply to the invalidity analysis I have conducted in this IPR. (Ex.
`
`1503, ¶¶ 13–26.) I have continued to apply those legal standards in preparing this
`
`Reply Declaration.
`
`V.
`
`PERSON OF ORDINARY SKILL IN THE ART
`
`27.
`
`In my first Declaration, I described what in my view are the
`
`qualifications of a POSITA to which the ’213 pertains. (Ex. 1503, ¶¶ 27–32.)
`
`Specifically, I provided the opinion that “a person of ordinary skill in the art in 1991
`
`related to the ’213 patent would be an individual that developed protein
`
`therapeutics.” This person would have a Ph.D. or equivalent (for example,
`
`knowledge gained through 4–5 years of work experience) in molecular biology,
`
`immunology, biochemistry or a closely related field, and may work as a member of
`
`a team. (Id., ¶ 29.)
`
`15
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 19
`
`
`
`
`
`28.
`
`I understand that, in its Preliminary Response, Genentech presented a
`
`different definition, specifically: “A person of ordinary skill for the ’213 patent
`
`would have had a Ph.D. or equivalent in chemistry, biochemistry, structural biology,
`
`or a closely related field, and experience with antibody structural characterization,
`
`engineering, and/or biological testing, or an M.D. with practical academic or
`
`industrial experience in antibody development.” (Paper 7 at 18.) The Board
`
`ultimately adopted Genentech’s definition. (Paper 40 at 8.)
`
`29. My opinions do not change under either definition. Indeed, Genentech
`
`itself noted that its definition was “similar” to mine (Paper 7 at 18), as did the Board
`
`(Paper 40 at 8). The only difference Dr. Wilson purports to identify is a
`
`disagreement “to the extent [my definition] implies that a person of ordinary skill in
`
`the art may include a laboratory technician with several years of work experience in
`
`molecular biology, without any specific experience in creating monoclonal
`
`antibodies, genetically engineering antibodies, or analyzing their three-dimensional
`
`structure.” (Ex. 2041, ¶ 98.) This is apparently an attempt to read Mr. Buss out of
`
`the definition. (Id., ¶ 99.) I disagree.
`
`30. First, Genentech’s definition explicitly allows for “a Ph.D. or
`
`equivalent” which, as reflected in my definition, can be gained through work
`
`experience (e.g., 4–5 years) rather than formal education. Furthermore, Dr. Wilson’s
`
`condition that the POSITA have “specific experience in creating monoclonal
`
`16
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 20
`
`
`
`
`
`antibodies, genetically engineering antibodies, or analyzing their three-dimensional
`
`structure” is not found in Genentech’s and the Board’s definition, which merely
`
`requires “experience with antibody structural characterization, engineering, and/or
`
`biological testing.” (Paper 40 at 8.) Nor is there anything in the definition that places
`
`a limit on the title of the POSITA, whether “lab technician” or otherwise.
`
`31.
`
`In this regard, I understand that Mr. Buss testified that, as of the ’213
`
`patent priority date, he had the “equivalent of a Ph.D.” in biochemistry with practical
`
`academic experience in antibody development, meeting Genentech’s and the
`
`Board’s definition of a POSITA. (Ex. 2040 34:19-25, 40:3–6; Ex. 1504 ¶¶ 4-6; Paper
`
`27 at 8.) I have no reason to doubt that testimony. Indeed, Mr. Buss’s skill set was
`
`refined in what was arguably the leading antibody engineering research group in the
`
`world (Winter’s).
`
`32. Dr. Wilson’s opinion also ignores the qualifications of the named
`
`inventors. Dr. Presta testified that, at the time he began work on the project leading
`
`to the ’213 patent, he had no experience with antibodies, much less their
`
`humanization, through his entire education and initial projects with Genentech, nor
`
`did he even have knowledge of the work that had been ongoing in the field. (Ex.
`
`1699 (Presta Tr.) at 14:22–21:18.)
`
`33. Moreover, as discussed further below, even if Mr. Buss did not meet
`
`the definition of the POSITA, that would not change any of my opinions, given that
`
`17
`
`PFIZER and SAMSUNG v. GENENTECH
`IPR2017-01489
`PFIZER EX. 1702, Page 21
`
`
`
`
`
`Mr. Buss’s opinion is limited to showing that POSITAs would have been motivated
`
`prior to the ’213 patent to humanize the 4D5 antibody described in the Hudziak
`
`(1987) reference, which is an opinion I independently reached based on my own
`
`review of the literature, including Hudziak, and which Dr. Wilson does