`
`
`
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`_____________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`_____________________
`
`
`PHIGENIX, INC.
`Petitioner
`
`v.
`
`IMMNUNOGEN, INC.
`Patent Owner
`
`_____________________
`
`Case IPR2014-00676
`Patent 8,337,856 B2
`_____________________
`
`IMMUNOGEN, INC.'S RESPONSE UNDER 37 C.F.R. § 42.120
`
`
`
`

`

`B. 
`
`C. 
`
`D. 
`
`C. 
`
`
`
`
`
`I. 
`
`
`
`
`IPR2014-00676
`Patent Owner's Response
`
`TABLE OF CONTENTS
`Claims 1-8 would not have been prima facie obvious ................................... 2 
`A.  Herceptin, HER2 immunoconjugates, and maytansinoids
`each raised toxicity concerns ................................................................. 3 
`Because Herceptin resistance was prevalent, a POSA would
`have been discouraged from using Herceptin in an anti-
`HER2 immunoconjugate ..................................................................... 14 
`The literature contradicts Phigenix's purported reasons to use
`Herceptin in an anti-HER2 immunoconjugate .................................... 15 
`Phigenix erroneously extrapolates the results of studies of
`combination therapies to conjugation of two or more
`components .......................................................................................... 23 
`Phigenix fails to show that it would have been obvious to use a non-
`II. 
`cleavable linker, as recited in Claims 6 and 8 ......................................................... 26 
`A. 
`Release of the drug from immunoconjugate was essential ................. 27 
`B. 
`A POSA would not have selected a non-cleavable linker
`because it would not have been expected to achieve the
`necessary release of the maytansinoid ................................................ 28 
`A POSA would have expected a maytansinoid linked to
`Herceptin via a non-cleavable linker to be ineffective ....................... 29 
`III.  Objective indicia overwhelmingly show nonobviousness of claim 8 .......... 34 
`A. 
`T-DM1 is unexpectedly superior to the closest prior art .................... 35 
`B. 
`T-DM1
`fulfilled a
`long-felt, unmet need
`for an
`immunoconjugate capable of
`targeting delivery of a
`cytotoxic agent to treat a solid tumor .................................................. 46 
`T-DM1 is praised in the field of breast cancer treatment ................... 53 
`C. 
`D.  Kadcyla®(T-DM1) has been a commercial success ............................ 57 
`IV.  Conclusion .................................................................................................... 60 
`
`
`
`
`- i -
`
`

`

`
`
`
`
`
`
`
`IPR2014-00676
`Patent Owner's Response
`
`In instituting trial, the Board preliminarily determined that it would have
`
`been obvious "to substitute the mouse monoclonal TA.1 antibody in the
`
`immunoconjugate of Chari 1992 with the humanized mAb huMAB4D5-8
`
`[Herceptin®] to produce the recited immunoconjugates…." Paper 11 at 12. But
`
`Phigenix's simple substitution argument cannot withstand scrutiny when it is
`
`viewed in light of the state of the art in March 2000– including art suggesting that
`
`such immunoconjugates would exhibit unacceptable levels of antigen-dependent
`
`toxicity in normal human liver tissue. Phigenix's arguments regarding claims 6 and
`
`8, which are limited to Herceptin-maytansinoid immunoconjugates linked with the
`
`non-cleavable linker SMCC, also cannot withstand scrutiny. Phigenix's expert
`
`admits that the maytansinoids in maytansinoid-based immunoconjugates must be
`
`released to have biological activity. But, Phigenix fails to establish why one would
`
`have nonetheless selected a non-cleavable linker—rather than a cleavable linker—
`
`to conjugate a maytansinoid to Herceptin. A skilled artisan would have expected a
`
`maytansinoid-based immunoconjugate containing a non-cleavable linker to be
`
`ineffective, and thus would have been dissuaded from making the
`
`immunoconjugates of claims 6 and 8.
`
`The invention's commercial embodiment, the ground-breaking cancer drug
`
`Kadcyla® (also known asTDM-1), exhibits results that were completely unexpected
`
`compared to the closest prior art. After several decades of research in an
`
`
`
`- 1 -
`
`

`

`
`
`
`unpredictable field, Kadcyla succeeded where others have repeatedly failed.
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`Kadcyla was the first, and is the only, FDA-approved antibody-drug conjugate for
`
`treating solid tumors. And Kadcyla's data presented to the American Society of
`
`Clinical Oncology (ASCO) "wowed the audience." For example, leading
`
`oncologist Hal Burnstein hailed Kadcyla as "incredible" and as providing
`
`"surprisingly positive" results in patients. By satisfying therapeutic needs that had
`
`long gone unmet, Kadcyla dramatically improves the lives of patients.
`
`Appropriately, given its safety and efficacy profile, Kadcyla enjoys tremendous
`
`commercial success. Consideration of all of the evidence reveals that Phigenix has
`
`failed to meet its burden to show obviousness by a preponderance of the evidence.
`
`I.
`
`Claims 1-8 would not have been prima facie obvious
`
`Immunoconjugates are comprised of an antibody conjugated to a toxic agent.
`
`EX1028 Abstract; EX2134 ¶14. While superficially a simple combination of
`
`elements—an antibody, a linker, and a cytotoxic agent—designing an efficacious
`
`immunoconjugate that exhibited an acceptable level of toxicity was fraught with
`
`obstacles and uncertainty in March 2000. EX2134 ¶15. In instituting trial, the
`
`Board cited Dr. Rosenblum's declaration, which alleged a person of skill in the art
`
`(POSA) would have been motivated to substitute the murine TA.1 antibody of the
`
`immunoconjugate of Chari 1992 with Herceptin. But Dr. Rosenblum posits
`
`motivations and expectations that the prior art has contradicted. Here, there would
`
`
`
`- 2 -
`
`

`

`
`
`
`not have been a reason to combine the claimed elements, and a person of ordinary
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`skill would not have had a reasonable expectation of success. Obviousness can be
`
`found only by ignoring highly-pertinent evidence in the prior art and resorting to
`
`hindsight.
`
`A. Herceptin, HER2 immunoconjugates, and maytansinoids each
`raised toxicity concerns
`
`An obviousness inquiry must consider the scope and content of the prior art
`
`and the differences between the invention and the prior art. Graham v. John Deere
`
`Co 383 U.S. 1, 17 (1966). Here, one must consider the scope and content of the art
`
`in March of 2000 when considering whether the art provided a reason to select
`
`Herceptin (from all the candidate anti-HER2 antibodies) and a maytansinoid (from
`
`all the candidate small molecule toxic agents and protein toxins) for conjugation
`
`with a reasonable expectation of success. Attempts to show obviousness may fail
`
`when there is a "broad selection of choices for further investigation available" or
`
`when "the challenges of [the] inventive process would have prevented one of
`
`ordinary skill in this art from traversing … multiple obstacles to easily produce the
`
`invention in light of the evidence available at the time of invention." Rolls-Royce,
`
`PLC v. United Techs. Corp., 603 F.3d 1325, 1339 (Fed. Cir. 2010); Ortho-McNeil
`
`Pharm., Inc. v. Mylan Labs., Inc., 520 F.3d 1358, 1364-65 (Fed. Cir. 2008). As
`
`explained below, a POSA would not had a reason to combine the claimed elements
`
`and would not have arrived at claims 1-8 with a reasonable expectation of success
`
`
`
`- 3 -
`
`

`

`
`
`
`in view of the literature available by March 2000.
`
`
`
`
`IPR2014-00676
`Patent Owner's Response
`
`Unacceptable HER2 antigen-dependent toxicity expected in normal
`
`liver tissue would have dissuaded targeting HER2 with an immunoconjugate:
`
`Phigenix's obviousness theory is fundamentally flawed because it fails to assess the
`
`state of the art as of March of 2000 concerning known reports of antigen-
`
`dependent toxicity associated with targeting the HER2 receptor with an
`
`immunoconjugate. EX2134 ¶¶49-61. Such concerns culminated in a study Pai-
`
`Scherf published in 1999, shortly before the '856 patent was filed ("Pai-Scherf
`
`1999"; EX2029). Pai-Scherf 1999 was not previously of record in this IPR.
`
`Pai-Scherf describes a Phase 1 study of patients receiving immunoconjugate
`
`(erb-38) comprising the anti-HER2 antibody e23 fused to the PE38 toxin (a
`
`fragment of the Pseudomonas exotoxin A that has decreased toxicity in
`
`unconjugated form). EX2029 Abstract; EX2134 ¶29. Though the authors initiated
`
`the study in humans based on "excellent antitumor activity and acceptable animal
`
`toxicities," unacceptable hepatotoxicity nevertheless was observed in all patients
`
`in the treatment group. EX2029 2311:2:2; EX2029 Abstract; EX2134 ¶¶50,54.
`
`The Pai-Scherf authors concluded that the toxicity was due to the presence
`
`of HER2 on hepatocytes (i.e., normal liver cells). EX2029 Abstract; EX2134 ¶54.
`
`Thus, despite the "very large" difference in HER2 expression on cancer cells
`
`compared to normal liver cells, normal liver cells were the initial site of the toxic
`
`
`
`- 4 -
`
`

`

`
`
`
`drug action. EX2029 2314:2:1; EX2134 ¶¶29,110. According to Pai-Scherf, this
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`occurred because there is a barrier to the immunoconjugate entering the tumor
`
`cells, as solid tumors lack lymphatics and connective flow. EX2029 2314:2:1;
`
`EX2134 ¶¶29,110. In addition, tumors are poorly vascularized. EX2029 2314:2:1;
`
`EX2134 ¶110; EX1029 583:2:2. Consequently, "mixing within tumors is solely by
`
`diffusion and, therefore, very slow." EX2029 2314:2:1; EX2134 ¶110. Given these
`
`results, the authors predicted that targeting tumors with other anti-HER2
`
`immunoconjugates would result in similar, unacceptable toxicities due to HER2
`
`expression on normal tissues. EX2029 2315:2:2; EX2134 ¶110.
`
`HER2, which Phigenix's Dr. Rosenblum agrees is a tumor-associated rather
`
`than tumor-specific antigen (EX2039 113:18-21), was known to be expressed on
`
`many additional types of normal cells including cells of the gastrointestinal,
`
`respiratory, reproductive, and urinary tracts. EX1018 871:1:2; EX1003 Abstract;
`
`EX1013 at 5870:2:2; EX2134 ¶¶29,52. As ImmunoGen's expert, Dr. Pietersz,
`
`explains, those in the field expressed significant concern about the potential
`
`toxicity of anti-HER2 immunoconjugates against normal HER2-expressing tissue,
`
`even before the Pai-Scherf study was reported. EX2134 ¶¶29,52. For example,
`
`Tecce stated that the "major constraints" of anti-HER2 immunoconjugates include
`
`limitations in the potential in vivo application due to their broad reactivity with
`
`normal tissues." EX 2063 125:1:3; EX2134 ¶52. Thus, Pai-Scherf validated
`
`
`
`- 5 -
`
`

`

`
`
`
`existing concerns about antigen-dependent toxicity being associated with
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`administering anti-HER2 immunoconjugates to patients. EX2134 ¶52.
`
`Indeed, in remarking on Pai-Scherf, Park et al. stated "[the HER2-dependent
`
`toxicity seen in Pai-Scherf] indicates that some immunotoxin strategies can be too
`
`potent, since even normal tissues with very low levels of antigen expression (such
`
`as HER expression in hepatocytes) can still be targeted by these otherwise
`
`nonspecific and exquisitely active cytotoxins." EX2057 226:1:31; EX2134 ¶55.
`
`According to Dr. Pietersz, "[t]hose in the field would have sought to avoid a
`
`target that was known to result in unacceptable antigen-dependent toxicity" as
`
`"numerous other potential targets for immunoconjugates in breast cancer cells had
`
`been identified." EX2134 ¶57. For example, additional potential target antigens
`
`included CEA, MAGE-1, p53, T/Tn, and MUC-1, each of which Hadden 1999 said
`
`"may be therapeutically useful as targets for immunotoxins." EX2153 Abstract;
`
`EX2154 Abstract; EX 2155 Abstract; EX2156 Abstract; EX2157 Abstract;
`
`EX2158 Abstract; EX2134 ¶57.
`
`
`1 While Park published after March 2000, later published reports can be used to
`
`indicate the state of the art existing at the time of filing of the '856 patent. See
`
`Plant Genetics Sys., NV v. DeKalb Genetics Corp., 315 F.3d 1335, 1344 (Fed. Cir.
`
`2003).
`
`
`
`- 6 -
`
`

`

`
`
`
`
`
`
`
`IPR2014-00676
`Patent Owner's Response
`
`Even assuming, arguendo, that there would have been a reason after Pai-
`
`Scherf to target HER2, a POSA would have had a clear motivation to avoid anti-
`
`HER2 antibodies, such as Herceptin, and cytotoxic agents that already exhibited
`
`significant liver toxicity prior to conjugation. As evidenced by its label, "hepatic
`
`failure" is a serious side effect of Herceptin EX1008 2:2:4; EX2134 ¶60. Also, the
`
`label includes a black-box warning that Herceptin is cardiotoxic. EX1008 1:1:1.
`
`What is more, the prevalence of toxic events greatly increased when Herceptin was
`
`used in combination with other agents, such as anthracycline and
`
`cyclophosphamide. Id. Table 3. Therefore, not only was Herceptin known to
`
`exhibit serious toxicity alone, those toxicities were exacerbated when Herceptin
`
`was used in a combination therapy with certain compounds. Id.
`
`As Dr. Pietersz explains, a POSA would have avoided Herceptin because it
`
`had the same toxicities as the anti-HER2 immunoconjugates of Pai-Scherf.
`
`EX2134 ¶¶60-61. As Dr. Rosenblum admits, "numerous" anti-HER2 antibodies
`
`were available in March 2000. EX1018 865:2:3. And, many candidate anti-HER2
`
`antibodies (including the TA.1 antibody disclosed in Chari 1992) were known not
`
`to have growth-inhibitory functions like Herceptin, which a POSA would have
`
`recognized would potentially make the other candidates less likely to have a
`
`compounding effect on toxicity when used in an immunoconjugate. EX2134 ¶33.
`
`Maytansinoids were also known to cause unacceptable liver toxicity in
`
`
`
`- 7 -
`
`

`

`
`
`
`unconjugated form. As Dr. Rosenblum himself points out, maytansinoids are
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`100- to 1000-fold more potent than some of the commonly-used anticancer drugs.
`
`EX1016 ¶24; see also EX1015 97:2:2; EX2134 ¶62. Numerous clinical trials
`
`involving testing of maytansinoids as anti-cancer agents revealed extreme
`
`toxicities, including liver damage. EX2002 3:2-4 and 10:1-7; EX2041 692:2:4-
`
`693:1:1; EX2042 Abstract; EX2043 203:4-205:1; EX2044 Abstract; EX2134 ¶¶66-
`
`67. In particular, maytansinoids caused a 20% incidence of hepatic injury in
`
`patients when given as weekly intravenous injections, and 40% when given by
`
`continuous infusion. EX2002 3:2-4 and 10:1-7; EX2041 692:2:4-693:1:1; EX2042
`
`Abstract; EX2043 203:4-205:1; EX2044 Abstract; EX2134 ¶¶66-67. In some
`
`cases, patients with previous liver dysfunction died in response to maytansinoid
`
`administration. EX2044 426:2:3; EX2134 ¶76.
`
`In fact, by March 2000, the immunoconjugate field was moving away from
`
`using cytotoxic agents that were toxic in unconjugated form. EX1013 5867:1:2-
`
`2:1; EX2045 7:1:1; EX2063 122:1:3; EX2292 at 2676:2:1; EX2134 ¶¶68-69. In
`
`particular, researchers were making use of protein toxins that were non-toxic until
`
`delivered to the target cell. EX1013 5867:1:2-2:1; EX2046 18327:2:1; EX2029
`
`2311:2:1; EX2063 122:1:3; EX1011 Abstract; EX2038 Abstract; EX2134 ¶¶68-69.
`
`Notably, Dr. Rosenblum's own work would have motivated a POSA to use the
`
`plant toxin gelonin, which, according to Dr. Rosenblum and his colleagues, "by
`
`
`
`- 8 -
`
`

`

`
`
`
`itself, cannot bind to cells and is relatively nontoxic" in unconjugated form.
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`EX2160 1971:2:4-1972:1:1; EX2134 ¶68.
`
`In sum, a POSA would have avoided conjugating two compounds with
`
`known independent toxicities to the liver— such as Herceptin and a maytansinoid
`
`— to produce an anti-HER2 immunoconjugate. This is especially true when, as
`
`here, clinical trials had already shown that anti-HER2 immunoconjugates exhibit
`
`unacceptable liver toxicity. EX2134 ¶¶66-70.
`
`Dr. Rosenblum's conclusion that anti-HER2 maytansinoid-based
`
`immunoconjugates were "substantially free" of toxicity is flawed because it is
`
`based on references that did not assess antigen-dependent toxicity: Dr.
`
`Rosenblum attempts to downplay the possibility of unacceptable antigen-
`
`dependent toxicity. EX1016 ¶¶41-44. In particular, Dr. Rosenblum discusses Trail
`
`1999 which warned of antigen-dependent toxicity; Trail stated, "[t]he use of
`
`extremely toxic drugs requires careful MAb selection as even low levels of
`
`expression of the targeted antigen by normal cells may lead to significant toxicity."
`
`EX1016 ¶41; EX1028 79:1:1. As discussed above, the concern expressed in Trail
`
`certainly held true for immunoconjugates targeting HER2, as reported in Pai-
`
`Scherf. EX2134 ¶¶50-61; EX2029 Abstract.
`
`Nonetheless, Dr. Rosenblum asserts that maytansinoid-based
`
`immunoconjugates targeting HER2 "were demonstrated to be substantially free of
`
`
`
`- 9 -
`
`

`

`
`
`
`toxicity" (EX1016 ¶42), citing several references. But, the references Dr.
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`Rosenblum cites—Chari 1992 (EX1012), Batra 1992 (EX1013), Liu 1996
`
`(EX1023), and Chari 1998 (EX1015) (EX1016 ¶¶42-44)—describe experiments
`
`performed in vitro or in xenograft mouse models and which fail to address antigen-
`
`dependent toxicity. EX2134 ¶¶19-27 and 62-65.
`
`Dr. Rosenblum argues that Chari 1992 (EX1012) shows "specific binding
`
`and highly selective concentration-dependent" effects and that Liu (EX1023)
`
`shows that a maytansinoid immunoconjugate (C242-DM1) is less active in
`
`antigen-negative cells than antigen-positive cells. EX1016 ¶¶42-43. However,
`
`these observations arose out of in vitro assays. EX2134 ¶¶62-65; EX1012 129:1:2-
`
`3; EX1023 8619:2:4-8620:1:2. It was well established that in vitro assays are
`
`highly unreliable in predicting antigen-dependent toxicity, in part because they are
`
`conducted under conditions that are not reflective of the in vivo environment and
`
`do not account for factors such as tissue accessibility and relative timing of
`
`exposure. EX2134 ¶62; EX2143 4530:2:1; EX1013 5870:2:2; EX1015 91:2:2.
`
`With respect to xenograft mouse models, in the references cited by Dr.
`
`Rosenblum, only the implanted human tumor cells express the target HER2
`
`antigen, whereas normal mouse cells lack HER2. EX2143 4530:2:1; EX1013
`
`5870:2:2; EX2134 ¶¶22-24; EX1023, Abstract. And Dr. Rosenblum admitted
`
`during his deposition that the HER2 antigen is a human antigen. EX2039 47:23-
`
`
`
`- 10 -
`
`

`

`
`
`
`25.As such, because an immunoconjugate targeting HER2 would not bind normal
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`mouse tissue, mouse xenograft models "do not provide any information about how
`
`the immunoconjugate may react with non-tumor human cells, tissues, and organs."
`
`EX2134 ¶¶22-27; EX1013 5870:2:2; EX2143 4530:2:1; EX1013 5870:2:2. In fact,
`
`Dr. Rosenblum admitted under cross-examination that he was aware that many
`
`immunoconjugates had failed in the clinic even after showing positive results in
`
`animal studies, at least in part because mice and humans are "fundamentally
`
`different." EX2039 63:14-16. Thus, in xenograft models, while cytotoxicity against
`
`the implanted HER2-expressing human tumor is antigen-dependent, measures of
`
`toxicity against the mice are measures of systemic or antigen-independent toxicity.
`
`EX2039 85:15-20; EX2134 ¶¶22-27; EX1013 5870:2:2; EX2143 4530:2:1;
`
`EX1013 5870:2:2. Indeed, Chari 1992 referred to "low systemic toxicity" rather
`
`than antigen-dependent toxicity when describing results from administering the
`
`A72 (-SS-May) immunoconjugate to mice. EX1012 130:1:2; EX2134 ¶42.
`
`Notably, Dr. Rosenblum also admitted that there are different types of
`
`toxicity: "I believe in general there are three different types of toxicity: Systemic
`
`toxicity to the intact animal. Specific toxicity to cells, isolated cells. And general
`
`
`2 Incidentally, the A7 antibody used in the in vivo experiment described in Chari
`
`1992 is not an anti-HER2 antibody. EX1012 129:1:3.
`
`
`
`- 11 -
`
`

`

`
`
`
`toxicity to isolated cells that have different attributes but the toxicity spans
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`different cells types…." EX2039 81:14-20. Dr. Rosenblum's acknowledgement
`
`that systemic toxicity and specific (i.e., antigen-dependent) toxicity are different
`
`types of toxicity undermines his assertion that lack of systemic toxicity in a mouse
`
`model is "at odds" with a concern about antigen-dependent toxicity in normal
`
`human tissue. EX1016 ¶44. In fact, Dr. Rosenblum's own paper, Rosenblum 1999,
`
`expresses concern about antigen-dependent toxicity: "[o]f concern in the
`
`development of therapies that target HER2/neu, however, is the expression of the
`
`target antigen on normal tissues." EX1018 871:1:2; EX2134 ¶52. Dr. Rosenblum
`
`also confirmed the importance of clinical trials in detecting antigen-dependent
`
`toxicity of HER2 immunoconjugates, noting that "the toxicity of the construct to
`
`normal tissues as a result of expression of the target antigen, is a common problem
`
`that is addressed in clinical trials." EX2111 8994:1:2; EX2134 ¶59.
`
`Finally, the authors of Batra 1992, another reference relied on by Dr.
`
`Rosenblum, state, "[i]t should be noted that a potentially important form of toxicity
`
`was not apparent in this [mouse xenograft model] experiment, as we do not expect
`
`the [anti-HER2 immunoconjugate] … to bind to murine gp185 erbB2 (neu) gene
`
`product." EX1013, 5870:2:2; EX2134 ¶52. And when asked during his deposition
`
`whether "the animal model test in Batra [would] have been a test of toxicity against
`
`normal human tissue that expresses HER2," Dr. Rosenblum unequivocally
`
`
`
`- 12 -
`
`

`

`
`
`
`admitted "No." EX2039 88:18-22. Thus, Dr. Rosenblum agrees that a mouse
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`xenograft model cannot predict antigen-dependent toxicity in a human. Indeed, as
`
`Dr. Pietersz explains, the antigen-dependent toxicity concerns expressed in the
`
`literature were realized for HER2-directed immunoconjugates in the Phase 1 study
`
`reported in Pai-Scherf. EX2029 Abstract; EX2134 ¶54 and 63-64.
`
`Phigenix and Dr. Rosenblum make no mention of Pai-Scherf in the Petition
`
`or in Dr. Rosenblum's declaration, apparently because Dr. Rosenblum did not
`
`review it when preparing his declaration. When asked "[i]s it fair to say that you
`
`don't recall reading the Pai-Scherf paper since it was published in 1999?", Dr.
`
`Rosenblum replied "Yes." EX2039 90:23-91:2. Further, Dr. Rosenblum revealed
`
`that he had not reviewed HER2 expression patterns in "at least 4 years," which is
`
`consistent with the fact that he could not even recall whether liver cells express
`
`HER2. Id. 12:16-13:15, 14:7-22, and 16:25-20:12.
`
`In sum, the very references Dr. Rosenblum relies on do not support his
`
`position that a POSA would have considered HER2-directed immunoconjugates to
`
`be "substantially free of toxicity" in humans based on experiments demonstrating
`
`low systemic toxicity in mice. Moreover, Dr. Rosenblum did not consider Pai-
`
`Scherf when formulating his opinion. But as discussed above, after Pai-Scherf,
`
`concerns about antigen-dependent toxicity in normal liver tissue would have
`
`dissuaded a POSA from targeting HER2 with an immunoconjugate.
`
`
`
`- 13 -
`
`

`

`
`
`
`
`
`
`
`IPR2014-00676
`Patent Owner's Response
`
`B.
`
`Because Herceptin resistance was prevalent, a POSA would have
`been discouraged from using Herceptin in an anti-HER2
`immunoconjugate
`
`Eighty-five percent of HER2-overexpressing breast cancer patients do not
`
`respond, or respond poorly, to Herceptin or they develop resistance to Herceptin
`
`quickly. See, e.g., EX1008 1:2:1 and Table 1; EX2134 ¶72. Herceptin resistance
`
`was believed to be due to factors that prevented Herceptin from interacting with
`
`HER2 on cancer cells. EX2134 ¶¶73-74. Thus, a POSA would have expected
`
`Herceptin to have the same or similar difficulties interacting with HER2 if
`
`Herceptin were part of an immunoconjugate. Id.
`
`Dr. Pietersz states that it was known that the extracellular domain of
`
`HER2—the target of Herceptin—was shed off the surface of breast cancer cells.
`
`EX2048 5123:2:2-4; EX2049 61:2:2; EX2134 ¶73. It was also known that the shed
`
`HER2 extracellular domain competes with full-length HER2 for antibody binding.
`
`EX2050 1718:1:3-4 and 1719:2:3; EX2134 ¶73. According to Dr. Pietersz, a
`
`POSA would have expected that presence of the soluble extracellular domain of
`
`HER2 in a patient's blood would neutralize the killing effect of anti-HER2
`
`antibodies on HER2-posititve tumors, resulting in Herceptin resistance in these
`
`patients. EX2048 5123:2:1; EX2134 ¶73.
`
`In addition, a naturally-occurring variant of HER2 had been discovered in
`
`human breast cancer cell lines. EX2134 ¶74. This truncated HER2 variant had
`
`
`
`- 14 -
`
`

`

`
`
`
`been shown to lack a portion of HER2 located in the same domain of HER2 that
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`Herceptin was believed to bind. EX2049 61:2:2; EX2051 62:2:1 and Fig. 1;
`
`EX2052 7:49-51. Thus, according to Dr. Pietersz, an inability of Herceptin to bind
`
`this HER2 variant was believed to be a second possible mechanism explaining
`
`Herceptin resistance in some patients. EX2134 ¶74.
`
`Because both proposed mechanisms of Herceptin resistance involved
`
`interference with Herceptin binding to the HER2 target, a POSA would not have
`
`had a reason to combine the Herceptin antibody with a maytansinoid to make the
`
`claimed immunoconjugates, because other candidate antibodies were available and
`
`at least 85% of the Herceptin target-patient population would not be expected to be
`
`receptive to treatment with an immunoconjugate containing Herceptin. EX2134
`
`¶¶43, 72, 75.
`
`C. The literature contradicts Phigenix's purported reasons to use
`Herceptin in an anti-HER2 immunoconjugate
`
`Phigenix asserts that a POSA would have been motivated to substitute
`
`Herceptin for TA.1 in the Chari 1992 immunoconjugates because (i) Herceptin is a
`
`humanized antibody (Paper 5 at 14-15; EX1016 ¶13); and (ii) Herceptin exhibits
`
`selective and high affinity binding to HER2 and is approved for use in humans
`
`(Paper 5 at 15; EX1016 ¶12). EX2134 ¶44.
`
`
`
`Herceptin's status as a humanized antibody would not have motivated a
`
`POSA to use it in an immunoconjugate: While it is true that humanized
`
`
`
`- 15 -
`
`

`

`
`
`
`antibodies are less likely than mouse antibodies to induce an immune response
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`when introduced into a patient, this feature of Herceptin would not have motivated
`
`a POSA to select Herceptin and overlook its disfavored features. Humanizing
`
`antibodies was routine in March 2000 and thus a POSA could have readily
`
`humanized any of the other candidate antibodies that were available and not known
`
`to have Herceptin's shortcomings. EX2134 ¶79. As early as 1992, researchers had
`
`noted that "[f]ull humanization of rodent antibodies with retention of most of their
`
`antigen binding activity is now a routine procedure." EX2056 111:2 (emphasis
`
`added); EX2134 ¶80. As confirmed by Dr. Rosenblum, by March 2000 contract
`
`research organizations provided on-demand antibody humanization. EX2040
`
`162:23-164:14; EX2083; EX2084 2:1; EX2134 ¶¶79-80. In fact, prior to March
`
`2000, several other anti-HER2 humanized antibodies had already been developed.
`
`See, e.g., EX1022; EX1018; EX2058; EX2134 ¶81. Therefore, as Dr. Pietersz
`
`explains, because of the routine nature of humanization, a POSA would not have
`
`considered an antibody’s status as humanized to be a significant factor in selecting
`
`it for developing an immunoconjugate. EX2134 ¶80, 82.
`
`
`
`Neither binding affinity nor approval in unconjugated form would have
`
`provided a reason to use Herceptin as a delivery vehicle in an anti-HER2
`
`immunoconjugate: As Dr. Pietersz states, more than 50 anti-HER2 antibodies and
`
`cytotoxic agents were available candidates for a HER2 immunoconjugate by 2000.
`
`
`
`- 16 -
`
`

`

`
`
`
`EX2134 ¶¶30-33. Thus, a POSA interested in designing an anti-HER2
`
`
`
`
`IPR2014-00676
`Patent Owner's Response
`
`immunoconjugate would have had a broad range of anti-HER2 antibodies to
`
`consider. EX1011; EX1013; EX1030; EX2032; EX2036; EX2037; EX2038;
`
`EX2061; EX2063; EX2065; EX2066; EX2067; EX2068; EX2069; EX2070;
`
`EX2071; EX2085; EX2086; EX2134 ¶33.
`
`
`
`Phigenix's petition and Dr. Rosenblum's declaration ignore this wide choice
`
`of anti-HER2 antibody candidates that were available in March of 2000. Dr.
`
`Rosenblum's disregard for such other antibodies is, at best, curious since his own
`
`1999 publication acknowledged: "[n]umerous monoclonal antibodies targeting the
`
`gp185 [erbB2] cell-surface domain have recently been developed." EX1018
`
`865:2:2. And of those developed, over 20 had been conjugated to various toxins.
`
`EX1012; EX1013; EX1018; EX1030; EX2038; EX2134 ¶33. For example, as Dr.
`
`Pietersz explains, Batra 1992 (EX1013) used anti-HER2 antibody to construct 5
`
`immunoconjugates usin 5 different anti-HER2 antibodies; Suzuki 1995 (EX1030)
`
`constructed an immunoconjugate with the Sv2-61 and SER4 anti-erbB2 antibodies;
`
`and Dr. Rosenblum's own anti-HER2 immunoconjugates used two antibodies –
`
`TAB-250 and BACH-250 (EX1018). EX2134 ¶33.
`
`
`
`Before the publication of Pai-Scherf in 1999, extensive research had been
`
`conducted to identify characteristics of anti-HER2 antibodies that would be
`
`predictive of success in immunoconjugates. Id. ¶32. This research revealed that the
`
`
`
`- 17 -
`
`

`

`
`
`
`field was highly unpredictable. EX2134 ¶109. For example, immunoconjugates
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`containing different anti-HER2 antibodies that bind the same antigen had widely
`
`variable activity. EX1013 Abstract and 5868:1:7; EX2038 525:1:1-1:3; EX 2134
`
`¶32. Boyer 1999 described comparative experiments involving immunoconjugates
`
`constructed with seven different anti-HER2 antibodies, including TA.1. EX2038
`
`526:1:1. Though TA.1 (unlike Herceptin) had been shown to be inactive in
`
`unconjugated form, an immunoconjugate containing TA.1 exhibited the greatest
`
`inhibitory effect against breast cancer cells. EX2038 528:2:2, Table 2, and Fig. 2;
`
`EX2134 ¶¶34, 76-78.
`
`
`
`Boyer concluded that "the growth inhibitory function of an anti-HER-2
`
`antibody does not affect the cytotoxic activity of the corresponding immunotoxin."
`
`EX2038 530:2:3; EX2134 ¶¶34, 77-78, 83. Boyer also concluded that activity of an
`
`immunoconjugate does not correlate with the binding affinity of the corresponding
`
`anti-HER2 antibody. EX2038 530:1:2; EX2134 ¶¶34, 77-78, 84. Thus, contrary to
`
`the picture Dr. Rosenblum attempts to paint, as of March 2000, a POSA would
`
`have understood that the binding affinity and unconjugated activity of an anti-
`
`HER2 antibody against cancer cells does not predict the success of an
`
`immunoconjugate containing that antibody. EX2134 ¶¶34, 77-78.
`
`The large number of reported failures in developing an effective
`
`immunoconjugate for treating solid tumors would have discouraged a POSA
`
`
`
`- 18 -
`
`

`

`
`
`
`from arriving at the claimed invention with a reasonable expectation of
`
`IPR2014-00676
`Patent Owner's Response
`
`
`
`
`success. Phigenix asserts that a POSA would have been motivated to make the
`
`invention because of its "clinical applications" (i.e., human applications), e.g.,
`
`"treating breast cancer" in "humans." Paper 5 at 14-16, 20, 31, and 37-40. For
`
`example, Phigenix states that a POSA would have been motivated to substitute
`
`Herceptin for TA.1 in Chari 1992 because humanized antibodies "exhibited
`
`reduced immunogenicity in human patients" and "were preferred over their mouse-
`
`derived counterparts for clinical applications." Paper 5 at 14 (emphasis added)3.
`
`But when the art as a whole is considered, a POSA would not have had a
`
`reason to attempt to make the claimed immunoconjugates because (i) anti-HER2
`
`immunoconjugates, Herceptin, and maytansinoids were all known to cause liver
`
`toxicity (see §I(A)) and (ii) a large portion of the patient population was non-
`
`responsive to Herceptin and would have been expected to be non-responsive to an
`
`
`3 Dr. Rosenblum presented both his motivation and reasonable expectation of
`
`success arguments in the context of clinical use of immunoconjugates, e.g. to
`
`treat breast cancer. (id. ¶¶13, 15). Moreover, each of the three allegations related
`
`to reasonable expectation of success asserted by Phigenix also reference clinical
`
`applicati