UNITED STATES PATENT AND TRADEMARK OFFICE
`___________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`___________
`
`MYLAN PHARMACEUTICALS INC.,
`Petitioner,
`
`v.
`
`ASTRAZENECA AB,
`Patent Owner.
`__________
`
`Case IPR2015-01340
`Patent RE44,186
`__________
`
`
`
`PATENT OWNER ASTRAZENECA AB’S PRELIMINARY RESPONSE
`
`
`
`
`
`

`
`I.
`
`II.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`
`TABLE OF CONTENTS
`
`Introduction ...................................................................................................... 1
`
`Background ...................................................................................................... 8
`
`A. DPP-IV Inhibitors.................................................................................. 8
`
`B.
`
`C.
`
`D.
`
`DPP-IV Inhibitors in the Late-90s Varied Greatly ............................... 9
`
`Structural Modifications of Known DPP-IV Inhibitors Yielded
`Unpredictable Results ......................................................................... 15
`
`Potential DPP-IV Inhibitors Faced Additional Unpredictable
`Hurdles Beyond Potency ..................................................................... 18
`
`E.
`
`The Invention of the RE’186 Patent and its Benefits .......................... 21
`
`III. The Challenged Claims ................................................................................. 27
`
`IV. Legal Standard ............................................................................................... 28
`
`V.
`
`Level of Ordinary Skill in the Art ................................................................. 31
`
`VI. Mylan’s Hindsight Analysis Ignores the Teachings of the Art and
`Focuses on Structures Divorced From Their Properties ............................... 31
`
`A. One of Skill in the Art Would Not Have Chosen Mylan’s Lead
`Compound ........................................................................................... 31
`
`B.
`
`Even if the Board Accepts Mylan’s Lead Compound, the
`Claimed Combination of Structural Features and Their
`Resulting Properties Would Not Have Been Obvious ........................ 35
`
`1.
`
`One of Skill in the Art Would Not Have Been Motivated
`to Add Cyclopropyl to Ashworth’s Compound 25 ................... 36
`
`2. Mylan Has Failed to Show That Narrower Claims
`Embodying Additional Features of Saxagliptin Are
`Likely Unpatentable .................................................................. 41
`
`3.
`
`The Saxagliptin Specific Claims Have Not Been Shown
`to Be Likely Unpatentable ........................................................ 41
`
`
`
`ii
`
`

`
`
`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`One of Skill in the Art Would Not Have Been
`Motivated to Add Cyclopropyl in the 4,5 cis
`Configuration to Ashworth’s Compound 25 .................. 42
`
`
`
`
`
`
`
`a.
`
`
`
`b.
`
`
`c.
`
`
`
`d.
`
`
`One of Skill in the Art Would Not Have Been
`Motivated to Substitute Adamantyl for the
`Cyclohexane of Ashworth’s Compound 25 ................... 44
`
`One of Skill in the Art Would Not Have Been
`Motivated to Hydroxylate the Adamantyl Group ........... 47
`
`Objective Evidence of Nonobviousness Supports
`Patentability .................................................................... 51
`
`VII. Conclusion ..................................................................................................... 55
`
`
`
`
`
`iii
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`
`TABLE OF AUTHORITIES
`
`Page(s)
`
`Cases
`Broadcom Corp. v. Emulex Corp.,
`732 F.3d 1325 (Fed. Cir. 2013) .......................................................................... 54
`
`In re Cyclobenzaprine Hydrochloride Extended-Release Capsule
`Patent Litigation,
`676 F.3d 1063 (Fed. Cir. 2012) .......................................................................... 55
`
`Daiichi Sankyo Co. v. Matrix Labs., Ltd.,
`619 F.3d 1346 (Fed. Cir. 2010) .......................................................................... 29
`
`Eli Lilly and Co. v. Zenith Goldline Pharms., Inc.,
`471 F.3d 1369 (Fed. Cir. 2006) .......................................................................... 30
`
`Graham v. John Deere,
`383 U.S. 1 (1966) ................................................................................................ 28
`
`In Touch Techs., Inc. v. VGO Commc’ns, Inc.,
`751 F.3d 1327 (Fed. Cir. 2014) .......................................................................... 30
`
`KSR Int’l Co. v. Teleflex Inc.,
`550 U.S. 398 (2007) ............................................................................................ 30
`
`Life Techs., Inc. v. Clontech Labs., Inc.,
`224 F.3d 1320 (Fed. Cir. 2000) .......................................................................... 30
`
`In re Meinhardt,
`392 F.2d 273 (C.C.P.A. 1968) ............................................................................ 53
`
`Otsuka Pharmaceutical Co. v. Sandoz, Inc.,
`678 F.3d 1280 (Fed. Cir. 2012) ...................................................................passim
`
`In re Papesch,
`315 F.2d 381 (C.C.P.A. 1963) ............................................................................ 28
`
`Pfizer, Inc. v. Apotex, Inc.,
`480 F.3d 1348 (Fed. Cir. 2007) .................................................................... 31, 43
`
`
`
`iv
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`
`Proctor & Gamble Co. v. Teva Pharms. USA, Inc.,
`566 F.3d 989 (Fed. Cir. 2009) ............................................................................ 31
`
`Takeda Chem. Indus., Ltd. v. Alphapharm Pty., Ltd.,
`492 F.3d 1350 (Fed. Cir. 2007) .......................................................................... 28
`
`Transocean Offshore Deepwater Drilling, Inc. v. Maersk Contractors
`USA, Inc.,
`617 F.3d 1296 (Fed. Cir. 2010) .................................................................... 51, 54
`
`Statutes
`
`35 U.S.C. § 103 ........................................................................................................ 30
`
`35 U.S.C. § 314(a) ..................................................................................................... 8
`
`Other Authorities
`
`37 C.F.R. § 42.107 ..................................................................................................... 1
`
`
`
`
`
`
`
`
`
`
`
`
`
`v
`
`

`
`I.
`
`Introduction
`
`Patent Owner AstraZeneca AB (“AstraZeneca”) submits this Preliminary
`
`Response under 37 C.F.R. § 42.107 to the Petition of Mylan Pharmaceuticals Inc.
`
`(“Mylan”) for Inter Partes Review (“Mylan’s Petition”) of claims 1-2, 4, 6-22, 25-
`
`30, 32-37, and 39-42 of U.S. Patent No. RE44,186 (“the RE’186 patent”), and
`
`respectfully requests that the Board deny institution.
`
`
`
`The subject matter of the RE’186 patent relates to a novel class of
`
`compounds that includes an active pharmaceutical ingredient called saxagliptin,
`
`the structure of which is shown here:
`
`
`
`Saxagliptin has a unique combination of structural features that include
`
`“cyclopropyl” (Δ) and “cyano” (CN) groups in a specific location and orientation
`
`on a nitrogen (N)-containing “pyrrolidine” ring (“the core”), a “primary amine”
`
`(NH2) in its backbone, and an “adamantyl” group that is “hydroxylated” (OH) on
`
`the other end of the molecule. These structural features cooperate to provide a
`
`constellation of desirable properties that make saxagliptin commercially useful as a
`
`
`
`
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`pharmaceutical drug product to inhibit the action of an enzyme known as
`
`
`
`
`
`
`
`
`
`dipeptidyl peptidase IV (“DPP-IV”) in the treatment of type II diabetes mellitus.
`
`At present, Mylan and nine other generic companies seek to make generic
`
`versions of AstraZeneca’s saxagliptin-containing products. Related litigation is
`
`ongoing in the U.S. District Court for the District of Delaware with trial scheduled
`
`to take place in 2016 and provides the impetus for Mylan’s Petition. That Petition,
`
`however, is impermissibly based on a hindsight-driven focus on isolated,
`
`“structural” features divorced from their combination with other claimed structural
`
`features and the actual properties of the resulting molecule as a whole.
`
`More specifically, Mylan’s entire analysis is based on the selection of an
`
`inappropriate “lead compound.” At the time of the invention, there were numerous
`
`DPP-IV inhibitors being developed with a diverse assortment of cores and
`
`backbones, each of which had an array of choices for structural modification. See
`
`infra § II.B. Many of these inhibitors had promising activity, were characterized
`
`by in vivo data, or had entered clinical trials. Mylan ignores these promising
`
`potential lead compounds in favor of what it believes, with perfect hindsight, to be
`
`the closest structural analog to saxagliptin it can find in the prior art: Compound 25
`
`from Ex. 1007, Ashworth et al., 6 Bioorg. & Med. Chem. Lett. 1163 (1996)
`
`(“Ashworth I”). In plucking Compound 25 from an ocean of promising
`
`compounds, Mylan ignores (1) that Ashworth I only reported in vitro data for
`
`
`
`2
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`Compound 25, while favorably reporting results from in vivo studies with another
`
`
`
`
`
`
`
`
`
`compound that showed “no acute toxicity when injected into mice” (id. at 1166);
`
`(2) that Ashworth I specifically stated that the result of further optimization with its
`
`molecules would be reported soon (id. at 1165); and (3) that Ashworth’s
`
`subsequent optimization publication found that the “optimum” core was not the
`
`“pyrrolidine” core of Ashworth Compound 25, but instead was a more potent
`
`“thiazolidide” core. Ex. 2001, Ashworth et al., 22 Bioorg. & Med. Chem. Lett.
`
`2745, 2746 (1996) (“Ashworth II”). If any of Ashworth’s work suggested a lead
`
`compound for further development, it would have been the structurally different
`
`compounds of Ashworth II, not Compound 25 from Ashworth I. See infra § VI.A.
`
`Even assuming that one of skill in the art would have chosen to start with
`
`Mylan’s lead compound, that compound still requires the combination of no less
`
`than three diverse structural modifications to arrive at saxagliptin: 1) addition of a
`
`cyclopropyl group to the core of the molecule; 2) substitution of an adamantyl
`
`group at the other end of the molecule; and 3) modification of that adamantyl
`
`group with a hydroxy group. Mylan finds those structural modifications in three
`
`additional references (two of which are not related to the subject matter of the
`
`RE’186 patent) and, again, focuses on the structures of pieces of prior art
`
`molecules divorced from the disclosed properties associated with those molecules.
`
`See infra § VI.B.
`
`
`
`3
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`Nowhere are the deficiencies in Mylan’s analysis more glaring than in its
`
`
`
`
`
`
`
`
`
`contentions relating to the “cyclopropyl” group fused to a pyrrolidine core, which
`
`is an important feature of the compounds encompassed by all of the challenged
`
`claims. Consistent with the fact that the DPP-IV inhibitor prior art is devoid of any
`
`suggestion to make this substitution, and actually strongly discourages increasing
`
`the size of such pyrrolidine rings, Mylan finds this feature in a non-DPP-IV
`
`inhibitor reference called Hanessian. Ex. 1010, Hanessian et al., 36 Angew. Chem.
`
`Int. Ed. Engl. 1881 (1997). Hanessian’s publications, however, describe the
`
`addition of a cyclopropyl ring to entirely different classes of compounds, wherein
`
`the ability of such modified molecules to bind to their targets is unpredictable and
`
`is sometimes actually damaged by cyclopropanation. Not only does Hanessian not
`
`motivate cyclopropanation of a DPP-IV inhibitor, it actually suggests lack of
`
`therapeutic success in doing so. Mylan’s flawed cyclopropyl argument thus
`
`amounts to another independent reasonin addition to its improper lead
`
`compound theorywhy the Board should find that Mylan has failed to show that
`
`any claims of the RE’186 patent are likely unpatentable. See infra § VI.B.1.a.
`
`Mylan’s additional proposed modifications needed to reach the specific
`
`structure of saxagliptin are similarly deficient. Mylan finds the “adamantyl” group
`
`attached to a structurally different compound described in Villhauer. Ex. 1008,
`
`WO 98/19998. In so doing, however, Mylan ignores that the Villhauer compounds
`
`
`
`4
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`are profoundly different from saxagliptin—saxagliptin has a “primary amine”
`
`
`
`
`
`
`
`
`
`(NH2) backbone whereas the Villhauer compounds have a “secondary amine”
`
`(NH) backbone that addresses an “intramolecular cyclization” stability problem
`
`noted in the art in connection with primary amine compounds like those of
`
`Ashworth. If one were to have taken a stability lesson from Villhauer, it would
`
`have been to abandon the primary amine backbone of Ashworth in favor of the
`
`secondary amine, not to add an adamantyl group. Moreover, Mylan’s alleged
`
`motivation to add bigger, bulkier groups to Ashworth Compound 25 is refuted by
`
`Ashworth’s actual data. For example, in the preferred Ashworth II “thiazolidide”
`
`series, Ashworth demonstrated that less bulk and less “beta branching” at this
`
`position affords greater stability. See infra § VI.B.3.b.
`
`In another departure from the DPP-IV literature, Mylan chooses Raag to
`
`support an alleged motivation to hydroxylate the adamantyl group. Ex. 1009, Raag
`
`and Poulos, 30 Biochemistry 2674 (1991). In essence, Mylan suggests that one
`
`might have speculated from Raag that a more complicated Ashworth I molecule
`
`with an adamantyl substituent might be metabolized in vivo into a hydroxylated
`
`molecule because Raag describes hydroxylation of the simple, unsubstituted
`
`“adamantane” molecule in the context of an in vitro reaction with a bacterial
`
`metabolizing enzyme. In so contending, however, Mylan (1) fails to explain why
`
`one skilled in the art would look to Raag in the context of DPP-IV inhibitors; (2)
`
`
`
`5
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`ignores the strong art preference to avoid metabolites (not incorporate them into
`
`
`
`
`
`
`
`
`
`lead compounds); (3) ignores art indicating that there are multiple metabolites for
`
`compounds containing an adamantyl group when administered in vivo to humans;
`
`and (4) ignores that drug metabolism is compound specific and highly
`
`unpredictable. See infra § VI.B.3.c.
`
`Even if a skilled artisan could rationalize one of the above-proposed
`
`modifications, Mylan fails to establish the predictability of implementing all of
`
`them together. That failure is especially fatal in this field, where the prior art
`
`demonstrates that even small structural changes can dramatically change the
`
`properties of a molecule, where different parts of DPP-IV inhibitors interact with
`
`each other in unpredictable ways, and where, as a result, individual components on
`
`complex molecules cannot be predictably substituted from one compound to the
`
`next. Mylan’s contrary views on the obviousness issue are supported only by the
`
`post-hoc opinions of its expert, Dr. Rotella.1 The weight to be accorded his current
`
`opinions
`
`is substantially undercut by his pre-litigation admissions and
`
`
`1 Dr. Rotella was a scientist at Bristol-Myers Squibb (“BMS”), working on back-up
`
`inhibitors to saxagliptin. See Ex. 1004, Curriculum Vitae of Dr. Rotella at 1.
`
`Researchers at BMS invented the subject matter of the RE’186 patent. That patent
`
`and others were subsequently purchased by AstraZeneca.
`
`
`
`6
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`contemporaneous activities in the field, which paint a far less predictable picture.
`
`
`
`
`
`
`
`
`
`See infra § II.C.
`
`Mylan also ignores the unexpected and superior properties resulting from
`
`saxagliptin’s unique combination of structural features. Contrary to the premise of
`
`independent operation of individual structural features underlying Mylan’s
`
`analysis, the structural elements of saxagliptin have been reported to work “in
`
`concert” with one another. Ex. 2002, Magnin et al., 47 J. Med. Chem. 2587, 2592
`
`(2004). The properties associated with that combination of distinct features proved
`
`to be surprising and could not have been predicted. These properties include
`
`synergistic potency, improved chemical stability, increased bioavailability, and
`
`superior and unpredictable binding characteristics. See infra § VI.B.3.d. The
`
`structural secret to saxagliptin’s unusual binding properties was understood only
`
`after the fact through later crystallographic studies showing a series of highly
`
`specialized, previously unknown, chemical contacts with the DPP-IV enzyme that
`
`saxagliptin inhibits. See Ex. 2003, Robl and Hamann, 4 RSC Drug Discovery
`
`Series 1, 13-14 (2011). Those unique properties manifest themselves in
`
`AstraZeneca’s commercially marketed saxagliptin-containing products, which sold
`
`over $800 million in 2014 alone. Ex. 2004, AstraZeneca Annual Report and Form
`
`20-F Information 2014 at 3. The combination of saxagliptin’s unique properties
`
`and resultant commercial success is exactly why Mylan now seeks to invalidate the
`
`
`
`7
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`RE’186 patent in order to profit from generic versions of AstraZeneca’s
`
`
`
`
`
`
`
`
`
`saxagliptin-containing pharmaceutical products, and that unexpected combination
`
`of properties and resulting objective evidence of nonobviousness is also why
`
`Mylan should not be permitted to do so. See infra § VI.B.3.d.
`
`For these and other reasons more fully explained below, Mylan’s
`
`obviousness position is legally insufficient to demonstrate a reasonable likelihood
`
`of prevailing and, accordingly, its Petition should be denied. See 35 U.S.C.
`
`§ 314(a).
`
`II. Background
`A. DPP-IV Inhibitors
`DPP-IV is an enzyme involved in glucose metabolism. DPP-IV plays a role
`
`in the regulation of insulin by interacting with a molecule called glucagon-like
`
`peptide 1 (GLP-1). Ex. 2005, Holst and Deacon, 47 Diabetes 1663, 1663 (1998).
`
`GLP-1 directly stimulates insulin release. Id. Accordingly, it is desirable to
`
`maintain levels of GLP-1 in the blood, particularly in the diabetic condition where
`
`natural levels of insulin are lower. Ex. 1001, RE’186 patent at 1:44-67.
`
`DPP-IV, on the other hand, acts to reduce insulin secretion and does so by
`
`inactivating GLP-1. See Ex. 2005, Holst and Deacon at 1663. As a result, one
`
`potential approach to preserving levels of intact GLP-1 in the bloodwhich would
`
`be beneficial for diabeteswould be to try to inhibit DPP-IV and its effect on
`
`
`
`8
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`GLP-1. Ex. 1001, RE’186 patent at 1:44-67. The prior art described a variety of
`
`
`
`
`
`
`
`
`
`putative DPP-IV inhibitors with varying potential to accomplish this goal. As of
`
`the date of the invention of the RE’186 patent, however, the safety and efficacy of
`
`this approach had yet to be established in appropriate clinical trials, and many
`
`unanswered questions remained. See Ex. 2005, Holst and Deacon at 1663.
`
`B. DPP-IV Inhibitors in the Late-90s Varied Greatly
`In the late 1990s, DPP-IV inhibition was associated with a variety of
`
`unknowns. The structure of DPP-IV itself was not yet available. Ex. 2006,
`
`Aertgeerts et al., 13 Protein Science 412, 413 (2004). Some putative DPP-IV
`
`inhibitors were designed with two groups, generally referred to as a “P1 group”
`
`and a “P2 group,” poised to interact with particular features of the DPP-IV active
`
`site. Ex. 2007, Augustyns et al., 6 Curr. Med. Chem. 311, 315-16 (1999). But
`
`without the benefit of knowing the precise structural features of DPP-IV, and how
`
`the “P1” and “P2” portions of a potential inhibitor affect each other and interact
`
`with the enzyme’s active binding sites, even this approach to DPP-IV inhibitor
`
`design was largely “unguided.” Ex. 2003, Robl and Hamann at 7. Moreover, good
`
`in vitro potency was not enough to yield a safe and effective drug, as many
`
`putative candidates failed for a variety of unpredictable reasons. See e.g., id. at 5-
`
`6. This uncertainty led to a wide variety of proposed structural motifs among
`
`potential inhibitors.
`
`
`
`9
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`Indeed, in the late-90s, a plethora of potential DPP-IV inhibitors were being
`
`
`
`
`
`
`
`
`
`explored with widely divergent structures and properties. Compounds containing
`
`“boronic acids” were among the most potent proposed DPP-IV inhibitors with
`
`inhibition values (“Ki” values2) in the low nanomolar (10-9 M) range. Ex. 2008,
`
`Flentke et al., 88 Pro. Nat’l Acad. Sci. 1556, 1557 (Table 1) (1991). They
`
`exhibited reversible, slow, tight-binding kinetics, and according to a patent
`
`application by Bachovchin at Tufts University (cited by Mylan), at least one such
`
`compound “clearly lower[ed] blood sugar based upon results from an oral glucose
`
`challenge in mice.” Ex. 1011, WO 99/38501 at 50:25-27; Ex. 2007, Augustyns at
`
`315-16.
`
`Another promising compound was “Isoleucine [Ile]-thiazolide,” which had
`
`highly specific, reversible inhibitory activity against DPP-IV. Ex. 2009, Pauly et
`
`al., 48 Metabolism 385, 385 (1999). By 1999, Ile-thiazolide was reported to
`
`successfully inhibit DPP-IV in an animal model resulting “in an earlier increase [in
`
`levels of] insulin and a more rapid clearance of blood glucose.” Id. The company
`
`
`2 The “Ki” represents the dissociation constant of the enzyme-inhibitor complex, or
`
`the reciprocal of the binding affinity of the inhibitor to DPP-IV. See Ex. 1025,
`
`Cheng et al., 22 Biochem. Pharm. 3099, 3099 (1973). Thus, an inhibitor with a Ki
`
`of 10-9 M is ten times more potent than one with a Ki of 10-8 M.
`
`
`
`10
`
`

`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`Probiodrug sought to develop this DPP-IV inhibitor as “P32/98” and, by 2000,
`
`
`
`
`
`
`
`
`
`
`
`reported improved glucose tolerance in diabetic patients after a single dose. Ex.
`
`2010, Demuth et al., 49 Diabetes 413-P, 413-P (2000). That these more advanced
`
`compounds had greatly different structures is apparent from Figure 1 below.
`
`
`Figure 1: Representative DPP-IV inhibitors in the Late-90s
`
`Neither compound in Figure 1 has been proposed by Mylan as a “lead compound”
`
`for its obviousness analysis. While possessing promising data at the time, neither
`
`of these compounds ultimately progressed to become a marketable DPP-IV
`
`inhibitor.
`
`In addition to these more advanced examples, there were many other DPP-
`
`IV inhibitor structures available to a person of skill in the art, each with varying
`
`degrees of functional data. For example, a single patent by the Ferring Research
`
`Institute (“Ferring”) reported over 150 structures of putative DPP-IV inhibitors.
`
`Ex. 2011, U.S. Patent No. 5,939,560 (“the ’560 patent”). That patent exemplified
`
`the syntheses of only seven compounds and provided Ki data for only fifteen
`
`compounds. Id. at Table 9, Experimental Details for Examples 1-7. Only four
`
`compounds had both synthesis and Ki data, each having a unique structure as
`11
`
`
`
`

`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`shown in Figure 2. Id. Significantly, Mylan’s proposed lead compound,
`
`
`
`
`
`
`
`
`
`
`
`Ashworth Compound 25, is disclosed in the ’560 patent, but there are no potency
`
`data for that compound and its synthesis is not disclosed.3 Id. at col. 13 (compound
`
`#16).
`
`Figure 2: Exemplified Compounds with Potency Data in the ’560 Patent
`
`
`
`In the meantime, an entirely different DPP-IV inhibitor had already
`
`advanced to the early stages of clinical trials by 2000. Novartis’s inhibitor NVP-
`
`DPP728 was reported to increase active GLP-1 levels in normal human subjects,
`
`supporting “the investigation of the glucose-lowering potential of NVP-DPP728
`
`for the treatment of type 2 diabetes.” Ex. 2012, Rothenberg et al., 49 Diabetes
`
`160-OR, 160-OR (2000). The chemical structure of this inhibitor, which had a
`
`secondary amine (NH) in its backbone instead of the primary amine (NH2) in
`
`saxagliptin, is shown in Figure 3.
`
`
`3 The Ashworth I and Ashworth II publications, and the compounds disclosed
`
`therein, were part of the work of the Ferring group.
`
`
`
`12
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`
`
`Figure 3: Novartis’s Clinically Advanced DPP-IV Inhibitor
`
`Notably, the Villhauer publication on which Mylan relies was from
`
`Novartis, but the original Novartis clinical trial candidate NVP-DPP728 was not
`
`the compound Mylan cites. Ex. 1008, WO 98/19998. Indeed, the only prior art
`
`Villhauer compound to be introduced commercially anywhere in the world,
`
`vildagliptin, shown in Figure 4 (Ex. 2013, U.S. Patent No. 6,166,063 (“the ’063
`
`patent”)), similarly had a secondary amine (NH) in its backbone.
`
`Figure 4: Vildagliptin
`
`
`
`Again, Mylan did not propose the exemplified compounds of the Ferring
`
`’560 patent, the Novartis compound NVP-DPP728, or vildagliptin as the “lead
`
`compound” for its obviousness analysis. As was true for the other more advanced
`
`DPP-IV inhibitors from this period in time, neither the Ferring compounds nor
`
`NVP-DPP728 became a commercial product, and vildagliptin was later approved
`
`only in Europe, but not in the United States.
`
`
`
`13
`
`

`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`In addition to these DPP-IV inhibitors, others in the art had also disclosed
`
`
`
`
`
`
`
`
`
`
`
`compounds that inhibited DPP-IV by a different, “irreversible” mechanism. See
`
`Ex. 2007, Augustyns at 316-317. Irreversible inhibitors were of interest since, at
`
`the time, it was unknown whether DPP-IV inhibition should be permanent or
`
`transient. Ex. 2005, Holtz and Deacon at 1668. One such irreversible inhibitor
`
`was reported to reduce in vivo levels of DPP-IV in the blood and peripheral tissues.
`
`Ex. 2007, Augustyns at 316-317. Another such compound, having high potency
`
`and high stability, was also reported to demonstrate “favorable results in vivo.” Id.
`
`The structures of these two irreversible inhibitors are shown in Figure 5.
`
`Figure 5: Irreversible DPP-IV inhibitors
`
`
`
`Neither was proposed as the lead compound by Mylan and, like most of the DPP-
`
`IV inhibitors that were being explored and tested in the art during this time period,
`
`these irreversible inhibitors never became approved drugs.4
`
`
`4 There were many, many more putative DPP-IV inhibitors reported in the
`
`literature. This discussion identifies only a representative sampling of the options
`
`available to those of skill in the art.
`
`
`
`14
`
`

`
`
`
`
`
`
`
`
`
`
`
`C.
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`Structural Modifications of Known DPP-IV Inhibitors Yielded
`Unpredictable Results
`
`As shown by the structural diversity of the DPP-IV inhibitors identified
`
`above that were never commercially marketed, there was no “recipe” or particular
`
`structure that predicted likely success. DPP-IV inhibitors took on a variety of
`
`structures, and each structural solution for binding DPP-IV was highly specific.
`
`Further challenging those of skill in the art, even slight structural modifications to
`
`promising candidates had potentially dramatic and unpredictable effects. This
`
`unpredictability is illustrated by numerous examples throughout the art.
`
`Particularly germane to Mylan’s allegation of the obviousness of adding a
`
`cyclopropyl ring to the Ashworth compounds, Ashworth itself reported that the
`
`addition of a methyl group in lieu of one of the hydrogens on the cyanopyrrolidine
`
`core greatly diminished activity. Figure 6; Ashworth II at 2747, Table I.
`
`
`Figure 6: Adding a Methyl Group Decreased Potency
`
`Equally pertinent to the alleged obviousness of increasing the size of the 5-
`
`membered pyrrolidine core of Ashworth by cyclopropanating it, the seemingly
`
`minor replacement of the five-membered pyrrolidine ring with a six-membered
`
`piperidine ring, while holding the rest of the molecule constant, resulted in a more
`
`
`
`15
`
`

`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`than 100-fold decrease in potency. Figure 7A; Ex. 2001, Ashworth II at 2747.
`
`
`
`
`
`
`
`
`
`
`
`Likewise, replacing the five-membered thiazolidine ring with a six-membered
`
`thiazinane ring resulted in a more than 10,000-fold decrease in potency. Figure
`
`7B; Ex. 2001, Ashworth II at 2747.
`
`
`Figure 7: Increasing Size by a Single Carbon Decreased Potency
`
`Other Ferring work showed that replacement of an “amide” bond with an
`
`“alkene,” thought to be “isosteric” (about the same size and spatial configuration),
`
`dramatically decreased potency from a Ki of 2.2 nM to 1700 nM. Figure 8;
`
`compare Ex. 1007, Ashworth I at 1166, with Ex. 2011, ’560 patent at Table 9.
`
`
`Figure 8: Other Single Replacements Decreased Potency
`
`
`
`16
`
`

`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`Mylan’s expert, Dr. Rotella was no stranger to these unpredictable effects.
`
`
`
`
`
`
`
`
`
`
`
`Based on his work on DPP-IV inhibitors, with full knowledge of saxagliptin, he
`
`noted that minor structural changes yielded significant differences in the potencies
`
`of the compounds. See Ex. 2014, Simpkins et al., 17 Bioorg. Med. Chem. Lett.
`
`6476, 6479 (2007). For example, in his work at BMS, Dr. Rotella studied some
`
`analogs of saxagliptin that did not have a cyano (CN) substitution. See generally
`
`id. As shown in Figure 9, simply removing the cyano group from saxagliptin
`
`yielded a compound with a Ki of 10 nM (Compound 48), but adding a “simple
`
`methyl substitution” to that molecule “essentially destroy[ed] all activity”
`
`(Compound 54 having a Ki of >10,000 nM). Id. at 6479.
`
`
`
`Figure 9: Dr. Rotella’s Single Methyl Substitution Destroys Potency
`
`Dr. Rotella’s work also confirmed the potentially dramatic adverse impact
`
`on potency of changing from a “primary amine” backbone, like that of saxagliptin,
`
`to a “secondary amine” backbone of the sort proposed in Villhauer. In a series of
`
`inhibitors lacking a cyano group, the saxagliptin primary amine analog had a Ki of
`
`10 nM (Compound 48), while the secondary amine analog (like Villhauer) had a
`
`dramatically worse Ki of 3081 nM (Compound 60). Id.; Figure 10.
`
`
`
`17
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`
`
`
`Figure 10: Dr. Rotella’s Secondary Amine Had Decreased Potency
`
`Given his personal experience, it is no wonder that Dr. Rotella freely opined,
`
`before being retained by Mylan for this case, that “[e]ven a minor modification of a
`
`drug structure can completely modify the properties of the molecule.” Ex. 2015,
`
`Fischer et al., Pioneer and Analogue Drugs, in Analogue-Based Drug Discovery
`
`III, 3, 3 (2013).
`
`D.
`
`Potential DPP-IV Inhibitors Faced Additional Unpredictable
`Hurdles Beyond Potency
`
`Potency was not the only consideration when searching for a suitable DPP-
`
`IV inhibitor. The challenge was to discover a molecule that had both good potency
`
`and good stability, in addition to other key properties, such as acceptable toxicity
`
`and selectivity for DPP-IV. In particular, chemical stability turned out to be a
`
`major problem for a number of putative DPP-IV inhibitors. As previously noted,
`
`proposed molecules with a “primary amine” group in the backbone, particularly if
`
`the core included a “cyano” group, were vulnerable to a reaction called
`
`“intramolecular cyclization,” in which the nitrogen in the primary amine became
`
`linked to the carbon of the cyano group. See Ex. 2007, Augustyns at 314.
`
`
`
`18
`
`

`
`
`
`
`
`
`
`
`
`Case No. IPR2015-01340
`Patent RE44,186
`The problem of intramolecular cyclization in primary amine backbones was
`
`
`
`
`
`
`
`
`
`not trivial. It caused numerous groups to abandon compounds susceptible to this
`
`type of reaction in favor of more stable alternatives. See, e.g., Ex. 1015, Lin et al.,
`
`95 Proc. Nat’l Acad. Sci. 14020, 14020-21 (1998). By way of example, at least
`
`three groups developed different structural solutions to reduce this problem and
`
`obtain useful compounds. Specifically, the Villhauer group (Novartis) used a
`
`backbone with a “secondary amine (NH)” and reported that “less than 1% of the
`
`compound is expected to cyclize during the time frame of the current
`
`investigations.” Figure 11A; Ex. 2016, Hughes et al., 38 Biochem. 11597, 11599
`
`(1999). This is the same compound that Novarti