UNITED STATES PATENT AND TRADEMARK OFFICE
`_____________________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`_____________________________
`
`
`
`MYLAN PHARMACEUTICALS INC.,
`WOCKHARDT BIO AG, TEVA PHARMACEUTICALS USA, INC.
`and AUROBINDO PHARMA U.S.A. INC.,
`Petitioners,
`
`v.
`
`ASTRAZENECA AB,
`Patent Owner.
`
`_____________________________
`
`Case IPR2015-01340
`Patent RE44,186 E1
`_____________________________
`
`
`
`PETITIONERS’ REPLY
`37 C.F.R. §42.24(c)
`
`
`
`1 Wockhardt (IPR2016-01209), Teva (IPR2016-01122) and Aurobindo (IPR2016-
`
`01117) have each been joined to this proceeding.
`
`

`
`TABLE OF CONTENTS
`
`I.
`
`II.
`
`Introduction .................................................................................................. 1
`
`Response to AstraZeneca’s arguments .......................................................... 2
`
`A. Dates of invention ............................................................................... 2
`
`B.
`
`Scope and content of the art ................................................................ 4
`
`1.
`
`2.
`
`3.
`
`4.
`
`5.
`
`Type-2 diabetes and DPP-4 inhibitors ....................................... 4
`
`Structural requirements for a safe and effective
`DPP-4 inhibitor were largely unknown ..................................... 5
`
`Stability concerns...................................................................... 6
`
`Different solution to intramolecular cyclization ........................ 7
`
`DPP-4 inhibitors in clinical studies ........................................... 8
`
`Invention of saxagliptin ...................................................................... 9
`
`Level of ordinary skill in the art ........................................................ 10
`
`Differences between the claimed invention and the prior art ............. 10
`
`1.
`
`Ashworth Compound 25 ......................................................... 10
`
`2. Modifications .......................................................................... 13
`
`3.
`
`Secondary considerations ........................................................ 18
`
`C.
`
`D.
`
`E.
`
`F.
`
`The other claims ............................................................................... 27
`
`III. Conclusion .................................................................................................. 27
`
`Appendix of Exhibits ............................................................................................ 29
`
`Certificate of Compliance ..................................................................................... 34
`
`Certificate of Service ............................................................................................ 35
`
`
`
`
`
`-i-
`
`
`
`
`
`
`
`

`
`I.
`
`INTRODUCTION
`
`Patent Owner (“AstraZeneca” for both the current owner and its
`
`predecessor) filed a response (Paper 28, POR) attacking the prima facie case of
`
`Petitioners (collectively, “Mylan”) and arguing secondary considerations. Much of
`
`AstraZeneca’s argument depends on its having invented an FDA-approved DPP-4
`
`inhibitor. AstraZeneca does not claim an FDA-approved DPP-4 inhibitor and
`
`cannot read such a limitation in from the specification because it did not disclose
`
`any basis for a POSA to think any claimed compound would be FDA-approved or
`
`even safe and effective in vivo. AstraZeneca disclosed and claimed DPP-4
`
`inhibitors that were plausible candidates for further study as type-2 diabetes
`
`treatments. At the time of filing, the prior art was awash with promising candidates
`
`that had established potency in vitro and sometimes more. While one of
`
`AstraZeneca’s many claimed compounds (saxagliptin) has since obtained FDA
`
`approval, it is not a remarkable member of its class (gliptins). It lagged another
`
`gliptin to approval, lags both that gliptin and a subsequently approved gliptin in
`
`sales, and has earned an FDA safety warning. In sum, judged from a POSA’s
`
`perspective at the time of filing, saxagliptin was just another dipeptide DPP-4
`
`inhibitor. Even judged in light of subsequent developments, saxagliptin is just
`
`another of several therapeutic gliptins and is not even particularly successful
`
`among them.
`
`-1-
`
`
`
`

`
`II. RESPONSE TO ASTRAZENECA’S ARGUMENTS
`
`A. Dates of invention
`
`AstraZeneca says saxagliptin has unique structural features in its P1 and P2
`
`groups. POR 4. Yet in 1993, Mentlein (EX2096) identified the key features of an
`
`ideal DPP-4 dipeptide substrate, all of which are in saxagliptin: proline in the P1
`
`position (at 829) with a “bulky” C-linked R
`
`group and an “obligate” free amine in the P2
`
`position (at 833; Fig. 4 (detail), right).
`
`AstraZeneca’s chemist-expert Dr. Weber concedes that adamantyl and
`
`hydroxyadamantyl would have been considered “bulky” groups. EX1073, 60:6-
`
`61:2. Ashworth I (EX1007) noted similar preferences for the substrate (H-X-
`
`Pro/Ala-Y-) and proceeded to investigate DPP-4 inhibitors with 2-cyano-proline
`
`and bulky R groups, reported good results for stability, and planned to modify the
`
`proline further for greater potency. While Mentlein did not expressly anticipate
`
`selection of hydroxyadamantyl as the R group and cyclopropyl as the proline
`
`modification, these selections were a linear progression from what a POSA had
`
`long known about the DPP-4 substrate.
`
`-2-
`
`
`
`

`
`AstraZeneca identifies October 2000 as the date of saxagliptin synthesis and
`
`testing for DPP-4 inhibitor function. POR 4.2 AstraZeneca’s response provides no
`
`evidence of human safety and efficacy results for its claimed compounds until after
`
`the involved patent’s filing date. To the extent that a “safe and effective” DPP-4
`
`inhibitor is the proper standard for a successful invention, AstraZeneca provides no
`
`evidence that even AstraZeneca itself knew it possessed such a compound, much
`
`less that a POSA would have appreciated it. In re Paulsen, 30 F.3d 1475, 1481 n.9
`
`(Fed. Cir. 1994) (explaining specification would not be enabling if held to same
`
`standard urged for reference). Conversely, if a POSA would have had any reason
`
`to make a promising DPP-4 inhibitor, then saxagliptin would have been an obvious
`
`candidate. Cf. In re Dillon, 919 F.2d 688, 692-93 (Fed. Cir. 1990) (en banc) (“the
`
`statement that a prima facie obviousness rejection is not supported if no reference
`
`shows or suggests the newly-discovered properties and results of a claimed
`
`structure is not the law.”).
`
`
`
`
`
`2 Despite citing evidence that saxagliptin was its most potent candidate in
`
`October 2000 (EX2189, 2), AstraZeneca never identified saxagliptin as its best
`
`mode.
`
`-3-
`
`
`
`

`
`AstraZeneca states the involved patent is “based” on a provisional
`
`application filed on 16 February 2001. POR 5. The provisional neither claims nor
`
`even discloses saxagliptin generically, much less specifically. EX1027. While the
`
`patent included saxagliptin as part of a Markush group in claim 8, AstraZeneca had
`
`to file a reissue application in 2011 to add claims specific to saxagliptin (and to
`
`cancel claim 23, which included treatment of immunological diseases).
`
`B. Scope and content of the art
`
`1. Type-2 diabetes and DPP-4 inhibitors
`
`AstraZeneca contends that, in the late 1990s, type-2 diabetes treatments had
`
`serious side effects and new treatments were needed. POR 5. Type-2 diabetes still
`
`has no cure and new treatments continue to be needed. EX1041, ¶31; EX1028,
`
`396:9-14. The physician-experts, Mylan’s Dr. Tanenberg and AstraZeneca’s Dr.
`
`Lenhard, confirm that the main 1990s treatments (including metformin and
`
`sulfonylureas) are still used today. EX1041, ¶¶18-21; EX1028, 395:15-396:8;
`
`EX1073, 105:15-20. The World Health Organization lists metformin and one
`
`sulfonylurea (but no gliptin) as “essential” medicines for type-2 diabetes. EX1041,
`
`¶¶19, 21; EX1043, 34. Diabetes specialists still prescribe metformin and
`
`sulfonylureas as first-line treatments. EX1041, ¶¶18-21; EX1028, 395:15-396:8;
`
`EX1073, 105:15-20.
`
`-4-
`
`
`
`

`
`AstraZeneca states that scientists proposed DPP-4 inhibitors as candidates
`
`for modulating insulin levels by inhibiting GLP-1 in 1998. POR 5. Mentlein
`
`discussed it in 1993. EX2096, abstract & 833.
`
`AstraZeneca reports concerns about using DPP-4 inhibitors in vivo because
`
`DPP-4 had other potential substrates and was known to play a role in the immune
`
`system. POR 5-6. Yet the involved patent has no in vivo data addressing those
`
`concerns and, before the 2011 reissue, itself claimed (without data or guidance)
`
`treatment of immune diseases. EX1001, claim 23. AstraZeneca provides no
`
`evidence that at the filing date saxagliptin had extensive characterization including
`
`in vivo evaluation for any purpose.
`
`2. Structural requirements for a safe and effective
`
`DPP-4 inhibitor were largely unknown
`
`AstraZeneca notes that the DPP-4 crystal structure was unknown so studies
`
`focused on structure-activity relationships, which led to many prior art dipeptide
`
`inhibitors that (based on in vitro evidence)3 were sufficiently potent. POR 6-7.
`
`AstraZeneca argues that the “greater hurdle was obtaining other necessary
`
`properties for a safe and effective therapeutic DPP-4 inhibitor.” Id. at 7.
`
`AstraZeneca does not claim—and did not disclose any basis to claim—a “safe and
`
`
`
`3 Note the sufficiency of in vitro data for AstraZeneca’s conclusion.
`
`-5-
`
`
`
`

`
`effective therapeutic” DPP-4 inhibitor. AstraZeneca provides no evidence of any
`
`“necessary properties” for saxagliptin at the time of filing.
`
`3. Stability concerns
`
`AstraZeneca argues that Ashworth I reported serious stability problems for
`
`C-linked (rather than N-linked) dipeptides. POR 7-9, 25. Yet AstraZeneca notes
`
`that Ashworth reported “excellent” stability for certain dipeptides with bulky C-
`
`linked alkyl groups (including Compound 25) and planned next to optimize the
`
`proline ring. POR 9. AstraZeneca notes that stability was studied at room
`
`temperature (rather than physiological temperature), but the response does not
`
`show whether this is a concern. POR 9, 25, 28. Obviously it was not a concern for
`
`the Ashworth group, which was so satisfied with their stability results that they
`
`advanced to another optimization. Note that AstraZeneca later proposes (POR 12)
`
`a different lead compound (NVP-DPP728) based on its room-temperature binding
`
`and stability data.
`
`AstraZeneca notes that Ashworth II favored substituting sulfur into the
`
`proline ring because even-larger rings adversely affected potency. POR 9-10.
`
`Similarly, AstraZeneca notes that Augustyns 1997 concludes the DPP-4 subsite for
`
`P1 ideally fits a saturated 5-membered saturated ring. POR 10. Both Ashworth
`
`articles disclose saturated 5-membered rings (including Compound 25). As Mylan
`
`expert Dr. Rotella explains, Ashworth’s sulfur-substituted ring maintains a
`
`-6-
`
`
`
`

`
`saturated 5-membered saturated ring and shows that DPP-4 allows some size
`
`increase at the 4-position. 4,5-Cyclopropanation also preserves the saturated 5-
`
`member ring. EX1074, ¶40.
`
`AstraZeneca notes Ashworth reported toxicity testing, but did not report
`
`further development of Compound 25. POR 10-11. Compound 25 is a natural
`
`evolution from Mentlein’s ideal: a dipeptide with a P1 proline and a P2 with a
`
`bulky C-linked group and a free amine. Ashworth did not identify any problem
`
`with Compound 25. Ashworth’s progress in a different direction is hardly a
`
`teaching away. As AstraZeneca’s Dr. Weber acknowledged, pharmaceutical
`
`companies may choose to pursue or not pursue promising candidates for business
`
`reasons. EX1073, 100:21-101:10. Ashworth II uses the best “natural” amino acid
`
`for further optimization studies. EX2001, 2746. Ashworth does not say why, but
`
`cost and availability would be sufficient reasons to experiment with a natural rather
`
`than synthetic substrate. EX1074, ¶13.
`
`4. Different solution to intramolecular cyclization
`
`AstraZeneca urges the art abandoned C-linkage for N-linkage to solve the
`
`stability problem. POR 11-13. Recall, however, that Ashworth reported “excellent”
`
`stability for several C-linked compounds, including Compound 25. Pursuing
`
`another path is hardly a teaching away.
`
`-7-
`
`
`
`

`
`5. DPP-4 inhibitors in clinical studies
`
`AstraZeneca argues that a POSA would have chosen as lead one of two
`
`compounds already in human clinical trials. POR 14-16. Again, pursuing other
`
`options is not a teaching away.
`
`While AstraZeneca’s Dr. Weber suggests that Ashworth’s 2-cyano group
`
`raised concerns about cyanide release, she does not corroborate her concerns or
`
`cite a real-world instance of such release by the mechanism she describes. POR 15,
`
`24. Ashworth expressly refutes the toxicity concern, both in vitro and in vivo.
`
`EX1007, 1166. When asked about her concern in cross examination, Dr. Weber
`
`avoided defending her testimony. EX1073, 51:2-54:14. Significantly, two of the
`
`three alternative leads that Dr. Weber proposes have a 2-cyano-proline, EX2056,
`
`¶156 (NVP-DPP728), ¶163 (“cyanothiazolidines of Ashworth-II”), so her
`
`argument is internally inconsistent.
`
`Vildagliptin, a prior art DPP-4 inhibitor shown at POR 15, also has 2-cyano-
`
`proline. AstraZeneca does not report cyanide release for vildagliptin. Instead,
`
`AstraZeneca argues that vildagliptin is not FDA approved (although it is approved
`
`in Europe), must be administered twice daily, and has a liver-screening
`
`requirement. POR 15. The involved claims do not specify a dosage regimen or a
`
`geographical location of treatment. Vildagliptin is available in a once-daily
`
`combination therapy. EX1041, ¶41. As for warning labels, saxagliptin itself has a
`
`-8-
`
`
`
`

`
`warning label for cardiac hospitalization. EX1041, ¶26 (and noting sitagliptin does
`
`not have a warning); EX1028, 399:8-402:1 (same); EX1032 (JTX-146); EX1033
`
`(JTX196). Vildagliptin serves diabetics’ needs outside the United States. EX1041,
`
`¶39; EX1028, 396:21-24. In terms of world-wide sales, vildagliptin has
`
`comparable success to saxagliptin. EX1060, ¶18.
`
`C. Invention of saxagliptin
`
`In describing research efforts leading to saxagliptin, AstraZeneca explains
`
`that they tried a vinylcyclopentyl analogue of saxagliptin and realized from its
`
`short half-life but prolonged effect that an active metabolite was likely; then found
`
`a similar effect for deshydroxysaxagliptin (saxagliptin without OH). POR 18-19.
`
`The story confirms that routine PK/PD testing would have led to an active
`
`metabolite. EX1074, ¶52. Moreover, the involved patent claims
`
`deshydroxysaxagliptin both specifically (claim 8) and generically, although
`
`AstraZeneca’s Dr. Weber says each of the compounds in claim 8 except
`
`saxagliptin is a failure because they are not FDA approved. EX1073, 19:12-34:18.
`
`Once again, AstraZeneca’s insistence that FDA approval is the only appropriate
`
`consideration contradicts its actual conduct in seeking patent protection.
`
`-9-
`
`
`
`

`
`D. Level of ordinary skill in the art
`
`AstraZeneca does not contest Mylan’s stated level of skill in the art.
`
`POR 21. While AstraZeneca tout’s Dr. Weber’s work on a different FDA-approved
`
`gliptin, POR 2 n.1, that experience is not required for a POSA.
`
`E. Differences between the claimed invention and the prior art
`
`1. Ashworth Compound 25
`
`AstraZeneca urges that case law requires a reason to select the proposed lead
`
`compound over “other compounds in the prior art.” POR 22. Mylan provided
`
`ample reason to select Compound 25 over other prior art compounds. For example,
`
`Ashworth itself discloses several compounds in the prior art over which it
`
`improves, e.g., boroprolines. EX1007, 1163. Ashworth next provides results for 25
`
`compounds and selects four (including Compound 25) as particularly good.
`
`EX1007, 1165. Ashworth even reports preliminary toxicity results for those four.
`
`EX1007, 1166. Dr. Rotella explained why the good potency and great stability of
`
`Compound 25 led him to choose it as a lead. EX1003, ¶¶101-104. Other choices
`
`might be possible but, as the Board has explained, Mylan need not show that
`
`Compound 25 is the only possible lead compound. That is not how pharmaceutical
`
`research actually proceeds. Dr. Weber herself proposed three alternative leads.
`
`EX2056, ¶¶155, 163.
`
`-10-
`
`
`
`

`
`AstraZeneca suggests that Dr. Rotella used improper hindsight in not
`
`choosing Dr. Weber’s proposed leads and imposes on Dr. Rotella two additional
`
`requirements without any authority: proving that no other lead was possible and
`
`proving that the lead itself was clinically successful. POR 24. Both of these
`
`additional requirements are legally and logically untenable. Dr. Rotella did not use
`
`hindsight in choosing Compound 25, but instead gave reasons solidly grounded in
`
`the prior art that a POSA would have readily understood. Pet. 24-26. The precedent
`
`on which AstraZeneca relies requires a reason for choosing a lead compound, not
`
`exhaustive reasons for excluding all conceivable alternatives.
`
`Moreover, AstraZeneca’s clinical-success standard (POR 25) is not met in
`
`the involved patent, which presents no data suggesting any clinical success by the
`
`filing date. Paulsen, 30 F.3d at 1481 n.9 (rejecting double standard for prior art
`
`disclosure).
`
`AstraZeneca describes cases requiring comparison with known drugs of the
`
`same class. POR 25-26. On the facts of this case, the requirement is a Catch-22: no
`
`gliptin (including saxagliptin) was already marketed or otherwise known to be safe
`
`and effective at the time. Logically, no such requirement governs a new drug class.
`
`Fortunately, the pharmaceutical industry does not proceed as AstraZeneca suggests
`
`or it would never be able to pick a lead compound for a new drug class.
`
`-11-
`
`
`
`

`
`AstraZeneca suggests that Ashworth II compounds would be better leads.
`
`POS 26-27. In fact, the Ashworth articles are not mutually exclusive. Ashworth I
`
`lauds Compound 25 as one of four top choices and then plans to optimize the
`
`proline. EX1007, 1165. Ashworth II explains that it is using a different P2 amino
`
`acid for its optimizations because it is the most potent natural amino acid.
`
`EX2001, 2745. Nothing in Ashworth teaches away from using Compound 25, so
`
`the likeliest explanation for the substitution is the cost and availability of a natural
`
`amino acid versus Compound 25’s synthetic amino acid. EX1074, ¶13.
`
` Compound 25 modified as Ashworth II proposes (a cyanothiazolidine
`
`analogue) is actually consistent with the further modification that Mylan has
`
`proposed. Sulfur substitution at the 4-position on the proline ring made the
`
`inhibitor more potent. Dr. Rotella explains that the sulfur substitution would have
`
`preserved a saturated 5-membered ring (as Augustyns 1997) requires.
`
`Cyclopropanation of the ring would also have preserved the saturated 5-membered
`
`ring. EX1074, ¶40. Thus, far from teaching away from Compound 25, Ashworth
`
`teaches a modification consistent with cyclopropanating proline.
`
`AstraZeneca cites Mylan as saying a free amino terminal is inherently
`
`unstable. POR 28-29. AstraZeneca overlooks Mylan’s reliance on Ashworth as
`
`adequately addressing the problem with bulky C-linked acyl groups. There is
`
`-12-
`
`
`
`

`
`neither inconsistency in Mylan’s position nor teaching away in the Ashworth
`
`articles.
`
`2. Modifications
`
`AstraZeneca argues Mylan failed to treat the modifications as an ordered
`
`whole. POR 54-57. The petition (Paper 1, Pet.) selects Compound 25 (at 24) as
`
`lead, then mirrors Ashworth’s order—optimizing P2, then P1—by selecting
`
`adamantyl as the P2 bulky group (at 25), hydroxylating the adamantyl (at 27), then
`
`cyclopropanating the P1 ring (at 28).
`
`a. Adamantyl
`
`AstraZeneca argues a POSA would have no reason to substitute adamantyl
`
`for Compound 25’s cyclohexyl ring because Compound 25 was already stable
`
`enough. POR 41. AstraZeneca does not cite support for its position. While
`
`Ashworth says that the stability of Compound 25 is “excellent”, Ashworth never
`
`says it could not be improved. EX1007, 1165; EX1074, ¶12. As AstraZeneca’s
`
`own exhibit shows, those in the art already knew that P2 should be “bulky”.
`
`EX2096, 833. Dr. Weber concedes that adamantyl and hydroxyadamantyl would
`
`have been considered “bulky” groups. EX1073, 60:6-61:2. As the petition
`
`explained, Villhauer was pursuing similar modifications on the adjacent P2
`
`nitrogen using bulkier substituents. Pet. 26.
`
`-13-
`
`
`
`

`
`Both Ashworth and Villhauer used cyclopentyl and cyclohexyl to impede
`
`cyclization, but Villhauer characterized adamantyl as even more preferred than
`
`unsubstituted cyclohexyl. Pet. 26, citing EX1008, 5. A POSA would have
`
`understood that adamantyl’s extra bulk must be the basis for this preference since
`
`both adamantyl and cyclohexyl are N-linked in Villhauer so the linkage cannot be
`
`the reason for Villhauer’s preference. Hence, a POSA would consider substituting
`
`adamantyl for cyclohexyl on Compound 25 both for its greater steric effect and for
`
`its greater bulk in DPP-4’s S2 subsite. EX1074, ¶¶19-21. While AstraZeneca notes
`
`(POR 45) that Villhauer provides data for compounds without adamantyl, it is not
`
`a teaching away from adamantyl. Villhauer invented a clinically successful gliptin
`
`(vildagliptin), which uses hydroxyadamantyl, before the involved patent’s filing
`
`date. EX2013, cover, Example 1; POR 15, EX1028, 396:15-24.
`
`AstraZeneca argues that Ashworth’s largest group
`
`(on Compound 28) decreases stability.
`
`POR 46 (see POR Fig. 20, right,
`
`comparing with cyclohexyl).
`
`As Dr. Rotella explained,
`
`Compound 28 has a large
`
`number of atoms, but that does
`
`not mean it has a similar steric effect. EX2174, 154:18-24. AstraZeneca makes no
`
`-14-
`
`
`
`

`
`showing that a POSA would have considered it to be “bulky”, whereas both
`
`Drs. Rotella and Weber testify that adamantyl and hydroxyadamantyl would have
`
`been considered “bulky.” EX1003, ¶115; EX1073, 60:6-61:2.
`
`AstraZeneca argues that Ashworth I actually teaches away from using an
`
`adamantyl because it has a quaternary carbon and Ashworth’s quaternary-carbon
`
`examples (both tert-butyl) were less potent. POR 47-48. AstraZeneca overlooks
`
`the petition’s rationale (Pet. 7, 31; EX1003, ¶115) that bulky is better. Cf. EX2096,
`
`833 (bulky P2 preferred since 1993). Adamantyl is manifestly bulkier than tert-
`
`butyl. EX1074, ¶23. AstraZeneca cannot manufacture a teaching away by
`
`substituting a strawman rationale.
`
`b. Hydroxyadamantyl
`
`AstraZeneca argues a POSA had no reason to hydroxylate the adamantyl
`
`because it was unnecessary for solubility or permeability. POR 49-51. Dr. Rotella
`
`explains, however, that a POSA would expect the hydrophobic adamantyl to offset
`
`the favorable solubility of the amine and would consider hydroxylation to
`
`compensate for this loss. EX1074, ¶26.
`
`AstraZeneca argues that hydroxylation through metabolism is unpredictable,
`
`with different enzymes producing different results. POR 51-53. AstraZeneca
`
`misses the point: a POSA would expect oxidation, most likely at a tertiary carbon.
`
`The adamantyl binds to the dipeptide at one tertiary carbon, leaving three
`
`-15-
`
`
`
`

`
`interchangeable tertiary carbons for oxidation. A POSA would have considered
`
`screening for metabolites routine. Not only would a POSA have reasonably
`
`expected oxidation on a tertiary carbon, but routine screening would have revealed
`
`it. EX1074, ¶¶31, 52; cf. EX2173, ¶¶12-13.
`
`AstraZeneca argues that the art favored a lipophilic group, teaching away
`
`from hydroxylation, which might disrupt binding. POR 53-54. AstraZeneca relies
`
`on a comparison between Ashworth I compounds 9 and 19, which have very small
`
`alkyl groups and bear no resemblance to adamantyl or even cyclohexyl. By
`
`contrast, when Villhauer 1998 teaches that adamantyl is even more preferred to
`
`unsubstituted cyclohexyl in the P2 position, Villhauer also prefers (C4-6)cycloalkyl
`
`monosubstituted in the 1-position with hydroxymethyl. EX1008, 5. Similarly,
`
`Villhauer’s patent teaches hydroxyadamantyl in the P2 position. EX2013, 7:15-25.
`
`Thus, far from teaching away from hydroxylating the bulky P2 moiety, a POSA
`
`would actually think it would be tolerated and even preferred.
`
`c. Cyclopropanation
`
`AstraZeneca critiques Dr. Rotella’s testimony on cyclopropanation, POR 30-
`
`41, but misapprehends his explanation and overlooks ample supporting evidence in
`
`the record. Dr. Rotella correctly described cyclopropanation as a known method
`
`for modulating a proline ring as part of drug studies. Hanessian specifically notes
`
`proline conformations are important to their biological function, constrained
`
`-16-
`
`
`
`

`
`prolines are an important investigative tool in peptidomimetics, and
`
`cyclopropanation provides highly controlled changes for studying 5- and 6-
`
`membered rings. EX1010, 1881-82. Dr. Weber admitted modifying peptides into
`
`small-molecule inhibitors is a type of peptidomimetic research. EX1073, 66:3-9.
`
`Ashworth contemplates modifying DPP-4 peptide substrates into dipeptide
`
`inhibitors. EX 1007, 1163. Hence, Hanessian is perfectly applicable to Ashworth’s
`
`suggested “optimisation of the pyrrolidine ring.” EX1007, 1165.
`
`Mylan explained that changing the proline conformation would change the
`
`2-cyano orientation, which could improve stability (reduce cyclization) or potency.
`
`Pet. 29. AstraZeneca notes that the prior art knew changing proline conformation
`
`with a double bond or standard 6-member ring reduces potency. POR 34. By at
`
`least 1997 the art also preferred a saturated 5-member ring. EX2151, 303; EX1073,
`
`89:6-11; 91:4-11. These teachings corroborate Dr. Rotella’s testimony that a POSA
`
`would have explored a ring-fusion strategy. EX1003, ¶55. Cyclopropanation
`
`preserves the saturated five-member ring while also providing controlled
`
`modifications to the ring’s conformation. EX1074, ¶¶39-42. Far from teaching
`
`away, the art’s knowledge is consistent with cyclopropanation providing a better
`
`approach to proline-ring modification.
`
`While AstraZeneca urges the art taught away from larger P1 structures, such
`
`as 6-member rings, POR 36, Dr. Rotella explains that Ashworth’s sulfur-
`
`-17-
`
`
`
`

`
`substituted ring shows that modest increases in ring size are tolerated and can even
`
`result in greater potency. EX1074, ¶39.
`
`AstraZeneca contends that Augustyns 1997 taught only isosteric additions
`
`were permitted on the proline ring. POR 35. In fact, Augustyns explained that only
`
`an isosteric substitution would be permitted at the ring’s 3-position. EX2151, 303.
`
`Augustyns does not teach away from 4,5-methanoproline; rather, it provides a
`
`reason for avoiding cyclopropanation at the 3-position. EX1074, ¶41. After 1997, a
`
`POSA would have been interested in only two conformations: 4,5-cis and 4,5-
`
`trans.
`
`3. Secondary considerations
`
`AstraZeneca bears the burden of production for secondary considerations.
`
`Prometheus Labs., Inc. v. Roxane Labs., Inc., 805 F.3d 1092, 1101-02 (Fed. Cir.
`
`2015). AstraZeneca’s evidence does not counter the strong prima facie case.
`
`a. Unpredictability and unexpected results
`
`AstraZeneca contends Compound 25 was one of many failures in the art.
`
`POR 58. Yet Ashworth recognized Compound 25 as a potent, stable and non-toxic
`
`DPP-4 inhibitor in vitro. EX1007, 1165-66. Rather than consider Compound 25 a
`
`failure, Ashworth considered four compounds (including Compound 25)
`
`sufficiently successful to turn to proline-ring optimization. The only support for
`
`“failure” is testimony from Dr. Weber, who testified everything not FDA-approved
`
`-18-
`
`
`
`

`
`for any reason is a failure, expressly including all but one of the many compounds
`
`claimed in the involved patent. EX2056, ¶251; EX1073, passim, esp. 19:12-34:18
`
`(claim 8), 34:19-37:20 (claim 10), 99:9-102:20 (Dr. Weber’s omarigliptin,
`
`approved for use in Japan, but not in the U.S.). Indeed, the involved patent
`
`disclosure does not provide any basis for believing saxagliptin or any of the other
`
`claimed compounds were any more promising than the Ashworth I compounds.
`
`Moreover, AstraZeneca provides no evidence that a POSA only pursued
`
`leads that were already FDA-approved or in clinical trials, particularly in a field
`
`with no established winners. AstraZeneca relies on cases involving actual failures
`
`to obtain FDA approval and attempts to improve existing drugs. POR 59-60.
`
`AstraZeneca shows no nexus between any prior art FDA failure and unexpected
`
`results. Specifically, AstraZeneca fails to show that the closest prior art
`
`(Compound 25) failed or would have failed to obtain FDA approval for clinical
`
`reasons. Dr. Weber conceded that a gliptin investigator may forgo FDA approval
`
`for non-clinical reasons. EX1073, 99:9-102:20.
`
`Far from being unexpected, saxagliptin reflects steady progress from
`
`Mentlein’s 1993 preferred DPP-4 dipeptide substrate (P1 proline and P2 with a
`
`bulky side group and a free amine), through Ashworth I’s substitution of a 2-cyano
`
`functional group and refinement of the bulky side group with further plans to
`
`optimize the proline. Saxagliptin’s differences from Compound 25 reflect further
`
`-19-
`
`
`
`

`
`optimizations of the same groups Ashworth was optimizing, with a reasonable
`
`expectation of success for the same reasons Ashworth expected to succeed.
`
`EX1074, ¶46.
`
`These further modifications came with a reasonable expectation of success.
`
`The Villhauer WO publication suggested that adamantyl or 1-
`
`hydroxymethylcyclohexyl were more preferred compared to cyclohexyl at the P2
`
`position and the Villhauer patent preferred 3-hydroxyadamantyl in the P2 position.
`
`EX1008, 5; EX2013, 2:10-16. Similarly, Ashworth promised further proline
`
`optimization and delivered sulfur substitution at the 4-position. EX2001, 2746.
`
`Ashworth’s sulfur substitution was consistent with the art’s knowledge that a
`
`saturated 5-member ring with hydrogen at the 3-position was preferred. EX2151,
`
`303. Hanessian provided an alternate approach to modifying the 4-position on a
`
`proline ring while preserving the saturated 5-member ring with a free 3-position.
`
`Choosing between 4,5-cis and 4,5-trans conformations would have been routine
`
`optimization. None of these steps required a departure from what the art expected
`
`would work. EX1074, ¶46. Expected beneficial results confirm obviousness. In re
`
`Skoll, 523 F.2d 1392, 1397 (CCPA 1975).
`
`AstraZeneca claims unexpectedly superior properties over other prescribed
`
`gliptins. POR 60-61. AstraZeneca fails to identify any claim limitation or support
`
`in the specification that might correlate with these properties. Moreover,
`
`-20-
`
`
`
`

`
`unexpected results must be differences of kind, not simply differences of degree. In
`
`re Aller, 220 F.2d 454, 456-57 (CCPA 1955) (finding no unexpected results where
`
`improved yields reflected a difference in degree, not in kind). To be clinically
`
`effective, a gliptin must be sufficiently stable and potent. Although AstraZeneca
`
`reports superior binding and stability, it has not demonstrated that the others lack
`
`sufficient binding or stability to be clinically effective. In fact, AstraZeneca’s
`
`physician-expert and economist-expert testified that the FDA-approved DPP-4
`
`inhibitors are essentially interchangeable. EX1028, 402:2-4 (relatively
`
`interchangeable); 399:4-7 (“no meaningful difference in the side effect profile”),
`
`398:22-25 (does not prescribe saxagliptin more than other DPP-4 inhibitors),
`
`398:4-21 (alleged unexpected results would also “apply to the other members of
`
`the DPP-4 class that have been FDA-approved”); EX1029, 445:16-23 (no more
`
`efficacious), 446:14-19 (no suggestion of better safety profile).
`
`AstraZeneca’s comparison with Ashworth compounds other than
`
`Compound 25 is inapposite: comparison with more distant prior art alternatives is
`
`entitled to less weight. Syntex (U.S.A.) LLC v. Apotex, Inc., 407 F.3d 1371, 1382-
`
`83 (Fed. Cir. 2005). Even assuming saxagliptin had some superior properties,
`
`unexpected properties do not negate reasons to modify based on expected
`
`properties. Bristol-Myers Squibb Co. v. Teva Pharms. USA, Inc., 752 F.3d 967, 976
`
`(Fed. Cir. 2014), citing Dillon, 919 F.2d at 693, 697.
`
`-21-
`
`
`
`

`
`Finally, AstraZeneca touts saxagliptin’s safety and efficacy compared to the
`
`earlier-invented vildagliptin, noting its twice-daily dosing and liver-toxicity
`
`warning. POR 64-65. Vildagliptin has a once-daily combination form. EX1041,
`
`¶44. Also, any (unclaimed) dosing differen