`______________________
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`______________________
`APOTEX INC., APOTEX CORP., ARGENTUM PHARMACEUTICALS LLC,
`ACTAVIS ELIZABETH LLC, TEVA PHARMACEUTICALS USA, INC., SUN
`PHARMACEUTICAL INDUSTRIES, LTD., SUN PHARMACEUTICAL
`INDUSTRIES, INC., AND SUN PHARMA GLOBAL FZE,
`Petitioners,
`V.
`NOVARTIS AG,
`Patent Owner.
`______________________
`Case IPR2017-008541
`U.S. Patent No. 9,187,405
`______________________
`FOURTH DECLARATION OF WILLIAM J. JUSKO, PH.D.
`
`
`Mail Stop Patent Board
`Patent Trial and Appeal Board
`U.S. Patent and Trademark Office
`P.O. Box 1450
`ALEXANDRIA, VA 22313-1450
`
`
`
`
` 1 Cases IPR2017-01550, IPR2017-01946, and IPR2017-01929 have been joined
`with this proceeding.
`
`
`
`
`
`Apotex v. Novartis
`IPR2017-00854
`NOVARTIS 2095
`
`
`
`I, William J. Jusko, Ph.D., declare as follows:
`
`
`
`Introduction
`I am the same William J. Jusko who submitted two prior declarations
`1.
`
`in this matter, Exhibits 2005 and 2024 plus a Third Declaration (Ex. 2076) served
`
`on December 5, 2017 as supplemental evidence. I submit this Fourth Declaration in
`
`support of Novartis’s sur-reply, and in particular to address certain opinions by Dr.
`
`Leslie Z. Benet (Ex. 1047). I use the same terms and abbreviations here that I used
`
`in my prior declarations.
`
`2.
`
`In my Second Declaration (Ex. 2024), I showed that a pharmacologist
`
`would have thought the invention claimed in the ’405 Patent unlikely to work based
`
`on information available in the art in June 2006. The Patent claims a method of
`
`treating various aspects of RRMS using a 0.5 mg daily dose of fingolimod. But
`
`available data in June 2006 suggested that only doses 1.0 mg or higher would work.
`
`3.
`
`Dr. Benet argues otherwise, relying in part on an “EAE” animal study—
`
`the “Kataoka” reference (Ex. 1029)—that he says would have led a pharmacologist
`
`to expect 0.5 mg daily to be effective in humans. Kataoka reports that 0.1 mg/kg
`
`was the lowest tested dose to have a therapeutic effect on EAE in mice and rats. Dr.
`
`Benet uses an FDA Guidance on how to extrapolate from animal to first-in-human
`
`doses to argue that 0.1 mg/kg in mice converts to approximately 0.5 mg in humans.
`
`
`
`
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`(Ex. 1047 ¶¶ 67-70.) Dr. Benet says that this analysis would have led a person of
`
`skill to expect 0.5 mg to be effective in humans.
`
`4.
`
`Dr. Benet’s analysis has two major flaws.
`
`• First, a person of skill in June 2006 would not have considered
`
`extrapolating from animal to human doses because extensive PK/PD
`
`data already existed in humans. The FDA Guidance is expressly
`
`designed only to identify a safe first-in-human dose before such data
`
`exists. But once human PK/PD data exists, that data would provide
`
`far more relevant information for estimating a dose’s effects than an
`
`estimate based on simple animal dose data. Accordingly, a person
`
`of skill would not have used the FDA Guidance to extrapolate a
`
`human dose from Kataoka’s lowest effective mouse dose.
`
`• Second, even if a person of skill in June 2006 would have considered
`
`extrapolating from animal to human doses, no pharmacologist
`
`would have limited the analysis to extrapolating from one animal
`
`using only one method. A pharmacologist would have made use of
`
`the other available human and animal data to scale doses. That more
`
`complete analysis would have pointed definitively toward doses of
`
`1.0 mg or higher—a range in line with what other prior art
`
`suggested.
`
`2
`
`
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`
` Analysis
`A. Animal Dosing Would Have Been Irrelevant
`Given the Existing Human PK/PD Data
`In its opening sentence, the FDA Guidance on which Dr. Benet relies
`
`5.
`
`describes its purpose: to “outline[] a process (algorithm) and vocabulary for deriving
`
`the maximum recommended starting dose (MSRD) for first-in-human clinical trials
`
`of new molecular entities in adult healthy volunteers, and recommends a
`
`standardized process by which the MSRD can be selected.” (Ex. 1049 at 1 (emphasis
`
`in original).) As the Guidance states in the next sentence, “[t]he purpose of this
`
`process is to ensure the safety of the human volunteers.” (Id.)
`
`6.
`
`In other words, the Guidance describes one process by which drug
`
`sponsors can use pre-clinical animal data to identify a dose to test in humans for the
`
`first time, before any human PK/PD data is available. The Guidance’s stated goal is
`
`safety for the first-in-human volunteers, not efficacy for any particular condition.
`
`That is why the Guidance suggests starting with “the highest dose level that does not
`
`produce a significant increase in adverse effects.” (Id. at 5.) That dose is then
`
`extrapolated to a human equivalent dose using an algorithm designed to be
`
`“conservative,” as I describe further below. The result is then reduced by a safety
`
`factor to further limit the risk to the first human volunteers.
`
`7.
`
`By June 2006, however, fingolimod had already been tested in humans
`
`in multiple trials for many years. Those trials had generated copious human PK/PD
`3
`
`
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`data, as reflected in Budde 2002 (Ex. 1008), Kahan 2003 (Ex. 1031), Park 2003 (Ex.
`
`2048), and Park 2005 (Ex. 1019). Given that fingolimod’s human safety had already
`
`been established, the FDA Guidance would no longer have been relevant or
`
`appropriate in June 2006.
`
`8.
`
`The focus then would have been on identifying effective doses for
`
`specific conditions.
`
` As
`
`the FDA Guidance states,
`
`identifying such
`
`“pharmacologically active” doses (PAD) “depends on many factors and differs
`
`markedly among pharmacological drug classes and clinical indications; therefore,
`
`selection of a PAD is beyond the scope of this guidance.” (Ex. 1049 at 12 (emphasis
`
`added).) The FDA Guidance observes that pharmacologically active doses could
`
`instead be “derived from appropriate pharmacodynamic models.” (Id.)
`
`9.
`
`That is exactly the analysis I performed in my Second Declaration.
`
`Webb identified a pharmacodynamic efficacy marker (a minimum of 70%
`
`lymphocyte suppression for “any efficacy”), and Kahan 2003 and Park 2005
`
`reported on that marker in humans at various doses (with 0.5 mg daily generally
`
`achieving less than 50% average suppression). As I pointed out in my Second
`
`Declaration (Ex. 2024 ¶ 75), pharmacologists assume that PD markers like
`
`lymphocyte suppression apply across species, absent evidence to the contrary. Dr.
`
`Benet does not disagree, or identify any evidence that a person of skill in June 2006
`
`would have thought the marker would not also apply to humans too. Just the reverse:
`
`
`
`4
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`Dr. Benet adopts the marker to argue that Webb, Kahan 2003, and Park 2005 would
`
`have pointed a person of skill toward 0.5 mg daily, if those references were to be
`
`read as he argues. (Ex. 1047 ¶¶ 38-62.)
`
`10. While I disagree with Dr. Benet’s reading of those references,2 his
`
`methodology in using them in this way is sound. Kataoka is not to the contrary.
`
`Kataoka says nothing about the degree of lymphocyte suppression needed for
`
`efficacy in mice or rats. But it does confirm that lymphocyte suppression and
`
`
` 2 Unlike Dr. Benet, I take Webb’s statement that “a threshold” of 70% suppression
`
`was needed for “any efficacy” at face value. Webb’s authors had ample data to
`
`reach that conclusion. And I further believe that, given RRMS’s chronic, life-
`
`long nature, a dose’s potential efficacy against the Webb’s benchmark would
`
`have been measured against the average human suppression levels in Kahan 2003
`
`and Park 2005, not the maximum suppression levels on which Dr. Benet focuses.
`
`Dr. Benet criticizes my use of average suppression numbers in Kahan 2003 rather
`
`than the maximum suppression over 28 days of about 60% as unsupported, but
`
`Dr. Benet is incorrect. A person of skill in June 2006 would have read Kahan
`
`2003 in light of Park 2005, which showed that suppression at 0.5 mg daily was
`
`highly variable after steady state concentrations were reached, but on average
`
`about 42%. (Park 2005 at 689.)
`
`
`
`5
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`efficacy are linked (Ex. 1029 at 446)—the same premise underlying Webb’s
`
`assessment that 70% suppression is needed for “any efficacy.”
`
`11.
`
`Indeed, Kataoka shows that the effective 0.1 mg/kg dose suppressed
`
`mouse lymphocytes to extremely low levels, in a manner entirely consistent with
`
`Webb’s finding. Figure 3 in Kataoka shows that 0.1 mg/kg of fingolimod or its
`
`phosphorylated form (FTY-P) suppressed T-cell lymphocytes from a baseline of
`
`over 3,100 cells per microliter to about 170 per microliter, and B-cells from over
`
`790 cells per microliter to about 240. (I calculated these numbers using Graph
`
`Digitizer version 1.9 created by N. Rodionov for the purpose of allowing more
`
`precise reading of graphs like Figure 3.)
`
`12. Taken together, that is a drop from a total of about 3,900 lymphocytes
`
`per microliter to about 450 per microliter, a more than 85% reduction. So the lowest
`
`effective dose in Kataoka was entirely consistent with Webb’s finding that efficacy
`
`depended on 70% or greater lymphocyte suppression.
`
`13. A pharmacologist assessing the possible efficacy of different doses
`
`would strongly prefer the pharmacodynamic information in Webb , Kahan 2003, and
`
`Park 2005 to raw animal dose extrapolations. Animal dose extrapolations are
`
`notoriously
`
`inconsistent and unreliable.
`
` Animal and human physiology,
`
`metabolism, and sensitivity to drugs can vary widely.
`
`
`
`6
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`14. Dr. Benet himself wrote a retrospective article on the history of animal
`
`scaling in 2011. (Ex. 2103.) He reviewed various scaling methodologies and found
`
`that “although some prediction methods are better than others, there currently is no
`
`universally outstanding method.” (Id. at 4048.) Moreover, “the prediction of human
`
`disposition kinetics following intravenous administration is much better than
`
`prediction of events following oral administration.” (Id.) Dr. Benet observed that
`
`“even the best methods could only predict events after oral administration, within a
`
`factor of two, in the order of 45% of the time, with a tendency to under predict the
`
`area under the curve.” (Id.) Dr. Benet observed that “[t]here are likely to be many
`
`reasons for this failure, including the complex physiology of the gastrointestinal tract
`
`coupled with the complex processes occurring during absorption[.]” (Id.)
`
`15. Fingolimod of course is administered orally, and thus is exactly the sort
`
`of drug that Dr. Benet himself contends is ill-suited for animal scaling. While some
`
`animal experiments tested fingolimod intravenously, the form used in all clinical
`
`trials was oral, and thus less likely to be amenable to effective scaling.
`
`16. The FDA Guidance itself cautions that many factors can confound
`
`animal to human dose scaling. Several of those factors were known in June 2006 to
`
`exist for fingolimod:
`
`• The FDA Guidance identifies “variable bioavailability” between
`
`species as a potential confounding factor in scaling from animal to
`
`
`
`7
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`human doses. (Ex. 1049 at 11.) For fingolimod, papers reported
`
`precisely such variation between species as of June 2006. Budde 2002
`
`reported that fingolimod “bioavilability is 60% to 90% in rats and
`
`dogs.” (Ex. 1008 at 1073.) Another reference states that fingolimod’s
`
`“oral bioavailability . . . exceeds 60% in dogs, 80% in rats, and 40% in
`
`subhuman primates.” (Troncoso 1998, Ex. 2101 at 372.) In other
`
`words, fingolimod was twice as bioavailable in rats as in subhuman
`
`primates. If all other factors were equal, primates would thus need an
`
`oral dose twice as large as rats to receive the same amount of drug.
`
`• The FDA Guidance identifies variation between different species’
`
`absorption, distribution, metabolism, and excretion (ADME) as another
`
`potential confounding factor in scaling from animal to human doses.
`
`(Ex. 1049 at 9.) For fingolimod, papers reported exactly such variation
`
`between species as of June 2006. For instance, Budde 2002 reported
`
`that fingolimod’s “half-life”—a measure of the amount of time the drug
`
`remains in the body—“in animals varies: 12 h in rats, 29 h in dogs and
`
`36 h in baboons.” (Ex. 1008 at 1073.) In humans, Budde 2002 reported
`
`fingolimod “has a very long half-life . . . , approximately 4 to 5 d.” (Id.
`
`at 1081; see also Ex. 2101 at 372 (measuring half-life of “12 h and 29
`
`h, respectively” in rats and dogs).) Different half-lives mean that
`
`
`
`8
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`different species would be exposed to different concentrations of the
`
`drug when all other factors are held constant.
`
`• The FDA Guidance identifies the presence of a steep dose response
`
`curve as another potential confounding factor in scaling from animal to
`
`human doses. (Ex. 1049 at 10.) For fingolimod, Webb reported just
`
`such a dose response: “In dose response experiments, a threshold of
`
`about 70% depletion of peripheral lymphocytes was required to see any
`
`efficacy, and thereafter, the dose-response relationship between clinical
`
`benefit and lymphopenia was very steep.” (Ex. 2014 at 118.)
`
`17. For these sorts of reasons, studies have shown that simple scaling under
`
`the Guidance can result in substantial errors, in many cases erring by a factor of more
`
`than 200%. (See, e.g., Ex. 2106 at 1299.)
`
`18. A pharmacologist in June 2006 accordingly would not have even
`
`considered the FDA Guidance (or any other animal to human dose extrapolation
`
`method) as a basis for identifying effective doses for RRMS. A pharmacologist
`
`would have focused on the known PD marker of lymphocyte suppression as a means
`
`of assessing the likelihood that different doses would work. Dr. Benet’s analysis of
`
`this issue is thus inapt, as are the arguments he bases on this analysis.
`
`
`
`
`
`9
`
`
`
`
`
`
`
`19.
`
`B.
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`If Used, Animal Dose Extrapolation Would Have Pointed
`to Human Doses of 1.0 mg or Higher—Just Like Webb,
`Kahan 2003, and Park 2005
`If a pharmacologist nonetheless were to consider scaling from Kataoka,
`
`the existing human and other animal PK/PD specific to fingolimod would not have
`
`been ignored. That data would have allowed a pharmacologist to develop more
`
`accurate and realistic scaling models than the rough assumptions used in the FDA
`
`Guidance—though even using those rough assumptions, Kataoka would point a
`
`pharmacologist to a human dose of 1.0 mg or higher.
`
`20. By way of background, it was well-known in June 2006 that most
`
`animal doses do not extrapolate to humans in a linear fashion based on weight. So
`
`a 0.1 mg/kg dose in a mouse would not usually be scaled to humans by simply
`
`multiplying that dose by a human’s weight (0.1 x 70 kg, which would equal a 7 mg
`
`dose in a 70 kg human). Linear extrapolation like this would usually produce a dose
`
`that is far too high.
`
`21. Differences in weight do not account for the fact that body metabolism
`
`(which determines how the animal processes the drug) does not remain constant
`
`between species. Smaller animals tend to have faster metabolisms than larger
`
`animals. Extrapolating animal to human doses must account for this fact.
`
`
`
`
`
`
`
`10
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`Scaling Based on Human and Rat PK Data
`i.
`22. The most realistic and accurate scaling would make use of the copious
`
`human and rat PK data that existed for fingolimod in June 2006. A pharmacologist
`
`would use that data to scale up from Kataoka’s 0.1 mg/kg effective rodent dose. In
`
`particular, a pharmacologist would utilize “clearance” data gathered from humans
`
`and animals to scale doses.
`
`23.
`
`“Clearance” measures the body’s rate of eliminating a drug from its
`
`system, and thus is a useful proxy for metabolism. It is calculated by dividing the
`
`dose amount by the area under the blood drug concentration-time curve (AUC),
`
`which represents the actual exposure of the body to the drug. With adequate
`
`clearance information on different species, clearance rates can be used to scale doses
`
`from one species to the next. Such data existed for fingolimod in June 2006.
`
`24. Kahan 2003 (Ex. 1031) reported on human clearance rates for different
`
`fingolimod doses administered orally over 28 days in Table 2. While the rates vary
`
`by dose and contain one extreme outlier at a lower dose (0.25 mg), they largely
`
`coalesced around a clearance of 10 liters per hour, which is 0.133 liters/hour/kg for
`
`a 75 kg person. (As I explain below (at ¶ 32), women are the primary victims of
`
`MS, and the average weight of an American woman in 2006 was about 75 kg
`
`according to the U.S. Centers for Disease Control.)
`
`
`
`11
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`25. Next, my article with Meno-Tetang (Ex. 2055) reported a rat clearance
`
`rate of 0.497 liters/hour/kg for intravenous doses. (Ex. 2055 at 1487, Table 4.)
`
`Using the oral bioavailability of fingolimod in rats reported in the Meno-Tetang
`
`article (71%), this i.v. clearance can be converted to an oral clearance rate of 0.7
`
`liters/hour/kg.3 The human oral clearance of 0.133 liters/hour/kg divided by the rat
`
`oral clearance of 0.7 liters/hour/kg yields a “Conversion Factor” (as that term is used
`
`in the FDA Guidance as a ratio of clearances) of 0.19. That conversion factor could
`
`then be applied to the lowest effective rat dose in Kataoka using the same formulaic
`
`steps as Dr. Benet to produce a human equivalent dose of about 1.4 mg (0.1 mg/kg
`
`from Kataoka x 75 kg human weight x 0.19 Conversion Factor = 1.43 mg).
`
`26. To summarize, if a pharmacologist were inclined to try to scale a human
`
`dose from the Kataoka animal data in June 2006, he would turn to actual clearance
`
`data in the relevant species. He would find the actual human and rat clearance data
`
`that was available at the time and would use it for scaling. As no mouse clearance
`
`
` 3 The oral clearance (CL/F) is considered to be Dose/AUC where the oral dose is
`
`divided by the exposure (AUC): CL/F = Dose/AUC. The intravenous clearance
`
`(CL), in which bioavailability is 100% and F is equal to 1, can therefore be
`
`converted to oral clearance by dividing by the oral bioavailability rate: 0.497
`
`liters/hour/kg ÷ 0.71 = 0.7 liters/hours/kg.
`
`
`
`12
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`data was available, he couldn’t scale from mouse. Scaling from rat to human would
`
`have pointed toward scaled doses for humans of 1.0 mg or higher. That is the same
`
`range that the pharmacodynamics analysis in my Second Declaration identifies as
`
`following from the Webb, Kahan 2003, and Park 2005 papers.
`
`The FDA Guidance Scaling Method
`ii.
`If a pharmacologist were inclined to ignore the existing human and rat
`
`27.
`
`data when scaling a dose from Kataoka as Dr. Benet does—and no pharmacologist
`
`would truly do so—then even under the FDA Guidance as a sole point of reference,
`
`a pharmacologist would be led to doses of 1.0 mg or higher in humans.
`
`28. The FDA Guidance employs a scaling method to be used before any
`
`human data exists and that is designed for any drug, not for fingolimod specifically.
`
`The Guidance accordingly uses average rules of thumb based on experience with
`
`many drugs to scale doses. In doing so, the Guidance, in an abundance of concern
`
`for human safety, errs toward lower doses.
`
`29. The Guidance’s scaling method is based on body surface area. Like
`
`metabolism, body surface area decreases relative to body weight as an animal gets
`
`bigger. Studies have shown a rough correlation between proportional changes in
`
`body surface area and proportional changes in metabolism. (See Ex. 1049 at 6.)
`
`(The reason for this relationship is not proven, but many think that body surface area
`
`and metabolism are linked in that both regulate the body’s disposition of heat.) This
`
`
`
`13
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`relationship between body surface area and metabolism is reflected in the “b”
`
`variable in the scaling equation set out in Appendix A of the FDA Guidance:
`
`“[Human Equivalent Dose] = animal [No Observed Adverse Effect Level] x
`
`([Weight]animal/[Weight]human)(1-b).” (Ex. 1049 at 16.)
`
`30. Under the Guidance, the No Observed Adverse Effect Level is the
`
`highest animal dose that showed no toxicity, and thus the dose being scaled to a
`
`“Human Equivalent Dose” or “HED” under the Guidance (which is focused on
`
`toxicity). As Appendix A of the Guidance describes, the “conventional[]” value for
`
`normalizing body weight and body surface area (“mg/m2”)—the “b” exponent in the
`
`equation—“would be 0.67[.]” (Id.) That is the exponent used to derive the scaling
`
`factor Dr. Benet used to scale Kataoka’s 0.1 mg/kg dose in mice to humans. As
`
`shown in Table 2 in Appendix A, that exponent yields a mouse conversion factor of
`
`0.075, which the Guidance simplifies to a “standard” of 0.081.
`
`31. Dr. Benet uses that factor in Paragraph 69 of his Declaration to scale
`
`from mouse to human: he multiplied the 0.1 mg/kg dose in Kataoka times the 0.08
`
`standard mouse scaling factor to derive a dose of 0.008 mg/kg dose for humans. For
`
`a human weight range of 50-80 kg, that yields a scaled dose of 0.4-0.64. (Dr. Benet’s
`
`human weight range artificially skews the dose lower; women are the predominant
`
`victims of MS, and the average female weight in the U.S. in 2006 was about 75 kg
`
`
`
`14
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`according to the U.S. Centers for Disease Control (Ex. 2104 at 7, Table 3), yielding
`
`a dose of 0.6 mg using Dr. Benet’s methodology.)
`
`32. Even under the FDA Guidance—and not accounting for other scaling
`
`methods—Dr. Benet’s analysis is highly selective. He does not address variants
`
`under the FDA Guidance that would extrapolate from Kataoka to higher human
`
`doses. Insofar as a pharmacologist sought to identify potential effective doses using
`
`the Guidance—contrary to the Guidance’s focus on safety, not efficacy—then the
`
`pharmacologist would weigh all of these results to assess the likelihood that
`
`Kataoka’s doses might be effective. The weight of those results points toward doses
`
`higher than 1.0 mg.
`
`33. First, a pharmacologist would analyze not only Kataoka’s mouse data,
`
`but also the paper’s rat data showing a lowest effective dose of 0.1 mg/kg. (Ex. 1029
`
`at 441.) Dr. Benet did not analyze that data in the part of his report arguing that
`
`Kataoka points toward the 0.5 mg daily dose. But Dr. Benet acknowledges
`
`elsewhere in his report (at ¶ 77) that the rat dose in Kataoka yields a human dose
`
`over 1.0 mg.
`
`34.
`
`In other words, scaling the rat dose to humans using the exact same
`
`methodology Dr. Benet uses for the mouse dose produces a human equivalent dose
`
`more than double in size. That is because the standard rat conversion factor is
`
`roughly double the standard mouse conversion factor. (See Ex. 1049, Appx. A Table
`
`
`
`15
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`2 (standard mouse scaling factor of 0.081 compared to standard rat scaling factor of
`
`0.162).) Dr. Benet never explains why he did not weigh the rat calculation in his
`
`scaling argument based on Kataoka. I can think of no persuasive reason to exclude
`
`the rat from the analysis.
`
`35.
`
`If scaling were being done for purposes of first-in-human testing, then
`
`using the mouse and not the rat might have made sense. The mouse is a more
`
`sensitive animal in this context, and the FDA Guidance recommends using the most
`
`sensitive species as the basis for scaling when the goal is to protect the human
`
`volunteers from harm. (Ex. 1049 at 9.)
`
`36. Here, however, Dr. Benet does not scale doses for first-in-human tests
`
`where safety is paramount. He is trying to scale doses for efficacy. He provides no
`
`reason to believe that mice would provide a better window into efficacy than rats in
`
`this context. I can think of none. As a result, a pharmacologist engaged in this
`
`hypothetical exercise would likely scale to human from both animals to provide a
`
`more complete picture.
`
`37. Second, Dr. Benet’s analysis relies mainly on the “conventional”
`
`normalizing exponent to generate the mouse scaling factor (“b” of 0.67). But studies
`
`have shown that a “b” value of 0.67 often does not hold for specific drugs. For
`
`instance, Tang 2005 shows in Table 1 that, for 60 drugs, the “b” value varies widely
`
`from 0.349 up to 1.196. (Ex. 2106 at 1298.)
`
`
`
`16
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`38. Accordingly, the Guidance provides another option too. As the
`
`guidance explains, “a number of studies . . . have shown that [maximum tolerated
`
`doses] scale best across species when b = 0.75.” (Ex. 1049 at 16.) A 2003 article
`
`the Guidance cites (at 2)—the “Mahmood” reference (Ex. 2102)—observes that the
`
`choice between these exponents in scaling exercises “is almost entirely arbitrary and
`
`without scientific justification.” (Id. at 692-93.) The FDA’s choice of 0.67 was not
`
`driven by scientific factors, but instead by the safety concerns that underlie the entire
`
`Guidance—0.67 yields systematically lower doses and thus “provides a more
`
`conservative conversion.” (Ex. 1049 at 16.)
`
`39. But Dr. Benet’s dose scaling analysis is not focused on safety. It is
`
`focused on efficacy. He argues that a person of skill would have predicted 0.5 mg
`
`daily would have been effective for RRMS based on Kataoka’s EAE data. (Ex. 1047
`
`¶¶ 67-70.) In that context, there is no “scientific justification” for using the 0.67
`
`exponent as opposed to 0.75. I accordingly would expect a pharmacologist seeking
`
`to use the Guidance’s methods to scale doses using both exponents, to develop a
`
`more sensitive estimate. Using the mouse conversion factor derived from the 0.75
`
`exponent (i.e. 0.141) again produces a dose of over 1.0 mg for a 75 kg woman (as I
`
`set out below).
`
`40. Third, and finally, combining the two observations above would lead a
`
`pharmacologist employing the FDA Guidance to also calculate a human equivalent
`
`
`
`17
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`dose based on Kataoka’s rat data and based on the higher 0.75 exponent. That
`
`calculation would produce a dose of 1.84 mg for an average American 75 kg woman.
`
`41. Taken together, the Kataoka mouse and rat data would thus produce a
`
`distribution of scaled human equivalent doses as follows (I use the actual conversion
`
`factors below rather than the “standard” factors for consistency’s sake, and the
`
`average US woman’s weight in 2006 of 75 kg):
`
`Kataoka Dose Body Surface
`Area Exponent
`
`Conversion
`Factor
`
`Formula
`
`0.1 mg/kg
`Mouse
`0.1 mg/kg
`Mouse
`0.1 mg/kg
`Rat
`0.1 mg/kg
`Rat
`
`
`
`0.67
`
`0.75
`
`0.67
`
`0.75
`
`Human
`Equivalent
`Dose
`0.56 mg
`
`0.075
`
`0.1 x 0.075 x 75
`
`0.141
`
`0.1 x 0.141 x 75
`
`1.05 mg
`
`0.156
`
`0.1 x 0.156 x 75
`
`1.17 mg
`
`0.245
`
`0.1 x 0.245 x 75
`
`1.83 mg
`
`42. A pharmacologist presented with this distribution in results would not
`
`expect the 0.5 mg daily dose to be effective in humans. The lowest effective doses
`
`in Kataoka’s mouse and rat models scale most often to human doses over 1.0 mg.
`
`From this analysis, a pharmacologist would expect the likely lowest effective dose
`
`in humans to be 1.0 mg daily or higher.
`
`43. The only result pointing lower is based on a methodology the FDA
`
`specifically designed to yield lower “conservative” doses for purposes of the first
`
`
`
`18
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`tests in humans. (Exhibit 1049 at 16.) As the Guidance points out, there is no
`
`expectation that this dose is pharmacologically active. (Id. at 12.) Instead it would
`
`be used to defined the maximum recommended start dose (MRSD), i.e., the starting
`
`point to test safety. If safe, the dose would later be gradually increased in humans
`
`to define a dose window in which testing for efficacy can be conducted. Here, the
`
`safety of human doses had already been established for fingolimod. (See Ex. 1008.)
`
`When determining a therapeutic dose for treating RRMS, a person of skill would
`
`instead take the opposite approach and err on the side of a higher dose. RRMS’s
`
`unpredictable, relapsing nature makes it impossible to determine within an
`
`individual patient whether a dose is having the desired therapeutic effect. To
`
`increase the chances that it is, a person of skill would want to err in favor of a larger
`
`dose, assuming it was safe. All of the doses above had been shown to be safe enough
`
`for human use already. Accordingly, a pharmacologist considering the FDA
`
`Guidance methods would not put much weight in the lower potential dose, and
`
`instead would weigh the other variations more heavily in predicting efficacy.
`
`44. Many of Dr. Benet’s arguments depend on this misguided scaling under
`
`the FDA Guidance, including at least the following from Dr. Benet’s declaration:
`
`That the data in Webb and Kataoka demonstrate efficacy of a 0.5 mg dose
`
`(Ex. 1047 ¶¶67-70);
`
`
`
`19
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`• That the Kappos 2005 results confirmed those of Kataoka rather than
`
`showing a surprising equivalency in effectiveness of the 1.25 mg and 5 mg
`
`doses (id. ¶¶ 71-78);
`
`• That the 0.5 mg dose would have been expected to be effective prior to
`
`Phase III clinical trials (id. ¶¶ 90-94);
`
`• That the clinical equivalency between the 0.5 mg and 1.25 mg dose was
`
`not unexpected (id. ¶¶ 95-99); and
`
`• That Dr. Steinman incorrectly scaled the 1.25 mg dose based on internal
`
`Novartis studies (id. ¶¶102-106).
`
`*
`
`
`
`
`
`
`
`
`
`*
`
`
`
`
`
`
`
`
`
`*
`
`45. Under penalty of perjury, all statements made herein of my own
`
`knowledge are true, and I believe all statements made herein on information and
`
`belief to be true. I have been warned and am aware that willful false statements and
`
`the like are punishable by fine or imprisonment or both under Section 1001 of Title
`
`18 of the United States Code.
`
`46.
`
`In signing this Declaration, I understand that it will be filed as evidence
`
`in a contested case before the Patent Trial and Appeal Board of the United States
`
`Patent and Trademark Office. I acknowledge that I may be subject to cross-
`
`examination in the case and that cross-examination will take place in the United
`
`
`
`20
`
`
`
`
`
`
`
`IPR2017-00854
`U.S. Patent No. 9,187,405
`States. If cross-examination is required of me, I will appear for cross-examination
`
`within the United States during the time allotted.
`
`
`
`
`
`21
`
`
`
`DATED: March 23, 2018
`
`1PR2017~00854
`
`US. Patent No. 9,187,405
`
`
`
` ‘
`William I JuskoP.
`
`22
`
`22
`
`