`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`
`
`
`MYLAN PHARMACEUTICALS INC., ACTAVIS
`LABORATORIES FL, INC., AMNEAL PHARMACEUTICALS LLC,
`AMNEAL PHARMACEUTICALS OF NEW YORK, LLC, DR. REDDY'S
`LABORATORIES, INC., DR. REDDY'S LABORATORIES, LTD.,
`SUN PHARMACEUTICALS INDUSTRIES, LTD.,
`SUN PHARMACEUTICALS INDUSTRIES, INC.,
`TEVA PHARMACEUTICALS USA, INC., WEST-WARD
`PHARMACEUTICAL CORP., and HIKMA PHARMACEUTICALS, LLC,
`Petitioner
`
`v.
`
`JANSSEN ONCOLOGY, INC.,
`
`Patent Owner
`
`Case IPR2016-013321
`Patent 8,822,438 B2
`
`
`
`
`
`REPLY DECLARATION OF IAN MCKEAGUE, Ph.D.
`IN SUPPORT OF PETITION FOR INTER PARTES REVIEW OF U.S.
`PATENT NO. 8,822,438
`
`
`
`
`1 Case IPR2017-00853 has been joined with this proceeding.
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 1
`
`

`

`TABLE OF CONTENTS
`
`
`Page
`
`
`INTRODUCTION ........................................................................................ 3
`I.
`II. EDUCATION AND PROFESSIONAL BACKGROUND ....................... 3
`III. MATERIALS CONSIDERED .................................................................... 6
`IV. LEGAL STANDARDS ................................................................................. 6
`V.
`SUMMARY OF OPINIONS........................................................................ 8
`VI. BACKGROUND ........................................................................................... 9
`A.
`Prostate Cancer And Abiraterone Acetate ........................................... 9
`B.
`Cougar’s Clinical Trials Investigating Abiraterone Acetate .............. 10
`C.
`Statistical Principles ........................................................................... 12
`VII. JANSSEN HAS NOT DEMONSTRATED UNEXPECTED
`RESULTS .................................................................................................... 13
`A. Dr. Rettig’s Median TTPP Data Point Comparisons Do Not
`Demonstrate Any Unexpected Clinical Efficacy Of The
`Abiraterone Acetate And Prednisone Combination ........................... 15
`1.
`The Confidence Intervals In Dr. Rettig’s Comparison
`Table Overlap ........................................................................... 15
`Other Data Reported By Janssen And Its Predecessor
`Conflict With The Data Reported In Dr. Rettig’s
`Declaration ............................................................................... 21
`Dr. Rettig’s Analysis Engages in an Improper Cross-
`Study Comparison .................................................................... 24
`The Phase III Studies Are Not Designed To Ascertain
`Any Effect Of Prednisone ........................................................ 29
`Janssen’s Selective Disclosure Of Four Instances Of Supposed
`“Reversal Of Resistance” Does Not Demonstrate Unexpected
`Clinical Efficacy ................................................................................. 30
`
`B.
`
`
`
`
`
`2
`
`
`
`2.
`
`3.
`
`4.
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 2
`
`

`

`
`
`I, Ian McKeague, Ph.D., do hereby declare:
`I.
`
`INTRODUCTION
`
`1.
`
`I am making this declaration at the request of Petitioners, in the matter
`
`of the Inter Partes Review (“IPR”) of U.S. Patent No. 8,822,438 (the “’438
`
`patent”), as set forth in the above caption.
`
`II. EDUCATION AND PROFESSIONAL BACKGROUND
`
`2.
`
`I received a B.A./M.A. and M.Math in Mathematics from the
`
`University of Cambridge in 1975 and 1976, respectively. I received a Ph.D. from
`
`the University of North Carolina at Chapel Hill in 1980, authoring a Ph.D. thesis
`
`titled Covariance Operators and Their Applications in Probability and Information
`
`Theory. My curriculum vitae is attached as Exhibit A.
`
`3.
`
`After completing my Ph.D., I took a position as an Assistant Professor
`
`in the Department of Statistics at Florida State University. I held that position
`
`from 1980 to 1986, when I was promoted to Associate Professor, a title I held until
`
`1991, when I was promoted to Professor. In 1996, I was promoted to Chairman of
`
`the Department of Statistics and I held that position until 1999. From 2000 to
`
`2004, I was the Ralph A. Bradley Professor of Statistics at Florida State
`
`University. In 2004, I left that university and took a position as Professor of
`
`Biostatistics at Columbia University, a position I hold to this day. In addition, I
`
`have been a visiting professor at University of Padua, Italy (1985), University
`
`
`
`3
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 3
`
`

`

`
`
`Joseph Fourier, Grenoble, France (1992, 2001), and the University of California,
`
`Berkeley (1991).
`
`4.
`
`I have received numerous professional honors, served as an editor on
`
`various journals, and have served on a number of professional committees. For
`
`example, I am a Fellow of the Institute of Mathematical Studies and a Fellow of
`
`the American Statistical Association. I am also an Associate Editor of the Journal
`
`of the American Statistical Association (1993–96, 2011–present), the International
`
`Journal of Biostatistics (2005–present), and Statistical Inference for Stochastic
`
`Processes (1998–present), and a former Associate Editor of the Annals of Statistics
`
`(1989–1995) and ESAIM: Probability and Statistics (2000–2005). At Florida State
`
`University, I received the Named Professorship Award (2000), the Graduate
`
`Teaching Award (1998), and the Professorial Excellence Program Award (1999). I
`
`have served on the Institute of Mathematical Statistics Fellows Committee (2008–
`
`2010); the ASA Section on Nonparametric Statistics Awards Committee (2010–
`
`2011); the NSF Statistics and Probability Program Panel (1997, 1999, 2000, 2002,
`
`2003, 2007, 2009, 2010, 2016), Special Meetings Panel (2005), Biocomplexity in
`
`the Environment Panel (2004), and Knowledge and Distributed Intelligence
`
`Program Panel (1998).
`
`5.
`
`As a biostatistician, I have more than twenty years of experience in
`
`clinical trial design and statistical analysis. I have consulted for a variety of
`
` 4
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 4
`
`

`

`
`
`entities regarding clinical trial design and analysis, including biopharmaceutical
`
`companies and nonprofit research centers. In particular, I have been involved in
`
`the design and analysis of numerous clinical trials, with special expertise in
`
`designing cohort studies. I have published dozens of papers analyzing clinical trial
`
`data, often including analysis regarding study methodology. I have also spoken
`
`widely at conferences, symposia, and seminars on clinical trials. For example, I
`
`was the Keynote Speaker at the Fourth Annual International Symposium on the
`
`Evaluation of Clinical Trial Methodologies and Applications in Beijing, China in
`
`2011, and I also was an invited speaker on clinical trials at the ICSA Applied
`
`Statistics Symposium in 2011.
`
`6. My research interests include functional data analysis, empirical
`
`likelihood, and non-standard asymptotics, to name a few. I am a named author on
`
`111 peer-reviewed articles, I have supervised eighteen doctoral students, and I
`
`have received grants from a number of organizations including the NIH and the
`
`NSF.
`
`7.
`
`In the past four years, I have testified in the following cases:
`
`• Casas v. Consolidated Edison Co. of New York, Inc., Index No.
`
`115106/04 (N.Y. Sup. Ct.);
`
`• In re Copaxone 40 mg Consolidated Cases, No. 1:14-cv-01171-GMS
`
`(consolidated) (D. Del.);
`
` 5
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 5
`
`

`

`
`
`• In re Certain Consolidated Zoledronic Acid Cases, No. 2:12-cv-
`
`03967-SDW-SCM (D.N.J.);
`
`• Gomez v. New York City Transit Authority, Index No. 33888/01 (N.Y.
`
`Sup. Ct.); and
`
`• The Medicines Company v. Mylan Inc., No. 1:11-CV-01285 (N.D.
`
`Ill.).
`
`8.
`
`I am being compensated at an hourly rate of $550/hour and am
`
`available to appear live for testimony in support of my opinions. My
`
`compensation in no way depends on the outcome of this proceeding. The opinions
`
`to which I will testify are based on the education, experience, training, and skill
`
`that I have accumulated in the course of my career as a biostatistician and
`
`researcher, as well as materials I have reviewed in connection with this case.
`
`III. MATERIALS CONSIDERED
`
`9.
`
`The list of materials I considered in forming the opinions set forth in
`
`this declaration is set forth in Exhibit B.
`
`IV. LEGAL STANDARDS
`
`10.
`
`I understand that this IPR involves U.S. Patent No. 8,822,438 (the
`
`“’438 patent”). Ex. 1001. I have been asked by Petitioners to review portions of
`
`the Declaration of Matthew B. Rettig, M.D. in Support of Janssen Oncology’s
`
`Patent Owner Response (the “Rettig Declaration”).
`
` 6
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 6
`
`

`

`
`
`11.
`
`I have been informed of the relevant legal principles as part of
`
`preparing and forming my opinions set forth in this declaration. I have applied my
`
`understanding of those principles in forming my opinions. My understanding of
`
`those principles is summarized below.
`
`12.
`
`I have been told that one of the four factual predicates in an
`
`obviousness inquiry is “secondary considerations.” I understand that one such
`
`secondary consideration is unexpected results. I further understand that, to show
`
`unexpected results, Patent Owner Janssen Oncology Inc. (“Patent Owner” or
`
`“Janssen”) must prove that the alleged invention produced benefits that were
`
`unexpected in light of the prior art, from the viewpoint of a person having ordinary
`
`skill in the field of technology of the patent at the patent’s priority date.
`
`13.
`
`I understand that evidence of secondary considerations such as
`
`unexpected results is only relevant to the obviousness analysis if the patentee can
`
`show a direct link, or nexus, between the secondary consideration and the claims of
`
`the patent, and that the evidence must be commensurate in scope with the asserted
`
`claims. I also understand that for results to be considered unexpected for these
`
`purposes, there must be a substantial difference from the prior art. In other words,
`
`a difference of kind, and not merely of degree.
`
`14.
`
`I have been asked to apply the definition of the person of ordinary
`
`skill in the art offered by Dr. Garnick:
`
` 7
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 7
`
`

`

`
`
`A person of ordinary skill in the art at the time of filing of this patent
`is someone who is a physician specializing in urology, endocrinology
`or oncology, or holds a Ph.D. in pharmacology, biochemistry or a
`related discipline. A related discipline may include, for example,
`pharmaceutical sciences. Additional experience could substitute for
`the advanced degree. To the extent necessary, a person of ordinary
`skill in the art may collaborate with one or more other persons of skill
`in the art for one or more aspects in which the other person may have
`expertise, experience and/or knowledge that was obtained through his
`or her education, industrial or academic experiences. A person of
`ordinary skill in the art may consult with an endocrinologist,
`oncologist or medical biochemist and thus may rely on the opinions of
`such specialists in evaluating the claims.
`
`15.
`
`I have also reviewed Patent Owner’s definition of the person of
`
`ordinary skill in the art. Ex. 2038 (Rettig Decl.) ¶ 78. My opinions are the same
`
`regardless of which definition is applied.
`
`V.
`
`SUMMARY OF OPINIONS
`
`16.
`
`It is my opinion that Janssen has not offered evidence of any
`
`unexpected results of the combination of abiraterone acetate and prednisone to
`
`treat metastatic castration-resistant prostate cancer (“mCRPC”). On the contrary,
`
`no reliable conclusions can be drawn regarding the combined anti-cancer effect of
`
`prednisone administered with abiraterone acetate based on the data presented in
`
`Dr. Rettig’s declaration. The data presented, and comparisons made, by Dr. Rettig
`
`in his declaration do not demonstrate unexpected results.
`
` 8
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 8
`
`

`

`
`
`VI. BACKGROUND
`
`A.
`
`17.
`
`Prostate Cancer And Abiraterone Acetate
`
`I understand that prostate cancer is an androgen-dependent disease
`
`driven by testosterone and its derivatives.2 Metastatic prostate cancer is cancer
`
`that is no longer localized and has spread beyond the prostate into other parts of the
`
`body.3 One strategy to treat metastatic prostate cancer is to lower testosterone
`
`levels, but patients frequently develop metastatic castration-resistant prostate
`
`cancer (“mCRPC”) when patients no longer respond to a reduction in testicular
`
`testosterone levels.4 mCRPC can be treated with “second-line” hormonal
`
`therapies, one of which is abiraterone acetate.5
`
`
`2 I am not an oncologist or urologist. My understanding of prostate cancer and its
`
`treatment comes from the references cited herein. Ex. 1092, Cancer.org (ACS),
`
`“Hormone
`
`Therapy
`
`for
`
`Prostate
`
`Cancer,”
`
`found
`
`at
`
`https://www.cancer.org/content/cancer/en/cancer/prostate-
`
`cancer/treating/hormone-therapy.html (accessed April 10, 2017).
`
`3 Ex.
`
`1093, Cancer.gov
`
`(NIH),
`
`“Metastatic
`
`cancer,”
`
`found
`
`at
`
`https://www.cancer.gov/types/metastatic-cancer (accessed April 14, 2017).
`
`4 Ex. 1092.
`
`5 Id.
`
` 9
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 9
`
`

`

`
`
`18. Prostate specific antigen (“PSA”) is a protein produced by the prostate
`
`gland.6 PSA levels can be measured and used to calculate time to PSA progression
`
`(“TTPP”).
`
`B. Cougar’s7 Clinical Trials Investigating Abiraterone Acetate
`
`19. Dr. Rettig’s unexpected results opinions rely on the results of
`
`disparate
`
`clinical
`
`trials
`
`conducted by
`
`Janssen’s predecessor, Cougar
`
`Biotechnology, Inc. (“Cougar”). These clinical trials include COU-AA-001, COU-
`
`AA-002, COU-AA-301, and COU-AA-302.
`
`20. COU-AA-001 was a single arm, open label, single center phase I/II
`
`study conducted in the United Kingdom to assess the safety, tolerability, and
`
`recommended abiraterone acetate dose for treating mCRPC patients. Ex. 2014
`
`(Attard 2008); Ex. 2015 (Attard 2009); see also Ex. 2038 (Rettig Decl.) ¶¶ 194–97,
`
`205 (representing that Attard 2008 and Attard 2009 describe the COU-AA-001
`
`study). The study administered single agent abiraterone acetate to chemotherapy-
`
`naïve patients. Id. The phase I portion included dose escalation, and 1000 mg
`
`abiraterone acetate was administered daily in the phase II portion. Ex. 2014
`
`6 Ex. 1094, Cancer.gov (NIH NCI), “Prostate-Specific Antigen (PSA) Test,” found
`
`at https://www.cancer.gov/types/prostate/psa-fact-sheet (accessed April 11, 2017).
`
`7 I understand that Johnson & Johnson, Janssen’s parent company, acquired
`
`Cougar Biotechnology, Inc. in 2009. Ex. 2101.
`
` 10
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 10
`
`

`

`
`
`(Attard 2008) at 4563, 4565; Ex. 2015 (Attard 2009) at 3743.8 Mineralocorticoid
`
`excess toxicities were managed with a mineralocorticoid receptor antagonist or
`
`dexamethasone 0.5 mg daily. Ex. 2014 (Attard 2008) at 4565; Ex. 2015 (Attard
`
`2009) at 3743. COU-AA-001 included an extension study in which 0.5 mg/day
`
`dexamethasone was added to patients who progressed on abiraterone acetate
`
`therapy. Id.
`
`21. COU-AA-002 was a single arm, open label, multicenter phase II study
`
`conducted at 5 sites in the United States. Ex. 2017 (Ryan 2011); see also Ex. 2038
`
`(Rettig Decl.) ¶ 205 (representing that Ryan 2011 describes the COU-AA-002
`
`study). The purpose of the COU-AA-002 study was to assess the efficacy of
`
`abiraterone acetate in treating mCRPC patients and investigate the frequency of
`
`bone scan results inconsistent with PSA and clinical response. Id. at 4854.
`
`Patients without previous chemotherapy or ketoconazole treatment were given
`
`1000 mg abiraterone acetate and 10 mg prednisone daily. Id.
`
`22. COU-AA-301 was a two arm, double-blind, randomized, multicenter
`
`phase III clinical trial in 1195 chemotherapy-refractory patients. Ex. 1034 (de
`
`Bono). The treatment arm administered 1000 mg abiraterone acetate and 10 mg
`
`8 Because there are no page numbers branded on the Janssen exhibits (all exhibits
`
`numbered Ex. 2001 and higher), I have cited to the internal page numbers of these
`
`exhibits.
`
` 11
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 11
`
`

`

`
`
`prednisone daily, and the placebo arm administered placebo and 10 mg prednisone
`
`daily. Id. at 1.
`
`23. COU-AA-302 was a two arm, double-blind, randomized, multicenter
`
`phase III clinical trial in 1088 chemotherapy-naïve patients. Ex. 1009 (Ryan
`
`2013); Ex. 2071 (Ryan 2015). The treatment arm administered 1000 mg
`
`abiraterone acetate and 10 mg prednisone daily, and the placebo arm administered
`
`placebo and 10 mg prednisone daily. Ex. 1009 (Ryan 2013) at 3; Ex. 2071 (Ryan
`
`2015) at 152.
`
`C.
`
`Statistical Principles
`
`24. Statistical analysis can be used to test hypotheses to assess whether
`
`they are supported by data. More generally, statistics can be used to interpret and
`
`draw inferences from data sets. Hypothesis testing is the process of using data and
`
`statistical tests to assess the weight of the evidence in support of a hypothesis.
`
`25. Statistically-based studies attempt to make inferences about a
`
`population from a representative sample of that population. Statistical power is a
`
`measurement of a study’s ability to accurately test a hypothesis. Power is affected
`
`by the sample size and the variance of the study population.
`
`26. A confidence interval indicates the precision available from a data set
`
`for estimating a quantity of interest. Particularly, if a point estimate (i.e., a median
`
`PSA value) is accompanied by a 95% confidence interval, there is a 95% chance
`
` 12
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 12
`
`

`

`
`
`that the true value falls within the reported confidence interval. A finding of non-
`
`overlapping confidence intervals provides evidence that the two population values
`
`are significantly different. However, in the case of overlapping confidence
`
`intervals, no statistically significant difference can be claimed.
`
`27.
`
`In the clinical trial context, studies can be designed to compare
`
`treatments. Phase III clinical trials typically have two arms—a treatment arm and
`
`a comparator arm. The comparator arm can comprise a placebo comparator or an
`
`active comparator, which allows the treatment arm to be compared with another
`
`effective intervention. A well-designed clinical trial allows the investigator to
`
`measure the effectiveness of a treatment over the placebo or active comparator.
`
`28.
`
`Ideally, researchers can obtain robust information about a treatment
`
`from a well-designed clinical trial. In limited circumstances, researchers may also
`
`make inferences from cross-study comparisons. One requirement for any reliable
`
`cross-study comparison is that the compared studies must be similar in patient
`
`populations and study procedures.
`
`VII. JANSSEN HAS NOT DEMONSTRATED UNEXPECTED RESULTS
`
`29. Dr. Rettig has advanced two arguments that the combination claimed
`
`in the ’438 patent produces unexpected results. First, Dr. Rettig compared point
`
`estimates across multiple phase I/II studies to argue that the abiraterone acetate and
`
`prednisone combination produces an anti-cancer effect more than twice as great as
`
` 13
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 13
`
`

`

`
`
`that produced by abiraterone acetate alone. Ex. 2038 (Rettig Decl.) ¶¶ 203–13.
`
`Second, Dr. Rettig reported data from certain individual patients participating in a
`
`phase I/II extension study involving dexamethasone to advance the argument that
`
`glucocorticoids reverse the resistance that patients develop to abiraterone acetate
`
`treatment. Id. ¶¶ 194–202.
`
`30. Upon investigation, both of Dr. Rettig’s arguments contain fatal flaws
`
`and do not support any conclusion of unexpected results. Dr. Rettig did not
`
`demonstrate that abiraterone acetate and prednisone combination therapy has a
`
`longer time to PSA progression (“TTPP”) than abiraterone acetate monotherapy
`
`for four independent reasons:
`
`(1) The confidence intervals surrounding the compared point estimates
`
`overlap;
`
`(2) Other reported data conflicts with the data used to support Dr. Rettig’s
`
`conclusions;
`
`(3) Dr. Rettig improperly compares point estimates across different
`
`clinical studies; and
`
`(4) The phase III studies cannot ascertain any effect of prednisone.
`
`31. The data from the phase I/II extension study also do not support a
`
`conclusion of unexpected results because the results are anecdotal and do not give
`
`a full picture of the COU-AA-001 extension study’s results.
`
` 14
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 14
`
`

`

`
`
`A. Dr. Rettig’s Median TTPP Data Point Comparisons Do Not
`Demonstrate Any Unexpected Clinical Efficacy Of The
`Abiraterone Acetate And Prednisone Combination
`
`32. Dr. Rettig’s comparison table does not establish that the combination
`
`of abiraterone acetate and prednisone has a median TTPP more than twice as long
`
`as abiraterone acetate alone, as Dr. Rettig asserts in paragraph 205 of his
`
`declaration. In fact, Dr. Rettig has not shown any statistically significant
`
`difference in median TTPP between the combination of abiraterone acetate and
`
`prednisone and abiraterone acetate treatment alone.
`
`1.
`
`The Confidence Intervals In Dr. Rettig’s Comparison Table
`Overlap
`
`33. Table 1 of Dr. Rettig’s declaration reports a median TTPP point
`
`estimate for four patient groups: (1) abiraterone acetate monotherapy (“AA”), (2)
`
`abiraterone acetate + dexamethasone at progression in dexamethasone naïve
`
`patients (“AA+dex – dex naïve”), (3) abiraterone acetate + dexamethasone at
`
`progression in patients who previously failed dexamethasone monotherapy
`
`(“AA+dex – previous dex treatment”), and (4) abiraterone acetate + prednisone
`
`from the start (“AA+pred”). Id. ¶ 205.
`
`34. Dr. Rettig failed to report the confidence intervals around the point
`
`estimates listed in Table 1, even though these confidence intervals are readily
`
`reported in the publications he cites in the very same table. Ex. 2015 (Attard 2009)
`
` 15
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 15
`
`

`

`
`
`at 3745–46; Ex. 2017 (Ryan 2011) at 4856–57. In fact, Dr. Rettig never even
`
`mentions these confidence intervals in his declaration.
`
`35. Dr. Rettig’s omission of the confidence intervals is particularly
`
`troubling because the reported confidence intervals actually overlap, meaning that
`
`Dr. Rettig has not shown there is a statistically significant difference between the
`
`median TTPPs for any of the four groups in his comparison table:
`
` 16
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 16
`
`

`

`
`
`Table 1: Adaptation of Rettig Table 1, With Confidence Intervals
`95% Confidence Interval9
`
`Treatment
`
`Median Time to PSA
`Progression (TTPP)
`7.5 months (225 days)
`AA
`AA + dex – dex-naïve 12 months (361 days)
`
`AA + dex – previous
`dex treatment
`AA + pred
`
`12.4 months (372 days)
`
`16.3 months10
`
`5.4–9.6 months (162–287 days)
`8.0–16.0 months
`(241–480
`days)
`1.8–22.9 months (55–688 days)
`
`9.2 months–upper
`estimable
`
`limit not
`
`Ex. 2015 (Attard 2009) at 3745–46; Ex. 2017 (Ryan 2011) at 4856–57.
`
`
`9 The confidence intervals reported in the Attard 2009 and Ryan 2011 publications
`
`are 95% confidence intervals surrounding a single point estimate. If Dr. Rettig
`
`wanted to demonstrate a statistically significant difference between two medians
`
`based on such confidence intervals, it would be necessary to use a higher level of
`
`confidence (namely 97.5%) for each interval to ensure an overall confidence level
`
`of 95%.
`
`10 I note that both of Janssen’s higher-powered phase III trials reported a lower
`
`median TTPP for patients treated with abiraterone acetate and prednisone. In
`
`COU-AA-301, the median TTPP in chemo-refractory patients treated with AA +
`
`pred was 10.2 months, and in COU-AA-302, the median TTPP in chemo-naïve
`
`patients treated with AA + pred was 11.1 months. Ex. 1034 (de Bono 2011) at 5;
`
`Ex. 1009 (Ryan 2013) at 7.
`
` 17
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 17
`
`

`

`
`
`Figure 1: Representation of Rettig Table 1 Data, With Confidence Intervals
`
`
`
`The overlap between confidence intervals is crucial because it indicates that there
`
`may be no statistically significant difference between any of the median TTPP
`
`point estimates listed above.
`
`36.
`
`I note that on clinicaltrials.gov, Janssen reports a higher AA median
`
`TTPP point estimate of 11 months (330 days), with a much broader 95%
`
`confidence interval of 6.6–17.7 months (197–530 days). Ex. 1095 (COU-AA-001
`
`synopsis) at 5. It is unclear why a longer median TTPP value appears on
`
`clinicaltrials.gov, but it may be because some COU-AA-001 patients were still
`
`being successfully treated on abiraterone acetate alone at the data cutoff. See Ex.
`
` 18
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 18
`
`

`

`
`
`2015 (Attard 2009) at 3745 (“[S]ix patients continued on single-agent abiraterone
`
`acetate at the date of data cutoff.”). Moreover, the abiraterone acetate
`
`monotherapy portion of the trial lasted only 12 months. Id. at 3. Accordingly, the
`
`median TTPP for AA monotherapy patients reported in the Attard 2009 paper
`
`could not have exceeded 12 months, because all patients successfully treated with
`
`abiraterone acetate after 12 months were given concomitant dexamethasone if they
`
`wanted to participate in the extension study. The clinicaltrials.gov figure may
`
`therefore be more representative of the AA monotherapy trial results than the data
`
`on which Dr. Rettig relies. Figure 2 compares the median TTPP and confidence
`
`interval reported on clinicaltrials.gov with the data presented in Dr. Rettig’s Table
`
`1:
`
` 19
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 19
`
`

`

`
`
`Figure 2: Comparison of Conflicting Janssen-Reported AA Data
`With Rettig Table 1 Data
`
`
`
`37. Simply comparing point estimates (here, the median TTPP) is not
`
`sufficient to demonstrate a meaningful difference between parameters. Indeed,
`
`where confidence intervals overlap, as here, comparing point estimates alone can
`
`be highly misleading. Dr. Rettig failed to perform the necessary statistical analysis
`
`to ensure that his comparison between point estimates is scientifically robust—
`
`namely, that there is a statistically significant difference between abiraterone
`
`acetate treatment alone and abiraterone acetate combined with a glucocorticoid.
`
`Any differences between the median TTPP estimates themselves, without more,
`
`cannot support any conclusion of unexpected results.
`
` 20
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 20
`
`

`

`
`
`38.
`
`It is not surprising to me that the confidence intervals surrounding the
`
`point estimates in Dr. Rettig’s Table 1 overlap. Confidence interval width is
`
`related to study power—in general, the higher a study’s power, the narrower the
`
`confidence intervals. Dr. Rettig chose to present data from a selection of small
`
`phase I/II studies, and the broad confidence intervals are largely due to the small
`
`sample sizes used in those trials. Ex. 2015 (Attard 2009) at 3745–46 (the AA point
`
`estimate was obtained from a sample of 42 patients, and the AA + dex point
`
`estimates were obtained from samples of 19 patients (dex-naïve group) and 11
`
`patients (previous dex treatment group)); Ex. 2017 (Ryan 2011) at 4856–57 (the
`
`AA + pred point estimate was obtained from a sample of 33 patients).
`
`39. Because the confidence intervals surrounding the median TTPP point
`
`estimates in Dr. Rettig’s comparison table overlap, Dr. Rettig’s conclusions were
`
`improper. Dr. Rettig presented no evidence of any difference—much less a
`
`doubling—in TTPP between treatment with abiraterone acetate alone and
`
`treatment with abiraterone acetate and prednisone. Therefore, the data in Dr.
`
`Rettig’s comparison table fail to support his conclusion that the claimed invention
`
`exhibited unexpected results.
`
`2.
`
`Other Data Reported By Janssen And Its Predecessor Conflict
`With The Data Reported In Dr. Rettig’s Declaration
`
`40. Dr. Rettig’s Table 1 compares median TTPP point estimates in two
`
`studies conducted in chemotherapy-naïve patient populations. However, Dr. Rettig
`
` 21
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 21
`
`

`

`
`
`omits median TTPP estimates from other Janssen-sponsored studies of abiraterone
`
`acetate that do not support his conclusion of unexpected results.
`
`41. For example, Janssen conducted phase II studies in patients who had
`
`previously been treated with chemotherapy (“chemo-refractory”). Ex. 1096 (Reid
`
`2010); Ex. 2016 (Danila 2010). Reid 2010 describes a multicenter phase II study
`
`(COU-AA-003) in which 1000 mg abiraterone acetate monotherapy was
`
`administered daily to 47 patients, and TTPP was a secondary endpoint. Ex. 1096
`
`(Reid 2010) at 1–2. Danila 2010 describes a multicenter phase II study (COU-AA-
`
`004) in which 1000 mg abiraterone acetate was administered daily to 58 patients
`
`along with 10 mg prednisone. Ex. 2016 (Danila 2010) at 1496–97; see also Ex.
`
`2038 (Rettig Decl.) ¶ 202 (representing that Danila 2010 describes the COU-AA-
`
`004 study). TTPP was also a secondary endpoint in the Danila 2010 study. Ex.
`
`2016 (Danila 2010) at 1498.
`
`42. Below, I compare the median TTPP results from the COU-AA-003
`
`and COU-AA-004 studies with the median TTPP results reported in Table 1 of the
`
`Rettig declaration:
`
` 22
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 22
`
`

`

`
`
`Table 2: Median TTPP for Omitted COU-AA-003 and -004 Trials
`
`Clinical Trial
`
`Patient
`Group
`COU-AA-003 Chemo-
`refractory
`
`COU-AA-004 Chemo-
`refractory
`
`Treatment
`
`Abiraterone
`acetate
`monotherapy
`Abiraterone
`acetate +
`prednisone
`from the start
`
`Median
`TTPP
`5.6 months
`(169 days)
`
`5.6 months
`(169 days)
`
`Source
`
`Ex. 1096
`(Reid 2010) at
`4.
`Ex. 2016
`(Danila 2010)
`at 1499.
`
`Rettig Table 1 (Ex. 2038 (Rettig Decl.) ¶ 205)
`
`Clinical Trial
`
`Patient
`Group
`COU-AA-001 Chemo-naïve Abiraterone
`acetate
`monotherapy
`
`Treatment
`
`COU-AA-002,
`Amendment 5
`
`
`
`Chemo-naïve Abiraterone
`acetate +
`prednisone
`from the start
`
`Median
`TTPP
`7.5 months
`(225 days)
`
`16.3 months
`
`Source
`
`Ex. 2015
`Attard et al.
`(2009), at
`3745.
`Ex. 2017,
`Ryan et al.
`(2011), at
`4856–4857.
`
`43. Unlike the median TTPP point estimates Dr. Rettig presented to
`
`support his unexpected results argument, the median TTPP point estimates from
`
`Cougar’s chemo-refractory phase II trials are identical for abiraterone acetate
`
`alone and combination therapy with prednisone (both 5.6 months/169 days).
`
`Accordingly, the data from the COU-AA-003 and -004 trials do not support the
`
`existence of a difference in median TTPP between abiraterone acetate alone and
`
`combination therapy with prednisone.
`
` 23
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 23
`
`

`

`
`
`3.
`
`Dr. Rettig’s Analysis Engages in an Improper Cross-Study
`Comparison
`
`44. Dr. Rettig relies on data points from two separate clinical studies,
`
`COU-AA-001 and COU-AA-002, to compare median TTPP point estimates
`
`between AA and AA+pred treatment groups. While Dr. Rettig points out a few
`
`choice similarities between these studies, he ignores a vast number of meaningful
`
`differences in patients, location, and study procedure that prevent any comparison
`
`between the two studies.
`
`45. The best way to assess a potential difference between two treatments
`
`is to test them together in a randomized, double blinded, controlled trial. See, e.g.,
`
`Ex. 2038 (Rettig Decl.) ¶ 206. Cross-study comparisons are possible in specific,
`
`limited circumstances, but the comparison must control for differences in the
`
`compared studies. Otherwise, one cannot ensure that the compared treatments,
`
`instead of a difference in the compared studies, is responsible for the finding of a
`
`difference.
`
`46. Dr. Rettig attempts to justify his improper cross-study comparison by
`
`selectively disclosing three similarities in patient populations and study design.
`
`Ex. 2038 (Rettig Decl.) ¶ 207. Dr. Rettig paints an incomplete picture by failing to
`
`disclose numerous differences in the two studies’ patients, endpoints, and study
`
`procedures.
`
` 24
`
`MYLAN PHARMS. INC. EXHIBIT 1091 PAGE 24
`
`

`

`
`
`47. Following is a non-exhaustive list of the differences in the COU-AA-
`
`001 and COU-AA-002 patients:11
`
`Table 3: Differences Between COU-AA-001 and -002 Patients
`
`COU-AA-001
`110 ng