`______________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`______________
`
`COALITION FOR AFFORDABLE DRUGS VII LLC
`Petitioner
`
`v.
`
`POZEN INC.
`Patent Owner
`______________
`
`Case No. IPR2015-01718
`Patent No. 8,945,621
`______________
`
`
`
`DECLARATION OF ROBERT W. MAKUCH, PH.D.
`
`IN SUPPORT OF PATENT OWNER’S RESPONSE
`
`Ex. 2021
`
`
`
`
`
`
`
`
`
`Page 1 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`Table of Contents
`
`
`Page
`
`
`
`I.
`
`II.
`
`III.
`
`IV.
`
`V.
`
`VI.
`
`INTRODUCTION AND SUMMARY OF QUALIFICATIONS...................................... 2
`
`MATERIALS CONSIDERED .......................................................................................... 5
`
`UNDERSTANDING OF THIS PROCEEDING ............................................................... 5
`
`UNDERSTANDING OF RELEVANT LEGAL PRINCIPLES ........................................ 6
`
`UNDERSTANDING OF THE TECHNICAL BACKGROUND...................................... 7
`
`THE ’621 PATENT ........................................................................................................... 8
`
`VII. PETITIONER’S OBVIOUSNESS CHALLENGES ......................................................... 9
`
`VIII. THE PRIOR ART DOES NOT RENDER CLAIMS 1-16 OF THE ’621 PATENT
`OBVIOUS ........................................................................................................................ 11
`A.
`GRAHAM ............................................................................................................ 11
`B.
`GOLDSTEIN ....................................................................................................... 16
`
`
`
`
`
`-
`
`
`
`i-
`
`Page 2 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`I, Robert W. Makuch, Ph.D., hereby declare and state as follows:
`
`I.
`
`INTRODUCTION AND SUMMARY OF QUALIFICATIONS
`
`1.
`
`2.
`
`I am over the age of eighteen and otherwise competent to make this declaration.
`
`I have been retained as an expert opinion witness on behalf of Patent Owners
`
`Pozen Inc. and Horizon Pharma Inc. for the above-captioned inter partes review (“IPR”). I am
`
`being compensated for my time in connection with this IPR at my standard consulting rate,
`
`which is $625 per hour.
`
`3.
`
`I have no financial interest in, or affiliation with, the Petitioner or the Patent
`
`Owners. My compensation is not dependent upon the outcome of, or my testimony in, the
`
`present inter partes review or any litigation proceedings.
`
`4.
`
`My background, qualifications, and experience relevant to the issues in this
`
`proceeding are summarized below. A full description of my background and qualifications is set
`
`forth in my curriculum vitae, attached hereto..
`
`5.
`
`I am a tenured Full Professor in the Department of Biostatistics at the Yale
`
`University School of Medicine. I am also currently Director of the Regulatory Affairs Track
`
`Yale University School of Medicine.
`
`6.
`
`I possess nearly 38 years of experience in a variety of areas including but not
`
`limited to identification, evaluation, and interpretation of clinical endpoints and other clinical
`
`events. My research has focused primarily on the design, conduct, analysis, and interpretation of
`
`clinical and pre-clinical data obtained from pre-clinical experiments, clinical trials, and
`
`epidemiologic studies.
`
`
`
`Ex. 2021
`
`Page 1
`
`Page 3 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`7.
`
`My activities include, but are not limited to, serving as: 1) Special Government
`
`Employee (SGE) to the U.S. Food and Drug Administration (FDA); 2) consultant/advisor to
`
`numerous government and pharmaceutical-sponsored clinical and pre-clinical studies; and 3)
`
`author or co-author of over 200 peer-reviewed publications in the medical and regulatory affairs
`
`literature.
`
`8.
`
`I received my M. Phil. and Ph.D. from Yale University in 1976 and 1977,
`
`respectively.
`
`9.
`
`I received my M.A. degree from University of Washington in 1974 and my B.A.
`
`degree from University of Connecticut in 1972.
`
`10.
`
`In 1977 I became an investigator and was subsequently promoted to Senior
`
`Investigator at the National Cancer Institute. In 1983 I was again promoted to Section Head of
`
`the biostatistics and data management section. I held that position until 1986. Many of my
`
`activities in these roles involved the design, conduct, analysis, and interpretation of clinical trials
`
`and pre-clinical studies.
`
`11.
`
`In 1986, I joined the faculty of Yale University in the Division of Biostatistics as
`
`an Associate Professor. I was tenured in 1990 and promoted to Full Professor in 1995 which I
`
`continue to hold today.
`
`12.
`
`From 1996-2003, I was the Acting Division Head, and then the Division Head, for
`
`the Division of Biostatistics, Yale University School of Medicine.
`
`13.
`
`I am also Director of the Regulatory Affairs Program and Chairman of the
`
`Regulatory Affairs Advisory Board at Yale University. I am Director of the Certificate Program
`
`in Regulatory Affairs for Senior Delegations of the Chinese Food and Drug Administration
`
`
`
`Ex. 2021
`
`Page 2
`
`Page 4 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`(CFDA), and an Invited Member of the Dean’s Advisory Board at the University of Connecticut
`
`for the College of Liberal Arts and Sciences.
`
`14.
`
`I currently teach two courses at the graduate level at Yale: Fundamentals of
`
`Clinical Trials and Introduction to Regulatory Affairs.
`
`15.
`
`I am an author or co-author on over 200 peer-reviewed publications, most of
`
`which relate to the design, conduct, analysis, and interpretation of clinical and pre-clinical
`
`studies, and have given numerous presentations on these topics at national and international
`
`meetings, universities, and government locations.
`
`16.
`
`I was nominated and appointed to serve as a Special Government Employee
`
`(SGE) to the FDA from 2002 to 2007. In this capacity, I performed a number of activities at the
`
`request of the FDA including serving on Advisory Committees and providing expert advice to
`
`the FDA on a variety of topics.
`
`17.
`
`Honors and awards I have received for teaching and research include the
`
`Connecticut Public Health Association Award for AIDS Research (1987). This award was given
`
`for my above-mentioned activities, and led to my selection to serve on The Stewart B. McKinney
`
`Foundation Advisory Board, a Foundation in honor of Congressman Stewart B. McKinney of
`
`Connecticut. I was also nominated and received the honor of becoming a Fellow of the
`
`American Statistical Association (ASA) (2003), a superlative honor that, under the Association
`
`by-laws, may be bestowed on no more than one-third of one percent of the total ASA
`
`membership (roughly 18,000 members currently). I was nominated and became a Fellow of
`
`Silliman College at Yale University. A complete list of my memberships on numerous academic,
`
`professional, and scholarly societies is found in my curriculum vitae.
`
`
`
`Ex. 2021
`
`Page 3
`
`Page 5 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`II. MATERIALS CONSIDERED
`18.
`
`In formulating my opinions, I have relied on my thirty-eight years of experience
`
`in biostatistics, clinical trials, epidemiology, and regulatory affairs. In addition to my expertise, I
`
`have specifically considered the ’621 patent, the Coalition for Affordable Drugs VII LLC’s
`
`(“Petitioner”) petition for inter partes review of the ’621 patent (the “Petition”) and associated
`
`exhibits. I have also considered the declaration of Dr. Leon Shargel (the “Shargel Decl.”), the
`
`transcript of the May 25, 2016 deposition of Dr. Shargel (the “Shargel Dep. Tr.”), and Patent
`
`Owner’s Preliminary Response.
`
`III. UNDERSTANDING OF THIS PROCEEDING
`19.
`I understand that this is an inter partes review (IPR) proceeding conducted before
`
`the Patent Trial and Appeal Board (“Board”) of the U.S. Patent and Trademark Office (“PTO”)
`
`to determine if claims 1-16 of the ’621 patent should be cancelled as unpatentable.
`
`20.
`
`I understand that Petitioner filed its Petition on August 12, 2015 asserting that
`
`claims 1-16 of the ‘621 patent are invalid as obvious over combinations of references. Pet. at 4-5.
`
`I further understand that the Petition was accompanied by a declaration of Dr. Leon Shargel.
`
`21.
`
`I understand that on November 23, 2015 Patent Owner submitted a Preliminary
`
`Response in opposition to the Petition.
`
`22.
`
`I understand that on February 22, 2016, the Board decided to institute an IPR and
`
`found that the Petition satisfied the threshold standard for institution by showing a “reasonable
`
`likelihood that Petitioner would prevail in establishing the unpatentability” of the challenged
`
`claims. Further, I understand that the Board has not made any final determination that claims 1-
`
`16 are obvious.
`
`
`
`Ex. 2021
`
`Page 4
`
`Page 6 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`IV. UNDERSTANDING OF RELEVANT LEGAL PRINCIPLES
`23.
`
`I understand that the ’621 patent must be considered from the viewpoint of a
`
`person of ordinary skill in the relevant art (“POSA”) as of June 25, 2009. A POSA is a
`
`hypothetical person who is presumed to be aware of all pertinent art.
`
`24.
`
`Further, I have been advised of the following legal principles, and have applied
`
`those principles to my analysis and conclusions as set forth in this declaration.
`
`25.
`
`It has been explained to me that that an invention may be unpatentable “if the
`
`differences between the subject matter sought to be patented and the prior art are such that the
`
`subject matter as a whole would have been obvious at the time the invention was made to a
`
`person having ordinary skill in the art to which said subject matter pertains.” 35. U.S.C. § 103.
`
`As I understand it, obviousness is a question of law that is based on the following factors: (i) the
`
`scope and content of the prior art, (ii) the claims at issue, (iii) the level of ordinary skill in the art,
`
`and (iv) objective evidence of secondary considerations. Routine experimentation does not
`
`render an otherwise obvious claim valid. .Obviousness only calls for a reasonable expectation of
`
`success, not a guarantee.
`
`26.
`
`I further understand that the Petitioner, in order to establish obviousness based on
`
`a combination of references, must provide sufficient rationale as to why, at the time of invention,
`
`a POSA would have been motivated to combine the teachings of those references to come up
`
`with the claimed subject matter. The Petitioner must also establish that a POSA would have had
`
`a reasonable expectation of success in combining said references. I further understand that a
`
`POSA cannot rely on hindsight to provide a motivation to select and combine prior art
`
`references.
`
`
`
`Ex. 2021
`
`Page 5
`
`Page 7 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`27.
`
`I also understand that obviousness cannot be shown where the art in any material
`
`respect “teaches away” from the claimed invention.
`
`V. UNDERSTANDING OF THE TECHNICAL BACKGROUND
`28.
`
`It has been explained to me that non-steroidal anti-inflammatory drugs
`
`(“NSAIDs”) are a class of drugs used to treat pain and inflammation associated with arthritis and
`
`other musculoskeletal disorders. NSAIDS include drugs such as naproxen and aspirin. While
`
`these drugs are widely used, their use has been associated with injury to the upper
`
`gastrointestinal (“GI”) tract. I understand that prior to the inventions claimed in the ‘621 patent,
`
`it was widely believed that a patient had an increased risk of upper GI injury, such as ulcer or
`
`bleeding, when taking multiple NSAIDs, including the combination of naproxen and low-dose
`
`aspirin (“LDA”).
`
`29.
`
`It has further been explained to me that the commercial embodiment of the ’621
`
`patent, the pharmaceutical drug product, VIMOVO®, was designed to reduce the risk of upper
`
`GI injury due to NSAID use. As I understand it, VIMOVO® consists of a combination of a
`
`delayed-release, enteric-coated NSAID core (naproxen) surrounded by an immediate-release acid
`
`inhibitor (esomeprazole magnesium). The acid inhibitor is released before the NSAID which
`
`allows the acid inhibitor’s gastroprotective effects to take hold before naproxen is released. This
`
`coordinated release of these two compounds reduces the potential for gastric damage.
`
`30.
`
`As stated above, I understand that before the invention claimed in the ‘621 patent,
`
`it was accepted that a patient’s concomitant use of an NSAID and LDA increased the risk of
`
`gastrointestinal injury. I further understand, however, the ’621 patent unexpectedly demonstrates
`
`the opposite: the administration on of a unit dose form of immediate-release esomeprazole and
`
`delayed-release naproxen is more effective at reducing the incidence of the NSAID-associated
`Ex. 2021
`
`Page 6
`
`
`
`Page 8 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`ulcers in patients taking LDA than in patients not taking LDA. Thus, rather than increasing the
`
`risk of occurrence of an NSAID-associated gastric ulcer, the concomitant usage of LDA with a
`
`unit dose form of immediate-release esomeprazole and delayed-release naproxen, surprisingly,
`
`reduces that risk.
`
`VI. THE ’621 PATENT
`31.
`
`I understand that the ’621 patent describes methods of reducing the incidence of
`
`gastric ulcers associated with use of NSAIDs in patients who are also taking low-dose aspirin.
`
`The claimed methods require that the patient receive a unit dosage form that is comprised of 20
`
`mg of esomeprazole and 500 mg of naproxen and that provides for coordinated release of the
`
`esomeprazole and the naproxen. Significantly, the claimed methods also require that the dosage
`
`form is more effective at reducing the incidence of NSAID-associated ulcers in patients taking
`
`LDA than in patients not taking LDA.
`
`32.
`
`It is my understanding that the ’621 patent is one of the patents listed in the
`
`FDA’s Approved Drug Products with Therapeutic Equivalence Evaluations (the “Orange Book”)
`
`as covering the pharmaceutical product VIMOVO®.
`
`33.
`
`The ’621 patent contains 16 claims. Claims 1, 8, 15, and 16 are independent
`
`claims
`
`34.
`
`Independent claim 1 is illustrative, and is copied below:
`
`1. A method of reducing the incidence of NSAID-associated
`gastric ulcers in patients taking low dose aspirin who are at risk of
`developing such ulcers, wherein the method comprises administering
`to said patient in need thereof a pharmaceutical composition in unit
`dose form comprising:
`
`Ex. 2021
`
`Page 7
`
`
`
`Page 9 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`(a) 20 mg of esomeprazole, or pharmaceutically acceptable salt
`thereof, in a form and route sufficient to raise the gastric pH
`of said patient to at least 3.5 upon the administration of one
`or more of said unit dosage forms, and
`(b) 500 mg of naproxen, or pharmaceutically acceptable salt
`thereof;
`wherein said unit dose form provides for coordinated release of
`the esomeprazole and the naproxen,
`wherein at
`least a portion of said esomeprazole, or
`pharmaceutically acceptable salt thereof, is released independent of
`the pH of the surrounding medium,
`wherein the unit dosage form releases less than 10% of the
`naproxen or a pharmaceutically acceptable salt thereof after 2 hours
`when tested using the USP Paddle Method in 1000 ml of 0.1N HCl at
`75 rpm at 37º C.+/-0.5º C.,
`wherein said pharmaceutical composition in unit dose form
`reduces the incidence of NSAID-associated ulcers in said patient
`and wherein administration of the unit dose form is more
`effective at reducing the incidence of the NSAID-associated
`ulcers in patients taking LDA than in patients not taking LDA
`who are administered the unit dose form.
`(Ex. 1001 at 26:61–27:20) (emphasis added).
`
`VII. PETITIONER’S OBVIOUSNESS CHALLENGES
`35.
`
`I understand that the Petitioner has raised two separate obviousness challenges to
`
`the ’621 patent.
`
`
`
`Ex. 2021
`
`Page 8
`
`Page 10 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`36.
`
`In Ground 1, the Petitioner asserts that claims 1-16 are obvious over the
`
`combination of U.S. Patent 6,926,907 (“Plachetka”) (Ex. 1004.); “Ulcer Prevention in Long-term
`
`Users of Nonsteroidal Anti-inflammatory Drugs,” by David Y. Graham, et al., published in the
`
`Archives of Internal Medicine, Vol. 162 on January 28, 2002. (“Graham”) (Ex. 1005); and
`
`“Ulcer Recurrence in High-Risk Patients Receiving Nonsteroidal Anti-Inflammatory Drugs Plus
`
`Low-Dose Aspirin: Results of a Post Hoc Subanalysis,” by Jay L. Goldstein, et al., published in
`
`Clinical Therapeutics, Vol. 26, No. 10, on October 2004. (“Goldstein”) (Ex. 1006).
`
`37.
`
`In Ground II, the Petitioner asserts that claims 1-16 are obvious over Plachetka
`
`alone.
`
`38. My analysis and opinions are focused on Ground I, more specifically, the
`
`combination of the Graham and Goldstein references.
`
`39.
`
`The Petitioner alleges that “Graham and Goldstein show that the combination of
`
`lansoprazole and an NSAID was more effective at reducing gastric ulcers in patients also taking
`
`low-dose aspirin (LDA) than in patients not also taking LDA.” (Pet. at 7.) Further, Petitioner
`
`asserts that “the results disclosed by Graham and Goldstein would indicate that the compositions
`
`disclosed in Plachetka could be used for treating patients on an LDA regimen to achieve the
`
`predictable result of lower gastric ulcer incidence.” (Pet. at 12.)
`
`40.
`
`I understand, however, that the Board disagreed with Petitioner and concluded:
`
`Petitioner has not established that Plachetka, Graham, and Goldstein
`teach or suggest that “administration of the unit dose form is more
`effective at reducing the incidence of the NSAID-associated ulcers in
`patients taking LDA than in patients not taking LDA who are
`administered the unit dose form,” as required by each of independent
`claims 1, 8, 15, and 16 (emphasis added).
`
`Ex. 2021
`
`Page 9
`
`
`
`Page 11 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`(Decision at 14).
`
`41.
`
`The Board further stated that the Petitioner and the Petitioner’s declarant, Dr.
`
`Shargel, failed to identify a basis for the conclusion that a POSA would have understood that a
`
`combination of lansoprazole (another acid inhibitor) and naproxen was more effective at
`
`reducing the incidence of the NSAID-associated ulcers in patients taking LDA than in patients
`
`not taking LDA. (Decision at 15.)
`
`42.
`
`As described more fully below, I agree with the Board’s conclusion on this point.
`
`It is my opinion that a POSA, after reviewing Plachetka, Graham, and Goldstein, would not be
`
`able to draw any conclusions about whether the administration of the claimed unit dose form is
`
`more effective in reducing the incidence of NSAID-associated ulcers in patients taking LDA
`
`than in patients not taking LDA.
`
`VIII. THE PRIOR ART DOES NOT RENDER CLAIMS 1-16 OF THE ’621
`PATENT OBVIOUS
`
`43. While my analysis below focuses on the Graham and Goldstein references, I
`
`understand that Plachetka recognizes that NSAID administration can lead to upper GI injury, and
`
`suggests that a POSA would have understood the risk of GI injury to patients taking both LDA
`
`and an NSAID. (Ex. 1004 at 1:23-25; 2:35-40.) In my review of Plachetka, however, I saw no
`
`mention of an unexpected reduction in the incidence of NSAID-associated upper GI injury in
`
`patients taking the claimed unit dosage form.
`
`A. GRAHAM
`44.
`
`Graham reports the results of a clinical trial comparing the efficacy of misoprostol
`
`(a synthetic prostaglandin that acts as a cytoprotective agent) and lansoprazole (a proton pump
`
`
`
`Ex. 2021
`
`Page 10
`
`Page 12 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`inhibitor that suppresses gastric acid production) in gastric ulcer prevention in long-term users of
`
`NSAIDs.
`
`45.
`
`Graham enrolled 537 patients who were long-term NSAID users with a history of
`
`endoscopically-documented gastric ulcer and were negative for H. pylori. (Ex. 1005 at 2.) The
`
`patients were randomized to one of four treatment groups: placebo, misoprostol, lansoprazole
`
`(15mg), and lansoprazole (30 mg). (Id.) Graham notes that patients were permitted to take “low
`
`dose aspirin for cardiovascular protection.” (Id.) Graham makes no other mention of aspirin use
`
`in the paper, except in passing with regards to ‘balance.’ He never used this factor as a variable
`
`in any analysis.
`
`46.
`
`It is significant that Graham does not describe how low dose aspirin use was
`
`actually tracked or usage even verified. It is unknown exactly how patients were asked about
`
`low-dose aspirin use or how frequently they were asked about low-dose aspirin use. For
`
`example, it is unclear whether patients were on low-dose aspirin for the duration of the study, or
`
`if patients stopped (or started) low-dose aspirin after starting the trial. Because the paper is silent
`
`on this fact, it is not possible to know when or how the clinical trial participants were provided or
`
`took low-dose aspirin. It is also not possible to know the actual dosage of low-dose aspirin that
`
`the study participants actually took. Because of the lack of even the most elementary data on
`
`low-dose aspirin use, Graham did not use this factor in any analyses. It is reasonable to conclude
`
`that aspirin data were very limited, and that pill count data were never obtained to ensure that
`
`patients actually took the low-dose aspirin.
`
`47.
`
`It is also noteworthy that low dose aspirin use is not listed in Table 2 as a
`
`characteristic of the intent-to-treat population. We do not know if there is LDA balance or not
`
`between the groups, although such data confirmation would be necessary to exclude
`
`Ex. 2021
`
`Page 11
`
`
`
`Page 13 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`‘confounding’ when interpreting the trial’s data. A ‘confounding variable’ is a variable that has
`
`an effect on the outcome. For example, in a hypothetical study looking at the relationship
`
`between physical activity and obesity levels, potential confounding variables would be age and
`
`gender. If a confounding variable is distributed unequally among the groups being compared
`
`(i.e., one treatment arm has a significantly higher number of elderly participants), this may
`
`introduce bias or suggest an alternative explanation to an observed correlation. Thus, in clinical
`
`study design, it is imperative to identify all possible confounding variables at the outset. In my
`
`experience, if a specific patient characteristic could possibly make a difference in the outcome, it
`
`would be reported across treatment arms. This is not apparent in the Graham study. As shown
`
`in Table 2, Graham reports the patient characteristics that were balanced across treatment arms—
`
`but did not include any information regarding low-dose aspirin use. Because no data are
`
`available to evaluate whether low-dose aspirin use was evenly balanced between treatment
`
`groups at baseline, no scientifically valid conclusions about aspirin users can be drawn from this
`
`paper.
`
`48.
`
`I have reproduced Tables 1 and 2 below. From Table 2, it appears that the
`
`treatment groups were well balanced at baseline (the “enrolled” patients) for the characteristics
`
`that Graham identified at the outset (including smoking, alcohol use, previous treatment, etc.).
`
`Table 1 demonstrates, however, that the patients who withdrew from the trial were not evenly
`
`distributed across the four treatment groups. A higher percentage of patients withdrew from the
`
`placebo and most notably, in the misoprostol arms. Even the authors noted this, reporting that
`
`early withdrawals were significantly higher in the misoprostol group. (Ex. 1005 at 3.)
`
`49. My experience in this disease area, reflected in my publications and in the general
`
`area of clinical trials methodology, would indicate that differential early patient dropouts
`
`Ex. 2021
`
`Page 12
`
`
`
`Page 14 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`between groups represent a major problem with the study, even putting the issues relating to
`
`LDA usage aside. Usually, these patient dropouts are a special group precisely because they
`
`dropped out early, and as a consequence, they are not similar to the remaining randomized
`
`patients in the study. These different rates of patient withdrawals result in treatment groups that
`
`are no longer comparable as a consequence of the randomization process. The reason that the
`
`intent-to-treat population, defined as all subjects randomized, is the primary pre-specifed
`
`population is that it carries the most weight of evidence. This is because the randomization
`
`process distributes patients randomly among the four groups, which insures that there is no bias
`
`among the groups at baseline. Bias is a form of systematic error introduced into a clinical trial
`
`by selecting one outcome over others. Because the removal of randomized patients can
`
`introduce bias, it is my opinion that any comparison of the resulting treatment groups is suspect
`
`because the imbalances could explain the results as opposed to the treatment group.
`
`50.
`
`Graham does not report any comparative treatment data stratified by LDA use. In
`
`other words, he does not compare treatment groups within LDA users and LDA non-users
`
`separately. Given these significant analytical limitations of the study, a POSA could not have
`
`relied on the Graham study to draw any conclusion about the interaction between NSAIDs,
`
`proton pump inhibitors, and LDA usage. Further, a POSA could reach a false conclusion by
`
`relying on the Graham study. For example, the non-compliance rate led to significant issues in
`
`the interpretation of the results. Graham points out that, despite “the statistical advantage of
`
`misoprostol over proton pump inhibitors for the prevention of ulcer relapse in long-term users of
`
`NSAIDS, there is little, if any, practical advantage”. This is a speculation on the part of Graham,
`
`and not supported by the data and its analysis as he himself admits.
`
`
`
`Ex. 2021
`
`Page 13
`
`Page 15 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`
`
`
`
`51.
`
`I’ve reviewed the declaration of Petitioner’s expert, Leon Shargel, Ph.D. He
`
`made no attempt to review or assess the methodology of the Graham study, which is a
`
`fundamental starting point in evaluating scientific publications.
`
`
`
`Ex. 2021
`
`Page 14
`
`Page 16 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`B. GOLDSTEIN
`52.
`
`The Goldstein publication represents a post-hoc subanalysis using the same data
`
`reported in the Graham paper. Thus, Petitioners did not rely on two independent studies for
`
`confirmatory purposes. Instead, they relied on one study with two populations, with the
`
`Goldstein paper based on a very small subset of the entire intent-to-treat population in the
`
`Graham paper. This reduces the weight of any conclusions drawn from this paper.
`
`53.
`
`Drawing from the group of 535 enrolled patients, Goldstein reports on just 70
`
`patients (13%) who reported concomitant LDA usage at the beginning of the Graham study. As
`
`described above, there is no indication how these patients were identified, how low-dose aspirin
`
`usage was tracked or verified, or how or whether dosage was estimated.
`
`54.
`
`Goldstein concludes that for patients taking NSAIDs with LDA usage,
`
`administration of misoprostol or lansoprazole reduced the risk of gastric ulcer compared to
`
`placebo. Goldstein, however, does not describe any comparison between patients taking NSAID,
`
`lansoprazole without LDA usage with patients taking NSAID, lansoprazole, but no LDA. Even
`
`ignoring all of the shortcomings described below, Goldstein could not address the question now
`
`being asked.
`
`55.
`
`Goldstein reports the following baseline demographics:
`
`
`
`Ex. 2021
`
`Page 15
`
`Page 17 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`
`
`56.
`
`It is noteworthy that the size of the four treatment groups ranged from 7 to 25.
`
`These sample sizes are far too small to draw any meaningful conclusions. Prospectively
`
`randomized clinical trials typically have hundreds of patients in each treatment arm. The reasons
`
`for this are many, and include: 1) being able to make precise statements about the Type 1 error
`
`(i.e., the false positive rate) and the Type 2 error (i.e., the false negative rate), and 2) being able
`
`to obtain results that are ‘robust’ (i.e., a few changes in the outcome event will not have a
`
`material effect on the outcome and interpretation of the study). For example, if there was only
`
`one more male in the lansoprazole group, (i.e., 6 men and 1 woman, rather than the 5 men and 2
`
`women reported in Table 2) then the proportion of males would increase from 71.4% to 85.7%.
`
`This treatment group would have 23-36% more males than the other treatment groups, a marked
`
`imbalance that would have a significant detriment on any scientifically valid comparison of these
`
`groups. Thus, a small change to the very small sample sizes reported in Goldstein would have a
`
`significant impact on the apparent results. As a result, one would be unable to differentiate
`
`
`
`Ex. 2021
`
`Page 16
`
`Page 18 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`whether any observed between-group differences were due to treatment group as compared to
`
`differences in gender.
`
`57.
`
`In designing a study, there is the ‘effect size’ that represents the ‘true difference’
`
`of interest to detect. One also has the Type 1 error (i.e., the false positive rate, defined as the
`
`probability of seeing a difference between groups although there is in truth no difference) and the
`
`Type 2 error rate (i.e., the false negative rate, defined as the probability of seeing no difference
`
`between groups although one truly exists). Type 1 error rates are usually set at 0.05 (1 in 20
`
`chance of getting a falsely positive result), and Type 2 error rates are usually set at 0.20 or 0.10.
`
`Taking the n = 7 for the lansoprazole group and n = 22 for the placebo group, detection of
`
`between-group treatment differences of 10% would have a Type 1, false positive error rate
`
`exceeding 85% (with a Type 2 error rate of 20%). And for a Type 1 error rate of 5%, the Type 2
`
`error rate (false negative rate) exceeds 90%. The conclusion is that the exceedingly small sample
`
`sizes in Goldstein provide totally unreliable results and no scientifically valid conclusions can be
`
`drawn. The probability of reaching a false positive conclusion is in excess of 90% in my
`
`example, an unacceptably high level. The error rates fall far outside the norms of accepted
`
`clinical trial design and scientifically meaningful results. Typically, Type 1 error rates of 5% or
`
`less, and Type 2 error rates of 20% or less, are required.
`
`58.
`
`The Type 2 error rate is the probability of saying that a treatment difference does
`
`not exist, when the truth is that a difference exists between the groups. Such errors can arise for a
`
`number of reasons, including biologic heterogeneity and the small sample sizes compared to all
`
`people who are possible users. For example, we use aspirin when we have a headache. The
`
`dosage is 2 full dose aspirin pills. But not everyone gets headache relief from taking two aspirin.
`
`Some people need only one, some need two, some need three, and others may not get any
`
`Ex. 2021
`
`Page 17
`
`
`
`Page 19 of 52
`
`Patent Owner Ex. 2021
`CFAD v. Pozen
`IPR2015-01718
`
`
`
`
`
`headache relief. So if a trial was done with a population drawn that never benefited form aspirin,
`
`or had to take three pills for relief, then the trial would show no difference between the aspirin
`
`users and non-users because the trial population by chance had mostly non-responders to aspirin.
`
`But this would be a false negative result because we know that aspirin works most of the time, in
`
`most people. It was only the biological heterogeneity of the particular population selected that
`
`led to the negative trial outcome.
`
`59.
`
`For the reasons discussed above, the sample sizes confirm the presence of a
`
`differential dropout rate, indicative of a corresponding imbalance among the groups with respect
`
`to LDA usage. Until an analysis is performed to show this is not true, the correct presumption is
`
`that an imbalance exists due to the substantial differences in sample size between treatment
`
`grou