Review Article
`
`Effects of Nonsteroidal Anti-inflammatory
`Drugs on Endogenous Gastrointestinal
`Prostaglandins and Therapeutic Strategies
`for Prevention and Treatment of Nonsteroidal
`Anti-inflammatory Drug—Induced Damage
`
`Byron Cryer, MD, Mark Fefdman, MD
`
`0 Although nonsteroidal anti-inflammatory drugs (NSAIDs)
`are effective for pain relief and treatment of arthritis, they
`can induce gastric and duodenal ulcers and life-threatening
`complications. The mechanisms of their anti-inflammatory
`action and their gastroduodenal toxic effects are related, in
`part, to inhibition of prostaglandin synthesis. This review
`article discusses prostaglandins, their functions in the gas-
`trointestinal tract, anti-inflammatory actions of NSAIDs,
`and mechanisms by which NSAIDs produce gastroduodenal
`ulcers. Also reviewed are risk factors associated with the
`development at NSAlD-related ulcers and pharmacologic
`strategies for the prevention and treatment of NSAID-
`induced ulcers.
`(Arch intern Med. 1992;1 52:1145-1155)
`
`Nonsteroidal anti-inflammatory drugs (NSAIDs) are
`widely used for pain relief and for treatment of
`arthritis {including rheumatoid arthritis, ankylosing
`spondylitis, osteoarthritis, and gouty arthritis). A partial
`list of NSAJDS is shown in the Table. Although NSAIDs
`are effective as therapeutic agents, their major toxic effect
`is induction of gastroduodenal ulcers. Mechanisms for
`their anti-inflammatory action and their gastmduodenal
`toxic effects are probably related to an inhibition of pros-
`taglandin synthesis. The purposes of this article are to re-
`view the effects of NSAIDs on the gastroduodenal mu-
`cosa, including their effects on mucosal prostaglandins,
`and to review the effects of therapeutic agents that can be
`used to prevent and treat NSAID-induced gastroduode-
`nal damage.
`PROSTAGLAN DINS AND RELATED COMPOUNDS
`
`Prostaglandins (PCs) are a family of related fatty acids
`that are produced by nearly all of the body‘ 5 cells. Pros-
`taglandins participate in a variety of activities, including
`mediation of inflammatory responses, protection of the
`
`Accepted ior publication November 16, 1991.
`From the Medical Service, Department of Veterans Affairs Med-
`ical Center, and Department of Internal Medicine, University of
`Texas Southwestern Medical Center at Dallas.
`Reprint requests to Dallas VA Medical Center (111), 4500 S Lan-
`caster Rd, Dallas, TX 75216 (Dr Feldman).
`
`Arch Intern Med—Vol 152, June 1992
`
`gastrointestinal mucosa against injury, and regulation of
`renal blood flow. The general chemical structure of PGS
`is an oxygenated, Sill-carbon, unsaturated fatty acid (ei-
`cosanoid) composed of a five-carbon ring, with two car-
`bon side chains, one composed of seven carbon molecules
`and the other composed of eight.1 Nomenclature used to
`describe individual PCs is based on two distinguishing
`features. First, the letter designation of PCs (ie,
`their
`family) is determined by the structure of the five-carbon
`ring. For example, all PGEs have a double-bonded oxygen
`(= O) at carbon 9 and a hydroxyl group (— 01-1) at carbOn
`11, while all PGFS have a hydroxyl group at both carbon
`9 and carbon 11.2 Second, the number of double bonds in
`the side chains determines PG classification as 1-, 2-, or
`3-series and is reflected by a subscript (eg, PGE, [2-series]
`or 6'kEt0'PGF1m 11-series]) (Figure).
`Prostaglandins are not stored within cells in any signif—
`icant quantities, but are stored as precursor molecules.
`Prostaglandjns of the 2-series are the most plentiful and
`biologically important and are derived from arachidonic
`acid, a component of phospholipids present in all cell
`membranes. In response to a mechanical or chemical per-
`turbation of the cell membrane, arachidonic acid is
`released from membrane phospholipids into the cyto-
`plasm of the cell
`through the action of a plasma
`membrane-bound enzyme, phospholipase A,. Once re-
`leased, arachidonic acid may be acted on by cyclo-
`oxygenase, a membrane-bound enzyme,
`resulting in
`synthesis of PCs; alternatively, it may be metabolized by
`another enzyme, 5-lipoxygenase, to a group of closely re-
`lated compounds, the leukotrienes (LTs) (Figure). The
`relative
`activities
`of
`the
`cyclo-oxygenase
`and
`5-lipoxygenase pathways, and. thus the relative amounts
`of eicosanoids produced, vary with cell type.3 in gastric
`and duodenal mucosa, most arachidonic acid is converted
`into PGEZ, PGFZM and P312."r6
`FUNCTION OF PROSTAGLANDINS IN THE
`GASTROINTESTINAL TRACT
`
`Although PCs were first identified in the human body
`in the 19305, it was not until the mid-19605 that PGs were
`identified in the gastrointestinal tract?9 The earliest rec-
`ognized effect of PCs on gastric mucosal function was an
`Effects of NSAle on Prostaglandins—Cryer 8: Feldman
`1145
`
`Downloaded From: http:ffal'chinte.jamanetworkxomr‘ by a Reprints Desk User on 03f20f2015
`
`Page 1 of 11
`
`Patent Owner Ex. 2002
`
`Mylan v. Pozen
`|PR2017-01995
`
`
`
`
`Page 1 of 11
`
`Patent Owner Ex. 2002
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`Partial list or Nomteroidal Anti-inflammatory may
`Salicylates
`Aspirin'
`Diflunisal (Dolobid)‘
`Salsalate [Disalcidl‘
`lndoles
`Indomethacin {Indocinl‘
`Sulindac tCIinoril)‘
`Tolmetl‘n {Tolectinl‘
`Zomepirac [Zomaxi
`Pyrazoles
`Apazone (Rheumoxl
`Fep razo ne (Meth razo nel
`Phenvlbutazone lButazolidin)‘
`Fenamates
`Flufenarnic acid lMeralen]
`Mefenamic acid {Ponstell‘
`Meclotenamate (Meclomeni‘
`Tolfenamic acid (Clotarnl
`Proprionic acid derivatives
`Carprolen [Rimadyll
`Fenbufen (Cinopal, Lederfen)
`Fenoprot’en (Nalfon, Fenopronl'
`Flurbiproten lAnsaid, Frobenl‘
`Ibuprofen (Motrin, Advil)“
`Ketoprofen {Orudi5)’
`Naproxen lNaprosyn, Anaproxl‘
`Pirprofen tRengasil)
`Phenylacetic acid derivatives
`Diclofenac (Voltaren, Voltaroll’
`Fenclolenac (Flenac)
`Oxicams
`lsoxicam {Maxicaml
`Piroxicam (Feldenel"
`‘Available in the United States in 1991.
`
`
`
`inhibition of gastric acid and pepsin secretion. ““3 Intra-
`venously administered [’65 of the E, F, and A classes and
`orally administered synthetic analogues of these com-
`pounds have potent antisecretory effects, PCs of the E
`class being the most potent.‘“"
`In the 19705, investigators began to demonstrate that
`PGs could protect the gastric mucosa from injury and ul—
`ceration against a wide variety of damaging agents, such
`as alcohol, bile salts, acid, hypertonic saline, boiling wa-
`ter, stress, aspirin, and other NSAIDsF"23 Robert et al21
`performed the earliest of these experiments, in which
`they demonstrated that pretreatment with PGs could
`prevent mucosal damage from various noxious agents in
`rats. It was demonstrated that mucosal protection could
`be observed at doses of PCs that did not inhibit acid se-
`
`cretion.21 This protective property of PCs was called "cy-
`toprotection. "39 Even though pretreatment with PGS may
`protect against macroscopic injury, there is usually mi-
`croscopic evidence of mucosal injury to surface epithelial
`cells after exposure to alcohol or other noxious agents.30
`Because of persistent surface cell damage despite PG pre-
`treatment, the term cytopratection is not entirely accurate
`and has for the most part been replaced by mucosal protec-
`tion. Mucosal protection by prostaglandins has not only
`been demonstrated in the stomach, but has been shown
`in the duodenum.“‘” Protection has been demonstrated
`
`with PCs of all classes and is separate from any effects the
`compounds may have on gastric acid secretion. In fact, in
`animals, mucosal protection has been demonstrated with
`PGs, such as 6-keto-PGFW that have no demonstrated
`1146 Arch Intern Med—Vol152, June 1992
`
`
`
`Thromboxane A:
`Poo2
`
`Poll
`
`PGE2
`
`PGFZG
`
`Membrane Phosphoiipids
`l Phospholipase A2
`Arachidonic Acid
`5-Lipoxy3Wnlo-oxvgenase
`S-HPETE —> 5-HETE
`P662
`1
`l
`LTA4
`PGH2
`”th
`4
`l4
`”Di
`l
`|_l'lE4
`
`Thromboxane B2
`
`sze'tu-PGFHz
`
`Leukorrienes
`
`Thromboxanes
`
`Prostaglandil‘ls
`
`Metabolism of arachidonic acid after its release from membrane
`ph ospholipids. HPETE indicates hydroperoxyeicosatetraenoic acid;
`HETE, hydroxyeicosaretraenoic acid; PG, prostagiandin; and LT,
`ieukotn’ene.
`
`effect on acid secretion.31 However, in humans, it is not
`certain that the protective effects of PCs are due to mech-
`anisms separate from inhibition of gastric acid secretion,
`since [’65, at doses employed in human trials, have
`antisecretory effects as well.
`How is mucosal protection by [’05 mediated? Integrity
`of the gastroduodenal mucosa is maintained by a balance
`between aggressive factors, such as acid and pepsin, and
`protective factors, such as bicarbonate and mucus?”
`When there is an imbalance between aggressive and pro-
`tective factors, such that the extent of mucosal protection
`is lowered in relation to the level of offending agents,
`mucosal
`injury ensues. Persistence of this imbalance
`could lead to mucosal erosions and ulceration. Some of
`
`several putative mechanisms proposed through which
`PGs may provide their mucosal protective effects include
`the following: stimulation of mucosal bicarbonate secre-
`tion, mucus secretion, increased blood flow, prevention
`of disruption of the gastric mucosal barrier, acceleration
`of cell proliferation, stimulation of cellular ionic transport
`processes, stimulation of cyclic adenosine monophos-
`phate production, promotion of formadon of surface—
`active phospholipids, maintenance of gastric mucosal
`sulfhydryi compounds,
`stabilization of cellular lyso-
`somes, and stabilization of cell membranesfimn‘m" 5011
`et a1“ categorized various protective mechanisms accord-
`ing to their location with respect to the surface epithelial
`cells. They have been accordingly described as pres-pi-
`thelial (mucus and bicarbonate secretion), epithelial (sur-
`face epithelial cell continuity and migration), and postep-
`ithelial (mucosal blood flow).
`INFLAMMATION AND ANTI-INFLAMMATORY ACTIONS
`0F NSAIDS
`
`Inflammatory cell recruitment is achieved through the
`release of a number of chemical mediators, such as PCs,
`LTs, histamine, serotonin, kinins, complement factors,
`and other peptides.‘7v‘3 Evidence implicating PCs in this
`process was not obtained until 1971, when Vane“9 pro-
`
`Effects of NSAle on Prostaglandfins—Cryer & Feidman
`
`Downloaded From: hflpdfarchintejamanetworkxomi‘ by a Reprints Desk User on 031’201‘2015
`
`Page 2 of 11
`
`Patent Owner Ex. 2002
`
`Mylan v. Pozen
`|PR2017-01995
`
`
`
`
`Page 2 of 11
`
`Patent Owner Ex. 2002
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`posed PGs as substances that could elicit an inflammatory
`response. Prostaglandins were demonstrated to be asso-
`ciated with inflammation in a variety of experimental sit-
`uations. For example, after subcutaneous PG administra-
`tion, edema and erythema as well as some of the
`histologic changes of inflammation were observedfi“ Af-
`ter administration of aspirin, biosynthesis of PCS de-
`creased in proportion to the decrease in the amount of in-
`flammation,“ 115‘ and then, if exogenous PCs were later
`administered, there would be a return of inflammation.“
`Experimental administration of PCs could induce fever53
`and potentiate pain,“ and subsequent administration of
`an NSAID could decrease fever and pain while also
`decreasing PG concentrations. It soon became clear that
`the anti-inflammatory effects of such drugs as aspirin
`could be explained by their suppression of PG synthesis
`and that such inhibition could also explain the actions of
`these drugs as analgesics and antipyretics.
`A5pin'n, an acetylated salicylate, was one of the first
`NSAIDs shown to be clinically effective as an anti-
`inflammatory agent.ss Although many other NSAIDs
`have since been introduced, aspirin remains one of the
`most effective anti-inflammatory agents.55 It is through
`the inhibition of cyclo-oxygenase that aspirin and other
`NSAIDs decrease PG synthesis. By acetylation of cyclo-
`oxygenase, aspirin inhibits this enzyme irreversibly,
`while other NSAIDs (flufenamic acid,
`ibuprofen, and
`sulindac, for example) inhibit cyclo—oxygenase in a re-
`versible, concentration-dependent manner.“»57 When
`cyclo—oxygenase is irreversibly inhibited within any par-
`ticular cell, the capacity for PG Synthesis does not return
`to normal until new enzyme can be synthesized.“ This
`may explain why aspirin,
`in compariscm with other
`NSAIDS, remains one of the most potent inhibitors of PG
`synthesis. [t is hypothesized that cyclo-oxygenase exists
`inmultiple forms throughout the body and that each form
`has its own drug specificity,” although this has not yet
`been verified by identification of structural
`cyclo—
`oxygenase variants. Cyclo-oxygenases obtained from dif-
`ferent tissues have different sensitivities to inhibition by
`a particular NSAID, and different NSAle have variable
`abilities to inhibit a particular cyclo-oxygertase.57-58 For ex-
`ample, acetaminophen is as effective as aspirin in the in-
`hibition of brain cyclo-oxygenase, but is not nearly as ef-
`fective as aspirin in the inhibition of cyclo-oxygenase from
`some peripheral sites.“ This may explain why acetami-
`nophen is an effective centrally acting antipyretic and an-
`algesic but is not an effective peripherally acting anti-
`inflammatory agent. This may also explain why
`acetaminophen does not cause gastroduodenal toxic ef-
`fects.
`
`The LTs also play a significant role in the inflammatory
`response. They increase vascular permeability,
`are
`chemotactic for neutrOphils, vasoc0nstrict arteries, stim-
`ulate bronchial wall constriction and mucus secretion,
`and increase intestinal inflammatioriflw Certain NSAIDs,
`in addition to inhibitingcyclo-oxygenase, also may inhibit
`5-lipoxygenase.5"5‘ The NSAIDs differ in their relative
`potencies to reduce inflammation,"2 and their anti-
`inflammatory effects do not always correlate with their
`ability to reduce PG synthesis. These observations may
`possibly be explained by different capacities of the various
`NSAle to inhibit cyclo-oxygenase, on the one hand, and
`5-lipoxygenase, on the other hand. An NSAID such as
`indomethacin is predominantly a cyclo-oxygenase inhib-
`Arch intern Med—Vol152,lune1992
`
`itor, while other experimental NSAIDs of the fenamate
`class are effective inhibitors of both enzymes."3 Whether
`differences in the relative amounts of cyclosoxygenasefi—
`lipoxygenase inhibition by NSAIDS is indeed related to
`differences in anti-inflammatory actions of NSAle is
`currently under investigation.
`Anti-inflammatory actions of NSAIDs are not only ex-
`plained by inhibition of eicosanoid synthesis. For exam-
`ple, NSAIDs inhibit PG synthesis in vivo and in vitro at
`concentrations much lower than those required to achieve
`anti-inflammatory effects.“ Moreover, some salicylates,
`including nonacetylated salicylates, are beneficial in in-
`flammatory diseasefl'w even though they do not inhibit
`PG synthesis?“9 Inhibition of neutrophil function has
`been suggested as a second mechanism by which NSAIDs
`can exert their anti-inflammatory el’fectsf'zrm71
`MECHANISMS OF GASTRODUODENA]. MUCOSAL
`INJURY BY NSAle
`
`The mechanisms by which aspirin can cause gas-
`trointestinal mucosa] damage can be grouped into two
`categories: those independent of and those dependent on
`cyclo—oxygenase inhibition. Within a few minutes of
`aspirin ingestion, denudation of surface epithelial cells
`and increased mucosa] permeability to sodium (Na+) and
`hydrogen (PP) ions can be observed,” reflected experi-
`mentally as a decrease in transmucosal potential differ-
`ence.73'7‘ Salicylic acid, the deacetylated metabolite of as-
`pirin, does not inhibit cyclo-oxygenase activity in the
`gastric mucosa,”5 yet it reduces transmucosal potential
`difference as much as aspirin does.”3 Thus, surface
`epithelial cell disruption and a decline in potential differ-
`ence are not dependent on cyclo-oxygenase inhibition,
`and epithelial cell disruption is not prevented by pre-
`treatment with PCs."
`Endoscopic observation of the gastric mucosa after 1 to
`2 weeks of enteric-coated aspirin therapy}??-78 or after 1
`week of emetic-coated naproxen therapy 9 revealed con-
`siderably less gastric mucosal damage than with plain,
`non—enteric-coated formulations. Although gastric injury
`from a topical effect is decreased with emetic-coated for-
`mulations, their use on a long-term basis will reSult in
`gastric ulcers (GU5),"” presumably the result of a systemic
`rather than topical effect. Gastric ulcers can be produced
`experimentally after NSAIDs are administered intrave-
`nouslym or by rectal suppository“ and without a change
`in gastric transmucosal potential difference.“33 It is likely
`that the NSAIDS were ulcerogenic because they reduced
`mucosal PG synthesis. This is supported by two observa—
`tions: (1) small nonantisecretory doses of exogenous PGs
`prevent NSAID-induced ulcersm‘ii‘is and (2) depletion of
`mucosal I’Gs by another mechanism, active or passive
`immunization with PG antibodies, leads to GUs.
`Although inhibition of PG synthesis contributes to
`NSAID-induoed mucosal injury, it is not. settled whether
`PG inhibition is the primary mechanism. In some studies,
`there has been poor correlation between gastric mucosal
`injury and PG suppression after NSAIDs.mg Other fac-
`tors probably work in combination with PG suppression
`to increase the propensity for mucosal injury by NSAle.
`For example, after indomethacln administration, gastric
`acid secretion has been shown to increase,” gastric mu-
`cosal blood flow to decrease,‘3c"91 and duodenal bicarbon-
`ate output to decrease.” Nonsteroidal anti-inflammatory
`drugs can also potentially affect mucus secretion, as
`Effects of NSAIDs on Frostagland'ins—Cryer 8.. Feldman
`1147
`
`Downloaded From: lrfl'pmlarchintejamanetworkcomi‘ by a Reprints Desk User on 03f201‘2015
`
`Page 3 of 11
`
`Patent Owner Ex. 2002
`
`Mylan v. Pozen
`|PR2017-01995
`
`
`
`
`Page 3 of 11
`
`Patent Owner Ex. 2002
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`to
`they have been shown to inhibit mucus synthesis,
`reduce incorporation of radiolabeled precursors into
`mucus glycoprotein, and to alter thickness of the mu-
`cus layer.2m
`It has been hypothesized that, as a consequence of
`cyclo-oxygenase inhibition, arachidonic acid metabolism
`could alternatively be shunted toward the lipoxygenase
`pathway, resulting in increased LT synthesis?“ The
`postulated mechanism by which increased activity of the
`5-1ipoxygenase pathway could enhance mucosa] injury is
`by LT-mediated vasoconstriction or by direct vascular in-
`jury by oxygen radicals produced in this pathway.”99 The
`relative importance of LTs in NSAlD-induced gastric mu-
`cosa] damage is still unclear.
`Since other pathogenetic mechanisms are potentially
`operative, one may ask whether significant NSAID-
`related mucosa] injury can occur in the absence of
`suppression of mucosal PGs. After administration of
`salsalate,
`a nonacetylated salicylate
`that
`is
`anti-
`inflammatory, mucosa] injury has been far less than after
`other l"~ISAIDs.1W"12 Salsalate does not significantly inhibit
`cyclo-oxygenase activity or reduce mucosa] PG con-
`tentw'm Thus, inhibition of PG synthesis is probably
`necessary but not sufficient for mucosa] injury.
`
`SHORT-TERM V5 lONG-TERM NSAID ADMINISTRATION
`
`After short-term administration, a variety of types of
`injury develop, ranging from petechia] hemorrhages. dif-
`fuse hemorrhages, superficial erosions, and, less com-
`monly, ulceration“ On the basis of such observations,
`many claims have been made as to the superiority of one
`NSAID over another regarding the incidence of mucosa]
`injury. Lanza112 reported the largest Experience with en-
`doscopic mucosa] observations after 7 days of NSAID in-
`gestion. High doses of aspirin had the highest incidence
`of acute gastric mucosa] injury, while the incidence of in-
`jury induced by other nonaspirin NSAIDs was less but
`also dose dependent. Among the nonaspirin NSAl'Ds. it
`was not easy to compare incidences of gastric mucosa]
`toxic effects because of difficulties in determining equiv-
`alent doses. By compiling all of his NSAID data, Lanza
`observed a 6.7% incidence of GU and a 1.4% incidence of
`duodenal ulcer (DU) after 1 week of NSAlD ingestion. The
`largest numbers of GUs were produced by aspirin and the
`lowest numbers by lower anti-inflammatory doses of ibu-
`profen.
`The evolution of mucosal injury over time after shutt-
`term NSAID therapy also has been an interest of investi-
`gation. After a single dose of aspirin (650 mg), gastric in-
`tramucosa] hemorrhages
`endoscopically visible
`as
`petechiae appear in as little as 15 minutes and gastric ero-
`sions in as little as 45 minutes. 11" Petechia] lesions become
`most pronounced by 1 to 2 hours1101“ and can occur inany
`locationin the stomach. 11” “4 After many repeated doses
`of aspirin, multiple erosions appear, mostly1n the an-
`h'um,su'106'11° but potentially in any gastric location. Endo-
`scopic gastric mucosal injury peaks within the first 3 days
`and then tends to decrease despite continued aspirin ad-
`ministration,“°“3”4 dempite the fact that mucosa] PG
`content remains low. 35 This phenomenon has been re-
`ferred to as gastric adaptation. “4 Increased epithelial cell
`regeneration and mitoses have been observed to occur in
`response to aspirin-induced injury.“5'118
`“References ?7-79, 82, 85, 8?, 88, 101-114.
`
`1148 Arch Intern Med—Vol 152, lune 1992
`
`Mucosal petechiae and erosions are comparatively triv-
`ial, transient lesions that have low risk for major unto-
`ward effects.“9 Acute mucosa] injury can be repaired rap-
`idly through processes of
`restitution and gastric
`adaptation. With continued and frequent aspirin admin-
`istration, the rate of mucosa] injury may be greater than
`the rate of mucosal repair, ultimately resulting in a
`persistent epithelial defectm Consequently, an erosion or
`an ulcer may develop, the distinction between the two
`being depth of damage.“1 An ulcer, once formed, has the
`potential to cause significant bleeding, lumina] obstruc-
`tion, or gastrointestinal perforation, all of which are
`not uncommon complications of
`long-term NSAID
`therapy?“23 Thus, the clinically important aspects of
`NSAID mucosa] damage are primarily the consequences
`seen after long—term rather than short-term therapy
`Although there may be considerable differences in in-
`cidences of injury after short-term administration ob-
`served between the various nonaspirin NSAle, these
`differences cannot be used to predict injury after longer-
`term administration. Drugs that produce slight acute
`muCOsa] injury can still produce ulcers when given on a
`long-term basis. For example, sulindac produces little
`mucosa] damage when given for a short termm but is as-
`sociated with one of the highest rates of NSAl'D—related
`upper gastrointestinal bleeding.122 Most data on conse-
`quences of long-term NSAlD therapy come from epide-
`miologic studies or from prospective trials of patients
`taking these medications for therapy for chronic rheu-
`matic diseases.
`
`Retrospective reviews of records of hospital admissions
`for upper gastrointestinal bleeding have provided further
`evidence that long-term aspirin use is associated with
`GUs.123 With the newer, nonaspirin NSAIDs, case-
`controlled studies also suggest that gastrointestinal bleed-
`ing from ulcers is strongly associated with NSAID
`usem-‘z‘t'31 However, the incidence of serious ulcer com-
`plications with nonaspirin NSAIDs is less than that with
`aspirin. Patients presenting with bleeding ulcers are three
`to five times as likely as controls to have taken an NSAID,
`and 13% to 60% have a recent history of NSAID
`consumption.u""3‘ Among subjects without a history of
`ulcer, patients taking NSAle have 1.5 times the risk of
`developing upper gastrointestinal bleeding than do con-
`trols not taking NSAIDs.127 A dose-response relationship
`between NSAID consumption and development of mu-
`cosa] ulcers may also exist. Cari-teron132 found that a pat-
`tern of regular aspirin consumption (>15 aspirin tablets
`per week) had a significantly higher association with GUs
`than patterns of occasional (14 or less per week) or no as-
`pirin consumption On the basis of the distribution of as-
`pirin use among these patients, it has been estimated that
`the relative risk of developing a GU rises dramatically
`above 15 to 20 aspirin tablets per week at an aspirin dose
`of 325 mgm
`Data on long-term mucosa] effects of NSAID consump-
`tion come mostly from endoscopic studies of patients with
`rheumatoid arthritis or osteoarthritis.‘i‘un'fl‘m‘s McCar-
`thy,”S by combining data from all available point preva-
`lence studies, estimated a GU oint prevalence of 13%
`and a DU point prevalence 0? 11% for patients with
`arthritis taking long-term NSAID therapy. Enteric-coated
`aspirin appears to be associated with fewer GUs than
`plain aspirin.“ However, incidences of DUs after use of
`either aspirin preparation are similar-.13“
`Effects of NSNDS on Prostagiano'ins—Cryer 5: Feldman
`
`Downloaded From: httpdiarchintedamauetworkcumi by a Reprints Desk User on 03f20i2015
`
`Page 4 of 11
`
`Patent Owner Ex. 2002
`
`Mylan v. Pozen
`|PR2017-01995
`
`
`
`
`Page 4 of 11
`
`Patent Owner Ex. 2002
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`Prospective, point-prevalence trials are limited by the
`fact that they look at the mucosa at only one point in time,
`after variable lengths of NSAID use.
`It
`is not certain
`whether the observed ulcer is truly a direct consequence
`of the NSAID or whether it was present before NSAID
`therapy began. To assess the risk of ulcer formation
`directly attributable to NSAIDS, a lesion-free mucosa
`needs to be observed at a zero time point, and the
`incidence of ulcers arising while the patient is taking
`NSAIDs as compared with placebo treatment is then re-
`corded. Caruso and Bianchi-I—‘orro‘33 observed new gastric
`lesions in 31% of patients after 3 months of NSAIDs. The
`incidence of ulcers among these "lesions" was not re-
`ported, and there was no placebo-treated group for com-
`parison. An alternative means to study the evolution of
`mucosal damage in long-term NSAID users is to use data
`from placebo-controlled trials of protective agents coad-
`ministered with NSAIDs. To date, there have been four
`large (ie, >100 subjects each] trials in which either a his-
`tamine, (Hz) blocker, a synthetic PG, or placebo was
`coadrninistered with one of various NSAIDs to patients
`with arthritis who were without mucosal abnormalities at
`initial endoscopyfimm Again, there was no group of pa-
`tients with arthritis who received placebo without
`NSAIDs. Nevertheless, it appears that, at least after 2
`months of NSAID therapy, a new GU may develop in a
`little greater than 10% of NSAID users, and a DU will de-
`velop in somewhere less than 10%. It is likely that gastric
`and duodenal ulceration with NSAID usage beyond 2 to
`3 months will continue to occur, since NSAlD-related ul-
`cer complications, such as bleeding or perforation, occur
`frequently in long-term NSAID users.m-13"“'
`RISK FACTORS FOR NSAID—INDUCED ULCERS
`Dose
`
`As the prescribed dose of an NSAI'D increases, the per-
`centage of patients presenting with upper gastrointestinal
`bleeding or hospitalized for ulcers increases.mm Griffin
`et al’“ recently reported that the relative risk of ulceration
`in older subjects who have consumed NSAIDs for less
`than 30 days is almost twice the risk for longer periods of
`consumption. ‘31 The authors stated that the estimated risk
`for development of an ulcer among an elderly individual
`who has recently begun a high dose of an NSAID is 10
`times that of a nonuser.‘3' Prospective data directly eval-
`uating dose—response or duration-reSponse relationships
`between long-term NSAID use and ulcer development are
`lacking.
`
`Ulcer History
`A history of idiopathic ulcer disease may increase the
`risk of ulceration during NSAID therapy. After 2 months
`of NSAID consumption, six of ‘11 patients with a history
`of peptic ulcers developed recurrent ulceration, compared
`with only 11 of 115 patients with no ulcer history.“3 More
`studies of this risk factor are required before previous ul»
`cer disease can be accepted as a definite risk factor.
`
`Age
`Age is one factor that has been consistently associated
`with an increased risk for NSAID-related ulcer complica-
`fions.‘3“"i5’mvi31 The risk of perforated ulcers may be high
`in elderly NSAID users, especially elderly women)“ and
`mortality from ulcer complications is also markedly ele-
`vated in the aged.“ One likely explanation for this asso-
`Arch Intern Med—Vol 152, june 1992
`
`ciation of greater age and risk of NSAID ulcerations is that
`NSAID use increases with advancing age, especially in
`those over 60 years old.‘2“'125 However, there may be other
`factors that predispose the elderly to damage by NSAIDs.
`For example, our group and a group from Ia an recently
`showed that both gastric“:-“3 and duodenal“ mucosa] PG
`concentrations decline with aging in humans. Thus, older
`patients, at baseline, may have an already compromised
`potential for mucosa] protection, perhaps placing this
`group at high risk for the development of NSAlD-induced
`ulcers.
`
`Smoking
`It is not known whether cigarette smoking influences
`the potential for NSAlD—induced ulceration. The ability of
`the gastroduodenal mucosa to protect itself against injury
`may be decreased in smokers, since smoking is associated
`with reduced mucosa] PG concentrations in humans.“145
`Use of NSAIDs by smokers should further reduce their
`already low mucosa] PG concentrations.
`PREVENTION OF NSAID—INDUCED ULCERS
`
`Initial attempts to lower gastroduodenal toxic effects
`seen with aspirin were directed toward development of
`alternative formulations. Newer NSAIDs, enteric—coated
`preparations, suppositories, and prodrugs disappoint-
`ingly continue to be associated with significant ulceration.
`None has demonstrated conclusive superiority to the
`others for decreased gastroduodenal toxic effects. Conse-
`quently, a major research interest has arisen to investigate
`other drugs that, when coadministered with NSAIDs, will
`either protect against or prevent mucosa] injury.
`Evaluation of the efficacy of a coadministered agent to
`prevent mucosa] damage is strongly influenced by the
`type of scale used to measure injury. Mucosal protection
`may or may not be observed, depending on which pattern
`of injury has been most heavily weighted in the scoring
`system. In a study of prevention of naproxen-induced
`acute gastroduodenal
`injury, cimetidine was demon-
`strated to be superior to placebo when a scale primarily
`reflective of mucosa] hemorrhage was used, but cimeti-
`dine was not different from placebo when a scale in which
`erosions were incorporated into the scoring system was
`used. 1“ Results of cotreatrnent trials are more reliably ap-
`plied to clinical practice when erosions and ulcers are used
`as end points to define response to therapy. Here again,
`findings of the short-term trials may not be relevant to
`long-term administration and, thus, the weight of our
`conclusions should be based on results of trials of ex-
`tended cotherapy in the long-term NSAID user.
`
`Hz-Reoeptor Antagonists
`it has become common practice to prescribe H2-
`antagonists, such as cimetidine (Tagamet), ranitidine
`(Zantac),
`famotidine (Pepcid), and nizatidine (Axid),
`along with NSAle for ulcer prophylaxis, even though
`supporting evidence from clinical trials is sparse. None—
`theless, coadministration of an antisecretory agent does,
`for the following reasons, have some theoretical merit: (1)
`after mucosa]
`integrity has been interrupted by an
`NSAID, further cellular damage can occur through the
`back diffusion of acid; (2) during NSAID therapy, acid se-
`cretion may increase?” possibly because of decreased
`mucosa] PG content; and (3) in animals, NSAID mucosa]
`damage in the presence of acid is greater than when mu-
`Effects of ~511le on Prostagl'andins—Cryer 8: Feldman
`1149
`
`Downloaded From: httpflarchintejamauetworkxomi‘ by u Reprints Desk User on flJiZOJ‘ZfllS
`
`Page 5 of 11
`
`Patent Owner Ex. 2002
`
`Mylan v. Pozen
`|PR2017-01995
`
`
`
`
`Page 5 of 11
`
`Patent Owner Ex. 2002
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`cosa is exposed to a higher pH.” On the other hand, the
`fact that Ids-antagonists have no known effects on gastric
`mucosal PGs theoretically argues against their possible
`benefit in a PG-deficient mucoea.““5°
`Coadmirrjstration of [dz-antagonists and aspirin, dosed
`for 7 days or less, to normal volunteers produced less
`gastric and duodenal mucosal injury than did adminis»
`tration of placebo. 51"” Thus,
`in the short term, H2-
`antagonists are effective for prophylaxis against gastric
`and duodenal NSAID-induced injury. During longer pe-
`riods of NSAID administration, til—antagonists, such as
`ranitidine (150 mg twice daily), are effective for NSAID-
`induced ulcer prevention in the duodenum but not the
`stomachJW’"
`There is limited information on ulcer prevention after
`an NSAID-induced ulcer has been healed by an H;-
`antagonist, assuming that coadministration of an H:-
`antagonist and the NSAID is continued.”“‘155 Available
`data suggest that recurrence rates during cotherapy are
`low.136 In these maintenance studies, however, repeated
`endoscopy was only performed fo