The Journal of Rheumatology
`
`Volume 29, no. 3
`
`Gastroprotective therapy and risk of gastrointestinal ulcers: risk reduction by
`COX-2 therapy.
`
`Frederick Wolfe, Janice Anderson, Thomas A Burke, Lester M Arguelles and Dan Pettitt
`
`J Rheumatol 2002;29;467-473
` http://www.jrheum.org/content/29/3/467
`
`
`
`1. Sign up for our monthly e-table of contents
`
`http://www.jrheum.org/cgi/alerts/etoc
`
`2. Information on Subscriptions
` http://jrheum.com/subscribe.html
`
`
`
`3. Have us contact your library about access options
`
` Refer_your_library@jrheum.com
`
`4. Information on permissions/orders of reprints
`
` http://jrheum.com/reprints.html
`
`The Journal of Rheumatology
` is a monthly international serial edited by Earl D.
`Silverman featuring research articles on clinical subjects from scientists working in
`rheumatology and related fields.
`
`
`Downloaded from Downloaded from
`
`
`RheumatologyRheumatology
`
`
`
`
`www.jrheum.orgwww.jrheum.org The Journal of on March 24, 2015 - Published by The Journal of on March 24, 2015 - Published by
`
`
`
`
`
`
`
`
`
`Page 1 of 8
`
`Patent Owner Ex. 2061
`Mylan v. Pozen
`IPR2017-01995
`
`   
`   
`   
`   
`

`

`Gastroprotective Therapy and Risk of Gastrointestinal
`Ulcers: Risk Reduction by COX-2 Therapy
`
`FREDERICK WOLFE, JANICE ANDERSON, THOMAS A. BURKE, LESTER M. ARGUELLES, and DAN PETTITT
`
`ABSTRACT. Objective. Proton pump inhibitors (PPI) and misoprostol decrease the risk of development of nons-
`teroidal antiinflammatory drug induced gastric ulcers and aid healing of upper gastrointestinal (GI)
`ulcers. H2 receptor antagonists (H2RA) are less effective for this task, but are widely used by
`patients and physicians for the treatment of GI symptoms and duodenal ulcers. Sucralfate is a
`weaker agent that is sometimes used for prophylaxis or treatment of upper GI ulcers. We investi-
`gated the effect of GI drugs and selective and nonselective NSAID on the incidence of GI ulcer
`development in a cohort of patients immediately after the release of celecoxib and rofecoxib to
`investigate the effect of confounding by indication when effective GI agents and cyclooxygenase 2
`(COX-2)-specific inhibitors are prescribed to a high risk population.
`Methods. During a 6 month period of observation 8547 NSAID users were evaluated by mailed
`questionnaire concerning NSAID drug use and ulcer development. In the first half of 1999, patients
`took 12,177 separate NSAID courses. GI therapy that followed the development of upper GI ulcers
`was excluded from analysis. Ulcer reports were confirmed by followup validation.
`Results. GI drugs were used concomitantly in this population by 42% of patients using an NSAID.
`GI drugs were associated with an increased risk of ulcer. But this risk was confined to PPI (OR 4.1,
`95% CI 2.95, 5.69), and not to other GI drugs. Overall, patients using nonselective NSAID
`compared to those taking COX-2-specific inhibitors had an increased risk of upper GI ulcers (OR
`2.12, 95% CI 1.43, 3.34). Patients taking nonselective NSAID plus PPI were also at increased risk
`for upper GI ulcers compared to those taking nonselective NSAID alone (OR 5.09, 95% CI 3.88,
`6.67). Similarly, the risk of upper GI ulcers was increased in the nonselective NSAID plus PPI group
`(OR 3.83, 95% CI 2.32, 6.31) compared to the COX-2 plus PPI group.
`Conclusion. PPI use, but not other GI drug use, is a marker for increased susceptibility to ulcers
`among NSAID users. This risk of upper GI ulcers is increased in PPI users regardless of which
`NSAID is used (nonselective or COX-2-specific inhibitor). Although COX-2 use is associated with
`greater risk factors for upper GI ulcers due to channeling bias, COX-2 users have significantly fewer
`ulcers than equivalent nonselective NSAID users regardless of concomitant PPI utilization.
`(J Rheumatol 2002;29:467–73)
`
`Key Indexing Terms:
`UPPER GASTROINTESTINAL ULCERS
`CYCLOOXYGENASE-2
`PROTON PUMP INHIBITORS H2 BLOCKERS
`RISK
`
`Nonsteroidal antiinflammatory drugs (NSAID) are associ-
`ated with the development of gastrointestinal (GI) ulcers1-9.
`Various GI therapies have been employed by physicians and
`patients for the prophylaxis and treatment of upper GI ulcers
`and symptoms in patients using NSAID, including hista-
`
`From the National Data Bank for Rheumatic Diseases – Arthritis
`Research Center Foundation and University of Kansas School of
`Medicine, Wichita, Kansas; Pharmacia Global Health Outcomes,
`Peapack, New Jersey; University of Illinois at Chicago, Chicago, Illinois;
`and Pfizer Outcomes Research, New York, New York, USA.
`Supported by a grant from Pharmacia Global Health Outcomes and
`Pfizer Outcomes Research.
`F. Wolfe, MD; J. Anderson, BA, National Data Bank for Rheumatic
`Diseases – Arthritis Research Center Foundation and University of
`Kansas School of Medicine; T.A. Burke, Pharm D, Pharmacia Global
`Health Outcomes; L.M. Arguelles, MS, University of Illinois at Chicago;
`D. Pettitt, DVM, MSc, Pfizer Outcomes Research.
`Address correspondence to Dr. F. Wolfe, Arthritis Research Center
`Foundation, 1035 North Emporia, Suite 230, Wichita, KS 67214.
`E-mail: fwolfe@arthritis-research.org
`Submitted March 27, 2001; revision accepted September 17, 2001.
`
`mine receptor antagonists (H2RA), proton pump inhibitors
`(PPI), barrier agents, and prostaglandin analogs. Overall, the
`prevalence of GI therapy among patients taking prescription
`NSAID has been reported to range from 26% in Canada to
`24–34% in the US10,11. H2RA and PPI are the most
`commonly used gastroprotective agents. In addition to
`NSAID, additional risk factors for gastroduodenal ulcers
`have been identified. These factors include age, history of
`previous upper GI ulcers, GI symptoms, decreased func-
`tional ability, corticosteroid and oral anticoagulant use, and
`heart disease, among others8,12-21. At least 2 types of risk
`factors have been identified and should be considered. The
`first is biologically based and is related to toxic effects of
`drugs or to host susceptibility. The use of NSAID and corti-
`costeroids and a history of previous ulcers are examples of
`this first type of risk factor.
`The second type of risk factors are confounder effects, as
`opposed to causal risk factors. One such confounder is the
`use of drugs to treat or prevent upper GI ulcers. This is para-
`
`Personal non-commercial use only. The Journal of Rheumatology Copyright © 2002. All rights reserved.
`
`The Journal of on March 24, 2015 - Published by
`www.jrheum.org
`Downloaded from
`Wolfe, et al: GI drugs and ulcers
`
`Rheumatology
`
`467
`
`Page 2 of 8
`
`Patent Owner Ex. 2061
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`doxical, of course, for treatments with misoprostol, PPI, and
`high dose H2RA have been shown to reduce the risk of
`upper GI ulcers in NSAID users22-26. Martin, et al studied
`19,087 patients in England who were prescribed meloxicam
`between December 1996 and March 1997, and inquired
`about adverse events experienced within 6 months of the
`first meloxicam prescription27. Patients receiving gastropro-
`tective agents had an increased rate ratio for peptic ulcers
`(2.9, 95% CI 1.0, 8.4) compared to those who were not. The
`definition of gastroprotective agents in this study included
`PPI, H2RA, and misoprostol. Singh and Ramey reported on
`1921 patients with rheumatoid arthritis (RA), among whom
`H2RA, sucralfate, or antacids were used by 34%28. They
`found no reduction in the risk of GI events by the use of
`these drugs, but suggested that “symptomatic patients
`started on antacid or H2 antagonist therapy have a higher
`risk of serious GI complications compared with those who
`did not take these medications.” They did not provide rates
`or confidence intervals, and they did not study PPI.
`Considering short and longterm use, PPI are the most effec-
`tive drugs in the prevention and treatment of NSAID
`induced ulcers, and along with misoprostol are the most
`effective cotherapy for the prevention of NSAID induced
`ulcer22,23,29.
`Confounders, such as GI drugs, are of particular interest
`because they help us understand factors that might make
`effective treatments appear ineffective. In addition,
`confounders can be used to stratify patients by their risk
`profile before examining the effect of biologically based
`risk factors. We recently studied channeling bias and
`confounding by indication following the introduction of
`celecoxib and rofecoxib. We showed that patients switched
`to COX-2-specific inhibitors had a history of more severe
`rheumatic symptoms, lifetime GI adverse events, and GI
`drug utilization at the time of switch compared to those who
`inhibitors30.
`were not switched
`to COX-2-specific
`Channeling bias is a form of allocation bias, and occurs
`when drugs with similar therapeutic indications are
`prescribed to groups of patients with prognostic differ-
`ences31,32. For example, in the early days of methotrexate
`(MTX) usage, MTX was prescribed to RA patients with the
`worst prognosis. Although MTX improved such patients,
`the underlying severity of their illnesses outweighed the
`effectiveness of MTX; MTX appeared not to work well and
`was a marker for poor outcome. Channeling may lead to
`another form of bias, confounding by indication33-36. This
`occurs when the indication for the drug prescription results
`in preferential identification of the patients with the condi-
`tion and, at the same time, increases the risk of the outcome
`under study.
`We investigated the association of upper GI ulcers with
`GI drugs and the interaction of GI drugs with COX-2 and
`nonspecific NSAID in the development of upper GI ulcers
`in 8547 patients with arthritis during 12,177 courses of
`
`therapy. We found that PPI, but not other GI drugs, are a risk
`marker for upper GI ulcers, and that COX-2-specific
`inhibitors reduce the risk of upper GI ulcers in those
`receiving and not receiving GI drugs.
`
`MATERIALS AND METHODS
`Study population. Patients in this study are participants in the National Data
`Bank for Rheumatic Diseases and were enrolled by 581 US rheumatolo-
`gists37. In this project 1342 patients were recruited from the practices of US
`rheumatologists during a 30 day enrollment period38; 3760 were enrolled
`from community rheumatologists who made their patient populations avail-
`able to us; 1759 were enrolled at the time they were prescribed leflunomide
`by community rheumatologists as part of their ordinary medical care; and
`1686 were patients followed in the Wichita data bank. The characteristics
`of the Wichita data bank have been described39-41. Patients in this study
`were 8547 patients with arthritis, including 6375 with RA, and 2172 with
`fibromyalgia (FM) or osteoarthritis (OA) who were participating in the data
`bank surveys. Diagnoses were made by the referring rheumatologists. The
`survey period covered January 1999 through June 1999, after the introduc-
`tion of celecoxib and in part after the introduction of rofecoxib to the US
`market.
`Demographic and clinical data. In the survey, patients were asked to list all
`drugs used during the study period. In addition, they listed the start and stop
`dates of drugs, and the specific side effects attributed to each drug, if any.
`Doses were recorded for all NSAID, but not for GI drugs. For the purposes
`of this study, upper GI ulcers were ulcers reported by patients as side effects
`to a specific medication. Each ulcer was subject to followup validation in
`which patients were contacted and supporting medical records were
`obtained. We were able to confirm the patient self-report in ~95% of cases
`by hospital, endoscopy, and physician report records. We found no
`instances in which the records refuted the patient self-report.
`Although the terminology associated with GI drugs differs among
`studies, in this report we define gastroprotective agents to include sucral-
`fate and misoprostol and separate these drugs from H2RA and PPI. We
`combined sucralfate and misoprostol because of their infrequent use,
`despite evidence of a superior efficacy of misoprostol42. The effect of these
`2 drugs on ulcer prevention is entirely different.
`As part of the survey assessment, demographic and utilization variables
`were collected. Study variables also included the Stanford Health
`Assessment Questionnaire functional disability index (HAQ disability)43,44,
`a visual analog scale (VAS) for pain, a VAS for global disease severity, and
`the SF-36 mental and physical component scales (MCS and PCS)45. The
`MCS and PCS scores have a range of 0 to 100 and were designed to have
`a mean score of 50 and a standard deviation of 10 in a representative
`sample of the US population. Scores above 50 represent better than average
`health. Additional details and normative data are available46. Except for the
`SF-36 scores, where higher scores mean better health, higher scores repre-
`sent worse health/more symptoms.
`Statistical methods. GI therapy that followed the development of upper GI
`ulcers was excluded from analysis. Because Arthrotec (diclofenac plus
`misoprostol) includes a GI protective agent, this drug was excluded from
`analysis, excepted as described in the text.
`In regression analyses in Table 6, adjustment was made for NSAID
`dose after transforming reported doses into a proportion of the recom-
`mended or usual drug dose.
`Data were analyzed using logistic regression. The rates of ulcers and GI
`therapy among the 4 referring sources were similar, and it was judged
`appropriate to pool the data sources for analysis. In these analyses, we
`adjusted for within-patient clustering using the Huber/White/sandwich esti-
`mator of variance since each patient contributed 6 units of observation for
`each month of followup. Clustering specifies that the observations are inde-
`pendent across groups (clusters), but not necessarily within groups47. All
`analyses were conducted using Stata software47. Because the 8547 patients
`
`Personal non-commercial use only. The Journal of Rheumatology Copyright © 2002. All rights reserved.
`
`The Journal of on March 24, 2015 - Published by
`www.jrheum.org
`Downloaded from
`The Journal of Rheumatology 2002; 29:3
`
`Rheumatology
`
`468
`
`Page 3 of 8
`
`Patent Owner Ex. 2061
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`had 12,177 individual courses of NSAID therapy, case weights were
`assigned such that the total weight given to each patient was 1. Statistical
`significance was set at 0.05 and all tests were 2 tailed.
`
`RESULTS
`Demographic and clinical status variables. Table 1 presents
`patients’ basic demographic and clinical status variables.
`Use of GI drugs. As shown in Table 2, use of GI drugs
`among the cohort of NSAID users was common. Excluding
`antacids that were used by 26% of patients, GI drugs were
`used by 42% of arthritis patients. The most common class of
`drug was H2RA (23.8%), followed by PPI (19.9%). Only a
`few patients were taking gastroprotective agents, including
`misoprostol 3.6% and sucralfate 1.2%. The distribution of
`drug usage was similar among the patients with RA and
`those with OA/FM.
`NSAID usage. During the 6 month period of observation the
`
`Table 1. Basic demographic and clinic status data on 8547 patients with
`arthritis.
`
`Variable
`
`Mean or %
`
`59.55
`Age, yrs
`20.30
`Sex, % male
`92.26
`White, (%)
`13.48
`Education level, yrs
`44,890.14
`Total income, $US
`1.05
`HAQ disability (0–3)
`4.05
`VAS pain (0–10)
`30.07
`SF-36 physical component score
`43.73
`SF-36 mental component score
`Lifetime history of upper GI ulcers, % 16.71
`Lifetime history of myocardial
`infarction,%
`Any GI symptoms, %
`Epigastric or abdominal pain
`Prednisone use, %
`
`5.24
`47.94
`23.55
`39.4
`
`SD
`
`12.72
`
`2.29
`28,147.31
`0.70
`2.75
`8.65
`13.57
`
`GI: gastrointestinal, HAQ: Health Assessment Questionnaire, VAS: visual
`analog scale.
`
`Table 2. GI agents used by 8547 patients with arthritis.
`
`GI Drug
`
`OA, FM
`RA,
`All,
`N = 8547,% N = 6375, % N = 2172, %
`
`Sucralfate
`Misoprostol
`Lansoprazole
`Omeprazole
`All gastroprotective agents
`All proton pump inhibitors
`H2RA
`Antacids
`Any one of PPI, H2RA,
`sucralfate, or misoprostol
`Diclofenac + misoprostol
`
`1.22
`3.63
`6.66
`14.23
`4.77
`19.89
`23.75
`26.09
`
`41.99
`6.59
`
`1.22
`4.17
`6.57
`13.96
`5.32
`19.53
`22.89
`24.22
`
`41.21
`5.68
`
`1.20
`2.03
`6.91
`15.01
`3.18
`20.95
`26.29
`31.58
`
`44.29
`9.25
`
`H2RA: H2 receptor antagonist, PPI: proton pump inhibitor.
`
`8547 NSAID users took 12,177 separate NSAID courses. A
`course is defined as the continued use of a particular NSAID
`until a switch occurs or the study period ends. Slightly more
`than 70% used one NSAID; 22.1% used 2 NSAID sequen-
`tially, 5.6% used 3 NSAID sequentially, and 2.9% used 4 or
`more NSAID sequentially.
`As shown in Table 3, the 4 most commonly used NSAID
`were celecoxib, ibuprofen, naproxen, and nabumetone.
`They accounted for 20.5%, 19.1%, 13.1%, and 9.1% of the
`12,177 courses, respectively. Celecoxib and rofecoxib
`together (COX-2-specific inhibitors) accounted for 3242
`courses (26.6%), and nonspecific NSAID accounted for
`8935 courses (73.4%). Median doses for each drug are
`displayed in Table 3.
`Upper GI ulcers and association with treatment variables.
`Upper GI ulcers occurred in 94 or 0.77% of courses, and in
`90 of the 8547 patients (1.05%). We evaluated the risk of
`upper GI ulcers, comparing patients taking GI drugs to those
`who were not (Table 4). We found no association between
`prior H2RA or gastroprotective agents and the risk of ulcers.
`However, there was a very strong risk (OR 4.1, 95% CI
`2.95, 5.69) with the use of PPI. This risk was carried over to
`GI drugs in general (OR 2.86, 95% CI 1.86, 4.42) compared
`to those no receiving GI drugs.
`To understand the relationship between the newer COX-
`2-specific inhibitors and GI drugs, we analyzed the various
`combinations of these agents. In doing these analyses we
`adjusted for NSAID dose (Table 3). First, COX-2 therapy
`compared to nonselective NSAID therapy was associated
`with reduced risk of upper GI ulcers (OR 0.45, 95% CI 0.30,
`0.70). We next considered the various combination of
`NSAID and GI drugs (Table 5). Using patients taking nons-
`elective NSAID and no GI drugs as the baseline category,
`
`Table 3. The use of NSAID in 12,177 courses by 8547 patients with
`arthritis between January 1999 and June 1999.
`
`NSAID
`
`% of Courses
`
`Median Dose, mg
`
`Celecoxib
`Ibuprofen
`Naproxen
`Nabumetone
`Diclofenac
`Oxaprozin
`Diclofenac + misoprostol
`Rofecoxib
`Etodolac
`Ketoprofen
`Sulindac
`Salsalate
`Piroxicam
`Indomethacin
`Flurbiprofen
`Tolmetin
`Meclofamate
`Fenoprofen
`
`22.50
`19.09
`13.06
`9.12
`5.54
`5.00
`4.62
`4.12
`4.04
`2.61
`2.52
`2.33
`2.09
`1.38
`0.93
`0.53
`0.32
`0.19
`
`200
`600
`750
`1000
`75
`1200
`75
`25
`500
`200
`200
`1500
`20
`75
`600
`600
`100
`600
`
`Personal non-commercial use only. The Journal of Rheumatology Copyright © 2002. All rights reserved.
`
`The Journal of on March 24, 2015 - Published by
`www.jrheum.org
`Downloaded from
`Wolfe, et al: GI drugs and ulcers
`
`Rheumatology
`
`469
`
`Page 4 of 8
`
`Patent Owner Ex. 2061
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`Table 4. The association of GI drugs with risk of GI ulceration among users of selective and nonselective NSAID.
`
`Drug Group
`
`No GI drugs
`Any GI drug(s)
`Proton Pump
`inhibitors
`H2RA
`Gastroprotective
`agents
`
`OR
`
`1.0
`2.86
`
`4.10
`1.07
`
`0.79
`
`SE
`
`—
`0.63
`
`0.69
`0.21
`
`0.29
`
`T
`
`—
`4.76
`
`8.42
`0.36
`
`–0.65
`
`p
`
`Lower 95% CI
`
`Upper 95% CI
`
`—
`0.000
`
`0.000
`0.722
`
`0.517
`
`—
`1.86
`
`2.95
`0.73
`
`0.38
`
`—
`4.42
`
`5.69
`1.57
`
`1.62
`
`H2RA: H2 receptor antagonists. Gastroprotective agents: misoprostol and sucralfate.
`
`Table 5. Distribution of COX-2 drugs and GI drugs.
`
`Combination
`
`Nonselective NSAID, no GI drug
`Nonselective NSAID + GI drugs
`COX-2-specific inhibitor, no GI drugs
`COX-2-specific inhibitor + GI drugs
`Non selective NSAID, no PPI
`Non selective NSAID + PPI
`COX-2-specific inhibitor, no PPI
`COX-2-specific inhibitor + PPI
`
`N
`
`5198
`3737
`1530
`1712
`7228
`1707
`2260
`982
`
`%
`
`42.69
`30.69
`12.56
`14.06
`59.36
`14.02
`18.56
`8.06
`
`Nonselective NSAID: nonselective COX agents containing varying degrees
`of nonselective NSAID and COX-2 activity. COX-2-specific inhibitors:
`celecoxib and rofecoxib. PPI: proton pump inhibitors. GI drugs: proton
`pump inhibitors, H2RA, sucralfate, misoprostol.
`
`Table 6 (2 regression analyses) shows that the use of PPI
`together with nonselective NSAID was associated with a
`strong increase in the risk of upper GI ulcers compared to
`those not taking PPI, with OR > 5. COX-2 specific
`inhibitors had a lower risk of ulcers, in the subgroups
`
`without GI drugs (OR 0.32, compared to NSAID users
`without GI drugs) and in the COX-2 subgroup without PPI
`(OR 0.52, compared to NSAID users without PPI).
`However, these reductions in ulcers did not reach statistical
`significance in this analysis. Compared with the COX-2 (+),
`PPI (+) group, the risk of upper GI ulcers was increased in
`the nonselective NSAID plus PPI group (OR 3.83, 95% CI
`2.32, 6.31). Rates of GI ulceration per 100 patients per 6
`month period are also described in Table 6.
`Results for all GI drugs combined were similar to the PPI
`analyses shown in Table 6, but were attenuated owing to the
`lack of contribution from the H2 and gastroprotective
`agents. Arthrotec (diclofenac plus misoprostol) was
`excluded from these analyses, but had a nonsignificant asso-
`ciation with upper GI ulcers (OR 1.70, 95% CI 0.81, 3.59).
`
`DISCUSSION
`The results of this study confirm reports that use of GI drugs
`is associated with an increased risk of upper GI ulcers, even
`though many of these agents are known to be effective in the
`
`Table 6. The association between NSAID, GI drugs, and the risk and rates of GI ulceration. Rates are per hundred patients per 6 month period.
`
`Drug Grouping
`
`OR
`
`p
`
`Lower 95% CI Upper 95% CI
`
`Rate
`
`Lower 95% CI
`
`Upper 95% CI
`
`Analysis 1
`Nonselective NSAID
`no GI drug
`Nonselective NSAID
`+ GI drugs
`COX-2 + no GI drugs
`COX-2 + GI drugs
`Standardized dose
`
`Analysis 2
`Nonselective NSAID
`no PPI
`Nonselective NSAID
`+ PPI
`COX-2 + no PPI
`COX-2 + PPI
`Standardized dose
`
`1.0
`
`2.93
`0.32
`1.27
`1.45
`
`1.0
`
`5.09
`0.52
`1.33
`1.48
`
`—
`
`0.000
`0.204
`0.144
`0.133
`
`—
`
`0.000
`0.216
`0.218
`0.128
`
`—
`
`1.90
`0.05
`0.92
`0.89
`
`—
`
`3.88
`0.19
`0.85
`0.89
`
`—
`
`4.50
`1.87
`1.74
`2.36
`
`0.004
`
`0.011
`0.001
`0.005
`
`0.003
`
`0.009
`0.000
`0.005
`
`—
`
`0.004
`
`0.003
`
`6.67
`1.46
`2.09
`2.44
`
`0.021
`0.002
`0.006
`
`0.017
`0.001
`0.003
`
`0.005
`
`0.014
`0.007
`0.006
`
`0.005
`
`0.026
`0.005
`0.010
`
`Nonselective NSAID: nonselective COX agents containing varying degrees of nonselective NSAID and COX-2 activity. COX-2-specific inhibitors: celecoxib
`and rofecoxib. PPI: proton pump inhibitors. GI drugs: proton pump inhibitors, H2RA, sucralfate, misoprostol.
`
`Personal non-commercial use only. The Journal of Rheumatology Copyright © 2002. All rights reserved.
`
`The Journal of on March 24, 2015 - Published by
`www.jrheum.org
`Downloaded from
`The Journal of Rheumatology 2002; 29:3
`
`Rheumatology
`
`470
`
`Page 5 of 8
`
`Patent Owner Ex. 2061
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`prevention and treatment of upper GI ulcers in the setting of
`randomized controlled trials (RCT). This seeming paradox
`is most likely the result of channeling bias and confounding
`by indication associated with nonrandom selection31, and
`confirms and expands upon the results of Martin, et al27.
`Simply put, patients with a high baseline risk for ulcers are
`prescribed gastroprotective therapy. Consequently, patients
`using GI drugs would be expected to have higher rates of
`upper GI ulcers than nonusers, all other factors being equal.
`Although GI drug use may reduce the risk of subsequent
`ulcer development, it does not overcome the higher baseline
`risk of GI drug users, relative to non-GI drug users. As a
`result, the overall risk of upper GI ulcers in patients using GI
`drugs remains higher than in those not selected to receive GI
`drugs.
`It is of some interest that we did not find an increased risk
`of upper GI ulcers with H2 and gastroprotective agents, but
`only with PPI. This is probably because changing percep-
`tions of GI drug efficacy and appropriateness of use have led
`to the use of the PPI compounds in those with the highest
`risk. Overall, the odds ratio for PPI use was 4.10 (95% CI
`2.95, 5.69). It is of interest that the overall risk associated
`with the use of GI drug in this study is the same as noted by
`Martin, et al in their study of meloxicam (2.9, 95% CI 1.0,
`8.4)27.
`COX-2-specific inhibitors uniformly reduced the risk of
`upper GI ulcers in RCT2,4,5,48. However, in agreement with
`Martin, et al27, we have shown that COX-2 prescription
`following the introduction of the COX-2-specific inhibitors
`is associated with channeling bias and confounding by indi-
`cation30. Thus COX-2 use might have been expected to be
`associated with increased risk of upper GI ulcers due to their
`higher baseline (i.e., nondrug) risk for ulcers. Our results,
`however, show that COX-2-specific inhibitor use overall
`was associated with greater than 2-fold reduction in upper
`GI ulcers compared to nonselective NSAID (OR 0.45, 95%
`CI 0.30, 0.70). When the analysis was stratified to patients
`using concomitant PPI, risk of upper GI ulcers associated
`with COX-2 usage compared to NSAID was reduced 4-fold
`(OR 0.26, 95% CI 0.15, 0.44). COX-2-specific inhibitors
`were also directionally safer in the category of no PPI use;
`however, the reduction was not statistically significant (OR
`0.53, 95% CI 0.19, 1.43).
`We also noted that there was no increase in risk of the
`COX-2 (+) PPI (+) patients compared to the nonselective
`NSAID patients alone, but that the risk of upper GI ulcers
`was greatly increased in those taking nonselective NSAID
`plus PPI (OR 5.09, 95% CI 3.88, 6.67). Although all the
`evidence is in favor of the protective effect of COX-2
`agents, it is possible that the biased assignment to these
`agents resulted in part of the observed effect. To test this
`hypothesis, we controlled for pain, global severity, age,
`number of medical visits, helplessness, and SF-36 mental
`component score, since these were among the items that
`
`differed between those who would receive COX-2-specific
`inhibitors in the future and those who would not in our
`previous research30. Addition of these covariates to our
`models did not result in any substantial change in the results
`(data not shown).
`This study reports the actual experience with NSAID and
`GI drugs in a large sample of patients receiving rheumatic
`disease care. While this is one of the strengths of our report,
`one of its limitations is that does not and cannot address
`whether GI drugs were prescribed or used for the correct
`indication or in the recommended way. It follows that we
`also do not know why GI drugs were prescribed. Even so,
`this study experience reflects actual rather than RCT study
`use. In contrast to clinical trials in which antiinflammatory
`drug use is continuous, use of antiinflammatory drugs in
`clinical practice is intermittent. This would explain the
`lower rates of ulcers observed in clinical practice as
`compared to clinical trials.
`The data from this study have specific clinical and
`research relevance. At the clinical level, the data show that
`use of PPI and/or COX-2 agents is not sufficient to eliminate
`upper GI ulcers in high risk patients. However, compared
`with nonselective NSAID, COX-2 NSAID were associated
`with reduced risk of ulcers in the presence of PPI use. It is
`likely that the use of PPI also reduced the ulcer risk in this
`high risk group, but this could not be determined in the
`current study. These data also explain, by confounding by
`indication and channeling bias, the common clinical obser-
`vation that drugs that are targeted to reduce certain adverse
`events may seem to be associated with an increase in these
`events. At the research level, the data underscore the diffi-
`culty of discerning true drug effectiveness in observational
`studies in the presence of confounding unless there is
`adequate adjustment using mechanisms such as propensity
`scores30. Finally, for both clinicians and researchers, these
`data underscore important differences between RCT, where
`drugs are tested in specific settings and for specific indica-
`tions, and the real-life use, where drugs are used in settings
`and for indications that are often quite different.
`The results of this study indicate the importance of
`confounding by indication and channeling bias in under-
`standing GI ulcer rates and the relation of PPI therapy, but
`also provide strong evidence for the effectiveness of COX-
`2 agents alone and in combination with PPI therapy in
`reducing the rate of upper GI ulcers.
`
`REFERENCES
`1. Wolfe MM, Lichtenstein DR, Singh G. Gastrointestinal toxicity of
`nonsteroidal antiinflammatory drugs [published erratum appears in
`N Engl J Med 1999;3417:548]. N Engl J Med 1999;340:1888-99.
`2. Simon LS, Weaver AL, Graham DY, et al. Antiinflammatory and
`upper gastrointestinal effects of celecoxib in rheumatoid arthritis: a
`randomized controlled trial. JAMA 1999;282:1921-8.
`3. Emery P, Zeidler H, Kvien TK, et al. Celecoxib versus diclofenac in
`long-term management of rheumatoid arthritis: randomised double-
`blind comparison. Lancet 1999;354:2106-11.
`
`Personal non-commercial use only. The Journal of Rheumatology Copyright © 2002. All rights reserved.
`
`The Journal of on March 24, 2015 - Published by
`www.jrheum.org
`Downloaded from
`Wolfe, et al: GI drugs and ulcers
`
`Rheumatology
`
`471
`
`Page 6 of 8
`
`Patent Owner Ex. 2061
`Mylan v. Pozen
`IPR2017-01995
`
`

`

`4. Goldstein JL, Silverstein FE, Agrawal NM, et al. Reduced risk of
`upper gastrointestinal ulcer complications with celecoxib, a novel
`COX-2 inhibitor. Am J Gastroenterol 2000;95:1681-90.
`5. Langman MJ, Jensen DM, Watson DJ, et al. Adverse upper
`gastrointestinal effects of rofecoxib compared with NSAIDs. JAMA
`1999;282:1929-33.
`6. MacDonald TM, Morant SV, Robinson GC, et al. Association of
`upper gastrointestinal toxicity of non-steroidal anti-inflammatory
`drugs with continued exposure: cohort study. BMJ 1997;315:1333-7.
`7. Lanza LL, Walker AM, Bortnichak EA, Dreyer NA. Peptic ulcer and
`gastrointestinal hemorrhage associated with nonsteroidal anti-
`inflammatory drug use in patients younger than 65 years. A large
`health maintenance organization cohort study. Arch Intern Med
`1995;155:1371-17.
`8. Tenenbaum J. The epidemiology of nonsteroidal anti-inflammatory
`drugs. Can J Gastroenterol 1999;13:119-22.
`9. Langman MJS, Brooks P, Hawkey CJ, Silverstein F, Yeomans N.
`Non-steroidal anti-inflammatory drug associated ulcer —
`epidemiology, causation and treatment. J Gastroenterol Hepatol
`1991;6:442-9.
`10. Simon LS, Zhao SZ, Arguelles LM, et al. Economic and
`gastrointestinal safety comparisons of etodolac, nabumetone, and
`oxaprozin from insurance claims data from patients with arthritis.
`Clin Ther 1998;20:1218-35.
`11. Hogan DB, Campbell NR, Crutcher R, Jennett P, MacLeod N.
`Prescription of nonsteroidal anti-inflammatory drugs for elderly
`people in Alberta. Can Med Assoc J 1994;151:315-22.
`12. Kurata JH, Nogawa AN, Noritake D. NSAIDs increase risk of
`gastrointestinal bleeding in primary care patients with dyspepsia.
`J Fam Practice 1997;45:227-35.
`13. Griffin MR, Piper JM, Daugherty JR, Snowden M, Ray WA.
`Nonsteroidal anti-inflammatory drug use and increased risk for
`peptic ulcer disease in elderly persons. Ann Intern Med
`1991;114:257-63.
`14. Silverstein FE, Graham DY, Senior JR, et al. Misoprostol reduces
`serious gastrointestinal complications in patients with rheumatoid
`arthritis receiving nonsteroidal antiinflammatory drugs. A
`randomized, double-blind, placebo- controlled trial. Ann Intern Med
`1995;123:241-9.
`15. Fries JF, Williams CA, Bloch DA, Michel BA. Nonsteroidal anti-
`inflammatory drug-associated gastropathy: incidence and risk factor
`models. Am J Med 1991;91:213-22.
`16. Gabriel SE, Jaakkimainen L, Bombardier C. Risk for serious
`gastrointestinal complications related to use of nonsteroidal anti-
`inflammatory drugs — a metaanalysis. Ann Intern Med
`1991;115:787-96.
`17. Piper JM, Ray WA, Daugherty JR, Griffin MR. Corticosteroid use
`and peptic ulcer disease: role of nonsteroidal antiinflammatory
`drugs. Ann Intern Med 1991;114:735-40.
`18. Griffin MR. Epidemiology of nonsteroidal anti-inflammatory drug-
`associated gastrointestinal injury. Am J Med 1998;104:23S-29S.
`19. Wolfe F, Hawley DJ. The comparative risk and predictors of adverse
`gastrointestinal events in rheumatoid arthritis and osteoarthritis: A
`prospective 13 year study of 2131 patients. J Rheumatol
`2000;27:1668-73.
`20. Weil J, Langman MJ, Wainwright P, et al. Peptic ulcer bleeding:
`accessory risk factors and interactions with nonsteroidal
`antiinflammatory drugs. Gut 2000;46:27-31.
`21. Singh G. Recent considerations in nonsteroidal anti-inflammatory
`drug gastropathy. Am J Med 1998;105:31S-38S.
`22. Agrawal NM, Campbell DR, Safdi MA, Lukasik NL, Huang B,
`Haber MM. Superiority of lansoprazole vs ranitidine in healing
`nonsteroidal antiinflammatory drug-associated gastric ulcers: results
`of a double-blind, randomized, multicenter study. NSAID-
`Associated Gastric Ulcer Study Group. Arch Intern Med
`
`2000;160:1455-61.
`23. Rostom A, Wells G, Tugwell P, Welch V, Dub