original article

Clopidogrel with or without Omeprazole in Coronary Artery Disease

Abstract
Background

Gastrointestinal complications are an important problem of antithrombotic therapy. Proton-pump inhibitors (PP s! are belie"ed to decrease the ris# of such complications$ though no randomized trial has pro"ed this in patients recei"ing dual antiplatelet therapy. %ecently$ concerns ha"e been raised about the potential for PP s to blunt the efficacy of clopidogrel.
Methods

&e randomly assigned patients with an indication for dual antiplatelet therapy to recei"e clopidogrel in combination with either omeprazole or placebo$ in addition to aspirin. 'he primary gastrointestinal end point was a composite of o"ert or occult bleeding$ symptomatic gastroduodenal ulcers or erosions$ obstruction$ or perforation. 'he primary cardio"ascular end point was a composite of death from cardio"ascular causes$ nonfatal myocardial infarction$ re"ascularization$ or stro#e. 'he trial was terminated prematurely when the sponsor lost financing.
Results

&e planned to enroll about ())) patients* a total of +,-+ were randomly assigned and +-./ were included in analyses. n all$ (/ patients had a gastrointestinal e"ent* the e"ent rate was /./0 with omeprazole and 1.20 with placebo at /,) days (hazard ratio with omeprazole$ ).+3$ 2(0 confidence inter"al 4C 5$ )./, to )..+* P6).))/!. 'he rate of o"ert upper gastrointestinal bleeding was also reduced with omeprazole as compared with placebo (hazard ratio$ )./+* 2(0 C $ ).)+ to ).(.* P 7 ).))/!. A total of /)2 patients had a cardio"ascular e"ent$ with e"ent rates of 3.20 with omeprazole and (.-0 with placebo (hazard ratio with omeprazole$ ).22* 2(0 C $ ).., to /.33* P 7 ).2.!* high-ris# subgroups did not show significant heterogeneity. 'he two groups did not differ significantly in the rate of serious ad"erse e"ents$ though the ris# of diarrhea was increased with omeprazole.
Conclusions

Among patients recei"ing aspirin and clopidogrel$ prophylactic use of a PP reduced the rate of upper gastrointestinal bleeding. 'here was no apparent cardio"ascular interaction between clopidogrel and omeprazole$ but our results do not rule out a clinically meaningful difference in cardio"ascular e"ents due to use of a PP . (8unded by Cogentus Pharmaceuticals* Clinical'rials.go" number$ 9C'))((-21/.!

n the basis of data from se"eral

studies$ clopidogrel has become the second most commonly used prescription drug worldwide./-2 Gastrointestinal hemorrhage is the most common serious bleeding complication from the use of long-term antiplatelet therapy./)$// Data

from randomized studies support the concept that therapies reducing acidity decrease gastrointestinal complications of antiplatelet therapy in"ol"ing aspirin$ though the data are largely based on endoscopic end points* obser"ational data also support this effect./1-/. %andomized$ controlled trials ha"e shown that proton-pump inhibitors (PP s! reduce the rate of recurrent gastrointestinal bleeding in high-ris# patients recei"ing aspirin./- Obser"ational studies$ howe"er$ ha"e suggested that there may be an interaction between clopidogrel and PP s that$ if real$ could ha"e significant clinical effects./,$/2 'hese studies ha"e been bolstered by results of e: "i"o analyses$ many of which ha"e shown inhibition of the antiplatelet effect of clopidogrel by PP s$ omeprazole most consistently.1)-11 n addition$ genetic polymorphisms ha"e been identified that could affect the response to clopidogrel and$ at least theoretically$ could increase the li#elihood of drug interactions mediated by cytochrome P3().1+-1- A number of other obser"ational studies$ howe"er$ did not show an interaction between clopidogrel and PP s.1,$12 Gi"en the conf licting data regarding a possible interaction$ the optimal care of patients who re;uire concomitant therapy with clopidogrel and PP s remains uncertain.+)-+3 &e initiated the Clopidogrel and the Optimization of Gastrointestinal <"ents 'rial (COG<9'! to assess the efficacy and safety of concomitant administration of clopidogrel and PP s in patients with coronary artery disease who are recei"ing clopidogrel plus aspirin. tion was performed centrally with the use of an interacti"e "oice-response system before the initiation of study treatment. Pare:el generated the randomization se;uence. All sites operated under appro"al from institutional re"iew boards or ethics committees$ and all patients ga"e written informed consent to participate in the trial. 'he study was conducted according to the study protocol (a"ailable at 9<=>.org!. At the conclusion of the trial$ the full database was transferred to an academic principal in"estigator. 'he analyses were performed independently of the sponsor$ by two academic authors. An academic principal in"estigator prepared the first draft of the manuscript$ which was then re"iewed and edited by the academic steering committee and other authors* all the academic authors made the decision to submit the paper for publication. 'here was no agreement made regarding confidentiality of the data between the sponsor and the academic authors or their institutions. 'he sponsor did not ha"e the right to appro"e the final manuscript. 'he academic principal in"estigators "ouch for the accuracy and integrity of the analyses and interpretation of the data. 'he initial planned sample size was +1)) patients$ with an accrual period of / year and a ma:imum follow-up period of 1 years. 'he target sample size was increased to 31)) and then to ())) to ensure an ade;uate number of gastrointestinal e"ents. 'he study was designed to end once /3+ gastrointestinal e"ents had occurred* howe"er$ the study ended prematurely$ when the sponsor suddenly and une:pectedly lost its financial bac#ing* the sponsor is now defunct. Patients COG<9' was an international$ randomized$ double-blind$ double-dummy$ placebo-controlled$ parallel-group$ phase + study of the efficacy and safety of CG'-1/.,$ a fi:ed-dose combination of

Study Conduct > e t h od s clopidogrel (-( mg! and omeprazole (1) mg!$ as compared with clopidogrel alone. %andomization was performed with the use of stratified 'he trial was designed by an academic steering committee (see the ?upplementary Appendi:$ a"ailable with the full te:t of this article at 9<=> .org! and the sponsor$ Cogentus Pharmaceuticals. 'he steering committee was responsible for the o"erall leadership of the trial. A clinical research organization$ Pare:el$ performed the data management and site monitoring. %andomizapermuted bloc#s. ?tratification was based on two baseline factors@ serologic findings for Helicobacter pylori (positi"e or negati"e! and concomitant use ("s. nonuse! of any nonaspirin nonsteroidal antiinf lammatory drug (9?A D!$ including agents selecti"e or nonselecti"e for cycloo:ygenase-1. All patients were to recei"e enteric-coated aspirin at a dose of -( to +1( mg daily. Alinded /2/) study #its and open-label enteric-coated aspirin were supplied by Pare:el to the in"estigators. Patients were eligible if they were 1/ years of age or older and if the use of clopidogrel therapy with concomitant aspirin was anticipated for at least the ne:t /1 months$ including patients presenting with an acute coronary syndrome or undergoing placement of a coronary stent. Patients were enrolled at +2+ sites in /( countries from =anuary 1)), through December 1)),. Bospitalized patients for whom discharge within 3, hours after randomization was not anticipated were e:cluded from the study. Additional e:clusion criteria were the need for shortterm or long-term use of a PP $ an B1receptor antagonist$ sucralfate$ or misoprostol* pree:isting erosi"e esophagitis or esophageal or gastric "ariceal disease or pre"ious nonendoscopic gastric surgery* receipt of clopidogrel or another thienopyridine for more than 1/ days before randomization* receipt of oral anticoagulation therapy that could not be safely discontinued for the duration of the study* or recent fibrinolytic therapy. End Points 'he prespecified primary gastrointestinal efficacy end point was the time from randomization to the first occurrence of a composite of upper gastrointestinal clinical e"ents@ o"ert bleeding of gastroduodenal origin (confirmed by means of upper endoscopy or radiography!$ o"ert upper gastrointestinal bleeding of un#nown origin$ bleeding of presumed occult gastrointestinal origin with a documented decrease in hemoglobin of 1 g per deciliter or more or in the hematocrit by /)0 or more from the baseline "alue$ symptomatic uncomplicated gastroduodenal ulcer (confirmed by means of endoscopy or radiography!$ cardio"ascular causes$ nonfatal myocardial infarction$ coronary re"ascularization$ or ischemic stro#e. 'here was no a priori sample-size calculation or e:plicit noninferiority hypothesis for the cardio"ascular end point. AdCudication of cardio"ascular e"ents was performed by an independent committee of cardiologists who were unaware of the study-drug assignments. 9ongastrointestinal bleeding e"ents were also recorded and adCudicated. Statistical Analysis >edians and inter;uartile ranges are reported for continuous "ariables$ and

counts and percentages for categorical "ariables. Analyses of time-to-e"ent "ariables were performed with the use of log-ran# statistics$ DaplanE>eier sur"i"al cur"es$ and Co: proportional-hazards models. 'he two maCor stratification "ariables were included in the models. 'hese analyses were performed on data for the adCudicated e"ents and secondarily on data for the e"ents ascertained by the site in"estigators@ the gastrointestinal composite e"ent$ the cardio"ascular composite e"ent$ myocardial infarction$ and re"ascularization. Other end points were analyzed descripti"ely. DaplanE >eier e"ent rates were calculated at /,) days$ co"ering appro:imately ,(0 of the total followup period. Bazard ratios with 2(0 confidence inter"als are reported. Gi"en the premature termination of the trial$ two academic authors separately analyzed the database and reconciled any discrepancies. All tests were two-sided. P "alues of less than ).)( were considered to indicate statistical signif icance. Analyses were performed with the use of ?tata software$ "ersion 2.1$ and % software$ "ersion 1.21 (% De"elopment Core 'eam$ 1))2!. persistent pain of presumed gastrointestinal origin with a duration of + days or more and with %esults fi"e or more gastroduodenal erosions (confirmed by means of endoscopy!$ obstruction$ or perforation. 'he time from randomization to the first occurrence of gastroesophageal ref lu: disease$ as e"idenced by symptomatic$ endoscopically confirmed erosi"e esophagitis$ was a predefined secondary end point. AdCudication of gastrointestinal e"ents was performed by an independent committee of gastroenterologists who were unaware of the studydrug assignments. 'he prespecified primary cardio"ascular safety end point was the composite of death from Study Participants A total of 3333 patients were screened for inclusion in the study (8ig. / in the ?upplementary Appendi:!. Of these patients$ +,-+ underwent randomization and (-/ did not. A total of +-./ patients were included in analyses@ /,-. in the omeprazole group and /,,( in the placebo group. 'he median duration of followup was /). days$ with a ma:imum of +3/ days (inter;uartile range$ (( to /..!. 'he two study groups were well matched with respect to baseline characteristics Gastrointestinal Outcomes 'here were (( adCudicated gastrointestinal e"ents$ with 3- patients ha"ing a single e"ent and 3 patients ha"ing two e"ents. A total of (/ first gastrointestinal e"ents were included in the time-to-e"ent analyses. 'he e"ent rate$ based on DaplanE>eier analysis$ for the primary gastrointestinal end point was reduced from 1.20 with placebo to /./0 with omeprazole at /,) days after randomization (P6).))/ by the log-ran# test! (8ig. /!. 'he hazard ratio deri"ed from the Co: model was ).+3 (2(0 confidence inter"al 4C 5$ )./, to )..+* P6).))/!. 'here were no significant interactions among subgroups or according to the stratification "ariables@ presence "ersus absence of H. pylori (P 7 ).3for interaction! and use "ersus nonuse of an 9?A D (P 7 ).2- for interaction! (8ig. 1 in the ?upplementary Appendi:!. 'here was a borderline significant interaction on the basis of se: (P 7 ).)( for interaction!. <"ent rates for indi"idual components of the composite gastrointestinal end point are listed in 'able 1. ?ignificant differences were seen between the

omeprazole group and the placebo group with regard to o"ert gastroduodenal bleeding (hazard ratio with omeprazole$ )./1! and o"ert upper gastrointestinal bleeding of un#nown origin (hazard ratio$ )./+!. 'he rate of the composite end point of o"erall (o"ert and occult! clinical gastrointestinal bleeding was also reduced with omeprazole as compared with placebo (hazard ratio$ ).+)* 2(0 C $ )./+ to )...* P 7 ).))/!$ as was the rate of the composite end point of o"ert gastroduodenal bleeding or o"ert upper gastrointestinal bleeding of un#nown origin@ from /.10 in the placebo group (occurring in /( of /,,( patients! to ).10 (occurring in 1 of /,-. patients! (hazard ratio with omeprazole$ )./+* 2(0 C $ ).)+ to ).(.* P 7 ).))/!. 'he number of patients who would need to be treated for . months to pre"ent one occurrence of an e"ent that was part of the primary gastrointestinal end point was (($ and the number needed to treat to pre"ent one occurrence of o"ert gastrointestinal bleeding was 2,. 'here was also a significant reduction in the number of patients with in"estigator-defined gastrointestinal e"ents with omeprazole (+2 patients! as compared with placebo (., patients! (hazard ratio with omeprazole$ ).(-* 2(0 C $ ).+, to ).,3* P7).))(!. 'he rate of symptoms of gastroesophageal reflu: disease at /,) days was ).10 in the omeprazole group and /.10 in the placebo group (hazard ratio$ ).11* 2(0 C $ ).). to ).-2* P 7 ).)/!. 'here was one case of gastrointestinal obstruction in each of the two groups$ with no perforations in either group. Cardiovascular and Other Outcomes 'here were /)2 adCudicated cardio"ascular e"ents ((3 in the placebo group and (( in the omeprazole group!$ with no significant difference in the rate of the primary cardio"ascular end point between the two groups (P 7 ).2, by the log-ran# test! (8ig. 1!. 'he e"ent rate at /,) days after randomization was (.-0 in the placebo group and 3.20 in the omeprazole group (hazard ratio with omeprazole$ ).22* 2(0 C $ ).., to /.33* P 7 ).2.!. Analysis of subgroups of patients with "arious forms of "ascular disease$ including pre"ious myocardial infarction$ did not show significant heterogeneity (8ig. + in the ?upplementary Appendi:!. 9or did the o"erall results differ significantly on the basis of serologic data regarding H. pylori or concomitant 9?A D use (P 7 ).31 and P 7 )..,$ respecti"ely$ for interaction!. 'he rates of indi"idual components of the cardio"ascular end point did not differ significantly between the two groups ('ables 1 and +!. 'here were two cases of definite or probable stent thrombosis in the placebo group and none in the omeprazole group. 9o signif icant difference was found between the two groups in the number of patients with in"estigator-defined cardio"ascular e"ents with omeprazole (./ patients! as compared with placebo ((, patients! (hazard ratio with omeprazole$ /.)+* 2(0 C $ ).-1 to /.3-*P 7 ).,2!. 'he rate of adCudicated nongastrointestinal bleeding e"ents did not differ significantly between the omeprazole group ().(0! and the placebo group ()./0! (hazard ratio with omeprazole$1.+1* 2(0 C $ )..) to ,.2,* P 7 ).1/!. Adverse Events 'he rate of serious ad"erse e"ents did not differ significantly between the two groups (/)./0 with omeprazole and 2.30 with placebo$ P 7 ).3,!$ nor did the rate of o"erall ad"erse e"ents (3/.+0 and 31.,0$ respecti"ely* P 7 ).++!. Diarrhea was reported in +.)0 of patients recei"ing omeprazole$ as compared with /.,0 of those recei"ing placebo (P 7 ).)/!. 9o patient had diarrhea caused

by infection with Clostridium difficile. 'here were no newly diagnosed cases of osteoporosis. One case of peripheral neuropathy was reported in the placebo group. 'here were no significant differences between the two groups in the rates of pneumonia$ headache$ nausea$ anemia$ or fracture. upper gastrointestinal bleeding$ in patients recei"ing dual antiplatelet therapy who were randomly assigned to also recei"e a PP . 8urthermore$ our prospecti"e$ double-blind$ randomized trial did not show any significant increases in the ris# of cardio"ascular e"ents with concomitant use of clopidogrel and omeprazole$ a finding that was consistent e"en in high-ris# subgroups and for indi"idual end points. Although pre"ious obser"ational studies ha"e yielded conf licting results in this regard$ the current study re"eals no signal of harm from concomitant clopidogrel and PP use. Gastrointestinal bleeding is an important potential complication of antithrombotic therapy. Pre"ious randomized studies ha"e shown that prophylactic use of PP s and B1-receptor antagonists reduces the ris# of endoscopically ascertained ulcers in patients recei"ing aspirin./+$/3'hese trials$ howe"er$ ha"e not been powered to Di s c u s s ion &e found a significant reduction in the ris# of gastrointestinal clinical e"ents$ including o"ert e"aluate clinical gastrointestinal e"ents$ nor ha"e they assessed the potential benefit for patients recei"ing combination antiplatelet therapy. 'rials showing the "alue of PP s in pre"enting recurrent gastrointestinal bleeding ha"e been conducted in populations at high ris# for gastrointestinal bleeding./- Our trial$ in which the study population was at least /) times as large as those in pre"ious randomized studies and was not selected to represent high-ris# patients$ showed a signif icant reduction in clinically manifested gastrointestinal bleeding e"ents$ including o"ert bleeding$ with a PP as compared with placebo. 'he number needed to treat would most li#ely be lower for a patient population at higher gastrointestinal ris# than our study population. 9ewer$ more potent antiplatelet agents are entering the clinical arena.+(-3) 9e"ertheless$ research into clopidogrel remains important$ gi"en that it has a wide range of uses and that the generic form is already a"ailable in certain parts of the world and may be more widely a"ailable relati"ely soon. n addition$ studies with higher doses of clopidogrel are ongoing. 'herefore$ e"aluation of possible drug interactions with clopidogrel remains important. 'he fact that se"eral (though not all! studies ha"e shown that PP s blunt the antiplatelet effect of clopidogrel$ e"en though there does not appear to be any significant clinical interaction between the drugs$ also calls into ;uestion the use of e: "i"o antiplatelet testing to alter clinical therapy.1, <: "i"o platelet assays ha"e already been shown to be potentially misleading$ particularly for assessing drug interactions. 8or e:ample$ initial concerns about e: "i"o manifestations of clopidogrelEstatin interactions were not borne out in clinical studies.3/$31 8urther wor# in the e"ol"ing area of platelet-function assays is clearly necessary. 'he potential for obser"ational studies to be misleading is also worth noting. 'here are limitations to our analysis. 8irst$ since the trial was terminated prematurely$ its power is limited$ owing to a smaller number of e"ents than had been anticipated. ?econd$ because the confidence inter"al around the hazard ratio for cardio"ascular e"ents is wide$ the absence of

interaction between clopidogrel and omeprazole cannot be "iewed as a definiti"e finding. Gi"en that 230 of the population was white$ the e:pected pre"alence of homozygosity for the loss-of-function cytochrome P-3() gene CYP2C19 was 1 to +0$ and in homozygous patients$ PP s may further reduce the le"el of the acti"e metabolite clopidogrel to a degree that does indeed blunt the effecti"eness of clopidogrel. A much larger study in"ol"ing genotyping would be necessary to determine whether this is the case. 9e"ertheless$ with respect to the cardio"ascular safety end point$ a greater number of patients than initially planned were enrolled$ and though the follow-up period was truncated$ the ris# of cardio"ascular e"ents would be e:pected to be greatest soon after the onset of an acute coronary syndrome or percutaneous coronary inter"ention. 'he absence of an effect on nongastrointestinal bleeding also supports the absence of an interaction between clopidogrel and omeprazole$ since if PP s diminish the antiplatelet effect of clopidogrel$ they should also decrease the rate of nongastrointestinal bleeding. An additional limitation of the study is that the single-pill formulation we used differs from generic omeprazole with respect to its release #inetics. 8inally$ this study was not designed to detect any differences among PP s with respect to a possible interaction$ though the PP most commonly and consistently implicated in e: "i"o studies has been omeprazole.1/$3+ n conclusion$ our randomized assessment of PP s "ersus placebo in patients with coronary artery disease who were recei"ing dual antiplatelet therapy pro"ides reassurance that there is no clinically significant cardio"ascular interaction between PP s and clopidogrel$ whereas there is a significant reduction in gastrointestinal bleeding with PP use as compared with placebo. 8urther research will be necessary to determine the optimal approach to reducing the ris# of gastrointestinal ad"erse e"ents among patients recei"ing potent antithrombotic therapy$ but prophylactic proton-pump inhibition appears to be promising.