Clinical Pharmacokinetics

, Volume 49, Issue 8, pp 509–533

Drug-Drug Interaction Profiles of Proton Pump Inhibitors


    • Department of PharmacotherapyMeiji Pharmaceutical University
  • Hirotoshi Echizen
    • Department of PharmacotherapyMeiji Pharmaceutical University
Review Article

DOI: 10.2165/11531320-000000000-00000

Cite this article as:
Ogawa, R. & Echizen, H. Clin Pharmacokinet (2010) 49: 509. doi:10.2165/11531320-000000000-00000


Proton pump inhibitors (PPIs) are widely prescribed for the treatment of gastric acid-related disorders and the eradication of Helicobacter pylori. In addition, they are routinely prescribed for the prevention of gastrointestinal bleeding in patients receiving a dual antiplatelet therapy consisting of clopidogrel and aspirin (acetylsalicylic acid) after myocardial infarction or percutaneous coronary intervention and stenting. Because PPIs are given to these patients for long periods, there is a concern about the potential for clinically significant drug-drug interactions (DDIs) with concomitantly administered medications. Because PPIs give rise to profound and long-lasting elevation of intragastric pH, it is not surprising that they interfere with the absorption of concurrent medications. Drug solubility may be substantially reduced at neutral pH compared with acidic conditions. In this context, PPIs have been shown to reduce the bioavailability of many clinically relevant drugs (e.g. ketoconazole, atazanavir) by 50% or more compared with the control values.

Soon after the introduction of omeprazole (a prototype PPI) into the market, it was reported that omeprazole was associated with 30% and 10% reductions in systemic clearance of diazepam and phenytoin, respectively. In vitro studies demonstrating the inhibitory effects of omeprazole on the metabolism of these drugs with human liver microsomes gave a mechanistic explanation for the DDIs. Numerous subsequent studies have been performed to investigate the DDI potential of PPIs associated with the metabolic inhibition of cytochrome P450 (CYP) enzyme activities; however, most such attempts have failed to find clinically relevant results.

Nevertheless, recent large-scale clinical trials have raised concerns about possible DDIs between PPIs and an antiplatelet drug, clopidogrel. It has been suggested that coadministration of PPIs with a dual antiplatelet therapy consisting of clopidogrel and aspirin may attenuate the anti-aggregation effects of those medications and augment the risk of cardiovascular ischaemic events. There is a possibility that PPIs may elicit detrimental effects by inhibiting CYP2C19-dominated metabolism of clopidogrel to its active metabolite. Further studies are urgently required to clarify themechanism of this DDI and to explore new aspects of theDDI potential of PPIs.

1. Background and Rationale of Drug-Drug Interactions (DDIs) of Proton Pump Inhibitors (PPIs)

Proton pump inhibitors (PPIs) are a group of substituted benzimidazole sulfoxide drugs with strong inhibitory effects on gastric acid secretion in the parietal cells of the stomach.[1] At present, five PPIs (omeprazole, esomeprazole, lansoprazole, pantoprazole and rabeprazole) are available on the market, and all of them, with the exception of esomeprazole (the S-enantiomer of omeprazole), are available as racemates.

PPIs are widely prescribed for the treatment of gastric acid-related disorders (e.g. gastric and duodenal ulcer, reflux oesophagitis and Zollinger-Ellison syndrome).[2] In addition, they are routinely prescribed for prevention of gastrointestinal bleeding in patients receiving a dual antiplatelet therapy consisting of clopidogrel and aspirin (acetylsalicylic acid) after myocardial infarction (MI) or percutaneous coronary intervention (PCI) and stenting. Furthermore, PPIs are considered an indispensable adjuvant component of a standard antibacterial regimen consisting of clarithromycin and amoxicillin for the eradication of Helicobacter pylori, because the profound acid-suppressing action of PPIs enhances the anti-H. pylori activities of the combined antibacterials.[3] Because PPIs are often given to patients with the above clinical indications, there is a concern about the potential for clinically significant DDIs with concomitantly administered medications.

Since their introduction into the global marketplace as a substitute for histamine H2-receptor antagonists, the safety profiles of PPIs have been of concern. This is because the prototype drug cimetidine (one of the H2-receptor antagonists possessing an imidazole ring structure) gave us harsh lessons on the danger of inhibition of hepatic drug metabolism caused by DDIs. Because PPIs are substituted benzimidazole prodrugs, there was a possibility that PPIs may also have potential for DDI, like cimetidine. Gugler and Jensen[4] were the first to report that omeprazole inhibited the elimination of phenytoin and diazepam, which are metabolized by cytochrome P450 (CYP) 2C9 and CYP2C19. Since then, numerous studies have been conducted and have reported somewhat controversial results.[5,6]

Recently, two retrospective studies[7,8] have suggested that patients with ischaemic heart disease who receive a PPI with clopidogrel after discharge may have an increased risk of adverse outcomes, compared with those receiving clopidogrel alone. Furthermore, recent studies have indicated that PPIs may alter the renal elimination of methotrexate.[9,10] In this context, we decided to undertake a comprehensive and updated review of DDIs attributable to PPIs, using a systematic review method. We performed a systematic search of the MEDLINE database (from 1964 to May 2010) for English- and Japanese-language articles with the keywords ‘proton pump inhibitors’ (MeSH) OR ‘omeprazole’ (MeSH) OR ‘rabeprazole’ (substance name) OR ‘lansoprazole’ (substance name) OR ‘pantoprazole’ (substance name) AND ‘drug interactions’ (MeSH) AND ‘humans’ (MeSH). We also carried out manual searches of relevant articles, based upon references in the retrieved articles. Of 372 retrieved articles, 144 were included in the current review. To ensure the reliability of the information in this review, we did not include data from abstracts or proceedings of academic conferences.

2. Pharmacokinetic Mechanisms of DDIs

2.1 Alteration of Drug Absorption

2.1.1 Elevation of Gastric pH

Gastric pH is an important parameter for drug absorption, because it influences the solubility of weakly acidic or basic drugs. The major cause of differences in absorption of a drug from various formulas is dissolution into the gastric fluid. Indeed, for many drugs, absorption is dissolution rate limited, because the rate and sometimes the extent of absorption from solid dosage forms is noticeably slower than from a liquid solution.

Previous studies demonstrated that coadministration of PPIs decreased the area under the plasma concentration-time curve (AUC) values of diverse classes of drugs administered orally.[11] We listed drugs in which oral absorption was shown to be altered significantly by coadministration of PPIs (table I). We tentatively considered that the mechanism of altered intestinal absorption of drugs with coadministration of PPIs could be attributable to changes in the solubility of affected (victim) drugs when their solubility is known to be pH sensitive, and coadministration not only of PPIs but also of H2-receptor antagonists and achlorhydria produced comparable changes in the AUCs of the drugs of interest. DDIs with two imidazole antifungal agents, ketoconazole and itraconazole, are classical examples. Because these antifungal drugs are weak bases with acid dissociation constant (pKa) values of >3.0, they are nearly insoluble at pH >4. Coadministration of a PPI was associated with significant decreases in the AUCs of these drugs.[39,53] Based upon these findings, coadministration of PPIs is not recommended with imidazole antifungal agents, particularly when given as tablets or capsules.[54] Interestingly, Johnson et al.[39] reported that administration of omeprazole with itraconazole oral solution did not alter the maximum plasma concentration (Cmax) or the AUC from 0 to 8 hours (AUC8) of itraconazole and its active metabolite, hydroxyitraconazole. Their findings would support our understanding that omeprazole-induced neutralization of gastric fluid would interfere with dissolution rather than absorption of itraconazole, as was reported for ketoconazole.[55] Interestingly, the PPI did not influence the absorption of fluconazole.
Table I

Effects of proton pump inhibitors (PPIs) on the area under the plasma concentration-time curve (AUC) values (fold increases) of orally administered concomitant drugs

A similar finding was reported for coadministration of pantoprazole with the immunosuppressant mycophenolate mofetil. Oral administration of pantoprazole at either 40 mg or 80 mg/day was shown to reduce the Cmax and AUC from 0 to 12 hours (AUC12) values of mycophenolic acid (MPA), an active metabolite of mycophenolate mofetil, by 57–78% and 27–30%, respectively, from the control values in healthy subjects and heart transplant recipients.[48,49] The solubility of mycophenolate mofetil at pH 7.0 (0.004 mg/L) was shown to be 100-fold less than that at pH 4 (4 mg/L). Interestingly, coadministration of pantoprazole with enteric-coated MPA sodium elicited no significant effects on the AUC12 of MPA.

DDIs between PPIs and protease inhibitors employed for the treatment of HIV infection have commanded attention (table I).[56] Coadministration of lansoprazole (60 mg/day) with a combination formula of atazanavir/ritonavir resulted in substantial decreases in the mean AUC value of atazanavir by 90% compared with the control value.[28] The mechanism of the DDI would be attributable to pH-dependent atazanavir solubility: the solubility of atazanavir is 1.1 mg/mL at pH 0.8 but reduces to 0.002 mg/mL at pH 5.4.[57] Omeprazole was also shown to reduce the AUC of orally administered atazanavir by 60%.[29] Because these changes may result in loss of the therapeutic effect or development of HIV resistance, the prescribing information for the drug recommends that administration of PPIs should not exceed a dose comparable to omeprazole 20 mg daily and the dose must be taken approximately 12 hours prior to the administration of atazanavir in treatment-naïve patients. The manufacturer also recommends that PPIs not be used in treatment-experienced patients receiving atazanavir.[58] Nevertheless, the clinical impact of this DDI on the anti-HIV activity of atazanavir has not been clearly demonstrated in clinical trials.[59,60] Coadministration of omeprazole was also associated with reductions in the AUCs of indinavir and nelfinavir by 60%[32] and 36%,[33] respectively.

In contrast, omeprazole was shown to increase the systemic availability of bismuth from tripotassium dicitrate bismuthate by as much as 3-fold the control value.[13] Because hypochlorhydria was associated with increased bioavailability of bismuth and the magnitude of the elevation in the AUC of bismuth was dependent on the degree of hypochlorhydria, the mechanism of the DDI between PPIs and bismuth may be attributable to a pH-dependent increase in its solubility or absorption. A similar finding was obtained from a clinical study that assessed the effect of ranitidine on bismuth absorption.[61] In addition, omeprazole (40 mg/day) as well as H2-receptor antagonists (e.g. cimetidine and ranitidine), were shown to increase the AUC of saquinavir (given orally as a saquinavir/ritonavir combination formula) by 70%.[35] It remains unclear whether the mechanism of these DDIs is solely attributable to an elevation in gastric pH.

PPIs may also alter the absorption of concomitantly administered drugs by mechanisms other than pH-dependent solubility. For instance, there is a possibility that PPIs may alter the oral absorption of a drug given by a pH-dependent release formula. This possibility has not been studied extensively. However, coadministration of omeprazole with a pH-controlled-release formulation of mesalazine (Asacol®) did not change the 24-hour urinary and faecal levels of 5-aminosalicylic acid (5-ASA) or its main metabolite, N-acetyl 5-ASA, suggesting that PPIs would not appreciably alter at least the extent of 5-ASA absorbed from the formula (table I).

The bioavailability of acid-labile drugs (e.g. penicillins) may be increased by coadministration of PPIs because the sustained elevation in gastric pH may prevent acidic hydrolysis of these drugs. Because one PPI is coadministered with amoxicillin and clarithromycin as a standard triple drug regimen against H. pylori, it would be of clinical interest to investigate whether PPIs increase the systemic availability or gastric mucosal concentrations of these drugs. Two previous studies, however, failed to demonstrate any appreciable changes in the AUCs of either amoxicillin or clarithromycin, compared with their respective control values.[42,62] Because the oral bioavailability of amoxicillin is high (i.e. 89–98%[63]), it would have been difficult to detect any pH-dependent improvement in its bioavailability, even if it existed.

2.1.2 Interaction with Efflux Transporter or Intestinal Cytochrome P450 Enzymes

Omeprazole,[15,64] but not pantoprazole,[16] was shown to increase the bioavailability of digoxin. Because digoxin is a substrate of P-glycoprotein[65] and undergoes little metabolic biotransformation, the mechanism of the DDI between omeprazole and digoxin may be attributable to omeprazole-induced inhibition of digoxin efflux into the intestinal lumen, mediated by P-glycoprotein. Nevertheless, this DDI would be of little clinical importance because the magnitude of the increase in the AUC of digoxin was at most 10%. In addition, omeprazole, but not pantoprazole, was shown to increase the AUC of oral nifedipine by 26%.[66,67] Because nifedipine is a substrate of CYP3A4 and P-glycoprotein, the precise mechanism of this DDI remains obscure. Nevertheless, the increase in the AUC of nifedipine did not translate into significant changes in concomitantly measured blood pressure or heart rate.[66] At present, none of the studies has clearly demonstrated that PPIs would elicit clinically relevant DDIs via inhibition of the P-glycoprotein drug transport system.

2.2 Alteration of Hepatic Drug Metabolism

PPIs may be involved in DDIs by altering the metabolism of concomitantly administered medications via either enzyme inhibition or induction. While enzyme inhibition by PPIs has been studied extensively, relatively fewer studies on enzyme induction have been reported.

2.2.1 In Vitro Studies on Enzyme Inhibition

All PPIs, except for rabeprazole, are extensively metabolized in the liver, mainly via CYP2C19 and CYP3A4, to pharmacologically inactive metabolites.[68] CYP2C19 is involved mainly in the hydroxylation of PPIs to form their respective hydroxylated metabolites, and CYP3A4 is involved mainly in the oxidation of the sulfoxide group to form the respective PPI sulfones. A significant contribution of CYP2C19 to the metabolism of these PPIs was well documented by the findings that genetic polymorphism of CYP2C19 is associated with a large interindividual difference in the systemic and oral clearance of these PPIs.[2,69] For instance, poor metabolizers (PMs) of CYP2C19, who lack CYP2C19 activity, showed mean AUCs of omeprazole that were 7.5-fold greater than those obtained from extensive metabolizers (EMs) of CYP2C19. Because CYP2C19 and CYP3A4 are CYP isoforms that are expressed abundantly in the human liver, and it is considered that as many as 70% of all therapeutic drugs are metabolized by these CYP enzymes,[70] there is a possibility that coadministration of PPIs with drugs that have high affinity for either CYP2C19 or CYP3A4 may cause a clinically relevant DDI. In contrast, rabeprazole is converted non-enzymatically to its thioether analogue and is thereby less susceptible to DDIs associated with inhibition of CYP activity.

Li et al.[71] studied the potency and specificity of five currently available PPIs with regard to their inhibitory effects on the activities of four major human CYP enzymes (CYP2C9, CYP2C19, CYP2D6 and CYP3A4), using human liver microsomes and recombinant CYP2C19. They reported that all PPIs they studied were competitive inhibitors of CYP2C19 with rather low (potent) inhibition constant (ki) values. Among the PPIs studied, lansoprazole was the most potent inhibitor, with ki values ranging from 0.4 to 1.5 µmol/L. In addition, omeprazole showed ki values ranging from 2 to 6 µmol/L, esomeprazole showed a ki value of 8 µmol/L, pantoprazole showed ki values ranging from 14 to 69 µmol/L, and rabeprazole showed ki values ranging from 17 to 21 µmol/L. Pantoprazole was also a competitive inhibitor of CYP2C9 and CYP3A4, with ki values of 6 and 22 µmol/L, respectively, indicating that pantoprazole was at least twice as potent an inhibitor as other PPIs towards these CYP isoforms. All PPIs were weak inhibitors of CYP2D6, with ki values of >200 µmol/L. Because S-omeprazole (esomeprazole) has a less potent inhibitory effect on CYP2C19 than R-omeprazole, esomeprazole appeared to be a less potent inhibitor than clinically available racemic omeprazole. Rabeprazole thioether, a non-enzymatically formed metabolite of rabeprazole, was a potent and competitive inhibitor of CYP2C9, CYP2C19 and CYP2D6, with ki values of 6 µmol/L, 2–8 µmol/L and 12 µmol/L, respectively.[71] These results imply that PPIs may be associated with clinically relevant DDIs if standard oral doses of the respective PPIs produce drug concentrations exceeding the ki values described above for the respective PPIs in the vicinity of microsomal CYP2C19 and other CYP molecules.

2.2.2 Prediction of In Vivo Consequences of In Vitro Metabolic DDIs

Considerable efforts have been made to devise a strategy to predict the in vivo consequences of in vitro metabolic DDIs in humans. According to the US FDA’s recent guidance for industry regarding drug interaction studies,[72] the likelihood of in vivo DDIs may largely be projected with a simplified equation comprising plasma concentrations of inhibitors (i.e. [I]) over ki values (i.e. [I]/ki) unless the time-dependent mechanism-based inhibition of drug metabolizing enzymes is involved in the DDI.[73] So far, no studies have demonstrated that PPIs are involved in time-dependent metabolic inhibition of drug-metabolizing enzymes. In this context, it cannot be totally negated that some of the PPIs with low ki values (e.g. lansoprazole) may be associated with clinically relevant DDIs.[71]

Caution should be exercised when predicting in vivo consequences of the inhibition of CYP activity by PPIs, because PPIs are rather rapidly eliminated from the body.[2] In EMs of CYP2C19, all PPIs are eliminated with half-lives of <2.0 hours. As a result, under the standard dosing interval (once daily), PPIs may only produce significant inhibitory effects on CYP2C19 for a short period, if at all, relative to their dosing interval. In addition, PPIs will not produce any inhibitory effects on CYP2C19 activity in PMs of CYP2C19, because they lack CYP2C19 activity by nature. Indeed, Yu et al.[74] found that omeprazole significantly inhibited the metabolism of a CYP2C19 substrate, moclobemide, in homozygous EMs of CYP2C19*1/*1 but not in PMs (*2/*2, *2/*3, *3/*3). Andersson et al.[75] also reported that omeprazole decreased clearance of diazepam only in EMs, who were phenotyped by urinary excretion of omeprazole, but not in PMs. These findings are consistent with our knowledge that moclobemide is eliminated largely by hepatic metabolism mediated by CYP2C19,[76] and omeprazole is a potent inhibitor of CYP2C19.[71] However, to our knowledge, no evidence is available as to whether PPIs would elicit clinically relevant inhibitory effects on the metabolism of CYP2C19 substrates in subjects who are heterozygous EMs of CYP2C19.

Recently, attempts have been made to predict metabolic DDIs in a quantitative manner with use of ‘in silico’ model simulation programs that allow prediction of the magnitude of DDIs by incorporating in vitro drug interaction parameters (e.g. the ki) and inhibitor concentrations in the vicinity of drug-metabolizing enzymes throughout the dosing interval, with use of whole-body physiological pharmacokinetic models.[58,7779] It is of interest to see if such approaches would give estimations similar to those discussed in the present article regarding the metabolic DDIs associated with PPIs.

2.2.3 Induction of Drug Metabolism

Shortly after the introduction of omeprazole into market, Diaz et al.[80] reported that omeprazole induced CYP1A2 in primary cultures of human hepatocytes at the levels of mRNA, protein and enzyme activity (e.g. phenacetin demethylase). In addition, they reported that administration of omeprazole 20 mg/day for 4 days produced 2- to 10-fold induction of the CYP1A2 protein and CYP1A-dependent activities in liver biopsies obtained from cancer patients. Because CYP1A2 is involved in the metabolism of not only clinically important drugs (e.g. xanthines, paracetamol [acetaminophen]) but also many procarcinogens,[81] considerable interest was aroused among researchers. Subsequent studies[8284] also reported that omeprazole significantly induced in vivo CYP1A2 activity measured by caffeine metabolism (e.g. the *C-N-3-demethylation breath test) in PMs of CYP2C19 at a standard therapeutic dose (40 mg daily). They also found that the induction by omeprazole would be concentration dependent because omeprazole induced CYP1A2 activity only at a high dose (120 mg daily) but not at a standard dose (40 mg daily) in EMs of CYP2C19.[83] In contrast to these data, subsequent studies[8588] performed by other groups of investigators showed that three PPIs (omeprazole, lansoprazole and pantoprazole) failed to elicit significant induction of CYP1A2 activity, using CYP1A2 marker drugs (caffeine and phenacetin) in subjects whose CYP2C19 phenotypes were undefined. The reason why these controversial results were obtained by these different groups of investigators remains unclear. At present, the clinical implications of CYP1A2 induction by omeprazole are still largely obscure.

2.3 Alteration of Renal Elimination

Two recent case studies have raised a possibility that coadministration of PPIs at standard doses (20 mg twice daily or 40 mg daily) may delay elimination of plasma methotrexate in patients receiving high-dose methotrexate therapy independently of concomitant renal dysfunction.[89,90] Recent observational studies have demonstrated that administration of one of the PPIs (omeprazole, lansoprazole, rabeprazole or pantoprazole) either immediately prior to or concomitantly with methotrexate would delay the elimination of methotrexate without causing any concomitant renal dysfunction.[9,10] In vitro studies have demonstrated that PPIs inhibited efflux of methotrexate from the membrane vesicles expressing breast cancer resistance protein (BCRP).[10] Further prospective, controlled clinical studies are definitely required to confirm this unexpected but clinically important DDI.

2.4 DDIs with Undetermined Mechanism(s)

Current guidelines recommend use of a PPI to decrease the risk of gastrointestinal bleeding in patients undergoing PCI and stenting who are taking clopidogrel plus aspirin to prevent recurrent cardiovascular events[91] on the basis that use of this dual antiplatelet therapy has been shown to be superior to use of aspirin alone.[92,93]

A recent, large-scale, retrospective, cohort study (n = 8205)[7] suggested that use of a PPI plus clopidogrel in patients with acute coronary syndrome (ACS) after discharge may be associated with an increased risk of the composite outcome of death or rehospitalization, compared with use of clopidogrel alone (adjusted hazard ratio [HR] 1.25; 95% CI 1.11, 1.41). Approximately 60% of the PPI prescriptions were for omeprazole, and approximately 50% of the patients underwent PCI during hospitalization. Nevertheless, in analyses of secondary outcomes, coadministration of PPI with clopidogrel was associated with increased risks of recurrent ACS and revascularization procedures but not all-cause mortality. In addition, there were substantial differences in baseline characteristics between those who received clopidogrel with PPIs and those who received clopidogrel alone: the former group included greater proportions of patients with diabetes mellitus, heart failure and peripheral vascular diseases than the latter group. Two retrospective observational studies were conducted to examine whether coadministration of a PPI with clopidogrel at discharge may have a deleterious effect on clinical outcomes in patients who had undergone emergent PCI for acute coronary events (acute MI, unstable angina pectoris or ACS) or elective PCI for stable angina pectoris, compared with administration of clopidogrel alone. Evanchan et al.[94] revealed that PPI users (n = 1369) had higher rates of rehospitalization for acute MI (odds ratio [OR] 1.78; 95% CI 1.55, 2.07) than nonusers (n = 4425) within 1 year of stenting. Gaglia et al.[95] reported that PPI users (n = 318) had a significantly greater HR of 1.8 (95% CI 1.1, 2.7) for an adjusted major adverse cardiac event endpoint (consisting of death, Q-wave MI, target vessel revasculization and stent thrombosis) than nonusers (n = 502) at 1 year after discharge. A retrospective, matched-cohort study performed in patients discharged from hospitals after acute MI or PCI with stenting indicated that patients receiving both PPI and clopidogrel (n = 1033) showed a greater risk of rehospitalization for acute MI within 1 year (adjusted HR 1.93; 95% CI 1.05, 3.54) than those receiving clopidogrel alone.[96] In addition, a population-based, nested case-control study (n = 13 636)[8] suggested that patients with MI receiving clopidogrel with one of the PPIs (except for pantoprazole) had an increased risk of reinfarction after discharge, compared with those receiving clopidogrel alone (OR 1.27; 95% CI 1.03, 1.57). That study did not disclose the breakdown of the PPIs, except for pantoprazole. Finally, a meta-analysis performed with the data retrieved from 23 studies (n = 93 278) reporting the impact of PPI exposure with clopidogrel on the risk of major cardiovascular events and mortality indicated that an analysis of observational data adjusted for confounders showed a significant increase in the risk of major cardiovascular events but not mortality, whereas an analysis of propensity-matched or randomized trial participants showed no increase in the cardiovascular risk (acute MI or ACS) with PPI exposure.[97]

In contrast, in a reanalysis of data from TRITON-TIMI 38 (the Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition with Prasugrel-Thrombolysis In Myocardial Infarction 38), O’Donoghue et al.[98] found no association between PPI use and the risk of the primary endpoint (consisting of a composite of cardiovascular death, MI or stroke) in 13 608 patients with ACS. A limitation of this study was that use of PPIs was not randomized. A retrospective study based on three large cohorts of patients older than 65 years (n = 18 565) revealed that there might be a small increase in the risk of hospitalization due to MI or death, based on the multivariate-adjusted relative risk (RR 1.32; 95% CI 1.08, 1.61) in patients receiving both clopidogrel and a PPI, compared with those receiving clopidogrel alone. Nevertheless, the analysis performed with the propensity-adjusted data was not significant (RR 1.22; 95% CI 0.99, 1.51).[99] Taken together, these data suggest that the magnitude of the risk increase with PPIs is unlikely to exceed 20%, if it exists. Another recent cohort study performed with the data obtained from 20 596 patients who were hospitalized for MI, coronary artery revascularization or unstable angina pectoris showed no difference in the hazard ratio for serious cardiovascular events (fatal or nonfatal MI, sudden cardiac death, stroke or other cardiovascular death) between concurrent PPI users and nonusers. However, there was a significant 50% reduction in the risk of hospitalization for gastroduodenal bleeding in concurrent PPI users compared with nonusers.[100] A reanalysis of the data obtained from a prospective study investigating the 1-year mortality and morbidity of patients undergoing coronary stenting showed that patients receiving omeprazole and dual antiplatelet therapy (clopidogrel plus aspirin) had no significant increase in the primary endpoint (consisting of cardiac death or rehospitalization for acute MI), compared with those receiving dual antiplatelet therapy alone.[101] Collectively, the results obtained from these outcome studies imply that the possibility of DDIs between PPIs and clopidogrel, leading to clinically relevant consequences, cannot totally be negated. In this context, the FDA and the European Medicines Agency recommend that concomitant use of a PPI (particularly omeprazole) and clopidogrel be avoided unless absolutely necessary.[102,103] However, Geisler et al.[104] claimed that non-genetic factors (e.g. age, diabetes mellitus, decreased left ventricular function, renal failure, ACS) are also important for predicting responsiveness to clopidogrel. The lines of available evidence on this topic are summarized in figure 1a.
Fig. 1

Changes in the pharmacokinetics of clopidogrel and its metabolites, pharmacodynamic responses to the drug and clinical outcomes as consequences of (a) drug-drug interactions (DDIs) with proton pump inhibitors (PPIs) and (b) genetic polymorphism of cytochrome P450 2C19 (CYP2C19). CYP2C19 activity is reduced in both situations. () The antiplatelet activity of clopidogrel is decreased by omeprazole (OME) but not by esomeprazole (ESO) or pantoprazole (PAN).[98,105107] Patients receiving clopidogrel with a PPI have increased risks of reinfarction[8,94,96] and major cardiovascular events[95] but unchanged risks of cardiovascular death, myocardial infarction or stroke.[98101] OME increases the risk of rehospitalization but not death in patients receiving clopidogrel.[7] (b) The concentration of clopidogrel is increased and that of the metabolite is decreased in poor metabolizers (PMs) of CYP2C19 substrates. The antiplatelet activity of clopidogrel is decreased in PMs of CYP2C19 substrates,[108110] who have been reported to have increased risks of all-cause mortality, nonfatal myocardial infarction or stroke,[111] the composite endpoint of cardiovascular death, myocardial infarction or stroke[109] and cardiovascular events overall.[110]

The mechanism(s) responsible for this newly recognized DDI remain incompletely defined. However, the lines of evidence indicate that PPIs may attenuate the antiplatelet effects of clopidogrel by interfering with the formation of an active metabolite of clopidogrel. Clopidogrel is a prodrug and must be biotransformed to confer its anti-aggregation properties[112] (figure 1). While 85% of the dose of clopidogrel is metabolized to its inactive carboxylic acid derivative (SR26334) by plasma and tissue esterase, the remaining 15% of the dose is metabolized to an active metabolite (R-130964) by two sequential CYP-dependent biotransformation steps.[113,114] The active metabolite irreversibly binds to platelet P2Y12 adenosine diphosphate receptors, thereby inhibiting platelet aggregation.[112] Previous studies have revealed that CYP2C19 and CYP3A4 appear to play major roles in the metabolic activation of clopidogrel.[108,115117] It was shown that PMs of CYP2C19 with two loss-of-function alleles (CYP2C19*2/*2 or CYP2C19*3/*3) or a combined heterozygous genotype (CYP2C19*2/*3) had lower AUCs of the active metabolite of clopidogrel and exhibited attenuated inhibition of platelet aggregation, compared with EMs with the CYP2C19*1/*1 genotype.[117,118] These data suggest that PPIs may cause significant DDIs with clopidogrel if they interfere substantially with CYP2C19 activity.

Because PPIs are potent inhibitors of CYP2C19,[71,119] there is a possibility that PPIs may interfere with the CYP2C19-mediating biotransformation that is responsible for the formation of the active metabolite of clopidogrel. Gilard et al.[105] demonstrated that coadministration of omeprazole with dual antiplatelet therapy consisting of aspirin and clopidogrel significantly attenuated the inhibitory effect on platelet P2Y12, as assessed by vasodilator-stimulated phosphoprotein (VASP), in a randomized, double-blind, placebo-controlled study of 140 patients undergoing coronary stenting. Cuisset et al.[106] showed that use of omeprazole with clopidogrel in patients undergoing coronary stenting for non-ST-segment elevation ACS (n = 104) was associated with a greater attenuation in antiplatelet effects, assessed by VASP, than use of pantoprazole with clopidogrel. In addition, reanalysing the data from the PRINCIPLE-TIMI 44 (Prasugrel in Comparison to Clopidogrel for Inhibition of Platelet Activation and Aggregation-Thrombolysis in Myocardial Infarction 44) trial, where 201 patients undergoing elective PCI were randomly assigned to either prasugrel (n = 102) or high-dose clopidogrel (n = 99), O’Donoghue et al.[98] found that the patients receiving clopidogrel plus a PPI showed a significantly attenuated mean inhibition of platelet aggregation (23.2 ± 19.5%), compared with those receiving clopidogrel alone (35.2 ± 20.9%) at 6 hours after a 600 mg clopidogrel loading dose. In contrast, Siller-Matula et al.[107] failed to find significant DDIs between either pantoprazole or esomeprazole and clopidogrel in antiplatelet activity assessed by VASP in a nonrandomized study. An in vitro study has demonstrated that omeprazole is a more potent CYP2C19 inhibitor than pantoprazole and esomeprazole.[71] Nevertheless, because none of the previous studies measured plasma concentrations of the active metabolite of clopidogrel in patients receiving the drug with or without PPIs, it is difficult to determine whether omeprazole, but not pantoprazole and esomeprazole, would be associated with the attenuated antiplatelet effects of clopidogrel (figure 1a).

Considering the mechanisms associated with possible DDIs between PPIs and clopidogrel, it is worth referring to the results obtained from previous studies of the effects of genetic polymorphism of CYP2C19 on the pharmacokinetics and pharmacodynamics of clopidogrel (figure 1b). Mega et al.[109] reported that carriers of at least one of the CYP2C19 loss-of-function alleles had a relative reduction of 32.4% in the AUCs of the active metabolite of clopidogrel and had significantly reduced platelet aggregation, compared with that in noncarriers, based upon reanalysis of the data obtained from TRITON-TIMI 38. In addition, they revealed that carriers had a relative increase of 53% in the composite outcome of the risk of death from cardiovascular causes, MI or stroke, and had a 3-fold increase in the risk of stent thrombosis, compared with noncarriers. Simon et al.[111] reported that carriers of any two loss-of-function alleles of CYP2C19 had a higher rate of adverse outcomes (consisting of all-cause death, nonfatal MI or stroke) than noncarriers (adjusted HR 1.98, 95% CI 1.10, 3.58). Among the 1535 patients who underwent coronary intervention, the rate of adverse outcomes was 3.58-fold higher (adjusted HR 3.58, 95% CI 1.71, 7.51) in patients with two CYP2C19 loss-of-function alleles than in noncarriers. In addition, Shuldiner et al.[110] reported that patients carrying at least one CYP2C19*2 allele had a diminished platelet response to clopidogrel treatment and poorer cardiovascular outcomes (including the composite of MI, ischaemic stroke, stent thrombosis, unplanned revascularization and hospitalization) than non-carriers. Collectively, the totality of the pharmacokinetics, pharmacodynamics and clinical outcomes in relation to the genetically reduced CYP2C19 activity strongly suggests that further studies are definitely required to determine whether possible DDIs between PPIs and clopidogrel would have clinically relevant implications.

3. Pharmacodynamic Interactions

PPIs are considered mainstays of the combination pharmacotherapy that is prescribed for the eradication of H. pylori, which is associated with the development and remission of peptic ulcer disease.[2] PPIs are considered to enhance the anti-H. pylori activities of concomitantly administered antibacterials (e.g. amoxicillin, clarithromycin) by pharmacokinetic and pharmacodynamic mechanisms. It has been demonstrated that PPIs not only attenuate the breakdown of acid-labile antibacterials in the gastric fluid[120] but also enhance their pH-dependent antibacterial activity towards H. pylori.[121] To our knowledge, few other clinically relevant pharmacodynamic DDIs have been reported for PPIs.

4. Metabolic DDI Profiles of Individual PPIs

4.1 Omeprazole

4.1.1 Omeprazole as a Metabolic Inhibitor

Omeprazole is almost completely metabolized in the liver to its pharmacologically inactive metabolites, 5-hydroxyomeprazole and omeprazole sulfone, the formation of which involves CYP2C19 and CYP3A4, respectively.[122] Thus omeprazole may have the potential to interact with concomitantly administered drugs, the metabolisms of which are mediated by these CYP isoforms. We have summarized previous in vivo studies where metabolic DDIs associated with omeprazole were assessed in healthy subjects and in patients.

Gugler and Jensen[4] were the first to report that the administration of omeprazole at an oral dose of 20 mg once daily for 7 days decreased the systemic clearance of diazepam by 20–30%. They also found that omeprazole inhibited in vitro diazepam metabolism with human liver micorosomes in a competitive manner. Their findings were confirmed by subsequent studies.[75,123125] As expected, such a DDI was observed in EMs, but not in PMs, of CYP2C19.[75,124] Interestingly, there was an interethnic difference in the inhibitory effect of omeprazole on the diazepam metabolism. Caraco et al.[125] studied the effects of omeprazole on diazepam clearance in EMs of S-mephenytoin and found that the inhibitory effect of omeprazole was significantly greater in Caucasians than in Asians, and the mean baseline diazepam oral clearance in Caucasians was marginally greater than that in Asians. Their data are compatible with the idea that the inhibitory effects of omeprazole on diazepam metabolism are mediated via inhibition of CYP2C19 activity, and their Asian subjects would have consisted of a greater number of heterozygous EMs of CYP2C19 than their Caucasian subjects. Asians are known to have greater frequencies of loss-of-function alleles of CYP2C19 than Caucasians. In addition, omeprazole was shown to increase the systemic exposure of other CYP2C19 substrates (e.g. etravirine, proguanil, moclobemide and voriconazole)[74,126128] by 30–120% (table II). Collectively, the inhibitory effects of omeprazole on the metabolism of CYP2C19 substrates do exist, but their clinical implications would largely be insignificant, except for a few drugs, because the resultant changes in the AUCs of the victim drugs were in most cases <2-fold.
Table II

Effects of omeprazole on the area under the plasma concentration-time curve (AUC) values (fold increases) of concomitant drugs

While Gugler and Jensen[4] reported that omeprazole elicited a small, albeit significant, reduction (i.e. 15% from the control value) in phenytoin clearance, our meta-analysis performed in the present review on the pooled data retrieved from three single-dose studies and one multiple-oral-dose study revealed that omeprazole increased the AUC of phenytoin by 10% at most, and the change did not reach a statistically significant level (p = 0.10), indicating that any inhibitory effect of omeprazole on phenytoin metabolism would be small. Phenytoin was shown to be metabolized mainly by CYP2C9 and to a lesser extent by CYP2C19.[158]

Interestingly, omeprazole was also shown to increase the AUCs of nifedipine and carbamazepine by 26% and 89%, respectively.[66,132] Because the metabolism of these drugs is considered to be mediated mainly by CYP3A4[159,160] rather than CYP2C19, these data imply that omeprazole may also have an inhibitory effect on CYP3A4 activity. The ki value of omeprazole determined for the CYP3A4 marker reaction of midazolam 1′-hydroxylation (41.9 µmol/L) was greater than those determined for other CYP marker reactions. These data imply that DDIs between these drugs and omeprazole would be unlikely during hepatic metabolism, but do not categorically exclude the possibility of DDIs during intestinal CYP metabolism. In this context, further studies are required to verify the reproducibility of such isolated reports.

4.1.2 Omeprazole as a Victim of Metabolic Inhibition

At present, none of the drugs have been shown to alter the absorption of omeprazole. Because omeprazole is metabolized mainly by CYP2C19 and CYP3A4,[122] there is a possibility that concomitant administration of drugs with strong inhibitory effects on either CYP2C19 or CYP3A4 may alter the systemic exposure of omeprazole. Table III summarizes relevant in vivo data retrieved from the literature. Previous studies have demonstrated that strong inhibitors of CYP3A4 (e.g. clarithromycin, ketoconazole, fluconazole)[22,166,168,171] and CYP1A2 (e.g. fluvoxamine)[169] significantly increased the AUCs of omeprazole by 20–530%. In addition, a CYP2C19 substrate, moclobemide, increased the AUCs of omeprazole on average by 2-fold.[172] However, it remains unclear whether the augmented systemic exposure of omeprazole by other drugs may be associated with any detrimental effects.
Table III

Effects of concomitant drugs on the area under the plasma concentration-time curve (AUC) values (fold increases) of omeprazole and esomeprazole

In contrast, a well known CYP3A4 inducer, St John’s wort (hypericum), has been reported to decrease the AUC of omeprazole by 50% in patients with the CYP2C19*1/*1 genotype.[174] Because omeprazole is metabolized to omeprazole sulfone by CYP3A4, the change is likely to be attributable to the augmented metabolism of omeprazole via the CYP3A4-mediated pathway. In addition, administration of other herbal medicines (e.g. Ginkgo biloba, yin zhi huang and artemisinin) has been shown to decrease the mean AUCs of omeprazole by 20–50% from the control values.[165,170,173] Yin zhi huang produced smaller effects in patients with at least one defective CYP2C19 allele (either CYP2C19*2 or *3) than in those with the homozygous wild-type genotype, CYP2C19*1/*1. Because yin zhi huang is a decoction of Artemisia capillaris and three other herbs, the substance(s) responsible for the apparent induction of the metabolism of omeprazole await further study. Whether the reduced bioavailability of omeprazole is clinically relevant or not remains unclear at present.

4.2 Esomeprazole

Clinically available omeprazole is a racemic mixture of S-omeprazole (esomeprazole) and R-omeprazole. In theory, DDIs that are associated with omeprazole would also be relevant to esomeprazole (tables II and III). Nevertheless, there are subtle differences in the pharmacokinetics and metabolism of esomeprazole compared with racemic omeprazole. Because esomeprazole has lower systemic clearance than that of the omeprazole racemate, the administration of an equivalent dose of esomeprazole would result in a higher AUC than the omeprazole racemate.[176] In addition, in vitro and in vivo studies demonstrated that esomeprazole is predominantly metabolized by CYP3A4, and consequently its elimination is less dependent on CYP2C19.[177] Whether such apparently trivial differences would have any relevant implications for the DDI profiles of esomeprazole remains to be studied.

4.3 Lansoprazole

4.3.1 Lansoprazole as a Metabolic Inhibitor

Most of the clinical studies performed to assess the DDI potential of lansoprazole with diverse classes of drugs (theophylline, phenytoin, prednisone, warfarin, diazepam, oral contraceptives, ivabradine and methotrexate) revealed that lansoprazole would cause no appreciable changes in the AUCs of these drugs (table IV). However, several reports have indicated that concomitant administration of lansoprazole may increase the AUC8 and the dose-normalized trough plasma tacrolimus concentrations (Cmin/D) of tacrolimus by approximately 50% after both single and multiple administrations.[184187] The magnitude of DDIs was greater in patients with at least one loss-of-function allele of CYP2C19 (consisting of intermediate metabolizers [IMs] and PMs) than in those with homozygous EM alleles; lansoprazole was shown to increase the mean AUC8 of tacrolimus by 81% in IM/PM patients, whereas it increased the corresponding AUC by only 29% in homozygous EM patients. These findings imply that lansoprazole may inhibit CYP3A4-mediated tacrolimus metabolism in relation to its plasma concentrations. Miura et al.[192] studied interindividual variability in the AUC12 of tacrolimus in 73 renal transplant recipients who also received either lansoprazole (n = 40) or rabeprazole (n = 30), and found that the dose-adjusted AUC12 values obtained from the patients who were PMs of CYP2C19 and homozygous for the CYP3A5*3 allele (n = 4 each for lansoprazole and rabeprazole) were higher than the corresponding values obtained from the patients with other genotypes. Unfortunately, the differences did not reach statistical significance, mainly because of an insufficient number of patients. It remains to be studied whether this possible DDI could be attributable to inhibition of tacrolimus metabolism either in the liver or in the intestinal wall.
Table IV

Effects of lansoprazole and its enantiomer, dexlansoprazole, on the area under the plasma concentration-time curve (AUC) values (fold increases) of concomitant drugs

4.3.2 Lansoprazole as a Victim of Metabolic Inhibition

Coadministration of a selective and time-dependent CYP3A4 inhibitor, clarithromycin,[193] increased the AUC from time zero to infinity (AUC) values of orally administered lansoprazole by 1.55-, 1.74- and 1.80-fold in patients with the homozygous EM genotype, those with heterozygous EM genotypes and those with PM genotypes of CYP2C19, respectively. In addition, coadministration of fluvoxamine, an inhibitor of multiple CYP isoforms (CYP2C9, CYP2C19, CYP2D6 and CYP3A4),[194] increased the AUC of lansoprazole in patients with homozygous and heterozygous EM genotypes of CYP2C19 by 3.83- and 2.50-fold, respectively, but not in those with the homozygous PM genotypes.[195,196] In addition, clarithromycin and fluvoxamine increased the AUCs of both the R- and S-enantiomers of lansoprazole (table V).[197,198] The clinical implications of these findings remain obscure.
Table V

Effects of concomitant drugs on the area under the plasma concentration-time curve (AUC) values (fold increases) of lansoprazole, pantoprazole and rabeprazole

Dexlansoprazole is an enantiomer of lansoprazole, which is under development in the US and Japan for the treatment of gastric acid-related disorders. Clinical trials are underway with an innovative new modified-release formulation designed to prolong the plasma dexlansoprazole concentration-time profile following once-daily oral administration. Dexlansoprazole is metabolized mainly by CYP3A4 and CYP2C19.[2] To date, four clinical studies have been conducted in healthy subjects, and they showed that dexlansoprazole 90 mg administered orally once daily for 9 or 11 days gave rise to no appreciable changes in the AUCs of four model drugs (diazepam for CYP2C19, phenytoin and warfarin for CYP2C9, theophylline for CYP1A2).[191]

4.4 Pantoprazole

Pantoprazole is well absorbed after oral administration and undergoes little first-pass metabolism. It is extensively metabolized by CYP2C19 to its O-demethylated metabolite and by CYP3A4 to pantoprazole sulfone.[207] Although many clinical studies have been performed to explore the DDI potential of pantoprazole, they found no clinically relevant DDIs (table VI). Interestingly, unlike omeprazole, pantoprazole did not have an altered AUC when coadministered with clarithromycin (table V).[166] Collectively, these data confirm that pantoprazole should have a low DDI potential.
Table VI

Effects of pantoprazole on the area under the plasma concentration-time curve (AUC) values (fold increases) of concomitant drugs

4.5 Rabeprazole

The major metabolic pathway of rabeprazole (unlike those of other PPIs) is non-enzymatic reduction to rabeprazole thioether,[215] and the contribution of CYP2C19 and CYP3A4 to this pathway appears minor. Thus genetic polymorphism of CYP2C19 and metabolic inhibition of CYP2C19 or CYP3A4 by concomitant medications may have less prominent effects on the pharmacokinetics of rabeprazole than on those of other PPIs. Rabeprazole was reported to be associated with no significant changes in the AUCs of theophylline, phenytoin, warfarin and diazepam in subjects whose CYP2C19 genotypes or phenotypes were unknown. Detailed data from these studies are unavailable[5] because they were published only in abstract form. However, Ishizaki et al.[124] reported that rabeprazole gave rise to a small (8%), albeit significant, increase in the AUC of diazepam only in PMs of CYP2C19. The authors attributed the observed changes to rabeprazole-induced inhibition of CYP3A4, which mediates the metabolism of diazepam to its demethylated metabolite (table VII).
Table VII

Effects of rabeprazole on the area under the concentration-time curve (AUC) values (fold increases) of concomitant drugs

As expected, the CYP3A4 inhibitors clarithromycin and verapamil do not alter the AUC of rabeprazole, irrespective of CYP2C19 genotypes (table V).[204] Unexpectedly, however, an inhibitor of multiple CYP isoforms, fluvoxamine,[194] has been shown to increase the AUCs of rabeprazole and rabeprazole thioether by 2.8- and 5.1-fold, respectively, in homozygous EMs of CYP2C19 and by 1.7- and 2.6-fold, respectively, in heterozygous EMs, whereas it elicited no significant changes in PMs.[205] Antacid did not significantly alter the AUC of rabeprazole in healthy subjects (table V).[207]

5. Conclusions

In the last 10 years, we have witnessed substantial advances in our understanding of mechanisms associated with clinically relevant DDIs. Based upon accumulated knowledge, the FDA has issued a comprehensive guidance for industry (which describes drug interaction studies, including the study design, data analysis and implications for dosing and labeling[72]) and a subsequent supplemental publication.[216] Sophisticated simulation models are now available for predicting in vivo consequences of a given drug from in vitro evaluation of its inhibitory effects on the activity of major CYP isoforms and P-glycoprotein-mediated transport.[58,7779]

Because PPIs are widely prescribed, extensive efforts have been made to characterize their DDI potential. Because of their profound inhibitory effects on gastric acid secretion, all PPIs are associated with interference with the intestinal absorption of many drugs of which the solubility is decreased at neutral pH compared with acidic conditions. Because PPIs reduce the oral bioavailability of certain HIV protease inhibitors (atazanavir, darunavir and fosamprenavir) by >50% from the control values via this mechanism, it would be prudent to avoid coadministration of PPIs with these drugs.

PPIs may also be involved in DDIs associated with metabolic inhibition of concomitantly administered drugs. Above all, because omeprazole and lansoprazole have lower ki values (0.45–69 µmol/L) for the marker reactions of CYP2C19 (e.g. S-mephenytoin 4′-hydroxylation) than those of other PPIs, these two PPIs are most likely to be involved in a DDI. In contrast, in vivo DDIs of pantoprazole and rabeprazole are considered to be less likely, because their ki values are higher than those of the above two PPIs. There are subtle but not ignorable differences between different PPIs regarding the selectively and potency of CYP inhibition profiles. Whether such knowledge would be useful in predicting DDIs associated with PPIs remains largely unknown.

Recent clinical studies have provided controversial data regarding possible DDIs between PPIs and clopidogrel. Some studies[7,8] have suggested that coadministration of PPIs in patients receiving clopidogrel following PCI for the treatment of acute MI or ACS may attenuate the beneficial effects of clopidogrel and thereby increase adverse clinical events (e.g. reinfarction and the composite endpoint of death or rehospitalization), but another study did not.[98] We are now encountering a therapeutic dilemma between Scylla and Charybdis[217] as to whether PPIs are worth prescribing routinely in such patients. Collectively, the long-lasting saga of the DDI potential of PPIs was once considered almost over, but in fact this is not quite so.


No sources of funding were used to assist in the preparation of this review. The authors have no conflicts of interest that are directly relevant to the content of this review.

Copyright information

© Adis Data Information BV 2010