Drug-drug interactions of antithrombotic agents as a mechanism underlying the risk of gastrointestinal bleeding

Introduction

Antithrombotic therapy is widely used in cardiovascular pathologies: atrial fibrillation (AF), ischemic heart disease (IHD), acute coronary syndrome (ACS), venous thromboembolism (VTE), and in the presence of mechanical heart valve prostheses [1–3]. The clinical efficacy of modern antithrombotic drugs has been confirmed in large randomized clinical trials. Thus, in AF, switching to direct oral anticoagulants (DOACs) is accompanied by a reduction in the risk of hemorrhagic stroke and systemic emboli compared with warfarin [4, 5]. Within secondary prevention of IHD, monotherapy with P2Y12 inhibitors demonstrates a significant advantage over aspirin monotherapy in terms of the risk of myocardial infarction and hemorrhagic stroke [6].

Despite the high clinical efficacy of antithrombotic drugs, their use is associated with an increased hemorrhagic risk, which occurs in 5–10% of patients annually [7]. Gastrointestinal bleeding (GIB) is the most common variant of hemorrhagic complications; however, the incidence of GIB may vary depending on the underlying disease, patient age, and concomitant pathologies [1, 8, 9]. During aspirin monotherapy, the annual incidence of GIB is 1–3%; switching to dual antiplatelet therapy is accompanied by an increase to 2–3% [3]. The prognostic significance of hemorrhagic complications is determined by high mortality rates: 30-day mortality after major bleeding in the general population is 10–11%, among comorbid patients 17.7%, and in individuals older than 60 years 12–25% [3, 9, 10]. The recurrence rate of hemorrhages varies from 7.3 to 32.5% depending on the initial pathology and management tactics [11]. The localization of bleeding changes predictably over time with antithrombotic drug intake. In the first year after initiation of antithrombotic drugs, bleeding from the stomach and duodenum is more common, whereas in subsequent years bleeding in the small intestine prevails [12]. The clinical course also differs depending on the source localization. Bleeding from the upper GI tract is more severe: intravenous administration of proton pump inhibitors (PPIs) is required in approximately 80% of patients, red blood cell transfusion in more than 80%, and in some cases surgery or embolization is necessary. In lower GI bleeding, transfusion is required less often—in approximately 60% [13].

The pathophysiology of GIB occurring during DOAC intake involves not only systemic suppression of hemostasis but also a local damaging effect of unabsorbed drug residues directly on the mucous membrane—this mechanism is particularly characteristic of DOACs [14]. The risk of such bleeding increases significantly in older patients burdened by multiple comorbidity [15, 16]. An additional risk factor is the intake of ulcerogenic drugs, in particular nonsteroidal anti-inflammatory drugs (NSAIDs) [15], as well as concomitant prescription of antiplatelet agents, which increases the probability of bleeding through pharmacodynamic interaction [15]. It has been separately shown that combined use of DOACs with amiodarone is associated with an increased risk of GIB due to inhibition of CYP3A4 and P-glycoprotein (P-gp) and the associated violation of anticoagulant dosing [15]. Management of patients with already developed GIB who nevertheless require continuation of antithrombotic therapy is associated with the main clinical problem: the need for a balanced trade-off between the risk of hemorrhagic complications, on the one hand, and thromboembolic complications, on the other [17, 18, 19]. If therapy is resumed prematurely, this may be associated with an increased risk of either recurrent bleeding or death [17, 18]. If drugs are discontinued altogether or the pause in treatment is unjustifiably prolonged, the probability of ischemic stroke, systemic embolism, and in certain high-risk groups, death increases [18, 19, 20].

Traditional modifiable risk factors explain only 37.3% of the population attributable excess risk of GIB [21]. Among pathogenetically significant predictors, polypharmacy, recorded in 40–77% of comorbid patients, acquires a special role [22]. Combined intake of antithrombotic agents with NSAIDs, selective serotonin reuptake inhibitors (SSRIs), and glucocorticoids acts as an independent predictor of hemorrhagic complications [3, 23]. A separate pathogenetic group consists of drug–drug interactions detected in 53% of patients: of these, 48% are accounted for by pharmacodynamic mechanisms and 5% by pharmacokinetic mechanisms [22, 23].

Despite the existence of modern clinical recommendations, prediction of GIB in patients on antithrombotic therapy remains a difficult task. Existing risk scores do not take into account drug interactions and do not allow identification of all modifiable risk factors. This determines the need to analyze additional clinical-pharmacological predictors of hemorrhagic complications.

Objective

The aim of this work was to review the literature on drug interactions of antithrombotic drugs with other groups of drugs as a pathogenetic basis of gastrointestinal bleeding, to systematize information on pharmacodynamic and pharmacokinetic mechanisms of these interactions, to assess their clinical significance, and to substantiate optimal strategies for reducing hemorrhagic risk in everyday clinical practice.

Materials and Methods

A bibliographic search was performed in PubMed and eLIBRARY.RU using the following keywords and their combinations: “antithrombotic therapy,” “drug–drug interactions,” “gastrointestinal bleeding,” “pharmacokinetic interactions,” “pharmacodynamic interactions.” Inclusion criteria were full-text publications in Russian and English—systematic reviews, meta-analyses, cohort studies, review articles, and case reports published between 2020 and 2026; duplicate publications were excluded. The subject of analysis was data on drug interactions of antithrombotic drugs with drugs capable of potentiating the risk of gastrointestinal bleeding.

Classification of Drug Interactions

The classification of drug interactions of DOACs by mechanism of action assumes their division into pharmacokinetic and pharmacodynamic [24]. A broader descriptive framework is provided by drug-related problems (DRPs)—an erroneous choice of drug, dose, or route of administration is one possible cause, whereas drug–drug interactions are only one variant of such a cause [25, 26]. In outpatient settings, among all identified drug–drug interactions, pharmacokinetic mechanisms predominated, accounting for 61.3% [26]. With respect to interactions involving potentially inappropriate medications for the elderly, this figure was even higher, reaching 71.1%, whereas in hospital settings pharmacodynamic mechanisms predominate, accounting for 59.8% and 54.9% of interactions involving potentially inappropriate drugs [26].

Pharmacokinetic Interactions

Pharmacokinetic interactions are understood as situations in which one drug affects the absorption, distribution, metabolism, or excretion of another, as a result of which its plasma concentration changes. For DOACs, these interactions are caused mainly by inhibition or induction of P-gp and, to a lesser extent, by the CYP3A4 isoenzyme of the cytochrome P450 system [24, 27, 28].

P-gp is a transport protein; it limits the absorption of xenobiotics in the intestine, including drugs, and simultaneously ensures their excretion through the kidneys and liver [24, 29, 30]. All DOACs without exception are substrates of this transporter [24, 29, 30, 31]. When P-gp inhibitors of varying strength—strong (ketoconazole, cyclosporine) and moderate (verapamil, amiodarone, clarithromycin)—block its function, the bioavailability and plasma concentration of DOACs noticeably increase, and with them the risk of serious bleeding [24, 29, 30, 31, 32]. Thus, dabigatran exposure when combined with verapamil may increase 2.5-fold, and the effect is most pronounced if verapamil is taken one hour before dabigatran [24, 29]. Similarly, the combination of rivaroxaban with amiodarone increases the risk of bleeding [29]. P-gp inducers—rifampicin, carbamazepine, phenytoin, St. John’s wort—produce the opposite effect. By activating the transporter, they, on the contrary, accelerate DOAC elimination, as a result of which their blood concentration decreases [24, 29, 31]. The result is insufficient anticoagulation, with the associated risk of thromboembolic complications and stroke [29].

The degree of dependence of different DOACs on CYP3A4 is not uniform. It is most pronounced for rivaroxaban and apixaban; however, published data on the contribution of CYP3A4 to their hepatic metabolism differ substantially and are estimated at 18–50% and 20–25%, respectively [29, 30, 31, 32]. Dabigatran and edoxaban are practically not metabolized by this pathway [29, 31].

That is why combined inhibitors of CYP3A4 and P-gp—ketoconazole, HIV protease inhibitors—make the combined use of these combinations contraindicated [24, 29]. The opposite picture is observed with combination with CYP3A4 inducers. Thus, rifampicin reduces the exposure of rivaroxaban, apixaban, and edoxaban by 50%, 54%, and 35%, respectively [31].

The main enzyme responsible for warfarin metabolism is CYP2C9 [33]. A number of antibiotics—fluoroquinolones (ciprofloxacin, levofloxacin), trimethoprim/sulfamethoxazole—act as its inhibitors [33, 34]. By slowing warfarin degradation, they provoke an increase in the international normalized ratio (INR) and prolongation of bleeding time [33]. Rifampicin has the opposite effect: being a strong inducer of hepatic enzymes, namely CYP3A4, it, on the contrary, accelerates warfarin metabolism and reduces INR, forming a hypercoagulable state that requires an increase in the drug dose [33].

Clopidogrel, being a prodrug, acquires pharmacological activity only after conversion to the active metabolite, which requires the participation of CYP2C19 [35]. PPIs—primarily omeprazole and esomeprazole—suppress its activity, as a result of which clopidogrel metabolism is impaired, its efficacy decreases, and, consequently, the incidence of adverse cardiovascular events increases [35]. In the overall cohort of patients, PPI intake itself did not lead to a significant increase in the incidence of cardiovascular events. However, a significant association was found in patients initially classified as normal metabolizers, that is, those without LOF alleles of CYP2C19 [35]. It was in this group that intake of omeprazole or esomeprazole was accompanied by a reliable increase in the incidence of adverse cardiovascular events, comparable in magnitude to the indicators of carriers of LOF alleles. This phenomenon is termed phenoconversion—under the influence of drug–drug interaction, normal metabolizers effectively acquire the properties of “slow” metabolizers, thereby imitating genetically determined enzyme deficiency [35].

The absorption of certain DOACs, in particular dabigatran and edoxaban, depends on gastric acidity. PPIs used can reduce the maximum concentration and area under the curve (AUC) of dabigatran by 20–30% [24]. However, in practice, such a pharmacokinetic change is not considered a clinically significant basis for dose correction; moreover, PPIs are often prescribed to such patients specifically for the prevention of gastric bleeding [24].

Antibiotics, primarily penicillins and cephalosporins, destroy intestinal bacteria that synthesize vitamin K, reduce its level in the body, and thereby enhance the pharmacological effect of warfarin; for cephalosporins and amoxicillin/clavulanate, this occurs even in the absence of interaction with the cytochrome P450 system [33]. In certain representatives of the penicillin class (nafcillin, dicloxacillin), on the contrary, induction of cytochrome P450 predominates, weakening the warfarin effect. It is the combination of these multidirectional mechanisms that ultimately leads to sharp fluctuations in INR, increasing the risk of both bleeding and thrombosis [33].

The dose-dependent nature of such interactions should also be noted. Interactions are classified by the degree of change in DOAC AUC: strong ≥5-fold change, moderate 2–5-fold, weak less than 2-fold [36]. Thus, dronedarone increases dabigatran AUC by 70–100% [24]. In elderly patients receiving DOACs, the prevalence of pharmacokinetic interactions is 16.9% at admission, increasing to 20.7% by discharge [28]; the most frequent are combinations of amiodarone with apixaban (59.7% and 69.7%, respectively) and with rivaroxaban (26.0% and 21.8%) [28]. About 85% of oncological drugs do not require DOAC dose adjustment due to the absence of significant pharmacokinetic influence [36]. Information on the main pharmacokinetic interactions is presented in Table 1.

Table 1. Pharmacokinetic interactions of antithrombotic drugs

Pharmacodynamic Interactions

The essence of pharmacodynamic interactions is that co-prescribed drugs affect the hemostatic system, platelet function, or gastric mucosa independently of the concentration of the anticoagulant itself. That is why the risk of bleeding with such combinations increases substantially and directly [24, 26, 28, 37, 38]. Among the agents involved in such interactions, NSAIDs, antiplatelet agents, and SSRIs play the leading role [24, 39, 40]. As shown by a Danish study, about 18.6% of patients take antiplatelet agents in parallel with DOACs [41]. Such a combination increases the probability of hemorrhagic complications by 25–50% depending on the specific antiplatelet agent [42]. Special attention should be paid to so-called “triple therapy”—a combination of an anticoagulant with dual antiplatelet therapy. Such a combination is usually prescribed after coronary artery stenting or in acute coronary syndrome and can increase the risk of bleeding up to 27% [43]. The choice of the specific antiplatelet agent also matters: replacing clopidogrel with prasugrel in the combination regimen increases the probability of bleeding more than fourfold (odds ratio [OR] = 4.35) [43]. In a cohort of patients with a mean age of 77 years, adding clopidogrel to a DOAC was accompanied by an increase in the risk of bleeding of any localization by 37% (OR = 1.37) and gastrointestinal bleeding by 44% (OR = 1.44) [42].

The mechanism of action of NSAIDs is related to impaired prostaglandin synthesis, as a result of which both the protection of the GI mucosa decreases and platelet function is suppressed [41]. The frequency of their use during anticoagulant therapy varies from 10.9% in Denmark to 13.4% in the USA [41, 42]. In several cohort studies, the combination of NSAIDs with anticoagulants was associated with a substantial increase in the risk of complications [41, 44], although in one study (Medicare, USA) no statistically significant association was found [42].

A large cohort study conducted using the British database revealed a threefold increase in the risk of gastrointestinal bleeding (hazard ratio [HR] = 3.01) and an increase in the risk of major bleeding by 2.77-fold (HR = 2.77) [44]. Moreover, the addition of NSAIDs can attenuate the therapeutic effect of anticoagulants. The risk of stroke, including ischemic and hemorrhagic subtypes, in such patients increases 2.71-fold (HR = 2.71). There was a trend toward an increased risk of systemic embolism (HR = 3.02), but the difference did not reach statistical significance (95% CI 0.82–11.07) [44].

As for SSRIs, their effect on hemostasis is mediated by serotonin, which is necessary for full activation and aggregation of platelets. By inhibiting its reuptake, fluoxetine, paroxetine, citalopram, and other representatives of this group deplete platelet serotonin stores and thereby impair platelet function [37]. In addition, they can enhance gastric acid secretion, provoking ulcer formation [37]. In parallel with DOACs, about 16% of patients take such antidepressants [42]. According to a large meta-analysis covering 32 studies and more than 1.84 million patients, adding SSRIs to anticoagulant therapy increases the risk of bleeding of any localization by 39% (OR = 1.39) [37]. At the same time, the risk of major bleeding increases by 39% (OR = 1.39), gastrointestinal bleeding by 34% (OR = 1.34), and intracranial hemorrhage by 31% (OR = 1.31) [37]. If antidepressants are prescribed simultaneously with antiplatelet agents, the probability of major bleeding increases by 45% (OR = 1.45) [37]. The use of SSRIs during DOAC therapy significantly increases the frequency of both gastrointestinal and overall hemorrhagic complications [42].

The adverse effect on the GI mucosa is also exerted by systemic glucocorticoids themselves [42]. Their combination with DOACs reliably increases the risk of bleeding of any localization by 26% (OR = 1.26) and gastrointestinal bleeding by 33% (OR = 1.33) [42].

Pharmacodynamic interactions are presented in Table 2.

Table 2. Pharmacodynamic interactions of antithrombotic drugs

Drug-Related Problems

Drug-related problems (DRPs) are understood as problems associated with drug therapy that actually or potentially hinder achievement of the desired therapeutic result [25]. They include errors in the choice of drug, dosing regimen, and route of administration, as well as insufficient patient adherence to treatment [25, 45]. Among cardiac patients receiving antithrombotic therapy, such problems are detected with high frequency—in more than 75% of patients, according to a structured pharmaceutical review of pharmacotherapy [25]. In a study involving 8,432 patients, DRPs associated with errors in DOAC prescribing were recorded in 9.8% of participants (n = 827) [45]. Most DRPs are related to therapy safety, whereas effects on the effectiveness of drug treatment occur significantly less often [25]. The most common causes are incorrect drug selection and errors in dose selection [25]. Among DRPs with DOAC intake, violations of the dosing regimen predominate, most often due to incorrect assessment of renal function, which leads both to unjustified dose reduction and to exceeding recommended doses [45]. At the same time, among cases involving apixaban, dose underestimation was more common than overdose, whereas in the cohort overall, patients receiving rivaroxaban most often received an underestimated dose (63.2%) [45].

Clinical Significance

The localization of hemorrhagic complications is largely determined by interclass and intraclass heterogeneity of antithrombotic drugs; the distribution of risks by mechanisms, risk factors, and drug interactions depending on the specific anticoagulant is presented in Table 1 [46, 47]. The identified differences necessitate the use of standard algorithms. The choice of drug is based on a strict balance between systemic prevention of thromboembolism and the risk of local bleeding during intake of a specific anticoagulant [46, 47, 49].

The probability of hemorrhagic complications additionally increases in comorbid patients. Polypharmacy aggravates hemorrhagic risks and contributes to the development of life-threatening conditions with high short-term mortality [54]. The increase in the number of complications is primarily due to drug–drug interactions against the background of combined use of anticoagulants and antiplatelet agents [50, 53]. From the standpoint of pharmacodynamics, concomitant intake of NSAIDs, corticosteroids, or SSRIs directly damages the mucosa and suppresses primary hemostasis [52, 54]. In parallel, pharmacokinetic shifts develop: the effect of other drugs on cytochrome P450 isoenzymes and the P-gp efflux pump provokes fluctuations in anticoagulant bioavailability. The combined influence of these factors prevents adequate control of systemic hemostasis. As a consequence, severe refractory bleeding develops, sharply worsening the clinical prognosis [54].

Under such conditions, the bleeding risk scores HAS-BLED and ORBIT demonstrate only moderate and mutually comparable predictive ability [55]. Their accuracy is limited by a static set of criteria. These algorithms take into account basic demographic and anamnestic indicators but completely exclude the influence of drug–drug interactions and transient episodes of physiological decompensation [52]. Ignoring pharmacokinetic shifts against the background of polypharmacy naturally reduces the prognostic value of scoring systems in the management of comorbid patients [52, 55]. Risk assessment using standardized scores alone is insufficient. Adequate prevention of bleeding requires an individualized clinical-pharmacological approach that flexibly adapts to changes in the patient’s somatic status and adjustments in concomitant therapy [52, 55].

Risk Minimization

The initial stage of reducing hemorrhagic risks consists in optimizing the prescription list and eliminating modifiable factors [56]. First of all, it is necessary to avoid the use of NSAIDs for routine relief of pain. Concomitant intake of NSAIDs with oral anticoagulants or antiplatelet agents multiply increases the frequency of life-threatening gastrointestinal bleeding [56].

To ensure adequate protection of the digestive mucosa without aggravating the problem of polypharmacy, gastroprotection with PPIs is required; it is justified in patients with an initially high risk of hemorrhage—older individuals and patients receiving dual antiplatelet therapy [56, 57, 58]. In these cohorts, preventive therapy has proven efficacy and justifies the risks of polypharmacy. If additional risk factors are absent, routine preventive PPI use is clinically inappropriate, which requires a strictly individualized approach to mucosal protection [57].

Another key condition for safe antithrombotic therapy is individual dose selection against the background of regular monitoring of renal function [56]. The DOAC dosing regimen should be flexibly adjusted taking into account age, body weight, and creatinine clearance [56]. Thus, age over 75 years or body weight less than 50–60 kg significantly increases the plasma concentration of dabigatran and apixaban, potentiating the risk of complications [56].

The effectiveness of antithrombotic therapy directly depends on drug–drug interactions in the cytochrome P450 system, which can cause phenoconversion [56, 58]. In particular, omeprazole and esomeprazole competitively block the CYP2C19 isoenzyme required for clopidogrel bioactivation; therefore, their combined prescription is not recommended [58]. In such situations, transferring patients to alternative P2Y12 receptor inhibitors—ticagrelor or prasugrel—is advisable. The antiplatelet activity of these molecules is not subject to the influence of genetic polymorphisms with loss of function of the CYP2C19 gene [58, 59]. If clopidogrel discontinuation is impossible, the drugs of choice for gastroprotection become pantoprazole or rabeprazole. Due to minimal affinity for CYP2C19 and metabolism via alternative pathways, they do not reduce the effectiveness of antiplatelet protection [58].

Safe management of such patients is based on multidisciplinary interaction between a cardiologist, gastroenterologist, and clinical pharmacologist [56, 57]. At the same time, assessment of hemorrhagic risks should be continuous. Regular review of HAS-BLED values is required in combination with screening for Helicobacter pylori infection. Timely eradication of the pathogen is a mandatory condition for primary and secondary prevention of ulcer bleeding [56, 60].

The success of treatment is also determined by high-quality patient education, which increases adherence to prescribed regimens. Patients must be explained the danger of independent intake of over-the-counter analgesics. Unauthorized discontinuation of antithrombotic therapy is unacceptable.

A brief risk minimization algorithm is presented in Fig. 1.

Fig. 1. Algorithm for minimizing the risk of gastrointestinal bleeding during antithrombotic therapy
Source: Teplova N.V. et al., 2026

1. Patient assessment

   • Age

   • Body weight

   • Renal function (CC)

   • Previous GIB/peptic ulcer disease

   • Bleeding risk (HAS-BLED)

2. Analysis of concomitant therapy

   • NSAIDs

   • Antiplatelet agents

   • GCs

   • SSRIs

   • Inhibitors/inducers of CYP3A4, P-gp, CYP2C19

3. Therapy correction

   • Choice of antithrombotic drug

   • Individual dose selection

   • Discontinuation of NSAIDs

   • Elimination of clinically significant interactions

4. Gastroprotection

   • PPIs for high-risk patients

   • During clopidogrel therapy — preferably pantoprazole or rabeprazole

5. Dynamic follow-up

   • Monitoring of renal function

   • Repeat HAS-BLED assessment

   • Diagnosis and eradication of H. pylori

   • Patient education

Conclusion

GIB during antithrombotic therapy represents a multifactorial complication. The development of fatal hemorrhages is more often due to polypharmacy and drug–drug interactions than to the direct isolated action of an individual drug. The pharmacokinetic shifts that arise are based on competition of molecules for cytochrome P450 isoenzymes (in particular, CYP3A4, CYP2C19) and the P-gp efflux pump. This leads to unpredictable drug accumulation or to the development of phenoconversion. An additional burden is created by pharmacodynamic synergism: concomitant prescription of ulcerogenic drugs, such as NSAIDs, corticosteroids, or SSRIs, enhances damage to the mucous membrane. Against the background of vulnerable somatic status—advanced age and renal dysfunction—the combination of these factors multiply increases the probability of severe bleeding.

When assessing such risks, standardized HAS-BLED scores have limited prognostic value, since their static algorithms do not take into account the dynamics of concomitant pharmacotherapy. Treatment safety requires individualized clinical-pharmacological control, discontinuation of unjustified prescriptions, and targeted gastroprotection. The choice of PPI is made strictly taking into account their drug interaction profile. Monitoring of renal clearance and readiness to replace the antithrombotic drug when there is a threat of adverse reactions remain mandatory conditions.

The development of new prognostic models and clinical decision support systems is impossible without the use of continuously updated clinical data. Calculation algorithms must incorporate parameters of pharmacokinetic interactions, the patient’s genetic profile, and information on transient organ dysfunction. Only a multidisciplinary approach can ensure a clinical balance between adequate prevention of thromboembolism and minimization of the risk of massive bleeding.