Gastroprotection in patients on antiplatelet and/or anticoagulant therapy: a position paper of National Association of Hospital Cardiologists (ANMCO) and the Italian Association of Hospital Gastroenterologists and Endoscopists (AIGO)

The main risk factors for UGIB while taking ASA are: concomitant therapy with another antiplatelet agent (OR 7.4; 95% CI: 3.5-15)[1], previous peptic ulcer disease (OR 6.5; 95% CI: 2-21.2), concomitant therapy with NSAIDs (OR 2.9, 95% CI: 1.7-4.8), and H. pylori infection (OR 4.7; 95% CI: 2-10.9) [, ]. The risk of UGIB in patients with H. pylori infection is also doubled during therapy with non-aspirin antiplatelet agents when compared to ASA []. As regards age, a recent meta-analysis of 13 trials – with a median age of trial participants of 62 years (range, 53-74) - showed that aspirin use was associated with an increased risk of bleeding events compared with no aspirin (23.1 per 10,000 participant-years with aspirin and 16.4 per 10,000 participant-years) with no aspirin - HR 1.43; absolute risk increase 0.47%) []. The presence of systemic comorbidities, alcohol use, smoking, and a high BMI are still controversial risk factors []. Finally, in a study of patients aged >70 years, ASA was reported to increase the incidence of bleeding from colonic diverticula, both when it was given as monotherapy (OR 1.7; 95% CI: 1.2-2.3) and, specifically, when it was used in combination with other antiplatelet agents (OR 2.7; 95% CI: 1.8-4.1) [].

The risk factors associated with GI bleeding in VKAs anticoagulant users are: 1) age> 65 years (RR 2.5; 95% CI: 1.2-5.5); 2) a previous GI bleeding (RR 5.1; 95% CI: 1.9-13.5); 3) liver cirrhosis (RR 6.9; 95% CI: 2-24.5); and 4) presence of diverticulosis of the colon for the lower digestive tract [, ]. The risk of UGIB is also increased by the concomitant use of lipid-lowering agents (RR 1.8; 95% CI: 1.4-2.4), NSAIDs (RR 8.7; 95% CI: 7.3-10.4), ASA (RR 6.9; 95% CI: 5.9-8.28), and COX-2 inhibitors (RR 5; 95% CI 1.2-8.9) [].

Oral anticoagulant with direct antagonist action offers several advantages, including stable dosages, rapidity of action, and a short half-life compared to classic anticoagulants. However, recent studies have shown that the risk of GI bleeding was similar between DOACs and VKAs [, ]. The main risk factors associated with GI bleeding are: 1) age> 75 years (OR 2.5; 95% CI: 1.1-3.2 - > 70 years for dabigatran); 2) the presence of systemic comorbidities; 3) kidney failure; 4) a history of peptic ulcer complicated by bleeding (OR 2.5; 95% CI: 1.3-4.7); 5) concomitant use of ASA or NSAIDs (OR 1.8; 95% CI: 1.01-8.9) [, , ]; and 6) drug interactions with medications that share cytochrome P450 metabolism, such as amiodarone, rifampicin, barbiturates and fluconazole []. The bleeding risk was also dependent by the type of DOAC used, with higher values ​​for rivaroxaban and lower values ​​for apixaban []. Table 1 summarizes the main gastrointestinal bleeding risk factors for the different classes of drugs [, , , , , , , , , , , ].

The risk/benefit ratio between prevention of ischemic events and increase of bleeding events due to aspirin therapy used as primary prevention is still controversial. The meta-analysis of the Antithrombotic Trialists' Collaboration published in 2009 - which included 95,000 patients - showed that aspirin allocation increased major (usually defined as a bleed requiring transfusion or resulting in death) gastrointestinal and extracranial bleeds (0.10% vs. 0.07% per year, p<0.0001; RR 1.54; 95% CI: 1.30-1.82) []. The incidence of major GI bleeding is even higher when LDA is used in association with clopidogrel in patients with previous GI bleeding or undergoing prolonged antiplatelet therapy []. Finally, the risk of bleeding is often proportional to cardiovascular risk. Although it does not act on cyclo-oxygenase, the harmful action of clopidogrel on the gastric mucosa has been widely demonstrated.

The CAPRIE study[15] showed that clopidogrel has a lower risk of gastric damage and GI bleeding than aspirin (not considering the use of PPI). Thus, clopidogrel, obviously with appropriate PPI association, appeared to be a safer antiplatelet monotherapy in patients with recent damage due to aspirin. Clopidogrel administration seems to be associated with both an impaired spontaneous healing process of gastric ulcers, and haemorrhagic diathesis due to the inherent antiplatelet effect []. However, the use of clopidogrel in patients with previous history of peptic ulcer disease (with or without complication) seems to be questionable [, ]. In two well-performed RCTs, patients who took aspirin to prevent vascular diseases and presented with ulcer bleeding, after the ulcers had healed, and H. pylori infection was eradicated if present, were randomized to receive either clopidogrel or LDA plus 20 mg of esomeprazole twice daily for 12 months. The cumulative incidence of recurrent bleeding during the 12-month period was significantly higher among patients who received clopidogrel compared to those treated with aspirin plus esomeprazole in both trials (difference for clopidogrel from 3.4 to 20.9) [, ]. Hence, the ACCF/ACG/AHA 2008 Expert Consensus Document on the prevention of GI risk of antiplatelet therapy did not recommend the substitution of clopidogrel for LDA in high-risk patients to reduce the risk of recurrence ulcer bleeding []. Finally, a very low dosage of aspirin (75 mg) could be also considered after GI bleeding in patients previously treated with higher dose of aspirin [].

In the CURE trial, the risk of major bleeding, and in particular GI bleeding, was clearly higher in patients undergoing DAPT with aspirin and clopidogrel compared to those treated with aspirin alone (1.3% vs. 0.7%) []. When compared with ticagrelor, clopidogrel seems to be safer with regard to GI-related risks, including fewer overall GI/anal bleeding events and spontaneous GI haemorrhagic episodes, less nausea, vomiting, dyspepsia and diarrhoea, and a lower rate of presence of H. pylori [, , ]. Prasugrel also causes a significant increase in digestive bleeding from 45 days after the start of therapy, but less data are available compared to ticagrelor and clopidogrel [, ]. Furthermore, it should be considered that the prolongation of DAPT beyond 12 months in selected cases can be indicated. While this leads to a significant reduction in ischemic events, it might confer an increased risk of GI bleeding. Therefore, it is desirable to carry out a correct stratification of bleeding risk without disregarding gastrointestinal symptoms that could be the expression of an unknown underlying disease. A recent meta-analysis of RCTs aimed to compare the risk of gastrointestinal bleeding (GIB) among users of third generation P2Y₁₂ inhibitors with clopidogrel []. The results showed that third generation P2Y₁₂ inhibitors were associated with higher risk of upper GI bleeding (RR 1.32, 95% CI 1.05‐1.67) and unspecified GI bleeding (RR 1.25, 95% CI 1.01‐1.53), but not lower GI bleeding (RR 1.25, 95% CI 0.95‐1.65).

The guidelines recommend the early use of risk stratification in patients with ACS. This stratification should include both an ischaemic risk assessment (preferably on the basis of the DAPT score), and a bleeding risk evaluation (with the PRECISE-DAPT score) [, ]. However, patients enrolled in RCTs are often different from those seen in the real practice in term of clinical characteristics, risk factors, and comorbidities. For these reasons, it is important to assess evidence coming also from clinical practice. In a Danish registry, 46,301 patients hospitalized with an acute myocardial infarction (AMI) (35% of whom were at high risk of GI bleeding) were treated with clopidogrel (76.2%), ticagrelor (20.3%), or prasugrel (3.5%) []. At 12 months, an episode of digestive bleeding occurred in 1.0% (95% CI: 0.9-1.1) of patients with low risk of bleeding, and in 1.7% (95% CI: 1.5-2.0) of those at high risk. In patients receiving PPIs, the absolute risk of bleeding was overall reduced by 0.44% (95% CI: 0.39-0.48), and by 0.47% (95% CI 0.43-0.51%) in those at high risk. In a Spanish observational register with 1,219 patients undergoing percutaneous angioplasty (PTA), 96.7% of whom were in DTA with LDA and clopidogrel, and the 76.6% on PPI therapy, 8 patients developed GI bleeding during hospitalization, and 27 during follow-up (1.52 bleeds per patient/year) []. Most GI bleeding (81.4%) occurred during the first year. Overall, 84.6% of patients were on long-term PPI at the time of the bleed, and indeed, lower GI bleeding occurred more frequently than upper GI bleeding (74% lower vs. 26% upper) []. A study performed in New Zealand on 66,500 patients aged ≥65 years showed no significant change in the incidence of GI bleeding in patients using ASA (ARR 0.84; 95% CI: 0.79-0.89), or with clopidogrel (ARR 0.97; 95% CI: 0.87-1.08) in monotherapy, while dual therapy resulted in an increased bleeding (ARR 1.34, 95% CI: 1.14-1.57) []. In addition, the incidence of bleeding was increased when an anticoagulant therapy was associated with ASA (ARR 1.79; 95% CI: 1.30-2.46), clopidogrel (ARR 6.36; 95% CI: 2.24-18.03), or dual therapy (ARR 4.85; 95% CI: 1:51 to 15:57) []. A case-control study involving 669,115 patients assessed the incidence of GI bleeding evaluating the age of the patients and use of different treatments []. The incidence of bleeding observed was 76 per 100,000 people/year, with a prevalence higher in men than in women (109 vs. 49 per 100,000 people/year; OR 2.23; 95% CI: 1.99-2.50), and with a progressive increase of incidence with age. The risk was significantly higher in patients with a previous episode of GI bleeding (RR 11.27, 95% CI: 8.35-15.20), in those using ASA (medium-high doses: RR 3.29; 95% CI: 1.42-7.62; low doses: RR 1.74; 95% CI: 1.37-2.21), and other antiplatelet agents (RR 1.73; 95% CI: 1.27-2.36). On the contrary, risk was not significantly higher in those using steroids (RR 1.11; 95% CI: 0.66-1.86), or paracetamol (RR 1.00; 95% CI: 0.82-1.23) []. Among the NSAIDs, aceclofenac (RR 1.02; 95% CI: 0.51-2.02), diclofenac (RR 1.32; 95% CI: 0.87-1.98), and ibuprofen (RR 1.33; 95% CI: 0.94-1.89) were the medications with the lower risk of bleeding if prescribed concomitantly with the antiplatelet agents compared to lornoxicam (RR 13.57; 95% CI: 4.61-29.93), piroxicam (RR 4.49; 95% CI: 2.36-8.54), and desketoprofen/ketoprofen (RR 3.79; 95% CI: 1.70-8.46). The results of a prospective, observational multicentre cohort trial (GRAPE), performed in patients with ACS undergoing PTA - 48% of whom treated with clopidogrel, 18% with prasugrel, and 35% with ticagrelor in dual therapy with LDA - showed that the rate of MACE was lower in prasugrel treated patients (4.4%) than in clopidogrel treated patients (10.1%) (HR 0.53; 95% CI: 0.30–0.91), although not significantly different between ticagrelor (6.8%) and clopidogrel groups (HR 0.78; 95% CI: 0.54–1.12) []. Any type of bleeding was more frequent in prasugrel-treated patients (51.2%) than in clopidogrel-treated patients (37.6%) (HR 1.61; 95% CI: 1.33–1.95), and more frequent in ticagrelor-treated patients (56.9%) than in clopidogrel-treated patients (HR 1.81; 95% CI: 1.55–2.10), but there was no significant difference for fatal events []. However, bleeding is often associated with a discontinuation of a DAT, with harmful consequences on ischemic events in the medium and long term []. Similar data emerged from the Swedish register SWEDEHEART, where 45,073 patients with ACS were enrolled []. At 24 months, the risk of ischemic events, death, and MI was lower with ticagrelor than to clopidogrel (11.7% vs. 22.3%, adjusted HR 0.85, 95% CI: 0.78–0.93; 5.8% vs. 12.9%, adjusted HR 0.83, 95% CI:0.75–0.92; 6.1% vs. 10.8%, adjusted HR 0.89, 95% CI: 0.78–1.01; respectively) []. However, re-admission with bleeding was lower in patients taking ticagrelor as compared to those taking clopidogrel: 5.5% vs. 5.2%; adjusted HR 1.20; 95% CI: 1.04–1.40. Furthermore, in a subset of patients undergoing PCI, the PCI-related in-hospital bleeding was higher in patients on ticagrelor compared to those in clopidogrel (7% vs.2.7%, adjusted OR 1.57, 95% CI: 1.30–1.90) []. Furthermore, a sub-analysis of data in patients with a reduced renal function, showed a correlation between clearance creatinine and bleeding events []. There are few studies in the literature comparing the prevalence of upper GI lesions between aspirin and clopidogrel users. In addition to the CAPRIE study [], in 2012, an observational study aimed to investigate the characteristics of endoscopic findings in clopidogrel or aspirin users undergoing endoscopy for upper gastrointestinal symptoms []. Authors found that gastroduodenal haemorrhagic spots were more common between clopidogrel users than aspirin users (10.1% vs. 25.5%, p= 0.004). Gastroduodenal erosions were barely more common between the aspirin users than clopidogrel users (53.2% vs. 38.7%; p=0.04), whilst gastroduodenal peptic ulcers were more common between the clopidogrel users than the aspirin users (38.7% vs. 23.9%; p= 0.027). Similarly, in 2015, a case-control study found that non-aspirin antiplatelet agents (mostly clopidogrel) were associated with an increased risk of UGIB that was similar to that observed with ASA []. Finally, in 2019, a retrospective study on 1119 ischemic stroke patients found that the HR for GIB was significantly in favour of clopidogrel (HR: 2.60; 95% CI: 2.82 - 3.70) [].

GI bleeding represents an adverse event associated also with oral anticoagulant therapy. In two Meta-analyses of phase 3 registration studies, DOACs increased the risk of major GI bleeding by 23-25% compared to warfarin in patients with AF, even if this difference was barely significant [, ]. In the RE-LY trial, when patients with AF with a mean age of 71 years were evaluated, the use of dabigatran 150 mg bid was associated with a higher risk of GI bleeding than warfarin (RR 1.50; 95% CI: 1.19-1.89), while there was no significant difference between dabigatran 110 mg bid and warfarin (RR 1.10; 95% CI: 0.86-1.41) (Fig. 1) []. However, the increased risk of bleeding observed with the higher doses of dabigatran in this study, could be attributed to blind administration of dabigatran 150 mg also in very fragile patients []. The European (EU) label recommends the dose of 150 mg bid in patients with AF who are aged < 80 years without an increased risk for bleeding (e.g., HAS-BLED score <3), and not on concomitant verapamil, whilst in other patients, the dose of 110 mg bid is recommended. []. In a post-hoc simulation using the RE-LY dataset, it was therefore assessed how dabigatran, used according to the EU label, would compare to well-controlled warfarin treatment (INR 2–3; median time in therapeutic range 67.3%). The results of this simulation showed that was no difference between the two medications in term of major GI bleeding (HR 1.23; 95% CI: 0.96–1.59), although this was barely not significant []. In the ROCKET-AF trial, patients randomized to rivaroxaban had a significantly higher incidence of GI bleeding than warfarin (3.15% vs. 2.16%, p <0.001; Fig. 1), although the frequency of fatal bleeding was comparable between the two []. In the ENGAGE AF-TIMI 48 study, there was an increased risk of major GI bleeding with the high dose of edoxaban (60 mg daily) compared to warfarin (HR 1.23; 95% CI: 1.02-1.5) although this was barely significant (Fig. 1) []. However, in patients who received the low dose of edoxaban (30 mg daily), the rate of G bleeding was significantly lower compared to warfarin (HR: 0.67; 95% CI: 0.53 to 0.83) (Fig. 1).

The only DOAC that does not appear to be associated with an increase in GI bleeding greater than warfarin is apixaban. Indeed, in the ARISTOTLE trial, there was a comparable incidence of major GI bleeding between apixaban 5 mg bid and warfarin (HR 0.89; 95% CI: 0.7-1.15) (Fig. 1) []. Therefore, the Working Group of the European Society of Cardiology suggests the use of apixaban in patients with AF at high risk of bleeding from the GI tract []. However, there are no RCTs where the different DOACs are compared directly. It is also useful to highlight that GI bleeding site can vary between the different DOACs [, ]. Bleeding events during dabigatran therapy affect the lower digestive tract in 53% of cases []. This could be due to the incomplete absorption of dabigatran in the upper GI tract, and to the greater availability of the medication in the colon, where it would induce bleeding from pre-existing lesions, such as angiodysplasias and erosions [, ]. On the contrary, with rivaroxaban, bleeding is more frequent in the upper tract (76% vs 24%), while with apixaban 10 mg/day there is no difference in the incidence between upper and lower bleeding []. Single administration of rivaroxaban results in a higher peak of the drug than fractional administration of apixaban [, ]. This may explain the higher incidence of rivaroxaban-induced bleeding compared to apixaban, although they have a similar bioavailability. The same mechanism has been proposed to elucidate the higher relative incidence of major GI bleeding observed with rivaroxaban, administered once daily, compared to dabigatran administered twice daily [, ]. Finally, in a meta-analysis aimed to assess the risk of GIB and clinically relevant bleeding in patients taking DOACs, it was found a total of 1.5% of GIB, 89% of which represented by major events []. The overall OR for GIB among patients taking DOACs, when compared to standard anticoagulant therapy, was 1.45 (95% CI:1.07 -1.97). Among the DOACs, only dabigatran (OR 1.58; 95% CI: 1.29-1.93) and rivaroxaban (OR 1.48; 95% CI: 1.21-1.82) were associated with a significantly increased risk of bleeding [].

The "real world" evidence show a higher incidence of GI bleeding with some DOACs compared to warfarin []. In a retrospective study, the risk of GI bleeding was assessed in a cohort of 92,816 patients taking anticoagulants (9.2% patients on dabigatran, 17.5% on rivaroxaban, and 73.2% on warfarin) []. The data allowed a head-to-head comparison between warfarin and each of the two DOACs for homogeneous subgroups of patients, with or without AF, and also by location of bleeding. Using a propensity score matched models, the risk of GI bleeding with DOACs was similar to that with warfarin in AF patients (dabigatran vs. warfarin: HR 0.79; 95% CI: 0.61 -1.03; rivaroxaban vs. warfarin: HR 0.93; 95% CI: 0.69 - 1.25) and in non-AF patients (dabigatran vs. warfarin: HR 1.14; 95% CI: 0.54 to 2.39; rivaroxaban vs. warfarin: HR 0.89; 95% CI: 0.60 - 1.32) []. However, the risk of bleeding increased with age, so that in patients aged ≥76 years, the risk exceeded that with warfarin among AF patients taking dabigatran (HR 2.49; 95% CI: 1.61 - 3.83) and in patients with and without AF taking rivaroxaban (HR 2.91; 95% CI: 1.65 - 4.81; and HR 4.58; 95% CI: 2.40 - 8.72; respectively) []. Recently, a retrospective analysis was conducted on patients with non-valvular AF treated with dabigatran, rivaroxaban or apixaban. Data from three patient cohorts were compared, including 31,574 case treated with rivaroxaban vs. dabigatran, 13084 cases in apixaban vs. dabigatran, and 13,130 patients in apixaban vs. rivaroxaban []. Gastrointestinal bleeding was more frequent, even if not statistically significant, in rivaroxaban patients compared to dabigatran (HR 1.20; 95% CI: 1.00-1.45), confirming the result of other observations. Apixaban, on the other hand, showed a lower risk than dabigatran (HR 0.39; 95% CI: 0.27-0.58; p <0.001) and rivaroxaban (HR 0.33; 95% CI: 0.22-0.49; p <0.001). In the three subgroups, the frequency of bleeding events increased in patients over 75 years of age []. Interestingly, a retrospective cohort study found that the incidence of hospitalization for upper gastrointestinal tract bleeding, among users of NOACs or warfarin, was lower among patients who were receiving PPIs co-therapy compared to those not receiving PPIs []. Apixaban showed a lower risk of GI bleeding also in elderly subjects compared to dabigatran (HR 0.45; 95% CI: 0.29-0.71) or rivaroxaban (HR 0.39; 95% CI: 0.25-0.61) (Table 2) [, ]. Other "real world" evidence raised from post-marketing pharmacovigilance data. A recent Italian study analysed data coming from the Adverse Reactions of Medicines Report System of the National Pharmacovigilance Network, between January and December 2017 for all DOACs [], and it also evaluated the locations of the reported bleeding adverse events (Fig. 2). The study showed that the percentage of overall bleeding events per site was substantially overlapped between the various DOACs, with a peak (12-15%) corresponding to the GI system, which was expected as a "class effect". Even if there was a slight increase in some of the minor locations considered, the lower rate of anaemia might suggest a low clinical relevance and an easy management of these bleedings.

In the absence of RCTs, a large retrospective observational study in patients taking warfarin showed that PPI co-therapy, among those concurrently using antiplatelet drugs or NSAIDs, significantly reduced the risk of hospitalization for UGIB (HR 0.55; 95% CI: 0.39- 0.77) compared to those not using PPI co-therapy, with a NNT of 78 []. On the contrary, in the absence of a concurrently use of antiplatelet drugs or NSAIDs, PPI co-therapy in patients taking warfarin did not significantly reduce the risk of hospitalization for UGIB (HR 0.86, 95% CI 0.70-1.06) []. A recent cohort study with 1,643,123 patients on anticoagulant therapy with an average age of 76.4 years - mainly for atrial fibrillation – compared the data of 754,389 person-years on anticoagulant treatment without PPIs (516,512 cases on warfarin) with those of 264,447 person-years in co-therapy with PPIs (183,929 on warfarin) []. The incidence of hospitalizations for GI bleeding was significantly reduced in patients receiving warfarin and PPIs (74 episodes per 10,000 person-years) compared to the group receiving warfarin but not treated with PPIs (113 episodes per 10,000 person-years), after adjustment for numerous covariates including also use of antiplatelet agents, steroids, and NSAIDs []. It should be noted that there is a pharmacological interaction between PPIs and warfarin due to their liver metabolism. Therefore, international normalized ratios should be monitored when initiating or discontinuing PPI therapy in patients taking VKAs [].

DOACs have a different pharmacokinetic from warfarin. Dabigatran also differs from other DOACs due to its low bioavailability and is administered as a prodrug (dabigratan etexilate). Since its absorption increases in an acidic environment, the capsules are intended for a release in the stomach and the drug is absorbed in the distal small intestine. The co-administration of PPIs involves a lowersolubility of dabigatran with a reduction of about 30% of its absorption, without however changing its clinical efficacy [, , ]. Furthermore, it has been observed that dyspeptic symptoms arising in some patients taking dabigatran can be successfully treated with PPIs without clinically relevant interactions []. Recently, a large retrospective study aimed to compare the incidence of hospitalization for UGIB in patients using individual anticoagulants (VKAs or DOACs) with and without PPI co-therapy, according to presence of GI risk factors []. In patients receiving anticoagulant treatment without PPI co-therapy, the adjusted incidence of hospitalization for UGIB for rivaroxaban was significantly greater than the incidence for apixaban (incidence rate ratio: 1.97; 95% CI: 1.73-2.25), dabigatran (incidence rate ratio: 1.19; 95% CI: 1.08-1.32), and warfarin (incidence rate ratio 1.27; 95% CI: 1.19-1.35]. For patients receiving anticoagulant treatment with PPI co-therapy, the adjusted incidence of hospitalization for UGIB was lower than the incidence in patients receiving treatment without PPI co-therapy (incidence rate ratio: 0.66; 95% CI: 0.62-0.69), and was significantly lower for each individual anticoagulant evaluated []. This raises the possibility that the use of PPIs might be able to reduce the risk of UGIB in all patients taking oral anticoagulants (VKAs or DOACs) regardless of the presence of risk factors. However, more studies are needed to make recommendations. The adjusted incidence of hospitalization for UGIB in patients with PPI co-therapy was most pronounced for dabigatran, probably explained by dabigatran-related upper gastrointestinal tract lesions that are potentially the result of direct mucosal injury by the drug's tartaric acid core [].

Concomitant treatment with gastroprotective agents - both PPI, which demonstrated a slightly better protective effect, and H₂RA - was associated with a 50% reduction of risk of GI bleeding, with the maximum effect when the two drugs were used in association []. This protective effect appears to be confined to the upper GI tract, and to patients with previous peptic disease or GI bleeding, being mediated by the reduction of bleeding from pre-existing ulcers []. The impact of gastroprotective agents is also significant in patients treated with dabigatran, but modest with rivaroxaban []. The COMPASS study, recently published, evaluated 17,598 patients with stable cardiovascular disease and peripheral arterial disease who were randomized to receive pantoprazole 40 mg daily or placebo, as well as rivaroxaban 2.5 mg twice daily with aspirin 100 mg once daily, rivaroxaban 5 mg twice daily, or aspirin 100 mg alone for 3 years []. There were no significant differences in clinical events (overt bleeding of gastroduodenal origin confirmed by endoscopy or radiography, overt upper GI bleeding of unknown origin, bleeding of presumed occult upper GI tract origin with documented decrease in Hb ≥2 g/dL, GI pain with underlying multiple gastroduodenal erosions, symptomatic gastroduodenal ulcer, upper GI obstruction or perforation). However, as upper GI events as defined in the study protocol were uncommon in this trial, in a post-hoc analysis the definition was broadened. This analysis showed that pantoprazole therapy significantly reduced overt bleeding of gastroduodenal origin with no need for active bleeding at time of investigation (HR 0.45; 95% CI: 0.270.74), gastroduodenal ulcer without the need for symptoms prior to diagnosis (HR 0.46; 95% CI: 0.25-0.83), and multiple gastroduodenal erosions without the need for symptoms prior to diagnosis (HR 0.33; 95% CI: 0.13-0.84), particularly in patients treated with ASA) []. On the other hand, the same study showed that there was a statistically significant difference between the pantoprazole and placebo groups for enteric infections (1.4% vs 1.0% in the placebo group; OR 1.33; 95% CI: 1.01–1.75). For all other safety outcomes (gastric atrophy, chronic kidney disease, dementia, pneumonia, fracture, chronic obstructive pulmonary disease, diabetes mellitus, cumulative incidence of combined and individual cardiovascular death, myocardial infarction, and stroke, hospitalization rates, and all-cause mortality) proportions were similar between groups except for C. difficile infection, which was approximately twice as common in the pantoprazole vs. the placebo group, although there were only 13 events, so this difference was not statistically significant []. In cirrhotic patients, current guidelines recommend all oral anticoagulants (VKAs or DOACs) in patients with Child-Pugh class A, while contraindicate their use in patients with liver disease associated with coagulopathy and clinically relevant bleeding risk, including Child-Pugh C. DOACs are variably recommended in Child-Pugh class B cirrhosis [, ]. In these patients, PPI should be recommended according to guidelines [, , ].

H. pylori infection and use of aspirin are two main risk factors in the pathogenesis of peptic ulcers and peptic ulcer complications. However, not all patients H. pylori positive or taking aspirin develop PUD, suggesting that individual susceptibility to the infection and drug toxicity are critical factors to the initiation of mucosal damage [, ]. In 2010, a systematic review aimed to determine the influence of the infection on the risk of UGIB in patients taking ASA or LDA. However, the heterogeneity of the studies and the controversial results did not allow for strong conclusions []. Afterwards, a case-control study showed that NSAIDs, LDA, and H. pylori infection were three independent risk factors for the development of PUB. An additive effect was also found between H. pylori infection and NSAID use, whilst no interaction was shown between LDA and the infection []. Analogous results were also reported in a recent German study [] (Table 3). As for the relationship between H. pylori and non-aspirin antiplatelet agents or oral anticoagulants, there are limited data in the literature. A recent cohort study found that, in H. pylori positive patients taking non-aspirin antiplatelet agents, there was an increased risk of haemorrhagic peptic disease (OR: 4.37; 95% CI: 1.28–14.99), while the risk was not increased in infected patients taking oral anticoagulants (OR: 0.95; 95% CI: 0.58-1.55) []. The uncertainty of evidence is mirrored in the recommendations of two recent international guidelines. In the European Maastricht V/Florence Consensus, it is stated that testing for H. pylori should be performed in patients with a history of peptic disease taking NSAIDs or aspirin []. But regarding LDA, as evidence was considered controversial, no specific recommendations were provided. Similarly, no specific advices were given for non-aspirin antiplatelet agents or oral anticoagulants []. The American guideline recommend to search for H. pylori infection in patients starting NSAID or LDA therapy []. However, it was highlighted that the basis for the LDA recommendation was weak in the absence of a prospective randomized study addressing H. pylori eradication in North American patients at increased risk for adverse cardiovascular outcomes. Furthermore, even in this consensus, there were no specific advices regarding non-aspirin antiplatelet agents or oral anticoagulants []. After these guidelines have been issued, two meta-analyses were published [, ]. The former showed that H. pylori infection increased the risk of gastroduodenal ulcers during LDA therapy by approximately 70% (OR: 1.65; 95% CI: 1.29-2.08) []. The latter found that the odds of UGIB in patients taking LDA was almost two-and-one-half times as frequent in H. pylori-positive (OR: 2.32; 95% CI: 1.25 - 4.33) []. Nevertheless, the Authors estimated that, with such an OR, the NNT to prevent one bleeding event annually was to be between 100 and more than 1,000 []. This raises doubts about the possible cost effectiveness of a test and treatment strategy for all subjects receiving LDA, also considering the increasing primary resistance to antimicrobial agents []. In addition, eradication was never adequately evaluated as a first-line strategy to prevent PUD in patients taking LDA with an average risk of GI bleeding. One study showed that incidence rates of ulcer bleeding were not significantly different between H. pylori eradicated patients with previous PUB and the average-risk cohort (i.e. LDA-naïve patients without a history of ulcer and with unknown H. pylori status) []. More clarifying data could be obtained from the Helicobacter Eradication Aspirin Trial which is ongoing []. A proposed strategy is to classify patients taking LDA into two groups, low and high risk of bleeding, eradicating H. pylori only in the latter []. In addition to a previous peptic disease - which represents an indication to eradicate always the infection [, ] - other risk factors are older age and the concomitant use of NSAIDs, oral anticoagulants, non-aspirin antiplatelet agents, and corticosteroids []. Since nearly all of the available studies are case-control or cohort studies -more prone to bias - well-designed RCTs are demanded as well as pharmaco-economic evaluations assessing the benefits and risks of H. pylori eradication in LDA, non-aspirin antiplatelet agents and oral anticoagulants users.

Data from two RCTs, published over 20 years ago, demonstrated the superiority of PPIs over ranitidine and misoprostol in the prevention of gastroduodenal erosions and/or ulcers induced by NSAIDs [, ]. A recent Dutch study showed that therapy with PPIs is cost-effective in reducing GI bleeding in patients aged ≥ 60 years taking antiplatelet agents []. As for ranitidine, it was observed that its protective action of gastroduodenal mucosa decreased significantly within 4 weeks starting treatment, and that, at 6 months, it was significantly less effective than PPIs []. Likewise, this depends on the onset of tachyphylaxis, typical of H₂RA, but not of PPIs. Indeed, a pharmacodynamic study clearly showed that, already on the third day of therapy, the efficacy of ranitidine in reducing gastric acid secretion was almost halved []. This aspect cannot be disregarded when long-term treatment is necessary, such as antiplatelet therapy in patients at risk of PUD or UGIB []. The interest in the use of ranitidine has renewed in recent years following the aforementioned report of a potential liver drug interaction with PPIs, especially between omeprazole and clopidogrel []. Although ranitidine, like pantoprazole, does not interfere with the antiplatelet power of clopidogrel in vitro [], it significantly attenuates the antiplatelet activity of the LDA, probably reducing its intestinal absorption []. This effect could be relevant in the gastroprotection of patients on DAT or in monotherapy with LDA. Compared to PPIs, gastroprotection with misoprostol is also significantly less effective in prolonged treatment, is more frequently burdened by adverse events (diarrhoea and abdominal pain), and is less effective in reducing NSAID-induced dyspepsia []. In addition, it requires the administration of at least 1-2 tablets twice daily, with possible negative implications for patient's compliance and costs compared to the single-administration of PPIs. The commercialization in Western countries of a new class of drugs, the P-CABs, of which vonoprazan is first, capable of effectively reducing gastric acid secretion, is expected in the short term. Therefore, these drugs could be useful in preventing gastroduodenal mucosal lesions induced by antiplatelet agents. A randomized double-blind study showed that peptic ulcer recurrence rates were significantly lower with vonoprazan 10 mg (p=0.039) compared with lansoprazole 15 mg in patients with a history of PUD who required long-term LDA therapy for cardiovascular and cerebrovascular protection []. Furthermore, the absence of any effect of vonoprazan on the platelet-aggregating inhibitory activity of aspirin was shown in vitro, as well as the lack of pharmacodynamic interaction between the two drugs in vivo []. However, data on the potential drug interaction of vonoprazan with clopidogrel, other antiplatelet agents, and oral anticoagulants are not yet available. Several other molecules are being evaluated in experimental models, such as potential NSAID gastroprotectors, including allantoin and rebamipide. The latter was found to be as effective as misoprostol in preventing gastroduodenal lesions from aceclofenac, meloxicam and nabumetone [], but not from naproxen [], and better tolerated. However, there are still no data in the literature to validate its efficacy in preventing gastrointestinal damage from antiplatelet agents or anticoagulants.

There are several guidelines on gastroprotection in patients taking antiplatelet and/or oral anticoagulant treatment, prepared by American, European and Italian societies of Cardiology, Gastroenterology, Pharmacology and General Practitioners [, , , , , , , , ]. The joint US guidelines of the American College of Cardiology Foundation, the American College of Gastroenterology, and the American Heart Association recommend the use of PPIs in patients taking DAT and in those who have multiple risk factors for GI bleeding []. The ESC 2017 guidelines on DAT recommend gastroprotection with PPIs in all patients [], while the recent ESC guidelines on non-ST-segment elevation acute coronary syndrome [] recommend (class I, level of evidence A), for pharmacological long-term management, concomitant use of PPIs in patients receiving aspirin monotherapy, DAPT, DAT, TAT or OAC monotherapy who are at high risk of gastrointestinal bleeding in order to reduce the risk of gastric bleeds. In particular, they suggest as a strategy to reduce bleeding risk related to percutaneous coronary intervention the use of PPIs in patients on DAPT at higher-than-average risk of gastrointestinal bleeds (i.e. history of gastrointestinal ulcer/haemorrhage, anticoagulant therapy, chronic non-steroidal anti-inflammatory drugs/corticosteroid use, or two or more of: age ≥ 65 years; dyspepsia; gastro-oesophageal reflux disease; H. pylori infection; chronic alcohol use []. In Italy, a position paper issued by Italian Society of Pharmacology, the Italian Association of Hospital Gastroenterologists, and the Italian Federation of General Practitioners, recommend the use of standard doses of PPIs for those taking antiplatelet agents (individually or in association) only if at least one risk factor is present (age> 65 years, concomitant use of steroids or anticoagulants, previous history of peptic ulcer) []. In addition, it was suggested the use of pantoprazole or rabeprazole if clopidogrel was used, while there were no restrictions on the choice of PPIs with prasugrel and ticagrelor []. The uncertainty of the evidence on the eradication of H. pylori is reflected in the recommendations of two recent international Consensus. In the European guideline, it is stated that the use of ASA and NSAIDs increases the risk of peptic disease in the presence of the infection and, therefore, the search for H. pylori should be performed in patients with a history of peptic disease who take NSAIDs or ASA []. However, there were no specific recommendations on LDA since the evidence was considered controversial, as well as no recommendations were provided for non-aspirin antiplatelet agents and/or oral anticoagulants []. In the Consensus of the American College of Gastroenterology [], despite the lack of prospective RCTs, it was suggested searching for the infection in patients starting ASA therapy, even if it was emphasized that the basis for this recommendation was weak. Furthermore, also in this case, there were no specific recommendations regarding non-aspirin antiplatelet agents and/or oral anticoagulants []. Therapy with oral anticoagulants alone has not been shown to be harmful to the gastroduodenal mucosa, and therefore does not require therapy with PPIs if patients are not taking steroid and/or antiplatelet therapy. Indeed, a consensus of experts recommends the use of PPIs only in patients taking warfarin in combination with antiplatelet therapy and/or NSAIDs [], while other guidelines do not address this aspect . According to 2020 ESC guidelines on the management of AF [], apixaban or dabigatran 110 mg b.i.d. are not associated with an excess of gastrointestinal bleeding compared with warfarin. From a regulatory point of view, in Italy the prescription of PPIs paid by the National Health Service, and regulated by the Prescriptive Note 1 of the Italian Medicines Agency, is limited to the prevention of serious complications of the upper GI tract in antiplatelet therapy with LDA provided one of the following bleeding risk conditions exists: history of past digestive bleeding or peptic ulcer not healed with H. pylori eradication therapy; concomitant anticoagulant or steroid therapy; old age. However, according to the Note 1, the use of ranitidine and misoprostol - as an equal alternative to PPIs - is also suggested, while rabeprazole cannot be prescribed for this purpose in Italy. However, the FDA has not approved ranitidine to prevent GI injury from NSAIDs. In the real world, a recent Danish study including more than 46,000 patients treated with DAT after an acute MI showed an underutilisation of gastroprotection with PPIs, against an annual risk of gastrointestinal bleeding of about 1% in the general population, and 1.7% in patients at high risk of bleeding []. The lack of gastroprotection in a portion of patients on antiplatelet therapy, at risk of GI bleeding, was also documented in Italy []. In a retrospective study conducted in Spain on patients hospitalized for ACS, the totality of the patients received gastroprotective agents at the moment of discharge, most of them with PPIs []. Nevertheless, in one every 3 cases of the patients who were on clopidogrel, the recommendation of the FDA and of the EMA were not followed as omeprazole or esomeprazole were prescribed [].

Antiplatelet and anticoagulant agents are effective for primary and secondary prevention of cardiovascular events in subjects at thrombotic risk. Unfortunately, both can cause bleeding in the GI tract, sometimes with a fatal outcome [, , , , , , , , , , , , , , , , , , , , , ]. Therefore, gastroprotection is required in patients at increased risk of GI bleeding (Fig. 3).