Atrial fibrillation (AF) is the most prevalent arrhythmia in adult patients worldwide and its prevalence is expected to rise due to increased longevity of the general population [1, 2, 3, 4, 5]. The incidence of AF in patients experiencing an acute coronary syndrome (ACS) stands at 2–23% and AF itself may be associated with a higher risk of myocardial infarction (MI) [6, 7]. Moreover, it is estimated that up to 15% of AF patients undergo percutaneous coronary intervention (PCI) for coronary artery disease (CAD) during their life [8].

Plaque rupture and thrombus formation induced by platelet aggregation and coagulation cascade activation are the main drivers of ACS and persistent activation of coagulation may last for several months, leading to increased risk of unfavourable outcome [9]. For this reason, oral anticoagulants (OACs) may exert a protective effect in the setting of CAD. OACs are more protective against stroke compared to antiplatelet agents in AF patients [10], whereas patients experiencing ACS or undergoing PCI are recommended to receive dual antiplatelet therapy (DAPT) to reduce the risk of ischemic events, namely cardiovascular death, recurrent MI and stent thrombosis [11, 12, 13]. Even though triple antithrombotic therapy (TAT) including OAC, acetylsalicylic acid and clopidogrel is recommended for AF patients experiencing ACS or undergoing PCI, the benefit of antithrombotic therapy regarding cardiovascular ischemic events must be carefully balanced against an increased risk of treatment-related major and minor bleedings [14, 15]. Literary data and European Society of Cardiology (ESC) guidelines suggest that AF patients at increased ischemic risk with a recent ACS or undergoing PCI may benefit from a short course of at least one week of TAT followed by a 6 to 12 months period of dual antithrombotic therapy (DAT) with OAC and an antiplatelet agent (preferably clopidogrel) according to the acute or chronic coronary setting [1, 16, 17, 18]. OAC monotherapy is to be continued afterwards unless there were recurrent ischemic events in this timeframe. Likewise, OAC monotherapy is also recommended after one year in patients with AF and chronic CAD with no PCI in the previous year [19].

Multiple trials and meta-analyses have flourished in recent years investigating and comparing different OAC molecules in the context of AF and CAD. We aimed to review the current available literature addressing the issue of OACs in AF patients with CAD and to provide comprehensive data on risks and benefits of individual OAC molecules in this complex scenario.

The efficacy and safety of OACs in the setting of CAD has long been investigated. Most studies in the late 20th century compared OAC therapy with or without aspirin against aspirin therapy alone in CAD patients without AF, to assess whether OACs could provide benefit over aspirin on its own in this setting. Earlier trials in the 1980s did not show different rates of reinfarction and mortality between patients given aspirin and those given vitamin K antagonists (VKA) [20, 21]; nevertheless, these studies were relatively small, the aspirin doses were high, and the intensity of anticoagulation was not adequately controlled, thus lessening the reliability of these trials’ results. Likewise, no benefit regarding cardiovascular events was observed when combination therapy with warfarin and aspirin was compared with aspirin alone in the later CHAMP study [22], CARS trial [23] and OASIS-2 [24], whereas bleeding rates were significantly increased in the combination therapy in the CHAMP study; all these studies, however, required low international normalized ratio (INR) targets and the positive effect of warfarin on hard ischemic endpoints might have been shadowed by the low intensity of the anticoagulation treatment. Conversely, later trials imposing a target INR higher than 2.0 consistently proved a significant benefit of aspirin plus OAC compared to aspirin alone regarding hard ischemic endpoints, as was the case for the Antithrombotic Therapy in Acute Coronary Syndromes (ATACS) trial in 1994 [25], the ASPECT-2 trial in 1999 [26] and the WARIS II trial in 2002 [27].

The WOEST study was an open-label, randomized controlled trial (RCT) conducted between 2008 and 2011 to assess whether patients on OACs undergoing PCI would benefit from receiving clopidogrel alone compared to clopidogrel plus aspirin [28]. 573 patients were enrolled; at 1-year follow-up bleeding events occurred in 19.4% of patients receiving DAT compared to 44.4% of those receiving TAT (hazard ratio [HR] 0.36, 95% confidence interval [CI] 0.26 to 0.50, p = 0.011). Moreover, the composite of death, MI, stroke, target-vessel revascularisation and stent thrombosis occurred significantly less frequently in the DAT group compared to the TAT group (11.1% vs 17.6%, respectively; HR 0.60, 95% CI 0.38 to 0.94, p = 0.025), even after baseline characteristics adjustments (adjusted HR [adj-HR] 0.56, 95% CI 0.35 to 0.91) [28]. The WOEST trial showed that clopidogrel administered to patients on OAC requiring PCI is associated with significantly less frequent bleeding events at 1 year than is the combined use of clopidogrel and acetylsalicylic acid. Specifically, gastrointestinal bleeding episodes occurred substantially less frequently in the DAT than in the TAT group, likely due to the local erosive effect of acetylsalicylic acid [29]; it should be pointed out, however, that proton pump inhibitors use was not mandatory in this trial, despite being recommended, and increased administration of these drugs might have lessened the number of gastrointestinal bleeding events. Notably, the WOEST trial also reported that the rate of thromboembolic events was not different between patients who received and did not receive acetylsalicylic acid [28]. The authors suggested that inhibition of thrombin with OACs and P2Y12 receptor inhibition with clopidogrel would reduce the impact of cyclo-oxygenase-1 inhibition by acetylsalicylic acid, as also stated by prior studies [26, 27].

Following the results of the WOEST trial, two registries comparing the safety and efficacy of TAT with warfarin, acetylsalicylic acid and clopidogrel vs different DAT regimens were published [15, 30]. The study by Lamberts et al. [15] included 12,165 patients with AF, hospitalized for MI or undergoing PCI and reported that, relative to TAT, no significant difference regarding the risk of coronary events was found for OAC plus clopidogrel, OAC plus acetylsalicylic acid or acetylsalicylic acid plus clopidogrel at 1-year follow-up, whereas the association of clopidogrel and acetylsalicylic acid was associated with a higher risk of ischemic stroke; moreover, OAC plus acetylsalicylic acid and acetylsalicylic acid plus clopidogrel significantly increased the risk of death, while the association of acetylsalicylic acid and either OAC or clopidogrel significantly lowered bleeding risk compared to TAT. The AFCAS registry enrolled 914 AF patients undergoing coronary stent implantation and showed that no significant differences in the rate of MACE and bleeding events among 3 different antithrombotic regimens (TAT with warfarin, acetylsalicylic acid and clopidogrel, DAT with warfarin plus clopidogrel and DAPT), even after propensity score adjustment, at 1-year follow-up [30]. Detailed data regarding the aforementioned studies on VKA molecules in the setting of CAD are reported in Table 1 (Ref. [15, 20, 21, 22, 23, 24, 25, 26, 27, 28, 30]).

Since 2009 non-VKA oral anticoagulants (NOAC), also known as direct oral anticoagulants, have emerged as an alternative to coumarin molecules and progressively took the place of warfarin as first choice regimen in several settings [31, 32, 33, 34, 35]. While dabigatran inhibits factor II of the coagulation cascade (thrombin), rivaroxaban, apixaban and edoxaban inhibit activated factor X; these molecules are characterized by rapid action onset and termination, absence of interference with dietary vitamin K intake, and fewer drug interactions than warfarin. In particular, four randomized trials, namely the RE-LY [31], ROCKET AF [32], ARISTOTLE [33], and ENGAGE AF-TIMI 48 [34] trials, first tested the safety and efficacy of these drugs in the setting of cerebrovascular and systemic thrombo-embolic event prevention in non-valvular AF. A landmark metanalysis by Ruff et al. [35] demonstrated how the “new oral anticoagulants”, as they were initially referred to, significantly reduced stroke, intracranial hemorrhage and mortality, with no difference in major bleedings but increased gastrointestinal bleedings compared to warfarin. Moreover, these results were consistent across different populations, regardless of age, sex, presence of diabetes, creatinine clearance, previous stroke or transient ischemic attack occurrence and CHADS${}_2$-score.

Unfortunately, when CAD presents in patients with AF, observational literary data as well as the 2018 ESC consensus paper on the management of antithrombotic therapy in AF patients presenting with ACS or requiring PCI report that CAD patients less likely receive optimal antithrombotic treatment if they have a history of AF, thus posing them at higher risk of death than patients without AF [14].

Four RCTs compared DAT with P2Y12 inhibitor plus each of the four NOAC molecules versus TAT consisting of aspirin, a P2Y12 inhibitor (P2Y12i) and VKA in the setting of AF patients undergoing PCI with the primary aim to analyze NOAC safety [31, 32, 33, 34].

The first RCT on this issue was published in 2016; it was the PIONEER-AF PCI, an open-label, randomized, multicenter study which randomly allocated 2124 patients to DAT with low-dose rivaroxaban (15 mg/die) plus a P2Y12 inhibitor for 12 months (Group 1), TAT with “very-low-dose” rivaroxaban (2.5 mg x2/die) plus DAPT for 1, 6, or 12 months (Group 2), or TAT with VKA plus DAPT for 1, 6, or 12 months (Group 3) in patients with nonvalvular AF undergoing PCI with stent deployment [36]. The study population consisted of about 30% of patients with ST-elevation MI (STEMI) and non-ST elevation MI (NSTEMI) and 20% with unstable angina; 93% of patients were on clopidogrel. Both rivaroxaban-based regimens were associated with lower rates of significant bleeding compared to TAT (16.8%, 18.0% and 26.7%, respectively in Group 1, 2 and 3; p 0.001). Moreover, the three groups did not differ significantly in efficacy, showing similar rates of cardiovascular death, MI, or stroke, even though the study was underpowered for efficacy. In addition, history of stroke was an exclusion criterion from the analysis, thus potentially leading to selection bias.

In 2017 the results from the open-label, multicenter, RE-DUAL PCI trial were published [37, 38]. Cannon et al. [37, 38] randomized 2725 patients with AF undergoing percutaneous coronary artery intervention to receive DAT with dabigatran (150 mg or 110 mg x2/die) and a P2Y12 inhibitor or TAT with warfarin, a P2Y12 inhibitor and aspirin for 1 or 3 months, based on the type of stent delivered. The primary endpoint was again a safety outcome, defined as major or clinically-relevant non-major bleeding according to the definition by the International Society on Thrombosis and Hemostasis (ISTH). The primary endpoint occurred in 15.4% of patients in the 110-mg DAT arm compared to 26.9% in the TAT arm (HR 0.52, 95% CI 0.42–0.63; p 0.001 for both non-inferiority and superiority), and 20.2% in the 150-mg DAT arm, compared with 25.7% in the TAT arm (HR 0.72, 95% CI 0.58–0.88; p 0.001 for non-inferiority). The intention-to-treat analysis provided consistent results with those of the on-treatment analysis and across major subgroups. The RE-DUAL PCI trial also tested for the non-inferiority of DAT with dabigatran compared to TAT with warfarin with respect to the composite efficacy endpoint of thromboembolic events (MI, stroke or systemic embolism), death, or unplanned revascularization. The incidence of this efficacy endpoint was 13.7% in the two DAT groups combined as compared with 13.4% in the TAT group (HR 1.04, 95% CI 0.84–1.29; p = 0.005 for non-inferiority). Notably, if analyzed separately, the 110 mg dabigatran regimen showed a non-significant excess in the number of ischemic events as compared with TAT. Overall, almost half of the trial population presented with an ACS and the vast majority (88%) of patients received clopidogrel as P2Y12i; no patient received prasugrel in RE-DUAL PCI.

More recently, the AUGUSTUS trial, published in 2019, was a prospective, multicenter, randomized clinical trial with a two-by-two factorial design, enrolling 4614 patients and concomitantly testing two hypotheses, namely (1) that apixaban 5 mg bid would have been at least noninferior to a VKA (open-label) and (2) that isolated P2Y12i would have been superior to DAPT with a P2Y12i and acetylsalicylic acid regarding clinically relevant bleedings in patients with AF and a recent ACS or PCI and planned concomitant antiplatelet therapy [39, 40]. At 6-month follow-up, the primary outcome occurred significantly less frequently in patients on apixaban compared to those receiving VKA (10.5% vs 14.7% respectively, HR 0.69, 95% CI 0.58 to 0.81, p 0.001). In the double-blind comparison of acetylsalicylic acid with placebo, acetylsalicylic acid led to a higher risk of bleeding compared to placebo (16.1% vs 9.0%; HR 1.89, 95% CI 1.59 to 2.24; p 0.001). From an efficacy point of view, at 6 months the apixaban group presented a 6.7% rate of the composite outcome of death or ischemic events as compared to a 7.1% rate in the VKA group and 6.5% of patients assigned to receive aspirin presented an event of as compared with a 7.3% of those assigned to placebo, with no statistically significant difference in either comparison. Despite not reaching statistical significance, ischemic events were numerically higher in the placebo group and, not surprisingly, the incidence of stent thrombosis was nearly twice as high in the placebo group as compared to the aspirin group. Moreover, the mean time-to-randomization in the AUGUSTUS trial was 7 days from the index ACS or PCI and patients were receiving aspirin in the interim, thus limiting assumptions on this period at high risk for coronary ischemic events. Interestingly, the AUGUSTUS trial was the first study to include medically managed ACS; this might be of great interest as nowadays this population is often represented by older and more fragile patients with very high bleeding risk.

Lastly, the ENTRUST-AF PCI trial was a randomized, multicenter, non-inferiority, open-label study published in 2019, randomly allocating 1506 patients with AF and a recent PCI to a DAT regimen with edoxaban 60 mg od (reduced to 30 mg od if creatinine clearance 15–50 mL/min or body weight $\le$60 kg or concomitant use of P-glycoprotein inhibitors) plus a P2Y12i or TAT with a VKA in combination with a P2Y12i (for 12 months) and aspirin (for a minimum of 1 month and up to 12 months) [41, 42]. At 12 months, major or clinically relevant non-major bleeding events occurred in 17% of patients in the DAT group compared to 20% in the VKA group (HR 0.83, 95% CI 0.65 to 1.05, p = 0.001 for non-inferiority), while the annualized event rate of the main efficacy outcome (cardiovascular death, stroke, systemic embolic events, MI and stent thrombosis) was similar between the two study arms (7% vs 6%, respectively; HR 1.06, 95% CI 0.71 to 1.69). The main characteristics of these trials are detailed in Table 2 (Ref. [1, 37, 38, 40, 42]).

Noteworthily, patients at high risk for ischemic events (e.g., previous stroke, complex PCI) were under-represented in these trials; moreover, clopidogrel was largely the most common P2Y12i used ($>$90%) both for DAT and TAT approaches and efficacy and safety of ticagrelor or prasugrel remain uncertain in these populations. Furthermore, none of these trials was specifically focused on urgent PCI and ACS patients’ prevalence ranged from 28% to 61%; this may further cloud judgment on how to manage extreme-risk patients. Finally, TAT was administered until randomization, and time to randomization varied from 72 h in the PIONEER-AF PCI up to 14 days in the AUGUSTUS trial, making assumptions on DAT effect in the hyper-acute phase after PCI unreliable.

The antithrombotic therapy indicated for AF patients after ACS or PCI comes at the cost of an increased bleeding risk. Hansen et al. [43] performed a cohort study using nationwide registries in Denmark to identify survivors of first-time hospitalization for AF between 1997 and 2006. After analysing data on 118,606 patients, the Authors reported the highest incidence rate of bleeding events for dual antithrombotic therapy with clopidogrel and warfarin (13.9% per patient-year) and TAT with warfarin, acetylsalicylic acid and clopidogrel (15.7% per patient-year), with a HR for the combined endpoint of non-fatal and fatal bleedings compared to warfarin monotherapy of 3.08 (95% CI 2.32 to 3.91) and 3.70 (95% CI 2.89 to 4.76) respectively. Toyoda et al. [44] later conducted the prospective, multicentre, observational BAT study in Japan to investigate the incidence and severity of bleedings in 4009 patients with stroke and cardiovascular diseases taking oral antithrombotic drugs. At a median 19-months follow-up, in 2008 the authors reported an annual incidence of Major bleedings of 1.21% in patients receiving a single antiplatelet agent, 2.00% in those on DAPT, 2.06% patients on warfarin alone and 3.56% in those taking warfarin plus an antiplatelet agent (p 0.001). Moreover, after adjusting for baseline characteristics, the BAT study showed that adding an antiplatelet agent to warfarin independently increased the risk of life-threatening or major bleeding events (relative risk: 1.76; 95% CI 1.05 to 2.95). Likewise, in 2009 Sørensen et al. [45] reported an increasing risk of hospital admissions for bleeding with the number of antithrombotic agents prescribed, with a four-fold increase of hospital readmission rate for patients taking TAT compared to aspirin alone during a mean follow-up of 476.5 days.

Uchida et al. [46] retrospectively explored the safety and effectiveness of DAPT and warfarin in 575 consecutive patients receiving drug-eluting stents; within a median 459-days follow-up, they found a 2.7% incidence of major bleeding events in patients receiving DAPT compared to 18.0% in the TAT group (p 0.001), thus highlighting that warfarin use is associated with an increased risk of major bleedings. Moreover, multivariate analysis also showed that renal impairment was an independent predictor of major bleeding in the TAT group [46]; this is not unexpected, as chronic kidney disease (CKD) is associated with prolonged bleeding time and platelet dysfunction which together lead to increased bleeding risk and ischemic events [47, 48].

According to the Academic Research Consortium for High Bleeding Risk, the long-term use of OAC by itself constitutes a major criterion defining a high bleeding risk after PCI [12, 49]; nevertheless, several bleeding risk factors must also be considered when deciding upon the optimal antithrombotic regimen for AF patients undergoing PCI, hereby including severe CKD, haemoglobin 11 g/dL, prior spontaneous bleeding requiring hospitalization and/or transfusion, platelet count 100.000/mcL, chronic bleeding susceptibility, liver cirrhosis with portal hypertension, active malignancy within the past year, previous spontaneous intracranial haemorrhage at any time (or within the past year if traumatic), the presence of a brain arteriovenous malformation, at least moderate ischaemic stroke within the past 6 months, recent major surgery or major trauma within one month prior to PCI and non-deferrable major surgery on DAPT [49]. The HAS-BLED score could also be calculated, as it may help correct modifiable bleeding risk factors and recognize patients at high bleeding risk (HAS-BLED score $\ge$3) who require more frequent clinical evaluation and follow-up; moreover, the HAS-BLED score may guide the clinical decision on the optimal NOAC dosage as well as the preferred TAT and DAT duration following PCI in each individual patient, as recommended by the latest ESC guidelines for AF [1, 50]. Nevertheless, an unmatched case-control study by Paciaroni et al. [51] described how both the HAS-BLED and CHA${}_2$DS${}_2$-VASc scores performed poorly in predicting intracranial haemorrhages in patients receiving NOACs for AF, while several other factors such as age, the presence of an active malignancy, high risk of fall, hyperlipidemia, low clearance of creatinine, were associated with an increased risk of intracranial haemorrhages. On the other hand, in a similar population atrial enlargement, hyperlipemia and high CHA${}_2$DS${}_2$-VASc score were associated with higher occurrence of cerebrovascular events [52].

Patients with AF undergoing PCI already represent a “special population”, despite being quite a common condition. Therefore, very few dedicated studies exist on subgroups commonly presenting as challenging scenarios and recommendations result from observational data or are inferred form subgroup analyses of major trials. The following paragraphs aim to only give a glimpse of the present state of the art and gaps in evidence.

The presence of concomitant AF and CAD warrants scrupulous antithrombotic therapy selection. The efficacy of NOAC molecules in preventing thromboembolic complications with relatively low bleeding risk have rapidly placed them as the mainstay of antithrombotic treatment for non-valvular AF, but their clinical use in patients with both AF and CAD is not yet thoroughly established. Based on the favourable literary data on NOAC-based combined antithrombotic therapy regarding thromboembolic and haemorrhagic risks, we believe that the Cardiologist prescribing TAT should weigh its treatment choice considering the individual AF burden (paroxysmal vs persistent vs permanent) whilst aiming at maintaining normal sinus rhythm for the longest time possible; transcatheter ablation of AF might be a valuable strategy with this regard. Conversely, left atrial appendage closure could be advised to those individuals at extreme haemorrhagic risk in whom the use of OAC therapy may be discouraged after risk/benefit trade-off evaluation. Furthermore, thorough knowledge of each patient’s coronary anatomy (single-vessel vs multivessel disease, presence of prognostic lesions, complex disease requiring the implantation of multiple stents) and type of stents implanted is important to optimize long-term antithrombotic therapy. In conclusion, as detailed in our review, attentive assessment of bleeding and ischemic risks is paramount in this scenario and the optimal antithrombotic treatment must be tailored on the clinical characteristics of each individual patient.

ET conceived the study. PPB and FA performed the research and wrote the first draft of the manuscript. All authors critically revised the manuscript and read and approved its final version.

Not applicable.

We would like to express our gratitude to all those who helped us during the writing of this manuscript.

This research received no external funding.

The authors declare no conflict of interest.