Defining Optimal Antithrombotic Therapy Post-TAVI

The Contribution of Atlantis

Felicita Andreotti; Massimo Massetti; Jurrien ten Berg


Eur Heart J. 2022;43(29):2798-2800. 

Graphical Abstract: Major outcomes of randomized controlled trials investigating direct oral anticoagulants in patients undergoing successful TAVI. 'Non-fatal stroke' refers to ischaemic stroke in GALILEO[5] and ENVISAGE-TAVI AF,[6] and to any stroke/transient ischaemic attack/systemic embolism in ATLANTIS.[13] 'Valve thrombosis' refers to RLM of >50% of ≥1 leaflet(s) (i.e. grade 3–4) in GALILEO,[8] to transprosthetic mean gradient ≥20 or ≥10 mmHg above previous measurements or HALT/RLM grade 3–4 in ATLANTIS,[13] and to thrombosis of haemodynamic relevance, symptomatic or completely reversible by high-intensity anticoagulation or to HALT/RLM or transprosthetic mean gradient ≥20 or ≥10 mmHg above previous measurements in ENVISAGE-TAVI AF.[6] Apix = apixaban; APT = antiplatelet therapy alone; bid = twice daily; Edox = edoxaban; HALT = hypo-attenuated leaflet thickening; od = once daily; Riva = rivaroxaban; RLM = reduced leaflet motion; VKA = vitamin K antagonist.

What is the best antithrombotic regimen for patients undergoing transcatheter aortic valve implantation (TAVI)? Recent randomized controlled trials (RCTs) have shown reduced bleeding risk with aspirin, compared to aspirin plus clopidogrel, for TAVI patients with no indication for oral anticoagulation (OAC), and with OAC alone, compared to OAC plus clopidogrel, for TAVI patients requiring OAC mostly because of atrial fibrillation (AF).[1,2] Not adding clopidogrel to aspirin or OAC did not appear to enhance thrombo-embolic risk. Although the trials were underpowered for endpoints such as stroke and myocardial infarction (MI), lifelong monotherapy post-TAVI with an antiplatelet or an anticoagulant agent (if indicated) has a class I recommendation in current guidelines.[3]

Recently, bioprosthetic valve thrombosis has gained interest. Valve thrombosis is subclinical in almost all cases and is diagnosed by computed tomography (CT) as hypo-attenuated thickening (HALT) or reduced leaflet motion (RLM) of one or more leaflets or by echocardiography as increased transprosthetic gradients.[4] Clinically manifest, symptomatic valve thrombosis, with symptoms of heart failure or thromboembolism, does occur but is rare (up to 1% in recent trials).[5,6] Valve thrombosis occurs more frequently in patients on antiplatelet than on OAC regimens.[5–8] It is typically reversed by 2–4 months of high-intensity anticoagulation.[3,4,9] Valve thrombosis may also resolve or develop over time while patients remain on antiplatelet therapy.7 Several, but not all reports, have associated subclinical valve thrombosis with increased risks of stroke and transient ischaemic attack (TIA),[7,9–12] prompting the US Food and Drug Administration to mandate CT substudies for two RCTs comparing TAVI to surgery.[4,7] Thus, endeavours to minimize adverse events, including subclinical valve thrombosis, are testing various antithrombotic regimens in TAVI patients with or without indication for OAC.

A first attempt to improve outcomes by adding OAC to routine antiplatelet therapy in TAVI patients without a clear indication for OAC failed. In the GALILEO RCT, 1644 such patients were randomized to either medium-dose rivaroxaban (10 mg daily) plus aspirin for 3 months, followed by rivaroxaban alone, or aspirin plus clopidogrel for 3 months, followed by aspirin alone.[5] The trial was halted after a median of 17 months for higher all-cause mortality in the rivaroxaban arm [hazard ratio (HR) 1.69, 95% confidence interval (CI) 1.13–2.53]. Life-threatening, disabling, or major bleeding and the composite of death and thromboembolic events were also higher in the rivaroxaban arm compared to the antiplatelet arm (HR 1.50, 95% CI 0.95–2.37 and HR 1.35, 95% CI 1.01–1.81, respectively). The rivaroxaban regimen compared to the antiplatelet one, on the other hand, reduced the prevalence of subclinical valve thrombosis assessed by CT at 90 days, from 10.9% to 2.1% (P = 0.01).[8] Symptomatic valve thrombosis occurred in three patients in the rivaroxaban group vs. seven in the antiplatelet group.[5]

The ENVISAGE-TAVI AF trial, instead, randomized 1426 AF patients after successful TAVI to edoxaban (60 mg daily) or vitamin K antagonist (VKA).[6] After a median of 18 months, the net primary endpoint of all-cause death, MI, ischaemic stroke, systemic thromboembolism, clinical valve thrombosis, and major bleeding was non-inferior in the edoxaban vs. the VKA group (HR 1.05, 95% CI 0.85–1.31), whereas major bleeds—but not fatal or intracranial bleeding alone—were significantly higher with edoxaban (HR 1.40, 95% CI 1.03–1.91) owing to more gastrointestinal bleeds (5.4 vs. 2.7 per 100 patient-years). No clinically manifest valve thrombosis was recorded; detection of subclinical valve thrombosis was not protocol mandated.

The ATLANTIS trial published in this issue of the European Heart Journal[13] adds information to the management of both types of patients (with or without clear indication for OAC). In 1500 subjects undergoing successful TAVI, the trial tested the superiority (followed by non-inferiority) for a net composite outcome of full-dose apixaban (5 mg twice daily) vs. standard of care (VKA or antiplatelet therapy, depending on an indication or not for OAC). Sample size was powered for an optimistic 33% relative risk reduction, with the non-inferiority margin set at an upper boundary of the 95% CI for the HR of 1.2. Treatment randomization occurred on average 3.6 days post-TAVI (range 0–33) and was stratified by presence (n = 451, stratum 1) or absence (n = 1049, stratum 2) of baseline indication for OAC, ensuring balanced characteristics across treatments within each stratum. Valve thrombosis at 3–6 months was assessed by echocardiography or CT in 89% and 59% of patients respectively. Overall, apixaban failed to show superiority at 1 year for the net primary endpoint of death, MI, stroke, TIA, systemic embolism, intracardiac or valve thrombosis, deep vein thrombosis, pulmonary embolism, and life-threatening, disabling, or major bleeding (HR 0.92, 95% CI 0.73–1.16). There was no evidence of interaction between treatment effects and strata (interaction P = 0.57). The overall HR for the primary safety endpoint (life-threatening, disabling, or major bleeding) was 1.02 (95% CI 0.72–2.44). Notably, in stratum 2 (no indication for OAC), although apixaban clearly reduced valve thrombosis compared to antiplatelet therapy (HR 0.19, 95 %CI 0.08–0.46), total and particularly non-cardiovascular mortality (HR 2.99, 95% CI 1.07–8.36) were higher in the apixaban arm.

What are the clinical implications of ATLANTIS? Although each stratum in ATLANTIS may appear underpowered for thromboembolic endpoints, the results of stratum 1 and 2 independently support those of ENVISAGE-TAVI AF and GALILEO, respectively, indicating true rather than chance findings. In stratum 1 (clear indication for OAC), there is no marked difference between full-dose apixaban and a VKA; the findings resemble those of ENVISAGE-TAVI AF, with the added benefit of no excess gastrointestinal bleeding. Thus, in TAVI patients with an indication for OAC, apixaban (or edoxaban with caution) can be used, and are favoured over VKAs by guidelines on the prevention of AF-related stroke.[14]

In contrast, the role of direct OACs (DOACs) in TAVI patients with no clear indication for OAC appears different. Both ATLANTIS stratum 2 and GALILEO data indicate harm with the investigated DOAC regimens compared to antiplatelet therapy alone. Therefore, we should not treat patients undergoing successful TAVI who have no evidence-based indication for OAC with full-dose apixaban (or a medium-dose rivaroxaban regimen). On the other hand, both DOAC regimens showed a markedly reduced prevalence of subclinical valve thrombosis compared to antiplatelet therapy alone, which might be associated with reduced rates of stroke and TIA. The latter association, however, did not emerge clearly, neither in ATLANTIS nor across the other DOAC-TAVI trials. While several reasons may contribute to the lack of association between valve thrombosis and stroke (e.g. only a subset of valve thromboses—the most severe and persistent—are associated with clinical thromboembolism;[7] most strokes occur in the very first days while the DOAC trials included only patients undergoing successful, uncomplicated TAVI; all DOAC-TAVI trials were underpowered for individual stroke outcomes), the clinical implications of subclinical valve thrombosis detected at 3–6 months post-TAVI remain uncertain.

The Graphical Abstract summarizes the major outcomes of the three recent RCTs testing DOACs in patients undergoing successful TAVI. Accepting the limitations of some indirect comparisons and of underpowered individual outcomes, several considerations can be made. (i) In TAVI patients with a clear indication for OAC, the composite annual rate of death, major bleeding, and non-fatal stroke appears fairly similar in the investigated DOAC arms compared with the VKA arms (excepting the increased gastrointestinal bleeding risk with edoxaban). (ii) The prevalence of largely subclinical valve thrombosis at 3–6 months is definitely lower with full-dose OAC (ATLANTIS and ENVISAGE-TAVI AF) or with a medium-dose rivaroxaban regimen (GALILEO) than with antiplatelet-only regimens. (iii) Among TAVI patients without a clear indication for OAC, the investigated DOAC regimens compared to the antiplatelet regimens carry higher mortality (>2% absolute increase per year), despite apparently comparable stroke rates and lower valve thrombosis prevalence at 3–6 months. This suggests that, on average, in the absence of an evidence-based OAC indication, the risks of the above OAC regimens exceed any benefit derived from reduced (largely subclinical) valve thromboses.

Several issues remain unanswered. Can the definitions, tools, and time points to assess valve thrombosis be harmonized and improved? The recent VARC-3 efforts certainly go in this direction.[4] Further, among patients with subclinical valve thrombosis and without a clear baseline indication for OAC, it is unclear whether OAC may be better than continuing aspirin, a question addressable through a randomized comparison of OAC vs. antiplatelet therapy.

In conclusion, ATLANTIS, far from representing a mythological lost land, contributes important evidence to the definition of optimal antithrombotic therapy post-TAVI. Based on ATLANTIS and ENVISAGE-TAVI AF, when OAC is clearly indicated,[3] this can now include a DOAC instead of a VKA. Equally, when anticoagulation is not indicated, ATLANTIS corroborates current recommendations against OAC.[3]