Long-term mortality and early valve dysfunction according to anticoagulation use

Selected in JACC by D. Giacoppo

Results of the FRANCE-TAVI registry

References

Authors

Overtchouk P, Guedeney P, Rouanet S, Verhoye JP, Lefevre T, Van Belle E, Eltchaninoff H, Gilard M, Leprince P, Iung B, Barthelemy O, Le Breton H, Souteyrand G, Vicaut E, Montalescot G and Collet JP

Reference

J Am Coll Cardiol. 2019 Jan 8;73(1):13-21

Published

January 2019

Link

Read the abstract

Reviewer

Daniele Giacoppo

@DanieleGiacoppo

Interventional cardiologist / Cardiologist

Catania, Italy

Latest contributions

NEO-MINDSET: Early withdrawal of aspirin after percutaneous coronary intervention for acute coronary syndrome PCRonline @ ESC Congress 2025: the interventional cardiology perspective! Insights into emerging prognostic predictors for patients with coronary artery disease

My Comment

Why this study – the rationale/objective?

Silent valve leaflet thrombosis after transcatheter aortic valve implantation is an emerging cause of concerns. This phenomenon seems to be relatively common but potentially associated with accelerated bioprosthetic valve deterioration and thromboembolism. Oral anticoagulation may prevent early bioprosthetic valve dysfunction but whether this approach is more effective than antiplatelet therapy alone is still unknown. This analysis from the FRANCE TAVI registry sought to identify independent predictors of long-term mortality and bioprosthetic valve dysfunction and define the impact of antithrombotic regimens at discharge on these outcomes.

How was it executed – the methodology?

The FRANCE TAVI is a prospective nation-wide observational study including 12,804 patients undergoing transcatheter aortic valve implantation at 48 centres in France from January 2013 through December 2015.
A total of 11,469 patients who did not die during index hospital stay and had known anti-thrombotic treatment at discharge were analysed by stepwise multivariable Cox regression analysis with respect to the endpoint of all-cause death.
A total of 2,555 patients who did not die during index hospital stay and had at least 2 bioprosthetic valve function echocardiographic evaluations with follow-up assessment at least 3 months after the procedure were analysed by stepwise multivariable logistic regression with respect to the endpoint of bioprostethic aortic valve dysfunction. Bioprosthetic aortic valve dysfunction was defined according to the European consensus statement. Missing values were handled by multiple imputations and complete cases were used as sensitivity analysis.

What is the main result?

Three-year survival by Kaplan-Meier method was 69.9% (67.9%-71.9%). One third of patients was discharged on oral anticoagulation mainly as a consequence of atrial fibrillation (70.8%) and median time to follow-up echocardiography was 12 (11-15) months.
After multivariable adjustment, the strongest associations with death were male sex (HR 1.63, 95% CI 1.44-1.84, p<0.001), history of atrial fibrillation (HR 1.41, 95% CI 1.23-1.62, p<0.001), and moderate-to-severe renal dysfunction (HR 1.37, 95% CI 1.23-1.53, p<0.001). NYHA class III or IV, diabetes mellitus, non-femoral access, prosthesis size ≤23 mm, auto-expandable prosthesis, and EuroSCORE I (percentage unit increase) were among factors showing a significant association with mortality though less pronounced.
Conversely, previous coronary artery bypass grafting (HR 0.64, 95% CI 0.54-0.77, p<0.001) and non-coronary artery bypass grafting cardiac surgery were associated with significant risk reductions (HR 0.59, 95% CI 0.46-0.76, p<0.001). Interestingly, neither aspirin nor clopidogrel was independently associated with all-cause death, while oral anticoagulation led to a mild risk increase (HR 1.18, 95% CI 1.04-1.35, p=0.013). However, different findings were observed in the analysis exploring bioprosthesis hemodynamic deterioration. After multivariable adjustment, anticoagulation at discharge (OR 0.54, 95% CI 0.35-0.82, p=0.005) and nonfemoral approach (OR 0.53, 95% CI 0.28-1.02, p=0.049) were independently associated with lower rates of bioprosthetic valve dysfunction, while moderate-to-severe renal dysfunction (OR 1.46, 95% CI 1.03-2.08, p=0.034), previous transcatheter aortic valve implantation (OR 2.96, 95% CI 1.15-7.64, p=0.019), prosthesis size ≤23 mm (OR 3.43, 95% CI 2.41-4.89, p < 0.001), and body mass index (1 kg/m² increment, OR 1.05, 95% CI 1.02-1.09, p=0.002) yielded higher risks.

Critical reading and the relevance for clinical practice

There is a paucity of data from randomised trials and large-scale registries exploring the effects of oral anticoagulation as adjunctive therapy for the prevention of subclinical leaflet thrombosis and thromboembolism after transcatheter aortic valve implantation.

The results of this subanalysis from the FRANCE TAVI registry indicates that oral anticoagulation after transcatheter aortic valve implantation may produce a remarkable reduction in the risk of accelerated bioprosthetic valve dysfunction. Although the 46% relative risk reduction observed at multivariable analysis cannot provide definitive answers against bioprosthetic valve thrombosis and potential thromboembolism-related events, it confirms in large-scale cohort findings previously emerged in small studies. However, in the FRANCE TAVI registry oral anticoagulation was also associated with mild but statistically-significant increase in the risk of long-term all-cause death. Whether this result is a consequence of a higher incidence of life-threatening bleeding or due to collinearity with atrial fibrillation requires further analyses.

The availability of only 22.3% of patients with functional assessment at follow-up and the absence of information about antithrombotic medication adherence and duration support the need for data with higher completeness and deriving from a protocol-guided systematic monitoring of therapeutic regimens. In addition, the identification of valve dysfunction by using mean gradient absolute or relative variations as assessed by transthoracic echocardiography cannot be used to prove a direct causality between oral anticoagulation and reduced bioprosthetic valve thrombosis which would be possible for example by multi-detector computed tomography. Finally, a prospective trial with random assignment of the therapy in non-atrial fibrillation patients would significantly reduce the influence of confounders and limit the possible impact of the tachyarrhythmia on mean valve gradient measurement.