Extensive research has established the efficacy and safety of direct oral anticoagulants (DOAC) in the prevention of stroke and systemic embolism in patients with atrial fibrillation (AF). DOAC therapy is currently recommended over treatment with vitamin K antagonists (VKA), especially for patients initiating anticoagulation.1–7 DOAC therapy has been approved for use in patients with nonvalvular atrial fibrillation (NVAF), a broad category based on the exclusion criteria of several randomized clinical trials (Table 1). However, this terminology has led to therapeutic confusion in clinical practice.8,9 The most recent clinical practice guidelines agree that the current definition of valvular AF relates to AF associated with moderate to severe mitral stenosis (usually rheumatic) or mechanical prosthetic valves.6,7,10

Heart valve disease and AF often co-occur and can be independent causes of morbidity and mortality. DOAC clinical trials have included a high proportion of patients with significant valve disease. This subgroup has an unfavorable clinical profile, with outcomes similar to VKA-treated patients.11–15 There is currently little evidence available from real-world experience in patients with heart valve disease and no contraindication for DOAC therapy. The goal of the present study was therefore to analyze the presence of heart valve disease in AF patients initiating DOAC therapy in a real-world setting and to assess its influence on the appearance of events.

In total, 2297 patients were included, of whom 915 (39.8%) received rivaroxaban, 419 (18.2%) dabigatran, 896 (39.0%) apixaban, and 67 (2.9%) edoxaban. Of the total study population, 499 patients (21.7%) had significant heart valve disease. Patients with valve disease tended to be older and have more comorbidity, and thus scored higher on thromboembolic and bleeding risk scales (CHA2DS2-VASc, 4.4 ±1.5 vs 4.0 ±1.7; P <.001; HAS-BLED, 2.6±0.9 vs 2.4±1.0; P <.001). The presence of heart valve disease showed no association with the type of DOAC prescribed (P>.05) (Table 2).

Clinical characteristics were analyzed according to the predominant valve disease (Table 3). The most frequent valve disease was mitral regurgitation (315 patients; 13.7%), followed by aortic stenosis (100 patients; 4.4%) and aortic regurgitation (84 patients; 3.7%). Population characteristics are stratified according to valve disease type in Table 3.

Clinical events recorded during follow-up (606 [474-731] days) are classified according to the presence and type of valve disease in Figure 1 and Figure 2. The presence of valve disease was associated with a higher total event rate: composite event, 17.99 vs 8.56 per 100 person-years (P>.001); death, 11.90 vs 5.27 per 100 person-years (P>.001); and major bleeding, 6.16 vs 2.52 per 100 person-years (P>.001). However, valve disease showed no association with thromboembolic events (ischemic stroke/TIA/ systemic embolism: 1.81 vs 1.83 per100 person-years; P>.05). Similarly, all types of heart valve disease analyzed were associated with higher rates for the composite event, death, and major bleeding, although for aortic regurgitation the association did not reach statistical significance. None of the embolic events analyzed showed an association with higher risk (Figure 2).

Figure 1.

Influence of co-occurring significant heart valve disease on clinical events in NVAF patients. TIA, transient ischemic attack.

(0.12MB).

Figure 2.

Influence of heart valve lesion type on clinical events in NVAF patients. TIA, transient ischemic attack.

(0.21MB).

The univariate Cox proportional risk analysis for the prediction of each event is summarized in Table 1 of the supplementary data. Multivariate adjustment revealed significant valvular heart disease as an independent predictor of the composite event (HR=1.54; 95% confidence interval [95%CI], 1.22-1.94; P <.001), death (HR=1.44; 95%CI, 1.09-1.91; P=.010), and major bleeding (HR=1.85; 95%CI, 1.23-2.79; P=.003). However, the presence of valve disease was not a predictor of stroke/TIA and/or systemic embolism (P>.05) (Table 4A). The composite event, death, and major bleeding were also independently predicted by mitral valve regurgitation and aortic valve stenosis (Table 4B). Finally, discrimination analysis of the regression models showed c-statistics between 0.71 and 0.80 (detail in Table 4A-B).

The present study evaluated the frequency of heart valve disease and its prognostic value in a contemporary cohort of NVAF patients initiating DOAC therapy. This patient population had a high prevalence of significant valve disease, and this was associated with an elevated risk of adverse events during follow-up. This Spanish multicenter registry is the first to evaluate the frequency and prognostic value of heart valve disease in NVAF patients in a real-world setting, outside the context of a randomized clinical trial. We therefore considered that it would provide important clinical information.

In recent years, the definition of “valvular” AF has provoked controversy, probably due in part to the widely heterogeneous exclusion criteria used in different DOAC clinical trials in AF patients (Table 1).21–24 The appropriateness of the term valvular AF remains a subject of debate. Some authors propose the alternative term ‘mechanical and rheumatic mitral atrial fibrillation’ to describe the condition of patients for whom DOAC therapy is not indicated. Similarly, European guidelines propose the replacement of the current terminology with a classification based on the underlying specific valve disease. In the proposed definition, valve diseases are classified into 2 groups based on the indicated anticoagulant therapy: type 1 disease corresponds to AF patients with heart valve disease requiring VKA therapy (moderate to severe rheumatic mitral stenosis or mechanical valve prosthesis); type 2 disease corresponds to AF patients with heart valve disease requiring therapy with a VKA or a DOAC, also taking into consideration the thromboembolic risk.9,10 Given that valve disease affects up to 30% of AF patients, it would seem prudent, in the absence of consensus, to use the definition provided in the clinical practice guidelines.6,7,10,25

In the present study, the presence of heart valve disease in AF patients was associated with a worse clinical profile, in line with clinical trial substudies.11–14 Similarly, AF patients with significant valve disease treated in real-world clinical practice tended to be older and were more likely than unaffected patients to have ischemic heart disease, a history of major bleeding, heart failure, and renal deterioration. Although this profile would seem to predict higher scores on the main risk scales, published clinical trial substudies present an inconsistent picture. In the ARISTOTLE and ENGAGE-AF trials, patients with heart valve disease had significantly higher scores on the CHADS2 scale than those who did not (2.2 vs 2.1 and 2.9 vs 2.8, respectively; P <.001); however, no such differences were found in the RE-LY and ROCKET-AF studies.11–14 Scores on the HAS-BLED scale in AF patients were recorded in only 2 trials, and the results were once again inconsistent. Whereas the ENGAGE-AF trial found an increased bleeding risk among AF patients with valve disease (2.6 vs 2.5; P=.018), the ROCKET-AF trial showed no effect (2.8 vs 2.8; P=.18).12,14 In the present study, patients with heart valve disease had a higher thrombotic risk and a statistically nonsignificant tendency toward a higher bleeding risk (CHA2DS2-VASc, 3.9±1.6 vs 4.4±1.6; P <.001; HAS-BLED, 1.6±0.9 vs 1.8±0.9; P=.060).

In the present series, the presence of valve disease was one of the most notable independent predictors of death and bleeding complications. In contrast, valve disease showed no association with the risk of thromboembolic events, even though these patients had high scores on thrombotic risk scales. These findings are in broad agreement with the results of clinical trial substudies, especially the ENGAGE-AF trial.14 Patients with heart valve disease in the RE-LY and ROCKET-AF trials had a similar risk of stroke/systemic embolism and death, with an increase in the bleeding risk.11,12 Moreover, the risk of stroke/systemic embolism and death was higher among patients with heart valve disease in the ARISTOTLE trial, whereas there was no evidence of a differential effect on bleeding risk.13 A subsequent meta-analysis including all 4 substudies produced results consistent with the present study.15

The close relationship between valvular heart disease and adverse events during follow-up suggests that valve functional status should be a factor in risk stratification and the design of strategies to improve prognosis in these patients. The mechanisms linking significant valve disease to clinical events remain unclear. AF worsens the prognosis of patients with severe heart valve disease; moreover, valve disease and AF mutually reinforce each other through volume-pressure overload and neurohormonal factors.26–28 It is for this reason, when valve dysfunction is severe, that AF can be considered a marker of progressive disease. Heart valve disease is a major independent predictor of mortality in AF patients, as observed in the present study; in light of this strong association, valve repair or replacement is recommended to reduce mortality in this patient subgroup.29 Moreover, aortic stenosis is known to be associated with gastrointestinal bleeding, which has a major impact on quality of life, hospital admissions, and mortality. Bleeding and angiodysplasia in Heyde syndrome are a consequence of acquired Von Willebrand factor deficiency secondary to the shear forces generated on the surface of the stenotic and calcified aortic valve; even moderate shear forces trigger von Willebrand factor proteolysis.30 Given the bleeding risk associated with oral anticoagulant therapy, it is of fundamental importance to stratify bleeding risk in patients initiating this therapy and to modify factors that increase bleeding risk in this population. Established factors that increase this bleeding risk include concomitant antiplatelet therapy, nonsteroidal antiinflammatory drugs, and poorly controlled hypertension.31

NVAF patients initiating DOAC therapy frequently have co-occurring heart valve disease, and this is associated with an elevated risk of death, stroke/TIA/systemic embolism, and bleeding complications. These findings not only confirm the results of clinical trials in similar patient populations, but also extend them to real-world clinical practice.

None declared.