Rivaroxaban in Rheumatic Heart Disease–Associated Atrial Fibrillation

Abstract

BACKGROUND

Testing of factor Xa inhibitors for the prevention of cardiovascular events in patients with rheumatic heart disease–associated atrial fibrillation has been limited.

METHODS

We enrolled patients with atrial fibrillation and echocardiographically documented rheumatic heart disease who had any of the following: a CHA₂DS₂VASc score of at least 2 (on a scale from 0 to 9, with higher scores indicating a higher risk of stroke), a mitral-valve area of no more than 2 cm², left atrial spontaneous echo contrast, or left atrial thrombus. Patients were randomly assigned to receive standard doses of rivaroxaban or dose-adjusted vitamin K antagonist. The primary efficacy outcome was a composite of stroke, systemic embolism, myocardial infarction, or death from vascular (cardiac or noncardiac) or unknown causes. We hypothesized that rivaroxaban therapy would be noninferior to vitamin K antagonist therapy. The primary safety outcome was major bleeding according to the International Society of Thrombosis and Hemostasis.

RESULTS

Of 4565 enrolled patients, 4531 were included in the final analysis. The mean age of the patients was 50.5 years, and 72.3% were women. Permanent discontinuation of trial medication was more common with rivaroxaban than with vitamin K antagonist therapy at all visits. In the intention-to-treat analysis, 560 patients in the rivaroxaban group and 446 in the vitamin K antagonist group had a primary-outcome event. Survival curves were nonproportional. The restricted mean survival time was 1599 days in the rivaroxaban group and 1675 days in the vitamin K antagonist group (difference, −76 days; 95% confidence interval [CI], −121 to −31; P<0.001). A higher incidence of death occurred in the rivaroxaban group than in the vitamin K antagonist group (restricted mean survival time, 1608 days vs. 1680 days; difference, −72 days; 95% CI, −117 to −28). No significant between-group difference in the rate of major bleeding was noted.

CONCLUSIONS

Among patients with rheumatic heart disease–associated atrial fibrillation, vitamin K antagonist therapy led to a lower rate of a composite of cardiovascular events or death than rivaroxaban therapy, without a higher rate of bleeding. 

Discussion

In patients with atrial fibrillation not related to rheumatic heart disease, treatment with rivaroxaban or other factor Xa inhibitors has been shown to be noninferior to warfarin therapy for stroke prevention, with large reduction in the risk of hemorrhagic stroke.⁴ The patients in the present trial were much younger (mean age, 50.5 years) and more likely to be women (72% of the patients) than in recent trials including only patients without rheumatic heart disease.⁴ A lower percentage of patients in this trial had hypertension (23%) than in previous trials. However, on the basis of previous trials, we expected to find a similar or higher risk of stroke because mitral stenosis has been associated with a high risk of stroke. We expected generally similar results in the present trial on the basis of the expectation that, despite a different underlying cardiac condition, the underlying stroke mechanism of embolism of left atrial thrombus would be similar in rheumatic heart disease as in other types of atrial fibrillation. We did not expect to observe a difference in mortality. The phase 3 trials confirming the safety and efficacy of the new oral anticoagulants, which included patients with atrial fibrillation not due to rheumatic heart disease and which compared non–vitamin K antagonist oral anticoagulants with vitamin K antagonists, showed consistent effects between patients with (nonrheumatic) valvular heart disease and those without such disease.⁹ A randomized trial comparing rivaroxaban with vitamin K antagonist therapy in patients with atrial fibrillation and bioprosthetic mitral valves showed a lower risk with rivaroxaban of stroke at 1 year of follow-up and no significant difference in mortality.¹⁰ Thus, the results of the present trial were unexpected.

Possible explanations for these findings include the reduced power of this trial for the outcome of stroke, because the rates of stroke in the two groups were lower than expected; in addition, the difference in the rate of stroke was modest, which suggests that the difference could be due to chance. The rate of the composite outcome of stroke or systemic embolism, which is widely accepted in trials of stroke prevention in patients with atrial fibrillation, did not differ between the two treatment groups. However, the difference in mortality was large and is therefore unlikely to be due to chance. Patients in the vitamin K antagonist group had more physician interactions than those in the rivaroxaban group because of the need for monthly monitoring of INR control. This situation could have resulted in better overall care and fewer strokes and deaths. It is possible that adherence to rivaroxaban therapy was worse than to vitamin K antagonist therapy because patients in the rivaroxaban group knew that they were not having the INR monitored. The difference in stroke rates between the two groups could also be, in part, related to the higher incidence of discontinuation of rivaroxaban, even though many of the patients who discontinued rivaroxaban then received a vitamin K antagonist. The most common reasons given for the discontinuation of rivaroxaban were hospitalization for valve surgery and patient decision. Some patients in the rivaroxaban group received mechanical valves, which necessitated a switch to vitamin K antagonist therapy to prevent valve thrombosis. However, the discontinuation of rivaroxaban does not explain the observed benefit of vitamin K antagonist therapy because the on-treatment analysis, which excluded any events that occurred more than 5 days after the discontinuation of trial treatment, showed results that were almost identical to those of the intention-to-treat analysis.

A mortality difference in favor of either treatment was not expected. In a meta-analysis of trials of non–vitamin K antagonist anticoagulants as compared with warfarin, mortality was 10% lower with non–vitamin K antagonist anticoagulants.⁴ This difference appears to be driven mostly by the large reductions in the risk of hemorrhagic stroke with the newer agents. The effect of vitamin K antagonist therapy on mortality in the present trial appears to be mostly unrelated to stroke prevention. The absolute number of strokes prevented was small (25 strokes), as compared with the absolute number of deaths prevented (110 deaths). The lower mortality that was observed with vitamin K antagonist therapy than with rivaroxaban therapy in this trial is also clearly not related to any effect on bleeding, given that bleeding was not less common with vitamin K antagonist therapy than with rivaroxaban therapy. Treatment with a vitamin K antagonist does not appear to have slowed the progression of heart-valve deterioration, because rates of valve replacement surgery or valvuloplasty were similar in the two groups. The use of a vitamin K antagonist led to a lower rate of death from vascular causes than rivaroxaban therapy, with lower rates of both sudden cardiac death and death from mechanical or pump failure. Thus, the lower rates of sudden cardiac death and of death from mechanical or pump failure with vitamin K antagonist therapy than with rivaroxaban therapy are not readily explained by effects on stroke, bleeding, or valve deterioration.

During the first 12 to 18 months of follow-up, little difference was seen between the vitamin K antagonist group and the rivaroxaban group (Figs. Figure 1 and Figure 2). After that, a lower rate of the primary composite outcome in the vitamin K antagonist group than in the rivaroxaban group became evident and was substantial beyond 3 years. We speculate that a delayed effect could be occurring, in part owing to improvement in the management of vitamin K antagonist therapy during the initial phase of the trial. It is also possible that there was a delay in the onset of the benefit from a vitamin K antagonist over rivaroxaban that is independent of INR control.

A delay in the onset of a benefit of vitamin K antagonist therapy was also seen in the Warfarin versus Aspirin in Reduced Cardiac Ejection Fraction (WARCEF) trial.¹¹ The WARCEF trial tested whether warfarin therapy would lead to a lower rate of death or stroke than aspirin among patients with heart failure due to a reduced ejection fraction without atrial fibrillation. The trial showed no benefit overall, but a time-varying Cox analysis showed a benefit of warfarin therapy that emerged only later in follow-up, as in the INVICTUS trial. The hazard ratio in favor of vitamin K antagonist therapy decreased by a factor of 0.89 per year (95% CI, 0.80 to 0.998; P=0.046) and became significant by year 4 (hazard ratio, 0.76; P=0.04).

No evidence suggests that rivaroxaban therapy increases mortality among patients with other heart conditions. Treatment with rivaroxaban reduces mortality substantially among patients with atherosclerotic vascular disease.¹² Thus, our data support the hypothesis that vitamin K antagonist therapy reduces the risk of death from vascular causes among patients with rheumatic heart disease; this effect appears to be independent of the prevention of atrial fibrillation–related stroke and suggests a direct effect on the disease process of rheumatic heart disease. Our trial showed that as compared with rivaroxaban, vitamin K antagonist therapy led to a lower rate of ischemic stroke among patients with rheumatic heart disease–associated atrial fibrillation and lower mortality due to vascular causes, without significantly increasing the rate of major bleeding. The results of this trial support current guidelines, which recommend vitamin K antagonist therapy for the prevention of stroke in patients with rheumatic heart disease in whom atrial fibrillation develops.

Supported by a nonrestricted grant from Bayer to the Population Health Research Institute.

Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.

This article was published on August 28, 2022, at NEJM.org.

A data sharing statement provided by the authors is available with the full text of this article at NEJM.org.

SOURCE: NEJM