Differences in outcomes between oral anticoagulation “new starters” and “switchers” in patients with nonvalvular atrial fibrillation: A pooled analysis of the AMADEUS and BOREALIS trials

Abstract Background To explore differences in outcomes between dose‐adjusted vitamin K antagonists (VKAs) “new starters” and “switchers” in patients with nonvalvular atrial fibrillation (AF). Methods A post hoc analysis was performed to assess the outcome differences between VKA “new starters” and “switchers” in AF patients using pooled individual patient data of AMADEUS and BOREALIS trials. Results A total of 4169 AF patients were included in the present analysis, which included 1383 “VKA new starters” and 2786 “VKA switchers”. VKA new starters had higher crude rates of all‐cause mortality (P = .035) and cardiovascular death (P = .047) compared to switchers. On multivariable Cox regression analysis, both “new starters” and “switchers” showed nonsignificant trends for different risks of stroke/systemic thromboembolism (SE) (hazard ratio (HR): 1.66, 95%CI: 0.95‐2.90, P = .08), major bleeding (HR: 1.25, 95% CI: 0.73‐2.16, P = .42), and all‐cause death (HR: 1.09, 95% CI: 0.75‐1.57, P = .65). On Kaplan‐Meier analysis, both groups had similar risks of stroke/systemic embolism (P = .09), major bleeding (P = .28), and all‐cause death (P = .06). Conclusions In this post hoc analysis of clinical trial patients with AF, “new starters” and “switchers” for VKA initiation had nonstatistically significant rates of trial‐adjudicated thromboembolism, major bleeding, and all‐cause mortality.


| INTRODUC TI ON
Despite the non-vitamin K antagonist oral anticoagulants (NOACs) being introduced for stroke prevention in atrial fibrillation (AF) in recent years, the vitamin K antagonists (VKAs) are still used as oral anticoagulants (OACs) in large number of patients. 1,2 Switching from one OAC drug to another is common, especially for VKAs to be changed to NOACs. [3][4][5][6] However, there is lack of evidence from direct comparisons on the effectiveness and safety of VKAs in those who have VKAs initiated following previous use of other OACs (ie, "switchers") and in those who are newly started on VKAs ("new starters").
The effectiveness and safety of different modes of VKA initiation (ie, VKA "switchers" vs VKA "new starters") have been based on evidence from "real-world" studies 7,8 and meta-analyses. 9 Patient willingness and compliance as well as physician adherence and compliance may impact on outcomes. 10,11 The aim of this analysis was to explore the differences in outcomes between VKA "new starters" and "switchers" in patients with nonvalvular AF. We performed a post hoc ancillary analysis using pooled individual patient data from two randomized, open-label trials (AMADEUS and BOREALIS), with negligible subjective intention of switching between OACs and trial-adjudicated outcomes.

| ME THODS
The full details of the designs of the AMADEUS and BOREALIS trials have been described previously, and are summarized in Supplemental Methods. 12,13 The AMADEUS and BOREALIS trials were approved by the institutional review boards and all patients provided written informed consent. Patients from the VKA arms were categorized into two groups according to their treatment after randomization, that is, "VKA new starters" (on VKA and without previous VKA treatment) or "VKA switchers" (on VKA and with previous VKA treatment).

| Study endpoints
In this pooled analysis, we included all outcomes collected from the initiation of the treatment to the end of the studies. The primary analysis of the both trials reported only outcomes collected during the on-treatment period. The primary efficacy outcome of this analysis was the composite of stroke and systemic thromboembolism (SE). Stroke was further classified into ischemic and nonischemic stroke based on brain imaging results. SE was confirmed by angiography, surgery, or autopsy in the original trials. The primary safety outcome was major bleeding, as defined on previously published criteria. 12,13 Other efficacy outcomes of venous thromboembolism, myocardial infarction, and safety outcomes of any clinically relevant bleeding, intracranial hemorrhage, were defined according to the original trials. 12,13 All-cause death and the subgroups of fatal stroke and cardiovascular death were also assessed in this analysis. 12,13 Only the first event of each outcome and their occurrence date were used for analysis. Suspected outcome events in both trials were adjudicated by each independent central adjudication committee that was blinded to the treatment assignment.

| Statistical analysis
Baseline characteristics were reported as percentages or mean ± SD.
On multivariable Cox regression analysis, with "VKA switchers" used as reference, "VKA new starters" had nonsignificant risks of stroke/SE, major bleeding, and all-cause death in Model I and Model II (Figure 1). Kaplan-Meier survival curves did not show significant difference in stroke/SE (P = .09), major bleeding (P = .28), and all-cause mortality (P = .06) between the two groups ( Figure 2).

| D ISCUSS I ON
In this ancillary analysis of the AMADEUS and BOREALIS trials, VKA "new starters" had a nonstatistically significant risk of stroke/SE and similar rates of major bleeding and all-cause death compared with VKA "switchers", after adjusting for associated comorbidities and risk factors. This is contrary to the perception that "new starters" and "switchers" were patients at higher risk of adverse outcomes. TA B L E 1 (Continued) need to be carefully assessed. 14 Relatively lower effectiveness and safety of "switchers" from VKA to NOACs compared to NOAC "new starters" have been reported, and have questioned whether the switching to NOACs is a marker of poor adherence and higher comorbidity. 8,10,15,16 As patients were randomly allocated to the study groups in the randomized trials, these concerns on adherence and comorbidities are reduced in the current analysis.
One previous study, 16  Previous "real-world" studies suggested that the benefits appeared to be decreased in NOAC switchers, with the assumption of poor compliance or residual confounding from comorbidities among NOACs switchers. 9,10 In the present ancillary pooled analysis of AMADEUS and BOREALIS trials on a group of VKA-treated patients only, those who were VKA starters had similar risks of stroke/SE, major bleeding, and all-cause death compared with VKA switchers. The main difference of the present study compared with previous comparisons lies in the fact that there was no drug switching, but not whether they had prior VKA treatment history before their strict trial-related follow-up. This meant that F I G U R E 2 Nelson-Aalen cumulative hazard ratios of stroke/SE, major bleeding, and all-cause death. SE, systemic thromboembolism; VKA, vitamin K antagonists benefits were not reduced and risks were not increased by prior VKA treatment history.
One point worth mentioning is that the higher risk of all-cause mortality rates in "VKA new-starters" was caused by higher burden of cardiovascular diseases. One possible reason could be that they were more likely to take antiplatelet agents (aspirin, clopidogrel) as their first antithrombotic choice, but not warfarin before randomization. However, this significant difference disappeared after multivariate adjustment.

| CON CLUS IONS
In this post hoc analysis of clinical trial patients with AF, "new starters" and "switchers" for VKA initiation had nonstatistically significant rates of trial-adjudicated outcomes of thromboembolism, major bleeding, and all-cause mortality. This is contrary to the perception of healthcare professionals that "new starters" and "switchers" were patients at higher risk of adverse outcomes.

ACK N OWLED G EM ENTS
This analysis was conducted independent of any industry collaboration or sponsorship.