Discussion
As few therapies to halt progression of cognitive decline and dementia have been ineffective, attention has been turned towards identification of moderate-high risk patients and early use of potential preventative treatments.10 In the setting of patients with AF, rhythm control and early use of anticoagulation are potential means to reduce the risk of cognitive decline and dementia.19
Before considering the findings of this study of anticoagulation for patients with AF, it is important to recognize this was a vanguard study and the event rates were lower than expected. As described above, we used event rates of prior observational studies to calculate a power to determine a difference in the primary endpoint. However observational trials have limitations and as such we prespecified in the protocol that this would be a vanguard study that would enroll 120 patients to “determine trends of cognition in the prospective study design that include incidence of dementia and the percent of patients that develop moderate decline in each study group to compare amongst the rates observed in the general population, the feasibility of recruitment, study design and subject retention, budget adherence and likelihood of obtaining the desired outcomes for potential future trial design and planning”. The neutrality of the outcomes and the biomarkers that showed very similar efficacy of both strategies were considered in the decision to not move forward with a larger, fully powered study based upon these randomized prospective data and their event rates.
Nonetheless, the current trial resulted in several key and interesting observations:
  1. Anticoagulation use in this prospective randomized trial, whether with warfarin or dabigatran was associated with very low dementia and stroke rates. These findings with contemporary management for other associated cardiovascular disease states suggest that a trial to detect any differences would need to be much larger and likely require an extended follow-up period beyond 2 years.
  2. Serial cognitive testing results after anticoagulation initiation in did not demonstrate cognitive decline and trended towards modest improvement. We hypothesized that the generalized cognitive scoring improvement may reflect recall bias from serial testing20, improved follow-up and management, and perhaps influence of the anticoagulant as it related to brain perfusion with a reduction in risk of ischemic brain injury over time.
  3. Anticoagulation use in patients results in a durable reduction of thrombotic and ischemic biomarkers that have been shown, when elevated, to be associated with the risk of stroke and vascular injury. However, similar to many trials of anticoagulation and population-based studies of anticoagulation use, long-term use and compliance remains a challenge.5-7 In this study approximately 1/3 discontinued their anticoagulation for a variety of reasons. Underuse has clearly been shown to increase risk of stroke. In this analysis, those at discontinued therapy also impacted our ability to study both therapies as to how they relate to anticoagulation.21
  4. Stroke, dementia, and cognitive outcomes were similar with both routine start prescription of dabigatran or CPAS-directed warfarin anticoagulation in patients with AF at 2 years.
We observed a lower rate of moderate-severe cognitive decline and dementia than expected based on findings from a routine population of patients with AF treated with warfarin.22 These lower rates may be from early and effective use of anticoagulation after AF diagnosis that has been shown to lower dementia risk as this trial involved all patients referred for AF management that included a goal of anticoagulation initiation.23 It is also possible that the relatively small patient population, with approximately 1/3 discontinuing anticoagulation, rendered the trial unable to detect a true incidence within the defined study follow-up period. Finally, in a prospective, randomized trial with frequent follow-up visits the outcomes may be different due to more frequent assessment of drug efficacy, identification and management of risk factors for AF-related comorbidities, more close management of anticoagulation, and selection bias that occurs in patients that are willing to enroll in these types of studies.
These data are complementary to those recently presented from the GIRAF (CoGnitive Impairment Related to Atrial Fibrillation) trial which was a two-year randomized, multi-center, prospective trial in Brazil that evaluated the effects of dabigatran and warfarin on cognitive and functional impairment, bleeding, and cerebrovascular complications in patients (>70 years of age) with atrial fibrillation.24 In this trial, patients underwent 90-minute cognitive and functional evaluations at the one-year and two-year follow-up visits as well as MRI brain imaging. At the conclusion of the trial, there were no patients that developed dementia and cognitive scoring assessments were identical in both treatment groups (<1 point difference). The study investigators concluded, similar to our study, that either strategy was acceptable as means to treat patients with AF to lower stroke risk and that neither can be used to preferentially impact risk of cognitive decline. These data combined with ours now comprise over 250 patients with serial testing of cognition after anticoagulation initiation with either dabigatran or warfarin.
Our data provide some additional mechanistic understanding into the equivalent outcomes when the biomarkers are considered. In our analysis, multiple markers of vascular injury, thrombosis, and inflammation were similar throughout the study in both treatment groups. This would suggest that both anticoagulation strategies were highly effective at the primary goal of reducing risk of micro and macro thromboembolism with reasonable safety of use. These data also support very well managed patients on warfarin anticoagulation. Highly effective warfarin can be accomplished in centers such as ours that use a dedicated \community pharmacy anticoagulation service (CPAS) guidance and administration; however in the general community both general warfarin use and compliance to DOAC therapies can be less optimal and the result in higher rates of brain injury.25, 26 In this community setting of less efficacy, a drug with a larger therapeutic window compared to warfarin may improve outcomes that were explored in this analysis.
The study has several limitations. First, this was a vanguard analysis and as such was underpowered to detect a significant difference in the endpoints considered in this analysis. This study design was pursued to inform on a subsequent study design, however the lack of any signal of difference and nearly equivalent outcomes between anticoagulation strategies in this analysis did not affirm a decision to perform a larger study; as such there was value in this study design and the results. Next, common to all anticoagulation studies, long-term adherence was suboptimal and reflects many decisions including a desire to not use anticoagulation therapy in general. In a recent community-based trial of predictors of contemporary oral anticoagulants use and adherence, persistence use at 2 years was approximately 80% and continued to steadily decline to approximately 70% at 4 years. In this analysis, lack of adherence significantly correlated with higher rates of stroke and stroke-related mortality.21 Finally, patients were enrolled that were willing and able to complete the battery of cognitive tests. This may have selected a patient population that would perform better on these tests compared to a general population and long-term results as well can be influenced by sequential testing and learned responses.