Detection of atrial fibrillation in real world setting in patients with cryptogenic stroke and an implantable loop recorder

Implantable loop recorders (ILR) are used to screen for atrial fibrillation (AF) in patients with cryptogenic stroke (CS). However, there is limited real‐world data regarding the long‐term rate of AF detection using ILR and management consequences in patients with CS. The objective is to assess the rate of AF detection in patients with CS in a real‐world study over 36 months of follow‐up and its consequences on stroke prevention.


INTRODUCTION
Approximately 795,000 individuals experience a stroke every year in USA. 1 Ischemic strokes represent 87% of all strokes and are a leading cause of significant disability and mortality in patients with cardiovascular disease. 1

Outcomes
The primary outcome was detection of AF in patients with ILR.

Statistical analysis
Means and standard deviation were used to represent continuous variables. Distribution of baseline characteristics were described using counts and percentages for categorical variables. Student's t-test or Fisher's exact test were utilized for assessment of differences between groups, as appropriate. For any statistically significant interactions, a post hoc comparison was performed with an unpaired t-test with Bonferroni correction to detect differences between groups and types.

Patient characteristics
Two hundred and twenty-five patients met inclusion criteria (Tables 1 and 2). The average patient age was 63.1 years (SD = 13.5) and 51.1% were female. A significant proportion of patients, 27.6%, had prior  The recurrence rate of stroke in patients diagnosed with AF was 9.3%.

Rate of AF detection after ILR implant
Due to the discrepancies in sample size between the patients who had ILRs are largely more effective for AF detection with longer monitoring times and higher sensitivities in comparison to external ECG monitors, with data from meta-analyses consistently finding higher rates of AF detection with ILRs compared to wearable devices. 9 In the CRYSTAL-AF trial, 441 patients with CS were randomized to long-term monitoring with an ILR or conventional follow up. By 36 months, AF was detected in 30% of patients with ILR as compared to only 3% of patients in the control group (hazard ratio 8.8, p < .001), a similar rate of detection to that observed in our study (Figure 1). 5 Our study cohort had a higher proportion of African American Despite the high proportion of African American patients in the current study, similar rates of AF were detected at 36 months when compared to CRYSTAL-AF, which had a predominantly white cohort.
However, a direct comparison is difficult as our cohort had significantly higher baseline vascular risk factors and CHADS 2 score as compared to the CRYSTAL participants ( Table 2). The increased burden of cardiovascular risk factors observed in our cohort may be contributing to the overall increased rates of AF incidence in African American patients.
In addition to the higher proportion of African American patients, our study cohort also had a higher proportion of female patients (51.1%) as compared to 35.7% in the CRYSTAL-AF cohort. 5 Ko et al.
conducted a literature review that investigated sex-specific differences in the epidemiology of AF; while females have overall lower prevalence and incidence of AF, females had higher incidence of hypertension and valvular heart disease. Additionally, there are studies suggesting possible sex-specific structural differences, such as smaller left atria and ventricles, decreased LV wall thickness, increased atrial fibrosis, nonpulmonary foci, and electrical activity differences attributed to both estrogen and membrane potential and action potential duration. 12 Overall, since females have an increased risk for thromboembolic events with AF and increased risk for mortality, further studies are needed to elucidate the sex-specific differences in AF risk and its impact on risk stratification of patients who should get ILR monitoring for CS assessment. 13 Another novel finding in our study is the high incidence of recurrent strokes occurring in patients without ILR detected arrhythmia (n = 27, 87.1% of total recurrent strokes). Previous European studies have reported variable rates of recurrent stroke following CS in the absence of ILR detected AF, ranging from around 50%-80%. 14,15 The higher proportion of recurrent strokes in our study may partly be accounted for by the high proportion of African American patients.
Despite African American patients having lower rates of AF as compared to white patients, the incidence of stroke, including CS, is increased in African-American patients, which suggests that the mechanisms underlying increased stroke risk in this patient population are primarily mediated by factors other than arrhythmia. 16 Furthermore, in patients with embolic stroke of undetermined source (ESUS), a sub-group of nonlacunar CS, there is often coexistence of multiple potential embolic sources besides AF. 17,18 It is possible that the higher burden of cardiovascular risk factors seen in our cohort may further increase the burden of potential embolic sources.
As demonstrated in our study, there is increasing recognition that risk of embolic stroke may be elevated even in the absence of AF. 19 Indeed, the direct causal relationship between ILR detected subclinical AF and stroke has been called into question by studies observing lack of a clear temporal relationship between detected AF and stroke event. 20,21 Furthermore, randomized trials to date have failed to show benefit with ILR implantation with regards to recurrent stroke reduction in CS patients, despite improved rates of AF detection and increased AC initiation. 22,23 In these patients without arrhythmia, an underlying atrial cardiomyopathy, characterized by atrial dilation fibrosis and dysfunction leading to an atrial thromboembolic substrate, is thought to play a key role in the pathogenesis of CS. 19  The aforementioned remodeling changes and structural differences seen in African American patients may be interesting to investigate further in our cohort to assess if the recurrent stroke incidence is more secondary to other cardiovascular risk factors or undiagnosed atrial cardiomyopathy that predisposes the patient to thromboembolic events. 24 The ongoing randomized ARCADIA (Atrial Cardiopathy and Antithrombotic Drugs in Prevention After Cryptogenic Stroke) trial will provide further evidence for this hypothesis by studying the effects of apixaban versus aspirin in ESUS patients with evidence of atrial cardiomyopathy. 25 The cumulative incidence of stroke recurrence following CS has been reported to be around 20% at 3 years. 26

Limitations
There are some limitations in our study that need to be considered.
First, this study is a retrospective study from one medical center and so results may not be generalizable to a broader population. Additionally, with AF detection on long-term ECG monitoring through ILR and the subsequent initiation of anticoagulation leading to overall stroke reduction, data points were extracted under the assumption that patients continued follow-up at the same hospital. There is a chance that some patients with subsequent strokes were not detected if they transferred their care to an outside facility. An additional limitation and consideration of this study design is that ILR implantation requires the presence of electrophysiologists at both the hospital and clinics for follow-up and monitoring given the substantial rates of false positive AF detections. Lastly, in the creation of the cohort, patients were excluded if they did not have any ILR data transmission reports in the Baylor Epic System or had no subsequent follow-ups with Baylor EP clinic for ILR monitoring.

CONCLUSION
Our study demonstrated similar rates of AF detection as compared to CRYSTAL-AF. However, in comparison to CRYSTAL-AF and other ILR AF studies, our population had a higher percentage of African American and female patients. Despite high detection rates of AF, most recurrent strokes after ILR implant occurred in patients without detected AF. Further studies are needed to delineate the optimal strategy for ILR implantation as secondary stroke prevention in CS patients.