Ischemic stroke causes disability, and can lead to death (1). Atrial fibrillation (AF) is a treatable condition that can lead to ischemic stroke (2) and is known to be associated with the highest severity of stroke (3). However, AF, frequently paroxysmal and asymptomatic also often remains undetected in patients with ischemic stroke, especially when conventional monitoring methods - i.e. 12-lead electrocardiography (ECG), 24-hour Holter ECG, or 24-hour continuous telemetry monitoring - are used (4-6). Patients with ischemic stroke caused by AF are moreover under increased risk of recurrent stroke (7). Therefore, current guidelines suggest initiating anticoagulant therapy after ischemic stroke in documented AF (8). It seems obvious that strategies which improve the diagnosis of atrial fibrillation will reduce reoccurrence of stroke. Indeed, several observational studies have shown that there is an increased rate of detection of AF by serial or prolonged (ECG) monitoring in patients with ischemic stroke (5, 9).
Based on this data, two new randomised controlled trials (RCTs) sought to determine the benefit of long term heart rhythm monitoring compared to conventional diagnostic methods (10, 11). The EMBRACE study (30-Day Cardiac Event Monitor Belt for Recording Atrial Fibrillation After a Cerebral Ischemic Event; set in Canada) investigated a noninvasive 30 day event-triggered loop recorder from Braemar (ER910AF Cardiac Event Monitor) (11), while CRYSTAL-AF (CRYptogenic STroke And underLying AF Trial; conducted in Europe, Canada, and the US) used an insertable cardiac monitor (ICM) from Medtronic (REVEAL XT).
I - Cryptogenic Stroke and detection of AF
Atherosclerosis, small vessel occlusion, and cardioembolism are the main causes of ischemic stroke. Less commonly, toxins, hypercoagulable disorders, and vasospasm can induce ischemic stroke. Each source carries its own risk of recurrence and prognosis (3, 12). The ASCO (atherosclerosis, small vessel disease, cardiac causes, other uncommon causes) (13) or the TOAST (Trial of ORG 10172 in acute stroke treatment) classification (14) and its enhancement the Causative Classification System for Ischemic Stroke (CCS) (15, 16) (available free for academic use at http://ccs.mgh.harvard.edu) were tools designed for subtyping ischemic stroke. If however, despite intensive diagnostic workup, the source cannot be determined, the stroke is labelled cryptogenic or etiologically undefined. Approximately 25-30 % of all ischemic strokes fall into this category (4, 10, 17).
The AF detection rate with 12-lead ECG ranges between 2 and 5% after ischemic stroke or transient ischemic stroke (TIA) (18), and between 2 and 6% if a 24-hour Holter device is used (19, 20). However, ECG is suspected to have limited sensitivity and limited negative predictive value for AF (6, 21).
2 - CRYSTAL-AF and EMBRACE
Rationale: Both CRYSTAL-AF and EMBRACE hypothesised that prolonged monitoring of patients with unexplained TIA or stroke increases the detection rate of AF and prescription of anticoagulant therapy on the promise from previous observational studies that ischemic stroke patients had shown an increased detected rate of AF by serial or prolonged ECG monitoring than conventional 24 hour ECG.
Trial design: Both trials are RCTs.
Randomisation: Both RCTs enrolled around 500 patients (447 in CRYSTAL-AF and 572 in EMBRACE) with cryptogenic stroke whose diagnosis was made by exclusion after intensive diagnostic work up. They were randomised in a 1:1 fashion into an intervention and a control group. In the intervention groups AF was mainly asymptomatic at the time of the first diagnosis.
Results (Table I): In both trials, detection rate of AF was significantly higher in the intervention group – whether using the implanted device or the belt.
|Patients enrolled - n||572||447|
|Mean age in years ±SD||72.5 ± 8.5||61.5 ± 11.3|
|Median CHADS2Score of all patients||3||3|
|Hypertension in intervention group - n/n total IG(%)||204/286 (71.3%)||144/221 (65.2%)|
|Diabetes mellitus in intervention group - n/n total IG(%)||55/286 (19.2%)||34/221 (15.4%)|
|Monitoring device in intervention group||30 day event-triggered loop recorder from Braemar (ER910AF Cardiac Event Monitor)||insertable cardiac monitor from Medtronic (REVEAL XT)|
|Invasive insertion of device required||No||Yes|
|Primary endpoint||newly detected AF lasting >30s within 90 days after randomisation||time to first detection of AF (lasting >30s) within 6 months|
|Follow up - months||3 mth||6 mth||12 mth||36 mth|
|AF Detection rate|
|Intervention group - n/n total IG*||45/280 (16.1%)$||19/221 (8.9%) §||29/221 (12.4%)#||42/221 (30.0%)&|
|Control group - n/n total CG *||9/277 (3.2%)||3/220 (1.4%)||4/220 (2.0%)||5/220 (3.0%)|
|Number needed to screen~ - n||8||14||10||4|
Table I. Main similarities and differences concerning patients methods, and results of EMBRACE and CRYSTAL-AF.
*Consisting only of patients included in primary analyses.
$ AF detection was significantly increased compared to the EMBRACE control group p<0.001; CI 8.0 – 17.6).
§ AF detection was significantly increased compared to the CRYSTAL-AF control group (p<0.001; HR 6.4; 95% CI 1.9 – 21.7).
# AF detection was significantly increased compared to the CRYSTAL-AF control group (p<0.001; HR 7.3; 95% CI 2.6 – 20.8).
& AF detection was significantly increased compared to the CRYSTAL-AF control group (p<0.001; HR 8.8; 95% CI 3.5 – 22.2).
~Number of patients needed to be evaluated with prolonged monitoring to detect a first episode of AF.
AF: Atrial fibrillation; HR: hazard ratio; CI: confidence interval; IG; intervention group; CG: control group.
Atrial fibrillation detection rates (primary endpoint) for
- CRYSTAL-AF groups: implanted device: 8.9% after 6 months, months. Control: 1.4% after 6 months;
- EMBRACE groups: belt: 16 % after 3 months. Control: 3.2% after 3 months.
Anticoagulant therapy: The mean CHADS2 Score was 3 (5.9% annual stroke risk) in both study populations. Patients in the intervention arms received oral anticoagulant therapy more often than in the conventional monitoring group.
- CRYSTAL-AF groups: implanted device: 10.1% after 6 months and 14.7% after 12 months. Control: 4.6% after 6 months and 6.0% after 12 months.
- EMBRACE groups: belt: 18.6% after 3 months. Control: 11.1% after 3 months.
B) Differences and limitations
Patient characteristics: mean age and number of patients with hypertension or diabetes in EMBRACE compared to CRYSTAL-AF was higher (71.3% vs. 65.2%; and 19.2% vs. 15.4%; 72.5±8.5 years versus 61.5±11.3 years).
Primary Endpoint: CRYSTAL-AF: time to first detection of AF at 6 months follow up whereas in EMBRACE it was set as newly detected AF or atrial flutter lasting 30 seconds or longer within 90 days after randomisation.
Complications: Of the 208 ICMs implanted in the CRYSTAL-AF population 5 ICMs had to be removed due to infection at the insertion site or pocket erosion.
3 - Patient characteristics challenge apparent superiority of non- invasive monitoring in detecting AF
Overall results: By increasing the detection rate of AF in patients with cryptogenic stroke by up to one third, CRYSTAL-AF and EMBRACE showed that long term heart rhythm monitoring is superior to conventional diagnostic methods that include 24 h ECG. These results confirm the findings of observational studies that had demonstrated a high rate of undetected AF in patients after cryptogenic stroke.
Monitoring device: The results of CRYSTAL-AF clearly showed a correlation between the duration of monitoring and an increased detection rate of AF for a period of up to 36 months. The detection rate of first episodes of AF increased from 8.9% after 6 months to 12.4% after 12 months up to 30.0% after 36 months in the intervention group. This seems to underline the need for a monitoring longer than the 3 months investigated in EMBRACE. Concerning extended monitoring a major advantage of an implanted device (CRYSTAL) compared to a noninvasive monitoring by belt (EMBRACE) is the straightforward possibility of extended monitoring without any extra inconvenience for the patient.
Apparent superiority: The trial results would seem to demonstrate superiority of the noninvasive method in detecting AF: In EMBRACE detection of first diagnosed AF was 16.1% within 3 months after randomisation by using a 30-day event-triggered recorder. Detection in the intervention group of CRYSTAL-AF was 8.9% after 6 months.
However, the mean age and the number of patients with hypertension or diabetes in EMBRACE was higher than in CRYSTAL-AF, which might explain lower detection in the intervention group of CRYSTAL-AF compared to that of EMBRACE. This is underpinned by the finding that the detection rates of AF in the control groups are also lower in the population of CRYSTAL-AF than in the population of EMBRACE (1.4% vs. 3.2%).
4 - Advantages of ICM
A study comparing noninvasive versus invasive monitoring would be needed to determine superiority of one method over the other. Treating physicians for the time being need to weigh the patient’s individual needs to determine the appropriate monitoring device. Advantages of ICM are that there is a) No need for patient compliance after ICM insertion, and b) Monitoring for several months is easily possible. The greatest disadvantage however is that implantation bears a risk of infection, pocket erosion and wound healing problems. Conversely, this is currently the only advantage of a noninvasive method.
5 - Anticoagulation during monitoring
Expert opinion is controversial when it comes to patients with increased risk factors but unproven AF (22, 23). On one hand, without effective anticoagulation these patients are at increased risk of recurrent ischemic stroke. However, effective anticoagulation in patients without yet detected AF may place the patient under an unnecessary increased risk of bleeding. An intense view into each patient’s medical history and especially in terms of history of bleeding and potential risk of bleeding is suggested. If the treating physician decides that early anticoagulation is required, anticoagulation only in combination with long term monitoring with an implantable device along - with prompt interruption of anticoagulation should AF remain undetected - is strongly recommended. This way potential overuse of anticoagulation is limited. However, it has to be mentioned that, if the patient is compliant a noninvasive monitoring system might be a potentially less invasive method.
6 - Embolic stroke of undetermined source (ESUS)
To overcome the uncertainty between antiplatelet and anticoagulant medication in the initial phase after the diagnosis of cryptogenic stroke, a new recent construct called embolic stroke of undetermined source (ESUS) has been offered for use (17). Patients with ESUS are a subgroup of patients with cryptogenic stroke. ESUS refers to patients with cryptogenic stroke who have undergone sufficient diagnostic assessment to rule out major risk cardioembolic sources, occlusive atherosclerosis, and lacunar stroke. The idea is that ESUS patients have a causal mechanism that is mostly embolism of unestablished source (17). Based on the ESUS construct a RCT comparing rivaroxaban and acetylsalicylic acid was initiated in December ‘14. This new trial may provide the physicians with the evidence based results necessary for determining whether anticoagulation is necessary in these patients or not.
7 - Anticoagulation after prolonged monitoring
In case of non-cardioembolic stroke, the current joint guidelines for the early management of patients with acute ischemic stroke of the American Heart Association (AHA) and the American Stroke Association (ASA) recommended antiplatelet therapy (2). In case of documented AF as plausible cause of ischemic stroke, the risk of recurrence of ischemic stroke is high (24). Furthermore, the stroke risk was found to be similarly high independently of whether AF was paroxysmal or sustained (25, 26). Additionally, investigating pacemaker patients have revealed that subclinical atrial fibrillation predicts the onset of clinically evident AF and recurrence of stroke (4, 27). Current guidelines therefore recommend initiating anticoagulant therapy after ischemic stroke when AF is documented.
Current AHA and ASA guidelines recommend performing an ECG in patients with cryptogenic stroke for 24 hours (or more – with no further detail). EMBRACE and CRYSTAL-AF showed that extended monitoring increased the detection rate of AF in patients with cryptogenic stroke by up to one third. We recommend considering cardiac event recorder implantation to ensure straightforward, objective cardiac monitoring for several months independently of patients compliance and motivation, or at least considering the use of a prolonged noninvasive monitoring method in patients with cryptogenic stroke, especially in patients who are known to be at increased risk of AF.