Completion of the Etiologic Workup: Roles for Advanced Cardiac Imaging and Long-Term Cardiac Monitoring

For vascular neurologists in both inpatient and outpatient settings, one of the primary goals in caring for people with ischemic stroke is determination of stroke etiology. We focus on identifying one of the recognized etiologic categories because doing so is critical to optimizing secondary stroke prevention measures. The most used system for classification of stroke etiologies is the Trial of Org 101072 in Acute Stroke Treatment (TOAST) criteria.¹ Developed in the 1990s for a trial of a heparinoid compound in treatment for acute ischemic stroke, the TOAST schema divided ischemic stroke into 5 etiologic subgroups: large-artery atherosclerosis, cardioembolism, small-vessel occlusion, stroke of other determined etiology, and stroke of undetermined etiology. Whereas the therapeutic trial was negative, the TOAST framework has remained a mainstay of stroke diagnostics and has defied attempts to replace it with more sophisticated classification systems.

Cryptogenic stroke (CS) is a term commonly applied to the 5th TOAST category: stroke of undetermined etiology. These are nonlacunar strokes without evidence of ipsilateral large-artery stenosis of ≥50%, a major cardioembolic source, or other specific mechanism of stroke despite a complete workup. The term embolic stroke with unknown source (ESUS) is often used interchangeably with CS (reported to account for between 25% and 40% of all ischemic strokes). As stroke etiologic workups have improved over time, the rate of strokes classified as cryptogenic has declined.^2,3 An important point for clinicians to consider in conducting workup for stroke etiology is that the risk of recurrent stroke increases after an initial stroke and recurrences frequently occur in the early phase. Most recurrent strokes are caused by the same etiology as the index stroke. Thus it is important to expedite workup for stroke etiology and implement secondary preventive therapies accordingly.^4,5

Inpatient vs Outpatient Evaluation

A basic evaluation with vessel and brain imaging and at least 24 hours of cardiac rhythm monitoring is recommended during inpatient evaluation for all people with ischemic stroke. Although it is common practice in the United States to obtain transthoracic echocardiography (TTE) routinely during stroke admission, data for the benefit of this practice are mixed and the American Heart Association guidelines are guarded, stating that “For prevention of recurrent stroke, the use of echocardiography is reasonable in some patients with AIS to provide additional information to guide selection of appropriate secondary stroke prevention.”⁶

Whereas studies that perform TTE routinely on all comers to the stroke unit tend to show low diagnostic yield, the role of cardiac imaging and long-term cardiac monitoring in people with stroke that is cryptogenic after initial workup is more justified, and in many instances, it is appropriate to defer this workup to the outpatient setting. Therefore, familiarity with the available modalities and their relative strengths and weaknesses is important for all neurologists who care for patients with stroke in the outpatient setting.

Cardioembolic Stroke

Cardioembolic stroke (CES) comprises 15% to 30% of ischemic stroke cases^7,8 and identification of a cardioembolic source usually dictates specific therapies to prevent recurrent stroke, most commonly oral anticoagulation. There is broad consensus in the stroke literature regarding entities that are considered to confer major risk for cardioembolic conditions; when these are found in a person with nonlacunar stroke who does not have significant vessel lesions, they are considered causative.⁹ In contrast, a number of cardiac conditions are thought to convey increased risk of stroke but have a less defined causal association. These are designated as minor risk. Detection of a minor-risk cardiac condition does not place the individual squarely in the category of CES, but may have treatment implications. Major and minor risk conditions are listed in Table 1.

Rhythm Monitoring

The most common cardiac cause of stroke by far is atrial fibrillation (AF), which is responsible for 15% to 25% of all strokes.^7,8 AF can be paroxysmal and asymptomatic and in the past decade randomized trials have proven the intuitive fact that more aggressive cardiac rhythm monitoring results in higher rates of AF detection. The pivotal trials were EMBRACE (30-Day Cardiac Event Monitor Belt for Recording Atrial Fibrillation After a Cerebral Ischemic Event), which randomized participants with CS to either 24-hour or 30-day noninvasive ambulatory electrocardiogram (ECG) monitoring, and CRYSTAL-AF (Study of Continuous Cardiac Monitoring to Assess Atrial Fibrillation After Cryptogenic Stroke), which randomized participants to implantable cardiac monitoring (ICM; specifically the Reveal device by Medtronic, Mounds View, MN) or intermittent ECG. EMBRACE detected AF in 16.1% of the intervention vs 3.2% of the control group and CRYSTAL-AF found AF at 6 months (defined as any AF episode ≥30 seconds) in 8.9% of participants in the ICM group and 1.4% of the control group. The investigators subsequently published long-term outcomes data, reporting that at 36 months AF was found in 30% of the ICM group and just 3% of the control group.⁹

The higher rate of AF detection in EMBRACE compared with CRYSTAL-AF is probably due primarily to the older age of the participants (mean 72.5 vs 61.5 years). Together, these studies provided clear and convincing evidence that monitoring for AF beyond the acute hospital stay is effective in detecting AF in a substantial minority of people with CS. Neither study, however, was powered to address the question of whether detecting asymptomatic AF and initiating anticoagulation prevents recurrent strokes in this population. In a long-term follow-up article, the CRYSTAL-AF authors reported 20 recurrent strokes in the ICM arm and 24 in the control arm, a difference that was not statistically significant.^10,11

LOOP Trial

The 2021 LOOP trial (Atrial Fibrillation Detected by Continuous ECG Monitoring) was designed to answer the essential clinical question of the impact of long-term AF monitoring on stroke recurrence, randomizing 6000 people to ICM or usual care. As in CRYSTAL-AF, LOOP detected AF (defined as an episode ≥6 minutes) in the ICM group at a significantly higher rate than in the control population (31.8% vs 12.2%). More than 90% of those with AF detected were treated with anticoagulation. However, although the rate of ischemic stroke was 20% lower in the ICM group (4.5% vs 5.4%), this difference did not reach statistical significance. Likewise, the rate of major bleeding events in the ICM group was higher, but not statistically significant.¹²

Although the LOOP trial complicates the landscape of evidence for long-term AF monitoring, it is a mistake to view it as proof that the practice is futile or misguided. First, the 12% AF rate in the control group was far higher than that seen in CRYSTAL-AF or in any previous trial, a fact that undoubtedly diluted the benefit of the monitoring program. This may reflect the increasing popularity of wearables such as the Apple Watch that have AF detection capabilities or a particularly medically engaged population in Denmark. Second, LOOP reignites the unsettled debate about what duration of AF should be considered a clinically meaningful stroke risk factor. The threshold of >6 minutes was derived from ASSERT (Asymptomatic Atrial Fibrillation and Stroke Evaluation in Pacemaker Patients and the Atrial Fibrillation Reduction Atrial Pacing Trial), which examined the relationship between detection of subclinical atrial tachycardia and subsequent stroke in people with ICDs (pacemakers or defibrillators).¹³ ASSERT reported an increase in stroke incidence in people with AF episodes >6 minutes, but a subsequent analysis showed that episodes of longer duration, especially those >24 hours, conveyed far greater risk than shorter spells.¹⁴ LOOP was likely underpowered given the lower risk of stroke in people with short-duration AF.

Summary and Recommendations. Overwhelming data demonstrate that cardiac rhythm monitoring to screen for asymptomatic AF beyond acute hospitalization increases the detection of AF. Major questions remain, however, about the threshold of AF duration that should be considered a true stroke risk factor and treated with anticoagulation. The optimal method of monitoring varies among individuals. For people with high pretest probability of AF, a 30-day monitor can be an excellent initial tool, which can be followed up by ICM if AF is not found and suspicion remains high. For people capable of high levels of adherence to monitoring protocols, strategies such as at-home twice-daily ECG monitoring may work well. For people who struggle with or reject more laborious monitoring strategies, ICM is an excellent option. Clinicians must, however, consider thoughtfully how to interpret and react to rare, short-duration AF episodes detected on ICM, with the LOOP study casting doubt on whether spells measured in minutes rather than hours warrant anticoagulation. More study is needed to settle the question of clinically meaningful duration of AF.

Advanced Cardiac Imaging

Whereas AF is the most common cause of CES, the other major and minor cardioembolic sources are structural and are diagnosed with cardiac imaging. There are a variety of imaging modalities available to clinicians, each of which has strengths and weaknesses in terms of detection of anomalies of interest, cost, and risk of complications. A summary of the diagnostic yield of each modality for specific disease entities is provided in Table 2.

Echocardiography

Transthoracic echocardiography (TTE) is traditionally the first choice for evaluation of cardiac output and detection of cardioembolic sources for acute ischemic stroke. TTE is generally very reliable for evaluation of chamber sizes, ejection fraction, and most valvular pathology; limitations include operator dependence and poor visualization with larger body habitus. TTE is noninvasive and readily available in most centers. TEE is a semi-invasive procedure that is generally safe in experienced hands. Nevertheless, it is time-consuming, carries physical discomfort, has limited availability, and is associated, albeit rarely, with potentially life-threatening complications such as esophageal rupture.

TTE is a very accurate modality for detecting left ventricular thrombus (sensitivity 95%, specificity 85%–90%); transesophageal echocardiography (TEE), which approaches the heart from the back, provides a limited image of the apex and is less sensitive for detecting left ventricular thrombus. TTE has low sensitivity for cardiac vegetations, especially for small vegetations or those on prosthetic valves. TEE, on the other hand, is very sensitive and specific for the diagnosis of infective endocarditis as well as nonbacterial thrombotic endocarditis. TTE has good sensitivity for detecting right-to-left shunt, but TEE is the gold standard and can better distinguish the type of the shunt (eg, patent foramen ovale [PFO], atrial septal defect [ASD], or pulmonary arteriovenous malformation [PAVM]).^15,16

Computed Tomography

Advantages of cardiac CT consist of short duration of study and high spatial resolution. Disadvantages include risks of radiation exposure and contrast-induced nephrotoxicity, poor assessment of the myocardium, and low sensitivity for patent foramen ovale.^{17–20,22,23} Advanced cardiac CT using delayed contrast-enhanced imaging has been shown to have a higher rate of detection of high-risk sources of embolism compared with TTE, particularly for diagnosis of left atrial thrombi, with sensitivity of 100% and specificity of 99%.²¹ CT has less resolution for detecting valvular vegetations compared with TEE but outperforms TEE for assessment of abscess and prosthetic valves. CT provides high sensitivity for detection of aortic atheroma, but cannot assess atheroma mobility.¹⁸

Magnetic Resonance Imaging

Current guidelines do not recommend cardiac MRI (CMR) as part of the common diagnostic workup for stroke, but CMR can be valuable in specific circumstances. CMR is considered the gold standard for diagnosis of cardiac tumors, cardiomyopathies, and left ventricular thrombus. CMR has good sensitivity in detecting left atrial and appendage thrombi and aortic atheroma but is inferior to cardiac CT for these indications. Both cardiac CT and MRI provide information about extracardiac chest structures with potential relevance, such as malignancy with cardiac invasion. Some of the limitations of CMR are higher cost, limited availability, contraindication to use in the presence of some cardiac devices, and the requirement that the patient be able to cooperate with breath-holding during the study.^24,25

Transcranial Doppler With Embolic Detection

Transcranial Doppler (TCD) is used for the diagnosis of right-to-left shunting, which can result from an intracardiac shunt such as PFO or ASD, or the less common PAVM. It is reported to have both high sensitivity (97%) and specificity (93%) for this indication. TCD is minimally invasive and has low cost. Some of its pitfalls are the potential for operator error and inability to differentiate between intracardiac and pulmonary shunts. There is a rough technique proposed to differentiate PFO from PAVM that consists of counting the number of cardiac cycles before detecting embolic signals (3 for PFO, 5 for PAVM), but this is not well-validated. Given that intracardiac shunts are considerably more common than PAVMs, it is reasonable to use noninvasive TCD as a screening tool, which can be followed by a confirmatory TEE if PFO closure is indicated.²⁶

PFO Closure

The mechanism by which PFO can cause stroke is paradoxical embolism, in which a venous embolus crosses from the right to the left atrium, skipping the lungs, and is pumped to the brain. PFO is common in general, but the prevalence of PFO in people with cryptogenic stroke is nearly twice that of the general population.²⁷ Current guidelines, based on the results of 6 randomized controlled trials, recommend PFO closure for a subgroup of people with PFO and cryptogenic stroke, namely those under age 60.⁶ Several of the positive PFO closure trials required high-risk PFO (large shunt or atrial septal aneurysm) and the benefit of PFO closure is likely highest in those individuals. The most significant procedural complication seen in the PFO closure trials was AF, primarily periprocedural. In one metaanalysis, the rate of persistent AF was found to be 1.8% in the pooled interventional arms compared with 0.2% in the medical therapy group. Despite this, the trials showed reduction in the rate of recurrent stroke from 4.7% with medical therapy to 2.0% with closure.²⁸ Currently there are not enough data on PFO closure in people >60 years of age and considering the higher rate of serious periprocedural complications, PFO closure is not recommended in this age group.^29,30

Summary and recommendations. The use of advanced cardiac imaging is valuable in determining stroke etiology in some people with CS. Choice of modality must be tailored to the individual, based upon the pretest probability of the various cardiac embolic sources and actionability of findings (ie, eligibility for PFO closure). Some authors have suggested combining TTE/TEE and cardiac CT to overcome the pitfalls of each modality.¹⁵ We suggest considering cardiac CT as a first-line test for people with CS in whom CES is strongly suspected. Cardiac CT provides a noninvasive alternative to TEE and has high sensitivity for most cardioembolic sources of stroke apart from PFO. For people with CS under age 60, we suggest a multimodal approach combining TCD and cardiac CT. Because it is the gold standard for diagnosis of left ventricular thrombus, CMR may be a good choice for a person with low ejection fraction or wall motion abnormality in whom detection of a thrombus would result in a switch from antiplatelet to anticoagulant therapy.

Conclusion

Diagnosis of a specific etiology of ischemic stroke is a major goal of stroke care because of the implications for secondary prevention. Approximately 25% of strokes are cryptogenic and in these individuals, thorough evaluation for cardioembolic sources is indicated and is often best accomplished by means of rapid follow-up in the outpatient setting. Long-term rhythm monitoring for AF can be done through invasive or noninvasive means and is high yield, particularly in people over age 70. Several modalities are available for advanced cardiac imaging, each with its own advantages and disadvantages. Cardiac CT has the most favorable profile overall, especially when combined with TCD in people under age 60, but individual factors must be considered.