Monitoring embolism in real time.

We have known since the 17th century that emboli can cause stroke but, despite their occasional visualisation in the retinal circulation, a diagnosis of embolic stroke is usually one of “guilt by association,” which is made by the detection of an appropriate embolic source in a patient with stroke. Frequently, .1 potential embolic source exists; determining which is clinically relevant may be impossible. The indirect diagnosis of embolic stroke also results in management difficulties in stroke prevention. For example, in atrial fibrillation, treatment failure can only be determined by the onset of stroke or systemic embolization. Recently, studies have evaluated a technique that allows the direct visualisation of circulating emboli. This offers exciting potential applications in both the diagnosis and management of patients at risk of cerebral and systemic embolism. Since the 1960s, we have known that gaseous emboli can be detected in blood using Doppler ultrasound. The large acoustic impedance difference between air and blood results in a scattering of ultrasound at the blood-air interface and a marked increase in received ultrasound intensity as the bubble passes. This results in a brief, high-intensity signal. This technique was applied to develop safe decompression limits in divers and to investigate air embolism during cardiopulmonary bypass. In 1990, while recording for air emboli during carotid endarterectomy, similar signals, but of lower intensity, were noted during manipulation of the carotid bifurcation.1 This was before arterial opening and, therefore, these signals could not represent air emboli; it was suggested that they represented thrombus and platelet emboli. Despite initial scepticism that these signals represented anything other than artifacts or flow turbulence, it has been clearly demonstrated experimentally that thrombus, platelet, and atheroma emboli result in these characteristic Doppler signals.2 The technique is highly sensitive and specific, and similar signals cannot be produced by flow turbulence or artifacts.2 The lower limit of detection remains unknown; the smallest thrombus and atheroma emboli that could be introduced experimentally were 200 to 400 mm in diameter.2 Confirmation was obtained that such signals represented emboli in a patient in whom a shower of embolic signals (ES) was noted in the middle cerebral artery coincident with the onset of contralateral hemiparesis and ipsilateral retinal emboli.3 ES have been reported in patients with a wide variety of potential embolic sources, including carotid stenosis, atrial fibrillation, and cardiac valvular disease, but they have not been found, or are rare, in age-matched controls.4 They are also common during interventional procedures, including cerebral and coronary angiography, carotid angioplasty, carotid endarterectomy, and cardiopulmonary bypass. ES appear as short-duration, unidirectional, high-intensity signals within the flow spectrum on the fast Fourier transform spectral display. The intensity increase is frequency-focused (ie, maximal over a narrow frequency range), and signals are accompanied by a characteristic chirping sound. In contrast, the intensity increase due to an artifact is bidirectional and maximal at low frequencies. This difference usually allows easy differentiation of ES from artifacts,5 except on rare occasions in which a predominantly unidirectional intensity increase occurs that is maximal at low frequency; the use of a multigate transducer allows unambiguous differentiation.5 Interobserver reproducibility studies have demonstrated excellent agreement between experienced observers in identifying ES above a certain intensity threshold, but less good agreement for very low intensity signals.6 This has led to the suggestion that an intensity threshold should be used as an additional criteria for ES detection.5 Technical aspects of both signal acquisition and processing can influence ES detection; recommended international standards have recently been published.5 This technique offers a unique window to study the pathogenesis of embolism in humans. One striking feature is the high frequency of asymptomatic ES in patients with potential embolic sources. During a single hour of recording, ES can be detected in 30% to 40% of patients with recently symptomatic carotid stenosis and in 4% to 20% of patients with asymptomatic stenosis.4 On reflection, this is perhaps less surprising in view of the frequency of finding small, clinically asymptomatic emboli at postmortem in the carotid arterial territory of patients who die of stroke.7 Embolism is, therefore, a dynamic process with a number of factors determining the clinical consequences of a single embolus. These include the characteristics of the embolus itself, the frequency of other emboli in the same vascular bed, and the nature of the recipient vascular bed, including collateral supply and perfusion pressure. A recent published debate presented divergent views on the clinical utility of this technique.8 Part of this confusion arises from the failure to distinguish between the clinical relevance of ES in different pathological conditions (in particular, where ES are caused predominantly by gaseous bubbles, as in patients with prosthetic heart valves, and conditions where they result from solid emboli, such as carotid stenosis). The The opinions expressed in this editorial are not necessarily those of the editors or of the American Heart Association. From the Department of Clinical Neuroscience, St George’s Hospital Medical School, London, UK. Correspondence to Hugh Markus, Clinical Neuroscience, St George’s Hospital Medical School, Cranmer Terrace, London, UK. SW17 ORE. E-mail [email protected] (Circulation. 2000;102:826-828.) © 2000 American Heart Association, Inc.