Reproducibility of Transcranial Doppler ultrasound in the middle cerebral artery

Ischaemic stroke remains a major global cause of disability and death that is associated with an enormous social and economic burden [1]. Up to 25% of ischaemic strokes are caused by atherosclerosis of the internal carotid artery [2, 3]. Carotid atherosclerosis is a complex disease that is characterised by the deposition of luminal atheroma that may rupture, thrombose and embolise [2]. The resulting thromboembolism can lead to a stroke or transient ischaemic attack (TIA) [4].

Transcranial Doppler is a well established real-time imaging modality that evaluates cerebral blood flow velocity and detects microembolic signals in patients who suffer from cerebral or retinal ischaemia [5]. Microembolic signals in symptomatic carotid artery stenosis are associated with an increased risk of a recurrent ipsilateral focal ischaemia [6‐14] and and correlate with a greater number of magnetic resonance imaging detectable cerebral infarcts when compared with patients free from microembolism [15‐18]. The intraoperative transcranial Doppler has enabled clinicians to lower the rate of the most serious post carotid endarterectomy complication such as thromboembolic stroke from 4 to 0.2% through detection of the middle cerebral artery flow cessation due to the intraluminal carotid artery thrombosis [19, 20]. Whereas, transcranial Doppler directed infusion of Dextran 40 has in some centres successfully erased the rate of postoperative thromboembolic cerebral ischaemia from 2.7 to 0% [8, 21]. Despite these benefits from transcranial Doppler, routine use has not been advocated amongst vascular specialists.

Although multiple studies have been conducted on flow velocities in basal cerebral arteries in both healthy volunteers and patients [22‐33], reproducibility data are limited to a hand full of reports. These include four articles involving healthy subjects [34‐37] and one study that recruited patients with clinical diagnosis of ischaemic stroke (n = 3) or TIA (n = 7) but provided no information regarding the clinical type of neurovascular event or underlying carotid artery stenosis [5]. In contrast, published data on microembolic signals detection in patients with symptomatic carotid artery stenosis includes systematic reviews, meta-analyses [38‐41] international multicenter reproducibility studies that have described the reproducibility of transcranial Doppler as sufficient for clinical use [10, 42, 43].

Our objective was to assess the intra-observer repeatability and reproducibility of transcranial Doppler for velocimetry measurements and microemboli detection in healthy volunteers and patients with symptomatic carotid artery stenosis that could form the basis for our future study investigating reliable identification of a vulnerable carotid plaque.

All participants tolerated transcranial Doppler examinations well and completed all assessments.

In this study, we have demonstrated that transcranial Doppler generates reproducible data regarding the velocity measurements. Transcranial Doppler utilises an acoustic temporal bone window through which the ultrasound beam can focus on the middle cerebral artery, which receives 80% of an ipsilateral internal carotid artery inflow [27]. The obtained middle cerebral artery insonation depth in both cohorts reflects published data [10, 45, 46].

In general, the success of transcranial Doppler imaging diminishes with an older age due to an increased temporal bone thickness that impairs the transmission of ultrasound waves through the skull [47, 48]. This has been observed primarily in approximately 10% of non-Caucasian elderly female participants [49]. However, others report temporal window failure in almost third of examined subjects [50].

Multiple studies described substantially different normal reference velocity values of cerebral arteries blood flow [22, 24, 25, 27, 29, 31] but the most frequently quoted normal middle cerebral artery velocity under resting condition ranges from 35 to 90 cm/sec with a mean of 60 cm/sec [29]. Our velocity values mirror the results published by others, except for the lower mean diastolic middle cerebral artery velocity. However, this could be explained by various physiological and technical factors that can affect velocity readings. First, physiological cardiovascular changes such as heart rate, blood pressure, respiratory rate, arterial carbon dioxide tension alter middle cerebral artery blood flow on a daily basis [51, 52]. Second, psychological factors (emotional state, fatigue) by influencing the above physiological cardiovascular autonomic responses can impact on the cerebral blood flow [51]. Unsurprisingly, changes in the cerebral metabolism due to cognitive activation also affect the middle cerebral artery blood flow. Some authors demonstrated that arithmetic activity produced very similar values to the resting blood flow values, whereas higher levels of arithmetic difficulty produced smaller changes in the blood flow [53]. Therefore, the above factors could have potentially influenced the obtained velocity values.

The main technical aspect that can impact on velocity measurement is the angle of insonation that is obtained between the middle cerebral artery and ultrasound beam [5, 34]. However, this is more relevant when large acoustic window such as the foramen of magnum is used, because it permits significant angle variation [34]. Fortunately, small temporal window with a sharp angle of insonation (0°-30°) that is relatively stable minimises any influence on obtained velocities values [5]. Hence, the maximum error has been estimated to be less than 15% [34, 35]. Finally, individual variability of the middle cerebral artery size, length and tortuosity are also contributing to the scattering of the velocity measurements [27, 29, 37].

In general, cerebral flow velocity decreases with age in a bimodal pattern with a first decline above the age of 40 years and a further reduction above 60 years of age [22, 26, 27, 54]. Unsurprisingly, our data demonstrate similar results with lower velocity values in patients cohort when compared with the healthy volunteers. Overall the obtained ICC values in our study represent a good repeatability and reproducibility in both cohorts. However, the peak and mean ICC repeatability values recorded in patients group reached an excellent agreement (ICC > 0.90). In both cohorts, the peak and mean ICC reproducibility values decrease with wider confidence intervals when compared to the repeatability values. This likely reflects the combination of technical variation and biological variation which will be much greater when measurements are conducted on separate days rather than within a day. For example, this could include probe displacement from the original middle cerebral artery segment that was sampled during the first study visit. This may also reflect the well described anatomical variability of the circle of Willis including diameter discrepancy of the individual parts of the middle cerebral artery [27, 29]. In effect, an over or under-estimated velocity values can be reported depending on the diameter of an insonated artery. The slightly higher ICC values obtained in the patients group could be explained by the lower range of physiological fluctuations and more consistent velocity measurements [28].

Finally, the equipment characterists such as head frame that supports the transducers could account for some differnces in velocity values. In our study we have used a professional head frame system (Marc 600 Spencer Technologies, USA) that minimises the motion and maintains a constant angle of insonation of the middle cerebral artery. Interestingly, no single reproducibility study on velocity measurements described any form of secure fixation of transducers during the examinations [5, 34‐37]. Similarly, systematic reviews and meta-analyses on microembolic signals detection provide no information on any head-frame systems used by individual studies [38‐41]. This raises many questions regarding the methodological aspects of these studies that have been conducted more than 20 years ago.

Although our data regarding transcranial Doppler velocities measurements echoes other researchers findings, it should be interpreted with caution owing to many methodological limitations of the published analyses including a limited number of reproducibility studies that contain small sample size and variable imaging protocols. Furthermore, evidence for the transcranial Doppler criteria to predict the degree of intracranial arteries stenoses remains inconclusive and controversial [18]. Several studies failed to demonstrate reproducible data on specific cut-off points for the velocities values with the percentage of stenosis [30, 32, 33, 55‐57]. Some authors have proposed middle cerebral artery velocity of > 80 cm/sec as a criterion for stenosis [57], whereas others used velocities > 100 cm/sec when diagnosing stenosing lesions [54]. In contrast, some researchers have highlighted the importance of additional measurements such as side-to-side differences in velocities (> 30%) or increase in velocity (> 50%) along with the assessment of collateral flow using temporary manual occlusions of the common carotid artery [30, 32]. However, one must remember potential pitfalls with such approach because high velocities in collateral circulation can indicate different diameters of the middle cerebral arteries on two sides [30]. In effect, high blood flow velocities may be caused simply by the smaller diameter of MCA despite otherwise normal anatomy [33]. Finally, the largest study (The Stroke Outcomes and Neuroimaging of Intracranial Atherosclerosis (SONIA) Trial) that attempted to validate transcranial Doppler findings with magnetic resonance angiography against the standard cerebral digital subtraction angiography regarding the identification of intracranial arterial disease revealed disappointingly low results of positive predictive values for transcranial Doppler (36%) and magnetic resonance angiography (59%) [55]. In effect, the transcranial Doppler’s clinical applicability regarding the abnormal velocity values assessment remains limited.

Despite complete 1-h transcranial Doppler assessments performed in our study, the lack of microembolic signals in patients with symptomatic carotid artery stenosis was disappointing. The reported incidence varied from 12 to 100% in individual studies [40, 58, 59]. Nevertheless, considerable differences regarding criteria for microembolic signals detection, timing after stroke, duration of monitoring and antithrombotic agents used have been identified among many studies [10, 40, 58]. Consequently, the majority of published data described microembolic signals in about 30–40% of individuals with symptomatic carotid artery stenosis when transcranial Doppler was performed for 1 h [39, 40, 59, 60].

Still, there are several potential explanations for the absent embolisation. Thromboembolism is a dynamic and random process with a generally reported low frequency of microembolic signals during 1-h long examination [60, 61]. Although 1-h long transcranial Doppler evaluation time is recommended for patients with symptomatic carotid artery stenosis, longer assessments increase the chances of successful emboli detection [60]. This was demonstrated by ambulatory recordings (greater than 5 h) with portable transcranial Doppler equipment that has yield greater number of microembolic signals when compared with the traditional 60 min approach [62, 63]. However, at present, an ambulatory transcranial Doppler recording remains primarily a research tool due to lack of a robust equipment.

Another possible explanation refers to the severity of carotid artery stenosis and plaque morphology. Microembolic signals are more common in patients with the higher degree of carotid artery stenosis, which in turn is associated with specific carotid plaque features reported histologically such as ulceration, intraplaque haemorrhage and surface thrombus [11, 18, 39, 49, 59, 64, 65]. These high-risk plaque features are more likely to lead to the development of stroke because they produce larger emboli that consist of thrombi [59]. Whereas, small embolic particles comprising of fibrin and platelets aggregates that lodge in small arteriolar branches, may be lysed by endogenous protective haemostatic defences, hence clinically may represent TIA [59]. The majority of our patients had a non-surgical grade of carotid artery stenosis and presented with TIA. Therefore, these factors could be potentially responsible for no detectable microembolic signals.

The various components of microembolic signals responds differently to treatment [39]. For example, antiplatelet agents are more effective for emboli originating from the symptomatic carotid artery stenosis, and reduce the rate of microembolic signals [11, 12]. On the other hand, anticoagulants deal more effectively with microembolic signals from a cardiac source [11, 39]. The majority of participants (90%) in our study have been on an antiplatelet agent at the time of the first transcranial Doppler assessment, and this could represent another potential confounder. Finally, microembolic signals are more likely to be detected within the first week after the index event, and in patients with recent stroke rather than with TIA [66]. Again, we have performed transcranial Doppler as soon as possible, but due to various logistic factors, only two patients had transcranial Doppler within seven days from their index event.

The main limitations of this study are the small sample size, and single-centre design. However, the main purpose of the study was to demonstrate reproducibility of the transcranial Doppler and this was achieved. At present, transcranial Doppler remains underutilised in clinical practice due to lack of human expertise, time-consuming recordings with the need for a continuous visual and audible evaluation [13, 60]. Furthermore, unsolved technical and methodological limitations of transcranial Doppler regarding the velocity assessments restrict its clinical applicability. However, its use during carotid surgery has shown that the clinical use of this non-invasive, non-ionising, portable and safe technique could be extended to vascular surgery specialists as part of the routine perioperative strategy that could reduce the risk of neurovascular events even further [20].

Our findings indicate that transcranial Doppler provides reproducible data on middle cerebral artery velocities. However, these findings should be interpreted with caution for the many technical and methodological limitations that the transcranial Doppler still presents. Larger studies with the colour transcranial Doppler may enable delivery of a robust data on velocity assessments along with the quantification of intracranial stenoses.