Transcranial doppler ultrasound for the diagnosis of large vessel occlusion in patients with acute ischemic stroke: A systematic review

Introduction: Transcranial Doppler (TCD) ultrasound may enhance the swift diagnosis of large vessel occlusion (LVO) in patients with a clinical suspicion of acute ischemic stroke (AIS). This is a comprehensive review of the literature on the use and performance of TCD ultrasound in diagnosing AIS caused by


Introduction
Acute ischemic stroke (AIS) accounts for a considerable morbidity and mortality worldwide carrying substantial health-economic and psychological impacts [1].
The current paradigm for handling patients with stroke before reaching a hospital, known as "drip and ship", concentrates on delivering thrombolytic treatment quickly by moving patients to the closest stroke center for treatment [2].Patients with large vessel occlusion (LVO) might need to be secondarily transferred to a comprehensive stroke center for mechanical thrombectomy (MT).This approach reduces waiting time for thrombolytic treatment administration but can lead to Abbreviations: AIS, Acute ischemic stroke; DV, diastolic velocity; FV, Flow velocity; LVO, Large vessels occlusion; MT, mechanical Thrombectomy; MV, Mean velocity; PI, Pulsatility Index; RSO2, regional saturation of oxygen; TCD, transcranial Doppler; SV, systolic velocity; VCI, Velocity curvature index; VAI, Velocity asymmetry index.delays in receiving MT for those with LVO-associated AIS [3,4].Alternatively, the "mothership" model involves transporting all patients directly to a comprehensive stroke center further away, possibly reducing MT wait times for suitable patients but extending overall wait times for those not eligible, and potentially putting strain on the resources of the comprehensive stroke centers [5].
In this context, discriminative field triage for patients based on LVO likelihood remains an unmet need, currently relying on clinical scales.The PRESTO study found the Rapid Assessment of Arterial Occlusion Scale (RACE) to be the most accurate prehospital tool for LVO diagnosis, with an AUC of 0.83 (95 % CI: 0.79; 0.86) [6].However, in the RACE-CAT trial, using the RACE scale, showed no significant difference in 90-day neurological outcomes between local stroke centers and thrombectomy-capable centers for suspected LVO stroke patients, partially due to the suboptimal capacity of clinical scales to accurately identify LVO [7].Mobile stroke units with embarked CT scanners can significantly reduce both thrombolysis and MT delays, but they are resource-intensive and may not be suitable for all settings, particularly in rural areas [8,9].
Transcranial Ultrasound Doppler (TCD) is an affordable, portable technology, and several Doppler parameters have been found to be effective in detecting LVO [10,11].TCD could potentially facilitate the correct identification of patients for direct transportation to centers capable of performing mechanical thrombectomy, but its application for detecting LVO has been limited [12][13][14].
Our objective is to provide a comprehensive review of the literature on the application and accuracy of TCD ultrasound in diagnosing AIS caused by anterior circulation LVO.Through a systematic review of existing literature, we aimed to give a comprehensive overview of the use and accuracy of TCD in diagnosing AIS caused by LVO.

Search strategy and study selection
Two authors (N.D.A.B and G.B) performed a literature search on PubMed and Google Scholar search engines, using the following keywords with Boolean operators: "Stroke," "Ultrasonography, Doppler, Transcranial," and "Proximal/large vessel occlusion".The search was restricted to articles in English or French published from January 1st, 2010 to January 25th, 2023.
Detailed search queries can be found in the supplemental material's Appendix.Two authors systematically assessed the search results to include articles reporting TCD use for LVO diagnosis in patients with suspected or confirmed AIS, and CT or MR as comparator.The screening process involved assessing the title, then abstract, and where appropriate full study text for relevance.Review articles and studies with unclear TCD usage, or post treating TCD waveform, for LVO or stroke diagnosis were excluded.

Data extraction and risk of bias assessment
For the included studies, two investigators (N.D.A.B and G.B) independently performed the data extraction using a predefined form including the following sections: 1) study design, 2) sample size, 3) country, 4) timing of TCD use, 5) Markers used for LVO diagnosis, 6) Sensitivity (Se), 7) Specificity (sp), 8) Negative predictive value (NPV)/ Positive predictive value (PPV), and 9) 95 confidential Interval (95 % CI) and overall accuracy.Study quality and risk of bias were assessed using the Cochrane "Tool to Assess Risk of Bias in Cohort Studies" (https://me thods.cochrane.org/).

Data analysis and report
After data curation, we recorded the type of study, available biomarkers for diagnosing anterior LVO in AIS cases, and their precision and effectiveness by presenting their sensitivity, specificity, negative predictive value, positive predictive value and accuracy where available.We subsequently scrutinized the primary pertinent biomarkers' efficiency, implications, and constraints in diagnosing and managing LVO.
The general characteristics of the included studies are summarized in Table 1.A majority of the studies were prospective (n=5, 71.4 %), and they were 2260 patients enrolled in the whole included studies.They were 49.3 % of females (n=1114) and were mean age of 65.3 +/-4.5 years.

Risk of bias assessment
All the seven studies had overall reported low risk of bias (Table 2).

TCD biomarkers used for LVO diagnosis
A total of six TCD based biomarkers were described in five studies among the seven studies included studies with a total of 2055 patients.See Table 1 for details.We describe below the most commonly reported TCD biomarkers.
PI is an objective measure of flow shape, calculated as PI = (SV-DV)/ MV, where SV is systolic velocity, DV is diastolic velocity, and MV is mean velocity over a cardiac cycle, PI normal value is 1 +/-0.2[17,18,20].
However, PI obtained by TCD is definitely operator dependent because PI calculation depend on the quality of the doppler waveform.Nevertheless, PI has also been used for recanalization scores to assess residual occlusion and patency arterial after recanalization thrombolysis and thrombectomy in AIS, with varying performances [19,23].

Discussion
In this comprehensive review we found that, in the context of AIS, various TCD based biomarkers, including pulsatility index (PI) demonstrated high accuracy for LVO identification.In the screened reports TCD was primarily employed after admission for evaluating persistent occlusion, we found no report of TCD usage for the diagnosis of LVO in the prehospital setting.
Amongst strength of TCD, it is a noninvasive, repeatable, and cost-effective screening tool, that appears to have high accuracy for detecting arterial occlusions in patients with AIS, most notably in the anterior circulation, where sensitivity is 100 % and specificity is 98.9 % with accuracy fluctuating between 94.5 % and 99.2 % comparing with CTA and MRA [17].Further, a model proposed by Daria Antipova and colleagues, which merges transcranial ultrasound with mean flow velocity (MFV) assessment and clinical examination (dubbed "TUCA"), shows superior potential for rural and remote communities in recognizing patients who may benefit from intravenous tissue plasminogen activator (IV tPA) and/or mechanical thrombectomy following AIS due to LVO.On the downside, TCD has limitations.For example, it does not offer any insights into the pathological characteristics of the intracranial lesions [16].Furthermore, several studies, including those by Gujjar et [19,20,22].
These challenges might be optimized with posttreatment waveform  Footnotes: PI (pulsatility Index), *(PAs were defined as direct admissions to the Reference Comprehensive Stroke Center RCSC, Emergency Department.The neurologist on call immediately performed a carotid ultrasound and a TCD examination to confirm the presence of an arterial occlusion), PSV (Peak systolic velocity), MV (mean velocity), DV (diastolic velocity), MFV (mean flow velocity), FV (Flow Velocity),), CTA (Computed tomograophy angiography), MRA (magnetique resonance angiography) DSA (digital substractive angiography), UK (United Kingdom), USA (United states of America).
from TCD which required the analysis of post treatment biomarkers such as velocity curvature index (VCI) with 88 % of accuracy or velocity index asymmetry (VAI) with 79 % of accuracy to diagnose LVO [24,25].VCI is biomarkers from posttreatment of TCD waveform which emerged as a potent diagnostic instrument being able to capture information such as minimal, blunted, and dampened flows, [24] especially relevant for LVO diagnosis in patients with AIS.VCI presents opportunities for fine-tuning by optimizing future diagnostic thresholds for different LVO levels, being easily calibrated with incoming data (Fig. 2).VCI is calculated by summing local curvature for each waveform's beat "canopy", which in turn illustrating the SV and DV which includes points where velocity exceeds a quarter of the total diastolic-systolic range.The specification of the quarter (0.25) is a free parameter in the metric computation to obtain the VCI value [14,24,25].VAI quantifies the difference between average cerebral blood flow velocity (CBFV) in a vessel within a cerebral hemisphere compared to the corresponding vessel in the opposite hemisphere (Fig. 3).[25] VAI can be computed using left/right average beat waveforms comparisons, with VAI as the minimum average velocity across hemispheres divided by the corresponding maximum [25].Researchers have proposed combining these markers (VCI and VAI) into a composite biomarker to improve the sensitivity for LVO diagnosis.[25] In their previous review exploring the progression of morphological analysis of TCD waveforms for the indication, localization, and monitoring of acute LVO, Dorn AY and colleagues discovered that TCD waveform morphology has been extensively investigated and validated for various applications in acute stroke management [26].They noted that this morphological analysis has significant clinical relevance for stroke triage, management, and prognosis prediction.However, they emphasized that there remains a need for TCD waveform analysis methods that can be utilized even when specialists aren't present.[26] This is the only additional review we found in the literature to date.Furthermore, it seems unclear the time which take to complete the TCD for itself, only one study in our review compared the delay from admission and TCD (15,5 +/-10,5 min) vs. Computed Tomography Angiography (CTA) (95,4 +/-75,2 min) [17].Among the reviewed reports, various authors also highlighted interesting observations regarding shifts in clinical management resulting from the employment of TCD in AIS settings.For example, Brunser et al. demonstrated that TCD could offer supplementary details such as collateral flow assessment (35/66), active micro embolism detection (10/66), identification of previously unseen occlusions by CTA (4/66), and information related to vessel patency (22/66).This allowed the authors to propose modifications in management for 17.4 % (95 CI. 9.4-25.5) of their AIS patients evaluated with TCD [19].Saqqur et al. showed in their study that TCD was capable of demonstrating enduring arterial occlusion in 38 % of a series of 349 AIS cases after tPA administration [23].Furthermore, Gomez et al. indicated that in 40 % of cases, TCD contributed extra clinical data to assist in the management of acute stroke patients, even after prior    CT-Angiography/Digital subtractive Angiography (CTA/DSA), and suggested an alternate treatment course in 22.7 % of cases [18].However, it's important to note that these studies were conducted in varied clinical contexts, which may have influenced the outcomes and observations reported.Future research could aim to standardize the contexts to offer more comparable.The validity and application of a systematic review might be impacted by the quality of each study included in our review.Nine studies had overall reported low risk and two had unclear risk of bias, because one of the studies was a case report and the second one was a retrospective cohort.
Despite our comprehensive literature search, the data at hand does not provide high-level evidence.This scarcity of available literature, however, underscores the current underutilization of TCD in diagnosing LVO in AIS cases.One of the primary shortcomings of TCD was the subpar temporal windows for data acquisition, which ranged between 13 % and 19.7 %.Additionally, our review was restricted to articles published in English from 2010 up to January 2023.Consequently, works published in other languages, or those released before the included timeframe, were not considered in our analysis.Further, due to the limited quality of available evidence, we opted not to proceed with a meta-analytic analysis, and remain descriptive in this review.
Our review has several strong points.First, it focuses on relevant studies from a reliable database regarding TCD's application for diagnosing LVO in the context of AIS.Second, it essentially scrutinizes performance parameters with the aim of assessing the accuracy of TCD in diagnosing LVO in the anterior circulation region pertinent to MT, from 2010 to the date of our review.

Conclusion
TCD is as a promising tool for LVO diagnosis in AIS cases, presenting opportunities to enhance early stroke management.PI demonstrate high accuracy in detecting LVO and should help to make the diagnosis on the field.This might help in the orientation of the patients toward a primary stroke center for thrombolysis or a comprehensive stroke center for thrombectomy.Despite operational challenges such as suboptimal temporal windows for data acquisition, the noninvasive, cost-effective, and repeatable nature of TCD potentially outweighs these limitations, positioning it as an effective option for widespread application in LVO diagnosis, with direct relevance in the prehospital diagnosis of LVO.Our review highlights the necessity of further relevant studies to validate TCD as a triage tool on field as well in hospital.

Criteria Used: 1 )
Were the selection criteria clearly defined?2) Similar point in the course of their disease?3) Was the study population representative of the population of interest?4) Was completeness of follow-up described?5) Was completeness of follow-up adequate?6) Was the outcome of interest clearly defined?7) Was the outcome determined appropriately?8) Study free of other aspects that have potential risk of bias?Rating: Yes (Y); No (N); Unclear (U).

Fig. 2 .
Fig. 2. "flow" versus time illustrating Velocity curve Index obtaining with SV and DV flow representation.The 2 circles represent Systolic Velocity (SV) and Diastolic Velocity (DV).

Fig. 3 .
Fig. 3. : VAI obtaining for LVO in Left hemispheric using the VCI in right hemispheric versus VCI in Left.

Table 1
General Characteristics of the included studies.

Table 2
Systematic appraisal web tables of study quality and risk of bias.