Direct Transfer to Angio-Suite Versus Computed Tomography–Transit in Patients Receiving Mechanical Thrombectomy: A Randomized Trial

The authors declare that all supporting data are available within the article and in the Data Supplement.

We performed a prospective, single-center, parallel-group, open label investigator initiated randomized controlled trial. In this trial, patients with a suspected acute ischemic stroke were randomized 1:1 to receive stroke imaging following either the CT transit pathway or the DTAS pathway, standardized according to institutional treatment protocols.

The trial was performed using written informed consent by the patient or the patient’s legal representative or, if the patient was not capable to consent and a legal representative was not known or available, preliminary inclusion was made by an independent physician who was not involved in the treatment of the patient. Patients were allocated to the randomization group after meeting trial entry criteria (Figure I in the Data Supplement). If not done previously, patients or the patient’s legal representative gave written informed consent for data collection, storage, and evaluation for the purpose of research within the first 72 hours of after endovascular stroke treatment to be included in the final analysis.

The trial was conducted at a university based comprehensive stroke center in Germany and the trial protocol was approved by the local ethics committee (Ethikkommission der Medizinischen Fakultät Heidelberg, research No. S-301/2017).

From December 2017 to February 2019, n=625 patients were admitted to the emergency room with acute stroke symptoms during screening hours, that is, while a neurointerventionalist was present in the hospital. Following clinical examination by a neurologist, n=430 (68.8%) patients were excluded because they did not fulfill the clinical inclusion criteria or they presented as a wake-up stroke and received MRI imaging. The remaining n=195 (31.2%) patients were judged to be eligible to randomization. Of these, n=63 (32.2%) were excluded because the angiosuite was not available for imaging because of concurrent ongoing treatment or severe agitation of the patient was anticipated rendering acquisition of flat-panel CT not feasible, and imaging using conventional CT was performed. The CTT pathway was proceeded in n=86 (44.1%), and n=46 (23.6%) patients received imaging using flat-panel CT in the DTAS pathway. The distribution of imaging findings, that is, patients with and without a large vessel occlusion, or with intracranial hemorrhage are shown in Figure I in the Data Supplement.

After n=60 patients who received mechanical thrombectomy provided written consent for final trial participation, a preplanned interim analysis was conducted. In this analysis n=34/60 (56.7%) patients were included who were treated according to the CT transit pathway and n=26/60 (43.3%) according to the DTAS pathway. Baseline demographics and clinical characteristics are shown in Table 2.

The mean (SD) age was 75 (13) years, balanced between groups. The baseline median (interquartile range) National Institutes of Health Stroke Scale score was 18 (11–20) with a trend to lower National Institutes of Health Stroke Scale scores in the DTAS group (CTT: 18 [13–20], DTAS: 16 [8–19], P value: 0.087). Following intubation before hospital admission, mechanical thrombectomy under general anesthesia was conducted in n=18 (52.9%) patients in the CT transit group and in n=8 (30.8%) in the DTAS group (P value: 0.227). No patient was intubated between hospital admission and the end of the endovascular stroke treatment.

In this randomized, single-center trial with 2 arms (CTT and DTAS) in ischemic stroke patients with large intracranial vessel occlusions requiring mechanical thrombectomy, we aimed to prospectively select a treatment pathway for patients receiving stroke imaging, which would decrease time between stroke imaging and groin puncture. The primary objective was to test the hypothesis that a direct transfer to the angiosuite would decrease time to endovascular treatment, compared with patients transferred to a CT scanner suite for conventional stroke imaging and a subsequent transfer in the angiosuite. The trial had to be terminated early after recruitment of 54.5% of the initially calculated sample size because of a delay of stroke imaging in the DTAS pathway.

The primary objective of this trial was met: there was a reduction in time from stroke imaging to groin puncture in patients randomized to DTAS compared with patients in the CTT pathway. This reduction in time was balanced out by a longer admission to imaging time in the DTAS arm and therefore could not translate into a shorter time to groin puncture or final angiographic reperfusion in this trial.

The later acquisition of stroke imaging in the DTAS arm is a key finding of the current trial. Possible reasons are (1) that transfer and positioning of a patient including attached monitoring devices is easier, more established, and thus faster in a CT scanner suite than in an angiosuite regarding the available space on the patient table or (2) that although the DTAS pathway was implemented and regularly exercised within our hospital before enrollment of the first patient, the CTT pathway was also throughout the trial the default pathway for all patients who did not fulfill in-/exclusion criteria and might thereby potentially be more familiar and thus faster for the personnel, or (3) that the mode of randomization might have introduced a selection bias and should be adapted to more practical means, for example, weekly clustering or block randomization.

Given the necessity to start treatment as fast and safe as possible and the delay in stroke imaging, and the restricted conspicuity of early signs of ischemia in patients randomized to the DTAS arm, the steering committee judged it unethical to randomize further patients into the trial without the possibility of showing a significant early intracranial revascularization.

Another finding of this trial was that successful recanalization was achieved more often in patients randomized to the CTT pathway. The imaging modality itself is unlikely to affect the reperfusion result. Therefore, this finding should not be exaggerated. However, imaging modality is a key part in patient selection. As early ischemic changes could not be adequately assessed in some patients randomized to the DTAS pathway, other restrictions concerning the assessment of intracranial arteries and perfusion imaging might apply as well.

How do the results of this trial fit into the landscape of previous publications suggesting a time and outcome benefit for patients treated according to the DTAS pathway?^8–10

First, previous studies were retrospective observational^8,10 and a case-control study with a historic patient population.⁹ As the case-control study’s primary patient cohort was treated between February 2016 and August 2017 and controls were recruited from a patient population reaching back to 2013, it is possible that changes in stroke care following the publication of the large randomized stroke trials might have influenced the results of the different study populations. There are several publications showing that implementation of new standard operating procedures streamlining stroke triage, communication, imaging and preprocedural patient preparation reduce the time between hospital admission and stroke imaging or groin puncture.^13–16 As acute stroke care has changed rapidly following the first publications of endovascular stroke trials, centers were urged to speed up the treatment process. Accordingly, previously reported significant reduction in hospital workflow times and better functional outcome might be multifactorial and have to be evaluated cautiously. Since all patients within the current trial were recruited over the same time period, potential confounding effects affected both pathways and the observed time difference between DTAS und CTT arm was exclusively caused by the randomized treatment pathway.

Second, in contrast to previous publications, presence of the neurointerventionalist on site was a prerequisite for randomization in the current trial. Since acquisition of FPCT is usually performed by angiosuite personnel, this measure was taken to avoid any delay in imaging acquisition and decision-making, thereby ensuring comparable conditions for randomized patients. It is not clear, if the observed longer time between imaging and groin puncture in CTT patients in previous studies was triggered because patients received conventional CT imaging as image acquisition in the angiosuite by angiosuite personnel and groin puncture was not possible in a timely manner (eg, out of hours). On the other hand, our data confirm that DTAS shortens time from imaging to groin puncture.

Third, the majority of patients who were transferred directly to the angiosuite reported by Ribo et al⁸ (n=31/40 [77.5%]) and Mendez et al⁹ (n=52/79 [65.8%]) received stroke imaging in a remote hospital previously and in the latter publication did not have in-house imaging. This means that the potential time savings reported in these previous studies might as well be caused by the circumstance that in-house imaging was only confirmatory or simply omitted and not because the patient was transferred directly to the angiosuite.

It is noteworthy that in previous studies some patients who received in-house stroke imaging according to the DTAS pathway (n=9/40 [22.5%]) only received a noncontrast enhanced FPCT but not a CT-angiography or perfusion imaging.⁸ Thus, n=7/40 (17.5%) did not show a treatable occlusion on the initial angiogram, which was significantly higher compared with patients receiving advanced stroke imaging. Mendez et al⁹ reported n=7/97 (7.2%) patients who did not show a treatable occlusion on the initial angiogram. This means that these patients underwent an invasive angiography to rule out an intracranial large vessel occlusion.

This trial is subject to bias as (1) we did not reach the full sample size, (2) followed a daily clustered randomization scheme, (3) performed a per protocol analysis, (4) excluded patients requiring general anesthesia and intubation between imaging and groin puncture, and (5) restricted the enrollment times to weekdays as the presence of the neurointerventionalist was a prerequisite for the randomization process.

The main limitation of this trial is the sample size, which was expected to be n=110 patients. However, since this trial was stopped prematurely, only n=60 patients were randomized (n=34 in CTT and n=26 in DTAS), making the trial potentially underpowered. The clustered randomization according to the date of stroke imaging might have introduced some bias as the observation period comprises 455 days (226 even numbered days and 229 uneven numbered days). However, as there was a higher number of holidays on uneven numbered working days (9 versus 4) during the observation period, the influence should be minimal. Perhaps, randomization on a daily basis introduced a bias, and weekly clusters might provide more continuity for the involved personnel. Excluding patients requiring general anesthesia and intubation between imaging and groin puncture, and restricting enrollment to times when the neurointerventionalist was present (ie, weekdays) limits the generalizability of our results. However, the latter reasons and the performed per protocol analysis allow us to provide results from an environment with seemingly optimal conditions for rapid imaging and endovascular stroke treatment in patients with ischemic stroke. Since DTAS did not show a meaningful time benefit even under these conditions, it seems unlikely that there would have been a time advantage for DTAS if inclusion of patients who required intubation after the imaging and before groin puncture or through treatments for which the neurointerventionalist needed to come to the hospital for imaging and treatment. Therefore, our observations contribute to the fact that data from previous studies may have to be assessed more critically.

In general, data reporting workflow improvements are influenced by local conditions. Therefore, the generalizability of our result is limited to similar conditions only. In hospitals with largely differencing distances or varying floors between arrival area of the emergency medical services, the emergency department, the CT scanner suite, and the angiosuite, our results might not be applicable. Furthermore, with the development and improvement of new imaging technologies, diagnostic reliability of FPCT will further improve in the future. Currently, FPCT-angiography is restricted to the cervical and intracranial vasculature, which poses the risk to delay the procedure in patients with unexpected anatomy of the aortic arch and the internal carotid arteries because of anatomic variants or severe tortuosity. Once image acquisition, post-processing, and image quality of FPCT is more or less equivalent to a conventional CT scanner in terms of speed, susceptibility to artifact, ease of use and availability, the DTAS pathway might be favorable over the CTT pathway. Therefore, further evaluation of the CTT and DTAS pathway in future trials is warranted. Based on our observations, in cases where the treating neurointerventionalist is not within the hospital and a flat-panel CT cannot be performed otherwise in a timely manner, it appears reasonable to conduct stroke imaging using a conventional CT-scanner to ensure imaging, decision-making, and administration of intravenous thrombolytics in eligible cases as fast as possible.

In conclusion, although the primary end point of the trial was met, that is, there was a significant reduction in time from stroke imaging to groin puncture in patients randomized to DTAS as opposed to the CTT pathway, the trial was stopped early with 54.5% of sample size randomized due to a delayed stroke imaging in the DTAS group. This preliminary trial could not show that DTAS or the CTT pathway is superior or inferior to one another, but that additional workflow improvements have to be implemented and evaluated cautiously and regularly with a special attention to local conditions.