Treatment and logistics of acute ischemic stroke : safety and outcomes

Abstract

Background: Stroke, both ischemic and hemorrhagic, accounts for over 20000 hospital admissions in Sweden yearly. The vast majority of patients (85%), suffer from ischemic stroke, an occlusion of a cerebral artery that can be treated pharmacologically with systemic intravenous thrombolysis (IVT) or mechanically with endovascular thrombectomy (EVT). Treatment with IVT carries the risk of symptomatic intracerebral hemorrhage (SICH) in 2-5% of cases, potentially leading to severe disability or death. There remain unanswered questions regarding the safety of IVT in specific subgroups: (1) patients without ischemic stroke, but who present with stroke-like symptoms, known as stroke mimics, and (2) patients suffering from stroke in the posterior cerebral circulation (PCS). EVT is an effective treatment, but only possible for patients suffering from stroke caused by a large artery occlusion (LAO stroke) and is restricted to thrombectomy capable centers with trained neurointerventionists. The Stockholm Stroke Triage System (SSTS) was implemented in 2017 to detect and route patients with suspected LAO stroke and eligible for EVT directly to the thrombectomy center.

The first half of this thesis concerns safety and outcomes after IVT in patients with stroke mimics (study I) and PCS (study II). The second half of the thesis concerns acute stroke logistics through use of the SSTS. Study III investigated if outcomes after EVT have improved after implementation of the SSTS. Study IV described patients incorrectly routed using the system and investigated if the system could be improved using statistical modeling. Studies I-II were performed using data from the Safe Implementation of Treatments in Stroke International Stroke Thrombolysis Registry (SITS-ISTR), an international database. Studies III-IV included prospectively recruited patients with suspected acute stroke transported by ambulance in the Stockholm region.

Study I included patients treated with IVT between 2003-2017 with MRI follow-up. Outcomes were parenchymal hematoma, SICH, and modified Rankin scale score and death at 3 months after treatment, with comparison between stroke mimics and ischemic stroke. Five parenchymal hemorrhages, and two SICH were identified in 429 stroke mimic patients treated with IVT. Functional symptoms, headache and seizure were the three most common mimicking conditions (>60% total). There was a higher proportion with excellent functional outcome and lower proportion of dead patients in the stroke mimic group. IVT treatment was concluded to be reasonable safe in patients with a stroke mimic diagnosis.

Study II included patients treated with IVT between 2013-2017 with available angiographic occlusion data. Outcomes were the same as in study I, with comparison between anterior and posterior circulation stroke. Adjustment of baseline differences using inverse probability treatment weighting, and a systematic review and meta-analysis were performed. Of ~5000 included patients, ~15% had PCS. Both the primary data and the metaanalysis showed fewer hemorrhagic complications in PCS. After adjustment, PCS patients had a slightly higher risk of death after treatment, with no differences in functional outcomes.

Study III included patients treated with EVT between October 2017 and October 2019 (during use of the SSTS) in the Stockholm region. Comparison was performed with historical controls treated during the two years prior. The main outcome was modified Rankin Scale scores with adjustment for baseline differences. Secondary outcomes were death, change in NIHSS 24 h post treatment, recanalization, and SICH. Time from onset to EVT was faster during SSTS by 69 minutes. Functional outcomes were better in the SSTS group with no differences in safety outcomes (hemorrhage and death).

Study IV included suspected stroke patients transported by ambulance between October 2017 and October 2018 in the Stockholm region. Three alternative triage algorithms were modelled using prehospital data and compared to the SSTS using decision curve analysis. All models included a test for hemiparesis and had similar sensitivity, specificity, and AUC. Comparison of net benefit, (correct routing of patients for EVT without increasing mistriage) showed that the SSTS was superior to the alternative models.