Vessel occlusion and functional outcome after acute stroke : prediction and evaluation

Abstract

In the acute setting of ischaemic stroke, two proven treatments are available: intravenous thrombolysis (iv-tPA) with alteplase and endovascular thrombectomy. The main aim of this thesis was to investigate the associations between symptom severity in the acute setting of ischaemic stroke and 1) the presence of arterial occlusions and 2) long-term functional outcome assessed at 3 months. The first 3 studies were based on patients registered in the large Safe Implementation of Treatments in Stroke - International Stroke Thrombolysis Register (SITS-ISTR). In the final project, conceptually separated from the first 3 studies, we aimed to investigate a novel method for assessing 3-month functional outcome after acute stroke.

Study 1. We aimed at finding thresholds for baseline stroke severity as measured by National Institutes of Health Stroke Scale (NIHSS) scores that predicted long-term functional outcome and baseline arterial occlusion. We analysed 44 331 iv-tPA treated ischaemic stroke patients with available functional outcome assessed by the modified Rankin Scale (mRS) at three-months and 11 632 patients with available computed tomography/magnetic resonance angiography data at baseline. For functional independency (mRS 0-2), NIHSS scores of 12 [area under the curve (AUC) 0.775] and for baseline arterial occlusion, NIHSS scores of 11 (AUC 0.678) were optimal threshold values. NIHSS thresholds decreased with time from stroke onset to imaging, with 2–3 points, respectively, if time to imaging exceeded three-hours. We concluded that an NIHSS threshold of 9 or 10 points could be considered in the pre-hospital selection of patients for immediate transfer to centres with arterial imaging and availability of endovascular thrombectomy.

Study 2. ASTRAL and DRAGON are two recently developed scores for predicting long-term functional outcome after acute stroke in unselected acute ischaemic stroke patients and in patients treated with iv-tPA, respectively. We aimed to perform external validation of these scores. We calculated the ASTRAL and DRAGON scores in 36131 and 33716 iv-tPA treated patients, respectively, registered in the SITS-ISTR between 2003 and 2013. The proportion of patients with death or dependency at 3 months (mRS 3-6) was observed for each score point and compared with the predicted proportion according to the risk scores. Predictive performance was assessed using the AUC of the receiver operating characteristic. The ASTRAL showed an AUC of 0.790 (95% CI, 0.786–0.795) and the DRAGON an AUC of 0.774 (95% CI, 0.769–0.779). We concluded that the ASTRAL and DRAGON scores show an acceptable predictive performance and may have a role for prognostication of outcome after acute ischaemic stroke.

Study 3. We aimed to assess the predictive value of various models based on baseline NIHSS sub-items, ranging from simple to more complex models, for predicting large arterial occlusions (LAO) in anterior circulation stroke. Patients registered in the SITS-ISTR with clinically defined anterior circulation stroke, and available NIHSS and radiological arterial occlusion data were analysed. We compared 1975 patients harbouring an LAO with 2036 patients having no/distal occlusions. Using binary logistic regression, we developed models ranging from a simple 1 NIHSS-sub-item to full NIHSS-sub-items models. Sensitivities and specificities of the models for predicting LAO were examined. The model with highest predictive value included all NIHSS subitems as well as other relevant parameters for predicting LAO (AUC 0.78), yielding a sensitivity and specificity of 74% and 72% respectively. The simplest model included only deficits in arm motor-function (AUC 0.72) for predicting LAO, yielding a sensitivity and specificity of 69% and 70% respectively. Differences between the models were not large. We concluded that assessing grade of arm-dysfunction along with an established strokediagnosis model may serve a surrogate measure of LAO status, thereby assisting in triage decisions.

Study 4. Assessment of long-term functional outcome is lacking in many settings. We aimed to investigate whether automatic assessment of the mRS based on a mobile phone questionnaire may serve as an alternative to mRS assessments at clinical visits after stroke. We enrolled 62 acute stroke patients admitted to our stroke unit during March to May 2014. Forty-eight patients completed the study. During the hospital stay, patients and/or caregivers were equipped with a mobile phone application in their personal mobile phones. Three months after inclusion, the mobile phone application automatically prompted the study participants to answer an mRS questionnaire in the mobile phones. A few days later, a study personnel performed a clinical visit mRS assessment. The 2 assessments were compared using quadratic weighing κ-statistics. We found a 62.5% agreement between clinical visit and mobile mRS assessment, weighted kappa 0.89 (95% CI 0.82–0.96), and unweighted kappa 0.53 (95% CI 0.36– 0.70). Mobile phone–based automatic assessments of mRS performed well in comparison with clinical visit mRS and may serve a supplementary role to traditional assessments, especially in settings where clinical follow-up visits are scarce because of economic and time-restraining factors.