Integrating GFAP Testing With Clinical Scores for Prehospital Large Vessel Occlusion Detection in Patients With Acute Stroke (DETECT LVO)

BACKGROUND: Clinical scores indicating large vessel occlusion (LVO) in acute stroke patients could streamline triage of patients with suspected LVO to endovascular centers. GFAP (glial fibrillary acidic protein) is a promising blood biomarker for indicating intracerebral hemorrhage in acute stroke. This study evaluates whether positive LVO score results combined with a prehospital negative GFAP test (thereby excluding intracerebral hemorrhage) could improve the accuracy of LVO detection. METHODS: This retrospective diagnostic accuracy study (DETECT LVO) is based on the prospective DETECT study (2022–2024, tertiary care hospital RKH Klinikum Ludwigsburg, Germany), which evaluated the rapid intracerebral hemorrhage detection in acute stroke by measuring prehospital plasma GFAP levels on a point-of-care platform (i-STAT Alinity Abbott). For DETECT LVO, 5 established LVO scores (Rapid Arterial Occlusion Evaluation, Field Assessment Stroke Triage for Emergency Destination, 3-Item Stroke Scale, Emergency Medical Stroke Assessment, Cincinnati Prehospital Stroke Scale) were retrospectively calculated from paramedic protocols. LVOs were diagnosed with CT angiography as follows: occlusion of the internal carotid artery, middle cerebral artery, and basilar artery. Diagnostic accuracy for LVO detection was determined using the area under the curve, sensitivity, specificity, positive predictive values, and negative predictive values. RESULTS: Three hundred fifty-three patients suspected of acute stroke (ischemic stroke, n=258; intracerebral hemorrhage, n=76; stroke mimics, n=19) with a mean age of 74.6 years were included. One hundred one patients with ischemic stroke suffered from LVO (internal carotid artery=23.8%; middle cerebral artery=64.4%; and basilar artery=11.9%). Integrating GFAP to LVO scores significantly increased area under the curve (95% CI) for LVO detection (Field Assessment Stroke Triage for Emergency Destination, 0.859 [0.818–0.893] to 0.899 [0.862–0.928]; Rapid Arterial Occlusion Evaluation, 0.845 [0.802–0.880] to 0.892 [0.855–0.923]; 3-Item Stroke Scale, 0.788 [0.741–0.829] to 0.865 [0.824–0.898]; Emergency Medical Stroke Assessment, 0.840 [0.796–0.875] to 0.870 [0.830–0.910]; Cincinnati Prehospital Stroke Scale, 0.827 [0.784–0.865] to 0.862 [0.821–0.896]; P <0.001). CONCLUSIONS: Integrating LVO scores combined with GFAP measurements into the prehospital work-up of patients with acute stroke improves diagnostic accuracy for LVO prediction. In the future, this could enable direct transfers of patients with suspected LVO to endovascular centers with reduced misdiagnosis rates. Independent replication in diverse prehospital cohorts is warranted to confirm these findings.