Plasma Glial Fibrillary Acidic Protein in the Differential Diagnosis of Intracerebral Hemorrhage

Background and Purpose— Plasma GFAP (glial fibrillary acidic protein) has recently emerged as a potential biomarker for the differentiation of acute intracerebral hemorrhage (ICH) from acute ischemic stroke (AIS). We prospectively assessed the diagnostic accuracy of GFAP in the differential diagnosis of ICH. Methods— Consecutive patients presenting to the emergency department within 6 hours from symptom onset were evaluated. All patients underwent extensive diagnostic work-up and were classified according to discharge diagnosis in AIS, ICH, subarachnoid hemorrhage, and stroke mimics. GFAP was also measured in healthy volunteers (controls). Baseline stroke severity was evaluated using National Institutes of Health Stroke Scale. Receiver operating characteristic curve analysis was used to identify the optimal cutoff point for the differentiation between subgroups. Correlation analyses of GFAP plasma concentrations with baseline National Institutes of Health Stroke Scale and onset to sampling time were performed with the nonparametric Spearman rank test and fractional polynomial regression, respectively. Results— Our study population consisted of 270 individuals (AIS: 121, ICH: 34, stroke mimics: 31, subarachnoid hemorrhage: 5, controls: 79). No differences on baseline stroke severity and onset to sampling time were detected between AIS and ICH. Higher median plasma GFAP values were documented in ICH compared with AIS, stroke mimics, and controls ( P P =0.515) and ICH ( P =0.387). In the fractional polynomial analysis, the association between GFAP concentration and onset to sampling time was best described by a J-shaped curve for AIS and an inverted U-shaped curve for ICH, with a peak at 2 hours. Conclusions— Plasma GFAP seems to be a sensitive and specific biomarker for the differentiation of ICH from both AIS and other acute neurological disorders, with the optimal diagnostic yield being present in the second hour from symptom onset.