Characterization of Clinical Phenotype to Glial Fibrillary Acidic Protein Concentrations in Alexander Disease

OBJECTIVE: To determine the concentration of glial fibrillary acidic protein (GFAP) in cerebrospinal fluid (CSF) and plasma in Alexander disease (AxD) and whether GFAP levels are predictive of disease phenotypes. METHODS: CSF and plasma were collected (longitudinally when available) from AxD participants and non-AxD controls. The concentration of GFAP was compared between groups using Wilcoxon rank sum test. The Kruskal-Wallis test was used to compare GFAP concentrations across multiple groups (clinical phenotype or common genetic variants occurring at p.Arg79, p.Arg88, p.Arg239). RESULTS: GFAP concentrations were significantly elevated at baseline in both AxD CSF (N = 44) compared to controls (N = 46, p < 0.0001) and AxD plasma (N = 96) compared to other leukodystrophy plasma controls (N = 67, p < 0.0001). CSF and plasma GFAP concentrations differed between AxD cerebral, intermediate, and bulbospinal phenotypes (Kruskal-Wallis test, CSF: p = 0.0235, plasma: p < 0.0001). The median GFAP change/year among patients sampled < 8 years of age at baseline was an increase of 91,100 [pg/mL]/year (IQR 136,050) in CSF and 2850 [pg/mL]/year (IQR 9500) in plasma compared to patients > 8 years whose medians decreased over time (-41,000 [pg/mL]/year [IQR 90,300], CSF; -843 [pg/mL]/year [IQR 3900], plasma). The fold change in GFAP from first to last sampling were significantly different before and after 8 years at baseline (Wilcoxon rank sum test, CSF: p = 0.0232, plasma: p = 0.0002). A significant association was not detected between GFAP variant and GFAP levels. INTERPRETATION: GFAP concentrations are higher in the cerebral phenotype and increase over time in young children. These data can be used to formulate biomarker qualification and context-of-use in AxD.