LAUSR

Autoimmune glial fibrillary acidic protein (GFAP) astrocytopathy was first described in humans in 2016 at the Mayo Clinic, USA, and has been confirmed independently in Europe and Asia within the last year [1–4]. This disorder manifests as a distinctive corticosteroid-responsive meningoencephalomyelitis or limited form thereof, with lymphocytic pleocytosis often accompanied by a hallmark brain linear perivascular radial gadolinium enhancement [1–4]. It can be accompanied by a neoplasm in 20% (usually teratoma) [2]. The disease spectrum is unified by the detection of GFAP-specific immunoglobulin G (GFAP-IgG) in patients’ cerebrospinal fluid (CSF) [2]. GFAP-IgG is detected with indirect immunofluorescence by the characteristic staining of mouse pia, subpia and enteric ganglia [1,2]. The molecular specificity is confirmed with cells transfected with plasmids encoding various GFAP isoforms. To date, the a isoform has had the greatest clinical utility [2]. A fatal necrotizing meningoencephalitis, with GFAP-IgG as biomarker, was described in pug dogs before the discovery of the human form [5]. In this issue, Long et al. describe the clinical presentation, serological and pathological findings in Chinese patients with autoimmune GFAP astrocytopathy [6]. The authors confirm the neurological and radiological spectrum with prominent inflammatory meningeal involvement, encephalitis and myelitis. The classical magnetic resonance imaging finding of radial perivascular enhancement was commonly encountered. A higher proportion of spinal cord lesions was noted than previously published [2]. Of critical importance, all patients in this study had GFAP-IgG detected in CSF. CSF antibody detection is recommended due to greater sensitivity and specificity [2]. Caution is advised regarding clinical utility or expansion of the autoimmune GFAP astrocytopathy spectrum based on serum positivity alone [3]. The patients in this study were less immunotherapy responsive (steroids and intravenous immunoglobulin) than prior reports and some had progression despite treatment [6]. This may reflect a faster steroid taper or a more severe disease course in this Asian population. In our experience, this disease is usually steroid responsive if treated promptly, although prolonged oral high-dose corticosteroids (for 3–6 months) followed by a slow taper and transition to a steroid-sparing agent may be necessary to avoid relapse. A novel finding in this report is IgG specificity for the early progenitor (GFAPe) isoform in some patients, where GFAPa-IgG was negative [6]. GFAPe-IgG sometimes coexisted in other published cases, but GFAPaIgG was positive in all [1–3]. Testing for GFAPe-IgG might add to diagnostic sensitivity where the characteristic murine brain immunohistochemical staining is detected, but specificity is not confirmed by GFAPaIgG testing. In our laboratory, a GFAPa-IgG negative, GFAPe-IgG positive case is yet to be found. The difference in the pattern of immunohistochemical staining noted in this study (staining of Bergman glia in the cerebellum) is consistent with our own interspecies experience (rat in this study, mouse in prior studies) [1–3]. The major addition of Long et al. to the current understanding of GFAP autoimmunity is the description of the neuropathological findings in four patients [6]. A necrotizing inflammatory process with cytotoxic T cells infiltrating the meninges has been previously described in one patient (leptomeningeal biopsy) [3]. Brain biopsies in four cases showed inflammatory infiltrates (lymphocytes, monocytes and neutrophils) surrounding blood vessels and microglial activation [6]. The infiltrating lymphocytes were both T and B cells and there were also abundant plasma cells. As predicted by GFAP’s intracellular location, this autoimmune astrocytopathy is probably mediated by antigen-specific cytotoxic T cells and the autoantibodies do not have pathogenic potential. In vivo evidence for the pathogenic potential of GFAP-specific cytotoxic T cells to cause an inflammatory central nervous system (CNS) disease come from a mouse model of GFAP autoimmunity, where GFAP-specific CD8+ cytotoxic T cells caused CNS inflammation in different forms depending on the environmental trigger (from relapsing remitting to severe progressive meningoencephalitis) [7]. In one patient in the Long et al. study [6], loss of the aquaporin-4 water channel and GFAP staining could be interpreted as astrocyte loss in the lesion as a result of cytotoxic T cell attack; neuronal loss may occur secondary to astrocyte death. Alternatively, a multifaceted immune response targeting aquaporin-4,