Clinical and neuroimaging phenotypes of autoimmune glial fibrillary acidic protein astrocytopathy: A systematic review and meta‐analysis

Abstract

Objective

This study was undertaken to provide a comprehensive review of neuroimaging characteristics and corresponding clinical phenotypes of autoimmune glial fibrillary acidic protein astrocytopathy (GFAP‐A), a rare but severe neuroinflammatory disorder, to facilitate early diagnosis and appropriate treatment.

Methods

A PRISMA (Preferred Reporting Items for Systematic Reviews and Meta‐Analysis)‐conforming systematic review and meta‐analysis was performed on all available data from January 2016 to June 2023. Clinical and neuroimaging phenotypes were extracted for both adult and paediatric forms.

Results

A total of 93 studies with 681 cases (55% males; median age = 46, range = 1–103 years) were included. Of these, 13 studies with a total of 535 cases were eligible for the meta‐analysis. Clinically, GFAP‐A was often preceded by a viral prodromal state (45% of cases) and manifested as meningitis, encephalitis, and/or myelitis. The most common symptoms were headache, fever, and movement disturbances. Coexisting autoantibodies (45%) and neoplasms (18%) were relatively frequent. Corticosteroid treatment resulted in partial/complete remission in a majority of cases (83%). Neuroimaging often revealed T2/fluid‐attenuated inversion recovery (FLAIR) hyperintensities (74%) as well as perivascular (45%) and/or leptomeningeal (30%) enhancement. Spinal cord abnormalities were also frequent (49%), most commonly manifesting as longitudinally extensive myelitis. There were 88 paediatric cases; they had less prominent neuroimaging findings with lower frequencies of both T2/FLAIR hyperintensities (38%) and contrast enhancement (19%).

Conclusions

This systematic review and meta‐analysis provide high‐level evidence for clinical and imaging phenotypes of GFAP‐A, which will benefit the identification and clinical workup of suspected cases. Differential diagnostic cues to distinguish GFAP‐A from common clinical and imaging mimics are provided as well as suitable magnetic resonance imaging protocol recommendations.

INTRODUCTION

In 2016, the discovery of a unique astrocyte‐specific IgG autoantibody present in both cerebrospinal fluid (CSF) and serum was reported in patients suffering from severe, corticosteroid‐responsive meningoencephalomyelitis [1, 2]. The antigen, glial fibrillary acidic protein (GFAP), an intermediate filament protein abundantly expressed in the cytoskeleton of mature astrocytes, led to the definition of this disorder as autoimmune GFAP astrocytopathy (GFAP‐A). To date, a detailed understanding of causative mechanisms and background factors for GFAP‐A is lacking.

Typically, the onset of GFAP‐A is acute or subacute, with prodromal symptoms of headache, fever, or other viral states. Over days and weeks, symptomatology may evolve to encompass movement disturbances, visual impairment, psychiatric manifestations, and autonomic dysfunction, among other symptoms of meningoencephalomyelitis. The clinical picture may be further complicated by coexisting neuronal autoantibodies and concurrent malignancies. However, GFAP‐A generally exhibits a positive response to immunotherapies, particularly high‐dose corticosteroids administered intravenously, even though a propensity for relapses and, in some cases, fatal outcomes have been reported [2, 3, 4, 5].

Magnetic resonance imaging (MRI) plays a pivotal role in the diagnosis of GFAP‐A. Pathological findings in the brain and/or spinal cord are present in a significant proportion of patients. A key diagnostic feature is perivascular contrast enhancement, recognized as an imaging hallmark of autoimmune GFAP‐A [1, 4, 6]. Additionally, manifestations such as longitudinally extensive myelitis and optic neuritis have been documented [7, 8, 9]. The diagnosis requires diligence, because many of its imaging findings overlap with other autoantibody‐associated diseases, and also various types of infectious meningitis/meningoencephalitis.

With the increasing number of reported cases in the literature, there is a need for high‐level evidence addressing imaging phenotypes of GFAP‐A. A systematic review and meta‐analysis of the neuroimaging characteristics and corresponding clinical phenotypes of GFAP‐A was, therefore, performed. The aim is to guide in distinguishing common differential diagnoses to facilitate early diagnosis and appropriate treatment. To further support this goal, recommendations for neuroimaging practices in the context of GFAP‐A are also provided.

METHODS

Protocol registration

The study protocol was registered in PROSPERO (International Prospective Register of Systematic Reviews; CRD42023392595, https://www.crd.york.ac.uk/PROSPERO/) and adhered to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta‐Analysis) guidelines [10].

Search strategy

A search for original studies published in full from January 2016 (the year of the first publications of GFAP‐A) up to June 2023 was performed in PubMed, Embase, and Web of Science. To maximize the sensitivity, a broad search string only encompassing the term “astrocytopathy” was employed.

Inclusion and exclusion criteria

Included were original studies reporting on one or more patients with GFAP‐A with documented GFAP‐IgG positivity, for which neuroimaging features were reported. This included case reports/series, cohort studies, randomized controlled trials, and case–control studies. Excluded were studies with only animal data, conference abstracts, non‐English articles, and studies that reiterated previously reported quantitative data. Reviews were excluded but retained as potential sources of additional records.

Study selection and data extraction

Titles and abstracts were screened for their relevance in Rayyan [11] by two independent reviewers (C.H. and B.V.I.), followed by full‐text screening. From eligible articles, the following data were extracted: title, authors, publication year, study design, number of subjects per group, mean/median age and sex of participants, age group (adults vs. children, i.e., <18 years old), study country, antibody status and detection method, clinical signs, MRI findings (brain, spinal cord, and optic nerves), advanced neuroimaging findings (e.g., positron emission tomography [PET]), and response to corticosteroid therapy.

Data synthesis and analysis

Findings were summarized narratively. In addition, for demographic parameters, neuroimaging findings, clinical phenotypes, and response to corticosteroid therapy, data were pooled to obtain summary measures. Category classes were predefined. Subgroup analyses were performed for neuroimaging phenotypes of paediatric patients.

For the meta‐analysis, studies describing clinical or imaging findings for ≥10 adult subjects each and reported by at least three individual studies were included, and only summary‐level data were used. Furthermore, a sensitivity analysis was performed on a subgroup level, consisting of only the CSF GFAP‐IgG‐positive patients, excluding patients presenting with seropositivity only. As the primary outcome, log‐transformed proportions were used. A random‐effects model was fitted to the data. The amount of heterogeneity, that is, τ ², was estimated using the DerSimonian–Laird estimator. The Q‐test for heterogeneity and the I ² statistic were calculated. A two‐tailed p‐value of <0.05 was considered statistically significant.

Publication bias

Publication bias was not assessed, as defined per protocol. However, the risk of bias assessment was qualitatively assessed for the included studies using an adjusted version of the Newcastle–Ottawa scale.

RESULTS

Eligible studies

In total, 641 studies were retrieved from our comprehensive database search. After deduplication, 499 references remained for title and abstract screening, of which 143 studies were eligible for full‐text search. After screening the full text of these studies, a total of 93 studies were included for the qualitative synthesis and 13 studies were eligible for a meta‐analysis (each comprising ≥10 patients). The flow chart for study selection is presented in Figure S1.

Study characteristics and demographics

The 93 included studies comprised a total of 681 patients (375 males, 55%; 306 females, 45%). The 13 studies eligible for the meta‐analysis comprised 535 patients (290 males, 54%; 245 females, 46%). Ages in all 93 included studies ranged from 1 to 103 years, with a median age of 46 years. Fifteen studies reported on a total of 88 paediatric cases (onset age between 1 and 17 years). Further details on the demography of the reported cases can be found in Table 1.

TABLE 1.

Demographics and clinical, laboratory, and neuroimaging findings of reported autoimmune GFAP astrocytopathy.

Demography and clinical features
Age, years	median = 46 (mean = 43), range = 1–103
Sex	Males 55% (375), females 45% (306)
GFAP‐IgG found in CSF	87% (593)
Simultaneous neuroautoantibodies	Overall found in 179/681 patients, 28%
NMDAR‐IgG	28% (50/179)
AQP4‐IgG	16% (29/179)
MOG‐IgG	8% (15/179)
Concomitant malignancy	14% (98/681)
Response to immunotherapy	88% (302/342)
Clinical phenotype
Meningoencephalomyelitis	31% (156/492)
Meningoencephalitis	23% (119/492)
Encephalitis	13% (60/492)
Encephalomyelitis	11% (58/492)
Myelitis	5% (26/492)
Meningitis	4% (20/492)
Specific symptoms
Fever	61% (243/398)
Movement disturbances	59% (294/496)
Headache	52% (254/492)
Decreased consciousness	38% (186/496)
Dysautonomia	38% (189/496)
Cognitive impairment	34% (168/496)
Visual symptoms	28% (142/611)
Psychiatric symptoms/psychosis	23% (115/496)
Nuchal rigidity and/or Kernig sign	23% (116/496)
Seizures	16% (80/496)
Respiratory failure/coma	12% (61/495)
Area postrema‐related symptoms (i.e., nausea, vomiting, hiccups)	11% (55/496)
Hyponatremia	7% (37/496, highlighted mainly in case reports)
High CSF opening pressure	3% (15/496)
Brain MRI findings
T2/FLAIR hyperintensities
Deep white matter/periventricular	45% (232/510)
Subcortical grey matter	37% (146/398)
Cortical/juxtacortical	21% (51/312)
Unspecified	11% (61/543)
Contrast enhancement
Perivascular linear	37% (146/399)
Leptomeningeal	33% (90/273)
Other, punctate/patchy	19% (74//399)
Involvement of brainstem	34% (119/353)
Involvement of cerebellum	19% (74/400)
Involvement of corpus callosum	5% (29/543)
Restricted diffusion	3% (15/543)
Imaging without specific findings	12% (63/543)
Spinal cord MRI findings
Longitudinally extensive myelitis	29% (118/403)
Short myelitis	10% (39/403)
Contrast enhancement	26% (89/336)
Leptomeningeal or central canal	16% (54/336)
Other, punctate/patchy	15% (50/336)
Cervical location	71% (70/98)
Thoracic location	65% (64/98)
Lumbar/conus/caudal location	23% (23/98)
Imaging without specific findings	21% (83/403)

Note: Percentages represent the proportion of patients with a specific feature with the parentheses conveying the number of patients with the feature and the number of patients where the data were available.

Abbreviations: AQP4, aquaporin‐4; CSF, cerebrospinal fluid; FLAIR, fluid‐attenuated inversion recovery; GFAP, autoimmune glial fibrillary acidic protein; Ig, immunoglobulin G; MOG, myelin oligodendrocyte glycoprotein; MRI, magnetic resonance imaging; NDMAR, N‐methyl‐d‐aspartate receptor.

Risk of bias assessment

Most included studies showed a low risk of bias for the selection domain. The selection domain was defined as whether patients presented with a clinical status of meningoencephalomyelitis (or some variant thereof), with GFAP‐IgG detected.

Clinical features of GFAP‐A

Clinical phenotype

GFAP‐A was associated with an acute or subacute onset of meningitis, encephalitis, and/or myelitis. Clinical phenotypes were reported for 492 of 681 patients. Among these 492 patients, meningoencephalomyelitis was most commonly observed (32%), followed by meningoencephalitis (24%), encephalitis (12%), encephalomyelitis (12%), myelitis (5%), and meningitis (4%). A comprehensive summary of the reported symptoms can be found in Table 1.

The predominantly reported clinical symptoms were fever (61% of patients), movement disturbances (59%), and headache (52%). Movement disturbances included gait disturbance, ataxia, tremor, limb weakness, and myoclonus. Autonomic dysfunction was less commonly reported (38%), including urinary or bowel dysfunction, erectile dysfunction, and blood pressure alterations.

Altered consciousness or confusion was observed in 38% of patients. Other cognitive disturbances, such as attention deficits, were noted in 34% of patients. In 23% of patients, there were overt signs of psychosis such as hallucinations or severe behavioural alterations. Impaired visual acuity or eye movement disturbances were documented in 28% of patients, and in approximately 50% of these there were findings of optic disc oedema or papillitis on clinical ophthalmological examination. Notably, 12% of patients deteriorated severely in their neurological status, resulting in coma and respiratory failure. Although a monophasic disease course was the most common clinical outcome, relapsing disease was also reported [4].

Prior infection and malignancy status

The meta‐analysis revealed that 45% (95% confidence interval [CI] = 31%–61%) of patients had a viral prodromal state (Figure 1a; six studies, 187 patients), mostly comprising upper respiratory tract or gastrointestinal infections but also rarer cases with human immunodeficiency virus and bacterial infections, for example, Mycobacterium tuberculosis or Brucella [12, 13].

FIGURE 1.

Forest plot of proportions of clinical and laboratory findings in autoimmune glial fibrillary acidic protein astrocytopathy (GFAP‐A). Pooled analyses of studies reporting the proportion of viral prodromal disease prior to GFAP‐A onset (a), concomitant neoplasm (b), coexisting autoantibodies (c), and response to immunotherapy (such as intravenous corticosteroids; d) are shown. Proportions were extracted and pooled using the random effects DerSimonian–Laird method. CI, confidence interval.

Coexisting tumours were reported in 98 of the patients (five of whom were paediatric patients), the most frequent of which was ovarian teratoma (32 patients). In the meta‐analysis, 18% (95% CI = 12%–28%) of patients had a concomitant neoplasm (Figure 1b; 10 studies, 428 patients). Accompanying neoplasms included ovarian teratoma, B‐cell lymphoma, meningioma, glioma, adenocarcinoma of the prostate and colon, carcinoma of the lung and urinary bladder, ductal breast cancer, and melanoma. In 25 reported cases of malignancy, there were simultaneous autoantibodies, most commonly N‐methyl‐d‐aspartate receptor (NMDAR)‐IgG in ovarian teratoma patients [2]. Notably, GFAP‐A onset was also reported to be associated with immune‐checkpoint inhibitor treatment [2, 14].

Laboratory findings and autoimmunity

GFAP‐IgG was by definition detected in all 681 patients included in this review. GFAP‐IgG was detected in CSF in a majority of cases (593 patients, 87%), most commonly by cell‐based assay. GFAP‐IgG was only detected in serum in 87 patients (13%) and found only by stereotactic biopsy of the brain in one case [8].

Coexisting autoantibodies were reported in 179 patients (Table 1). The meta‐analysis showed that up to 45% (95% CI = 32%–59%) of patients can present with coexisting autoantibodies (Figure 1c; nine studies, 212 patients). Among reported coexisting autoantibodies, the most frequent was NMDAR‐IgG (28%), followed by aquaporin‐4 (AQP4)‐IgG (16%) and myelin oligodendrocyte glycoprotein (MOG)‐IgG (8%). Less frequent coexisting autoantibodies were antinuclear antibodies, anti‐Yo‐antibodies/Purkinje cell cytoplasmic autoantibody type 1, thyroxine peroxidase antibodies, ganglioside antibodies, anti‐Sjögren syndrome‐related antigen A/B antibodies, and antineutrophil cytoplasmic antibodies.

Response to immunotherapy

Treatment with immunotherapy, high‐dose intravenous corticosteroids in particular (less commonly plasma exchange and intravenous immunoglobulin), was reported for 342 patients, of whom 302 patients (88%) responded well with a complete or partial remission. The imaging response to intravenous corticosteroids was often noted to be prompt, with rapid resolution of contrast enhancement as a sign of improved blood–brain barrier function, followed by the resolution of T2/fluid‐attenuated inversion recovery (FLAIR) hyperintensities at a slower pace [15]. The meta‐analysis on treatment response was in line with this finding, with 83% of patients (95% CI = 69%–91%) showing complete or partial remission upon immunotherapy (Figure 1d; seven studies, 171 patients).

Neuroimaging phenotypes of GFAP‐A

Brain MRI findings

Brain MRI findings were reported in 543 patients. A detailed summary of imaging findings can be found in Table 1. In the meta‐analysis, normal brain MRI findings were reported in 21% of patients (95% CI = 15%–28%; Figure S2a; eight studies, 172 patients).

The most common neuroimaging manifestation was T2/FLAIR hyperintensities, reported in 74% of patients (95% CI = 56%–87%) in the meta‐analysis (Figure 2a; seven studies, 203 patients). These hyperintensities were predominantly periventricular, extensive, confluent, and hazy. Conversely, smaller and demarcated T2/FLAIR hyperintensities in the deep white matter were less frequent. Representative examples of neuroimaging findings in an adult case with GFAP‐A from our institution are presented in Figure 3, with a graphical illustration in Figure 4.

FIGURE 2.

Forest plot of proportions of brain and spinal cord imaging findings in glial fibrillary acidic protein astrocytopathy (GFAP‐A). Pooled analyses of studies reporting the proportion of the presence of cerebral (fluid‐attenuated inversion recovery [FLAIR]‐)/T2‐weighted (T2w) hyperintensities (a), the presence of gadolinium‐enhancing lesions (b), perivascular radial gadolinium enhancement (c), and the presence of (lepto)meningeal enhancement (d) are shown. Proportions were extracted and pooled using the random effects DerSimonian–Laird method. CI, confidence interval.

FIGURE 3.

Neuroimaging findings in an adult case of autoimmune glial fibrillary acidic protein astrocytopathy (GFAP‐A). A male in his mid‐60s with a history of hypertension presented to the emergency room with headache and fever. He subsequently developed nystagmus, diplopia, ataxia, tremor, and myoclonic seizures. He further progressed with a loss of consciousness and apnoeas, and was intubated and treated in the intensive care unit. Brain and spinal cord magnetic resonance imaging revealed lesions and characteristic leptomeningeal enhancement in a perivascular radial distribution in the centrum semiovale (a–c), basal ganglia (d, e), pons, and cerebellum (g). There was also leptomeningeal enhancement around the conus (f). Notably, bilateral lesions in the pons in proximity to the middle cerebellar peduncles had restricted diffusion (h, i). Suspicion of autoimmune GFAP astrocytopathy was raised, and cerebrospinal fluid testing confirmed the presence of GFAP‐IgG autoantibodies. No coexisting malignancy was found, and no other neuronal autoantibodies were detected. There was a prompt response to corticosteroid treatment and continuous remission after initiation of anti‐CD20 therapy with rituximab.

FIGURE 4.

Lesion distribution of neuroinflammatory disorders. Schematic lesion distribution maps in the central nervous system of neuroinflammatory disorders that may overlap clinically and neuroradiologically with autoimmune glial fibrillary acidic protein astrocytopathy (GFAP‐A) are shown. MOGAD, myelin oligodendrocyte glycoprotein antibody disease; MS, multiple sclerosis; NMOSD, astrocyte aquaporin‐4‐positive neuromyelitis optica spectrum disorder.

Subcortical grey matter involvement was also prevalent, with 37% of patients presenting with T2/FLAIR hyperintensities and/or contrast enhancement in the basal ganglia or thalami, frequently bilaterally. Cortical/juxtacortical involvement was detected in 16% of patients. Mirroring the white matter manifestations, the grey matter lesions were typically diffuse and/or hazy.

Contrast enhancement was observed in 58% of patients (95% CI: 47%–68%) in the meta‐analysis (Figure 2b; 10 studies, 305 patients), most frequently located in areas with T2/FLAIR hyperintensities. Perivascular linear contrast enhancement was reported in 45% of patients (95% CI = 31%–59%) in the meta‐analysis (Figure 2c; nine studies, 234 patients). This enhancement typically extended radially from the lateral ventricles but was also reported in the cerebellum, brainstem, and basal ganglia [2, 9, 15, 16, 17]. Leptomeningeal contrast enhancement was present in 30% of patients (95% CI = 18%–46%) in the meta‐analysis (Figure 2d; eight studies, 166 patients).

Brainstem pathology was reported in 34% of patients and cerebellar pathology in 19% of patients, mostly described as T2/FLAIR hyperintensities. Area postrema lesions were highlighted in a few studies [18, 19, 20].

Corpus callosum involvement was rather rare, reported in only 5% of the patients, and often associated with reversible splenial lesion syndrome [21, 22, 23, 24]. Findings of restricted diffusion were found in only 3% of the patients, commonly in the corpus callosum. Cerebral haemorrhage was reported in one single patient, located in the thalamus [25].

Spinal cord MRI findings

Spinal cord MRI findings were commonly observed, with 49% of patients (95% CI = 40%–62%) having an abnormal spinal cord MRI in the meta‐analysis (Figure S2b; five studies, 143 patients; see also Table 1). Details on spinal lesion topography were specified for 98 patients, of whom 71% had cervical involvement, 65% had thoracic involvement, and 23% had conus and/or cauda equina involvement.

Longitudinally extensive myelitis, defined as lesions extending over three vertebral levels, was evident in 29% of patients. Conversely, smaller spinal lesions with an extension of less than three vertebral levels were noted in 10% of patients, frequently manifesting in a multifocal or patchy pattern. The lesions were predominantly located centrally within the cord, affecting the grey matter, and characterized as hazy, subtle, or diffuse on T2‐weighted imaging [1, 2, 7, 26, 27]. Notably, three case reports described the lesions as bilateral longitudinal with an eccentric location [28, 29, 30].

Spinal contrast enhancement was detected in 26% of patients, specified as leptomeningeal in 16% and intraparenchymal (characterized as patchy, punctate, speckled, or scattered) in 15% of patients. Interestingly, a couple of studies reported the hallmark pattern of perivascular contrast enhancement within the spinal cord [17, 31], as well as notable instances of contrast enhancement adjacent to the central canal [2, 7, 32].

The meta‐analysis suggested that spinal cord MRI scans with no particular features were also relatively common, with 41% of patients (95% CI = 26%–58%) having a normal spinal cord MRI (Figure S2c; eight studies, 172 patients).

Optic nerve MRI findings

Despite visual symptoms and optic disc oedema being described in several case reports [33], MRI findings of optic neuritis were only specifically reported in 19 patients. Bilateral optic neuritis was reported in 10 cases [8, 14, 34, 35, 36, 37]. Unilateral optic neuritis was reported in six cases, whereas the rest of the cases were unspecified [4, 15, 38]. Two of the cases were reported as having extensive lesions involving the optic chiasm [15, 36].

Advanced and nuclear neuroimaging findings

Advanced and nuclear neuroimaging modalities were reported in 21 patients.

Fluorodeoxyglucose PET findings were reported in 16 patients. Whereas some authors reported normal findings [25], even in areas with pathology on MRI [39, 40], others reported increased or decreased uptake in different regions [27]. A few studies reported high intramedullary metabolism associated with extensive myelitis [29, 37, 41].

Single‐photon emission computed tomography was applied in two patients, demonstrating a decreased blood flow in the frontal lobes in one case and an increased blood flow in the basal ganglia, thalamus, and corona radiata in the other patient [42, 43].

MRI spectroscopy was evaluated in two patients, showing low levels of myo‐inositol (an astrocyte marker) in one of the patients [23]. In the other patient, an abnormally high choline peak (a marker of increased cellular membrane turnover) and low N‐acetylaspartate peak (a neuronal marker) were reported.

Paediatric clinical and neuroimaging phenotypes of GFAP‐A

There were 88 paediatric cases with reports on neuroimaging phenotypes (54 males, 61%; 34 females, 39%), the largest study comprising 35 patients [35]. In the paediatric population, GFAP‐IgG was detected in CSF in 66 patients (75%), and in serum only in the remaining 22 patients (25%).

In the paediatric cases, simultaneous autoantibodies were detected in the CSF or serum in 19 patients (24%), and in five (7%) patients there were reports of coexisting malignancy (two cases of yolk sac tumour, paraganglioma, retroperitoneal tumour, and ependymoma). The most common clinical phenotype was meningoencephalitis (32%), followed by meningoencephalomyelitis (26%), encephalomyelitis (10%), encephalitis (8%), meningitis (2%), and myelitis (2%). In 20% of patients, there was no specified clinical phenotype defined as the above. Visual symptoms were present in 14%. Most publications evaluated treatment response, and in 90% of the patients a good or partial response to immunosuppressive treatment (most commonly corticosteroids) was reported.

All 88 paediatric patients underwent brain MRI. Normal findings were reported to a low extent (20%), and signal abnormalities were located in the basal ganglia (38%), deep white matter (33%), brainstem (23%), cerebellum (16%), cortical/juxtacortical regions (10%), and corpus callosum (15%). Restricted diffusion in the corpus callosum was infrequently reported (5%).

Nearly all reported brain MRI scans (94%) included gadolinium‐based contrast agents. Intracranial contrast enhancement was less common than in the pooled, predominantly adult, population. Leptomeningeal enhancement was reported as slightly more frequent (19%) than perivascular linear enhancement (13%). Unspecified contrast enhancement was also reported (9%).

Spinal cord MRI was performed in 71 paediatric patients, often with normal findings (52%). Lesions occurred all along the spinal cord, including longitudinally extensive myelitis (25%), whereas only one case presented with short and patchy lesions. Leptomeningeal enhancement was the most frequent enhancement pattern (11%).

Sensitivity analysis

A complementary meta‐analysis was performed for cases that had confirmed GFAP antibodies in CSF (Figures S3–S5), that is, excluding studies of cases with only seropositivity. This revealed consistent results regarding overall effect sizes, for both clinical and imaging characteristics, emphasizing the reliability of the findings.

DISCUSSION

The clinical spectrum of autoantibody‐mediated neuroimmunological conditions has recently been expanded to include a new entity, GFAP‐A. GFAP‐A is a rare but severe neuroinflammatory disorder characterized by the presence of GFAP‐IgG autoantibodies, predominantly detected in the CSF. Based on a systematic review and meta‐analysis, covering data from the definition of the disease in 2016 up until June 2023, we have generated a structured synthesis of the clinical and neuroimaging spectrum encountered with GFAP‐A.

The most notable imaging findings include the hallmark sign of perivascular linear perivascular contrast enhancement, deep grey matter signal abnormalities, hazy white matter hyperintensities, and longitudinally extensive myelitis. This consolidation of the evidence base may assist in raising the suspicion of GFAP‐A in patients with acute onset neuroinflammation, and testing for GFAP‐IgG will distinguish these patients from those with other conditions, thereby facilitating an early diagnosis and initiation of appropriate treatment.

Clinical phenotypes and response to treatment

Considering GFAP is an intracellular protein lacking a surface antigen, the pathophysiological pathway leading to disease is obscure. It is hypothesized that GFAP‐IgG might be a surrogate marker for an underlying cytotoxic T‐cell‐mediated autoimmune response [6].

Despite uncertain pathological mechanisms, the clinical phenotype, with prodromal infectious states, fever, and headache in conjunction with a rather acute onset of psychiatric symptoms as well as movement disorders, was unifying for a majority of the patients. Interestingly, GFAP‐A cases have been reported in a wide age span, including paediatric cases, and there is no clear sex predominance.

Visual impairment and eye movement disturbance are rather common, despite few reports of MRI‐confirmed optic neuritis. Notably, optic disc oedema and papillitis are relatively frequent, although often asymptomatic, with reports of normal or only slightly elevated intracranial pressure at lumbar puncture. The pathophysiological mechanism of this phenomenon remains unclear, but interestingly, GFAP is expressed in the retina, and it has been suggested that the finding may be a result of inflammatory vasculopathy with papillitis [33].

High‐dose intravenous corticosteroids are the most frequent treatment in the acute stage, often with prompt and efficacious response, although as many as every 10th patient requires intensive care during the disease course. In terms of the long‐term prognosis, previous studies have shown that relapses occur in up to 28% of patients within a 19‐month follow‐up [4], often during steroid tapering, underlining the need for monitoring and consideration of maintenance therapy.

The high frequency of other neuronal autoantibodies and malignancies further emphasizes the need for careful diagnostic workup [35, 44, 45]. Future studies are needed to show to what degree such findings can explain clinical heterogeneity and possibly identify subforms of GFAP‐A.

Neuroimaging phenotypes

The GFAP‐A hallmark of perivascular linear contrast enhancement is reported in 45% of all patients in our meta‐analysis, in agreement with previous larger publications [2, 14, 46, 47]. It is noteworthy that this pattern has been observed not only in the brain but also in the brainstem, cerebellum, and spinal cord. This finding is dynamic and may potentially be underreported, because linear perivascular enhancement was sometimes reported to be discovered in retrospect after the detection of GFAP‐IgG. This indicates that the perivascular enhancement pattern may be subtle and that awareness of it and optimized imaging protocols are essential. Temporal aspects are also crucial in the evaluation of neuroimaging findings in GFAP‐A [15]; however, details on the timing of neuroimaging are lacking in many of the larger case series, impeding a quantitative assessment of the temporal significance in this systematic review.

Other consistent neuroimaging findings include diffuse signal abnormalities in both subcortical grey and white matter of the brain [2, 47, 48], as well as longitudinally extensive myelitis. Contrast enhancement of the spinal cord is to a large extent heterogeneous. Nevertheless, several studies reported a specific finding of central canal contrast enhancement, corresponding to GFAP‐enriched regions in the rodent cord [2]. Potentially, this enhancement pattern could further add to the hallmark findings of GFAP‐A.

Currently, there are few studies on advanced neuroimaging and nuclear medicine modalities in GFAP‐A, and the diagnostic and prognostic value of such modalities remains to be determined.

Paediatric perspective

Paediatric GFAP‐A tends to present with more subtle or even absent imaging findings in the brain and spinal cord, although definite subgroup analyses were not feasible due to the pooling of data. Larger prospective studies are therefore needed.

Differential diagnostic cues

GFAP‐A, with its diverse manifestations, can resemble many other diseases, including other neuroinflammatory disorders, infectious diseases (e.g., M. tuberculosis, of which there are noteworthy examples), neurosarcoidosis, small vessel vasculitis, and lymphoma. Clinical and laboratory features, as well as the usually prompt responsiveness to corticosteroids, may implicate differential diagnosis in the field of autoimmune neuroinflammatory disorders. There are, however, key clinical and imaging findings that may facilitate the diagnostic workup of GFAP‐A; these are summarized in Table 2, and a schematic of lesion distribution compared to other neuroinflammatory disorders can be found in Figure 4.

TABLE 2.

Comparison of clinical and neuroimaging phenotypes in GFAP‐A, MOGAD, AQP4+ NMOSD, and MS.

	GFAP‐A	MOGAD	NMOSD	MS
Clinical features and demographics
Median age at onset, years	46	23	38	32
Males	55%	72%	17%	33%
Myelitis	+++	++	+++	++
Para‐/tetraplegia	+	+	++	−
Autonomic dysfunction	++	+	+	+
Headache	+++	+	+	+
Fever	+++	+	+	−
Cognitive impairment at onset	++	++	+	+
Altered consciousness	++	++	+	−
Psychosis	++	−	−	−
Visual symptoms	+	++	+++	++
Nuchal rigidity	++	−	−	−
Respiratory failure	+	+	+	−
Area postrema symptoms	+	+	+++	−
Hyponatremia	+	N/A	N/A	N/A
Brain imaging findings
Periventricular white matter lesions	++	+	+++	+++
White matter lesions: confluent, hazy	+++	++	−	+
White matter lesions: focal, distinct	++	++	+	+++
Presence of Dawson fingers	−	−	+	+++
Cortical/juxtacortical lesions	++	+	−	+++
Subcortical grey matter lesions	+++	++	−	+
Corpus callosum lesions	+	++	+	+++
Brainstem lesions	++	+++	+++	++
Area postrema lesions	+	+	+++	−
Contrast enhancement, perivascular	+++	−	−	−
Contrast enhancement, leptomeningeal	++	++	+	+
Optic nerve imaging findings
Optic neuritis	+	+++	+++	++
Optic disc oedema	++	++	N/A	−
Spinal cord imaging findings
Longitudinal extensive myelitis	++	++	+++	−
Short lesions (<3 vertebral levels)	+	++	+	+++
Spinal lesions: hazy, speckled	+++	++	+	+
Spinal cord oedema	+	+	+++	+
Cervical spinal cord involvement	+++	++	+++	+++
Thoracic spinal cord involvement	+++	+++	+++	++
Conus involvement	+	+++	+	−
Centrally located spinal lesions	++	++	+++	+
Laterally located spinal lesions	+	+	++	+++
Contrast enhancement	++	+	+++	+
Central canal contrast enhancement	++	−	−	−
Leptomeningeal contrast enhancement	++	+	+	−

Note: Summary based on the current literature review as well as previous reports in Carandini et al. [S51], Xiao et al. [4], and Carnero Contentti et al. [S52]. The number of + symbols represents how frequent/typical the clinical feature or neuroimaging finding is per diagnosis.

Abbreviations: AQP4, aquaporin‐4; GFAP‐A, autoimmune glial fibrillary acidic protein astrocytopathy; MOGAD, myelin oligodendrocyte glycoprotein antibody disease; MS, multiple sclerosis; N/A, data not available; NMOSD, AQP4‐positive neuromyelitis optica spectrum disorder.

Clinically, GFAP‐A shows a resemblance with MOG antibody‐associated disease (MOGAD) and acute disseminated encephalomyelitis (ADEM), with its prodromal state and psychiatric symptoms in conjunction with altered consciousness. Although GFAP‐A causes longitudinally extensive myelitis, permanent para‐ or tetraplegia was seldom reported in GFAP‐A, as opposed to AQP4‐positive neuromyelitis optica spectrum disorder (NMOSD) [49, 50]. Similarly, visual involvement in GFAP‐A is typically painless and of a mild character compared to AQP4‐positive NMOSD, with only rare cases of long‐term visual impairment or blindness [8, 33, 36].

Neuroradiologically, the hallmark imaging finding of perivascular linear contrast enhancement in GFAP‐A is distinctive from other neuroinflammatory disorders, although it is not pathognomonic, because it may also be present in, for example, neurosarcoidosis, small vessel vasculitis, and intravascular lymphoma [1]. The frequent diffuse involvement of both subcortical grey and white matter is another similarity with ADEM and MOGAD. Although subcortical grey matter involvement occurs in MS, it is typically focal in the thalami, and in AQP4‐positive NMOSD it is typically located in periventricular structures and the hypothalamus. In GFAP‐A, spinal cord lesions are often longitudinally extensive, but usually more subtle, with less generalized spinal cord oedema and swelling compared to AQP4‐positive NMOSD. Furthermore, leptomeningeal and central canal enhancement, as seen in GFAP‐A, is less characteristic of AQP4‐positive NMOSD.

Recommendations on neuroimaging

Based on the findings in this review, we provide suggestions for MRI protocols in Table 3. We recommend imaging of the entire neuroaxis with the administration of gadolinium‐based contrast agents. Although three‐dimensional (3D) T1‐weighted imaging is used for detecting contrast enhancement, 3D T2‐weighted FLAIR after gadolinium may facilitate the detection of leptomeningeal enhancement. The high frequency of spinal cord involvement in GFAP‐A underlines the importance of spinal cord MRI.

TABLE 3.

Recommended neuroimaging protocol for diagnostics and monitoring of autoimmune glial fibrillary acidic protein astrocytopathy.

Brain MRI	Comment	Spinal cord MRI	Comment
Sagittal 3D T2‐weighted FLAIR	White and grey matter hyperintensities, encephalitis	Sagittal T2‐weighted TSE and/or STIR	Spinal cord lesions; Dixon could also be used
Sagittal 3D T1‐weighted GRE IR/TSE	White and grey matter hypointensities, precontrast image for comparison	Sagittal T1‐weighted (FLAIR) TSE	Precontrast image for comparison
Axial 3D SWI	Microbleeds, subarachnoid haemorrhage, cortical superficial siderosis (ruling out differential diagnosis)
Coronal T2‐weighted TSE or STIR	Optical neuritis
GBCA administration	Unless contraindicated	GBCA administration	Unless contraindicated
Axial T2‐weighted TSE	White and grey matter changes, encephalitis	Axial T2‐weighted TSE	Spinal cord lesion topography
Axial DWI	Infarcts, encephalitis (ruling out differential diagnosis)	Sagittal T1‐weighted (FLAIR) TSE	Blood–brain barrier disruption, leptomeningeal enhancement
Sagittal 3D T1‐weighted GRE IR/TSE with GBCA	Blood–brain barrier disruption, leptomeningeal enhancement	Axial T1‐weighted (FLAIR) TSE	Coverage over suspected contrast enhancement
Sagittal 3D T2‐weighted FLAIR with GBCA	Leptomeningeal enhancement, perivascular linear contrast enhancement

Note: Brain 3D imaging should ideally have an isotropic voxel size, preferably ≤1 mm. The spinal cord sagittal imaging should cover the entire spinal cord and conus. Ideally, the axial T2‐weighted images should have the same coverage. Sagittal images should have a slice thickness of ≤3 mm and axial images ≤4 mm. T1‐weighted TSE can preferably be performed as FLAIR, if possible. Some sequences have been placed after GBCA administration to allow for appropriate contrast distribution before T1‐weighted imaging after contrast.

Abbreviations: 3D, three‐dimensional; DWI, diffusion‐weighted imaging; FLAIR, fluid‐attenuated inversion recovery; GBCA, gadolinium‐based contrast agent; GRE IR, gradient‐recall echo with an inversion pulse; MRI, magnetic resonance imaging; STIR, short tau inversion recovery; SWI, susceptibility‐weighted imaging; TSE, turbo spin echo.

Limitations

In our comprehensive study, we included patients presenting with GFAP‐IgG in CSF and/or serum, but there are studies indicating that the latter might be an unspecific finding [2, 48]. Notably, a sensitivity analysis of only CSF‐positive patients revealed similar results. Additionally, the presence of coexisting autoantibodies and overlapping clinical and neuroimaging features, in for example MOGAD and NMOSD, limits the ability to tease out GFAP‐A‐specific findings in the current literature and discriminate it from overlapping disorders. Furthermore, we report only patients who were tested for GFAP‐IgG, and therefore the suspicion of GFAP‐A astrocytopathy would have an inherent bias toward clinical and MRI features already known to associate with this diagnosis. Finally, inevitable heterogeneity existed in the available data for the meta‐analysis, emphasizing the diversity of included cohorts and employed clinical care.

CONCLUSIONS

GFAP‐A can present with a range of neuroimaging and clinical findings. A high clinical awareness of GFAP‐A is therefore necessary in the diagnostic workup of patients with noninfectious encephalitis and meningeal features, and prompt testing of GFAP‐IgG is recommended. Neuroradiological findings of perivascular contrast enhancement, deep grey matter involvement, and longitudinally extensive myelitis may be indicative of GFAP‐A. Detection of coexisting autoantibodies and/or concomitant malignancy are important factors to consider in the diagnostic workup of suspected cases. Future studies should elaborate on the described clinical, laboratory, and imaging features, explore the paediatric panorama of GFAP‐A, and evaluate the role of advanced MRI in the diagnostics and management of GFAP‐A.

AUTHOR CONTRIBUTIONS

Benjamin V. Ineichen: Funding acquisition; supervision; conceptualization; methodology; writing – original draft; project administration; formal analysis; visualization; data curation; investigation. Caroline Hagbohm: Conceptualization; investigation; writing – original draft; data curation; formal analysis; validation. Russell Ouellette: Conceptualization; visualization; writing – review and editing; investigation; validation. Eoin P. Flanagan: Investigation; validation; writing – review and editing; conceptualization. Dagur I. Jonsson: Investigation; validation; writing – review and editing. Fredrik Piehl: Validation; writing – review and editing; investigation. Brenda Banwell: Writing – review and editing; validation; investigation. Ronny Wickström: Validation; writing – review and editing; investigation. Ellen Iacobaeus: Validation; writing – review and editing; investigation. Tobias Granberg: Funding acquisition; supervision; conceptualization; writing – original draft; project administration; investigation.

FUNDING INFORMATION

This work was supported by Region Stockholm and Karolinska Institutet through ALF Medicine (No. ALF 20200224 to T.G.) and CIMED (CIMED FoUI‐976444 to TG) as well as grants of the Swiss National Science Foundation (No. P400PM_183884, to BVI), and the UZH Alumni (to B.V.I.). The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication.

CONFLICT OF INTEREST STATEMENT

C.H. reports no disclosures. R.O. is funded by the Swedish Society for Medical Research's Postdoctoral Grant (No. PG‐22‐0440) and has received speaker honoraria from Novartis. D.J. reports no disclosures. F.P. has received research grants from Janssen, Merck KGaA, and UCB; and fees for serving on data monitoring committees in clinical trials with Chugai, Lundbeck, and Roche, and preparation of an expert witness statement for Novartis. B.B. has served as a consultant for Novartis, Roche, UCB, Sanofi‐Genzyme, and Biogen Idec. E.P.F. has served on advisory boards for Alexion, Genentech, Horizon Therapeutics, and UCB. He has received research support from UCB. He has received speaker honoraria from Pharmacy Times. He has received royalties from UpToDate. E.P.F. is a site principal investigator in a randomized clinical trial of rozanolixizumab for relapsing myelin oligodendrocyte glycoprotein antibody‐associated disease run by UCB. E.P.F. is a site principal investigator and a member of the steering committee for a clinical trial of satralizumab for relapsing myelin oligodendrocyte glycoprotein antibody‐associated disease run by Roche/Genentech. E.P.F. has received funding from the NIH (R01NS113828). E.P.F. is a member of the medical advisory board of the MOG project. E.P.F. is an editorial board member of the Journal of the Neurological Sciences and Neuroimmunology Reports. A patent has been submitted on DACH1‐IgG as a biomarker of paraneoplastic autoimmunity. R.W. has received honoraria for serving on advisory boards for UCB, GW Pharma, and Octapharma and speaker's fees from Eisai, Jazz Pharma, and Sanofi‐Genzyme. He has received funding from Region Stockholm Clinical Research Appointment. E.I. has received honoraria for serving on advisory boards for Biogen, Sanofi‐Genzyme, and Merck and speaker's fees from Biogen and Sanofi‐Genzyme. She has received funding from Region Stockholm Clinical Research Appointment (No. 108291). T.G. is funded by Region Stockholm and Karolinska Institutet (Nos. ALF 20200224, ALF Medicine FoUI‐987826, MedTechLabs FoUI‐991015, and CIMED FoUI‐976444), the Swedish Research Council (No. 2023–03146), the Swedish Society for Medical Research's Big grant (No. S19‐0227), Alzheimersfonden (Nos. AF‐939915, AF‐968591, AF‐994629), and Merck's Grant for Multiple Sclerosis Innovation. B.V.I. is funded by the Swiss National Science Foundation (No. P400PM_183884 and 407940_206504), the UZH FAN Alumni Fellowship, and the Digital Entrepreneur Fellowship from the University of Zürich.

Supporting information

DATA S1.

Hagbohm C, Ouellette R, Flanagan EP, et al. Clinical and neuroimaging phenotypes of autoimmune glial fibrillary acidic protein astrocytopathy: A systematic review and meta‐analysis. Eur J Neurol. 2024;31:e16284. doi: 10.1111/ene.16284

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available in Github at https://github.com/Ineichen‐Group/GFAP_astrocytopathy_SR.