Most Cases of a Rare Autoimmune Brain Disease Are Misdiagnosed as Infection, Chinese Study Finds

Nearly two-thirds of patients with autoimmune glial fibrillary acidic protein astrocytopathy wait more than two weeks for a correct diagnosis, and more than four in five are initially misdiagnosed, according to a retrospective cohort study conducted at two tertiary hospitals in southern China. The most frequent wrong diagnosis was infectious meningoencephalitis, which accounted for 23 of the 34 misdiagnosed patients, or 54.76 percent of the cohort. The findings, drawn from 42 patients treated between July 2020 and March 2026, offer one of the first quantitative portraits of how often this rare autoimmune disorder slips past clinicians and why the delay happens.

GFAP astrocytopathy was formally established as a distinct autoimmune disease of the central nervous system in 2016. It is identified by antibodies against glial fibrillary acidic protein, detected in patients’ cerebrospinal fluid. The clinical picture is notoriously varied: fever, headache, impaired consciousness, epileptic seizures, limb tremors, ataxia, behavioral changes, visual impairment, and urinary or bowel dysfunction all appear among the reported initial symptoms. On brain MRI, the classic signature is a pattern of linear radial enhancement tracing the blood vessels of the cerebral white matter, and spinal imaging often shows long-segment cord lesions. But the disease frequently announces itself with fever, headache, and meningeal irritation, a combination that closely mimics viral or tuberculous meningitis and pushes infectious disease to the top of the differential.

The study, led by neurologists at Yuebei People’s Hospital in Shaoguan and Shantou Central Hospital in Guangdong Province, screened consecutive patients with suspected autoimmune central nervous system disorders at both centers. Of 58 patients whose cerebrospinal fluid tested positive for GFAP-IgG, 16 were excluded because of incomplete data, alternative confirmed diagnoses, or insufficient clinical manifestations, leaving 42 for analysis. The researchers defined delayed diagnosis as more than 14 days from the first symptom to the final diagnosis, a threshold borrowed from autoimmune encephalitis studies but not specifically validated for this disease. Misdiagnosis was defined more strictly: an incorrect initial diagnosis had to have prompted disease-specific treatment or caused a clinically meaningful delay of more than seven days before GFAP astrocytopathy was considered. A suspected infection did not count as a misdiagnosis if the treating physician had documented autoimmune causes in the differential and ordered GFAP-IgG testing within 72 hours of admission.

By these definitions, 27 of the 42 patients, or 64.29 percent, experienced diagnostic delay, and 34, or 80.95 percent, were misdiagnosed. Beyond infectious meningoencephalitis, the roster of wrong diagnoses ranged widely: two cases each of multiple sclerosis and acute myelitis, and single cases of subacute combined degeneration of the spinal cord, community-acquired pneumonia, cerebral infarction, posterior circulation ischemia, Guillain-Barré syndrome, incomplete intestinal obstruction, and spinocerebellar ataxia. The misdiagnosis rate was nearly identical in both groups, at 81.48 percent among delayed-diagnosis patients and 80.00 percent among those diagnosed promptly, indicating that being misdiagnosed at first contact did not by itself determine whether the overall diagnosis was delayed.

Where a patient first landed in the hospital mattered. Only 25.93 percent of the delayed-diagnosis group was first seen by a neurology department, compared with 53.33 percent of the non-delayed group, a statistically significant difference. The symptom profile at presentation also diverged. Patients diagnosed promptly were significantly more likely to show neck stiffness, present in 80.00 percent of that group versus 33.33 percent of the delayed group, and impaired consciousness, present in 60.00 percent versus 7.41 percent. As initial manifestations specifically, neck stiffness and impaired consciousness were recorded in 40.00 percent and 26.67 percent of the non-delayed group respectively, and in none of the delayed group. These dramatic signs appear to function as an alarm bell that accelerates referral and testing.

Conversely, patients whose disease began quietly fared worse. At the first medical consultation, 59.26 percent of the delayed-diagnosis group had only isolated fever and/or headache without any localizing neurological syndrome, meaning no meningeal, encephalic, myelitic, autonomic, or optic features, compared with 13.33 percent of the non-delayed group, a difference that was statistically significant. At symptom onset the same pattern was more common in the delayed group, 59.26 percent versus 33.33 percent, but this difference did not reach statistical significance. Autonomic dysfunction was prominent among delayed patients, with urinary difficulty in 51.85 percent and abdominal pain or distension in 44.44 percent, although these differences between groups were not significant. The authors suggest that such non-specific, non-localizing prodromes delay clinical recognition and the subsequent referral to neurology.

Timeline analysis sharpened the point. The median time from initial diagnosis to a positive cerebrospinal fluid antibody result was 16 days in the delayed group versus 5 days in the non-delayed group, a significant difference. Yet the median time from neurology admission to antibody positivity was nearly identical, at 5 days versus 4 days, and not statistically different. In other words, the bottleneck sits before the specialist: once patients reach neurologists, the diagnostic workup proceeds at a comparable pace regardless of how long they waited beforehand. Routine cerebrospinal fluid biochemistry offered little help in flagging the disease earlier. Both groups showed inflammatory profiles with elevated white cell counts and protein and normal or slightly low glucose and chloride, values that overlap extensively with viral and tuberculous meningitis. Although the median white cell count in the non-delayed group was roughly 4.6 times higher, the difference was not statistically significant, and the authors conclude that conventional CSF indicators cannot reliably distinguish patients who will experience delay.

Imaging and testing strategy added further signals. The characteristic perivascular radial enhancement appeared in only 22.22 percent of the delayed group on early MRI, against 60.00 percent of the non-delayed group, a significant difference, while meningeal enhancement and longitudinally extensive spinal cord lesions did not differ significantly. In an exploratory multivariable logistic regression limited to three prespecified covariates, the presence of perivascular radial enhancement was associated with lower odds of delayed diagnosis, with an odds ratio of 0.122, as were first presentation to neurology, odds ratio 0.153, and simultaneous submission of GFAP-IgG testing with the initial CSF examination, odds ratio 0.234. Notably, 34 of the 42 patients had GFAP-IgG submitted concurrently with the first lumbar puncture, and their median time from neurology admission to antibody positivity was 4 days versus 10 days for those tested only after pathogen screening came back negative. The authors are careful to stress that this comparison is observational and vulnerable to confounding by indication, since clinicians were more likely to defer antibody testing in patients with atypical presentations and lower initial suspicion, and the regression model could not adjust for center effects, calendar year, or disease severity. The wide confidence intervals and the absence of correction for multiple comparisons further mark these results as hypothesis-generating rather than confirmatory.

The study has other limits the authors acknowledge candidly. The sample was small and drawn entirely from two tertiary referral hospitals in Guangdong Province, so the referral patterns and levels of disease awareness may not generalize to primary care settings or other health systems. No control group of patients with infectious meningoencephalitis or other autoimmune neurological diseases was included, so the specificity of any clinical, laboratory, or imaging feature for distinguishing GFAP astrocytopathy from its mimics could not be assessed. The retrospective design carries risks of documentation bias, and the 14-day delay threshold has not been clinically validated. Even so, the analysis is presented as the first quantitative, multi-center characterization of diagnostic delay in this disease in a Chinese population, and it points to a concrete, testable idea: for patients presenting with fever and headache accompanied by subtle autonomic symptoms, or without clear localization, particularly when empirical anti-infective treatment fails, GFAP-IgG testing could be performed concurrently with the initial lumbar puncture rather than deferred until infections have been exhaustively excluded. If prospective studies validate that approach, the authors suggest, the time to diagnosis might shrink from weeks to days, allowing earlier immunomodulatory treatment. For now, the message is narrower but firm: the biggest obstacle to timely diagnosis of GFAP astrocytopathy lies outside the neurology ward, and closing that gap will require clinicians who first meet these patients to think of an autoimmune disease that so convincingly wears the mask of meningitis.

Subject of Research: Delayed diagnosis and misdiagnosis of autoimmune GFAP astrocytopathy in a dual-center retrospective cohort in China

Article Title: Delayed diagnosis and misdiagnosis of autoimmune glial fibrillary acidic protein astrocytopathy: a dual-center retrospective cohort study in China

Article References: Lin, S., Huang, Y., Huang, C., & Zheng, X. (2026). Delayed diagnosis and misdiagnosis of autoimmune glial fibrillary acidic protein astrocytopathy: a dual-center retrospective cohort study in China. Frontiers in Neurology, 17, Article 1968815. https://doi.org/10.3389/fneur.2026.1968815

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1968815

Keywords: GFAP astrocytopathy, misdiagnosis, diagnostic delay, GFAP-IgG, infectious meningoencephalitis, cerebrospinal fluid, autoimmune neurological disease, MRI, neuroimmunology, retrospective cohort study, China, Delayed