Treatment-refractory autoimmune glial fibrillary acidic protein astrocytopathy: Clinical features, predictive models, therapy

@article{zhang2026treatmentrefractory,
  title = {{Treatment-refractory autoimmune glial fibrillary acidic protein astrocytopathy: Clinical features, predictive models, therapy}},
  author = {Shifeng Zhang and Huilu Li and Haiyan Yao and Ting Li and Wendong Lai and Shiyu He and Peihao Lin and Jie Yang and Yaqing Shu and Dayang Hui and Li Huang and Cong Gao and Xiaorong Liu and Youming Long},
  journal = {Neurobiology of Disease},
  year = {2026},
  month = oct,
  doi = {10.1016/j.nbd.2026.107653},
  url = {https://doi.org/10.1016/j.nbd.2026.107653},
  note = {CC BY},
}

TY  - JOUR
TI  - Treatment-refractory autoimmune glial fibrillary acidic protein astrocytopathy: Clinical features, predictive models, therapy
AU  - Shifeng Zhang
AU  - Huilu Li
AU  - Haiyan Yao
AU  - Ting Li
AU  - Wendong Lai
AU  - Shiyu He
AU  - Peihao Lin
AU  - Jie Yang
AU  - Yaqing Shu
AU  - Dayang Hui
AU  - Li Huang
AU  - Cong Gao
AU  - Xiaorong Liu
AU  - Youming Long
PY  - 2026
DA  - 2026/10/30
JO  - Neurobiology of Disease
DO  - 10.1016/j.nbd.2026.107653
UR  - https://doi.org/10.1016/j.nbd.2026.107653
AB  - Background and objectives: A subset of glial fibrillary acidic protein astrocytopathy experiences treatment-refractory disease course, yet the clinical and immunological features defining this subgroup remain incompletely characterized. Methods: Patients were classified as treatment-refractory ( n = 23) or treatment-responsive ( n = 30). Clinical, imaging, and outcomes were compared. For subgroup analyses, refractory patients were divided into three groups: no immunoadsorption/plasma exchange (IA/PE) ( n = 9), IA/PE monotherapy ( n = 4), and IA/PE plus rituximab ( n = 10), with functional outcomes compared across groups. Immunoglobulin G (IgG) from refractory patients was applied to a passive transfer mouse model, and single-cell RNA sequencing was performed to explore underlying pathogenic mechanisms. Results: The treatment-refractory group had higher proportion of the meningoencephalomyelitis phenotype (82.6% vs. 53.3%, P = 0.026) and more frequent disturbances of consciousness (65.2% vs. 30.0%, P = 0.011), paraplegia (78.3% vs. 33.3%, P = 0.001), and peripheral nervous system involvement (89.5% vs. 22.2%, P < 0.001). Refractory patients had elevated cerebrospinal fluid glial fibrillary acidic protein and neurofilament light chain levels. Refractory patients more frequently exhibited spinal cord lesions involving three or more segments (65.2% vs. 21.4%, P = 0.002). Disability at last follow-up was greater in the refractory group (median modified Rankin Scale score 3 vs. 1, P < 0.0001). Firth penalized regression model identified peripheral nervous system involvement as the independent predictor of treatment refractoriness (odds ratio = 10.74, P = 0.006). Among refractory patients, the IA/PE-plus-rituximab group had significantly lower long-term mRS scores than the no-IA/PE group (adjusted P = 0.031). GFAPA-IgG from refractory patients induced CD45 + cell infiltration in mouse brain. Single cell analysis demonstrated activation of complement and inflammatory pathways, with astrocytes adopting a pro inflammatory state. Conclusion: Treatment refractory disease is marked by severe central and peripheral nervous system injury. Convergent clinical and experimental evidence provides a rationale for B cell targeted therapies.
SN  - 1095-953X
SN  - 0969-9961
ER  -