Rising GFAP After Thrombolysis Flags Early Stroke Deterioration

Among adults with minor ischemic stroke who received intravenous thrombolysis, a marked increase in serum glial fibrillary acidic protein (GFAP) within 24 hours was associated with early neurological deterioration in a prospective multicenter cohort study conducted in Northeast China. Patients who deteriorated had a median 3.57-fold increase in GFAP after treatment, compared with 1.57-fold among those who did not. The association remained statistically significant after adjustment for vascular risk factors, stroke characteristics and other clinical variables, indicating that a dynamic blood marker may provide information beyond the initial severity of a seemingly mild stroke.

The researchers performed a post hoc analysis of data from a prospective multicenter cohort involving patients treated for minor stroke. Between September 2016 and April 2023, patients receiving intravenous thrombolysis were approached at the First Hospital of Jilin University, and additional recruitment occurred at 15 hospitals in Northeast China from September 2021 to January 2022. A total of 347 patients received intravenous thrombolysis. After 28 patients who did not receive the standard dose of alteplase or bridging therapy were excluded, 319 met the study inclusion criteria. Two patients refused to participate, 39 had incomplete or unqualified blood samples, and six were lost to follow-up at 90 days, leaving 272 patients for analysis.

Eligible participants were older than 18 years, received thrombolysis within 4.5 hours of symptom onset, had an admission National Institutes of Health Stroke Scale (NIHSS) score of no more than 5, and had a pre-stroke Modified Rankin Scale score of 0 or 1. Minor stroke was therefore represented by mild neurological deficits rather than absence of disability. Blood was collected at baseline and 24 hours after thrombolysis. Early neurological deterioration was defined as an increase in the NIHSS score of at least four points within 24 hours after treatment. Twenty of the 272 analyzed patients, or 7.35%, met that definition; 252 did not experience early neurological deterioration.

Baseline GFAP concentrations did not differ significantly between the deterioration and no-deterioration groups. Before thrombolysis, median GFAP was 10.25 pg/mL among patients who later deteriorated, with an interquartile range of 6.81–19.24 pg/mL, compared with 8.54 pg/mL, with an interquartile range of 5.18–14.27 pg/mL, among those who did not. The groups separated more clearly after treatment: median GFAP at 24 hours was 40.05 pg/mL in the deterioration group, with an interquartile range of 22.59–126.24 pg/mL, versus 13.57 pg/mL, with an interquartile range of 8.74–33.80 pg/mL, in the no-deterioration group. The difference was significant at p<0.001. The change from baseline was therefore more closely associated with subsequent deterioration than the single pre-treatment measurement.

The investigators defined the relative GFAP change, ∆GFAP, as the post-thrombolysis concentration divided by the baseline concentration. Median ∆GFAP was 3.57 in the deterioration group, with an interquartile range of 1.40–23.39, and 1.57 in the no-deterioration group, with an interquartile range of 1.00–2.94; the group difference was significant at p=0.006. In univariable logistic regression, each unit increase in the log-transformed change measure was associated with an odds ratio of 3.495 for early neurological deterioration, with a 95% confidence interval of 1.672–7.306. The association remained statistically significant in models adjusted for age and sex, vascular risk factors, and clinical variables. In the most extensively adjusted model, a 2.54-fold increase in ∆GFAP was associated with a 4.64-fold increase in the odds of deterioration, with a 95% confidence interval of 1.86–11.539 and p<0.001.

GFAP is a protein in the cytoskeleton of astrocytes, which contribute to the structural support of neighboring cells and to maintenance of the blood–brain barrier. The source study notes that elevated GFAP during acute ischemic stroke can arise from extensive release associated with astrocyte necrosis in the infarct core and from increased expression associated with reactive gliosis in surrounding tissue. Infarct volume was similar in the two deterioration groups: 1.14 mL, with an interquartile range of 0.70–4.03 mL, among patients who deteriorated and 1.32 mL, with an interquartile range of 0.42–2.80 mL, among those who did not, with p=0.203. On that basis, the authors proposed that the larger GFAP rise may reflect differences in reactive gliosis rather than substantially different amounts of tissue death. This was a biological interpretation proposed by the researchers, not a mechanism directly tested in the cohort.

To assess whether GFAP change added information to conventional clinical variables, the researchers built two prediction models. Model A contained age and admission NIHSS score. Model B added log-transformed ∆GFAP. The clinical-only model had an area under the receiver operating characteristic curve of 0.695, with a 95% confidence interval of 0.584–0.805, while the model including GFAP had an area of 0.770, with a 95% confidence interval of 0.674–0.886. Under 10-fold cross-validation, the corresponding areas were 0.61 and 0.72. A Hosmer–Lemeshow test did not indicate poor model conformity, because its p value was greater than 0.05. The comparison between the two areas under the ROC curves was not statistically significant, but the GFAP-containing model produced significantly higher net reclassification and integrated discrimination values: the net reclassification index was 70.00%, with p=0.003, and the integrated discrimination improvement was 6.62%, with p=0.038. Decision-curve analysis also showed greater net benefit after adding ∆GFAP across a range of threshold probabilities.

Patients who experienced early neurological deterioration had a substantially poorer 90-day outcome profile. At follow-up, 5 of 20 patients in the deterioration group, or 25.0%, had a favorable outcome, defined as a Modified Rankin Scale score of 0 or 1. By comparison, 169 of 252 patients in the no-deterioration group, or 67.1%, had a favorable outcome; the difference was significant at p<0.001. Favorable outcomes also became less common across increasing thirds of ∆GFAP: 76.7% in the lowest third, 63.0% in the middle third and 55.6% in the highest third, with p=0.016. These observations describe an association in this cohort and do not demonstrate that GFAP elevation caused either neurological worsening or poorer functional recovery.

Repeated GFAP assessment was technically feasible with the assay used in the study. Serum GFAP was measured using a magnetic-particle-based automated chemiluminescent enzyme-linked immunosorbent assay system that required 60–80 microliters of serum and had a reported turnaround time of approximately 28 minutes from sample loading to result reporting. The lower limit of quantification was 5 pg/mL, and the assay coefficient of variation was below 8.0%. Laboratory technicians were blinded to patients’ clinical data, and test results were not shared with healthcare providers. The results support further evaluation of a repeated biomarker measurement as part of risk assessment, but the study did not test whether using GFAP to guide surveillance or treatment would prevent deterioration or improve outcomes.

Several limitations constrain the conclusions and the immediate generalizability of the prediction model. This was a post hoc analysis of a prospective multicenter cohort, and the final analysis included 272 of the 347 patients who initially received thrombolysis, after 28 exclusions and further losses from refusal, unusable or incomplete blood samples, and missing 90-day follow-up. The investigators did not distinguish between disabling and non-disabling minor strokes. There was no control group of patients receiving standard medical care without intravenous thrombolysis, so the findings do not isolate the effect of thrombolysis or establish whether the GFAP response would have differed without treatment. The relatively small number of events also prevented more precise prediction of hemorrhagic and ischemic causes of deterioration. External validation in larger, independent prospective cohorts is needed before the model can be considered broadly applicable, and future studies must determine whether biomarker-guided monitoring or intervention can actually reduce early neurological deterioration and improve long-term outcomes.

Subject of Research: Prospective cohort analysis of dynamic serum GFAP for predicting early neurological deterioration after thrombolysis in minor stroke.

Article Title: Dynamic Changes in Serum Glial Fibrillary Acidic Protein Levels Predicts Early Neurological Deterioration After Intravenous Thrombolysis in Minor Stroke

Article References: Zhou, J., Qu, Y., Abuduxukuer, R., Zhang, P., Wang, T., Gao, J.-H., Pan, D.-W., Yuan, Z.-M., Li, C.-Y., Chen, L.-P., Chen, J.-H., Gu, J.-L., Wang, A.-Y., Jiang, L.-G., Liu, L., Jiang, Y.-F., Wang, C.-F., Jia, Y., Xin, H., … the Biomarkers of Brain Injury Investigator Study Group (2026). Dynamic Changes in Serum Glial Fibrillary Acidic Protein Levels Predicts Early Neurological Deterioration After Intravenous Thrombolysis in Minor Stroke. Annals of Clinical and Translational Neurology, Article acn3.70552. https://doi.org/10.1002/acn3.70552

Image Credits: AI Generated

DOI: 10.1002/acn3.70552

Keywords: minor ischemic stroke, intravenous thrombolysis, glial fibrillary acidic protein, GFAP, astrocyte injury, early neurological deterioration, blood biomarker, stroke prognosis, prediction model, Chinese cohort, Dynamic, Changes