Summary

A medRxiv systematic review and meta-analysis found that the blood biomarker GFAP distinguished intracerebral hemorrhage from ischemic stroke, with pooled specificity of 87.2%. Accuracy was higher in samples taken within two hours, but sensitivity remained limited, particularly for smaller hemorrhages.

A systematic review and meta-analysis posted as a medRxiv preprint reports that blood levels of glial fibrillary acidic protein (GFAP) can help distinguish intracerebral hemorrhage (ICH) from acute ischemic stroke (AIS). Across the diagnostic studies included in the analysis, GFAP showed a pooled area under the curve (AUC) of 0.86, sensitivity of 74.4% and specificity of 87.2% for separating the two stroke subtypes.

The analysis also found stronger discrimination when blood was sampled soon after symptoms began. For samples collected within two hours, the pooled AUC reached 0.96 and specificity was 96.4%. Sensitivity remained 76.7%, however, indicating that early testing could be better at supporting identification of bleeding than at excluding it.

Contents

What the review evaluated

GFAP is a structural protein associated with astrocytes, cells that support neurons in the central nervous system. Brain tissue injury can release GFAP into the bloodstream, making it a candidate blood biomarker for acute neurological damage.

The clinical distinction examined in the review is between ICH, in which a blood vessel ruptures and bleeding occurs within the brain, and AIS, in which a blocked blood vessel reduces blood flow. These conditions require different acute diagnostic and treatment pathways, so a blood-based signal that helps identify hemorrhage could be useful alongside established clinical assessment and imaging.

The researchers searched five databases in March 2026 and independently screened 6,336 studies. Sixty-four studies were included in the systematic review, while 18 contributed data to the diagnostic meta-analysis. The review was registered with PROSPERO before data extraction, and the authors assessed diagnostic-study bias with the QUADAS-2 tool. Overall risk of bias among the diagnostic studies was reported as low.

For ICH compared with AIS, the pooled AUC was 0.86, with sensitivity of 74.4% and specificity of 87.2%. AUC, or area under the receiver operating characteristic curve, summarises how well a test separates two groups across possible decision thresholds. Sensitivity measures how often the test detects ICH when it is present; specificity measures how often it correctly identifies people without ICH.

Earlier sampling produced stronger discrimination

The review analysed GFAP results according to the time between stroke onset and blood sampling. Diagnostic performance improved in earlier time windows. Among samples collected within six hours, the pooled AUC was 0.91, sensitivity was 75.0% and specificity was 92.7%. Within two hours, the corresponding values were 0.96, 76.7% and 96.4%.

The researchers also assessed GFAP for distinguishing ICH from undifferentiated suspected stroke, where the initial clinical picture does not yet identify the subtype. In that comparison, the pooled AUC was 0.88, sensitivity was 75.4% and specificity was 90.5%.

These results give GFAP its clearest potential as a rule-in aid for ICH: a high specificity means a positive result is more useful for supporting the presence of bleeding. The sensitivity figures show why a negative result cannot reliably exclude ICH, particularly when the bleed is small.

GFAP was linked more consistently to bleeding than to ischemic injury

The review found a relationship between GFAP levels and ICH volume in 11 of 12 studies that examined this outcome. Smaller hematomas, especially those below 10 mL, were frequently associated with GFAP values below the limits of detection. This finding is important because it places a practical boundary on the biomarker's ability to identify all cases of brain bleeding.

Evidence for using GFAP to estimate the extent of ischemic injury was much thinner. Only three studies suggested that higher GFAP levels were associated with greater ischemic injury in AIS. The authors therefore identified assessment of ischemic injury extent as an area requiring further research.

The findings support continued evaluation of GFAP as an early blood-based complement to stroke imaging, especially for identifying ICH. The report is a preprint based on previously published studies, and its strongest evidence concerns stroke-subtype discrimination and the relationship between GFAP and hemorrhage volume. Its sensitivity remains limited, while the biomarker's role in measuring ischemic damage is still uncertain.

Sources