Diagnostic Performance of GFAP, UCH-L1, and S100B in addition to NICE Head Injury Guidelines in Head-Injured Patients
Presenter: Dr Alice Rogan, Emergency Specialist, University of Otago, Wellington
Authors List
Rogan, A., University of Otago, Wellington Regional Hospital, Wellington, New Zealand
Beck, S., University of Otago, Dunedin Hospital, Dunedin, New Zealand
Daniels, A., Dunedin Hospital, Dunedin, New Zealand
Faragasso, D., Taranaki Base Hospital, Taranaki, New Zealand
Joyce, L.R., University of Otago, Christchurch Hospital, Christchurch, New Zealand
Than, M., Christchurch Hospital, Christchurch, New Zealand
Jones, P.G., University of Auckland, Auckland City Hospital, Auckland, New Zealand
Larsen, P., University of Otago, Wellington, New Zealand
Background
Computed tomography (CT) head imaging is frequently performed in emergency department (ED) patients with head injury, but evidence supporting the incremental benefit of blood biomarkers in addition to established clinical decision rules remains limited.
Aim
To compare the diagnostic performance of S100B, GFAP, and UCH-L1 when used in addition to the NICE Head Injury Guidelines (HIG) in patients presenting to ED within 24 hours of head injury.
Methods
This prospective multicentre observational study enrolled adults (≥18 years) meeting NICE HIG criteria for CT head across six New Zealand EDs between May 2020 and June 2024. Serum S100B (≥0.1 µg/ml), GFAP (≥35 pg/ml), and UCH-L1 (≥400 pg/ml) were measured at predefined, validated thresholds; the reference standard was acute traumatic pathology on CT head.
Results
Of 673 included patients, 85 (12.6%) had a positive CT head. GFAP demonstrated the highest diagnostic performance (AUC 0.80, 95% CI 0.74–0.85; sensitivity 95.3%; specificity 25.2%; NPV 97.4%. S100B showed moderate performance (AUC 0.63, 95% CI 0.58–0.69; sensitivity 69.4%; specificity 51.4%; NPV 92.1%). UCH-L1 had the lowest sensitivity (AUC 0.63, 95% CI 0.58–0.69; sensitivity 43.5%; specificity 71.4%; NPV 89.8%). Combined GFAP/UCH-L1 improved sensitivity to 97.7% and NPV to 98.4% but reduced specificity to 20.9%.
Conclusion
These findings suggest that biomarkers may have a role as adjuncts to NICE-based pathways, although the trade-off between sensitivity and specificity, and the relevance of small missed injuries, remain important considerations. Further prospective studies are needed to confirm safety, refine thresholds, and clarify how this approach could be implemented in
Rogan, A., University of Otago, Wellington Regional Hospital, Wellington, New Zealand
Beck, S., University of Otago, Dunedin Hospital, Dunedin, New Zealand
Daniels, A., Dunedin Hospital, Dunedin, New Zealand
Faragasso, D., Taranaki Base Hospital, Taranaki, New Zealand
Joyce, L.R., University of Otago, Christchurch Hospital, Christchurch, New Zealand
Than, M., Christchurch Hospital, Christchurch, New Zealand
Jones, P.G., University of Auckland, Auckland City Hospital, Auckland, New Zealand
Larsen, P., University of Otago, Wellington, New Zealand
Background
Computed tomography (CT) head imaging is frequently performed in emergency department (ED) patients with head injury, but evidence supporting the incremental benefit of blood biomarkers in addition to established clinical decision rules remains limited.
Aim
To compare the diagnostic performance of S100B, GFAP, and UCH-L1 when used in addition to the NICE Head Injury Guidelines (HIG) in patients presenting to ED within 24 hours of head injury.
Methods
This prospective multicentre observational study enrolled adults (≥18 years) meeting NICE HIG criteria for CT head across six New Zealand EDs between May 2020 and June 2024. Serum S100B (≥0.1 µg/ml), GFAP (≥35 pg/ml), and UCH-L1 (≥400 pg/ml) were measured at predefined, validated thresholds; the reference standard was acute traumatic pathology on CT head.
Results
Of 673 included patients, 85 (12.6%) had a positive CT head. GFAP demonstrated the highest diagnostic performance (AUC 0.80, 95% CI 0.74–0.85; sensitivity 95.3%; specificity 25.2%; NPV 97.4%. S100B showed moderate performance (AUC 0.63, 95% CI 0.58–0.69; sensitivity 69.4%; specificity 51.4%; NPV 92.1%). UCH-L1 had the lowest sensitivity (AUC 0.63, 95% CI 0.58–0.69; sensitivity 43.5%; specificity 71.4%; NPV 89.8%). Combined GFAP/UCH-L1 improved sensitivity to 97.7% and NPV to 98.4% but reduced specificity to 20.9%.
Conclusion
These findings suggest that biomarkers may have a role as adjuncts to NICE-based pathways, although the trade-off between sensitivity and specificity, and the relevance of small missed injuries, remain important considerations. Further prospective studies are needed to confirm safety, refine thresholds, and clarify how this approach could be implemented in
