Prospective comparison of capillary and venous brain biomarker S100B: capillary samples have large inter-sample variation and poor correlation with venous samples

Abstract Background Guidelines for the emergency management of mild traumatic brain injury have been used for over a decade and are considered safe. However, they recommend computerized tomography for at least half of these patients. The Scandinavian Neurotrauma Committee guideline uses serum S100B protein level to rule out intracranial hemorrhage. Analysis of capillary serum S100B protein level has not yet been employed for this purpose. The primary aim of this study was to investigate the correlation and agreement of capillary and venous serum S100B protein level over a spectrum of concentrations typical for mild traumatic brain injury. Methods Eighteen patients with traumatic intracranial hemorrhage and 39 volunteers without trauma to the head within the past 7 days were recruited. Blood was sampled from patients with intracranial hemorrhage daily up to four consecutive days and healthy volunteers were sampled once during the study. One venous and two capillary samples were drawn at each sampling event. Samples were analyzed using the Cobas e411 S100 electrochemiluminescence assay. Results Median serum S100B protein level of capillary sampling 1 was 0.12 (IQR 0.075–0.21) μg/l and median serum S100B protein level of capillary sampling 2 was 0.13 (IQR 0.08–0.22) μg/l. Median serum S100B protein level of all venous samples was 0.05 (IQR 0.03–0.07) μg/l. Correlation plots of capillary and venous samples showed poor correlation and Bland-Altman plots showed a large dispersion of samples and wide limits of agreement. Conclusion The results of this study indicate that correlation and agreement between capillary and venous samples are low, and because of this, we cannot recommend studies on capillary serum S100B protein level to rule out intracranial hemorrhage in mild traumatic brain injury. Given the limitations of the current sampling and analysis methods of capillary protein S100B protein level, we conclude that evaluating its predictive ability to rule out intracranial hemorrhage should be withheld until more reliable methods can be incorporated into the study design.