Clinical cutoffs for plasma biomarkers to identify intracranial pathology in traumatic brain injury patients with Glasgow Coma Scale scores 13 to 15: a TRACK-TBI study.

Objective

The authors sought to identify plasma glial fibrillary acidic protein (GFAP) and ubiquitin c-terminal hydrolase-L1 (UCH-L1) cutoffs corresponding to a range of high positive predictive values (PPVs) for detecting intracranial pathology on CT and detecting pathology on MRI (when CT is negative) in patients who present to a level I trauma center for TBI evaluation with Glasgow Coma Scale (GCS) scores of 13-15.

Methods

This is a secondary analysis of prospectively collected data from the Transforming Research and Clinical Knowledge in Traumatic Brain Injury (TRACK-TBI) study, which is a longitudinal, observational study of patients with TBI who presented to one of 18 US level I trauma centers within 24 hours of injury. Blood samples were collected within 24 hours. Plasma samples were analyzed for GFAP (pg/mL) and UCH-L1 (pg/mL). The CT or MRI scan was considered positive (CT+, MRI+) if there was evidence of acute intracranial pathology related to TBI. Receiver operating characteristic curve analysis (AUC) was performed to evaluate the ability of day-1 GFAP and UCH-L1 to discriminate CT+ versus CT-, and CT-/MRI+ versus CT-/MRI-. A K-fold cross-validation procedure was used to identify GFAP cutoffs for predicting CT+ among patients with GCS scores of 13-15 and for predicting MRI+ among patients with GCS scores of 13-15 and CT- subjects, based on the adjusted PPV thresholds.

Results

Among the 1718 TBI participants with GCS scores 13-15, 633 (37%) had CT+ findings. Of the 1085 CT- participants, 608 underwent MRI at 2 weeks postinjury, of whom 154 (25%) had MRI+ findings. Day-1 GFAP had an area under the curve (AUC) of 0.82 (95% CI 0.80-0.84), and UCH-L1 had an AUC of 0.60 (95% CI 0.57-0.63) for discriminating CT+ from CT-. Day-1 GFAP had an AUC of 0.76 (95% CI 0.71-0.81) while UCH-L1 had an AUC of 0.60 (95% CI 0.55-0.65) for discriminating CT-/MRI+ from CT-/MRI-. For CT+ prediction, assuming a prevalence of 0.36 (from our data) and 0.1 (from the literature), the median GFAP cutoffs to achieve an adjusted PPV of 0.9 were 2879 (95% CI 2195-4704) and 5191 (95% CI 3779-5426), respectively. For MRI+ prediction among GCS 13-15 CT- patients, the median GFAP cutoffs were 1382 (95% CI 1145-2424) to achieve an adjusted PPV of 0.9 and 2317 pg/mL (95% CI 1262-2551 pg/mL) to achieve an adjusted PPV of 0.98.

Conclusions

These threshold values can be used as an aid-in-diagnosis tool and assist in appropriate triaging of patients who have sustained TBI to urgent neuroimaging, potential escalation in care, and interventions as clinically indicated.