Intracranial pressure in traumatic brain injury is dynamic and influenced by factors like injury patterns, treatments, and genetics. Existing studies use time invariant summary intracranial pressure measures thus potentially losing critical information about temporal trends. We identified longitudinal intracranial pressure trajectories in severe traumatic brain injury and evaluated whether they predicted outcome. We further interrogated the model to explore whether ABCC8 polymorphisms (a known cerebraledema regulator) differed across trajectory groups.Design:
Prospective observational cohort.Setting:
Single-center academic medical center.Patients:
Four-hundred four severe traumatic brain injury patients.Interventions:
None.Measurements and Main Results:
We used group-based trajectory modeling to identify hourly intracranial pressure trajectories in days 0–5 post traumatic brain injury incorporating risk factor adjustment (age, sex, Glasgow Coma Scale 6score, craniectomy, primary hemorrhage pattern). We compared 6-month outcomes (Glasgow Outcome Scale, Disability Rating Scale, mortality) and ABCC8 tag-single-nucleotide polymorphisms associated with cerebral edema (rs2237982, rs7105832) across groups. Regression models determined whether trajectory groups predicted outcome. A six trajectory group model best fit the data, identifying cohorts differing in initial intracranial pressure, evolution, and number/proportion of spikes greater than 20 mm Hg. There were pattern differences in age, hemorrhage type, and craniectomy rates. ABCC8 polymorphisms differed across groups. GOS (p = 0.006), Disability Rating Scale (p = 0.001), mortality (p < 0.0001), and rs2237982 (p = 0.035) differed across groups. Unfavorable outcomes were surprisingly predicted by both low intracranial pressure trajectories and sustained intracranial hypertension. Intracranial pressure variability differed across groups (p < 0.001) and may reflect preserved/impaired intracranial elastance/compliance.Conclusions:
We employed a novel approach investigating longitudinal/dynamic intracranial pressure patterns in traumatic brain injury. In a risk adjusted model, six groups were identified and predicted outcomes. If validated, trajectory modeling may be a first step toward developing a new, granular approach for intracranial pressure phenotyping in conjunction with other phenotyping tools like biomarkers and neuroimaging. This may be particularly relevant in light of changing traumatic brain injury demographics toward the elderly.