T60. Continuous EEG after moderate to severe traumatic brain injury: Seizures, ictal-interictal patterns and neurophysiologic predictors of functional outcome

Abstract

Traumatic brain injury is common and the leading source of disability in the US. International guidelines for the use of EEG in those requiring ICU admission include indications to detect nonconvulsive seizures in those with altered mental status or in those at high-risk, although evidence is lacking for which patients might be at risk. Further, there is no evidence for the prognostic importance of continuous EEG (cEEG) in the context of established clinical indicators. We performed a post hoc analysis of data from the INTREPD2566 study (NCT00805818), a randomized controlled clinical trial of NNZ-2566, a synthetic neurotrophic IGF-1 derivative for moderate to severe TBI, defined as a Glasgow Coma Scale (GCS) score 4–12. This study failed to achieve its secondary end-point of modifying EEG. Per protocol, all patients received study drug or placebo within 8 h of injury; cEEG was initiated as soon as possible and continued for 72 h. Glasgow Outcome Scale-Extended data were collected at 3 months and dichotomized as good (GOSE 5–8) or poor (GOSE 1–4). A total n = 261 were initially enrolled; n = 38 were subsequently excluded from the study and n = 141/222 (63.5%) underwent cEEG monitoring. The mean age was 34 ± 15 years and 127/141 (90.1%) were men. The median post-resuscitation GCS was 7 (Interquartile Range 6–8) with a mean Injury Severity Score (ISS) of 23.6 ± 13.3. Mortality was 8.5% (12/141) at 3 months. The mean time from trauma to cEEG was 13.8 ± 9.8 h. Of those with data at 72 h, n = 24/120 (20%) had electrographic seizures, lateralized periodic discharges, or lateralized rhythmic discharges (SZ/LD); 4/120 had seizures alone. Only the presence of mass lesion on CT was associated with SZ/LD (OR 4.0, 95 CI 0.05–0.58; p = 0.01). Although n = 3/4 patients with seizures had poor outcome, there was no statistical association with outcome (OR2.9, 95 CI 0.22–154.60; NS); similarly, SZ/LD were observed equally between those with good and poor outcome. By contrast, the presence of a posterior dominant rhythm or sleep transients were associated with good outcome, while background delta frequencies were associated with poor outcome; superimposed background alpha/beta and burst-suppression (usually associated with sedation) were not associated with outcome. On multivariate analysis, only the ISS (OR 1.05, 95 CI 1.02–1.09) and an absence of sleep transients within 24 h (OR 4.61, 95 CI 2.14–9.94) were independently associated with poor outcome; the odds for poor outcome increased to 6.19 (95 CI 2.54–15.07) at 72 h without sleep transients. In this prospective cohort, we confirmed the incidence of seizures and focal periodic or rhythmic discharges and found an association with mass lesions on initial imaging. While seizures did not predict poor functional outcome, we found an independent association with the absence of identifiable sleep transients at 24 and 72 h.