Prognostic use of cognitive event-related potentials in acute consciousness impairment

Abstract

Electroencephalography (EEG) provides a powerful tool for evaluating the extent of intact cognitive functions in comatose patients at the bedside, mostly by recording responses to sensory stimuli, exemplified by the so-called event-related potentials (ERPs). Different ERP paradigms are informative of various levels of cognitive functions, ranging from basic auditory processing and auditory discrimination (i.e., mismatch negativity (MMN)) to novelty detection and detection of complex sound sequences. Among them, the MMN paradigm has proven especially useful, as the presence of an MMN response has been repeatedly associated with a favorable clinical prognosis. The high predictive power of MMN is basically driven by the absence of such a response in patients who fail to regain consciousness, mainly assessed several weeks or months after coma onset. However, the use of this paradigm in clinical routine has been limited so far, possibly due to the difficulty of assessing the presence of an MMN at the level of single patients. Multivariate EEG decoding methods provide a powerful tool for quantifying the degree of auditory discrimination, at the single-patient level with minimal a priori inclusion criteria, with very promising results in terms of prognostication of awakening. Moreover, recent evidence in patients in acute postanoxic coma treated with therapeutic hypothermia shows that auditory discrimination might still be intact during the first days of coma, irrespective of patients’ outcome. Here, we propose a general framework for assessing the degree of auditory discrimination over time, as a process that degenerates in non-survivors and that improves in patients who eventually awake.