Nonlinear EEG changes in postanoxic encephalopathy

Application of concepts from nonlinear dynamics to EEG has suggested the existence of two different types of pathophysiology in diffuse encephalopathies. The first is characterized by decreased cortical activation. The EEG is then linear stochastic. The second is characterized by neuronal hyperexcitability and hypersynchronous oscillations. The corresponding EEG contains nonlinear deterministic structure. We examined the hypothesis that anoxic encephalopathy belongs to the second category. EEGs were recorded in 20 healthy controls (mean age 63.4 year; SD 3.7; 12 females, 8 males) and 15 patients (mean age 65.6 year; SD 17.7; 5 females, 10 males) with postanoxic encephalopathy following an episode of cardiac arrest. EEG epochs (16 seconds; sample frequency 250 Hz; 16 bit A-to D) were examined for nonlinear structure with the algorithm of MNLCP (multi channel nonlinear cross prediction). This algorithm characterizes each epoch in terms of its predictability pred, amplitude asymmetry ama and time irreversibility tir. Significant amplitude asymmetry or time irreversibility indicate nonlinear dynamics. EEGs of patients were significantly better predictable and more amplitude asymmetric than those of controls. Stronger evidence for nonlinear structure corresponded with less reactivity of the EEG to painful stimuli and a worse outcome. Evidence for nonlinear brain dynamics in postanoxic encephalopathy was found. This lends further support to the concept of 'linear stochastic' and 'nonlinear deterministic' encephalopathies characterized respectively by decreased cortical excitability and increased neuronal excitability and their associated EEG patterns. Nonlinear EEG analysis may provide information which has pathophysiological relevance. Also, nonlinear dynamics is associated with a worse prognosis in anoxic encephalopathy, which may be clinically relevant.