EEG criteria for nonconvulsive status epilepticus

Nonconvulsive status epilepticus (NCSE) represents a prolonged state of seizure(s) (status epilepticus—SE) without marked motor manifestations. NCSE typically encompasses an ictal impairment of cognition, subtle facial or limb twitches, eyes-open mutism, head or eye deviation, automatisms, and behavioral change. Owing to the pleomorphic clinical features that are consistent with diagnoses other than NCSE, confirmation of NCSE requires an EEG showing seizure activity. Further, the determination of SE on EEG is subjective. In most publications on these entities, authors make the “leap” to a diagnosis of SE without providing objective numeric criteria of frequency, amplitude, morphology, and evolution of EEG characteristics. Hence, “I know it when I see it.” Illustrative figures are often absent, may be ambiguous, or show the most evident cases. With these caveats in mind, this paper will review literature on EEG patterns reported to represent NCSE, explore borderline para-ictal patterns, comment on EEG response to benzodiazepines (BZPs), and association with seizures; and offer the author's opinions. NCSE is traditionally divided into (a) focal NCSE (complex partial status epilepticus—CPSE), or (b) generalized NCSE (GNSE) often called absence status. Typical EEG features are: typical spike-and-wave (TSW) at 3–3.5 Hz, rhythmic, generalized, synchronous, and symmetric; atypical spike and wave (ASW) which lack one or more of these features, e.g., frequency <3 Hz or asymmetric; multiple spike-and-wave (MSW)-repetitive complexes of two or more spikes followed by a slow wave; and rhythmic delta with intermittent spikes (RDIS)-high amplitude, repetitive, rhythmic, focal, or generalized delta activity with intermixed spikes or sharp waves (Granner and Lee 1994). These discharges may be continuous; persistent with brief pauses of a few seconds; or intermittent (with pauses lasting several seconds or more). GNSE should have no significant EEG lateralization and no history of focal epilepsy. GNSE patterns are varied (Thomas et al., 1992) and may be more ambiguous, with more blunted morphologies at <1.5 Hz resembling TWs (Kaplan, 1996a, 1996b). Over time, these patterns may evolve in morphology, amplitude and frequency, wax and/or wane, often with periods of normal or more normal background recording. In one study, most GNSE were ASW, had frequencies of 2.2 ± 0.6 Hz, with a persistent or continuous pattern (Granner and Lee, 1994). Young et al. (1988, 1996) have enlarged the primary EEG criteria of frequency and morphological features in NCSE to include context and evolution. A large minority of NCSE tracings lack pathognomonic features. In these borderline situations, it is often the rapid clinical and EEG regression shortly after parenteral BZPs that confirms the ictal nature of the disorder, yet epidemiological norms might exclude response to treatment from a definition of a disease. The EEG borderlands of what represent seizures versus what are postictal or interictal patterns, periodic epileptiform discharges (PEDs), and triphasic waves (TWs), have been the subject of some study (Pohlmann-Eden, 1996; Young et al. 1996; Chong and Hirsch, 2005) with incomplete consensus. PEDs have been referred to as an irritative pattern found in temporal proximity to seizures proper; the footprints rather than the animal itself. Such distinctions may dichotomize diagnosis, prognosis, and hence the intensity of patient clinical management. PED nomenclature, and clinical and EEG categorization have largely arisen from case series reports, which are subject to “lumping,”“splitting,” and often the absence of a control population. PEDs that may occur without frequent clinical motor correlates include (Brenner and Schaul, 1990) PLEDs, BIPLEDs, PLEDS-plus, GPEDs, and more recently SIRPIDs. Periodic lateralized epileptiform discharges (PLEDs) are surface-negative bi-, tri-, or polyphasic discharges consisting of spike, sharp, polyspike components, variably with slow-wave complexes lasting 60–600 msec (mean 200 msec), of 50–150 uV (sometimes 300 uV) in amplitude, usually occur at 0.5–2.0 Hz (ranging from 0.2 to 3 Hz). They must last a minimum of 10 min, and typically, the entirety of a 20-min recording. A total of 83–87% of patients had seizures in the course of the illness (Chatrian et al., 1964; Snodgrass et al., 1989). PLEDs usually occur 1–4 days after clinical seizure activity. In patients with bilateral independent synchronous PLEDs (BIPLEDs), 78% had seizures (de la Paz and Brenner, 1981). With PLEDs-plus (PLEDs with transitional rhythmic discharges), 74% had seizures (Reiher et al., 1991). With generalized periodic discharges (GPEDs), 32–90% had seizures (Husain et al., 1999; Yemisci et al., 2003), and those consistent with NCSE had higher amplitude (110 v 80 uV); duration (0.5 v 0.3 sec); and inter-GPED amplitude (34 uV v 17 uV). They were viewed as “end-stage SE” (Husain et al., 1999). An aggregate of case series (Snodgrass et al., 1989) documents the relationship between PLEDs and seizures (concordance is 74–90%) and between PLEDs and SE (10–66%); 94% in hospitalized patients, leading to their conclusion also that PLEDs were “equivalent to the terminal phase of SE”. Stimulus-induced rhythmic periodic or ictal discharges (SIRPIDs) described in the critically ill also lie inexactly along an ictal–interictal continuum when cases lack a motor clinical correlate, but perhaps edge more toward a determinable ictal phenomenon when SIRPIDs can be elicited together with facial or limb jerking (a minority of patients) (Hirsch et al., 2004). Clearly this determination is arbitrary to the degree that seizures in motor (frontal) areas may induce more apparent clinical features. Conversely, other authors opine that PLEDs are not ictal, and distinguish nonictal (but epileptiform) PLEDs from seizures by their bi-, tri-, or polyphasic morphology; and their relatively static, nonevolving rhythmicity (Young et al., 1988; Pohlmann et al., 1996). Other articles state that PLEDs in the context of SE are ictal by analyzing the temporal relationship between clinical seizures, the timing of PLEDs, the clinical course, the PLED response to BZPs; and an EEG evolution of frequency and amplitude change, terminating in a more typical seizure pattern (Garzon et al., 2001). Depending on their context, PEDs may represent different points along an ictal–interictal continuum (Pohlmann-Eden et al., 1996; Chong and Hirsch, 2005). Epileptic encephalopathies such as Lennox-Gastaut, Landau-Kleffner, and ESES syndromes include patients with clinical and EEG patterns that appear to merge and blur the ictal/interictal distinction. This increases the importance of comparison to “baseline” recordings and of observable clinical changes. TWs comprising bursts of moderate- to high-amplitude (100–300 uV) rhythmic complexes, usually at 1–2 Hz, occurring in clusters, consist of a blunted, low-amplitude initial negative-phase; a dominant positive second phase and slow rising; and a slow-wave-appearing third phase. TWs appear in toxic and metabolic encephalopathies (TMEs); and frequently occur in the context of behavior, cognition, tone, and motor disturbances clinically indistinguishable from NCSE. TMEs include hepatic and renal insufficiency; intoxication with lithium, baclofen, tiagabine, ifosfamide, cefapime; neuroleptic and serotonin syndromes. TWs may be distinguished from NCSE pharmaco-therapeutically, when EEG rhythmic activity resolves after BZPs, but without concomitant clinical improvement. This lack of clinical improvement may be due to the underlying encephalopathic state causing the TWs, rather than to conclusive evidence of a nonictal state. Recent study provides working EEG distinctions between TWs and NCSE (Boulanger et al., 2006). Comparing TWs with NCSE, there is EEG concordance with a clinical diagnosis of NCSE when the epileptiform discharges were of higher frequency (mean = 2.4 vs 1.8Hz), shorter phase one duration, multispikes (69% vs 0%), and less background slowing (15.1% vs 91.1%). TWs had a predominant phase two (40.8% vs 0%) and phase lag in 40.8% vs 0% in NCSE; and were increased with stimulation in 51% vs 0% in NCSE. These distinctions remain in contention. Recent work by an ACNS working group has developed nomenclature categorizing the above spectrum of epileptiform patterns to facilitate multicenter research on the significance of these patterns (Hirsch et al., 2005). In so doing, it is hoped that older terms (e.g., TWs, PLEDs) will cede to research-usable terminology free of prior assumptions. The recent studies noted above have tendered criteria for the EEG diagnosis of NCSE, and may contribute toward moderating semiological borderline disputes among NCSE, PEDs, and TWs. The addition of clinical context, temporal EEG evolution, and both clinical and EEG response to BZPs appear warranted. To resolve and refine the objective EEG criteria for NCSE, the nomenclature for PEDs and TWs should be replaced by standardized descriptors (Hirsch et al., 2005). The Table incorporates and modifies these diagnostic features abstracted from the literature. This synthesis may generate working criteria for future prospective studies that might, in turn, address these diagnostic Gordian knots.