Periodic discharges (PD) are thought to represent synchronized, aberrant firing of dysfunctional populations of cortical neurons with prolonged recovery times, or refractory periods, analogous to that observed after the paroxysmal depolarizing shift that characterizes epileptiform discharges [6–8]. In a model of the hippocampal CA3 region, spontaneous excitatory postsynaptic potentials (EPSPs) trigger neuronal bursting by depolarization of postsynaptic neurons in an excitable post-refractory state or even (if the quantity of EPSPs is great enough) neurons still in a refractory state [9]. Decreased interneuronal inhibition of adjacent neuronal column coupling, resulting in excess excitation, may result from frequent depolarizations [10] and manifest as periodic, phase-locked, super-positioned bursts [11]. By contrast, increased inhibitory tone may make otherwise continuous ictal activity appear periodic [12]. In cortical slice models, a silent interval between discharges has been associated with discharge-induced extracellular alkalization and potassium concentrations with intracellular acidification, inhibiting synchronized neuronal firing by uncoupling gap junctions. Interestingly, when the intracellular acidification was prevented, periodicity was replaced with continuous ictal activity [13]. More recently, the observed spatial dynamics of ictal discharges suggest that small cortical areas of tonic firing (termed the ictal wavefront) generate much larger regions of synchronized, rhythmic discharges distinct from the source of ictal activity [14, 15] and that these lower frequency discharges counterintuitively reflect desynchronized neuronal bursting and eventual seizure termination. For this reason, periodic discharges may be recorded on scalp EEG as a result of either multifocal or aberrantly propagating ictal wavefronts, or as a result of increasing refractoriness to a persistent ictal focus. Evidence exists that smaller foci of ictal activity may underlie surface EEG that demonstrates only interictal, periodic, or rhythmic discharges (Fig. 5.2) [16, 17].

Periodic or rhythmic discharges in some cases may not be ictal, but a manifestation of neuronal death [18, 19] or of an anatomic disconnection between cortical and subcortical regions [5], both of which create conditions for abnormally synchronous cortical oscillations and may contribute to cortical hyperexcitability. Animal studies have demonstrated that while the application of excitatory substances to intact cortex produces irregular rhythmic bursts, much more regular rhythmic bursts occur with thalamectomized cortex [20]. Moreover, nearly regular periodic bursts with interburst suppression are seen upon stimulation of completely undercut perfused cortex. In humans, one case-control study found radiographic evidence of white matter changes and subcortical atrophy (but not cortical atrophy) associated with encephalopathy and generalized periodic discharges with triphasic morphology [21]. Autopsy-based studies, however, have demonstrated that periodic discharges may be recorded distant to a pathologic lesion or from intact cortex [22]. After generalized convulsive status epilepticus (GCSE), periodic discharges are often observed on an attenuated background [23, 24], and others have suggested that periodic discharges particularly represent a “fatigued” state of clonic firing [25]. Periodic discharges have also been recorded in ischemic penumbra of animals undergoing occlusion of the middle cerebral artery [26]; contralateral to the occlusion, rhythmic delta activity was observed, suggesting that periodic and rhythmic discharges reflect broader changes in the balance between inhibition and excitation. Whether periodic discharges reflect aberrant ictal activity or an alteration in inhibitory and excitatory cortical inputs, cortical dysfunction remains the common denominator. Direct cortical injury, such as hypoxia, or widespread metabolic dysfunction, such as acute hepatic failure, may equally result in a similar EEG phenotype, a concept reflected in the term “interictal-ictal continuum,” used to describe periodic or rhythmic patterns that may or not be ictal depending on their complex, underlying contributors [27].

Both periodic and rhythmic discharges are highly associated with the development of clear, unequivocal seizures. In a retrospective analysis of 625 consecutive hospitalized patients undergoing >18 h of C-EEG, periodic discharges were recorded during the first 30 min in 10% (60/625) of patients, 27% (16/60) of whom had a seizure at some point thereafter [28]. Of 570 hospitalized patients undergoing C-EEG, lateralized periodic discharges (LPDs) were associated with seizures only after >24 h of monitoring, with an odds ratio of 3.1 [29], while in a case-control study of generalized periodic discharges (GPDs), only those with GPDs exhibited seizures after >48 h of monitoring compared with controls without GPDs [30]. Currently, guidelines from the Neurocritical Care Society recommend at least 48 h of C-EEG for comatose patients to rule out NCSE [31]. A 2010 survey found that the majority of adult and pediatric neurologists would monitor C-EEG in patients with periodic discharges for >24 h (40% for 24 h, 29% for 48 h, and 15% for 72 h) [32]. We recommend up to 48 h of C-EEG for all hospitalized patients with newly diagnosed periodic or rhythmic discharges where resources are available. Where C-EEG is not available, we recommend repeated EEG evaluation for patients with periodic or rhythmic discharges who exhibit intermittent or persistent changes in neurologic examination.

Focal or multifocal regions of periodic discharges are termed LPDs, bilateral independent periodic discharges (BIPDs), or multifocal periodic discharges (MfPDs). LPDs were initially described as repetitive, lateralized, spike- or sharp-wave complexes followed by a slow-wave and lasting from 0.3 to 2 s with repetition rates of 0.2 to 3 Hz [22, 25, 27, 33]. LPDs may be focal or spread broadly over one hemisphere, sometimes with extension to the homologous area on the contralateral side (albeit to a lesser extent). Most often, LPDs and BIPDs have maximum voltage in the frontocentral regions [34]. LPDs tend to vary from patient to patient in their periodicity (0.5–4 s) [27], morphology (e.g., biphasic, triphasic, or polyphasic spike- and sharp-waves) [35], voltage (50–300 μV), and duration (60–1000 ms) [36]. LPDs do not always exist continuously over the entirety of a recording, although LPDs have persisted for months and even years [33, 37, 38]. More commonly, LPDs dissipate between 9 and 16 days (range 1–31 days) after clinical symptom onset, even in the face of progressive neurologic disease (e.g., progressive brain tumor) [22, 25, 33, 39]. Their natural progression proceeds with decreasing frequency, prolongation of discharge duration, and evolution to paroxysmal delta and finally theta frequency slowing [33].

LPDs have been described in 0.4–1% of routine EEGs [27] and in 6–22% of C-EEG series [29, 40, 41]. Clinically, LPDs are often accompanied by focal neurologic deficits and alterations in mental status [33]. LPDs may be the electrographic correlate of focal seizures, such as epilepsia partialis continua [42], focal sensory-motor seizures [43], or complex partial seizures [22, 33, 44]. More often, LPDs occur in association with an acute or subacute focal neurologic disease, usually involving the cortex, capable of generating symptoms independent of LPDs [38, 45]. Table 5.2 lists etiologies associated with LPDs [22, 25, 33, 35–38, 45–64]. The most common etiology underlying LPDs is ischemic stroke [22, 33, 35, 65, 66]. In a meta-analysis of 586 patients from several earlier studies of LPDs (1964–1987), 35% had a primary etiology of ischemic stroke [25]. Interestingly, one in ten had a metabolic etiology, suggesting a role for metabolic stressors in lowering the threshold for LPDs to develop in much the same way that the seizure threshold may be lowered in patients with epilepsy. Studies utilizing C-EEG have confirmed that one-quarter have remote or progressive brain disease [67]. Some have focused on specific disease processes: LPDs were seen in 13% of those with intracerebral hemorrhage (ICH), particularly when bleeding was closer to the cortex (<1 mm) and when ICH was greater than 60 cc 24–72 h post-bleed, and less frequently in the presence of midline shift 3–7 days post-bleed [68]. LPDs were also observed in 20% of those with subarachnoid hemorrhage (SAH) [69]. Within 24–48 h of suspected herpes simplex encephalitis, LPDs were observed in 90% with positive HSV PCRs (at symptom onset) and 30% with negative HSV PCRs; the sensitivity of EEG decreased after 48 h [70]. Even cohorts of patients undergoing C-EEG may under-estimate the frequency of periodic discharges in those with critical neurologic illness such as traumatic brain injury or subarachnoid hemorrhage, because scalp EEG recordings can miss some focal periodic discharges seen only on direct cortical recordings [16, 17]. For instance, in a series of patients with traumatic brain injury (TBI), 38.2% (13/34) had periodic discharges recorded during electrocorticography, of whom 8/13 had no periodic discharges on surface EEG [17].

LPDs are frequently associated with seizures, found in 49–90% of patients with LPDs [25, 27, 46], although in many studies using routine EEG, recordings were not made until a seizure was suspected clinically, likely enriching the cohort of patients with LPDs to include those at highest risk for seizures. In more recent C-EEG series, 27–49% of those with LPDs on C-EEG developed seizures [28, 29, 40]. The addition of “plus” characteristics (superimposed faster frequencies or rhythmic waveforms in conjunction with LPDs) increases the risk for seizures. LPDs+ account for between 17 and 60% of all LPDs, and LPDs+ and BIPDs+ are associated with seizures in 85 and 100% of cases, respectively, [36, 46].

BIPDs have been described since the earliest mention of LPDs [22], but there have been few studies of BIPDs, in part because they are less commonly recorded. BIPDs have been reported in approximately 0.1% of patients undergoing routine or outpatient EEG, and in 1–5% of C-EEG [34, 46, 102]. Table 5.4 lists the etiologies associated with BIPDs. Whereas LPDs have a stronger association with focal seizures (albeit coexisting at times with BIPDs), BIPDs have a greater association with generalized seizures [102]. BIPD+ patterns are invariably associated with seizures [46]. Clinically, BIPDs are less often associated with focal neurologic deficits and more with coma, likely as a result of bilateral cortical involvement. While BIPDs have traditionally been associated with substantial mortality (52–61% according to most series [46, 102]), thus far there have been no dedicated case-control series to confirm an independent association. The substantial mortality seen in patients with BIPDs likely reflects the severity of the underlying etiology, but more than 20% of patients with BIPDs in one series were living independently at a mean follow up of 18 months [34], and relatively benign BIPDs have been documented after bilateral strokes [103]. When controlling for disease severity after SAH or ICH, BIPDs have not been found to confer an independent risk for poor outcome, in contrast to LPDs [68, 69].

GPDs are described as synchronous, bihemispheric periodic discharges [30]. GPDs may co-occur with LPDs (21.5%, vs. 10% of controls matched for age, etiology, and level of arousal) and BIPDs (10.5% vs. 1.5%) [30]. Clinically, GPDs are associated with lethargy or coma in 92% of cases. Myoclonus is sometimes seen in conjunction with specific disease processes associated with GPDs, such as Creutzfeldt–Jakob disease (CJD) or anoxic ischemic injury. Other causes of GPDs are listed in Table 5.5 [30, 34, 44, 60, 104–112]. GPDs have been variably subclassified. (“triphasic morphology,” the most common subtype of GPDs, will be discussed separately.) Prior reviews have distinguished periodic short-interval diffuse discharges (PSIDDs), which occurs every 0.5–4 s, from periodic long-interval diffuse discharges (PLIDDs), which are polyphasic and occur every 4–30 s [113]. The standardized ACNS terminology regards polyphasic discharges as bursts with >4 phases, and PLIDDs would now be classified as either a burst–suppression pattern or, when background activity is preserved, continuous with high-amplitude bursts [4]. PLIDDs have a close association with subacute sclerosing panencephalitis (SSPE), a now rare form of progressive degenerative post-measles encephalitis, and have been observed with ketamine or PCP toxicity. While anoxia is clearly associated with burst–suppression on EEG, “burst–suppression with identical bursts” is a regular, recurrent polyphasic bursting pattern seen after severe diffuse anoxia and may be considered part of the PLIDD category. A retrospective blinded study of 101 cardiac arrest patients following return of spontaneous circulation found this pattern in 20, all of whom had poor six month neurologic outcome (CPC 3–6), compared with 10 of 28 patients with poor outcome and a more typical suppression–burst pattern [114]. This chapter will focus on short-interval GPDs.