Description

Circadian rhythms and epilepsy. The endogenous circadian system is an important modulator of the sleep-wake rhythm. The circadian oscillator is located in the suprachiasmatic nucleus of the anterior hypothalamus and, independently of other factors, potentiates wakefulness at one phase of the diurnal cycle and facilitates sleep at the opposite phase. This autonomous pacemaker determines sleep propensity in relation to other biological functions and, in particular, to the rhythm of deep body temperature. Laboratory studies have demonstrated that the timing of peaks and troughs of sleep propensity during the day is highly correlated with the thermal curve: besides the high sleep propensity during the nocturnal hours (primary sleep gate concomitant to the descending branch of deep body temperature), a less prominent peak of sleep propensity is observed in the midafternoon (secondary sleep gate), about 10 hours after the temperature minimum clock time. In contrast, besides the trough of sleep propensity in the late morning (ascending branch of deep body temperature), there is a period of low sleep propensity in the early evening (forbidden zone for sleep), about 8 hours before the temperature minimum time (36).

Homeostatic process. The variations of sleep propensity over time are not only an effect of the biological clock but are also influenced by the subject’s prior sleep-wake history. Extensive studies on sleep deprivation have ascertained a positive relationship between the duration of pre-sleep wakefulness and the spectral energy enhancement of EEG slow-wave activities (0.5 to 4 Hz). According to the synaptic homeostasis hypothesis, slow-wave activities, which reflect NREM sleep intensity, are an active mechanism that restore the metabolic and functional efficiency of the brain through the proportional reduction (downscaling) of synaptic weights after the synaptic potentiation gathered during wakefulness (109). The slope of slow waves during NREM sleep best reflects this “downscaling.”

Sleep stages 3 and 4, ie, N3 according to the new scoring criteria (38), are mostly concentrated in the early portions of sleep, whereas they are practically absent in the final hours. The priority of recuperation of N3 after sleep onset and the progressive decline of N3 along the night suggests the involvement of a compensatory mechanism based on the accumulation when awake of some unknown factor, which undergoes a sort of dissipative process during sleep. In short, prolonged wakefulness hastens sleep onset and proportionally potentiates slow-wave activities, regardless of the circadian phase (19). The increased intensity of sleep after prolonged wakefulness indicates that the characteristics of sleep recovery respond to mechanisms of homeostatic regulation.

NREM and REM sleep: the ultradian cycle. A third mechanism of sleep regulation is the NREM-REM cycle, characterized by periods of sustained high-voltage, slow-wave synchronized EEG patterns (NREM sleep) that are periodically replaced by sustained periods of low-voltage, fast-wave desynchronized EEG rhythms (REM sleep). The NREM-REM cycle is conventionally composed of a descending branch (from light to deep NREM sleep), a trough (the deepest stage of the sleep cycle), and an ascending branch (from the deepest NREM stage to REM sleep). Experimental investigation has ascertained that the intrinsic alternation between NREM and REM sleep is under the control of an oscillatory process generated by a particular rhythm of neurotransmission and by the reciprocal interaction between two neuronal groups with REM-on and REM-off activities (51).

The two main sleep states have different physiological components and contrasting effects on generalized ictal and interictal discharges. The hypothalamic and brainstem generators of sleep and arousal have diffuse ascending and descending projections (41) that give rise to a number of distinguishing physiological characteristics, called “components,” and influence the likelihood that an electrographic or clinical seizure will occur. The most salient state-specific tonic and phasic components affecting epilepsy seem to be the degree to which cellular discharge patterns are synchronized and alterations in antigravity muscle tone (11; 90).

Sleep microstructure: the cyclic alternating pattern (CAP). Polysomnography is the major method of sleep analysis and the main diagnostic tool in sleep medicine. The standard interpretation of polysomnography recordings describes their macrostructure in terms of sleep stages, defined according to the conventional scoring criteria. Sleep stages are regarded as stable referential states, each corresponding to a specific depth of vigilance. However, sleep is an extremely dynamic process with a variety of EEG phasic activities (spindles, delta bursts, vertex waves, arousals), suggesting the existence of an underlying dimension beneath the macrostructural framework (35). In NREM sleep, these transient EEG events are organized in long-lasting sequences that assume a CAP and reflect a sustained instability of the arousal level (67). CAP is the main physiological component of NREM sleep microstructure (101), and it is characterized by sequences of transient electrocortical events that are distinct from background EEG activity and recur at up to 1-minute intervals (105). CAP sequences are composed of two or more CAP cycles.

Each CAP cycle is the EEG translation of a transient activation (phase A) followed by a rebound inhibition (phase B). The complementary pattern, defined as non-CAP and consisting of a rhythmic EEG background with few, randomly distributed arousal-related phasic events, represents a stable sleep condition. Intensive though subwaking perturbation delivered during non-CAP determines the prompt appearance of a CAP sequence. Variations during CAP involve, to different degrees, muscle tone, heart rate, and respiratory activity, which increase during phase A and decrease during phase B.

On the contrary, non-CAP sleep periods are associated with regular neurovegetative activities.

Three subtypes of A phases corresponding to different levels of neurophysiological activation can be classified:

In the physiological architecture of sleep, the A1 subtypes prevail in the build-up and maintenance of deep NREM sleep, whereas the A2 and A3 subtypes dominate in light sleep that precedes the onset of desynchronized REM sleep.

CAP time is the temporal sum of all CAP sequences and can be calculated throughout total NREM sleep. The percentage ratio of CAP time to sleep time is referred to as CAP rate. In human sleep, CAP rate is a measure of arousal instability that shows a particular evolution along the lifespan (65) and correlates with the subjective appreciation of sleep quality. In particular, the higher the CAP rate, the poorer the quality of sleep.

Clinical applications

Impact of the 24-hour cycle. Several studies have investigated the influence of the vigilance states on the occurrence of epileptic manifestations, showing that the variations of arousal level across the 24-hour rhythm play an important role in the modulation of epileptic events (40). The sleep-wake cycle can significantly affect the appearance of interictal EEG discharges and epileptic seizures. Overall, epileptic manifestations are more likely to appear when the level of vigilance is low (relaxation, drowsiness, sleep) and unstable (arousal fluctuations), when frequency and morphology of epileptiform EEG patterns may depend on sleep stages. In particular, the two main neurophysiological states that characterize sleep (NREM and REM) have opposite consequences on interictal abnormalities and on critical manifestations, with NREM sleep endowed with activating properties and REM sleep exerting an inhibitory action (89).

Although sleep may modulate epileptogenicity through cortical excitation or other mechanisms, its contribution to circadian seizure patterns varies with the pathophysiology of the underlying epileptic syndrome (and above all with the localization of epileptic focus) because of the differing sensitivity of brain regions to circadian modulation (37). Both experimental epilepsy models (90) and studies in symptomatic partial epilepsies (80) provide evidence that the main vigilance states differentially facilitate seizures depending on the location of the epileptic foci. In particular, kindling models of partial epilepsy have proportionately more seizures during wakefulness compared to generalized seizure models (90). Temporal lobe seizures peak significantly from 15.00 to 19.00 hours (73). Seizures in patients with extratemporal lobe epilepsies are distributed in a pattern not statistically distinguishable from a uniform 24 hours rate, whereas seizures of patients with mesial temporal lobe epilepsy show a peak of occurrence in the mid-to-late portion of the light phase of the day (80).

In animal models of limbic epilepsy, spontaneous seizures occur predominantly during the light half of the cycle. This daily pattern persists when epileptic rats are allowed to free-run in constant darkness. Even without time cues, seizures continue to recur in an identical pattern, endogenously mediated by the circadian rhythm of body temperature (80). Other physiological cyclic functions, including the hypothalamic-pituitary-adrenal axis and the hypothalamic-pituitary-gonadal axis, might be equally important in the modulation of seizure occurrence (80). Moreover, the pattern of EEG activity may be deeply influenced by the increased engagement in physical and mental performances during the diurnal hours. In other words, the genuine distribution of EEG bursts over the 24-hour period can be obscured by several masking factors. This hypothesis can be tested by having patients suffering from epilepsy lie in bed under constant conditions and instructing them to refrain from any type of activity. The distribution of EEG paroxysms might then reveal a different picture.

Impact of sleep depth and sleep deprivation. It is known that increased depth of sleep is associated with increased interictal epileptiform discharges: NREM sleep has been shown to activate interictal epileptiform discharges in patients with partial epilepsy, with the maximal spiking rates occurring during the deeper stages (N3) of sleep and, less frequently, occurring during the lighter sleep stages 1 and 2 (48). Sammaritano and Saint-Hilaire observed maximal spiking rates during N3 in three patients with extratemporal lobe epilepsy referred for presurgical evaluation (86). In children with benign epilepsy of childhood with rolandic spikes, maximal spike rates are related to deeper stages of sleep and, in general, to the first cycle (10). However, one study carried out on children with benign centrotemporal spikes showed a lower amount of spikes per hour during N3 compared to N2 (08).

Sleep deprivation increases slow-wave activity. As a result, it may have an overall activating effect on epileptic phenomena. Rodin and colleagues noted high-voltage paroxysmal activity in the EEG of normal subjects after 120 hours of sleep deprivation and concluded that prolonged loss of sleep is associated with increased cerebral irritability, which may result in epileptic-like manifestations in certain predisposed individuals (84). Further studies showed that sleep-deprived EEG activates epileptiform abnormalities in 30% to 50% of people with epilepsy. This activation method has a high specificity. On the basis of these observations, sleep deprivation has become an established activating method to elicit EEG epileptiform activity.

Whether sleep deprivation in epileptic patients has genuine activating effects on EEG or whether it acts through sleep induction remains an open question. Some studies support the premise that sleep deprivation may activate interictal epileptiform discharges independently of the activating effects of sleep (13; 24). In juvenile myoclonic epilepsy seizures commonly occur following nights of sleep deprivation (17). In idiopathic generalized epilepsy, continuous long-term EEG monitoring indicates that sleep has an activating effect on interictal epileptiform discharge densities being the highest during NREM stages 1 and 2. Interictal epileptiform discharge densities increase in all vigilance states after sleep deprivation but remain the highest in stages 1 and 2. The authors assume that fine-graded vigilance fluctuations, which are more frequent after sleep deprivation, have an essential role in spike-wave discharges activation in superficial sleep or even in wakefulness (32).

Impact of NREM and REM sleep. During NREM sleep, virtually every cell in the brain discharges synchronously (95). Synchronous synaptic effects, whether excitatory or inhibitory, could augment the magnitude and propagation of postsynaptic responses, including epileptic discharges. Background EEG effects seem to be exacerbated by sudden surges of afferent stimulation associated with transient, synchronous phasic arousal events. Generalized seizures, particularly generalized tonic-clonic or myoclonic convulsions, tend to occur during NREM sleep or transitional arousal periods characterized by background EEG synchronization, often with phasic events that include sleep EEG transients such as sleep spindles, K-complexes, and ponto-geniculo-occipital waves. In the majority of patients with primary generalized epilepsy, frequent brief bursts of spikes, polyspikes, and spike-wave-like discharges are associated with K-complexes or spindles, which are specific phasic EEG patterns of NREM sleep (92).

In West syndrome, the characteristic high amplitude “hypsarrhythmic” interictal EEG pattern is seen most prominently in early NREM sleep (30). Although infantile spasms in West syndrome can occur during drowsiness before sleep onset, they are uncommon during sleep itself (42). In contrast, the tonic seizures of Lennox-Gastaut syndrome (LGS) occur much more frequently during NREM sleep than during wakefulness. Associated with characteristic paroxysmal fast activity on EEG, sleep-related tonic seizures are seen in over 90% of patients with LGS (26). In addition, NREM sleep modulates the interictal slow spike-wave pattern of LGS (31; 20). Sleep represents an activating background for symptomatic and cryptogenetic epilepsies. In particular, frontal lobe seizures have a strong preponderance to occur during sleep, and occur more commonly during sleep compared to temporal lobe seizures. Most individuals with frontal lobe epilepsy will have a significant proportion of their events during sleep, and many will have their seizures exclusively during sleep. When 90% of an individual’s seizures arise during sleep, the individual is said to have nocturnal frontal lobe epilepsy (78).

A rare epileptic disorder that occurs in children is the continuous spike waves during slow-wave sleep syndrome. It is diagnosed by a special type of EEG pattern called “electrical status epilepticus during sleep” (ESES), which is characterized by near-continuous spike-wave discharges at a frequency of 1.5 to 3.5 Hz on the EEG in N3 (44). The diagnostic criteria include continuous EEG abnormalities that occupy more than 85% of the slow-sleep tracing on three recordings performed over a period of more than one month. Although continuous spike waves during slow-wave sleep syndrome has been used interchangeably with ESES in the recent literature, the syndrome is more likely to be defined by a constellation of clinical signs, such as gradual cognitive and behavioral deterioration and acquired language impairment, whereas ESES is a defining EEG feature of several epileptic syndromes. Continuous spike waves during slow-wave sleep syndrome, acquired epileptic aphasia or Landau-Kleffner syndrome, and benign epilepsy of childhood with centrotemporal spikes are three different epileptic syndromes that share some similar clinical manifestations during the periods of ESES, including seizures and psychomotor disturbances, and sometimes they are regarded as forming a clinical spectrum (25).

Comparison of the time course of the slope of EEG slow waves between healthy controls and patients with ESES associated with regression or stagnation of cognitive functions showed that the expected decrease of the slope was found only in the control group from the first to the last hour of sleep (17.2% decrease, p< 0.001) (07). In contrast, ESES patients showed no significant change in slope across the night, suggesting a disruption of the downscaling process during sleep, which may contribute to the developmental regression in these children.

A typical feature of the continuous spike waves during slow-wave sleep syndrome is the nonpersistence of ESES during the REM stage of sleep. This is a typical behavior of most epileptic syndromes as REM sleep, with its asynchronous cellular discharge patterns (41) and skeletal motor paralysis, is resistant to propagation of epileptic EEG potentials and to clinical motor accompaniment (93), even though spontaneous phasic activity and focal EEG discharges persist at this time and may be evoked by photic stimulation. Although antigravity muscle tone is preserved in NREM sleep and waking, thus, permitting seizure-associated movement, profound lower motor neuron inhibition occurs in REM, creating virtual paralysis and preventing seizure-related movement.

These conclusions are supported by other experimental and clinical findings indicating that substrates of state-specific components rather than integrity of the state per se can be salient determinants of seizure propagation, regardless of epileptic syndrome. Agents that synchronize the EEG, such as cholinergic or noradrenergic antagonists, have proconvulsant effects. Conversely, agents that desynchronize the EEG discourage epileptic EEG discharge propagation. Finally, pharmacologic manipulations that induce atonia, such as carbachol infusion into the brainstem, also block clinical motor accompaniment (110).

Collectively, these findings confirm that neural cell discharge patterns and alterations in muscle tone can affect electrographic and clinically evident seizure manifestations in different epileptic syndromes (91). Phasic activity can provoke epileptiform discharges, but the extent of EEG discharge propagation and clinical motor accompaniment depend on tonic EEG and motor sleep components.

REM sleep, due to cortical neuronal asynchrony typical of this brain-state, spatially constrains epileptiform discharges and reduces their propagation away from the source generator. Accordingly, REM sleep may be helpful to localize the seizure onset zone in patients with epilepsy (52).

CAP and epilepsy.

Interictal discharges. Pioneering contributions in the early 1990s (106; 103) focused attention on the dynamic relationship between epileptic paroxysms and EEG phasic events during sleep. These studies confirmed sleep as a major physiological activator of epileptic manifestations, highlighting the relevance of arousal instability as an important triggering factor. In primary generalized epilepsy, interictal discharges are commonly activated during unstable sleep, with a number of EEG paroxysms per minute of sleep significantly higher during CAP compared to non-CAP. Phase A has a significant activation influence, whereas phase B exerts a powerful and prolonged inhibitory action, especially considering that its mean duration (16 sec) is twice the phase A length (8 sec).

Identical CAP-related influences are found in juvenile myoclonic epilepsy, but in the presence of a lower all-night amount of EEG discharges (28). Compared to normal controls, the occurrence of interictal epileptiform discharges in patients with primary generalized epilepsy has no remarkable consequences on sleep macrostructure, but it produces significant effects on arousal stability as the epileptic patients show higher CAP-rate values (103). Within the epileptic group, the CAP cycles including at least one interictal epileptiform discharge are significantly longer than those without interictal epileptiform discharges. In primary generalized epilepsy patients, the selective lengthening of CAP cycles (only those with interictal epileptiform discharges) and the increase of CAP rate support the hypothesis that CAP and interictal epileptiform discharges share common anatomical pathways and behave as a concerted pattern that links a regular physiological phenomenon (CAP) to a random pathological event (interictal epileptiform discharges). In the dynamic interplay between EEG spiking and arousal modulation, the CAP sequence (especially its activating swings) triggers the paroxysmal burst; the latter may, in turn, promote the generation of a phase A or increase the instability of sleep up to full wakefulness. Though extremely short-lived, interictal epileptiform discharges are an activating event traveling along the same pathways of normal cerebral communication. The reciprocal support of the two activating processes determines the high probability of a simultaneous occurrence of both phenomena (71). Accordingly, the low amount of arousal-related phasic events that characterizes the non-CAP condition makes this an unfavorable background for epileptic discharges.

In primary generalized epilepsy, 70% of all the phase A subtypes are A1, 24% are A2, and 6% are A3. The equivalent distribution of interictal epileptiform discharges throughout the three phase A subtypes clearly indicates that none of the A-phase subtypes plays an attractive or repulsive action on primary generalized epilepsy paroxysms. However, when the position of primary generalized epilepsy interictal abnormalities within each phase-A subtype is determined, it can be noticed that there is a striking preference for the portions characterized by EEG synchrony. In particular, the EEG paroxysms tend to occur throughout the entire length of subtypes A1 (totally expressed by EEG synchronized patterns), whereas the interictal discharges are mostly concentrated in the initial portions of the A2 and in the A3 subtypes that almost invariably start with a K-complex or a delta burst (70).

A powerful activating effect of CAP-related events has also been described for temporal lobe epilepsy with a spike frequency significantly higher in phase A compared to phase B (46). This study also revealed an intermediate activating effect of non-CAP between phase A and phase B.

Patients with focal lesional frontotemporal epilepsy show significant interictal epileptiform discharge differences between CAP and non-CAP, between phase A and phase B, and between phase A and non-CAP, but not between phase B and non-CAP. As occurs with primary generalized epilepsy, the presence of focal lesional interictal epileptiform discharges impairs the stability of sleep.

Ninety-one percent of secondarily generalized focal lesional bursts are collected in CAP, whereas 96% of all the generalized interictal epileptiform discharges found in CAP occur during phase A (106).

On the contrary, despite the high burst frequency during NREM, interictal epileptiform discharges in benign epilepsy with rolandic spikes (BERS) are not modulated by the arousal-related mechanisms of CAP. The CAP independence of BERS cannot exclude a possible relationship between interictal discharges and other microstructural events, particularly EEG features such as sleep spindles, which are generally disjointed from the CAP (especially phase A) patterns. A spectral EEG-polysomnography study in nine patients with BERS showed a significant positive correlation between interictal epileptiform discharges during sleep and sigma (12 to 16 Hz) activity (60).

According to these results, the occurrence of interictal epileptiform discharges during NREM sleep may depend on a number of complex factors, including the degree of integration between the epileptogenic focus and the neurophysiological circuits involved in the production of phasic events. Consolidated data indicate that some basic elements of CAP, particularly K-complexes and delta bursts, are generated in the thalamocortical circuits in which a pivotal role is played by the thalamic reticular nucleus. Interictal epileptiform discharges represent the epileptic variant of the complex thalamocortical system function, which is the substrate of NREM sleep EEG phenomena (34). The activation of EEG discharges could interfere with the function of these pathways. Hypothetically, in primary generalized epilepsy and frontal lobe epilepsy, the firing area and these thalamocortical circuits respectively coincide or extensively overlap, and this could explain why, in these types of epilepsy, the EEG abnormalities are significantly triggered during phase A. In functional epilepsy such as BERS, the focus is probably detached from the CAP-related mechanisms. Accordingly, CAP rate is increased in the sleep recordings of primary generalized epilepsy and frontal lobe epilepsy, even without nocturnal ictal manifestations. In contrast, CAP-rate values are reduced in BERS (08), in which interictal epileptiform discharges are associated with spindling activity that abounds during non-CAP.

Ictal events. Seizures cannot be regarded in isolation but require a process of changes in brain dynamics that start long before its manifestation. Analysis of preictal synchronizations indicates that epileptic seizures do not occur in a behavioral vacuum, but that the functioning of the brain before the seizure occurs is critical. Seizure foci are surrounded by pools of neurons functioning in local and large-scale interactions and are “pulled” into the seizure discharge once the seizure has started. The preictal period may reflect a state of increased susceptibility for pathologic synchronization, which acts as a route to the seizure. In a study carried out by Navarro and colleagues, the dynamics of the EEG signal were analyzed independently by a nonlinear measure of similarity between a reference period and successive 30 s segments of the recording (55). In order to study the spatiotemporal dynamics of the changes, the statistical significance of these deviations from the baseline was calculated, and dynamic changes up to five standard deviations were represented. The similarity indices shifted to reduced levels for several minutes before the seizure occurred. The time course of the similarity indices also showed high deviations from the baseline during the seizure and the postictal state, but the EEG dynamics then returned to a state comparable to the reference. The mean anticipation time for the 34 seizures was 7.54 ± 1.15 min. No differences according to the location of the epileptogenic focus in the brain were observed. Preictal changes were detected as efficiently from scalp EEG as from intracerebral EEG electrodes. The arousal state of the patient during the recordings did not modify the ability to anticipate seizures (16 of 19 seizures anticipated during sleep, and 18 of 22 seizures anticipated during wakefulness), and mean anticipation times in the two states were not statistically different. Finally, the type of seizure (simple partial, complex partial, secondary generalized) did not influence seizure anticipation.

These findings are in line with investigation conducted so far using the CAP parameters in epileptic disorders. In a study carried out on patients affected by focal epilepsy, 43 out of 45 nocturnal partial motor seizures occurred during NREM sleep. Among the 43 NREM seizures, 42 appeared in CAP, and always during a phase A (104). A later study that analyzed a total of 56 nocturnal partial attacks showed that all seizures occurred during NREM sleep, more frequently in CAP than in non-CAP sleep and in phase A than in phase B (p < 0.001). In addition, seizures occurring in clusters were more frequently associated with CAP sleep (p < 0.05) than isolated seizures (50). These findings indicate that a NREM sleep condition of highly fluctuating vigilance constitutes a favorable substrate for the occurrence of focal epileptic seizures.

Scoring and analysis of arousals and CAP parameters were conducted in a sleep study involving patients with temporal lobe epilepsy (TLE). Three age-matched and sex-matched groups were investigated: drug-naive temporal lobe epilepsy (20), temporal lobe epilepsy on carbamazepine (20), and healthy controls (40) (56). All groups underwent overnight polysomnography. REM arousal indices and overall CAP rates were higher in patients with temporal lobe epilepsy (group 1, p < 0.001; group 2, p < 0.001) compared to controls. Furthermore, the overall CAP rate was higher in patients on carbamazepine. Finally, an increase in the percentage of CAP subtypes A2 emerged in patients with temporal lobe epilepsy (group 1, p < 0.011; group 2, p < 0.011). The authors suggest that antiepileptic drugs, such as carbamazepine, may augment arousal instability in patients with temporal lobe epilepsy, and hence worsen sleep quality and continuity.

The same authors assessed the effects of valproate in patients with juvenile myoclonic epilepsy (57). Three age- and gender-matched groups (N = 20 in each group): (1) drug naive juvenile myoclonic epilepsy; (2) juvenile myoclonic epilepsy on valproate; and (3) healthy controls underwent overnight polysomnography. REM arousal indices were higher in juvenile myoclonic epilepsy patients, whereas the overall and NREM arousal indices were comparable between the three groups. CAP rate was higher in juvenile myoclonic epilepsy patients as compared to controls (p < 0.001). Duration of phase A and its subtypes (p < 0.001) was reduced in drug-naive patients as compared to the valproate group and controls. Finally, percentage of phase A1 (p = 0.003) was decreased and A3 (p = 0.045) was increased in drug naive patients as compared to valproate group and controls. As many of the described alterations were not seen in the valproate group, the authors suggest that antiepileptic medications, such as valproate, may beneficially modulate arousal instability in juvenile myoclonic epilepsy patients, and hence promote sleep quality and continuity.

The controversial impact of antiepileptic medication on arousal instability is discussed extensively in the session on sleep-related hypermotor epilepsy (SHE).

Sleep-related hypermotor epilepsy. In 2016, the syndrome previously known as nocturnal frontal lobe epilepsy (NFLE) was renamed as sleep-related hypermotor epilepsy (SHE) (107). Clinically, sleep-related hypermotor epilepsy is characterized by short-lasting seizures (less than 2 minutes) patterned by repetitive and stereotyped motor events in the same subject. Besides complex hypermotor seizures, paroxysmal arousals and minor motor events are the most common clinical manifestations, occurring most commonly during the night. Most seizures occur periodically during non-REM sleep. Differential diagnosis between sleep-related hypermotor epilepsy and parasomnias can represent a challenging issue.

In the large majority of patients, neurologic and neuropsychological assessment is normal (78). Although the name nocturnal frontal lobe epilepsy has historical significance, three critical issues justified the change. First, the term nocturnal was considered misleading because it implies a chronobiological pattern of seizure occurrence, whereas evidence indicates that occurrence in sleep is the most important characteristic, whether at night or during daytime naps. Second, the emphasis on localization to the frontal lobe was considered misleading in view of evidence that the characteristic seizures may also arise from other cerebral regions. Third, the original name did not specify the typical clinical semiology involved, which consists primarily of hypermotor seizures and may also include attacks with predominantly tonic or dystonic features (107).

Diagnosis of sleep-related hypermotor epilepsy cannot be excluded even when interictal and ictal features are lacking in the EEG, both during sleep and wakefulness. Clinical history is the starting point for the diagnosis of sleep-related hypermotor epilepsy. According to the new definition, diagnostic criteria of sleep-related hypermotor epilepsy are based on three levels of certainty: witnessed (possible) sleep-related hypermotor epilepsy, video-documented (clinical) sleep-related hypermotor epilepsy, and video-EEG-documented (confirmed) sleep-related hypermotor epilepsy.

New insights into the biology of sleep-related hypermotor epilepsy occurred with the discovery of an autosomal dominant form and identification of the first gene, CHRNA4, encoding a neuronal nicotinic receptor subunit (87).

Further investigation has ascertained that sleep-related hypermotor epilepsy shows similar clinical features both in familial and sporadic cases, although autosomal dominant inheritance is characterized by a marked intrafamilial variation in severity (74; 94; 12).

A retrospective study including 139 sleep-related hypermotor epilepsy patients with a 16-year median follow-up showed a poor outcome after a long follow-up, pointing to a symptomatic etiology as the most consistent determinant affecting prognosis. Lesional cases were underestimated in the cohort because of the limits of conventional neuroimaging in detecting subtle focal cortical dysplasia (45).

Although sleep-related hypermotor epilepsy may have diverse etiologies, it is defined by clinical manifestations (hypermotor seizures) possibly resulting from shared downstream mechanisms occurring during sleep/wake oscillation changes, suggesting a unique pathogenic network (96; 66).

Because ictal discharges may arise not only from frontal lobe (61; 82), but also from various extrafrontal areas (59), this might suggest the ictal involvement of common cortico-subcortical networks (05) or a release phenomenon of a stereotyped inborn fixed motor pattern (66).

Tassinari and colleagues claim that genetically determined motor behaviors essential for survival (feeding, locomotion, reproduction, etc.) are under the control of central pattern generators (CPGs) and neuronal aggregates located in the brain stem and spinal cord (97). During sleep, transient arousals triggered by epileptic events or brain dysfunction can “activate” or “release” CPGs responsible for involuntary behavioral patterns. In other words, activation or release of a CPG, whatever the nature of the trigger, leads to a common motor semiology.

Considering the CPG theory, it is possible that the genetic or sporadic alteration resides in the mechanism controlling the arousal system, explaining why sometimes, or in some subjects, we can expect arousal disorders (parasomnias) or epileptic seizures by the same activation but with different triggers: epileptic abnormalities or sleep-related dysfunctions (97; 61). In support of this hypothesis, Tinuper and colleagues collected some data on familial aggregation of patients with diagnosed hypermotor epilepsy and found a higher frequency of arousal parasomnias (clinically defined) in hypermotor epilepsy probands and their relatives compared with a control population (108; 115).

In NREM sleep, where muscle tone is still operative, EEG synchrony allows multiple levels of expression (from N1 to N3), and a variety of motor events can take place from seizures to parasomnias. Whether the outcome is a muscle jerk or a major epileptic attack will depend on a number of ongoing factors (sleep stage, delta power, motor chain, body position, etc.), but all events will share the common trait of arousal-activated phenomena (64). These findings suggest that arousal during sleep is the common condition for the onset of motor patterns that are already written in the brain codes (GPG) but require a certain degree of activation (arousal) to become visibly apparent. In this case, arousal acts as a trigger, releasing or facilitating an encoded “kinetic melody” (97; 96; 68). Exploiting stereo-EEG investigation during sleep, Gibbs and colleagues state that “involvement of anterior frontal regions gives rise to integrated gestural motor behaviors, distal stereotypies, manipulation/utilization and fixed facial expression when the rostral prefrontal ventrolateral regions and the rostral cingulate gyrus are involved early. Indeed, these areas have been shown to be essential for organizing and controlling goal-directed behaviors and emotional responses” (27).

A common feature is the onset of all episodes during NREM sleep, with different distribution with the sleep stages. Major attacks prevail in N3 leading abruptly to a wake condition. Paroxysmal arousals and minor motor events may recur every night, sometimes several times per night, arising mainly from CAP in stage 2 (99).

Patients affected by sleep-related hypermotor epilepsy frequently report a poor sleep quality, fatigue, and excessive daytime sleepiness ascribed to the presence of recurrent seizures and motor episodes during sleep. In these patients, enhanced sleep fragmentation and higher percentages of wakefulness are common polysomnography findings, as well as increased amounts of CAP rate (117). The presence of an epileptic focus in the frontal regions likely represents an internal disturbance that interferes subcontinuously with the stability of NREM sleep. Though extremely short-lived, the interictal epileptiform discharges in sleep-related hypermotor epilepsy is an activating event sharing the same pathways of arousal instability.

PSG recordings show that increased sleep instability is very common in sleep-related hypermotor epilepsy, particularly when multiple events occur during sleep. Some studies suggest that macrostructural sleep disturbances and arousal instability are part of the syndrome (66). In sleep-related hypermotor epilepsy, stereotyped manifestation not only involves motor behaviors, but also polysomnographic patterns (PSG) patterns (66a). Baseline PSG recordings of 40 patients (20 males and 20 females; mean age: 31 ± 10 years) with a diagnosis of sleep-related hypermotor epilepsy were compared with those of 24 age- and gender-balanced healthy subjects without sleep complaints (controls). Sleep patients showed a significant increase in wake after sleep onset, slow wave sleep duration, and REM latency, whereas REM sleep was significantly lower. Sleep-related hypermotor epilepsy patients also showed a significant increase of CAP time, CAP rate (72% vs. 32% in the control group), CAP cycles, and mean duration of CAP sequences. These findings were associated with a significant enhancement of all subtypes of the A phases of CAP (mainly subtype A1). A total of 139 epileptic motor events supported by video-polysomnography evidence were counted. NREM sleep included 98% of all seizures, which were especially abundant in the first sleep cycles, decreasing in frequency together with the progressive decline of deep sleep. Ninety percent of total NREM seizures occurred during a CAP sequence, and CAP-related seizures occurred in association with a phase A.

In sleep-related hypermotor patients, the robust enhancement of sleep instability was associated with an increased amount of all phase A subtypes of CAP, especially phase A1, without relevant changes of respective percentages. This feature differs from other sleep disorders with high values of CAP rates, such as obstructive sleep apnea syndrome, where an increase of subtypes A2 and A3 and, conversely, a significant reduction of phase A1, are observed (69; 72).

Studies of scalp mapping obtained with electromagnetic tomography during sleep show that slow wave activity (0.25 to 2.5 Hz), a main component of A1 phases of CAP, has a major representation over the frontal and prefrontal regions, whereas fast activities (7 to 12 Hz), mainly associated with phases A2 and A3, have a parieto-occipital prevalence (21). These findings suggest a primary role of the high-amplitude low-frequency CAP phase A components, that is, EEG delta activity or K-complexes, on the occurrence of ictal episodes (88; 102; 77; 62; 98; 99). In this perspective, subtypes A1 of CAP could be considered as frontal lobe arousals periodically causing an increase of cerebral excitability that triggers the onset of frontal seizures.

According to these premises, diagnosis of sleep-related hypermotor epilepsy based exclusively on video-recorded phenomena, without EEG support, is not satisfactory, especially for paroxysmal arousals and minor motor events (112). Moreover, a number of sleep parasomnias may occur during NREM sleep, and differentiating these conditions from sleep-related hypermotor epilepsy only through visual patterns can be problematic (116; 16; 18; 06). Therefore, only the integrated evaluation of nocturnal video-EEG features, PSG metrics, and underlying regulatory sleep processes can help to clarify diagnostic uncertainties. These integrative criteria for the diagnosis of sleep-related hypermotor epilepsy can supply precious information on the effects of antiepileptic treatment. Twenty patients with a clinical and video-PSG diagnosis of sleep-related hypermotor epilepsy underwent a clinical follow-up and performed a second video-PSG after effective antiepileptic treatment lasting for at least six months (15). Conventional sleep measures, CAP parameters, and objective video-PSG seizures were assessed in sleep-related hypermotor epilepsy patients before and after treatment and compared with of the results of 20 age- and gender-matched control subjects.

Antiepileptic therapy in sleep-related hypermotor epilepsy subjects included carbamazepine as single therapy in 12 subjects in association to topiramate in one subject or to levetiracetam in three subjects. Two subjects were only treated with topiramate, and two subjects were only treated with levetiracetam. All antiepileptic drugs were taken at bedtime in single daily administration. After six months of antiepileptic therapy, all sleep-related hypermotor epilepsy subjects treated with carbamazepine showed blood concentration of the drug within the therapeutic range. Seven out of 20 subjects were seizure free, whereas the remaining 13 reported a reduction of at least 50% of seizure frequency. Antiepileptic treatment determined a partial reduction of objectively recorded seizures (both major attacks and minor motor events) of approximately 25% compared to baseline condition. Most conventional sleep measures (ie, REM latency, wake after sleep onset, sleep efficiency) recovered normal values, but NREM sleep instability remained pathologically high (CAP rate +26% compared to controls), and were associated with persistence of daytime sleepiness (mean Epworth Sleepiness Scale score before therapy, 13 ± 3; after, 11 ± 3). Attenuation of sleep fragmentation due to major seizures in the first part of the night was the main factor leading to a more physiologic sleep structure during treatment. On the contrary, most CAP parameters continued to be pathologically high and far above control values in treated nocturnal frontal lobe epilepsy subjects. The residual high NREM sleep instability was probably related to the persistence of epileptic discharges that act as internal triggers of subcontinuous arousal fluctuations during NREM sleep. In turn, these arousal swings promote a gait effect on the occurrence of nocturnal motor events, especially in the form of minor motor events, which could be the expression of stereotyped innate motor sequences through arousal facilitation codified by central pattern generators both in untreated patients and during antiepileptic medication (97; 98).

The subjective effectiveness of antiepileptic treatment in sleep-related hypermotor epilepsy subjects could be mostly related to the partial reduction of the longer and more complex nocturnal seizures, which are more likely to be perceived (78). In contrast, the persistence of residual objective polysomnography epileptic phenomena and a high level of unstable NREM sleep indicate a partial resistance of both seizures and disturbed arousal system to the therapeutic action of the antiepileptic treatment.

So far, the principal target of antiepileptic drugs has been addressed to eliminate or decrease major epileptic events. However, despite the reduction of nocturnal seizures, the normalization of sleep efficiency and wake after sleep onset, and the high values of slow wave sleep, residual epileptic events, high levels of unstable NREM, impaired daytime vigilance persists under drug treatment, with negative impact also on the neurovegetative balance. An evaluation of the cardiovascular system during different A-phases of CAP during sleep in sleep-related hypermotor epilepsy patients and normal controls showed that the autonomic response is similar for all types of A-phases (A1, A2, A3). Regardless of the phase A subtype, activation patterns present a similar latency (4 seconds) at the minimum of the RR interval with respect to the A-phase onset. These findings indicate a biological price paid by inadequate control of arousal stability fueled by EEG paroxysmal discharges (29).

In animal models of sleep-related hypermotor epilepsy, where nicotinic acetylcholine receptors (nAChRs) are involved in seizure mechanisms, chronic administration of fenofibrate in the diet for 14 days significantly reduces or abolishes behavioral and EEG expressions of nicotine-induced seizures (79). Whether these findings may be confirmed in human sleep-related hypermotor epilepsy remains a stimulating challenge. Meanwhile, interest in orexin receptor antagonism as a novel mechanism of action against epilepsy is increasing. Indeed, loss of orexinergic activity is associated with REM sleep onset, and REM sleep is generally protective against seizures (58).

Sleep-related hypermotor epilepsy and NREM parasomnias: a continuum. NREM parasomnias or disorders of arousal encompass heterogeneous motor behaviors including sleepwalking, sleep terror, and confusional arousals, arising as a result of incomplete awakening. Differential diagnosis between disorders of arousal and sleep-related hypermotor epilepsy often represents a clinical challenge. The two conditions may be indistinguishable from a semiological point of view and the scalp video-polysomnography is often uninformative. In both disorders variable hypermotor manifestations range from major events to fragments of a hierarchical continuum of increasing intensity, complexity, and duration. Besides their semiological overlap, disorders of arousal and sleep-related hypermotor epilepsy share a number of common features in the sleep texture.

A study explores the commonalities and differences between sleep-related hypermotor epilepsy and disorders of arousal (54). In particular, disorders of arousal and sleep-related hypermotor epilepsy show similar amounts of sleep efficiency, light sleep (N1+N2), deep sleep (N3), REM sleep, and CAP subtypes (A1, A2, A3). Compared to healthy controls, both conditions also present slow wave sleep fragmentation (with higher amount of CAP rate in slow wave sleep) and an increased representation of stage N3 in the second part of the night. Whether the elevated percentages of N3 are an intrinsic feature (trait) of both disorders of arousal and sleep-related hypermotor epilepsy or the compensatory by-product of a nonconsolidated SWS due to motor events and/or EEG paroxysms occurring in NREM sleep (state) remains an open question. Probably, both assumptions stem from a common neurophysiological background.

Patients with disorders of arousal show more frequent and longer arousals and awakenings from N3 compared to controls. Impairment of sleep intensity and depth determines vulnerable and discontinuous slow waves during stage N3. Disorders of arousal sleep recordings are characterized by an excessive fragmentation of slow wave sleep independent of concomitant parasomniac manifestations. In the sleep-related hypermotor epilepsy recordings, the supplement of unstable sleep is fueled by a number of disturbing factors including ictal and interictal EEG paroxysms acting as noise equivalents on the neural circuitry.

Overall, the excess and/or an abnormal distribution of CAP interferes with the build-up and maintenance of slow wave sleep. Accordingly, the high amounts of slow wave sleep in the second half of the night in both sleep-related hypermotor epilepsy and disorders of arousal are probably the result of an adaptive intra-night homeostatic recovery of stage N3 due to disturbed and inadequate SWS consolidation occurring in the initial sleep cycles.

In disorders of arousal recordings, all motor episodes arose from NREM sleep: 37% during stages N1 or N2 and 63% during stage N3 (simple arousal movements: 94%). In sleep-related hypermotor epilepsy recordings, 57% of major attacks occurred during stage N3. According to a study the occurrence of at least one minor event during stage N3 is highly suggestive for disorders of arousal whereas the occurrence of at least one major event outside stage N3 is highly suggestive for sleep-related hypermotor epilepsy (76). However, in the same study, the number of major events in stage N3 per subject coincided in both disorders of arousal and sleep-related hypermotor epilepsy patients, suggesting that sleep staging is not a major element for the differential diagnosis. Probably, a different modulation of NREM stages on major and minor motor events in disorders of arousal and sleep-related hypermotor epilepsy patients is a more plausible statement. A close relation between minor motor events and arousal fluctuations is also a consolidated issue (88; 99). Therefore, besides classifying disorders of arousal and sleep-related hypermotor epilepsy according to the stage-distribution of nocturnal episodes, perhaps a greater attention on the unstable balance between arousal-promoting and sleep-promoting forces may provide additional information regardless of the ongoing sleep stage.

The impressive analogies between disorders of arousal and sleep-related hypermotor epilepsy suggest the existence of an underestimated continuum across the conditions, linked by increased levels of sleep instability, higher amounts of slow wave sleep, and NREM/REM sleep imbalance. The only discriminating elements involve the amounts of sleep length (more reduced in NREM parasomnias) and sleep instability (more elevated in sleep-related hypermotor epilepsy). Supportive data are provided by a study that showed an increased gamma band relative power and a lower slope of the EEG aperiodic component during stages N2 and N3 in sleep-related hypermotor epilepsy recordings compared to disorders of arousal (63). Overall, three pathophysiological hypotheses have been formulated to explain commonalities between sleep-related hypermotor epilepsy and disorders of arousal:

(1) Liberation: functional deinactivation of frontal lobe by subcortical nuclei (central pattern generators) due to variable external or internal stimuli (97);

(2) Dissociation: the simultaneous mixture of sleep and awake state becomes a major determinant for clinical manifestations (02; 100);

(3) Pathological: harmonizing the previous models, assumes the existence of a gain-of-function of frontal cortical acetilcholine receptors in both sleep-related hypermotor epilepsy and disorders of arousal, explaining their semiological differences according to underlying facilitatory circumstances (33).

Given their clinical and epidemiological overlap, a common genetic background is also hypothesized. In such a perspective, we suggest that the consolidated dichotomy disorders of arousal versus sleep-related hypermotor epilepsy should be reappraised.

Seizure outcomes, epilepsy prognosis, and sleep. Sleep quality has been associated with the degree of seizure control in patients affected by epilepsy (75). Insomnia prevalence ranges from 15% to 55% in adults affected with epilepsy, whereas poor sleep was reported in up to 72% of patients (39; 114).

Interestingly, there is evidence suggesting that melatonin in add-on with valproate might reduce seizure frequency while it improves sleep-quality in patients with generalized epilepsy (111; 47).

Sleep microstructure has been explored as a response biomarker in operated drug-resistant temporal lobe epilepsy (85). In fact, in basal conditions patients affected with temporal lobe epilepsy presented higher amounts of CAP oscillations, higher CAP rate in N2, higher A3, higher mean duration of A1, A3 index, and cycles in sequences compared to healthy controls. They also presented lower phase B and cycle mean duration. Starting from one month after epilepsy surgery, although no appreciable variations were noticed at sleep macrostructure level, a stable attenuation of sleep microstructure was observed. Hence, intriguingly, the amelioration of sleep features can be adopted to verify the efficacy of treatment in some forms of drug resistant epilepsy.

Sleep and antiepileptic drugs. Antiepileptic drugs (AEDs) might affect sleep architecture at various levels; however, most of available data are historical and sometimes lead to controversial results. Furthermore, the effect of antiepileptic drugs on sleep architecture had not always been explored in epileptic patients, and many data had been collected from healthy volunteers.

Carbamazepine (CBZ) initially reinforces SWS and decreases the representation of REM sleep (113). In chronic patients, the drug effects are less predictable, as in some cases it seems to favor sleep instability, whereas other observations suggest that it does not significantly affect sleep features (49; 28).

Valproate (VPA) monotherapy influences sleep microstructure in patients with juvenile myoclonic epilepsy; in particular, it apparently restores the physiologic duration of CAP cycles, which is shorter in drug-naive patients (56).

Lacosamide apparently reduces sleep fragmentation, lowering the arousal index, and it does not contribute to daytime sleepiness in treated patients (23).

Evidences on levetiracetam (LEV) are highly heterogenous. In a double-blind crossover study on 12 healthy volunteers assuming 2000 mg of levetiracetam/day, it seems that the drug significantly increases total sleep time, sleep efficiency, and time spent in non-rapid eye movement (NREM) sleep stages N2 and SWS. In parallel, stage shifts and wake after sleep onset (WASO) were significantly reduced (09).

The utilization of perampanel (PER) as add-on medication apparently ameliorates sleep architecture in adults affected with drug-resistant focal epilepsy (83). Benefits included increase in total sleep time, decrease in sleep latency, increase in sleep efficiency, and drop of wake after sleep onset. SWS was reinforced.

Pregabalin is among the commonest prescribed drugs to treat comorbid chronic insomnia in patients affected with epilepsy (03; 53). A double-blind randomized study in patients with focal epilepsy documented a significant reduction in nocturnal awakenings, decrease in wake after sleep onset, lastly yielding to reinforcement in sleep continuity and subjectively-perceived amelioration in sleep quality (14).

Finally, cannabidiol, which has been shown to be effective in reducing seizure frequency in Dravet syndrome, Lennox-Gastaut syndrome, and tuberous sclerosis, has been investigated for its role with respect to sleep architecture (43). The drug restored the physiological representation of graphoelements of lighter sleep stages (vertex waves, sleep spindles, and K-complexes), which are usually reduced in affected patients.

Overall, it seems that newer antiepileptic drugs present less interference with sleep structure compared to classic drugs, such as carbamazepine. Due to the key role of sleep in the dynamic course of epilepsy, randomized double-blind trials are urgently needed to clarify the role of these widely adopted drugs in sleep architecture.

Conclusions. Sleep is a powerful enhancer of epileptic features. For this reason, it is commonly used in clinical practice for diagnostic purposes. As a major agreement, it is well-established that synchronized NREM sleep facilitates seizures, whereas desynchronized REM sleep discourages seizure occurrence. The presence of nocturnal seizures affects the regular profile of the sleep architecture. In most cases, the immediate effect of an epileptic attack corresponds to an upward shift towards either awakening or a more superficial sleep stage. Enhanced sleep fragmentation and higher percentages of wakefulness and light sleep with a decrease in N3 and REM are common polysomnography findings. In addition, marked sleep instability is often observed in epileptic patients, even in the absence of nocturnal seizures. Overall, sleep-related epileptic attacks mostly affect the conventional sleep measures, whereas nocturnal interictal discharges basically have a destabilizing impact on CAP parameters (22).

Clinical vignette 1. A 50-year-old man arrived at our sleep disorders center following a pulmonary consultation. The patient referred snoring and sudden awakenings during the night associated with snoring. Daytime sleepiness was a heavy burden (Epworth Sleepiness Scale score: 13), and ENT evaluation showed deviated septum and enlarged turbinates. The hypothesis of a sleep apnea syndrome was explored by means of ambulatory monitoring and in spite of an apnea-hypopnea index (AHI) of 4/hour; a CPAP device was applied during nocturnal sleep. The patient continued to complain of disturbed sleep with several awakenings accompanied by choking gasps and facial grimaces, often followed by immediate CPAP mask removal. The report of snoring and the patient’s complaint of nonrestorative sleep persisted.

A neurologic consultation suggested a complete video-polysomnographic evaluation, which allowed diagnosis of nocturnal frontal lobe epilepsy triggered by a left frontocentral focus. Sleep efficiency was 77%. CAP rate was 58%. AHI was 3.5/hour. Most of the respiratory-related motor events were actually paroxysmal arousals misinterpreted as gasps and evidence of inadequate compliance to the CPAP treatment. Video-polysomnographic assessment also showed minor dystonic episodes of the right side of the body. Clinical and neuroradiological investigation was normal. The patient was regularly treated with carbamazepine 400 mg at bedtime. After a 5-year follow-up, nocturnal episodes and daytime sleepiness (Epworth Sleepiness Scale score: 8) were well-controlled by medication.

Clinical vignette 2. A paper describes the case of a 22-year-old male affected by NFLE reporting paroxysmal RLS-like symptoms (01). The patient was referred to the sleep center due to nocturnal paresthesias and cramps involving the left leg and leading to sleep fragmentation. At the age of 4, the patient presented with secondary generalized seizures preceded by left leg discomfort, controlled on carbamazepine. After successive therapy discontinuation, leg symptoms built up in frequency and duration until a secondary generalized seizure reoccurred. On prompt carbamazepine resumption, no further grand mal seizures occurred albeit persistence of nighttime frequent cramps and paraesthesia. Sleep EEG demonstrated asymmetric interictal sharp theta on the right posterior frontal areas, whereas brain MRI results were consistent with a Taylor type right frontal cortical dysplasia. Carbamazepine augmentation and add-on therapy with levetiracetam led to further frequency reduction of sensory symptoms.