The occurrence of diurnal patterns in epilepsy has been recognized for centuries, and challenging sleep homeostasis by partial or total sleep deprivation is a common method used to improve the detection of interictal abnormalities [13, 14]. Since both the circadian clock and sleep–wake homeostasis regulate cortical excitability [15, 16] by influencing sleep–wake timing and the level of EEG activation during wake and sleep, it is not surprising that these factors may modulate EEG abnormalities. Thus, from a CTS perspective, detection of interictal epileptiform discharges (IEDs) or the transition from interictal to seizure onset during wakefulness should be minimal during the wake maintenance zone and maximal during circadian sleep promotion in the early morning (Fig. 2). On the other hand, from an SWH perspective, both IEDs and seizure probabilities should increase continuously with hours of wakefulness, especially in frontal brain areas. However, there are two caveats to linking cortical excitability to IEDs and seizure probability: (1) Cortical excitability is not formally defined in neuroscience and can be measured by various methods such as transcranial magnetic stimulation (TMS)–EEG (as in [15, 16]), but also by probing the network response to pulse stimulation (for example, [17]); (2) there is no simple association, such that when cortical excitability increases, so do IEDs and seizures.

To the best of our knowledge, only one study has separated the circadian and SWH influences on IEDs in five patients with idiopathic generalized epilepsy in a forced desynchrony protocol [18]. Figure 3 shows a patient with sufficient IEDs across the circadian cycle during the scheduled wake and sleep episodes from this study. In fact, the circadian waveform of IEDs, although small in amplitude, showed minima at the onset of melatonin (i.e., wake maintenance zone) and maxima at the peak of melatonin secretion (i.e., circadian sleep promotion). Interestingly, this was not the case for IEDs during non-REM sleep in this patient. First, IEDs were dramatically higher during scheduled non-REM sleep than during scheduled wakefulness episodes, confirming the IED-promoting effect of non-REM sleep. This is likely to be due to progressive synchronization within the thalamocortical network that enables the generation of non-REM sleep oscillations. Similar circuits are thought to be involved in the generation of spike-wave discharges in patients with generalized epilepsy [19, 20]. This increased synchronization during non-REM sleep eases the propagation of IEDs, increasing both their magnitude and spread [21]. Second, the percentage of IEDs was three times higher during daytime non-REM sleep than during nighttime non-REM sleep. In other words, if the patient was scheduled to sleep outside the window of endogenous melatonin secretion, the risk of IEDs tripled [18]. This underscores the importance of proper circadian entrainment (i.e., aligning sleep–wake timing with endogenous circadian rhythms) in people with idiopathic generalized epilepsy. Unfortunately, the authors of this study did not disaggregate the IED data by time awake or time asleep to confirm whether IEDs progressively increase with time awake or decrease with time asleep. Although IEDs are mainly favored by sleep, the transition from interictal to seizure onset also seems to be modulated by ultradian factors (i.e., non-REM–REM sleep cycles), with effects related to the location of the epileptic network or the type of epilepsy [22].

In everyday life, of course, both the CTS and SWH processes are not separate but occur simultaneously and interact on EEG activity. A 29‑h sleep deprivation study of TMS-evoked responses in the frontal cortex revealed nonlinear changes in gamma-band evoked oscillations, consistent with an influence of circadian timing on inhibitory interneuron activity [15]. Neural modeling indicated that the balance between brain excitation and inhibition is strongly influenced by circadian rhythms, demonstrating a daily regulation of brain balance dependent on circadian timing and sleep–wake history. At a time when circadian sleep promotion coincided with a significant amount of prior wakefulness (i.e., 22–26 h), maximal TMS-evoked gamma EEG activity occurred, indicating a high level of cortical excitability that could potentially increase the likelihood of interictal discharges and seizures in epileptic patients. Thus, sleep restriction or sleep deprivation combined with circadian sleep promotion in the early morning hours could have a negative effect on IEDs and seizures. This may underlie the observed increase in seizures, especially after long-haul flights arriving at an airport without optimal sleep [23]. In fact, a sleep restriction study of people with juvenile myoclonic epilepsy who were deprived of sleep from midnight until morning showed reduced intracortical inhibition and increased facilitation measured by TMS at 9 a.m. compared with normal controls, both of which are measures of increased excitation in the primary motor cortex [24].

Advances in technology, particularly long-term EEG recordings of neuronal activity, have confirmed diurnal and potentially circadian rhythms in epilepsy [25]. Periodogram analyses of interictal epileptiform discharge rhythms in the landmark study by Baud et al. [25] showed rather strong 24‑h and 12‑h rhythms in all patients with focal epilepsy, while weekly and monthly rhythms were less pronounced. Although it is not entirely clear from these studies whether the 24‑h component is diurnal rather than circadian and could also be sleep-related, as sleep–wake rhythms are also diurnal, the strong harmonic at 12 h suggests that these rhythms may have been driven also by the CTS rather than sleep alone. In addition, post hoc analyses of seizure occurrence in clinical settings showed a profound diurnal modulation, depending on the type of seizure, the brain location of seizure onset, and whether seizures occurred during wakefulness or sleep [21, 26, 27]. Absence seizures peaked from 9 a.m. to noon and 6 p.m. to midnight, predominantly during wakefulness. Atonic seizures peaked from noon to 6 p.m. and predominantly during wakefulness. Epileptic spasms peaked between 6 a.m. and 9 a.m. and between 3 p.m. and 6 p.m., also during wakefulness. Tonic-clonic seizures peaked from 6 a.m. to 9 a.m., although interestingly there were no significant differences between sleep and wakefulness [26].

An intracranial EEG study showed evidence of peaks from 11 a.m. to 5 p.m. in temporal lobe seizures, 11 p.m. to 5 a.m. in frontal lobe seizures, and 5 p.m. to 11 p.m. in parietal seizures [27]. When grouping by sleep–wake state, most seizures during wakefulness occurred from 11 a.m. to 5 p.m., while seizures during sleep peaked from 11 p.m. to 5 a.m. (Fig. 4; [27]). Although these temporal patterns show strong diurnal variation, it is not clear to what extent these modulations are influenced by the CTS but also by the timing and pharmacokinetics of antiepileptic drugs (AEDs) and individual sleep–wake timing (i.e., chronotype). Again, these patients probably slept less during the day than at night, as the CTS tends to have a strong effect on the sleep–wake cycle, promoting wakefulness during the day. In sum, all these findings and observations in healthy individuals and epilepsy patients highlight the important influence of circadian rhythm, sleep–wake history, and vigilance states on the balance between cortical inhibition and excitation, and the need to measure markers of CTS and SWH for better seizure prediction.

Diurnal temporal distribution of seizures while awake and during sleep. Data were extracted from Fig. 2 of Hofstra et al. [27] and averaged across different brain regions (i.e., mesial temporal, neocortical temporal, frontal, and parietal). Averaged number of seizures (while awake: n = 41 during 05:00–11:00; n = 94 during 11:00–17:00; n = 67 during 17:00–23:00; n = 20 during 23:00–05:00; while asleep: n = 54 during 05:00–11:00; n = 33 during 11:00–17:00; n = 45 during 17:00–23:00; n = 96 during 23:00–05:00)