Epileptic seizures do not occur randomly in time but rather their occurrence is confined to specific time windows, which in turn are shaped by slow brain rhythms related to the circadian clock.  This is the important novel insight, which was borne out of a collaboration between McGill University, the University of Nicosia Medical School, the Cyprus Institute of Neurology and Genetics, and the University of Cyprus and was published in the journal Human Brain Mapping (Mitsis et al, 2020).

Avgis Hadjipapas, Associate Professor in Neuroscience at the University of Nicosia Medical School and one of the authors of the publication, discusses this result. “We formed a multidisciplinary team comprised of engineers, neuroscientists and clinical neurologists. We analyzed electroencephalographic (EEG) data of patients with epilepsy, which were recorded continuously over periods ranging from multiple hours to multiple days. The length of the continuous recordings enabled us to investigate whether the EEG in the period between seizures was relatively constant and thus consistent with a single brain state, the so-called interictal state (the state between seizures). A relatively constant interictal state had been an implicit assumption in many studies, which had attempted to predict seizures from the patient’s past EEG. The approach was largely based on examining the EEG for only brief periods before the seizure. The current study, however, shows that the EEG in the interictal period is not constant but rather exhibits large intrinsic variations.  These variations can be attributed to circadian rhythms, the ubiquitous, roughly 24-hour long cycles found in most physiological processes of living beings, and their harmonic rhythms. These fluctuations did not only affect the EEG signals themselves but also to the pattern of correlations between EEG signals, which in turn reflect functional connections between different regions of the brain.  Strikingly, we found that these rhythmic variations were much larger than the seizure-related changes per se. We next asked whether these large variations could also influence the occurrence of seizures. To that aim, the timings of seizures were analyzed with respect to these rhythmic variations. Indeed, we found that seizure onsets occurred preferentially at particular time windows delineated by these rhythms.  These can be seen as windows of opportunity for seizures to surface, whereas outside these windows seizures are much less likely to occur. Importantly, it was the rhythmic variation in the functional connections between EEG signals that correlated well with seizure onsets, whereas no reliable correlation could be obtained with the signals themselves. What this might mean, is that the global, collective state of the brain is more important for the timing of seizures than the characteristics of the local EEG signal. This result opens avenues for investigating the role of this global state in the pathophysiology of seizures in greater depth. In addition, this work delivers an important insight in relation to seizure prediction from EEG, namely, that these slow rhythmic variations must be taken into account for accurate seizure prediction. Our results pave the way for future studies that in the first instance must replicate the results in larger samples of patients and seizures. In such studies, one could further establish whether the timing of seizures in individual patients can indeed be predicted reliably within a margin of hours, as hinted by our data in a relatively small sample of patients and seizures”.

 

Mitsis, GD, Anastasiadou, MN, Christodoulakis, M, Papathanasiou, ES, Papacostas, SS, Hadjipapas, A.  Functional brain networks of patients with epilepsy exhibit pronounced multiscale periodicities, which correlate with seizure onset. Hum Brain Mapp. 2020; 1– 18. https://doi.org/10.1002/hbm.24930